GMN4AD: Graph Matching Network for Alzheimer's Disease Diagnosis with Test-Time Domain Adaptation using Multi-centered Structure Magnetic Resonance Imaging

The article introduces the Graph Matching Network for Alzheimer's Disease Diagnosis (GMN4AD), which utilizes a novel approach to model interactions between heterogeneous brain graphs from structural Magnetic Resonance Imaging (sMRI) data. GMN4AD incorporates a test-time domain adaptation strategy that employs contrastive learning to address domain shifts, significantly improving diagnostic performance across three public AD datasets compared to existing methods. This advancement is crucial for practitioners as it enhances the accuracy of early Alzheimer's diagnosis, particularly during the mild cognitive impairment stage, by effectively handling variability in neuroimaging data.