Utilising artificial intelligence to identify surgical anatomy during laparoscopic donor nephrectomy - a validation and feasibility study

Comprehensive Summary

Deep learning (DL)-based computer vision (CV) has the potential to be applied in performing laparoscopic donor nephrectomies (LDN) with the aim to further decrease intraoperative risks and complications. With the goal of recognizing key anatomical structures and preventing operative injuries, the DL model was trained with thousands of annotated images and the YOLO v11x DL network. Performance metrics were calculated to reveal promising results in the machine learning design’s ability to accurately identify vital anatomical structures in LDN.

Outcomes and Implications

The deep learning (DL) based computer vision’s (CV) ability to accurately identify anatomical structures for intraoperative procedures serves as a stepping stone for future artificial intelligence (AI) models to be used in decreasing operative risk, increasing efficiency, and improving post-operative outcomes. Specifically, the ability to identify vital structures during high-risk surgical procedures provides an avenue for intra-operative guidance, education, and post-hoc operative analysis and standards evaluation.