With AI Assistance in Colonoscopy, Benign Lesions More Often Removed (GI & Endoscopy News)

A study at the Minneapolis VA Medical Center found that AI-assisted colonoscopies (AIAC) led to a 4% absolute increase in the resection of benign lesions compared to unassisted procedures. AIAC uses computer-aided detection (CADe) to enhance adenoma detection rates (ADR) and reduce miss rates but may also result in unnecessary removal of non-neoplastic lesions.

Researchers analyzed 1,040 colonoscopies—599 with AI assistance and 441 without—conducted for screening, surveillance, or positive fecal tests. Both groups had balanced demographics, and the same eight endoscopists, all new to AI technology, performed the procedures.

The findings highlight AI’s potential to improve detection while raising concerns about over-resection of benign lesions, emphasizing the need to refine AI algorithms to better distinguish lesion types.