| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Musculoskeletal Imaging |
1 Institute of Diagnostic Radiology, University Hospital Zurich, Raemistrasse
100, Zurich 8091, Switzerland.
2 Division of Trauma Surgery, University Hospital Zurich, Zurich 8091,
Switzerland.
Received November 25, 2003;
accepted after revision April 28, 2004.
Address correspondence to J. E. Roos
(justus.roos{at}usz.ch).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Eighty-two patients with multiple injuries underwent MDCT for the chest and abdomen using a standardized 4-MDCT trauma protocol (collimation, 4 x 2.5 mm). Secondary reconstructions targeted to the spine were performed (slice width, 3 mm; reconstruction interval, 1.5 mm). All spinal fractures were additionally scanned using a collimation of 4 x 1 mm, and these images served as the standard of reference for fracture classification. An additional 50 patients with no spinal fracture served as the control group. A total of 65 major spinal fractures were present in 55 of the patients with multiple injuries. Two observers (observer 1 and observer 2) independently evaluated all CT data for spinal fractures using a 5-point confidence scale, classified the different fracture types, and rated the image quality of spinal structures on axial images and multiplanar reformations.
RESULTS. Image quality for axial images was excellent in 80% and in
68% using 4 x 1 mm and 4 x 2.5 mm collimation, respectively. Image
quality of the multiplanar reformations was excellent in 75% and good in 65%
using 4 x 1 mm and 4 x 2.5 mm collimation, respectively. Spinal
fractures were detected by observer 1 and observer 2 with a sensitivity and
specificity of 98% and 97% and of 97% and 97%, respectively. Interobserver
agreement regarding the confidence scale for fracture detection was
substantial (
= 0.80), and agreement between the different imaging
protocols for fracture classification was excellent for observer 1 ( =
0.95) and observer 2 ( = 0.97).
CONCLUSION. Accurate evaluation of the thoracolumbar spine is possible with targeted image reconstruction based on a standardized 4-MDCT trauma protocol of the chest and abdomen.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In most trauma centers, conventional radiographs are obtained as the initial imaging technique for the evaluation of thoracic and lumbar spine integrity in patients with multiple injuries [13]. If the findings from these radiographs are indeterminate or show a fracture that needs further evaluation or if the patient has symptoms of a fracture but the findings on conventional radiographs are negative, CT is usually performed. In the acute management of patients with multiple injuries, it would be desirable if CT for the secondary evaluation of spinal injuries could be combined with CT assessment of abdominal and thoracic organ injuries in a single data acquisition. The simultaneous assessment of organ and spinal injuries would result in a significant reduction in scanning time, which may be of paramount importance in maintaining the "golden hour" [14] in the management of patients with multiple injuries.
Several authors have reported their experience with a standardized trauma CT protocol for assessing organ and spinal injuries of the chest and abdomen using single-detector CT [3, 7, 15]. These protocols use a slice collimation of 5–10 mm. However, an accurate assessment of thoracic and lumbar spine injuries, including fracture classification and determination of fracture stability, often requires thinner slice collimation especially for high-quality multiplanar reformations. Sagittal multiplanar reformations are, for example, mandatory for a complete evaluation of possible fractures running in the plane of the X-ray beam because they are often missed on axial images [12, 16, 17]. Therefore, when single-detector CT is being used in clinical practice, additional targeted CT scans of the spine in case of questionable or definite vertebral fractures are obtained with thinner slice collimation. Obtaining these images may prolong the total scanning time, which is crucial under emergency conditions. Therefore, the introduction of MDCT, allowing rapid acquisition of a large scanning volume with a thin slice collimation, is of special interest in the initial CT assessment of the patient with multiple injuries [18–20].
In this study, we evaluated the accuracy and diagnostic performance of a standardized trauma protocol using a 4-MDCT scanner for the detection of thoracic and lumbar spine fractures in the initial assessment of patients with multiple injuries.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
We used the computer databases of the department of trauma surgery and the department of radiology to retrospectively review the clinical and radiologic charts of all patients with multiple injuries who were treated in our level I trauma center during the period mentioned earlier. For the purpose of this study, we included all those patients with multiple injuries who fulfilled all the following criteria: presence of a major spinal fracture of the thoracic or lumbar spine (or both) based on the radiologic workup or strong clinical suspicion for the presence of a spinal fracture, no visible fracture on initial radiologic workup, and referral for MDCT assessment of simultaneous organ or osseous injuries of the chest or abdomen.
