Modified FCM using genetic algorithm for segmentation of MRI brain images
S. Jansi, P. Subashini · 2014
Magnetic Resonance Imaging (MRI) is generally a medical imaging technique nearly everyone used in radiology to visualize the structure and role of the body. MRI gives complete images of the body in several planes. Skull Stripping is a major preprocessing phase and an essential part in neuro-imaging applications it refers to the removal of non-cerebral tissues. Various methods have been worked for medical image segmentation such as clustering based methods, thresholding based methods, region based methods, classifiers, deformable model, markov random model etc. This paper mainly concentrates on clustering methods, especially K-Means, Fuzzy C-Means clustering algorithm for segmentation of Gray Matter, White Matter and Cerebrospinal Fluid tissues in MRI brain images. FCM is more effective to the fuzzy boundary region segment, but the biggest disadvantage is that there is no better way to find the centroid clustering value. So it will converge to the local minimum point easily. To overcome this limitation, a Genetic Algorithm is integrated along with Fuzzy Clustering Method for determining the global centroid value. Experimental outcome shows Genetic Algorithm based FCM segmentation gives better performance compared with existing methods by using evaluation metrics such as Under Segmentation, Over Segmentation and Incorrect Segmentation.