Surgical Navigation and Lesion Detection System

The research team collaborated with the renowned Da Vinci surgery team at Tri-Service General Hospital (Dr. Wang Yu-chi) to develop a clinical surgical navigation and lesion detection system using semantic segmentation technology in deep learning. Due to the difficulty in labeling and collecting data for organs and lesions, the mainstream research goal is currently to detect surgical instruments. However, the target boxes for detection are all rectangles (bounding boxes) and cannot completely depict the edges of organs or tumors. Therefore, using semantic segmentation technology can segment each pixel in the image to achieve precise pixel-wise segmentation. The team has recently added recognition and segmentation of surgical instruments and malignant tumors (such as uterine fibroids and ovarian tumors) and hopes to further develop in the direction of surgical navigation or teaching.