Zurich Open Repository and
Archive
University of Zurich
Main Library
Strickhofstrasse 39
CH-8057 Zurich
www.zora.uzh.ch
Year: 2005
Coronal thick CT reconstruction: an alternative for initial chest radiography
in trauma patients
Alkadhi, Hatem ; Baumert, Bernhard ; Wildermuth, Simon ; Bloch, Konrad E ; Marincek, Borut ;
Boehm, Thomas
Abstract: It has been proposed that the imaging workup of trauma patients be accelerated by omitting
the initial chest radiography (CR) and directly performing a computed tomography (CT); however, the
baseline CR is then lacking. The purpose of this study was to assess if coronal thick reconstructions
generated from chest CT could present an adequate alternative for CR. Sixty trauma patients underwent
bedside CR and multidetector row chest CT in the emergency room. The image quality of thoracic
anatomical structures, the diagnostic accuracy for chest pathology, and the depiction of indwelling devices
were assessed on both modalities. Main pulmonary arteries and perihilar bronchi were equally visualized
with both modalities. Central bronchi, retrocardial lung parenchyma, diaphragm, descending aorta, and
vertebral pedicles were better visualized on thick CT reconstructions, whereas peripheral lung vessels were
better depicted on CR (p<0.05). The accuracy to delineate various pathological findings did not differ
between both modalities, except for a higher sensitivity to diagnose bronchial cuffing on CR (p<0.05).
The location of indwelling devices was similarly and correctly depicted with both modalities. Coronal
thick CT reconstructions provide a similar image quality and diagnostic accuracy compared with CR.
These reconstructions may serve as an equivalent baseline image in trauma patients in whom emergency
radiological evaluation has to be accelerated
DOI: https://doi.org/10.1007/s10140-005-0432-1
Posted at the Zurich Open Repository and Archive, University of Zurich
ZORA URL: https://doi.org/10.5167/uzh-156179
Journal Article
Published Version
Originally published at:
Alkadhi, Hatem; Baumert, Bernhard; Wildermuth, Simon; Bloch, Konrad E; Marincek, Borut; Boehm,
Thomas (2005). Coronal thick CT reconstruction: an alternative for initial chest radiography in trauma
patients. Emergency Radiology, 12(1-2):3-10.
DOI: https://doi.org/10.1007/s10140-005-0432-1
Emerg Radiol (2005) 12: 3–10
DOI 10.1007/s10140-005-0432-1
ORIGINA L ARTI CLE
Hatem Alkadhi . Bernhard Baumert .
Simon Wildermuth . Konrad E. Bloch .
Borut Marincek . Thomas Boehm
Coronal thick CT reconstruction: an alternative for initial
chest radiography in trauma patients
Received: 25 November 2004 / Accepted: 15 June 2005 / Published online: 9 November 2005
# Am Soc Emergency Radiol 2005
Abstract It has been proposed that the imaging workup of
trauma patients be accelerated by omitting the initial chest
radiography (CR) and directly performing a computed
tomography (CT); however, the baseline CR is then
lacking. The purpose of this study was to assess if coronal
thick reconstructions generated from chest CT could
present an adequate alternative for CR. Sixty trauma
patients underwent bedside CR and multidetector row chest
CT in the emergency room. The image quality of thoracic
anatomical structures, the diagnostic accuracy for chest
pathology, and the depiction of indwelling devices were
assessed on both modalities. Main pulmonary arteries and
perihilar bronchi were equally visualized with both
modalities. Central bronchi, retrocardial lung parenchyma,
diaphragm, descending aorta, and vertebral pedicles were
better visualized on thick CT reconstructions, whereas
peripheral lung vessels were better depicted on CR
(p<0.05). The accuracy to delineate various pathological
findings did not differ between both modalities, except for
a higher sensitivity to diagnose bronchial cuffing on CR
(p<0.05). The location of indwelling devices was similarly
and correctly depicted with both modalities. Coronal thick
CT reconstructions provide a similar image quality and
diagnostic accuracy compared with CR. These reconstrucH. Alkadhi . B. Baumert . S. Wildermuth .
