Artificial Intelligence Assisted Augmented Reality in Minimally Invasive Adrenal Surgery

Abstract

Abstract Background The integration of advanced technologies into surgical practice has revolutionized minimally invasive (MIS) procedures, enhancing precision and outcomes. This study explores the innovative use of patient-specific 3D models rendered in an augmented reality setting in adrenal surgery by using AI models as a guidance. Method The deep learning models that were previously developed to recognize anatomical structures in MIS transabdominal left and right sided adrenal surgeries performed in a tertiary center were used in this study to integrate patient specific 3D models of anatomical structures created by computed tomography (CT) images. Results A total of 2000 images extracted from minimally invasive transabdominal videos were used to develop deep learning models. Two pilot patients were selected from left and right sided surgeries to develop 3D models by using ExtremePACS tools. Both models showed successful auto-synchronization with real time surgery videos. Conclusion These models may enable enhanced preoperative planning and intraoperative navigation by providing surgeons with real-time, spatially accurate visualizations of adrenal anatomy and adjacent structures.