Severity Classification of Alzheimer Dementia Based on MRI Images Using Deep Neural Network

Abstract

Alzheimer's dementia (AD) is the most common type of dementia, usually characterized by memory loss followed by progressive cognitive decline and functional impairment. AD is one of the leading causes of death and cannot be cured, but proper medical treatment can delay the severity of the disease. Early detection of AD can detect early and prevent the disease from getting worse. So, we need a system that can detect AD as a means of support for the clinical diagnosis. In this study, a system was designed to classify the severity of AD using the Convolutional Neural Network (CNN) method with VGG-16 and VGG-19 modeling. From the simulation results with a total of 4,160 MRI datasets, the highest accuracy rate was 98.28% with VGG-19 architecture using Adam's Optimizer for the classification of 3 classes, namely no dementia (normal), mild dementia, and moderate dementia. It is hoped that this study can support clinical diagnosis in assessing the severity of AD.