Fuzzy C-Means(FCM) is a popular clustering algorithm and has been widely used in fuzzy segmentation of Magnetic Resonance(MR) images.However

the segmented results using the conventional FCM when dealing with noisy MR images are not satisfying because FCM takes no spatial information of images into account.Generally an ideal MR images is assumed to be a piecewise constant.We present an improved model of conventional FCM algorithm using membership smoothing constraint.The proposed algorithm can reasonably use the spatial information of images and improve the accuracy of segmentation.The segmentation of simulated brain MR images with different noise level and real brain MR image are presented in the experiments.The results of experiments show that the proposed algorithm is more powerful than many other segmentation algorithms.