Multi-channeled MR brain image segmentation: A new automated approach combining BAT and clustering technique for better identification of heterogeneous tumors

Segregation of tumor region in brain MR image is a prominent task that instantly provides easier tumor diagnosis, which leads to effective radiotherapy planning. For decades together, several segmentation methods for a brain tumor have been presented and until now, enhanced tumor segmentation procedure tends to be a challenging task because, MR images are mostly inbred with varied tumor dimensions of disproportioned boundaries. To address this issue, we develop an improved brain image segmentation technique called BAT based Interval Type-2 Fuzzy C-Means (BAT-IT2FCM) clustering. The BAT algorithm is utilized to find out the optimal cluster location from which the clustering operation by Interval Type-2 Fuzzy C-Means (IT2FCM) is performed. The optimal cluster location pointed/identified by the BAT algorithm helps in easing the clustering operation performed by IT2FCM algorithm, and thereby reducing computational complexity. The efficient outcome from BAT-IT2FCM methodology was affirmed using the performance metrics such as computational time, Peak Signal to Noise Ratio, Mean Squared Error, Jaccard Tanimoto Co-efficient Index and Dice Overlap Index. Also, segmentation results of clinical brain MR images produced by the proposed methodology were evaluated with the support from radiologists (Gold Standard). The suggested BAT based fuzzy related clustering produces sensitivity and specificity values of 98.56±1.2 and 97.67±1.3, respectively, which are better than the existing techniques used for brain image segmentation. Heterogeneous tumor types of different grade levels and tissue structures present in the brain MR slices of three different axes are precisely segmented by the proposed methodology for better visualization of oncologists.