B. Marincek . T. Boehm
Institute of Diagnostic Radiology,
University Hospital Zurich,
Zurich, Switzerland
K. E. Bloch
Pulmonary Division, Department of Internal Medicine,
University Hospital Zurich,
Zurich, Switzerland
T. Boehm (*)
Department of Radiology, Spitäler Chur AG,
Loestrasse 170,
CH-7000 Chur, Switzerland
e-mail: Thomas_Boehm@gmx.net
Tel.: +41-81-2566452
Fax: +41-81-2566685
tions may serve as an equivalent baseline image in trauma
patients in whom emergency radiological evaluation has to
be accelerated.
Keywords Emergency . Chest radiography . Computed
tomography . Coronal thick reconstruction
Introduction
The emergency management of a trauma patient relies on
the concept of the advanced trauma life support (ATLS),
which has been developed in response to the need for a
safe, consistent, standardized, and effective way to initially
evaluate and resuscitate patients with multiple injuries
[1–3]. The term “golden hour” characterizes the fact that
the morbidity and mortality of trauma patients are affected
if care is not instituted within the first hour after injury. To
comply with this concept, radiological examinations must
be fast, systematic, and as complete as possible.
In patients with acute chest trauma, bedside chest
radiography (CR) is the most common initial imaging
workup method in the emergency room [4]. CR is used to
screen for the presence of life-threatening conditions that
necessitate immediate intervention, to evaluate injuries that
require intervention without further diagnostic workup, to
assess abnormalities that require further investigation, or to
document the positioning of indwelling devices. Especially
if the presumed injury did not severely affect the chest, a
CR might suffice, obviating the need for a chest computed
tomography (CT). Hence, chest CT is indicated only in a
subset of patients depending on the severity and site of
injury and on the hemodynamic stability of the patient.
Chest CT should be also performed when the image quality
of bedside CR is low, having limited diagnostic accuracy,
which is often the case in the emergency setting [5–7].
Since time is crucial in the emergency workup of the
trauma patient, it would be desirable to gain all necessary
information by a single imaging method. Therefore,
omitting CR in patients who anyway undergo CT of the
chest as part of their imaging evaluation was suggested by
4
several authors [8–10]. This would also result in a reduction of patient discomfort, costs, and irradiation [11]. On
the other hand, the initial CR serves as an important baseline examination for further treatment monitoring and is
considered a necessary tool by most clinicians. Therefore,
it is still performed as a part of the initial workup in most
trauma centers.
New-generation multidetector row CT (MDCT) scanners acquire high-resolution data with nearly isotropic
voxels, allowing multiplanar reconstructions (MPR) of the
CT data in any arbitrary plane. When performing coronal
thick MPR from MDCT data of the chest, images can be
obtained, which qualitatively resemble a conventional
bedside CR. When it has been decided that the imaging
evaluation of trauma patients be accelerated by immediately performing a CT (and to omit the CR), such coronal
thick MPR could serve as an alternative equivalent for the
initial bedside CR. Time could be saved in the first critical
hour of emergency management, without lacking a baseline examination for further treatment monitoring and
follow-up. We therefore assessed the image quality and
diagnostic accuracy of coronal thick MPR reconstructed
from MDCT data and compared them with the initial
bedside CR of trauma patients in the emergency room.
Materials and methods
From June to October 2003, 60 trauma patients (26 women,
34 men; age range 17–67 years, mean age 48 years) were
referred to our emergency department, a level I European
trauma center, for an MDCT of the chest. Causes of injuries
were traffic accidents (n=29), falls (n=17), and domestic
(n=9) and occupational accidents (n=5). All patients
initially received a bedside CR and, afterwards, a chest
MDCT within the first hour of their stay in the emergency
room. The patients had the following extrathoracic injuries:
head injuries in 46%; renal, splenic, and liver injuries in
27%; sacral, lumbar, or cervical spine fractures in 71%;
and upper or lower extremity fractures in 78% of the
patients. Fifty-six patients (93%) underwent, in addition,
an abdomino–pelvic CT, and 51 patients (85%), an additional head CT. The radiological workup in all patients
was not performed for study purposes but for clinical
indications. The study protocol was approved by the local
ethics committee.
a storage phosphor plate system (ADC Compact with ADC
MD40 phosphor plates, Agfa-Gevaert, Mortsel, Belgium).
For image postprocessing, the routine parameters defined
at our institution for bedside CR were used.
