Cross-Scale Causality and Information Transfer in Simulated Epileptic Seizures

Area of Science:

• Neuroscience
• Computational Neuroscience
• Information Theory

Background:

• Understanding causal interactions in neural oscillations is crucial for deciphering brain function and dysfunction.
• Epileptic seizures involve complex dynamics across multiple time scales, necessitating advanced analytical methods.

Purpose of the Study:

• To develop and validate an information-theoretic framework for detecting causality and information transfer between oscillatory components in neural time series.
• To apply this methodology to simulated epileptic seizures generated by the Epileptor model to identify cross-scale causal interactions.

Main Methods:

• Wavelet transform was employed to decompose neural time series into oscillatory components at different time scales.
• Conditional mutual information estimation was used to quantify information transfer between these components.
• Surrogate data testing was applied to ensure the statistical significance of the detected causal interactions.

Main Results:

• The analysis successfully identified three main time scales and their directional causal interactions within the simulated epileptic seizures.
• These findings were consistent with the known interactions between variables in the Epileptor model.
• The methodology demonstrated the ability to detect all directional (causal) interactions between the identified time scales.

Conclusions:

• The developed information-theoretic approach, combining wavelet transform and conditional mutual information, is effective for identifying causal interactions in neural data.
• This method is robust and suitable for application to experimental and clinical neural data, including EEG and MEG.
• The findings provide a foundation for investigating complex causal relationships in brain activity, particularly in conditions like epilepsy.