Maximal Stable Extremal Region Extraction of MRI Tumor Images Using Successive Otsu Algorithm

Abstract

The computer-aided diagnosis of magnetic resonance imaging (MRI) images using computer vision techniques is a facilitating area of interest. The segmentation model tries to differentiate tumor regions in an MRI section. The MRI is better than computerized tomography (CT) scan images since there is no radiation in the magnetic resonance operation, and furthermore, the outcomes obtained utilizing MR images are more useful and clear when compare with CT scan image. In this paper, we proposed a novel method of successive Otsu algorithm applied it into hierarchical levels to define segmentation threshold for MRI. Maximal stable extremal region (MSER) is more appropriately used for objective detection as well as tracking, as it detects the major MSER from both the depth image and the intensity image. Appling MSER initially, the input image is changed from a pixel-based to a region-based model by utilizing the MSER. The segmentation performance is evaluated by using four stability parameters, namely probability randomness indexing, variation often information, globally consistency errorless, PSNR measures. The purpose of segmentationing method is carried out by utilizing the successive Otsu algorithm. The outcome of segmented MRI images is used to extract the tumor area by introducing the MSER.