Abstract
Studies of brain functional connectivity have provided a better understanding of the organization and integration of large-scale brain networks. Functional connectivity using resting-state functional magnetic resonance imaging (fMRI) is typically based upon the correlations of the low-frequency fluctuation of fMRI signals. While spatial maps using the amplitude of low-frequency fluctuations (ALFF) are reproducible, the influence of ALFF on functional connectivity measures remains poorly understood. This study analyzed resting-state fMRI data from 79 healthy older individuals using spatial independent component analysis (ICA) and region-of-interest (ROI) based connectivity analysis. The results demonstrate a consistent correlation between ALFF variability and network strengths across all resting-state networks. Furthermore, the connectivity of several key regions—including the anterior cingulate cortex, medial prefrontal cortex, and precuneus—was found to be significantly modulated by their own regional ALFF. These findings suggest that fluctuation amplitudes may reflect functional significance or physiological noise, necessitating their control in future resting-state connectivity research.
Introduction
In resting-state functional magnetic resonance imaging (rs-fMRI), researchers typically measure "functional connectivity" by calculating the correlation between low-frequency blood-oxygen-level-dependent (BOLD) signal fluctuations in different brain regions. While we have reproducible maps of these networks, the amplitude of these fluctuations (ALFF) is often ignored. Theoretically, a correlation coefficient should be independent of the signal's scale; however, in biological data, the reliability of these signals is often tied to their power relative to noise. This study investigates whether the local "volume" of brain activity (ALFF) influences the measured "sync" (connectivity) between regions, a factor that could confound comparisons between healthy and diseased brains.
Methods
The researchers employed two primary analytical strategies on rs-fMRI data from 79 healthy older male individuals (mean age 80.3 years):
- Spatial Independent Component Analysis (ICA): This data-driven approach decomposes brain activity into spatially distinct networks. Voxel-matched inter-subject correlations were performed to see if network strength at a specific point related to the ALFF at that same point.
- ROI-based Connectivity Analysis: 160 Regions of Interest (ROIs) were defined across the brain. The connectivity between every pair of ROIs was calculated and tested for correlation with local ALFF variance across the subject group.
Standard preprocessing was performed using the SPM8 toolbox, with ALFF calculated specifically within the 0.01–0.08 Hz frequency band.
Key Results and Insights
The study found a consistent correlation between ALFF variability and network strengths across all major resting-state networks. This suggests that the "loudness" of the neural signal is intrinsically tied to how well that region integrates into a network.
The influence of ALFF was not uniform across the brain. It was most prominent in "hub" regions, including the anterior cingulate cortex (ACC), medial prefrontal cortex (MPFC), precuneus, insula, basal ganglia, and thalamus. These areas are heavily involved in the Default Mode Network (DMN) and saliency processing.
One of the most striking findings was the "asymmetry" of connectivity: the ALFF of one region might influence its connection to another, but the second region's ALFF might have no effect on that same connection. In the MPFC and ACC, higher ALFF generally led to stronger positive connectivity. Conversely, in the precuneus, higher ALFF was associated with more negative connectivity (stronger anti-correlations) with task-positive networks.
Discussion and Limitations
These findings suggest that ALFF should be used as a covariate in future functional connectivity studies. If a disease state shows "reduced connectivity," researchers must verify if it is actually a reduction in signal amplitude before concluding there is a network breakdown.
Limitations include the demographic specificity (elderly males) and the persistent ambiguity regarding whether high ALFF represents meaningful neural activity or physiological noise from large blood vessels. Regardless of its origin, the modulation of connectivity by signal amplitude is a critical factor that must be controlled to ensure the validity of rs-fMRI findings.
Summary updated