A novel segmentation algorithm for MRI brain tumor images

Abstract

A novel segmentation algorithm for MRI Brain tumor images is proposed. The proposed algorithm is compared with Thresholding and Region Grow methods. Testing was performed by generating two datasets of real MRI images of brain tumors. Criteria for assessment of the quality of the segmentation results were: the Dice score, sensitivity, specificity and accuracy. Analysis of results obtained using this algorithm to solve the brain tumor MRI image segmentation task showed levels of sensitivity and specificity of 91% to 99%, which is evidence that assessment of the position and boundaries of brain pathology is highly effective.