Constructing High-Order Functional Connectivity Networks With Temporal Information From fMRI Data.

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Data Science

Background:

• Functional connectivity analysis of functional magnetic resonance imaging (fMRI) data is complex.
• Current methods for functional connectivity networks (FCNs) often neglect temporal dynamics, limiting accuracy.
• Temporal information is crucial for understanding blood oxygenation level-dependent signal changes.

Purpose of the Study:

• To develop a novel framework for extracting temporal dependencies from fMRI data.
• To infer high-order functional connectivity (FC) by incorporating temporal information.
• To enhance FCNs using hypergraph-based manifold regularization and causal modeling.

Main Methods:

• Developed a framework to extract temporal dependencies from fMRI data.
• Inferred high-order FCNs by considering the current state, previous state, and hypergraph-based manifold regularization.
• Employed causal modeling for dynamic brain system analysis to obtain directed FC.

Main Results:

• The proposed framework achieved an average of 12% higher accuracy compared to non-temporal and low-order FCNs.
• The method demonstrated efficient processing times.
• Identified key, discriminative regions of interest (ROIs) consistent with prior research.

Conclusions:

• Integrating temporal information into FCN analysis significantly enhances accuracy in fMRI studies.
• The framework provides a robust method for dynamic brain connectivity analysis.
• This approach facilitates deeper insights into cognitive and behavioral processes through improved ROI identification.