J. Anat. (2004) 204, pp465–474 Cerebral cortex three-dimensional profiling in human fetuses by magnetic resonance imaging Blackwell Publishing, Ltd. Andrea Sbarbati,1 Francesca Pizzini,2 Paolo F. Fabene,1 Elena Nicolato,1 Pasquina Marzola,1 Laura Calderan,1 Alessandro Simonati,3 Laura Longo,4 Antonio Osculati5 and Alberto Beltramello2 1 Department of Morphological–Biomedical Sciences, Section of Anatomy and Histology, and Department of Neurological and Visual Sciences, Section of Neurology, University of Verona, Italy 2 Service of Neuroradiology, Verona City Hospital, Italy 4 Natural Science Museum, Verona, Italy 5 Department of Medicine and Public Health, University of Insubria, Italy 3 Abstract Seven human fetuses of crown /rump length corresponding to gestational ages ranging from the 12th to the 16th week were studied using a paradigm based on three-dimensional reconstruction of the brain obtained by magnetic resonance imaging (MRI). The aim of the study was to evaluate brain morphology in situ and to describe developmental dynamics during an important period of fetal morphogenesis. Three-dimensional MRI showed the increasing degree of maturation of the brains; fronto-occipital distance, bitemporal distance and occipital angle were examined in all the fetuses. The data were interpreted by correlation with the internal structure as visualized using high-spatial-resolution MRI, acquired using a 4.7-T field intensity magnet with a gradient power of 20 G cm−1. The spatial resolution was sufficient for a detailed detection of five layers, and the contrast was optimized using sequences with different degrees of T1 and T2 weighting. Using the latter, it was possible to visualize the subplate and marginal zones. The cortical thickness was mapped on to the hemispheric surface, describing the thickness gradient from the insular cortex to the periphery of the hemispheres. The study demonstrates the utility of MRI for studying brain development. The method provides a quantitative profiling of the brain, which allows the calculation of important morphological parameters, and it provides informative regarding transient features of the developing brain. Key words development; fetal; neuroimaging; neuronal migration; neuroradiology; subplate zone. Introduction Two- or three-dimensional (2D or 3D) ultrasound is the standard imaging technique used for prenatal diagnosis of suspected fetal abnormalities, but studies have also shown the utility of magnetic resonance imaging (MRI) for the assessment of fetal organs (McCarthy et al. 1985; Mintz et al. 1987; Schierlitz et al. 2001). This approach has proved useful in volumetric analysis of important Correspondence Professor Andrea Sbarbati, Dipartimento di Scienze MorfologicoBiomediche, Sezione di Anatomia ed Istologia, Strada le Grazie 8, 37134 Verona, Italy. T: +39 45 8027155; F: +39 45 8027163; E: andrea.sbarbati@univr.it Accepted for publication 25 March 2004 © Anatomical Society of Great Britain and Ireland 2004 structures like the germinal matrix and lateral ventricles (Kinoshita et al. 2001). However, 3D MRI studies on large populations are still lacking and as a consequence it is difficult to assess the normal pattern of a developing brain. There are large ongoing projects devoted to MRI analysis of fetal tissues (Cohen, 2002), but developmental data should be obtained from different collections of fetuses, to avoid bias due to racial differences or preservation methods. Sectional MRI data provide information on neuronal migration processes, revealing a layered appearance of the cortex and germinal matrix (Kostovic et al. 2002). In the developing brain, seven transient layers are present in the cerebral cortex (i.e. ventricular, periventricular, subventricular, intermediate, subplate zones, cortical plate and marginal zone) and good correlation has been found between histology and MRI data on T1-weighted (T1W) sequences 466 3D MRI of fetal brain, A. Sbarbati et al. at 2.0 T (Kostovic et al. 2002). Five laminar compartments of varying MRI signal intensity have been detected. However, detailed visualization of some layers composing the developing cortex has not yet been achieved. In particular, specific methods for detection of subplate and marginal zones are lacking. High-field-intensity MRI permits a better signal-tonoise ratio with respect to standard MRI using clinical magnets, and could generate better imaging of the thin cerebral layers. Such data could include 3D images, providing a more complete description of the developmental process. In the present