EEG Spectral Coherence Analysis in Nocturnal Epilepsy.

Abstract

Electroencephalography (EEG) is widely employed in the study of sleep disorders. This paper exploits the identification of cyclic alternating patterns (CAPs), a periodic ubiquitous phenomenon nested in the sleep stages, to analyze the EEG spectral coherence in subjects affected by nocturnal frontal lobe epilepsy (NFLE) and healthy controls. For each EEG recording, we extracted several CAP A1 subtype 4 s time series. We analyze the coherence between each pair of electrodes for each individual to obtain its distribution for each frequency range of interest to investigate differences between cases and controls. In addition, the imaginary and real parts of the spectral coherence were calculated and plotted to assess their likelihood of segregation into different classes and anatomical regions. The results of this study suggest a relevant frontal-temporal neural circuitry difference between individuals affected by epilepsy and controls. This supports the observation that, though highly variable, a broad range of executive, cognitive and attentional deficit observed in subjects affected by NFLE might depend on frontal-temporal altered networking. The investigation of EEG activity in the domain of the complex sleep architecture represents a challenging topic in neurophysiology and needs new methods to explore the manifold aspects of sleep. This work aims to provide a simple method to distinguish NFLE from healthy subjects from a functional connectivity point of view and to explore the possibility of using a smaller EEG channel set to support diagnosis.