Unveiling Alzheimer's Stages Through Transfer Learning and Custom CNN Layers

A neuro degenerative condition called Alzheimer's is a complex disease which involves the progressive decline of cognitive abilities. Alzheimer's disease gradually destroys brain cells, disrupting the intricate networks that enable thinking, memory, and behavior. The various stages of Alzheimer's disease are categorized as Non-Dementia, Very Mild Dementia, Mild Dementia, and Moderate Dementia. A Pre-trained model and a 10-layer CNN model were used for the classification of various stages of Alzheimer's using brain MRI scans. The system utilizes a dataset of MRI images including Mild Demented, Moderate Demented, NonDemented, and Very Mild Demented classes. Pretrained transfer learning models are compared to achieve accurate classification, including Efficient Net B0, Inception V3, ResNet 50, MobileNet V3, VGG16, and DenseNet. The Convolutional Neural Network (CNN) Model with 10 different layers is designed to classify the different stages of the disease accurately. The results show that the segregation of images using pre-trained ResNet50 Model without preweights and additional layers is 97% and this outperforms the CNN's classification rate.