Automatic rib fracture detection on postmortem CT data using deep learning.

Rib fractures: elusive, but important

Rib fractures at postmortem computed tomography (PMCT) validated against the autopsy

Few studies have compared the sensitivities of autopsy and post mortem computed tomography (PMCT) in detecting rib fractures caused by cardiopulmonary resuscitation (CPR). We aimed to compare the characteristics between both modalities for accurately detecting CPR-related rib fractures. This single-centre observational study included adult patients with autopsy records and PMCT scans at our institution from January 2013 to March 2019. CPR-related rib fractures were evaluated using autopsy and PMCT findings. In 62 patients enrolled, 339 rib fractures were detected on autopsy and/or PMCT (222 fractures on both PMCT and autopsy, 69 on PMCT alone, and 50 on autopsy alone). The agreement of detection for both modalities was substantial (kappa coefficient, 0.78). In the logistic regression model, incomplete fractures detected by PMCT and age <75 years were significantly associated with findings that were negative on autopsy but positive on PMCT, while rib number (ribs 1–3 and 7–12) and fracture location (posterolateral and paravertebral) were significantly associated with negative PMCT findings but positive autopsy findings. Autopsy and PMCT showed complementary roles, and are thus necessary in accurately detecting CPR-related rib fractures. Combining both modalities may contribute to improved CPR quality and better understanding of discrepancy in characteristics between the two modalities.

“Buckle” rib fracture: An artifact following cardio-pulmonary resuscitation detected on postmortem CT

Forensic pathologists often encounter cases of acute subdural hematoma (SDH) due to trauma, whereas those attributable to endogenous causes are rare. Here, we report a case of the latter type in a 42-year-old man who was found dead at home after several months of fever and malaise. Postmortem computed tomography (PMCT) and autopsy were undertaken to clarify the cause of death. PMCT images revealed a fatal SDH and a localized hyper-density area in the right parietal lobe; macroscopic and microscopic examinations revealed SDH due to rupture of a mycotic aneurysm (MA) associated with meningitis. The PMCT images also indicated thickening and calcification of the mitral valve, while autopsy demonstrated infective endocarditis (IE). In addition, PMCT demonstrated a low-density area in the spleen, which was shown to be a splenic abscess at autopsy. PMCT also demonstrated tooth cavities. Based on the findings of autopsy, the cause of death was considered to be SDH due to rupture of the MA resulting from meningitis with IE and splenic abscess. Although PMCT was unable to clarify the significance of any individual feature, a retrospective review of the PMCT images might have suggested IE, bacteremia, or ruptured MA leading to SDH. This case suggests that, instead of interpreting individual features demonstrated on PMCT images, integrated interpretation of overall PMCT findings might provide clues for identifying causes of death, despite the fact that PMCT lacks diagnostic accuracy for infectious diseases such as IE and meningitis.

Purpose: The aim of the present study was to assess time-dependent changes in postmortem images of mouse skeletal muscle that were obtained with a small animal CT scanner. Materials and methods: Mice were euthanized and time-dependent changes in postmortem computed tomography (PMCT) images of skeletal muscle were recorded for 11 elapsed intervals: at the confirmed time of cardiac arrest, and at 0.5, 1, 2, 3, 6, 12, 24, 36, 48, and 72 hours after death. After acquiring PMCT images, skeletal muscle was removed and prepared for histological evaluation. Histological images and PMCT images of the skeletal muscle were then compared. Hounsfield unit (HU) measurements were obtained for the skeletal muscle PMCT images at each interval and time-dependent changes were examined. Results: Time-dependent changes were observed in histological images of skeletal muscle, but not in PMCT images. Time-dependent changes were detectable by HU values of skeletal muscle. Conclusion: Postmortem changes in skeletal muscle tissue appear as decreases in HU values on PMCT.

