Comparative Study of Wavelet-Based Unsupervised Ocular Artifact Removal Techniques for Single-Channel EEG Data
Saleha Khatun1, Ruhi Mahajan1, Bashir I Morshed1
1Department of Electrical and Computer Engineering The University of Memphis Memphis TN 38152 USA.
IEEE Journal of Translational Engineering in Health and Medicine
|August 24, 2016
Summary
This study introduces an effective unsupervised method for removing ocular artifacts from single-channel electroencephalogram (EEG) data using wavelet transform. The findings highlight optimal combinations for real-time artifact removal in minimalistic EEG systems.
Area of Science:
- Neuroscience
- Signal Processing
Background:
- Electroencephalogram (EEG) records brain activity using scalp electrodes.
- Artifacts like eye blinks corrupt EEG signals, especially in minimalistic systems for natural environments.
Purpose of the Study:
- To investigate unsupervised and effective ocular artifact (OA) removal from single-channel streaming raw EEG data.
- To evaluate the wavelet transform (WT) decomposition technique for OA removal in single-channel EEG.
Main Methods:
- Analyzed seven raw EEG datasets using two WT methods: Discrete Wavelet Transform (DWT) and Stationary Wavelet Transform (SWT).
- Evaluated four WT basis functions (haar, coif3, sym3, bior4.4) with universal and statistical threshold (ST) methods.
- Quantified OA removal using correlation coefficients, mutual information, signal-to-artifact ratio, normalized mean square error, and time-frequency analysis.
Main Results:
- The optimal combination for OA removal was identified as DWT with ST, using either the coif3 or bior4.4 basis function.
- Temporal and spectral analysis confirmed the effectiveness of the selected WT combinations.
Conclusions:
- Wavelet transform is a viable tool for unsupervised OA removal from single-channel EEG data.
- The study provides optimal WT parameters for real-time artifact reduction in minimalistic EEG applications.