Among the 285 patients with multiple injuries, 82 patients (14 women, 68 men; mean age, 44 years; age range, 16–85 years) fulfilled the inclusion criteria and were the focus of this study. The mechanisms of trauma in these 82 patients with multiple injuries were fall in 31 patients (accidental fall in 27 patients, fall as suicide attempt in four patients), motor vehicle crash in 26 patients, accident at work in 12 patients, and automobile–pedestrian collision in 13 patients.
All 82 patients underwent a standardized 4-MDCT trauma protocol of the thorax and abdomen that used a 4 x 2.5 mm collimation. The fractured spinal thoracic and lumbar levels were also scanned using a dedicated thin-section spinal MDCT protocol with a 4 x 1 mm collimation to obtain a standard of reference for fracture detection and classification. If no fracture was visible on conventional radiographs but there was a strong clinical suspicion for the presence of a vertebral fracture, suspected vertebral levels were scanned using the thin-section spinal CT protocol. Thus, in all 82 patients, a CT data set with a 4 x 2.5 mm collimation of the entire thoracic and lumbar spine and a CT data set with a 4 x 1 mm collimation of the fractured levels or the suspected fracture levels were available for analysis.
To evaluate the specificity of fracture detection, we selected a control group of 50 patients (14 females, 36 males; mean age, 51 years; age range, 15–86 years) with no clinical history of a spinal fracture who underwent MDCT of the chest and abdomen in a technically identical manner as the patients with multiple injuries.
Institutional review board approval was obtained for this study. Informed consent was not necessary according to our institutional review board because obtaining informed consent is a standard protocol for clinically indicated diagnostic procedures in the trauma setting. Informed consent was obtained from all patients in the control group.
Imaging Technique
All CT examinations were performed with a 4-MDCT scanner (Somatom
VolumeZoom, Siemens Medical Solutions) that is installed in a CT suite
adjacent to our emergency department. An IV contrast agent (Visipaque 320
[iodixanol], Nycomed Imaging) was administrated in a cubital vein at an
injection rate of 3 mL/sec using a power injector (Envision CT Injector,
Medrad). Further acquisition parameters were as follows: 180 mAs; table feed,
12.5 mm/rotation; pitch, 5; and gantry rotation time, 0.5 sec. The CT sections
were obtained from the lower cervical level to the level of the lesser
trochanter of the femur. In addition, a thin-section spinal MDCT protocol was
applied to all fractured levels or suspected spinal fractures during the same
examination. This protocol included a 4 x 1 mm collimation covering the
fractured spinal level and two vertebral levels cranially and caudally. If no
fracture was visible on the initial radiographs, a range defined by the trauma
surgeons was scanned using this thin-section spinal CT protocol. As a result
of tube heating, this range was limited but covered at least five spinal
levels. Because not the entire thoracolumbar spine could be assessed in this
thin-section spinal MDCT protocol, the clinical charts and all imaging
examinations that had been performed during the follow-up were reviewed for
probable spinal fractures that were not diagnosed using the standardized
4-MDCT trauma protocol at admission.
The 50 patients of the control group were examined on a technically identical MDCT unit (Somatom VolumeZoom) as the patients with multiple injuries were scanned. This control group was mainly recruited from patients who underwent CT angiography of the thoracoabdominal aorta with a biphasic contrast injection protocol that consisted of arterial phase (4 x 1 mm collimation) and venous phase (4 x 2.5 mm collimation) imaging.
Image Reconstruction
The entire data set from the standardized 4-MDCT trauma protocol was
reconstructed with a slice width of 3 mm and a reconstruction interval of 2 mm
using a medium smooth kernel (B30). On the basis of the same raw data set, an
additional axial reconstruction step was performed specifically targeted to
the spine with a field of view just covering the entire vertebral column.
Subsequent axial source images with a slice width of 3 mm and a reconstruction
interval of 1.5 mm were reconstructed with a bone edge-enhancement kernel
(B50). In five (3.8 %) of 132 patients with scoliosis or kyphosis of the
thoracic or lumbar spine (or both), the targeted axial image set did not cover
the entire vertebral column with the standard field of view used for
reconstruction. In these patients, a second reconstruction with a larger field
of view was necessary. For multiplanar reformations in the sagittal and
coronal planes, the basic data set (B30) acquired with the standardized 4-MDCT
trauma protocol was used and was secondarily magnified to the vertebral column
on the workstation's view screen (Advantage Windows 4.0, GE Healthcare). The
concept of this standardized 4-MDCT trauma protocol and its reconstruction
order is illustrated in Figure
1. The raw data of the spinal thin-section MDCT protocol acquired
with a collimation of 4 x 1 mm were reconstructed for axial images with
a slice width of 1.25 mm, a reconstruction interval of 0.8 mm, and a bone
edge-enhancement kernel (B50).
