Fuzzy c-means technique with histogram based centroid initialization for brain tissue segmentation in MRI of head scans

Segmentation plays an important role in biomedical image processing. It is often the starting point for other processes like analysis, visualization and quantization. In brain diagnostic system, segmentation is essential to study many brain disorders. Several popular clustering techniques for segmentation are available. Fuzzy c-means (FCM) is one such soft segmentation technique applicable for MRI brain tissue segmentation. The performance of this method to obtain an optimal solution depends on the initial positions of the centroids of the clusters. In the existing FCM, the centroids are initialized randomly. This leads to increase in time to reach the optimal solution. In order to accelerate the segmentation process an application specific knowledge is used to initialize the centers of required clusters. To segment brain portion, we use the knowledge about the MRI intensity characteristics of brain regions to initialize the centroids. The performance of existing FCM and the proposed approach with centroid initialization is evaluated by applying the methods on several datasets. The comparison is done in terms of processing time and the values obtained as final centroids. The proposed approach produced the optimal results within 14–18 iterations in 2.5–11 sec/slices while the existing FCM took 3.5–15 sec/slice. The results indicate that the knowledge about the datasets to be clustered can be used effectively to initialize the centroids for FCM algorithm. The results reveal that the proposed method with 14 iterations is sufficient to segment the normal brain volumes.