Identification of Alzheimer's disease using Convolutional Neural Network

Abstract

Alzheimer's disease (AD) as a brain disease has caused a progressive, devastating effect on the memory and general mental and physical coordination of victims. The impact on victims is irreversible, and the cause has yet to be identified. The treatment at full-blown can be difficult, but it could be properly managed in the early phase. Hence, there is a need for an efficient and effective early diagnosis. Machine learning techniques have proved to be successful in image classification. It was on this premise that this paper adopted a machine learning approach. The approach used a convolutional neural network with transfer learning to classify structural Magnetic Resonance Images (sMRI) into a multi-classification of 3 classes. The classes were Normal Cognitive (NC), Mild Cognitive Impairment (MCI) and Alzheimer's Disease (AD). K-fold cross validation was employed to validate the test set. The sMRI subjects included 97 NC, 57 MCI, and 24 AD patients. The proposed method achieved an overall accuracy of 94% on classification based on the multiclass classification.