Contour segmentation of echocardiographic images

Segmentation of echocardiographic image has important application in medical imaging. Echocardiography is widely used for the diagnosis of heart diseases. However, ultrasound image segmentation for boundary detection of the target object is a very difficult task because of its inherent drawbacks, because of the uncertainty in segmentation of boundary caused by speckle noise, relatively low SNR, and low contrast. We propose and demonstrate two powerful, accurate segmentation methods with high stability and sensitivity.The first method uses closed curve for extracting the boundary. However, the conventional segmentation technique with the arc length penalty term often fails to reach the desired segmentation, because it has inherent difficulty in controlling the fitting parameter to encourage smoothness of curves and preventing separation.The second method uses clustering mask based on the Gaussian mixture model (GMM). GMM clustering allows independent classification of pixels into their own layers. This allows us to reduce the influence of speckle noise and degradation, making our approach more robust.