A robust region based level set framework for medical image segmentation

Abstract

In this paper, we present new developments in the formulation of a new class of level set method for medical image segmentation. In this work, a new speed function of level set framework is proposed. The region statistical information, instead of the conventional image gradient information, is fused into the level set fundamental model to improve the robustness of the segmentation for medical images. The new method has some advantages over classical level set methods especially in the situations where edges are weak and fuzzy. A number of experiments on MR, US and CT images were performed to evaluate the new method. Experimental results are given to illustrate the effectiveness and robustness of the method. © Springer-Verlag Berlin Heidelberg 2006.