Automatically detection of multi-class Alzheimer disease using Deep Siamese Convolutional Neural Network

Abstract

Alzheimer's disease (AD) is a gradual, lifelong dementia that typically affects elderly adults. Alzheimer's disease affects memory, listening, and other cognitive skills. Early Alzheimer's diagnosis is difficult for clinicians. Machine learning and deep convolution neural network (CNN) based techniques can handle brain imaging data processing difficulties. Clinical studies have employed MRI to detect Alzheimer's. In the proposed work we are using a deep Siamese-based neural network to automatically diagnose Alzheimer's disease from a Brain MRI images. Each MRI image of the brain is separated into two segments, which are sent into a network that compares their symmetric structure and infection levels. We are using the Kaggle dataset to train and evaluate for Alzheimer's model. This algorithm could help doctors to identify Alzheimer's from MRI images. The model exceeds the state-of-the-art in every output metric, indicating reduced bias and better generalization.