Integrating Stable diffusion for Improved Alzheimer's Disease Classification: Insights from MRI Images and YOLOv8

Alzheimer's disease (AD) is the most common form of dementia, characterized by progressive neurodegeneration. Structural changes in the brain associated with AD can be visualized using magnetic resonance imaging (MRI). However, acquiring a sufficient number of MRI images from patients with Alzheimer's disease is challenging. In this study, we employ stable diffusion to simulate MRI scans of patients with AD. We train the stable diffusion model from scratch to generate new MRI images as part of data augmentation alongside the original images. The model is capable of producing synthetic MRI images of AD. We employ t-distributed Stochastic Neighbor Embedding (t-SNE) visualization to evaluate the quality and diversity of the generated images by comparing their distribution to that of actual MRI scans. Subsequently, a modified YOLOv8-cls model is retrained using the generated images for AD classification. Our findings demonstrate that this approach effectively produces realistic synthetic biological images suitable for training deep learning models with 10% improvement in accuracy.