THE ROLES OF DIFFERENT COMPONENTS OF EEGS FOR SEIZURE PREDICTION -WAVELET ENERGY EVALUATION

Epilepsy, a chronic disorder of the nervous system affecting 1% of the population, is characterized by the abnormal synchronized firing of a large number of neurons. Alerting a patient and/or his attending staff to an impending epileptic seizure has obvious clinical importance. A lot of attempts at epileptic prediction have been made, some based on sharp-transient detection and some tracked changes in background activity. Wavelet transform was applied to 8 channel scalp EEGs recording from 3 epileptic patients of partial seizures secondarily generalized seizures. The data were sampled a couple of minutes or tens of minutes prior to the seizure onset. For each record and channel, the data was decomposed at 7 scales. Spike/sharp and slow wave components of EEGs can be highlighted at different scales. Energy of the spike/sharp and slow wave components was calculated from detail signal at different scales, respectively. Result: The energy of slow waves increased among 8 channels ahead of seizure onset several minutes hi all the 3 patients, but the energy of spike/sharp components had no trends. Conclusion: Slow wave components of EEGs are well suited for seizure prediction in partial type secondarily generalized seizures. High-amplitude slow waves of EEGs may be an important factor for seizure transition.