Myocardium segmentation using a priori knowledge of shape and a spatial relation

摘要

We propose a new method to segment the myocardium based on Active Shape Model - ASM and a spatial distance relation. The main idea is to take advantage from statistical a priori knowledge of the shape that exists in ASM, in order to model the shape of both structures to be segmented, and integrate a new a priori knowledge about the variation of a spatial distance relation between them. This knowledge is estimated during a training step, then, the obtained models are used to guide the segmentation process. The proposed method is applied to endocardial and epicardial contours segmentation in axial CT scan slices of the heart. The obtained results are encouraging and show the performance of the proposed method.