TSS Group

Translational System & Signal Processing Group

Artifact Reduction from Scalp EEG for Epilepsy

Seizure Monitoring

Md Kafiul Islam and Zhi Yang

Abstract

This research presents a method to reduce artifacts from scalp EEG recordings to facilitate seizure diagnosis/detection for epilepsy patients. The proposed method is primarily based on stationary wavelet transform and takes the spectral band of seizure activities (i.e. 0.5 - 30 Hz) into account to separate artifacts from seizures. It requires a reference seizure epoch of N-sec which can either be generated from a patient-specific seizure database (if available) or can be simulated by a simple mathematical model of seizure. The purpose of the algorithm is to reduce as much artifacts as possible without distorting the desired seizure events to be detected/diagnosed. Different artifact templates have been simulated to mimic the most commonly appeared artifacts in real EEG recordings. The algorithm is applied on three sets of synthesized data: fully simulated, semi-simulated and real data to evaluate both the artifact removal performance and seizure detection performance. The EEG features responsible for detection of seizures from non-seizure epochs have been found to be easily distinguishable after artifacts are removed and consequently reduces the false alarms in seizure detection. Results from an extensive experiment with these datasets prove the efficacy of the proposed algorithm and hence this algorithm (with some modifications) is expected to be a future candidate for artifact removal not only in epilepsy diagnosis applications but also in other applications (e.g. BCI or other neuroscience studies).