Brain Tumor Identification using FCM Threshold Method and Morphological Area Selection

Abstract

Brain tumors segmentation has become a popular research topic in the last five years, proved by the emergence of many methods proposed to segment brain tumors accurately. In this study, the authors propose a brain tumor segmentation method based on the FCM method with a modification of the threshold value, which will later be used to convert an MRI image to a binary image with only the tumor area detected. The segmentation process divided into three stages, with steps is preprocessing segmentation and post-processing. In the preprocessing stage, the skull bones from MRI images are removed, then the noise is removed using Wiener filters, then proceed with the segmentation stage using FCM Thresh, and finally applying morphological area selection to select areas from segmentation results. From a total of 100 positive tumor MRI images that we acquire from the BRATS 2015 dataset, we obtained an average similarity of 0.7592. We achieved an improvement of 0.06 in term of SSIM value from the previous method.