Methods of EEG analysis for prediction of epileptic seizures

This paper considers epileptic seizures prediction methods based on electroencephalograms analysis. Basic methods of signal analysis for seizure prediction and the results of their application are presented. Methods of signal analysis in time domain (signal energy and curve length), methods based on signal transformation (spectral and wavelet analysis) and nonlinear methods (entropy analysis and synchronization analysis) were considered. Main problems that arise in epileptic seizures prediction are low noise resistance of methods and their unsuitability for EEG signals measured noninvasively. Recommendations for future research directions in seizure prediction are given.Ref. 23, figs. 1.