work, we studied a series of human fetuses ranging from the 12th to the 16th week of gestational age (GA). During this period important maturational events take place in the developing brain (Chi et al. 1977). We used a paradigm based on 3D reconstruction of the brain obtained by MRI at different field intensities. The data were interpreted by comparison with the internal structure obtained by 4.7T MRI and with histological specimens. The major aim of the study was to evaluate brain morphology in situ, avoiding artefacts due to removal from the cranial cavity, and to describe developmental dynamics during an important period of fetal morphogenesis. A further aim was to relate the 3D morphology of the brain to internal structures, optimizing MRI pulse sequences for visualization of transient layers present during development, such as the subplate and the marginal zone. Materials and methods Seven human fetuses of 65, 85, 100, 105, 110, 116 and 125 mm crown–rump length (CRL), corresponding to GAs ranging from the 12th to the 16th week, were studied. The fetuses were selected from Medical Faculty and Natural Science Museum (Verona, Italy) collections for their good state of preservation. Some of the fetuses had been used in previous MRI studies (Sbarbati et al. 1998). The cadavers were fixed by immersion in 20% formalin, for routine examination, and conserved in this solution. They were removed from this solution only for the time of the scan. Magnetic resonance images at 1.5 T were acquired using a Symphony system (Siemens). Entire fetuses were positioned within a knee coil for a better signalto-noise ratio. Multiplanar reconstruction (MPR) T1W sagittal localizer and fast spin echo (SE) MPR T1W images were acquired. The acquisition parameters were: repetition time (TR) = 1900 ms; echo time (TE) = 4.09 ms; slab thickness = 72 mm; matrix size and fieldof-view (FOV) were adjusted to obtained a voxel size of 0.6 × 0.4 × 0.5 mm3. The total imaging time was 20–25 min. Data were processed on a Sun Advantage window 4.0 workstation. A 3D reconstruction was made of the fetal brains. Subsequently, the 3D brains were visually analysed in the conventional anatomical views (superior, inferior, anterior and posterior). Images in views other than the axial, coronal and sagittal were also taken into consideration. Fronto-occipital distance and bitemporal distance were calculated on the 3D reconstruction, using brains orientated in the axial and coronal planes, respectively. The fronto-occipital distance was the linear distance between the anterior extremity of the frontal pole and the posterior extremity of the occipital pole. The bitemporal distance was the linear distance between the most protruding points of each temporal pole. The interobserver variability of the measurements was less than 1%. For each fetus, the fronto-occipital/bitemporal ratio was calculated. Statistical analysis was carried out using Pearson’s test. The occipital angle was calculated on a lateral projection using two lines tangential to the inferior and superior borders of the occipital lobe. For MRI at 4.7 T, the specimens were removed from formalin, sealed in a polyethylene container and examined using a Biospec Tomograph (Bruker, Karlsruhe, Germany) equipped with a horizontal magnet (Oxford Ltd, Oxford, UK) having a 33-cm bore with a gradient power of 20 G cm−1, and a Silicon Graphics O2 computer. Proton MRI was performed with all the fetuses positioned in a 72-mm internal diameter birdcage coil. Then, 2D and 3D images were acquired. Two-dimensional images were obtained in different planes at high space resolution and using SE sequences. Moderately T2-weighted (T2W) images were acquired with the following parameters: TR = 2000 ms; TE = 60 ms; number of averages (NEX) = 8; FOV = 6 × 6 cm2; slice thickness = 1 mm, matrix size = 256 × 256 corresponding to an in-plane resolution of 0.234 × 0.234 mm2. Strongly T2W images were acquired with: TR = 3000 ms; TE = 80 ms; NEX = 16; FOV = 6 × 6 cm2; slice thickness = 1 mm; matrix size = 512 × 512 corresponding to an inplane resolution of 0.117 × 0.117 mm2. T1W SE images were acquired with TR = 800 ms; TE = 25 ms; NEX = 4; FOV = 6 × 6 cm2; slice thickness = 1 mm and matrix size = 512 × 512. © Anatomical Society of Great Britain and Ireland 2004 3D MRI of fetal brain, A. Sbarbati et al. 467 Three-dimensional images were acquired using a spoiled gradient-echo sequence with the following parameters: TR = 80 ms; TE = 6 ms; FOV = 12 × 6 cm2, slab thickness = 6 cm; matrix size = 256 × 128 × 128 corresponding to