Longitudinal Structural Brain Development in Externally Preterm

Longitudinal brain
development
in extremely preterm
newborns
Voor mijn ouders

C o lo p hon
L o n gi t u d i n a l b ra i n d eve lo pme nt in ex tre me ly pre te rm newb or ns
T h e si s , U tre c h t U n i ve rs ity, w ith a s ummary in Dutch

P roe f s c h r i ft, U n i ve rs i teit Utre ch t, me t e e n s ame nvatting in het Ned er la nd s
A u t ho r
Kar i n a J o s e p h a K e rs b erg e n
Cove r de sign & il l u st rat i o n

E st h e r B e e k m a n ( w w w.e sth e ro ntwe rpt.nl )
G rap h ic D e sign
R o n ald B o i te n & I re n e Me s u
(www.b o i te n g ra fi s c h o ntwe rpe rs .nl )
B ook P ro d u c t io n
B oi ten b o e k p ro j e c te n
P u bl i ca ti o n o f th i s th e sis w as s po ns o re d by th e Div is io n of Per ina tology of the University Med ica l C enter
U t re c h t, T h e S u rg i ca l Co mpany, Ch ips o f t, BMA BV ( MOSOS), Gilles Hond ius Found a tion a nd
Toshi b a M e d i ca l S y ste m s Ne de rl and.
A l l r i g h ts re s e r ve d . N o part o f th is th e s is may be re pro duced or tra nsmitted in a ny for m b y a ny mea ns,

wi t h o u t p e r m i ss i o n i n writing o f th e co pyrig h t ow ne r.
T h e co p y r i g h t o f th e a rticle s th at h ave be e n publ is h e d h as b een tra nsfer red to the resp ec tive jour na ls.
ISBN 97 8 - 90 - 3 93 - 62 8 8-4
longitudinal
brain development
in extremely
preterm
newborns
Longitudinale hersenontwikkeling bij extreem te vroeg geboren kinderen

(met een samenvatting in het Nederlands)
Proefschrift
ter verkrijging van de graad van doctor aan de Universiteit Utrecht op
gezag van de rector magnificus, prof.dr. G.J. van der Zwaan, ingevolge het
besluit van het college voor promoties in het openbaar te verdedigen op
dinsdag 24 maart 2015 des middags te 4.15 uur
door
Karina Josepha Kersbergen
geboren op 29 april 1986
te Assen
promotor e n
Pr of. d r . L . S . d e V r i e s
Pr of. d r . i r . M . A . V i e rg e v e r
Co pro motor e n
Dr . M. J. N . L . B end er s
Dr . F. G r oenenda a l
c on t e n ts
6 CHAPTER 1
Introduction
24 part 1 108 part 3

Qualitative assessment of magnetic White matter structure, diffusion
resonance imaging in relation to weighted imaging and network analysis
outcome to study brain development
26 chapter 2 110 chapter 6

Different patterns of punctate white matter Microstructural brain development between 30
lesions in serially scanned preterm infants and 40 weeks corrected age in a longitudinal
cohort of extremely preterm infants
46 part 2
Morphological changes during brain 140 chapter 7
development Corticospinal tract injury precedes thalamic
volume reduction in preterm infants with cystic
48 chapter 3 periventricular leukomalacia
Longitudinal brain development and clinical risk
factors in preterm infants 156 chapter 8
The neonatal connectome during preterm brain
72 chapter 4 development
Hydrocortisone treatment for bronchopulmonary
dysplasia and brain volumes in preterm infants 190 chapter 9
Summarizing discussion, conclusions and
84 chapter 5 directions for future research
Is early cortical folding related to clinical risk
factors and neurodevelopmental outcome? A 204 chapter 10
longitudinal MRI study of preterm newborns Nederlandse samenvatting
216 List of abbreviations

218 Co-authors and their affiliations
220 List of publications
221 Curriculum vitae
222 Dankwoord – Acknowledgements
5
1
ch a p t er
introduction
Part of this introduction has been adapted from

Manon J.N.L. Benders
Karina J. Kersbergen
Linda S. de Vries
Neuroimaging of white matter injury, intraventricular and cerebellar haemorrhage
Clinics in Perinatology 2014, 41:69-82
chapter 1
I
mprovements in neonatal care have led to increasing survival of even the most premature
infants. In the Netherlands, 547 infants were born alive between 24 and 28 weeks of
gestation in 2012 and the survival rate beyond 28 days for these infants was 74%.1 However,
long-term neurological sequelae are still common in survivors of preterm birth. Motor
impairments, most notably cerebral palsy (CP), have been reported in 3-20% and are most
often seen in the presence of severe white matter injury.2-7 At term equivalent age (TEA), the
development of CP can be reliably predicted with MRI, using a combination of conventional
T1- and T2-weighted imaging.8 Motor outcome has improved over the years, as shown by the
decreasing incidence and severity of CP across several cohorts, and this is thought to be
related to the decreasing incidence of severe white matter injury, especially cystic
periventricular leukomalacia (c-PVL).9,10 The percentage of infants diagnosed with milder, non-
cystic forms of white matter injury has decreased over the years as well.11
Impairments is cognitive and behavioural outcome and executive functions are nowadays the
most common sequelae from preterm birth and unfortunately, their incidence remains high,
with rates up to 50% in extremely preterm infants.2,5,6,12-14 These deficits will become more
apparent with age, as more complicated skills are needed. During childhood and into
adolescence and early adulthood, children born preterm show poorer academic achievement,
a higher need for special education and higher rates of behavioural problems compared to
term born peers.15-17 This seems to be inversely related to gestational age, with infants born at
the lowest gestational ages showing the highest rates of impairment.5,18 A relation with brain
abnormalities in the neonatal period is often found, but even those children without
neurosensory impairment and with a normal intelligence will still show more problems
compared to their term born peers.19,20 Although the negative predictive value of MRI for
motor impairments is high, predicting which infants will suffer from cognitive and behavioural
impairments in the absence of white matter injury is still an area of active research.21
Nowadays, the general consensus is that these impairments are likely the result of more
diffuse disturbances in brain development, that are not reliably predicted by visual evaluation
on conventional MRI.13,
The third trimester of a normal pregnancy is one of major brain development and maturation.
A four- to five-fold increase in overall brain growth, the development of the majority of gyri
and sulci, formation of the thalamo-cortical and cortico-cortical connections and the start of
myelination all occur during these 15 weeks.22-27 The brain is therefore extremely vulnerable
for disturbances during this period, such as premature birth and the subsequent treatment on
a Neonatal Intensive Care Unit (NICU). Many of these disturbances may not be readily visible
with conventional imaging and quantitative imaging techniques are therefore required to get
a better understanding of extra-uterine brain development during the third trimester.21,28 In
addition, longitudinal imaging during this period is of importance. Longitudinal imaging will
allow in vivo visualisation of the order and timing of brain development and its deviations,
makes it possible to study brain growth and can identify the differences between direct and
8
chapter 1
remote effects of brain injury. Rather than only studying the resulting effects of preterm birth
at term equivalent age, adding earlier imaging may show individual developmental
trajectories for each infant.
Over the last decade, several image post-processing techniques have been developed or were
adjusted for neonatal imaging, such as volumetric segmentations29-31 and techniques to
quantify cortical folding and thickness.25,32 Differences have been shown in brain volumes
between preterm born infants at TEA and term born peers, and these were still visible in
studies of ex-preterms in childhood and were related to outcome at that stage.33-35 Cortical
folding seems to be delayed in preterm infants and again, shows a relation to outcome.25,36
These initial results warrant further exploration, and extension to data before TEA.
In addition, more sophisticated MRI techniques, such as susceptibility weighted imaging
(SWI), diffusion weighted imaging (DWI) and functional MR-imaging (fMRI), have become
available for neonatal use.37-43 The SWI sequence is very sensitive for susceptibility changes
caused by e.g. blood, iron and calcifications, and can therefore be used to determine the
difference between haemorrhagic and non-haemorrhagic lesions. DWI and fMRI can provide
in vivo brain imaging on a more microstructural level, looking into grey and white matter
integrity and brain function, respectively.
A combination of these techniques with rigorous clinical and follow-up data may help to
further elucidate the underlying neural mechanisms associated with disrupted brain
development in preterm infants, and may make it possible to identify those infants at highest
risk of neurodevelopmental impairments for early neurodevelopmental intervention
programs to try and improve outcome, as well as for future neuro-protective studies.44
In this chapter, the most common forms of brain injury in preterm infants, and the way to
identify those on neuroimaging, will be described. At the end of this chapter, the aims of this
thesis will be outlined.
Neuroimaging of white matter injury, intraventricular

haemorrhage and cerebellar haemorrhage
Neuroimaging of preterm infants has become part of routine clinical care in the NICU.
Cranial ultrasound (cUS) is still considered the method of first choice for a sick preterm
infant in the NICU. Besides the advantage of being a bedside technique, another advantage
of cUS is that the examination can be performed as often as indicated, which allows
visualization of the evolution of the lesion. Without sequential cUS, it may be difficult to
make a distinction between a cyst following a periventricular haemorrhagic infarction
(PVHI) (Fig. 1) and c-PVL. In the past, performing sequential cUS has also helped to
recognize risk factors, including hypercapnia in germinal matrix–intraventricular
haemorrhage (GMH-IVH) and hypocarbia in c-PVL.45,46 MRI is increasingly performed in
extremely preterm infants at least once during the admission period, either during the
acute phase following stabilization, at discharge, and/or at TEA. MRI may show lesions,
I ntroduction 9
c h ap te r 1
which are not always easy to interpret and may be a cause of concern for parents. 47
Several studies have compared cUS and MRI contemporaneously. All investigators concluded that
cUS is highly effective in diagnosing severe white matter lesions such as c-PVL and PVHI, but that
MRI is superior in the identification of smaller cerebellar haemorrhages (CBH) and more subtle
Figure 1 MRI, T1-weighted image at term equivalent age and at 2 years of age in a preterm born infant (gestational age 26
weeks). A single cyst is seen adjacent to but not communicating with the lateral ventricle. A PVHI is noted on sequential cUS. Asym-
metry in myelination is seen (A). At 2 years, an interruption of the posterior limb of the internal capsule is seen as well as an area
of low signal intensity adjacent to the mildly dilated right ventricle (B, C). The child developed a mild unilateral spastic cerebral
palsy.
white matter injury (WMI), such as punctate white matter lesions (PWML) or diffuse extensive
high signal intensity (DEHSI), which may further improve prediction of outcome.48-51
I n t r av e n tr icu l a r h a em o r r h age
GMH-IVH is a frequent complication of preterm birth, occurring in 10-20% of preterm
infants.52,53 GMH-IVH will most often occur within the first 24 hours after birth and by 72 hours,
approximately 90% can be seen on cUS.52,53 Many different risk factors have been identified and
in addition to the well-known fluctuations in cerebral blood flow to the immature germinal
matrix microvasculature, multiple environmental and genetic factors may also affect the risk
for GMH-IVH through different pathways.52 Classification of GMH-IVH is generally based on
cUS and in addition, a difference between low grade IVH (grade I-II according to Papile,54 with
a haemorrhage in the germinal matrix or extending in the ventricle but without ventricular
distension) and high grade IVH (grade III-IV) is often made. According to Papile and
colleagues,54 a severe haemorrhage, grade III and IV, can be reliably diagnosed with cUS, even
though associated WMI and/or cerebellar lesions may not be detected. As the outcome of a
grade III haemorrhage (a large GMH-IVH with a blood clot filling the ventricle >50% and
resulting in acute dilatation of the lateral ventricle) is considerably better than that of a PVHI
10
c h ap te r 1
(sometimes still called a grade IV), these two should not be taken together. A large GMH-IVH,
complicated by posthaemorrhagic ventricular dilation (PHVD) and a PVHI, carries an increased
risk of adverse neurologic sequelae. PVHI may look like a large globular lesion communicating
with the lateral ventricle, with subsequent evolution into a porencephalic cyst (PC) (Fig. 2).
Figure 2 MRI, T2-weighted sequence. (A) Preterm infant, gestational age (GA) 30 weeks, small GMH-IVH on the right. (B, C) Pre-
term infant, GA 27 weeks showing a bilateral GMH-IVH, larger on the left and associated with a PVHI on the left. Also note a len-
ticulostriate infarct (open arrow) seen as a wedge shaped lesion in the left thalamus (B).
However, a PVHI may be more triangular with the apex of the lesion adjacent to, but not
necessarily in communication with, the lateral ventricle. This type of PVHI does not evolve into
a PC, but tends to evolve into one or more cysts, which remain separate from the lateral
ventricle (see Fig. 1).
As the amount of blood is an important risk factor for the development of PHVD, the chance of
developing PVHD is greater in infants with grade III than in those with a PVHI.55 The size of the
lateral ventricles can be measured with sequential cUS, which will help to determine when
best to intervene. The most commonly used measures are the anterior horn width (AHW) and
thalamo-occipital distance (TOD), according to Davies and colleagues,56 as well as the
ventricular width, according to Levene (Fig. 3).57
New reference values for the ventricular width, AHW, and TOD were recently published with
an age range of 24 to 42 weeks.58 When PHVD is severe, it may become difficult to access the
periventricular white matter with cUS (see Fig. 3C) because there will be compression of the
white matter and the white matter may no longer fit within the field of view. Because
associated WMI may be missed in infants with severe PHVD, MRI will be of additional value.
I ntroduction 11
chapter 1
Figure 3 Preterm infant, gestational age 25 weeks. cUS on day 3 (A) and day 12 (B, C) showing a large left-sided IVH and PVHI (A). PHVD
developed with severe dilatation of the left occipital horn (31 mm, C). Dotted lines A and C in figure B are measurements of ventricular width,
while B and D are measurements of the AHW. The dotted line in figure C is a measurement of the TOD.
Figure 4 Preterm infant, GA 25 weeks. MRI, T2-weighted

sequence coronal view, showing large bilateral cerebellar
haemorrhages on the scan performed at 30-weeks
postmenstrual age (A). The MRI performed at TEA, axial view
still clearly shows the haemorrhages as well as cerebellar
atrophy, especially on the right side (B). Another preterm
infant, GA 34 weeks, MRI, T2- weighted sequence, coronal
view with a few punctate lesions (C). Many more punctate
lesions were, however, identified in the cerebellum using
susceptibility weighted imaging (D).
Ce r e b e l l a r h a em o r r h age
The combination of cUS and MRI has led to an increased recognition of cerebellar injury in the
preterm infant, especially by a better recognition of subtle punctate lesions (Fig 4). The
cerebellum is rapidly developing (fivefold increase in volume from 24 to 40 weeks corrected
age) and, may therefore be especially vulnerable.59 Injury to the cerebellum is often related to
cognitive and behavioural deficits later in life.60,61 Previous studies have shown that large CBH,
detected on cUS, affect the sickest and most immature infants.62
The pathogenesis of large and small CBHs may be different. For the large CBH, it is described
that circulatory factors, such as a patent ductus arteriosus and the need for inotropic support,
12
chapter 1
in the presence of impaired cerebrovascular autoregulation, exposes the immature capillary
beds of the subependymal and subpial germinal matrix to fluctuations in arterial pressure. This
may lead to rupture of immature vessels.62 These circulatory risk factors were not found for
small CBHs. Primary CBH in preterm infants may originate in the external granular layer, from
the germinal matrices that are present in the subependymal region around the fourth
ventricle. The vulnerable capillaries, present in these structures, can easily rupture, especially
with the circulatory changes in the vulnerable preterm infant.63 Besides the actual lesion in the
cerebellar tissue, there are additional deleterious effects such as the toxicity of blood
breakdown products (e.g., haem and iron) in the cerebrospinal fluid because of
(supratentorial) haemorrhage. This iron accumulation and free radical attack is associated
with cerebellar atrophy.61,64,65
MRI studies are particularly valuable in the identification of cerebellar injury in premature
infants. CBH is more often seen in the presence of supratentorial cerebral lesions, such as a
large IVH with or without ventricular dilatation, PVHI, or WMI.65,66 Other investigators have
noted that unexplained ventricular dilatation can be seen in the context of CBH.67 It is
Figure 5 Preterm infant,

GA 26 weeks, T2-weighted
image, coronal views at 30
and 40 weeks postmenstru-
al age, with a large IVH with
right-sided PVHI, as seen on
the early scan (A). The sub-
sequent contralateral cere-
bello-cerebral diaschisis is
clearly visible on the
T2-weighted sequence at
40-weeks postmenstrual
age (A). Also note the deve-
lopment of a PC (B).
described that these supratentorial lesions are related to reduced cerebellar volumes due to
interruption of the cerebello-thalamo-cortical pathways. These pathways project from the
cerebellum to the contralateral cerebral cortex and vice versa. Disruption of supratentorial
neural systems, leading to crossed cerebello-cerebral diaschisis, might play a role in long-term
neurodevelopmental impairment (Fig. 5).68 Cerebellar volume of the preterm infant, assessed
at TEA, is associated with cognitive developmental outcome at 2 years corrected age.69
I ntroduction 13
c h ap te r 1
Figure 6 Preterm infant, GA

32+1 weeks, who developed
c-PVL. MRI, T2-weighted
sequence, coronal views,
performed on day 24. Note
both the PWML and
ventriculomegaly due to white
matter loss (A) and the
extensive cystic lesions (B).
Using diffusion tensor
imaging, the direction
encoded colour (DEC) map at
TEA shows an abnormal
development of the posterior
limb of the internal capsule
(PLIC) and optic radiation (C).
For comparison, a DEC image
with normal appearance of
the PLIC in a preterm infant at
TEA (D).
W hi t e matt er i nj u r y
Neuroimaging studies indicate that WMI in its various forms is by far the most common type of
injury in preterm infants, occurring in 50% or more of very low birth weight (VLBW) infants.70 In
the literature, WMI is often divided into two types: cystic and diffuse PVL.13 The cystic type
consists of focal necrosis deep in the periventricular white matter involving the loss of cellular
elements. Within several weeks, this generally evolves to multiple cystic lesions, known as
c-PVL, clearly visualized by cUS (Fig. 6). Currently, this severe lesion is less often observed (1-
3%) in VLBW infants.10,10 In noncystic WMI, which has also shown a decreasing incidence, injury
is most likely to be a more diffuse process.11 White matter necrosis is more often microscopic
14
c h ap te r 1
with paucity of myelinating (mature and immature) oligodendrocytes throughout the cerebral
white matter, which might evolve over several weeks to astrogliosis and microgliosis. Often,
this type of WMI is not reliably detected by cUS, although it may sometimes be suggested by
the presence of inhomogeneous (“patchy”) echogenicity. This form is referred to as noncystic
WMI.71
Performing sequential cUS is very important in detecting cystic WMI because the cysts will
become apparent after 2 to 4 weeks but they are often no longer seen at TEA. Instead,
ventriculomegaly suggestive of white matter loss can be seen on the TEA MRI. However, the
role of cUS is limited in detecting noncystic WMI. MRI, a more sensitive neuroimaging method,
shows more subtle and diffuse WMI.72,73 Noncystic WMI is often referred to as DEHSI and
PWML. White matter atrophy also is a frequent finding. Several recent papers have shown that
a very large proportion of preterm neonates show DESHI on their TEA MRI, especially when
imaged on a 3 Tesla scanner. When imaged after a postmenstrual age of 50 weeks, DESHI was
no longer identified. In addition, no relation with neurodevelopmental outcome is yet
found.74-76 Therefore, it is likely that DESHI is a prematurity-related developmental
phenomenon, instead of WMI, even though increased apparent diffusion coefficient (ADC)
values were reported by the group who coined the term DEHSI.72 PWML is suggested by
inhomogeneous echogenicity seen on cUS, but can only be reliably detected with MRI.
Pun ctate w hi t e m at t e r l e s i o ns
PWML are small, focal patches of different signal intensity, most often seen in the
periventricular white matter or intermediate white matter. Typically, they are seen as
hyperintense focal lesions on T1-weighted imaging and hypointense lesions on T2-weighted
imaging. The reported incidence ranges from around 20% to more than 50% of premature
infants.49,51,73,74,77,78
Several appearances are described. Lesions are arranged either linear, isolated, or in clusters,
and are located in the border of the ventricle, adjacent to the ventricles, in the intermediate
white matter or deeper in the centrum semiovale (Fig 7).48,77,79 The number of lesions can differ
greatly and severity is often judged by a cut-off. In infants imaged before term, more than
three lesions or lesions greater than 2 mm, but involving less than 5% of the hemisphere, are
considered moderate. When more than 5% of the hemisphere is involved, they are considered
severe. For infants imaged at TEA, a cut-off of more than six lesions is most often used to define
PWML as moderate-severe.48,51,73,80
The underlying pathophysiology of PWML in not yet completely understood and there is not a
lot of histologic material available.48,77,81-83 Concurrent with the differences in appearance, it is
reasonable to suspect several possible underlying mechanisms. Childs and colleagues
described pathologic states in two cases, with focal areas of gliosis corresponding with the site
of the PWML on MRI in one infant, whereas the other showed focal areas of ischemic
damage.48 Other histologic studies showed clusters of activated microglia and focal areas of
I ntroduction 15
c h ap te r 1
Figure 7 Examples of two

preterm infants with punc-
tate white matter lesions. In
A (age at scan 32+3 weeks),
lesions adjacent to the ven-
tricle are seen, as well as
some signs of venous conge-
stion on this T2-weighted
image. In B (age at scan 31
weeks), lesions are some-
what larger and more roun-
ded, and partially located
deeper in the white matter.
These lesions are clearly visi-
ble on the early T1-weighted
image.
hemorrhage.77,82,83 These limited data suggest that PWML is caused by both hypoxic-ischemic
and haemorrhagic processes. Apart from premature infants, PWML are also often found in
infants with congenital heart disease, even before cardiac surgery is performed.84,85 A larger
pathology study in this group showed diffuse gliosis in the white matter, suggestive of hypoxia-
ischaemia as an underlying cause.86
Neurodevelopmental outcome in infants with PWML so far seems favourable, although
outcome data beyond 2 years of age are still lacking. Lesion load and origin seem to play a role
in determining outcome. Infants with supposedly haemorrhagic lesions and with fewer than
six lesions of any kind at TEA are reported to perform well.77,78,80 Lower developmental indices
are reported in infants with supposedly ischemic lesions, a high lesion burden of more than six
lesions at TEA, and/or associated white matter lesions.74,76,77,79
D i ffu si on w e igh t e d a nd f u nct i o nal im agi ng

Apart from conventional T1-weighted and T2-weighted imaging, additional imaging
techniques can further increase sensitivity and may help improve the prediction of outcome.
In particular, diffusion-weighted imaging (DWI) and diffusion tensor imaging (DTI) are
increasingly being used. DWI is especially useful in the acute stage for detection of cytotoxic
oedema, which is present for about a week after the insult and may help to predict subsequent
cystic evolution.87 DTI can identify altered microstructure in the white matter of the
corticospinal tracts by evaluating fractional anisotropy (FA) values or fibertracking.88 DTI has
also been used to show brain maturation in preterm infants,40,89 and the differences in FA
between preterm infants at TEA and term born controls.40,90 These measurements seem to
correlate with outcome at two years of age.91 In childhood and adulthood, ex-preterm
16
c h ap te r 1
children still show lower FA values and these are correlated with cognitive outcome as
measured by IQ.92,93 In addition, structural connectivity can be studied using DTI and the
impact of preterm birth on for example thalamo-cortical connections has been shown using
this technique.23,24,28
Neonatal fMRI, and especially resting-state fMRI, has made it possible to study functional brain
activity, as well as the relation between structure and function and the brain connectome. The
development of resting state networks in the neonatal brain during the last trimester of
pregnancy has been described and the presence of those resting state networks at term age,
although still immature, has been shown by several groups.28,38,39,94 How resting state networks
mature during the last trimester of pregnancy and what influences preterm birth has on
functional connectivity, still needs to be determined.
Summa r y
In the VLBW infant, GMH-IVHs are still common with an incidence of 10% to 20%.55,56 Although
the incidence of severe WMI such as c-PVL in extremely preterm infants is declining, other
lesions such as CBH and PWML are commonly recognized, especially with the increased use of
MRI. Development of CP is explained by a severe haemorrhage (PVHI) or c-PVL seen on
neonatal cUS or MRI. The size and, especially, the site of the lesion will reliably predict whether
the infant is likely to develop neurologic sequelae.95,96 For the prediction of cognitive outcome
after preterm birth, more advanced MRI techniques are required.97 In recent years, a diverse
spectrum of neuronal-axonal disturbances involving thalamus, basal ganglia, and cerebral
cortex is commonly found in premature infants, usually in association with WMI.51 It is now
apparent that an additional and clinically important component of this neuronal-axonal
constellation is the involvement of the cerebellum.98 The relation between milder WMI such as
PWML and neurodevelopmental outcome remains uncertain and this must be evaluated in
serial cohort studies.
MRI enables non-invasive exploration of acute injury and post-injury alterations in structural
and functional connectivity associated with preterm birth. This might be of crucial
importance, not only for assessing long-term neurodevelopmental impairment, but also for
neuroprotective intervention trials. Advanced imaging techniques will provide data on the
disturbances of neuronal-axonal constellation by studying both connectivity and cortical
development in relation to WMI. Functional MRI connectivity, combined with advanced
imaging analysis tools, such as voxel-based morphometry and morphology-based analysis of
cortical folding, may provide complementary data for understanding development, injury, and
recovery in the developing brain.
Currently, MRI is the most sensitive non-invasive neuroimaging technique to assess brain
development, anatomic integrity, microstructural connectivity, and functional performance.
Therefore, it should be the method of choice to evaluate the effect of (future) neuroprotective
treatments, ranging from pharmacologic drugs to stem cell therapies.
I ntroduction 17
c h ap te r 1
Aims and outline of the thesis

The general aim of this thesis was to study longitudinal brain development in extremely
preterm infants. We conducted a prospective cohort study, including all infants born below 28
weeks gestation and admitted to our NICU between June 2008 and March 2013. For this study,
infants underwent MR imaging twice: once – if clinically stable – around 30 weeks of gestation
and again around term equivalent age. Conventional T1- and T2-weighted imaging, as well as
DWI and DTI were included in the MRI protocol and in part of the cohort SWI or resting-state
fMRI were also performed. All infants were seen in the follow-up clinic and were or will be
tested with the Bayley Scales of Infant and Toddler Development, third edition99 by a single
developmental specialist. Part of these infants was included in a European multicentre study
(Neobrain100) and was therefore seen at a corrected age of 2.5 years, and the remaining infants
were seen at two years’ corrected age.
In Chapter 2 two different patterns of punctate white matter lesions in preterm infants are
described. The addition of diffusion and susceptibility weighted imaging and their help in
discrimination between the patterns is discussed.
In Chapter 3 third trimester extra-uterine volumetric brain growth is evaluated with a newly
developed segmentation method that segments the brain into 50 brain regions. Detailed
growth data, as well as the influence of clinical risk factors on the longitudinal data, are
provided.
In Chapter 4 the influence of hydrocortisone, a corticosteroid often given to premature infants
at risk of developing chronic lung disease, on brain volumes at term equivalent age is assessed.
Chapter 5 describes the evaluation of the development of several sulci on longitudinal MRI
data and their relation to both clinical risk factors and outcome at two years’ corrected age.
For this study, sulci are automatically identified and a differentiation into specific sulci is
therefore possible.
In Chapter 6 serial DTI data are used to describe maturational changes on a microstructural
level in a sub-cohort of preterm infants without brain injury and with a normal
neurodevelopmental outcome at 15 months.
Chapter 7 describes the differences between infants with c-PVL and preterm born controls for
both tractography of the corticospinal tract and volumes of the thalamus throughout the
neonatal period. These measures are also correlated to later motor outcome and specifically
the development of CP.
In Chapter 8 the combination of longitudinal resting state fMRI and DTI data is used to study
the development of the macroscale connectivity network in preterm born neonates. The
spatial layout of structural and functional connectivity patterns is examined.
In Chapter 9 the findings of this thesis are summarized. Possible implications of this work as
well as suggestions for future research are discussed.
Finally, Chapter 10 provides a Dutch summary.
18
c h ap te r 1
References
1 Stichting Perinatale Registratie Nederland. 14 Wood NS, Marlow N, Costeloe K, Gibson AT,
Perinatale Zorg in Nederland 2012. Utrecht: Wilkinson AR. Neurologic and developmental
Stichting Perinatale Registratie Nederland; 2013. disability after extremely preterm birth. EPICure
2 Allen MC. Neurodevelopmental outcomes of Study Group. N Engl J Med 2000;343:378-84.
preterm infants. Curr Opin Neurol 2008;21:123-8. 15 Hille ET, Weisglas-Kuperus N, van Goudoever JB,
3 de Vries LS, van Haastert IL, Rademaker KJ, Koopman Jacobusse GW, Ens-Dokkum MH, de GL et al.
C, Groenendaal F. Ultrasound abnormalities Functional outcomes and participation in young
preceding cerebral palsy in high-risk preterm adulthood for very preterm and very low birth
infants. J Pediatr 2004;144:815-20. weight infants: the Dutch Project on Preterm and
4 Hart AR, Whitby EW, Griffiths PD, Smith MF. Small for Gestational Age Infants at 19 years of age.
Magnetic resonance imaging and developmental Pediatrics 2007;120:e587-e595.
outcome following preterm birth: review of current 16 Johnson S, Wolke D, Hennessy E, Marlow N.
evidence. Dev Med Child Neurol 2008;50:655-63. Educational outcomes in extremely preterm
5 Larroque B, Ancel PY, Marret S, Marchand L, Andre children: neuropsychological correlates and
M, Arnaud C et al. Neurodevelopmental disabilities predictors of attainment. Dev Neuropsychol
and special care of 5-year-old children born before 2011;36:74-95.
33 weeks of gestation (the EPIPAGE study): a 17 Saigal S, Doyle LW. An overview of mortality and
longitudinal cohort study. Lancet 2008;371:813-20. sequelae of preterm birth from infancy to
6 Latal B. Prediction of neurodevelopmental outcome adulthood. Lancet 2008;371:261-9.
after preterm birth. Pediatr Neurol 2009;40:413-9. 18 Lagercrantz H. The hard problem. Acta Paediatr
7 Wood NS, Costeloe K, Gibson AT, Hennessy EM, 2008;97:142-3.
Marlow N, Wilkinson AR. The EPICure study: 19 Nosarti C, Giouroukou E, Micali N, Rifkin L, Morris
associations and antecedents of neurological and RG, Murray RM. Impaired executive functioning in
developmental disability at 30 months of age young adults born very preterm. J Int Neuropsychol
following extremely preterm birth. Arch Dis Child Soc 2007;13:571-81.
Fetal Neonatal Ed 2005;90:F134-F140. 20 Saigal S, Hoult LA, Streiner DL, Stoskopf BL,
8 de Vries LS, van Haastert IC, Benders MJ, Rosenbaum PL. School difficulties at adolescence in
Groenendaal F. Myth: cerebral palsy cannot be a regional cohort of children who were extremely
predicted by neonatal brain imaging. Semin Fetal low birth weight. Pediatrics 2000;105:325-31.
Neonatal Med 2011;16:279-87. 21 Kwon SH, Vasung L, Ment LR, Huppi PS. The role of
9 Hamrick SE, Miller SP, Leonard C, Glidden DV, neuroimaging in predicting neurodevelopmental
Goldstein R, Ramaswamy V et al. Trends in severe outcomes of preterm neonates. Clin Perinatol
brain injury and neurodevelopmental outcome in 2014;41:257-83.
premature newborn infants: the role of cystic 22 Back SA, Miller SP. Brain injury in premature
periventricular leukomalacia. J Pediatr neonates: A primary cerebral dysmaturation
2004;145:593-9. disorder? Ann Neurol 2014;75:469-86.
10 van Haastert IC, Groenendaal F, Uiterwaal CS, 23 Ball G, Boardman JP, Rueckert D, Aljabar P, Arichi T,
Termote JU, Heide-Jalving M, Eijsermans MJ et al. Merchant N et al. The effect of preterm birth on
Decreasing incidence and severity of cerebral palsy thalamic and cortical development. Cereb Cortex
in prematurely born children. J Pediatr 2012;22:1016-24.
2011;159:86-91. 24 Ball G, Boardman JP, Aljabar P, Pandit A, Arichi T,
11 Gano D, Andersen SK, Partridge JC, Bonifacio SL, Xu Merchant N et al. The influence of preterm birth on
D, Glidden DV et al. Diminished White Matter Injury the developing thalamocortical connectome.
over Time in a Cohort of Premature Newborns. J Cortex 2013;49:1711-21.
Pediatr 2014. 25 Dubois J, Benders M, Cachia A, Lazeyras F, Ha-Vinh
12 Marret S, Marchand-Martin L, Picaud JC, Hascoet JM, LR, Sizonenko SV et al. Mapping the early cortical
Arnaud C, Roze JC et al. Brain injury in very preterm folding process in the preterm newborn brain.
children and neurosensory and cognitive disabilities Cereb Cortex 2008;18:1444-54.
during childhood: the EPIPAGE cohort study. PLoS 26 Kinney HC, Brody BA, Kloman AS, Gilles FH.
One 2013;8:e62683. Sequence of central nervous system myelination in
13 Volpe JJ. Brain injury in premature infants: a human infancy. II. Patterns of myelination in
complex amalgam of destructive and autopsied infants. J Neuropathol Exp Neurol
developmental disturbances. Lancet Neurol 1988;47:217-34.
2009;8:110-24. 27 Volpe JJ. Neurology of the Newborn. Chapter 2:
I ntroduction 19
c h ap te r 1
Neuronal Proliferation, Migration, Organization, and human newborn cerebral white matter assessed in
Myelination. 51-118. 2008. Philadelphia, Saunders, vivo by diffusion tensor magnetic resonance
Elsevier Inc. imaging. Pediatr Res 1998;44:584-90.
28 Counsell SJ, Ball G, Edwards AD. New imaging 41 Huppi PS, Dubois J. Diffusion tensor imaging of brain
approaches to evaluate newborn brain injury and development. Semin Fetal Neonatal Med
their role in predicting developmental disorders. 2006;11:489-97.
Curr Opin Neurol 2014;27:168-75. 42 Miller SP, Vigneron DB, Henry RG, Bohland MA,
29 Anbeek P, Isgum I, Van Kooij BJ, Mol CP, Kersbergen Ceppi-Cozzio C, Hoffman C et al. Serial quantitative
KJ, Groenendaal F et al. Automatic segmentation of diffusion tensor MRI of the premature brain:
eight tissue classes in neonatal brain MRI. PLoS One development in newborns with and without injury. J
2013;8:e81895. Magn Reson Imaging 2002;16:621-32.
30 Makropoulos A, Gousias I, Ledig C, Aljabar P, Serag A, 43 Neil JJ, Shiran SI, McKinstry RC, Schefft GL, Snyder
Hajnal J et al. Automatic Whole Brain MRI AZ, Almli CR et al. Normal brain in human
Segmentation of the Developing Neonatal Brain. newborns: apparent diffusion coefficient and
IEEE Trans Med Imaging 2014;PP:Epub ahead of diffusion anisotropy measured by using diffusion
print. tensor MR imaging. Radiology 1998;209:57-66.
31 Weisenfeld NI, Warfield SK. Automatic 44 Van Hus JW, Jeukens-Visser M, Koldewijn K, Geldof
segmentation of newborn brain MRI. Neuroimage CJ, Kok JH, Nollet F et al. Sustained Developmental
2009;47:564-72. Effects of the Infant Behavioral Assessment and
32 Hill J, Dierker D, Neil J, Inder T, Knutsen A, Harwell J Intervention Program in Very Low Birth Weight
et al. A surface-based analysis of hemispheric Infants at 5.5 Years Corrected Age. J Pediatr
asymmetries and folding of cerebral cortex in term- 2013;162:1112-9.
born human infants. J Neurosci 2010;30:2268-76. 45 Shankaran S, Langer JC, Kazzi SN, Laptook AR, Walsh
33 Peterson BS, Vohr B, Staib LH, Cannistraci CJ, M. Cumulative index of exposure to hypocarbia and
Dolberg A, Schneider KC et al. Regional brain hyperoxia as risk factors for periventricular
volume abnormalities and long-term cognitive leukomalacia in low birth weight infants. Pediatrics
outcome in preterm infants. JAMA 2000;284:1939- 2006;118:1654-9.
47. 46 Vela-Huerta MM, Amador-Licona M, Medina-
34 Peterson BS, Anderson AW, Ehrenkranz R, Staib LH, Ovando N, Aldana-Valenzuela C. Factors associated
Tageldin M, Colson E et al. Regional brain volumes with early severe intraventricular haemorrhage in
and their later neurodevelopmental correlates in very low birth weight infants. Neuropediatrics
term and preterm infants. Pediatrics 2003;111:939- 2009;40:224-7.
48. 47 Pearce R, Baardsnes J. Term MRI for small preterm
35 Inder TE, Warfield SK, Wang H, Huppi PS, Volpe JJ. babies: do parents really want to know and why has
Abnormal cerebral structure is present at term in nobody asked them? Acta Paediatr 2012;101:1013-
premature infants. Pediatrics 2005;115:286-94. 5.
36 Rathbone R, Counsell SJ, Kapellou O, Dyet L, Kennea 48 Childs AM, Cornette L, Ramenghi LA, Tanner SF,
N, Hajnal J et al. Perinatal cortical growth and Arthur RJ, Martinez D et al. Magnetic resonance and
childhood neurocognitive abilities. Neurology cranial ultrasound characteristics of periventricular
2011;77:1510-7. white matter abnormalities in newborn infants. Clin
37 Deipolyi AR, Mukherjee P, Gill K, Henry RG, Partridge Radiol 2001;56:647-55.
SC, Veeraraghavan S et al. Comparing 49 Debillon T, N’Guyen S, Muet A, Quere MP, Moussaly
microstructural and macrostructural development F, Roze JC. Limitations of ultrasonography for
of the cerebral cortex in premature newborns: diagnosing white matter damage in preterm infants.
diffusion tensor imaging versus cortical gyration. Arch Dis Child Fetal Neonatal Ed 2003;88:F275-F279.
Neuroimage 2005;27:579-86. 50 Maalouf EF, Duggan PJ, Counsell SJ, Rutherford MA,
38 Doria V, Beckmann CF, Arichi T, Merchant N, Groppo Cowan F, Azzopardi D et al. Comparison of findings
M, Turkheimer FE et al. Emergence of resting state on cranial ultrasound and magnetic resonance
networks in the preterm human brain. Proc Natl imaging in preterm infants. Pediatrics
Acad Sci U S A 2010;107:20015-20. 2001;107:719-27.
39 Fransson P, Skiold B, Horsch S, Nordell A, Blennow 51 Miller SP, Cozzio CC, Goldstein RB, Ferriero DM,
M, Lagercrantz H et al. Resting-state networks in the Partridge JC, Vigneron DB et al. Comparing the
infant brain. Proc Natl Acad Sci U S A diagnosis of white matter injury in premature
2007;104:15531-6. newborns with serial MR imaging and transfontanel
40 Huppi PS, Maier SE, Peled S, Zientara GP, Barnes PD, ultrasonography findings. AJNR Am J Neuroradiol
Jolesz FA et al. Microstructural development of 2003;24:1661-9.
20
c h ap te r 1
52 McCrea HJ, Ment LR. The diagnosis, management, Barkovich AJ et al. Cerebellar development in the
and postnatal prevention of intraventricular preterm neonate: effect of supratentorial brain
hemorrhage in the preterm neonate. Clin Perinatol injury. Pediatr Res 2009;66:102-6.
2008;35:777-92, vii. 65 Tam EW, Miller SP, Studholme C, Chau V, Glidden D,
53 Volpe JJ. Neurology of the Newborn. Chapter 11: Poskitt KJ et al. Differential effects of intraventricular
Intracranial Hemorrhage: Germinal Matrix- hemorrhage and white matter injury on preterm
Intraventricular Hemorrhage of the Premature cerebellar growth. J Pediatr 2011;158:366-71.
Infant. 517-588. 2008. Philadelphia, Saunders, 66 Argyropoulou MI, Xydis V, Drougia A, Argyropoulou
Elsevier Inc. PI, Tzoufi M, Bassounas A et al. MRI measurements of
54 Papile LA, Burstein J, Burstein R, Koffler H. Incidence the pons and cerebellum in children born preterm;
and evolution of subependymal and intraventricular associations with the severity of periventricular
hemorrhage: a study of infants with birth weights leukomalacia and perinatal risk factors.
less than 1,500 gm. J Pediatr 1978;92:529-34. Neuroradiology 2003;45:730-4.
55 Brouwer A, Groenendaal F, van Haastert IL, 67 Ecury-Goossen GM, Dudink J, Lequin M, Feijen-Roon
Rademaker K, Hanlo P, de Vries L. M, Horsch S, Govaert P. The clinical presentation of
Neurodevelopmental outcome of preterm infants preterm cerebellar haemorrhage. Eur J Pediatr
with severe intraventricular hemorrhage and 2010;169:1249-53.
therapy for post-hemorrhagic ventricular dilatation. 68 Tien RD, Ashdown BC. Crossed cerebellar diaschisis
J Pediatr 2008;152:648-54. and crossed cerebellar atrophy: correlation of MR
56 Davies MW, Swaminathan M, Chuang SL, Betheras findings, clinical symptoms, and supratentorial
FR. Reference ranges for the linear dimensions of diseases in 26 patients. AJR Am J Roentgenol
the intracranial ventricles in preterm neonates. Arch 1992;158:1155-9.
Dis Child Fetal Neonatal Ed 2000;82:F218-F223. 69 Van Kooij BJ, Benders MJ, Anbeek P, van Haastert IC,
57 Levene MI. Measurement of the growth of the de Vries LS, Groenendaal F. Cerebellar volume and
lateral ventricles in preterm infants with real-time proton magnetic resonance spectroscopy at term,
ultrasound. Arch Dis Child 1981;56:900-4. and neurodevelopment at 2 years of age in preterm
58 Brouwer MJ, de Vries LS, Groenendaal F, Koopman C, infants. Dev Med Child Neurol 2012;54:260-6.
Pistorius LR, Mulder EJ et al. New reference values 70 Volpe JJ. The encephalopathy of prematurity--brain
for the neonatal cerebral ventricles. Radiology injury and impaired brain development inextricably
2012;262:224-33. intertwined. Semin Pediatr Neurol 2009;16:167-78.
59 Chang CH, Chang FM, Yu CH, Ko HC, Chen HY. 71 Back SA. Perinatal white matter injury: the changing
Assessment of fetal cerebellar volume using three- spectrum of pathology and emerging insights into
dimensional ultrasound. Ultrasound Med Biol pathogenetic mechanisms. Ment Retard Dev Disabil
2000;26:981-8. Res Rev 2006;12:129-40.
60 Limperopoulos C, Bassan H, Gauvreau K, Robertson 72 Counsell SJ, Allsop JM, Harrison MC, Larkman DJ,
RL, Jr., Sullivan NR, Benson CB et al. Does cerebellar Kennea NL, Kapellou O et al. Diffusion-weighted
injury in premature infants contribute to the high imaging of the brain in preterm infants with focal
prevalence of long-term cognitive, learning, and and diffuse white matter abnormality. Pediatrics
behavioral disability in survivors? Pediatrics 2003;112:1-7.
2007;120:584-93. 73 Leijser LM, de Bruine FT, Steggerda SJ, van der Grond
61 Tam EW, Rosenbluth G, Rogers EE, Ferriero DM, J, Walther FJ, van Wezel-Meijler G. Brain imaging
Glidden D, Goldstein RB et al. Cerebellar findings in very preterm infants throughout the
hemorrhage on magnetic resonance imaging in neonatal period: part I. Incidences and evolution of
preterm newborns associated with abnormal lesions, comparison between ultrasound and MRI.
neurologic outcome. J Pediatr 2011;158:245-50. Early Hum Dev 2009;85:101-9.
62 Limperopoulos C, Benson CB, Bassan H, Disalvo DN, 74 de Bruine FT, van den Berg-Huysmans AA, Leijser LM,
Kinnamon DD, Moore M et al. Cerebellar Rijken M, Steggerda SJ, van der Grond J et al. Clinical
hemorrhage in the preterm infant: ultrasonographic implications of MR imaging findings in the white
findings and risk factors. Pediatrics 2005;116:717- matter in very preterm infants: a 2-year follow-up
24. study. Radiology 2011;261:899-906.
63 Volpe JJ. Neurology of the Newborn. Chapter 10: 75 Hart A, Whitby E, Wilkinson S, Alladi S, Paley M,
Intracranial Hemorrhage: Subdural, Primary Smith M. Neuro-developmental outcome at 18
Subarachnoid, Cerebellar, Intraventricular (Term months in premature infants with diffuse excessive
Infant), and Miscellaneous. 495-501. 2008. high signal intensity on MR imaging of the brain.
Philadelphia, Saunders, Elsevier Inc. Pediatr Radiol 2011;41:1284-92.
64 Tam EW, Ferriero DM, Xu D, Berman JI, Vigneron DB, 76 Jeon TY, Kim JH, Yoo SY, Eo H, Kwon JY, Lee J et al.
I ntroduction 21
c h ap te r 1
Neurodevelopmental outcomes in preterm infants: SD et al. Early detection of periventricular

comparison of infants with and without diffuse leukomalacia by diffusion-weighted magnetic
excessive high signal intensity on MR images at resonance imaging techniques. J Pediatr
near-term-equivalent age. Radiology 2012;263:518- 1999;134:631-4.
26. 88 Bassi L, Chew A, Merchant N, Ball G, Ramenghi L,
77 Cornette LG, Tanner SF, Ramenghi LA, Miall LS, Boardman J et al. Diffusion tensor imaging in
Childs AM, Arthur RJ et al. Magnetic resonance preterm infants with punctate white matter lesions.
imaging of the infant brain: anatomical Pediatr Res 2011;69:561-6.
characteristics and clinical significance of punctate 89 Berman JI, Mukherjee P, Partridge SC, Miller SP,
lesions. Arch Dis Child Fetal Neonatal Ed Ferriero DM, Barkovich AJ et al. Quantitative
2002;86:F171-F177. diffusion tensor MRI fiber tractography of
78 Dyet LE, Kennea N, Counsell SJ, Maalouf EF, Ajayi- sensorimotor white matter development in
Obe M, Duggan PJ et al. Natural history of brain premature infants. Neuroimage 2005;27:862-71.
lesions in extremely preterm infants studied with 90 Anjari M, Srinivasan L, Allsop JM, Hajnal JV,
serial magnetic resonance imaging from birth and Rutherford MA, Edwards AD et al. Diffusion tensor
neurodevelopmental assessment. Pediatrics imaging with tract-based spatial statistics reveals
2006;118:536-48. local white matter abnormalities in preterm infants.
79 Miller SP, Ferriero DM, Leonard C, Piecuch R, Neuroimage 2007;35:1021-7.
Glidden DV, Partridge JC et al. Early brain injury in 91 Van Kooij BJ, de Vries LS, Ball G, van Haastert IC,
premature newborns detected with magnetic Benders MJ, Groenendaal F et al. Neonatal tract-
resonance imaging is associated with adverse early based spatial statistics findings and outcome in
neurodevelopmental outcome. J Pediatr preterm infants. AJNR Am J Neuroradiol
2005;147:609-16. 2012;33:188-94.
80 Sie LT, Hart AA, van HJ, de GL, Lems W, Lafeber HN et 92 Allin MP, Kontis D, Walshe M, Wyatt J, Barker GJ,
al. Predictive value of neonatal MRI with respect to Kanaan RA et al. White matter and cognition in
late MRI findings and clinical outcome. A study in adults who were born preterm. PLoS One
infants with periventricular densities on neonatal 2011;6:e24525.
ultrasound. Neuropediatrics 2005;36:78-89. 93 Eikenes L, Lohaugen GC, Brubakk AM, Skranes J,
81 Felderhoff-Mueser U, Rutherford MA, Squier WV, Haberg AK. Young adults born preterm with very
Cox P, Maalouf EF, Counsell SJ et al. Relationship low birth weight demonstrate widespread white
between MR imaging and histopathologic findings matter alterations on brain DTI. Neuroimage
of the brain in extremely sick preterm infants. AJNR 2011;54:1774-85.
Am J Neuroradiol 1999;20:1349-57. 94 Smyser CD, Inder TE, Shimony JS, Hill JE, Degnan AJ,
82 Niwa T, de Vries LS, Benders MJ, Takahara T, Nikkels Snyder AZ et al. Longitudinal analysis of neural
PG, Groenendaal F. Punctate white matter lesions in network development in preterm infants. Cereb
infants: new insights using susceptibility-weighted Cortex 2010;20:2852-62.
imaging. Neuroradiology 2011;53:669-79. 95 Rademaker KJ, Groenendaal F, Jansen GH, Eken P, de
83 Rutherford MA, Supramaniam V, Ederies A, Chew A, Vries LS. Unilateral haemorrhagic parenchymal
Bassi L, Groppo M et al. Magnetic resonance imaging lesions in the preterm infant: shape, site and
of white matter diseases of prematurity. prognosis. Acta Paediatr 1994;83:602-8.
Neuroradiology 2010;52:505-21. 96 Roze E, Harris PA, Ball G, Elorza LZ, Braga RM, Allsop
84 Block AJ, McQuillen PS, Chau V, Glass H, Poskitt KJ, JM et al. Tractography of the corticospinal tracts in
Barkovich AJ et al. Clinically silent preoperative infants with focal perinatal injury: comparison with
brain injuries do not worsen with surgery in normal controls and to motor development.
neonates with congenital heart disease. J Thorac Neuroradiology 2012;54:507-16.
Cardiovasc Surg 2010;140:550-7. 97 Ment LR, Hirtz D, Huppi PS. Imaging biomarkers of
85 Ortinau C, Beca J, Lambeth J, Ferdman B, outcome in the developing preterm brain. Lancet
Alexopoulos D, Shimony JS et al. Regional Neurol 2009;8:1042-55.
alterations in cerebral growth exist preoperatively 98 Volpe JJ. Cerebellum of the premature infant: rapidly
in infants with congenital heart disease. J Thorac developing, vulnerable, clinically important. J Child
Cardiovasc Surg 2012;143:1264-70. Neurol 2009;24:1085-104.
86 Kinney HC, Panigrahy A, Newburger JW, Jonas RA, 99 Bayley N. Bayley Scales of Infant and Toddler
Sleeper LA. Hypoxic-ischemic brain injury in infants Development, 3rd edition. San Antonio, TX: Harcourt
with congenital heart disease dying after cardiac Assessment. 2006.
surgery. Acta Neuropathol 2005;110:563-78. 100 https://www.i-med.ac.at/neobrain. 28-4-2014.
87 Inder T, Huppi PS, Zientara GP, Maier SE, Jolesz FA, di
22
chapter 1
I ntroduction 23
1
pa r t
Qualitative
assessment of
magnetic resonance
imaging in relation
to outcome
2
ch a p t er
Different patterns
of punctate white
matter
lesions in
serially
scanned preterm
infants
Floris Groenendaal
Corine Koopman-Esseboom
Rutger A.J. Nievelstein
Ingrid C. van Haastert
Linda S. de Vries
PLoS One 2014 9:e108904

Abstract
Background and Purpose With the increased use of MRI in preterm infants, punctate white
chapter 2
matter lesions (PWML) are more often recognized. The aim of this study was to describe
the incidence and characteristics of these lesions as well as short-term outcome in a cohort
of serially scanned preterm infants, using both conventional imaging, diffusion (DWI) and
susceptibility (SWI) weighted imaging.
Materials and Methods 112 preterm infants with 2 MRIs in the neonatal period, with
evidence of punctate white matter lesions, were included. Appearance, lesion load,
location, and abnormalities on DWI and SWI were scored and outcome data were
collected.
Results Different patterns of punctate white matter lesions did appear: a linear appearance
associated with signal loss on SWI, and a cluster appearance associated with restricted
diffusion on DWI on the first MRI. Cluster and mixed lesions on the first scan changed in
appearance in over 50% on the second scan, whereas linear lesions generally kept their
appearance. Lesions were only visible on the early scan in 33%, and were only seen at term
equivalent age in 20%. Nine infants developed cerebral palsy, due to additional overt white
matter lesions in six.
Conclusion Two patterns of punctate white matter lesions were identified: one with loss of
signal on SWI in a linear appearance, and the other with DWI lesions with restricted
diffusion in a cluster appearance. These different patterns are suggestive of a difference in
underlying pathophysiology. To reliably classify PWML in the preterm infant in either
pattern, an early MRI with DWI and SWI sequences is required.
28 part 1 Q ualitative assessment of magnetic resonance imaging in relation to outcome

Introduction
W
ith the increased use of MRI in preterm infants, more subtle punctate lesions and
c h ap te r 2
diffuse white matter signal abnormalities are increasingly recognized.1,2 The
incidence of these punctate white matter lesions (PWML) varies widely, from less
than 10 to over 50%.1,3-7 Identification of risk factors responsible for PWML has proven to be
difficult and data on outcome are controversial.8-10 Most studies so far have either used
conventional T1-weighted and T2-weighted imaging or diffusion tensor imaging to study
PWML. Only a few studies have utilized additional sequences more sensitive to blood, blood
products, or gliosis.11-13
Obtaining a better understanding of the extent, appearance and possible underlying
pathophysiology of PWML is important for accurate counselling of parents of preterm infants,
and for a better selection of those infants that may benefit from neuroprotective strategies or
early rehabilitation services.
The aim of this study was to describe the appearance, location and evolution of PWML in a
cohort of serially scanned preterm infants utilizing conventional imaging as well as diffusion
weighted imaging (DWI) and susceptibility weighted imaging (SWI), and to correlate these
findings with neurodevelopmental outcome.
Methods
For this study, all infants born below a gestational age (GA) of 34 weeks and admitted to the
Neonatal Intensive Care Unit of the Wilhelmina Children’s Hospital between March 2007 and
March 2013, with at least two MRI scans in the neonatal period, were retrospectively
identified. The group of included infants consisted of two subgroups. Infants with a GA <28
weeks were routinely scanned based on their extreme prematurity, whereas infants with a GA
≥28 weeks were only scanned when serial cranial ultrasound examinations were suggestive of
brain injury. Infants were included in the present study if they had been reported to have
PWML on one or both scans. A total of 119 infants were identified. Seven infants had
congenital anomalies and were therefore excluded. The remaining 112 infants were eligible
for this study. Permission from the medical ethical review committee of the University Medical
Centre Utrecht (MERC UMC Utrecht) for the current study and informed parental consent for
the MRI was obtained. Patient data were anonymized prior to analysis. Since this was a
retrospective study, using MRIs performed as part of standard clinical care, oral consent for the
MRI was obtained by the treating physician and any questions or remarks were noted in the
charts. No written consent was deemed necessary. The MERC UMC Utrecht waived the need
for parental consent for both the use of medical data and for publication of medical images.
different patterns of punctate white matter lesions in serially scanned preterm infants 29
Magn e t ic R e s o na nc e Im agi ng
In the subgroup with a GA <28 weeks, MRI was performed around 30 weeks postmenstrual
age and again around term equivalent age (TEA). The subgroup with a GA ≥28 weeks was
c h ap te r 2
scanned as soon as possible after birth and again around TEA.

MRI was performed on either a 1.5 Tesla ACS-NT system (early n=9, TEA n=3) or a 3 Tesla (early
n=103, TEA n=109) whole-body Achieva system (Philips Medical Systems, Best, the
Netherlands). On the 1.5T magnet, the routine protocol included conventional inversion
recovery-weighted imaging and T2-weighted imaging (30 and 40 weeks: inversion recovery-
weighted repetition time [TR] 4147 ms; inversion time [TI] 600 ms; echo time [TE] 30 ms; slice
thickness 2 mm and T2-weighted TR 7656 ms; TE 150 ms; slice thickness 2 mm), as well as DWI
(single-shot echo planar imaging in 3 orthogonal directions; 25 slices; slice thickness 4 mm; TR
3700–5200 ms; TE 89 ms; b-values of 0 and 1000 mm2/s, no gap).
On the 3T magnet, the routine protocol included conventional 3D T1-weighted and T2-
weighted imaging (30 weeks: 3D T1-weighted TR 9.4 ms; TE 4.6 ms; slice thickness 2 mm and
T2-weighted TR 10085 ms; TE 120ms; slice thickness 2 mm; 40 weeks: 3D T1-weigthed TR 9.5
ms; TE 4.6 ms; slice thickness 1.2-2 mm and T2-weighted TR 4847-6293 ms; TE 120-150 ms;
slice thickness 1.2-2 mm), as well as DWI (single-shot echo planar imaging in 3 orthogonal
directions; 25 slices; slice thickness 4 mm; TR 2393 ms; TE 68 ms; b-values of 0 and 800 mm2/s,
no gap).
From 2008 onwards, the SWI sequence became available (3D gradient-echo sequence with
flow compensation, multishot echo-planar imaging; at 1.5T: TR 82 ms; TE 40 ms, at 3T: TR 52
ms; TE 30 ms, slice thickness 2 mm and EPI factor 3 in both). Since this sequence was not
included in the standard protocol, use was at the discretion of the attending neonatologist
and dependent on the time left in the scanning timeslot.
Infants were sedated using oral chloral hydrate 30 to 60 mg/kg. Heart rate, transcutaneous
oxygen saturation (Nonin Medical Incorporated, Minneapolis, MN, USA) and respiratory rate
were monitored. For hearing protection Minimuffs (Natus Medical Incorporated, San Carlos,
CA, USA) and Earmuffs (EM’s 4 Kids, Brisbane, Australia) were used. A neonatologist or
physician assistant was present throughout the examination.
C l i n ica l a nd fo l low - u p data

Clinical data were retrieved from the medical records. All infants were seen according to
clinical protocol in our follow-up clinic. At 15 months corrected age, the Griffiths Mental
Development Scales were used to assess neurodevelopmental outcome.14 Around 24 months
corrected age, either the Griffiths Mental Development Scales or, in infants with a GA<28
weeks, the Bayley Scales of Infant and Toddler Development, third edition were used.15 The
presence of cerebral palsy (CP), epilepsy, visual or hearing problems was also recorded.

Re-e va l u ati on o f t h e data
All scans were re-evaluated (KJK) using a combination of T1-weighted imaging, T2-weigthed
imaging, DWI and –when available– SWI. PWML were defined as small areas of high signal
c h ap te r 2
intensity (SI) on T1-weighted imaging and low SI on T2-weighted imaging. The number of
PWML as well as the appearance, location and laterality were scored, with a system based on
that of Cornette.5 For early MR scans, the number of lesions was divided in <3, <6 and ≥6 or
containing over 5% of a hemisphere, in agreement with the literature.3 For TEA scans, a division
of more or less than 6 lesions was used.
Appearance was divided in lesions with a linear appearance, an appearance of (solitary)
clusters or a mixed appearance containing a combination of both. Linear appearance was
defined as multiple lesions in close relation to another, with a linear organization, often in
close approximation to the ventricles. A cluster appearance was defined as more solitary
lesions, with a rounded shape and somewhat larger in size, which were often, although not
exclusively, located deeper in the white matter (see Figure 1 and Figure 2 for examples).
Location was divided in anterior, mid and posterior to the lateral ventricles and laterality was
scored as unilateral or bilateral involvement. The DWI sequences were evaluated on the
presence of lesions with restricted diffusion. When available, the SWI sequences were
evaluated on the presence of foci of signal loss outside the ventricles or the ventricular wall,
which did not have continuity suggestive of veins or venous congestion.
Scans were also scored for the presence of intraventricular haemorrhage (IVH) and other overt
white matter lesions. In infants with additional lesions, such as a unilateral periventricular
haemorrhagic infarction (PVHI), only PWML outside the area of the parenchymal lesion were
scored.
Stati st ica l a n a ly s e s
Statistical procedures were performed using R version 2.15.3 (www.r-project.org). First, chi
square tests were used to assess the relation between findings of PWML on the DWI or SWI
sequence and the appearance of PWML, the relation between IVH and appearance, and to
assess the effects of GA, and thereby the possible inclusion bias due to the two subcohorts.
The subcohorts, defined as having a GA above or below 28 weeks, were tested against findings
of PWML on the DWI or SWI sequence and the appearance, as well as the evolution of the
lesions between both scans and the occurrence of additional lesions. Possible differences in
clinical characteristics between the appearances were tested with one-way ANOVA tests. Next,
since the presence of lesions on DWI can be largely influenced by the time of scanning, day of
life of the first MRI scan, defined as more or less than 10 days after birth, was included in two
different logistic regression analyses along with cluster and/or linear appearance as
independent variables and with the presence of abnormalities on DWI or SWI as dependent
variables. Finally, both appearance and number of lesions were tested against outcome
measures. For this linear regression analysis, the presence of additional lesions was also taken
into account. A significance level of p<0.05 was used.
Results
A total of 112 infants were eligible for the study. In 62 infants with a GA <28 weeks, MRIs were
c h ap te r 2
part of routine clinical care. For the other 50 infants, with a GA ≥28 weeks, abnormal findings
on cranial ultrasound examinations were the reason to perform an MRI. Both types of PWML
Ta bl e 1. C linica l character i sti cs
Variable Total Linear Cluster Mixed p-value
N = 112 N = 67 N = 26 N = 19
Clinical parameters
Sex (male/female) 56/56 (50%) 34/33 (49%) 13/13 (50%) 9/10 (47%) 1.0
27.8 [24.4 – 28.2 [24.7 – 29.7 [26.1 –

Gestational age (weeks, mean [range]) 28.2 [24.4 –33.4] <0.01
33.4] 31.6] 32.7]
1101 [515 – 1164 [695 – 1354 [865 –

Birth weight (grams, mean [range]) 1158 [515 –2100] 0.02
2100] 1940] 2100]
Birth weight z-score (mean [range]) 0.4 [-1.7 – 2.0] 0.4 [-1.7 – 2.0] 0.4 [-1.6 – 1.5] 0.3 [-0.7 – 1.8] 0.9
Mechanical ventilation (no/yes) 33/79 (71%) 17/50 (75%) 9/17 (65%) 7/12 (63%) 0.5
Bronchopulmonary dysplasia (no/yes) 92/20 (18%) 54/13 (19%) 22/4 (15%) 16/3 (16%) 0.9
Patent ductus arteriosus (no/yes) 76/36 (32%) 44/23 (34%) 15/11 (42%) 17/2 (11%) 0.07
Necrotizing enterocolitis requiring

103/9 (8%) 63/4 (3%) 23/3 (12%) 17/2 (11%) 0.6
surgical intervention (no/yes)
Culture proven sepsis (no/yes) 85/27 (24%) 51/16 (24%) 18/8 (31%) 16/3 (16%) 0.5
Intraventricular haemorrhage (no/ 56/7/15/12/22 31/4/8/9/15 18/1/3/2/2

7/2/4/1/5 (63%) 0.08
grade I/ grade II/ grade III/ grade IV) (50%) (54%) (31%)
Posthaemorrhagic ventricular
dilatation requiring intervention (no/ 95/17 (15%) 53/14 (21%) 26/0 (0%) 16/3 (16%) 0.04
yes)
Scan parameters
Postmenstrual age at first scan 31.0 [26.6 – 31.2 [29.6 – 31.8 [29.6 –
31.2 [26.6 –34.7] 0.1
(weeks, mean [range]) 34.7] 33.7] 34.6]
Postmenstrual age at TEA scan (weeks, 41.2 [40.0 – 41.1 [40.1 – 41.1 [40.0 –
41.2 [40.0 – 43.6] 0.7
mean [range]) 42.9] 43.1] 43.6]
Postnatal age at first scan (days) 21 [0-50] 22 [0-50] 21 [5-44] 14 [2-44] 0.04
Lesions besides PWML (no/yes) 67/45 (40%) 35/32 (48%) 22/4 (15%) 10/9 (47%) 0.01
PWML on first scan (no/yes) 21/91 (81%) 16/51 (76%) 3/23 (79%) 2/17 (90%) 0.2
PWML on TEA scan (no/yes) 36/76 (68%) 28/39 (58%) 8/18 (69%) 0/19 (100%) <0.01
PWML on both scans (no/yes) 57/55 (49%) 44/23 (34%) 11/15 (58%) 2/17 (90%) <0.01
were clearly seen on 1.5 Tesla and 3 Tesla magnets, although only a small group of infants was
c h ap te r 2
imaged on a 1.5 Tesla magnet. Clinical characteristics, general scan parameters and outcome
data of all infants can be found in Table 1.
Follow-up parameters
Corrected age at 1st visit (months, 16.1 [12.6 – 16.5 [13.8 – 16.3 [14.7 –
16.2 [12.6 – 21.0] 0.7
mean [range]) 20.4] 19.4] 21.0]
N = 100 N = 59 N = 24 N = 17
Griffiths developmental quotient 1st

102 [9] 102 [8]# 103 [8] 102 [11] 0.8
visit (mean [SD])
Corrected age at 2nd visit (months, 25.6 [23.7 – 25.4 [23.2 – 24.1 [22.6 –
25.3 [22.6 – 31.3] 0.1
mean [range]) 31.3] 30.1] 26.4]
N = 83 N = 52 N = 18 N = 13
Griffiths developmental quotient 2 nd

99 [14] 93 [9] 117 [17] 93 [5] 0.004
visit (mean [SD])
N = 17 N=7 N=4 N=6
Bayley cognitive composite score 2nd

105 [13] 104 [13] 106 [15] 108 [10] 0.6
visit (mean [SD])
N = 65 N = 44 N = 14 N=7
Bayley total motor composite score

108 [12] 108 [10] 110 [16] 106 [14] 0.7
2nd visit (mean [SD])
N = 65 N = 41^ N = 13^ N=7
Cerebral palsy (no/yes) 103/9 55/4 22/4 18/1 0.01
With c-PVL 1 0 1 0
With PVHI 4 3 0 1
With PHVD 1 1 0 0
Without additional lesions 3 0 3 0
Infants are scored according to the appearance of their lesions on the early scan. For the 21 infants that only showed lesions at TEA,
the appearance at TEA is used.
#
In one patient, with a hemiplegia, the Griffiths test was not completed.
În four patients (three with linear lesions, one with cluster lesions) the motor subtest of the Bayley scales could not reliably be
completed.
A pp e a r a n c e
An overview of appearances, location, laterality and lesion load is given in Table 2 and of the
chapter 2
findings on DWI and SWI in Table 3, and in more detail in supplemental Table S1.
Ta b le 2 . PWM L characteri sti cs
Appearance Location
Linear Cluster Mixed Anterior-mid Posterior
Early (n=91) 51 (56) 23 (25) 17 (19) 85 (93) 1 (1)
TEA (n=76) 53 (70) 9 (12) 14 (18) 71 (93) 1 (1)
The values between brackets represent the percentages

Overall = diffusely located lesions, i.e. in all 3 of the mentioned regions.
Ta b le 3 . P W ML id e nti fi cati on u si ng addi ti onal i magin g metho ds
Early MRI (n=91)
SWI and
Appearance DWI + SWI + No SWI available
DWI +*
Linear 6 (7) 29 (32) 22 (24) 3 (3)
Cluster 10 (11) 6 (7) 10 (11) 1 (1)
Mixed 11 (12) 11 (12) 4 (4) 6 (7)

*The patients included in this column are also included in the columns ‘DWI +’ and ‘SWI +’
L inear P WML
The most common type of PWML was linear, in 56% of the early and 70% of the TEA scans
(Figure 1). Lesions were often more clear on T2-weighted than T1-weighted imaging. Location
of the lesions was anterior or adjacent to the ventricles in all infants. For lesions with low SI on
SWI, there was a significant association with a linear appearance (p=0.002) and with a scan
beyond the first 10 days after birth (p=0.02). At TEA, low SI lesions on SWI were less likely to
have a cluster appearance (p=0.004). The presence of an IVH was borderline significantly
associated with a linear appearance (p=0.047). Low SI lesions on early SWI were found more
often in infants with a GA <28 weeks (p=0.0007) but no relation between GA and linear
lesions on either scan was found.

