Non-sinusoidal waves in the EEG and their simulated effect on anaesthetic quantitative EEG monitors

The effect of anaesthetic drugs on the cortex are commonly estimated from the electroencephalogram (EEG) by quantitative EEG monitors such as the Bispectral Index (BIS). These monitors use ratios of high to low frequency power which assumes that each neurological process contributes a unique frequency pattern. However, recent research of the effect of deep brain stimulation on EEG beta oscillations suggests that wave shape, a non-sinusoidal feature that is only measurable in the time-domain, can change the frequency 'signature' of a neurological rhythmical process by the inclusion or removal of harmonic frequencies. If wave shape variations are present in the EEG of anaesthetised patients, then quantitative EEG monitors likely overestimate the anaesthetic drug effect. The purpose of this paper is to investigate alpha-wave shape in the EEG of anaesthetised patients and demonstrate the effect of wave shape on the frequency ratios that are commonly utilised in the BIS quantitative EEG monitor. EEG data, demographic information, and surgery details were collected prospectively from 305 patients undergoing a general anaesthetic for elective surgery. Alpha-wave shape was categorised by triangularity of the EEG extrema, a measure of how peaked (towards a sawtooth wave) or flat (towards a square wave) the extremum was. The alpha-wave was then artificially modified to either a sawtooth wave or square wave, and BetaRatio and PowerFastSlow metrics calculated. Age was found to be the only significant predictor of alpha wave triangularity. The artificially modified square-alpha waves increased the power in the frequency spectrum at 26 Hz by 1-5 dB, and increased the BetaRatio by 0.7. The alpha-wave of anaesthetised patients contains non-sinusoidal components which likely impact depth of anaesthesia calculations.