Atlas of intracranial EEG complexity across cortical regions and global states of consciousness in the human brain

Background Electroencephalography (EEG) signal complexity reflects the richness of brain activity and is considered a marker of consciousness. However, its normative values across physiological conscious states using intracranial EEG (iEEG) across cortical areas remain poorly defined. We aimed at referencing iEEG complexity in the human brain and its links with global states of consciousness. Methods We analysed 5703 iEEG recordings from healthy cortices in 106 participants during wakefulness, N2, N3, and REM sleep, using complementary markers: Kolmogorov complexity, permutation entropy and spectral entropy. An independent dataset comprising 474 recordings was used for validation and added data on epileptic cortices and propofol anaesthesia. Results iEEG complexity decreased with reduced consciousness, with highest values in wake/REM, and lowest under propofol. Complexity was more variable during N2/N3 sleep. Fronto-parietal regions exhibited the highest complexity in conscious states. Epileptic cortex showed lower complexity than healthy cortex during wake and REM. A supervised machine learning model reliably classified sleep stages using complexity markers. Discussion Markers of iEEG signal complexity reliably index global states of consciousness. This study provides reference values for iEEG signal complexity across cortical regions. These normative data support the use of complexity as a marker of physiological consciousness, cortical integrity and sleep stage detection.