C luster PWM L
Clusters of solitary lesions were seen in 25% and 12% on the early and TEA MRI, respectively
chapter 2
(Figure 2). In infants with a cluster appearance on the early MRI, MRI at TEA showed lesions
that were generally more florid on T1-weighted imaging as compared to T2-weighted imaging.
Laterality Lesion load
Overall Unilateral Bilateral 1-3
5 (6) 15 (16) 76 (84) 33 (36)
4 (5) 11 (14) 65 (86) 50 (66)
Term equivalent
MRI (n=76)
DWI + SWI + No SWI available
0 (0) 22 (29) 27 (36)
0 (0) 1 (1) 2 (3)
1 (1) 6 (8) 3 (4)
Location of the lesions was anterior or adjacent to the ventricles. For the early MRI, logistic
regression analysis showed a significant correlation between lesions with restricted diffusion
on DWI and both a cluster appearance (p=0.001) and a scan within the first 10 days after birth
(p<0.0001).
The presence of an IVH was not associated with lesions in a cluster appearance (p=0.3). On the
early scan, infants with a GA ≥28 weeks were more likely to have lesions in a cluster
appearance (p=0.006) and also more likely to have lesions with restricted diffusion on DWI
(p<0.0001).
c h ap te r 2
Figure 1 Punctate white matter lesions with a linear pattern. Early (A-C) and term equivalent (D-F) scans in an infant of 32
weeks gestation with lesions in a mixed, though mainly linear pattern. T2-weighted imaging shows bilateral lesions in the white
matter adjacent to the ventricles (A). Lesions are also visible on T1-weighted imaging, additionally showing a small subdural
haemorrhage (B). The arrow indicates one of the lesions, visible on all images. Also note the grade III intraventricular
haemorrhage. SWI shows signal loss in the areas of the lesions, suggestive of a haemorrhagic origin (C). Additionally, some small
frontal lesions can be identified, that do not show up on SWI, and show a cluster appearance. At term equivalent age, lesion load
has greatly diminished and the intraventricular haemorrhage has largely resolved on T2- and T1-weighted imaging (D, E). A
subcutaneous reservoir has been inserted to treat post-haemorrhagic ventricular dilatation. SWI still shows signal loss with
blood residue being most clearly visible on this sequence (F). Outcome was favourable with a cognitive composite score of 115
and a total motor composite score of 124 on the Bayley scales at two years corrected age.
Co m b inatio n o f lin e ar and clu ster PW M L

A combination of both types of lesions was seen in 19% and 18% on the early and TEA MRI,
respectively. Location of the lesions was anterior or adjacent to the ventricles in most, but in 5
infants, all with a high lesion load, lesions were also found in the posterior part of the brain. No
association with IVH was found in this group (p=0.3). Infants with a GA ≥28 weeks were more
likely to have lesions in a mixed appearance on the early scan (p=0.005).

c h ap te r 2
Figure 2 Punctate white matter lesions with a cluster pattern. Early (A-C) and term equivalent (D-F) scans in an infant of 31+3
weeks gestation with lesions in a cluster pattern. T2-weighted imaging shows multiple lesions throughout the white matter (A)
that are also clearly visible on inversion recovery imaging (B). The apparent diffusion coefficient -map shows restricted diffusion
at the site of the lesions (C). The SWI sequence (not shown) did not show any signal loss. The arrow indicates one of the lesions,
visible on all images. At term equivalent age, lesion load has diminished. Lesions are now better appreciated on T1-weighted
imaging (E), compared with T2-weighted imaging (D). Again, SWI does not show signal loss (F). Outcome was favourable with a
developmental quotient of 94 on the Griffiths scales at 18 months corrected age.
Evol u t i on
PWML were seen in 91 infants on the early MRI. Of those 91, PWML were still visible in 55
infants at TEA. In more than half of the infants with the lowest lesion loads on the early scan,
lesions could no longer be identified at TEA. The majority of these lesions had a linear
appearance. In an additional 21 infants, PWML were only seen at TEA, with <6 lesions in 18 and
≥6 lesions in three. For all infants with a high lesion load on the early scan, PWML were still
present at TEA. Lesion load was still high in 21 patients (75%). In the infants with the highest
lesion loads, appearance of PWML was cluster or mixed in 79%. Lesions on both scans were
more often identified in the group of infants with a GA ≥28 weeks (p=0.001, see also Table 2)
chapter 2
and in infants with lesions with a mixed appearance (p<0.001).

In the group of 55 infants with PWML on both scans, appearance differed between the scans in
24, as represented in Figure 3. Linear lesions only changed appearance in two out of 22 infants.
One infant had only one lesion left at TEA and was thus scored as solitary and the other
suffered from a parechovirus infection two weeks before his TEA scan, and showed lesions with
a mixed appearance.16 This was the only infant with lesions with restricted diffusion on DWI at
TEA. In the majority of infants (85%) PWML were bilateral at both time points. Infants with
unilateral lesions had either a very low lesion load or a unilateral PVHI with PWML on the
contralateral side.
A ddi ti on a l l es i o ns
Additional lesions were identified in 45 infants. These were post-haemorrhagic ventricular
dilatation in 8, PVHI in 25, cystic periventricular leukomalacia in 1, significant cerebellar
Figure 3 Changes in PWML appearance

over time. This figure depicts the changes
in appearance of PWML between both
scans. Early imaging is represented on the
left and term equivalent imaging on the
right. The number of infants per category
is depicted by the thickness of the arrow
and also printed above each arrow.
Especially cluster and mixed lesions will
change appearance, whereas linear
lesions remain linear in the majority of
the infants, or are no longer visible at
TEA. At the right side of the figure, infants
are depicted that only showed PWML at
TEA.

haemorrhage in 7, lenticulostriate infarction in 3, and an arachnoid cyst in 1 infant. More
lesions were found in the group with a GA ≥28 weeks (p=0.002), mainly due to the higher
incidence of PVHI. This can be explained by the difference in inclusion criteria. Infants with
chapter 2
lesions in a cluster appearance less often showed additional lesions compared to the other
two groups (p=0.01).
Neu r od e v e lop m enta l o u tco m e

Follow-up data at 15 months corrected age were available for 100 infants. Seven infants were
not seen and the remaining five were not yet 15 months of age. Follow-up was available for 83
infants around two years corrected age. Six infants were lost to follow-up and the remaining
23 were not yet two years of age. Outcome data are presented in Table 1. The majority of
infants performed within the normal range (defined as above -1 standard deviation) on both
motor and cognitive outcome. Both the relation between initial appearance (i.e., linear or
cluster) and between lesion load (i.e., < or ≥6 lesions) and neurodevelopmental outcome as
tested with the Bayley or Griffiths scales were explored. The only significant relation found was
between a cluster appearance and a higher developmental quotient on the Griffiths scales at
24 months corrected age (p=0.004). However, numbers were really small and this relation was
Figure 4 PWML in an infant without

additional lesions who developed
cerebral palsy. T1-weighted images at
early (born at 31 weeks, postmenstrual
age at scan 33 weeks, (A)) and term
equivalent age (B) in an infant with
lesions in a cluster pattern. Note that
the lesions seem to be directly in the
path of the corticospinal tracts (see
arrow for an example). Lesions show
high signal intensity, suggestive of
restricted diffusion, on DWI of the early
scan (C) and these are visible in the
same location on T1-weighted imaging
at term equivalent age (D). No signal
intensity changes were seen on SWI on
either scan and no additional lesions
were identified. At term equivalent age,
the posterior limb of the internal
capsule appears to be less well
developed on the right side. This infant
subsequently developed a mild
asymmetric bilateral spastic cerebral
palsy, with her left leg being most
severely affected.
skewed due to one outlier, an infant with a developmental quotient of 135. After exclusion of
this infant, no significant differences were found.
Nine infants developed mild CP. Seven had an asymmetrical posterior limb of the internal
c h ap te r 2
capsule on T1-weighted imaging at TEA and six had additional overt white matter injury. All
three infants without additional lesions who did develop CP had a cluster appearance on their
early MRI, with a high lesion load (Figure 4). Lesions showed restricted diffusion on DWI in all.
In two infants SWI was available, both without abnormalities. At TEA, lesions converted to a
mixed appearance in two and a linear appearance in one, and lesion load was still high in all.
Discussion
This is, to the best of our knowledge, the largest study so far describing sequential data of
PWML in preterm infants and the first to make a distinction between two patterns of PWML
with differences in appearance and evolution over time. These differences correlate with
findings on DWI and SWI, with a relation between a cluster appearance and restricted
diffusion on DWI, and a linear appearance and lesions with low SI on SWI.
The first linear pattern shows an often mild lesion load and lesions with low SI on SWI. The
second cluster or mixed pattern shows more often a high lesion load, lesions without low SI on
SWI and, depending on the time of scanning, restricted diffusion on DWI. In the infants with a
GA ≥28 weeks, the cluster and mixed patterns were more often identified compared to the
linear pattern.
These two patterns may reflect differences in underlying pathology. In this study, lesions with a
cluster appearance, similar to most lesions described in the literature, were found to have
normal SI on SWI, suggesting a non-haemorrhagic origin. When imaged within the first 2
weeks after birth, DWI findings often showed high SI with concurrent low SI on the apparent
diffusion coefficient map, suggesting an underlying restricted diffusion, with a possible
inflammatory or ischemic origin. At TEA, lesions with a cluster appearance on the early MRI
were most florid on T1-weighted imaging and generally more florid compared to lesions with
a linear appearance. This is in agreement with lesions described in some of the previous
studies with imaging at TEA, where lesions were also described to be most florid on T1-
weighted imaging, possibly due to early gliosis.3,7 In contrast, lesions with a linear appearance
were often found to have low SI on SWI, suggestive of a haemorrhagic origin. Only linear
punctate lesions adjacent to the medullary veins in the white matter, suggesting more than
just venous congestion, were scored. These linear lesions were seen more clearly on T2,
especially on early imaging. This may be due to haemoglobin degradation products, which
cause T2 shortening before they cause T1 shortening.5,17 Again, this is in agreement with some
of the previous studies.5,12,18 A few studies have already described a difference in types of
PWML and our data confirm these findings.11,19
Although findings on DWI are limited to the early scan, since a restricted diffusion will only be
visible during the first 10 to 14 days after injury, findings on SWI were visible on both scans.

Often, lesions were better visible on SWI and sometimes more lesions were seen, not
recognized on T1 or T2. Since restricted diffusion on DWI can also be caused by signal
distortion due to haemorrhage, the combination of both DWI and SWI will be most
c h ap te r 2
informative. Therefore, these sequences should become part of the standard protocol when
imaging very preterm infants.
In agreement with previous studies, lesion load differed between the two scans.1,6,20 Although
the majority of the lesions, especially those with a high lesion load, were already visible on the
early MRI, lesions had resolved in 33% of the infants and were first seen on the TEA scan in 20%.
Lesion load was generally lower in the infants with a GA <28 weeks, whose early scan was
performed later after birth compared to the older age group. Some lesions could already have
resolved before the early scan. Six out of 21 infants in whom lesions appeared between the
two scans had a sepsis or underwent surgery in the period between both scans, which may be
a possible explanation, comparable to late-onset cystic periventricular leukomalacia.21 Of
these six infants, four showed a linear, one a cluster and one a mixed pattern of PWML. In
infants with a high lesion load, lesions were often less clear or decreased in number between
the scans, with a milder pattern at TEA compared to the first MRI. Also, as shown in Figure 3,
the appearance of the lesions changed, especially for the lesions with a cluster and mixed
appearance on the early scan. This evolution may be due to the tissue reaction or partial
resolution of the lesions. These two findings suggest that serial MR imaging is important to be
able to identify the full extent and initial pattern of PWML, with the first MRI showing the full
extent of the PWML and the second scan the evolution and distribution in relation to the
corticospinal tracts.
Most lesions in this study were located adjacent to the ventricles, in the corona radiata and
centrum semiovale. Only five infants, with a high lesion load, also showed lesions in the
posterior white matter or along the optic radiation. This is somewhat different from previous
studies, where posterior lesions were reported in 23-53% of cases.5,20,22 A clear explanation for
this difference could not be found.
Due to the heterogeneity in inclusion criteria, neurodevelopmental outcome data in infants
with PWML differ between studies. High rates of CP have been reported in some studies,4
whereas others did not find this relation and described the presence and severity of additional
lesions to be most indicative of an abnormal outcome at two years.5,20 Whether the presence
of additional lesions was taken into account was not reported in all previous PWML related
studies and this may explain at least some of the differences in outcome. In our study, six out
of nine infants who developed CP had overt additional white matter lesions. In the other three,
PWML are likely to be responsible for the development of their CP. Lesions in those infants
appeared to be directly in the path of the corticospinal tracts and lesion load was high in all.
Also, all lesions had a cluster appearance with restricted diffusion on DWI, suggesting ischemia
as the underlying cause. None of the infants with a linear pattern developed CP.
PWML may have a more extensive effect on the development of the brain than can be seen on
conventional imaging. Two studies have shown a lower fractional anisotropy in infants with
PWML, suggesting an altered microstructure extending beyond the immediate area of injury,
possibly indicating a more diffuse injury to these tracts which is not visible on conventional
c h ap te r 2
MRI.22,23 Sie and colleagues described a follow-up imaging study in infants with PWML on
neonatal imaging, who were scanned again at 18 months of age. When only a few PWML were
present at TEA, no abnormalities on imaging were seen at 18 months and those children all
had a normal neurodevelopmental outcome. However, in the presence of more than 6 lesions
at TEA, the MRI at 18 months showed gliosis at the site of the lesions and some of those infants
developed CP, similar to the findings in this study.18 Lesion load may therefore be important
for prediction of neurodevelopmental outcome. Outcome data beyond two years of age are
currently lacking. Several authors have speculated that PWML may be related to the milder
forms of cognitive and behavioural problems found at school age in preterm infants and this
seems a likely hypothesis.5,18,20
There are several limitations to this study. First, due to the retrospective nature of our study,
our cohort consisted of two subgroups. All infants with a GA <28 weeks were scanned,
whereas those ≥28 weeks were only scanned when cranial ultrasound examination was
indicative of white matter injury, most often inhomogeneous periventricular echogenicity or
overt white matter pathology, such as PVHI. Therefore, we may well have missed the mild
lesions in infants ≥28 weeks gestation which may have skewed our findings. Also, infants <28
Ta b le S 1 . P W M L i magi ng characteri sti cs of the total co ho rt ,
a nd s eparately for i nfa nts ab ove and b elow 28 week s o f gestatio n
Appearance Location
Anterior-
Linear Cluster Mixed Posterior Overall
mid
Early (n=91) 51 (56) 23 (25) 17 (19) 85 (93) 1 (1) 5 (6)
<28 wks (n=49) 34 (70) 11 (22) 4 (8) 49 (100) 0 (0) 0 (0)
≥28 wks (n=42) 17 (40) 12 (29) 13 (31) 36 (86) 1 (2) 5 (12)
TEA (n=76) 53 (70) 9 (12) 14 (18) 71 (93) 1 (1) 4 (5)
<28wks (n=35) 26 (74) 5 (14) 4 (11) 34 (97) 0 (0) 1(3)
≥28 wks (n=41) 27 (66) 4 (10) 10 (24) 37 (90) 1 (2) 3 (7)

weeks were scanned around 30 weeks postmenstrual age and thus beyond the optimal timing
to see restricted diffusion on DWI.
Thirdly, SWI was introduced during the study period and was not done in all infants, limiting
c h ap te r 2
the number of infants in which a comparison between DWI and SWI could be made. Finally,
outcome data are not yet available for the total cohort and are restricted to two years of age.
In conclusion, two different patterns of PWML were seen in a large population of serially
scanned preterm infants on T1-weighted and T2-weighted imaging, with discordant findings
on DWI and SWI. These two patterns are suggestive of a difference in underlying pathology.
Evolution of the lesions between the two scans often showed a decrease in lesion load and a
change in appearance of the lesions. An early scan is therefore needed to be informed about
the full extent of the lesion load. For a more complete classification of PWML, both
conventional imaging and the combination of DWI and SWI are required. Long-term follow-up
studies are needed to determine whether this difference in patterns will also explain
differences in outcome.
Acknowledgements
The authors would like to thank the physician assistants and MR technicians for their help in
performing all MRIs, and Dr. Tetsu Niwa for his help in developing and implementing the SWI
sequence.
Laterality Lesion load
Unilateral Bilateral 1-3 4-6 >6 or >5%
15 (16) 76 (84) 33 (36) 29 (32) 29 (32)
8 (16) 41 (84) 30 (61) 12 (24) 7 (14)
7 (17) 35 (83) 3 (7) 17(40) 22 (52)
11 (14) 65 (86) 50 (66) 26 (34)
7 (20) 28 (72) 31 (89) 4 (10)
4 (10) 37 (90) 19 (46) 22 (54)
References
1 Debillon T, N’Guyen S, Muet A, Quere MP, with perinatal clinical data. Early Hum Dev
c h ap te r 2
Moussaly F, Roze JC. Limitations of 2009;85:111-5.

ultrasonography for diagnosing white matter 9 McCrea HJ, Ment LR. The diagnosis, management,
damage in preterm infants. Arch Dis Child Fetal and postnatal prevention of intraventricular
Neonatal Ed 2003;88:F275-F279. hemorrhage in the preterm neonate. Clin
2 Maalouf EF, Duggan PJ, Counsell SJ, Rutherford Perinatol 2008;35:777-92, vii.
MA, Cowan F, Azzopardi D et al. Comparison of 10 Benders MJNL, Kersbergen KJ, de Vries LS.
findings on cranial ultrasound and magnetic Neuroimaging of White Matter Injury,
resonance imaging in preterm infants. Pediatrics Intraventricular and Cerebellar Hemorrhage. Clin
2001;107:719-27. Perinatol 2014;41:69-82.
3 Miller SP, Cozzio CC, Goldstein RB, Ferriero DM, 11 Rutherford MA, Supramaniam V, Ederies A, Chew
Partridge JC, Vigneron DB et al. Comparing the A, Bassi L, Groppo M et al. Magnetic resonance
diagnosis of white matter injury in premature imaging of white matter diseases of prematurity.
newborns with serial MR imaging and Neuroradiology 2010;52:505-21.
transfontanel ultrasonography findings. AJNR Am 12 Niwa T, de Vries LS, Benders MJ, Takahara T,
J Neuroradiol 2003;24:1661-9. Nikkels PG, Groenendaal F. Punctate white matter
4 de Bruine FT, van den Berg-Huysmans AA, Leijser lesions in infants: new insights using
LM, Rijken M, Steggerda SJ, van der Grond J et al. susceptibility-weighted imaging. Neuroradiology
Clinical implications of MR imaging findings in 2011;53:669-79.
the white matter in very preterm infants: a 2-year 13 de Bruine FT, Steggerda SJ, van den Berg-
follow-up study. Radiology 2011;261:899-906. Huysmans AA, Leijser LM, Rijken M, van Buchem
5 Cornette LG, Tanner SF, Ramenghi LA, Miall LS, MA et al. Prognostic value of gradient echo T2*
Childs AM, Arthur RJ et al. Magnetic resonance sequences for brain MR imaging in preterm
imaging of the infant brain: anatomical infants. Pediatr Radiol 2014;44:305-12.
characteristics and clinical significance of 14 Griffiths R. The abilities of young child ren. A
punctate lesions. Arch Dis Child Fetal Neonatal Ed comprehensive system of mental measurement for
2002;86:F171-F177. the first eight years of life. London: The test agency
6 Dyet LE, Kennea N, Counsell SJ, Maalouf EF, Ajayi- Ltd. 1984.
Obe M, Duggan PJ et al. Natural history of brain 15 Bayley N. Bayley Scales of Infant and Toddler
lesions in extremely preterm infants studied with Development, 3rd edition. San Antonio, TX:
serial magnetic resonance imaging from birth Harcourt Assessment. 2006.
and neurodevelopmental assessment. Pediatrics 16 Verboon-Maciolek MA, Krediet TG, Gerards LJ, de
2006;118:536-48. Vries LS, Groenendaal F, van Loon AM. Severe
7 Leijser LM, de Bruine FT, Steggerda SJ, van der neonatal parechovirus infection and similarity
Grond J, Walther FJ, van Wezel-Meijler G. Brain with enterovirus infection. Pediatr Infect Dis J
imaging findings in very preterm infants 2008;27:241-5.
throughout the neonatal period: part I. 17 Gomori JM, Grossman RI, Goldberg HI,
Incidences and evolution of lesions, comparison Zimmerman RA, Bilaniuk LT. Intracranial
between ultrasound and MRI. Early Hum Dev hematomas: imaging by high-field MR. Radiology
2009;85:101-9. 1985;157:87-93.
8 Leijser LM, Steggerda SJ, de Bruine FT, van der 18 Sie LT, Hart AA, van HJ, de GL, Lems W, Lafeber HN
Grond J, Walther FJ, van Wezel-Meijler G. Brain et al. Predictive value of neonatal MRI with
imaging findings in very preterm infants respect to late MRI findings and clinical outcome.
throughout the neonatal period: part II. Relation A study in infants with periventricular densities

on neonatal ultrasound. Neuropediatrics periventricular leukomalacia in premature
c h ap te r 2
2005;36:78-89. infants: a threat until term. Am J Perinatol
19 Raybaud C, Ahmad T, Rastegar N, Shroff M, Al NM. 2001;18:79-86.
The premature brain: developmental and lesional 22 Bassi L, Chew A, Merchant N, Ball G, Ramenghi L,
anatomy. Neuroradiology 2013;55 Suppl 2:23-40. Boardman J et al. Diffusion tensor imaging in
20 Miller SP, Ferriero DM, Leonard C, Piecuch R, preterm infants with punctate white matter
Glidden DV, Partridge JC et al. Early brain injury in lesions. Pediatr Res 2011;69:561-6.
premature newborns detected with magnetic 23 Miller SP, Vigneron DB, Henry RG, Bohland MA,
resonance imaging is associated with adverse Ceppi-Cozzio C, Hoffman C et al. Serial
early neurodevelopmental outcome. J Pediatr quantitative diffusion tensor MRI of the
2005;147:609-16. premature brain: development in newborns with
21 Andre P, Thebaud B, Delavaucoupet J, Zupan V, and without injury. J Magn Reson Imaging
Blanc N, d’Allest AM et al. Late-onset cystic 2002;16:621-32.
2
pa r t
Morphological
changes during
brain development
3
ch a p t er
Longitudinal
brain development
and clinical
risk
factors
in preterm
infants
Antonios Makropoulos
Paul Aljabar
Floris Groenendaal
Linda S. de Vries
Serena J. Counsell
Submitted
Abstract
Introduction A better understanding of brain development in extremely preterm infants
may elucidate reasons explaining the high prevalence of neurodevelopmental deficits in
this population. The aim of this study was to investigate third trimester extra-uterine brain
growth and to correlate this to clinical risk factors.
Methods Preterm infants (gestational age <28 weeks) underwent brain MRI around 30
chapter 3
weeks postmenstrual age and again around term equivalent age (TEA). MRIs were
segmented with a recently developed method in 50 different regions covering the entire
brain. Multivariable regression analysis was used to determine the influence of sex, birth
weight z-score, brain injury and clinical variables on volumes at both scans and volumetric
growth.
Results MRI at TEA was available for 210 and serial data for 131 infants. Growth over these
10 weeks was shown for all regions and was largest for the cerebellum, with an increase of
258%. Sex, birth weight z-score and prolonged mechanical ventilation showed a global
effect on brain volumes on both scans. The effect of brain injury on ventricular size was
already visible at 30 weeks, whereas growth data and volumes at TEA mainly revealed the
effect of brain injury on the cerebellum.
Discussion This study provides data about third trimester extra-uterine volumetric brain
growth in preterm infants in 50 brain regions. Both global and local effects of several
common clinical risk factors were found to influence serial volumetric measurements,
highlighting the vulnerability of the human brain, especially in the presence of brain injury,
in this period.
50 part 2 morphological changes during brain development

Introduction
T
he last trimester of gestation is one of immense foetal brain growth and development.
Brain volume increases about five-fold and the majority of sulci and gyri formation takes
place in the last 15 weeks of pregnancy.1 Preterm infants will spend this period outside
the uterus, in a neonatal intensive care environment where they are exposed to a myriad of
factors that can possibly disturb these processes.2 Over the last decade, manual or (semi-)
automatic segmentation of brain volumes using T1-weighted and T2-weighted MRI has
c h apt e r 3
become available for neonatal imaging. Cross-sectional studies have shown a sharp linear
increase of brain volumes between 25 and 40 weeks of gestation, with a difference in regional
growth rates.3,4 Brain volumes at term equivalent age (TEA) are shown to be influenced by a
several clinical factors of which brain injury (BI), intra-uterine growth retardation (IUGR),
chronic lung disease and the use of dexamethasone are most often reported.5,6 A few studies,
with limited sample sizes, have studied the preterm brain longitudinally and have shown a
difference in growth rates for grey versus white matter, as well as regional growth
differences.7,8
The aim of this study was to investigate third trimester extra-uterine volumetric brain growth
and to correlate volumes with clinical risk factors in a longitudinally scanned cohort of
extremely preterm born infants.
Methods
Cl inica l data
Between June 2008 and March 2013, all preterm infants with a gestational age below 28
weeks, admitted to the level three Neonatal Intensive Care Unit of the Wilhelmina Children’s
Hospital were consecutively enrolled in a prospective neuroimaging study. For this study,
infants were scanned twice: once –if clinically stable- around 30 weeks gestation and again
around TEA (39-43 weeks). A total of 265 infants were eligible for inclusion. Figure 1 represents
the final inclusion of all infants.
Perinatal data were obtained prospectively. Birth weight z-scores (BWZ) were computed
according to the Dutch Perinatal registry reference data.9 Surgery (for necrotizing
enterocolitis, patent ductus arteriosus, insertion of a Rickham reservoir, retinopathy of
prematurity) was scored, taking timing of surgery into account in relation to when the MRI was
obtained, e.g. if the first surgery took place after the first MRI was done, it was only scored as
such for the TEA scan. Serial cranial ultrasound (cUS) was obtained and reported as part of
standard clinical care. Intraventricular haemorrhage (IVH) grading on cUS was scored
according to Papile.10 Permission from the medical ethics review committee and parental
consent for the MRI were obtained.
longitudinal brain development and clinical risk factors in preterm infants 51

chapt e r 3
Figure 1 Inclusion criteria. Flowchart showing the

inclusion of infants for this study.
MR I
MR imaging was performed on a 3.0 Tesla MR system (Achieva, Philips Medical Systems, Best,
the Netherlands). At 30 weeks, infants were scanned in an MRI compatible incubator (Dräger
MR Incubator, Lübeck, Germany and later Nomag® IC 3.0, Lammers Medical Technology
GmbH, Lübeck, Germany, with a dedicated neonatal head coil), while at TEA the sense head
coil was used. The protocol included T2-weighted imaging in the coronal plane (30 weeks:
repetition time [TR] 10085 ms; echo time [TE] 120 ms; slice thickness 2 mm and TEA: TR 4847
ms; TE 150 ms; slice thickness 1.2 mm).
After evaluation by a paediatric radiologist, all scans were re-assessed by two neonatologists
(LdV and MB) with more than 10 years of experience in neonatal neuro-imaging. The presence
of IVH, periventricular haemorrhagic infarction (PVHI), post-haemorrhagic ventricular
dilatation (PHVD, defined as a ventricular index 4 mm >97th centile, according to Levene11),
cystic periventricular leukomalacia (c-PVL), punctate white matter lesions, central or cortical
grey matter infarctions and punctate or larger lesions in the cerebellum were scored. The
presence of significant BI was defined as the presence of one or more of the following: IVH
grade III, PHVD, PVHI, c-PVL, large cerebellar haemorrhages destroying more than half of one
hemisphere, or infarctions of the central or cortical grey matter.

Figure 2 Example of the volu-
metric segmentation. Examples
of the automatic segmentation
method at 30 weeks (A) and
term equivalent age (B). The co-
lours depict the different brain
regions.
c h apt e r 3
Vol u me t r ic se g m e ntat i o n
T2-weigthed scans of both 30 week and TEA MRIs were segmented using the segmentation
method of Makropoulos and colleagues.12 This method utilizes an expectation-maximization
scheme that combines manually labelled, age-dependent atlases with intensity information
from the image to be segmented and has shown reliable results among infants scanned at
post-menstrual ages ranging between 28 and 44 weeks (see figure 2 for an example).12,13 All
segmentations were manually checked and small corrections were performed if necessary.
Stati st ica l a n a ly s i s
Statistical procedures were performed using Matlab (MATLAB and Statistics Toolbox Release
2013b, The MathWorks Inc., Natick, MA, USA) and R version 2.15.3 (www.r-project.org).
Growth was determined as the difference between the volume of a brain region at TEA and at
30 weeks, divided by the difference in scan age in weeks and subsequently multiplied by 10.
For the analyses at 30 weeks and TEA, all data were corrected for post-menstrual age at time of
scan. All analyses at 30 weeks and TEA were done with both absolute volumes and with
relative volumes corrected for total brain volume (TBV). First, a canonical correlation analysis
was performed to determine which clinical factors accounted for the largest variance in the
total set of 50 volumes, calculating the percentage of explained variance. For variables in the
clinical variate, results were compared across the time points and for the data with and
without correction for TBV. This led to a set of clinical variables being most influential across all
analyses, which was subsequently validated by repeating the analysis with random selections
of about half of the patients.
With the restricted group of clinical variables, multiple linear regression analysis was
performed to determine the effect of each clinical variable. To correct for the 50 different
regions that were analysed, a p-value of 0.001 (0.05/50) was considered significant.

Ta bl e 1. C linica l character i sti cs of al i nclu ded i nfan ts
Characteristic Total Serial scans

N = 210 N = 131
Sex (male/female) 110/100 66/65
Gestational age (wks, median [range]) 26.4 [23.9 – 28.0] 26.6 [24.4 – 27.9]
Birth weight (gr, median [range]) 878 [455 – 1450] 910 [460 – 1450]
chapter 3
Birth weight z-score (median [range]) 0.4 [-2.5 – 2.4] 0.5 [-2.5 – 2.3]
Small for gestational age (BW <p10) 16 (8%) 8 (6%)
Hypotension requiring inotropes 74 (35%) 43 (33%)
Patent ductus arteriosus requiring treatment 109 (52%) 61 (47%)
Mechanical ventilation >7 days of life 100 (48%) 50 (38%)
Chronic lung disease 81 (39%) 34 (26%)
Culture proven sepsis 79 (38%) 42 (32%)
Necrotizing enterocolitis requiring surgical intervention 18 (9%) 12 (9%)
Surgery during NICU admission 61 (29%) 31 (24%)
Treatment with morphine 127 (61%) 68 (52%)
Brain injury* 45 (21%) 27 (21%)

IVH III, no PHVD 8 4
PHVD 23 13
PVHI 14 12
c-PVL 1 1
Large cerebellar haemorrhages 4 2
Grey matter infarctions^ 8 8
Scan parameters
Postmenstrual age at 30wk scan (wks, median [range]) 30.6 [28.7 – 33.3] 30.6 [28.7 – 33.1]
Postmenstrual age at TEA scan (wks, median [range]) 41.0 [39.3 – 43.7] 41.0 [40.0 – 42.7]
Weight at TEA scan (gr, median [range]) 3315 [1685-4715] 3365 [2045 – 4715]
Head circumference at TEA scan (cm, median [range]) 35.0 [30.0-39.0] 35.2 [31.0 – 38.5]
* Some patients had several diagnoses of brain injury

^ One infant had a larger cortical grey matter infarct, all others had lenticulostriate infarctions.

Results
Clinical characteristics of the included infants are summarized in table 1. Infants that were
scanned serially had lower rates of pulmonary problems and sepsis, since they needed to be
clinically stable to be scanned early. After canonical correlation analysis, the 6 clinical factors
that were most influential were sex, BWZ, mechanical ventilation for > 7 days (MV7), use of
morphine, surgery and BI.
chapter 3
Grow th
For the 131 infants with serial scans, growth data could be calculated. Figure 3 shows the
percentage of growth between the two scans for each studied brain region, and also for TBV.
The cerebellum showed the largest increase (258%) whereas total brain volume increased by
140% and the ventricles only by 61%. Generally, the central regions showed less growth
compared to the cortical regions. Brain regions in the temporal and occipital lobes showed
more growth compared to the frontal and parietal regions.
Figure 3 Brain growth. This figure shows the percentage of growth between the two scans for each brain region. Data are
sorted by percentage increase in descending order. The increase in total brain volume is depicted at the bottom. The bilateral
cerebellum shows the largest increase, whereas the ventricles increased the least. In general, cortical regions seem to increase
more compared to central regions.

chapter 3
Figure 5 Effect of brain injury. This figure shows

the brain regions significantly affected by brain
injury. On the top left side, the brain regions in
which growth is affected (A) are depicted, with the
differences in percentage of increase on the x-axis.
The absolute volumes for 30 weeks (B) and term
equivalent age (C) are shown on the middle row,
with differences in cubic centimetres shown on the
x-axis. The bottom row shows the data for brain
regions corrected for total brain volume (D, E),
with the difference in percentages on the x-axis.
The bar colours represent patients with (blue),
and patients without brain injury (orange).

Figure 4 Effect of clinical variables on brain growth. The
effect of clinical variables on growth between serially
acquired scans. For all significant brain regions, the
colour indicates effect size, with the axis representing the
effect size in percentages. Sex (A, with girls having
smaller brains) influences growth mainly in the occipital
regions. Long duration of mechanical ventilation (B) is
significant around the insula and in the
parahippocampal, left temporal and cingulate gyri.
Surgery (C) is only significant in the left amygdala.
chapter 3

Ta bl e 2. S ig n if ica nt b rai n re gi ons after correcti on fo r total brain vo lume
Characteristic Brain region Effect size (%) p-value
Growth data
Surgery Left amygdala -18.8 0.0001
Brain injury Left hippocampus -16.5 0.0006
Left cerebellar hemisphere -12.6 <1*10-5

chapter 3
Right cerebellar hemisphere -15.4 <1*10-5
Left thalamus -8.4 0.0002
Corpus callosum -13.9 0.0003
After exclusion of infants with BI
Sex Left hippocampus -12.2 <1*10-5
Right anterior part of the lateral occipito-temporal gyrus – gyrus

-8.9 0.0002
fusiformis
Corpus callosum -10.4 0.0006
MV7 Right hippocampus -9.6 0.001
Right posterior part of the lateral occipito-temporal gyrus – gyrus

-9.1 0.0002
fusiformis
Left anterior part of the cingulate gyrus -10.9 0.0009
30 weeks
BWZ Right hippocampus -4.8 <1*10-5
Right insula 2.3 <1*10-5
MV7 Right frontal lobe -1.5 0.0005
Surgery Left frontal lobe -3.5 <1*10-5
BI Left caudate nucleus -11.1 0.0001
Left lentiform nucleus -4.3 0.0009
Corpus callosum 9.9 <1*10-5
BWZ Right insula 2.3 <1*10-5
Right hippocampus -4.8 <1*10-5
Right lateral part of the anterior temporal lobe -5.1 0.0004
Morphine Right lateral part of the anterior temporal lobe 10.3 0.0006

Term equivalent age
Sex Left posterior part of the parahippocampal and ambiens gyri 2.6 0.0007
BWZ Left lateral part of the anterior temporal lobe -4.7 0.0005
Right anterior part of the parahippocampal and ambiens gyri -2.2 0.0001
MV7 Right occipital lobe 3.1 0.001
Right posterior part of the medial and inferior temporal gyri 2.6 0.0004
chapter 3
BI Left cerebellar hemisphere -8.0 <1*10-5
Right cerebellar hemisphere -9.5 <1*10-5
Right anterior part of the medial and inferior temporal gyri 4.2 0.0001
Left posterior part of the parahippocampal and ambiens gyri -5.1 <1*10-5
Right posterior part of the parahippocampal and ambiens gyri -5.4 <1*10-5
Left caudate nucleus -6.1 0.0004
BWZ Left lateral part of the anterior temporal lobe -4.7 0.0009
Right anterior part of the parahippocampal and ambiens gyri -2.2 0.0002
Sex Left subthalamic nucleus -2.8 0.0005
Right subthalamic nucleus 3.4 0.0001
Morphine Right posterior part of the medial and inferior temporal gyri 2.6 0.0008
Note that at TEA, some of the brain regions show effects in the direction opposite from those of the absolute volumes, possibly
indicating specific local alterations, although effect sizes were relatively small. Also note that at 30 weeks, the corpus callosum
appeared to be larger in the infants with BI. However, growth was significantly less in these infants and no volumetric differences
were found at TEA.

Ta bl e 3. Mu lt ivar ia b le analy si s b etween the cli ni cal variabl es an d
the tota l b ra in vo lu mes
Term equivalent age
Total cohort (N = 210) Infants without brain injury (N = 165)
Variable Estimate p-value Adjusted R 2

Estimate p-value Adjusted R2
0.38 0.40
PMA at scan* 16.4 1.2 * 10-6 16.6 3.4 * 10-6

chapt e r 3
Sex -28.2 7.5 * 10-11 -29.4 5.6 * 10-10
BW z-score 16.2 2.0 * 10-11 17.0 6.5 * 10-11
MV7 -24.9 4.0 * 10-9 -23.7 2.7 * 10-7
30 weeks
Variable Estimate p-value Adjusted R 2

Estimate p-value Adjusted R2
0.49 0.39
PMA at scan* 12.1 3.3 * 10-14 10.6 5.7 * 10-8
Sex -10.5 7.8 * 10-6 -10.7 1.4 * 10-4
BW z-score 10.2 9.1 * 10-12 9.3 5.5 * 10-8
MV7 -9.5 7.8 * 10-5
Growth
Variable Estimate p-value Adjusted R2
0.15 No significant factors
Sex -13.8 4.8 * 10-4
MV7 -14.8 2.8 * 10-4
* PMA at scan was defined as the difference from 40 or 30 weeks PMA, in weeks.
Abbreviations: PMA = postmenstrual age; MV7 = mechanical ventilation for >7 days

c h apt e r 3
Figure 6 Effect of clinical variables on brain volumes at the early scan. This figure shows the effect of the clinical variables on
the absolute volumes at the early scan. For all significant brain regions, the effect size is indicated by the colour, with the axes
representing the effect size for each region in percentages. Sex (A, with girls having smaller brains) and birth weight z-score (B)
show an overall effect on the brain. Note again that most brain regions not significant for long duration of mechanical
ventilation (C) are significant for surgery (D).
Cl inica l r i sk facto r s a nd b r ai n vo l u m es
Effect o n g r owth
After multivariable regression analyses for each of the 50 brain regions (see Supplementary
table 1 for details), MV7 showed a significant effect on growth in 16 regions and sex (with
females showing smaller volumes) in 6 (figure 4). The effects for surgery and BI are shown in

table 2 and figure 5A. For TBV alone, sex (-6.4%, 0.0005) and MV7 (-6.9%, 0.0003) showed a
significant negative effect on growth (table 3).
After exclusion of the infants with BI, only a few regions remained significantly correlated with
growth, as represented in table 2. No significant correlations with TBV were found.
E f f ect s ee n at 3 0 w e e k s
For the 131 infants with serial data, multivariable regression analyses for each of the 50 regions
chapt e r 3
showed sex (females showing smaller volumes) to be significantly associated with volumes in
36, and BWZ in 43 brain regions (see Supplementary table 1 for details). Surgery was significant
in 10 regions and MV7 in 7 (figure 6). In most brain regions in which surgery was significant,
MV7 was not. Morphine was significant in the left posterior part of the superior temporal
gyrus, and BI was significant for bilateral ventricular size (figure 5B).
When performing the multivariable analysis for TBV, the factors sex, BWZ and MV7 reached
significance (table 3). The significant brain regions after correction for TBV can be found in
table 2. The significance of only the left-sided central grey nuclei can be explained by the larger
percentage of infants with left sided BI. When repeating the analysis after exclusion of infants
with BI, the effects diminished with sex now being significant in 26, BWZ in 40 and MV7 in 4
regions. Surgery and morphine were no longer found to be significant. The model for TBV
showed significant effects of sex and BWZ only, and a trend for MV7 (coefficient -8.8 cc,
p=0.002). Significant regions after correction for TBV are represented in table 2.
E f f ect s ee n at t e rm e q u i val e nt age

The effect of clinical risk factors on brain volumes at TEA could be studied for all 210 infants.
After multivariable regression analyses for the 50 separate brain regions sex, BWZ and MV7
reached significance in 43, 44 and 35 regions, respectively (for details, see Supplementary table
1). Surgery was significant in 10 regions and in 9 of these regions, MV7 was not significant. The
interaction between MV7 and surgery, when added to the model, was significant instead of the
separate factors in 23 regions (16 for MV7, 7 for surgery). Figure 7 shows the effect sizes of the
significant factors for all brain regions. BI was significant in the cerebellum and ventricles, both
bilateral (-8-10% and 34-37%, respectively, figure 5C). Excluding infants with primary cerebellar
injury did not change these findings significantly.
When performing multiple regression analysis for TBV alone, sex, BWZ and MV7 reached
significance (table 3). Adding surgery to the model showed a trend (coefficient -14.4 cc,
p=0.005). Excluding infants that were classified as small for gestational age (birth weight
below the 10th percentile) did not affect the influence of BWZ.
After correction for TBV sex, BWZ, MV7 and surgery lost their significance in many regions. The
remaining significant regions can be found in table 2 and for BI also in figure 5E.
When repeating the analyses after exclusion of all infants with BI, the remaining 165 infants
showed similar results. Sex was now significant in three additional brain regions, meaning it

Figure 7 Effect of clinical characteristics on brain volumes at term equivalent age. This figure shows the effect of the clinical c h apt e r 3
variables on the absolute volumes at term equivalent age. For all significant brain regions, the effect size is given in colour, with
the axes representing the effect size of each region in percentages. Sex (A, with girls having smaller brains) and birth weight
z-score (B) show an overall effect on the brain, with the strongest effect in the occipital regions. Note again that most brain
regions not significant for long duration of mechanical ventilation (C) are significant for surgery (D).
was significant in all regions except the ventricles and the medial part of the anterior temporal
lobes, bilaterally. BWZ was significant in the same regions as before. For MV7 and surgery,
some small differences were seen with MV7 being no longer significant in 2 regions, and
replacing surgery in another 2. Surgery was no longer significant in 5 regions but became
significant in 2. The model for TBV showed a significant effect of sex, BWZ and MV7, with effect
sizes fairly similar to the model including all infants (table 3). The significant brain regions after
correction for TBV can again be found in table 2.

Discussion
This study describes the growth of 50 brain structures in an unselected cohort of extremely
preterm infants using automatic segmentations. It gives detailed information about growth in
this crucial period of brain development. Sex, birth weight z-score, prolonged mechanical
ventilation and surgery showed a global effect on the brain, whereas BI showed a more
localized effect.
So far, only a few small studies described neonatal serial volumetric measurements covering
chapt e r 3
the entire brain. Growth patterns were shown to differ between tissue types and brain
regions.7,8 Due to the small size of these studies, multiple clinical risk factors could not be
taken into account. The current study size allows for this comparison and adds more detailed
information about the development of the entire brain in relation to clinical risk factors in this
unselected and therefore non-biased population.
In this study, the cerebellar hemispheres showed the largest growth between scans, with a
258% increase in this period, or a growth rate of 1.7 cm3/week. BI caused a decrease of 14% in
cerebellar growth and 9% in corrected volume at TEA, also in the absence of primary cerebellar
injury. The effect of supratentorial BI on infratentorial volume at TEA has been shown
previously.14,15 The cerebellar volume increases rapidly during the last trimester of
pregnancy.14,16 It is therefore extremely vulnerable to disturbances in normal development,
which are likely to occur in premature infants. Not only primary injury to the cerebellum, but
also remote effects, such as impaired trans-synaptic trophic effects, infection and
inflammation, and the negative effects of hemosiderin-degradation products are thought to
play an important role.14,16 Even in the absence of primary cerebellar injury, cerebellar growth
is impaired in preterm infants.17
We now demonstrate that these effects are not yet present at the early scan, but mainly seem
to influence growth of the cerebellum in the period between 30 weeks and TEA, without
influencing total brain growth. This delayed effect suggests that the processes of remote injury
are indeed a prominent factor on cerebellar growth and may indicate a window of opportunity
for intervention. In contrast with the cerebellum, ventricular enlargement after BI was already
clearly visible at 30 weeks and at TEA, the effect remained. The enlargement of the ventricles
usually takes place in the acute phase after an IVH, as a result of obstruction by blood clots and
impaired reabsorption of the cerebrospinal fluid and was therefore likely already present at
the first MRI.18
In general, the occipital regions of the brain showed a larger growth compared to the frontal
ones. This is in agreement with previous studies and is consistent with the well-known
maturational pattern of the brain in the occipital-frontal direction.19,20 Between 30 and 40
weeks postmenstrual age, a crucial part of human brain development and maturation takes
place. In the white matter, the oligodendrocyte lineage is undergoing major maturational
changes. Premyelinating late oligodendrocyte progenitors (preOLs) are abundant in this
period. Due to cell intrinsic properties, these preOLs are highly susceptible to hypoxia-
ischemia and thereby early ischemic cell death. In response to the acute cell death, the

remaining progenitor cells will proliferate and differentiate, but fail to myelinate properly,
leading to a disrupted maturation and thereby to disrupted brain growth. These disturbances
occur at a critical window for establishment of neuronal connections and the effect may
therefore persist and be translated into neurodevelopmental impairments later in life.
Multiple molecular mechanisms are involved and these may offer opportunities for
intervention, as has been tentatively shown in experimental studies.21,22 Clinical factors may
exacerbate these disturbances in cerebral maturation.23
c h apt e r 3
Sex and BWZ show a global effect on the brain, both at 30 weeks and at TEA, with girls and
infants with a relatively lower birth weight showing smaller volumes. Girls are known to be
generally smaller than boys in both weight and head circumference, and have been shown to
have smaller brain volumes from birth onwards.9,19,24 The findings in this study, with an effect of
-8% in girls at TEA, are in agreement with data found later in life.24
The effect of BWZ is in agreement with previous studies,25 and does not seem to be restricted
to infants small for gestational age (SGA) as it was retained after exclusion of these infants
from the analysis. The effect seems to be static, as it is already present at the early scan and
does not influence growth. This characteristic thus seems to represent generally smaller
infants with therefore smaller brains. The global effect of lung disease, defined as the need for
prolonged mechanical ventilation in this study, on brain volumes at TEA has been described
before.6,26 The current study additionally showed that the effect of prolonged mechanical
ventilation is already visible at 30 weeks, and also affects growth in multiple brain regions,
resulting in a slightly larger effect at TEA (-6.7 vs -7.0%, respectively). The underlying
mechanism of this effect is poorly understood, although experimental studies suggest a diffuse
and ongoing effect of lung disease on brain development.27
Not much is known about the effect of surgery on neonatal brain volumes. A small study
showed higher rates of BI in infants needing surgery,28 and this is in agreement with our data,
with 85% of the infants that underwent surgery showing BI, present before the first surgery,
versus only 13% of those without surgery (p=3.2*10-8). Hypotension, hypocarbia and hyperoxia
during surgery may exacerbate brain damage.29 For infants undergoing abdominal surgery,
relative undernutrition could be a factor. Inflammation may also play an important role and
this finding could indicate an ongoing process of alterations in brain growth. Additionally,
anaesthetic agents used during surgery may also cause neurodegenerative changes, i.e. due to
neurotoxicity, in the developing brain.30 Alternatively, it could be that the infants that had to
undergo surgery represent the most severely ill infants, and therefore showed volumetric
restrictions.
As described above, brain development is a complex progress, influenced by a myriad of
factors that are only partly understood. A combination of volumetric studies with studies
looking at cortical folding and those using diffusion and functional MR imaging, together with
long-term neurodevelopmental follow-up, at school age and older, may help to further
elucidate the underlying mechanisms.
There were some limitations to this study. First, the more severely ill children did not get an

early scan and the results for the early scan and the growth data may therefore be biased
towards the healthier part of the cohort. Second, despite the fairly large size of our cohort, it
was not large enough to reliably perform a more detailed analysis between the different forms
of surgery or BI. Third, outcome data are not available for the entire cohort, as part of the
infants are not yet two years of age and we can therefore not study correlations between the
volumetric and outcome data. Finally, no healthy control infants were included and no
conclusions regarding the difference in volumes between preterm and term infants can be
drawn.
chapter 3
In conclusion, this study showed volumetric growth, using longitudinal data, as well as the
effect of different clinical parameters on brain volumes in a large, unselected cohort of
preterm infants. Growth data showed differences between brain regions in percentage of
growth, with the cerebellar volume increasing the most and thereby also being most
vulnerable for developmental disturbances. A global effect on the brain was found for sex,
BWZ, long duration of mechanical ventilation and surgery, whereas BI showed localized effects
on most notably the ventricles and the cerebellum. Further studies are needed to assess the
relation of our findings with long-term neurodevelopmental outcome, and to examine the
association between segmentation of the brain and other MR techniques such as diffusion
weighted imaging and functional MRI.
Acknowledgements
This study includes infants participating in the Neobrain study (LSHM-CT-2006-036534), and
infants from a study funded by the Wilhelmina Research Fund (10-427).
Suppl eme ntar y ta b le 1 . Si gni fi ca nt var i ab les per b rain regio n
AREAS 40 weeks (n=210) 40 weeks / TBV
Hippocampus right Gender BWZ MV7
Hippocampus left Gender BWZ Surgery
Amygdala right Gender BWZ MV7
Amygdala left Gender BWZ Surgery
Anterior temporal lobe medial part right MV7
Anterior temporal lobe medial part left
Anterior temporal lobe lateral part right Gender MV7
Anterior temporal lobe lateral part left Gender BWZ
Gyri parahippocampalis et ambiens anterior part right Gender BWZ MV7 BWZ
Gyri parahippocampalis et ambiens anterior part left Gender BWZ MV7
Superior temporal gyrus middle part right Gender BWZ MV7
Superior temporal gyrus middle part left Gender BWZ MV7
Medial and inferior temporal gyri anterior part right Gender BWZ MV7 Brain injury

Medial and inferior temporal gyri anterior part left Gender BWZ MV7
Lateral occipito-temporal gyrus gyrus fusiformis anterior part right Gender BWZ MV7
Lateral occipito-temporal gyrus gyrus fusiformis anterior part left Gender BWZ MV7
Cerebellum right Gender BWZ MV7 Brain injury Brain injury
Cerebellum left Gender BWZ MV7 Brain injury Brain injury
Brainstem Gender BWZ MV7 Surgery
Insula left Gender BWZ MV7
chapter 3
Insula right Gender BWZ MV7
Occipital lobe left Gender BWZ MV7
Occipital lobe right Gender BWZ MV7
Gyri parahippocampalis et ambiens posterior part left BWZ MV7 Gender, brain injury
Gyri parahippocampalis et ambiens posterior part right BWZ MV7 Brain injury
Lateral occipito-temporal gyrus gyrus fusiformis posterior part left Gender BWZ MV7
Lateral occipito-temporal gyrus gyrus fusiformis posterior part right Gender BWZ MV7
Medial and inferior temporal gyri posterior part left Gender BWZ Surgery
Medial and inferior temporal gyri posterior part right Gender BWZ Surgery MV7
Superior temporal gyrus posterior part left Gender BWZ Surgery
Superior temporal gyrus posterior part right Gender BWZ Surgery
Cingulate gyrus anterior part left Gender BWZ MV7
Cingulate gyrus anterior part right Gender BWZ MV7
Cingulate gyrus posterior part left Gender BWZ MV7
Cingulate gyrus posterior part right Gender BWZ Surgery
Frontal lobe left Gender BWZ MV7
Frontal lobe right Gender BWZ MV7
Parietal lobe left Gender BWZ Surgery
Parietal lobe right Gender BWZ MV7
Caudate nucleus left BWZ MV7 Brain injury
Caudate nucleus right Gender BWZ MV7
Thalamus left Gender BWZ Surgery Brain injury
Thalamus right Gender BWZ MV7
Subthalamic nucleus left Gender BWZ MV7
Subthalamic nucleus right Gender BWZ MV7
Lentiform Nucleus left Gender BWZ MV7
Lentiform Nucleus right Gender BWZ MV7
Corpus Callosum Gender BWZ
Lateral Ventricle right Brain injury
Lateral Ventricle left Brain injury

AREAS 30 weeks (n=131) 30 weeks / TBV Growth (n=131)
Hippocampus right Gender BWZ Gender, MV7
Hippocampus left Gender BWZ Brain injury
Amygdala right Gender BWZ MV7
Amygdala left Gender BWZ Surgery Surgery
Anterior temporal lobe medial part right
Anterior temporal lobe medial part left
Anterior temporal lobe lateral part right Gender
Anterior temporal lobe lateral part left
Gyri parahippocampalis et ambiens anterior part right BWZ Gender,MV7
Gyri parahippocampalis et ambiens anterior part left BWZ MV7

chapter 3
Superior temporal gyrus middle part right Gender BWZ
Superior temporal gyrus middle part left Gender BWZ Surgery
Medial and inferior temporal gyri anterior part right Gender BWZ
Medial and inferior temporal gyri anterior part left Gender BWZ
Lateral occipito-temporal gyrus gyrus fusiformis anterior part right Gender BWZ Gender
Lateral occipito-temporal gyrus gyrus fusiformis anterior part left BWZ
Cerebellum right BWZ MV7 Brain injury
Cerebellum left BWZ MV7 Brain injury
Brainstem BWZ Surgery
Insula left Gender BWZ MV7
Insula right Gender BWZ BWZ MV7
Occipital lobe left Gender BWZ Gender
Occipital lobe right Gender BWZ Gender
Gyri parahippocampalis et ambiens posterior part left BWZ MV7
Gyri parahippocampalis et ambiens posterior part right BWZ MV7
Lateral occipito-temporal gyrus gyrus fusiformis posterior part left Gender BWZ MV7 MV7
Lateral occipito-temporal gyrus gyrus fusiformis posterior part right Gender BWZ MV7
Medial and inferior temporal gyri posterior part left Gender BWZ Gender, MV7
Medial and inferior temporal gyri posterior part right Gender BWZ
Superior temporal gyrus posterior part left Gender BWZ Morphine
Superior temporal gyrus posterior part right Gender BWZ
Cingulate gyrus anterior part left Gender BWZ MV7
Cingulate gyrus anterior part right Gender BWZ MV7
Cingulate gyrus posterior part left Gender BWZ MV7
Cingulate gyrus posterior part right Gender BWZ MV7
Frontal lobe left Gender BWZ Surgery Surgery
Frontal lobe right Gender BWZ MV7 MV7
Parietal lobe left Gender BWZ Surgery
Parietal lobe right Gender BWZ Surgery
Caudate nucleus left BWZ Surgery Brain injury
Caudate nucleus right BWZ MV7
Thalamus left Gender BWZ Surgery Brain injury Brain injury
Thalamus right Gender BWZ Surgery
Subthalamic nucleus left BWZ MV7
Subthalamic nucleus right BWZ
Lentiform Nucleus left Gender BWZ Surgery Brain injury MV7
Lentiform Nucleus right Gender BWZ MV7
Corpus Callosum Gender BWZ Brain injury Brain injury
Lateral Ventricle right Brain injury
Lateral Ventricle left Brain injury

References
1 Dubois J, Benders M, Cachia A, Lazeyras F, Ha- 11 Levene MI. Measurement of the growth of the
Vinh LR, Sizonenko SV et al. Mapping the early lateral ventricles in preterm infants with real-
cortical folding process in the preterm newborn time ultrasound. Arch Dis Child 1981;56:900-4.
brain. Cereb Cortex 2008;18:1444-54. 12 Makropoulos A, Gousias I, Ledig C, Aljabar P,
2 Volpe JJ. Brain injury in premature infants: a Serag A, Hajnal J et al. Automatic Whole Brain
complex amalgam of destructive and MRI Segmentation of the Developing Neonatal
developmental disturbances. Lancet Neurol Brain. IEEE Trans Med Imaging 2014;PP:Epub
2009;8:110-24. ahead of print.
3 Huppi PS, Warfield S, Kikinis R, Barnes PD, 13 Išgum I, Benders MJNL, Avants B, Cardoso MJ,
c h apt e r 3
Zientara GP, Jolesz FA et al. Quantitative Counsell SJ, Fischi Gomez E et al. Evaluation of
magnetic resonance imaging of brain automatic neonatal brain segmentation
development in premature and mature algorithms: the NeoBrainS12 challenge. Medical
newborns. Ann Neurol 1998;43:224-35. Image Analysis 2014;Accepted for publication.
4 Nishida M, Makris N, Kennedy DN, Vangel M, 14 Limperopoulos C, Soul JS, Gauvreau K, Huppi PS,
Fischl B, Krishnamoorthy KS et al. Detailed Warfield SK, Bassan H et al. Late gestation
semiautomated MRI based morphometry of the cerebellar growth is rapid and impeded by
neonatal brain: preliminary results. Neuroimage premature birth. Pediatrics 2005;115:688-95.
2006;32:1041-9. 15 Tam EW, Miller SP, Studholme C, Chau V, Glidden
5 Keunen K, Kersbergen KJ, Groenendaal F, Isgum I, D, Poskitt KJ et al. Differential effects of
de Vries LS, Benders MJ. Brain tissue volumes in intraventricular hemorrhage and white matter
preterm infants: prematurity, perinatal risk injury on preterm cerebellar growth. J Pediatr
factors and neurodevelopmental outcome: a 2011;158:366-71.
systematic review. J Matern Fetal Neonatal Med 16 Volpe JJ. Cerebellum of the premature infant:
2012;25 Suppl 1:89-100. rapidly developing, vulnerable, clinically
6 Thompson DK, Warfield SK, Carlin JB, Pavlovic M, important. J Child Neurol 2009;24:1085-104.
Wang HX, Bear M et al. Perinatal risk factors 17 Haldipur P, Bharti U, Alberti C, Sarkar C, Gulati G,
altering regional brain structure in the preterm Iyengar S et al. Preterm delivery disrupts the
infant. Brain 2007;130:667-77. developmental program of the cerebellum. PLoS
7 Mewes AU, Huppi PS, Als H, Rybicki FJ, Inder TE, One 2011;6:e23449.
McAnulty GB et al. Regional brain development 18 Robinson S. Neonatal posthemorrhagic
in serial magnetic resonance imaging of low-risk hydrocephalus from prematurity:
preterm infants. Pediatrics 2006;118:23-33. pathophysiology and current treatment
8 Zacharia A, Zimine S, Lovblad KO, Warfield S, concepts. J Neurosurg Pediatr 2012;9:242-58.
Thoeny H, Ozdoba C et al. Early assessment of 19 Gilmore JH, Lin W, Prastawa MW, Looney CB,
brain maturation by MR imaging segmentation Vetsa YS, Knickmeyer RC et al. Regional gray
in neonates and premature infants. AJNR Am J matter growth, sexual dimorphism, and cerebral
Neuroradiol 2006;27:972-7. asymmetry in the neonatal brain. J Neurosci
9 Visser GH, Eilers PH, Elferink-Stinkens PM, Merkus 2007;27:1255-60.
HM, Wit JM. New Dutch reference curves for 20 Kinney HC, Brody BA, Kloman AS, Gilles FH.
birthweight by gestational age. Early Hum Dev Sequence of central nervous system myelination
2009;85:737-44. in human infancy. II. Patterns of myelination in
10 Papile LA, Burstein J, Burstein R, Koffler H. autopsied infants. J Neuropathol Exp Neurol
Incidence and evolution of subependymal and 1988;47:217-34.
intraventricular hemorrhage: a study of infants 21 Scafidi J, Hammond TR, Scafidi S, Ritter J,
with birth weights less than 1,500 gm. J Pediatr Jablonska B, Roncal M et al. Intranasal epidermal
1978;92:529-34. growth factor treatment rescues neonatal brain

injury. Nature 2014;506:230-4. 26 Boardman JP, Counsell SJ, Rueckert D, Hajnal JV,
22 Wood TL, Loladze V, Altieri S, Gangoli N, Levison Bhatia KK, Srinivasan L et al. Early growth in brain
SW, Brywe KG et al. Delayed IGF-1 administration volume is preserved in the majority of preterm
rescues oligodendrocyte progenitors from infants. Ann Neurol 2007;62:185-92.
glutamate-induced cell death and hypoxic- 27 Loeliger M, Inder T, Cain S, Ramesh RC, Camm E,
ischemic brain damage. Dev Neurosci Thomson MA et al. Cerebral outcomes in a
2007;29:302-10. preterm baboon model of early versus delayed
23 Back SA, Miller SP. Brain injury in premature nasal continuous positive airway pressure.
neonates: A primary cerebral dysmaturation Pediatrics 2006;118:1640-53.
chapt e r 3
disorder? Ann Neurol 2014;75:469-86. 28 Filan PM, Hunt RW, Anderson PJ, Doyle LW, Inder
24 Brain Development Cooperative Group. Total TE. Neurologic outcomes in very preterm infants
and regional brain volumes in a population- undergoing surgery. J Pediatr 2012;160:409-14.
based normative sample from 4 to 18 years: the 29 McCann ME, Schouten AN, Dobija N, Munoz C,
NIH MRI Study of Normal Brain Development. Stephenson L, Poussaint TY et al. Infantile
Cereb Cortex 2012;22:1-12. postoperative encephalopathy: perioperative
25 Tolsa CB, Zimine S, Warfield SK, Freschi M, factors as a cause for concern. Pediatrics
Sancho RA, Lazeyras F et al. Early alteration of 2014;133:e751-e757.
structural and functional brain development in 30 Mellon RD, Simone AF, Rappaport BA. Use of
premature infants born with intrauterine growth anesthetic agents in neonates and young
restriction. Pediatr Res 2004;56:132-8. children. Anesth Analg 2007;104:509-20.

chapter 3

ch a p t er
4
Hydrocortisone
treatment for
bronchopulmonary
dysplasia
and brain
volumes in preterm
infants
Linda S. de Vries
Britt J.M. van Kooij
Ivana Išgum
Karin J. Rademaker
Frank van Bel
Petra S. Hüppi
Jessica Dubois
Floris Groenendaal
Journal of Pediatrics 2013 163:666-71 73

Abstract
Objective To assess whether there was an adverse effect on brain growth after
hydrocortisone (HC) treatment for bronchopulmonary dysplasia (BPD) in a large cohort of
infants without dexamethasone exposure.
Study design Infants who received HC for BPD between 2005 and 2011 and underwent
magnetic resonance imaging at term equivalent age were included. Control infants born in
Geneva (2005-2006) and Utrecht (2007-2011) were matched to the infants treated with
HC according to segmentation method, sex, and gestational age. Infants with overt
parenchymal pathology were excluded. Multivariable analysis was used to determine if
there was a difference in brain volumes between the 2 groups.
chapter 4
Results Seventy-three infants treated with HC and 73 matched controls were included.
Mean gestational age was 26.7 weeks and mean birth weight was 906 g. After correction
for gestational age, postmenstrual age at time of scanning, the presence of intraventricular
haemorrhage, and birth weight z-score, no differences were found between infants treated
with HC and controls in total brain tissue or cerebellar volumes.
Conclusions In the absence of associated parenchymal brain injury, no reduction in brain

tissue or cerebellar volumes could be found at term equivalent age between infants with
or without treatment with HC for BPD.

