Artificial Intelligence-Driven 3D Simulation System for Enhanced Preoperative Planning in Gastric Cancer Surgery: A Retrospective Validation Study.
Authors
Affiliations (6)
Affiliations (6)
- Department of Surgery, Shiga University of Medical Science, Otsu, Shiga, Japan. Electronic address: [email protected].
- Department of Radiology, Shiga University of Medical Science, Otsu, Shiga, Japan.
- Division of Information Science, Nara Institute of Science and Technology, Ikoma, Nara, Japan.
- Department of Artificial Intelligence in Diagnostic Radiology, The University of Osaka Graduate School of Medicine, Osaka, Japan.
- Institute for Radiation Science, The University of Osaka / Graduate School of Medicine, The University of Osaka, Osaka, Japan.
- Department of Surgery, Shiga University of Medical Science, Otsu, Shiga, Japan.
Abstract
Few studies have developed artificial intelligence systems for automatic recognition of the anatomy of the stomach, a dynamic organ capable of expansion and contraction. This study aimed to create a three-dimensional simulation to assist gastric cancer surgery by combining artificial intelligence models to visualize the positional relationships among the stomach, surrounding organs, and blood vessels. We developed a deep learning-based model using an artificial intelligence system to segment abdominal organs and detect blood vessels, including mid-artery level structures from contrast-enhanced computed tomography images. Surgical structures including the stomach, pancreas, and arteries were extracted using a blood vessel detection model. Two surgeons and two radiologists evaluated 51 three-dimensional images for structural detection confidence using a five-point scale and compared them to standard computed tomography images. A retrospective analysis of 51 cases of preoperative patients with gastric cancer demonstrated that artificial intelligence-generated images provided clear visualization of the spatial relationships between blood vessels and organs. Structures, including the left hepatic-left-gastric artery, common duct and its branches, and the short gastric artery distinct from the splenic artery, were clearly identified. These findings were useful for surgical planning. The reliability score for detecting blood vessels was significantly higher (P < 0.05) for the artificial intelligence images compared to the computed tomography images, with good agreement among the evaluators. Automatic organ recognition systems are promising, valuable tools for gastric cancer surgery, improving preoperative planning and potentially reducing operative time and complications.