Computer techniques play important role in medical fields, especially in the classification and disease diagnoses. In this study, Brain MRI is classified using many techniques. Back propagation neural network, K-nearest and K-means used for classifying these images into normal and abnormal after clustering images using fuzzy C-means and minimizing the extracted features by GLCM using genetic algorithm. The system shows high efficiency through practical experiments that proves that the accuracy of system is reached to 100% through back propagation neural network.