A Wavelet-Based Artifact Reduction From Scalp EEG for Epileptic Seizure Detection

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Scalp electroencephalogram (EEG) recordings are crucial for diagnosing epilepsy.
• Artifacts in EEG data can significantly hinder accurate seizure detection and diagnosis.
• Existing artifact removal methods may not effectively isolate seizure-specific activity.

Purpose of the Study:

• To develop and evaluate a novel method for reducing artifacts in scalp EEG recordings.
• To enhance the accuracy of seizure detection in epilepsy patients by improving EEG signal quality.
• To validate the proposed artifact removal technique on simulated and real EEG datasets.

Main Methods:

• The proposed method utilizes stationary wavelet transform (SWT).
• It specifically targets the spectral band of seizure activities (0.5–29 Hz) for artifact separation.
• Artifacts are simulated using templates to mimic real-world noise in EEG.

Main Results:

• The algorithm effectively separates artifacts from seizure activities in EEG signals.
• Post-artifact removal, EEG features for seizure detection become more distinguishable.
• A significant reduction in false alarms for seizure detection was observed.

Conclusions:

• The proposed stationary wavelet transform-based method demonstrates high efficacy in EEG artifact removal.
• This technique improves seizure detection performance, aiding in epilepsy diagnosis.
• The algorithm has potential applications in neuroscience research and other EEG-based studies.