Introduction
B
ronchopulmonary dysplasia (BPD) remains a common complication of extremely
preterm birth. Risk factors for the development of BPD, such as difficulty in weaning an
infant from the ventilator or prolonged oxygen requirement in the first weeks of life,
are even more frequently encountered. Treatment options are limited. If conservative care
with less aggressive ventilator settings, treatment of a hemodynamic significant patent ductus
arteriosus, fluid restriction, and/or diuretics is not sufficient, a decision can be made to treat
with corticosteroids, with dexamethasone used most commonly. Although the short term
effects of dexamethasone prescription on pulmonary function are satisfactory, effects on long
term neurodevelopmental outcome are not.1 Preterm infants treated with dexamethasone
had a higher rate of cerebral palsy and cognitive impairment and more often needed special
education at early school age.2-4 The origin of these adverse sequelae may be represented by
chapt e r 4
smaller brain volumes at term equivalent age.5;6 Therefore, treatment with dexamethasone is
not recommended and should be restricted to the most severe cases.
Hydrocortisone (HC) is an alternative treatment option. Although somewhat less potent, if
given moderately early (between 5-25 days after birth), the effects on pulmonary function are
similar. There is not much research on long term outcomes after the use of HC, but several
studies have not shown any difference between HC-treated infants and controls regarding
rates of cerebral palsy and other neuromotor deficits and cognitive development.7-10 Two
studies have reported on the effect of HC on brain volumes at term equivalent age. Benders et
al described a small cohort of preterm infants without associated brain injury and did not find
any differences between HC-treated infants and controls.11 However, Tam et al described a
larger cohort and found a difference in cerebellar size at term equivalent age after treatment
with HC.12 Important drawbacks of the study by Tam et al were that part of these infants also
received dexamethasone and infants with parenchymal brain lesions were included.
Our aim was to assess whether there was an adverse effect on brain volume at term equivalent
age after HC treatment for BPD in a cohort of preterm infants without dexamethasone exposure.
Methods
A combined cohort was formed containing children from Geneva and Utrecht. Infants from
Utrecht, who received HC for BPD between 2005 and 2011 and had magnetic resonance
imaging (MRI) at term equivalent age, were included. Infants born between 2005-2006 have
been described previously.11 For those previously described infants, parental informed consent
was given. For the other infants, clinically indicated magnetic resonance (MR) images were
used with permission from the ethical review board of our institution. HC was given starting at
a postnatal age of ≥7 days, in ventilator-dependent infants with need for supplemental
oxygen, if this could not be accounted for by an infection or a patent ductus arteriosus.
Standard clinical dosage schemes were followed, starting with a dose of 5 mg/kg/d divided in 4
doses for 1 week and a subsequent tapering course every 5 days, leading to a total treatment
hydrocortisone treatment for bronchopulmonary dysplasia and brain volumes in preterm infants 75
duration of 22 days and a standard cumulative dosage of 72.5 mg/kg. This scheme could be
adjusted at the discretion of the attending neonatologist. Infants treated with HC were
matched to control infants born in both Geneva (2005-2006) and Utrecht (2007-2011).
Control infants were matched for sex and gestational age. Matching was performed in
subgroups, taking into account that infants were matched with controls scanned with the
same imaging protocol and segmented with the same automatic method. Clinical variables
were extracted from patient charts. Cerebral lesions were diagnosed on the basis of serial
cranial ultrasound and MRI results. The presence of an intraventricular haemorrhage (IVH),
graded according to Papile et al13, white matter damage, and large cerebellar haemorrhages
was recorded. Infants with a large parenchymal haemorrhage (1 infant) or a large cerebellar
haemorrhage (1 infant) were excluded. In all infants with posthaemorrhagic ventricular
dilatation, a stable situation with a decrease in ventricular size was reached soon after the
chapt e r 4
initiation of treatment. Three infants in the HC-treated group had a reservoir inserted but
none required a permanent shunt. There were no infants with evidence of cystic
periventricular leukomalacia on their cranial ultrasound and MRI examinations.
Mag n e tic R es o na nc e Im agi ng

MRI was performed around term equivalent age in all infants. Infants born in 2005 and 2006
were scanned with use of a 1.5-T MR system (Philips Medical Systems, Best, The Netherlands).
The protocol included a 3-dimensional (3D) T1 fast-gradient echo sequence (repetition time
[TR] 15 ms, echo time [TE] 4.4 ms, slice thickness 1.5 mm) and a T2 fast-spin echo sequence
(TR 3500 ms, TE 30/150 ms, slice thickness 1.5 mm), both in the coronal plane.
Infants born from in 2007 or later were scanned on a 3.0-T MR system (Achieva; Philips
Medical Systems) using the sense head coil. Between 2007 and June 2008, the imaging
protocol contained axial 3D T1-weighted and T2-weighted images (TR 9.4 ms, TE 4.6 ms, and
slice thickness 2.0 mm; and TR 6293 ms, TE=120 ms, slice thickness 2.0 mm, respectively).
From June 2008 onwards, 3D T1-weighted and T2-weighted images were acquired in the
coronal plane (3D T1-weigthed: TR 9.5 ms, TE 4.6 ms, slice thickness 1.2 mm; T2-weighted: TR
4847 ms, TE 150 ms, and slice thickness 1.2 mm).
Infants were sedated using oral chloral hydrate 50-60 mg/kg. Heart rate, transcutaneous
oxygen saturation, and respiratory rate were monitored. Minimuffs (Natus Medical Inc, San
Carlos, California) were used for hearing protection. All MR examinations were re-evaluated by
2 experienced neonatologists. Lesions seen on conventional imaging were scored. Infants with
evident tissue loss on MRI (e.g., porencephaly due to a periventricular haemorrhagic infarction
or a large cerebellar lesion destroying more than one-half of the cerebellar hemisphere) were
excluded (2 patients, as described earlier).
Vol u me t r ic m e a s u r em ent s
Brain tissue and cerebellum were segmented automatically. Considering that infants were
scanned using 2 different imaging protocols on scanners of a different field strength, we chose

to use the segmentation method best fitted for the acquisition of the data per imaging
protocol. Therefore, segmentations were performed using 2 different segmentation methods.
For the children born before 2007 in both Utrecht and Geneva, tissues were segmented using
the method described by Warfield et al.14 The method segments cortical gray matter , deep
gray matter, unmyelinated white matter, myelinated white matter and cerebrospinal fluid, but
it does not allow separate delineation of the cerebellum. Therefore, cerebellar tissue was
manually outlined in T2-weighted scans and volumes were corrected accordingly. Details from
this method have been described previously.11 For the infants born in 2007 or later, the
algorithm of the method of Anbeek et al was adjusted to segment 3T scans.15;16 In addition to
the mentioned tissues, this algorithm delineates cerebellum, basal ganglia, brainstem and
separation of the cerebrospinal fluid in the ventricles from cerebrospinal fluid outside the
brain. Cerebellar segmentations in this group were thus automatically generated. Total brain
chapt e r 4
tissue volume and intracranial volume were calculated from the segmentations. The Anbeek
et al method was developed for axially acquired images.15;16 However, we tested for a
difference in volumes between coronal and axial acquired images in a subgroup of 5 infants
and did not find any differences in volumes (paired t tests: cerebellum, p = 0.245; total brain
volume, p = 0.783). In addition, to confirm the quality of the automatically obtained
segmentations, results were visually inspected. This allowed us to combine these sets into 1
cohort.
Statistical procedures were performed using both IBM SPSS Statistics version 20 (SPSS Inc,
Chicago, Illinois) and R version 2.15.0 (www.r-project.org).17 Baseline characteristics between
the 2 groups were compared using independent-sample t tests. For multivariable analysis,
general linear modelling was used with brain volume as dependent variable. A cut-off value of
p = 0.05 was used. All patients with HC were included, and a matched sample of all other
patients without HC use and with available MRI. Brain volumes were analyzed with individual
patients in the regression model. Earlier studies reported a decrease in brain volume of 20% for
cerebellar volume and 10%-30% for total cerebral tissue volume after the use of
dexamethasone.5;6 Sample size calculations were performed, taking into account the ability to
show the presence or absence of an effect one-half the size of dexamethasone, so a 10%
difference in brain volumes. Sample size calculations demonstrated that this difference of 10%
in brain volumes could be demonstrated with an α of 0.05, a power of 0.9, and a sample size of
55 patients in each arm.
Because postmenstrual age at time of scanning, gestational age, and birth weight influence
total brain volumes, we included these variables in our model.18-20 We also included the
presence of an IVH. Separate analysis of infants with mild posthaemorrhagic ventricular
dilatation, minor punctate white matter lesions, or punctate cerebellar haemorrhages
revealed no significant differences in volumes. Therefore, these patients were included in the
final analysis. Corticosteroid use was analyzed in 2 ways: binomial (yes/no) and as total
cumulative dosage per kilogram up to the day of MRI. Because distribution of HC was skewed,
we used the natural logarithm of this variable. Also, to screen whether a high dose of HC
would have an effect where a low dose did not, we repeated the analysis with HC divided as
none, a cumulative dosage <50 mg/kg and a cumulative dosage >50 mg/kg. Interaction of
gestational age and HC was tested in the model. Postmenstrual age at time of scanning was
calculated as the difference from 40.0 weeks’ postmenstrual age and gestational age as the
difference from 24.0 weeks’ (our youngest included infants were born at that gestation). Birth
weights are represented using a z-score. All statistical analyses were also performed within
each subgroup (e.g., infants segmented with the Warfield et al method14 and with the Anbeek
et al method15;16) to further quantify any possible differences between groups.
chapter 4
Ta bl e 1. C linica l var ia b les
Infants treated with HC Control infants

Variable p-value
(n=73) (n=73)
Gestational age, wk (SD, range) 26.6 (1.35, 24.3-30.4) 26.9 (1.34, 24.3-30.4) 0.12
Sex, No. male/female 40/33 36/37 0.51
Birth weight, g (SD) 863 (205) 948 (166) 0.01
Birth weight z-score 0.06 0.40 0.02
Antenatal steroids, % 70 77 0.70
Mechanical ventilation, median d (25-75 percentile) 16 (12-21) 4 (0-7) 0.05
Grade RDS, No.

No 4 21
Grade 1 2 3
0.00
Grade 2 15 31
Grade 3 28 11
Grade 4 24 7
Postmenstrual age at time of scanning, wk (SD) 41.1 (0.76) 41.1 (0.75) 0.67
Weight at time of scanning, g (SD) 3182 (571) 3280 (459) 0.25
Presence of IVH, No.

No 41 60
Grade 1-2 24 12 0.001
Grade 3 7 1
Grade 4 0 0
Cerebellar haemorrhage, No.

No 69 68
0.79
<6 Punctate lesions 3 4
>6 Punctate lesions 1 1
Posthaemorrhagic ventricular dilatation requiring

intervention, No.
0.05
Lumbar punctures 3 1
Placement of a reservoir 3 0

Results
A total of 75 infants treated with HC and 75 control infants were included. Two infants treated
with HC had evident tissue loss on their MRI (1 porencephaly and 1 large cerebellar
haemorrhage) and thus were excluded. The remaining 73 infants with HC treatment and 73
control infants were eligible for analysis. Despite matching, infants treated with HC had a
significantly lower birth weight (p = 0.01, Table 1). Weight at scan, however, did not differ
significantly (p = 0.25). Concurrent with the more severe pulmonary problems, the grade of
respiratory distress syndrome21 and days of mechanical ventilation were significantly higher in
infants treated with HC compared with controls (p = 0.00 and p = 0.05, respectively).
Because HC was given as part of clinical practice at the discretion of the attending
neonatologist, not all infants received the same cumulative dosage. In some children, HC was
chapter 4
Ta bl e 2 . Bra in vo lu mes for the di fferent ti ssue cl asses, per segme n tatio n
meth o d , u nc o rrected for postmenstr u al age at time o f scan n in g, gestatio n al
age o r b irth we ig ht
Warfield et al14 method Anbeek et al15,16 method
HC (n=19) Control (n=19) HC (n=54) Control (n=54)
Cortical grey matter, ml (SD) 172 (32) 183 (38) 153 (22) 158 (17)
Unmyelinated white matter, ml (SD) 177 (27) 194 (33) 145 (23) 150 (21)
Cerebellum, ml (SD) 23 (4) 24 (5) 28 (4) 30 (3)
Cerebrospinal fluid, ml (SD) 49 (18) 57 (27) 101 (18) 101 (14)
Total brain tissue volume, ml (SD) 377 (57) 403 (64) 355 (39) 369 (34)
Intracranial volume, ml (SD) 426 (64) 460 (73) 455 (52) 471 (44)
tapered sooner because of significant improvement or side effects such as hypertension. In

other infants, a longer treatment or a slower tapering course was given because of a relapse
after the first tapering steps. This led to a wide range in the cumulative dosage of HC, between
16 and 216 mg/kg with a mean of 77 mg/kg. Also, although children with evidence of tissue
loss were excluded, the presence of an IVH did show a significant difference (p = 0.00, Table 1).
The presence of an IVH may have an adverse effect on cerebellar volumes.22 Therefore, the
presence of a grade 3 IVH was included as a factor in the multivariable analysis. There were no
infants with evidence of cystic periventricular leukomalacia on their cranial ultrasound and
MRI in this cohort.
Uncorrected brain volumes are shown in Table 2. The results of the multivariable analysis are
shown in Table 3. There was no significant difference in total brain tissue volume between HC-
treated and control infants (p = 0.08). Only postmenstrual age at MRI, birth weight z-score,
and gestational age at birth were significant factors in the multivariable model. Repeating the
analysis using the cumulative dosage of HC gave the same results (p = 0.1), as did the analysis
with HC divided into high and low dosages (p = 0.11 and p = 0.28, respectively).
We also performed these analyses for cerebellar volumes, because previous literature
suggested that this brain structure may especially be affected by the use of hydrocortisone.12
Multivariable analysis showed postmenstrual age at time of scan, birth weight z-score, and
grade 3 IVH to be of significant influence on cerebellar volumes at term equivalent age. Again,
neither HC as a binomial factor (p = 0.39), nor the cumulative dosage of HC (p = 0.07), or the
divided dosages (p = 0.26 for the high and p = 0.82 for the low dosage), were significant.
Although infants with large cerebellar haemorrhages were excluded, we did include infants
with punctate lesions in the cerebellum. When repeating the analysis after exclusion of these
children, the resulting models were the same. Therefore, these children remained included in
chapt e r 4
the analysis.
The results of the subgroup analysis were identical to those of the entire cohort and are
therefore not represented separately.
Ta bl e 3. Mu lt ivar ia b le analy si s
Model/factor Coefficient, ml 95% CI p-value
Total brain tissue volume: final model
Post menstrual age at scan, wk 18.2 9.2 – 27.3 0.00

Birth weight z-score 11.2 3.3 – 19.2 0.00
Gestational age, wk 6.8 1.7 – 12.0 0.01
HC, yes/no -12.5 -26.3 – 1.4 0.08
Cerebellar volume: final model
Postmenstrual age at scan, wk 2.1 1.3 – 2.9 0.00

Birth weight z-score 2.2 1.6 – 2.9 0.00
Grade 3 IVH -2.2 -4.4 – 0.1 0.06
HC, yes/no -0.53 -1.8 – 0.7 0.39
In the model, postmenstrual age is taken as the difference from 40 wk, and gestational age as the difference from 24 wk. IVH is
graded according to Papile et al.13
Discussion
In this cohort of 146 preterm infants, HC treatment for BPD was not associated with a
reduction in total brain tissue volume or cerebellar volume. Only postmenstrual age at time of
scanning, gestational age, and birth weight z-score had a significant effect on total brain tissue
volume. Postmenstrual age at time of scanning, birth weight z-score, and the presence of a
grade 3 IVH had a significant effect on the cerebellar volumes at term equivalent age. These
are all known factors to adversely affect brain volumes. 18-20,22

The lack of effect of HC on brain volumes is consistent with previous reports from our group
describing short- and long-term effects of HC. Benders et al described a small subset of this
cohort and were unable to find any differences in brain volumes.11 This absence of volumetric
differences seems to persist into childhood. Lodygensky et al showed in a cohort of 60 preterm
children that brain volumes and neurocognitive outcome at 8 years did not differ between
children treated with HC and preterm controls.23 The lack of effect of HC on long-term
neurodevelopmental outcome has also been shown in several follow-up studies.7-10;24 In
contrast, Tam et al showed that in their cohort of 172 infants, cerebellar volume was
negatively affected by the use of both dexamethasone and HC.12 However, only 31 subjects
received postnatal HC and 11 of those received both HC and dexamethasone. Also, infants
with large cerebellar haemorrhages were included in the analysis, although multivariable
correction was used. Tam et al found a negative effect of 8% on cerebellar volumes after the
chapt e r 4
use of HC.12 Allin et al previously reported an 8% decrease of cerebellar volume in preterm
infants compared with healthy term controls at 15 years of age. This decrease had a negative
correlation with long-term cognitive outcome.25 An 8% difference, therefore, seems to have
clinical relevance. An additional power calculation demonstrated that in our 146 patients, the
power to detect a difference of 8% in brain volumes with an α of 0.05 would be 0.85. In
addition, based on the observations of dexamethasone, the potential effect of HC is to
diminish and not to increase brain volumes, a 1-sided test could be used. In this case, the
power of our study of 146 infants divided over 2 groups of 73 would be 0.92, which is high.
Previous studies on volumetric changes after dexamethasone use have shown decreases in
cerebellar volume of 20% and in total cerebral tissue volume of 10%-30%.5;6 The effect sizes
found in this study are, besides not being significant, very small. The effect size of HC on
cerebellar volume is -0.5 ml. With a mean cerebellar volume of 28.4 ml in the control infants,
this would mean a decrease of 1.8%. For total brain volume, the same calculation would lead to
a decrease of 3.3% (mean volume 378.2 ml, effect size -12.5 ml). These values are much smaller
than those presented by Tam et al,12 and this difference may be partially explained by the
concomitant use of dexamethasone and HC in part of their cohort.
Dexamethasone has long been known to have a detrimental effect on brain growth and
maturation in the preterm infant and has been extensively studied in animal experiments.26;27
There are several hypotheses on why there is such a difference in adverse long-term effects
between dexamethasone and HC. First, dexamethasone mainly interacts with the
glucocorticoid receptor of the brain, whereas HC primarily uses the mineralocorticoid
receptors.28 Activation of the glucocorticoid receptor leads to adverse neuronal effects.29
Second, during gestation, the enzyme 11β-hydroxysteroiddehydrogenase-2 is widely
expressed in the foetal or preterm brain. This enzyme will metabolize the active HC to the
inactive 11-dehydro form, thus lowering the active dosage and protecting the developing
brain from excessive corticosteroid exposure.30;31 Dexamethasone is a synthetic
glucocorticoid, and there is no equivalent enzyme for dexamethasone in the human brain.
Third, the effective dosage of HC given is often lower than the comparable dosage
dexamethasone, resulting in a lower cumulative dosage that can cause toxic effects.32;33 Our
standard dosage scheme of a total 72.5 mg/kg HC is substantially higher than in most other
studies reporting on HC use, where total doses between 6 and 30 mg/kg were given.33
However, the equivalent dosage of dexamethasone in our patients would be approximately
2-3 mg/kg. Most trials reporting on dexamethasone used higher dosage schemes.32 Another
possible cause for a difference in accumulation is the shorter biological half-life of HC (8-12
hours) compared with dexamethasone (36-72 hours).34 Again, this could explain a difference
in toxic effects.
There are several limitations to our study. First, the study was retrospective, limiting our ability
to detect confounders. Second, 2 different acquisition protocols were used; however, infants
were matched within protocol. Third, we excluded infants with cystic parenchymal brain
chapter 4
lesions, which may have selected healthier infants but did avoid the possible effect of large
white matter lesions on cerebellar volumes.22;35 Fourth, we cannot comment on the underlying
microstructure of the brain, which may be altered by HC exposure. And last, this was a short-
term, structural study; we did not evaluate the relationship of brain volume to
neurodevelopmental outcomes.
In conclusion, we could not demonstrate any differences in brain volumes between children
treated with HC for BPD and nontreated controls at term equivalent age in this cohort.
Combined with the earlier reported lack of effect on long-term developmental outcome, HC
seems to be a safer alternative than dexamethasone in the treatment of BPD.
Acknowledgments
The authors would like to thank the medical students Mila Rast and Kristin Keunen for their
help in collecting data from the medical charts.
References
1 Halliday HL, Ehrenkranz RA, Doyle LW. Early (< 8 asone therapy for lung disease of prematurity. N Engl
days) postnatal corticosteroids for preventing chron- J Med 2004;350:1304-13.
ic lung disease in preterm infants. Cochrane Database 5 Murphy BP, Inder TE, Huppi PS, Warfield S, Zientara
Syst Rev 2010;CD001146. GP, Kikinis R, et al. Impaired cerebral cortical gray
2 O’Shea TM, Washburn LK, Nixon PA, Goldstein DJ. Fol- matter growth after treatment with dexamethasone
low-up of a randomized, placebo-controlled trial of for neonatal chronic lung disease. Pediatrics
dexamethasone to decrease the duration of ventila- 2001;107:217-21.
tor dependency in very low birth weight infants: neu- 6 Parikh NA, Lasky RE, Kennedy KA, Moya FR, Hochhaus-
rodevelopmental outcomes at 4 to 11 years of age. er L, Romo S, et al. Postnatal dexamethasone therapy
Pediatrics 2007;120:594-602. and cerebral tissue volumes in extremely low birth
3 Shinwell ES, Karplus M, Reich D, Weintraub Z, Blazer weight infants. Pediatrics 2007;119:265-72.
S, Bader D, et al. Early postnatal dexamethasone 7 Rademaker KJ, Uiterwaal CS, Groenendaal F, Venema
treatment and increased incidence of cerebral palsy. MM, Van Bel F, Beek FJ, et al. Neonatal hydrocortisone
Arch Dis Child Fetal Neonatal Ed 2000;83:F177-F181. treatment: neurodevelopmental outcome and MRI at
4 Yeh TF, Lin YJ, Lin HC, Huang CC, Hsieh WS, Lin CH, et school age in preterm-born children. J Pediatr
al. Outcomes at school age after postnatal dexameth- 2007;150:351-7.

8 van der Heide-Jalving M, Kamphuis PJ, van der Laan hemorrhage and white matter injury on preterm cer-
MJ, Bakker JM, Wiegant VM, Heijnen CJ, et al. Short- ebellar growth. J Pediatr 2011;158:366-71.
and long-term effects of neonatal glucocorticoid 23 Lodygensky GA, Rademaker K, Zimine S, Gex-Fabry M,
therapy: is hydrocortisone an alternative to dexa- Lieftink AF, Lazeyras F, et al. Structural and functional
methasone? Acta Paediatr 2003;92:827-35. brain development after hydrocortisone treatment
9 Watterberg KL, Shaffer ML, Mishefske MJ, Leach CL, for neonatal chronic lung disease. Pediatrics
Mammel MC, Couser RJ, et al. Growth and neurode- 2005;116:1-7.
velopmental outcomes after early low-dose hydro- 24 Karemaker R, Heijnen CJ, Veen S, Baerts W, Samsom J,
cortisone treatment in extremely low birth weight in- Visser GH, et al. Differences in behavioral outcome
fants. Pediatrics 2007;120:40-8. and motor development at school age after neonatal
10 Rademaker KJ, de Vries LS, Uiterwaal CS, Groenendaal treatment for chronic lung disease with dexametha-
F, Grobbee DE, Van Bel F. Postnatal hydrocortisone sone versus hydrocortisone. Pediatr Res
treatment for chronic lung disease in the preterm 2006;60:745-50.
newborn and long-term neurodevelopmental follow- 25 Allin M, Matsumoto H, Santhouse AM, Nosarti C, AlA-
up. Arch Dis Child Fetal Neonatal Ed 2008;93:F58-F63. sady MH, Stewart AL, et al. Cognitive and motor func-
11 Benders MJ, Groenendaal F, Van Bel F, Ha VR, Dubois J, tion and the size of the cerebellum in adolescents
Lazeyras F, et al. Brain development of the preterm born very pre-term. Brain 2001;124:60-6.
neonate after neonatal hydrocortisone treatment for 26 Almeida OF, Conde GL, Crochemore C, Demeneix BA,
chapter 4
chronic lung disease. Pediatr Res 2009;66:555-9. Fischer D, Hassan AH, et al. Subtle shifts in the ratio
12 Tam EW, Chau V, Ferriero DM, Barkovich AJ, Poskitt KJ, between pro- and antiapoptotic molecules after acti-
Studholme C, et al. Preterm cerebellar growth im- vation of corticosteroid receptors decide neuronal
pairment after postnatal exposure to glucocorticoids. fate. FASEB J 2000;14:779-90.
Sci Transl Med 2011;3:105ra105. 27 Kreider ML, Aldridge JE, Cousins MM, Oliver CA, Sei-
13 Papile LA, Burstein J, Burstein R, Koffler H. Incidence dler FJ, Slotkin TA. Disruption of rat forebrain devel-
and evolution of subependymal and intraventricular opment by glucocorticoids: critical perinatal periods
hemorrhage: a study of infants with birth weights less for effects on neural cell acquisition and on cell sign-
than 1,500 gm. J Pediatr 1978;92:529-34. aling cascades mediating noradrenergic and cholin-
14 Warfield SK, Kaus M, Jolesz FA, Kikinis R. Adaptive, ergic neurotransmitter/neurotrophic responses. Neu-
template moderated, spatially varying statistical clas- ropsychopharmacology 2005;30:1841-55.
sification. Med Image Anal 2000;4:43-55. 28 De Kloet ER, Vreugdenhil E, Oitzl MS, Joels M. Brain
15 Anbeek P, Vincken KL, Groenendaal F, Koeman A, van corticosteroid receptor balance in health and dis-
Osch MJ, van der Grond J. Probabilistic brain tissue ease. Endocr Rev 1998;19:269-301.
segmentation in neonatal magnetic resonance imag- 29 Sapolsky RM, Uno H, Rebert CS, Finch CE. Hippocam-
ing. Pediatr Res 2008;63:158-63. pal damage associated with prolonged glucocorti-
16 Van Kooij BJ, Benders MJ, Anbeek P, van Haastert I, de coid exposure in primates. J Neurosci 1990;10:2897-
Vries LS, Groenendaal F. Cerebellar volume and pro- 902.
ton magnetic resonance spectroscopy at term, and 30 Wyrwoll CS, Holmes MC, Seckl JR. 11beta-hydroxys-
neurodevelopment at 2 years of age in preterm in- teroid dehydrogenases and the brain: from zero to
fants. Dev Med Child Neurol 2012;54:260-6. hero, a decade of progress. Front Neuroendocrinol
17 R: A Language and Environment for Statistical Com- 2011;32:265-86.
puting [computer program]. Vienna, Austria: 2012. 31 Holmes MC, Sangra M, French KL, Whittle IR, Paterson
18 Inder TE, Warfield SK, Wang H, Huppi PS, Volpe JJ. Ab- J, Mullins JJ, et al. 11beta-Hydroxysteroid dehydroge-
normal cerebral structure is present at term in pre- nase type 2 protects the neonatal cerebellum from
mature infants. Pediatrics 2005;115:286-94. deleterious effects of glucocorticoids. Neuroscience
19 Peterson BS, Anderson AW, Ehrenkranz R, Staib LH, 2006;137:865-73.
Tageldin M, Colson E, et al. Regional brain volumes 32 Doyle LW, Ehrenkranz RA, Halliday HL. Dexametha-
and their later neurodevelopmental correlates in sone treatment after the first week of life for bron-
term and preterm infants. Pediatrics 2003;111:939- chopulmonary dysplasia in preterm infants: a system-
48. atic review. Neonatology 2010;98:289-96.
20 Tolsa CB, Zimine S, Warfield SK, Freschi M, Sancho RA, 33 Doyle LW, Ehrenkranz RA, Halliday HL. Postnatal hy-
Lazeyras F, et al. Early alteration of structural and drocortisone for preventing or treating bronchopul-
functional brain development in premature infants monary dysplasia in preterm infants: a systematic re-
born with intrauterine growth restriction. Pediatr Res view. Neonatology 2010;98:111-7.
2004;56:132-8. 34 Rademaker CMA. Geneesmiddelen-Formularium
21 Giedion A, Haefliger H, Dangel P. Acute pulmonary X- voor Kinderen. Utrecht: 2007.
ray changes in hyaline membrane disease treated 35 Srinivasan L, Allsop J, Counsell SJ, Boardman JP, Ed-
with artificial ventilation and positive end-expiratory wards AD, Rutherford M. Smaller cerebellar volumes
pressure (PEP). Pediatr Radiol 1973;1:145-52. in very preterm infants at term-equivalent age are as-
22 Tam EW, Miller SP, Studholme C, Chau V, Glidden D, sociated with the presence of supratentorial lesions.
Poskitt KJ, et al. Differential effects of intraventricular AJNR Am J Neuroradiol 2006;27:573-9.
ch a p t er
5
Is early cortical folding
related to clinical
risk factors and
neurodevelopmental outcome?
A longitudinal
MRI study of
preterm newborns
François Leroy
Floris Groenendaal
Linda S. de Vries
Nathalie H.P. Claessens
Pim Moeskops
Clara Fischer
Jean-Francois Mangin
Ivana Išgum
Max A. Viergever
Jessica Dubois
In preparation 85
Abstract
Introduction Cortical folding mainly takes place in the third trimester of pregnancy and
may therefore be influenced by preterm birth. The aim of this study was to evaluate the
development of specific cortical structures between 30 and 40 weeks postmenstrual age in
extremely preterm infants and to correlate this to clinical characteristics and outcome at
two years of age.
Methods 71 preterm infants with a gestational age at birth <28 weeks and two scans in the
neonatal period, one around 30 weeks and again around 40 weeks postmenstrual age,
were included. The central sulcus (CS), lateral fissure (LF), insula (INS), superior temporal
sulcus (STS), postcentral sulcus (PCS), superior (SFS) and inferior frontal sulcus (IFS) were
identified with Brainvisa®. Asymmetry in folding and growth between both scans were
assessed. Multivariable analysis was performed to test the influence of clinical
characteristics and neurodevelopmental outcome.
Results CS, LF and INS were present at 30 weeks in all infants, whereas the other sulci were
only seen in part of the infants. Relative growth was largest in the SFS. A rightward
chapter 5
asymmetry of the surface area was seen in development except for the LF, which showed a
leftward asymmetry at both time points. Lower birth weight z-score, multiple pregnancy
and prolonged mechanical ventilation showed negative effects on cortical folding of the
CS, LF, INS, STS and PCS, mainly on the early scan. The correlation with outcome was
strongest for receptive language.
Discussion Cortical folding in preterm infants proceeds from the central sulci towards the
occipital and frontal poles, with the right hemisphere developing before the left. Sulci
developing the earliest were most affected by clinical factors. A clear correlation between
cortical folding and neurodevelopmental outcome at two years was found.
86 part 2 qualitative assessment of magnetic resonance imaging in relation to outcome

Introduction
D
uring the third trimester of pregnancy, large morphological changes occur in the
human brain. Not only an impressive volumetric growth, but also the majority of
cortical gyrification and sulcation takes place during this period of brain development,
changing the human brain from its largely lisencephalic appearance at 24 weeks of gestation
to a brain folded similarly to an adult brain at term age.1-4 Many different mechanisms to
explain cortical folding have been proposed and no clear consensus has been reached so far.
Two of the most popular theories are the tension-based theory, that states that tension along
growing axons causes folding,5 and the differential growth hypothesis, which states that
folding is a result of the slower growth rate of the subcortical layers compared to the cortex.6 A
recent paper integrated both theories and showed a model of the human brain with a
morphogenetically growing outer surface and a stretch-driven growing inner core.7 This model
was calibrated on preterm MRIs and seemed to be able to predict cortical folding between 27
and 32 weeks of gestation.
Over the last decades, magnetic resonance imaging (MRI) has been used to visualize these
changes in vivo and to describe the standard order in which the gyri and sulci develop.8-10 With
foetal MR imaging, although it still has its methodological challenges mainly related to motion
chapter 5
of the infant, normal development can be studied in vivo. Studies using foetal MRI have shown
the characteristic pattern of folding, with primary folds actively developing from 25 weeks, and
secondary folds starting to delineate from 30 weeks onwards.2,9 Preterm infants imaged ex
utero show a similar pattern of folding.8 Infants born extremely preterm spend this critical
period of brain development outside the womb, in a neonatal intensive care environment in
which they are exposed to a multitude of potentially damaging factors. This may lead to
disturbances in the normal folding processes taking place simultaneously,8,11 which may be a
partial explanation for the impairments in neurodevelopmental outcome of especially the
cognitive and behavioural domains that are frequently seen in this population.12,13 So far,
studies evaluating cortical folding in preterm infants up to term equivalent age (TEA) have
been mainly cross-sectional, and measurements were applied to the overall brain or large
brain regions.8,12,14 Little is known about the longitudinal development of specific sulci and
their relationship with neurodevelopmental outcome.
The aim of this study was to evaluate cortical folding of specific sulci in extremely preterm
infants and to correlate this with both clinical characteristics and outcome at two years of age,
in a longitudinally scanned cohort of preterm infants.
Methods
Cl inica l data
Between June 2008 and March 2013, preterm infants with a gestational age at birth between
24 and 28 weeks, admitted to the level three neonatal intensive care unit of the Wilhelmina
is early cortical folding related to clinical risk factors and neurodevelopmental outcome ? 87
a longitudinal mri study of preterm newborns
Children’s Hospital, were consecutively included in a prospective neuroimaging study. For this
study, infants were scanned twice: once –if clinically stable- around 30 weeks gestation (28.7-
32.7 weeks) and again around TEA (40.0-42.7 weeks). From the 265 infants born during this
period, serial imaging data were acquired in 137. 43 of those 137 infants were not yet two
years of corrected age at the time of this study and outcome data were therefore not available.
From the remaining 94, severe motion on either of the scans was reason for exclusion in 10
infants and in an additional 13 infants, segmentation errors were too severe to correctly
process the data, leading to a final 71 infants to be included in this study. Figure 1 shows a
flowchart with the final inclusion of all infants.
Perinatal data were obtained by chart review. Birth weight z-scores (BWZ) were computed
according to the Dutch Perinatal registry reference data.15 Corrected weight at scan was
defined based on the z-score of the absolute weight at scan, thus correcting for the intrinsic
differences between boys and girls. Socioeconomic status was determined based on maternal
educational level.16 Serial cranial ultrasound (cUS) was obtained and reported as part of
standard clinical care. Intraventricular haemorrhage (IVH) grading on cUS was scored
according to Papile and post-haemorrhagic ventricular dilatation (PHVD) was defined as a
ventricular index 4 mm > 97th percentile, according to Levene.17,18 Permission from the
cha p t e r 5
medical ethics review committee was obtained.
Figure 1 Flowchart of inclusion. Flowchart

showing which infants were included, as well
as the reasons for exclusion.

MR I acqu i si ti on
MR imaging was performed on a 3.0 Tesla MR system (Achieva, Philips Medical Systems, Best,
the Netherlands). At 30 weeks, infants were scanned in an MRI compatible incubator (Dräger
MR Incubator, Lübeck, Germany and later Nomag® IC 3.0, Lammers Medical Technology GmbH,
Lübeck, Germany, with a dedicated neonatal head coil), while at TEA the 8-channel sense head
coil was used. The protocol included T2-weighted imaging in the coronal plane (turbo spin
echo, at 30 weeks: repetition time [TR] 10085 ms; echo time [TE ]120ms; slice thickness 2 mm,
in-plane spatial resolution 0.35x0.35 mm; at TEA: TR 4847 ms; TE 150 ms; slice thickness 1.2
mm, in-plane spatial resolution 0.35x0.35mm). After evaluation by a paediatric radiologist, all
scans were re-assessed by two neonatologists (LdV and MB) with over 10 years of experience
in neonatal neuro-imaging. The presence of IVH, periventricular haemorrhagic infarction,
PHVD, cystic periventricular leukomalacia, punctate white matter lesions, central or cortical
grey matter infarctions and punctate or larger lesions in the cerebellum were scored.
MR I p ost -p r oc e s s i ng
T2-weighted images were segmented with a recently developed segmentation method,
defining masks of the cortical grey matter, white matter and cerebrospinal fluid.19 These
cha p t e r 5
segmentations were then used to obtain meshes of the inner cortical surface of both
hemispheres, by adapting the anatomical pipeline of the Brainvisa® software with additional
software dedicated to the baby brain.20,21 Where necessary, individual segmentations and
meshes were manually corrected. From these meshes, using statistical probabilistic anatomy
map modelling (SPAM), implemented for the adult brain, automated sulcal detection and
recognition was possible.22 The accuracy of the sulcal identification was checked visually and if
necessary, adjusted for the sulci of interest in each hemisphere. Left and right hemispheres, as
well as early and TEA scans of each infant, were checked together to provide intra-subject
reliability of the sulci labelling across hemispheres and ages, and to be able to assess inter-
individual differences in sulcus formation. Since not all sulci were visible at the first scan, we
decided to analyse only those sulci that were visible at the first scan in over 50% of infants.
Additionally, due to the difficulty of carefully segmenting the medial part of the brain in
particular in the occipital region, only sulci of the lateral surface were taken into account.
Since the pre-central sulcus has been shown to develop from the fusion of the posterior parts
of the superior (SFS) and inferior (IFS) frontal sulcus and can therefore be hard to distinguish
at the early scan,8 it was considered with either SFS or IFS at both time points. Only the primary
fold of each sulcus was labelled, taking into account the part of the fold that was visible at 30
weeks. Secondary and tertiary folds and branches were not taken into account to make
comparison between 30 and 40 weeks more reliable. An example of the selected sulci at both
time points is shown in figure 2.
Pa r a me t e rs c h a r act e r i z i ng co r t ical fo ldi ng
For each sulcus, sulcal surface area and mean geodesic depth were extracted from the data
using Brainvisa®.20,21 Surface area was chosen, since it is not affected by either voxel size or the
level of fold opening, as the sulcal volume would be. For the lateral fissure, sulcal length was
also computed to be able to study sulcal asymmetries. For the Insula (INS), only surface area
was computed, since depth of the INS provides no information. The INS as computed with
Brainvisa® does not refer to the sulcus but to the whole structure, as sulcal detection in this
area is not possible with this method at this age. For the sake of clarity, we will however keep
referring to the INS as a sulcus as well throughout the rest of the paper. Spherical brain hulls
were computed for each hemisphere and surface areas were computed as a measure of overall
hemisphere size by computing the morphological closing of the brain segmentation.23,24 To
correct for inter-individual variability in brain size, a sulcal index per sulcus was calculated by
dividing the surface area of the given sulcus with the surface area of the hemisphere hull.
Asymmetry indices were calculated for all sulci as [surface left hemisphere – surface right
hemisphere]/[surface left hemisphere + surface right hemisphere]. For the LF, length was
calculated as well to be able to study whether asymmetry in LF length, that has been shown
before to exist in healthy adults,25 is already present at this age. For both surface area and
cha p t e r 5
mean depth, the absolute growth rate between both scans was defined in each hemisphere as
the difference between the TEA and the early scan measures, divided by the difference in
postmenstrual age in weeks (PMA) between both scans. Relative growth rate was defined as
the absolute growth rate per week, divided by the corresponding measure at 40 weeks
corrected age.
N e u r od e v e lo p m e nta l o u tco m e
Neurodevelopmental outcome was assessed at either 24 or 30 months’ corrected age (CA, i.e.
corrected for gestational age at birth), dependent on inclusion in a European study.26
Neurodevelopmental assessment was performed by a single developmental specialist (ICH)
using the cognitive, fine motor and gross motor subtests of the Bayley Scales of Infant and
Toddler Development, third edition (BSITD-III).27 In the infants seen at 30 months’ CA, the
expressive and receptive language subtests were also performed. Composite and scaled scores
corrected for gestational age were calculated (mean [standard deviation] in a normative
population: 100 [15] and 10 [3], respectively). No differences in scores were found between
infants tested at 24 (n=34) and 30 months (n=37) and data of both groups were therefore
combined.
S tati st ica l a na ly s i s
Statistical procedures were performed using Matlab (MATLAB and Statistics Toolbox Release
2013b, The MathWorks Inc., Natick, MA, USA) and R version 2.15.3 (www.r-project.org). Paired
samples t-tests were used to compare the left and right surface areas and mean depths. We

first evaluated the impact of clinical factors on cortical folding. General linear modelling was
performed to correlate brain morphological measures (as dependent variables) with the
clinical characteristics of the infants (as independent variables). For each sulcus, absolute
surface area, mean depth and sulcal index at 30 and 40 weeks, and absolute growth rate of
surface area and depth between both scans were studied. All models were corrected for PMA
at MRI. For the models at 30 weeks, the infants in which the sulcus was not yet identifiable
were excluded. Sex and multiple pregnancy were included in the model based on previous
literature.8,28,29 Since only two infants were small for gestational age (SGA, defined as a birth
weight below the 10th percentile), birth weight z-score was included instead of SGA. Severe
IVH (grade III-IV) was included as a measure for brain injury and finally prolonged mechanical
ventilation (>7 days) was chosen as an overall parameter measuring severity of illness. To
assess whether there was an intrinsic effect of sex on cortical folding, not explained by the
difference in weight at scan, general linear modelling was repeated with the interaction
between sex and corrected weight at scan in the model instead of birth weight z-score.
Next, we evaluated whether outcome parameters around two years of age may be explained
by cortical folding measures in the neonatal period. General linear modelling was performed
to correlate outcome parameters (as dependent variables) and morphological measures (as
cha p t e r 5
independent variables: absolute surface area, mean depth and sulcal index at 30 and 40 weeks,
and growth between both scans for each sulcus). In these analyses, all data were corrected for
PMA, socio-economic status (as measured by educational level of the mother) and sex. For the
cognitive subscale all sulci were taken into account (one after the other), because we had no
hypothesis on specific brain regions to be involved in cognitive development. For the gross
motor subscale the CS, SFS and IFS were considered, since the pre-central sulcus was included
in both the SFS and IFS. For the fine motor subscale, the PCS was added to the CS, SFS and IFS
because of the possible role of the parietal lobe on praxia. Since peri-sylvian regions are
activated early on by speech perception in preterm born infants,30 the LF, INS, STS and IFS were
tested for the language subscales. To correct for the multiple sulci assessed, a p-value of
≤0.007 (0.05/7 sulci) was considered significant in the multivariable analyses.
Results
Pro g r e ssi on of co r t ica l fo l di ng
At the first scan, the CS, LF and INS were present in all infants. For the other sulci, the presence
at the first scan differed, as can be seen in table 2. Of those sulci, the STS was most often
present, followed by the IFS, PCS and SFS. When a sulcus was only present at one side, this was
more often in the right than the left hemisphere (p<0.004 for all sulci). As shown in the graphs
in figure 2 and 3, large individual differences were seen between infants scanned at the same
PMA, with a wider variation in surface area at 40 weeks compared to 30 weeks. Indeed, at 30
weeks morphological measures correlated strongly with PMA at scan for all sulci, whereas at
40 weeks these correlations were no longer significant (figure 2). Surface area at 30 weeks was
Ta b le 1 . C lin i ca l character i sti cs of all i nfan ts
Clinical characteristics Mean (range) or N (percentage)
Gestational age (weeks) 26.5 (24.4 – 27.9)
Birth weight (grams) 903 (460 – 1350)
Birth weight z-score 0.4 (-2.5 – 1.8)
Sex (male/female) 36/35 (51/49%)
Multiple pregnancy (single/multiple) 49/22 (69/31%)
Small for gestational age 2 (3%)
>7 days of mechanical ventilation 25 (35%)
PDA requiring treatment 35 (49%)
Chronic lung disease 20 (28%)
Culture proven sepsis 25 (35%)
Necrotizing enterocolitis requiring surgery 7 (10%)
IVH (0/1/2/3/4) 44/8/11/5/3 (62/11/15/7/4%)
PHVD +4mm > 97th centile 7 (10%)

chapter 5
Postmenstrual age at early MRI (weeks) 30.7 (28.7 – 32.7)
Week of post-natal life at early MRI (weeks) 4.1 (0.9-8.1)
Weight at early MRI (grams) 1205 (620-1785)
Postmenstrual age at TEA MRI (weeks) 41.2 (40.0 – 42.7)
Week of post-natal life at TEA MRI (weeks) 14.7 (12.3-17.1)
Weight at TEA MRI (grams) 3326 (2045 – 4315)
Follow-up (FU)
Maternal education (low/middle/high) 19/24/28 (27/34/39%)
Corrected age (CA) at FU (months) 27.2 (23.2 – 30.8)

Subgroup seen at 24 months (n=35) 24.2 (23.2-27.0)
Subgroup seen at 30 months (n=36) 30.1 (29.5-30.8)
Cognitive composite score 104 (85 – 130)
Fine motor scaled score 13 (9 – 19)
Gross motor scaled score 10 (3 – 17)
Total motor composite score 109 (79 – 148)
Expressive language scaled score (N=34) 12 (7-17)
Receptive language scaled score (N=35) 11 (8-16)
Clinical characteristics, MRI parameters and outcome data of all infants are summarized in table 1.
Mean gestational age at birth was 26.5 weeks, mean postmenstrual age at first scan 30.7 weeks and at the term equivalent scan
41.2 weeks.
Ta bl e 2 . P re s e n ce o f su lci at 3 0 weeks
Sulcus Presence
Both sides Only right side Only left side Neither side
Central sulcus 71 (100%) - - -
Lateral fissure 71 (100%) - - -
Insula 71 (100%) - - -
Superior temporal sulcus 54 (77%) 5 (7%) 5 (7%) 7 (10%)
Inferior frontal sulcus 34 (48%) 16 (23%) 7 (10%) 14 (20%)
Postcentral sulcus 28 (39%) 12 (17%) 7 (10%) 24 (34%)
Superior frontal sulcus 27 (38%) 11 (15%) 4 (6%) 29 (41%)
predictive for surface area at 40 weeks in all sulci (p<0.02) independently from the delay
between both scans and depth at 30 weeks predicted depth at 40 weeks for all sulci except
the bilateral LF and right INS (p<0.006). For all sulci, correlations between morphological
measurements (surface area and depth) and overall brain size (defined as the surface of the
brain hull, per hemisphere) were also found for both time points, while taking into account
cha p t e r 5
weight at scan.
All sulci showed a clear growth between both scans, both in surface area (figure 2) and in
mean geodesic depth. Absolute and relative growth rates are shown in figure 4. The absolute
growth shows a clear increase of all studied sulci with the increase in overall brain size being
about tenfold as large. When studying relative growth, the individual sulci grow relatively
more compared to the overall brain. The SFS showed the largest relative growth ratio,
followed by the PCS, IFS and STS. The three sulci that develop first and were present at all 30
week scans (CS, LF and INS) showed a relatively slower growth, although absolute growth was
largest in these sulci (p<0.001).
Sulcal a sy mme t r i es
Mean asymmetry indices and standard deviations are represented in figure 5. Significant inter-
hemispheric differences in surface area were found with a rightward direction for the STS at
both time points, for the IFS at 30 weeks and for the INS at 40 weeks. A leftward asymmetry
was found for the LF (both surface area and sulcal length) at both time points. Although
significant, asymmetry indices were below the 5% threshold that is often used to describe
asymmetry31, for the LF (surface area and length at 30 weeks, surface area at 40 weeks), and
the INS (surface area at 40 weeks). Inter-hemispheric differences were also significant for
sulcal depth with a right-ward direction for the STS and IFS at both time points, for the CS at
30 weeks and for the SFS at 40 weeks, and a left-ward direction for the LF at 40 weeks.
Cl i n ica l c h a r act er i s t ics i nf l u e nci ng co rt ical fo ldi ng
Table 3 shows the results of the multivariable analyses combining sulcal surface area and depth
at both 30 and 40 weeks, and absolute growth rate with clinical characteristics, including
effect sizes. No correlation was found for any sulcal index (surface area divided by overall
hemispheric hull area).
When assessing overall brain size (i.e. bilateral hull surface area) at 30 weeks, we found a
positive effect of BWZ and high grade IVH, and a negative effect of prolonged mechanical
ventilation. At 40 weeks, the effects of BWZ and prolonged ventilation persisted. Overall brain
growth was influenced by multiple pregnancy, BWZ, prolonged ventilation and high grade IVH
for the left hemisphere, but not for the right hemisphere.
In the next paragraphs, the results of Table 3 are summarized by clinical characteristic and not
by sulcus, to highlight the reproducible significant effects. First, maternal education did not
have a direct effect on the morphological measures for any sulcus.
No significant effects of sex were found in the multivariable model for any sulcus. To further
assess an intrinsic effect of sex on cortical folding, not explained by a difference in weight,
another multivariable model was tested which included weight at scan corrected for age at
scan and sex, as well as sex and the interaction between both parameters (birth weight z-score
cha p t e r 5
was not considered in this model). Using this model, the left STS and right IFS of boys showed
an initially lower surface area at 30 weeks, with a steeper growth and a larger surface area at
40 weeks (p=0.01 for all analyses). For the left PCS, the surface area of boys was already larger
at 30 weeks and growth was slightly steeper (p=0.02), leading to larger surface areas at 40
weeks (p=0.04). For the left LF, this effect was seen for growth (p=0.02), but no significant
differences between boys and girls with a similar weight were found at 40 weeks. To
summarise, we observed that some sulci showed higher surface are or growth in boys
compared to girls with a similar weight. However, none of these differences did survive the
correction for multiple comparisons.
A higher birth weight z-score had a significant positive influence on surface area of the
bilateral LF and INS at 30 weeks. At 40 weeks, this effect was observed for the surface area of
the left INS and bilateral CS, and for depth of the left STS. Birth weight z-score had a positive
effect on absolute growth rate of right CS surface area and depth.
Infants from a multiple pregnancy showed smaller surfaces compared to singletons for the
bilateral INS, and a smaller depth of the left STS, with the effect being more pronounced at 30
weeks (no effect for the right INS at 40 weeks).
Infants requiring prolonged mechanical ventilation, used as a measurement for illness severity,
had smaller surface areas of the bilateral LF and INS at 30 weeks, as well as a smaller depth of
the right CS. At 40 weeks, these effects were no longer found, but the right STS surface area and
left PCS depth were lower in infants that were ventilated longer than 7 days. Absolute growth
rate of the left PCS surface area was also negatively influenced by prolonged ventilation.

Ten infants had a grade III-IV IVH. Presence at 30 weeks of the STS, PCS, SFS and IFS was lower
in those infants compared to the 61 infants without severe IVH. Indeed, in infants with severe
IVH, the absence of the sulcus was about two times higher for the bilateral STS, IFS, left PCS
and right SFS, and a third higher for the left SFS. Only the right PCS was more frequently
present in infants with severe IVH (70%), than in those without (54%). None of these
differences were however significant, and the effect of high grade IVH did not reach
significance in the multivariable models of sulci folding measurements (Table 3). Taking the
side of the IVH into account (left, right or bilateral) did not yield significant differences either.
Rel ati on ships b et w e e n co r t ica l fo l din g and

n eu r od e v e lop m enta l o u tco m e
The relation between outcome during toddlerhood (at 24-30 months) and cortical folding
during the neonatal period (at 30 and 40 weeks) was tested per subscale of the BSITD-III. Table
4 shows an overview of all results from the multivariable analyses.
No correlations between gross and fine motor or cognitive outcome and brain size
(hemispheric hull surface area) were found. Gross motor outcome could be predicted by the
early development of the left IFS (bilateral surface area, right depth and left sulcal index) and
cha p t e r 5
the depth of the right SFS at 30 weeks. Fine motor outcome was significantly correlated with
right PCS surface area and right SFS depth at 40 weeks. The cognitive subscale showed a
correlation with absolute growth rates of left LF and SFS depth. Although these differences
were significant, effect sizes were small for all correlations (≤3%, Table 4). Maternal educational
level had a significant influence on outcome across all models at 30 and 40 weeks (but not for
growth), with an effect size between 7 and 18%. Again, sex was not significant in any of these
models. When infants with brain injury (IVH grade III-IV) were excluded, only the correlation
between cognitive outcome and absolute growth rates of the left SFS depth remained
significant.
In a subgroup of infants, the expressive and receptive language subtests were administered as
well. At 30 weeks, correlations between receptive language and folding measures were only
observed for the left IFS (surface area and sulcal index). At 40 weeks, effects were found for the
bilateral hemispheric size, bilateral LF ( surface area, depth and sulcal index), IFS (depth), STS (
right: surface area, depth, and sulcal index, left: depth) and left INS (surface area). The
expressive language subscale correlated with the left IFS at 30 weeks (surface area) and left LF
at 40 weeks (depth). Maternal education remained significant in all analyses. Again, effect sizes
were small. When repeating the analyses after exclusion of the infants with brain injury (IVH
grade III-IV), only the correlation between receptive language and the left IFS (surface area and
sulcal index) and between receptive language and the left IFS (surface area) lost significance.
All other correlations remained significant.
cha p t e r 5
Figure 2 Longitudinal development of sulci. Example of meshes showing the selected sulci in one infant. Meshes from the left (A,
C) and right (B, D) hemispheres are shown for the same infant at 30 weeks (top) and at 40 weeks (bottom). The following
colours represent the sulci: red – central sulcus; blue – lateral fissure; light green – insula; vale pink – superior temporal sulcus;
purple – postcentral sulcus; green – superior frontal sulcus; pink – inferior frontal sulcus. Around these examples, graphs are
depicted showing the absolute surface areas of each sulcus at 30 and 40 weeks for all infants as a function of postmenstrual age
at scan. Note the large variation in sulcal surface area at 40 weeks for all sulci, independently from age at MRI, while around 30
weeks, sulcal surfaces show increases with an increasing age at scan.
Figure 3 Inter-individual variability in folding progression at 30 weeks. Mesh of the right hemispheres from five different
infants, all scanned at a postmenstrual age of 29.86 weeks. Note the differences in the number of sulci present between the
infants, and the difference in trajectories of especially frontal sulci. The same colour representation as in figure 2 is used.

Figure 4 Sulcal growth rates. This figure shows the absolute and relative growth rates of the sulci. Growth is represented as
either absolute (in mm2) or relative (%) surface increase per week, compared to the surface at the term equivalent scan. Note
that relative growth rate is largest in the superior frontal sulcus, followed by the postcentral, inferior frontal and superior
temporal sulci. The central sulcus, lateral fissure and insula, that are already well established at the first scan, show a smaller
relative growth rate, although absolute growth rates are high.
cha p t e r 5
Figure 5 Sulcal asymmetries. The asymmetry indices of the surface area of all sulci, and of the sulcal length of the lateral fissure
are represented in this figure for the scans at 30 (A) and 40 (B) weeks. The square represents the mean value, with the whiskers
extending 1 standard deviation. Significant differences are indicated as follows: <0.05 ‘*’, <0.01 ‘**’, <0.001 ‘***’.
Discussion
This study describes extra-uterine third trimester cortical folding in a cohort of extremely
preterm infants with serial MRI, showing a rightward asymmetry in brain development and
regional growth differences. Additionally, the influence of several clinical characteristics and
relationship with outcome were studied for specific primary sulci, providing valuable
information on the deviations caused by preterm birth.
Ta b le 3 . C lin i ca l character i sti cs i nflu encin g co rt ica l fo l din g
Sulcus 30 weeks
Surface area or Depth Characteristic P
Mult
BWZ 0.0001
Hull surface Area left
MV7 0.0001
IVH34 0.005
BWZ 0.0001
Area right MV7 <0.0001
IVH34 0.0054
BWZ
Area left
BWZ
Central sulcus Area right
MV7
Depth right 0.0019
BWZ
BWZ 0.0013
Area left
MV7 0.0021
cha p t e r 5
Lateral fissure
BWZ 0.0002
Area right
MV7 0.0015
Mult 0.0046
Area left BWZ <0.0001
MV7 0.0005
Insula
Mult 0.0012
Area right BWZ 0.0001
MV7 0.0051
MV7
Area right
Superior temporal sulcus Mult
Depth left 0.001
BWZ
Area left MV7

Postcentral sulcus
Depth left MV7
Only significant correlations are presented (p<0.0007). The corrections for postmenstrual age at MRI are not shown.
Abbreviations: BWZ = birth weight z-score; IVH34 = intraventricular haemorrhage grade III or IV; Mult = multiple pregnancy;
MV7 = mechanical ventilation >7 days.
Fol di n g p ro g r es s i o n a nd i nt er - h em is phe ric asy m m e t rie s

Previous studies have shown that sulci maturation progresses in a specific order, starting with
the central structures,2-4,8 progressing to peripheral structures in an occipital to frontal
direction.4,32,33 The current results corroborate those data, as shown by differences in the sulci
presence or absence on the early scan. The most central sulci (CS, LF and INS) were present in

40 weeks Absolute growth rate
Effect size Effect size Effect size

P P
(%) (%) (%)
0.0046 -7.5
5.4 0.0011 5.2
0.0002 4.2
-8.9 0.0005 -9.3
0.0019 -5.7
8.5 0.0028 11.2
5.3
0.0001 4.7
-8.9
0.0012 -6.1
8.3
0.0015 7.2 0.0004 10.8

0.0026 6.1
-8.9
0.006 7.7
18.8
-27.6
cha p t e r 5
18.4
-23.5
-22.1
20.9
-26.4 0.0062 -11.4
-23.0 0.0067 6.6
18.0
-19.0
0.0007 -21.0
0.0022 -10.1
-18.1
0.0049 5.4
0.0012 -19.6
0.0045 -12.9
all infants at the first scan, around 30 weeks of gestation. Then folding progressed from the
temporal to the frontal and parietal lobes: the STS was seen in 90%, followed by the IFS (80%),
PCS (66%) and finally the SFS (59%). These data suggest a slightly later development of the sulci
compared to in vivo studies of fetal brain development, where these sulci are shown to
develop between 28 and 30 weeks and thus should be present in all infants.1,2,9,34 In agreement
Ta b le 4 . Cort i ca l foldi ng mea su re s i nfluen cin g n euro develo pmen tal o utc o me
Outcome measure Sulcus Surface area or Depth
All infants (n=71)
Fine motor Postcentral sulcus Area right

Superior frontal sulcus Depth right
Gross motor Superior frontal sulcus Depth right
Inferior frontal sulcus Area left
Sulcal index left
Area right
Depth right
Cognition Lateral fissure Depth left
Superior frontal sulcus Depth left
Infants with language

subscores (n=35)
chapter 5
Receptive language Hull surface Left

Right
Lateral fissure Area left
Area right
Depth left
Depth right
Sulcal index left
Sulcal index right
Insula Area left
Sulcal index left
Superior temporal sulcus Area right
Depth left
Depth right
Sulcal index right
Depth left
Depth right
Sulcal index left
Expressive language Lateral fissure Depth left
* Effect size calculated as the absolute increase in scaled scores when surface or depth would increase 10%.

