Tracking dynamic brain networks using high temporal resolution MEG measures of functional connectivity

Area of Science:

• Neuroscience
• Computational Neuroscience
• Signal Processing

Background:

• Functional brain connectivity fluctuates rapidly, on the millisecond timescale.
• Conventional connectivity metrics lack the temporal resolution to capture these fast fluctuations.
• Time-resolved analysis using fixed sliding windows introduces arbitrary limitations.

Purpose of the Study:

• To evaluate high temporal resolution metrics for detecting fast fluctuations in functional connectivity.
• To identify temporally evolving brain subnetworks using non-negative tensor factorization.
• To introduce and assess a novel metric: instantaneous amplitude correlation.

Main Methods:

• Utilized phase difference derivative, wavelet coherence, and instantaneous amplitude correlation.
• Employed recurrence plots and pair-wise orthogonalization to handle noisy MEG data and signal leakage.
• Validated metrics using dynamically coupled neural mass models and real magnetoencephalography (MEG) data.

Main Results:

• High temporal resolution metrics significantly outperformed conventional static connectivity metrics in simulations.
• Analysis of post-movement beta rebound MEG data revealed time-locked sensorimotor subnetworks.
• Resting-state MEG analysis demonstrated robust and consistent spatial patterns across the evaluated metrics.

Conclusions:

• The developed high temporal resolution metrics effectively detect rapid changes in functional brain connectivity.
• These methods offer a robust approach for analyzing dynamic brain networks in cognitive tasks.
• The techniques can be applied to temporal graph analysis for deeper insights into brain network topology.