Electroencephalogram (EEG) is a record of brain electrical activities along the scalp which contains precious information related to different brain states. The aim of this study is to investigate the complexity of EEG signals and distinguish the EEG signals with different rhythm.

The EEG time series data used in this study come from single neural mass model. The model with the different parameters produces the EEG signals with different rhythm such as normal EEG signals and epileptiform spikes. The modified permutation-entropy is used to extract the complexity features of the normal EEG signals and epileptiform spikes and is compared with the permutation-entropy algorithm and Rényi permutation-entropy algorithm. The statistical analysis is used to characterize the differences between the signals.

The signals corresponding to   and   are easy to distinguish according to the average values and the standard deviations of the modified permutation-entropy. The statistical results reveal that group differences of the modified permutation-entropy values for the signals corresponding to   and   are significant, manifested as   and  . Compared with the permutation-entropy and Rényi permutation-entropy, F-values of the modified permutation-entropy are the greatest and p-values of the modified permutation-entropy are the smallest.

There are significant differences between the complexity features of normal EEG signals and epileptiform spikes signals. The signals can be distinguished under different lengths of time series, embedding dimensions and delay times by using the modified permutation-entropy. The modified permutation-entropy is an effective complexity measure to quantify the EEG signals. Compared with the permutation-entropy and Rényi permutation-entropy, the modified permutation-entropy is more effective to distinguish the EEG signals with different rhythm.