The paper proposes a computer aided detection method to diagnose brain tumor in its early stage the use of Mathematical Morphological Reconstruction (MMR). Image is pre-processed to cast off noise and artefacts after which segmented to find areas of interest with probably tumor. A big number of textural and statistical functions are extracted from the segmented photo to classify whether the mind tumor within the photo is benign or malignant. Sobel facet detection approach is used to find out the brain shape and tumor form is then discovering the usage of shade threshold in grey scale photograph. Experimental results show that the segmented photos have a excessive accuracy at the same time as substantially decreasing the computation time. The look at shows that the proposed solution can be used to diagnose mind tumor in patients with a excessive success rate.