|
Image Analysis
To reduce bias during image analysis caused by the fact that patients with
more severe thoracic or abdominal injuries (or both) are more likely to
present with concomitant fracture of the spine, the study coordinator
segmented all CT data sets (Advantage Windows 4.0) before the review session
by manually cropping in a fashion that only the bony structures of the
thoracolumbar spine with the surrounding soft tissues were visible. The study
coordinator also reviewed all clinical charts. Image analysis was then divided
into three different image analysis sessions that were separated by a 3-week
interval between each session. The image analysis of all data was
retrospective.
During the first image analysis session, the standardized 4-MDCT trauma protocol data set (4 x 2.5 mm collimation) of the patients with multiple injuries randomly mixed with the data set of the control group patients was assessed independently by two radiologists (observer 1 and observer 2) experienced in emergency radiology (both observers, 5 years' experience) on an interactive workstation (Advantage Windows 4.0). The observers were blinded to all clinical data about the patients including age and sex. They were only aware of the fact that some patients underwent MDCT after trauma. The readout was based on targeted axial source images and multiplanar reformations. All images were viewed with bone windowing settings (window level, 450 H; window width, 1,500 H). The observers were asked to evaluate each vertebral level for the presence of any spinal fracture and to rate their confidence on a 5-point Likert scale according to the following scores: 1, fracture definitely absent; 2, fracture probably absent; 3, equivocal; 4, fracture probably present; and 5, fracture definitely present. The number, level, and fracture type were categorized and anatomically described by both observers. For categorization of spinal fractures, observers were asked to classify major spinal fractures using the classification system described by McAfee et al. [23] and Denis [24]. For our study, we used five distinctive fracture and injury types for classification of thoracolumbar spinal fractures: wedge compression fracture, burst fracture, Chance fracture (i.e., seatbelt fracture), flexion–distraction injuries, and translational injuries. Wedge compression fractures included only fractures of the anterior column. In addition, minor fractures such as fractures of the transverse or spinous process were noted if present. Fractures of the ribs were not analyzed.
During the second image analysis session, a consensus panel review session was organized by the study coordinator. This session consisted of a senior spinal trauma surgeon and the same two radiologists who participated in the first image interpretation session. The purpose of this consensus panel was to classify all spinal fractures to establish a standard of reference for spinal fracture classification. For this purpose, the imaging data of the standardized 4-MDCT trauma protocol and the data from the spinal thin-section MDCT protocol of all patients were available on the same interactive workstation. The consensus panel was asked to note the presence or absence of spinal fractures and to classify the major spinal fractures using a classification system similar to that used during the individual review sessions by observers 1 and 2. The final diagnosis was checked by surgical or autopsy findings if available.
During the third image analysis session, the same radiologists who participated in sessions 1 and 2 assessed in consensus the quality of the images obtained using standardized 4-MDCT trauma protocol and the spinal thin-section 4-MDCT protocol of patients with major spinal fractures as defined by the consensus review panel. The observers were blinded to the imaging parameters. Four anatomic structures were evaluated both on axial and on sagittal and coronal multiplanar reformations acquired with the spinal thin-section standardized 4-MDCT trauma protocol: inner and outer vertebral cortex, facet joints, and cancellous bone of two nonfractured vertebrae adjacent to the level of the fracture in the spine. The image quality for diagnostic purposes was classified in terms of visibility of these spinal anatomic details on a 4-point scale: score of 1, poor, nondiagnostic; 2, fair, but still diagnostic; 3, good; and 4, excellent. Excellent (score of 4) was assigned when the image was 100% sharp and virtually free from degradation. Fair (score of 2) represented blurring of the vertebral cortex and cancellous bone that, however, did not result in nondiagnostic images. Poor image quality (score of 1) did not allow an evaluation of the anatomic structures sufficient for diagnosis. Good (score of 3) represented a partially blurred vertebral cortex and cancellous bone and was assigned on the basis of the radiologists' subjective judgment between fair (score of 2) and excellent (score of 4) image quality. A final score of each assessed anatomic structure was summarized for the two evaluated nonfractured vertebrae and assigned to each patient.
