Automatic Cardiac Segmentation Using Triangle and Optical Flow

Abstract

Cardiac function assessment plays an important role in daily cardiology and ultrasound. Full automatic cardiac segmentation is a challenging study because cardiac ultrasound imaging has low contrast and irregular moves. In this research, full automatic cardiac segmentation for cardiac diseases is presented. The technique used Initial Center Boundary, Pre-processing, Triangle Segmentation and Optical Flow. The first step is determining the initial center boundary. The second step is using Pre-processing to eliminate noise. The third step is Triangle Segmentation to detect cardiac boundary and reconstruct the accurate border. The last step is applying Optical Flow method to detect and track the border for every frame in a cardiac video. The performance segmentation for assessment errors cardiac cavity obtained an average triangle 8.18%, snake 19.94% and watershed 15.97%. The experiments showed that triangle method is able to find and improve the segmentation of cardiac cavity images with accurate. The result can be seen that error between system and average of users is only less than 5.6%. This indicates that this method is effective to segment and tracking cardiac cavity in a cardiac video.