Advancing Humanitarian Efforts in Alzheimer's Diagnosis Using AI and MRI Technology

Abstract

Alzheimer’s disease (AD) is a serious neurodegenerative condition that damages brain cells and causes dementia, or the permanent loss of memory. Since there is no cure for this illness, AD causes loss of precious human lives every year. The age group of 65 and above has the highest prevalence of AD. Early detection and diagnosis of this disease using artificial intelligence (AI) can be of high clinical value, medical, and humanitarian significance. Many of these types of medical conditions can be identified and treated using deep learning (DL) methods like convolutional neural networks (CNN). This work proposes a new CNN-based approach to identify the type of AD from the brain magnetic resonance imaging (MRI) images. Experiments were performed on MRI images dataset to classify into very mild, mild, moderate, and severe AD. The pre-processing of data between the unbalanced categories was using SMOTE and other standard image augmentation techniques. The proposed work achieves a testing accuracy of 95.16% and training accuracy of 99.68% and outperforms the Inception-v3 model. The observed results were examined and compared with standard image recognition models like AlexNet, GoogLeNet, VGG-16, and ResNet-18, which report that GoogLeNet model provided the optimum outcomes with a training accuracy of 99.84% and a testing accuracy of 98.25%.

Ähnliche Arbeiten