Tracking of Heart Wall Motion using Unscented Kalman Filter

Abstract

We propose a semiautomatic method based on unscented Kalman filter (UKF) to track left ventricular heart wall motion from a series of echocardiographic images. Our method requires manual drawing of contour on the first frame of the image sequence to represent the left ventricle boundary. Heart wall tracking from echocardiographic image is a difficult task because of poor signal-to-noise ratio and presence of speckle noise. For speckle noise removal adaptive weighted median filter is used. Tracking results obtained by using our proposed UKF based method are compared with results obtained by using Kalman filter (KF) based tracker. The root mean square error (RMSE) between actual and tracked boundary of the heart wall is calculated for each pixel using both these methods. It is observed that RMSE produced by UKF based tracker is less than that produced by KF based tracker