Complexity Analysis of Brain Electrical Activity

The characterization of brain electrical activities in terms of neuronal complexity has recently received great attention. However, traditional complexity measures which are maximized for random sequences fail to quantify the inherent long-range correlation in brain dynamics. The recently introduced multiscale entropy (MSE) analysis accounts for the complexity over multiple time scales and therefore can reveal the complex structure of the brain electrical signal. The aim of this study is to test the applicability of MSE for electroencephalogram (EEG) signal characterization and classification. Besides, the MSE method has been tested against two other entropy metrics: permutation entropy (PermEn) and Lempel-Ziv entropy (LZEn) along with sample entropy (SampEn) in the context of discriminating EEG epochs recorded from different recording regions and for different brain states. It is shown that the MSE method with sample entropy as the complexity estimator outperforms other entropy measures for the cases considered and its potential is demonstrated as a feature extraction method for more sophisticated EEG signal classification scheme.