AI-Assisted Cholecystectomy: A Novel Framework for Real-Time Critical View of Safety Verification

Abstract

The Critical View of Safety (CVS) is the accepted standard for safe laparoscopic cholecystectomy, yet its achievement is inconsistently documented and surgeon-dependent. We developed and describe an AI-based framework using computer vision to verify CVS in real time during laparoscopic cholecystectomy. The system analyzes laparoscopic video frames and provides a binary CVS-achieved/not-achieved classification with a confidence score. In initial validation across 45 recorded procedures, the AI achieved 89% agreement with expert consensus CVS assessment (κ=0.76). The framework offers a potential safety checkpoint that could reduce the incidence of bile duct injury, the most feared complication of laparoscopic cholecystectomy.