The abstract is read at the publisher; the summary is JClub's.
Classifier-guided diffusion anomaly detection on postmortem brain CT identifies fatal intracranial haemorrhage with AUROC 0.991 (95% CI 0.984–0.997) in 33 cases.
- Why it matters: Forensic PMCT interpretation is manual and supervised deep learning is limited by scarce annotated datasets, so methods that learn from weak labels are needed.
- What they did: Using retrospectively collected PMCT from a single forensic institute (2011–2025), diffusion models were trained on 265 healthy subjects, with anomaly scores from slice reconstruction error.
- The result: The weakly supervised classifier-guided model achieved AUPRC 0.960 (95% CI 0.855–0.994), supporting automated triage tool development for postmortem brain imaging.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.