Remote detection of the critical view of safety in pediatric laparoscopic cholecystectomy using artifitial intelligence

Abstract

Background Laparoscopic cholecystectomy (LC) is increasingly performed in pediatric patients. Bile duct injury remains one of its most serious complications. The critical view of safety (CVS) aims to reduce this risk, but its identification is subjective. Artificial intelligence (AI) has shown promise in adult surgery for CVS detection but it has not been applied in pediatrics. Remote implementation of AI could reduce subjectivity and improve access to advanced tools. Methods A prospective, observational, cross-sectional, single-blind study was conducted between May and August 2025. An AI algorithm trained with 346 validated adult LC cases was tested live in 50 pediatric patients. Surgeries were transmitted in real time via teleconferencing to a second center where the algorithm processed the surgical image and identified CVS structures. Two expert surgeons, blinded to the algorithm, assessed CVS presence independently. Results A total of 50 patients (38 females and 12 males were classified as middle childhood to early adolescence) were included. The mean body mass index was 29.3±5.8. CVS was detected fully in 37 cases. In 13 cases, one or more elements were absent. Agreement between the algorithm and surgeons assessments was 100%. No postoperative complications were reported. Conclusion Remote AI-assisted CVS detection in pediatric LC is feasible, safe and consistent with expert evaluation.