To determine the frequency of cardiopulmonary resuscitation (CPR)-related injuries and factors involved in their occurrence, data based on forensic autopsy and postmortem computed tomography (PMCT) during implementation of the 2010 American Heart Association Guidelines for CPR were studied. We retrospectively evaluated data on adult patients with non-traumatic deaths who had undergone manual CPR and autopsy from January 2012 to December 2014. CPR-related injuries were analyzed on autopsy records and PMCT images and compared with results of previous studies. In total, 180 consecutive cases were analyzed. Rib fractures and sternal fractures were most frequent (overall frequency, 66.1 and 52.8%, respectively), followed by heart injuries (12.8%) and abdominal visceral injuries (2.2%). Urgently life-threatening injuries were rare (2.8%). Older age was an independent risk factor for rib fracture [adjusted odds ratio (AOR), 1.06; 95% confidence interval (CI), 1.04-1.08; p<0.001], ≥3 rib fractures (AOR, 1.06; 95% CI, 1.02-1.09; p=0.002), and sternal fracture (AOR, 1.03; 95% CI, 1.01-1.05; p<0.001). Female sex was significantly associated with sternal fracture (AOR, 2.08; 95% CI, 1.02-4.25; p=0.04). Chest compression only by laypersons was inversely associated with rib and sternal fractures. Body mass index and in-hospital cardiac arrest were not significantly associated with any complications. The frequency of thoracic skeletal injuries was similar to that in recent autopsy-based studies. Implementation of the 2010 Guidelines had little impact on the frequency of CPR-related thoracic skeletal injuries or urgently life-threatening complications. Older age was the only independent factor related to thoracic skeletal injuries.

To review studies on deep learning (DL) models for classification, detection, and segmentation of rib fractures in CT data, to determine their risk of bias (ROB), and to analyse the performance of acute rib fracture detection models. Research articles written in English were retrieved from PubMed, Embase, and Web of Science in April 2023. A study was only included if a DL model was used to classify, detect, or segment rib fractures, and only if the model was trained with CT data from humans. For the ROB assessment, the Quality Assessment of Diagnostic Accuracy Studies tool was used. The performance of acute rib fracture detection models was meta-analysed with forest plots. A total of 27 studies were selected. About 75% of the studies have ROB by not reporting the patient selection criteria, including control patients or using 5-mm slice thickness CT scans. The sensitivity, precision, and F1-score of the subgroup of low ROB studies were 89.60% (95%CI, 86.31%-92.90%), 84.89% (95%CI, 81.59%-88.18%), and 86.66% (95%CI, 84.62%-88.71%), respectively. The ROB subgroup differences test for the F1-score led to a p-value below 0.1. ROB in studies mostly stems from an inappropriate patient and data selection. The studies with low ROB have better F1-score in acute rib fracture detection using DL models. This systematic review will be a reference to the taxonomy of the current status of rib fracture detection with DL models, and upcoming studies will benefit from our data extraction, our ROB assessment, and our meta-analysis.

Three-dimensional (3D) documentation is increasingly being utilized in forensic investigations to record injuries and reconstruct crime scenes accurately. Although photogrammetry offers a low-cost and accessible method for capturing surface details, its integration with postmortem CT (PMCT) data and virtual reality (VR) can further enhance spatial understanding. We report a fatal case of a stair-related fall of a man in his 70s, in which we visualized the injuries by combining PMCT data with surface models of the body and the staircase. A 3D model of the bones was created from the PMCT data, whereas surface models of the body and staircase were generated using photogrammetry. The reconstructed scene was visualized in VR using a MetaQuest 3 headset. The reconstructed scene clearly demonstrated the spatial relationship between the stair edges and injury sites, such as the occipital region, midthoracic spine, and sacrum. The vertical distances between the injuries closely matched the staircase step depth, supporting the interpretation of stair-related falls. This method provides an intuitive and immersive understanding of injury mechanisms. Our approach demonstrates the feasibility and utility of integrating CT, photogrammetry, and VR in forensic death investigations, offering enhanced documentation and visualization that can benefit not only forensic experts but also legal professionals and juries.

