Intelligent Computing for the Analysis of Brain Magnetic Resonance Images

A technique for intelligent processing is proposed for the analysis of brain magnetic resonance images. This paper presents segmentation and detection technique of tumor, edema and healthy tissues from fluid attenuated inversion recovery magnetic resonance images of brain with the help of composite feature vectors comprising of empirically developed functions of higher order wavelets and statistical parameters. The main contributions of paper is that clustering of feature vectors, corresponding to blocks of size 4×4pixels from intra cranial brain image in three useful clusters of tumor, edema and healthy tissues is done with fuzzy C-mean algorithm. From clustered feature vectors, segmented image is obtained, pseudocoloring is done and then analysis of brain for pathological tissues and healthy tissues are done. Training of artificial neural network using fuzzy back propagation algorithm is done for detection of tumor and edema. The performance error is coming as low as 7.8485e-015.