Classification of EEG Signals in the Improved Complete Ensemble EMD Domain

Abstract

Epilepsy is one of the major neurological problems the today's world is facing. In this paper we have introduced a method for the detection of epileptic seizure in the improved complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) domain. Here, firstly we have applied improved CEEMDAN on the electroencephalogram (EEG) signals to get the modes (Ms) and on these modes we will apply fractal dimension, generalized Hurst exponent and higher order statistical moments. All the calculated features of the EEG dataset which is available online are feed to two classifiers namely support vector machine (SVM) and k-nearest neighbors algorithm (kNN). The classification result of both the classifiers show a good accuracy of 100% in order to classify the normal and ictal as well as interictal and ictal EEG signals.