Artificial Intelligence and Its Role in Endoscopic Adenoma and Cancer Detection

Abstract Colorectal cancer incidence and mortality have declined over time, due in part to high-quality screening and surveillance colonoscopy. Nevertheless, postcolonoscopy colorectal cancer (PCCRC) occurs in up to 7% of cases and is inversely related to examination quality. Artificial intelligence-assisted colonoscopy aims to improve performance metrics and, ultimately, patient outcomes. Multiple randomized trials show that computer-aided polyp detection (CADe) increases adenoma detection, predominantly for diminutive lesions (≤5 mm). Computer-aided polyp characterization (CADx) enables real-time optical diagnosis, potentially shifting management of diminutive polyps by supporting resect-and-discard and diagnose-and-leave in situ strategies. Computer-aided quality assessment (CAQ) systems monitor key metrics—including cecal intubation rate, withdrawal time, speed, and mucosal exposure. Whether CADe alone leads to a reduction in PCCRC or cancer-related mortality remains to be determined; in the near term, a combined approach using CADe, CADx, and CAQ is most likely to deliver the greatest improvements in patient outcomes.