Multidetector row CT
CT scans were performed on a 16-channel MDCT scanner
(Sensation 16, Siemens Medical Solutions, Forchheim,
Germany), which is situated in the emergency room. If
possible, CT was acquired with inspiratory breath-hold and
with elevated arms. For the whole body examination,
150 ml nonionic, iodinated, low-osmolar contrast agent
(Visipaque 270, Amersham Health, Buckinghamshire, UK)
was injected into a cubital vein at a flow rate of 3 ml/s and
subsequently flushed by 30 ml saline fluid using a power
injector (Envision CT, Medrad Inc., Indianola, PA, USA).
The imaging volume extended from the apices of the lungs
to the diaphragm. The scanning parameters were the
following: collimation, 16×1.5 mm; table feed, 24 mm/
rotation; increment, 1 mm; tube potential, 140 kV; tube
current, 140 mA. Two sets of axial images were reconstructed: The first set, with a slice width of 5 mm and a
reconstruction interval of 4 mm, was used for filming and
hospital-wide electronic image distribution, and the second
set, with a slice width of 2 mm and a reconstruction increment of 1 mm, was used for image reading and postprocessing. Both sets of axial images were reconstructed twice
using a medium soft kernel (BF30) for soft tissue imaging
and a high-resolution kernel (BF60) for lung imaging.
Coronal thick MPR of the chest (MPR thickness, 500 mm)
was reconstructed from the 2 mm/1 mm BF60 axial
datasets in the coronal plane using the standard 3D postprocessing software of the secondary CT console (Sensation Wizard, software version VA70, Siemens Medical
Solutions). The large standardized MPR thickness of
500 mm was used to take into account differences in
patient size and morphology and to avoid losing time for
individual patient adjustment. The CT table, which would
otherwise interfere with the patients’ structures, was removed from the dataset using the standard 3D image editing
tools. Window settings for MPR were adjusted manually
by two radiologists in consensus to obtain an image appearance that is similar to a correctly exposed bedside CR.
Both CR and coronal thick MPR were printed out on
hard copy films (size, 14×11 cm) using a laser imaging
system (Drystar 5500, Agfa-Gevaert).
Chest radiography
The bedside CR was obtained in the supine position with
a focus-film distance of 100 cm using a ceiling-mounted
x-ray setup in the emergency room (Siemens Mobilett XT,
Siemens, Erlangen, Germany). A fixed tube voltage of
75 kV and a tube current of 3.0 to 3.6 mA were manually
adjusted by the x-ray technician according to the patient’s
size. No automatic exposure control and no antiscatter
grids were employed. When possible, the bedside CR was
taken in inspiration. Image acquisition was performed with
Image reading
Image readout was performed in two steps. First, image
reading was performed by one experienced radiologist
using the axial CT images and all other available
images, including follow-up evaluations and all clinical
data. The results of this reading were used as gold
standard for the assessment of diagnostic accuracy with
CR and thick MPR. This assessment was limited to a
5
binary scale (i.e., pathological changes of a certain
anatomical structure present or not). In a second step,
10 weeks after the first reading, the hardcopies of the
CR and thick MPR were presented in a random order to
two independently working chest radiologist. They were
blinded to the clinical data, the results of the gold
standard reading, and to the results of the other viewing
modality. The reviewers were asked to rate each image
on a defined Likert scale with respect to the exposure
quality, windowing, artifacts, visibility of critical anatomic structures, pathological findings, and indwelling
devices. More specifically, the following scores were
applied: exposure quality of CR: 1=normal exposure;
2=underexposure; 3=overexposure. Stair step artifacts of
the bronchi on thick MPR were scored as the following:
1=no stair step artifacts; 2=stair step artifacts less than
1 mm; 3=stair step artifacts more than 1 mm; 4=severe
distortion of the bronchial wall due to stair step artifacts.
The anatomical landmarks, including central and perihilar bronchi, lung vessels in the lung periphery, central
pulmonary arteries, retrocardial lung parenchyma, diaphragm, descending aorta, and pedicles of the thoracic
spine, were rated on both modalities with the following
scores: 1=excellent visibility; 2=good visibility; 3=fair
visibility; 4=bad visibility; 5=anatomical structure not
visible.