a space resolution of 0.469 × 0.469 × 0.469 mm3. MRI data were analysed by comparison with histological sections obtained from ten paraffin-embedded, Nissl- or haematoxylin–eosin-stained fetuses of corresponding gestational age from the collection of the Department of Neurological and Visual Sciences, Section of Neurology, University of Verona. The thickness of the cortex was calculated in both the hemispheres of each fetus using a sectional MRI data set. We analysed 146 axial slices encompassing the whole brain, out of the 256 slices acquired. In each slice, the thickness of the cortex was evaluated at 12 points chosen at regular intervals according to a systematized protocol using Paravision (Bruker) software. In total, 1752 points were evaluated in each cerebral hemisphere. The measurements of the thickness of the cortex were projected on to the telencephalic lateral surface of a 3D anatomical image of the same brain and then displayed in pseudo-colour. Fig. 1 (A) Parasagittal 4.7-T MR image ( T1W) of a representative fetus (CRL 116 mm). The wide space (b) separating the telencephalic vesicles (d) from the skull wall is visible. The ganglionic eminence (c) appears as a high-intensity crescent on the floor of the lateral ventricle (a). (B) MRI of fetal brain (T1W, transverse section at the level of the line shown in A). The medial wall of the hemispheres is the thinnest and shows the fronto-parietal incisure (arrows). (C) Fronto-occipital and bitemporal distances in the examined fetuses. (D) Occipital angle of the examined fetuses. © Anatomical Society of Great Britain and Ireland 2004 Results Three-dimensional morphology of the hemispheres High-spatial-resolution (4.7 T) MRI showed that a wide space, in which developing meninges were visible, separates the developing brain from the cranial wall (Fig. 1). Considering that formalin fixation modifies anatomical volumes, the exact dimension of this space was difficult to calculate. However, visual inspection showed that it was widest at the supero-lateral surface of the brain. In the 3D reconstruction, all the examined brains appeared lissencephalic. The longitudinal cerebral fissure separated the telencephalic vesicles, and the lateral cerebral sulcus (sylvian fissure) was visible on their lateral surface (Fig. 2). The brains showed a similar surface morphology even though their degree of maturation was variable. All the parameters measured (i.e. fronto-occipital distance, bitemporal distance and occipital angle, Fig. 1C,D) were significantly correlated to CRL (correlation coefficient R = 0.965, 0.915 and − 0.902, respectively; P < 0.005). The plot in Fig. 1(C) shows that, in the interval of gestational age considered, both parameters show a linear increase, although with 468 3D MRI of fetal brain, A. Sbarbati et al. Fig. 2 Three-dimensional reconstruction of a series of fetal brains (MRI 1.5T). (A) Superior view. (B) Lateral view. The fronto-occipital distance of the brains has been normalized at the same value for all the fetuses. In Figs 2–4, the CRL is reported for each fetus to allow comparison of their morphological evolution. In B, the arrow indicates the parieto-occipital fossa. The dotted line surrounds the fronto-parietal prelobe, which appears as a single structure. The region between the dotted line and the arrow belongs to the fronto-parietal prelobe. The occipital angle is also shown. different rates. The elongation of the fronto-occipital distance exceeded that of the bitemporal distance. As a consequence, the fronto-occipital/bitemporal ratio progressively increased with gestational age (from 1.1 to 1.2, 1.2, 1.2, 1.2, 1.3 and 1.4). In 65–105 mm fetuses, the longitudinal cerebral fissure (Fig. 2) was relatively wide and shaped like an hourglass, due to the presence of a minimal interhemispheric distance around its median segment. Two incisures (i.e. fronto-parietal and parieto-occipital) were visible at the medial border of the hemispheres (Fig. 1B). In 110–125 mm fetuses, the longitudinal fissure appeared more regular and narrower. The two incisures remained visible but were less deep. The presence of these incisures was confirmed in high-resolution sectional images, in which they appeared as two notches on the medial wall of the telencephalic vesicle (Fig. 1B). In 65–105 mm fetuses, it was impossible to distinguish a clear separation between the frontal and parietal lobes: for descriptive purposes, we identified a single structural unit that could be termed the frontoparietal prelobe (Fig. 2). This crescent-shaped structure surrounding