Bypassing the volume conduction effect by multilayer neural network for effective connectivity estimation.

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Biomedical Engineering

Background:

• Differentiating true brain interactions from spurious ones in neuroimaging is challenging.
• Volume conduction (VC) in electroencephalographic (EEG) data often causes spurious interactions between recording sites.

Purpose of the Study:

• To develop a method that jointly models causal brain relationships and volume conduction effects.
• To improve the accuracy of identifying true causal interactions in EEG data.

Main Methods:

• A multilayer neural network was developed to simultaneously model VC effects (time-invariant linear equation) and causal brain interactions (nonlinear multivariate autoregressive process).
• The network structure maps sensor data to source space, models source interactions in hidden layers, and returns to sensor space.
• Causality coefficients, derived from network weights and parameters, estimate causal interactions.

Main Results:

• The proposed method was validated using simulated data and applied to real EEG signals from a memory retrieval task.
• The results demonstrate the method's ability to eliminate volume conduction interference.
• This leads to a higher accuracy in identifying true causal interactions between brain regions.

Conclusions:

• The novel neural network approach effectively mitigates volume conduction effects in EEG analysis.
• This method enhances the accuracy of estimating causal brain connectivity.
• The findings have significant implications for understanding brain function through neuroimaging.