Upcoming Event: PhD Dissertation Defense

Uncovering spatiotemporal dynamic of tau-Abeta-atrophy for Alzheimer’s Disease

Zheyu Wen,

Abstract

Misfolded tau and amyloid-beta are central biomarkers of Alzheimer's disease. We develop computational models to characterize the spatiotemporal dynamics of these proteins and predict subsequent neurodegeneration. The proposed framework is calibrated using PET imaging data. The evolution of abnormal tau and amyloid-beta is governed by diffusion, reaction, clearance, and latent-state dynamics. In particular, the diffusion process is modeled through a learnable graph Laplacian that is both individualized and disease-progression dependent. We introduce a cohort-level inversion framework that jointly estimates cohort-shared parameters and subject-specific parameters by fitting the model to longitudinal PET scans. The proposed approach is evaluated on the ADNI and A4 datasets using multiple brain atlases. Our model demonstrates strong performance in forecasting future PET scans. Uncertainty quantification is performed via analytical Hessian.

Biography

Zheyu Wen is dedicated to developing computational models in medical imaging and inverse problems. He earned his bachelor’s degree in Communication Engineering from the University of Electronic Science and Technology of China and his master’s degree in Electrical and Computer Engineering from the University of Michigan, Ann Arbor.