Combinative multi-scale level set framework for echo cardiographic image segmentation

In the automatic segmentation of echocardiographic images, a priori shape knowledge is used to compensate poor features in ultrasound images. The shape knowledge is often learned via off-line training process, which requires tedious human effort and is unavoidably expertise-dependent. More importantly, a learned shape template can only be used to segment a specific class of images with similar boundary shapes. In this paper, we present a multi-scale level set framework for echo image segmentation. We extract echo image boundaries automatically at a very coarse scale. These boundaries are then not only used as boundary initials at finer scales, but also as an external constraint to guide contour evolutions. This constraint functions similar to a traditional shape prior. Experimental results validate this combinative framework.