Dementia Detection from MRI

This paper introduces a novel DCGAN-based Augmentation and Classification (D-BAC) model to detect and classify dementia, including early onset (Mild Cognitive Impairment), using MRI scans. The model achieved a prediction accuracy of 74% for MCI and can classify dementia into four categories based on MRI prominence. The proposed approach, using a GAN-augmented dataset and various CNN architectures, demonstrated high accuracy, with a training accuracy of 97% and testing accuracies of up to 87% using VGG-19.