Effects of model-based physiological noise correction on default mode network anti-correlations and correlations

Area of Science:

• Neuroscience
• Cognitive Neuroscience
• Functional Magnetic Resonance Imaging (fMRI)

Background:

• Previous research indicates an inverse relationship between the default-mode network (DMN) and task-positive network (TPN) activity during rest.
• Existing studies often use global signal normalization, which may introduce spurious negative correlations.
• The true extent of DMN-TPN anti-correlation without global signal processing remains unclear.

Purpose of the Study:

• To investigate DMN-TPN correlations using physiological noise correction instead of global signal removal.
• To determine if negative correlations persist and are enhanced by accounting for cardiac and respiratory noise.
• To assess the impact of physiological noise correction on both negative and positive correlations within brain networks.

Main Methods:

• Applied model-based physiological noise correction, including RETROICOR for cardiac/respiratory artifacts and regression of low-frequency variations.
• Analyzed resting-state fMRI data to examine correlations between the DMN and TPN.
• Compared results with and without physiological noise correction.

Main Results:

• Negative correlations between the DMN and TPN were observed in most subjects, irrespective of physiological noise correction.
• Physiological noise correction amplified the spatial extent and magnitude of these negative correlations.
• Physiological noise correction reduced false positives by decreasing spurious positive correlations within the DMN.

Conclusions:

• Negative correlations between the DMN and TPN are a genuine phenomenon, not solely an artifact of global signal processing.
• Physiological noise correction provides a more accurate characterization of intrinsic brain network interactions.
• This approach offers a robust alternative to global signal regression for analyzing resting-state fMRI data.