the lateral cerebral sulcus displayed a smooth regular surface. Its anterior extremity was conical with a rather pointed apex. The extremities of the two hemispheres were convergent (Fig. 3). In 110– 125 mm fetuses, the frontal poles had rounded extremities that, in a frontal view, showed approximately parallel axes. Sectional MRI (Figs 4 and 5) showed that the wall is thicker in areas surrounding the ganglionic eminence and progressively thinner distally (i.e. near the longitudinal cerebral fissure and at the apex of the frontal pole). In all the fetuses, the parietal lobe appeared as a slightly convex area with a smooth surface. On the lateral surface of the brain (Fig. 2) a depressed area was visible between the developing parietal and occipital regions, in correspondence with the dorsal–posterior border of the lateral sulcus (parieto-occipital fossa, Fig. 2B). Owing to the presence of the lateral sulcus and of the parieto-occipital fossa, the developing telencephalon appeared divisible into two main components that could be identified as the fronto-parietal and temporo-occipital prelobes (Fig. 2B). In 65–105 mm fetuses, the temporal pole appeared as a curved tube, with a medial concavity and a thin rostral extremity (Figs 2 and 3). The axes of the poles were rostrally convergent. In 110–125 mm fetuses, the temporal poles progressively enlarged and their rostral extremities showed a rounded shape. The axes of the poles were caudally convergent. In the 65–100 mm fetuses in our series, the occipital lobe was poorly developed, appearing as a posterior extension of the temporal lobe (Fig. 4). In fetuses of 105–125 mm CRL, it showed a round shape and a smooth regular surface. Its growth was the main reason for the greater length of the antero-posterior axis compared with the bitemporal axis. Sectional MRI showed that the occipital cortex was thin, in particular at the distal end of the pole and in areas near the inferior surface of the brain. © Anatomical Society of Great Britain and Ireland 2004 3D MRI of fetal brain, A. Sbarbati et al. 469 Fig. 3 Three-dimensional reconstruction of a series of fetal brains (MRI 1.5T). (A) Anterior view. (B) Inferior view. The lateral cerebral sulcus was visible in all the fetuses (Fig. 2). Comparing fetuses at different degrees of maturity, a clear difference was noted in the morphology of the lateral sulcus. In 65–100 mm fetuses, the sulcus appeared as a light depression of the lateral surface limited by a curved border, and the opercula were not yet formed. © Anatomical Society of Great Britain and Ireland 2004 The parieto-occipital fossa reached the medial border of the hemisphere, separating the parietal from the occipital lobes. In some fetuses the borders of the developing opercula formed an angle of about 30°, opening ventrally. Sectional images showed that at the level of the future insular cortex there was a thickening 470 3D MRI of fetal brain, A. Sbarbati et al. Fig. 4 (A) Three-dimensional reconstruction of a series of fetal brains, posterior view (MRI 1.5T). (B) Sectioned 3D reconstruction of a series of fetuses, anterior view (MRI 1.5T). The coronal section of the brain shows the decreasing thickness of the cortex, starting from the ganglionic eminence. of the lateral wall of the hemisphere. This latter structure can be subdivided into a medial and a lateral portion. The medial portion showed a high intensity signal, thus corresponding to the germinal matrix. The lateral portion showed an intensity similar to that of the developing cortex and was composed of basal nuclei (Figs 4B, 5 and 6). In 105–110 mm fetuses, the development of the lateral sulcus was characterized by an expansion of the opercula. The future insular cortex displayed a square © Anatomical Society of Great Britain and Ireland 2004 3D MRI of fetal brain, A. Sbarbati et al. 471 Fig. 5 Correlation of the 3D morphology with internal structures (fetus 125 mm CRL). Using a sectional data set (A) the thickness of the cortex has been projected on to the telencephalic lateral surface and is displayed in pseudo-colour (B). shape with four margins (i.e. anterior, dorsal, posterior and inferior). The surface corresponding to the developing insular cortex showed a regular surface with a slight central elevation. In the fetus of 125 mm CRL, the lateral sulcus appeared as a deep depression and the developing insular cortex was triangular in shape and delimited by a superior– anterior and an inferior–posterior margin. The two margins formed a ventrally opening angle of about 18°. Correlation of the 