30 weeks 40 weeks Absolute growth rate
Effect size
P Effect size P Effect size P
0.006 0.1
0.005 0.12
0.007 -0.61
0.006 -0.17
0.001 -0.27
0.0008 -0.24
0.002 -0.64
0.004 0.21
<0.0001 0.55
chapter 5
0.001 0.21
0.001 0.21
0.002 0.11
0.004 0.11
0.0005 0.14
0.002 0.13
0.002 0.12
0.004 0.12
0.004 0.12
0.003 0.14
0.004 0.11
0.0009 0.14
0.001 0.13
0.004 0.12
0.005 -0.02
0.0008 0.12
0.001 0.13
0.005 -2.1
0.006 0.13
0.006 -0.02
with previous studies,2,8 this is probably related to the post-processing technique used to
identify the sulci, which failed to detect small dimples on the inner cortical surface.
Relative growth was largest for the sulci that were least developed at 30 weeks, highlighting
the intense folding taking place in the last 10 weeks of pregnancy, whereas those sulci that
develop first were already well developed before the first measurement. These regional
differences in growth rate seem to continue after birth. Hill and colleagues compared
morphology measures of 12 term-born infants with 12 healthy young adults, and found
regional differences in the degree of expansion, with expansion being twofold higher in the
areas of the four sulci that were least developed at 30 weeks.35 The higher expanding regions
were less mature at term gestation on a cellular level as well.35
The development of the human brain is known to be asymmetrical, with a slightly earlier
development of the right hemisphere compared to the left. 4,32,33
In this study, several
hemispheric asymmetries were found. For most sulci (INS, STS, IFS), a rightward asymmetry
was appreciated. Only the lateral fissure showed an asymmetry towards the left hemisphere.
For STS and LF the asymmetry was seen at both ages, while the IFS only showed a significant
asymmetry at 30 weeks and the INS at 40 weeks. Thus the three central sulci (CS, LF, INS) that
folded early on, barely changed in asymmetry index from 30 to 40 weeks. In contrast, the
cha p t e r 5
disappearance of rightward asymmetry for IFS, that developed later on, suggests that a delay in
left hemisphere development is a likely explanation for this asymmetry, with the left
hemisphere catching up with the right during this period.
The STS rightward asymmetry has been described in many previous studies of brain
development before4 and has been found in foetuses,1,36 preterm8,28 and term born infants.37-39
The STS remains asymmetrical in adults, suggesting an underlying difference in structure
rather than a delay in folding to be responsible for this asymmetry.40-42 The left-ward
asymmetry of the LF surface area has also been found before in studies of preterm infants,
mostly focused on the planum temporale.4,43,44 Studies in infancy also showed this
asymmetry.37,39,44 In addition, sulcal length was larger on the left side as well, and asymmetry
increased from 30 to 40 weeks. This is in agreement with the age-related increase observed in
a cross-sectional study of children, adolescents and young adults45 and with an earlier study in
healthy adults.25 A larger LF on the left side has also been found in chimpanzees.46,47 This left-
sided asymmetry of the LF surface area therefore probably has an old origin in the primate
lineage, which may have been amplified in humans due to the early language acquisition that
occurs predominantly left-sided, as shown in functional MRI studies of preterm newborns30
and 2 month old infants.48
I n t e r -i n di v id u a l va r iabi l i t y i n co rt ical fo ldi ng and t he im pact

of c l i n ica l c h a r act er i s t ics
Inter-individual differences were noted between the infants in both size and morphology of
the folds, in agreement with earlier studies, suggesting that genetic factors partly influence the

folding process.40,49 The large inter-individual differences and the increase of these variations
towards term equivalent age, may also be the explanation for the few effects of clinical risk
factors on cortical growth that were found in this cohort.
The CS, LF and INS were the sulci most influenced by clinical characteristics. These sulci are all
thought to form before 28 weeks.2 In other words, these sulci are actively folding during the
period an extremely preterm infant is most severely ill (i.e. the first weeks ex utero, and
possibly also the last weeks in utero, depending on the foetal history and reason for preterm
birth). It is therefore not unreasonable to think that developmental disturbances related to
preterm birth and severity of illness will have a larger impact on these early developing sulci
compared to the later developing sulci, whose active folding will occur during a more stable
period for most of the extremely preterm infants.50
Intrauterine growth retardation has been described as a risk factor for abnormal cortical
folding before.28 In this population, only two infants were small for gestational age, which was
probably due to an selection bias, as infants born small for gestational age during the inclusion
period more often died before admission to the neonatal intensive care unit or did not receive
an early scan. Proper correlations could therefore not be made. Nevertheless, the effect of
birth weight z-score was obvious throughout analyses for most sulci. Because sulcal indices
cha p t e r 5
were not affected by this factor, it suggests that an overall reduction in brain size led to
reductions in surface area and depth for most sulci.
Lower folding rates of the INS and STS were seen in this study in infants born after multiple
pregnancy compared with singletons, particularly at 30 weeks and to a lesser extent at 40
weeks. An harmonious delay of folding in preterm infants from twin pregnancies has been
reported at birth28 and on autopsy,4 and this effect seems largely diminished by TEA.28 The large
negative effect of prolonged mechanical ventilation, used in this study as a surrogate for
overall severity of illness, on sulcal measures particularly at 30 weeks suggests that illness
severity in the immediate perinatal period strongly affects cortical folding.51 At 40 weeks, this
effect was less pronounced, suggesting that folding may catch up during the period between
both scans. Alternatively, the diminishing effects may be related to the increasing inter-
individual variability, as shown in figure 2. The duration of the insult to the white matter is
likely to be subacute or chronic in preterm infants, rather than acute, as seen in term
newborns with brain injury. The duration of the insult and the age at onset (e.g. prenatal or
not), possibly related to the underlying cause for preterm birth, may influence cortical folding
as well. A larger cohort of preterm newborns with careful antenatal history is needed to study
these effects.
Only ten infants in this cohort had a severe IVH (grade III-IV), and these infants seemed to
show delayed folding, as suggested by the less frequent presence of STS, PCS, SFS and IFS. This
delay may be due to increased intracranial pressure in infants with post-haemorrhagic
ventricular dilatation, or to disturbed connectivity in infants with a periventricular
haemorrhagic infarction.14 However, differences were not significant in the clinical models.
Other white matter and cerebellar pathologies were sparse in this population and were
therefore not taken into account. In another study, which included part of the infants of this
study as well, infants with brain injury showed a decrease of global folding measures and an
increase of cortical thickness, which was more pronounced at 40 weeks.52 Further elucidating
the effect of the different types of brain injury on cortical folding may offer options for
intervention.
F u n cti on a l o u tco m e p r e dict ed b y e arly co rt ical fo ldi ng

Outcome measures during toddlerhood were related to the folding of several sulci during
infancy. Especially many correlations were found for receptive language and perisylvian sulci,
as well as for total brain size. Most correlations were also significant for sulcal indices, thus
partly excluding the global effect of total brain size on these relationships. This is in agreement
with an earlier study of a British population that showed a relation between growth of the
surface area between 24 and 44 weeks and outcome at two and six years for cognitive
functions including language development, although the effect sizes reported in that study
were larger compared to our results.13 The smaller effects in our study may be explained by the
young age at follow-up (2-2.5 years) or may be due to the use of other neurodevelopmental
cha p t e r 5
tests. In our cohort with a relatively unimpaired outcome at two years of age, outcome at a
later age may be needed to show if the effects of cortical folding on cognitive and language
functions will increase with age. Additionally, the large influence of maternal educational level
suggests that both genetic and environmental factors contribute to early development.
Some limitations of this study should be noted. First, the method for sulci identification
required some manual correction of the segmentations and meshes of the inner cortical
surfaces, particularly for the 40 week scans in regions where sulci were very narrow. This was
however done in a systematic way. Further improvement of the tissue segmentation may
reduce the need for manual corrections and allow the inclusion of the medial and occipital
parts of the brain. Second, we focused on those infants with serial imaging and inclusion may
therefore have been biased towards the healthier part of the cohort as the more ill infants
were not stable enough to be scanned at 30 weeks. Third, outcome data were restricted to 2.-
2.5 years of age, and the effects on neurodevelopment during childhood could therefore not
be shown. Finally, the cohort was not large enough to extensively study clinical risk factors, and
interactions between prolonged mechanical ventilation and surgery for example could not be
studied.
In conclusion, this study shows that cortical folding in preterm infants, as measured with
longitudinal data, proceeds asynchronously from the central sulci towards the temporal,
parietal and frontal lobes, with the right hemisphere developing earlier than the left. By term
equivalent age these differences have largely disappeared and only the known asymmetries of

the STS and LF were still present, despite larger inter-individual variability in folding measures.
Multiple pregnancy, a lower birth weight z-score and prolonged mechanical ventilation were
related to a delay in cortical folding mainly on the early scan, particularly for sulci developing
the earliest. A clear correlation between cortical folding and neurodevelopmental outcome at
two years was also highlighted. Longer term neurodevelopmental follow-up and a
combination of these data with functional imaging are needed to further investigate these
findings.
Funding
infants from a study funded by the Wilhelmina Research Fund (10-427).
cha p t e r 5
References
1 Habas PA, Scott JA, Roosta A, Rajagopalan V, Kim affects the developmental synergy between
K, Rousseau F et al. Early folding patterns and cortical folding and cortical connectivity
asymmetries of the normal human brain observed on multimodal MRI. Neuroimage
detected from in utero MRI. Cereb Cortex 2014;89:23-34.
2012;22:13-25. 15 Visser GH, Eilers PH, Elferink-Stinkens PM, Merkus
2 Clouchoux C, Kudelski D, Gholipour A, Warfield HM, Wit JM. New Dutch reference curves for
SK, Viseur S, Bouyssi-Kobar M et al. Quantitative birthweight by gestational age. Early Hum Dev
in vivo MRI measurement of cortical 2009;85:737-44.
development in the fetus. Brain Struct Funct 16 Divisie Sociale en ruimtelijke statistieken SSt.
2012;217:127-39. Standaard Onderwijsindeling 2006; Editie
3 Battin MR, Maalouf EF, Counsell SJ, Herlihy AH, 2013/’14. Den Haag/Heerlen: Centraal Bureau
Rutherford MA, Azzopardi D et al. Magnetic voor de Statistiek; 2014.
resonance imaging of the brain in very preterm 17 Papile LA, Burstein J, Burstein R, Koffler H.
infants: visualization of the germinal matrix, early Incidence and evolution of subependymal and
myelination, and cortical folding. Pediatrics intraventricular hemorrhage: a study of infants
1998;101:957-62. with birth weights less than 1,500 gm. J Pediatr
4 Chi JG, Dooling EC, Gilles FH. Gyral development 1978;92:529-34.
of the human brain. Ann Neurol 1977;1:86-93. 18 Levene MI. Measurement of the growth of the
5 Van Essen DC. A tension-based theory of lateral ventricles in preterm infants with real-
morphogenesis and compact wiring in the time ultrasound. Arch Dis Child 1981;56:900-4.
central nervous system. Nature 1997;385:313-8. 19 Chita SM, Benders MJNL, Moeskops P, Kersbergen
6 Xu G, Knutsen AK, Dikranian K, Kroenke CD, Bayly KJ, Viergever MA, Isgum I. Automatic
PV, Taber LA. Axons pull on the brain, but tension segmentation of the preterm neonatal brain with
does not drive cortical folding. J Biomech Eng MRI using supervised classification. Proceedings
cha p t e r 5
2010;132:071013. of SPIE 2013;8669.

7 Budday S, Raybaud C, Kuhl E. A mechanical model 20 Leroy F, Mangin JF, Rousseau F, Glasel H, Hertz-
predicts morphological abnormalities in the Pannier L, Dubois J et al. Atlas-free surface
developing human brain. Sci Rep 2014;4:5644. reconstruction of the cortical grey-white
8 Dubois J, Benders M, Cachia A, Lazeyras F, Ha-Vinh interface in infants. PLoS One 2011;6:e27128.
LR, Sizonenko SV et al. Mapping the early cortical 21 Mangin JF, Riviere D, Cachia A, Duchesnay E,
folding process in the preterm newborn brain. Cointepas Y, Papadopoulos-Orfanos D et al. A
Cereb Cortex 2008;18:1444-54. framework to study the cortical folding patterns.
9 Garel C, Chantrel E, Brisse H, Elmaleh M, Luton D, Neuroimage 2004;23 Suppl 1:S129-S138.
Oury JF et al. Fetal cerebral cortex: normal 22 Perrot M, Riviere D, Mangin JF. Cortical sulci
gestational landmarks identified using prenatal recognition and spatial normalization. Med
MR imaging. AJNR Am J Neuroradiol Image Anal 2011;15:529-50.
2001;22:184-9. 23 Fischer C, Operto G, Laguitton S, Perrot M,
10 van der Knaap MS, van Wezel-Meijler G, Barth PG, Denghien I, Rivière D, Mangin J-F. Morphologist
Barkhof F, Ader HJ, Valk J. Normal gyration and 2012: the new morphological pipeline of
sulcation in preterm and term neonates: BrainVISA. 2012.
appearance on MR images. Radiology 24 Germanaud D, Lefevre J, Toro R, Fischer C, Dubois
1996;200:389-96. J, Hertz-Pannier L et al. Larger is twistier: spectral
11 Ajayi-Obe M, Saeed N, Cowan FM, Rutherford MA, analysis of gyrification (SPANGY) applied to adult
Edwards AD. Reduced development of cerebral brain size polymorphism. Neuroimage
cortex in extremely preterm infants. Lancet 2012;63:1257-72.
2000;356:1162-3. 25 Foundas AL, Faulhaber JR, Kulynych JJ, Browning
12 Kapellou O, Counsell SJ, Kennea N, Dyet L, Saeed CA, Weinberger DR. Hemispheric and sex-linked
N, Stark J et al. Abnormal cortical development differences in Sylvian fissure morphology: a
after premature birth shown by altered quantitative approach using volumetric magnetic
allometric scaling of brain growth. PLoS Med resonance imaging. Neuropsychiatry
2006;3:e265. Neuropsychol Behav Neurol 1999;12:1-10.
13 Rathbone R, Counsell SJ, Kapellou O, Dyet L, 26 https://www.i-med.ac.at/neobrain. 28-4-2014.
Kennea N, Hajnal J et al. Perinatal cortical growth 27 Bayley N. Bayley Scales of Infant and Toddler
and childhood neurocognitive abilities. Development, 3rd edition. San Antonio, TX:
Neurology 2011;77:1510-7. Harcourt Assessment. 2006.
14 Melbourne A, Kendall GS, Cardoso MJ, Gunny R, 28 Dubois J, Benders M, Borradori-Tolsa C, Cachia A,
Robertson NJ, Marlow N et al. Preterm birth Lazeyras F, Ha-Vinh LR et al. Primary cortical

folding in the human newborn: an early marker of 40 Van Essen DC. A Population-Average, Landmark-
later functional development. Brain and Surface-based (PALS) atlas of human cerebral
2008;131:2028-41. cortex. Neuroimage 2005;28:635-62.
29 Vasileiadis GT, Thompson RT, Han VK, Gelman N. 41 Ochiai T, Grimault S, Scavarda D, Roch G, Hori T,
Females follow a more “compact” early human Riviere D et al. Sulcal pattern and morphology of
brain development model than males. A case- the superior temporal sulcus. Neuroimage
control study of preterm neonates. Pediatr Res 2004;22:706-19.
2009;66:551-5. 42 Leroy F, Caj Q, Bogart S, Dubois J, Coulon O,
30 Mahmoudzadeh M, Dehaene-Lambertz G, Monzalvo K et al. A new human-specific brain
Fournier M, Kongolo G, Goudjil S, Dubois J et al. landmark: The depth asymmetry of the superior
Syllabic discrimination in premature human temporal sulcus. Proc Natl Acad Sci U S A
infants prior to complete formation of cortical 2014;accepted.
layers. Proc Natl Acad Sci U S A 2013;110:4846- 43 Dubois J, Benders M, Lazeyras F, Borradori-Tolsa C,
51. Leuchter RH, Mangin JF et al. Structural
31 Galaburda AM, Corsiglia J, Rosen GD, Sherman GF. asymmetries of perisylvian regions in the preterm
Planum temporale asymmetry, reappraisal since newborn. Neuroimage 2010;52:32-42.
Geschwind and Levitsky. Neuropsychologia 44 Glasel H, Leroy F, Dubois J, Hertz-Pannier L,
1987;25:853-68. Mangin JF, Dehaene-Lambertz G. A robust
32 Kersbergen KJ, Leemans A, Groenendaal F, van der cerebral asymmetry in the infant brain: the
Aa NE, Viergever MA, de Vries LS et al. rightward superior temporal sulcus. Neuroimage
Microstructural brain development between 30 2011;58:716-23.
and 40 weeks corrected age in a longitudinal 45 Sowell ER, Thompson PM, Rex D, Kornsand D,
cohort of extremely preterm infants. Neuroimage Tessner KD, Jernigan TL et al. Mapping sulcal
2014;103C:214-24. pattern asymmetry and local cortical surface gray
cha p t e r 5
33 Kinney HC, Brody BA, Kloman AS, Gilles FH. matter distribution in vivo: maturation in
Sequence of central nervous system myelination perisylvian cortices. Cereb Cortex 2002;12:17-26.
in human infancy. II. Patterns of myelination in 46 Gannon PJ, Holloway RL, Broadfield DC, Braun AR.
autopsied infants. J Neuropathol Exp Neurol Asymmetry of chimpanzee planum temporale:
1988;47:217-34. humanlike pattern of Wernicke’s brain language
34 Fogliarini C, Chaumoitre K, Chapon F, Fernandez area homolog. Science 1998;279:220-2.
C, Levrier O, Figarella-Branger D et al. Assessment 47 Hopkins WD, Nir TM. Planum temporale surface
of cortical maturation with prenatal MRI. Part I: area and grey matter asymmetries in
Normal cortical maturation. Eur Radiol chimpanzees (Pan troglodytes): the effect of
2005;15:1671-85. handedness and comparison with findings in
35 Hill J, Inder T, Neil J, Dierker D, Harwell J, Van ED. humans. Behav Brain Res 2010;208:436-43.
Similar patterns of cortical expansion during 48 Dehaene-Lambertz G, Montavont A, Jobert A,
human development and evolution. Proc Natl Allirol L, Dubois J, Hertz-Pannier L et al. Language
Acad Sci U S A 2010;107:13135-40. or music, mother or Mozart? Structural and
36 Kasprian G, Langs G, Brugger PC, Bittner M, Weber environmental influences on infants’ language
M, Arantes M et al. The prenatal origin of networks. Brain Lang 2010;114:53-65.
hemispheric asymmetry: an in utero 49 Im K, Pienaar R, Lee JM, Seong JK, Choi YY, Lee KH
neuroimaging study. Cereb Cortex 2011;21:1076- et al. Quantitative comparison and analysis of
83. sulcal patterns using sulcal graph matching: a
37 Hill J, Dierker D, Neil J, Inder T, Knutsen A, Harwell twin study. Neuroimage 2011;57:1077-86.
J et al. A surface-based analysis of hemispheric 50 Wright R, Kyriakopoulou V, Ledig C, Rutherford
asymmetries and folding of cerebral cortex in MA, Hajnal JV, Rueckert D et al. Automatic
term-born human infants. J Neurosci quantification of normal cortical folding patterns
2010;30:2268-76. from fetal brain MRI. Neuroimage 2014;91:21-32.
38 Leroy F, Glasel H, Dubois J, Hertz-Pannier L, 51 Kaukola T, Kapellou O, Laroche S, Counsell SJ,
Thirion B, Mangin JF et al. Early maturation of the Dyet LE, Allsop JM et al. Severity of perinatal
linguistic dorsal pathway in human infants. J illness and cerebral cortical growth in preterm
Neurosci 2011;31:1500-6. infants. Acta Paediatr 2009;98:990-5.
39 Li G, Nie J, Wang L, Shi F, Lyall AE, Lin W et al. 52 Moeskops P, Benders MJNL, Kersbergen KJ,
Mapping longitudinal hemispheric structural Groenendaal F, de Vries LS, Viergever MA et al.
asymmetries of the human cerebral cortex from Development of cortical morphology evaluated
birth to 2 years of age. Cereb Cortex with longitudinal MR brain images of preterm
2014;24:1289-300. infants. submitted 2014.
3
pa r t
White matter
structure,
diffusion weighted
imaging and
network analysis
to study brain
development
109
ch a p t er
6
Microstructural
brain development
between 30 and
40 weeks corrected
age in a
longitudinal
cohort of extremely
preterm infants
Alexander Leemans
Floris Groenendaal
Niek E. van der Aa
Max A. Viergever
Linda S. de Vries
Neuroimage 2014 103:214-224 111

Abstract
Diffusion tensor imaging (DTI) is frequently used to assess brain development in preterm
infants. This study investigates maturational changes in diffusivity measures in 122 regions
of the brain between 30 and 40 weeks postmenstrual age (PMA) using the neonatal atlas of
Oishi and colleagues.1 Forty infants without cerebral injury and with normal
neurodevelopmental outcome were selected from a cohort of preterm infants (gestational
age <28 weeks), scanned longitudinally at 30 and 40 weeks PMA. Fractional anisotropy
(FA) changed significantly in 84 brain regions, with the largest increase in the central brain
regions; by contrast, the cortical brain regions showed a decrease in FA. Mean, radial and
axial diffusivity all showed a clear decrease in the majority of brain regions. This study
provides longitudinal reference diffusivity values in a cohort of extremely preterm infants,
showing a central to peripheral and posterior to anterior directed gradient, in line with our
current understanding of brain maturation, and adding to this knowledge. This study
further elucidates brain maturation in preterm infants during the last 10 weeks prior to
term equivalent age. The presented values can be used as a reference for assessing brain
development in other cohorts, when investigating the effects of brain injury in this
vulnerable period, and to evaluate the effect of future neuroprotective strategies.
chapter 6
112 part 3 white matter structure , diffusion weighted imaging and network analysis to study brain development
Introduction
D
uring the second half of gestation, the human brain undergoes major growth and
development. In recent years, MRI, and more recently diffusion tensor imaging (DTI),
has become an important tool to evaluate microstructural changes in vivo in preterm
infants.2
DTI has shown significant changes in white matter tracts in the second trimester in post-
mortem foetuses and large changes in fractional anisotropy (FA) and mean diffusivity (MD)
values between 29 weeks and term age,3-6 which suggests an ongoing maturation of the brain.
This maturational process continues after term equivalent age, with an ongoing organization
of axons and myelination of tracts until childhood and adulthood.7,8
In particular the FA measure has often been used to quantify brain maturation. It has been
extensively shown that FA values in the posterior limb of the internal capsule (PLIC) are
positively related to increasing gestational age.6,9,10 This FA increase with a simultaneous MD
decrease is thought to be due to a combination of factors related to brain development,
including a decrease in water content of the brain, greater cohesiveness and compactness of
the fibre tracts, reduced extra-axonal space and an increase in myelination, restricting water
diffusion perpendicular to the axons.11
Several methods have been applied to quantify diffusivity measures, of which the manual
drawing of regions of interest (ROI), tractography and tract based spatial statistics (TBSS) are
most commonly used (for a review, see Deprez et al12). To avoid the subjective nature of
manual ROI-based measurements and to provide the opportunity to study the entire brain
within a manageable time frame, an automated, atlas based approach can be used. Because of
cha p t e r 6
the rapid development of the neonatal brain, adult brain atlases are not suitable for
implementation in this population. Recently, Oishi and colleagues published a neonatal brain
atlas for DTI, where they identified 122 brain regions according to the Talairach atlas.1,13
The aim of our study was to gain more information about maturational changes of the preterm
brain, by computing the main diffusivity measures (i.e., FA, MD, and also the radial (RD) and
axial (AD) diffusivities) at 30 and 40 weeks and the change in these measures during this
period in a unique group of longitudinally scanned, preterm born infants without brain injury
and with a normal neurodevelopmental outcome at 15 months.
Material and Methods

Cl inica l data
From a cohort of extremely preterm infants with a gestational age below 28 weeks,
consecutively admitted to the neonatal intensive care unit of the Wilhelmina Children’s
Hospital between June 2008 and June 2011, all 59 infants with serial MR imaging and available
outcome data at 15 months were selected for initial inclusion. According to the clinical
protocol, these infants are scanned around 30 weeks gestational age, if clinically stable, and
again at term equivalent age with DTI as part of the protocol. Standardized follow-up is
microstructural brain development between 30 and 40 weeks corrected age in a longitudinal 113
performed in our out-patient department at a corrected age of 15 months, using the Griffiths
Mental Development Scales.14 Infants were considered to have a normal developmental
follow-up if their Griffiths developmental quotient (DQ) was above 88, which is -1 standard
deviation (SD) with respect to the mean of 100. Infants at risk for abnormal
neurodevelopment (scoring below -1 SD) were hereby excluded. Infants with major
congenital anomalies, chromosomal abnormalities, structural brain abnormalities or severe
haemorrhagic-ischemic MR lesions (as determined with the T1- and T2-weighted images:
details below) and infants with corrupt DTI data were excluded, leaving 40 infants eligible for
final inclusion. Permission from our medical ethical review board and parental consent for the
MRI were obtained.
Mag n e tic r es o na nc e i m agi ng

MR imaging was performed on a 3.0 T MR system (Achieva, Philips Medical Systems, Best, The
Netherlands). The protocol included conventional T2-weighted and 3D T1-weighted imaging
in the coronal plane (30 weeks: 3D T1-weighted TR 9.4 ms; TE 4.6 ms; in plane FOV 130 x 101
mm; acquisition matrix 140 x 108 mm; slice thickness 2 mm; total scan time 4.44 min and T2-
weighted TR 10,085 ms; TE 120ms; in plane FOV 130 x 104 mm; acquisition matrix 244 x 162
mm; slice thickness 2 mm; total scan time 6.23 min, and 40 weeks: 3D T1-weighted TR 9.5 ms;
TE 4.6 ms; in plane FOV 200 x 200 mm, acquisition matrix 256 x 220 mm; slice thickness 1.2
mm; total scan time 7.02 min and T2-weighted TR 4847 ms; TE 150 ms; in plane FOV 180 x 180
mm; acquisition matrix 232 x 202 mm; slice thickness 1.2 mm; total scan time 5.05 min). The
DTI protocol at both time points consisted of a single-shot spin-echo echoplanar imaging
cha p t e r 6
sequence (echoplanar imaging factor 55, TR/TE 5685/70 ms, field of view 180 x 146 mm,
acquisition matrix 128 x 102 mm, reconstruction matrix 128 x 128 mm, 50 slices with 2 mm
thickness without gap). Images were acquired in the axial plane with diffusion gradients
applied in 32 non-collinear directions with a b-value of 800 s/mm2 and one non-diffusion
weighted image, with a total scan time of 4.28 min.15 Infants were sedated using oral
chloralhydrate 50 to 60 mg/kg, according to clinical protocol. Heart rate, transcutaneous
oxygen saturation and respiratory rate were monitored. For hearing protection Minimuffs
(Natus Medical Incorporated, San Carlos, CA, USA) and Earmuffs (EM’s 4 Kids, Brisbane,
Australia) were used. A neonatologist was present throughout the examination.
After evaluation by a radiologist, all scans were re-assessed by two neonatologists (LSdeV and
MJNLB) with more than 10 years experience in neonatal neuro-imaging, using the MRI scoring
system as published by Kidokoro.16 Only infants scored as no or mild injury for both white
matter and global score remained included in the present study. Excluded were all infants with
an intraventricular haemorrhage (IVH) grades III-IV according to Papile,17 post-haemorrhagic
ventricular dilatation requiring intervention, parenchymal damage, and severe lesions to the
deep gray matter or the cerebellum. This yielded a unique group of 40 extremely preterm
infants with a normal longitudinal scan and normal neurodevelopmental outcome.
Imag e p r oc e ssi ng
DTI data were analyzed using the diffusion MR toolbox ‘ExploreDTI’,18 and consisted of the
following steps: (i) correction for subject motion and eddy current induced distortions;19 (ii)
tensor estimation using the REKINDLE approach for outlier detection20 with iteratively
reweighted linear least squares estimation after identification and removal of data outliers;21
and (iii) automated atlas based analysis with the neonatal atlas developed in the paper of
Oishi et al1 (template and parcellated atlas regions are publically available at http://cmrm.med.
jhmi.edu/) using affine and elastic registration based on ‘elastix’.22 All DTI data were visually
checked in terms of quality of tensor estimation and quality of registration. After these
preprocessing steps, FA, AD, RD, and MD values were calculated in the 122 brain regions that
are provided by the Oishi atlas.1
Stat i st ica l a n a ly s i s
Statistical procedures were performed using both IBM SPSS Statistics version 20 and R version
2.15.0.23 To avoid a bias from the exact scan time on the FA estimates of the brain regions, all
values were calculated back to a postmenstrual age of 30 and 40 weeks. A linear increase in FA
was hereby assumed, as has been described previously.6,24 More specifically, for each brain
region the difference in FA between both time points was calculated by subtracting the
corrected FA value at 30 weeks from the corrected FA value at 40 weeks. Mixed effect
modelling was used to determine whether the change in FA between the two measurements
was significant, with the original FA value as independent variable, PMA as fixed effect, and
patient number as random factor. A paired sample t-test was performed to identify left-right
cha p t e r 6
differences. The same procedure was then used for MD, RD and AD values.
Next, all brain regions were tested with mixed effect modelling for significant effects on FA
values of gestational age at birth and of gender. For this analysis the original FA-value was used
as independent factor, postmenstrual age at time of scan as an additional fixed factor, and
patient number as random factor. Again, the same procedure was then used for MD, RD and
AD values.
Finally, to be able to combine brain regions with a similar change in FA, principal component
analysis was performed using R, followed by plotting clusters (library ‘cluster’) using K-means
clustering with three clusters against the first two principal components. This identified
clusters of brain regions with comparable FA changes. Subsequently, MD, AD and RD changes
in these clusters were analyzed.
To correct for multiple comparisons, a Bonferroni correction was used with a testwise alpha of
0.05 and a familywise alpha of 244 (2 measurements, FA and MD, over 122 brain regions).
Since AD and RD are derived from MD, we chose not to correct for these values separately. This
correction led to a p-value below 0.0002 to be considered as significant for all analyses.
Results
Clinical characteristics of the infants are shown in table 1. For each brain region, the average FA
value of the 40 infants is shown at 30 and 40 weeks in figures 1 A and B, respectively. Figures 2,
3 and 4 show the equivalent maps for the MD, AD and RD, respectively.
Ta ble 1 . Dem o g raphi cs of the 4 0 i nfa nts
Characteristics N=40
Gestational age (weeks, mean [range]) 26.4 [24.4 – 27.9]
Gender (male/female) 21/19
Birth weight (grams, mean [range]) 880 [650 – 1135]
Birth weight z-score (mean [range]) 0.3 [-1.2 – 1.9]
Bronchopulmonary dysplasia* 8 (20%)
Patent ductus arteriosus requiring treatment 18 (45%)
Necrotizing enterocolitis requiring surgery 1 (3%)

cha p t e r 6
Culture proven sepsis 12 (30%)
Intraventricular haemorrhage#
No 31 (78%)
Grade 1 8 (20%)
Grade 2 1 (3%)
Post haemorrhagic ventricular dilatation

No 37 (93%)
Yes, mild (<2SD Levene index), no treatment needed 3 (8%)
MRI scoring classification^ (normal/mild) 27/13
Postmenstrual age at 1st scan (weeks, mean [range]) 30.8 [29.3 – 32.7]
Postmenstrual age at 2nd scan (weeks, mean [range]) 41.0 [40.0 – 42.7]
Corrected age at follow-up (months, mean [range]) 16.3 [14.1-19.9]
Griffiths DQ score (mean [range]) 104 [88-131]
*Bronchopulmonary dysplasia is defined as oxygen dependency at 36 weeks postmenstrual age

#
IVH is defined according to Papile17
^Scoring classification as described by Kidokoro16
Figure 1 Average across the 40 subjects of the FA values in all 122 atlas brain regions at 30 (A) and 40 (B)
weeks. The colour legend at the right side of the figure represents the FA values.
cha p t e r 6
Figure 2 Average across the 40 subjects of the MD values in all 122 atlas brain regions at 30 (A) and 40 (B)
weeks. The colour legend at the right side of the figure represents the MD values (mm2/s).
Figure 3 Average across the 40 subjects of the AD values in all 122 atlas brain regions at 30 (A) and
40 (B) weeks. The colour legend at the right side of the figure represents the AD values (mm2/s).
cha p t e r 6
Figure 4 Average across the 40 subjects of the RD values in all 122 atlas brain regions at 30 (A) and
40 (B) weeks. The colour legend at the right side of the figure represents the RD values (mm2/s).
Figure 5 depicts the change in the average FA values across the 40 infants between the two
scans. In 70 brain regions, a significant FA increase and in 14 brain regions, a significant FA
decrease was seen between the first and the second scan. Rates of increase differed between 8
and 76%, rates of decrease between -8 and -18%. The largest increase in FA was seen in the
posterior limb of the internal capsule, the cerebral peduncles, the sagittal stratum and the
corona radiata. Other central structures of the brain also showed a clear increase. A decrease
in mean FA was seen in the cortical brain regions, particularly in the temporal and occipital
parts. Figures 6, 7 and 8 show the equivalent changes for MD, AD and RD, respectively. Here, a
decrease in values was observed for all significant brain regions, except for AD values in the
cerebral peduncles. Rates of decrease were -5 to -27%, -4 to -25% and -5 to -32% for MD, AD
and RD respectively. Changes were significant in 112 brain regions for MD, 104 for AD and 114
cha p t e r 6
Figure 5 Average change in FA across the 40 subjects between 30 and 40 weeks postmenstrual age, represented for the 14
brain regions with significant FA decrease (A) and the 70 brain regions with significant FA increase (B). For 12 representative
brain regions (all left-sided), boxplots are shown, with whiskers representing the complete range of measurements. The
percentage of change is given above each boxplot. When the change was not significant, this is indicated with NS. FA values are
represented on the y-axis in all plots.
for RD. The pattern of change for MD, AD and RD was similar to that of FA, with a larger
decrease in the occipital and temporal regions as compared with the frontal regions. AD
measurements showed the largest decreases. All FA, MD, AD and RD values and standard
deviations with corresponding p-values for both the change over time and the left-right
differences are shown online only. (Inline Supplementary Table S1 can be found online at http://
dx.doi.org/10.1016/j.neuroimage.2014.09.039).
chapter 6
Figure 6 Average change in MD across the 40 subjects between 30 and 40 weeks postmenstrual age, represented for the 112
significant brain regions (A). For the same 12 brain regions as in figures 5, 7 and 8, boxplots are shown, with whiskers
representing the complete range of measurements. The percentage of change is given above each boxplot. When the change was
not significant, this is indicated with NS. MD values (in mm2/s) are represented on the y-axis in all plots.
To be able to combine brain regions with a similar change in diffusivity, a principal component
analysis was performed. After principal component analysis, four clusters of brain regions
could be formed based on the first component. The second component did not add further
distinction. The four clusters are shown in figure 9 and the brain regions belonging to each
cluster are shown online only (Inline Supplementary Table S1).
chapter 6
Figure 7 Average change in AD across the 40 subjects between 30 and 40 weeks postmenstrual age, represented for the 104
not significant, this is indicated with NS. AD values (in mm2/s) are represented on the y-axis in all plots.
cha p t e r 6
Figure 8 Average change in RD across the 40 subjects between 30 and 40 weeks postmenstrual age, represented for the 114
not significant, this is indicated with NS. RD values (in mm2/s) are represented on the y-axis in all plots.
The two clusters with the largest increase in FA include the central structures. The degree of
increase for all central structures was similar, although FA values at both 30 and 40 weeks
were significantly higher for the major motor pathways as compared with the other central
structures. The cortical structures were divided in two clusters, a frontal cluster showing the
lowest FA at 30 weeks, with no significant change to a slight increase at 40 weeks, and a
temporal-occipital cluster showing a slightly higher FA at 30 weeks with a subsequent slight
decrease towards 40 weeks. In all four clusters, a decrease was seen for MD, AD and RD, as
represented in figure 9. The degree of change was similar between the clusters, although
absolute values differed.
chapter 6
Figure 9 Results of principal component analysis. Clusters found with principal component analysis and the corresponding
changes in FA, MD, AD and RD between the two scans, resulting from multiple regression analysis. Clusters are represented as
follows: cluster 1 = dark blue, cluster 2 = light blue, cluster 3 = yellow, cluster 4 = red. Brain regions belonging to each cluster
are represented in the brain image and can found in more detail in online only table 1.
Discussion
In this study, the longitudinal change of FA, MD, AD and RD in the preterm brain over the
period of 30 to 40 weeks postmenstrual age has been analyzed for infants with a normal
neurodevelopmental outcome at 15 months corrected age. FA changed significantly in 84 out
of 122 atlas-based brain regions, with the largest increase in the central brain regions, up to
76% in a period of 10 weeks. In contrast, cortical brain regions showed a small decrease in FA.
For the other diffusivity measures a decrease was noted in both central and cortical brain
regions. This is the first study demonstrating, with the use of longitudinal data, that for infants
with a normal neurodevelopmental outcome at 15 months corrected age major maturational
changes in FA, MD, RD and AD occur throughout the brain before term equivalent age.
Several authors have previously provided measurements of FA at different gestational ages,
showing an increase in FA with an increasing gestational age measured in selected parts of the
brain, most often the motor tracts or corpus callosum.4,9,25,26 Apparent diffusion coefficient
(ADC) or MD measurements show a simultaneous decrease, although this can be disrupted in
the presence of brain injury.9,27 Studies of healthy foetuses scanned in utero showed similar FA
increases and ADC decreases in the white matter and major motor tracts.28,29 The findings of
this study are in line with these previous findings inasmuch as the diffusivity measures in this
selected group of preterm infants without brain injury and with normal neurodevelopmental
outcome reflect normal maturation.
In contrast with most neonatal DTI studies, our study shows longitudinal DTI data and uses a
global automated approach, covering the whole brain rather than manually selecting specific
white matter structures. By doing so, we gained a more detailed insight into differential
regional brain development. In particular, we showed that FA values increased and the other
diffusivity measures decreased in central-peripheral and occipital-frontal directions,
consistent with the well-known maturation pattern of the human brain as shown by
histology30-32 and corroborated by measurements of specific fibre bundles, using several MR
parameters.26,33,34 Furthermore, histological analyses have shown that axonal myelination
proceeds in a proximal to distal manner as seen from the neuron of origin.35 Different regions
of the brain will start this maturational process at different time points, and the rate at which it
continues can also differ.32 For instance, the PLIC and corpus callosum are known to start early
and to mature fast, which is represented by a higher FA and concurrent lower MD at 30 weeks
cha p t e r 6
and a sharp change towards term equivalent age. The anterior limb of the internal capsule
(ALIC) also shows a clear increase in FA, but matures somewhat more slowly compared to the
PLIC, showing the occipital-frontal fashion of maturation.36 In contrast, in the frontal white
matter, microscopic myelin can only be found after term equivalent age and myelination in the
frontal pole is not complete until after two years of age, showing the central to peripheral
direction of maturation.32 Noteworthy in this context is that the occipital-frontal gradient of
maturation has also been shown in perfusion and metabolic studies of brain development.37,38
The increasing FA in white matter structures in this study also included brain regions that are
not yet myelinated at term equivalent age. It has been described before that an increase in
anisotropy can precede myelination of these structures. This is likely due to the ‘pre-
myelination’ encasement of axons by immature oligodendrocytes, which is taking place
throughout the third trimester of gestation, beginning with the shorter axons in central
structures of the brain.30,39 In several animal studies it has also been shown that such FA
changes can precede histological changes associated with myelination by several weeks.40-42
Apart from myelination, the increase in FA and concurrent decrease in MD can be attributed to
a combination of factors representing the advancing microstructural development of the
brain, such as an overall decrease in brain water content due to proliferation and functional
maturation of glial cell bodies and prolongations, as well as the proliferation of intercellular
compartments, and an increase in membrane density.11,43 Also, a sharp increase in the number
of axons and a variation in axonal diameter of the corpus callosum was found during the foetal
stages in a rhesus monkey model, with myelination not starting until after birth.44 On the one
hand, since DTI measures are highly sensitive (e.g., see reviews Jones et al.45 and Deprez et
al.12), it is likely that the increased FA and decreased diffusivities found in this study are indeed
a representation of brain maturation. On the other hand, it is well-known nowadays that DTI
measures are nonspecific, that is, any observed changes in such measures are probably due to
a combination of several processes that cannot be disentangled or attributed to a single
microstructural constituent (e.g., axonal filaments, myelin, etc.). New advances in diffusion
MRI modelling and processing may provide us with more precise characterizations of axonal
microstructure in the near future2,12,46,47 and will allow a more detailed understanding of these
processes.
In this study, an increase in FA with a corresponding decrease in MD, AD and RD was mainly
found in the central structures. In the cortical brain regions, no difference or a slight decrease
in FA was noted. MD, AD and RD measurements did show a significant decrease in most
cortical brain regions as well. This difference in FA change between brain regions may be
explained in part by the difference in diffusivity properties between cortex and white matter.
Several studies have shown an initial increase in FA in the cortex until a gestational age of 26-
28 weeks, followed by a decreasing FA towards term age, with a less uniform pattern in the
frontal cortex as opposed to the other parts.48-51 This is consistent with the pattern of
synaptogenesis found in human autopsy studies, where synaptogenesis, expressed as synapse
cha p t e r 6
density, increased more rapidly in the auditory cortex compared to the prefrontal cortex.52
This regional heterogeneity in cortical development, again in the occipital-frontal direction,
may be a representation of the different functions of those brain regions. The more
pronounced decrease in FA in the occipital cortex as opposed to the FA decrease in the frontal
cortex could reflect the earlier use of visual functions in a preterm infant towards term
equivalent age.49 Alternatively, the frontal cortex may have a different maturational process
and may therefore show a less uniform pattern of FA change.50,51 The decrease in anisotropy in
the developing cortex is thought to be due to an increasing organization of the microstructure
along multiple and distinctly different orientations. Innervation of the cortex by
thalamocortical axons, cortico-cortical connections, the arborisation of basal dendrites from
immature cortical neurons, early synaptogenesis and proliferation of glial cells with the
disappearance of the radial glia, all take place between 25 and 40 weeks of gestation.53 These
processes will hinder water motion in all directions, and will thereby gradually diminish the
anisotropic nature of diffusion.3,48 In our study, cortical brain regions also included part of the
white matter, as predefined by the brain labels of the Oishi atlas, which may explain these
results. Note, however, that the FA decrease in the temporal-occipital parts of the cortex can
still be observed, as well as the less uniform pattern in the frontal cortex.
Left to right differences in brain maturation were seen in a minority of brain regions from this
study, and absolute differences were small. Generally, a larger change in FA, MD, AD or RD was
seen on the left compared to the right side, while the corresponding differences in absolute
values were smaller at 40 weeks than at 30 weeks. These data suggest a slightly earlier
maturation of the right side, which is compensated by a larger change in diffusivity measures
on the left side. This is in agreement with previous studies on brain maturation, that showed
an earlier maturation of the right side of the brain, which starts early in the second trimester
and continues during the third trimester.54-56 Although data are limited, the left-right
asymmetry seems to become less noticeable towards term equivalent age, especially in the
large white matter bundles.26,33 This may be different for cortical areas, as shown in a study
describing a right-ward asymmetry around term equivalent age in the superior temporal and
lateral occipito-temporal gyrus.3
Although it cannot be excluded that the preterm birth of the infants used in this study may
have affected cortical development, it has previously been described that preterm birth does
not change the rate of synaptogenesis, but instead leads to distinct changes in size, type and
laminar distribution of synapses in the visual cortex of rhesus monkeys.57 The selected
population included in this study may therefore still reflect normal neurogenesis, comparable
to the third trimester of pregnancy.
The MD, AD and RD measures showed a significant decrease in most brain regions, with a
similar regional heterogeneity in degree of decrease. Relative decreases in RD were larger than
those in AD. Changes in RD are thought to be the best reflection of changes in myelination,
since the diffusion perpendicular to the axons is considered to be more sensitive to myelin
cha p t e r 6
changes than diffusion parallel to the axons.7,34 The changes in RD found in this study may
therefore represent the process of myelination taking place between 30 weeks and term
equivalent age.
The atlas used has been described by Oishi and colleagues.1 This atlas has been constructed
using scans of 25 healthy, term born neonates and was subsequently applied to a group of 22
healthy, full-term newborns with a post-conceptional age between 37 and 53 weeks. In their
cohort, a trend towards a decreasing MD and increasing FA with age was found, similar to our
results. Also, the slope of FA increase and MD decrease was comparable with our results,
although our FA values at 40 weeks seem to be slightly lower compared with their estimated
40 week FA values. These lower FA values at term equivalent age in premature infants as
compared with term infants are in agreement with several other studies.6,58 By using a DTI
based atlas,1 the subjective and time-consuming nature of manual ROI measurements for each
subject has been avoided. This lack of subjectivity is likely to increase comparability and
accuracy of the measurements. Additionally, diffusivity measurements cover the whole brain
instead of selected structures.
It has been shown that gender differences were likely caused by a larger percentage of boys
showing brain lesions at term equivalent age.58-60 The same is true for gestational age, where
the presence of brain injury and co-morbid conditions seems more important than gestational
age at birth itself.3,9 By selecting only infants with a normal neurodevelopmental outcome,
these gender differences were not present in this study.
The measurements shown in this study are limited to FA, MD, RD and AD values. Whether
these measures will also have a direct correlation with the topological structure of the
macroscopic structural brain network and involve an increasing capacity of the network to
integrate information between subparts remains to be elucidated. A combination of DTI and
resting state functional MRI could provide additional information to study connectivity as well
as structure in the developing brain, providing more insight into both normal and pathological
processes taking place.
There are several limitations to this study. First, outcome data are limited to 15 months
corrected age. We do not yet know whether these children will continue to have a normal
development later in life. Second, the diffusivity values computed in the brain regions are
strongly dependent on the accuracy of the alignment with the Oishi atlas.1 Since preterm
brains are generally small, they are particularly sensitive to partial volume effects, caused by
the relatively large voxel size compared with the size of the structures measured.61 Also, the
atlas of Oishi and colleagues was constructed with data from a different population and with a
different type of MR scanner, which may bias our results.62,63 Notwithstanding these well-
known atlas-based limitations, we checked the quality of the registrations and corresponding
diffusivity values for all scans. A very precise segmentation of cortex and white matter with a
proper registration to the atlas would be needed to divide between gray and white matter in
the atlas-defined brain regions, but to achieve this in a reliable way remains challenging to
cha p t e r 6
date. Then, the atlas was created for brains at term (equivalent) age and may therefore be
suboptimal for use at earlier gestational ages. Separate atlases per gestational age week,
especially for the younger ages, ideally created using more infants than the 25 of the Oishi
atlas, would be preferable above the approach used in this study. Unfortunately, such atlases
require good quality DTI data of many infants and are therefore difficult to acquire. To the best
of our knowledge, no such atlases are currently freely available and we did not have DTI data of
enough infants to create such atlases ourselves. However, careful registration and visual
quality checks assured the accuracy of the registration and the corresponding diffusivity values
of both the early and term equivalent age scans. Finally, it is well-known that DTI cannot
unambiguously characterize microstructural tissue properties in voxels with multiple fibre
populations, which may complicate the interpretation of the observed findings.46,64
Nevertheless, despite this lack of specificity, DTI can still be considered to be a quantitative
approach that is very sensitive to detect changes in brain tissue.
In conclusion, this study provides longitudinal reference values of FA, MD, AD and RD
estimates, covering the whole brain for a relatively large cohort of preterm infants with a
normal neurodevelopmental outcome. The values provided are indicative for the progression
of regional human brain maturation and can be used in future studies to identify factors
influencing diffusivity at preterm and term equivalent age in preterm infants, such as clinical
risk factors or brain injury, and to study the effect of neuroprotective interventions.
Funding
infants from a study funded by the Wilhelmina Research Fund (OZF2008-19). The research of
A.L. is supported by VIDI Grant 639.072.411 from the Netherlands Organisation for Scientific
Research (NWO).
Suppl eme n tar y ta b le S 1
FA 30 weeks FA 40 weeks MD 30 weeks (mm²/s) MD 40 weeks (mm²/s)

