|Home | About | Journals | Submit | Contact Us | Français|
The aim of this study was to compare the efficiency of multiplanar reformatted images and three‐dimensional images created after multidetector computed tomography examination in detecting acute post‐traumatic osseous pathology of the skeletal system.
Between October 2006 and December 2008, 105 patients with a history of acute trauma were referred to our service. Patients were evaluated with multidetector computed tomography using multiplanary reconstructed images initially (R‐I), and six months after this initial evaluation, three‐dimensional images were assessed of each patient (R‐II). Axial images were used for guiding as a reference Data obtained was recorded and graded according to importance levels of the pathologies.
The R‐II score was higher in the non‐articular and highest in periartricular fractures of the extremities, and thoracic and pelvic cage injuries. For the spinal column, while R‐I data was more significant In patients referred with polytrauma, R‐II data, was more statistically significant, for short processing and adaptation time to acquiring immediate critical information. For all cases it was seen that three dimensional scans were more efficient in providing the orientation, within a short time.
By dual source multidedector tomography systems trauma patients may be evaluated by multiplanary and three dimensionally reconstructed images. When used correctly, three dimensional imaging is advantageous and can help determine the exact nature and extension and also importance of osseous injuries.
With recent developments in computerized tomography technology, multidetector computed tomography (MDCT) enabled better volume imaging and high‐speed data acquisition,1 allowing both increased coverage and improved resolution with less exposure to radiation, particularly in trauma patients.
Parallel advances in image processing software and hardware warranted designing better examination protocols that optimize image quality for easy and quick interpretation and minimize radiation dose.2
The aim of this study was to compare the efficiency of multiplanar reformatted (MPR) images and three‐dimensional (3D) images created after multidetector computed tomography (MDCT) examination in detecting acute post‐traumatic osseous pathology of the skeletal system. We also set out to form a short guide to the use of 3D MDCT on trauma patients that would save this method from arbitrary application.
Between October 2006 and December 2008, 105 consecutive patients (average age: 37, F = 30, M = 75) referred to our clinic with a history of acute trauma were retrospectively evaluated. MDCT was performed on a 16‐MDCT scanner (Somatom Definition, Siemens Medical Solutions, Germany). with tube voltage, 120‐140 kV; effective tube current, 240‐280 mAs; section thickness 2 mm; reconstruction interval: 0.75 mm; and collimation 0.6 mm. During the scan, the automatic dose control system was activated to prevent excessive radiation exposure.
Average scan time was between 11 and 23 seconds according to the different region of interest (ROI) settings. In 3 cases, 1 with a suspicion of pathological fracture after the first examination, and 2 with identified vascular injury, an intravenous non‐ionic contrast agent (iohexol 1‐1.2 ml/kg) was used. All the other examinations were completed without administration of contrast. Scans with contrast were applied in emergency settings, no bolus tracking method was applied, and scans were obtained in the arterial phase with 25‐30 sec as delay time.
After axial scanning, thin section reconstructions were also generated by reconstruction in both soft tissue (B10‐soft tissue) and bony kernels (B 70‐sharp osteo). In both convolution kernels, single data with a slice thickness of 2 mm (1.5 mm reconstruction increment) were used to form the reconstructed images. These reconstructed thin section images were transferred to a special workstation (Leonardo Running Inpace, Siemens Medical Solutions) and then multiplanar reformat (MPR), maximum intensity projection (MIP) and 3D images were generated.
Assessments were performed by a radiologist on sophisticated multidisplay screens. Axial and MPR images were evaluated immediately after each shot, and data were recorded as record I (R‐I).Six months after the initial assessment of each patient, only the axial and 3D images were regenerated and evaluated again by the same radiologist. The data obtained were recorded as record II (R‐II).
During R‐I evaluation, MPR images were examined in coronal, sagittal and if needed in curved or oblique planes. In R‐II evaluation, 3D images were examined by rotating the image fully around the coronal and horizontal axis (if needed in oblique positions) at least once. 3D images were first produced from the skin surface to the soft tissues, and then to the bony structure surface by changing the window settings. Both the volume rendered (VR) and surface shaded display (SSD) methods were used for 3D imaging. If needed, virtual extraction of foreign materials superposed to the region of interest was used. It was also possible to evaluate joint surface congruity with virtual exarticulation to remove superposed structures during some R‐II evaluation.