3D morphology with internal structures Comparing sectional data with the 3D reconstruction, it was possible to project the thickness of the developing brain structures on to the telencephalic lateral surface (Fig. 5). The maps showed that an area of thick cerebral cortex surrounds the projection of the ganglionic eminence on the surface. Thinner areas were found at the apex of the frontal lobe and in the temporal occipital lobe. Areas of intermediate thickness were found between the dorsal ending of the lateral sulcus and the longitudinal fissure. Fig. 6 (A) MRI (4.7T) of fetal brain ( T2W, transverse section, CRL 85 mm). The asterisk indicates the ganglionic eminence. Cortical areas marked by squares are enlarged in B and C. a, liquor; b, marginal zone; c, cortical plate; d, subplate zone; e, intermediate (fetal white matter), subventricular periventricular zones; f, ventricular zone. ( D) High-resolution T2W MRI of the fetal cortex (left panel) in comparison with a low-magnification histological section (right panel, haematoxylin–eosin staining). 1, ventricular zone; 2, periventricular zone; 3, subventricular zone; 4, intermediate zone; 5, subplate zone; 6, cortical plate; 7, marginal zone. Interpretation of the layers was made in accordance with Kostovic et al. (2002). A, ×3.4; B and C, ×13.6; D, ×34. © Anatomical Society of Great Britain and Ireland 2004 Optimization of sequence for analysis of the developing cortex In our paradigm, a quite high spatial resolution was obtained using a 4.7-T field intensity magnet with a gradient power of 20 G cm−1. At these conditions, the spatial resolution was sufficient for detailed visualization of the marginal and subplate zones. The contrast was optimized using sequences with different degrees of T2 weighting. In particular, for the marginal and 472 3D MRI of fetal brain, A. Sbarbati et al. subplate zones, a moderately T2W pulse sequence appeared capable of providing considerable detail with respect to the adjacent layers (Fig. 6). Strongly T2W sequences resulted in images of lesser quality and did not seem adequate for detection of marginal zones, due to the strong intensity of the signal emitted by cerebrospinal fluid. Using the proposed moderately T2W sequence, the signal emitted by the liquor was reduced and the thickness of the marginal layer was quite evident. Layers in the developing cortex; T1W images Using T1W images, the more cellular layers generated a more intense signal (Fig. 1). As a consequence, the ventricular zone and the germinal matrix of the ganglionic eminence, both characterized by a cellular pattern, appeared clearly in the images. The subventricular zone appeared to be of intermediate intensity. Strips of lowintensity material were visible in some areas, probably due to bundles of axons. The subplate zone emitted a signal of low intensity. The cortical plate, characterized by a cellular pattern, appeared clearly in T1W images. The marginal zone emitted a signal of low intensity. Layers in the developing cortex; optimized T2W images Using the proposed T2W pulse sequence a detailed analysis of the developing cortex was achieved (Fig. 6). The ventricular zone and the germinal matrix of the ganglionic eminence, both characterized by a cellular pattern, appeared dark in images. The subventricular zone appeared of intermediate intensity. Strips of lowintensity material were visible in some areas, probably due to bundles of axons. The subplate zone emitted a signal of strong intensity (Fig. 6C, layer marked d ). The cortical plate, characterized by a cellular pattern, appeared dark in images (Fig. 6C, layer marked c). The marginal zone emitted a signal of strong intensity (Fig. 6C, layer marked b). The intense emitted signal makes it possible to detect small irregularities in the layer. Discussion Advantages and methodological limitations of 3D MRI analysis The development of the brain is a complex event, the spatial organization of which is difficult to study using conventional anatomical methods, and difficult to explain in simple mechanical terms (Sidman & Rakic, 1982). In fetuses, the brain parenchyma is a delicate, almost gelatinous structure and its removal from the cranial cavity may cause deformation, hampering the quantitative evaluation of important parameters relevant to the stage of maturation. In particular, the dissection of immature human fetal specimens may be difficult because the partially myelinated brain does not fix well, even after preservation in formaldehyde for several years, due to its high water content (Kier & Truwit, 1996). The present study demonstrates the utility of MRI in the study of brain development. The 3D reconstruction obtained is quite similar to those provided by necropsy but with an absence of mechanical artefacts. The method also provides a quantitative profiling of the brain, allowing the calculation of important morphological parameters. In addition, 3D MRI reconstruction appears to provide important informative regarding transient features of the developing brain. The data clearly demonstrate some transient features such as the parieto-occipital fossa or the incisures visible between the lateral and medial walls of the hemispheres. However, caution is necessary in examination of the aldehyde-fixed brain, considering the changes induced in volumes and conformation. In addition, it is important to avoid overestimation of data obtained from collection material, the normality of which is not always evident. A study based on gross photographs of fixed brain by Chi et al. (1977) reported difficulty in distinguishing between an early gyral depression and an artefactual tear or fold. In situ MRI studies could be useful to minimize this kind of artefact. In their study Chi et al. (1977) did not show photographs, only very schematic drawings; however, our data seem to confirm almost entirely their classical description. The only differences concern description of the development of the lateral sulcus. Chi et al. (1977) state that the superior border of the insula can be designated the circular sulcus, and the inferior border the proper sylvian fissure. According to their description the circular sulcus extends rostrally first, then curves around in a circular pattern and is joined by the posterior end of the sylvian fissure. In our specimens, this progression is not visible, and in 100 – 125 mm fetuses the inferior border appears more evident than the superior border. © Anatomical Society of Great Britain and Ireland 2004 3D MRI of fetal brain, A. Sbarbati et al. 473 Mapping of sectional parameters on 3D reconstruction We correlated 3D MRI morphology with parameters obtained by sectional analysis, mapping cortical thickness on to the hemispheric surface and thereby clearly showing the thickness gradient from the insular cortex to the periphery of the hemispheres. It is likely that this approach, together with improved MRI visualization of the cortex, could allow mapping of features such as the thickness of single layers, thus providing new data on the evolution of the neuronal migratory pathway. Visualization of layers in the developing cortex Aldehyde fixation changes the relaxation characteristics of the tissue, and therefore the sequences usually employed for in vivo study of the brain must be reconsidered in studies on necropsy material. To obtain a complete visualization of the developing cortex in the age window taken into consideration, we propose using two pulse sequences. As has been amply demonstrated, the cell-rich layers (i.e. germinal matrix, ventricular layer and cortical plate) emit an intense signal with T1W pulse sequence. Water-rich subplate and marginal zones emit an intense signal with T2W sequence. This sequence also provides good visualization of the subventricular layer. The signal originating from several cellular layers is markedly reduced in T2W images, whereas it is strong in T1W images. In the present study, the results obtained by this approach confirm and extend previous demonstrations of the layered appearance of the fetal cerebral wall (Girard & Raybaud, 1992; Brisse et al. 1997). In particular, we achieved a detailed visualization of the subplate zone. This transient structure is the 5th layer of the developing cortex. In both primates and humans, it has been shown that thalamocortical afferents accumulate in the subplate zone before entering the cortical plate. The transient subplate zone is still well developed in the newborn frontal cortex (Kostovic et al. 1989). At histology, this zone is characterized by a strongly hydrophilic matrix that is rich in acid mucopolysaccharides. At histochemistry, the subplate zone is stained by Alcian blue (Kostovic et al. 2002). By contrast, adjacent zones (i.e. intermediate zone/fetal white matter and cortical plate) are Alcian negative. Structurally, the subplate zone contains a large extracellular space through which neurons and growth cones migrate. In © Anatomical Society of Great Britain and Ireland 2004 a detailed study of the developing human cerebrum, Kostovic et al. (2002) found that changes in the MRI lamination pattern