Human EEG dimensionality and depth of anesthesia

Although the brain is the target organ of anesthesia, the electroencephalogram (EEG) is not routinely monitored during anesthetic procedures. This is due primarily to the difficulty of interpreting changes in the complex EEG waveform with respect to anesthetic conditions. Most attempts at developing EEG derived variables have been based on spectral analysis. In this study the EEG is examined as a non-linear dynamic system that may exhibit chaotic behavior. Using this approach the EEG signal can be characterized by its fractal dimension based on a phase space geometric reconstruction of the EEG time series. Eight human subjects were studied at three anesthetic levels (light, nominal, and deep anesthesia). The dimensionality of EEG samples from these subjects is shown to decrease with increasing anesthetic depth. This property may prove important in the classification of brain activity and may have clinical utility as a diagnostic tool