SC-TauPath maps Alzheimer's tau spread via brain connectivity
New AI reveals how structural brain connections guide tau protein propagation in Alzheimer's.
SC-TauPath, developed by Jing Zhang and colleagues, offers a novel approach to understanding tau propagation in Alzheimer's disease by leveraging structural connectivity (SC) data from DTI and tau PET from 234 ADNI participants. The framework uses a Network Diffusion Model (NDM)-augmented multilayer perceptron, then applies gradient×input attribution to score each SC edge's contribution to tau prediction. These scores are translated into multi-scale pathway maps—backbone edges, high-traffic routes, and hub ROIs—which validate established Braak staging anatomy.
The method achieves strong cross-validated tau prediction and demonstrates that structural connections encode spatially specific information about regional tau distribution, moving beyond biophysical models that lack interpretability. This work provides a data-driven tool to trace how misfolded tau travels along neural highways, potentially enabling earlier detection of Alzheimer's progression and personalized treatment targeting specific propagation pathways. The framework is accessible via arXiv:2606.04066.
- Combines NDM-augmented MLP with gradient×input attribution to quantify each structural connection's role in tau spread
- Validated on 234 ADNI subjects using paired DTI and 18F-Flortaucipir PET data
- Produces pathway maps consistent with Braak staging, confirming known tau progression anatomy
Why It Matters
Offers interpretable, data-driven mapping of Alzheimer's tau spread, aiding early diagnosis and targeted therapies.