Level Set Segmentation Using a Point-Based Statistical Shape Model Relying on Correspondence Probabilities
Abstract
In order to successfully perform automatic segmentation in medical images containing noise and intensity inhomogeneities, modern algorithms often rely on a priori knowledge about the respective anatomy. This is often introduced by statistical shape models (SSMs) which are typically based on one-to-one point correspondences. In this work, we propose a unified statistical framework for image segmentation with shape prior information. The shape prior is an explicitely represented probabilistic SSM based on point correspondence probabilities, and the segmentation contour is implicitly represented by the zero level set of a higher dimensional surface. These two aspects are unified in a Maximum a Posteriori (MAP) estimation where the level set is evolved to converge towards the boundary of the organ to be segmented based on the image information while taking into account the prior given by the SSM information. The optimization of the MAP formulation leads to an alternate update of the level set and an update of the fitting of the SSM. We demonstrate the efficiency of our new algorithm with soft tissue segmentation where adaptive weights ensure that the SSM constraint is optimally exploited. Our experimental results show the well-posedness of the approach on noisy kidney CT data impaired by breathing artefacts.
Domains
Computer Science [cs]
Origin : Files produced by the author(s)