Medical Image Segmentation Using GA-Based Modified Spatial FCM Clustering

Abstract

This chapter proposes a unique method for unsupervised segmentation of medical images using genetic algorithm (GA)-based spatial fuzzy C-means (SFCM) clustering. The aim of the algorithm is to segment the medical image into an appropriate number of clusters, whereby the required number of clusters is computed automatically. SFCM takes into account the effect of neighborhood pixels on a central pixel, and thus the set of clusters obtained by SFCM forms the basis for a genetic algorithm where different genetic operators are used to further calibrate the centroids. A validity index is used to obtain the optimal number of clusters. The experimental results of the proposed method are compared with existing methods for further validation.