Fast eigenvector centrality mapping of voxel-wise connectivity in functional magnetic resonance imaging:...

Area of Science:

• Neuroimaging
• Network Neuroscience
• Computational Neuroscience

Background:

• Eigenvector Centrality Mapping (ECM) spatially characterizes brain connectivity by assigning network properties to voxels.
• A major limitation of ECM in functional magnetic resonance imaging (fMRI) is the computational expense and storage requirements of the connectivity matrix.

Purpose of the Study:

• To present fast ECM (fECM), an efficient algorithm for estimating voxel-wise eigenvector centralities from fMRI time series.
• To demonstrate the validity and applicability of fECM for high-resolution functional neuroimaging analyses.

Main Methods:

• fECM computes matrix-vector products directly from fMRI data, avoiding explicit storage of the connectivity matrix.
• The algorithm's performance and validity were tested using synthetic data with known connectivities and in vivo resting-state fMRI data.
• A method for generating time series with prescribed covariances was used to create synthetic datasets for validation.

Main Results:

• fECM achieves significant acceleration in computing voxel-wise centralities for fMRI data at standard and high resolutions.
• Validation on synthetic data showed results comparable to the theoretical gold standard.
• Analysis of resting-state fMRI data revealed fECM-detectable connectivity changes after repetitive transcranial magnetic stimulation, with a moderate influence of motion parameters.

Conclusions:

• fECM provides a computationally efficient and statistically sensitive method for brain connectivity analysis in fMRI.
• The algorithm enables high-resolution and multimodality neuroimaging analyses on standard hardware.
• fECM is a promising tool for characterizing network properties in functional neuroimaging data.