Application of the sample entropy for discrimination between seizure and seizure-free EEG signals

Abstract

The electroencephalogram (EEG) is an invaluable measurement for the purpose of assessing brain activities. The detection of epileptic seizures based on EEG signal is very useful in diagnostics. In this paper, we present a new method for discrimination between seizure and seizure-free EEG signals. The proposed method is based on empirical mode decomposition (EMD) process. We investigated that the sample entropy of the intrinsic mode functions (IMFs) generated by EMD process have potential to discriminate seizure from the seizure-free EEG signals. We have shown that the sample entropy measurement of IMFs is able to characterize the irregularity of the seizure EEG signals. The sample entropy measured from the IMFs has been used as a feature in order to discriminate seizure and seizure-free EEG signals. The sample entropy measurement of IMFs has provided better discrimination performance. The proposed approach based on EMD and sample entropy is better than sample entropy based approach for EEG signal discrimination.