An epileptic seizure classifier using EEG signal

The brain is always active with exchange of electrical signals, which may be trapped by electroencephalograph (EEG). In the EEG the wave pattern reflects the neurons' electrical activity. These EEG signals due to environmental factors are contaminated with noises. There is a severe need to remove these artifacts before using the signals as inputs to any diagnostic systems. We propose the usage of daubechies wavelet (db8) to remove the artifacts and noises. Five frequency bands are extracted at the 8th level of decomposition. The features like mean, standard deviation, mode, minimum, maximum, frequency, energy, variance, entropy, median, phase, skewness, magnitude, kurtosis are extracted from reconstructed signal. These features are given as input for the classifier which identifies whether the person is affected by epilepsy.