Multimodal learned features classify prostate lymph-node involvement
Source check: Source checked. Not independent scientific confirmation.
Finding: A stacked model combined pretrained MRI image features, radiomics and clinical information to classify pathology-referenced pelvic lymph-node involvement. Its AUC was 0.92 in 71 internal test patients and0.76 in 50 second-center patients. This retrospective validation does not establish avoided surgery or routine clinical benefit.
Technology: Pretrained image-network embeddings; random-forest signatures; AutoGluon stacked ensemble
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- ImageNet-pretrained networks embed tumor/peritumoral MRI regions; forests select/learn imaging signatures and an ensemble integrates them with clinical and reader variables.
- Test and result
- 401 selected surgical patients:280 development,71 internal test and50 second-center test. Only six second-center patients had positive nodes; sensitivity there was50% at the development-selected threshold.
- Limitation / still unknown
- Retrospective surgical selection, small external positive count and site-specific readers limit transfer. Tumor segmentation used radiological-pathological correlation; preprocessing and feature-selection isolation are not completely specified. Source feature-importance description contains unrelated linker terminology. Surgery avoidance and decision-curve benefits are hypothetical.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC8167242/
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: 21 Nov 2026. Review scope: Actual primary clinical/model Methods2.1–2.8, diagnostic Results3.2 and limitations; fresh exact PubMed date/notice metadata. No independent implementation or supplementary detail reproduction..
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 2/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.