Generalized Predictive Control of a Surgical Robot for Beating-Heart Surgery Under Delayed and Slowly-Sampled Ultrasound Image Data

Abstract

Operating on a beating heart would offer many benefits to patients. The risks associated with heart-lung machines used in arrested-heart surgery would be eliminated and the effectiveness of reconstructive procedures could be judged immediately. However, the heart's fast beating motions make operating on a beating heart impossible for the surgeon. With advances in surgical robotics, we can now envision a robot-assisted surgical system that first synchronizes the surgical robot with the beating heart motion and then lets the surgeon operate through teleoperation on a seemingly motionless point on the heart. This letter presents such a system that relies on both motion prediction and predictive control to overcome the delays introduced in acquiring the beating heart's position from ultrasound images. Also, slowly sampled position data originating from low-frame-rate ultrasound images is treated with cubic interpolation and extended Kalman filter-based prediction. The results of a user study involving a task based on mitral valve annuloplasty are presented to show the proposed method's efficacy in terms of synchronizing the surgical robot to the beating heart motion.