Nonlinear Noise Reduction and Predictability of Epileptic Seizures

R. Venugopal, A. Prasad, K. Narayanan, A. Spanias, and L.D. Iasemidis (USA)


Biomedical signals, Time series analysis, Nonlinear noise reduction, EEG, Seizure predictability.


In this work we address the problem of nonlinear filtering of noise in EEG (electroencephalographic) signal using the centroid correction method, in an attempt to improve the predictability of epileptic seizures long prior to their occur rences. We analyzed multi-channel EEG recordings con taining 15 successive seizures in one epileptic patient with focal temporal lobe epilepsy. The total EEG spanned 3 days. The results show an improvement in the predictabil ity time in most of the seizures in addition to improvement in the number of predictable seizures (from 10 to 12) after filtering and dynamical analysis of the EEG.

Important Links:

Go Back