The pathological findings, including pneumothorax,
bronchial cuffing, perihilar congestion, atelectasis, consolidation/lung contusion, pleural effusion/hemothorax, pleural capping, soft tissue emphysema, fractures of the ribs
and shoulder girdle, mediastinal widening, pericardial effusion, pneumomediastinum/pneumopericardium, and vertebral fractures, were assessed on both modalities using the
following scores: 1=definitely present; 2=probably present;
3=indefinite; 4=probably not present; 5=definitely not
present.
The indwelling devices, including endotracheal tube,
central venous catheter, and pleural drainage tube, were
evaluated on both modalities using the following scores:
1=definitely correct location; 2=probably correct location;
3=indefinite, assessment not possible; 4=probably incorrect location; 5=definitely incorrect location.
Statistical analysis
Statistical analysis was performed using a commercially
available software (SSPS 11.5 for Windows, SPSS Inc.,
Chicago, IL, USA). Interobserver agreement between both
readers was calculated by using κ statistics. According to
Landis and Koch [12], a κ value of 0 indicated poor
agreement; a κ value of 0.01–0.20, slight agreement; a κ
value of 0.21–0.40, fair agreement; a κ value of 0.41–0.60,
moderate agreement; a κ value of 0.61–0.80, good agreement; and a κ value of 0.81–1.00, excellent agreement.
Subjective certainty was calculated by reclassifying the
initial scores: former scores 1 and 5 (definitely present/
definitely not present) were pooled to score 1 (high
diagnostic certainty), and scores 2 and 4 (most probably
present/most probably not present) were pooled to score 2
(medium diagnostic certainty); score 3 (assessment not
possible) remained score 3 (low diagnostic certainty).
Comparisons of these scores from both viewing modalities
were made using the Wilcoxon signed-rank test. For the
evaluation of the diagnostic accuracy of pathological
findings and location of indwelling devices, scores 1 and
2 (definitely and probably present/correct location) were
categorized as pathologic/correct location, and scores 4 and
Table 1 Mean scores±standard deviations and interobserver agreement for image quality of anatomical landmarks
Anatomical landmarks
MPR better
Retrocardial lung parenchyma
Right diaphragm
Left diaphragm
Descending aorta
Pedicles of the thoracic spine
Right main bronchus
Left main bronchus
CR better
Peripheral lung vessels—upper
Peripheral lung vessels—lower
Peripheral lung vessels—upper
Peripheral lung vessels—lower
CR and MPR equal
Right main pulmonary artery
Left main pulmonary artery
Right perihilar bronchi
Left perihilar bronchi
CR
right
right
left
left
MPR
p Value
CR κ
MPR κ
2.94±0.96
2.75±0.99
3.16±0.73
3.47±0.96
3.75±1.04
3.35±0.72
3.51±0.78
1.72±0.87
1.95±0.97
2.13±0.89
2.51±0.97
2.24±0.58
2.08±0.58
2.27±0.66
0.015
0.048
0.015
0.006
0.013
0.012
0.016
0.845
0.799
0.811
0.848
0.749
0.798
0.844
0.951
0.837
0.858
0.936
0.819
0.858
0.901
1.79±0.63
1.94±0.51
1.89±0.48
1.72±0.42
2.69±0.54
2.42±0.46
2.33±0.53
2.24±0.56
0.016
0.049
0.018
0.016
0.914
0.897
0.921
0.933
0.851
0.879
0.743
0.833
2.85±0.79
2.66±0.58
3.14±0.84
3.25±0.59
3.27±0.62
3.19±0.45
2.24±0.61
2.31±0.73
0.564
0.317
0.705
0.092
0.791
0.806
0.879
0.835
0.898
0.829
0.917
0.897
CR Chest radiograph, MPR multiplanar reconstruction, CR κ and MPR κ interobserver agreement
6
5 (probably and definitely not present/incorrect) were
categorized as not pathologic/incorrect location. Sensitivity, specificity, accuracy, positive predictive values (PPV),
and negative predictive values (NPV) were then calculated
with the results from axial CT image reading and all other
available images, including follow-up evaluations and all
clinical data as gold standard. Differences between MPR
and portable chest film were computed from chi-square
tests of contingency. Statistically significant differences
were indicated by p values of less than 0.05.
