Abstract
A NON-INVASIVE DETECTION OF EPILEPTIC SEIZURE IN HUMAN SCALP ELECTROENCEPHALOGRAM USING DISCRETE WAVELET TRANSFORM
Opiarighodare D. K., *Ajala F. A., Oludipe O. and Ojebamigbe V. I.
ABSTRACT
This study shows the effectiveness of Discrete Wavelet Transform (DWT) technique in the analysis of Electroencephalogram (EEG) by performing decomposition of the EEG signals and extracting unique features for efficient diagnosis of epilepsy using standard benchmarked EEG datasets. The total dataset comprising 250 segments was grouped into five sets (A?E). The datasets used in this technique were chosen from records of EEG after artifacts caused by muscle movements and blinking of the eyes were removed. The datasets were used as inputs of the system. The technique applied includes signals preprocessing which involves synthesizing the signals which were originally represented in ASCII form, decomposition of the signals and extraction of features characterizing seizures. The researchestablished that though the decomposed EEG signals were imprecise, the wavelet detail coefficients computedgave an acceptable representation thatreveals the energy distribution of the EEGsignals in both time and frequency domain while the wavelet approximate coefficients computed gave a clearer view of spikes which indicate the presence of epileptiform activities in the EEG signals sub - band.
[Full Text Article] [Download Certificate]