TL;DR
This study reveals that the relationship between brain activity measured by fMRI and physiological signals like heart rate and respiration becomes stronger as a person's vigilance decreases. This suggests that changes in autonomic physiology play a significant role in shaping fMRI signals, especially during states of reduced alertness, impacting our understanding of brain function.
What Problem Does This Paper Address?
Functional Magnetic Resonance Imaging (fMRI) is a powerful tool for studying brain activity, but its signals are influenced by more than just neural activity. Physiological processes, particularly those related to the autonomic nervous system (like heart rate and breathing) and vigilance (our level of alertness), can significantly affect fMRI readings. While previous research has studied these influences separately, it's unclear how they interact and collectively shape fMRI signals, especially as vigilance naturally fluctuates over time, from full alertness to drowsiness or light sleep. This paper aims to disentangle these influences by investigating how the covariation between fMRI and autonomic signals changes as vigilance gradually decreases, both during rest and during a task.
Methods
The researchers employed a multi-modal approach, simultaneously recording:
- fMRI: To measure brain activity via blood-oxygen-level-dependent (BOLD) signals.
- EEG: To index vigilance levels by analyzing the ratio of alpha to theta band power.
- Respiration and Pulse Oximetry (PPG): To capture peripheral autonomic signals like respiratory volume (RV), heart rate (HR), and pulse wave amplitude (PWA).
Key methodological aspects include:
- Simultaneous recordings: This allowed for direct comparison of signals from different modalities.
- Sliding window analysis: Data was analyzed in non-overlapping windows (126 seconds) to capture gradual changes in vigilance and its impact on signal covariation.
- Autonomic and EEG preprocessing: Standardized pipelines were used to clean and extract relevant features from physiological and electrophysiological data.
- fMRI preprocessing: Multi-echo independent component analysis (ICA) was used for denoising, and standard steps like slice-timing correction and motion correction were applied.
- Regression and correlation analyses: These were used to quantify the relationships between fMRI signals, autonomic measures, and vigilance levels.
- Spatiotemporal analysis: Cross-correlations were used to examine the timing and spatial distribution of fMRI-autonomic relationships across different vigilance states.
- Functional Connectivity (FC) analysis: The study investigated how vigilance and autonomic influences affect the correlations between different brain networks.
Datasets
The study utilized data collected from 14 healthy, right-handed adult participants. This dataset included both resting-state scans and data from a psychomotor vigilance task. No new benchmarks were introduced; the focus was on analyzing existing, simultaneously acquired multimodal data.
Key Insights and Ideas
- Vigilance as a modulator: The study strongly suggests that vigilance isn't just a passive state but actively modulates the relationship between physiological signals and brain activity measured by fMRI.
- Autonomic influence amplifies with lower vigilance: The core finding is that the connection between fMRI signals and autonomic measures (heart rate, respiration) becomes significantly stronger as vigilance decreases. This implies that during states of reduced alertness, our physiological state has a more pronounced impact on brain imaging data.
- Spatiotemporal dynamics matter: The timing and location of these fMRI-autonomic correlations are not uniform. They show distinct patterns, with earlier correlations in grey matter and later ones in ventricles and white matter, and these patterns are more pronounced during lower vigilance.
- Fast EEG fluctuations also link to autonomic signals: Even rapid, seconds-level changes in EEG power (reflecting neural activity) show stronger correlations with autonomic signals as overall (minutes-level) vigilance decreases.
- Autonomic variance explains vigilance-related FC changes: A significant portion of how functional connectivity between brain networks changes with vigilance can be attributed to variations in autonomic physiology.
Key Results
- Increased fMRI-autonomic covariation: Regression analyses showed that the percentage of fMRI variance explained by autonomic signals increased significantly as baseline vigilance decreased. This effect was observed in both resting-state and task conditions.
- Resting-state shows stronger effects: The relationship between vigilance and fMRI-autonomic covariance was generally stronger during resting-state scans compared to the vigilance task.
- Network-specific effects: Several brain networks, including the default mode, salience/ventral attention, and limbic networks, showed significant vigilance-related changes in fMRI-autonomic covariance.
- Spatiotemporal patterns: Cross-correlation analyses revealed that autonomic signals influenced fMRI signals with specific temporal lags, with these effects being more widespread and pronounced during low vigilance.
- Functional connectivity changes: As vigilance decreased, functional connectivity between most brain networks strengthened, particularly during resting-state scans. Removing autonomic variance from fMRI signals largely attenuated this effect.
- Heart rate and pulse wave amplitude are key: Heart rate (HR) and pulse wave amplitude (PWA) showed a significant increase in covariance with fMRI as vigilance decreased, while respiratory volume (RV) showed a less consistent relationship.
Limitations and Future Work
- Autonomic signal variability: The study notes that the absolute values of respiratory volume (RV) and pulse wave amplitude (PWA) can be arbitrary due to variations in measurement setup, which is why they were excluded from some analyses.
- Task vs. Rest differences: The differences observed between task and resting-state conditions are attributed to factors like the task requiring more alertness, the potential for stimuli to elicit rapid autonomic responses that decouple from gradual vigilance changes, and the difficulty in perfectly modeling stimulus effects.
- Stimulus modeling complexity: The paper highlights the challenge of fully disentangling stimulus effects from vigilance and autonomic effects in task-based fMRI, especially given the collinearity between stimulus and autonomic regressors.
- Future directions: The authors suggest further research is needed to:
- Investigate the precise mechanisms through which vigilance, autonomic fluctuations, and neural activity interact.
- Explore the role of more rapid electrophysiological fluctuations and their contribution to fMRI signals.
- Develop more sophisticated models to better represent the covariance between these signals.
- Examine the impact of sleep events and stages on fMRI signals.
- Investigate how vigilance-related effects in fMRI signals differ across timescales.
- Assess the generalizability of these findings with larger sample sizes.
Summary updated