Image quality
The exposure quality of CR was rated as normal in 44
patients (73%) by reader 1 and in 46 (77%) by reader 2, as
underexposed in 12 patients (20%) by reader 1 and in 11
patients (18%) by reader 2, and as overexposed in 4
patients (7%) by reader 1 and in 3 patients (5%) by reader
2. Thick MPR showed, on average, stair step artifacts of the
both main bronchi less than 1 mm (mean scoreMPR, both
readers, right=1.53±0.47, mean scoreMPR, both readers, left=1.58±
0.27).
Results
Anatomical landmarks
All 60 CR and MDCT examinations were performed
without any technical complications. The mean time
interval from the completion of the CR through the
completion of the MDCT scan was 28±10 min (±standard
deviation). The reconstruction of thick MPR required an
average time of 31 s (range 19–41 s).
Fig. 1 A 47-year-old male patient after motorcycle accident.
There was no pathological finding demonstrating a similar
image quality, with only slight
differences between chest radiography (CR; a–c) and coronal
thick multiplanar reconstruction
(MPR; d–f). The trachea and
both main bronchi were better
visualized on thick MPR when
compared with CR (b and e,
enlarged images), whereas peripheral pulmonary vessels are
better depicted on CR (c and f,
enlarged images). Retrocardial
lung parenchyma, descending
aorta (arrow), and pedicles of
the thoracic spine (arrowhead)
are superiorly visualized on
thick MPR (d). The location of
the endotracheal tube can be
sufficiently assessed with both
modalities (b and e, enlarged
images)
Table 1 summarizes the results for visualization of critical
anatomical landmarks using CR and thick MPR. The
ratings did not differ significantly between the two readers
for all anatomical structures (p>0.05). The interobserver
agreement ranged from good to excellent. The rating
7
Table 2 Sensitivity, specificity, accuracy, positive predictive value (PPV), negative predictive value (NPV), and interobserver agreement of
pathologic findings and location of indwelling devices
Pathology/Indwelling devices
Equal accuracy
Pneumothorax
Perihilar congestion
Atelectasis
Consolidation/Contusion
Pleural effusion /hemothorax
Chest wall emphysema
Rib fracture
Shoulder girdle fracture
Thoracic spine fracture
Mediastinal widening
Pneumomediastinum
Pericardial effusion
Endotracheal tube
Central venous catheter
Pleural drainage tube
CR more accurate
Bronchial cuffing
N
Sensitivity
Specificity
Accuracy
PPV
CR
CR
CR
CR
MPR
MPR
MPR
NPV
MPR
CR
κ
MPR
CR
MPR
17
31
22
17
15
12
11
8
4
6
8
3
42
19
11
60
68
84
81
70
97
80
63
75
63
50
66
98
95
100
53
77
90
68
93
78
90
57
25
75
63
33
95
94
100
100
82
97
100
87
95
88
94
97
94
100
96
100
100
100
97
76
100
97
97
96
91
92
98
96
96
96
100
100
100
90
75
90
73
74
88
86
92
98
91
88
95
98
97
100
89
77
96
68
80
93
92
88
93
93
91
93
97
98
100
100
80
94
81
81
77
80
63
100
71
57
50
100
100
100
88
78
100
74
93
90
71
50
50
75
71
33
100
100
100
91
71
93
91
85
90
94
96
98
94
92
98
86
96
100
86
75
95
85
97
93
97
94
95
96
94
96
93
98
100
0.943
0.843
0.931
0.916
0.753
0.775
0.915
0.876
0.913
0.901
0.902
0.855
0.972
0.920
0.925
0.884
0.882
0.820
0.914
0.863
0.845
0.832
0.771
0.719
0.965
0.813
0.797
0.954
0.989
0.971
29
86
71
93
87
90
78
92
83
87
75
0.867
0.817
CR Chest radiograph, MPR multiplanar reconstruction, N number of pathologic findings, κ interobserver agreement
results were therefore grouped together, and the mean
scores±standard deviations of both readers were calculated.
Thick MPR performed better in visualizing both main
bronchi, diaphragm, retrocardial lung parenchyma, descending aorta, and thoracic spine pedicles. CR visualized
peripheral lung vessels in all four quadrants with a higher
quality than thick MPR. The visualization of perihilar
bronchi and central pulmonary arteries did not differ
between the two techniques. An example for the visualTable 3 Diagnostic certainty
for the assessment of pathology
and location of indwelling
devices for both readers
CR Chest radiograph, MPR
multiplanar reconstruction
ization of chest anatomy with CR and thick MPR is
demonstrated in Fig. 1.
