Predicting Parkinson's Disease Using MRI Scans and Spiral Images: A Machine Learning Approach

Parkinson's disease (PD) is a progressive neurodegenerative disorder that requires early and accurate diagnosis to facilitate effective treatment and clinical management.Conven-tional clinical evaluation is generally non-specific and does not identify the disease at the onset stage.This study proposes a new method of PD prediction based on multimodal data-MRI scans and spiral drawingsand using cutting-edge deep learning methods.We used CNNs and LSTM networks for sequence pattern recognition and feature extraction, and Vision Transformers (ViTs) for high-level spatial analysis.We also investigated multimodal fusion models to fuse information from different data sources.The experiments showed that multimodal models are superior to single-modality counterparts and provide increased accuracy and robustness.This research highlights the promise of combining cutting-edge imaging technologies with deep learning to make early, non-invasive, low-cost, and scalable diagnosis of Parkinson's disease possible, leading to AI-powered healthcare innovations.