Detection of Alzheimer’s Disease using Structural Brain Network and Convolutional Neural Network

Alzheimer’s disease (AD) is an incurable and progressive disease, and for delaying or preventing disease progression the early detection of Alzheimer’s disease is important. Structural brain networks from Diffusion MR Imaging convey a huge amount of information that has not been fully explored previously and become a major diagnostic tool of AD. Recently, significant attention has been paid to employ deep learning methods for the automatic identification of AD with the various neuroimaging modalities and types of extracted features. An efficient convolutional neural network (CNN) architecture is presented here to obtain high-quality features from structural brain networks to classify individuals into normal control (NC) subjects and AD patients. To the best of our knowledge, no study has been conducted to apply the structural brain networks with deep learning to identify AD. The proposed approach is examined on the Alzheimer’s Disease Neuroimaging Initiative (ADNI) project dataset with 60 NC subjects and 62 AD patients. The proposed method efficiently shows discriminative brain network characteristics and offers an accurate Alzheimer’s detection classifier. The proposed CNN based approach achieves 95% prediction accuracy, which means that our prediction model is more robust and accurate compared to the previous approaches. Our work explores the benefits of deep learning by classifying high-dimensional clinical data and could assist in the early prediction and prevention of Alzheimer’s disease.