2D anomaly detection of fatal cerebral hemorrhage on postmortem CT
- 1University of Basel
Abstract
Postmortem computed tomography (PMCT) is increasingly used in forensic medicine, yet image interpretation remains largely manual and time-consuming, and supervised deep learning is constrained by the scarcity of annotated forensic datasets. This study investigated whether diffusion-based generative models can detect fatal cerebral haemorrhage on PMCT through reconstruction-based anomaly detection, using solely image-level labels derived from the documented cause of death. Postmortem brain CT examinations acquired at a single forensic institute between 2011 and 2025 were retrospectively analyzed, comprising 265 healthy subjects and 33 cases of fatal intracranial haemorrhage. After automated brain extraction, slice selection and intensity normalization, two diffusion frameworks were trained: an unsupervised denoising diffusion probabilistic model (AnoDDPM) trained on healthy anatomy alone, and a weakly supervised classifier-guided diffusion model exploiting image-level labels. Anomaly scores were derived from the reconstruction error between each slice and its pseudo-healthy reconstruction, and detection was assessed at the image level using 95% bootstrap confidence intervals. AnoDDPM achieved an AUROC of 0.928 (95% CI 0.812-0.989) and an AUPRC of 0.794 (95% CI 0.429-0.946), whereas the classifier-guided model reached an AUROC of 0.991 (95% CI 0.984-0.997) with an AUPRC of 0.960 (95% CI 0.855-0.994). These results demonstrate excellent feasibility of diffusion-based anomaly detection for forensic PMCT and indicate that weak, image-level supervision substantially improves detection, thereby supporting the future development of automated triage tools for postmortem brain imaging.
Related
- Multimodal Postmortem Investigation of Fatal Thoracic Metallic Projectile Trauma in a Jindo Dog: Radiographic, Computed Tomographic, Necropsy, and Histopathological Findings.
- Do Early Prognostic Measures Differentiate Mechanisms of ICU Death in Aneurysmal Subarachnoid Hemorrhage? An Analysis of Fatal Outcomes
- Association of postmortem mitragynine positivity with fatal drug toxicity in polysubstance exposures—a restrospective analysis
- A fatal case of cerebral scedosporiosis caused by Scedosporium boydii in a kidney transplant recipient: A case report and literature review
- PINK1 agonist MTK458 attenuates cerebral edema by protecting blood-brain barrier integrity in intracerebral hemorrhage mice
BibTeX
@misc{zirn2026anomaly,
title = {{2D anomaly detection of fatal cerebral hemorrhage on postmortem CT}},
author = {A. Zirn and E. Scheurer and C. Lenz},
howpublished = {medRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.14.26363002},
url = {https://www.medrxiv.org/content/10.64898/2026.09.14.26363002v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - 2D anomaly detection of fatal cerebral hemorrhage on postmortem CT
AU - A. Zirn
AU - E. Scheurer
AU - C. Lenz
PY - 2026
DA - 2026/09/14
PB - medRxiv
DO - 10.64898/2026.09.14.26363002
UR - https://www.medrxiv.org/content/10.64898/2026.09.14.26363002v1
AB - Postmortem computed tomography (PMCT) is increasingly used in forensic medicine, yet image interpretation remains largely manual and time-consuming, and supervised deep learning is constrained by the scarcity of annotated forensic datasets. This study investigated whether diffusion-based generative models can detect fatal cerebral haemorrhage on PMCT through reconstruction-based anomaly detection, using solely image-level labels derived from the documented cause of death. Postmortem brain CT examinations acquired at a single forensic institute between 2011 and 2025 were retrospectively analyzed, comprising 265 healthy subjects and 33 cases of fatal intracranial haemorrhage. After automated brain extraction, slice selection and intensity normalization, two diffusion frameworks were trained: an unsupervised denoising diffusion probabilistic model (AnoDDPM) trained on healthy anatomy alone, and a weakly supervised classifier-guided diffusion model exploiting image-level labels. Anomaly scores were derived from the reconstruction error between each slice and its pseudo-healthy reconstruction, and detection was assessed at the image level using 95% bootstrap confidence intervals. AnoDDPM achieved an AUROC of 0.928 (95% CI 0.812-0.989) and an AUPRC of 0.794 (95% CI 0.429-0.946), whereas the classifier-guided model reached an AUROC of 0.991 (95% CI 0.984-0.997) with an AUPRC of 0.960 (95% CI 0.855-0.994). These results demonstrate excellent feasibility of diffusion-based anomaly detection for forensic PMCT and indicate that weak, image-level supervision substantially improves detection, thereby supporting the future development of automated triage tools for postmortem brain imaging.
ER -