Model-based and data-driven methods for modelling and removing physiologically-driven signals from BOLD fluctuations

COEX 

Modelling physiological effects in fMRI data can increase the precision of fMRI for measuring local neural activity, as well as open new avenues for studying brain physiology and brain-body interactions in health and disease. In this educational talk, I will describe existing techniques for extracting systemic physiological information (cardiac, respiration, CO2) from concurrent physiological recordings (i.e. model-based) or directly from fMRI data (i.e. data-driven). I will also explain how to use this information for accounting for physiological confounding signals in fMRI data analysis (i.e. denoising), including both faster-scale oscillations time-locked to the respiratory/cardiac cycles and slower-scale variations (< 0.1 Hz) that overlap with the spectrum of intrinsic neuronal hemodynamic activity, and gaining further insights in the interpretation of fMRI findings for broad cognitive and clinical applications.