Comparison of Frequency Bands Using Spectral Entropy for Epileptic Seizure Prediction

Abstract

Under the hypothesis that the uncontrolled neuronal synchronization propagates recruitingmore andmore neurons, the aim is to detect its onset as early as possible by signal analysis.This synchronization is not noticeable just by looking at the EEG, somathematical tools are needed for its identification.Objective.The aimof this study is to compare the results of spectral entropies calculated in different frequency bands of the EEG signals to decide which bandmay be a better tool to predict an epileptic seizure. Materials andMethods. Invasive ictal recordswere used.Wemeasured the Fourier spectrumentropy of the electroencephalographic signals 4 to 32minutes before the attack in low, mediumand high frequencies.Results.The high-frequency band shows amarkedly rate of increase of the entropy, with positive slopes and low correlation coefficient.The entropy rate of growth in the low-frequency band is practically zero, with a correlation around 0.2 and mostly positive slopes. The mid-frequency band showed both positive andnegativeslopeswithlowcorrelation.Conclusions. The entropy in the high frequencies could be predictor, because it shows changes in the previous moments of the attack. Its main problem is the variability, whichmakes it difficult to set the threshold that ensures an adequate prediction.