Among cerebrum tumors, gliomas territory unit the preeminent normal and forceful, bringing about a dreadfully short life in their most noteworthy evaluation. In this manner, treatment structuring might be a key stage to upgrade the nature of life of oncologic patients. Attractive Resonance Imaging (MRI) might be a widely utilized imaging procedure to evaluate these tumors. Anyway, the enormous amount of learning made by attractive reverberation imaging anticipates manual division in an entirely reasonable time, constraining the utilization of exact quantitative estimations inside the clinical apply. Along these lines, programmed and dependable division ways territory unit required; be that as it may, the enormous spatial and basic changeability among cerebrum tumors make programmed division a troublesome downside. In this paper, we will, in general, propose a programmed division technique dependent on Statistical Region Merging. To ensure the tumor segmentation based on the merging predicate of the neighborhood pixels to accurately determine the tumor. So effective results are achieved when compared to manual segmentation where the segmentation lacks accuracy.