Modelling the effect of cardiac and respiratory fluctuations on the central autonomic network in a novel test-retest dataset.

Changes in physiological state corresponding to fluctuations in heart rate and respiration drive non-neuronal contributions to the BOLD fMRI signal, complicating investigation of regions of the brain which participate in and process autonomic regulation: the central autonomic network (CAN). The estimation of physiological response functions (PRFs) provides a tool to interrogate and minimize the effects of these noise processes on fMRI connectivity. In this study, we explore the reproducibility of cardiac and respiratory response functions used to denoise resting and task data acquired with 3T MRI and their effect on the test-retest reliability of connectivity within the CAN. We characterize group-level PRFs during rest, fast-paced breathing and breath-holds, and a cold-pressor task and show that cardiac response dynamics vary significantly across scan conditions and subjects. Comparing physiological nuisance signals with indices of sympathetic and parasympathetic activity used to map the CAN, we further demonstrate that PRFs may provide an opportunity to disentangle neuronal and non-neuronal correlates of cardiac activity in fMRI data. Finally, we evaluate the effect of denoising on the test-retest reliability of connectivity between regions associated with the CAN, shedding light on the uses and limitations of PRFs for fMRI studies of brain-body interactions.