Use of artificial intelligence in colonoscopy

Le résumé fourni par la source

Summary: Background: Artifi cial intelligence (AI) has emerged as a powerful tool to enhance and standardise the quality of colonoscopy. Its benefi ts include improved lesion detection, support for examination quality, and potential cost reduction by limiting the need for histological analyses and improving colorectal cancer screening strategies. Purpose: This review aims to evaluate current applications of AI in colonoscopy, focusing on computer-aided detection (CADe), computer-aided dia gnosis (CADx), assessment of infl ammatory bowel dis ease (IBD), quality indicators of colonoscopy, and an overview of expert society recommendations. Conclusions: AI has been shown to improve adenoma detection rates, especially among less experienced endoscopists. CADx may aid in optical dia gnosis of diminutive polyps, although its accuracy often falls short of thresholds required to replace histology. In the context of IBD, AI has demonstrated potential in standardising dis ease activity scoring and dysplasia detection, although clinical application remains experimental. Additional uses include automated assessment of bowel preparation, mucosal visualisation, and caecal intubation. Economic and regulatory aspects remain critical to broader implementation. AI holds promise in supporting quality improvement in colonoscopy, particularly through lesion detection and procedural standardisation. However, limitations persist – such as lack of standardised training data, inconsistent performance across AI models, and insuffi cient real-world data on long-term epidemiological impact. Implementation should be approached with transparency, awareness of limitations, and consideration of health system constraints. Key words: artifi cial intelligence – colonoscopy – CADe – CADx – infl ammatory bowel dis ease – quality of endoscopy

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.