Quantifying the causal interactions in the brain using a measure of directed transinformation

Area of Science:

• Neuroscience
• Computational Neuroscience
• Signal Processing

Background:

• Cognitive functions depend on coordinated neuronal interactions across brain regions.
• Quantifying temporal interactions in neuroimaging data, like electroencephalogram (EEG), is crucial.
• Existing measures for neural interactions include linear and nonlinear methods, but directionality is key.

Purpose of the Study:

• To propose and evaluate a directed transinformation (T measure) for quantifying causal interactions between neuronal sources.
• To assess the T measure's ability to capture both linear and nonlinear dependencies in neural signals.
• To demonstrate the application of the T measure to simulated and real EEG data.

Main Methods:

• Development of a directed transinformation (T measure) as a generalization of Granger causality.
• Application of the T measure to analyze temporal interactions in simulated and real electroencephalogram (EEG) data.
• Evaluation of the T measure's sensitivity to signal dependencies.

Main Results:

• The proposed T measure effectively quantifies causal interactions between neuronal sources.
• The T measure successfully captures both linear and nonlinear relationships in the data.
• The measure demonstrated sensitivity to dependencies when applied to simulated and real EEG signals.

Conclusions:

• The directed transinformation (T measure) provides a robust method for quantifying directed causal interactions in neural data.
• This approach enhances the analysis of brain connectivity by incorporating directionality.
• The T measure is a valuable tool for understanding information flow in the brain using EEG.