V14-05 Artificial Intelligence-Based Intraoperative Guidance: Demonstration of Real-Time Segmentation and Feature Tracking in Robotic Surgery

You have accessJournal of UrologyAdrenal/Renal Oncology II (V14)1 May 2024V14-05 ARTIFICIAL INTELLIGENCE-BASED INTRAOPERATIVE GUIDANCE: DEMONSTRATION OF REAL-TIME SEGMENTATION AND FEATURE TRACKING IN ROBOTIC SURGERY Luca A. Morgantini, Rebecca Canneto, Rogerio Garcia Nespolo, Yannek I. Leiderman, and Simone Crivellaro Luca A. MorgantiniLuca A. Morgantini , Rebecca CannetoRebecca Canneto , Rogerio Garcia NespoloRogerio Garcia Nespolo , Yannek I. LeidermanYannek I. Leiderman , and Simone CrivellaroSimone Crivellaro View All Author Informationhttps://doi.org/10.1097/01.JU.0001008704.74547.03.05AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookLinked InTwitterEmail Abstract INTRODUCTION AND OBJECTIVE: Artificial Intelligence (AI) has the potential to greatly enhance the precision and efficiency of robotic surgery can be significantly by providing real-time feature tracking and segmentation. Additionally, AI can also provide an accurate measure of a surgeon's performance and real-time guidance, ensuring optimal patient outcomes. We propose an innovative approach for real-time tracking and segmenting key anatomical structures during robotic surgery. METHODS: We adapted a state-of-the-art instance segmentation tool, YOLACT++, and trained it with segmented frames from single port retroperitoneal surgical videos. Our emphasis was on the accurate recognition and segmentation of the renal artery, renal vein, inferior vena cava, ureter, kidney, psoas muscle, and the bipolar and monopolar forceps. RESULTS: Our AI model, post-training, demonstrated excellent in tracking and segmentation of the anatomical structures of interest. This video demonstrates real-time feature identification, with the model differentiating between anatomical structures and surgical instruments. CONCLUSIONS: This preliminary video suggest the role that AI may play in augmenting robotic surgical accuracy via real-time feedback. More studies are needed to generalize results to non-segmented videos and test the model's performance in real-time surgical settings. Source of Funding: N/A © 2024 by American Urological Association Education and Research, Inc.FiguresReferencesRelatedDetails Volume 211Issue 5SMay 2024Page: e1228 Advertisement Copyright & Permissions© 2024 by American Urological Association Education and Research, Inc.Metrics Author Information Luca A. Morgantini More articles by this author Rebecca Canneto More articles by this author Rogerio Garcia Nespolo More articles by this author Yannek I. Leiderman More articles by this author Simone Crivellaro More articles by this author Expand All Advertisement PDF downloadLoading ...