Examining the Potential of Deep Learning in the Early Diagnosis of Alzheimer's Disease using Brain MRI Images

Abstract

Alzheimer's disease is a severe public health problem affecting millions worldwide. Deep Learning (DL) models can aid in detecting the disease using MRI data, and we evaluated three DL models for this purpose. We used detailed MRI images of Alzheimer's patients and healthy controls to train these models. A convolutional neural network (CNN) with two convolutional and two fully connected layers was employed in the initial model, which had a 95% accuracy rate. The second model, which included a leaky ReLU activation function, more fully connected layers, and a bigger kernel size, was an enhanced version of the previous one and had a 96% accuracy rate.  The third model was a transfer learning model with two dense layers built on top of the VGG16 architecture, achieving an accuracy of 80%. Our findings imply how neural network models may assist with MRI data-based the disease assessment via evaluations of reliability, precision, recollection, and the F1 ranking. For enhancing the precision and usability of these gadgets for therapeutic usage, more study must be conducted.