Wavelet Transform for Seizures Detection in EEG Records

In this paper, we propose an algorithm based on wavelet transform for seizures detection in EEG records. Epilepsy is a disorder that affects about 1% of the population. Seizure detection is important in patient monitoring and in diagnostic. For this work were used records from 23 channels EEG scalp. The method consists in signal processing by using DWT (discrete wavelet transforms). The signals were passed through low pass filter and decomposed using high-pass filter. The outputs are the detail coefficients and approximation coefficients. The decomposition was repeated for increasing the frequency resolution, approximation coefficients were decomposed with high and low-pass filters. After signal processing for seizure detection we used neural network for classify the EEG seizure segments by finding the onset and offset points.