Identification of Physiological Response Functions to Correct for Fluctuations in Resting-State fMRI related to Heart Rate and Respiration

Abstract

Functional magnetic resonance imaging (fMRI) is widely viewed as the gold standard for studying brain function due to its high spatial resolution and non-invasive nature. However, it is well established that changes in breathing patterns and heart rate strongly influence the blood oxygen-level dependent (BOLD) fMRI signal and this, in turn, can have considerable effects on fMRI studies, particularly resting-state studies. The dynamic effects of physiological processes are often quantified by using convolution models along with simultaneously recorded physiological data. In this context, physiological response function ( ) curves (cardiac and respiratory response functions), which are convolved with the corresponding physiological fluctuations, are commonly employed. Initially, these curves were assumed to be identical across subjects, but more recently, the use of subject-specific curves has been suggested (derived by e.g. using the global fMRI signal). In the present study, we propose a novel framework for the robust estimation of curves and use this framework to rigorously examine the implications of using population-, subject-, session-and scanspecific curves. The proposed framework was tested on resting-state fMRI and physiological data from the Human Connectome Project. Our results suggest that curves vary significantly across subjects and, to a lesser extent, across sessions from the same subject. These differences can be partly attributed to physiological variables such as the mean and variance of the heart rate during the scan. The proposed methodological framework can be used to obtain robust scan-specific curves from data records with duration longer than 5 minutes, exhibiting significantly improved performance compared to previously defined canonical cardiac and respiration response functions. Besides removing physiological confounds from the BOLD signal, accurate modeling of subject-(or session-/scan-) specific curves is of importance in studies that involve populations with altered vascular responses, such as aging subjects.

Physiological and head motion signatures in static and time-varying functional connectivity and their subject discriminability

“…The signature induced by SLFOs remained after the MildA, MildB and FIX pipelines were applied, but was greatly reduced by the WM 50 and WM 200 strategies ( Figure 2E ). The observation that FIX, which is a rather aggressive preprocessing strategy, was unable to remove most of the SLFOs is consistent with recent studies showing that global artifactual fluctuations are still prominent after FIX denoising (Burgess et al, 2016; Glasser et al, 2018; Kassinopoulos and Mitsis, 2019b; Power et al, 2018, 2017). Notably, GSR seemed to be an effective technique for removing the physiological signature from SLFOs on static FC, albeit for some scans it appeared to introduce a negative correlation between the SLFOs and “neural” FC matrices ( Figure 2E ).…”

“…4E ). Notably, FIX denoising without GSR was unable to remove the confounds introduced by SLFOs, which is consistent with recent studies showing that global artifactual fluctuations are still prominent after FIX denoising (Burgess et al, 2016; Glasser et al, 2018; Kassinopoulos and Mitsis, 2019b; Power et al, 2018, 2017).…”

“…Finally, we evaluated the addition of model-based nuisance regressors to the preprocessing strategies. Specifically, we added the physiological regressors used to model SLFOs (Kassinopoulos and Mitsis, 2019b), cardiac pulsatility and breathing motion (Glover et al, 2000). We found that including the regressor that models SLFOs reduces their effect on static FC for all preprocessing strategies apart from WM 50 and WM 200 , but, in contrast to GSR, the similarity remains well above chance levels ( Supp.…”

“…The practice of removing the GS from fMRI data (i.e. GSR) has been adopted by many fMRI investigators as it has been linked to head motion artifacts and fluctuations in heart rate and breathing patterns (Birn et al, 2006; Byrge and Kennedy, 2018; Chang and Glover, 2009a; Falahpour et al, 2013; Kassinopoulos and Mitsis, 2019b; Power et al, 2018, 2014; Shmueli et al, 2007). Further, GSR has been shown to increase the neuronal-hemodynamic correspondence of FC measures extracted from BOLD signals and electrophysiological high gamma recordings (Keller et al, 2013), as well as strengthen the association between FC and behaviour (Li et al, 2019b).…”

“…8 ). These time-series were used in the current study as the physiological regressor related to SLFOs. Cardiac pulsatility: Pulsatility of blood flow in the brain can cause pronounced modulations of the BOLD signal (Noll and Schneider, 1994), which tend to be localized along the vertebrobasilar arterial system and the sigmoid transverse and superior sagittal sinuses (Dagli et al, 1999; Kassinopoulos and Mitsis, 2019b). We modelled fluctuations induced by cardiac pulsatility using 6 regressors obtained with 3 rd order RETROICOR (Glover et al, 2000), based on the pulse oximeter signal of each scan. Breathing Motion: Chest movements during the breathing cycle generate head motion in the form of head nodding by mechanical linkage through the neck, but also factitious motion (also known as pseudomotion) through small perturbations of the B0 magnetic field caused by changes in abdominal volume when air enters the lungs (Power et al, 2019; Raj et al, 2001; Van de Moortele et al, 2002).…”