Recursive dynamic functional connectivity reveals a characteristic correlation structure in human scalp EEG

Area of Science:

• Neuroscience
• Computational Neuroscience
• Signal Processing

Background:

• Traditional sliding window analysis uses lower-order statistics to study time-varying brain activity.
• This approach has limitations in capturing the full complexity of dynamic functional connectivity.

Purpose of the Study:

• Introduce recursive dynamic functional connectivity (rdFC) to analyze neurophysiological data at multiple temporal scales.
• Incorporate higher-order statistics for a more comprehensive understanding of brain connectivity patterns.
• Investigate the universality and scale-invariance of these patterns across diverse subjects.

Main Methods:

• Developed rdFC, a technique that builds hierarchical graphs across temporal scales.
• Analyzed over a million rdFC patterns from electroencephalograms (EEGs) of 2378 subjects.
• Utilized spatiotemporal evaluation to identify dominant connectivity patterns.

Main Results:

• Identified three dominant, mathematically equivalent connectivity patterns present across subjects and scalp locations.
• Demonstrated that these patterns exhibit spatial scale-invariance.
• Found a link between the number of connectivity patterns and the number of nodes used.
• Observed that temporal changes in rdFC patterns correlate with seizure dynamics.

Conclusions:

• rdFC offers a more advanced method for exploring time-varying neurophysiological activity.
• The identified connectivity patterns represent a universal, scale-invariant brain correlation structure.
• rdFC has potential applications in understanding neurological health and dynamics, including seizure prediction.