fMRI in the presence of task-correlated breathing variations
Introduction
A central challenge in functional magnetic resonance imaging (fMRI) is that the neuronal activity induced blood oxygenation level dependent (BOLD) signal changes are small compared to various sources of noise — including thermal noise, scanner instabilities, subject motion, and the subject's respiration and heart beat. The pulsation of the blood, for example, causes in-flow effects leading to signal changes throughout gray matter with the largest signal changes near large vessels and highly vascular regions (Biswal et al., 1996, Dagli et al., 1999, Glover et al., 2000, Lund et al., 2006). The movement of the chest during respiration causes magnetic field changes which can distort the acquired images, particularly in inferior regions and edges of the brain, as well as the CSF (Brosch et al., 2002, Raj et al., 2001). In addition to these fluctuations occurring at the cardiac and respiratory frequencies (approximately 1 Hz and 0.3 Hz, respectively) and their harmonics, slower fMRI signal changes can be induced by variations in cardiac rate (Bianciardi et al., 2008a, Bianciardi et al., 2009, Chang et al., 2009, Shmueli et al., 2007). The respiration variation induced changes are believed to be mediated in part by changes in arterial levels of CO2, a potent vasodilator. This effect can be most clearly seen in response to a breath-holding challenge, where even a brief breath hold can result in a signal increase of several percent (Abbott et al., 2005, Kastrup et al., 1999a, Kastrup et al., 1999b, Stillman et al., 1995, Thomason et al., 2005). However, even subtle changes in breathing depth and rate, which occur naturally during rest at very low temporal frequencies (∼ 0.03 Hz) (Van den Aardweg and Karemaker, 2002), can lead to significant signal changes (Bianciardi et al., 2008a, Bianciardi et al., 2009, Birn et al., 2006, Chang et al., 2009, Wise et al., 2004). Such fluctuations can either mask, or appear similar to, BOLD signal changes induced by neuronal activity. These findings therefore raise a concern for fMRI, particularly for studies where heart rate or breathing changes are associated with the task, as may be seen in some cognitive tasks or in response to some emotional stimuli. The focus of the study presented here is to investigate the issue of task-correlated breathing changes.
The influence of physiological fluctuations has been a particular focus in studies of “functional connectivity,” a technique that attempts to derive information about functional connections in the brain based on the correlation of time series fluctuations between different regions of the brain. Networks of brain regions could be falsely identified as functionally connected based on the similarity of cardiac- or respiratory-induced fluctuations to brain activation related changes. Care must therefore be taken in these studies to remove the influence of physiological noise.
In contrast to investigations of functional connectivity, the role of physiological noise in task activation studies has generally been of little concern, except in studies of regions known to be severely affected, such as the brain stem (Guimaraes et al., 1995). On average over the whole brain, cardiac and respiratory fluctuations account for less than 10% of the noise (Bianciardi et al., 2008a, Bianciardi et al., 2009, Shmueli et al., 2007). Likely because of this relatively small increase in noise, many researchers do not expend the additional time and effort to monitor and correct for physiological changes. While this does improve statistical power, others improve their statistics by increasing the amount of data acquired from each subject, or by including more subjects in the analysis.
A potentially more serious problem, however, occurs when the heart beat and respiration changes are correlated with the task. The concern is that blood flow and blood oxygenation changes induced by variations in the heart rate or breathing rate and depth, and mediated by non-neuronal mechanisms such as changes in the levels of arterial CO2, may be mistaken for neuronal-induced signal changes. Furthermore, the additional noise cannot be reduced by averaging more data. In fact, with averaging, these artifactual effects can increase. The correlation of physiological changes with the task cannot always be avoided. For example, tasks which are cognitively demanding, requiring focused or sustained attention, or emotional stimuli are likely to cause changes in breathing and heart rate. These changes are frequently used as behavioral measures (e.g. Lane et al., 2009, Napadow et al., 2008).
Previous studies have investigated the spatial and temporal patterns of respiration induced signal changes, and its influence on functional connectivity (Birn et al., 2006, Birn et al., 2008a, Birn et al., 2008b). In this study, we focus on the special case were the physiological noise is task-correlated. The goal of this study is to provide information that can help in determining whether or not task activation maps are influenced by task-correlated physiological noise, particularly task-correlated breathing changes, and to compare various strategies of dealing with this noise. Based on previous research, our hypothesis is that respiration-induced signal changes can be identified and distinguished from neuronal-induced BOLD signal changes because of differences in the spatial and temporal characteristics. We therefore compare various strategies that aim to exploit this difference in order to reduce the detrimental influence of physiological noise on two different data sets. The first data set involves a lexical decision making task, where some subjects showed task-correlated respiration changes. In the second data set, subjects were instructed to hold their breath during a presentation of a contrast-reversing checkerboard. This second data set was collected in order to test our methods for separating task correlated physiologic effects from activation for a more extreme case of task-correlated physiological noise.
