Detection of epileptic seizures from EEG signals with Hilbert Huang Transformation

Abstract

Epilepsy is a significant neurological disease that occurs due to abnormal activities of a particular portion of brain neurons. Electroencephalography (EEG) signals are mainly used to detect this disease. Epilepsy can be diagnosed automatically by measuring and analyzing the non-linearity and non-stationary properties of EEG signals. In this study, the Hilbert Huang Transformation (HHT) is proposed to extract the distinctive features from EEG signals for epileptic seizure detection. Research work, firstly, the mean Instantaneous Amplitude (IA) and mean Instantaneous Frequency (IF) data were extracted from EEG signals with Hilbert Huang Transformation (HHT) as a feature. Then, these features were classified with Extreme Learning Machine (ELM). Classification results indicated that epileptic seizures are detected with high accuracy. In addition, the performance evaluation of the proposed method was compared with some other techniques studied by using the same dataset recently. According to the experimental results, HHT based approach has 0.5-1% better classification accuracy than current studies and higher accuracy in detecting epileptic seizures than similar studies.