Functional Connectivity Across the Brain and Temporally Enhanced Non-neuronal Low-Frequency Oscillation (SLFO) Blood Flow Signal Based on Functional Magnetic Resonance Imaging Scans

In the field of neuroscience, a core question is how the brain’s connectivity reconfigures over time to support adaptive functional changes. These dynamically changing neuronal variations can be measured in humans through functional magnetic resonance imaging (fMRI) by estimating functional connectivity (FC). FC quantifies the degree of coordination in neuronal activity across different brain regions, thereby reflecting the strength of neural connections.

The precise measurement of FC and its temporal changes relies on the reliability, validity, and specificity of blood-oxygen-level-dependent (BOLD) signals based on fMRI, which is the fundamental input for quantifying neural activity. This study shows that fMRI-based brain functional connectivity estimates are artificially inflated at spatially heterogeneous rates during resting-state and task-based scans. This results in false-positive connectivity strength changes and spatial distortion of brain connectivity maps. The data demonstrate that this artifact is driven by temporal inflation of SLFO blood flow signals in fMRI scans, which standard denoising processes cannot resolve. Our evidence indicates that SLFO inflation reflects brain blood flow disturbances caused by respiratory and heart rate changes during scans, although the mechanisms of this pathway are not fully understood. Finally, we show that adding a dedicated SLFO denoising process to the fMRI processing pipeline can mitigate artificial inflation of functional connectivity, enhancing the validity and intra-scan reliability of fMRI findings.

Received: September 13, 2023
Accepted: May 3, 2024
Published Online: Anonymous
Updated Check

The paper was authored by scientists from the Department of Psychiatry at Harvard Medical School, the Brain Imaging Center at McLean Hospital, the Neuroimaging Research Branch at the National Institute on Drug Abuse, and the National Institutes of Health, and it was jointly supervised by Cole Korponay, Amy C. Janes, and Blaise B. Frederick. This study provides a method to address the inherent issue of non-neuronal signal enhancement in fMRI-derived functional connectivity assessments.

1. Overview of Research Process: By investigating the temporal changes in FC between the insula and the striatum during resting-state and task-based scans, the authors found that blood flow disturbances from individual respiratory and heart rate changes were the cause of artifacts. Further whole-brain analyses confirmed that these time-dependent FC increases were a structural, task-independent phenomenon across the entire brain. Additional tests in replicated and independent samples demonstrated the generalizability of this conclusion.

2. Main Research Findings: The findings indicate that SLFO signals spatially increase unevenly as scan time progresses. This phenomenon is present in all human fMRI datasets and is associated with a decrease in individuals’ arousal levels during scans, although this does not always lead to changes in sleep states.

3. Research Conclusions and Significance: The study concludes that non-neuronal SLFO signals are the direct cause of FC and GMS inflation. By removing SLFO signals, FC inflation can be significantly reduced. By comparing the effectiveness of different denoising pipelines, the Riptide denoising pipeline was shown to stabilize the average brain-wide FC, completely removing FC inflation.

4. Research Highlights: This study addresses the issue of resting-state fMRI reliability by presenting an innovative perspective on the temporal growth of non-neuronal physiological signals. The SLFO denoising procedure developed in the study provides a powerful tool for improving fMRI data denoising and the precise measurement of dynamic FC changes.

This research was published in “Nature Human Behaviour” and was conducted jointly by Harvard University, McLean Hospital, and the National Institute on Drug Abuse. The study provides new insights into understanding the dynamic changes in FC and how the brain coordinates activity across different regions during rest or task performance. Additionally, the research findings will contribute to more accurate application of fMRI technology in brain function organization and disease studies, offering scientific evidence for the development of relevant clinical diagnostic and therapeutic strategies.