Nonlinear analysis of the electroencephalogram in depth of anesthesia

Abstract

Digital signal processing of the electroencephalogram (EEG) became important in monitoring depth of anesthesia (DoA) being used to provide a better anesthetic technique. The objective of this work was to conduct a review about nonlinear mathematical methods applied recently to the analyses of nonlinear non-stationary EEG signal. A review was conducted showing time- and frequency-domain nonlinear mathematical methods recently applied to EEG analysis: Approximate Entropy, Sample Entropy, Spectral Entropy, Permutation Entropy, Wavelet Transform, Wavelet Entropy, Bispectrum, Bicoherence and Hilbert Huang Transform. Some algorithms were implemented and tested in one EEG signal record from a patient at The Sabana University Clinic. Recently published results from different methods are discussed. Nonlinear techniques such as entropy analysis in time domain and combination with wavelet transform, and Hilbert Huang transform in frequency domain have shown promising results in classifications of depth of anesthesia stages.