Enhancing k-means and kernelized fuzzy c-means clustering with cluster center initialization in segmenting MRI brain images

Clustering is the process of organizing data objects into a set of disjoint classes called clusters. The objective of this paper is to develop an enhanced k-means and kernelized fuzzy c-means for a segmentation of brain magnetic resonance images. Performance of iterative clustering algorithms which converges to numerous local minima depend highly on initial cluster centers. In general the clustering algorithm chooses the initial centers in random manner. In this paper we propose a new center initialization algorithm for measuring the initial centers of the proposed clustering algorithms. This algorithm is based on maximum measure of the distance function which is found for cluster center detection process. More recently clustering is an effective tool in segmenting medical images for further treatment plan. In order to solve the problems of clustering performance affected by initial centers of clusters, this paper introduces a specialised center initialization method for executing the proposed algorithms in segmenting medical images. Experiments are performed with real brain images to access the performance of the proposed methods. Further the validity of clustering results are obtained using silhouette method and compares the results with the results of original k-means and fuzzy c-means clustering algorithms. The experimental results show the superiority of the proposed clustering results.