Brain region
Mean SD Mean SD Mean SD Mean SD
Left corpus callosum 0.25 0.03 0.29 0.03 1.67E-03 1.60E-04 1.47E-03 1.16E-04
Right corpus callosum 0.25 0.03 0.28 0.03 1.62E-03 1.70E-04 1.45E-03 9.82E-05
Left anterior limb of internal capsule 0.18 0.03 0.24 0.02 1.44E-03 1.63E-04 1.15E-03 6.11E-05
Right anterior limb of internal capsule 0.20 0.03 0.25 0.02 1.40E-03 1.03E-04 1.15E-03 6.77E-05
Left posterior limb of internal capsule 0.27 0.04 0.40 0.02 1.31E-03 9.83E-05 1.04E-03 4.84E-05
Right posterior limb of internal capsule 0.28 0.04 0.39 0.02 1.31E-03 1.39E-04 1.04E-03 4.66E-05
Left retrolenticular part of internal capsule 0.25 0.03 0.30 0.03 1.43E-03 1.54E-04 1.15E-03 5.43E-05
chapter 6
Right retrolenticular part of internal capsule 0.26 0.03 0.30 0.02 1.43E-03 1.37E-04 1.17E-03 5.12E-05
Left anterior corona radiata 0.13 0.02 0.18 0.02 1.62E-03 9.32E-05 1.36E-03 1.04E-04
Right anterior corona radiata 0.14 0.02 0.18 0.02 1.60E-03 8.90E-05 1.38E-03 1.08E-04
Left superior corona radiata 0.13 0.02 0.22 0.03 1.67E-03 1.08E-04 1.33E-03 1.18E-04
Right superior corona radiata 0.13 0.02 0.23 0.03 1.65E-03 1.07E-04 1.31E-03 1.10E-04
Left posterior corona radiata 0.15 0.03 0.26 0.03 1.69E-03 1.33E-04 1.31E-03 1.17E-04
Right posterior corona radiata 0.17 0.03 0.27 0.03 1.69E-03 1.28E-04 1.31E-03 1.12E-04
Left cingulum cingular part 0.24 0.03 0.19 0.02 1.40E-03 1.18E-04 1.21E-03 6.21E-05
Right cingulum cingular part 0.22 0.04 0.18 0.02 1.36E-03 1.07E-04 1.20E-03 5.67E-05
Left cingulum hippocampal part 0.18 0.02 0.19 0.03 1.39E-03 1.17E-04 1.22E-03 6.05E-05
Right cingulum hippocampal part 0.19 0.03 0.18 0.02 1.38E-03 1.19E-04 1.23E-03 6.95E-05
Left fornix 0.20 0.03 0.23 0.02 1.89E-03 2.50E-04 1.76E-03 2.79E-04
Right fornix 0.20 0.02 0.25 0.02 1.94E-03 2.04E-04 1.69E-03 1.74E-04
Left stria terminalis 0.19 0.04 0.25 0.03 2.01E-03 3.72E-04 1.71E-03 3.03E-04
Right stria terminalis 0.21 0.04 0.26 0.03 1.72E-03 2.89E-04 1.57E-03 2.22E-04
Left tapetum 0.14 0.03 0.22 0.04 2.22E-03 3.40E-04 1.93E-03 2.57E-04
Right tapetum 0.16 0.03 0.23 0.04 2.00E-03 2.84E-04 1.80E-03 2.11E-04
Left superior longitudinal fasciculus 0.12 0.02 0.17 0.02 1.78E-03 1.09E-04 1.43E-03 1.11E-04
Right superior longitudinal fasciculus 0.13 0.02 0.16 0.01 1.72E-03 1.10E-04 1.40E-03 1.02E-04
Left external capsule 0.15 0.02 0.20 0.02 1.45E-03 1.17E-04 1.20E-03 6.13E-05
Right external capsule 0.16 0.02 0.20 0.02 1.44E-03 8.75E-05 1.20E-03 5.64E-05
Left posterior thalamic radiation 0.17 0.03 0.27 0.02 1.93E-03 2.67E-04 1.42E-03 1.08E-04
Right posterior thalamic radiation 0.20 0.03 0.28 0.02 1.76E-03 2.11E-04 1.38E-03 9.04E-05
Left sagittal stratum 0.17 0.02 0.27 0.03 1.98E-03 2.43E-04 1.45E-03 1.01E-04
Right sagittal stratum 0.19 0.03 0.28 0.02 1.81E-03 2.27E-04 1.41E-03 7.52E-05
Left thalamus 0.17 0.02 0.21 0.01 1.47E-03 1.86E-04 1.14E-03 7.92E-05
Right thalamus 0.17 0.01 0.20 0.02 1.45E-03 1.58E-04 1.18E-03 7.51E-05
Left putamen 0.13 0.02 0.15 0.02 1.40E-03 1.17E-04 1.10E-03 5.52E-05
Right putamen 0.13 0.04 0.14 0.02 1.37E-03 1.11E-04 1.10E-03 5.31E-05
Left globus pallidus 0.12 0.03 0.17 0.02 1.32E-03 8.80E-05 1.10E-03 5.25E-05
Right globus pallidus 0.12 0.03 0.16 0.02 1.32E-03 1.21E-04 1.11E-03 5.44E-05
Left caudate nucleus 0.12 0.02 0.13 0.02 1.83E-03 2.21E-04 1.68E-03 2.76E-04
Right caudate nucleus 0.12 0.02 0.13 0.02 1.74E-03 2.29E-04 1.62E-03 1.83E-04
Left cerebral peduncle 0.25 0.03 0.35 0.02 1.25E-03 1.44E-04 1.27E-03 9.15E-05
Right cerebral peduncle 0.25 0.03 0.35 0.02 1.22E-03 9.80E-05 1.25E-03 1.02E-04
Left superior fronto-occipital fasciculus 0.13 0.03 0.21 0.04 1.62E-03 1.48E-04 1.39E-03 2.10E-04
Right superior fronto-occipital fasciculus 0.15 0.04 0.21 0.03 1.57E-03 1.30E-04 1.32E-03 1.30E-04
Left inferior fronto-occipital fasciculus 0.20 0.03 0.26 0.03 1.43E-03 8.40E-05 1.19E-03 6.17E-05
Right inferior fronto-occipital fasciculus 0.19 0.02 0.25 0.02 1.43E-03 8.50E-05 1.21E-03 5.87E-05
Left corticospinal tract 0.19 0.03 0.27 0.03 1.24E-03 1.23E-04 1.11E-03 1.17E-04
Right corticospinal tract 0.18 0.03 0.24 0.03 1.26E-03 2.06E-04 1.14E-03 1.19E-04
Left superior cerebellar peduncle 0.17 0.03 0.29 0.02 1.33E-03 1.54E-04 1.18E-03 7.81E-05
Right superior cerebellar peduncle 0.17 0.03 0.28 0.02 1.33E-03 1.47E-04 1.22E-03 7.30E-05
Left middle cerebellar peduncle 0.19 0.02 0.31 0.03 1.39E-03 1.26E-04 1.17E-03 9.64E-05
Right middle cerebellar peduncle 0.21 0.02 0.30 0.02 1.38E-03 1.45E-04 1.25E-03 1.22E-04
Left inferior cerebellar peduncle 0.16 0.03 0.26 0.03 1.25E-03 1.50E-04 1.09E-03 7.06E-05
Right inferior cerebellar peduncle 0.18 0.03 0.27 0.03 1.23E-03 1.91E-04 1.00E-03 9.18E-05
chapter 6
Left pontine crossing tract 0.19 0.03 0.25 0.03 1.17E-03 1.08E-04 1.00E-03 8.73E-05
Right pontine crossing tract 0.18 0.02 0.22 0.03 1.20E-03 1.26E-04 1.01E-03 9.33E-05
Left uncinate fasciculus 0.19 0.03 0.23 0.03 1.38E-03 1.22E-04 1.19E-03 5.39E-05
Right uncinate fasciculus 0.20 0.03 0.23 0.03 1.32E-03 1.10E-04 1.18E-03 6.49E-05
Left midbrain 0.14 0.02 0.23 0.02 1.29E-03 1.52E-04 1.05E-03 8.59E-05
Right midbrain 0.14 0.02 0.23 0.02 1.32E-03 1.51E-04 1.08E-03 8.91E-05
Left pons 0.18 0.02 0.24 0.02 1.25E-03 1.33E-04 1.01E-03 6.55E-05
Right pons 0.17 0.02 0.23 0.02 1.32E-03 1.96E-04 1.02E-03 7.67E-05
Left medial lemniscus 0.20 0.03 0.28 0.03 1.20E-03 1.65E-04 9.41E-04 6.24E-05
Right medial lemniscus 0.20 0.03 0.26 0.02 1.20E-03 1.97E-04 9.63E-04 7.04E-05
Left medulla oblongata 0.16 0.02 0.22 0.03 1.22E-03 2.56E-04 1.11E-03 1.42E-04
Right medulla oblongata 0.16 0.03 0.22 0.03 1.25E-03 4.13E-04 1.14E-03 1.52E-04
Left superior frontal gyrus 0.14 0.02 0.16 0.01 1.46E-03 9.68E-05 1.38E-03 1.36E-04
Right superior frontal gyrus 0.15 0.02 0.15 0.01 1.47E-03 1.01E-04 1.44E-03 1.37E-04
Left middle frontal gyrus 0.14 0.02 0.15 0.01 1.46E-03 8.20E-05 1.36E-03 1.11E-04
Right middle frontal gyrus 0.14 0.02 0.15 0.01 1.46E-03 8.79E-05 1.32E-03 8.23E-05
Left inferior frontal gyrus 0.15 0.02 0.16 0.01 1.45E-03 9.12E-05 1.28E-03 7.93E-05
Right inferior frontal gyrus 0.15 0.02 0.16 0.01 1.45E-03 1.07E-04 1.29E-03 5.91E-05
Left medial fronto-orbital gyrus 0.14 0.02 0.16 0.01 1.42E-03 8.73E-05 1.34E-03 8.38E-05
Right medial fronto-orbital gyrus 0.15 0.02 0.16 0.02 1.41E-03 1.01E-04 1.34E-03 8.20E-05
Left lateral fronto-orbital gyrus 0.14 0.01 0.16 0.01 1.48E-03 9.33E-05 1.34E-03 7.99E-05
Right lateral fronto-orbital gyrus 0.14 0.02 0.16 0.01 1.46E-03 8.51E-05 1.34E-03 7.13E-05
Left gyrus rectus 0.16 0.02 0.16 0.01 1.35E-03 1.06E-04 1.38E-03 1.09E-04
Right gyrus rectus 0.15 0.02 0.16 0.01 1.39E-03 1.04E-04 1.37E-03 1.15E-04
Left precentral gyrus 0.14 0.02 0.17 0.01 1.49E-03 1.27E-04 1.24E-03 7.99E-05
Right precentral gyrus 0.15 0.02 0.16 0.01 1.46E-03 1.17E-04 1.22E-03 6.65E-05
Left postcentral gyrus 0.15 0.02 0.16 0.01 1.50E-03 1.27E-04 1.22E-03 7.56E-05
Right postcentral gyrus 0.15 0.02 0.17 0.01 1.48E-03 1.13E-04 1.20E-03 6.71E-05
Left superior parietal gyrus 0.13 0.01 0.15 0.01 1.72E-03 1.28E-04 1.34E-03 9.21E-05
Right superior parietal gyrus 0.13 0.02 0.15 0.02 1.69E-03 1.16E-04 1.31E-03 8.75E-05
Left precuneus 0.15 0.02 0.14 0.02 1.68E-03 1.80E-04 1.40E-03 1.24E-04
Right precuneus 0.16 0.02 0.14 0.02 1.67E-03 1.85E-04 1.40E-03 1.29E-04
Left cingular gyrus 0.16 0.02 0.15 0.01 1.45E-03 9.99E-05 1.26E-03 7.08E-05
Right cingular gyrus 0.16 0.02 0.15 0.01 1.43E-03 8.45E-05 1.32E-03 7.38E-05
Left supramarginal gyrus 0.15 0.02 0.15 0.01 1.54E-03 1.26E-04 1.28E-03 6.91E-05
Right supramarginal gyrus 0.15 0.02 0.14 0.01 1.46E-03 9.62E-05 1.24E-03 5.68E-05
Left angular gyrus 0.14 0.02 0.14 0.01 1.64E-03 1.08E-04 1.33E-03 6.94E-05
Right angular gyrus 0.14 0.02 0.14 0.01 1.56E-03 1.13E-04 1.29E-03 6.73E-05
Left superior temporal gyrus 0.16 0.02 0.15 0.01 1.46E-03 1.29E-04 1.33E-03 7.77E-05
Right superior temporal gyrus 0.16 0.02 0.15 0.01 1.40E-03 1.30E-04 1.30E-03 6.49E-05
Left middle temporal gyrus 0.17 0.02 0.15 0.01 1.48E-03 1.07E-04 1.29E-03 6.06E-05
Right middle temporal gyrus 0.17 0.02 0.15 0.01 1.44E-03 1.05E-04 1.28E-03 5.64E-05
Left inferior temporal gyrus 0.14 0.02 0.15 0.01 1.49E-03 9.56E-05 1.30E-03 5.80E-05
Right inferior temporal gyrus 0.15 0.02 0.15 0.01 1.46E-03 9.58E-05 1.30E-03 5.89E-05
Left fusiform gyrus 0.14 0.02 0.15 0.01 1.69E-03 1.45E-04 1.31E-03 6.25E-05
Right fusiform gyrus 0.15 0.02 0.15 0.01 1.59E-03 1.39E-04 1.29E-03 5.76E-05
chapter 6
Left parahippocampal gyrus 0.16 0.02 0.15 0.01 1.57E-03 2.06E-04 1.28E-03 9.30E-05
Right parahippocampal gyrus 0.17 0.02 0.15 0.01 1.43E-03 1.43E-04 1.24E-03 6.57E-05
Left entrhinal cortex 0.14 0.02 0.15 0.02 1.37E-03 1.44E-04 1.21E-03 9.63E-05
Right entrhinal cortex 0.14 0.02 0.15 0.01 1.37E-03 1.77E-04 1.20E-03 6.29E-05
Left superior occipital gyrus 0.15 0.02 0.14 0.02 1.72E-03 1.90E-04 1.31E-03 9.29E-05
Right superior occipital gyrus 0.15 0.02 0.15 0.02 1.66E-03 2.19E-04 1.32E-03 1.02E-04
Left middle occipital gyrus 0.16 0.02 0.15 0.01 1.53E-03 1.32E-04 1.26E-03 7.39E-05
Right middle occipital gyrus 0.16 0.03 0.14 0.01 1.48E-03 1.35E-04 1.24E-03 6.40E-05
Left inferior occipital gyrus 0.16 0.02 0.14 0.01 1.58E-03 1.32E-04 1.29E-03 8.42E-05
Right inferior occipital gyrus 0.17 0.03 0.14 0.01 1.50E-03 1.43E-04 1.26E-03 6.95E-05
Left cuneus 0.16 0.02 0.13 0.01 1.62E-03 1.62E-04 1.31E-03 7.97E-05
Right cuneus 0.16 0.03 0.13 0.02 1.52E-03 1.53E-04 1.28E-03 6.76E-05
Left lyngual gyrus 0.15 0.02 0.13 0.01 1.56E-03 1.55E-04 1.29E-03 8.82E-05
Right lyngual gyrus 0.16 0.03 0.13 0.01 1.50E-03 1.28E-04 1.27E-03 7.63E-05
Left amygdala 0.16 0.02 0.16 0.02 1.30E-03 1.13E-04 1.09E-03 6.49E-05
Right amygdala 0.18 0.02 0.18 0.02 1.31E-03 1.10E-04 1.09E-03 5.79E-05
Left hippocampus 0.15 0.02 0.16 0.02 1.57E-03 2.04E-04 1.29E-03 1.05E-04
Right hippocampus 0.16 0.02 0.16 0.02 1.49E-03 1.67E-04 1.28E-03 9.17E-05
Left cerebellar hemisphere 0.15 0.02 0.16 0.02 1.52E-03 1.22E-04 1.17E-03 6.91E-05
Right cerebellar hemisphere 0.15 0.02 0.17 0.02 1.48E-03 1.09E-04 1.15E-03 6.50E-05
Left insular cortex 0.15 0.02 0.15 0.01 1.42E-03 1.48E-04 1.29E-03 7.99E-05
Right insular cortex 0.15 0.02 0.16 0.01 1.33E-03 8.81E-05 1.24E-03 8.58E-05
Brain region AD 30 weeks (mm²/s) AD 40 weeks (mm²/s) RD 30 weeks (mm²/s) RD 40 weeks (mm²/s)
Mean SD Mean SD Mean SD Mean SD
Left corpus callosum 2.13E-03 1.67E-04 1.95E-03 1.43E-04 1.45E-03 1.62E-04 1.23E-03 1.09E-04
Right corpus callosum 2.06E-03 1.74E-04 1.89E-03 1.29E-04 1.39E-03 1.73E-04 1.22E-03 9.25E-05
Left anterior limb of internal capsule 1.74E-03 1.65E-04 1.46E-03 6.80E-05 1.29E-03 1.65E-04 9.90E-04 6.33E-05
Right anterior limb of internal capsule 1.72E-03 1.01E-04 1.48E-03 7.14E-05 1.24E-03 1.08E-04 9.86E-04 6.86E-05
Left posterior limb of internal capsule 1.72E-03 1.10E-04 1.56E-03 5.74E-05 1.11E-03 1.08E-04 7.89E-04 4.95E-05
Right posterior limb of internal capsule 1.74E-03 1.49E-04 1.54E-03 5.65E-05 1.09E-03 1.43E-04 7.97E-04 4.58E-05
Left retrolenticular part of internal capsule 1.84E-03 1.60E-04 1.55E-03 7.72E-05 1.23E-03 1.54E-04 9.54E-04 5.23E-05
Right retrolenticular part of internal capsule 1.85E-03 1.39E-04 1.57E-03 7.66E-05 1.22E-03 1.42E-04 9.65E-04 4.65E-05
Left anterior corona radiata 1.85E-03 9.31E-05 1.62E-03 1.07E-04 1.51E-03 9.64E-05 1.23E-03 1.05E-04
Right anterior corona radiata 1.85E-03 8.93E-05 1.63E-03 1.07E-04 1.47E-03 9.11E-05 1.25E-03 1.11E-04
Left superior corona radiata 1.89E-03 9.62E-05 1.65E-03 1.07E-04 1.55E-03 1.16E-04 1.18E-03 1.24E-04
Right superior corona radiata 1.88E-03 1.01E-04 1.63E-03 1.02E-04 1.53E-03 1.13E-04 1.15E-03 1.16E-04
Left posterior corona radiata 1.97E-03 1.22E-04 1.68E-03 1.22E-04 1.55E-03 1.43E-04 1.12E-03 1.17E-04
Right posterior corona radiata 2.00E-03 1.20E-04 1.69E-03 1.13E-04 1.54E-03 1.37E-04 1.12E-03 1.14E-04
Left cingulum cingular part 1.77E-03 1.44E-04 1.46E-03 7.21E-05 1.22E-03 1.13E-04 1.09E-03 5.94E-05
Right cingulum cingular part 1.68E-03 1.55E-04 1.43E-03 6.52E-05 1.19E-03 9.36E-05 1.09E-03 5.56E-05
Left cingulum hippocampal part 1.66E-03 1.26E-04 1.47E-03 8.08E-05 1.26E-03 1.16E-04 1.10E-03 5.84E-05
Right cingulum hippocampal part 1.65E-03 1.22E-04 1.46E-03 8.44E-05 1.24E-03 1.21E-04 1.11E-03 6.75E-05
Left fornix 2.29E-03 2.52E-04 2.19E-03 3.02E-04 1.69E-03 2.51E-04 1.55E-03 2.69E-04
Right fornix 2.36E-03 2.06E-04 2.13E-03 1.97E-04 1.74E-03 2.07E-04 1.46E-03 1.64E-04
Left stria terminalis 2.38E-03 3.58E-04 2.13E-03 3.32E-04 1.82E-03 3.81E-04 1.49E-03 2.91E-04
Right stria terminalis 2.09E-03 2.79E-04 1.99E-03 2.43E-04 1.54E-03 2.95E-04 1.35E-03 2.14E-04
Left tapetum 2.53E-03 3.60E-04 2.36E-03 2.61E-04 2.06E-03 3.34E-04 1.71E-03 2.59E-04
chapter 6
Right tapetum 2.32E-03 2.84E-04 2.23E-03 2.22E-04 1.84E-03 2.87E-04 1.59E-03 2.10E-04
Left superior longitudinal fasciculus 2.00E-03 1.11E-04 1.68E-03 1.17E-04 1.68E-03 1.10E-04 1.31E-03 1.09E-04
Right superior longitudinal fasciculus 1.96E-03 1.11E-04 1.63E-03 1.12E-04 1.61E-03 1.13E-04 1.29E-03 9.85E-05
Left external capsule 1.66E-03 1.31E-04 1.44E-03 7.12E-05 1.34E-03 1.11E-04 1.08E-03 5.84E-05
Right external capsule 1.67E-03 9.67E-05 1.44E-03 6.78E-05 1.32E-03 8.70E-05 1.07E-03 5.31E-05
Left posterior thalamic radiation 2.26E-03 2.63E-04 1.85E-03 1.18E-04 1.76E-03 2.71E-04 1.21E-03 1.05E-04
Right posterior thalamic radiation 2.14E-03 2.05E-04 1.80E-03 1.11E-04 1.58E-03 2.18E-04 1.17E-03 8.31E-05
Left sagittal stratum 2.31E-03 2.36E-04 1.88E-03 1.08E-04 1.82E-03 2.48E-04 1.24E-03 1.02E-04
Right sagittal stratum 2.15E-03 2.31E-04 1.85E-03 9.13E-05 1.63E-03 2.28E-04 1.19E-03 7.30E-05
Left thalamus 1.73E-03 2.04E-04 1.38E-03 9.57E-05 1.34E-03 1.77E-04 1.02E-03 7.24E-05
Right thalamus 1.70E-03 1.75E-04 1.42E-03 8.84E-05 1.32E-03 1.49E-04 1.06E-03 7.01E-05
Left putamen 1.59E-03 1.36E-04 1.27E-03 6.86E-05 1.31E-03 1.11E-04 1.02E-03 5.38E-05
Right putamen 1.56E-03 1.61E-04 1.26E-03 6.43E-05 1.28E-03 9.45E-05 1.03E-03 5.16E-05
Left globus pallidus 1.48E-03 1.12E-04 1.29E-03 6.12E-05 1.24E-03 8.49E-05 1.01E-03 5.39E-05
Right globus pallidus 1.48E-03 1.39E-04 1.29E-03 6.39E-05 1.24E-03 1.15E-04 1.01E-03 5.55E-05
Left caudate nucleus 2.08E-03 2.47E-04 1.92E-03 3.07E-04 1.71E-03 2.11E-04 1.56E-03 2.62E-04
Right caudate nucleus 1.97E-03 2.57E-04 1.86E-03 1.99E-04 1.62E-03 2.18E-04 1.51E-03 1.76E-04
Left cerebral peduncle 1.59E-03 1.36E-04 1.77E-03 1.08E-04 1.07E-03 1.49E-04 1.02E-03 8.71E-05
Right cerebral peduncle 1.57E-03 9.08E-05 1.74E-03 1.17E-04 1.05E-03 1.07E-04 1.01E-03 9.70E-05
Left superior fronto-occipital fasciculus 1.86E-03 1.41E-04 1.71E-03 2.15E-04 1.50E-03 1.59E-04 1.24E-03 2.11E-04
Right superior fronto-occipital fasciculus 1.83E-03 1.39E-04 1.62E-03 1.35E-04 1.44E-03 1.36E-04 1.17E-03 1.32E-04
Left inferior fronto-occipital fasciculus 1.74E-03 9.26E-05 1.53E-03 7.61E-05 1.27E-03 8.80E-05 1.02E-03 6.27E-05
Right inferior fronto-occipital fasciculus 1.73E-03 9.18E-05 1.55E-03 7.31E-05 1.28E-03 8.86E-05 1.04E-03 5.69E-05
Left corticospinal tract 1.51E-03 1.37E-04 1.44E-03 1.32E-04 1.11E-03 1.23E-04 9.51E-04 1.12E-04
Right corticospinal tract 1.50E-03 2.31E-04 1.43E-03 1.39E-04 1.14E-03 1.96E-04 9.93E-04 1.13E-04
Left superior cerebellar peduncle 1.56E-03 1.56E-04 1.55E-03 8.83E-05 1.21E-03 1.57E-04 1.00E-03 7.50E-05
Right superior cerebellar peduncle 1.57E-03 1.59E-04 1.57E-03 8.35E-05 1.21E-03 1.46E-04 1.04E-03 6.99E-05
Left middle cerebellar peduncle 1.67E-03 1.39E-04 1.57E-03 9.89E-05 1.24E-03 1.25E-04 9.71E-04 9.78E-05
Right middle cerebellar peduncle 1.68E-03 1.56E-04 1.66E-03 1.39E-04 1.23E-03 1.43E-04 1.05E-03 1.16E-04
Left inferior cerebellar peduncle 1.46E-03 1.52E-04 1.38E-03 9.13E-05 1.15E-03 1.51E-04 9.39E-04 6.71E-05
Right inferior cerebellar peduncle 1.45E-03 2.04E-04 1.30E-03 1.21E-04 1.12E-03 1.86E-04 8.54E-04 8.09E-05
Left pontine crossing tract 1.40E-03 1.17E-04 1.27E-03 1.13E-04 1.06E-03 1.07E-04 8.67E-04 7.82E-05
Right pontine crossing tract 1.42E-03 1.37E-04 1.25E-03 1.23E-04 1.09E-03 1.22E-04 8.93E-04 8.20E-05
Left uncinate fasciculus 1.66E-03 1.33E-04 1.50E-03 7.24E-05 1.23E-03 1.23E-04 1.03E-03 5.87E-05
Right uncinate fasciculus 1.63E-03 1.29E-04 1.48E-03 8.21E-05 1.17E-03 1.10E-04 1.03E-03 6.64E-05
Left midbrain 1.47E-03 1.56E-04 1.30E-03 1.04E-04 1.20E-03 1.50E-04 9.25E-04 7.90E-05
Right midbrain 1.51E-03 1.55E-04 1.34E-03 1.06E-04 1.23E-03 1.50E-04 9.51E-04 8.34E-05
Left pons 1.48E-03 1.40E-04 1.27E-03 8.73E-05 1.13E-03 1.33E-04 8.79E-04 5.88E-05
Right pons 1.56E-03 2.11E-04 1.28E-03 1.06E-04 1.20E-03 1.91E-04 8.93E-04 6.47E-05
Left medial lemniscus 1.45E-03 1.66E-04 1.23E-03 8.73E-05 1.07E-03 1.69E-04 7.94E-04 5.61E-05
Right medial lemniscus 1.45E-03 2.09E-04 1.24E-03 9.14E-05 1.07E-03 1.94E-04 8.26E-04 6.33E-05
Left medulla oblongata 1.42E-03 2.94E-04 1.37E-03 1.81E-04 1.12E-03 2.39E-04 9.84E-04 1.25E-04
Right medulla oblongata 1.46E-03 4.55E-04 1.41E-03 1.98E-04 1.14E-03 3.93E-04 1.00E-03 1.32E-04
Left superior frontal gyrus 1.68E-03 1.11E-04 1.60E-03 1.54E-04 1.35E-03 9.56E-05 1.27E-03 1.28E-04
Right superior frontal gyrus 1.70E-03 1.31E-04 1.66E-03 1.52E-04 1.36E-03 9.67E-05 1.32E-03 1.31E-04
Left middle frontal gyrus 1.66E-03 8.36E-05 1.57E-03 1.24E-04 1.36E-03 8.45E-05 1.25E-03 1.06E-04
Right middle frontal gyrus 1.66E-03 8.52E-05 1.52E-03 9.20E-05 1.36E-03 9.18E-05 1.22E-03 7.85E-05
chapter 6
Left inferior frontal gyrus 1.67E-03 9.55E-05 1.48E-03 8.49E-05 1.34E-03 9.14E-05 1.17E-03 7.73E-05
Right inferior frontal gyrus 1.67E-03 1.16E-04 1.51E-03 6.62E-05 1.34E-03 1.05E-04 1.19E-03 5.67E-05
Left medial fronto-orbital gyrus 1.64E-03 9.02E-05 1.57E-03 8.83E-05 1.30E-03 8.72E-05 1.22E-03 8.28E-05
Right medial fronto-orbital gyrus 1.64E-03 1.07E-04 1.57E-03 8.61E-05 1.29E-03 9.89E-05 1.23E-03 8.15E-05
Left lateral fronto-orbital gyrus 1.69E-03 9.91E-05 1.56E-03 8.79E-05 1.38E-03 9.13E-05 1.23E-03 7.69E-05
Right lateral fronto-orbital gyrus 1.68E-03 8.62E-05 1.56E-03 8.00E-05 1.35E-03 8.62E-05 1.23E-03 6.80E-05
Left gyrus rectus 1.57E-03 1.12E-04 1.61E-03 1.19E-04 1.24E-03 1.05E-04 1.27E-03 1.05E-04
Right gyrus rectus 1.61E-03 1.16E-04 1.59E-03 1.25E-04 1.28E-03 1.02E-04 1.26E-03 1.11E-04
Left precentral gyrus 1.71E-03 1.36E-04 1.45E-03 8.60E-05 1.38E-03 1.24E-04 1.13E-03 7.80E-05
Right precentral gyrus 1.67E-03 1.17E-04 1.42E-03 7.17E-05 1.35E-03 1.18E-04 1.11E-03 6.52E-05
Left postcentral gyrus 1.73E-03 1.46E-04 1.42E-03 8.38E-05 1.39E-03 1.20E-04 1.11E-03 7.25E-05
Right postcentral gyrus 1.70E-03 1.18E-04 1.41E-03 7.37E-05 1.37E-03 1.12E-04 1.10E-03 6.47E-05
Left superior parietal gyrus 1.95E-03 1.45E-04 1.55E-03 1.07E-04 1.61E-03 1.22E-04 1.23E-03 8.62E-05
Right superior parietal gyrus 1.91E-03 1.23E-04 1.51E-03 1.01E-04 1.58E-03 1.14E-04 1.20E-03 8.27E-05
Left precuneus 1.95E-03 2.00E-04 1.61E-03 1.57E-04 1.54E-03 1.73E-04 1.29E-03 1.10E-04
Right precuneus 1.96E-03 2.16E-04 1.61E-03 1.67E-04 1.53E-03 1.73E-04 1.29E-03 1.12E-04
Left cingular gyrus 1.69E-03 1.12E-04 1.45E-03 7.51E-05 1.32E-03 9.60E-05 1.16E-03 6.94E-05
Right cingular gyrus 1.68E-03 9.53E-05 1.53E-03 7.70E-05 1.31E-03 8.21E-05 1.22E-03 7.30E-05
Left supramarginal gyrus 1.79E-03 1.54E-04 1.47E-03 7.82E-05 1.42E-03 1.15E-04 1.18E-03 6.57E-05
Right supramarginal gyrus 1.69E-03 9.84E-05 1.43E-03 6.44E-05 1.35E-03 9.70E-05 1.15E-03 5.48E-05
Left angular gyrus 1.87E-03 1.21E-04 1.53E-03 7.68E-05 1.52E-03 1.04E-04 1.23E-03 6.70E-05
Right angular gyrus 1.79E-03 1.13E-04 1.48E-03 7.29E-05 1.45E-03 1.15E-04 1.19E-03 6.61E-05
Left superior temporal gyrus 1.70E-03 1.41E-04 1.54E-03 8.81E-05 1.33E-03 1.24E-04 1.22E-03 7.34E-05
Right superior temporal gyrus 1.63E-03 1.42E-04 1.50E-03 7.56E-05 1.28E-03 1.25E-04 1.20E-03 6.10E-05
Left middle temporal gyrus 1.74E-03 1.12E-04 1.48E-03 6.61E-05 1.35E-03 1.06E-04 1.19E-03 5.91E-05
Right middle temporal gyrus 1.68E-03 1.07E-04 1.47E-03 6.24E-05 1.31E-03 1.06E-04 1.18E-03 5.47E-05
Left inferior temporal gyrus 1.71E-03 9.14E-05 1.51E-03 6.17E-05 1.38E-03 9.93E-05 1.20E-03 5.73E-05
Right inferior temporal gyrus 1.69E-03 9.18E-05 1.51E-03 6.18E-05 1.35E-03 9.91E-05 1.20E-03 5.84E-05
Left fusiform gyrus 1.93E-03 1.52E-04 1.52E-03 7.33E-05 1.57E-03 1.43E-04 1.20E-03 5.88E-05
Right fusiform gyrus 1.83E-03 1.45E-04 1.50E-03 6.31E-05 1.47E-03 1.37E-04 1.19E-03 5.59E-05
Left parahippocampal gyrus 1.82E-03 2.17E-04 1.49E-03 1.12E-04 1.45E-03 2.03E-04 1.18E-03 8.48E-05
Right parahippocampal gyrus 1.68E-03 1.62E-04 1.43E-03 7.55E-05 1.31E-03 1.36E-04 1.14E-03 6.27E-05
Left entrhinal cortex 1.58E-03 1.63E-04 1.40E-03 1.09E-04 1.27E-03 1.37E-04 1.12E-03 9.12E-05
Right entrhinal cortex 1.56E-03 2.06E-04 1.38E-03 7.20E-05 1.27E-03 1.64E-04 1.11E-03 6.05E-05
Left superior occipital gyrus 2.00E-03 2.41E-04 1.51E-03 1.08E-04 1.58E-03 1.68E-04 1.21E-03 8.78E-05
Right superior occipital gyrus 1.93E-03 2.42E-04 1.52E-03 1.16E-04 1.52E-03 2.10E-04 1.21E-03 9.66E-05
Left middle occipital gyrus 1.78E-03 1.39E-04 1.45E-03 8.41E-05 1.40E-03 1.32E-04 1.17E-03 7.02E-05
Right middle occipital gyrus 1.72E-03 1.31E-04 1.43E-03 7.08E-05 1.36E-03 1.38E-04 1.15E-03 6.19E-05
Left inferior occipital gyrus 1.84E-03 1.33E-04 1.48E-03 9.94E-05 1.45E-03 1.33E-04 1.19E-03 7.79E-05
Right inferior occipital gyrus 1.74E-03 1.33E-04 1.44E-03 7.77E-05 1.37E-03 1.49E-04 1.17E-03 6.68E-05
Left cuneus 1.89E-03 1.87E-04 1.49E-03 9.21E-05 1.49E-03 1.53E-04 1.22E-03 7.59E-05
Right cuneus 1.78E-03 1.68E-04 1.45E-03 7.97E-05 1.39E-03 1.49E-04 1.19E-03 6.40E-05
Left lyngual gyrus 1.80E-03 1.60E-04 1.46E-03 1.02E-04 1.43E-03 1.54E-04 1.20E-03 8.31E-05
Right lyngual gyrus 1.74E-03 1.31E-04 1.44E-03 8.70E-05 1.38E-03 1.30E-04 1.19E-03 7.21E-05
Left amygdala 1.53E-03 1.39E-04 1.28E-03 8.47E-05 1.19E-03 1.03E-04 9.99E-04 5.77E-05
Right amygdala 1.55E-03 1.44E-04 1.29E-03 7.17E-05 1.18E-03 9.71E-05 9.85E-04 5.37E-05
Left hippocampus 1.83E-03 2.25E-04 1.50E-03 1.18E-04 1.45E-03 1.95E-04 1.18E-03 1.00E-04
chapter 6
Right hippocampus 1.75E-03 1.97E-04 1.49E-03 1.07E-04 1.36E-03 1.55E-04 1.17E-03 8.66E-05
Left cerebellar hemisphere 1.76E-03 1.47E-04 1.37E-03 8.87E-05 1.40E-03 1.12E-04 1.07E-03 6.22E-05
Right cerebellar hemisphere 1.72E-03 1.22E-04 1.35E-03 8.13E-05 1.37E-03 1.05E-04 1.05E-03 5.95E-05
Left insular cortex 1.64E-03 1.65E-04 1.49E-03 9.42E-05 1.32E-03 1.41E-04 1.18E-03 7.38E-05
Right insular cortex 1.53E-03 9.74E-05 1.44E-03 1.02E-04 1.22E-03 8.54E-05 1.14E-03 7.93E-05
Brain region Cluster Delta Left-Right difference
FA MD AD RD FA MD AD RD
p-value p-value p-value p-value p-value p-value p-value p-value
Left corpus callosum 3 <0.0001 <0.0001 <0.0001 <0.0001 0.001 0.172 0.849 0.066
Right corpus callosum 3 0.0001 <0.0001 <0.0001 <0.0001
Left anterior limb of internal capsule 4 <0.0001 <0.0001 <0.0001 <0.0001 0.408 0.005 0.010 0.009
Right anterior limb of internal capsule 4 <0.0001 <0.0001 <0.0001 <0.0001
Left posterior limb of internal capsule 3 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.717 0.016 0.050
Right posterior limb of internal capsule 3 <0.0001 <0.0001 <0.0001 <0.0001
Left retrolenticular part of internal capsule 3 <0.0001 <0.0001 <0.0001 <0.0001 0.657 0.483 0.576 0.472
Right retrolenticular part of internal capsule 3 <0.0001 <0.0001 <0.0001 <0.0001
Left anterior corona radiata 1 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.173 <0.0001
Right anterior corona radiata 1 <0.0001 <0.0001 <0.0001 <0.0001
Left superior corona radiata 4 <0.0001 <0.0001 <0.0001 <0.0001 0.168 0.474 0.282 0.627
Right superior corona radiata 4 <0.0001 <0.0001 <0.0001 <0.0001
Left posterior corona radiata 4 <0.0001 <0.0001 <0.0001 <0.0001 0.203 0.903 0.316 0.698
Right posterior corona radiata 4 <0.0001 <0.0001 <0.0001 <0.0001
Left cingulum cingular part 2 <0.0001 <0.0001 <0.0001 <0.0001 0.357 0.014 0.039 0.007
Right cingulum cingular part 2 <0.0001 <0.0001 <0.0001 <0.0001
Left cingulum hippocampal part 2 0.4460 <0.0001 <0.0001 <0.0001 0.028 0.213 0.786 0.087
Right cingulum hippocampal part 2 0.1171 <0.0001 <0.0001 <0.0001
Left fornix 4 <0.0001 0.0022 0.0118 0.0009 <0.0001 <0.0001 <0.0001 <0.0001
Right fornix 4 <0.0001 <0.0001 <0.0001 <0.0001
Left stria terminalis 4 <0.0001 <0.0001 0.0001 <0.0001 0.295 0.006 0.006 0.006
Right stria terminalis 4 <0.0001 0.0005 0.0212 0.0001

chapter 6
Left tapetum 4 <0.0001 <0.0001 0.0037 <0.0001 0.529 0.05 0.085 0.041
Right tapetum 4 <0.0001 0.0001 0.0451 <0.0001
Left superior longitudinal fasciculus 1 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.011 0.263 <0.0001
Right superior longitudinal fasciculus 1 <0.0001 <0.0001 <0.0001 <0.0001
Left external capsule 1 <0.0001 <0.0001 <0.0001 <0.0001 0.012 0.603 0.468 0.289
Right external capsule 1 <0.0001 <0.0001 <0.0001 <0.0001
Left posterior thalamic radiation 4 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.001 0.043 <0.0001
Right posterior thalamic radiation 4 <0.0001 <0.0001 <0.0001 <0.0001
Left sagittal stratum 4 <0.0001 <0.0001 <0.0001 <0.0001 0.165 <0.0001 <0.0001 <0.0001
Right sagittal stratum 4 <0.0001 <0.0001 <0.0001 <0.0001
Left thalamus 4 <0.0001 <0.0001 <0.0001 <0.0001 0.144 0.001 0.001 <0.0001
Right thalamus 2 <0.0001 <0.0001 <0.0001 <0.0001
Left putamen 1 0.0002 <0.0001 <0.0001 <0.0001 0.023 <0.0001 0.058 <0.0001
Right putamen 1 0.1630 <0.0001 <0.0001 <0.0001
Left globus pallidus 1 <0.0001 <0.0001 <0.0001 <0.0001 0.001 0.687 0.634 0.470
Right globus pallidus 1 <0.0001 <0.0001 <0.0001 <0.0001
Left caudate nucleus 1 0.2767 <0.0001 <0.0001 <0.0001 0.318 0.231 0.204 0.259
Right caudate nucleus 1 0.0656 <0.0001 0.0002 <0.0001
Left cerebral peduncle 3 <0.0001 0.362 <0.0001 0.0272 0.205 0.739 0.879 0.536
Right cerebral peduncle 3 <0.0001 0.1135 <0.0001 0.0097
Left superior fronto-occipital fasciculus 1 <0.0001 <0.0001 <0.0001 <0.0001 0.018 0.18 0.016 0.535
Right superior fronto-occipital fasciculus 1 <0.0001 <0.0001 <0.0001 <0.0001
Left inferior fronto-occipital fasciculus 4 <0.0001 <0.0001 <0.0001 <0.0001 0.615 0.044 0.011 0.222
Right inferior fronto-occipital fasciculus 4 <0.0001 <0.0001 <0.0001 <0.0001
Left corticospinal tract 4 <0.0001 <0.0001 0.0093 <0.0001 0.063 0.795 0.882 0.607
Right corticospinal tract 4 <0.0001 0.0007 0.0471 <0.0001
Left superior cerebellar peduncle 4 <0.0001 <0.0001 0.8056 <0.0001 <0.0001 0.29 0.921 0.100
Right superior cerebellar peduncle 4 <0.0001 <0.0001 0.7655 <0.0001
Left middle cerebellar peduncle 3 <0.0001 <0.0001 0.0002 <0.0001 <0.0001 <0.0001 0.007 <0.0001
Right middle cerebellar peduncle 3 <0.0001 <0.0001 0.3429 <0.0001
Left inferior cerebellar peduncle 4 <0.0001 <0.0001 0.0048 <0.0001 0.826 0.046 0.041 0.055
Right inferior cerebellar peduncle 4 <0.0001 <0.0001 <0.0001 <0.0001
Left pontine crossing tract 4 <0.0001 <0.0001 <0.0001 <0.0001 0.004 0.464 0.139 0.874
Right pontine crossing tract 4 <0.0001 <0.0001 <0.0001 <0.0001
Left uncinate fasciculus 4 <0.0001 <0.0001 <0.0001 <0.0001 0.015 0.008 0.452 <0.0001
Right uncinate fasciculus 4 0.0006 <0.0001 <0.0001 <0.0001
Left midbrain 4 <0.0001 <0.0001 <0.0001 <0.0001 0.105 0.748 0.939 0.599
Right midbrain 4 <0.0001 <0.0001 <0.0001 <0.0001
Left pons 4 <0.0001 <0.0001 <0.0001 <0.0001 0.959 0.008 0.009 0.009
Right pons 4 <0.0001 <0.0001 <0.0001 <0.0001
Left medial lemniscus 4 <0.0001 <0.0001 <0.0001 <0.0001 0.001 0.278 0.695 0.148
Right medial lemniscus 4 <0.0001 <0.0001 <0.0001 <0.0001
Left medulla oblongata 4 <0.0001 0.0195 0.3417 0.0018 0.911 0.885 0.908 0.872
Right medulla oblongata 4 <0.0001 0.0859 0.4407 0.0247
Left superior frontal gyrus 1 0.0011 0.0001 0.0002 <0.0001 0.014 <0.0001 <0.0001 <0.0001
Right superior frontal gyrus 1 0.1154 0.0729 0.0949 0.0791
Left middle frontal gyrus 1 0.0008 <0.0001 <0.0001 <0.0001 0.746 <0.0001 <0.0001 <0.0001
chapter 6
Right middle frontal gyrus 1 0.0020 <0.0001 <0.0001 <0.0001
Left inferior frontal gyrus 2 0.0945 <0.0001 <0.0001 <0.0001 0.066 0.171 0.063 0.293
Right inferior frontal gyrus 1 0.0042 <0.0001 <0.0001 <0.0001
Left medial fronto-orbital gyrus 1 <0.0001 <0.0001 <0.0001 <0.0001 0.172 0.344 0.599 0.251
Right medial fronto-orbital gyrus 1 <0.0001 <0.0001 <0.0001 <0.0001
Left lateral fronto-orbital gyrus 1 <0.0001 <0.0001 <0.0001 <0.0001 0.221 0.107 0.350 0.055
Right lateral fronto-orbital gyrus 1 <0.0001 <0.0001 <0.0001 <0.0001
Left gyrus rectus 2 0.3339 0.1757 0.1391 0.2118 0.705 0.0120 0.011 0.015
Right gyrus rectus 2 0.5305 0.2525 0.2367 0.2801
Left precentral gyrus 1 <0.0001 <0.0001 <0.0001 <0.0001 0.001 0.186 0.449 0.108
Right precentral gyrus 1 <0.0001 <0.0001 <0.0001 <0.0001
Left postcentral gyrus 1 <0.0001 <0.0001 <0.0001 <0.0001 0.008 0.662 0.301 0.976
Right postcentral gyrus 1 <0.0001 <0.0001 <0.0001 <0.0001
Left superior parietal gyrus 1 <0.0001 <0.0001 <0.0001 <0.0001 0.002 0.931 0.733 0.730
Right superior parietal gyrus 1 <0.0001 <0.0001 <0.0001 <0.0001
Left precuneus 2 0.0013 <0.0001 <0.0001 <0.0001 <0.0001 0.772 0.557 0.437
Right precuneus 2 <0.0001 <0.0001 <0.0001 <0.0001
Left cingular gyrus 2 <0.0001 <0.0001 <0.0001 <0.0001 0.175 <0.0001 <0.0001 <0.0001
Right cingular gyrus 2 <0.0001 <0.0001 <0.0001 <0.0001
Left supramarginal gyrus 2 0.0293 <0.0001 <0.0001 <0.0001 0.604 0.002 0.007 0.001
Right supramarginal gyrus 2 0.0121 <0.0001 <0.0001 <0.0001
Left angular gyrus 1 0.0927 <0.0001 <0.0001 <0.0001 0.009 0.003 0.008 0.003
Right angular gyrus 1 0.9038 <0.0001 <0.0001 <0.0001
Left superior temporal gyrus 2 0.0411 <0.0001 <0.0001 <0.0001 0.748 0.008 0.006 0.012
Right superior temporal gyrus 2 0.0533 <0.0001 <0.0001 <0.0001
Left middle temporal gyrus 2 <0.0001 <0.0001 <0.0001 <0.0001 0.249 <0.0001 <0.0001 <0.0001
Right middle temporal gyrus 2 <0.0001 <0.0001 <0.0001 <0.0001
Left inferior temporal gyrus 1 0.0114 <0.0001 <0.0001 <0.0001 <0.0001 0.003 0.140 <0.0001
Right inferior temporal gyrus 2 0.2685 <0.0001 <0.0001 <0.0001
Left fusiform gyrus 1 0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001
Right fusiform gyrus 1 0.0810 <0.0001 <0.0001 <0.0001
Left parahippocampal gyrus 2 0.0862 <0.0001 <0.0001 <0.0001 0.054 0.002 0.009 0.001
Right parahippocampal gyrus 2 0.0001 <0.0001 <0.0001 <0.0001
Left entrhinal cortex 1 0.0116 <0.0001 <0.0001 <0.0001 0.223 0.742 0.926 0.638
Right entrhinal cortex 1 <0.0001 <0.0001 <0.0001 <0.0001
Left superior occipital gyrus 2 0.0107 <0.0001 <0.0001 <0.0001 0.168 0.033 0.027 0.043
Right superior occipital gyrus 2 0.1883 <0.0001 <0.0001 <0.0001
Left middle occipital gyrus 2 0.0007 <0.0001 <0.0001 <0.0001 0.726 0.031 0.020 0.045
Right middle occipital gyrus 2 0.0016 <0.0001 <0.0001 <0.0001
Left inferior occipital gyrus 2 <0.0001 <0.0001 <0.0001 <0.0001 0.024 <0.0001 0.002 <0.0001
Right inferior occipital gyrus 2 <0.0001 <0.0001 <0.0001 <0.0001
Left cuneus 2 <0.0001 <0.0001 <0.0001 <0.0001 0.169 <0.0001 0.001 <0.0001
Right cuneus 2 <0.0001 <0.0001 <0.0001 <0.0001
Left lyngual gyrus 2 <0.0001 <0.0001 <0.0001 <0.0001 0.055 0.051 0.069 0.046
Right lyngual gyrus 2 <0.0001 <0.0001 <0.0001 <0.0001
Left amygdala 2 0.8077 <0.0001 <0.0001 <0.0001 0.796 0.546 0.597 0.533
Right amygdala 2 0.9733 <0.0001 <0.0001 <0.0001
Left hippocampus 2 0.2353 <0.0001 <0.0001 <0.0001 0.023 0.007 0.029 0.004
Right hippocampus 2 0.3141 <0.0001 <0.0001 <0.0001

chapter 6
Left cerebellar hemisphere 1 0.0003 <0.0001 <0.0001 <0.0001 0.019 0.141 0.087 0.218
Right cerebellar hemisphere 1 <0.0001 <0.0001 <0.0001 <0.0001
Left insular cortex 1 0.0964 <0.0001 <0.0001 <0.0001 0.927 0.001 0.001 0.001
Right insular cortex 1 0.1207 <0.0001 <0.0001 <0.0001
References
1 Oishi K, Mori S, Donohue PK, Ernst T, Anderson L, 13 Talairach J, Tournoux P. Co-planar stereotaxic
Buchthal S et al. Multi-contrast human neonatal atlas of the human brain: 3-dimensional
brain atlas: application to normal neonate proportional system: an approach to cerebral
development analysis. Neuroimage 2011;56:8- imaging. New York: Thieme. 1988.
20. 14 Griffiths R. The abilities of young children. A
2 Tournier JD, Mori S, Leemans A. Diffusion tensor comprehensive system of mental measurement
imaging and beyond. Magn Reson Med for the first eight years of life. London: The test
2011;65:1532-56. agency Ltd. 1984.
3 Aeby A, Van Bogaert P, David P, Baleriaux D, 15 Jones DK, Leemans A. Diffusion tensor imaging.
Vermeylen D, Metens T et al. Nonlinear Methods Mol Biol 2011;711:127-44.
microstructural changes in the right superior 16 Kidokoro H, Neil JJ, Inder TE. New MR Imaging
temporal sulcus and lateral occipitotemporal Assessment Tool to Define Brain Abnormalities in
gyrus between 35 and 43 weeks in the preterm Very Preterm Infants at Term. AJNR Am J
brain. Neuroimage 2012;63:104-10. Neuroradiol 2013.
4 Berman JI, Mukherjee P, Partridge SC, Miller SP, 17 Papile LA, Burstein J, Burstein R, Koffler H.
Ferriero DM, Barkovich AJ et al. Quantitative Incidence and evolution of subependymal and
diffusion tensor MRI fiber tractography of intraventricular hemorrhage: a study of infants
sensorimotor white matter development in with birth weights less than 1,500 gm. J Pediatr
premature infants. Neuroimage 2005;27:862-71. 1978;92:529-34.
5 Huang H, Xue R, Zhang J, Ren T, Richards LJ, 18 Leemans A, Jeurissen B, Sijbers J, Jones DK.
Yarowsky P et al. Anatomical characterization of Explore DTI: a graphical toolbox for processing,
human fetal brain development with diffusion analyzing, and visualizing diffusion MR data.
tensor magnetic resonance imaging. J Neurosci Hawaii, USA 2009 p. 3537.
2009;29:4263-73. 19 Leemans A, Jones DK. The B-matrix must be
6 Huppi PS, Maier SE, Peled S, Zientara GP, Barnes rotated when correcting for subject motion in
PD, Jolesz FA et al. Microstructural development DTI data. Magn Reson Med 2009;61:1336-49.
of human newborn cerebral white matter 20 Tax CM, Otte WM, Viergever MA, Dijkhuizen RM,
assessed in vivo by diffusion tensor magnetic Leemans A. REKINDLE: Robust extraction of
resonance imaging. Pediatr Res 1998;44:584-90. kurtosis INDices with linear estimation. Magn
7 Lebel C, Walker L, Leemans A, Phillips L, Beaulieu Reson Med 2014.
C. Microstructural maturation of the human 21 Veraart J, Sijbers J, Sunaert S, Leemans A,
chapter 6
brain from childhood to adulthood. Neuroimage Jeurissen B. Weighted linear least squares
2008;40:1044-55. estimation of diffusion MRI parameters:
8 Miller JH, McKinstry RC, Philip JV, Mukherjee P, strengths, limitations, and pitfalls. Neuroimage
Neil JJ. Diffusion-tensor MR imaging of normal 2013;81:335-46.
brain maturation: a guide to structural 22 Klein S, Staring M, Murphy K, Viergever MA,
development and myelination. AJR Am J Pluim JP. elastix: a toolbox for intensity-based
Roentgenol 2003;180:851-9. medical image registration. IEEE Trans Med
9 Bonifacio SL, Glass HC, Chau V, Berman JI, Xu D, Imaging 2010;29:196-205.
Brant R et al. Extreme premature birth is not 23 R: A Language and Environment for Statistical
associated with impaired development of brain Computing [computer program]. Vienna,
microstructure. J Pediatr 2010;157:726-32. Austria: 2012.
10 Neil JJ, Shiran SI, McKinstry RC, Schefft GL, Snyder 24 Miller SP, Vigneron DB, Henry RG, Bohland MA,
AZ, Almli CR et al. Normal brain in human Ceppi-Cozzio C, Hoffman C et al. Serial
newborns: apparent diffusion coefficient and quantitative diffusion tensor MRI of the
diffusion anisotropy measured by using diffusion premature brain: development in newborns with
tensor MR imaging. Radiology 1998;209:57-66. and without injury. J Magn Reson Imaging
11 Beaulieu C. The basis of anisotropic water 2002;16:621-32.
diffusion in the nervous system - a technical 25 Aeby A, Liu Y, De Tiège X, Denolin V, David P,
review. NMR Biomed 2002;15:435-55. Baleriaux D et al. Maturation of thalamic
12 Deprez S, Billiet T, Sunaert S, Leemans A. radiations between 34 and 41 weeks’ gestation:
Diffusion tensor MRI of chemotherapy-induced a combined voxel-based study and probabilistic
cognitive impairment in non-CNS cancer tractography with diffusion tensor imaging. AJNR
patients: a review. Brain Imaging Behav Am J Neuroradiol 2009;30:1780-6.
2013;7:409-35. 26 Partridge SC, Mukherjee P, Henry RG, Miller SP,
Berman JI, Jin H et al. Diffusion tensor imaging: Arterial Spin Labeling MRI in neonates. Eur J
serial quantitation of white matter tract maturity Radiol 2013;82:538-43.
in premature newborns. Neuroimage 39 Back SA, Luo NL, Borenstein NS, Volpe JJ, Kinney
2004;22:1302-14. HC. Arrested oligodendrocyte lineage
27 Cheong JL, Thompson DK, Wang HX, Hunt RW, progression during human cerebral white matter
Anderson PJ, Inder TE et al. Abnormal white development: dissociation between the timing
matter signal on MR imaging is related to of progenitor differentiation and
abnormal tissue microstructure. AJNR Am J myelinogenesis. J Neuropathol Exp Neurol
Neuroradiol 2009;30:623-8. 2002;61:197-211.
28 Bui T, Daire JL, Chalard F, Zaccaria I, Alberti C, 40 Kroenke CD, Bretthorst GL, Inder TE, Neil JJ.
Elmaleh M et al. Microstructural development of Diffusion MR imaging characteristics of the
human brain assessed in utero by diffusion developing primate brain. Neuroimage
tensor imaging. Pediatr Radiol 2006;36:1133-40. 2005;25:1205-13.
29 Kasprian G, Brugger PC, Weber M, Krssak M, 41 Prayer D, Barkovich AJ, Kirschner DA, Prayer LM,
Krampl E, Herold C et al. In utero tractography of Roberts TP, Kucharczyk J et al. Visualization of
fetal white matter development. Neuroimage nonstructural changes in early white matter
2008;43:213-24. development on diffusion-weighted MR images:
30 Volpe JJ. Neurology of the Newborn. Chapter 2: evidence supporting premyelination anisotropy.
Neuronal Proliferation, Migration, Organization, AJNR Am J Neuroradiol 2001;22:1572-6.
and Myelination. 51-118. 2008. Philadelphia, 42 Wimberger DM, Roberts TP, Barkovich AJ, Prayer
Saunders, Elsevier Inc. LM, Moseley ME, Kucharczyk J. Identification of
31 Kostovic I, Vasung L. Insights from in vitro fetal “premyelination” by diffusion-weighted MRI. J
magnetic resonance imaging of cerebral Comput Assist Tomogr 1995;19:28-33.
development. Semin Perinatol 2009;33:220-33. 43 Dubois J, Dehaene-Lambertz G, Perrin M, Mangin
32 Kinney HC, Brody BA, Kloman AS, Gilles FH. JF, Cointepas Y, Duchesnay E et al. Asynchrony of
Sequence of central nervous system myelination the early maturation of white matter bundles in
in human infancy. II. Patterns of myelination in healthy infants: quantitative landmarks revealed
autopsied infants. J Neuropathol Exp Neurol noninvasively by diffusion tensor imaging. Hum
1988;47:217-34. Brain Mapp 2008;29:14-27.
33 Nossin-Manor R, Card D, Morris D, 44 LaMantia AS, Rakic P. Axon overproduction and
Noormohamed S, Shroff MM, Whyte HE et al. elimination in the corpus callosum of the
cha p t e r 6
Quantitative MRI in the very preterm brain: developing rhesus monkey. J Neurosci
assessing tissue organization and myelination 1990;10:2156-75.
using magnetization transfer, diffusion tensor 45 Jones DK, Knosche TR, Turner R. White matter
and T(1) imaging. Neuroimage 2013;64:505-16. integrity, fiber count, and other fallacies: the do’s
34 Rose J, Vassar R, Cahill-Rowley K, Guzman XS, and don’ts of diffusion MRI. Neuroimage
Stevenson DK, Barnea-Goraly N. Brain 2013;73:239-54.
microstructural development at near-term age 46 Jeurissen B, Leemans A, Tournier JD, Jones DK,
in very-low-birth-weight preterm infants: An Sijbers J. Investigating the prevalence of complex
atlas-based diffusion imaging study. Neuroimage fiber configurations in white matter tissue with
2014;86:244-56. diffusion magnetic resonance imaging. Hum
35 Keene MF, Hewer EE. Some Observations on Brain Mapp 2013;34:2747-66.
Myelination in the Human Central Nervous 47 Raffelt D, Tournier JD, Rose S, Ridgway GR,
System. J Anat 1931;66:1-13. Henderson R, Crozier S et al. Apparent Fibre
36 van den Heuvel MP, Kersbergen KJ, de Reus MA, Density: a novel measure for the analysis of
Keunen K, Kahn RS, Groenendaal F et al. The diffusion-weighted magnetic resonance images.
Neonatal Connectome During Preterm Brain Neuroimage 2012;59:3976-94.
Development. Cereb Cortex 2014. 48 Ball G, Srinivasan L, Aljabar P, Counsell SJ,
37 Chugani HT. A critical period of brain Durighel G, Hajnal JV et al. Development of
development: studies of cerebral glucose cortical microstructure in the preterm human
utilization with PET. Prev Med 1998;27:184-8. brain. Proc Natl Acad Sci U S A 2013;110:9541-6.
38 De Vis JB, Petersen ET, de Vries LS, Groenendaal F, 49 deIpolyi AR, Mukherjee P, Gill K, Henry RG,
Kersbergen KJ, Alderliesten T et al. Regional Partridge SC, Veeraraghavan S et al. Comparing
changes in brain perfusion during brain microstructural and macrostructural
maturation measured non-invasively with development of the cerebral cortex in
premature newborns: diffusion tensor imaging 58 Anjari M, Srinivasan L, Allsop JM, Hajnal JV,
versus cortical gyration. Neuroimage Rutherford MA, Edwards AD et al. Diffusion
2005;27:579-86. tensor imaging with tract-based spatial statistics
50 Gupta RK, Hasan KM, Trivedi R, Pradhan M, Das V, reveals local white matter abnormalities in
Parikh NA et al. Diffusion tensor imaging of the preterm infants. Neuroimage 2007;35:1021-7.
developing human cerebrum. J Neurosci Res 59 Rose J, Butler EE, Lamont LE, Barnes PD, Atlas SW,
2005;81:172-8. Stevenson DK. Neonatal brain structure on MRI
51 McKinstry RC, Mathur A, Miller JH, Ozcan A, and diffusion tensor imaging, sex, and
Snyder AZ, Schefft GL et al. Radial organization of neurodevelopment in very-low-birthweight
developing preterm human cerebral cortex preterm children. Dev Med Child Neurol
revealed by non-invasive water diffusion 2009;51:526-35.
anisotropy MRI. Cereb Cortex 2002;12:1237-43. 60 Miller SP, Cozzio CC, Goldstein RB, Ferriero DM,
52 Huttenlocher PR, Dabholkar AS. Regional Partridge JC, Vigneron DB et al. Comparing the
differences in synaptogenesis in human cerebral diagnosis of white matter injury in premature
cortex. J Comp Neurol 1997;387:167-78. newborns with serial MR imaging and
53 Sidman RL, Rakic P. Neuronal migration, with transfontanel ultrasonography findings. AJNR Am
special reference to developing human brain: a J Neuroradiol 2003;24:1661-9.
review. Brain Res 1973;62:1-35. 61 Vos SB, Jones DK, Viergever MA, Leemans A.
54 Chi JG, Dooling EC, Gilles FH. Gyral development Partial volume effect as a hidden covariate in DTI
of the human brain. Ann Neurol 1977;1:86-93. analyses. Neuroimage 2011;55:1566-76.
55 Dubois J, Benders M, Cachia A, Lazeyras F, Ha- 62 van Hecke W., Sijbers J, D’Agostino E, Maes F, de
Vinh LR, Sizonenko SV et al. Mapping the early Backer S., Vandervliet E et al. On the construction
cortical folding process in the preterm newborn of an inter-subject diffusion tensor magnetic
brain. Cereb Cortex 2008;18:1444-54. resonance atlas of the healthy human brain.
56 Sun T, Patoine C, Abu-Khalil A, Visvader J, Sum E, Neuroimage 2008;43:69-80.
Cherry TJ et al. Early asymmetry of gene 63 van Hecke W., Leemans A, Sage CA, Emsell L,
transcription in embryonic human left and right Veraart J, Sijbers J et al. The effect of template
cerebral cortex. Science 2005;308:1794-8. selection on diffusion tensor voxel-based
57 Bourgeois JP, Jastreboff PJ, Rakic P. analysis results. Neuroimage 2011;55:566-73.
Synaptogenesis in visual cortex of normal and 64 Vos SB, Jones DK, Jeurissen B, Viergever MA,
preterm monkeys: evidence for intrinsic Leemans A. The influence of complex white
cha p t e r 6
regulation of synaptic overproduction. Proc Natl matter architecture on the mean diffusivity in
Acad Sci U S A 1989;86:4297-301. diffusion tensor MRI of the human brain.
Neuroimage 2012;59:2208-16.
ch a p t er
7
Corticospinal
tract injury
precedes thalamic
volume reduction
in preterm
infants with
cystic
periventricular
leukomalacia
Linda S. de Vries
Floris Groenendaal
Andrew T.M. Chew
Antonios Makropoulos
Sarah Dawson
Frances M. Cowan
Serena J. Counsell
Submitted 141
Abstract
Objectives Cystic periventricular leukomalacia (c-PVL) in preterm infants affects the
corticospinal tracts (CST) as well as the central grey matter nuclei. The aim of our study
was to measure both fractional anisotropy (FA) values in the CST and volume of the
thalami in preterm infants with c-PVL and to compare these measurements to control
infants.
Study design Preterm infants with c-PVL and controls with MRI data acquired between
birth and term equivalent age (TEA) were retrospectively identified in two centres.
Tractography of the CST and segmentation of the thalamus were performed, and values
from infants with c-PVL and controls were compared.
Results Thirty-three subjects with c-PVL and 31 preterm controls were identified. All had at
least one scan up to TEA and multiple scans were performed in 31 infants. A significant
difference in FA values of the CST was found between cases and controls on the scans both
before and at TEA. Thalamic volumes were significantly reduced at TEA but not on the
earlier scans. Data acquired in infancy showed lower FA values in infants with c-PVL.
Discussion Damage to the CST can be identified on the early scan and persists, whereas the
changes in thalamic volume develop in the weeks between the early and term equivalent
MRI. This may reflect the difference between acute and remote effects of the extensive
injury to the white matter caused by c-PVL.
chapter 7
Introduction
D
espite the decreasing incidence, cystic periventricular leukomalacia (c-PVL) remains
one of the leading causes of cerebral palsy (CP) after preterm birth.1,2 Early recognition
of infants at risk of developing CP is important, both for accurate counselling of parents
as well as for a possible selection of infants that may benefit from early behavioural
interventions or rehabilitation services.3 Assessment of the myelination of the posterior limb
of the internal capsule (PLIC) on conventional magnetic resonance imaging (MRI) at term
equivalent age (TEA) can reliably predict the development of CP in most infants.4 However, the
severity of CP as well as cognitive and behavioural outcomes and the development of epilepsy,
are still hard to predict.
Over the last decade, the use of diffusion tensor imaging (DTI) has made the in vivo assessment
of early human brain development possible on a microstructural level, providing a better
understanding of both the direct and remote effects of injury to the white or grey matter.5
Several studies have shown a correlation between the presence and severity of CP after PVL and
fractional anisotropy (FA) in the corticospinal tract (CST) as a measure of white matter injury in
childhood. FA measurements were lower in children with CP compared to healthy controls, and
also lower in children that were more severely impaired compared to those with mild,
ambulatory CP.6-8 It is however unclear when these differences develop, since these studies
were performed in older children and studies in the neonatal period are few, and often include
infants with non-cystic PVL.9 This is also the case for neonatal studies assessing thalamic
involvement in children with CP after PVL. Thalamic involvement is thought to occur as a
secondary process to white matter injury and thalamic volumes were reduced in children with
CP compared to healthy controls.10,11 Again, this appears to be related to the severity of CP.12,13
The aim of our study was to measure both FA values in the CST and thalamic volumes in
preterm infants with c-PVL, scanned during the early neonatal period and again around term
equivalent age (TEA), and to compare these measurements to control infants. Additionally, we
aimed to assess the association between these measures and gross motor outcome in infancy.
ch a pt er 7
Methods
For this study, all infants born below a gestational age (GA) of 36 weeks with a clinical
diagnosis of c-PVL on cranial ultrasound and at least one MRI during the neonatal period were
retrospectively identified. Infants born between March 2005 and September 2013 from both
the Wilhelmina Children’s Hospital in Utrecht, the Netherlands and Queen Charlotte’s and
Chelsea Hospital in London, United Kingdom, were eligible for inclusion. PVL-grading was
determined on sequential cranial ultrasound, according to de Vries et al.14 Scans were
obtained within the first few weeks after the insult occurred as assessed from cranial
ultrasound scans in most patients and were repeated at TEA. Preterm control infants without
significant brain injury on their MRI and normal motor development at follow-up around 15
corticospinal tract injury precedes thalamic volume reduction in preterm infants with cystic 143
periventricular leukomalacia
months corrected age were randomly selected from our preterm population and were
matched for gender. Scans in the controls were done around 30 weeks post-menstrual age
(PMA) for infants with a GA <28 weeks and soon after birth on clinical indication in those with
a GA >28 weeks, and were repeated at TEA. For a subgroup of c-PVL patients, scans in infancy
were also available. Outcome data were collected from the patient charts.
Neurodevelopmental outcome was formally assessed in the out-patient follow-up clinic and
all children had a standard neurological examination. Severity of CP was graded according to
the gross motor function classification system (GMFCS).15
Magn e t ic r e s o na nc e i m agi ng
For the infants from Utrecht, MRI was performed on either a 1.5 Tesla ACS-NT system or a 3
Tesla whole-body Achieva system (Philips Medical Systems, Best, the Netherlands) with the
phased array head coil. On the 1.5T magnet, the routine protocol included conventional
inversion recovery-weighted imaging and T2-weighted imaging (30 and 40 weeks: inversion
recovery-weighted repetition time [TR] 4147 ms; inversion time [TI] 600 ms; echo time [TE] 30
ms; slice thickness 2 mm and T2-weighted TR 7656 ms; TE 150 ms; slice thickness 2 mm) and
on the 3T magnet, the routine protocol included conventional 3D T1-weighted and T2-
weighted imaging (30 weeks: 3D T1-weighted TR 9.4 ms; TE 4.6 ms; slice thickness 2 mm and
T2-weighted TR 10085 ms; TE 120ms; slice thickness 2 mm; 40 weeks: 3D T1-weigthed TR 9.5
ms; TE 4.6 ms; slice thickness 1.2-2 mm and T2-weighted TR 4847-6293 ms; TE 120-150 ms;
slice thickness 1.2-2 mm). The DTI protocol consisted of a single-shot spin-echo echo-planar
imaging sequence (echo-planar imaging factor 55, TR/TE 5685/70 ms, field of view 180x146
mm, acquisition matrix 128x102 mm, reconstruction matrix 128x128 mm, 50 slices with 2
mm thickness without gap). Images were acquired in the axial plane with diffusion gradients
applied in 32 non-collinear directions with a b-value of 800 s/mm2 and one non-diffusion
weighted image, with a total scan time of 4.28 min. In infancy, all children were scanned on a
1.5T magnet. A similar single-shot echo-planar imaging sequence was used but with an echo-
cha p te r 7
planar imaging factor of 51, TR of 8382 ms, and field of view of 192x192 mm. Infants from
London were scanned on a Philips 3 Tesla system (Philips Medical Systems, Best, The
Netherlands) using an 8-channel phased array head coil. T2-weighted fast-spin echo MRI was
acquired using TR 8670 ms; TE 160 ms; slice thickness 1 mm and T1-weighted imaging using
TR 17 ms, TE 4.6 ms, slice thickness 0.8 mm. Single shot EPI DTI was acquired in the transverse
plane in 32 non-collinear directions using the following parameters: TR 8000 msec; TE 49
msec; slice thickness 2 mm; field of view 224 mm; matrix 128×128 (voxel size 1.75×1.75×2
mm); b-value 750 s/mm2; SENSE factor of 2.
Infants in Utrecht were sedated for the first two scans, using oral chloral hydrate 50 to 60 mg/
kg, and in London only for the TEA scan (30-55 mg/kg), according to the local clinical protocol.
A neonatologist was present throughout all examinations. In infancy, children in Utrecht
received general anaesthesia and were monitored by an anaesthetist, while those in London
were sedated with oral chloral hydrate (50-80 mg/kg) and monitored by an experienced
paediatrician.
For all MRIs obtained around TEA, the presence and quality of myelination in the PLIC were
scored retrospectively (FC, LdV) as either normal or abnormal/absent, using both T1- and T2-
weighted sequences. Scoring was performed blinded to outcome data.
Po s t -p r oc e ssi ng o f vo l u m et r ic data
T2-weighted scans were segmented using the neonatal specific segmentation method of
Makropoulos and colleagues.16 This method utilizes an expectation-maximization scheme that
combines manually labelled atlases17 with intensity information from the image to be
segmented and has shown reliable results among infants scanned at post-menstrual ages
ranging between 28 and 44 weeks.16,18 All segmentations were manually checked and small
corrections were performed if necessary. Bilateral thalamic volumes and total brain volume
(TBV) were extracted for all scans.
Po s t -p r oc e ssi ng o f DT I data
DTI data were analysed using the FMRIB Software Library (FSL).19 Probabilistic tractography of
the bilateral CST was performed as described previously.20 A seed mask (early scan 17 voxels,
TEA scan 20 voxels, late scan 24 voxels) was drawn in the cerebral peduncle, with waypoint
masks in the PLIC (early scan 17 voxels, TEA 21 voxels, late scan 24 voxels) and around the
central sulcus. Connectivity distributions were generated from every voxel in the seed masks,
and only those paths that went through the waypoint masks were retained. The tracts were
normalized by the number of samples going from the seed mask through the waypoints.21
These connectivity distributions were then thresholded at 1%.
Statistical procedures were performed using R version 2.15.3.22 FA values and thalamic
ch a pt er 7
volumes from infants with c-PVL and controls, as well as the relation between FA and
volumetric measurements and both PLIC scoring (defined as normal or abnormal) and
outcome for the infants with c-PVL, were compared by multiple linear regression modelling.
PMA at time of scan and the interaction between the FA value or volume and PMA were
included in the model. Since FA values can differ per DTI-protocol, only infants and controls
scanned with the same protocol were included in the statistical analysis, and data from
Utrecht and London were analysed separately. 95% Confidence intervals for the figures were
calculated using GraphPad Prism version 6.05 software (GraphPad Software Inc., La Jolla, CA,
USA). Left-right differences were tested with paired-samples t-tests. The relation between the
PLIC scoring and outcome was calculated with a Fisher’s exact test.
Results
Thirty-three infants, 26 from Utrecht and 7 from London, with c-PVL were eligible for inclusion.
The majority of patients had grade III c-PVL (23 from Utrecht, 5 from London).14 Two patients
from Utrecht and two from London had grade II c-PVL and one patient from Utrecht had
widespread pre-cystic lesions but died before these could evolve into cysts. Lesion appearance
was largely symmetrical. Clinical data on the included infants are presented in table 1. An
overview of the number of available scans, and the extent of serial imaging data, is shown in
table 2.
S u r v i va l a nd m oto r o u tco m e
Three of the 33 infants with c-PVL died during the neonatal period. Four of the infants did not
develop CP, and three have an evolving motor disorder but are too young to be reliably
classified with the GMFCS. The remaining 23 infants all developed CP, which was classified as
level I in four infants, level II in two infants, level III in ten infants, level IV in 5 infants and level V
in two infants. Seven infants with c-PVL developed epilepsy: five with CP level IV-V, one without
CP but with severe visual impairment, and one with an evolving motor disorder. None of the
control infants developed CP or epilepsy.
Ta bl e 1. Pat ie n t characteri sti cs
Characteristic Infants with c-PVL (n = 33) Control infants (n=31) p-value
Gestational age (wks,

30.1 [24.7 – 35.4] 27.9 [25.6 – 31.0] <0.001
median [range])
Sex (m/f) 24/9 19/12 0.27
PMA early scan (wks, mean

33.7 [29.3 – 38.0] 31.3 [30.3 – 34.7] 0.01
[range])
cha p te r 7
PMA TEA scan(wks, median

41.1 [38.0 – 43.7] 41.4 [39.3 – 43.3] 0.50
[range])
PMA late scan (months,

31.4 [21.2 – 76.2] 26.2 [26.2 – 34.2] 0.38
median [range])
Visu a l a n a ly s i s o f t h e m y el i nat i on o f t he P LI C o n s cans at TEA

Of the 33 infants with c-PVL, T1- and T2-weighted imaging at TEA was available in 29. The other
four infants were either deceased (n=3), or did not have imaging around TEA (n=1). Of those
29, 8 had myelination of the PLIC that appeared normal and in 21 infants, the PLIC myelin
Ta bl e 2 . Ava ila b le MRI data
Utrecht London
Cases Controls Cases Controls

(n=26) (n=21) (n=7) (n=10)
DTI
Early 12 10 3 4
TEA 19 18 6 10
Infancy 7 0 4 3
Early + TEA 7 7 1 2
Early + Infancy 1 0 0 0
TEA + Infancy 5 0 2 1
Early + TEA + Infancy 1 0 2 2
Volumes
Early 14 10 3 4
TEA 21 18 7 10
Early + TEA 9 7 3 4
appeared abnormal or was not seen. All control infants had myelin in the PLICs that appeared
normal.
GMFCS grading was available for 26 of these 29 case infants and three, all with abnormally
ch a pt er 7
appearing PLICs, were too young for accurate CP classification using the GMFCS. When
correlated to outcome at 15 months corrected age and beyond, 7/8 infants with normally
appearing PLICs and 3/18 with abnormally appearing PLICs had a favourable outcome (no CP
or GMFCS I-II). In contrast, 1/8 infants with normally appearing PLICs and 15/18 infants with
abnormal PLICs had an unfavourable outcome (GMFCS III-V), leading to an Odds ratio of 35
(95% confidence interval: 3.1 – 399.4).
No clear correlation between the appearance of the PLIC and FA measurements of the CST at
TEA could be found. A significant correlation between thalamic volumes at TEA and the
appearance of the PLICs was found for both the absolute volumes (left p=0.03, right p=0.01)
and the thalamic volume/TBV ratio (left p=0.046, right p=0.02).
Figure 1 FA values of the left corticospinal tract. This figure
shows the mean FA values of the left corticospinal tract for
infants with c-PVL and control infants, with dotted lines
representing the 95% confidence intervals of the cases and
controls scanned with the same protocol. Since different DTI
protocols were used, data are shown separately for the
infants from Utrecht (A) and London (B). Additionally, the
infants with c-PVL from Utrecht are separated per imaging
protocol. Although the absolute FA values differ, the same
pattern with lower FA values can be observed for the infants
with c-PVL across the different protocols. FA values of the
infants with c-PVL scanned again during infancy show that
values are predominantly below 0.5 (C). Here, the open
figures and dotted lines represent the infants with no or
ambulatory CP and the solid figures and lines represent the
infants with severe CP.
cha p te r 7
Cor t icospi na l t r acto g r aph y

