Region of Interest-guided Unsupervised Anomaly Detection of Forceps Force

• Published in: Transactions of Japanese Society for Medical and Biological Engineering Vol. Annual62; no. Abstract; p. 125_2
• Main Authors: 庄, 文輝, 中尾, 恵, 粂, 直人, 増井, 仁彦
• Format: Journal Article
• Language: Japanese
• Published: 公益社団法人 日本生体医工学会 2024; Japanese Society for Medical and Biological Engineering
• ISSN: 1347-443X; 1881-4379
• DOI: 10.11239/jsmbe.Annual62.125_2

The lack of haptic feedback in robotic surgical systems can lead to unintended tissue damage due to excessive mechanical force. In this study, we propose an approach by formulating the force feedback problem as an image-based anomaly detection task. Our method employs Bidirectional Generative Adversarial Networks (BiGANs) integrated with a Region of Interest (ROI) localization module to enhance the model's attention on the contact area between surgical instruments and tissue. The model is trained to distinguish normal and abnormal force samples based on deformation features in an organ. Extensive experiments on an ex vivo porcine spleen manipulation dataset demonstrate the efficacy of our proposed method in accurately detecting abnormal forces.