Alzheimer's disease detection using convolutional neural networks and transfer learning based methods

Alzheimer's disease (AD) remains a major public health problem. This neurodegenerative pathology affects generally old people. Its symptoms are loss of memory followed over the years by more hard ability of expression and various handicaps. Therefore, early detection of AD is become an active research area in recent years. In this paper, we propose a deep based method for the detection of AD (i.e. classify brain images into normal brain or brain with AD). The proposed method contains two main steps. The first step is region of interest extraction; it is based on the partition of the image into separate blocks to extract only the part that contains the hippocampus of the brain. The second step is the classification of images using two deep based techniques namely convolutional neural network (CNN) and Transfer Learning. In one hand, CNN allows extracting the characteristics from brain images, then classifies them into normal brain or AD brain. Transfer Learning, in the other hand, consists of using features acquired from the Alexnet architecture to classify the images. We have assessed the proposed method on Oasis dataset (Open Access Series of Imaging Studies). The obtained results show that the classification of images using Transfer Learning with 92.86% outperformed the CNN's classification rate.