Mean FA values at the different PMAs from the left CST are depicted in figure 1A-C. No
significant left-right differences were found. Since different DTI protocols were used, analyses
were performed separately on the London and Utrecht cohorts, and also only between
patients and controls tested with the same DTI protocol. However, as figure 1A shows, a similar
pattern could be seen for the infants scanned with the different protocols. A significant
difference was found between patients and controls for both left and right CST when early and
TEA scans were combined (London: left p=7.39*10-4, right p=0.011, Utrecht: left p=6.84*10-10,
right p=8.02*10-7). For the left CST of the Utrecht data, the model including the interaction
between PMA and the factor case/control was also significant (case/control p=0.2, interaction
p=0.03). When separated according to time of scanning, these differences persisted except for
the right CST of the Utrecht data at the early scan (early scan London: left p=0.03, right
p=0.03, Utrecht: left p=0.01, right p=0.16 and TEA scan London: left p=6.8*10-7, right
p=9.19*10-4, Utrecht: left p=4.3*10-8, right p=8.6*10-6).
In the infants with serial data, an increase in FA was seen for both cases and controls with
increasing PMA, similar to the cross-sectional data in figure 1A-B. FA increased further after TEA,
as shown in the infants with a repeat scan in infancy (figure 1C). Supplementary figure 1A and 1B
show the correlation between FA values at TEA and the severity of CP for each centre. No
significant differences could be found, possibly due to the relatively small sample size.
Th al a mic vol u m e s
Figure 2 shows the thalamic volumes for the left hemisphere, as well as the thalamic volumes
corrected for TBV. No significant left-right differences or differences between infants from
Utrecht and London were found. Therefore, data from Utrecht and London were combined for
the analysis. When volumes across both early and TEA scans were combined, a significant
difference was found between cases and controls for both the left and right thalamus, with the
Figure 2 Thalamic volumes. This figure shows the

difference in absolute (A) and relative (B) thalamic
volumes between infants with c-PVL and control
infants. Both centres are shown together. The
dotted lines represent the 95% confidence intervals.
At 30 weeks post-menstrual age, the difference
between both groups was not yet significant,
whereas at 40 weeks, a clear difference could be
observed.
ch a pt er 7
interaction between PMA and the factor case/control also being significant (left case/control
p=1.3*10-5 and interaction p=7.6*10-8, right p=3.7*10-5 and p=2.8*10-7). When separated
according to time of scanning, the early data (PMA<34.8 weeks, since no controls were
scanned between 35-38 weeks) no longer showed a significant difference. At TEA, the
difference between infants with c-PVL and controls was highly significant, while no interaction
with PMA was found (left p=1.08*10-13 and right p=2.7*10-13). After correction for TBV, the
combined data again showed a correlation between the thalamus and both the difference
between infants with c-PVL and controls as well as the interaction with PMA (left p=0.02 and
p=4.7*10-4, right p=0.05 and p=0.001). When the data were separated according to time of
scan, the early data already showed significant differences (left p=0.01 and p=0.009, right
p=0.03 and p=0.02).
Highly significant differences were found at TEA between cases and controls, with the
interaction no longer being significant (left p=1.5*10-15, right p=<2.0*10-16). For infants with
serial data (n = 23, 12 cases, 11 controls), a significant difference in increase was seen for both
the left and right thalamus between patients with c-PVL and controls (left p=0.002, right
p=5.14*10-4). For some infants with c-PVL, a decrease in absolute volumes was noted. In
Supplementary figure 1C and 1D, the correlation between the thalamic volumes and the
severity of CP is shown. Infants with severe CP (GMFCS level III-V) had significantly smaller
thalamic volumes compared to those with no or mild CP (absolute volumes left p=0.02, right
p=0.03; relative volumes left p=0.02, right p=0.04). Infants with c-PVL that developed epilepsy
had significantly smaller relative thalamic volumes (left p=0.02, right p=0.04) compared to
those that did not.
Discussion
In this study, we have identified differences in both FA values of the CST and thalamic volumes
in the neonatal period between preterm infants with c-PVL and preterm control infants. In the
CST, a significant difference in FA values was observed prior to TEA, which persisted at TEA and
cha p te r 7
in infancy. The differences in absolute thalamic volumes between c-PVL and control groups
were not significant in the early neonatal period but became highly significant at TEA,
suggesting an ongoing process of disturbed brain development.
Previous studies have shown the CST to be affected in children with CP during childhood and
adolescence.6-8,23 This held true for both children with c-PVL and for children with CP due to
other causes, such as hypoxic-ischaemic encephalopathy, suggesting that the damage to the
CST can be caused by different mechanisms. The current study confirms the effect of extensive
white matter damage on the CST at a very early stage. This effect can be found soon after the
insult, as the difference in FA values was already present on the first scan, obtained within the
first weeks after the insult in most patients. This suggests acute damage to the white matter as
a result of the ischaemic and inflammatory processes taking place in c-PVL.24 This damage to
the white matter persisted, as shown in the lower FA values at TEA and in infancy. The lasting
effect of damage to the CST is likely to result in impaired gross motor outcome in most infants.
Indeed, our data in infancy suggest that maturation of the CST does not completely stop in
infants with c-PVL, but continues at a reduced rate and will not reach the values found in
healthy infants of the same age. Murakami and colleagues have previously shown that a FA
value of the CST below 0.5 predicts the development of CP in children ranging from 9 to 41
months of age.25 All infants with c-PVL who were imaged in infancy or early childhood in this
study developed CP and, except for one infant with a mild CP (GMFCS level I), FA values of the
CST were indeed below 0.5.
In contrast to the CST, absolute thalamic volumes were not significantly different between
cases and controls on the early MRI obtained prior to TEA. They were however highly
significant on the TEA scan, both in absolute volumes as well as after correction for overall
brain size. This suggests a more remote effect of injury to the thalamus, which develops over
time. The white matter damage in c-PVL may cause secondary trophic effects in the thalamus,
thereby impairing normal development of the thalamocortical connections that are being
formed during the last trimester of pregnancy.24 Our findings are in agreement with
histological studies that have shown extensive damage to the thalamus in infants with c-PVL,
26,27
and a functional MRI study showing diminished thalamic connectivity to the caudate
nucleus, cingulate cortex and cerebellum.6 Smaller thalamic volumes in infants with c-PVL
have been found before on ultrasound 21 days after birth and on MRI performed in infancy
and these seem to relate to both motor and cognitive outcome.10-12,28,29 The decrease in the
thalamus/TBV ratio between the early and TEA scan in the infants with c-PVL in this study, and
for some infants even a decrease in absolute thalamus volumes, suggest that the remote
effects develop over time as a result of impaired input to the thalamus due to extensive white
matter damage.24 Thalamic volumes were related to gross motor outcome and the
development of epilepsy, as those infants that developed severe CP (GMFCS level III-V) and/or
epilepsy had smaller thalamic volumes at TEA.
Brain injury caused by c-PVL is extensive and involves more structures than just the CST and
ch a pt er 7
the thalamus. Many other brain structures have previously been shown to be affected, such as
the corpus callosum, optic radiation, thalamic radiation and sensory tracts.23,30-32 The
combination of deficits seen in survivors of c-PVL is therefore likely caused by a global pattern
of injury. The fact that thalamic volumes, but not FA values of the CST predicted the severity of
CP in the current study suggests that the extent of the lesions, likely reflected in the thalamic
volumes, is an important contributor to gross motor outcome. However, not all infants with
c-PVL will develop motor impairment and as shown in this study, this could be predicted by
the appearance of myelination of the PLIC at TEA, similar to previous reports.4,33
Most infants with c-PVL were born after a gestation of more than 28 weeks, with a median
gestational age of 30 weeks. Rather than extreme prematurity, factors such as infection and
inflammation seem the most likely cause of the extensive brain damage in these infants.34,35
The majority of controls in the present study however were extremely preterm infants, with a
lower gestational age than the infants with c-PVL, as only infants with a gestational age below
28 weeks had routine MRI at TEA. It is known that both FA values and volumetric
measurements can be affected by preterm birth. When compared to healthy, term born
infants, preterm infants at TEA generally show lower FA values and smaller volumes, although
this is influenced by the severity of illness during the neonatal period.36-39 The difference
between infants with c-PVL and healthy, term born controls is therefore likely to be even larger
than shown in this study. Due to ethical regulations, we were unable to scan a sufficient
number of healthy control infants, specifically from Utrecht. We do however feel, that by
selecting only those infants without overt brain injury and with a normal motor outcome,
controls represent infants with a likely normal, or approaching normal, brain development.
There were several limitations to this study. First, since this was a retrospective study using
data acquired over a number of years, different DTI protocols were used. FA values were
therefore not comparable between all cases and controls, although the pattern of FA decrease
remained. Secondly, as described above, all control infants were born prematurely and this
may have affected their brain development as well. However, if anything, their FA
measurements would have been lower compared to term born infants and the results of this
study may thus underestimate but not overestimate the differences between infants with
c-PVL and controls. Third, some of the studied infants with c-PVL were still very young and the
severity of their CP could not yet be determined, or may change when they grow older.
However, a previous study showed that the majority of infants with c-PVL grade III will develop
severe CP (GMFCS level III-IV) and will remain stable during childhood.40
In conclusion, this study showed that damage to the CST in infants with c-PVL is already
detectable on early MRI scans, well before TEA, whereas the damage to the thalamus develops
in the weeks between birth and TEA. This may reflect the difference between acute and remote
effects of extensive injury to the white matter and these findings suggest that measuring the
remote effects at TEA can is recommended in order to assess any effects from future medical
interventions initiated after the diagnosis of c-PVL. More long-term outcome after advanced
cha p te r 7
neonatal imaging may further elucidate the underlying mechanisms causing

neurodevelopmental impairment in these infants.
Acknowledgements
SJC, AC and AM acknowledge financial support from: the Department of Health via the
National Institute for Health Research (NIHR) Comprehensive Biomedical Research Centre
award to Guys & St Thomas NHS Foundation Trust in partnership with Kings College London
and Kings College Hospital NHS Foundation Trust; the Medical Research Council (UK).
Supplementary figure 1. Relation between FA values, thalamic volumes and gross motor outcome in the infants with c-PVL. In
this figure FA values (A Utrecht, B London), thalamic volumes (C) and thalamic/total brain volume ratios (D) of the infants with
c-PVL are correlated to gross motor outcome as measured with the GMFCS. Open symbols are used for those infants without CP
or with ambulatory CP, and closed symbols for those with non-ambulatory CP. For the DTI data, only infants with the 32
directions DTI protocols are shown. For thalamic volumes both centres are shown together.
ch a pt er 7
References
1 Hamrick SE, Miller SP, Leonard C, Glidden DV, white-matter integrity and cognitive outcome at
Goldstein R, Ramaswamy V et al. Trends in severe school age. Pediatr Res 2012;71:354-60.
brain injury and neurodevelopmental outcome 12 Lin Y, Okumura A, Hayakawa F, Kato K, Kuno T,
in premature newborn infants: the role of cystic Watanabe K. Quantitative evaluation of thalami
periventricular leukomalacia. J Pediatr and basal ganglia in infants with periventricular
2004;145:593-9. leukomalacia. Dev Med Child Neurol
2 van Haastert I, Groenendaal F, Uiterwaal CS, 2001;43:481-5.
Termote JU, van der Heide-Jalving, Eijsermans MJ 13 Zubiaurre-Elorza L, Soria-Pastor S, Junque C,
et al. Decreasing incidence and severity of Segarra D, Bargallo N, Mayolas N et al. Gray
cerebral palsy in prematurely born children. J matter volume decrements in preterm children
Pediatr 2011;159:86-91. with periventricular leukomalacia. Pediatr Res
3 Van Hus JW, Jeukens-Visser M, Koldewijn K, 2011;69:554-60.
Geldof CJ, Kok JH, Nollet F et al. Sustained 14 de Vries LS, Eken P, Dubowitz LM. The spectrum
Developmental Effects of the Infant Behavioral of leukomalacia using cranial ultrasound. Behav
Assessment and Intervention Program in Very Brain Res 1992;49:1-6.
Low Birth Weight Infants at 5.5 Years Corrected 15 Palisano R, Rosenbaum P, Walter S, Russell D,
Age. J Pediatr 2013;162:1112-9. Wood E, Galuppi B. Development and reliability
4 de Vries LS, van Haastert IC, Benders MJ, of a system to classify gross motor function in
Groenendaal F. Myth: cerebral palsy cannot be children with cerebral palsy. Dev Med Child
predicted by neonatal brain imaging. Semin Fetal Neurol 1997;39:214-23.
Neonatal Med 2011;16:279-87. 16 Makropoulos A, Gousias I, Ledig C, Aljabar P,
5 Huppi PS, Dubois J. Diffusion tensor imaging of Serag A, Hajnal J et al. Automatic Whole Brain
brain development. Semin Fetal Neonatal Med MRI Segmentation of the Developing Neonatal
2006;11:489-97. Brain. IEEE Trans Med Imaging 2014;PP:Epub
6 Lee JD, Park HJ, Park ES, Oh MK, Park B, Rha DW et ahead of print.
al. Motor pathway injury in patients with 17 Gousias IS, Edwards AD, Rutherford MA, Counsell
periventricular leucomalacia and spastic SJ, Hajnal JV, Rueckert D et al. Magnetic
diplegia. Brain 2011;134:1199-210. resonance imaging of the newborn brain:
7 Rha DW, Chang WH, Kim J, Sim EG, Park ES. manual segmentation of labelled atlases in term-
Comparing quantitative tractography metrics of born and preterm infants. Neuroimage
motor and sensory pathways in children with 2012;62:1499-509.
periventricular leukomalacia and different levels 18 Išgum I, Benders MJNL, Avants B, Cardoso MJ,
of gross motor function. Neuroradiology Counsell SJ, Fischi Gomez E et al. Evaluation of
2012;54:615-21. automatic neonatal brain segmentation
8 Yoshida S, Hayakawa K, Yamamoto A, Okano S, algorithms: the NeoBrainS12 challenge. Medical
Kanda T, Yamori Y et al. Quantitative diffusion Image Analysis 2014;Accepted for publication.
tensor tractography of the motor and sensory 19 Jenkinson M, Beckmann CF, Behrens TE, Woolrich
cha p te r 7
tract in children with cerebral palsy. Dev Med MW, Smith SM. FSL. Neuroimage 2012;62:782-
Child Neurol 2010;52:935-40. 90.
9 Miller SP, Vigneron DB, Henry RG, Bohland MA, 20 Bassi L, Chew A, Merchant N, Ball G, Ramenghi L,
Ceppi-Cozzio C, Hoffman C et al. Serial Boardman J et al. Diffusion tensor imaging in
quantitative diffusion tensor MRI of the preterm infants with punctate white matter
premature brain: development in newborns with lesions. Pediatr Res 2011;69:561-6.
and without injury. J Magn Reson Imaging 21 Johansen-Berg HB, Behrens TEJ. Diffusion MRI:
2002;16:621-32. From Quantitative Measurement to In-Vivo
10 Nagasunder AC, Kinney HC, Bluml S, Tavare CJ, Neuroanatomy. London: Elsevier. 2009;333-353.
Rosser T, Gilles FH et al. Abnormal 22 R: A Language and Environment for Statistical
microstructure of the atrophic thalamus in Computing [computer program]. Vienna,
preterm survivors with periventricular Austria: 2012.
leukomalacia. AJNR Am J Neuroradiol 23 Trivedi R, Agarwal S, Shah V, Goyel P, Paliwal VK,
2011;32:185-91. Rathore RK et al. Correlation of quantitative
11 Zubiaurre-Elorza L, Soria-Pastor S, Junque C, sensorimotor tractography with clinical grade of
Fernandez-Espejo D, Segarra D, Bargallo N et al. cerebral palsy. Neuroradiology 2010;52:759-65.
Thalamic changes in a preterm sample with 24 Volpe JJ. Brain injury in premature infants: a
periventricular leukomalacia: correlation with complex amalgam of destructive and
developmental disturbances. Lancet Neurol 33 Mirmiran M, Barnes PD, Keller K, Constantinou
2009;8:110-24. JC, Fleisher BE, Hintz SR et al. Neonatal brain
25 Murakami A, Morimoto M, Yamada K, Kizu O, magnetic resonance imaging before discharge is
Nishimura A, Nishimura T et al. Fiber-tracking better than serial cranial ultrasound in
techniques can predict the degree of neurologic predicting cerebral palsy in very low birth weight
impairment for periventricular leukomalacia. preterm infants. Pediatrics 2004;114:992-8.
Pediatrics 2008;122:500-6. 34 Andre P, Thebaud B, Delavaucoupet J, Zupan V,
26 Ligam P, Haynes RL, Folkerth RD, Liu L, Yang M, Blanc N, d’Allest AM et al. Late-onset cystic
Volpe JJ et al. Thalamic damage in periventricular periventricular leukomalacia in premature
leukomalacia: novel pathologic observations infants: a threat until term. Am J Perinatol
relevant to cognitive deficits in survivors of 2001;18:79-86.
prematurity. Pediatr Res 2009;65:524-9. 35 Verboon-Maciolek MA, Truttmann AC,
27 Pierson CR, Folkerth RD, Billiards SS, Groenendaal F, Skranes J, Dollner H, Hunt RW et
Trachtenberg FL, Drinkwater ME, Volpe JJ et al. al. Development of cystic periventricular
Gray matter injury associated with leukomalacia in newborn infants after rotavirus
periventricular leukomalacia in the premature infection. J Pediatr 2012;160:165-8.
infant. Acta Neuropathol 2007;114:619-31. 36 Anjari M, Srinivasan L, Allsop JM, Hajnal JV,
28 Yokochi K. Thalamic lesions revealed by MR Rutherford MA, Edwards AD et al. Diffusion
associated with periventricular leukomalacia and tensor imaging with tract-based spatial statistics
clinical profiles of subjects. Acta Paediatr reveals local white matter abnormalities in
1997;86:493-6. preterm infants. Neuroimage 2007;35:1021-7.
29 Fukuda S, Yokoi K, Suzuki S, Goto H. Serial 37 Boardman JP, Counsell SJ, Rueckert D, Hajnal JV,
ultrasonographic observation of bilateral thalami Bhatia KK, Srinivasan L et al. Early growth in brain
in low birth weight infants with periventricular volume is preserved in the majority of preterm
leukomalacia. Brain Dev 2011;33:394-9. infants. Ann Neurol 2007;62:185-92.
30 Hoon AH, Jr., Stashinko EE, Nagae LM, Lin DD, 38 Keunen K, Kersbergen KJ, Groenendaal F, Isgum I,
Keller J, Bastian A et al. Sensory and motor de Vries LS, Benders MJ. Brain tissue volumes in
deficits in children with cerebral palsy born preterm infants: prematurity, perinatal risk
preterm correlate with diffusion tensor imaging factors and neurodevelopmental outcome: a
abnormalities in thalamocortical pathways. Dev systematic review. J Matern Fetal Neonatal Med
Med Child Neurol 2009;51:697-704. 2012;25 Suppl 1:89-100.
31 Koerte I, Pelavin P, Kirmess B, Fuchs T, Berweck S, 39 Ball G, Counsell SJ, Anjari M, Merchant N, Arichi T,
Laubender RP et al. Anisotropy of transcallosal Doria V et al. An optimised tract-based spatial
motor fibres indicates functional impairment in statistics protocol for neonates: applications to
children with periventricular leukomalacia. Dev prematurity and chronic lung disease.
Med Child Neurol 2011;53:179-86. Neuroimage 2010;53:94-102.
32 Nagae LM, Hoon AH, Jr., Stashinko E, Lin D, Zhang 40 van Haastert IC, de Vries LS, Eijsermans MJ,
ch a pt er 7
W, Levey E et al. Diffusion tensor imaging in Jongmans MJ, Helders PJ, Gorter JW. Gross motor
children with periventricular leukomalacia: functional abilities in preterm-born children
variability of injuries to white matter tracts. AJNR with cerebral palsy due to periventricular
Am J Neuroradiol 2007;28:1213-22. leukomalacia. Dev Med Child Neurol
2008;50:684-9.
ch a p t er
8
The neonatal
connectome
during
preterm
brain
development
Martijn P. van den Heuvel

Marcel A. de Reus
Kristin Keunen
René S. Kahn
Floris Groenendaal
Linda S. de Vries
Cerebral cortex 2014 pii: bhu095 157

Abstract
The human connectome is the result of an elaborate developmental trajectory. Acquiring
diffusion-weighted imaging and resting-state fMRI, we studied connectome formation
during the preterm phase of macroscopic connectome genesis. In total, 27 neonates were
scanned at week 30 and/or week 40 gestational age (GA). Examining the architecture of the
neonatal anatomical brain network revealed a clear presence of a small-world modular
organization before term birth. Analysis of neonatal functional connectivity (FC) showed
the early formation of resting-state networks, suggesting that functional networks are
present in the preterm brain, albeit being in an immature state. Moreover, structural and
FC patterns of the neonatal brain network showed strong overlap with connectome
architecture of the adult brain (85 and 81%, respectively). Analysis of brain development
between week 30 and week 40 GA revealed clear developmental effects in neonatal
connectome architecture, including a significant increase in white matter microstructure
(P < 0.01), small-world topology (P < 0.01) and interhemispheric FC (P < 0.01).
Computational analysis further showed that developmental changes involved an increase
in integration capacity of the connectivity network as a whole. Taken together, we
conclude that hallmark organizational structures of the human connectome are present
before term birth and subject to early development.
chapter 8
Introduction
T
he macroscopic brain comprises a large number of anatomically and functionally
distinct regions, linked by a complex web of structural white matter pathways, known as
the human connectome.1 Adopting network science as a general mathematical
framework to visualize and examine the connectome’s complex network structure, recent
structural and functional studies have reported several hallmark features of adult connectome
organization, including the presence of a small-world modular organization with functionally
coupled communities, short communication pathways, and the formation of central
connectivity hubs facilitating efficient global communication.2-4 During what developmental
period these hallmark features arise remains however unknown. The adult cerebral brain
network is the result of a complex developmental trajectory. From the prenatal formation of
the first neurons, throughout the first years of life and all the way into late adolescents, the
brain undergoes an elaborate developmental trajectory. Its developmental period involves an
initial exuberant increase in the number of network parts, being neurons, axons, and
dendrites, as well as an elaborate growth and transformation of its large-scale wiring pattern.
Studies have suggested that the first “predecessor” neurons appear from neuroepithelial cells
as early as around the 5th week after conception (see for a detailed review on the
development of the embryonic cortex Bystron et al.5). During the following period, the vast
majority of 10 billion cortical neurons are being formed, migrate to their final position, and
start to grow their axonal and dendritic projections from which they connect to other
neurons, with a steady, but relative slow growth of synaptic connections.6-9 It is during the last
weeks of gestation (week >24)10,11 that a period of exuberant synapse growth starts, with a
synaptic “big bang” occurring around birth,8 indicating the start of a period of exuberant
formation of synaptical connections, with a peak of synaptic connections occurring around
year 2 of life.7,8,10-13 This period around birth thus involves a sudden increase in total physical
connections and interactions between neurons of the cerebral cortex, and thus involves an
important period in cortical connectome genesis.14
The far majority of these neural interactions and synaptic connections take place between
neurons on the microscopic scale, with axonal projections linking neurons by means of
intracortical and short-range white matter axonal projections.15,16 However, a relatively small
fraction involves long-range connections involved in neural communication processes
ch a pt e r 8
between distant parts of the brain.15,16 It is the early development of these long-distance
cortico-cortical connections, to some extent accessible on the macroscopic scale to
noninvasive neuroimaging techniques, that is the primary focus of our study. Structural
neuroimaging studies have shown that as early as 2 years, and likely even earlier, the vast
majority of these long-range cortico-cortical projections are present.17-20 In parallel, functional
studies examining the functional dynamics and functional interactions between brain regions
of the young brain have suggested that most functional networks are present in the infant
the neonatal connectome during preterm brain development 159

brain,21-23 this with later development of the child and adolescent brain mostly involving the
transformation of functional connectivity (FC) from a segregated, clustered structure early in
life, to a more global and integrated system during adult life.24-27 These neuroimaging studies
have elucidated network changes in the postnatal young and adolescent brain, but important
open questions about the development of the brain’s macroscopic wiring circuitry still remain,
in particular about developmental processes during and after the first phases of connectome
genesis.
Here, we aimed to study the macroscale connectivity network of the neonatal brain during
week 30 and week 40 of gestational age (GA) in preterm born neonates. Using diffusion-
weighted and resting-state fMRI data to provide an estimate of the macro-scale connections of
the brain, we examined the spatial layout of the structural and FC patterns of the neonatal
brain during the early phases of preterm macroscopic connectome development.
Materials and methods

Pa r t icipan t s
Twenty-seven preterm infants born with a GA of around (on average) 27 weeks were scanned
between week 30 and week 42 of corrected GA (see Table 1 for demographics). Participants
underwent a 35-min scanning session in a 3-Tesla Philips Achieva system (Philips Medical
systems, Best, The Netherlands). A scanning session included the acquisition of an anatomical
scan (T1 weighted), a diffusion-weighted imaging (DWI) scan, and/or a resting-state functional
MRI scan (rs-fMRI). All neonates were sedated during imaging (see below). DWI data were
acquired of in total 23 neonates; in 21 neonates rs-fMRI was acquired (see Table 1). Of 7
neonates, longitudinal DWI scan (i.e., 1 scan at 30 weeks and 1 scan at 40 weeks) was available.
Of 7 neonates, longitudinal resting-state fMRI data were available. Of 21 datasets, both resting-
state fMRI and diffusion imaging was available, allowing for an examination of a structure–
function relationship. For the cross-sectional analysis, all datasets were included into one
structural dataset to maximize statistical power, as regularly performed in infant imaging
studies.28-30 All data were acquired as part of standard clinical care. For this study, all scans were
evaluated by a team of neonatologists and radiologists with extensive experience in neonatal
brain imaging. MRI imaging was acquired as part of standard clinical care to examine for
potential gray and white matter damage, graded according to the injury scores of Woodward
et al.31 Only neonates that scored normal/mild on the Woodward scale were included in our
c ha pt er 8
study, to reflect normal processes of healthy brain development as much as possible. Clinical
description and the availability of DWI/ rs-fMRI data of the included neonates are presented in
Table 1.
Scan P r e pa r at i o n a nd S e dat i o n
Prior to MR scanning all infants were sedated; chloral hydrate was administered orally (50–60
mg/kg). To minimize motion during scanning, all neonates were immobilized by wrapping
them into a vacuum cushion. MiniMuffs (Natus Europe, Münich, Germany) and earmuffs (EM’s
4 kids, Everton Park, Australia) were used to reduce noise and for further stabilization.
Throughout the examination, the following parameters were monitored; heart rate,
respiratory rate, and transcutaneous oxygen satur ation. A neonatologist was present at all
times during the examination.
An atomica l T 1-Weigh t e d Im age

Acqu isitio n
A T1-weighed image (3D Fast Field Echo using parallel imaging; TR/TE 9 ms/4.6 ms; field of view
(FOV) 200 × 200 mm, 110 slices, 0.78 × 0.78 × 1.2 mm voxelsize, 30 weeks: TR/TE 9.4 ms/4.6
ms, 0.94 × 0.94 × 2.0 mm voxelsize) was acquired for anatomical reference.
I mag e pro c essin g

A neonate anatomical template was registered to the neonatal T1 image, and the 56 cortical
regions of the template were selected as regions of interest.32 A detailed description of this
neonatal template, including information and illustrations on the 56 regions and an open
access copy of the template, is presented in the paper of Oishi et al.32
Diffu si on W e ig h t e d Im agi ng
Acqu isitio n
A DWI set of 32 weighted diffusion scans (b = 800 s/mm2) and an unweighted B0 scan (b = 0 s/
mm2) were acquired (DWI-MR using parallel imaging SENSE P-reduction 2; TR = 5764 ms, TE =
70 ms, FOV 180 × 146.25 mm, voxel size 1.4 × 14 × 2.0 mm, 50 slices).
I mag e pro c essin g

Processing included the following steps: DWI images were 1) corrected for eddy-current
distortions and small-head movements34 and realigned to the b=0 image. 2) The diffusion
profile of each voxel was fitted a tensor using a robust tensor fitting method,35 and 3) the
principal eigenvector of the eigenvalue decomposition of the fitted tensor was taken as the
main diffusion direction within each voxel. 4) Using the results of the eigenvector
decomposition, several metrics of microstructural organization of white matter were
computed,36,37 including the level of fractional anisotropy (FA), a metric of microstructural
ch a pt e r 8
organization and often used as an estimate of the level of white matter integrity, the level of
mean diffusion (MD), an estimate of the total level of diffusion within a voxel, the level of
transverse diffusion (TD), reflecting the amount of restricted diffusion in the perpendicular
direction of the first eigenvector and, in the adult brain, often interpreted as an (inverse)
estimate of the level of myelination of white matter, and the level of parallel diffusion (PD), a
metric of the level of local bra orientation.

Neon Gender GA PMA PMA DWI RS Twin BW IUGR
30 40 30 40 30 40
1 f 27.9 41.7 X X 780

chapter 8
Ta b le 1 . Demographi cs
Imaging data of 36 neonatal datasets from 27 neonates with MRI around week 30 and week 40 postmenstrual age (PMA)
were included; serial measurements at both 30 and 40 weeks PMA were available for 9 neonates. Neon, neonate; GA,
gestational age at birth; PMA, postmenstrual age (weeks) at time of scan; DWI, diffusion-weighted imaging data were
acquired and of sufficient quality; RS, resting-state data were acquired and of sufficient quality; Twin, part of a twin; BW,
body weight at birth (in grams); IUGR, intrauterine growth restriction defined as birth weight below p10; HYP,
HYP PDA BPD Sepsis IVH GMI WMI Other lesions
chapter 8
hypotension requiring inotropes and/or vasopressors; PDA, patent ductus arteriosus requiring therapy; BPD,
bronchopulmonary dysplasia (defined as oxygen dependency at 36 weeks PMA); sepsis, blood culture proven
sepsis; IVH, intraventricular haemorrhage (scored according to Papile et al.33); WMI, white matter injury; GMI,
gray matter injury; mild, mild injury; PWML, punctate white matter lesions; CBH, punctate cerebellar
haemorrhages; No, number of patients; SD, standard deviation; X, yes; F, female; M, male. GMI and WMI were
graded according to the white and gray matter injury scores of Woodward et al.31

R ec o nstru cti o n o f W hit e M att e r Tracts
Streamline tractography based on the Fiber Assignment by Continuous Tracking algorithm
(FACT) was used to reconstruct the white matter pathways of the neonatal brain.38 To this end,
within each voxel eight seeds were started, and the main diffusion direction was followed from
voxel to voxel, until one of the stopping criteria was met, being when the streamline reached a
voxel with a low FA value (<0.1), when a streamline left the brain mask or when a
reconstructed streamline made a sharp turn (>60°).
R e st i n g -S tat e f MRI
Ac q uisiti o n
Measurements of endogenous brain activity of the neonatal brain were acquired through
means of the acquisition of 6.71 min (256 volumes) of resting-state fMRI (parameters: 2D EPI
SENSE, TR/TE 1600/45 ms, flip-angle 50; SENSE reduction 2, 2.5 mm isotropic voxel size, 22
consecutive slices covering the whole neonate cortex).
I mag e Pr o c e ssin g
Imaging processing steps were performed using SPM8 (www.fil.ion.ucl.ac.uk/spm/software/
spm8). Time-series were realigned and co-registered with the T1 image to ensure overlap with
the cortical region of interest (ROI) template. Voxel-wise time-series were 1) de-trended
(removing linear trends and first order drifts), 2) corrected for global effects (regressing out
white matter, ventricle, and global mean signals). To minimize potential influences of motion-
induced spurious effects into FC the effects of motion (as assessed through means of the 6
motion parameters resulting from image realignment) were also regressed out of the resting-
state time-series. No significant differences were observed in the motion parameters between
the 30-week and 40-week datasets (all P > 0.05). Considering recent concerns of motion-
related artefacts introducing spurious effects into FC, several steps were undertaken to
minimize the effects of motion in our FC analysis39-41: All neonates were sedated during
scanning, which minimizes the effects of voluntary head movement, and all neonates were
fixated in place by means of foam paddings, straps, and/or a vacuum cushion. Across datasets,
no significant differences were found in motion parameters between 30- and 40-week
neonates. The functional time-series were “scrubbed” before FC computation (i.e., removing
scans that showed a frame-wise displacement of >0.5 mm, as proposed by Power et al.39
chapter 8
Scrubbing resulted in the exclusion of (mean/ standard) 9.75%/13%). In 2 neonates, more than
33% of the scans showed more than 0.5 mm displacement (52 and 35%). Removing these 2
neonatal datasets did not change the nature of the resting-state findings. 3) Time-series were
band-pass filtered (0.01–0.1 Hz) to select resting-state frequencies of interest.
Struct ural Co nn e cto m e R e c o nstru cti on
The brain’s network was mathematically described as a graph, describing a collection of nodes,
and a collection of connections (also called edges) of the network. Nodes were selected as the
cortical regions of interest of the neonate template,32 resulting in a parcellation of the whole
neonate cerebral cortex into 56 regions (25 cortical regions covering each hemisphere,
together with bilateral amygdala, hippocampus, and cerebellum). Network nodes were
defined as being structurally connected when a set of tractography streamlines was present in
the total collection of reconstructed streamlines that interconnected them, with the number
of reconstructed streamlines taken as an indirect measure of white matter volume. In addition
to using absolute streamline count as a connectivity metric, additional analysis were
performed in which more qualitative weights of connectivity were taken, including the level of
FA as a metric of overall white matter integrity (e.g. van den Heuvel et al.).42 The neonatal (and
adult, see below) brain network was reconstructed at a relative low spatial resolution,
examining structural connectivity (SC) and FC between large-scale brain regions, of varying
sizes. Other studies have shown the feasibility of reconstructing and examining brain networks
at higher resolutions.3,43-47 The optimal resolution for MRI-based connectome reconstruction
and examination remains, however, unknown.48 At lower resolutions, subtle connectivity
effects might be missed due to spatial averaging, but higher resolution parcellation schemes
have been shown to involve higher levels of intersubject variability, making the accumulation
of information into group-averaged connectome maps and perform between-group analysis
more difficult.48
Fu ncti o nal Co nne cti v ity and F u nctio nal Ne tworks

For each neonatal dataset, the level of FC between all pairs of nodes in the network was
assessed through means of a correlation analysis. To this end, for all regions a regional
averaged resting-state time-series was computed, as the average time-series of all voxels
within a region. Next, between all pair of cortical regions the level of correlation between their
region time-series was computed, resulting in FC matrix describing the level of cortico-cortical
interregional functional couplings between each pair or regions in the neonatal brain.
Negative correlations were included and interpreted as low functional interactions between
brain regions.3,49,50
Next, a group-averaged FC matrix was created by averaging the individual FC matrices across
chapter 8
all individuals, and the formation of functional communities -also referred to as resting-state
networks- was examined by means of modular decomposition of this group-averaged FC
matrix.51,52 For this analysis, negative weights were set to 0.
Gr ou p -Av erag ed C o nne cto m e s

Group-averaged structural and functional connectomes were computed for the total group of
neonates, and for the week 30 and week 40 groups separately. Structural group connectomes

were formed by including structural connections that were presented in 60% or more of the
total group,53 with the weights of the connections in the group connectome taken as the mean
of the values across the included individuals. Similar, group-averaged functional connectomes
were computed as the average of the FC values across the group of subjects.
Gr aph T he o r y
Characteristic graph metrics were computed to describe the topological organization of the
neonatal structural connectome. Graph metrics were computed based on both the individual
unweighted networks (main analysis) and on the weighted networks (FA-weighted, additional
analysis). Metrics included (see also Supplementary Materials):
1. Clustering coefficient C, describing the level of topological local connectedness of a node. The
overall clustering coefficient C was computed as the mean of Ci over all nodes i in the network.
The normalized clustering coefficient γ was computed as the ratio between C and the mean
clustering coefficient Crandom of a set of random networks (1000 networks, randomizing the
connections of the original network, keeping the degree sequence intact).54
2. Shortest path length L, describing the average minimal travel distance between nodes in the
network. The overall path length L was computed as the mean of shortest path length Li over
all nodes i in the network. The normalized path length λ was computed as the ratio between L
and the mean shortest path length Lrandom of a set of random networks (1000 networks,
randomizing the connections of the original network, keeping the degree sequence intact).54
3. The small-world index SW was computed as the ratio the normalized clustering coefficient γ
and the normalized path length λ, with SW >1 indicating a small-world organization of a
network.55
4. Modularity Q, describing the tendency of the network to be formed out of distinct, separated
communities, each characterized by high within-community connectivity and relative low
connectivity to other communities. Modularity was computed on basis of Newmann’s
community detection algorithm.56
5. Rich club organization, describing the tendency of high-degree nodes in the network to be
densely interconnected, above what one would expect on the basis of their degree alone.57
Following recent reports of rich club organization of the adult connectome (e.g. van den
Heuvel and Sporns4), rich club organization of the neonatal connectome was computed as
follows: For the group-averaged neonatal connectome, for each level of degree k the rich club
chapter 8
coefficient Φ(k) was computed as the ratio between the number of connections present
within the subset of nodes E>k with a degree of >k, and the number of connections that could
maximally be present between the nodes in E>k. A network displays a rich club organization if,
for a range of k, Φ(k) exceeds the rich club coefficient Φrandom(k) of a set of random graphs
exceeds 1. The normalized rich club coefficient Φnorm(k) is defined as the ratio between Φ(k)
and Φrandom(k). In this study, a set of 1000 random networks was formed by randomizing the
connections of the group-average neonatal connectome, maintaining an identical level of
overall sparsity, as well as preserving the degree sequence of the network (i.e., hubs remain
hubs), but with the underlying network organization randomized.
I nte g ratio n Capacit y B e twe e n R e stin g -S tate Comm u nitie s

Integration capacity. The level of integration capacity of a network measures the capacity of the
network to integrate information between its communities. Given a network X, with n
functional modules (here taken as the 4 functional networks, see Results), the level of
integration I(X) of the network as a whole can be computed through means of information
theory. A mathematical description of I(X) is given in the Supplemental Materials and in
Zamora-Lopez et al..58 In short, I(X) measures the capacity of the network to integrate
information between its segregated communities. A system with total independence between
its modules satisfies I(X) = 0. Increasing levels of I(X) indicate a higher capacity of the structural
connections of the system to integrate information between its subparts. As communities
were defined on basis of module decomposition of the FC matrix, I(X) here reflects the level of
integration between functional domains of the brain system.
Adu lt Con n e ctom e

The neonatal connectome (both structurally and functionally) was compared to the adult
connectome. The adult connectome was taken from a dataset as described in detail in Collin
et al.14 This dataset included data from 26 males and 16 females (age mean/standard: 29/8.0 in
years). The adult connectome resulted from a reconstruction of the adult brain network on
basis of DWI, resulting in a macroscopic connectome describing all cortical-cortico white
matter connections between 68 cortical regions. For comparison to the neonatal connectome,
the 68 regions of the adult connectome were manually overlapped with the regions provided
in the Oishi atlas.32 The adult connectome included a more detailed parcellation of the human
cortex (68 cortical regions vs. 50 cortical regions (subcortical and cerebellum were excluded
when comparing to the adult cortical connectome) in the neonatal connectome). To match
the cortical regions as defined in the neonatal atlas, cingulate regions (posterior, anterior,
caudal, isthmus cingulate), rostral frontal regions (rostral middle frontal, rostral caudal middle
frontal, frontal pole), lateral frontal (parsopercularis, parsorbitalis, parstriangularis), and lateral
temporal (superior temporal, transversetemporal and temporal pole) were merged together
into single cortical regions, respectively.
chapter 8
Stati st ica l A n a ly s i s
C ompariso n o f th e Layo u t o f the N e o natal and Adu lt Conne ctome
The connectivity layout of the group-averaged neonatal SC and FC brain networks were
compared with the connectivity layout of the SC and FC adult brain networks. An adjusted
version of the Mantel test for comparison of matrices was used.59 First, for the SC and FC
matrices, the overlap between the binary neonatal and adult group-averaged matrices was

compared, yielding a “distance matrix” (DIS) indicating for each cell entry whether the cell
entries were similar or different across the neonatal and adult matrix. When cell entries were
overlapping across the neonatal and adult connectome, this was marked as a 1 in the DIS
matrix, and 0 otherwise. Second, the total level of overlap O was expressed as the level of
sparsity of the DIS matrix (i.e., the percentage of entries being a 1 present in the DIS matrix),
ranging between 0% (no overlap between the neonatal and adult matrix, showing maximal
distance between the 2 matrices) and 100% (complete overlap between the neonatal and
adult matrix, indicating minimal distance between the 2 matrices). Third, similar to the Mantel
test, permutation testing was used to obtain a null-distribution of distance scores that are
present under the null-hypothesis of no overlap being present between the 2 matrices. To this
end, for each permutation, the entries of neonatal and adult connectome were randomized
(obtaining a similar degree distribution54) and a DIS* matrix was computed, together with the
overlap score O*. This was repeated for 10 000 permutations, yielding a null-distribution of O*.
Fourth, the original overlap score of the neonatal and adult matrix was assigned a P-value as
the percentage of the obtained null-distribution that exceeded the original overlap score.42,49
For the comparison of the FC matrices, suprathreshold FC correlations of >T were used.
Multiple settings of T (0 < T < 0.2) were tested (see Results).
Dev elo pm e ntal E f f e cts

Developmental effects of white matter changes and graph metrics were examined through
means of cross-sectional analysis using linear regression, with GA as the independent variable
and white matter integrity/graph metrics as the dependent variable of interest. Longitudinal
effect was examined by means of two-sample tests.
Results
In what follows, we describe results pertaining the architecture of the neonatal connectome,
including a description of its structural and functional organization, followed by results
pertaining the development of neonatal connectivity during the first weeks (week 30 to week
40 corrected GA) of cerebral connectome genesis.
Top ologica l O r gani z at i o n o f t h e St ruct ural Ne o natal

chapter 8
Con n e ctom e
For each individual dataset, a structural brain network was formed on the basis of the diffusion-
weighted data, reconstructing the cortico-cortical tracts of the neonatal brain. Examining the
(binary) topological organization of the resulting network, revealed high levels of binary
clustering (mean/ standard: 0.66/0.033) and normalized clustering (1.46/0.25, when compared
with 1000 random networks), suggesting a high level of local organization, higher than one can
expect on basis of a random topology (Fig. 1a). Furthermore, networks revealed overall short
communication paths (mean/standard: 1.70/0.11, normalized: 1.04/0.030, 1000 networks).
Taken together, such an efficient local and global organization is indicative of a small-world
organization of the preterm neonatal brain (small-world index, mean/standard: 1.40/0.20, 1000
random networks) (Fig. 1). Incorporating information on the weights of the connections on the
basis of the number of streamline count (NOS) and FA revealed similar findings, showing above
chance levels of clustering and overall short communication paths in the neonatal brain.
Consistent with the observed clustering, the neonatal structural connectome revealed high
levels of modularity Q (mean/standard: 0.27/0.06), significantly higher than a level of
modularity that one would expect on the basis of a random network (P < 0.001).
C onn ecti v ity H u b s

In addition to a small-world modular topology, the neonatal connectome revealed a right-
tailed degree distribution (Fig. 1b), suggestive of the existence of a small number of densely
connected “hub nodes.” Such an organization overlaps with reports of hub formation in the
adult brain.60 The top 12 (20% of total) of highest connected nodes included left superior
frontal cortex, left lateral front-orbital gyrus, left precentral gyrus, left postcentral gyrus, left
and right superior parietal lobule, left and right cingular gyrus, left and right angular gyrus, and
left and right fusiform gyrus (Fig. 1c). Notably, the network betweenness centrality of these
high-degree nodes was found to be significantly higher than that of nonhigh-degree nodes (P
< 0.001, 10 000 permutations), suggesting a central position of these nodes in the overall
network topology. Besides being individually “rich” in connectivity, these putative brain hubs
showed an above chance level of interconnectivity, which suggest the presence of a densely
connected central “core” or “rich club” in the neonatal brain, a type of network organization
reported for the adult human connectome.4 Figure 1d shows the rich club curve of the
neonatal connectome (group-average) with the normalized rich club coefficient Φnorm(k) > 1
for the range of k > 17 and k > 24, indicative of a dense level of connectivity between high-
degree nodes of the neonatal brain and suggestive of the formation of an early rich club
collective in the neonatal connectome.
Fu n cti on a l Con ne ct i v i t y a nd F u nct i o nal Ne t wo rks o f t he

Neon ata l B r ai n
Functional connectivity of the neonatal brain was examined by means of correlation analysis
chapter 8
between resting-state fMRI time-series of cortical brain regions. Cortical regions were identical
to the nodes of the structural network, allowing for a direct overlap between the SC and FC
connections of the neonatal brain. Figure 2a shows a side-by-side comparison of the group-
averaged FC matrix of the neonates, together with the group-averaged SC matrix. Rows and
columns of both the FC and SC matrix are arranged in the same ordering, with the nodes
arranged according to FC module participation (Fig. 2b), showing a clear overlap between the
neonate’s functional and structural connectivity layout.

Figure 1 Network descriptive of the neonatal connectome. (a) “Small-world organization of the neonatal brain.” Figure shows
the clustering coefficient and shortest path length values of the individual neonatal connectome (dark gray bars) and levels of a
set of comparable networks (light gray). Error bars depict variation (standard deviations) across the individual datasets. The
small-world index SW was significantly higher than 1, suggesting a small-world organization of the neonatal connectome at
term. (b) “Degree distribution.” Figure shows a right-tailed distribution of the group-averaged connectome, indicating the
existence of high-degree hub nodes in the neonatal brain. (c) “Hubs.” Figure illustrates the locations of the top 12 high-degree
hub nodes in the group-averaged neonatal connectome (locations are shown on one hemisphere only, yielding 8 unique hub
locations, see Results). (d) “Rich club organization.” Rich club analysis revealed above rich club coefficient levels >1, indicating
dense levels of connectivity between high-degree nodes, as previously reported for the adult connectome.
F uncti o nal N e two r k s

Functional networks showed a modular organization (Q =0.40). Across the group of neonates,
4 functional modules could be identified, describing a (bilateral) 1) frontal network, a 2)
motor/sensory network, an 3) occipital visual network and a 4) temporal/auditory network.
Str uct ural – F u nctio nal C o u pling

c ha pt er 8
The level of structural–functional coupling was examined through means of Pearson

correlation analysis. Structural weights (NOS count) of the non-zeros entries of the network
were rescaled to a Gaussian distribution and correlated with their functional counterparts,
resulting in a single structure– function coupling (SC-FC) metric.3,14,61 A positive SC-FC coupling
was found to be present in all neonates, suggesting that higher levels of structural connectivity
are associated with higher levels of functional coupling (SC-FC coupling, mean/standard:
0.31/0.089).
Figure 2 Structural and functional connectivity matrices of the neonatal connectome. Figure shows the structural and
functional connectivity matrices of the group-averaged neonatal brain. Rows and columns include all cortical regions,
expressing the level of structural and functional connections between each pair of regions in the neonatal brain. The structural
connectivity matrix (SC) is shown in the left panel, with the levels of connectivity expressed as the number of reconstructed
streamlines (NOS) between each pair of regions. Higher levels of SC are indicated by larger dots. Levels of positive functional
coupling (i.e., functional connectivity, higher levels of FC are indicated by larger dots) between each pair of regions are shown in
the functional connectivity (FC) matrix as shown in the right panel. The columns and rows (i.e., nodes of the network) are
ordered according to the community assignment as based on community detection on the FC matrix (see Materials and
Methods) revealing the existence of 4 functional communities in the neonatal brain. These communities overlapped temporal
(auditory regions), occipital (visual regions), central (motor and sensory regions) and frontal cortex. Distribution of the nodes
across the cortex are illustrated in the lower panel. Figure illustrates clear overlap between the SC and FC matrix of the neonatal
brain, indicating a structural basis of functional connectivity in the neonatal brain. To aid visual comparison between the SC and
FC matrices, positive values are shown in the FC matrices with all negative correlations represented as empty cells.
Co mpa r i son of t h e N eo nata l Br ai n N e t wo rk

to the A d u lt Con necto m e
ch a pt e r 8
Struct ural Co nn e cto m e

To examine whether particular structures of the adult human connectome are already present
at (term equivalent) birth, the organization of the neonate’s connectivity circuitry was
compared with the organization of the connectome of a group of adults. An adult connectome
was reconstructed based on DWI data from a group of healthy adult participants (see
Materials and Methods for details). Figure 3a shows a side-by-side comparison of the neonatal
connectome and a group-averaged adult connectome, illustrating a high level of overlap

Figure 3 Comparison of the group-averaged neonatal and adult connectome. Upper panels show the structural connectivity
(SC) matrices (NOS weighted) of the neonatal (left) and adult (right) group-averaged connectome. Rows and columns (i.e.,
nodes of the network) are ordered according to the community structure of the adult FC matrix. Lower panels show the level of
positive functional coupling (functional connectivity FC) between each pair of regions in the neonatal (left) an adult connectome
(right). Columns and rows (nodes) of the matrices are ordered according to module decomposition of the adult functional
connectome, revealing 4 distinct functional modules, overlapping an occipital (visual), temporal (auditory), central (sensory,
motor network) and the functional default mode network. Figure shows overlap between both the SC matrices of the neonatal
and adult group (overlap: 81%, see Results), confirming our observations that overall SC structure is present before term
equivalent birth. Furthermore, overlap between the FC matrices (overlap: 79%, see Results) suggests the presence of outlines of
functional modules in the neonatal brain, albeit being in an immature state. Note that the rows and columns (i.e., nodes of the
network) are ordered according to the community structure of the adult FC matrix, and thus different from the ordering used in
Figure 2.
between the network of the neonatal brain (left panel) and that of the adult brain (right panel)
on SC. For visual comparison, regions of the neonate and adult template are placed into the
chapter 8
same arrangement to allow a visual comparison between the 2 networks (see Materials and
Methods), showing clear overlap between the structural of the neonatal and adult
connectome. Quantitative comparison of the group-averages structural neonatal and adult
connectome (comparing the existence and nonexistence of each of the entry in the matrix
between the group-averaged neonatal and adult structural connectivity matrix) revealed
84.6% overlap between the 2 matrices (P < 0.001, 1000 permutations). Importantly, focusing
on the intrahemispheric connections in the neonatal and adult connectome (i.e., cortico-
cortical connections within a single hemisphere, not taking into account interhemispheric
corpus callosal tracts) also showed a high level of overlap (76% averaged over the 2
hemispheres, P < 0.00, 1000 permutations; left hemisphere: 78% right hemisphere: 75%).
These results are indicative of the majority of all large-scale pathways (both intra as well as
interhemispheric tracts) to be present at term, together with an adult-like small-world
modular architecture.
Fu ncti o nal N etwo r k s

Figure 3b shows a side-by-side comparison of the FC matrix of the group of neonates and the
group of adults. Rows and columns (i.e., nodes of the network) are arranged in the same
ordering as the matrices in Figure 3a. Side-by-side comparison again shows that the FC matrix
of the neonatal group shows the same contours of the community structure of the adult
matrix, suggestive of the notion that the outlines of “adult-like” functional networks are
already present (albeit in an underdeveloped form) directly from the first phases of
connectome formation. Quantitative comparison of the existence and nonexistence of
(suprathreshold FC connections using a T > 0.1) functional connections in the adult and
neonatal connectome revealed 81% overlap between the 2 matrices (P < 0.001, 1000
permutations). Intrahemispheric FC connections was also found very consistent across adult
and neonatal networks (77% averaged over the 2 hemispheres, P < 0.001, 1000 permutations;
left hemisphere: 77%, right hemisphere: 77%). Different thresholds T resulted in similar
findings (whole brain: T > 0: 62% P < 0.001, T > 0.2: 86% P < 0.001; intrahemisphere (averaged
over 2 hemispheres): T > 0: 62% P < 0.001, T > 0.2: 86% P < 0.01, all tests survived Bonferroni
correction for multiple testing).
Deve lop me n t of t h e N eo nata l Co nnecto m e

Acquisition of MR data at 2 different periods of GA (28–32 and 40–42 weeks) allowed for an
exploratory analysis of early developmental changes of the connectome. Our cross-modal
structural–functional study design allowed for the examination of at least 4 aspects of
development, including an examination of 1) developmental changes in microstructural white
matter organization, 2) effects of white matter changes on the topological organization of the
macroscopic neonatal connectome, 3) developmental influences on the formation of
functional networks, and 4) conjectural implications of connectome development on the
chapter 8
communication and integration efficacy of the neonatal brain network.
Pret e r m D e v e lo p m e nt o f W h i t e Mat t er St ruct ure

Examination of the white matter markers between week 30 and week 40 postconceptional age
revealed changes in prenatal white matter microstructure. First, FA showed a significant
increase over time (r = 0.90, P < .001, Fig. 4a), suggesting a development of the local
microstructure of the brain’s structural connections. Second, other markers of white matter

Figure 4 Development of structural and functional connectivity (a) Panels show individual levels of fractional anisotropy (FA),
mean diffusivity (MD) and transverse diffusivity (TD) of each included neonatal dataset in relationship to age. Panels show a
clear increase in white matter development between week 30 and week 40 gestational age (GA), illustrated by an increase in FA
and a decrease in MD and TD over time. (b) Panel shows individual levels of interhemisperic functional connectivity. (c)
Developmental change of structural-functional coupling between week 30 and week 40 GA.
efficacy also showed a clear association with age, including a developmental decrease in the
MD (r = 0.88, P < 0.001, Fig. 4a), a decrease in TD coefficient (r = 0.88, P < 0.001, Fig. 4a), and
decrease in the PD coefficient (r = 0.82, P < 0.001). As the age range was not normally
distributed, additional statistical testing was performed to verify the significance of these
findings: Directly testing group differences on FA, MD, TD, and PD between the group of 30-
week GA neonates (<32 GA) and the group of 40-week GA neonates (>38 GA) using
nonparametric permutation testing of the 2 group means all revealed significant differences
chapter 8
between week 30 GA and week 40 GA neonates (P < 0.001, 10 000 permutations). (All tests
survived Bonferroni correction for multiple testing).
Lo n g itudinal A naly sis

The increase in white matter microstructure was confirmed in the longitudinal analysis (7
neonates were scanned both around 30 weeks and around 40 weeks), showing a significant
increase in FA (P < 0.001, Fig. 6) and a significant decrease in MD (P = 0.005, Fig. 6), TD (P <
0.001), and PD (P = 0.0014). All tests survived Bonferroni correction for multiple testing.
Deve lop me n t of t h e To p o lo gica l Ar c h it e ct ure o f t he
Neo n ata l Con n e cto m e
Next, we examined the effects of white matter changes on the development of the topological
properties of the brain’s network circuitry. A positive increase in clustering coefficient C was
observed (r = 0.78, P < 0.001, Fig. 5a), together with shorter communication pathways, as
shown by a decrease in total path length L (r = 0.83, P < 0.001, Fig. 5a). Normalized to the
organization of a set of 1000 random networks, white matter development coincided with an
increase in the overall small-world organization of the network (r = 0.47, P = 0.006, Fig. 5b),
suggestive of the growth of the neonatal connectome to a more overall efficient organization
during development. In addition, coinciding with the increase in clustering, also the level of
modularity of the network was found to increase between 30 and 40 weeks (r = 0.50, P =
0.0037, Fig. 5b). The tests of C, L, and SW survived Bonferroni correction for multiple testing.
Figure 5 Development of network

characteristics. Figure shows individual
levels of several network characteristics of
the (unweighted) structural neonatal brain
network between 30 and 40 weeks
gestational age (GA). (a) The level of
clustering C was found to significantly
increase over time. Furthermore, the level
of path length L was found to decrease over
time, indicating the formation of shorter
communication pathways in the neonatal
brain. (b) The increase in clustering and
decrease in path length resulted in a
significant increase in small-world topology
of the neonatal brain between week 30 and
week 40. Neonatal development was also
found to involve an increase in network
modularity, suggesting a sharper definition
of structural modules in the neonatal brain
over time. (c) Hub nodes (see Materials and
Methods, and Fig. 1) in the neonatal brain
network were only found to show, at best, a
marginal increase in centrality over time.
This effect was not significant (P = 0.13 ns).
(d) The level of Integration Capacity (as
chapter 8
estimated by mathematical modelling on

basis of the structural connectivity
network) was found to increase between
week 30 and week 40 GA.

Lo n g itudinal A naly sis
The impact of development of white matter microstructure on network topology was
confirmed in the longitudinal analysis, showing a modest increase in the network’s small-
world organization of the neonatal connectome between 30 and 40 weeks (P = 0.006, Fig. 6),
together with an increase in normalized clustering (P = 0.014).
P r e te r m D ev elo p m e nt o f F u nct i o n al Re s t ing-Stat e Ne t wo rks

a n d S tr u ct u r a l - F u nct i o na l Co u p ling
No clear developmental increase in overall FC (i.e., averaged over all connections in the FC
matrix) was observed between week 30 and week 40 GA. A significant increase in
interhemispheric FC (aggregated over all homotopic left-right hemispheric region pairs, as an
often examined metric of positive functional couplings in resting-state fMRI studies) was
observed between week 30 and week 40 GA (r = 0.69, P = 0.001, Fig. 4b). Intrahemispheric FC
(aggregated over all intrahemispheric connections) was not found to significantly change
during week 30 and week 40 GA (P > 0.05). Interestingly, also the level of structural–functional
coupling revealed a significant association with age (r = 0.53, P = 0.0137, Fig. 4c), suggesting
an increase in coherence between structural and FC of the neonatal brain during prenatal
development.
Figure 6. Development of structural

connectivity and network characteris-
tics (longitudinal analysis). Of 7 neo-
nates, longitudinal assessment of
brain development was possible, on
the basis of a structural and function-
al MRI measurement at both
(around) week 30 and week 40 gesta-
tional age (GA). Longitudinal analysis
confirmed developmental changes in
structural and functional properties
of the neonatal brain network. White
matter microstructure showed a
clear development between week 30
and week 40 GA in all 4 neonates, in-
cluding a significant increase in frac-
chapter 8
tional anisotropy (FA), a decrease in

mean diffusivity (MD) of white matter
and an increase in small-world (SW)
properties of the network. Further-
more, confirming the cross-sectional
findings, integration capacity showed
a significant increase during week 30
and week 40 GA.
Long itudinal A naly sis
Due to the low sample size of longitudinal FC measurements, only an exploratory analysis on
possible longitudinal aspects of FC and SC-FC coupling was possible. Longitudinal analysis
revealed a positive increase in FC (P = 0.0467) and SC-FC coupling (P = 0.0401). These findings
are consistent with the cross- sectional observations.
Co n je ctu r a l I mp l icat i o ns o f W h i t e Mat t e r D e v e lo pm e nt o n

Co mmu n icat i on E f F Icacy N e t wo r k a nd Int e grat io n Capaci t y
Communication hubs in the neonatal brain (12 nodes, selected as described above) showed a
trend-level increase in betweenness centrality—indicating the level of central placement of
hubs in the overall network—over time (r = 0.28, P = 0.15 ns, Fig. 5c), suggestive of a possible,
but modest increase in the level of centrality of these nodes in the overall network during
development. Fig. 5d shows this increase in integration capacity between the functional
communities between 30 and 40 weeks. With the functional modules describing a modular
partitioning of the neonatal brain, the capacity of the underlying structural brain network to
integrate information between these subsystems was found to significantly increase over time
(r = 0.45, P = 0.0072).
Long itudinal A naly sis

Longitudinal analysis revealed an increase in integration capacity between week 30 and week
40 (P = 0.0286, Fig. 6), supporting the notion of a growth of the connectome toward a more
global and integrative network over time.
Discussion
Our findings show the early presence of key organizational features of the human connectome,
albeit in an immature state and subject to clear developmental changes starting at least as early
as in the third trimester of gestation. Collecting and combining structural and functional
neuroimaging data in 27 neonates scanned between week 30 and 40 GA, our study reports on 5
properties of formation and transformation of the early human macroscopic connectome.
First, cortico-cortical white matter pathways of the preterm brain revealed characteristics of a
small-world, modular, and hub-including network architecture. Studies have reported on such
chapter 8
topological properties of the baby, child, and adult brain, advocating that a small-world
modular architecture may be a hallmark feature of macroscopic cortical connectivity.2,3,62-65
Our findings now suggest that these organizational principles are likely to be present from the
earliest phases of cortical connectome formation, before term birth. Such an early architecture
of the structural connectome is consistent with recent studies showing a small-world
organization of the infant functional brain network derived from resting-state fMRI66 and EEG
recordings.67

Second, overlapping recent reports of the existence of functional hubs in the term neonatal
brain based on resting-state fMRI recordings,66 our study shows the existence of a hub
structure of the neonatal anatomical connectome. Observed high-degree hub regions overlap
the functional hubs as reported by Fransson et al.,66 classifying (among other regions) the
cingulate cortex, superior frontal and superior parietal regions as high-degree hub regions.
Interestingly, as mentioned by Fransson et al., the spatial location of these putative hubs tend
to show large overlap with both structural and functional hub regions as reported in the adult
brain,3,42,47,60,68 suggesting an early presence of connectivity hubs and a neural rich club in the
human brain.
Third, our findings suggest an early start of developmental changes in the microstructure of
white matter connections. The mean diffusivity and FA values of cortico-cortical tracts
displayed clear changes during week 30 and week 40 GA, pointing into the direction of an
increase in white matter structure and thus an increase in the efficacy of long-distance
neuronal signal transmission in the neonatal brain. These observed developmental changes in
white matter microstructure of the neonatal connectome are in clear support of studies
reporting global FA and MD changes in the (preterm) neonatal brain,29,69-71 as well as with
numerous studies showing developmental effects of white matter microstructure during later
(>2 years) brain development.17,72,73
Fourth, outlines of functional communities were observed in the neonatal brain overlapping
functional networks as commonly reported for the child and adult brain.74-77 No clear
developmental changes in these networks could be observed during the examined period
here, supporting theories suggesting that most functional networks are present in the early
brain, but in an immature state and subject to functional reorganization during later brain
development.25,26,66,78
Fifth, changes in white matter microstructure showed to have a clear impact on the
topological structure of the macroscopic structural brain network, revealing an increase in
small-world network characteristics. Such a developmental gradient is supported by the
results of the performed computational modelling analysis, showing an increase in the
network’s integration capacity between week 30 and week 40 GA. These findings suggest that
connectome development is accompanied by an increase of the brain system to consolidate
information from different functional subnetworks, which is in support of modern theories
hypothesizing that development of the child and adolescent brain involves an increasing
chapter 8
capacity of the network to integrate information between its sub-parts.78,79

