A Novel Fast Fuzzy C-Means Clustering Technique for Segmentation of Human- Brain Magnetic Resonance Images

In medical applications all effectual agents in patient health must be fast, even medical algorithms such as clustering ones. In this paper an optimized technique is presented to decrease execution time and iterations of standard Fuzzy C-Means (FCM) alghorythm. New approach calculates cluster center in each iteration by new formula. Applying proposed method decreases the complexity of FCM algorithm. A type of averaging among cluster centers is applied in each iteration step however, membership function is a fuzzy coefficient. By proposed approach intensity of pixels in a cluster is averaged in each time step. Simulation results show that the proposed Fast FCM (FFCM) spends moderately half time of standard FCM and decreases iteration numbers. Moreover, to decrease the time of convergence considerably and decline the number of iterations significantly, cluster centroids are initialized by an algorithm. FCM and FFCM techniques are applied to segment magnetic resonance (MR) images. Accuracy of the proposed approach is significantly same as standard FCM. Applying fuzzy validity functions to quantitative assessment of FFCM in comparison with FCM verifies efficient performance of the proposed approach.