Temporal and spatial characterization of physiological noise in rs-fMRI at a high temporal resolution

Abstract

Abstract The fMRI signal contains “noise” components resulting from physiological processes that interfere with the component due to neuronal activation and obscure the understanding of brain function. Having the neuronal activation component as the unknown, it is crucial to characterize the spatial and temporal aspects of the signal component that is due to cardiac and breathing cycles. 17 fMRI exams were performed in 9 subjects with a 3T Philipps MRI scanner with high sampling frequency (repetition time 125 ms, one slice). Photoplethysmography was used to track the cardiac cycle, and a pneumatic thoracic respiration transducer was used to measure the breathing cycle. The impact of the physiological signal on fMRI signal was evaluated in four regions of interest: (1) a region encompassing all areas below the subarachnoid space (Global Signal - GS), (2) ventricles (CSF), (3) venous return (VR), and (4) the region surrounding Middle Cerebral Arteries (MCA). Physiological noise is found heterogeneous across the ROIs in terms of amplitude and phase. The impact of respiration on the fMRI signal is greater than that of the heart, and it may be mediated by modulation of the venous return.