Pathology
Table 2 lists the sensitivity, specificity, accuracy, PPV, and
NPV for the detection of pathological findings and for the
localization of indwelling devices with CR and thick MPR.
Pathologic findings/indwelling devices
Diagnostic certainty
CR
MPR better
Atelectasis
Pleural effusion/Hemothorax
Chest wall emphysema
Mediastinal widening
Pneumomediastinum
CR better
Pneumothorax
Bronchial cuffing
Consolidation/Contusion
Shoulder girdle fracture
Thoracic spine fracture
CR and MPR equal
Perihilar congestion
Rib fracture
Pericardial effusion
Endotracheal tube
Central venous catheter
Pleural drainage tube
MPR
p Value
1.56±0.68
1.60±0.66
1.49±0.47
1.49±0.51
1.63±0.67
1.22±0.59
1.25±0.39
1.19±0.39
1.30±0.42
1.43±0.60
0.022
0.026
0.038
0.049
0.042
1.54±0.34
1.29±0.49
1.43±0.56
1.20±0.59
1.29±0.32
2.13±0.45
1.94±0.83
2.08±0.71
1.79±0.55
1.63±0.53
0.012
0.008
0.008
0.009
0.024
1.71±0.52
1.31±0.27
1.32±0.49
1.24±0.38
1.29±0.25
1.31±0.27
1.54±0.34
1.45±0.73
1.23±0.42
1.40±0.38
1.39±0.30
1.38±0.44
0.052
0.081
0.171
0.259
0.452
0.231
8
Fig. 2 CR (a) and coronal thick MPR (b) images demonstrating
different traumatic chest injuries of a 20-year-old male patient after a
15-m downfall. Lung contusions (large arrows) and mediastinal
emphysema (small arrows) were similarly visualized on both CR and
thick MPR. The ventral pneumothorax on the left side (arrowheads)
is better depicted on thick MPR and was initially missed on CR by
both readers, possibly because of an increase in the time interval
between CR and computed tomography (CT). The pleural drainage
on the right and the endotracheal tube were introduced before CT
Statistically significant differences were only present for
the sensitivity for the detection of bronchial cuffing, with a
better performance of CR compared with MPR. The
interobserver agreement for the assessment of pathology
ranged from good to excellent for both imaging modalities
(see Table 2).
Table 3 lists the diagnostic certainty and corresponding
p values of both methods for all pathological findings and
for the localization of indwelling devices. The diagnostic
certainty was higher for thick MPR for the diagnosis of atelectasis, pleural effusion/hemothorax, chest wall emphysema, mediastinal widening, and mediastinal emphysema.
Diagnostic certainty was higher for CR for diagnosing pneumothorax, bronchial cuffing, consolidation/contusion, and
fractures of the shoulder girdle and spine. The readers were
equally certain with both methods in diagnosing perihilar
congestion, pericardial effusion, and rib fractures (see Table 3).
Figure 2 demonstrates the visualization of some pathological findings with CR and thick MPR.
should be limited to a single CT examination, thereby
saving important time [8–10]. This study demonstrates that
coronal thick MPR generated from MDCT data depicts
anatomy and pathology and documents the location of
indwelling devices with a similar quality when compared
with conventional bedside CR. Consequently, in trauma
patients in whom the initial CR is omitted and who directly
undergo MDCT, an adequate equivalent to the omitted CR
can be reconstructed for subsequent treatment monitoring
on the intensive care unit or ward.
CT is nowadays a widely used and versatile imaging
modality for emergency patient management [8–10, 13–
15]. Major recent developments have led to the introduction of MDCT in the emergency setting, with fast data
acquisition and tailored MDCT protocols, allowing even
severely injured and hemodynamically unstable patients to
have a CT [13, 14]. Besides the high sensitivity for the
detection of various traumatic chest injuries [16], MDCT
enables one to scan, with a sub-millimeter resolution, large
areas of the body during a single breath-hold. This high
z-plane resolution results in nearly isotropic voxels (i.e.,
voxels with almost the same dimensions in the x, y, and z
axes), which allow one to perform high-quality 2D and 3D
reconstructions in any plane and desired thickness.
