Soft-computing assisted tool to extract tumor section from brain MR images

Brain tumor is one of the chief life threatening diseases in human community. The mainstream of brain malignancy will initiate in the segment of the brain tissue. Mainly the brain malignancy is due to the abnormal lump in the brain tissue called a tumor and it can be detected using procedures, such as Magnetic Resonance (MR) imaging and Computed Tomography (CT). Normally, the MR imaging is widely considered initial imaging procedure due to its multi-modality nature and it is also efficient in providing superior information compared with CT. Premature recognition may reduce the severity of sickness and also diminish the death rate. In this work, heuristic approach supported automated tool is proposed to mine and analyze the tumor part from the Magnetic Resonance Images (MRI) of brain. Proposed tool consist of the following stages; i. Skull stripping, ii. Social Group Optimization (SGO) assisted Kapur's multi-thresholding, iii. Watershed segmentation, and iv. Evaluation of the extracted tumor. The proposed tool is experimentally investigated using various bench mark brain MRI dataset. The results confirm that, proposed tool is efficient in extracting the abnormal region from the considered brain MR test images. The competence of the proposed technique is established by computing the image likeliness measures, such as Jaccard, Dice, sensitivity, specificity and accuracy. The implemented procedure confirms its clinical significance by contributing superior values of image similarity and statistical measure values. In future, it can be considered to examine the real time clinical brain MRI database registered with varied modalities.