Automatic segmentation of the heart muscle from echocardiographic images

This paper presents a fully automated method for endocardial and epicardial contour estimation from sequences of echocardiographic images using a short axis view of the left ventricle. The segmentation is performed by maximum likelihood estimation, separating the Rayleigh modes that model the regions of interest (inside endocardium, between endocardium and epicardium, or outside epicardium). The presented method leads to a fully automatic myocardial segmentation to be further used for strain estimation.