Secondary Segmentation Algorithm for Magnetic Resonance Brain Image Based on Local Entropy Minimization

Medical image segmentation plays a very important role in medical image processing, particularly in the clinical analysis of magnetic resonance (MR) brain images. Intensity inhomogeneity in MR images, which can change the local statistical characteristics of the images, becomes a major obstacle to any automatic method for MR images. In order to reduce the influences of intensity inhomogeneity during segmentation, a secondary segmentation algorithm is presented for MR brain images based on local entropy minimization. By making use of the tissue-based method and local entropy minimization, the clustering blocks of brain image segmentation are gotton, and then the dynamic search is implemented by taking each clustering block as central region. For each dynamic searching-window, the fuzzy C-means algorithm is used to segment the images. Comparing all the segmentation results with them of original clustering block, secondary segmentation is made for the pixels which hold the conditions of secondary segmentation. The segmentation results by using both simulated and real MR images show that the proposed secondary segmentation algorithm is accurate and reliable.