MR Image segmentation based on a new hybrid level set evolution

In this paper, a new hybrid model for active contour image segmentation is proposed. The model is a combination of an edge and region based active contour. To make more efficient noisy images segmentation, the proposed method is separated into two stages. The first one is a pre-processing step consisting of a morphological contrast enhancement followed by a de-noising process using an anisotropic diffusion filter. In the second stage, segmentation is performed using a level set based on a hybrid energy minimization. Various experimental results on medical and synthetic images are presented. Segmentation tests show that the proposed method is efficient, accurate, fast and robust.