Segmentation of Brain Subjects for the classification of Alzheimer’s disease in MR Images using hybrid Classifier

Abstract

Alzheimer’s disease (AD) is generally detected from the structural variation in brain subjects. Grey Matter (GM) decrease and reduction in hippocampus are the essential estimation parameters for classifying the nature of disease. Earlier detection of AD is very helpful to the physicians in the diagnostic,which is possible with volumetric measure of brain subjects. Magnetic Resonance imaging (MRI) is preferable imaging technique among various modalities, because of its better visualization and higher resolution. Segmentation plays a crucial task inmedical application to identify different stages of disease. Four different labels are assigned to cluster various brain tissue category depending upon the similarity of pixels. Intuitionistic Fuzzy algorithm is employed to segment GM, White Matter (WM), Hippocampus region and the cerebrospinal Fluid (CSF) regions. Essential features are selected from the pre and post segmented brain image using Grey Level Co-occurrence Matrix (GLCM). The severity of the disease has been classified using the chosen features. Stable and progressive Mild Cognitive Impairment (MCI) as well as the AD subjects are classified using hybridSupport Vector Machines (SVM) and naïve bayesclassifier. The results of our proposed approach is analyzed with previous works and the performance of classification approach is 95.2%.