Computer Aided Methods for the Detection of Brain Tumors from MRI Images

Abstract

Tumors can develop at any brain location with uneven boundaries and shapes. They newlineincreased rapidly, so their size doubled in approximately twenty-five days. If they are newlinenot recognized in earlier phases, patients may suffer from various medical problems, newlineincluding death. Therefore, the identification of brain tumors in the earlier stages is newlineone of the critical aspects. In addition, an effective imaging sequence also plays a vital newlinerole in tumor diagnosis. Magnetic Resonance Imaging (MRI) is widely used among the newlineavailable scanning approaches. Researchers have recently developed numerous meth- newlineods; however, they yield limited accuracy due to the variations in tumor characteristics, newlineimage noise variations, irregular boundary pixels, and intensity non-uniformity in MRI newlineimages. Hence, this thesis suggests three Computer-Aided Diagnosis (CAD) frame- newlineworks for classifying and detecting brain tumors from MRI images. newlineThe first framework introduces a novel methodology based on adaptive methods and newlinelocal texture descriptors. We began with a median filter to improve the brightness of newlinebrain MRI images. Then, we employ Bi-dimensional Empirical Mode Decomposition newline(BEMD) and Modified Quasi-bivariate Variational Mode Decomposition (MQBVMD) newlineto obtain significant subimages. Then, to get relevant features from these sub-images, newlinewe applied an Entropy-based Local Directional Pattern (ELDP) feature descriptor and newlinethen fed it to a Support Vector Machine (SVM) to classify given brain MRI images as newlinelow-grade (LG) or high-grade (HG). Finally, we used adaptive K-means clustering and newlinemorphological operations to identify the infected region of these gliomas. newlineThe second framework introduces a new approach for detecting brain tumors from newlineMRI images based on histogram-based feature descriptors. The Anisotropic Diffusion newlineFilter (ADF) is primarily employed to minimize noise without losing essential details incorporated in brain MRI images. Then, to separate the abnormal region from the enhanced MRI images, Spatial Fuzzy C-Means Thresholding (SFCM