In this study, the accuracy of coronal thick MPR in
identifying thoracic anatomy, pathology, and indwelling
devices was comparable with that of CR. However, some
differences in image quality and diagnostic accuracy
between modalities were noted, which may be explained
on methodological grounds.
Respiratory motion artifacts may decrease the image
quality of axial MDCT images and, consequently, also the
reconstructed thick MPR. They affect the image quality of
CR in a much lesser degree because of the shorter
acquisition time (10 ms for CR vs approximately 9 s for
MDCT). If possible, the patient is asked to hold his breath
Indwelling devices
Both CR and thick MPR similarly and correctly visualized
the position of the endotracheal tube, central venous
catheter, and pleural drainage tube. The accuracy and
diagnostic certainty did not differ, and the interobserver
agreement was excellent for both imaging modalities (see
Tables 2 and 3).
Discussion
The radiological workup of trauma patients has to be
accomplished timely, efficiently, and accurately. It has been
therefore proposed that imaging in the emergency setting
9
during the CT scan, or if mechanically ventilated, the
ventilation is stopped during the scan. However, in the
emergency setting, the patient’s compliance to hold breath
is often limited, and the marginal respiratory and cardiovascular function may not allow a pause in ventilation.
The currently used axial CT resolution (2 mm slice
width) leads to an impaired visualization of small pulmonary structures such as peripheral lung vessels and may
result in stair step artifacts of the central bronchi. Increasing
the axial resolution, however, is currently not feasible in the
emergency setting due to more time-consuming reconstructions and because of the longer image loading and
viewing times of larger CT datasets.
On the other hand, some anatomical structures were
better visualized on thick MPR than on CR. This may be
explained by the comparably higher radiation dose applied
during CT (approximately 3 mSv for MDCT vs 0.1 mSv
for CR), resulting in a better definition of structures that are
located behind each other, such as the retrocardial lung
parenchyma, both main bronchi, and the pedicles of the
thoracic spine.
In addition, geometric distortion in conventional imaging, which is due to an unequal magnification of different
structures, may also contribute to a reduction of image
quality in CR. This distortion is a well-known and accepted
limitation for accurate measurements of thoracic organs, a
phenomenon that does not occur in reconstructed images
based on volumetric CT datasets.
A more simple explanation can be given for the better
visualization of the descending aorta on thick MPR, which
is most likely due to the intravenous contrast material
administered for CT.
Finally, differences between both modalities concerning
pathological findings may also be the result of the time
delay between the two examinations. Although keeping
this delay as short as possible (less than 1 h), various drugs
supporting the patient’s cardiovascular function or indwelling devices, such as chest tubes, may interfere with the
imaging appearance on the corresponding modality. For
example, bronchial cuffing often caused by edema may
rapidly improve when the patient is stabilized before
undergoing CT. Similarly, fractures could be dislocated or
readjusted during the patient’s transport from the emergency table to the CT scanner, and also, a pneumothorax
may change during this time interval.
Concerning the relatively low accuracy of thick MPR
for the detection of certain pathological findings, it is
necessary to bear in mind that the most accurate diagnostic
information is obtained anyway from axial CT images.
Since the idea of this study is the replacement, in the
future, of the initial bedside CR by thick MPR, the most
important statistical comparison is that of the diagnostic
accuracy between modalities, which, in fact, showed a
high concordance.
This study does not advocate to omit CR in the emergency setting but to provide an alternative image resembling CR when performing chest CT anyway. Therefore,
the unquestionable higher radiation dose of CT when compared with CR is not an argument against the suggested
thick MPR method. In contrast, when replacing the initial
CR by thick MPR, the totally applied radiation dose to the
individual patient is even slightly reduced.
We have to acknowledge the following limitations.
Despite the relatively large patient number, the incidence of
certain pathological findings (such as fractures or mediastinal pathology) is relatively low. This may limit the
calculated descriptive statistics regarding sensitivity, specificity, PPV, NPV, and diagnostic accuracy. On the other
hand, this does not apply for the assessment of other
pathological findings and for anatomical structures that
were assessed for the entire patient population in a high
number. We did not assess the value of CT scanograms as
an alternative method for CR instead of thick MPR. The
scanograms obtained in this study were acquired in a single
frontal view and with a low radiation dose, and although
not performing a direct quantitative comparison, we
considered their image quality as being insufficient to be
a reliable alternative for CR.
