🤖 AI Summary
This study investigates the relationship between structural connectivity and the spatial propagation of tau pathology in Alzheimer’s disease. We propose a novel framework integrating a network diffusion model (NDM) with a multilayer perceptron, and for the first time apply gradient×input attribution at the level of structural connectivity to quantify the contribution of each white matter tract to tau distribution prediction. This approach enables the construction of a multiscale propagation map encompassing backbone edges, high-flux pathways, and hub regions. Leveraging DTI and 18F-Flortaucipir PET data from 234 ADNI participants, our model achieves robust tau prediction performance. The resulting propagation map aligns closely with Braak staging, demonstrating that structural connectivity encodes the spatial specificity of tau spread and offering both predictive power and neurobiological interpretability.
📝 Abstract
Understanding how structural connections are associated with tau propagation in Alzheimer's disease (AD) remains a central open question, yet existing computational models either rely heavily on biophysical assumptions or lack neurobiologically interpretable pathway maps. We present SC-TauPath, a structural connectivity (SC) attribution framework that maps tau propagation pathways from in vivo neuroimaging data. SC-TauPath combines a Network Diffusion Model (NDM)-augmented multilayer perceptron with gradient $\times$ input attribution to score each SC edge's contribution to tau prediction, then translates these attribution scores into multi-scale pathway maps (backbone edges, high-traffic routes, and hub ROIs), which validates established Braak staging anatomy. Applied to 234 ADNI participants with paired DTI SC and 18F-Flortaucipir PET, SC-TauPath achieves strong cross-validated tau prediction and yields attribution-based pathway maps consistent with established Braak staging anatomy, demonstrating that SC encode spatially specific information about regional tau distribution in AD.