Novelty Detection in Catheter-Aorta Interaction
Yohannes Kassahun, Bingbin Yu, Abraham Temesgen Tibebu, Emmanuel Vander Poorten
In Co-SuR 2013 - Workshop on Cognitive Surgical Robotics: From Virtual Fixtures to Advanced Cooperative Control, (IROS-2013), 07.11.-13.11.2013, Tokyo, o.A., Nov/2013.
Abstract
:
A recently EU-funded FP7 project CASCADE (Cognitive AutonomouS CAtheter operating in Dynamic Environments, http://www.cascade-fp7.eu/), investigates autonomous catheter control and explores machine learning techniques to learn the input-output behavior of the catheter inside vessels of artificial mock-ups. The results from this study should enhance the understanding of catheter motion and control also during real interventions. An envisioned cognitive assistance system would employ this understanding to provide more intelligent guidance cues and shared control actions to the surgeon. Improved operational safety and performance in this complex procedure is the overall aim. Novelty detection [1] plays an important role for the cognitive assistance system envisioned in the project. During the learning phase, the system can detect novel catheter-aorta interactions. The detected interactions can either be learned autonomously, or be notified to the surgeon for verification before they are learned. During the operation, the system can detect anomalies (unexpected situations) and initiate the replanning module, which will be developed in the project, to guarantee safety.