Section snippets
Subjects and imaging parameters
In the first part of this study, we used data acquired from 10 healthy right-handed adult volunteers (ages: 21–44 yrs, average age 30.8 ± 8.2 yrs). The Institutional Review Board (IRB) of the National Institute of Mental Health approved the protocol, and all subjects provided informed consent. Other analyses of this data were previously published in Birn et al., 2006, Birn et al., 2008a, Birn et al., 2008b). Time series of T2⁎-weighted echo-planar MR images were acquired on a 3 T General Electric
Lexical task
The lexical decision task resulted in activations in the left and right precentral gyrus, middle occipital gyrus, fusiform gyrus, and inferior frontal gyrus (see Fig. 2). Decreases in the fMRI signal during task performance relative to fixation were observed in the anterior cingulate, posterior cingulate, precuneus, and the superior occipital gyrus — areas typically part of the “default mode network.” These areas are believed to be generally more active in the absence of a specific task (i.e.
Discussion
Changes in the respiration or heart rate that occur during the performance of a task are an important behavioral measure for a variety of studies. In addition, the sites of neural activity changes associated with these physiological changes are of great interest (e.g. Evans et al., 1999, Lane et al., 2009). A potential concern is that respiration or heart rate variations can produce hemodynamic changes that are not induced by the variations in neuronal activity. For example, changes in the
Conclusions
MRI signal changes correlated with variations in breathing depth and rate have a characteristic spatial and temporal profile that is different from the typical activation-induced BOLD response. Respiration related signal changes are evident throughout gray matter, particularly in the sagittal and transverse sinuses, and regions that have a high baseline blood volume, such as the primary sensory regions. In addition, respiration-related signal changes are delayed relative to activation induced
Acknowledgment
This research was funded by the National Institutes of Mental Health Intramural Research Program.
References (35)
- et al.
The inferential impact of global signal covariates in functional neuroimaging analyses
Neuroimage
(1998) - et al.
Modulation of spontaneous fMRI activity in human visual cortex by behavioral state
Neuroimage
(2009) - et al.
Separating respiratory-variation-related fluctuations from neuronal-activity-related fluctuations in fMRI
Neuroimage
(2006) - et al.
The respiration response function: The temporal dynamics of fMRI signal fluctuations related to changes in respiration
Neuroimage
(2008) - et al.
Mapping the MRI voxel volume in which thermal noise matches physiological noise—implications for fMRI
Neuroimage
(2007) - et al.
Influence of heart rate on the BOLD signal: the cardiac response function
Neuroimage
(2009) AFNI: software for analysis and visualization of functional magnetic resonance neuroimages
Comput. Biomed. Res.
(1996)- et al.
Localization of cardiac-induced signal change in fMRI
Neuroimage
(1999) - et al.
Regional variability of cerebral blood oxygenation response to hypercapnia
Neuroimage
(1999) - et al.
Neural correlates of heart rate variability during emotion
Neuroimage
(2009)
Functional MRI of human brain during breath holding by BOLD and FAIR techniques
Neuroimage
Non-white noise in fMRI: does modelling have an impact?
Neuroimage
The impact of global signal regression on resting state correlations: are anti-correlated networks introduced?
Neuroimage
Brain correlates of autonomic modulation: combining heart rate variability with fMRI
Neuroimage
Low-frequency fluctuations in the cardiac rate as a source of variance in the resting-state fMRI BOLD signal
Neuroimage
Functional MRI of brain during breath holding at 4 T
Magn. Reson. Imaging
Breath holding reveals differences in fMRI BOLD signal in children and adults
Neuroimage
Cited by (127)
Meditation attenuates default-mode activity: A pilot study using ultra-high field 7 Tesla MRI
2023, Brain Research BulletinPhysiological brain pulsations
2023, Advances in Resting-State Functional MRI: Methods, Interpretation, and ApplicationsAn orderly sequence of autonomic and neural events at transient arousal changes
2022, NeuroImageCitation Excerpt :In addition to neuro-vascular source, it appears that autonomic mechanisms also contribute to the observed fMRI signal changes. For example, the timing of the late, widespread reductions in fMRI signal at the end of the cascade (Fig. S6) relative to that of the autonomic arousal (evidenced from the joint drop in PPG-amplitude and increase in HR and RV) are consistent with the sympathetic mechanism described previously (Birn et al., 2009; Özbay et al., 2019). A similar autonomic-fMRI relationship was observed in a recent study of microsleeps defined by periodic eye closure, which is related to brain arousal state changes (Soon et al., 2021).