Class imbalance learning–driven Alzheimer’s detection using hybrid features

Alzheimer’s is the main reason which leads to memory loss of a human being. The living style of an affected person also varies. This variation in the lifestyle creates difficulties for a person to spend a normal life. The detection of Alzheimer’s disease in the initial stage accurately has a great significance in the early treatment of the disease. Therefore, it is considered as a challenging task. An efficient method is proposed to detect the Alzheimer’s disease in the early stage using magnetic resonance imaging. k-Means clustering is used to develop the proposed method for efficient segmentation of the white matter, cerebrospinal fluid, and grey matter. The amalgam feature vectors are formed using the grey-level co-occurrence matrix and speeded-up robust features based on textural feature extraction. Statistical and histogram of gradients is used from shape-based features. Fisher linear discriminant analysis is used for dimensionality reduction and the resampling method is used to handle the class imbalance problem. The algorithm is evaluated using the three classifiers k-nearest neighbour, support vector machine and random forest. The dataset of OASIS is used to assess the outcomes. The proposed approach achieved the optimum accuracy of 92.7%.