AlzClassNet: A Hybrid CNN-Transformer Model for Alzheimer’s Disease Detection and Staging

Alzheimer’s disease (AD) presents a significant global health challenge, and early diagnosis is crucial to improving patient care and outcomes. Although no definitive cure exists, timely identification allows for better management and intervention strategies, potentially slowing disease progression and giving families more time to adjust. In this study, we propose an advanced model, AlzClassNet, which combines a Convolutional Neural Network (CNN) with Transformers for detecting and staging Alzheimer’s disease (AD) from brain MRI scans. The CNN component of AlzClassNet excels at extracting detailed spatial features such as edges and textures from these MRI scans, while the Transformer captures broader spatial relationships and temporal patterns crucial for differentiating between various disease stages. By leveraging the complementary strengths of both CNNs and Transformers, AlzClassNet ensures more reliable and early diagnosis of AD. The experimental results on the real-world benchmark data set ADNI (Alzheimer’s Disease Neuroimaging Initiative) validate our findings and shows that our proposed model outperforms the baselines on standard metrics.