To evaluate complexity of intracranial EEG as a biomarker for consciousness during seizures, awake and sleep.

Seizures are characterized by abnormally excessive neuronal activity, often resulting in altered consciousness. Prior studies in patients emerging from anesthesia demonstrate correlations between loss of scalp EEG complexity and consciousness. However, intracranial EEG (iEEG) better captures higher frequencies with less noise, providing a more robust signal for evaluation of complexity.

Eight patients with medically refractory epilepsy with intracranial depth electrodes to localize epileptic foci were included in this study. iEEG recordings sampled at 1000 Hz were segmented into random 20-second clips for awake, sleep, and seizure. Seizures were further subdivided into start, middle, and end. Multiscale Entropy and Correlation Dimension quantified iEEG complexity across multiple time scales. A repeated measures ANOVA was conducted on the complexity measure most correlated with the others to assess statistical significance across all states, followed by Wilcoxon Signed Rank Sum tests to evaluate differences between individual states.

The seizure state had the lowest signal complexity, particularly at the end, followed by sleep (intermediate) and awake (highest). Multiscale Entropy with time scale of 31 was most correlated to the other complexity measures. Primary analyses revealed significant differences (p<0.05) in complexity between all states. Secondary analyses revealed significant differences between awake and sleep (p=0.008), awake and seizure (p=0.016), but not between sleep and seizure (p=0.078). Complexity at seizure end was significantly lower than all other states (p<0.05).

iEEG complexity decreases significantly during seizures and sleep, nominating it as a potential biomarker for altered consciousness. Future studies specifically on temporal dynamics of complexity may advance understanding of mechanism underlying impaired consciousness during seizures.