Multimodal Approach for Alzheimer’s Disease Diagnosis using MRI and Clinical Data

Abstract

significant advancements in diagnostics and patient care. Specifically, the early and precise detection of Alzheimer’s disease (AD) is crucial for effective intervention and management. This thesis aims to develop an accurate, multimodal diagnostic system for AD that integrates Magnetic Resonance Imaging (MRI) and clinical data to enhance clinical decision-making. In this thesis, we propose a multimodal approach for AD diagnosis using MRI and clinical data. In this approach, for MRI analysis, we employed an ensemble deep learning model based on the EfficientNetB3 and InceptionV3 architectures, achieving an accuracy of 95.06%. Simultaneously, clinical data were evaluated using a hyperparameter-tuned XGBoost model, reaching an accuracy of 86.07%. By integrating the predictions from both models, our proposed multimodal system achieved an improved overall accuracy of 96.35%. To enhance the interpretability and trustworthiness of our proposed approach, Explainable AI (XAI) techniques were employed to provide insights into model decision-making processes.