of the human fetal cerebral wall are predominantly caused by changes in the subplate zone. Histochemical staining of the extracellular matrix enabled selective visualization of the subplate zone and correlation with an increase in MRI signal intensity in the subplate zone and ingrowth and accumulation of thalamocortical and corticocortical afferents and their subsequent relocation to the cortical plate. The authors concluded that dynamic changes in the MRI appearance of the subplate zone and histochemical staining of its extracellular matrix can be used as indirect parameters for an assessment of normal vs. disturbed unfolding of crucial histogenetic events that are involved in prenatal shaping of the human cerebral cortex. However, in the paper by Kostovic et al. (2002), the subplate zone appeared as a layer of low MRI signal intensity at all the gestational ages studied. Our work expands the results of Kostovic et al. (2002), demonstrating that, using a 4.7-T magnet (with a smaller bore and a higher signal-to-noise ratio with respect to clinical 2-T instruments), it is possible to obtain higher spatial resolution (117 vs. 406 µm) reducing partial volume artefacts in a structure with seven layers and a total thickness in large areas inferior to 3 mm. In addition, we demonstrate that the signal intensity of the subplate zone is dependent on the pulse sequence used, with T2W images proving more effective in visualizing the structure. Final considerations The present results demonstrate that MRI technology allows brain profiling in human fetuses from the 12th to the 16th week of gestational age. In the same specimens, high-field-intensity MRI provides a detailed analysis of the layered cerebral cortex. If large virtual collections of developing brains were available, this would facilitate comparisons between data from different countries and vertebrate species. Such data could be useful in understanding neuronal migration disorders (Barkovich, 1995; Barkovich et al. 1996). References Barkovich AJ (1995) Pediatric Neuroimaging. New York: Raven Press. Barkovich AJ, Kuzniecky RI, Dobyns WB, Jackson GD, Becker LE, Evrard P (1996) A classification scheme for malformations of cortical development. Neuropediatrics 27, 59 – 63. 474 3D MRI of fetal brain, A. Sbarbati et al. Brisse H, Fallet C, Sebag G, Nessmann C, Blot P, Hassan M (1997) Supratentorial parenchyma in the developing fetal brain: in vitro MR study with histologic comparison. AINR Am. J. Neuroradiol. 18, 1491–1497. Chi JG, Dooling EC, Gilles FH (1977) Gyral development of the human brain. Ann. Neurol. 1, 86 – 93. Cohen J (2002) Embryology. Embryo development at the click of a mouse. Science 297, 1629. Girard NJ, Raybaud CA (1992) In vivo MRI of fetal brain cellular migration. J. Comput. Assist. Tomogr. 16, 265 –267. Kier EL, Truwit CL (1996) The normal and abnormal genu of the corpus callosum: an evolutionary, embryologic, anatomic, and MR analysis. AJNR Am. J. Neuroradiol. 17, 1631–1641. Kinoshita Y, Okudera T, Tsuru E, Yokota A (2001) Volumetric analysis of the germinal matrix and lateral ventricles performed using MR images of postmortem fetuses. AJNR Am. J. Neuroradiol. 22, 382–388. Kostovic I, Lukinovic N, Judas M, Bogdanovic N, Mrzljak L, Zecevic N, et al. (1989) Structural basis of the developmental plasticity in the human cerebral cortex: the role of the transient subplate zone. Metab. Brain Dis. 4, 17–23. Kostovic I, Judas M, Rados M, Hrabac P (2002) Laminar organization of the human fetal cerebrum revealed by histochemical markers and magnetic resonance imaging. Cereb. Cortex 12, 536 –544. McCarthy SM, Filly RA, Stark DD, Hricak H, Brant-Zavadsky MN, Callen PW, et al. (1985) Obstetrical magnetic resonance imaging: fetal anatomy. Radiology 154, 427– 432. Mintz MC, Grossman RD, Isaacson G, Thickman DI, Kundel H, Joseph P, et al. (1987) MR imaging of the fetal brain. J. Comput. Assist. Tomogr. 11, 120 –123. Sbarbati A, Marzola P, Simonati A, Nicolato E, Osculati F (1998) High-field magnetic resonance imaging of the developing human brain from the 10th to the 16th week of gestational age. Acta Anat. 163, 39 – 46. Schierlitz L, Dumanli H, Robinson JN, Burrows PE, Schreyer AG, Kikinis R, et al. (2001) Three-dimensional magnetic resonance imaging of fetal brains. Lancet 357, 1177–1178. Sidman RL, Rakic P (1982) Development of the human central nervous system. In Histology and Histopathology of the Nervous System (eds Haymaker W, Adams RD), pp. 3–133. Springfield: Charles. Thomas Publisher. © Anatomical Society of Great Britain and Ireland 2004