Multiresolution Analysis of Epileptic Seizure Signal to Eliminate EEG Artifacts

An electroencephalogram (EEG) is utilized to assess activity in the brain and find various epileptic conditions of the brain. But, EEG signals are regularly corrupted with different types of physiological artifacts. Due to this reason, epileptic seizure detection and further analysis of the seizure information becomes challenging and leads to ambiguous decisions made by neuro-specialists. Hence, it is important to remove these EEG physiological artifacts in the pre-processing phase itself. This research paper discusses EEG physiological artifact removal based on multiresolutional analysis and thresholding. Multiresolutional analysis is employed using discrete wavelet transform (DWT). Using DWT, the signal containing the seizure information is separated in to two components namely, high frequency and low frequency. Threshold is applied to these separated frequency components to eliminate artifacts from the original EEG seizure signal. The outcome of the investigation proved that the proposed technique removes the physiological disturbances in the seizure EEG signal by retaining the significant seizure EEG data. The examination also investigates the performance of the wavelet function to remove physiological artifacts in terms of signal-to-noise ratio and cross relation coefficient. The analysis proves that the proposed technique is able to conserve seizure information by removing artifacts in the examined datasets which is then verified using a simulated noisy signal.