Entropy and complexity of the electroencephalogram (EEG) have recently been proposed as measures of depth of anesthesia and sedation. Using surrogate data of predefined spectrum and probability distribution we show that the various algorithms used for the calculation of entropy and complexity actually measure different properties of the signal. The tested methods, Shannon entropy (ShEn), spectral entropy, approximate entropy (ApEn), Lempel-Ziv complexity (LZC), and Higuchi fractal dimension (HFD) are then applied to the EEG signal recorded during sedation in the intensive care unit (ICU). It is shown that the applied measures behave in a different manner when compared to clinical depth of sedation score--the Ramsay score. ShEn tends to increase while the other tested measures decrease with deepening sedation. ApEn, LZC, and HFD are highly sensitive to the presence of high-frequency components in the EEG signal.

Spectral entropy and approximate entropy of EEG are two totally different measures. They change similarly in deepening anaesthesia due to an increase in slow activity. In some cases, however, they may change in opposite directions when the EEG signal properties change during anaesthesia. Failure to understand the behaviour of these measures can lead to misinterpretation of the monitor readings or study results if no reference to the raw EEG signal is taken.

In this paper 5 methods for the assessment of signal entropy are compared in their capability to follow the changes in the EEG signal during transition from continuous EEG to burst suppression in deep anesthesia. To study the sensitivity of the measures to phase information in the signal, phase randomization as well as amplitude adjusted surrogates are also analyzed. We show that the selection of algorithm parameters and the use of normalization are important issues in interpretation and comparison of the results. We also show that permutation entropy is the most sensitive to phase information among the studied measures and that the EEG signal during high amplitude delta activity in deep anesthesia is of highly nonlinear nature.

The ability of two easy-to-calculate nonlinear parameters, the Higuchi fractal dimension (HDf) and spectral entropy, to follow the depth of sedation in the intensive care unit is assessed. For comparison, the relative beta ratio is calculated. The results are evaluated using clinical assessment of the Ramsay score. The results show that the HD/sub f/ discriminates well between Ramsay scores 2-4 while beta ratio is superior for deeper levels of sedation. The value of the HD/sub f/ correlates highly with the cutoff frequency of the low-pass prefilter while spectral entropy is sensitive to the length of the analysis window.

A novel algorithm for the detection and tracking of rhythmic patterns in the EEG signal is presented. The algorithm includes the following steps: 1) linear filtering using symmetric impulse response, 2) calculation of the first intrinsic mode of the filter output and 3) calculation of instantaneous frequency and amplitude using the Hilbert transform. The linear filter is adapted according to the instantaneous frequency. The algorithm is shown to perform well in tracking the alpha rhythm (the alpha coma pattern) in critically ill patients sedated with midazolam.