Skip to Main Content
Level-sets methods have successfully been used for segmentation of the endocardial border in cardiac US images. Robust methods have been proposed within a Bayesian framework. However, segmentation of the whole myocardial wall is still challenging given that the epicardial boundary is highly heterogeneous and discontinuous. The presence of papillary muscles may also complicate endocardial boundary detection. We propose a novel level-set technique for combined endo- and epicardial boundaries segmentation. Hereto a localizing approach is used to overcome limitations associated with global techniques. A comparison with such a global level-set approach presented in is made.