This study aimed at comparing two main existing diagnostic methods for the detection of lung infection in post-mortem cases of sudden adult death. Post-mortem computed tomography (PMCT) imaging of lungs and histopathology examination (HPE) of lung tissue were selected to compare their sensitivity and specificity. This retrospective case control study was conducted in a tertiary referral centre on 220 deceased individuals who underwent PMCT imaging prior to autopsy and had histology sampling during autopsy in the years 2016 - 2019. The bodies were examined with PMCT prior to conventional autopsy. Histology sampling were taken in those selected cases as part of medicolegal investigations. The reports and images of PMCT, and HPE reports with slides were retrieved and re-evaluated. Findings of PMCT and HPE were divided into pneumonia and non-pneumonia groups. Only PMCT images of chest and histology slides of lungs were accessed and evaluated. The result showed that the sensitivity and specificity of PMCT in the diagnosis of lung infection was 98.2% and 36.4%, whereas HPE showed a sensitivity and specificity of 97.3% and 100%. The accuracy of PMCT and HPE were 67.2 % and 98.6 % respectively. In conclusion, HPE had better accuracy compared to PMCT with almost similar sensitivity but higher specificity. PMCT may act as a good screening tool for pneumonia but is insufficient to substitute conventional autopsy in the diagnosis of pneumonia. Routine histology sampling during autopsy should be practised whenever dealing with sudden death.

The purpose of the present study was to evaluate the ranges of Hounsfield unit (HU) found in body fluids, putrefaction fluids, and blood on postmortem CT and how these ranges are affected by postmortem interval, temperatures, and CT beam energy. Body fluids, putrefaction fluids, and blood from a total of 53 corpses were analyzed to determine the ranges of HU values from postmortem CT images that were taken prior to autopsy. The fluids measured in CT images were obtained at autopsy and examined in terms of macroscopic and microscopic appearances. Body fluids and blood were also collected in plastic bottles, which were subjected to CT scans at different beam energies (80-130 kV) and at various fluid temperatures (4 to 40 °C). At a postmortem interval of 1 to 4 days, the ranges of HU values of the serous fluids (13-38 HU) and the nonsedimented blood (40-88 HU) did not overlap. In the sedimented blood, the upper serum layer exhibited HU value ranges that overlapped with those of the serous fluids. The putrefaction fluids exhibited a range of HU values between 80 and -130 HU. Elevated HU values were observed in fluids with accretive cell impurities. HU values decreased slightly with increasing temperature and CT beam energy. We concluded that serous fluids and blood in fresh corpses can be characterized and differentiated from each other based on HU value ranges. In contrast, body fluids in decomposed corpses cannot be differentiated by their HU value ranges. Different beam energies and corpse temperatures had only minor influences on HU value ranges and therefore should not be obstacles to the differentiation and characterization of body fluids and blood.

Objective. To determine the accuracy of postmortem computed tomography (PMCT) for the assessment of causes in nontraumatic deaths in children. Study Design. We enrolled cases of nontraumatic deaths of infants and children who underwent PMCT at a single center. The presumed cause of death determined by PMCT was prospectively compared with the clinical and pathological diagnoses of deaths. Results. Thirty-eight cases were enrolled for analysis. Among them, seven cases also underwent conventional medical autopsy. PMCT revealed an identifiable cause of death in accordance with the clinical diagnosis of death in 16 cases of the 38 cases (the concordance rate was 42%) and in accordance with the autopsy cause of death in four of the seven autopsy cases (the concordance rate was 57%). Among eight cases with unknown cause of death by clinical diagnosis, four cases (50%) were identified with cardiac tamponade as a cause of death (one case) and intracranial hemorrhage suggesting abuse (3 cases). Conclusions. PMCT seems to be a promising technique that might serve as a substitute for conventional medical autopsy and give us the complementary information to clinical diagnoses particularly in cases of child abuse. Larger multicenter trials are worthwhile to validate the general feasibility of PMCT.

