A new algorithm for neural connectivity estimation of EEG event related potentials

Area of Science:

• Neuroscience
• Signal Processing
• Computational Biology

Background:

• Estimating neural connectivity is crucial for understanding brain function.
• Event-related potentials (ERPs) in electroencephalography (EEG) offer insights into neural processing.
• Existing methods for dynamic connectivity estimation during ERPs have limitations.

Purpose of the Study:

• To introduce a novel algorithm for dynamic neural connectivity estimation during ERPs.
• To validate the proposed algorithm's performance using simulated data.

Main Methods:

• The algorithm involves two key steps: time-varying multivariate autoregressive (MVAR) model estimation and generalized partial directed coherence (gPDC) calculation.
• MVAR estimation utilizes an adapted Nuttall-Strand algorithm, a multivariate extension of Burg's method.
• Connectivity is assessed between EEG channels using gPDC.

Main Results:

• The algorithm successfully estimated neural connectivity in simulated ERP data.
• Validation demonstrated the algorithm's capability to capture dynamic network changes.
• Simulations incorporated physiologically relevant ERP features for robust testing.

Conclusions:

• The proposed algorithm provides a robust method for estimating neural connectivity during ERPs.
• This approach can advance the analysis of brain network dynamics in response to stimuli.
• The validated algorithm holds potential for applications in various neuroscience research areas.