External frozen-image study evaluates AI colorectal-polyp characterization
Source check: Source checked. Not independent scientific confirmation.
Finding: An external study tested a ResNet152-based colorectal-polyp characterization system on 115 images from 66 patients, using histology as the reference. For diminutive polyps, reported negative predictive values were 80% with white light and 90.9% with image enhancement, with wide confidence intervals; surveillance-interval agreement exceeded 90%. Images came from colonoscopy videos but assessment used frozen images. Real-time procedural benefit, safety of avoiding histology and broader clinical performance were not established.
Technology: NEC CADx; ResNet152
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- ResNet152 trained on 55,890 histologically labelled images from 11,068 lesions; distinct external clinical image evaluation. Frozen-image histology classification, not demonstrated real-time procedure outcomes.
- Test and result
- External 66-patient frozen-image diagnostic evaluation against polyp histology and endoscopist interpretation.
- Limitation / still unknown
- Real-time colonoscopy performance, safe pathology omission and generalizability remain unestablished.
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/PMC9614381/
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 full article Methods, Results and limitations plus Europe PMC bibliographic metadata; PMC full article..
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.