Michele Guindani. (University of California, Irvine)
Title: Bayesian Methods for Studying Heterogeneity in Brain Imaging Experiments.
Abstract: Dr. Guindani is a Professor in the Department of Statistics, University of California, Irvine. Before joining UCI, he has held faculty positions in the Department of Biostatistics, University of Texas MD Anderson Cancer Center and the Department of Mathematics and Statistics at the University of New Mexico. He has received his Ph.D. in Statistics from Università Bocconi, Milan, Italy in Spring 2005. He is currently a Co-Editor for Bayesian Analysis, the official journal of the International Society for Bayesian Analysis (ISBA) and he has been serving as Editor-in-Chief of the same journal from January 2019 to December 2021. He is also an Associate Editor for Biometrics.
An improved understanding of the heterogeneity of brain mechanisms is considered key for enabling the development of interventions based on imaging features. In this talk, we will discuss some examples of heterogeneity in animals’ and humans’ experiments. More specifically, we will first discuss the analysis of neuronal responses to external stimuli in awake behaving animals through the analysis of intra-cellular calcium signals. We propose a nested Bayesian finite mixture specification that allows for the estimation of spiking activity and, simultaneously, reconstructs the distributions of the calcium transient spikes' amplitudes under different experimental conditions. The proposed model borrows information between experiments and discovers similarities in the distributional patterns of neuronal responses to different stimuli. In the second part of the talk, we will discuss a computationally efficient time-varying Bayesian VAR approach for studying dynamic effective connectivity in functional magnetic resonance imaging (fMRI). The proposed framework employs a tensor decomposition for the VAR coefficient matrices at different lags. Dynamically varying connectivity patterns are captured by assuming that at any given time the VAR coefficient matrices are obtained as a mixture of only an active subset of components in the tensor decomposition. We show the performances of our model formulation via simulation studies and data from real fluorescence microscopy and fMRI studies.
Via Zoom: https://mcgill.zoom.us/j/85978187693?pwd=WWtJZUpnb0JXK3o5SStnOFcxK3FFUT09