Tuberculosis (TB) caused by Mycobacterium tuberculosis remains a serious public health threat with the World Health Organisation (WHO) reporting 5.8 million cases and 1.3 million deaths in the year 2020 due to TB. TB can be diagnosed by imaging, histopathological and bacteriological methods with culture remaining the gold standard. This study was performed to look at the sensitivity and specificity of post-mortem computed tomography (PMCT) imaging when compared to culture in diagnosing pulmonary tuberculosis. This was a retrospective comparative study looking at post mortem cases where lung tissue samples sent for TB culture at Hospital Kuala Lumpur were compared against PMCT imaging. Exclusion criteria included contaminated samples, decomposed cases, immunocompromised subjects and those below 18 years of age. Subjects included 80 medico-legal autopsy cases at the National Institute of Forensic Medicine, Hospital Kuala Lumpur, Malaysia who had whole body PMCT done in accordance with the Institute's protocol and tissue samples sent for bacteriology culture for tuberculosis. PMCT findings were positively associated with acid-fast organisms in 23.5 out of 33 cases (71.2%). Our study also showed that PMCT had a sensitivity of 71.3% and specificity of 54.3% (95% CI: 39.5-68.4) in diagnosing TB based on the protocol set in this study. This study showed that there was relatively good agreement between radiological PMCT findings and bacterial culture, suggesting that radiological examination is a relatively reliable tool for preliminary screening and possible diagnosis of TB prior to a postmortem examination which would be beneficial in reducing the risk of transmission of TB to health workers during autopsy.

ObjectivesFire deaths are challenging fatalities for forensic pathologists, as the main question of whether death was due to the fire or not needs to be answered. In this retrospective study, we assessed whether post-mortem computed tomography (PMCT) has an added value prior to a forensic autopsy of burned victims.MethodsFrom 2008 to 2016, a PMCT was performed in 50 burned corpses prior to a complete forensic autopsy. In retrospect, all 50 PMCT scans were systematically assessed by a forensically experienced radiologist, masked from the autopsy reports. Subsequently, the PMCT findings were compared with the autopsy reports.ResultsHeat fractures, contractions and destruction of extremities, subcutaneous emphysema and post-mortem gas collections were easier to detect by PMCT compared to autopsy. Alterations by penetrating and blunt trauma and the presence of foreign bodies were easy to detect by PMCT as well by autopsy. PMCT was, however, not successful in detecting signs of vitality during the fire, detection of superficial thermal injuries and to answer the main question of the forensic autopsy, which is to investigate the cause of death.ConclusionsPMCT prior to autopsy is a valuable add-on in the post-mortem forensic investigation of burned victims for detection of hidden signs of trauma, gas collections and foreign bodies. However, since PMCT cannot answer the two main questions in forensic examination—determining the cause of death and detecting signs of vitality during the fire—it cannot replace an autopsy.Key Points• Post-mortem CT (PMCT) in burned victims shows hidden signs of trauma.• Foreign bodies and gas collections can easily be detected.• Cause of death and vitality signs cannot be assessed by PMCT.

In cases of decomposition, carbonization, or drowning of bodies, dental comparison is key for forensic identification. To this end, post-mortem computed tomography (PMCT), a non-invasive and reproducible technique, enables detailed visualization of dental structures. However, metallic artefacts and density saturation may impair image quality. This study developed and validated a decision tree-based scoring scale to determine the most efficient PMCT filter and reconstruction modality for forensic dental analysis. PMCT scans from ten autopsies, each with an intraoral reference examination, were analysed. Six filters (Bone and Soft Tissue, with or without Iterative Metal Artifact Reduction (IMAR) and Extended Hounsfield Unit (EHU) filters) and four reconstruction modalities (Multiplanar Reconstruction (MPR) and Maximum Intensity Projection (MIP) at 2-, 5-, and 10-mm thickness) were compared. Each body's dental restoration was evaluated using a decision tree assigning an image quality score from 0 to 4. Inter- and intra-observer agreement was assessed using the weighted Cohen's Kappa coefficient. Two hundred dental treatments were analysed. Intra-observer agreement was substantial to almost perfect across all filters (k > 0.61), highest for the EHU bone filter (k = 0.81). Inter-observer agreement ranged from fair (IMAR soft tissues, k = 0.33) to good (EHU soft tissue, k = 0.75). The best concordance was obtained for amalgam and composite restorations using EHU filters. The decision-tree scoring scale improved reproducibility and objectivity in PMCT interpretation. The EHU filter significantly enhanced dental image quality and could be integrated into forensic identification protocols, particularly in mass disaster contexts.