A Local Likelihood-based Level Set Segmentation Method for Brain MR Images

Abstract

Automatic segmentation of brain structures in MR images is of great importance in studying neurodegenerative disease quantitatively. In this paper, we propose a fully automatic solution to this problem through the utilization of a probabilistic atlas built from a set of training MR images. As a generalization of the Chan-Vese piecewise-constant model, our model uses Bayesian a posterior probabilities as the driving forces for curve evolution. Prior intensity distribution can be seamlessly integrated into the level set evolution procedure. Unlike other region-based active contour models, our solution relaxes the global piecewise constant assumption, and uses locally varying Gaussians to better account for intensity inhomogeneity and local variations existing in many brain MR images. Our model shows particular strength in distinguishing low contrast structures from surrounding tissues. Accurate and robust segmentation is therefore achieved. Experiments conducted on synthetic and real brain MRIs demonstrate the improvement made by our model.