Applications of random field theory to functional connectivity

Area of Science:

• Neuroimaging
• Statistical analysis
• Brain connectivity

Background:

• Functional connectivity in brain imaging is typically assessed using 3D correlation analysis of voxel measurements.
• Existing methods may struggle to differentiate true correlations from background noise in complex datasets.
• Positron Emission Tomography (PET) Cerebral Blood Flow (CBF) and Blood-Oxygen-Level-Dependent (BOLD) functional Magnetic Resonance Imaging (fMRI) are common neuroimaging modalities.

Purpose of the Study:

• To develop a novel method for analyzing functional connectivity across the entire 6D correlation matrix of all voxels.
• To introduce a new theoretical framework for identifying significant local maxima within this 6D matrix.
• To provide a robust statistical method for distinguishing true brain signal from background noise in neuroimaging data.

Main Methods:

• Proposed a 6D analysis framework examining correlations between all pairs of voxels.
• Searched for 6D local maxima within the computed correlation matrix.
• Derived a new theoretical formula based on random field theory to calculate the p-value for these local maxima.

Main Results:

• Developed a statistically rigorous method to identify significant functional connectivity patterns.
• The new random field theory-based formula effectively distinguishes true correlations from background noise.
• The method is applicable to both autocorrelations within a single imaging set and crosscorrelations between different imaging sets.

Conclusions:

• The proposed 6D analysis and associated statistical formula offer a powerful new tool for brain connectivity research.
• This approach enhances the reliability of functional connectivity findings derived from PET CBF and BOLD fMRI.
• The methodology can be applied to diverse neuroimaging analyses, including task-based activation studies and resting-state connectivity.