Within the infant and child brain, white matter development has been suggested to mostly
involve ongoing myelination of axonal projections, a developmental process that is suggested
to continue at least throughout adolescence.3,80 Postmortem myelogenetic maps have shown
myelination of primordial regions at birth, while myelination of tertiary association regions
and their tracts suggested to start after birth (see for review, Catani et al.81). Animal studies of
white matter development have indeed suggested that large-scale myelination processes of
cortico-cortical association tracts do not start before the onset of (term) birth,82,83 with
prenatal white matter development predominantly involving an increase in total number of
axons and an increase in the variation of axonal diameter of axons in white matter pathways.82
Our neuroimaging observations of increasing FA and decreasing MD/TD signal between week
30 and week 40 GA may thus also reflect, to some extent, developmental changes in the
diameter of axons and an ongoing growth in axonal count of the examined cortico-cortical
pathways.84 To the best of our knowledge, no previous reports have been made on in vivo
changes in axonal count or axonal diameter of macroscopic white matter bundles in the
immature human brain on basis of neuroimaging observations. A clear delineation of the
different processes of microstructural white matter development is out of range of current
imaging techniques, but new advances in neuroimaging might provide more detailed
measurements of axonal microstructure in the near future,85-87 allowing further understanding
of the underlying processes of formation, growth, and transformation of macroscopic brain
connectivity during this crucial period of brain development.
The early presence of a small-world and rich club topology may suggest the existence of an
underlying “connectivity blue- print” of the human brain. High consistency of connectivity
patterns across mammalian species,15,48,88-90 together with findings of a genetic background of
brain networks50,91,92 indeed suggest the existence of a somewhat predefined, likely genetically
driven, layout of macroscopic connectivity of the human brain. Our findings do not, however,
necessarily imply that the trajectories of white matter pathways are completely
predetermined. Tract tracing data of the developing kitten brain have revealed spatially
widespread termination zones of long-range visual corpus callosum tracts in kittens at post-
term day 4, followed by a relative short period (days to weeks) in which spatial specialization
of tracts and their termination zones occurs (see for review, Innocenti and Price6). Such a,
potential nonspecific, overgrowth of macroscopic connectivity is further supported by
observations of corpus callosal and associative tracts in young rhesus monkeys, revealing the
highest number of axonal connections at birth, followed by a 70% decrease in axonal count in
the first postnatal weeks.83,93 Thus, despite the presence of a putative small-world type
architecture of the human brain at term birth, one might speculate about the notion of a, to
some extent, initially random-shaped layout of cortico-cortical connections at connectome
genesis, with rapid pruning and specialization of white matter connections after birth.6,82,93
Our results are inherently limited by both the nature of the noninvasive imaging techniques
involves, as well as by the group of participants. First, as mentioned, the majority of
chapter 8
connections in the human brain involve intracortical and very short intracortical axonal
projections,15,16 connections that are well out of reach for current day neuroimaging diffusion
and resting-state fMRI technology that operate at the millimetre to centimetre resolution. In
contrast, only a fraction of all neural interactions involve long-range associative and
commissural white matter projections and form the “macroscale connectome”.94,95 It is this
class of connections that is, at the macroscale, within the reach of modern neuroimaging
techniques, as also used in this study. However, in this context, it is important to note that
non-invasive neuroimaging techniques only provide “reconstructions” of large bundles of

neural connections, and do not directly measure or asses individual neural projections. As a
result, both diffusion imaging and functional MRI techniques suffer from many limitations, like
the complex reconstruction of projections where the diffusion signal is not uniform as in
points were multiple fibre bundles cross or pass,37,86,87 the inclusion of false-positive and false-
negative connection reconstructions53 and a known influence of postprocessing steps on the
reconstructions and subsequent graph analytical analysis of the data.39,47,48,63 Despite these
limitations, at the macroscale, studies have shown consistent results of key features of
connectome organization across different acquisition and analysis techniques, as well across
different species.46,96-99 Together, these findings suggest that diffusion imaging and functional
MRI are useful approaches to map and examine the topological organization of human
connectome at the macro-scale62,95,100,101 and study the interaction between anatomical
connectivity and functional brain dynamics.3,14,46,102,103 Furthermore, it provides a promising
platform to examine the genetic contribution to the formation of macroscale brain
connectivity,50,91,104 examine structural and functional dysconnectivity effects in neurological
and psychiatric disorders,63,105-107 and in particular study how disrupted and/or deviating
development of brain connectivity may relate to the development of brain disorders.24,108-110
Second, concerning the nature of the included population, all of our data were acquired post
birth, with neonatal development continuing in an incubator environment. Third, as most
studies concerning neonatal development, also our study contains a relative small group of
subjects when compared with imaging studies performed in older children and adults. This is
particular the case for our developmental longitudinal analysis, which included only a small
subset of inclusions, and the results of this analysis should therefore be interpreted as
exploratory and preliminary. Fourth, the acquisition of neonatal data is a specialized task, and
similar to most neonatal studies, scans were acquired as part of standard clinical care. As a
result, the included group is rather heterogenous (birth weight varied between 460 and 1360
g, with 19 neonates showing mild brain injury as shown on a conventional image, see Table 1).
Furthermore, preterm born neonates often have brain lesions, developed in the perinatal
period, which may lead to a deviating pattern of brain development111 and affected cognitive
functioning later in life.112,113 To reflect the processes of normal development as much as
possible, only neonates without major morbidity on their conventional MR-imaging were
included in our study. Fifth, scan-time is limited in a clinical setting, and as a result the diffusion
imaging sequence involved a relative low number of directions (32 directions), when
c ha pt er 8
compared with adult studies. And sixth, all data were acquired under sedation which may
influence the acquisition of the resting-state fMRI data.114 Nevertheless, neonatal FC patterns
showed relative high overlap with patterns as observed in non-sedated, awake adults. Others
have indeed reported that sedation during scanning does not have a strong influence on
global FC patterns and the formation of resting-state networks.115
This report demonstrates the presence of an early small- world modular architecture of the
human neonatal connectome. Our findings suggest that a development toward a more
efficient and more integrative brain network forms one of the key biological algorithms that
shape macroscopic connectivity in the early human connectome.
Supplementary Material
Grap h t he or y
Characteristic graph metrics were computed to describe the topological organization of the
neonatal structural connectome. Graph metrics were computed based on both the individual
unweighted networks (main analysis) and on the weighted networks (FA-weighted, additional
analysis). Examined metrics included:
(i) clustering coefficient C, describing the level of topological local connectedness of a node,
given as:
# edges inGi
Ci =
#1edges
k i (ki inG
1) i
C = 2
i
1 1)
k (k 1)
2 i i
with Gi the local subgraph of nodes connected to node i. The overall clustering coefficient C
was computed # edges inGi
Ci as
= the mean of Ci over all nodes i in the network. The normalized clustering
1
k (k 1) as the ratio between C and the mean clustering coefficient
coefficient γ was computed
2 i i
Crandom of a set of random networks (1,000 networks, randomizing the connections of the
original network, keeping the degree sequence intact).54
(iii) shortest path length L, describing the average minimal travel distance between nodes in
the network, given as,
1
L= d(i, j)
N(N1 1) i j,i, j G
L= d(i, j)
N(N 1) i j,i, j G
2)
1
with d(i,j) the shortest distance between node i and j in the network. The overall path length L
L= d(i, j)
N(N 1)
was computed as the mean i j,i, j G
of Li over all nodes i in the network. The path length λ was
computed as the ratio between L and the mean shortest path length Lrandom of a set of
ch a pt e r 8
random networksSW (1,000

= networks, randomizing the connections of the original network,
SW =
keeping the degree sequence intact).54
the small-world index SW was computed as the ratio the normalized clustering coefficient γ
and the normalized path length λ
SW =
2E >k 3)
(k) =
(N>k>k 1)
N>k 2E
(k) =
N>k ( N>k 1)
2E >k
(k) =
N>k ( N>k (k)1)
1
L => 1 indicatingi aj,i,small-world
with SW d(i, j) organization of a network.55
N(N 1) j G
(v) modularity Q, describing the tendency of the network to be formed out of distinct,
separated communities, each characterized by high within-community connectivity and
relative low connectivity to other communities. Modularity was computed on basis of
1
L = community
Newmann’s d(i, j) algorithm.56
detection
N(N 1) i j,i, j G
(iv) rich club organization, describing the tendency of high degree nodes in the network to be
densely interconnected, above what one would expect on the basis of their degree alone.57
Following recent reports of rich club organization of the adult connectome (e.g. van den
SW =
Heuvel and Sporns4), rich club organization of the neonatal connectome was computed as
follows: For the group-averaged neonatal connectome, for each level of degree k the rich club
coefficient f (k) was computed as the ratio between the number of connections present
within the subset of nodes E>k with a degree of >k, and the number of connections that could
SW =
maximally be present between the nodes in E>k . Formally,
2E >k
(k) =
N>k ( N>k 1)
4)
2E >k
(k) =
With E>k expressing the number of connections present between the subset of nodes with a
N>k ( N>k 1)
degree >k, and N>k the (k)number of nodes with a degree >k. A network displays a rich club
norm (k) =
of k, f (k) exceeds the rich club coefficient f random(k) of a set of
random (k)
organization if, for a range
random graphs, or, equivalent, if the ratio
(k)
(k) =
random (k)
norm

5)
n
I(X) =
exceeds 1. In this
H(X i ) aH(X)
i=1 study, set of 1,000 random networks was formed by randomizing the
connections of the group-average neonatal connectome, maintaining an identical level of
overall sparsity, as well as preserving the degree sequence of the network (i.e. hubs remain
hubs), but with the underlying network organization randomized.
c ha pt er 8
n
I(X) = H(X i ) H(X)
i=1
Integration capacity between resting-state communities

Integration capacity. The level of integration capacity of a network measures the capacity of
the network to integrate information between its communities. Given a network X, with n
functional modules (here taken as the 4 functional networks, see results), the level of
integration I(X) of the network as a whole can be computed through means of information
theory, as given in Zamora-Lopez et al.58 In short, the integration I of the system X can be
described in terms of a potential decrease in entropy when its subparts X1, X2, …, Xn are
combined, formally described as:
n
I(X) = H(X i ) H(X)
i=1
6)
where the analytically solvable entropy H(Xi) of a network Xi measures the statistical
independence of the nodes of the network (assuming simple linear dynamics with Gaussian
noise and a normalized coupling strength g of 0.5).58 As such, I(X) measures the capacity of
the network to integrate information between its segregated communities. A system with
total independence between its modules satisfies I(X) = 0. Increasing levels of I(X) indicate a
higher capacity of the structural connections of the system to integrate information between
its subparts. As communities were defined on basis of module decomposition of the
functional connectivity matrix, I(X) here reflects the level of integration between functional
domains of the brain system.
Funding
MPvdH is supported by a Fellowship of the Brain Center Rudolf Magnus and a VENI grant of the
Dutch Council for Research (VENI: 451-12-001 NWO). Funding to pay the Open Access
publication charges for this article was provided by the Dutch Research Council (NWO).
ch a pt e r 8

References
1 Sporns O, Tononi G, Kotter R. The human the Human Cerebral Cortex. Cereb Cortex 2013
connectome: A structural description of the April 3.
human brain. PLoS Comput Biol 2005 15 Markov NT, Misery P, Falchier A, Lamy C, Vezoli J,
September;1(4):e42. Quilodran R, Gariel MA, Giroud P, Ercsey-Ravasz
2 Bassett DS, Bullmore E. Small-world brain M, Pilaz LJ, Huissoud C, Barone P, Dehay C,
networks. Neuroscientist 2006 Toroczkai Z, Van Essen DC, Kennedy H,
December;12(6):512-23. Knoblauch K. Weight consistency specifies
3 Hagmann P, Cammoun L, Gigandet X, Meuli R, regularities of macaque cortical networks. Cereb
Honey CJ, Wedeen VJ, Sporns O. Mapping the Cortex 2011 June;21(6):1254-72.
structural core of human cerebral cortex. PLoS 16 Schuz A, Braitenberg V. The human cortical white
Biol 2008 July 1;6(7):e159. matter: quantitative aspects of cortico-cortical
4 van den Heuvel MP, Sporns O. Rich-club long-range connectivity. In: Schuz A, Miller R,
organization of the human connectome. J editors. Cortical Areas: Unity and Diversity. 2002.
Neurosci 2011 November 2;31(44):15775-86. 17 Hagmann P, Sporns O, Madan N, Cammoun L,
5 Bystron I, Blakemore C, Rakic P. Development of Pienaar R, Wedeen VJ, Meuli R, Thiran JP, Grant
the human cerebral cortex: Boulder Committee PE. White matter maturation reshapes structural
revisited. Nat Rev Neurosci 2008 connectivity in the late developing human brain.
February;9(2):110-22. Proc Natl Acad Sci U S A 2010 November
6 Innocenti GM, Price DJ. Exuberance in the 2;107(44):19067-72.
development of cortical networks. Nat Rev 18 Hwang K, Hallquist MN, Luna B. The
Neurosci 2005 December;6(12):955-65. development of hub architecture in the human
7 Kostovic I, Jovanov-Milosevic N, Rados M, functional brain network. Cereb Cortex 2013
Sedmak G, Benjak V, Kostovic-Srzentic M, Vasung October;23(10):2380-93.
L, Culjat M, Rados M, Huppi P, Judas M. Perinatal 19 Khundrakpam BS, Reid A, Brauer J, Carbonell F,
and early postnatal reorganization of the Lewis J, Ameis S, Karama S, Lee J, Chen Z, Das S,
subplate and related cellular compartments in Evans AC. Developmental changes in
the human cerebral wall as revealed by organization of structural brain networks. Cereb
histological and MRI approaches. Brain Struct Cortex 2013 September;23(9):2072-85.
Funct 2014 January;219(1):231-53. 20 Yap PT, Fan Y, Chen Y, Gilmore JH, Lin W, Shen D.
8 Lagercrantz HE. The newborn brain: neuroscience Development trends of white matter
and clinical applications. Cambridge: Cambridge connectivity in the first years of life. PLoS One
University Press; 1998. 2011;6(9):e24678.
9 Stiles J, Jernigan TL. The basics of brain 21 Fransson P, Skiold B, Engstrom M, Hallberg B,
development. Neuropsychol Rev 2010 Mosskin M, Aden U, Lagercrantz H, Blennow M.
December;20(4):327-48. Spontaneous brain activity in the newborn brain
10 Bruer J. Neural connections: some you use, some during natural sleep--an fMRI study in infants
you lose. The myth of the first three years. Simon born at full term. Pediatr Res 2009
& Schuster, Free Press.; 1999. September;66(3):301-5.
11 Huttenlocher PR, Dabholkar AS. Regional 22 Smyser CD, Snyder AZ, Neil JJ. Functional
differences in synaptogenesis in human cerebral connectivity MRI in infants: exploration of the
cortex. J Comp Neurol 1997 October functional organization of the developing brain.
20;387(2):167-78. Neuroimage 2011 June 1;56(3):1437-52.
12 Petanjek Z, Judas M, Simic G, Rasin MR, Uylings 23 Stevens MC, Pearlson GD, Calhoun VD. Changes
HB, Rakic P, Kostovic I. Extraordinary neoteny of in the interaction of resting-state neural
c ha pt er 8
synaptic spines in the human prefrontal cortex. networks from adolescence to adulthood. Hum
Proc Natl Acad Sci U S A 2011 August Brain Mapp 2009 August;30(8):2356-66.
9;108(32):13281-6. 24 Dosenbach NU, Nardos B, Cohen AL, Fair DA,
13 Rakic P, Bourgeois JP, Eckenhoff MF, Zecevic N, Power JD, Church JA, Nelson SM, Wig GS, Vogel
Goldman-Rakic PS. Concurrent overproduction AC, Lessov-Schlaggar CN, Barnes KA, Dubis JW,
of synapses in diverse regions of the primate Feczko E, Coalson RS, Pruett JR, Jr., Barch DM,
cerebral cortex. Science 1986 April Petersen SE, Schlaggar BL. Prediction of
11;232(4747):232-5. individual brain maturity using fMRI. Science
14 Collin G, Sporns O, Mandl RC, van den Heuvel 2010 September 10;329(5997):1358-61.
MP. Structural and Functional Aspects Relating 25 Fair DA, Cohen AL, Power JD, Dosenbach NU,
to Cost and Benefit of Rich Club Organization in Church JA, Miezin FM, Schlaggar BL, Petersen SE.
Functional brain networks develop from a “local 36 Beaulieu C, Allen PS. Water diffusion in the giant
to distributed” organization. PLoS Comput Biol axon of the squid: implications for diffusion-
2009 May;5(5):e1000381. weighted MRI of the nervous system. Magn Reson
26 Fair DA, Dosenbach NU, Church JA, Cohen AL, Med 1994 November;32(5):579-83.
Brahmbhatt S, Miezin FM, Barch DM, Raichle ME, 37 Jones DK. Studying connections in the living
Petersen SE, Schlaggar BL. Development of human brain with diffusion MRI. Cortex 2008
distinct control networks through segregation September;44(8):936-52.
and integration. Proc Natl Acad Sci U S A 2007 38 Mori S, van Zijl PC. Fiber tracking: principles and
August 14;104(33):13507-12. strategies - a technical review. NMR Biomed 2002
27 Power JD, Fair DA, Schlaggar BL, Petersen SE. The November;15(7-8):468-80.
development of human functional brain 39 Power JD, Barnes KA, Snyder AZ, Schlaggar BL,
networks. Neuron 2010 September 9;67(5):735- Petersen SE. Spurious but systematic correlations
48. in functional connectivity MRI networks arise
28 Adams E, Chau V, Poskitt KJ, Grunau RE, Synnes A, from subject motion. Neuroimage 2012 February
Miller SP. Tractography-based quantitation of 1;59(3):2142-54.
corticospinal tract development in premature 40 Satterthwaite TD, Wolf DH, Loughead J, Ruparel
newborns. J Pediatr 2010 June;156(6):882-8, K, Elliott MA, Hakonarson H, Gur RC, Gur RE.
888. Impact of in-scanner head motion on multiple
29 Berman JI, Mukherjee P, Partridge SC, Miller SP, measures of functional connectivity: relevance
Ferriero DM, Barkovich AJ, Vigneron DB, Henry for studies of neurodevelopment in youth.
RG. Quantitative diffusion tensor MRI fiber Neuroimage 2012 March;60(1):623-32.
tractography of sensorimotor white matter 41 Van Dijk KR, Sabuncu MR, Buckner RL. The
development in premature infants. Neuroimage influence of head motion on intrinsic functional
2005 October 1;27(4):862-71. connectivity MRI. Neuroimage 2012 January
30 Deipolyi AR, Mukherjee P, Gill K, Henry RG, 2;59(1):431-8.
Partridge SC, Veeraraghavan S, Jin H, Lu Y, Miller 42 van den Heuvel MP, Mandl RC, Stam CJ, Kahn RS,
SP, Ferriero DM, Vigneron DB, Barkovich AJ. Hulshoff Pol HE. Aberrant frontal and temporal
Comparing microstructural and macrostructural complex network structure in schizophrenia: a
development of the cerebral cortex in graph theoretical analysis. J Neurosci 2010
premature newborns: diffusion tensor imaging November 24;30(47):15915-26.
versus cortical gyration. Neuroimage 2005 43 Bassett DS, Brown JA, Deshpande V, Carlson JM,
September;27(3):579-86. Grafton ST. Conserved and variable architecture
31 Woodward LJ, Anderson PJ, Austin NC, Howard K, of human white matter connectivity. Neuroimage
Inder TE. Neonatal MRI to predict 2011 January 15;54(2):1262-79.
neurodevelopmental outcomes in preterm 44 Eguiluz VM, Chialvo DR, Cecchi GA, Baliki M,
infants. N Engl J Med 2006 August 17;355(7):685- Apkarian AV. Scale-free brain functional
94. networks. Phys Rev Lett 2005 January
32 Oishi K, Mori S, Donohue PK, Ernst T, Anderson L, 14;94(1):018102.
Buchthal S, Faria A, Jiang H, Li X, Miller MI, van Zijl 45 van den Heuvel MP, Stam CJ, Boersma M,
PC, Chang L. Multi-contrast human neonatal Hulshoff Pol HE. Small-world and scale-free
brain atlas: application to normal neonate organization of voxel-based resting-state
development analysis. Neuroimage 2011 May functional connectivity in the human brain.
1;56(1):8-20. Neuroimage 2008 November 15;43(3):528-39.
33 Papile LA, Burstein J, Burstein R, Koffler H. 46 van den Heuvel MP, Sporns O. An anatomical
ch a pt e r 8
Incidence and evolution of subependymal and substrate for integration among functional
intraventricular hemorrhage: a study of infants networks in human cortex. J Neurosci 2013
with birth weights less than 1,500 gm. J Pediatr September 4;33(36):14489-500.
1978 April;92(4):529-34. 47 Zalesky A, Fornito A, Harding IH, Cocchi L, Yucel
34 Andersson JL, Skare S. A model-based method for M, Pantelis C, Bullmore ET. Whole-brain
retrospective correction of geometric distortions anatomical networks: does the choice of nodes
in diffusion-weighted EPI. Neuroimage 2002 matter? Neuroimage 2010 April 15;50(3):970-83.
May;16(1):177-99. 48 de Reus MA, van den Heuvel MP. The
35 Chang LC, Jones DK, Pierpaoli C. RESTORE: robust parcellation-based connectome: limitations and
estimation of tensors by outlier rejection. Magn extensions. Neuroimage 2013 October
Reson Med 2005 May;53(5):1088-95. 15;80:397-404.

49 van den Heuvel MP, Sporns O, Collin G, Scheewe Neuroimage 2012 October 1;62(4):2296-314.
T, Mandl RC, Cahn W, Goni J, Hulshoff Pol HE, 64 Ratnarajah N, Rifkin-Graboi A, Fortier MV, Chong
Kahn RS. Abnormal rich club organization and YS, Kwek K, Saw SM, Godfrey KM, Gluckman PD,
functional brain dynamics in schizophrenia. JAMA Meaney MJ, Qiu A. Structural connectivity
Psychiatry 2013 August;70(8):783-92. asymmetry in the neonatal brain. Neuroimage
50 van den Heuvel MP, van Soelen IL, Stam CJ, Kahn 2013 July 15;75:187-94.
RS, Boomsma DI, Hulshoff Pol HE. Genetic 65 Tymofiyeva O, Hess CP, Ziv E, Tian N, Bonifacio SL,
control of functional brain network efficiency in McQuillen PS, Ferriero DM, Barkovich AJ, Xu D.
children. Eur Neuropsychopharmacol 2013 Towards the “baby connectome”: mapping the
January;23(1):19-23. structural connectivity of the newborn brain.
51 Rubinov M, Sporns O. Complex network PLoS One 2012;7(2):e31029.
measures of brain connectivity: uses and 66 Fransson P, Aden U, Blennow M, Lagercrantz H.
interpretations. Neuroimage 2010 The functional architecture of the infant brain as
September;52(3):1059-69. revealed by resting-state fMRI. Cereb Cortex 2011
52 Rubinov M, Sporns O. Weight-conserving January;21(1):145-54.
characterization of complex functional brain 67 Omidvarnia A, Fransson P, Metsaranta M,
networks. Neuroimage 2011 June 15;56(4):2068- Vanhatalo S. Functional Bimodality in the Brain
79. Networks of Preterm and Term Human
53 de Reus MA, van den Heuvel MP. Estimating false Newborns. Cereb Cortex 2013 May 5.
positives and negatives in brain networks. 68 Gong G, He Y, Concha L, Lebel C, Gross DW, Evans
Neuroimage 2013 April 15;70:402-9. AC, Beaulieu C. Mapping anatomical connectivity
54 Maslov S, Sneppen K. Specificity and stability in patterns of human cerebral cortex using in vivo
topology of protein networks. Science 2002 May diffusion tensor imaging tractography. Cereb
3;296(5569):910-3. Cortex 2009 March;19(3):524-36.
55 Humphries MD, Gurney K, Prescott TJ. The 69 Bonifacio SL, Glass HC, Chau V, Berman JI, Xu D,
brainstem reticular formation is a small-world, Brant R, Barkovich AJ, Poskitt KJ, Miller SP,
not scale-free, network. Proc Biol Sci 2006 Ferriero DM. Extreme premature birth is not
February 22;273(1585):503-11. associated with impaired development of brain
56 Newman ME. Modularity and community microstructure. J Pediatr 2010
structure in networks. Proc Natl Acad Sci U S A November;157(5):726-32.
2006 June 6;103(23):8577-82. 70 Huppi PS, Maier SE, Peled S, Zientara GP, Barnes
57 Colizza V, Flammini A, Serrano MA, Vespignani A. PD, Jolesz FA, Volpe JJ. Microstructural
Detecting rich-club ordering in complex development of human newborn cerebral white
networks. Nat Phys 2006;2:110-5. matter assessed in vivo by diffusion tensor
58 Zamora-Lopez G, Zhou C, Kurths J. Cortical hubs magnetic resonance imaging. Pediatr Res 1998
form a module for multisensory integration on October;44(4):584-90.
top of the hierarchy of cortical networks. Front 71 Neil JJ, Shiran SI, McKinstry RC, Schefft GL, Snyder
Neuroinform 2010;4:1. AZ, Almli CR, Akbudak E, Aronovitz JA, Miller JP,
59 Mantel N. The detection of disease clustering Lee BC, Conturo TE. Normal brain in human
and a generalized regression approach. Cancer newborns: apparent diffusion coefficient and
Res 1967 February;27(2):209-20. diffusion anisotropy measured by using diffusion
60 van den Heuvel MP, Sporns O. Network hubs in tensor MR imaging. Radiology 1998
the human brain. Trends Cogn Sci 2013 October;209(1):57-66.
December;17(12):683-96. 72 Mukherjee P, Miller JH, Shimony JS, Philip JV,
c ha pt er 8
61 Honey CJ, Sporns O, Cammoun L, Gigandet X, Nehra D, Snyder AZ, Conturo TE, Neil JJ,
Thiran JP, Meuli R, Hagmann P. Predicting human McKinstry RC. Diffusion-tensor MR imaging of
resting-state functional connectivity from gray and white matter development during
structural connectivity. Proc Natl Acad Sci U S A normal human brain maturation. AJNR Am J
2009 February 10;106(6):2035-40. Neuroradiol 2002 October;23(9):1445-56.
62 Bullmore E, Sporns O. Complex brain networks: 73 Schneiderman JS, Buchsbaum MS, Haznedar MM,
graph theoretical analysis of structural and Hazlett EA, Brickman AM, Shihabuddin L, Brand
functional systems. Nat Rev Neurosci 2009 JG, Torosjan Y, Newmark RE, Tang C, Aronowitz J,
March;10(3):186-98. Paul-Odouard R, Byne W, Hof PR. Diffusion
63 Fornito A, Zalesky A, Pantelis C, Bullmore ET. tensor anisotropy in adolescents and adults.
Schizophrenia, neuroimaging and connectomics. Neuropsychobiology 2007;55(2):96-111.
74 Damoiseaux JS, Rombouts SA, Barkhof F, diffusion tensor and T(1) imaging. Neuroimage
Scheltens P, Stam CJ, Smith SM, Beckmann CF. 2013 January 1;64:505-16.
Consistent resting-state networks across healthy 85 Alexander DC, Hubbard PL, Hall MG, Moore EA,
subjects. Proc Natl Acad Sci U S A 2006 September Ptito M, Parker GJ, Dyrby TB. Orientationally
12;103(37):13848-53. invariant indices of axon diameter and density
75 Salvador R, Suckling J, Coleman MR, Pickard JD, from diffusion MRI. Neuroimage 2010 October
Menon D, Bullmore E. Neurophysiological 1;52(4):1374-89.
architecture of functional magnetic resonance 86 Jbabdi S, Johansen-Berg H. Tractography: where
images of human brain. Cereb Cortex 2005 do we go from here? Brain Connect
September;15(9):1332-42. 2011;1(3):169-83.
76 van den Heuvel MP, Mandl RC, Kahn RS, Hulshoff 87 Johansen-Berg H, Rushworth MF. Using diffusion
Pol HE. Functionally linked resting-state imaging to study human connectional anatomy.
networks reflect the underlying structural Annu Rev Neurosci 2009;32:75-94.
connectivity architecture of the human brain. 88 Goulas A, Uylings HB, Stiers P. Mapping the
Hum Brain Mapp 2009 October;30(10):3127-41. hierarchical layout of the structural network of
77 Zuo XN, Ehmke R, Mennes M, Imperati D, the macaque prefrontal cortex. Cereb Cortex
Castellanos FX, Sporns O, Milham MP. Network 2014 May;24(5):1178-94.
centrality in the human functional connectome. 89 Harriger L, van den Heuvel MP, Sporns O. Rich
Cereb Cortex 2012 August;22(8):1862-75. club organization of macaque cerebral cortex
78 Dosenbach NU, Fair DA, Miezin FM, Cohen AL, and its role in network communication. PLoS One
Wenger KK, Dosenbach RA, Fox MD, Snyder AZ, 2012;7(9):e46497.
Vincent JL, Raichle ME, Schlaggar BL, Petersen SE. 90 Shi F, Yap PT, Gao W, Lin W, Gilmore JH, Shen D.
Distinct brain networks for adaptive and stable Altered structural connectivity in neonates at
task control in humans. Proc Natl Acad Sci U S A genetic risk for schizophrenia: a combined study
2007 June 26;104(26):11073-8. using morphological and white matter networks.
79 Fair DA, Cohen AL, Dosenbach NU, Church JA, Neuroimage 2012 September;62(3):1622-33.
Miezin FM, Barch DM, Raichle ME, Petersen SE, 91 Fornito A, Zalesky A, Bassett DS, Meunier D,
Schlaggar BL. The maturing architecture of the Ellison-Wright I, Yucel M, Wood SJ, Shaw K,
brain’s default network. Proc Natl Acad Sci U S A O’Connor J, Nertney D, Mowry BJ, Pantelis C,
2008 March 11;105(10):4028-32. Bullmore ET. Genetic influences on cost-efficient
80 Peper JS, Schnack HG, Brouwer RM, Van Baal GC, organization of human cortical functional
Pjetri E, Szekely E, van LM, van den Berg SM, networks. J Neurosci 2011 March 2;31(9):3261-
Collins DL, Evans AC, Boomsma DI, Kahn RS, 70.
Hulshoff Pol HE. Heritability of regional and 92 Glahn DC, Winkler AM, Kochunov P, Almasy L,
global brain structure at the onset of puberty: a Duggirala R, Carless MA, Curran JC, Olvera RL,
magnetic resonance imaging study in 9-year-old Laird AR, Smith SM, Beckmann CF, Fox PT,
twin pairs. Hum Brain Mapp 2009 Blangero J. Genetic control over the resting
July;30(7):2184-96. brain. Proc Natl Acad Sci U S A 2010 January
81 Catani M, Thiebaut de SM, Slater D, Dell’Acqua F. 19;107(3):1223-8.
Connectomic approaches before the 93 Price DJ, Ferrer JM. The incidence of bifurcation
connectome. Neuroimage 2013 October 15;80:2- among corticocortical connections from area 17
13. in the developing visual cortex of the cat. Eur J
82 LaMantia AS, Rakic P. Axon overproduction and Neurosci 1993 March 1;5(3):223-31.
elimination in the corpus callosum of the 94 Sporns O. The human connectome: a complex
ch a pt e r 8
developing rhesus monkey. J Neurosci 1990 network. Ann N Y Acad Sci 2011
July;10(7):2156-75. April;1224:109-25.
83 LaMantia AS, Rakic P. Axon overproduction and 95 Sporns O. Discovering the human connectome.
elimination in the anterior commissure of the Cambridge: MIT Press; 2012.
developing rhesus monkey. J Comp Neurol 1994 96 de Lange SC, de Reus MA, van den Heuvel MP.
February 15;340(3):328-36. The Laplacian spectrum of neural networks. Front
84 Nossin-Manor R, Card D, Morris D, Comput Neurosci 2014 January 13;7:189.
Noormohamed S, Shroff MM, Whyte HE, Taylor 97 Markov NT, Ercsey-Ravasz M, Van Essen DC,
MJ, Sled JG. Quantitative MRI in the very preterm Knoblauch K, Toroczkai Z, Kennedy H. Cortical
brain: assessing tissue organization and high-density counterstream architectures.
myelination using magnetization transfer, Science 2013 November 1;342(6158):1238406.

98 Park HJ, Friston K. Structural and functional brain Hulshoff Pol HE, Agosta F, Comi G, Rocca MA.
networks: from connections to cognition. Science Assessment of system dysfunction in the brain
2013 November 1;342(6158):1238411. through MRI-based connectomics. Lancet Neurol
99 Bassett DS, Bullmore ET, Meyer-Lindenberg A, 2013 December;12(12):1189-99.
Apud JA, Weinberger DR, Coppola R. Cognitive 108 Boersma M, Kemner C, de Reus MA, Collin G,
fitness of cost-efficient brain functional Snijders TM, Hofman D, Buitelaar JK, Stam CJ, van
networks. Proc Natl Acad Sci U S A 2009 July den Heuvel MP. Disrupted functional brain
14;106(28):11747-52. networks in autistic toddlers. Brain Connect
100 Jbabdi S, Behrens TE. Long-range connectomics. 2013;3(1):41-9.
Ann N Y Acad Sci 2013 December;1305:83-93. 109 Dennis EL, Thompson PM. Typical and atypical
101 van den Heuvel MP, Hulshoff Pol HE. Exploring brain development: a review of neuroimaging
the brain network: a review on resting-state fMRI studies. Dialogues Clin Neurosci 2013
functional connectivity. Eur September;15(3):359-84.
Neuropsychopharmacol 2010 110 van den Heuvel MP, Fornito A. Brain networks in
August;20(8):519-34. schizophrenia. Neuropsychol Rev 2014
102 Honey CJ, Sporns O. Dynamical consequences of March;24(1):32-48.
lesions in cortical networks. Hum Brain Mapp 111 Pandit AS, Robinson E, Aljabar P, Ball G, Gousias
2008 July;29(7):802-9. IS, Wang Z, Hajnal JV, Rueckert D, Counsell SJ,
103 Senden M, Deco G, de Reus MA, Goebel R, van Montana G, Edwards AD. Whole-Brain Mapping
den Heuvel MP. Rich club organization supports of Structural Connectivity in Infants Reveals
a diverse set of functional network Altered Connection Strength Associated with
configurations. Neuroimage 2014 August Growth and Preterm Birth. Cereb Cortex 2013
1;96:174-82. March 31.
104 Jahanshad N, Rajagopalan P, Hua X, Hibar DP, Nir 112 Aarnoudse-Moens CS, Weisglas-Kuperus N,
TM, Toga AW, Jack CR, Jr., Saykin AJ, Green RC, Duivenvoorden HJ, van Goudoever JB, Oosterlaan
Weiner MW, Medland SE, Montgomery GW, J. Executive function and IQ predict
Hansell NK, McMahon KL, de Zubicaray GI, mathematical and attention problems in very
Martin NG, Wright MJ, Thompson PM. Genome- preterm children. PLoS One 2013;8(2):e55994.
wide scan of healthy human connectome 113 Mulder H, Pitchford NJ, Hagger MS, Marlow N.
discovers SPON1 gene variant influencing Development of executive function and
dementia severity. Proc Natl Acad Sci U S A 2013 attention in preterm children: a systematic
March 19;110(12):4768-73. review. Dev Neuropsychol 2009;34(4):393-421.
105 Bassett DS, Bullmore ET. Human brain networks 114 Liu X, Pillay S, Li R, Vizuete JA, Pechman KR,
in health and disease. Curr Opin Neurol 2009 Schmainda KM, Hudetz AG. Multiphasic
August;22(4):340-7. modification of intrinsic functional connectivity
106 Buckner RL, Sepulcre J, Talukdar T, Krienen FM, of the rat brain during increasing levels of
Liu H, Hedden T, Andrews-Hanna JR, Sperling RA, propofol. Neuroimage 2013
Johnson KA. Cortical hubs revealed by intrinsic December;83:581-92.
functional connectivity: mapping, assessment of 115 Fransson P, Skiold B, Horsch S, Nordell A,
stability, and relation to Alzheimer’s disease. J Blennow M, Lagercrantz H, Aden U. Resting-state
Neurosci 2009 February 11;29(6):1860-73. networks in the infant brain. Proc Natl Acad Sci U
107 Filippi M, van den Heuvel MP, Fornito A, He Y, S A 2007 September 25;104(39):15531-6.
c ha pt er 8
chapter 8

ch a p t er
9
Summarizing
discussion,
conclusions and
directions
for future
research
191
T
o unravel the pathophysiology underlying the large percentage of preterm born infants
that will demonstrate neurodevelopmental impairments during childhood, a better
understanding of brain development during what would have been the third trimester
of pregnancy is needed. With the increasing survival of even the most extremely preterm
infants, the burden for both the parents and immediate environment as well as for the society
needs to be considered, and improvements in outcome would not only benefit the child, but
also the society as a whole.1,2 Identifying deviations in brain development, taking place while
preterm infants are cared for in a neonatal intensive care environment with all its stressors and
complications, may help in finding better predictive strategies and protective therapies to
improve cognitive and behavioural outcome for this vulnerable population. The aim of this
thesis was to study longitudinal brain development in extremely preterm infants, utilizing
serial MRI scans of preterm infants born before a gestation of 28 weeks.
In Chapter 1 an introduction to preterm birth is given and the most important patterns of
preterm brain injury are described. Research describing visual evaluation of conventional brain
imaging has helped in recognizing the patterns of injury in preterm infants as well as the
differences in severity. Correlations with outcome have been made and especially for motor
outcome, MRI seems to be a reliable predictor.3 Over time, injury patterns have changed and
severe white matter injury is fortunately becoming rare.4,5 Combined with the increased use of
MRI around term equivalent age, different patterns of white matter injury are nowadays
identified, and punctate white matter lesions (PWML) are among the most commonly
identified lesions, occurring in 20 to 50% of preterm infants.6 PWML can only reliably be
diagnosed on MRI and identification of these lesions is largely dependent on image quality. In
Chapter 2 we described the different patterns of PWML that could be seen in a cohort of
infants with serial scans in the neonatal period, supported by the additional use of diffusion
(DWI) and susceptibility (SWI) weighted imaging. Two different patterns were identified with
different findings on DWI and SWI. The linear pattern consisted of lesions in a linear form,
most often in close proximity to the ventricles, which did also show up on SWI but not on DWI.
These lesions are therefore thought to be haemorrhagic in origin. The cluster pattern
consisted of more solitary placed, somewhat larger and more rounded lesions, located deeper
in the white matter and not visible on SWI. On early imaging, when the scan was made within
10 days after birth, these lesions could be identified on DWI, suggestive of an ischaemic origin.
No pathology samples are available to confirm our findings, but these different patterns likely
represent different types of injury and care should therefore be taken when these patterns are
grouped with regard to outcome. Although it has been hypothesized that PWML may be
responsible for milder cognitive and behavioural deficits found at school age, we did not find a
relation with outcome up to two years of age.
These findings underline the limitations of predicting outcome by visual evaluation of
ch a pt e r 9
conventional imaging, even when performed in a systematic way.7 To be able to get a more
reliable prediction of cognitive and behavioural outcome as well as to get a better
understanding of the underlying pathophysiology, quantitative techniques are required to
study MRIs of preterm infants.
In Part 2 of the thesis, several of these quantitative techniques used on conventional T1- and
T2-weighted imaging are described. In Chapter 3 we describe the use of a recently developed
segmentation method, which divides the brain into 50 brain regions. Both the early scans and
scans at TEA could be segmented with this method, allowing us to study third trimester extra-
uterine volumetric brain growth. Growth data showed differences between regions, with the
cerebellum showing the largest (258%) and the ventricles the smallest (61%) increase in
volume over this time. Girls had smaller brains than boys and infants with lower birth weight
z-scores and those needing prolonged mechanical ventilation or surgery, showed a global
decrease in volumes at both scans. The effect of brain injury showed more localized effects,
with the ventricles already being larger at the first scan, while the decreased volume of the
cerebellum was only visible in the growth data and the volumes at TEA. These data highlight
the vulnerability of the preterm brain and also show that while some effects are already visible
at 30 weeks, others will develop over time.
There are more risk factors that have been described to influence brain volumes at TEA and
one of those is the use of corticosteroids for chronic lung disease. Dexamethasone, the drug
most widely used, has been shown to have a detrimental effect on both brain volumes at TEA
as well as outcome in childhood.8-11 These findings have led to lower and more careful dosage
schemes that may reduce these effects.12 However, there are alternatives available and in
Utrecht, hydrocortisone instead of dexamethasone has been the clinical standard for decades.
Hydrocortisone is said to be less potent, but has also shown fewer side effects and no effect on
neurodevelopmental outcome.13,14 In Chapter 4, we compared brain volumes at TEA between
infants receiving hydrocortisone and control infants matched for gestational age and sex.
Infants with larger parenchymal injury were excluded. After correcting for gestational age,
postmenstrual age at time of scan, birth weight z-score and the presence of IVH, no differences
could be found for both total brain volume and cerebellar volume between the infants
receiving hydrocortisone and control infants. Hydrocortisone therefore seems to be a safer
alternative than dexamethasone for the treatment of a developing chronic lung disease.
Apart from studying the volumetric effects on brain growth and development, and the
influence of clinical risk factors, volumetric segmentations can also be used as input data for
further quantitative analysis, such as the creation of white matter meshes to study cortical
folding. The majority of gyrification and sulcation takes place during the third trimester of
pregnancy and the development of these folds has been studied with MRI, showing the order
of development,15-17 influences of clinical characteristics such as intra-uterine growth
retardation, twin pregnancy, sex, gestational age at birth, and asymmetries in sulcal
development.18,19 Chapter 5 describes the use of a dedicated pipeline that allows identification
ch a p te r 9
of specific folds at both early and TEA scans. This made it possible to study the development of
these specific sulci in our population and to correlate these measures with both clinical
summarizing discussion , conclusions and directions for future research 193

characteristics and outcome at two years of age. Seven different regions were studied: the
central sulcus (CS), lateral fissure (LF), insula (INS), superior temporal sulcus (STS), postcentral
sulcus (PCS), superior frontal sulcus (SFS) and inferior frontal sulcus (IFS). The three most
central sulci (CS, LF and INS) were already visible on the first MRI in all infants, while the other
four were less well developed and could not yet be identified around 30 weeks of gestation.
The relative growth of these four sulci was larger in the period between both scans, with the
SFS showing the largest increase in surface area. The well-known right-ward asymmetry in
brain development was evident in the data at 30 weeks. At TEA, right-left differences were
most clear for the STS (right>left) and the LF (left>right), in accordance with literature on
older subjects.20-22 Multiple pregnancy, a lower birth weight z-score and prolonged mechanical
ventilation were correlated with a delay in cortical folding, more pronounced on the early than
on the TEA scan. The sulci that develop earliest were most affected by these factors. A positive
association with motor, cognitive and language outcome at 2-2.5 years’ corrected age was
found, suggesting that cortical folding is an important determinant of neurodevelopmental
outcome. Maternal education showed a strong association with outcome as well. Maternal
education represents a combination of genetic influences and socio-economic status of the
environment the child grows up in, and should therefore invariably be considered when
studying neurodevelopmental outcome.
Next to conventional imaging, diffusion weighted and functional MR imaging are increasingly
used to study brain development and the deviations occurring in prematurity in more detail.
Some of these results are described in Part 3 of this thesis. DWI, and more specifically the often
used diffusion tensor imaging (DTI), allows the visualization of the Brownian motion of water
molecules within the brain, providing information about microstructural integrity and
maturation. By studying the different diffusion parameters, more insight about brain
development can be obtained. In Chapter 6, we describe diffusion measurements estimated
with an automated, atlas-based approach covering the entire brain at both 30 and 40 weeks,
as well as the growth in between. For this study, 40 infants with good quality serial DTI data
and a normal neurodevelopmental outcome at 15 months were selected, to mimic normal
brain development as much as possible. Fractional anisotropy (FA) showed an increase with a
central to peripheral and posterior to anterior directed gradient, with corresponding
decreases in mean, radial and axial diffusivity. Cortical regions showed no increase or a
decrease in FA. These findings are in line with the current understanding about brain
maturation, first described in histological studies and later confirmed in vivo.23-25 The posterior
to anterior directed gradient could also be noted in the results of Chapter 4, where the
occipital regions showed a slightly larger increase compared to the frontal regions. The
central-peripheral gradient is similar to the findings in Chapter 5, where the central sulci were
the first to develop. The diffusivity values of this study can be used as a reference for future
ch a pt e r 9
studies assessing brain development in other cohorts, or in studies evaluating brain injury and
the effects of neuroprotective strategies.
Besides the atlas-based approach towards DTI data used in Chapter 6, multiple other ways of
assessing DTI data are available. Techniques most often used include manual region of interest
(ROI) analysis, tract based spatial statistics (TBSS), and both deterministic and probabilistic
tractography.26 Atlas-based and to a lesser extent ROI-based approaches are valuable for
assessing the overall brain or larger regions. TBSS is an effective way of determining differences
between cohorts or cases and controls and shows those voxels that differ significantly.27
Tractography can be done both for the whole brain and subsequently be used as an input for
connectivity measures, or can be done to study diffusivity values of specific fibre bundles such
as the corticospinal tract (CST). In Chapter 7, we used probabilistic tractography of the CST to
show differences between preterm infants with cystic periventricular leukomalacia (c-PVL)
and preterm controls without brain injury, for both FA values of the CST and thalamic volumes.
c-PVL is one of the most severe types of white matter injury and the leading cause of cerebral
palsy (CP) after preterm birth.4,5 In children with CP after c-PVL, both the CST and thalamus
were affected when those children were scanned in childhood and compared to normal
developing controls.28,29 However, it is still not clear when those differences exactly develop.
We now showed that injury to the CST can already be identified on the first scan, performed
within weeks after the insult. On the contrary, thalamic volumes are not yet affected at the first
scan, but show a clear decrease in growth with lower volumes around TEA. These data fit with
the hypothesis that the cysts, which often are directly in the path of the CST, will directly affect
the CST and other white matter bundles. As a consequence, the afferent and efferent inputs to
the thalamus will be affected and thereby hinder normal thalamic development, leading to
smaller thalamic volumes over the course of several weeks.30 The thalamus is known to be an
important factor for outcome and this was also true for this study, in which the severity of CP
was correlated with thalamic volumes at TEA.31
This relation between structural differences or impairments and functional outcome is an area
of active research. Over the last couple of years, resting state functional MR imaging (rs-fMRI)
has become available for use in the neonatal population.32,33 The combination of DTI and rs-
fMRI has allowed the compilation of the human connectome, in other words allowed a
comparison of the structural wiring of the human brain to neural activity. Chapter 8 describes
connectome formation in 27 neonates with longitudinal scans, including DTI and rs-fMRI of
sufficient quality at both time points. The DTI scans showed small-world organization to be
clearly present already at the first scan, with the formation of immature resting-state networks
found on fMRI. Structural and functional connectivity networks found at 30 weeks showed
large similarities to adult brain architecture, with over 80% overlap between both. The
networks found early developed towards TEA in microstructure, small-world topology and also
interhemispheric functional connectivity, resulting in an increase in integration capacity of the
connectivity network as a whole. The basic structure of the human connectome thus seems to
ch a p te r 9
develop very early on, as it is already clearly visible around 30 weeks gestation. However, the
development between 30 and 40 weeks as found in this study suggests once again that active

brain development in the last trimester of pregnancy may be disrupted by preterm birth.
These functional and to a lesser extent also the structural connections are not discernible with
the human eye and only come to light using advanced MRI techniques. Conventional imaging
has been inadequate to predict cognitive and behavioural outcome, exactly those domains
that are impaired in almost half of the preterm infants. These data suggest that the negative
consequences of preterm birth on neurodevelopmental outcome may well be –partly– caused
by deviations in normal development of the human connectome. Correlating these data to
neurodevelopmental outcome therefore seems of the utmost importance, as is true for all the
chapters of this thesis.
Conclusions
The following conclusions can be drawn from this thesis:
- Punctate white matter lesions can be divided into two distinct patterns, suggestive of a
difference in underlying pathophysiology. An early MRI including DWI and SWI
sequences is required to reliably identify these patterns (chapter 2).
- Volumetric brain growth differs regionally and is largest for the cerebellum. Female sex,
a lower birth weight z-score and prolonged mechanical ventilation negatively influence
brain volumes. The effect of brain injury is regional and most pronounced in the
ventricles at 30 weeks, while the cerebellum shows reduced growth and smaller
volumes at TEA after brain injury (chapter 3).
- Treatment with hydrocortisone for developing chronic lung disease does not cause
reductions in overall brain size or cerebellar volume at TEA in preterm infants without
parenchymal brain injury (chapter 4).
- Cortical folding in preterm infants occurs in a central to occipital and frontal direction,
with the right hemisphere developing slightly before the left. Cortical depth and
surface area are smaller in infants from a multiple pregnancy, with a lower birth weight
z-score, and after prolonged mechanical ventilation and show a correlation with
outcome at two years, including linguistic scales (chapter 5).
- Longitudinal diffusivity values of the preterm brain show a central-peripheral and
occipital-frontal directed gradient. While fractional anisotropy increases in the white
matter, it decreases in the cortex. Mean, radial and axial diffusivity show a decrease
across the brain (chapter 6).
- Damage to the corticospinal tracts in infants with c-PVL occurs within weeks after the
insult, while the remote effects on thalamic volume develop over time and become
clear at TEA (chapter 7).
- Immature functional networks are already present at 30 weeks and are similar to adult
networks. Both structural and functional connectivity networks develop between 30
ch a pt e r 9
weeks and TEA leading to an increase in integration capacity (chapter 8).

- The central-peripheral and occipital-frontal gradient of brain development can be
confirmed with multiple MRI techniques (chapter 3, chapter 5, chapter 6, chapter 8).
- Prolonged mechanical ventilation, whether as a representative of lung damage or of
overall severity of illness, has a negative influence on brain development (chapter 3,
chapter 5).
- Neurodevelopmental outcome around two years of age is probably too early to find
minor deficits caused by mild brain injury (chapter 2, chapter 6).
- The remote effects of brain injury will become evident in the period between 30 and 40
weeks gestation, whereas the direct effects are already established by 30 weeks
(chapter 3, chapter 7).
Directions for future research

With the increased use of MRI in preterm born infants, large steps have been made in
unravelling preterm brain development and the deviations from term born infants. However,
about half of the extremely preterm infants will still develop neurodevelopmental difficulties
later in life and the precise mechanisms remain poorly understood. In this thesis, we described
visual analysis of conventional imaging, the use of advanced post-processing techniques on
conventional imaging and the use of diffusion weighted and functional imaging in serial data
of extremely preterm infants. We have shown that longitudinal imaging will improve our
knowledge of brain development and affirmed the importance of early imaging to determine
the order and timing of developmental processes. In chapter 8, we show that the relation
between structure and function can give important new insights into the consequences of the
structural deviations found in the previous chapters.
This thesis contains only a small part of the increasing amount of information becoming
available about brain development and its deviations due to preterm birth. Many questions
remain and a large amount of further studies will be necessary to unravel this intricate
problem. Several thoughts on possibilities for future research are suggested below.
Foeta l MR I
Studying preterm born infants will always introduce a possible bias and truly normal brain
development over this period should be studied in utero. Improving the quality and quantity
of foetal MRI research is therefore essential. Normal development of cortical folding has been
described in vivo using foetal MRI, although 3D measurements are largely lacking.16,34-36 A few
studies using foetal DTI and even foetal functional imaging have been published, but sufficient
quality remains a problem.37-39 Developing further 3D measurements on foetal MRI as well as
increasing the use and usability of both diffusion weighted and functional imaging may further
enhance our knowledge about ‘true’ normal brain development.
ch a p te r 9

Ce l l u l a r d e v elo p m e nt
Besides research utilizing MRI, more insight on a cellular level will also result in important
additional knowledge. Both human and experimental studies can be used to show the
response of different cell types to preterm birth.30,40 Especially the late oligodendrocyte
progenitor cells are an important target, as they are most affected by injury after preterm
birth. This may lead to a cascade of other disturbances in normal cerebral growth and
maturation seen after preterm birth.40,41 A better understanding of the cellular basis of white
matter injury, such as the different patterns of punctate white matter lesions described in
chapter 2 or the thalamic involvement in infants with c-PVL described in chapter 7, may also
help in understanding the differences in outcome between individual infants and may give
further insight into possible preventive strategies.
R e l ati on b et w e e n b r ai n s t r u ct u re and f unct io n

As chapter 8 of this thesis already describes, an important next step in unravelling the
disturbances caused by preterm birth will be to correlate structural development and its
deviations with brain function. The combination of advanced diffusion weighted imaging and
functional MRI in larger cohorts, especially when studied longitudinally and/or in correlation
with long-term neurodevelopmental outcome, will give more insight into whether the
deviations found after preterm birth are actually responsible for impairments in outcome later
in life. Advanced DWI, such as the use of higher order models to adjust for crossing fibres and
to better visualize non-dominant tracts, the acquisition of high-angular resolution DWI,
neurite orientation dispersion and density imaging or diffusion spectrum imaging, and the use
of apparent diffusion coefficient modelling such as diffusion kurtosis imaging, will provide a
more detailed understanding of microstructural development.42,43 Resting state functional MRI
on the other hand measures neural activity in the brain and can thus be used to study loss and
gain of function in the developing brain and the presence and development of resting state
networks.43,44 Deviations in normal neural function have already been shown in preterm
infants with periventricular haemorrhagic infarction and depended upon injury severity.45,46
Longitudinal use of these techniques may also help in elucidating whether the remote effects,
that seem to develop over the last trimester (as shown after brain injury in chapters 3 and 7),
are presenting us with a window of opportunity for intervention or whether the damage is
done and this period is merely needed for the impairments to become significant.
Both functional and diffusion imaging can be used in the study of human connectome
formation. Whole-brain connectome approaches integrate the distribution of neural systems,
either structurally and/or functionally, and will allow for further understanding of brain
development and plasticity after injury. Studies done so far have shown diminished
ch a pt e r 9
thalamocortical integration47 and reduction of connection strength48 in preterm born infants

and this will likely extend further when larger cohorts become available. Another advantage of
whole-brain approaches is that studies can integrate all brain regions and not only focus on
grey and white matter. The cerebellum and brain stem are nowadays recognized as important
contributors to normal development as well and a further focus on integration between the
different parts of the brain is necessary.49,50 Additionally, these imaging techniques can be
combined with or compared to other measures of brain development, such as near-infrared
spectroscopy, positron emission tomography or MR spectroscopy, further integrating our
knowledge on brain structure and function.
Neu r op r ote cti v e s t r at e gie s

While a better understanding of why the preterm brain is different will greatly help in
elucidating underlying mechanisms, damage caused by preterm birth is not easily reversible.
We still have a long way to go before efficient therapies preventing or repairing brain injury
will be clinically available. Different targets and different moments in time should be
considered. First of all, preventing extremely preterm birth is probably the most effective way
of preventing its consequences and further research into the underlying reasons for preterm
birth remains important. Second, when preterm birth is apparent, preventive strategies may
be considered. The use of antenatal magnesium sulphate has for example been suggested to
correlate with reduced white matter injury in preterm infants, although the results of a
multicentre trial did not reach significance.51 Additionally, therapy given in the immediate
postnatal period may prevent white matter injury. A reduction in brain injury on MRI has been
shown when high-dose erythropoietin was administered within 42 hours after birth,52 and a
meta-analysis showed better neurodevelopmental scores in preterm infants at 18-22 months
after therapeutic recombinant human erythropoietin, without severe adverse side effects.53 A
correlation between prolonged administration of indomethacin and lower rates of white
matter injury has also been found, although this did not improve neurodevelopmental
outcome scores.54,55 Several nutritional interventions to optimize growth and brain
development during the neonatal period have also been tested.56 Increasing protein intakes
are correlated with a better growth over the neonatal period,57 and better growth is in turn
related to better performance at school age and may even continue to have an effect in
adolescence.58,59 Additionally, specific nutritional interventions may counteract inflammatory
effects and modulate immune responses.56,60
Third, once damage is done, repair mechanisms should be identified. The first steps towards
this goal have been made in experimental studies. Stimulating epidermal growth factor
receptor activation trough intranasally administered epidermal growth factor decreased
oligodendrocyte death and promoted oligodendrocyte regeneration and functional recovery
after diffuse white matter injury in a mouse model.61 Insulin-like growth factor, counteracting
the effect of glutamate, has been shown to increase oligodendrocyte survival and promote
myelination as well.62 Additionally, the use of stem cells may be a real possibility in the not too
ch a p te r 9
distant future. Experimental studies in term hypoxic-ischaemic brain injury have shown a
decrease in lesion volume and improvements in long-term motor and cognitive outcome after

administration of mesenchymal stem cells in mice.63,64 Although the aetiology in preterm
infants is different from those born at term, with more subacute and/or chronic insults,
hypoxia-ischaemia is one of the important underlying factors of preterm brain injury. Studies
into optimal administration therapies,65 as well as studies in larger primates to further test
safety and long term outcomes will be needed to determine the viability of stem cell therapy
in preterm infants.
Finally, stimulating brain development in preterm infants should not stop once they leave the
hospital. Studies have shown that behavioural intervention programs teaching effective
parenting techniques may have a positive effect on neurodevelopmental outcome up to five
years of age, although the effects do not seem to persist at school age so far.66,67 This does
however show that effective parenting is important and this is in agreement with our findings
in chapter 6 of the large influence of maternal education, in which parenting techniques may
well be reflected. Optimizing the child’s environment after discharge home may therefore be
an important strategy as well.
Lon g te r m neu r o d e v elo p m e nta l outco m e

Not many studies have described the correlation between neonatal MRI findings with
outcome in childhood, adolescence and beyond. With the increasing use of more
sophisticated techniques, it will become even more important to test if significant MRI
findings noted in the neonatal period will actually have long term neurodevelopmental
consequences. This type of research is complicated, since it will require great efforts from the
research team to track and follow the children all the way into adulthood. Also, by the time
preterm infants born today will reach adulthood, improvements in care and imaging
techniques may have led to differences between those ex-preterm adults and the preterm
born infants admitted at that moment. However, improvements in neonatal care over the last
decade have not led to an improvement in outcome and even older imaging data can
therefore still be very useful.68
Preterm infants are known to grow into their deficits and the results of this thesis have shown
that neurodevelopmental outcome at two years of age is too young to reliably predict
cognitive and behavioural impairments in the absence of severe brain injury. Deviations in
folding patterns and brain volumes have been shown to correlate with childhood psychiatric
disorders such as autism and attention deficit hyperactivity disorder, both known to occur
more frequently in survivors of preterm birth.69-71 MRI studies performed during childhood in
survivors of preterm birth also show clear correlations with neurodevelopmental outcome.72
An integration of neonatal findings and later follow-up may thus further elucidate the
underlying pathophysiology. Asking the infants of the cohort described in this thesis back to
perform neurodevelopmental assessment at 8-10 years of age in combination with a new MRI,
ch a pt e r 9
and ideally again around 15-17 years, may prove whether the findings of this thesis correlate
with long-term neurodevelopmental outcome.
References
1 Hodek JM, von der Schulenburg JM, Mittendorf T. prematurity. N Engl J Med 2004;350:1304-13.
Measuring economic consequences of preterm 12 Watterberg KL. Policy statement--postnatal
birth - Methodological recommendations for the corticosteroids to prevent or treat
evaluation of personal burden on children and bronchopulmonary dysplasia. Pediatrics
their caregivers. Health Econ Rev 2011;1:6. 2010;126:800-8.
2 Blencowe H, Cousens S, Oestergaard MZ, Chou 13 Rademaker KJ, Uiterwaal CS, Groenendaal F,
D, Moller AB, Narwal R et al. National, regional, Venema MM, Van Bel F, Beek FJ et al. Neonatal
and worldwide estimates of preterm birth rates hydrocortisone treatment: neurodevelopmental
in the year 2010 with time trends since 1990 for outcome and MRI at school age in preterm-born
selected countries: a systematic analysis and children. J Pediatr 2007;150:351-7.
implications. Lancet 2012;379:2162-72. 14 van der Heide-Jalving M, Kamphuis PJ, van der
3 de Vries LS, van Haastert IC, Benders MJ, Laan MJ, Bakker JM, Wiegant VM, Heijnen CJ et al.
Groenendaal F. Myth: cerebral palsy cannot be Short- and long-term effects of neonatal
predicted by neonatal brain imaging. Semin Fetal glucocorticoid therapy: is hydrocortisone an
Neonatal Med 2011;16:279-87. alternative to dexamethasone? Acta Paediatr
4 Hamrick SE, Miller SP, Leonard C, Glidden DV, 2003;92:827-35.
Goldstein R, Ramaswamy V et al. Trends in severe 15 Chi JG, Dooling EC, Gilles FH. Gyral development
brain injury and neurodevelopmental outcome of the human brain. Ann Neurol 1977;1:86-93.
in premature newborn infants: the role of cystic 16 Clouchoux C, Kudelski D, Gholipour A, Warfield
periventricular leukomalacia. J Pediatr SK, Viseur S, Bouyssi-Kobar M et al. Quantitative
2004;145:593-9. in vivo MRI measurement of cortical
5 van Haastert IC, Groenendaal F, Uiterwaal CS, development in the fetus. Brain Struct Funct
Termote JU, Heide-Jalving M, Eijsermans MJ et al. 2012;217:127-39.
Decreasing incidence and severity of cerebral 17 Dubois J, Benders M, Cachia A, Lazeyras F, Ha-
palsy in prematurely born children. J Pediatr Vinh LR, Sizonenko SV et al. Mapping the early
2011;159:86-91. cortical folding process in the preterm newborn
6 Benders MJNL, Kersbergen KJ, de Vries LS. brain. Cereb Cortex 2008;18:1444-54.
Neuroimaging of White Matter Injury, 18 Dubois J, Benders M, Borradori-Tolsa C, Cachia A,
Intraventricular and Cerebellar Hemorrhage. Clin Lazeyras F, Ha-Vinh LR et al. Primary cortical
Perinatol 2014;41:69-82. folding in the human newborn: an early marker
7 Kidokoro H, Neil JJ, Inder TE. New MR Imaging of later functional development. Brain
Assessment Tool to Define Brain Abnormalities in 2008;131:2028-41.
Very Preterm Infants at Term. AJNR Am J 19 Kapellou O, Counsell SJ, Kennea N, Dyet L, Saeed
Neuroradiol 2013. N, Stark J et al. Abnormal cortical development
8 Murphy BP, Inder TE, Huppi PS, Warfield S, after premature birth shown by altered
Zientara GP, Kikinis R et al. Impaired cerebral allometric scaling of brain growth. PLoS Med
cortical gray matter growth after treatment with 2006;3:e265.
dexamethasone for neonatal chronic lung 20 Ochiai T, Grimault S, Scavarda D, Roch G, Hori T,
disease. Pediatrics 2001;107:217-21. Riviere D et al. Sulcal pattern and morphology of
9 O’Shea TM, Washburn LK, Nixon PA, Goldstein DJ. the superior temporal sulcus. Neuroimage
Follow-up of a randomized, placebo-controlled 2004;22:706-19.
trial of dexamethasone to decrease the duration 21 Sowell ER, Thompson PM, Rex D, Kornsand D,
of ventilator dependency in very low birth Tessner KD, Jernigan TL et al. Mapping sulcal
weight infants: neurodevelopmental outcomes pattern asymmetry and local cortical surface
at 4 to 11 years of age. Pediatrics 2007;120:594- gray matter distribution in vivo: maturation in
602. perisylvian cortices. Cereb Cortex 2002;12:17-
10 Parikh NA, Lasky RE, Kennedy KA, Moya FR, 26.
Hochhauser L, Romo S et al. Postnatal 22 Van Essen DC. A Population-Average, Landmark-
dexamethasone therapy and cerebral tissue and Surface-based (PALS) atlas of human
volumes in extremely low birth weight infants. cerebral cortex. Neuroimage 2005;28:635-62.
Pediatrics 2007;119:265-72. 23 Chugani HT. A critical period of brain
ch a p te r 9
11 Yeh TF, Lin YJ, Lin HC, Huang CC, Hsieh WS, Lin CH development: studies of cerebral glucose
et al. Outcomes at school age after postnatal utilization with PET. Prev Med 1998;27:184-8.
dexamethasone therapy for lung disease of 24 De Vis JB, Petersen ET, de Vries LS, Groenendaal F,

