An anatomy-informed CT phenotype estimates extrapancreatic perineural-invasion risk in pancreatic cancer
Source check: Source checked. Not independent scientific confirmation.
Finding: The locked PROPHET model used tumor-vessel spatial features from CT to estimate pathologically confirmed extrapancreatic perineural invasion (EPNI) in resected pancreatic ductal adenocarcinoma. Across 1,854 patients, reported AUCs included 0.92 in prospective external testing and 0.85 in retrospective external testing; in a reader study, score-plus-visualization assistance increased mean sensitivity from 82% to 93% and reduced reading time. This is an association and diagnostic-support result, not evidence of treatment benefit or direct visualization of nerves. It concerns a related but distinct imaging study from the active catalog record on MRI prediction of perineural invasion (2026-10-machine-learning-models-integrating-2d-and-3d-intratumo--frai-2026-1975752); overlap in clinical endpoint is possible, while modality and reported model result differ.
Technology: PROPHET is a locked interpretable machine-learning model deriving an anatomy-informed CT spatial phenotype associated with pathologically confirmed EPNI.
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- PROPHET is a locked interpretable machine-learning model deriving an anatomy-informed CT spatial phenotype associated with pathologically confirmed EPNI.
- Test and result
- Multicenter testing included prospective and retrospective external cohorts, neoadjuvant and internal cohorts, and TCIA; six radiologists were assessed with and without combined score-and-visualization assistance.
- Limitation / still unknown
- The population comprised resected cases; the observational survival association does not establish causality or treatment benefit. TCIA specificity was low (2/19). Reader improvement cannot be attributed to the score or visualization separately. Similar perineural-invasion endpoint to active MRI record noted; distinct source and CT-based result.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://pubs.rsna.org/doi/10.1148/ryai.260357
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.
Review date: 22 Nov 2026. Review scope: Primary publisher full text, including Methods, Results, author affiliations and publication history..
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 4/5 · Validation 2/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.