A Comparison of fMRI Data-Derived and Physiological Data-Derived Methods for Physiological Noise Correction

Abstract

Physiological noise has been shown to have a large impact on the quality of functional MRI data, especially in areas close to fluid-filled cavities and arteries, such as the brainstem. Commonly, physiological recordings during scanning are transformed with methods such as RETROICOR and used as nuisance regressors in general linear models to remove variance associated with cardiac and respiratory cycles from the data. In contrast, modern pre-processing pipelines such as fMRIPrep, have created easy access to streamlined data-driven noise reduction methods such as aCompCor and ICA-AROMA. In combination, these methods have shown efficacy in correcting for motion, scanner as well as physiological artifacts. Given the ease of usability, it has to be questioned, whether there is any added benefit to applying logistically demanding methods such as RETROICOR. To answer this question, we applied RETROICOR, ICA-AROMA and aCompCor to a resting-state data set and compared variance explained by the respective methods and improvements in temporal signal-to-noise ratio throughout different regions of interest in the brain. In line with previous literature, RETROICOR significantly explains variance throughout the brain with peaks around areas of strong cardiac pulsations. ICA-AROMA and aCompCor largely account for the same variance. Nonetheless, RETROICOR retains unique explanatory power in individual participants. Further analysis points towards a pattern of unreliability of ICA-AROMA and aCompCor to consistently remove physiological noise across recordings, which is compensated by RETROICOR. While some of this inconsistency could be attributed to misclassifications in the noise selection models of ICA-AROMA, most is likely the consequence of secondary factors such as fMRI sequence parameters (e.g. long TR) limiting the efficiency of aCompCor and ICA-AROMA. Thus, it is advisable to additionally apply RETROICOR, especially when assuming regionally high levels of physiological noise.