Speaker: Zhengwu Zhang, PhD, Associate Professor of Statistics and Operations Research, University of North Carolina
This talk presents a series of studies demonstrating how generative modeling can address key challenges in large-scale neuroimaging, including motion artifacts, site effects, high dimensionality, and limited clinical sample sizes. Using structural and functional connectomics from large cohorts such as ABCD and HCP, representation learning of brain networks is shown to yield more robust and reproducible brain–behavior associations. Motion-invariant modeling and unpaired MRI harmonization approaches are described to improve data quality and cross-site consistency. Finally, integrating large normative datasets and jointly modeling multimodal connectivity are highlighted as strategies for enhancing prediction and discovery in data-limited clinical studies.
The Department of Epidemiology & Biostatistics welcomes all participants to our events. If you need a reasonable accommodation to participate in this event because of a disability, please contact Liz Buggs ([email protected]).