High performance EEG feature extraction for fast epileptic seizure detection

Abstract

Epilepsy is a neurological disorder that affects around 70 million people worldwide. Early detection of epileptic seizures has the potential to help patients in improving their quality of life. Electroencephalogram (EEG) has been used to record the brain's electrical activities associated with seizures. This paper presents a fast method for selecting EEG features that are relevant to early detection of epileptic seizures. The feature extraction model is based on LASSO regression and is applied to the EEG spectrum to recognize the EEG spectral features pertinent to seizures. These features are then selected and fed into a random forest (RF) classifier for epileptic seizure recognition. Compared to the state-of-the-art methods, the proposed scheme achieves the highest detection performance of 100% sensitivity, 100% specificity, 100% classification accuracy, and 1.18 Sec detection delay. Furthermore, our model has proven to be robust in noisy and abnormal conditions.