Towards “real-time” tool-tissue interaction detection in robotically assisted laparoscopy

The real-time detection of surgeonpsilas action could be very useful for intelligent intraoperative assistance. Robotic assisted surgery provides quantified information that can be used to detect these actions. In this paper we present our attempt at spotting tool-tissue interactions from real surgery data, in the context of robotic assisted laparoscopy with the daVincitrade system, using both kinematic information and visual information. Our early results are encouraging in this very challenging context.