The death rate due to tumor has been increasing enormously over the past three decades. This fact increases the importance of research in the medical field, to identify brain pathologies for tumor segmentation and detection, which helps the neurosurgeon to diagnose brain disease and to set up most suitable treatment for the pathology. Normally, manual segmentation of brain image is a tedious and laborious task due to the processing of large amount of data and also due to the presence of minute brain lesions. Thus a completely automated segmentation system has become a real challenging in medical image processing, which has fascinated many researchers in this field, in recent years. The recommended system focus on all possible outcomes, that can be used to address the brain segmentation problems in multi modality MR images, which is widely used imaging technique for its high quality. More precisely, the classifier used in this recommended system is Support Vector Machine (SVM), which is the most prevalent classification method used recently, along with a combination of Wrapper based Genetic Algorithm.