Tracking of dynamic functional connectivity from MEG data with Kalman filtering

Area of Science:

• Neuroscience
• Biophysics
• Signal Processing

Background:

• Magnetoencephalography (MEG) and electroencephalography (EEG) offer millisecond temporal resolution crucial for studying dynamic functional connectivity in the human brain.
• Current methods for estimating functional connectivity from MEG/EEG involve a two-step process: solving the inverse problem for source activity and then estimating connectivity.
• Existing techniques lack the capability for real-time tracking of functional connectivity changes.

Purpose of the Study:

• To develop a novel method for simultaneous estimation of source activities and dynamic functional connectivity from MEG/EEG data.
• To overcome the limitations of current two-step approaches in capturing rapid functional connectivity dynamics.
• To enable real-time tracking of functional connectivity at the millisecond scale.

Main Methods:

• A Kalman filter-based approach was developed for the simultaneous estimation of source activities and dynamic functional connectivity.
• The method was validated using simulations, including scenarios with low signal-to-noise ratio (SNR < 1).
• Empirical MEG data from visual stimuli were analyzed to assess performance on real-world data.

Main Results:

• The proposed method reliably estimates source activities and resolves time-varying interactions, even at low SNR.
• Dynamic patterns of functional connectivity changes were successfully captured in empirical MEG data.
• The approach demonstrated the capability to track functional connectivity changes at millisecond resolution.

Conclusions:

• The developed Kalman filter method enables simultaneous and accurate estimation of brain activity and dynamic functional connectivity.
• This approach overcomes limitations of traditional methods, offering real-time tracking of functional connectivity.
• The technique is well-suited for leveraging the millisecond temporal resolution of MEG/EEG for advanced brain network analysis.