Alzheimer’s Disease Diagnosis Using CNN Based Pre-trained Models

Abstract

Memory loss and impairment are signs of Alzheimer's disease (AD), which may also cause other issues. It has a significant impact on patients' lives and is incurable, but rapid recognition of Alzheimer's disease can be useful to initiate appropriate therapy to avoid further deterioration to the brain. Previously, Machine Learning methodswere used to detect Alzheimer's disease. In recent times, Deep Learning algorithms have become more popular for pattern recognition. This workconcentrates on the recognition of Alzheimer's disease at a preliminary phase using advanced convolutional neural network models. As the disease advances, they steadily forget everything. It is critical to detect the disease as quickly as possible. The proposed model usespre-trained models that uses magnetic resonance imaging of the brain to determine if a person has very mild, mild, moderate, or non-dementia. The models used for classification are VGG16, VGG19, and ResNet50 architectures and provide performance comparison.