Brain MR images that are quickly and accurately detected are crucial for medical analysis. Tumor categorization is a topic that has been covered extensively in the literature. In this study, we offer a technique for automatically segmenting MR brain images to identify and characterize any abnormality segments (such as those caused by tumor) present in the picture. The initial phase in the process of extracting features from an input picture is proposed in this study, and it involves using a DWTdiscrete wavelet transform. To do this, principal component analysis (PCA) is used on the feature picture to minimize the number of dimensions. Kernel support vector machine (KSVM) is used to process the down sampled image of retrieved features. There are a total of 90 MR pictures of the brain in this data collection, representing seven different prevalent illnesses. For the KSVM procedure, these pictures are necessary. The suggested classification approach employs a Gaussian Radial Basis (GRB) kernel, which, in comparison to a linear kernel, achieves a maximum accuracy of 98%. (LIN). The results of the study demonstrated that the GRB kernel technique was superior to the previously used approaches. When an aberrant MR picture containing a tumors is detected using this classification, the appropriate portion is extracted and segmented using a thresholding method.