Computational Neuroscience, Neuroimaging, Complex brain networks
Our lab focuses on the development and identification of computational models of brain activity (electrical, metabolic, and hemodynamic) in order to clarify how the signals we record in Neuroimaging (fMRI, PET, DWMRI) and Electrophysiology (EEG) are generated. Our research comprises two complementary approaches: Forward Problem: we develop computational models of the generation of EEG rhythms and fMRI signals, and study how these models can be coupled. These theoretical biophysical models allow us to study the link between metabolism (glucose and oxygen consumption), cerebral blood flow, electrical activity (postsynaptic potentials and action potentials) and the Blood-oxygenation level dependent (BOLD) response, and the specific role of excitation and inhibition in this coupling. Inverse problem: Here we fit to real data, the models we previously developed. We focus on estimating effective connectivity, and other biophysical parameters. The use of a Bayesian model comparison framework provides us with a tool for selecting the best between different plausible models. Additionally, we are interested in the study of the statistical properties of anatomical (using DWMRI data) and functional brain networks (using EEG and fMRI data). Results from these studies allow us to introduce anatomical and functional constraints in the models. For instance, in the generative EEG and EEG/fMRI models we have proposed, the brain areas were connected using an average anatomical connectivity matrix estimated from DWMRI data.