CT-based AI risk support for cystic renal masses: multicenter validation, reader study, and a randomized implementation study
Source check: Source checked. Not independent scientific confirmation.
Finding: In a randomized implementation phase involving 5,869 adults at one hospital, the post-hoc intention-to-treat proportion with benign pathology among all randomized patients was lower with AI-assisted reporting (0.95% vs 3.54%; risk difference −2.59 percentage points, 95% CI −3.38 to −1.85). Among treated lesions, benign pathology was 16.4% vs 40.3%, but the intended-use cystic renal mass subgroup difference was not statistically significant. Retrospective development followed by external validation at five hospitals (AUROC 0.913–0.999; one external set had only three malignant cases), a six-radiologist reader study (AUROC 0.825 unaided to 0.973 aided; mean reading time 9.7 to 5.5 minutes), and a randomized implementation study. Caveats: single-centre, developer-run implementation; treated-lesion endpoint conditions on post-randomization treatment; subgroup result not significant; no long-term cancer outcome or safety evidence.
Technology: KITE-CRM is a deep-learning pathway using corticomedullary-phase contrast-enhanced CT as its sole model input to estimate lesion-level malignancy risk and support radiologist reporting; clinicians retained treatment decisions.
Publication: Peer-reviewed paper · Test context: prospective or independent check. Catalog categories, not an evidence grade.
- AI contribution
- KITE-CRM is a deep-learning pathway using corticomedullary-phase contrast-enhanced CT as its sole model input to estimate lesion-level malignancy risk and support radiologist reporting; clinicians retained treatment decisions.
- Test and result
- In a randomized implementation phase involving 5,869 adults at one hospital, the post-hoc intention-to-treat proportion with benign pathology among all randomized patients was lower with AI-assisted reporting (0.95% vs 3.54%; risk difference −2.59 percentage points, 95% CI −3.38 to −1.85). Among treated lesions, benign pathology was 16.4% vs 40.3%, but the intended-use cystic renal mass subgroup difference was not statistically significant.
- Limitation / still unknown
- Retrospective development followed by external validation at five hospitals (AUROC 0.913–0.999; one external set had only three malignant cases), a six-radiologist reader study (AUROC 0.825 unaided to 0.973 aided; mean reading time 9.7 to 5.5 minutes), and a randomized implementation study. Caveats: single-centre, developer-run implementation; treated-lesion endpoint conditions on post-randomization treatment; subgroup result not significant; no long-term cancer outcome or safety evidence.
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/PMC13755345/
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 source full text Methods and Results; version/date checked against bibliographic record..
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 3/5 · Validation 4/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.