Artificial intelligence is steadily moving from hype to hands-on utility in endoscopy suites. In this edition of Sharma’s Endoscopy Insights, two recent studies explore how AI could enhance both optical diagnosis and quality measurement during colonoscopy.
The first study evaluated computer-aided diagnosis (CADx) in identifying sessile serrated lesions (SSLs)—a known blind spot in colorectal cancer prevention. Endoscopists demonstrated higher diagnostic accuracy when CADx supported optical classification, particularly in distinguishing SSLs from hyperplastic polyps. However, current systems still face limitations in reliably characterizing these lesions, highlighting the need for further refinement.
The second study focused on withdrawal time, a key colonoscopy quality metric. Here, AI outperformed physicians in accurately calculating withdrawal time, offering a more standardized and objective approach to quality assessment.