Joint segmentation of right and left cardiac ventricles using multi-label graph cut

Abstract

Segmenting the left ventricle (LV) and the right ventricle (RV) in magnetic resonance (MR) images is required for cardiac function assessment. In particular, the segmentation of the RV is a difficult task due to low contrast with surrounding tissues and high shape variability. To overcome these problems, we introduce a fully automatic segmentation method based on multi-label graph cuts, that makes use of a probabilistic shape model. The shape model is obtained by merging several atlases after their non-rigid registration on the unseen image. This prior is then incorporated into the multi-label graph cut framework in order to guide the segmentation. Our automatic segmentation method has been applied on 754 MR images. We show that encouraging results can be obtained for this challenging application.

Multilabel statistical shape prior for image segmentation

“…In the following, we will apply the proposed multilabel shape prior segmentation method to cardiac MR images with four labels (LV, RV, myocardium, background). A preliminary version of this work has appeared in [26].…”