Statistical Analysis
Descriptive statistics were calculated concerning image quality for all
evaluated anatomic spinal structures on axial source images and multiplanar
reformations. Differences regarding image quality between both methods were
analyzed using the paired sign test due to the nonparametric nature of these
ordinal data. Correlation between the patient's age and image quality of the
CT images based on the spinal thin-section and the standardized 4-MDCT trauma
protocol was assessed with the Spearman's rank correlation test (SPSS version
10, Statistical Package for the Social Sciences) for Windows (Microsoft). A
p value of less than 0.05 was considered to be statistically
significant.
Individual sensitivity and specificity values for each observer were determined for the diagnosis of major spinal fracture with a decision threshold of 3 or above (based on Likert scoring) as positive for spinal fracture. Neither positive nor negative predictive values were computed because of the arbitrary size of the control group. Likert scores of 1 and 2 were considered as absence of fracture.
Determination of the agreement between the imaging protocols (standardized 4-MDCT trauma protocol vs thin-section spinal 4-MDCT protocol) for fracture type detection and interobserver variability regarding the confidence rating was achieved by computation of kappa values (< 0.20, poor; 0.21–0.40, fair; 0.41–0.60, moderate; 0.61–0.80, substantial; 0.81–1.00, excellent) using SPSS version 10 (Statistical Package for the Social Sciences) for Windows (Microsoft) [25].
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
= 0.95) and for observer 2 ( = 0.97). The mean
confidence score (Likert scale) for the diagnosis of a major spinal fracture
was 4.5 and 4.6 for observer 1 and observer 2, respectively. In patients with
no fracture, the mean confidence score (Likert scale) was 1.2 and 1.2,
respectively. Interobserver agreement between observers 1 and 2 regarding the
confidence scale was substantial for all evaluated data ( = 0.80).
|
|
|
|
|
|
|
|
|
|
|
|
|
All 55 minor fractures were detected by both observers using the standardized 4-MDCT trauma protocol.
The image quality concerning spinal anatomic details for diagnostic purposes is summarized in Table 2. The image quality for axial source images was rated in 55 evaluated patients as excellent (score of 4) in 80% (44 patients) with the spinal thin-section MDCT protocol and in 68% with the standardized 4-MDCT trauma protocol. Multiplanar reformation images were classified as excellent (score of 4) in 74% with 4 x 1 mm collimation and as good (score of 3) in 65% (36 patients) with 4 x 2.5 mm collimation (Fig. 6). The mean values of all four summarized anatomic details regarding image quality showed no statistically significant difference between the spinal thin-section 4-MDCT protocol (median, 3.75) and the standardized 4-MDCT trauma protocol (median, 3.75) for axial images (p > 0.061). In contrast, the lower mean values for image quality on multiplanar reformation images resulted from statistically increased image degradation with the standardized 4-MDCT trauma protocol (median, 2.75) compared with the thin-section 4-MDCT data set (median, 3.75) (p < 0.001) (Fig. 6). Spearman's rank correlation test revealed a statistically significant correlation between decreasing image quality and increasing age of the patients in the group of multiplanar reformation images acquired with the standardized 4-MDCT trauma protocol in three of four anatomic details (outer vertebral cortex, p = 0.037; inner vertebral cortex, p = 0.042; cancellous bone, p = 0.001). No statistically significant correlation was found between image quality of all four anatomic details and the age of patients for the group of multiplanar reformation images based on the spinal thin-section 4-MDCT protocol and for both groups of axial images based either on the spinal thin-section or on the standardized 4-MDCT trauma protocol.
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The ability for postprocessing of the acquired data set is a major advantage of MDCT scanners. The versatility of the data set for secondary reconstruction of images using different slice thicknesses and field of views during image reconstruction is especially helpful in the evaluation of the thoracolumbar spine. With the standardized 4-MDCT trauma protocol as proposed in this study, it is possible to evaluate thoracic, abdominal, and pelvic organs as well as soft-tissue and bone structures based on one CT data set. As shown in this study, the entire thoracic and lumbar spine can be reconstructed in a smaller field of view targeted to the spine based on a raw data set with a collimation of 4 x 2.5 mm. For that reason, we chose an algorithm that provided the optimal combination of good image quality and short reconstruction time with one additional reconstruction step (targeted axial reconstruction of the vertebral column) after acquiring the standardized 4-MDCT data set for trauma patients. The image quality of the secondarily reconstructed data set using the proposed algorithm for discerning spinal details was rated excellent in 68% of the axial images and good in 65% of the multiplanar reformations. However, we acknowledge that when using the standardized 4-MDCT trauma protocol, the overall image quality may be adversely affected in elderly patients with more osteopenic bone structures—in particular, when multiplanar reformations are performed. This fact is underlined by a significant negative correlation between patient age and image quality in the group of multiplanar reformations acquired with the standardized 4-MDCT trauma protocol. However, this correlation did not result in nondiagnostic images in any of the patients evaluated for this study.
