Multimodal Image Fusion and Real-Time Alzheimer’s Diagnosis using 3D CNN

Alzheimer’s disease (AD) is a progressive neurological disorder and the leading cause of dementia, requiring early detection for effective management. This study presents a deep learning-based dual-modality classification model using MRI and PET imaging to differentiate AD from cognitively normal (CN) individuals. The OASIS-3 dataset was used to acquire MRI and PET scans, which were pre-processed and organized into patient-specific folders. The proposed model consists of two convolutional neural networks (CNNs) for feature extraction from MRI and PET images, followed by a fusion layer and fully connected layers for classification. Data augmentation and normalization techniques were applied to improve generalization. The trained model demonstrated the potential of deep learning in neuroimaging-based AD detection, aiding in early diagnosis and clinical decision-making. Future work will focus on optimizing the network architecture and expanding datasets to enhance robustness and reliability.