A New Fuzzy Clustering Algorithm Using Uncertainty-based Entropy for Brain MR Image Segmentation

In this paper, we propose a new fuzzy clustering algorithm using uncertainty-based entropy for brain MR image segmentation. To overcome this problem, we introduce an uncertainty measure, which defines a degree of improbable of a pixel for becoming a member of a cluster. In addition, we use this uncertainty measure to calculate the entropy for a particular pixel. The pixels in the tissue boundaries have higher entropy as the region is blurred due to noise and intensity inhomogeneity. Finally, a similarity measure, which is characterized by a Gaussian density function, is integrated both with this uncertainty measure and the fuzzy membership function. We suitable formulate the objective function of the proposed algorithm by integrating the above parameters; thereby addressing the limitations of the standard fuzzy c-means (FCM) clustering algorithm. The simulation results of the proposed algorithm suggest that it is suitable for segmentation of brain MR images, especially in the presence of high percentage of noise and intensity inhomogeneity and even superior to some of the state-of-the art methods.