Local region based medical image segmentation using J-divergence measures

In this paper, we propose a novel variational formulation. The originality of our formulation is on the use of J-divergence (symmetrized Kullback-Leibler divergence) for the dissimilarity measure between local and global regions. The intensity of a local region is assumed to follow Gaussian distribution. Thus, two features - mean and variance of the distribution of every voxel are introduced to ensure the robustness of the algorithm when noise appeared. Then, J-divergence is used to measure the "distance" between two distributions. The proposed method is verified on synthetic and real medical images. The experimental results are very encouraging for medical image segmentation.