Analysing epileptic events online by neural nets, special preprocessing methods included

Normally EEG-signatures are analysed in the frequency domain while a common FFT- or wavelet-transformation transforms the time signals of the electromagnetic activities of the (wet) neurons into the frequency domain. We show that better and faster results can be obtained if instead of these common algorithms the FD-operation is applied to the pure time-signal and a difference-autopower-spectrum (DAS) calculation for eliminating the noisy contributions is done next. Especially we compare the signatures of some epilepsy events analysed by common signal analysis and by the class-oriented DAS- and FD-operations and discuss the results