An extension of various available clustering algorithms has been serving as a solution to serve many current problems by the researchers. The Fuzzy C Means (FCM) algorithm that has been in use all these days is extremely noise sensitive. Hence it fails to provide the desired results. This was solved to an extent with the introduction of spatial fuzzy c means. This included a spatial function which was the summation of all the membership values of the neighbors of the pixel considered for study. This paper proposes a new and better modification of the spatial fuzzy c means(sFCM) by introducing kernel distance metric. This groups the objects into clusters which are not separable linearly.Here radial basis kernel function is applied for sFCM clustering. The proposed clustering algorithm is tested on MRI image and noise induced MRI image. The results reveal that kernel based spatial fuzzy c means (sKFCM) is better than Euclidean based spatial fuzzy c means. © 2016, Institute of Integrative Omics and Applied Biotechnology. All rights reserved.