Exploiting Macro- And Micro-structural Brain Changes For Improved Parkinson’s Disease Classification From Mri Data

This article discusses the development of an accurate and explainable deep learning model for the image-guided diagnosis support of Parkinson’s disease using high-resolution, multimodal MRI data. The model was trained and tested on a large database and outperformed traditional machine learning approaches. Saliency map explanations were used to identify the most important regions for the decision task.