In conclusion, coronal thick MPR generated from
MDCT data provide chest images with a quality and
diagnostic accuracy comparable with the bedside CR in
trauma patients. They may be used as an equivalent
baseline image for subsequent follow-up in trauma
patients, in whom the initial CR is omitted to accelerate
the diagnostic workup. Further prospective outcome
studies have to assess the practical usefulness of thick
MPR when replacing the initial CR in the clinical setting.
Acknowledgements This research has been supported by the
National Center of Competence and Research, Computer Aided, and
Image Guided Medical Interventions (NCCR CO-ME) of the Swiss
National Science Foundation.
References
1. Bell RM, Krantz BE, Weigelt JA (1999) ATLS: a foundation
for trauma training. Ann Emerg Med 34:233–237
2. Olson CJ, Arthur M, Mullins RJ, Rowland D, Hedges JR,
Mann NC (2001) Influence of trauma system implementation
on process of care delivered to seriously injured patients in rural
trauma centers. Surgery 130:273–279
3. American College of Surgeons Committee on Trauma (1997)
Student course manual. Advanced trauma life support for
doctors. American College of Surgeons, Chicago
4. Schweiberer L, Nast-Kolb D, Duswald KH, Waydhas C, Muller
K (1987) Polytrauma-treatment by the staged diagnostic and
therapeutic plan. Unfallchirurg 90:529–538
5. Massarutti D, Berlot G, Saltarini M, Trillo G, D’Orlando L,
Pessina F, Modesto A, Meduri S, Da Ronch T, Carchietti E
(2004) Abdominal ultrasonography and chest radiography are
of limited value in the emergency room diagnostic work-up of
severe trauma patients with hypotension on the scene of
accident. Radiol Med (Torino) 108:218–224
6. Collins JA, Samra GS (1998) Failure of chest X-rays to
diagnose pneumothoraces after blunt trauma. Anaesthesia
53:74–78
7. Neff MA, Monk JS Jr., Peters K, Nikhilesh A (2000) Detection
of occult pneumothoraces on abdominal computed tomographic
scans in trauma patients. J Trauma 49:281–285
8. Leidner B, Adiels M, Aspelin P, Gullstrand P, Wallen S (1998)
Standardized CT examination of the multitraumatized patient.
Eur Radiol 8:1630–1638
10
9. Low R, Duber C, Schweden F, Lehmann L, Blum J, Thelen M
(1997) Whole body spiral CT in primary diagnosis of patients
with multiple trauma in emergency situations. Rofo 166:382–
388
10. Boehm T, Alkadhi H, Schertler T, Baumert B, Roos J, Marincek
B, Wildermuth S (2004) Application of multislice spiral CT
(MSCT) in multiple injured patients and its effect on diagnostic
and therapeutic algorithms. Rofo 176:1734–1742
11. Diederich S, Lenzen H (2000) Radiation exposure associated
with imaging of the chest: comparison of different radiographic
and computed tomography techniques. Cancer 89:2457–2460
12. Landis JR, Koch GG (1977) The measurement of observer
agreement for categorical data. Biometrics 33:159–174
13. Novelline RA, Rhea JT, Rao PM, Stuk JL (1999) Helical CT in
emergency radiology. Radiology 213:321–339
14. Linsenmaier U, Krotz M, Hauser H, Rock C, Rieger J,
Bohndorf K, Pfeifer KJ, Reiser M (2002) Whole-body computed tomography in polytrauma: techniques and management.
Eur Radiol 12:1728–1740
15. Voggenreiter G, Aufmkolk M, Majetschak M, Assenmacher S,
Waydhas C, Obertacke U, Nast-Kolb D (2000) Efficiency of
chest computed tomography in critically ill patients with
multiple traumas. Crit Care Med 28:1033–1039
16. Alkadhi H, Wildermuth S, Desbiolles L, Schertler T, Crook D,
Marincek B, Boehm T (2004) Vascular emergencies of the
thorax after blunt and iatrogenic trauma: multi-detector row CT
and three-dimensional imaging. Radiographics 24:1239–1255