All data obtained from both evaluations were summarized in a detailed table (Table I). Osseous pathologies determined by the R‐I and R‐II were scored by means of a special grading system. In this system, coordination was set up with the clinician who referred the cases to our service and the data collected were categorized as important or unimportant. The data classified as important were essential for diagnosis and might potentially change the approach to the treatment. Minor findings that did not affect the treatment algorithm were classified as unimportant category (Table 2). Data obtained in both evaluations were scored according to their importance. Under this scheme, when no major pathology could be determined (between R‐I and R‐II) a value of “0” was assigned; when an abnormality was both detected and described, the value assigned was “1”; when unimportant data were determined the other mode; the value was “2”; and if a different finding in the important category was reported, the value was “3”.
The average score (OIS: Osseous Injury Score) was calculated and the statistical difference between the points of the two methods was measured with the Chi‐square test.
The patients were informed before the examinations, and consent was obtained. Neural parenchymal injuries, mediastinal or pulmonary parenchymal injuries and thoracoabdominal solid/luminal organ injuries, effusions or collections or other soft tissue injuries were reported with both evaluations (R‐I and R‐II) but in accordance with the aim of the study, only osseous findings were reflected to the statistical analyses.
Most of the cases had suffered traffic accidents (out‐of‐vehicle traffic accident=OVTA, n=34; in‐vehicle traffic accident=IVTA, n=29), The remaining cases were gunshot wounds (GSW, n=3), falling accidents at work (Fall, n=30), sports‐related injuries (Sport trauma, n=2), and assault (Beating, n=7).
Extremity+joint injuries (n=42) and maxillofacial injuries (n=38) were mostly seen, the osseous spinal axis was scanned in 10 subjects, the thoracic cage in 3 subjects, the pelvic cage in 5 subjects, and the whole body in 7 polytraumatic subjects.
The total time needed for the evaluation of the osseous pathologies of the 105 cases was calculated as approximately 1600 minutes for R‐I, and 1200 minutes for R‐II.
In agreement with the principles stated in Table 1; in 8 cases unimportant findings, and in 6 cases important findings, were noted by R‐I as distinct from the other method. On the other hand, R‐II examination of the 3D scans recorded important findings in 23 patients and unimportant findings in 52.
Data recorded by using two methods in the other 11 cases were scored at the same level.
In 7 patients (one case with non‐displaced femoral neck fracture, one case with a minimal impacted femoral neck fracture, 4 cases with orbital blow‐out and/or rim fractures and a case with a foreign body in the spinal canal) were recorded as “2 or 3 points” in R‐I, while R‐II scores were “0 or 1 points” (Figure 1).
Otherwise, the differences recorded as important in R‐II were related to 8 cases with upper extremity injuries, 9 cases with lower extremity injuries, and 5 cases with thoracic cage/pelvic cage/spinal injuries (Figure 2).
The R‐II score was higher in the non‐articular fractures of the extremities, with an average score of 2 points. On the other hand, in the articular‐periarticular osseous injuries, R‐II were assigned 3 points on average by more oriented images (Figure 3).
While the R‐II data were determined to be in the more important category for the thoracic cage (at the 3 points level), more data were provided compared to R‐I (unimportant category‐weighted) with a 2 point average for the pelvic cage (Figure 4).
For the spinal column, R‐I data were more significant, with 2 points in the definition of the anterior compartment configuration (listhesia, compression fracture detection and orientation to fracture configuration). Also, for intracanalicular pathologies (posterior elements or fragmentous displacements to the canalicular surface, foreign bodies and ‐ despite not being included in the statistical evaluations ‐ epidural soft tissue components) R‐I was scored with 3 points compared the “1 or 2” points scores of R‐II. However, R‐II data were again more significant in the injury or extracanalicular displacement of the posterior elements of the osseous spinal column (Figure 5, Figure 6).
Also, in another two patients referred with polytrauma, R‐II data, which were found to be more statistically significant, were scored with 3 points higher on average than R‐I in the aspects of orientation, short processing and adaptation time (5 mins versus 11 mins) for acquiring critical information (Figure 6, Figure 7).
