Ictal EEG Classification based on State Space Modeling of Intrinsic Mode functions

The Electroencephalogram signals are characterized by mental status of human body. The electrical activity of this signal is non-linear and non-stationary in nature. In this work, the EEG signal is to undergo empirical mode decomposition which yields a set of intrinsic mode functions. Taking into account the non-linear nature of the EEG signal, state space analysis of the IMFs is carried out, which eventually leads to an outstanding classification into ictal EEG and healthy EEG class. Here Kalman filter is used to take the state estimation of each IMFs. Then temporal and statically features of each IMFs are extracted. The adaptive neuro-fuzzy inference system is used to classify ictal EEG and healthy EEG status. The study shows Teager energy and Kurtosis are 100% accuracy in classification. Apart from this, combination of Kurtosis and standard deviation features shows 100% in sensitivity, specificity and accuracy achieved in classification of different ictal stages.