Mutual Information Analysis of Brain-Heart Interactions in Epileptic Children

In this work we apply the network physiology paradigm to retrieve information from central and autonomic nervous systems before focal epileptic seizure, represented respectively by electroencephalogram (EEG) signals and R-R intervals (RRI), and investigate on the presence and strength of brain-heart interactions by computing mutual information (MI) measures. Statistical significance of MI values was tested through surrogate time series generated with the random shuffle approach. Our results suggest that the proposed method for aligning signals representing brain and heart activity measured with different sampling rates, is capable of revealing coupling between RRI representing heart system, and aligned averaged power spectrum of brain processes, measured with EEG, resulting in significant MI. For electrodes C3, Fp2, Cz, and T4 in correspondingly α, β, γ, and total frequency bands, we obtain significantly smaller values of MI in the pre-ictal period in comparison with baseline period, as well as general decrease of significant and all estimated MI values before the focal seizure can be observed.