Monitoring changes in wakefulness level using spectral power-based EEG-indices

There is a need for empirical indicators that can monitor subtle changes in wakefulness levels with high temporal resolution. We aimed to assess the applicability in this regard of several indices based on the average spectral power of EEG rhythms, as well as the BIS index used in anesthesiology. 26 volunteers participated in an experiment involving forced awakenings from the slow-wave stage of daytime sleep: immediately after an alarm sound, they had to performance visual-motor and arithmetic tasks. From the EEG recordings, we isolated artifact-free segments with different levels of wakefulness: “sleep”, “awakening”, “partial wakefulness” (when task performance was still difficult) and “full wakefulness” (when the ability to perform tasks correctly was restored). EEG indices were calculated for this segments and an analysis was conducted to determine the ability of each index to distinguish between these states. The results obtained revealed that the most indicative indices were Gamma/Beta, Beta/Delta, Gamma/Delta, Complex index ((Alpha + Beta)/(Delta + Theta)) and BIS index. Then, for these indices, an assessment was made of the dependence of their values on muscle and eye movement artifacts, as well as how much their values change when opening or closing the eyes. Muscle artifacts had the greatest impact on the Gamma/Beta index, and eye movement artifacts had the greatest impact on the Beta/Delta, Gamma/Delta and Complex indexes. Cleaning up artifacts using filtering and ICA transformation significantly improved indexes performance. As a result, the BIS index proved to be the most informative – it was less affected by both muscle and eye movement artifacts. Our findings suggest that EEG indices may be a useful tool for monitoring subtle changes in alertness; however a combination of several different EEG indices may improve the accuracy of the results.