Kersbergen KJ, Alderliesten T et al. Regional 36 Wright R, Kyriakopoulou V, Ledig C, Rutherford
changes in brain perfusion during brain MA, Hajnal JV, Rueckert D et al. Automatic
maturation measured non-invasively with quantification of normal cortical folding patterns
Arterial Spin Labeling MRI in neonates. Eur J from fetal brain MRI. Neuroimage 2014;91:21-
Radiol 2013;82:538-43. 32.
25 Kinney HC, Brody BA, Kloman AS, Gilles FH. 37 Bui T, Daire JL, Chalard F, Zaccaria I, Alberti C,
Sequence of central nervous system myelination Elmaleh M et al. Microstructural development of
in human infancy. II. Patterns of myelination in human brain assessed in utero by diffusion
autopsied infants. J Neuropathol Exp Neurol tensor imaging. Pediatr Radiol 2006;36:1133-40.
1988;47:217-34. 38 Kasprian G, Brugger PC, Weber M, Krssak M,
26 Deprez S, Billiet T, Sunaert S, Leemans A. Krampl E, Herold C et al. In utero tractography of
Diffusion tensor MRI of chemotherapy-induced fetal white matter development. Neuroimage
cognitive impairment in non-CNS cancer 2008;43:213-24.
patients: a review. Brain Imaging Behav 39 Thomason ME, Brown JA, Dassanayake MT,
2013;7:409-35. Shastri R, Marusak HA, Hernandez-Andrade E et
27 Smith SM, Jenkinson M, Johansen-Berg H, al. Intrinsic functional brain architecture derived
Rueckert D, Nichols TE, Mackay CE et al. Tract- from graph theoretical analysis in the human
based spatial statistics: voxelwise analysis of fetus. PLoS One 2014;9:e94423.
multi-subject diffusion data. Neuroimage 40 Back SA, Miller SP. Brain injury in premature
2006;31:1487-505. neonates: A primary cerebral dysmaturation
28 Lee JD, Park HJ, Park ES, Oh MK, Park B, Rha DW et disorder? Ann Neurol 2014;75:469-86.
al. Motor pathway injury in patients with 41 Volpe JJ, Kinney HC, Jensen FE, Rosenberg PA. The
periventricular leucomalacia and spastic developing oligodendrocyte: key cellular target
diplegia. Brain 2011;134:1199-210. in brain injury in the premature infant. Int J Dev
29 Nagasunder AC, Kinney HC, Bluml S, Tavare CJ, Neurosci 2011;29:423-40.
Rosser T, Gilles FH et al. Abnormal 42 Tournier JD, Mori S, Leemans A. Diffusion tensor
microstructure of the atrophic thalamus in imaging and beyond. Magn Reson Med
preterm survivors with periventricular 2011;65:1532-56.
leukomalacia. AJNR Am J Neuroradiol 43 Counsell SJ, Ball G, Edwards AD. New imaging
2011;32:185-91. approaches to evaluate newborn brain injury
30 Volpe JJ. Brain injury in premature infants: a and their role in predicting developmental
complex amalgam of destructive and disorders. Curr Opin Neurol 2014;27:168-75.
developmental disturbances. Lancet Neurol 44 van den Heuvel MP, Kersbergen KJ, de Reus MA,
2009;8:110-24. Keunen K, Kahn RS, Groenendaal F et al. The
31 Yokochi K. Thalamic lesions revealed by MR Neonatal Connectome During Preterm Brain
associated with periventricular leukomalacia and Development. Cereb Cortex 2014.
clinical profiles of subjects. Acta Paediatr 45 Arichi T, Counsell SJ, Allievi AG, Chew AT,
1997;86:493-6. Martinez-Biarge M, Mondi V et al. The effects of
32 Fransson P, Skiold B, Horsch S, Nordell A, hemorrhagic parenchymal infarction on the
Blennow M, Lagercrantz H et al. Resting-state establishment of sensori-motor structural and
networks in the infant brain. Proc Natl Acad Sci U functional connectivity in early infancy.
S A 2007;104:15531-6. Neuroradiology 2014;56:985-94.
33 Smyser CD, Snyder AZ, Neil JJ. Functional 46 Smyser CD, Snyder AZ, Shimony JS, Blazey TM,
connectivity MRI in infants: exploration of the Inder TE, Neil JJ. Effects of white matter injury on
functional organization of the developing brain. resting state fMRI measures in prematurely born
Neuroimage 2011;56:1437-52. infants. PLoS One 2013;8:e68098.
34 Garel C, Chantrel E, Brisse H, Elmaleh M, Luton D, 47 Ball G, Boardman JP, Aljabar P, Pandit A, Arichi T,
Oury JF et al. Fetal cerebral cortex: normal Merchant N et al. The influence of preterm birth
gestational landmarks identified using prenatal on the developing thalamocortical connectome.
MR imaging. AJNR Am J Neuroradiol Cortex 2013;49:1711-21.
2001;22:184-9. 48 Pandit AS, Robinson E, Aljabar P, Ball G, Gousias
35 Habas PA, Scott JA, Roosta A, Rajagopalan V, Kim IS, Wang Z et al. Whole-brain mapping of
ch a pt e r 9
K, Rousseau F et al. Early folding patterns and structural connectivity in infants reveals altered
asymmetries of the normal human brain connection strength associated with growth and
detected from in utero MRI. Cereb Cortex preterm birth. Cereb Cortex 2014;24:2324-33.
2012;22:13-25. 49 Parvizi J. Corticocentric myopia: old bias in new
cognitive sciences. Trends Cogn Sci 2009;13:354-9. growth factor treatment rescues neonatal brain
50 Volpe JJ. Cerebellum of the premature infant: injury. Nature 2014;506:230-4.
rapidly developing, vulnerable, clinically 62 Wood TL, Loladze V, Altieri S, Gangoli N, Levison
important. J Child Neurol 2009;24:1085-104. SW, Brywe KG et al. Delayed IGF-1 administration
51 Marret S, Marpeau L, Zupan-Simunek V, Eurin D, rescues oligodendrocyte progenitors from
Leveque C, Hellot MF et al. Magnesium sulphate glutamate-induced cell death and hypoxic-
given before very-preterm birth to protect infant ischemic brain damage. Dev Neurosci
brain: the randomised controlled PREMAG trial*. 2007;29:302-10.
BJOG 2007;114:310-8. 63 Donega V, van Velthoven CT, Nijboer CH, van BF,
52 Leuchter RH, Gui L, Poncet A, Hagmann C, Kas MJ, Kavelaars A et al. Intranasal mesenchymal
Lodygensky GA, Martin E et al. Association stem cell treatment for neonatal brain damage:
between early administration of high-dose long-term cognitive and sensorimotor
erythropoietin in preterm infants and brain MRI improvement. PLoS One 2013;8:e51253.
abnormality at term-equivalent age. JAMA 64 Donega V, Nijboer CH, van TG, Dijkhuizen RM,
2014;312:817-24. Kavelaars A, Heijnen CJ. Intranasally
53 Wang H, Zhang L, Jin Y. A Meta-analysis of the administered mesenchymal stem cells promote
Protective Effect of Recombinant Human a regenerative niche for repair of neonatal
Erythropoietin (rhEPO) for Neurodevelopment in ischemic brain injury. Exp Neurol 2014;261:53-
Preterm Infants. Cell Biochem Biophys 2014. 64.
54 AlFaleh K. Indomethacin prophylaxis revisited: 65 Li J, McDonald CA, Fahey MC, Jenkin G, Miller SL.
changing practice and supportive evidence. Acta Could cord blood cell therapy reduce preterm
Paediatr 2011;100:641-6. brain injury? Front Neurol 2014;5:200.
55 Miller SP, Mayer EE, Clyman RI, Glidden DV, 66 Orton J, Spittle A, Doyle L, Anderson P, Boyd R.
Hamrick SE, Barkovich AJ. Prolonged Do early intervention programmes improve
indomethacin exposure is associated with cognitive and motor outcomes for preterm
decreased white matter injury detected with infants after discharge? A systematic review. Dev
magnetic resonance imaging in premature Med Child Neurol 2009;51:851-9.
newborns at 24 to 28 weeks’ gestation at birth. 67 Van Hus JW, Jeukens-Visser M, Koldewijn K,
Pediatrics 2006;117:1626-31. Geldof CJ, Kok JH, Nollet F et al. Sustained
56 Keunen K, van Elburg RM, van BF, Benders MJ. developmental effects of the infant behavioral
Impact of nutrition on brain development and its assessment and intervention program in very
neuroprotective implications following preterm low birth weight infants at 5.5 years corrected
birth. Pediatr Res 2014. age. J Pediatr 2013;162:1112-9.
57 Cormack BE, Bloomfield FH. Increased protein 68 Groenendaal F, Uiterwaal C. Long term follow-up
intake decreases postnatal growth faltering in of extremely preterm neonates. BMJ
ELBW babies. Arch Dis Child Fetal Neonatal Ed 2012;345:e8252.
2013;98:F399-F404. 69 Schaer M, Ottet MC, Scariati E, Dukes D, Franchini
58 Franz AR, Pohlandt F, Bode H, Mihatsch WA, M, Eliez S et al. Decreased frontal gyrification
Sander S, Kron M et al. Intrauterine, early correlates with altered connectivity in children
neonatal, and postdischarge growth and with autism. Front Hum Neurosci 2013;7:750.
neurodevelopmental outcome at 5.4 years in 70 Wolosin SM, Richardson ME, Hennessey JG,
extremely preterm infants after intensive Denckla MB, Mostofsky SH. Abnormal cerebral
neonatal nutritional support. Pediatrics cortex structure in children with ADHD. Hum
2009;123:e101-e109. Brain Mapp 2009;30:175-84.
59 Isaacs EB, Morley R, Lucas A. Early diet and 71 D’Onofrio BM, Class QA, Rickert ME, Larsson H,
general cognitive outcome at adolescence in Langstrom N, Lichtenstein P. Preterm birth and
children born at or below 30 weeks gestation. J mortality and morbidity: a population-based
Pediatr 2009;155:229-34. quasi-experimental study. JAMA Psychiatry
60 van den Akker CH, Te Braake FW, Weisglas- 2013;70:1231-40.
Kuperus N, van Goudoever JH. Observational 72 Peterson BS, Vohr B, Staib LH, Cannistraci CJ,
Outcome Results Following A Randomized Dolberg A, Schneider KC et al. Regional brain
Controlled Trial of Early Amino Acid volume abnormalities and long-term cognitive
ch a p te r 9
Administration in Preterm Infants. J Pediatr outcome in preterm infants. JAMA

Gastroenterol Nutr 2014. 2000;284:1939-47.
61 Scafidi J, Hammond TR, Scafidi S, Ritter J,
Jablonska B, Roncal M et al. Intranasal epidermal

10
ch apter
Nederlandse
samenvatting
205
O
m te kunnen begrijpen waarom een groot percentage van de prematuur geboren
kinderen ontwikkelingsproblemen laat zien, is een beter begrip nodig van de
hersenontwikkeling die plaatsvindt gedurende de periode die het derde trimester van
de zwangerschap zou zijn geweest. Met de toenemende overlevingskans van zelfs de meest
prematuur geboren kinderen moet rekening gehouden worden met de gevolgen van extreme
vroeggeboorte voor zowel ouders en de onmiddellijke omgeving, als voor de maatschappij.
Verbeteringen in neuromotorische uitkomst komen niet alleen ten goede van het kind, maar
ook van de gehele maatschappij.1,2 Het identificeren van afwijkingen in de hersenontwikkeling
-die plaatsvindt terwijl een prematuur geboren kind wordt verzorgd in een intensive care
omgeving met alle stressoren en complicaties vandien- kan helpen in het vinden van betere
voorspellingsmethodes en het ontwikkelen van behandelingen om de cognitieve en
gedragsmatige ontwikkeling van deze kwetsbare populatie te verbeteren. Het doel van dit
proefschrift was om de longitudinale hersenontwikkeling van extreem te vroeg geboren
kinderen te bestuderen door middel van seriële MRI scans van kinderen geboren na een
zwangerschapsduur van minder dan 28 weken.
In Hoofdstuk 1 wordt een introductie van het begrip prematuriteit gegeven en worden de
belangrijkste patronen van hersenschade in prematuur geboren kinderen beschreven.
Onderzoek naar de visuele evaluatie van conventionele beeldvorming van de hersenen heeft
geleid tot een betere herkenning van de patronen van hersenschade in prematuur geboren
kinderen en de verschillen in de mate van ernst hiervan. Correlaties met ontwikkeling zijn
gelegd en in het bijzonder voor motorische ontwikkeling lijkt MRI een betrouwbare
voorspelling te kunnen geven.3 In de afgelopen tientallen jaren zijn de patronen van
hersenschade veranderd en ernstige witte stof schade wordt gelukkig niet vaak meer gezien.4,5
Mede dankzij het toegenomen gebruik van MRI onderzoek rond de uitgerekende datum
worden nu andere patronen van witte stof schade herkend. Puntlaesies van de witte stof
(PWML) behoren nu tot de meest voorkomende laesies, die gezien worden in 20 tot 50% van
de prematuur geboren kinderen.6 PWML kunnen alleen met behulp van MRI betrouwbaar
worden gediagnosticeerd en het herkennen van deze laesies is in grote mate afhankelijk van
de kwaliteit van de beeldvorming.
In Hoofdstuk 2 beschrijven we verschillende patronen van PWML die werden gezien in een
cohort van kinderen met seriële scans in de neonatale periode, waarbij tevens gebruik werd
gemaakt van diffusie gewogen (DWI) en susceptibiliteit gewogen (SWI) beelden. Twee
verschillende patronen werden gevonden met ook verschillen op DWI en SWI. Het lineaire
patroon bestond uit lineaire laesies, meestal rondom de ventrikels, die ook zichtbaar waren op
SWI maar niet op DWI. Van deze laesies wordt daarom gedacht dat ze hemorrhagisch van
origine zijn. Het cluster patroon bestond uit meer solitaire, wat grotere en rondere laesies, die
zich dieper in de witte stof bevonden en niet zichtbaar waren op SWI. Op vroege
beeldvorming, als de scan was gemaakt binnen 10 dagen na de geboorte, konden deze laesies
ook gezien worden op de DWI, hetgeen een ischemische origine suggereert. Er is geen
ch a p te r 1 0
pathologie materiaal voorhanden om onze bevindingen te bevestigen, maar deze
verschillende patronen staan waarschijnlijk voor verschillende types schade. Hiermee moet
rekening gehouden worden als deze patronen worden samengevoegd bij het evalueren van
de neuromotorische ontwikkeling. Hoewel er gesuggereerd is dat PWML verantwoordelijk
kunnen zijn voor de mildere cognitieve en gedragsmatige problemen die gevonden worden
op de schoolleeftijd, vonden wij geen relatie met ontwikkeling op 2 jaar.
Deze bevindingen onderstrepen de limitaties van het voorspellen van ontwikkeling door de
visuele beoordeling van conventionele beeldvorming, zelfs wanneer dit systematisch wordt
gedaan.7 Om een betrouwbaarder voorspelling van cognitieve en gedragsmatige ontwikkeling
te krijgen en beter te begrijpen wat de onderliggende pathofysiologie is, zijn kwantitatieve
technieken nodig om de MRI’s van te vroeg geboren kinderen te bestuderen.
In Deel 2 van dit proefschrift worden een aantal van deze kwantitatieve technieken beschreven
die zijn toegepast op T1- en T2- gewogen beeldvorming. In Hoofdstuk 3 beschrijven we de
toepassing van een recent ontwikkelde segmentatie methode, die de hersenen in 50 regio’s
verdeeld. Zowel de vroege scans als die rond de uitgerekende datum konden gesegmenteerd
worden met behulp van deze methode, wat ons de mogelijkheid gaf om de groei van het brein
in het derde trimester, buiten de baarmoeder, te bestuderen. De mate van groei verschilde
tussen de regio’s, met de grootste groei in het cerebellum (258%) en de kleinste toename in de
ventrikels (61%). Meisjes hadden kleinere hersenen dan jongens en kinderen met een lagere
geboortegewicht z-score en diegenen die langdurige mechanische beademing nodig hadden
of chirurgie ondergingen, lieten een globaal kleiner volume op beide scans zien. Hersenschade
toonde een meer lokaal effect, waarbij de ventrikels al groter waren op de eerste scan, terwijl
het kleinere volume van het cerebellum zichtbaar was in de groei en volumes rond de
uitgerekende datum. Deze data tonen hoe kwetsbaar het premature brein is en laten ook zien
dat sommige effecten al zichtbaar zijn op 30 weken, terwijl andere effecten na verloop van tijd
pas duidelijk worden.
Er zijn meer risicofactoren beschreven die hersenvolumes rond de uitgerekende datum
beïnvloeden en één daarvan is het gebruik van corticosteroïden voor een chronisch longbeeld.
Er is aangetoond dat dexamethason, het meest wijdverbreid gebruikte medicijn, een
schadelijk effect heeft op zowel hersenvolumes rond de uitgerekende datum als ontwikkeling
in de kindertijd.7-10 Deze bevindingen hebben geleid tot lagere en voorzichtiger doseringen,
die deze effecten mogelijk verminderen.11 Er zijn echter alternatieven beschikbaar en in
Utrecht wordt sinds jaar en dag hydrocortison in plaats van dexamethason gebruikt. Van
hydrocortison wordt gezegd dat het een minder sterke werking heeft, maar de bijwerkingen
zijn ook minder en er is geen effect aangetoond op de neuromotorische ontwikkeling.12, 13 In
Hoofdstuk 4 hebben we de hersenvolumes op de uitgerekende datum vergeleken tussen
kinderen die hydrocortison kregen en controle kinderen die dit niet kregen, met dezelfde
zwangerschapsduur en van hetzelfde geslacht. Kinderen met grote parenchymschade werden
c h a pt er 10
nederlandse samenvatting 207

geëxcludeerd. Na correctie voor zwangerschapsduur, leeftijd ten tijde van scan,
geboortegewicht z-score en de aanwezigheid van een intraventriculaire bloeding, werden
geen verschillen gevonden voor zowel het totale hersenvolume als de volumes van het
cerebellum tussen kinderen die hydrocortison kregen en controles. Hydrocortison lijkt dus
een veiliger alternatief te zijn voor dexamethason in de behandeling van een zich
ontwikkelend chronisch longbeeld.
Naast het bestuderen van de volumetrische effecten op hersengroei en ontwikkeling en de
invloed hierop van klinische risicofactoren, kunnen de volumetrische segmentaties ook
gebruikt worden als begin voor verdere kwantitatieve analyses, zoals het ontwikkelen van
witte stof maskers om de ontwikkeling van hersenwindingen te bestuderen. Het grootste deel
van de ontwikkeling van de gyri en sulci vindt plaats tijdens het derde trimester van de
zwangerschap en de ontwikkeling van deze windingen is bestudeerd met behulp van MRI,
hetgeen de volgorde van ontwikkeling,14-16 invloed van klinische factoren zoals intra-uteriene
groei retardatie, tweelingzwangerschap, geslacht en zwangerschapsduur, en asymmetrieën in
sulci ontwikkeling, heeft laten zien.17, 17, 18, 18 Hoofdstuk 5 beschrijft het gebruik van een speciaal
programma dat specifieke windingen op zowel de vroege als term-equivalente scans kan
identificeren. Hierdoor was het mogelijk de ontwikkeling van deze specifieke sulci te
bestuderen in onze populatie en deze metingen te correleren met zowel klinische factoren als
ontwikkeling op 2 jaar. Zeven verschillende regio’s werden bekeken: de sulcus centralis (CS),
fissura lateralis (LF), insula (INS), sulcus temporalis superior (STS), sulcus post centralis (PCS),
sulcus frontalis superior (SFS) en sulcus frontalis inferior (IFS). De drie meest centrale sulci (CS,
LF en INS) waren zichtbaar op de eerste MRI bij alle kinderen, terwijl de andere vier minder
goed ontwikkeld waren en niet bij alle kinderen gevonden werden rond een postmenstruele
leeftijd van 30 weken. In deze laatste vier sulci was de relatieve groei in de periode tussen
beide scans groter, waarbij de SFS de grootste toename in oppervlakte liet zien. De bekende
rechtszijdige asymmetrie in hersenontwikkeling kon duidelijk herkend worden in de data op
30 weken. Rond de uitgerekende datum waren de rechts-links verschillen het duidelijkst voor
de STS (rechts>links) en de LF (links>rechts), hetgeen overeenkomt met bevindingen op
latere leeftijd.19-21 Meerlingzwangerschap, een lagere geboortegewicht z-score en langdurige
beademing waren gecorreleerd met een vertraging in de ontwikkeling van hersenwindingen,
het duidelijkst waarneembaar op de vroege scan. De sulci die het eerst ontwikkelden waren
het meest aangedaan door deze factoren. Een positieve associatie met motorische, cognitieve
en taal ontwikkeling op 2-2.5 jaar gecorrigeerde leeftijd werd gevonden. Het opleidingsniveau
van moeder liet ook een sterke associatie met neuromotorische ontwikkeling zien. Het
opleidingsniveau van moeder staat voor een combinatie van genetische invloeden en socio-
economische status van de omgeving waarin het kind opgroeit, en moet daarom altijd in acht
genomen worden wanneer neuromotorische ontwikkeling wordt bestudeerd.
ch a p te r 1 0
Naast conventionele beeldvorming, worden diffusie gewogen en functionele MRI steeds meer
gebruikt om herseontwikkeling en de afwijkingen ten gevolge van prematuriteit in meer detail
te bestuderen. Een aantal van deze resultaten wordt beschreven in Deel 3 van dit proefschrift.
DWI, en meer specifiek het vaak gebruikte diffusie tensor beeldvorming (DTI), visualiseert de
Browniaanse beweging van watermoleculen in het brein, hetgeen informatie geeft over de
microstructurele integriteit en maturatie. Door de verschillende diffusie parameters te
bestuderen kan meer inzicht in hersenontwikkeling verkregen worden. In Hoofdstuk 6
beschrijven we diffusie metingen van een geautomatiseerde, atlas gebaseerde methode die
het gehele brein op zowel 30 als 40 weken omvat en ook de groei ertussen laat zien. Voor deze
studie werden 40 kinderen met seriële DTI data van goede kwaliteit en een normale
neuromotorische ontwikkeling op 15 maanden geselecteerd, om de normale
hersenontwikkeling zo goed als mogelijk te benaderen. Een toename van fractionele
anisotropie (FA) werd gemeten met een centraal naar perifeer en posterieur naar anterieure
gradiënt, met daarbij behorende afname van de gemiddelde, radiale en axiale diffusiviteit.
Corticale regio’s lieten minder toename of een afname van FA zien. Deze bevindingen passen
bij het huidige begrip van hersenmaturatie, die eerst beschreven werd in histologische studies
en later in vivo bevestigd is.22-24 De posterieur naar anterieure gradiënt komt overeen met de
resultaten van hoofdstuk 4, waar de occipitale regio’s een wat grotere toename lieten zien dan
de frontale regio’s. De centraal naar perifere gradiënt komt overeen met de bevindingen van
hoofdstuk 5, waar de centraal gelegen sulci het eerst ontwikkelden. De diffusie metingen van
deze studie kunnen gebruikt worden als referentie voor toekomstige studies om
hersenontwikkeling in andere cohorten te bestuderen, of in studies die hersenschade en het
effect van neuroprotectieve behandelingen evalueren.
Naast de atlas gebaseerde methode om DTI data te bestuderen die gebruikt werd in hoofdstuk 6,
zijn verschillende andere methodes beschikbaar. De meest gebruikte technieken zijn
handmatige region of interest (ROI) analyse, tract-based spatial statistics (TBSS) en zowel
deterministische als probabilistische tractografie.25 Atlas gebaseerde en in mindere mate ROI
gebaseerde methodes zijn waardevol voor het bestuderen van het gehele brein of grotere
regio’s. TBSS is een effectieve methode om de verschillen tussen cohorten of tussen patiënten
en controles vast te stellen en laat die voxels zien waarin de verschillen tussen beiden
significant zijn.26 Tractografie kan gedaan worden voor zowel het gehele brein en dan daarna
gebruikt worden als input voor het meten van connectiviteit, of kan gedaan worden om
diffusie waarden van een specifieke witte stof bundel, zoals de corticospinale baan (CST), te
meten. In Hoofdstuk 7 hebben we probabilistische tractografie gebruikt om de verschillen aan
te tonen tussen te vroeg geboren kinderen met cysteuze periventriculaire leukomalacie
(c-PVL) en te vroeg geboren controles zonder hersenschade, voor zowel de CST als de volumes
van de thalamus. c-PVL is een van de meest ernstige soorten van witte stof schade en is de
meest voorkomende oorzaak voor de ontwikkeling van een cerebrale parese (CP) na
vroeggeboorte.4, 5
Wanneer kinderen met CP veroorzaakt door c-PVL werden gescand
c h a pt er 10

gedurende de kindertijd, waren zowel de CST als de thalamus aangedaan, vergeleken met
controle kinderen die zich normaal ontwikkelden.27, 28 Het is echter nog steeds niet duidelijk
wanneer deze verschillen precies ontstaan. Met deze studie hebben we laten zien dat schade
aan de CST al op de eerste scan geïdentificeerd kan worden, wanneer deze gemaakt wordt
binnen de eerste weken na het ontstaan van de schade. In tegenstelling tot de CST, waren de
thalamusvolumes nog niet aangedaan op de eerste scan. Deze lieten echter wel een duidelijke
vermindering van groei zien, met kleinere volumes op de scan rond de uitgerekende datum.
Deze data passen bij de hypothese dat de cysten, die vaak midden in de baan van de CST
liggen, een direct effect hebben op de CST en andere witte stof banen. Hierdoor ontstaat een
verandering in de afferente en efferente signalen naar de thalamus en wordt de normale
ontwikkeling van de thalamus gehinderd. In de loop van een aantal weken worden hierdoor
kleinere thalamusvolumes zichtbaar.29 De thalamus is een belangrijke factor voor de
neuromotorische ontwikkeling en dit gold ook voor deze studie, waar de ernst van de CP
gecorreleerd was met de thalamusvolumes rondom de uitgerekende datum.30
Deze relatie tussen structurele veranderingen of schade en functionele uitkomstmaten is een
gebied waarin veel onderzoek wordt verricht. Gedurende de laatste jaren is resting state
functionele MRI (rs-fMRI) beschikbaar gemaakt voor gebruik in neonatale studies.31, 32 De
combinatie van DTI en rs-fMRI heeft het mogelijk gemaakt het menselijke connectoom samen
te stellen of met andere woorden, heeft het mogelijk gemaakt de structurele vorm van de
hersenen te vergelijken met de neurale activiteit. Hoofdstuk 8 beschrijft de ontwikkeling van
het connectoom in 27 neonaten met longitudinale scans inclusief zowel DTI als rs-fMRI van
voldoende kwaliteit op beide tijdspunten. De DTI scans lieten zien dat de small-world
organisatie al duidelijk zichtbaar was op de eerste scan, met de vorming van immature resting-
state netwerken op de fMRI. Netwerken van structurele en functionele connectiviteit die
gevonden werden op de scans op 30 weken lieten een grote overeenkomst zien met de
volwassen architectuur van het brein, met meer dan 80% overlap. De netwerken die
geïdentificeerd werden op de vroege scan lieten rond de uitgerekende datum een
ontwikkeling qua microstructuur, small-world topologie en interhemisferische functionele
connectiviteit zien, waardoor een toename van de integratie capaciteit van de connectiveit als
geheel werd gezien. De basisstructuur van het menselijke connectoom lijkt zich dus al vroeg in
de ontwikkeling te vormen, aangezien het al duidelijk zichtbaar is rond 30 weken
zwangerschapsduur. De ontwikkeling tussen 30 en 40 weken, die gevonden werd in deze
studie, suggereert echter opnieuw dat de actieve ontwikkeling van het brein gedurende het
laatste trimester van de zwangerschap verstoord kan worden door vroeggeboorte. Deze
functionele en in mindere mate ook de structurele verbindingen zijn niet zichtbaar met het
blote oog en worden alleen duidelijk als geavanceerde MRI technieken gebruikt worden.
Conventionele beeldvorming is onvoldoende gebleken om cognitieve en gedragsmatige
ontwikkeling te voorspellen, precies die gebieden die verstoord zijn bij bijna de helft van de te
vroeg geboren kinderen. Deze data suggereren dat de negatieve gevolgen van vroeggeboorte
ch a p te r 1 0
op neuromotorische ontwikkeling –deels– veroorzaakt worden door afwijkingen in de
normale ontwikkeling van het menselijke connectoom. Het correleren van deze data met
neuromotorische uitkomstmaten lijkt daarom van groot belang te zijn, hetgeen geldt voor alle
hoofdstukken uit dit proefschrift.
Conclusies
De volgende conclusies kunnen getrokken worden uit dit proefschrift:
- Puntlaesies van de witte stof kunnen onderverdeeld worden in twee verschillende
patronen, die een verschil in onderliggende pathofysiologie suggereren. Een vroege MRI
inclusief DWI en SWI is een vereiste om deze patronen betrouwbaar te kunnen
identificeren (hoofdstuk 2).
- Volume groei van de hersenen laat regionale verschillen zien, met de grootste groei in het
cerebellum. Vrouwelijk geslacht, een lagere geboortegewicht z-score en langdurige
mechanische beademing hebben een negatieve invloed op hersenvolumes. Het effect van
hersenschade is regionaal en het duidelijkst in de ventrikels op 30 weken, terwijl het
cerebellum na hersenschade verminderde groei en kleinere volumes rond de uitgerekende
datum laat zien (hoofdstuk 3).
- Behandeling met hydrocortison voor een zich ontwikkelend chronisch longbeeld zorgt
niet voor een kleiner totaal hersenvolume of cerebellum volume op de uitgerekende
datum in te vroeg geboren kinderen zonder parenchymschade van de hersenen (hoofdstuk
4).
- De ontwikkeling van de hersenwindingen van te vroeg geboren kinderen vindt plaats in
een centraal naar occipitale en frontale richting, waarbij de rechterhemisfeer iets eerder
ontwikkelt dan de linker. De diepte en oppervlakte van de hersenwindingen zijn kleiner in
kinderen die deel zijn van een tweelingzwangerschap, een lagere geboortegewicht z-score
hebben en na langdurige mechanische beademing en laten een correlatie zien met de
neuromotorische ontwikkeling op 2 jaar, inclusief de taalontwikkeling (hoofdstuk 5).
- Longitudinale diffusiviteitswaarden van het premature brein laten een centraal-perifere en
occipitaal-frontale gradiënt zien. Fractionele anisotropie laat een toename zien in de witte
stof maar een afname in de cortex. De gemiddelde, radiale en axiale diffusiviteit laten een
afname zien over het gehele brein (hoofdstuk 6).
- Schade aan de corticospinale banen in kinderen met c-PVL ontwikkelt zich binnen een paar
weken na het insult, terwijl de secundaire effecten op het volume van de thalamus pas
rond de uitgerekende datum zichtbaar worden (hoofdstuk 7).
- Immature functionele netwerken zijn al zichtbaar op 30 weken en tonen een grote
gelijkenis met volwassen netwerken. Zowel de structurele als functionele
connectiviteitsnetwerken ontwikkelen zich tussen 30 weken en de uitgerekende datum
waardoor een toename van de integratie capaciteit ontstaat (hoofdstuk 8).
- De centraal-perifere en occipitaal-frontale gradiënt van hersenontwikkeling kan bevestigd
c h a pt er 10

worden met verschillende MRI technieken (hoofdstuk 3, hoofdstuk 5, hoofdstuk 6, hoofdstuk
8).
- Langdurige mechanische beademing, of het nu een uiting is van chronische longschade of
een overkoepelende meting van de algehele mate van ziek zijn, heeft een negatieve
invloed op hersenontwikkeling (hoofdstuk 3, hoofdstuk 5).
- Meting van de neuromotorische ontwikkeling rond twee jaar is waarschijnlijk te vroeg om
de kleine afwijkingen te vinden die door milde hersenschade veroorzaakt worden
(hoofdstuk 2, hoofdstuk 6).
- De secundaire effecten van hersenschade worden duidelijk in de periode tussen 30 en 40
weken zwangerschapsduur, terwijl de directe effecten al zichtbaar zijn op 30 weken
(hoofdstuk 3, hoofdstuk 7).
ch a p te r 1 0
Referenties
1 Hodek JM, von der Schulenburg JM, Mittendorf T. dexamethasone therapy for lung disease of
Measuring economic consequences of preterm prematurity. N Engl J Med 2004;350:1304-13.
birth - Methodological recommendations for the 12 Watterberg KL. Policy statement--postnatal
evaluation of personal burden on children and corticosteroids to prevent or treat
their caregivers. Health Econ Rev 2011;1:6. bronchopulmonary dysplasia. Pediatrics
2 Blencowe H, Cousens S, Oestergaard MZ, Chou 2010;126:800-8.
D, Moller AB, Narwal R et al. National, regional, 13 Rademaker KJ, Uiterwaal CS, Groenendaal F,
and worldwide estimates of preterm birth rates Venema MM, Van Bel F, Beek FJ et al. Neonatal
in the year 2010 with time trends since 1990 for hydrocortisone treatment: neurodevelopmental
selected countries: a systematic analysis and outcome and MRI at school age in preterm-born
implications. Lancet 2012;379:2162-72. children. J Pediatr 2007;150:351-7.
3 de Vries LS, van Haastert IC, Benders MJ, 14 van der Heide-Jalving M, Kamphuis PJ, van der
Groenendaal F. Myth: cerebral palsy cannot be Laan MJ, Bakker JM, Wiegant VM, Heijnen CJ et al.
predicted by neonatal brain imaging. Semin Fetal Short- and long-term effects of neonatal
Neonatal Med 2011;16:279-87. glucocorticoid therapy: is hydrocortisone an
4 Hamrick SE, Miller SP, Leonard C, Glidden DV, alternative to dexamethasone? Acta Paediatr
Goldstein R, Ramaswamy V et al. Trends in severe 2003;92:827-35.
brain injury and neurodevelopmental outcome 15 Chi JG, Dooling EC, Gilles FH. Gyral development
in premature newborn infants: the role of cystic of the human brain. Ann Neurol 1977;1:86-93.
periventricular leukomalacia. J Pediatr 16 Clouchoux C, Kudelski D, Gholipour A, Warfield
2004;145:593-9. SK, Viseur S, Bouyssi-Kobar M et al. Quantitative
5 van Haastert IC, Groenendaal F, Uiterwaal CS, in vivo MRI measurement of cortical
Termote JU, Heide-Jalving M, Eijsermans MJ et al. development in the fetus. Brain Struct Funct
Decreasing incidence and severity of cerebral 2012;217:127-39.
palsy in prematurely born children. J Pediatr 17 Dubois J, Benders M, Cachia A, Lazeyras F, Ha-
2011;159:86-91. Vinh LR, Sizonenko SV et al. Mapping the early
6 Benders MJNL, Kersbergen KJ, de Vries LS. cortical folding process in the preterm newborn
Neuroimaging of White Matter Injury, brain. Cereb Cortex 2008;18:1444-54.
Intraventricular and Cerebellar Hemorrhage. Clin 18 Dubois J, Benders M, Borradori-Tolsa C, Cachia A,
Perinatol 2014;41:69-82. Lazeyras F, Ha-Vinh LR et al. Primary cortical
7 Kidokoro H, Neil JJ, Inder TE. New MR Imaging folding in the human newborn: an early marker
Assessment Tool to Define Brain Abnormalities in of later functional development. Brain
Very Preterm Infants at Term. AJNR Am J 2008;131:2028-41.
Neuroradiol 2013. 19 Kapellou O, Counsell SJ, Kennea N, Dyet L, Saeed
8 Murphy BP, Inder TE, Huppi PS, Warfield S, N, Stark J et al. Abnormal cortical development
Zientara GP, Kikinis R et al. Impaired cerebral after premature birth shown by altered
cortical gray matter growth after treatment with allometric scaling of brain growth. PLoS Med
dexamethasone for neonatal chronic lung 2006;3:e265.
disease. Pediatrics 2001;107:217-21. 20 Ochiai T, Grimault S, Scavarda D, Roch G, Hori T,
9 O’Shea TM, Washburn LK, Nixon PA, Goldstein DJ. Riviere D et al. Sulcal pattern and morphology of
Follow-up of a randomized, placebo-controlled the superior temporal sulcus. Neuroimage
trial of dexamethasone to decrease the duration 2004;22:706-19.
of ventilator dependency in very low birth 21 Sowell ER, Thompson PM, Rex D, Kornsand D,
weight infants: neurodevelopmental outcomes Tessner KD, Jernigan TL et al. Mapping sulcal
at 4 to 11 years of age. Pediatrics 2007;120:594- pattern asymmetry and local cortical surface
602. gray matter distribution in vivo: maturation in
10 Parikh NA, Lasky RE, Kennedy KA, Moya FR, perisylvian cortices. Cereb Cortex 2002;12:17-
Hochhauser L, Romo S et al. Postnatal 26.
dexamethasone therapy and cerebral tissue 22 Van Essen DC. A Population-Average, Landmark-
volumes in extremely low birth weight infants. and Surface-based (PALS) atlas of human
Pediatrics 2007;119:265-72. cerebral cortex. Neuroimage 2005;28:635-62.
11 Yeh TF, Lin YJ, Lin HC, Huang CC, Hsieh WS, Lin CH 23 Chugani HT. A critical period of brain
et al. Outcomes at school age after postnatal development: studies of cerebral glucose
c h a pt er 10

utilization with PET. Prev Med 1998;27:184-8. 28 Lee JD, Park HJ, Park ES, Oh MK, Park B, Rha DW et
24 De Vis JB, Petersen ET, de Vries LS, Groenendaal F, al. Motor pathway injury in patients with
Kersbergen KJ, Alderliesten T et al. Regional periventricular leucomalacia and spastic
changes in brain perfusion during brain diplegia. Brain 2011;134:1199-210.
maturation measured non-invasively with 29 Nagasunder AC, Kinney HC, Bluml S, Tavare CJ,
Arterial Spin Labeling MRI in neonates. Eur J Rosser T, Gilles FH et al. Abnormal
Radiol 2013;82:538-43. microstructure of the atrophic thalamus in
25 Kinney HC, Brody BA, Kloman AS, Gilles FH. preterm survivors with periventricular
Sequence of central nervous system myelination leukomalacia. AJNR Am J Neuroradiol
in human infancy. II. Patterns of myelination in 2011;32:185-91.
autopsied infants. J Neuropathol Exp Neurol 30 Volpe JJ. Brain injury in premature infants: a
1988;47:217-34. complex amalgam of destructive and
26 Deprez S, Billiet T, Sunaert S, Leemans A. developmental disturbances. Lancet Neurol
Diffusion tensor MRI of chemotherapy-induced 2009;8:110-24.
cognitive impairment in non-CNS cancer 31 Yokochi K. Thalamic lesions revealed by MR
patients: a review. Brain Imaging Behav associated with periventricular leukomalacia and
2013;7:409-35. clinical profiles of subjects. Acta Paediatr
27 Smith SM, Jenkinson M, Johansen-Berg H, 1997;86:493-6.
Rueckert D, Nichols TE, Mackay CE et al. Tract- 32 Fransson P, Skiold B, Horsch S, Nordell A,
based spatial statistics: voxelwise analysis of Blennow M, Lagercrantz H et al. Resting-state
multi-subject diffusion data. Neuroimage networks in the infant brain. Proc Natl Acad Sci U
2006;31:1487-505. S A 2007;104:15531-6.
chapter 1 0
214
List of abbreviations
Co-authors and their affiliations
List of publications
Curriculum vitae
Dankwoord – Acknowledgements
215
L ist o f a b bre vi at i o n s
3D 3-Dimensional
AD Axial diffusivity
ADC Apparent diffusion coefficient
AHW Anterior horn width
ALIC Anterior limb of the internal capsule
BI Brain injury
BPD Bronchopulmonary dysplasia
BSITD-III Bayley Scales of Infant and Toddler Development, third edition
BWZ Birth weight z-score
C Clustering coefficient
CA Corrected age
CBH Cerebellar haemorrhage
CP Cerebral palsy
c-PVL Cystic periventricular leukomalacia
CS Central sulcus
CST Corticospinal tract
cUS Cranial ultrasound
DEHSI Diffuse extensive high signal intensity
DIS Distance matrix
DQ Developmental quotient
DTI Diffusion tensor imaging
DWI Diffusion weighted imaging
FA Fractional anisotropy
FACT Fiber assignment by continuous tracking
FC Functional connectivity
fMRI Functional magnetic resonance imaging
FOV Field of view
FU Follow-up
GA Gestational age
GMFCS Gross motor function classification system
GMH-IVH Germinal matrix haemorrhage – intraventricular haemorrhage
HC Hydrocortisone
IFS Inferior frontal sulcus
INS Insula
IUGR Intra-uterine growth retardation
IVH Intraventricular haemorrhage
L Shortest path length
LF Lateral fissure
MD Mean diffusion / Mean diffusivity
MR Magnetic resonance
MRI Magnetic resonance imaging
Mult Multiple pregnancy
216
MV7 Mechanical ventilation for >7 days
N Number of patients
NICU Neonatal intensive care unit
No. Number of infants
NOS Number of streamline count
O Total level of overlap
PC Porencephalic cyst
PCS Postcentral sulcus
PD Parallel diffusion
PHVD Post-haemorrhagic ventricular dilatation
PLIC Posterior limb of the internal capsule
PMA Postmenstrual age
preOLs Premyelinating late oligodendrocyte progenitors
PVHI Periventricular haemorrhagic infarction
PWML Punctate white matter lesions
Q Modularity
RD Radial diffusivity
RDS Respiratory distress syndrome
ROI Region of interest
Rs-fMRI Resting state functional MRI
SC Structural connectivity
SC-FC Structure-function coupling
SD Standard deviation
SFS Superior frontal sulcus
SGA Small for gestational age
SI Signal intensity
STS Superior temporal sulcus
SW Small-world index
SWI Susceptibility weighted imaging
TBSS Tract based spatial statistics
TBV Total brain volume
TD Transverse diffusion
TE Echo time
TEA Term equivalent age
TI Inversion time
TOD Thalamo-occipital distance
TR Repetition time
VLBW Very low birth weight
WMI White matter injury
Some of the chapters have been slightly adjusted from their published version to have consistent use of layout and language
throughout the document.
217
C o -a u tho r s a n d the i r aff i li at i o ns
N ie k E . va n d er Aa, M D P hD
Department of Perinatology, Wilhelmina Children’s Hospital and Brain Center Rudolf Magnus, University Medical Center
Utrecht, the Netherlands
Pa u l A l ja b ar , P hD
Centre for the Developing Brain, Division of Imaging Sciences and Biomedical Engineering, King’s College London, St. Thomas’
Hospital, London, United Kingdom
F ra n k va n Be l, M D P hD
Ma no n J .N .L . Bender s, M D P hD
A n d rew T.M. C hew, M D

Centre for the Developing Brain, Division of Imaging Sciences and Biomedical Engineering, King’s College London, United
Kingdom
Natha lie H . P. Claessens, M D

Seren a J . C o u nsell, P hD
F ra n ce s M. C owa n, P hD M RCP CH
Department of Paediatrics, Hammersmith Campus, Imperial College London, United Kingdom
Sarah Daw s o n, BA M BBS

Je ss ica D u b o is, P hD
INSERM, Cognitive Neuroimaging Unit U992; CEA, Neurospin Center, Saclay, France; University Paris Sud, Orsay, France
C lara F is cher, M Sc
CEA, UNATI, NeuroSpin, Saclay, France
F lo r is Gro e n endaal, M D P hD
I ng r id C . va n H aastert, M A P hD
Mart ij n P. va n den H eu vel, P hD

Department of Psychiatry and Brain Center Rudolf Magnus, University Medical Center Utrecht, The Netherlands
218
P etra S . H ü ppi, MD
Department of Pediatrics, Geneva University Hospitals, Switzerland
I va n a I š g u m , P hD
Image Sciences Institute, University Medical Center Utrecht, The Netherlands
Re n é S . Kah n, MD P hD
Kri s t in Ke u ne n, MD
Bri tt J . M. va n Ko o ij , M D P hD
Cor i n e Ko o pman - E sseb oom, M D P hD

Utrecht,the Netherlands
Jean -F ra nc o is Ma n gi n, P hD
CEA, UNATI, NeuroSpin, Saclay, France
Al e x a n d er Leema ns , P hD
Fra n ç o is Ler oy , P hD
INSERM, Cognitive Neuroimaging Unit U992; CEA, Neurospin Center, Saclay, France; University Paris Sud, Orsay, France
An tonio s Ma k r o p o u los, P hD
Biomedical Image Analysis Group, Department of Computing, Imperial College London, United Kingdom
P i m Mo e s ko p s , MSc
Rutger A . J . N ie v e ls tei n, M D P hD
Department of Radiology, University Medical Center Utrecht, the Netherlands
Kari n J . R a d ema k er , M D P hD
M arce l A . d e R e u s , MSc
M a x A . V ier g e v er , P hD
L i n da S . d e Vr ie s , MD P hD
219
Li s t of p u b l icat i o n s
Kersbergen KJ, Benders MJNL, Groenendaal F, Koopman-Esseboom C, Nievelstein RAJ, van Haastert IC, de Vries LS. Different patterns
of punctate white matter lesions in serially scanned preterm infants. PLoS One 2014 9:e108904.
Kersbergen KJ, Leemans A, Groenendaal F, van der Aa NE, Viergever MA, de Vries LS, Benders MJ. Microstructural brain development
between 30 and 40 weeks corrected age in a longitudinal cohort of extremely preterm infants. Neuroimage 2014 103C:214-224.
Išgum I, Benders MJ, Avants B, Cardoso MJ, Counsell SJ, Gomez EF, Gui L, Hűppi PS, Kersbergen KJ, Makropoulos A, Melbourne A,
Moeskops P, Mol CP, Kuklisova-Murgasova M, Rueckert D, Schnabel JA, Srhoj-Egekher V, Wu J, Wang S, de Vries LS, Viergever MA.
Evaluation of automatic neonatal brain segmentation algorithms: The NeoBrainS12 challenge. Med Image Anal. 2014 pii: S1361-
8415(14)00158-3.
van den Heuvel MP, Kersbergen KJ, de Reus MA, Keunen K, Kahn RS, Groenendaal F, de Vries LS, Benders MJ. The Neonatal
Connectome During Preterm Brain Development. Cereb Cortex 2014 pii: bhu095.
De Vis JB, Hendrikse J, Groenendaal F, de Vries LS, Kersbergen KJ, Benders MJ, Petersen ET. Impact of neonate haematocrit variability
on the longitudinal relaxation time of blood: Implications for arterial spin labelling MRI. Neuroimage Clin. 2014 4:517-25.
Blok CA, Kersbergen KJ, van der Aa NE, van Kooij BJ, Anbeek P, Isgum I, de Vries LS, Krediet TG, Groenendaal F, Vreman HJ, van Bel F,
Benders MJ. Unmyelinated white matter loss in the preterm brain is associated with early increased levels of end-tidal carbon
monoxide. PLoS One 2014 9:e89061.
Benders MJ, Kersbergen KJ, de Vries LS. Neuroimaging of white matter injury, intraventricular and cerebellar hemorrhage. Clin
Perinatol. 2014 41:69-82.
Anbeek P, Išgum I, van Kooij BJ, Mol CP, Kersbergen KJ, Groenendaal F, Viergever MA, de Vries LS, Benders MJ. Automatic
segmentation of eight tissue classes in neonatal brain MRI. PLoS One 2013 8:e81895.
Hoff GE, Van den Heuvel MP, Benders MJ, Kersbergen KJ, De Vries LS. On development of functional brain connectivity in the young
brain. Front Hum Neurosci. 2013 7:650.
Nijman J, Gunkel J, de Vries LS, van Kooij BJ, van Haastert IC, Benders MJ, Kersbergen KJ, Verboon-Maciolek MA, Groenendaal F.
Reduced occipital fractional anisotropy on cerebral diffusion tensor imaging in preterm infants with postnatally acquired
cytomegalovirus infection. Neonatology 2013 104:143-50.
De Vis JB, Petersen ET, Kersbergen KJ, Alderliesten T, de Vries LS, van Bel F, Groenendaal F, Lemmers PM, Hendrikse J, Benders MJ.
Evaluation of perinatal arterial ischemic stroke using noninvasive arterial spin labeling perfusion MRI. Pediatr Res. 2013 74:307-13.
Kersbergen KJ, de Vries LS, van Kooij BJ, Išgum I, Rademaker KJ, van Bel F, Hüppi PS, Dubois J, Groenendaal F, Benders MJ.
Hydrocortisone treatment for bronchopulmonary dysplasia and brain volumes in preterm infants. J Pediatr. 2013 163:666-71.
Kersbergen KJ, de Vries LS, Leijten FS, Braun KP, Nievelstein RA, Groenendaal F, Benders MJ, Jansen FE. Neonatal thalamic hemorrhage
is strongly associated with electrical status epilepticus in slow wave sleep. Epilepsia 2013 54:733-40.
De Vis JB, Petersen ET, de Vries LS, Groenendaal F, Kersbergen KJ, Alderliesten T, Hendrikse J, Benders MJ. Regional changes in brain
perfusion during brain maturation measured non-invasively with Arterial Spin Labeling MRI in neonates. Eur J Radiol. 2013 82:538-
43.
Lopriore E, Slaghekke F, Kersbergen KJ, de Vries LS, Drogtrop AP, Middeldorp JM, Oepkes D, Benders MJ. Severe cerebral injury in a
recipient with twin anemia-polycythemia sequence. Ultrasound Obstet Gynecol. 2013 41:702-6.
Keunen K, Kersbergen KJ (shared first author), Groenendaal F, Isgum I, de Vries LS, Benders MJ. Brain tissue volumes in preterm
infants: prematurity, perinatal risk factors and neurodevelopmental outcome: a systematic review, J Matern Fetal Neonatal Med.
2012 Suppl 1:89-100.
Kersbergen KJ, Groenendaal F, Benders MJNL, de Vries LS, Neonatal cerebral sinovenous thrombosis: Neuro-imaging and long-term
follow-up. Journal of Child Neurology 2011 26:1111-20.
Kersbergen KJ, Groenendaal F, Benders MJNL, van Straaten HLM, Niwa T, Nievelstein RAJ, de Vries LS, The spectrum of associated brain
lesions in cerebral sinovenous thrombosis (CSVT): relation to gestational age and outcome. Arch Dis Child Fetal Neonatal Ed. 2011
96:F404-9.
Berfelo FJ, Kersbergen KJ, van Ommen CH, Govaert P, van Straaten HL, Poll-The BT, van Wezel-Meijler G, Vermeulen RJ, Groenendaal
F, de Vries LS, de Haan TR, Neonatal cerebral sinovenous thrombosis from symptom to outcome. Stroke 2010 41:1382-8.
Kersbergen KJ, de Vries LS, van Straaten HLM, Benders MJNL, Nievelstein RAJ, Groenendaal F, Anticoagulation therapy and imaging in
neonates with a unilateral thalamic hemorrhage due to cerebral sinovenous thrombosis. Stroke 2009 40:2754-60.
Cu rri c u lu m vi tae
K
arina Kersbergen was born on April 29th 1986 in Assen,
the Netherlands. She grew up with two brothers and one
sister in Driehuis, where she attended grammar school at
the Gymnasium Felisenum. After graduating cum laude in 2004,
she moved to Utrecht to start medical school at Utrecht
University. As part of her medical training she performed two of
her internships abroad, gynaecology in Garissa, Kenia, and social
health in Savannakhet, Laos. In 2008 she started her research at
the department of Neonatology at the Wilhelmina Children’s
Hospital, as part of the honours program. After graduating
medical school in 2010, she continued at the department of
Neonatology with a new research project on longitudinal MRI in
preterm infants under supervision of prof. dr. L.S. de Vries, prof.
dr. M.A. Viergever, dr. M.J.N.L. Benders and dr. F. Groenendaal, the
results of which are presented in this thesis. In February 2013
she spent two weeks in Paris, France at Neurospin Center to
learn a post-processing technique to study cortical folding,
under supervision of dr. Jessica Dubois. In May of that year, she
moved to London where she spent six months studying and
applying a new neonatal segmentation technique at the Centre
for the Developing Brain, King’s College London, under
supervision of professor Serena Counsell. Since November 2014
she is working as a paediatric resident in the Flevoziekenhuis in
Almere.
221
Dan k wo o r d – Ac k n ow le dg eme n t s
E
indelijk is het zover: na ruim vier jaar onderzoek is het tijd om mijn dankwoord te schrijven.
Een kans om iedereen te bedanken die op wat voor manier dan ook heeft bijgedragen aan
mijn promotieonderzoek en die wil ik dan ook graag gebruiken.
Prof. dr. L.S. de Vries, beste Linda. De eerste keer dat we elkaar ontmoetten was in de kamer van
Floris, toen ik als student net kwam kijken. Dat bleek het begin van een mooie samenwerking waarin
ik ongelooflijk veel van je heb geleerd. Ik heb enorm respect voor je kennis, de manier waarop je
met je vak en je patiënten omgaat, en heb genoten van je drive om jonge onderzoekers het vak aan
te leren. Daarnaast heb je ook echt oog voor ons als persoon, zoals wel bleek uit de vele gesprekken
die we hebben gevoerd tijdens congressen, MRI’s of gewoon tussendoor op de onderzoekskamer.
Prof. dr. M.A. Viergever, beste Max. Als hoofd van het ISI was jij betrokken bij alle perikelen rondom
het ontwikkelen van segmentatie en folding methodes specifiek voor het neonatale brein. Dank
voor je overzicht, je constructieve bijdrage aan de manuscripten en de randvoorwaarden die het
mogelijk maakten om de samenwerking tot een succes te maken.
Dr. F. Groenendaal, beste Floris. Bij jou begon mijn onderzoeksavontuur bij de neonatologie met het
onderzoek naar sinustrombose. En ook bij dit promotieonderzoek heb ik dikwijls bij je aangeklopt
met allerhande statistiek en andere vragen. Dank voor je hulp daarbij en ook voor je nuchtere kijk
op de zaken waarbij je telkens alles weer in het juiste perspectief wist te plaatsen en mij weer stevig
met beide benen op de grond kon zetten.
Dr. M.J.N.L. Benders, beste Manon. Als directe begeleider was jij het meest betrokken bij mijn
onderzoek. We hebben samen heel wat meegemaakt en geleerd hoe het is als dingen niet direct
lopen zoals we zelf graag zouden willen. Dankjewel voor alle steun en voor de mogelijkheden die je
me geboden hebt om alles uit het onderzoek te halen. Ik heb enorme bewondering voor jouw
enthousiasme en drive waarmee je je werk doet. De hoeveelheid aan onderzoeksprojecten die jij op
weet te starten en de mooie resultaten die je daarmee behaald zijn bijna niet bij te houden. Veel
succes in je nieuwe positie!
Prof. dr. H. Hulshoff Pol, prof. dr. E. Nieuwenhuis, prof. dr. M. Joëls en prof. dr. A. Bos wil ik hartelijk
danken voor het plaatsnemen in de leescommissie. Also a word of thanks to dr. C. Nosarti for being
part of the reading committee.
Beste Ivana, Pim, Nelly en Sabina, dank voor al jullie inzet om de neonatale segmentatie en cortical
folding mogelijk te maken. En vooral ook dank voor alle uitleg en begrip om deze ‘domme dokter’ in
te wijden in de wereld van de beeldverwerking en alles wat ermee samenhangt.
Beste Alexander, ook jij hebt deze domme dokter het nodige geleerd over DTI, wat geleid heeft tot
een prachtig artikel. Dank daarvoor!
Dear Jessica and Francois, thank you for the possibility to come to Paris. Our collaboration on the
cortical folding data may have taken a while but I think we have a truly wonderful dataset now to
work with. Thanks for your patience with me and for answering all my questions about the program
and the scripts.
222
Colleagues and supervisors in London. Dear Serena, thank you so much for the opportunity to come
to London. It was a most wonderful experience to spend time in your research group and I learned
so much about all aspects of image post-processing. An extra word of appreciation for your
presence during my defence. Antonis and Paul, thank you for guiding me through the steps of the
segmentations itself and the statistics thereafter.
Dear Georgia, Prachi, Kathryn, Rui, Piergiorgio, Fran, David, Anthony, Tom Bibi, Michiel, Emer, Nora,
Vito, and all others, thank you for making my time in London a blast! I enjoyed all our lunch
conversations, dinners and outings. You all made me feel completely at home.
A special word of thanks to all co-authors, both in the Netherlands and abroad, who made the
manuscripts of this thesis possible.
Beste MR laboranten, beste Barbara, Maurice en Bianco, (en iedereen die er verder bij betrokken
was), hartelijk dank voor al jullie tijd en extra inzet bij het maken van alle MRI’s, al dan niet met extra
couveuse, sequenties of patches. Door jullie was het maken van de MRI’s altijd gezellig.
Beste Ineke, Karin, Hanneke, Marian en Anneke, dank voor jullie hulp bij de planning en
administratie van alle MRI’s en alle zaken eromheen.
Ook alle neonatologen, PA’s en verpleegkundigen wil ik graag bedanken voor hun hulp de afgelopen
4 jaar.
Beste (oud)-roomies, de meeste tijd bracht ik natuurlijk met jullie door. Dankjulliewel voor alle
gezelligheid, uitgebreide gesprekken, gedeelde frustraties, plezier tijdens de congressen en al het
andere wat we in de loop der jaren hebben gedeeld. Inge-Lot, als kameroudste blijf jij de stabiele
factor tussen alle wisselingen door. Dankjewel voor je luisterend oor, welwillendheid om altijd mee
te denken en niet te vergeten natuurlijk alle lekkere cakejes en bastogne koeken waarmee je ons
verwende. Britt, mijn onderzoek was eigenlijk het vervolg op dat van jou en hoe vaak ik ‘de nieuwe
Britt’ genoemd ben, zeker op de MRI, valt niet te tellen. Dank voor het delen van je kennis en
ervaring, en niet te vergeten je ‘pixelpoets’-skills. Niek, als enige man in het vrouwenbolwerk kon al
dat gekwebbel je soms gestolen worden, maar stiekem was het toch ook wel erg gezellig. Dank voor
je hulp bij al de computerproblemen en ook de meer technische kanten van de beeldverwerking en
de DTI. Margaretha, onze eerste ontmoeting was op Schiphol toen we samen naar de SPR vlogen en
dat was gelijk een goed begin. Ook nu werken we nog mooi samen op. Ik heb enorm respect hoe jij
alles weet te combineren. Heel veel succes met het afronden van je promotie! Johanneke, wij zaten
niet alleen naast elkaar maar bleken ook vlak bij elkaar te wonen, waardoor we onze gesprekken ook
op de fiets naar huis gewoon door konden zetten. Over de jaren heen hebben we bijna alles wel
besproken en heb ik op jouw ervaring vaak een beroep mogen doen. Gelukkig kunnen we nog
steeds regelmatig bijkletsen. Hilde, jij zat iets verder weg maar dat ging niet ten koste van de
theemomentjes, dr Phil sessies en verdere gezelligheid. Jouw laatste eindsprint ging precies 4
maanden voor de mijne uit en daar hebben we samen veel om kunnen lachen. Lauren, toen ik terug
kwam uit Londen was jij mijn nieuwe buurvrouw en dat was gelijk gezellig. Dank voor alle leuke
223
gesprekken het afgelopen jaar en nog veel succes met het vervolg van je promotie! Laura, toen jij
begon ging ik al bijna weg, maar het was de laatste maanden gezellig om een kamer met je te delen.
Beste Thomas, Joppe, Julia, Kristin, Lisanne, dear Rita, Caterina, Maria Luisa, Simona and Sylvia,
although we did not share the same room, we certainly had a lot of fun together during lunch
breaks, conferences and dinners. Thanks for all the nice times we had together. I wish all of you the
best of luck in completing your work. Beste Nienke, Nathalie en Juliette, dank voor jullie hulp bij
mijn onderzoek en veel succes!
Beste kinderartsen, arts-assistenten en verpleegkundigen in Almere. Bij jullie begon ik na ruim 4 jaar
onderzoek weer in de kliniek en ik had me geen betere plek kunnen wensen. Dankjulliewel!
Lieve Rachel, Jacinthe, Frea, Nathalie, Karlijn, Bram, Iris, Linda (en aanhang), op allerlei verschillende
manieren kennen we elkaar, en het is altijd gezellig. Dank voor jullie vriendschap.
Groepje 2, Bart-Jan, Jonathan, David, Annette, Rachel en Amani, hier is dan boekje nummer 3.
Dankjulliewel voor alle support, diepgaande gesprekken, gezellige etentjes en uitgebreide
whatsapp-conversaties. Ik hoop dat we samen nog veel mooie dingen gaan meemaken!
Lieve paranimfen, Frea en Marjolein, niet alleen tijdens mijn verdediging staan jullie achter me maar
ook in de voorbereiding hebben jullie me erg geholpen.
Frea, in jaar 1 kwamen we elkaar toevallig tegen op het perron en bleken onze ouders bij elkaar in de
buurt te wonen. Dat was het begin van een mooie vriendschap. Dankjewel voor alle gezelligheid en
steun. Grappig hoe ook in een heel andere ziekenhuis toch een hoop dingen hetzelfde blijken te
gaan en we onze ervaringen dus goed kunnen delen.
Marjolein, je bent mijn kleine zusje maar tegelijk kan ik op veel gebieden van je leren. Dankjewel
voor al je hulp bij het afronden van mijn promotie en ook voor alle gezelligheid daarbuiten.
Lieve familie Kersbergen en Pieters, het was niet altijd makkelijk te begrijpen wat ik nu precies deed
tijdens dat promotieonderzoek, maar dat maakte jullie belangstelling er niet minder op. Hier is het
tastbare resultaat!
Lieve Wouter, Reinier, Marjolein, Rachel en Marlous. In de tijd dat ik mijn onderzoek deed, zijn jullie
allemaal ook op je plek terecht gekomen. Ik vind het geweldig om te zien hoe jullie in het leven
staan en wat jullie allemaal al bereikt hebben. Dankjulliewel voor jullie interesse en hulp, zowel voor
dit boekje als daarbuiten.
Lieve papa en mama, zonder jullie was dit boekje er niet geweest. Dankjulliewel voor jullie continue
steun, lieve woorden, praktische tips, aandacht en liefde. Van jongs af aan hebben jullie me
gestimuleerd om altijd het beste uit mezelf te halen en op zoek te gaan naar nieuwe uitdagingen.
Ook tijdens het onderzoek stonden jullie altijd vierkant achter me. Ik heb ontzettend veel van jullie
geleerd en nog veel meer aan jullie te danken. Dit boekje is voor jullie.
224

Longitudinal Structural Brain Development in Externally Preterm

Uploaded by

Copyright:

Available Formats

Longitudinal Structural Brain Development in Externally Preterm

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Longitudinal Structural Brain Development in Externally Preterm

Uploaded by

Copyright:

Available Formats

Longitudinal brain

Voor mijn ouders

L o n gi t u d i n a l b ra i n d eve lo pme nt in ex tre me ly pre te rm newb or ns

T h e si s , U tre c h t U n i ve rs ity, w ith a s ummary in Dutch

Cove r de sign & il l u st rat i o n

A l l r i g h ts re s e r ve d . N o part o f th is th e s is may be re pro duced or tra nsmitted in a ny for m b y a ny mea ns,

Longitudinale hersenontwikkeling bij extreem te vroeg geboren kinderen

24 part 1 108 part 3

resonance imaging in relation to weighted imaging and network analysis

outcome to study brain development

26 chapter 2 110 chapter 6

216 List of abbreviations

Part of this introduction has been adapted from

Neuroimaging of white matter injury, intraventricular

Figure 4 Preterm infant, GA 25 weeks. MRI, T2-weighted

Figure 5 Preterm infant,

Figure 6 Preterm infant, GA

Figure 7 Examples of two

D i ffu si on w e igh t e d a nd f u nct i o nal im agi ng

Aims and outline of the thesis

Neurodevelopmental outcomes in preterm infants: SD et al. Early detection of periventricular

PLoS One 2014 9:e108904

28 part 1 Q ualitative assessment of magnetic resonance imaging in relation to outcome

scanned as soon as possible after birth and again around TEA.

C l i n ica l a nd fo l low - u p data

30 part 1 Q ualitative assessment of magnetic resonance imaging in relation to outcome

Ta bl e 1. C linica l character i sti cs

Variable Total Linear Cluster Mixed p-value

27.8 [24.4 – 28.2 [24.7 – 29.7 [26.1 –

1101 [515 – 1164 [695 – 1354 [865 –

Necrotizing enterocolitis requiring

Intraventricular haemorrhage (no/ 56/7/15/12/22 31/4/8/9/15 18/1/3/2/2

Griffiths developmental quotient 1st

Griffiths developmental quotient 2 nd

N = 17 N=7 N=4 N=6

Bayley cognitive composite score 2nd

Bayley total motor composite score

N = 65 N = 41^ N = 13^ N=7

Cerebral palsy (no/yes) 103/9 55/4 22/4 18/1 0.01

Without additional lesions 3 0 3 0

Ta b le 2 . PWM L characteri sti cs

Linear Cluster Mixed Anterior-mid Posterior

Early (n=91) 51 (56) 23 (25) 17 (19) 85 (93) 1 (1)

TEA (n=76) 53 (70) 9 (12) 14 (18) 71 (93) 1 (1)

The values between brackets represent the percentages

Ta b le 3 . P W ML id e nti fi cati on u si ng addi ti onal i magin g metho ds

Early MRI (n=91)

Linear 6 (7) 29 (32) 22 (24) 3 (3)

Cluster 10 (11) 6 (7) 10 (11) 1 (1)

Mixed 11 (12) 11 (12) 4 (4) 6 (7)

The values between brackets represent the percentages

34 part 1 Q ualitative assessment of magnetic resonance imaging in relation to outcome

Laterality Lesion load

Overall Unilateral Bilateral 1-3

5 (6) 15 (16) 76 (84) 33 (36)

4 (5) 11 (14) 65 (86) 50 (66)

DWI + SWI + No SWI available

0 (0) 22 (29) 27 (36)

0 (0) 1 (1) 2 (3)