Left ventricle motion classification in the medial surface shape space

The wall thickness is known as a valuable measure for the cardiac diagnosis. From the geometric point of view, it can be considered as a function defined on the 2D manifold of the medial surface. This paper presents a novel classification method based on medial representation to diagnose and detect the myopathic regions on the left ventricle. A shape space is proposed and constructed based on the changes of the left ventricle wall thickness, in which two shape descriptors are introduced which show remarkable performance to distinguish normal and abnormal left ventricle deformations. The experimental results show that this method can automatically classify the healthy and myopathic subjects and detect myopathic regions on the left ventricle well.