Use of wavelet entropy to compare the EEG background activity of epileptic patients and control subjects

Abstract

In this study, we processed the EEG background activities of five epileptic patients by means of the wavelet packet transform (WPT), and compared them with the EEGs of five control subjects. The WPT provides a fast way of choosing the best out of a large number of time-frequency bases according to features of the signal to be analysed. From the wavelet packet coefficients, we calculated an information cost function (ICF), which evaluates the information associated with the energy distribution of the signal: the wavelet entropy. Results suggested that the entropy or degree of disorder in the energy distribution of control subjects' EEG was higher. Applying the ANOVA test, we verified that there was a significant difference (p-ANOV A<0.01 ) between EEG frames of both groups.