Quantification of depth of anesthesia by nonlinear time series analysis ofbrain electrical activity

We investigate several quantifiers of the electroencephalogram (EEG) signal with respect to their ability to indicate depth of anesthesia. For 17 pati ents anesthetized with sevoflurane, three established measures (two spectra l and one based on the bispectrum), as well as a phase space based nonlinea r correlation index were computed from consecutive EEG epochs. In the absen ce of an independent way to determine anesthesia depth, the standard was de rived from measured blood plasma concentrations of the anesthetic via a pha rmacokinetic/pharmacodynamic model for the estimated effective brain concen tration of sevoflurane. Tn most patients, the highest correlation is observ ed for the nonlinear correlation index D*. In contrast to spectral measures , D* is found to decrease monotonically with increasing (estimated)depth of anesthesia, even when a "burst-suppression" pattern occurs in the EEG. The findings show the potential for applications of concepts derived from the theory of nonlinear dynamics, even if little can be assumed about the proce ss under investigation.