A unified statistical model for the human electrocorticogram

Abstract

Objective: Extracellular field potentials (ECFs) generated in the cerebral cortex span a vast range of spatiotemporal scales. The process(es) leading to this large dynamic range remain debatable. Here we propose a novel statistical description of the amplitude spectrum of the human electrocorticogram (ECoG).

Methods: Spectral analysis was performed on long-term recordings from epilepsy patients undergoing pre-surgical evaluation with intracranial electrodes. Amplitude spectra were fit with a multi-component Gaussian model on semi-logarithmic axes.

Results: The Gaussian formulation provided excellent fits to the data. It also suggested how the changes accompanying the sleep-wake cycle and certain epileptiform transitions could be understood by variation in the parameters of the model.

Conclusions: The proposed continuum model synthesizes several previous observations regarding the statistical structure of the resting human ECoG. It offers a conceptual platform for understanding the EEG changes accompanying the sleep-wake cycle and pathologically hypersynchronous behaviour.

Significance: Statistical characterisation of the spectral distribution of field potentials yield insight into the cortico-cortical interactions that underlie the summated cortical ECFs comprising the ECoG. Such insight is relevant for a synoptic understanding of major state changes in the brain that are diagnosed in clinical practice by visual inspection of the ECoG.

Keywords: EEG; Epilepsy; Log normal distribution; Power law; Synchronisation.