Artificial Intelligence for Automated Recognition of Hepatocystic Anatomy During Laparoscopic Cholecystectomy: Current Evidence, Clinical Readiness, and Future Directions

Background and Objectives: Bile duct injury remains a major safety concern during laparoscopic cholecystectomy, and reliable interpretation of hepatocystic anatomy is fundamental to safe dissection. Artificial intelligence (AI)-based computer vision may support anatomical recognition, critical-view-of-safety (CVS) assessment, and intraoperative decision support. This narrative review synthesized the current evidence, clinical applications, readiness for implementation, and future requirements of anatomy-aware AI during laparoscopic cholecystectomy. Materials and Methods: Five bibliographic databases were searched through 15 August 2026, supplemented by citation tracking and targeted searches. Studies were classified according to clinical task, dataset, reference standard, validation design, performance metrics, real-time capability, human-factor assessment, and clinical-readiness stage. The evidence base comprised 105 verified references: 50 primary AI reports and 55 contextual or methodological sources; 39 direct model-development or evaluation reports were characterized in detail. Results: Investigated applications included landmark detection, semantic segmentation, CVS assessment, safe- and hazard-zone mapping, multimodal analysis incorporating indocyanine green fluorescence, automated documentation, education, and real-time perceptual prompting. Among the 39 direct reports, 24 remained at the offline proof-of-concept or internal-validation stage, eight achieved temporal, external, or multicenter validation, six demonstrated prospective operating-room feasibility, and one reached post-deployment surveillance. The latter evaluated a surgical-process outcome rather than patient morbidity. Generalizability was constrained by dataset overlap, heterogeneous reference standards and metrics, domain shift, and underrepresentation of difficult cholecystectomy. No included study demonstrated reduced bile duct injury or other patient-level benefit. Conclusions: Anatomy-aware AI has progressed from technical feasibility to early clinical translation, with the strongest near-term rationale in documentation, video triage, education, coaching, and quality assurance. Surgeon-facing deployment should remain adjunctive, staged, and risk proportionate, requiring multicenter and temporal validation, explicit uncertainty and abstention mechanisms, human-factor evaluation, comparative-effectiveness studies, and continuous post-deployment surveillance before routine clinical adoption.