Graph-Based Multimodal ASL-fMRI Coupling Analysis for Distinguishing Significant Memory Concerns from Normal Controls

Motivation: Motivated by the complementary insights of ASL and fMRI in measuring cerebral perfusion and neural activity, this study aims to enhance the detection of significant memory concerns (SMC) from normal controls (NC). Goal(s): We developed a graph-based ASL-fMRI coupling framework to differentiate SMC from NC using graph convolutional networks. Approach: The approach involves preprocessing ASL and fMRI data, extracting graph-based features, and integrating them into a new multimodal representation for classification and visualization. Results: Results show significant connectivity differences between SMC and NC, improving sensitivity in distinguishing SMC abnormalities, which contributes to advancing early Alzheimer's disease diagnosis. Impact: This approach provides enhanced diagnostic sensitivity for detecting preclinical Alzheimer's disease, offering a valuable tool for early intervention by analyzing combined neurovascular and functional connectivity through ASL-fMRI integration.