Real-Time Computer-Aided Detection During Colonoscopy-Effects on Advanced, Diminutive, and Non-Neoplastic Lesion...

Computer-aided detection (CADe) increases adenoma detection rate (ADR), but whether this improvement reflects greater detection of clinically consequential neoplasia remains uncertain. We evaluated the effects of real-time CADe on advanced, large, diminutive, and non-neoplastic lesions as distinct outcomes.

We searched PubMed, Embase, CENTRAL, ClinicalTrials.gov, and WHO ICTRP through August 27, 2026, supplemented by citation searching. Randomized controlled trials comparing real-time CADe-assisted with non-CADe colonoscopy were eligible. Patient-level risk ratios (RRs) were pooled using random-effects meta-analysis with restricted maximum likelihood estimation and Hartung-Knapp confidence intervals. Risk of bias was assessed with RoB 2 and certainty of evidence with GRADE.

Ninety eligible randomized trial families were identified; 64 contributed to at least one primary quantitative synthesis. CADe did not demonstrate clear evidence of increased advanced adenoma detection (25 trials; RR 1.03, 95% CI 0.97-1.10; I² = 0%) or adenomas ≥ 10 mm (5 trials; RR 1.18, 95% CI 0.92-1.53; I² = 0%). A definition-restricted advanced-adenoma analysis was also null (16 trials; RR 1.00, 95% CI 0.95-1.04). CADe increased overall ADR on average (63 trials; RR 1.16, 95% CI 1.12-1.20; I² = 61.2%) and non-neoplastic lesion detection (9 trials; RR 1.16, 95% CI 1.03-1.31). CADe also increased diminutive adenoma detection in the primary analysis (5 trials; RR 1.22, 95% CI 1.05-1.42; moderate-certainty evidence), although the confidence interval crossed the null after exclusion of one high-risk result (4 trials; RR 1.24, 95% CI 0.97-1.57). Advanced colorectal neoplasia detection showed no clear difference (5 trials; RR 1.02, 95% CI 0.95-1.09).

Real-time CADe increases overall ADR on average and increases non-neoplastic lesion detection, with moderate-certainty evidence suggesting increased diminutive adenoma detection. However, it did not demonstrate clear evidence of increased advanced adenoma, adenoma ≥ 10 mm, or advanced colorectal neoplasia detection. Higher ADR alone should not be assumed to represent greater detection of clinically consequential neoplasia.