The measured average score for R‐I for all of the subjects was 1.25; while it was recorded as 1.99 for R‐II. Thus, a statistically significant difference (p=0.000005) was shown in the Chi‐square test, which proved that findings described by the R‐II method defined posttraumatic pathologies more efficiently. In the all evaluated data, it was seen that 3D scans were fast and more efficient in providing good orientation for clinicians.
To our knowledge, this is the first article about the standardization of postprocessing operations in the literature in English. With this study, our aim was to display the usefulness and differences of both (MPR versus 3‐D) modalities, with their advantages and disadvantages. In this way, we intended to make a first step towards the standardization of the postprocessing algorithm, at least for osseous trauma patients.
Traumatized patients usually have injuries in several anatomical regions or organs, and CT assessment of such patients should be systematic, complete, and accurate. Another key factor is the time required for radiological examination, which, because the chance of survival increases the sooner trauma care is initiated, must be as short as possible.3,5
The reliability and workflow of MDCT for emergency purposes have been supported by the results of several studies.5-9 MDCT scanners are widely used because they rapidly produce high‐resolution images of large areas, offering short examination times for multiple body regions under emergency conditions.6-11 With new software programs and systems which have 64 or more detectors, 3D scanning and also processing time has been reduced significantly.4 Additionally, reduced radiation exposure rates have made the MDCT as preferred method in the trauma setting especially in unstable and uncooperative cases. Sometimes, it is possible to examine the affected area can be completed in subsecond time with new generation machines. So, it can exclude any requirement for sedation which may be used as a first line examination method especially in pediatric trauma cases.5,12 Seven pediatric cases in this study had been diagnosed directly with MDCT examination but without need for MRI or direct radiography correlation.
Orientation can be facilitated with MPR and 3D reconstruction in addition to the conventional axial images without changing routine MDCT examination protocols for trauma. Although there are many studies about the CT or MDCT protocols in trauma cases; there is no other record which emphasizes the standardization of the postprocessing algorithm or efficiency of 3‐D images objectively.3,9,11 In this study, while high resolution axial plus MPR images may be enough, we answer the question as to whether or not the formation and evaluation of the 3D images is needed.
It is also a known fact that besides the difficulty of quick evaluation of a high number of axial images consecutively, there are also difficulties regarding time and cost in the storage and retrieval of this information when needed.13 Forming and storing 3D images may partially alleviate this problem. Likewise, another important but overlooked point is the opportunity for processing the same data to produce 3D images from remote computers. This makes virtual inspection by teleradiologists possible. Although transferring high resolution and numerous images consumes time, a small number of focused 3D images can be transferred in a short time.14,15
It must be remembered that images reconstructed in the bone window give higher contour resolution in the sectional images, whereas thin axial images reconstructed in soft tissue kernels should be used for 3D evaluation16 (Figure 8).
The average time needed for each case for R‐I and R‐II protocols was 15 minutes, and 11,5 minutes respectively. At emergency setting, the time spared with R‐II protocol is critical for each patient. One of the facilities in the evaluation of 3D images is the ability to orient the reader's attention directly to the injury site by turning around the skeleton to reveal any defect 16,23 (Figure 14). In this way, it is possible to detect abnormalities easily in a short time.24,25 It is also is possible to reduce the artifacts or and to remove superposition (missed foreign bodies, pathological fractures, superposed implants or catheters) by this method.
With R‐I protocol eight extra unimportant findings were identified where as in R‐II protocol 52 additional unimportant findings was found. In terms of important findings, R‐I noted 6 and R‐II recorded 23 important findings. In only 11 cases, both methods scored the same important findings.
R‐I protocol was thought to be the appropriate protocol for the evalaution of basicranium, orbital rim (blow‐out), temporal bone, spinal canal, vertebral body internal contour injuries and intra‐articular fragments due to their complex anatomical structures and superpozitions. R‐II protocol was used to evaluate facial bones, calvarium, ribs, the posterior elements of the spinal column, long bones and articular surface integrity because it provided quick and cumulative information. It is obvious that usage of improper modality in inappropriate conditions leads time consumption
In some conditions, one of these techniques may be mandatory due to different characteristics of patients. For example, non‐displaced but minimally impacted porotic femoral neck fractures can only be identified by the evaluation of axial+3D images. In porotic patients, 3D images, even with suitable reconstruction, were not able to show the pathologies like fissures or nondisplaced fracture lines.
