Detection of epileptic seizure in EEG signals using phase space reconstruction and euclidean distance of first-order derivative

Abstract

A most common disorder of human brain “Epilepsy”, which is a neurological disorder and is identified as unexpected and transient electrical disturbance of the brain. EEG is a widely used method of signal recording for detection of epileptic seizures. A modified method for classification of ictal (Epileptic seizure) and seizure-free EEG signals is proposed in this paper. The technique is employed for an epoch across the channel for feature extraction. The First order derivative (FOD) shows the rate of variability of the signal while the phase space reconstruction (PSR) shows the evolution of a system and the Euclidean distance measures the dispersion of the points in the 3-D PSR, these shows a better feature for ictal and normal EEG signals classification. The interquartile range of Euclidean distance has been used for feature selection due to its insensitivity towards the outlier. KNN classifier is used for classification of ictal and normal EEG signals. The methodology resulted detection of epileptic seizure in 0.1 second with the degree of performance, sensitivity-100%, Specificity-100%, and the Accuracy-100%.