Motion prediction for computer-assisted beating heart surgery.

Abstract

Off-pump totally endoscopic coronary artery bypass grafting is a milestone for cardiac surgery, and still a technical challenge. Indeed, the fast and complex cardiac motion makes this operating method technically demanding. Therefore, several robotic systems have been designed to assist the surgeons by compensating for the cardiac motion and providing a virtually motionless operating area. In the proposed systems, the servoing schemes often take advantage of a prediction algorithm that supplies the controller with some future heart motion. This prediction enlarges the control-loop bandwidth, thus allowing a better motion compensation. Obviously, improving the prediction accuracy will lead to better motion-compensation results. Thus, a current challenge in computer-assisted cardiac surgery research is the design of efficient heart-motion-prediction algorithms. In this paper, a detailed survey of the main existing prediction approaches is given and a classification is provided. Then, a novel prediction technique based on amplitude modulation is proposed, and compared with other techniques using in vivo collected datasets. A final discussion summarizes the main features of all the proposed approaches.