Here we took a set of images,applied CLAHE to them to remove the noise. Then after we applied Otsu Thresholding to segment gray and white matter. We calculated the GLCM parameters of the segmented images and applied them to train the SVM. We tested the rest of the images and we found better results. Through a machine learning method, we hope to achieve better precision and accuracy in interpreting a normal and abnormal brain mri image.So we can detect the brain tumors,diseases and all other defects.