Knowledge-Based Segmentation of Medical Images

Abstract

Knowledge-based medical image segmentation provides applicationspecific context by constructing prior models and incorporating them into the segmentation process. In this chapter, we present recent work that integrates intensity, local curvature, and global shape information into level set based segmentation of medical images. The object intensity distribution is modeled as a function of signed distance from the object boundary, which fits naturally into the level set framework. Curvature profiles act as boundary regularization terms specific to the shape being extracted, as opposed to uniformly penalizing high curvature. We describe a representation for deformable shapes and define a probability distribution over the variances of a set of training shapes. The segmentation process embeds an initial curve as the zero level set of a higher dimensional surface, and evolves the surface such that the zero level set converges on the boundary of the object to be segmented. At each step of the surface evolution, we estimate the pose and shape of the object in the image, based on the prior shape information and the image information. We then evolve the surface globally, towards the estimate, and locally, based on image information and curvature. Results are demonstrated on magnetic resonance (MR) and computed tomography (CT) imagery.