Artificial intelligence-assisted prediction of bile duct bifurcation site in pure laparoscopic donor right hepatectomy: a retrospective feasibility study.

Purpose: Accurate identification of the bile duct bifurcation site is crucial in pure laparoscopic donor right hepatectomy (PLDRH) for living donor liver transplantation. This study aimed to develop and evaluate a deep learning model to predict the bile duct bifurcation site for surgical precision. Methods: We retrospectively analyzed 55 PLDRH procedures conducted between August 2021 and April 2022. A deep learning model combining UNet with a MiT-B3 encoder was trained on 150 manually annotated frames. We then incorporated expert-reviewed pseudo-labels from an additional 901 frames to refine the model. Performance was evaluated using 5-fold cross-validation and an independent test set. Results: Clinical evaluation showed a 97% accuracy in 5-fold cross-validation and 93.3% accuracy in the independent test set. From the initial to final model, dice similarity coefficient improved from 0.392 to 0.472, intersection over union from 0.279 to 0.339, and sensitivity from 0.487 to 0.643, while specificity remained consistent at 0.993. Conclusion: The proposed artificial intelligence (AI) model demonstrated strong clinical performance in predicting the bile duct bifurcation site during PLDRH. Despite modest quantitative scores, the high clinical accuracy highlights the potential of integrating AI for precise donor hepatectomy.