Using Permutation Entropy to Measure the Electroencephalographic Effects of Sevoflurane

Abstract

Approximate entropy (AE) has been proposed as a measure of anesthetic drug effect in electroencephalographic data. Recently, a new method called permutation entropy (PE) based on symbolic dynamics was also proposed to measure the complexity in an electroencephalographic series. In this study, the AE and PE were applied to electroencephalographic recordings for revealing the effect of sevoflurane on brain activity. The dose-response relation of PE during sevoflurane anesthesia was compared with that of AE. Nineteen patients' electroencephalographic data were collected during the induction of general anesthesia with sevoflurane. PE and AE were applied to the electroencephalographic recordings, and the performance of both measures was assessed by pharmacokinetic-pharmacodynamic modeling and prediction probability. To ensure an accurate complexity measure of electroencephalographic recordings, a wavelet-based preprocessor was built in advance. Both PE and AE could distinguish between the awake and anesthetized states and were highly correlated to each other (r = 0.8, P = 0.004). The pharmacokinetic-pharmacodynamic model adequately described the dose-response relation between PE and AE and sevoflurane effect site concentration. The coefficient R between PE and effect site concentration was 0.89 +/- 0.07 for all patients, compared with 0.60 +/- 0.14 for AE. Prediction probabilities of 0.86 +/- 0.04 and 0.79 +/- 0.09 for PE and AE showed that PE has a stronger ability to differentiate between the awake and anesthetic states. The results show that PE can estimate the sevoflurane drug effect more effectively than AE. This method could be applied to design a new electroencephalographic monitoring system to estimate sevoflurane anesthetic drug effect.