Improvement of Automated Identification of the Heart wall by Reducing Stationary Clutter in Ultrasonic Echoes

Currently, both cardiac wall motion and strain rate can be quantatively evaluated by means of ultrasound Doppler shift or speckle motion. In most of these methods, the heart wall, which is the object to be analyzed, is manually identified by an operator. However, this task is very time-consuming and suffers from interand intraobserver variability. With the aims of facilitation of analysis and elimination of operator dependence, automated identification of the heart wall needs to be realized. We have developed a method for automated identification of the heart wall region throughout an entire cardiac cycle by tracking the points classified in an initial frame. In our method [1], the multiple features, such as echogenicity and temporal phase changes of echo signals, are extracted to accurately identify heart wall in the initial frame. RF echo signals in echocardiography contain the undesirable stationary clutters which are the components of echoes from external tissue such as the ribs [2]. In our previous study [1], to reduce this stationary clutter, high-pass filtering was applied to RF signals before extracting features, corresponding to the moving target indicator (MTI) filtering [3]. However, the magnitude-squared coherence (MSC) function of echoes, which is one of the features, still contained the effect of the stationary clutter. In this study, we improved the performance of the MSC by reducing the influence of the stationary clutter.