EEG Signal Analysis based Detection and Prediction of Epileptic Seizure: A Review

This paper presents a review on the analysis of Electroencephalogram (EEG) for detection and prediction of seizures in patients who are suffering from epilepsy. Epilepsy analysis requires the patient to be on bed for acquisition of the EEG and these signals are studied by practitioners to identify its preictal, ictal and interictal stages which help in the diagnosis of epileptic seizures. With an automation of the detection of these stages and the spikes that are an indication of seizure occurrences, it would be helpful in the early and timely prediction of seizure and also with the continuous monitoring of patients in the ICU. Based on previous studies that have been conducted to develop an automated seizure prediction mechanism from EEG signals it is evident that early prediction of epileptic spikes is crucial. Literature revels that the signals from an EEG are first preprocessed, to remove noise and other artifacts and then the signals are subject to the feature extraction and classification. The spikes in the signal are identified based on the frequency and amplitude of the signals. Most of the works in this area is centered on the use of Power Spectrum Analysis and Machine learning approaches for analysis and processing of the signals.