In this study, 3D images were examined by turning image fully around the vertical and horizontal axis (if needed in oblique positions) at least once. 3D image windows were produced from skin surface to the soft tissues at first, and then to the bony structure examined in more suitable colors. Because of the scope of the study, we emphasized the efficiency of the 3D images in the acute post‐traumatic osseous changes. However, superficial anatomy, tendons, periarticular ligaments, muscles and meniscal structures can be seen clearly with the colored 3D‐VR images without administering any contrast agent.17,18 Thus, this method will be crucial in cases where MRI examination is contradicted10 (Figure 9).
Patients with huge defects in their skin or other soft tissues but without any osseous injury being reported as normal with conventional axial or MPR images because of obscurence in sectional slices. But this may lower the confidence of readers. However, those injuries can be determined easily by 3D application examination. With the pointing out of the primary injuried sites, 3D application is an interesting tool which can be equal to diffusion weighted imaging considering that it has an effect that accelerates diagnosis.
In our cases, MPR images presented more useful data especially in spinal column injuries, and in intracanalicular pathologies (posterior vertebral contour instability, epidural lesions, disc pathologies, non‐osseous pathologies, and foreign bodies). This is probably related to the curved spinal column anatomy which prevents optimal virtual 3D evaluation through the spinal canal. But in some circumstances (such as with foreign bodies) the evaluation of intracanaliculary structure with 3D images may provide clearer and sufficient information in the detection of pathologies, and 3D method may yield good anatomical orientation to clinicians (19‐22) (Figure 7).
Articular and periarticular traumas can be evaluated confidently by means of different view angles in 3D images. The affected joint surfaces can also be evaluated by virtual‐3D exarticulations from different angles to view the articular surface integrity (Figure 10). Foreign bodies, fixation devices, plaster casts and splints which may hamper the image quality also can be virtually removed.22 In pediatric patients, evaluation of the epiphyseal lines may be impossible with axial images due to misleading variations. Also, it may be difficult to reveal the fracture orientation only with MPR sections. All the epiphyseal lines can be easily evaluated by the 3D images, so that injury characterization (according to Salter‐Harris categorization) can also be made confidently21 (Figure 11).
Due to the high resolution, it is possible to differentiate avulsion nonavulsion, separated fragments and ossified components. Osseous contours must be examined clearly in the 3D images especially confirming the conventional axial slices. In all cases, especially in maxillofacial traumas, whole images formed from the skin to the bone surface should be examined step by step to to assess the primary injury site (Figure 12, Figure 13).
Considering all the points in Table I, the measured average score for R‐I was 1.25; while it was recorded as 1.99 for R‐II. And also to the all evaluated data, it was seen that 3D scans were fast and more efficient in providing good orientation for clinicians.
In summary, 3D images must be adjusted from the skin to the bony surfaces by changing the window settings. The whole image should be examined by being turned fully at least once around the axial and coronal planes. If necessary, artifact‐producing implants, splints or fixators can be extracted in the virtual settings. Porotic bones, deep structures (temporal, facial, basicranial bony structure), or the spinal column should not be evaluated by 3D images only. These areas must be examined mainly with MPR images because of their superiority to 3D imaging in these regions. Injured articular surface integrity can be examined with virtually exarticulated images. It can also be said that examining the ligamentous or meniscal anatomy with coloured VR‐MPR images may provide more detailed information (especially in patients whom MRI contrendicated for) than 3D imaging.
In conclusion, imaging the injured skeleton in a trauma patient can be evaluated in twelve minutes with dual source MDCT sysem. It must be rotated by it self in whole axes at least once. Virtual inspection, virtual exarticulation is possible with 3D images. 3D‐CT imaging is a perfect method for evaluating unstable or MRI‐incompatible cases, and does not need direct graphy correlation. Because of these numerous utilities, 3D images must be created and evaluated for all trauma cases.