RESUMO
PURPOSE: The aim of this study was to assess the performances of unenhanced post-mortem computed tomography (CT) to detect thoracic injuries in violent death. MATERIALS AND METHODS: Retrospectively, we conducted a review of unenhanced CT scans of 67 victims of violent deaths with thoracic injuries and compared CT findings with the results of clinical autopsy. Our gold standard was a comparison of CT scans with autopsy discussed in a monthly forensic radiology multidisciplinary team meeting (MDTM). The data were collected by organ system: heart, pericardium, aorta, lungs, pleura, bone, and diaphragm and performance indices (sensitivity, specificity, accuracy) were calculated. RESULTS: Pleural (59/67) and bone (55/67) injuries detected on CT were also found at autopsy and confirmed by the MDTM (sensitivity and specificity 100%). Seventeen out of 67 diaphragmatic lesions were visible on CT. Eighteen out of 67 were confirmed during MDTM after autopsy, yielding overall sensitivity of 94% and specificity of 98%. Forty out of 67 lung contusions were found on CT with two false positives and one false negative yielding 95% sensitivity for CT with a specificity of 96%, and accuracy of 95%. Fourteen out of 67 aortic injuries were found on CT compared to 19 confirmed during MDTM (sensitivity 74%, specificity 85%, accuracy 82%). In terms of pericardial lesions, 19/67 were found on CT and 20 on autopsy and confirmed during MDTM (sensitivity 80%, specificity 94%, accuracy 85%). Ten out of 10/67 cardiac lesions were visible on CT imaging and 15 found on autopsy and confirmed during MDTM (sensitivity 57%, specificity 94%, accuracy 81%). CONCLUSION: Unenhanced post-mortem CT performs well to detect pleural, pulmonary, bone and diaphragmatic injuries but less well to identify cardiac and aortic injuries, for which the use of indirect signs is essential.