Nonparametric Dynamic Granger Causality based on Multi-Space Spectrum Fusion for Time-varying Directed Brain Network...

Area of Science:

• Neuroscience
• Computational Neuroscience
• Signal Processing

Background:

• Analyzing dynamic brain communication is crucial for understanding neurological function.
• Existing model-driven methods have limitations in capturing transient network organization.
• Reliable time-frequency representations are essential for accurate causality inference.

Purpose of the Study:

• To propose a novel nonparametric method for estimating time-varying directed brain networks.
• To enhance the reliability of dynamic causality inference through robust spectral representations.
• To assess the method's performance in simulations and real-world applications.

Main Methods:

• Developed nonparametric dynamic Granger causality based on Multi-space Spectrum Fusion (ndGCMSF).
• Integrated complementary spectrum information from different spaces for spectral representation.
• Employed systematic simulations and validations to test the method's efficacy.

Main Results:

• ndGCMSF demonstrated superior noise resistance compared to existing methods.
• The method effectively captured subtle dynamic changes in directed brain networks.
• Revealed specific patterns of hemispheric laterality during instruction response movements in hemiplegia.

Conclusions:

• ndGCMSF provides a powerful tool for analyzing dynamic brain networks in changing operational settings.
• The method's findings offer reliable features for distinguishing hemiplegia types and assessing motor function.
• Contributes to a deeper understanding of dynamical and directed communication in the brain.