Deep-Learning-Based Classification and Diagnosis of Alzheimer's Disease

Deep-Learning-Based Classification and Diagnosis of Alzheimer's Disease

Rekh Ram Janghel (NIT Raipur, India)
Copyright: © 2018 |Pages: 25
DOI: 10.4018/978-1-5225-5775-3.ch011

Abstract

Alzheimer's is the most common form of dementia in India and it is one of the leading causes of death in the world. Currently it is diagnosed by calculating the MSME score and by manual study of MRI scan. In this chapter, the authors develop and compare different methods to diagnose and predict Alzheimer's disease by processing structural magnetic resonance image scans (MRI scans) with deep learning neural networks. The authors implement one model of deep-learning networks which are convolution neural network (CNN). They use four different architectures of CNN, namely Lenet-5, AlexNet, ZFNet, and R-CNN architecture. The best accuracies for 75-25 cross validation and 90-10 cross validation are 97.68% and 98.75%, respectively, and achieved by ZFNet architecture of convolution neural network. This research will help in further studies on improving the accuracy of Alzheimer's diagnosis and prediction using neural networks.
Chapter Preview
Top

Introduction

Convolution Neural Network (CNN) is a deep learning algorithm which helps in classification to extract low to high-level features. In this paper, various different architectures of Convolution Neural Network have been used to classify Alzheimer’s Disease. This kind of medical data is classified to potentially develop a model which can predict or system that can recognize the type disease from normal subjects or to estimate the stage of the disease. Classification of Alzheimer's disease has always been a challenging task and most difficult task has been to select the most different features. CNN helps to extract low to high level features automatically by learning features

Complete Chapter List

Search this Book:
Reset