The Critical View of Safety (CVS) is the established method to minimize bile duct injury risk during Laparoscopic Cholecystectomy (LC). This investigation assessed the precision and real-world functionality of a pre-trained AI model for confirming successful CVS attainment in elective LC surgeries within an actual operating theatre environment. A deep learning neural network, earlier trained on annotated footage from 700 LC procedures to recognize CVS elements, was connected directly to the laparoscopic imaging system in the operating room. This allowed ongoing, instantaneous assessment of CVS during elective cases. AI-generated feedback was systematically compared against the surgeon’s intraoperative evaluation, alongside documentation of overall CVS success rates. Across 40 successive LC operations, surgeons indicated CVS achievement in 34 cases (85%). In the 6 procedures where CVS could not be obtained owing to intense inflammation or distorted anatomy, the AI system aligned completely with the surgeons’ judgment and correctly refrained from confirming CVS. Of the 34 cases reported as successful by surgeons, the AI model correctly detected CVS in 33 instances. This yielded a specificity of 100% [95%-CI 98.1%, 100%] and sensitivity of 97% [95%-CI 96.1%, 98.2%] for CVS detection. The trained AI model accurately identified CVS in elective LC with exceptional reliability in real-time operating room scenarios. Its deployment may promote greater consistency in CVS attainment, leading to safer procedures and better patient outcomes.
