Bayesian filtering to improve the dynamic accuracy of electromagnetic tracking

Abstract

Tracking systems are essential components for many computer assisted interventions because they enable the doctor to visualize anatomical information, derived from preoperative or intraoperative images, registered with respect to the actual patient anatomy. This paper presents two applications of Bayesian filters: Particle Filter (PF) and Extended Kalman Filter (EKF) to obtain accurate dynamic tracking performance from an electromagnetic tracking (EMT) system, even if the EMT cannot provide the full measurement state at each sampling interval (for example, when transmit coils are driven sequentially and/or receive coils are not sampled simultaneously). Experiments are performed with a custom EMT system, consisting of a transmitter coil array and one or more receiving coils, to demonstrate that the proposed method provides good dynamic tracking accuracy at different velocities.