Development of an explainable AI model for early classification of MCI and Alzheimer’s disease using T1 MRI scans

This research project focuses on the development of an explainable artificial intelligence (XAI) model for the classification of patients with Mild Cognitive Impairment (MCI) and Alzheimer’s Disease (AD). The goal is to combine the accuracy of convolutional neural networks (CNN) with the transparency and interpretability that are crucial in clinical settings. The data used come from the ADNI (Alzheimer Disease Neuroimaging Initiative) database, with a particular focus on T1 MRI MPRAGE scans, extracted using the FastSurfer tool. A CNN was trained on these images for disease classification, while an explainability technique based on Grad-CAM was integrated to highlight the brain areas relevant to classification, enhancing model interpretability. The project is funded by the AD-RICAMI program, Spoke 2, PNRR, within the RAISE program, aiming to provide a diagnostic tool that is not only accurate but also understandable and usable by healthcare professionals. Initial results show promising clinical applications for the early diagnosis of neurodegenerative diseases.