BrainMap: B.T. Thomas Yeo, PhD; Hierarchical Bayesian Models of Brain Function and Disorder

Thomas Yeo, PhD

Assistant Professor, Electrical & Computer Engineering and Clinical Imaging Research Centre, National University of Singapore

Affiliated Faculty, SINAPSE, Duke-NUS & Athinoula A. Martinos Center for Biomedical Imaging (Harvard University)

Talk title: Hierarchical Bayesian Models of Brain Function and Disorder

Abstract: The concept of hierarchy is prevalent in system neuroscience models of brain functions and disorders. However, most of these models are not mathematical. The advantages of mathematical models are that underlying assumptions are made explicit and model parameters might be estimated given the appropriate data. In this talk, I will present two hierarchical Bayesian models of brain function and disorder. The first model encodes the notion that behavioral tasks engage multiple cognitive processes, in turn supported by multiple brain regions. Application to a meta-analytic database (N = 10,449) identified association regions ranging from being highly specialized to highly flexible. Flexible association regions were selectively connected to multiple specialized networks suggesting that they might bind or integrate information across specialized brain networks. The second model encodes the notion that AD dementia patients exhibit multiple factors of overlapping atrophy patterns. Application to the ADNI database (N = 810) demonstrates for the first time that most AD dementia patients and at-risk nondemented participants express multiple latent atrophy factors to varying degrees. These atrophy factors are also associated with distinct cognitive decline trajectories across the preclinical and clinical stages.