Artificial intelligence for polyp detection during colonoscopy: a systematic review and meta-analysis

Abstract

Background and study aims: Artificial intelligence (AI)-based polyp detection systems during colonoscopy aim at increasing lesion detection and improving colonoscopy quality. Patients and methods: We performed a systematic review and meta-analysis of prospective trials to determine the value of AI-based polyp detection systems for detection of polyps and colorectal cancer. We performed systematic searches in MEDLINE, EMBASE and Cochrane CENTRAL. Independent reviewers screened studies and assessed eligibility, certainty of evidence, and risk of bias. We compared colonoscopy with and without AI by calculating relative and absolute risk and mean differences for detection of polyps, adenomas, and colorectal cancer. Results: Five randomized trials were eligible for analyses. Colonoscopy with AI increased adenoma detection rates (ADRs) and polyp detection rates (PDRs) compared to colonoscopy without AI: ADR with AI 29.6% [95% confidence interval, 22.2-37.0], versus 19.3% [12.7-25.9] without AI (relative risk (RR) 1.52, [1.31-1.77], high certainty); PDR 45.4% [41.1-49.8] with AI, versus 30.6% [26.5-34.6] without AI (RR 1.48, [1.37-1.60], high certainty). There was no difference in detection of advanced adenomas between the two groups (mean number of advanced adenomas per colonoscopy 0.03 for each, high certainty). Mean number of adenomas per colonoscopy were higher for small adenomas (≤5 mm) with AI compared to non-AI colonoscopy (mean difference 0.15, [0.12-0.18]), but not for larger adenomas (>5-≤10mm: mean difference 0.03, [0.01-0.05]; >10 mm: mean difference 0.01, [0.00-0.02], high certainty). Data on cancer are unavailable. Conclusions: AI-based polyp detection systems during colonoscopy increase detection of small, non-advanced adenomas and polyps, but not advanced adenomas.

Artificial intelligence in colonoscopy ‐ Now on the market. What's next?

“…CADe is designed to help endoscopists to avoid missing polyps, which might contribute to increasing their ADRs, while CADx is designed to help endoscopists to interpret polyp histology with greater accuracy than endoscopists themselves normally do. The performance of both CADe and CADx has been assessed in several prospective studies, with surprisingly encouraging results, whereas retrospective benchmark tests are considered as suboptimal for accurate assessment of AI tools because of selection bias 7,8 . Additionally, an increasing number of AI products designed for colonoscopy are already available on the market, especially in Europe and Asia.…”

“…This number is surprisingly large considering that only around 10 RCTs have been published in the whole medical AI field 16 . Subsequently, two meta‐analyses were published to interpret the impact of these RCTs 7,17 . These meta‐analyses reported that the use of CADe during colonoscopy is likely to increase ADR by ~50%, which may have great benefits in cancer prevention.…”

“…The performance of both CADe and CADx has been assessed in several prospective studies, with surprisingly encouraging results, whereas retrospective benchmark tests are doi:10.1111/jgh.15339 considered as suboptimal for accurate assessment of AI tools because of selection bias. 7,8 Additionally, an increasing number of AI products designed for colonoscopy are already available on the market, especially in Europe and Asia. Therefore, physicians, patients, academic societies, and industries have high expectations of broader use of AI in colonoscopy.…”

“…16 Subsequently, two meta-analyses were published to interpret the impact of these RCTs. 7,17 These meta-analyses reported that the use of CADe during colonoscopy is likely to increase ADR by~50%, which may have great benefits in cancer prevention. Additionally, no major drawbacks to the use of AI, including serious adverse events, were confirmed in the trials.…”

Can artificial intelligence help to detect dysplasia in patients with ulcerative colitis?

“…UC-associated dysplasia is often flat with an unclear boundary from the surrounding tissues, making it difficult to detect [2]. AI-based polyp detection systems are used during colonoscopy to increase lesion detection [3]. The EndoBRAIN-EYE system can reportedly identify colorectal lesions with high accuracy in non-UC patients [4,5].…”