Modulatory Interactions of Resting-State Brain Functional Connectivity

Xin Di, Bharat B. BiswalPLoS 2013 · DOI

Abstract

Functional brain connectivity studies typically rely on the synchronization of resting-state fMRI signals, often assuming stable relationships over time. However, recent evidence suggests that the strength and spatial distribution of these connections are dynamic. This study investigates how a third brain region modulates the connectivity between two other regions using the physiophysiological interaction (PPI) technique. By identifying eight brain networks and conducting voxel-wise PPI analysis, we demonstrate that regions within the same functional system generally exhibit positive modulatory interactions, while competing networks—specifically the Default Mode Network (DMN) and the Executive Network—show mutual negative modulation. These findings validate PPI as a tool for exploring the complex, non-linear dynamics of resting-state functional connectivity.

Introduction

In neuroimaging, functional connectivity (FC) is defined as the temporal correlation between brain regions during rest. While traditionally viewed as a static property, it is now understood that these relationships fluctuate. The mechanism driving these fluctuations remains a key question: is connectivity between two regions (A and B) regulated by a third region (C)? This study explores "modulatory interactions," where one brain region acts as a regulator for the communication between others.

Methods

The researchers adapted the Physiophysiological Interaction (PPI) method for resting-state fMRI (rs-fMRI). Traditionally used in task-based paradigms, PPI here identifies how the activity of one region modulates the connectivity between two others.

The study utilized the Beijing_Zang dataset (1000 Functional Connectomes Project), split into a Discovery Sample (n=64) and a Replication Sample (n=64). After identifying eight intrinsic networks via Spatial Independent Component Analysis (ICA), two representative seed regions (ROIs) were selected for each. A regression model was then applied where voxel activity was predicted by the interaction of the two seed regions' signals. Crucially, a deconvolution algorithm was used to estimate underlying neural activity from the BOLD signal, ensuring more robust results.

Key Results and Insights

The analysis revealed distinct patterns of modulation across the brain's functional architecture, which were consistently observed in both the discovery and replication samples.

Synergy and Competition

Most positive modulatory interactions occurred within the same functional systems. For instance, in the visual (extrastriate) network, additional visual regions enhanced the connectivity between primary seeds, suggesting a cooperative internal dynamic.

Conversely, the study found significant negative modulation between "competing" networks. Specifically, activity in the Executive Network was associated with decreased internal connectivity within the DMN, and vice versa. This provides a mechanistic basis for the well-known anti-correlation between these two systems.

Figure 4: Voxel-wise PPI analysis showing modulatory interactions between DMN seeds and Executive regions across discovery and replication samples. Blue clusters indicate negative modulation, illustrating the competitive relationship between these networks.

Attention Network Dynamics

Modulatory interactions were also observed in the superior parietal lobule and fusiform gyrus for the dorsal attention network. These findings indicate that third-party regions within the attention system help coordinate the synchronization of its primary nodes.

Figure 3: Modulatory interactions within the dorsal attention network, showing consistent spatial patterns across discovery and replication samples at z=50.

The Role of the Caudate

An interesting finding was the role of the left caudate body, which showed positive modulatory interactions with the DMN. This suggests that the basal ganglia may play a regulatory role in coordinating DMN activity.

Methodological Validation

The use of deconvolution was validated as it produced more statistically significant and robust clusters compared to raw BOLD signals, despite a high correlation between the two methods ($r \approx 0.6$). This confirms that brain connectivity is fundamentally non-linear and non-additive.

Discussion and Limitations

While PPI effectively identifies modulatory relationships, it does not establish causal directionality—it cannot distinguish which region is the "modulator" versus the "modulated." Future research could expand this work using connectome-wide approaches to avoid ROI selection bias and investigate how these modulatory "volume knobs" shift during cognitive tasks compared to the resting state.

Summary updated

Abstract

The functional brain connectivity studies are generally based on the synchronization of the resting-state functional magnetic resonance imaging (fMRI) signals. Functional connectivity measures usually assume a stable relationship over time; however, accumulating studies have reported time-varying properties of strength and spatial distribution of functional connectivity. The present study explored the modulation of functional connectivity between two regions by a third region using the physiophysiological interaction (PPI) technique. We first identified eight brain networks and two regions of interest (ROIs) representing each of the networks using a spatial independent component analysis. A voxel-wise analysis was conducted to identify regions that showed modulatory interactions (PPI) with the two ROIs of each network. Mostly, positive modulatory interactions were observed within regions involved in the same system. For example, the two regions of the dorsal attention network revealed modulatory interactions with the regions related to attention, while the two regions of the extrastriate network revealed modulatory interactions with the regions in the visual cortex. In contrast, the two regions of the default mode network (DMN) revealed negative modulatory interactions with the regions in the executive network, and vice versa, suggesting that the activities of one network may be associated with smaller within network connectivity of the competing network. These results validate the use of PPI analysis to study modulation of resting-state functional connectivity by a third region. The modulatory effects may provide a better understanding of complex brain functions.
BibTeX
@inproceedings{di2013modulatory,
  author = {Xin Di and Bharat B. Biswal},
  title = {{Modulatory Interactions of Resting-State Brain Functional Connectivity}},
  booktitle = {PLoS},
  year = {2013},
  doi = {10.1371/journal.pone.0071163},
}
Figure