Recurrent networks classify cancer-bearing mouse ovary images
Source check: Source checked. Not independent scientific confirmation.
Finding: A convolutional LSTM classified optical-coherence tomography of excised ovaries from sixteen mice, half from an ovarian-cancer model and half wild-type controls. Tissue pathology supported the disease labels. Across repeated leave-one-animal-out evaluations, mean AUC was 0.81; the best run reached 0.98. This is a small retrospective preclinical imaging validation, not human ovarian-cancer screening or evidence of earlier detection or improved treatment outcomes.
Technology: Convolutional long short-term memory; VGG and three-dimensional CNN comparisons
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- The convolutional LSTM learns sequential relationships between depth-resolved OCT slices and classifies tissue from cancer-model versus wild-type animals, using pathology-supported labels.
- Test and result
- Existing ex-vivo ovarian OCT recordings from eight cancer-model and eight wild-type mice, with pathological verification and leave-one-animal-out testing repeated ten times. Mean convolutional-LSTM AUC .81 is distinguished from the selected peak .98.
- Limitation / still unknown
- Human diagnostic transfer, clinical screening, early-stage detection and benefit remain unestablished. Only sixteen animals and one held-out animal per fold limit uncertainty estimates; peak-run performance is not representative average performance.
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/PMC8785933/
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: Primary relevant animal-cohort/reference, preprocessing, learned architecture, animal-wise validation Methods, Results/Table2 and limitations; cited earlier experimental protocols not independently re-reviewed. Exact primary identity/date and fresh primary PubMed correction/retraction relationships checked..
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 1/5 · 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.