Detection of spinal injuries is of paramount clinical importance in the management of patients with multiple injuries. In addition, clinical evaluation of patients with multiple injuries is frequently difficult because these patients are often unconscious or intubated. Early knowledge of a significant spinal injury may be helpful even when standard precautions are taken. The results of our study show that, on the basis of the standardized 4-MDCT trauma protocol, both observers detected all major and minor spinal fractures with the exception of a single anterior wedge fracture that was not detected by either observer. The results of our study are in accordance with the recently published study of Wintermark et al. [29]. Those researchers found that MDCT can replace conventional radiography for the detection of thoracolumbar fractures in patients who have sustained severe trauma. In addition to the study of Wintermark et al., our study indicates that spinal fractures may be detected with a high sensitivity and specificity using a standardized thoracoabdominal 4-MDCT trauma protocol and that the fractures can be classified with a high interobserver agreement as well as a high agreement between the imaging protocols when compared with a spinal thin-section MDCT protocol.
Classification of spinal injuries is considered important with regard to surgical management, although different trauma centers use different classification systems for radiologic classification of thoracolumbar spinal fractures. For study purposes, we followed the fracture classification system of McAfee et al. [23] and Denis [24], which describes thoracolumbar fractures using CT. While using this classification concept, both observers were able to correctly determine the fracture type, assess vertebral stability, and provide relevant information for further patient management after interpretation of the standardized 4-MDCT trauma protocol (Table 1). Of 65 fractures, one subtle anterior wedge compression fracture was misdiagnosed as no fracture and one vertebra showing no fracture was incorrectly classified as an anterior wedge fracture in three patients. No vertebral fracture interpreted as stable using the standardized 4-MDCT trauma data set was subsequently classified as unstable by the following reference panel. On basis of the data from this study, it may be concluded that a standardized 4-MDCT trauma protocol with a collimation of 4 x 2.5 mm is sufficient for the detection and classification of thoracic and lumbar fractures in the trauma setting. To what extent thinner collimations of 1 mm or even submillimeter slice width, the latter applicable with the most recent generation of 16-MDCT scanners, will further strengthen the accuracy in evaluating patients with multiple injuries with regard to injuries of the thoracic and lumbar spine will be the focus of future studies.
This study had directly influenced our CT protocol in the emergency department. Since we finished this study, we routinely perform the described standardized thoracoabdominal trauma 4-MDCT trauma protocol with secondary reconstruction of the spine when CT is required in patients with multiple injuries. Additional thin-section spinal MDCT is not performed routinely anymore in the presence of fractures of the thoracolumbar spine.
This study has several limitations. Although the readout with regard to image quality has been performed blinded to acquisition parameters and patient sex and age, the visual difference between the multiplanar reformation images based on the standardized and thin-section spinal 4-MDCT data set was obvious. This may have resulted in a bias. The fact that several patients had more than one fractured vertebral level might have implied that the observers screened the other vertebral bodies more carefully because fractures occur simultaneously in adjacent levels with a relatively high frequency. Although all minor fractures revealed by the consensus panel were detected by both observers analyzing the standardized 4-MDCT trauma protocol, it remains unclear if the standardized 4-MDCT trauma protocol is sufficient for detection of minor vertebral fractures. Because of the chosen study design, in which the control group consisted of patients not being at risk for spinal injury, a probable overestimation of the diagnostic accuracy may have occurred [30]. However, for the purposes of the study, we preferred to have comparable CT data sets for which both collimations (4 x 2.5 mm and 4 x 1 mm) were available. Finally, another limitation is the fact that the thin-section imaging protocol was not available for all spinal levels in patients with multiple injuries. However, because we reviewed all patients' charts and all imaging that was performed during the hospitalization, this bias should have been minimized.
In conclusion, we have shown that MDCT allows accurate evaluation of the thoracolumbar spine with regard to fracture detection and classification based on a standardized 4-MDCT trauma protocol with a collimation of 4 x 2.5 mm.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  K. P. Daly, C. P. Ho, D. L. Persson, and S. B. Gay Traumatic Retroperitoneal Injuries: Review of Multidetector CT Findings RadioGraphics, October 1, 2008; 28(6): 1571 - 1590. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
