Minimally invasive robotic surgery is subjected to serious intraoperative complications caused by accidental damage to delicate organs, arteries or veins. One of the possible risk factors falls on the surgeon’s skills. Therefore, assistive systems able to supervise the operations and prevent such complications can potentially bring significant improvements in terms of safety during robotic surgery. To prove this hypothesis, we are developing novel vision systems to robustly track safety regions on real-time 3D surgical videos. In addition, we are developing and embedding supervisory systems in real surgical robots to automatically protect delicate structures in the operating field.

Another important area of computer vision research within our lab is that of intraoperative tissue classification. Automatic image-based tissue classification can be a valuable solution to provide decision support and context awareness to clinicians. Therefore, we are researching and developing novel vision-based solutions to these problems. Specifically, we are applying machine learning techniques to achieve robust and reliable tissue segmentation and classification in challenging medical images. Applications include the detection of early tumors in endoscopic video, steatosis assessment in liver grafts, classification of abdominal organs, and blood vessel segmentation during neurological interventions.