A permutation Lempel-Ziv complexity measure for EEG analysis

Abstract

ObjectiveIn this study we develop a new complexity measure of time series by combining ordinal patterns and Lempel-Ziv complexity (LZC) for quantifying the dynamical changes of EEG. MethodsA neural mass model (NMM) was used to simulate EEG data and test the performance of the permutation Lempel-Ziv complexity (PLZC) in tracking the dynamical changes of signals against different white noise levels. Then, the PLZC was applied to real EEG data to investigate whether it was able to detect the different states of anesthesia and epileptic seizures. The Z-score model, two-way ANOVA and t-test were used to estimate the significance of the results. ResultsPLZC could successfully track the dynamical changes of EEG series generated by the NMM. Compared with the other four classical LZC based methods, the PLZC was most robust against white noise. In real data analysis, PLZC was effective in differentiating the different anesthesia states and sensitive in detecting epileptic seizures. ConclusionsPLZC is simple, robust and effective for quantifying the dynamical changes of EEG. SignificanceWe suggest that PLZC is a potential nonlinear method for characterizing the changes in EEG signal.