A New Approach for Investigating the Complexity of Short Term EEG Signal Based on Neural Network

The nonlinear quality of electroencephalography (EEG), like other irregular signals, can be quantified. Some of these values, such as Lyapunovchr('39')s representative, study the signal path divergence and some quantifiers need to reconstruct the signal path but some do not. However, all of these quantifiers require a long signal to quantify the signal complexity.

In this study, we present a new approach to investigate the complexity of turbulent signals in short term and use this method to investigate the complexity of EEG. This method is based on signal modeling and we compared this model with the real signal. The importance of this method is its ability to estimate the complexity of short-term signals, especially in signals whose dynamics change rapidly.

To quantify the appropriateness of the proposed method, this method was calculated on an EEG signal and also the values ​​of Lyapunov view were calculated by Wolf and Rosenstein and the correlation of the value obtained from the proposed method and two Lyapunov views were calculated. This value was 90% compared to Wolf method and 83% compared to Rosenstein method.

The method used in current study, can estimate the complexity of signals in short periods. This quantifier feature is of great help for tracking rapid changes and tracking the time sequence of this change. This quantifier can also be used to detect other disturbed signals.