Main Article Content
The term "Alzheimer's Disease (AD)" relates to a brain disorder had been affecting a large number of individuals throughout the world annually. It may be dangerous to the individual if left undiagnosed or untreated, also this disorder has no successful therapy. There is a lot of interest in developing novel ways to detect AD more often. In the classification of AD using brain "Magnetic Resonance Imaging (MRI)" images, the "K-Nearest Neighbor (KNN)" classification was among the most successful technique. This algorithm compares the similarity between the training data and new instances before classifying them. However, its accuracy gets lacking when dealing with a complex dataset. To enhance its accuracy level in large datasets in this research we had enhanced the traditional KNN with FuzzyLogic by proposing "Enhanced Fuzzy-KNN (EFKNN)". A fuzzy degree of membership in the problem classes was calculated using the EFKNN. As a consequence, the boundaries between classes are smoother. The conventional KNN technique is unable to deal with large datasets because it lacks a fuzzy variation. The class-membership calculations, however, entail an additional computational burden, making them ineffective for dealing with huge datasets due to large storage requirements and higher running time. The primary goal of this work is to use "structural-MRI (sMRI)" images for obtaining the hippocampus volume area to automatically learn and categorize Alzheimer's disease. Here the proposed model consists of various stages namely "Pre-Processing", "Segmentation", "Feature Extraction", and "Classification". The EFKNN has been shown that it has not only a lower error in the classification of subjects but also more faith in the classification taking advantage of the FuzzyLogic principle. In this research, an EFKNN classifier is implemented for evaluating the subject of sMRI brain images as "Cognitive Normal (CN)", "Mild Cognitive Impairment (MCI)" or "Pure Alzheimer's Disease (AD)" classes during classification. The proposed EFKNN approach proves its efficiency in terms of detection and classification with its accuracy more than the existing KNN approach.