Skip to Main Content
Permutation entropy (PE) is a new complexity measure which can extract important information from long, complex and high-dimensional time series. The advantages of this measure such as its fast calculation and robustness with respect to additive noise make it suitable for biomedical signal analysis. In this paper the ability of PE for characterizing the normal and epileptic EEG signals is investigated. Classification is performed using discriminant analysis. The effect of additive Gaussian noise on the discrimination performance is also studied and some parameters derived from PE are suggested to improve the classification accuracy when the signal is contaminated with noise. The results indicate that the proposed measures can distinguish normal and epileptic EEG signals with an accuracy of more than 97% for clean EEG and more than 85% for highly noised EEG signals.