Early Detection of Alzheimer’s Disease Using Multi-feature Fusion and an Ensemble of Classifiers

Abstract

Listen

Translate article

Computerized detection of Alzheimer’s disease (AD) can assist medical practitioners to a greater extent in the early diagnosis of the onset of Dementia in elderly people. Structural Magnetic Resonance Imaging (MRI) of the human brain is a key modality that aids in the early diagnosis of the disease. We have come up with an automatic identification of the AD examining the structural MRI using machine learning. The proposed approach aids to identify persons with Alzheimer’s disease using a multi-feature fusion approach. Multi-feature fusion is performed by Support Vector Machine using Feature elimination in a recursive manner where an optimal subset of features is obtained. An ensemble of classifiers with Multi-Layer Perceptron (MLP), Support Vector Machine (SVM) and J48 are used. The performance of the individual classifiers and the ensemble of classifiers with combined features are measured. A better performance is achieved through the multi-feature approach with ensemble of classifiers. The CAD system provided a maximum accuracy of 93.8%.