Removal of muscle artifact from EEG data: comparison between stochastic (ICA and CCA) and deterministic (EMD and wavelet-based) approaches

Abstract

Electroencephalographic (EEG) recordings are often contaminated with muscle artifacts. This disturbing myogenic activity not only strongly affects the visual analysis of EEG, but also most surely impairs the results of EEG signal processing tools such as source localization. This article focuses on the particular context of the contamination epileptic signals (interictal spikes) by muscle artifact, as EEG is a key diagnosis tool for this pathology. In this context, our aim was to compare the ability of two stochastic approaches of blind source separation, namely independent component analysis (ICA) and canonical correlation analysis (CCA), and of two deterministic approaches namely empirical mode decomposition (EMD) and wavelet transform (WT) to remove muscle artifacts from EEG signals. To quantitatively compare the performance of these four algorithms, epileptic spike-like EEG signals were simulated from two different source configurations and artificially contaminated with different levels of real EEG-recorded myogenic activity. The efficiency of CCA, ICA, EMD, and WT to correct the muscular artifact was evaluated both by calculating the normalized mean-squared error between denoised and original signals and by comparing the results of source localization obtained from artifact-free as well as noisy signals, before and after artifact correction. Tests on real data recorded in an epileptic patient are also presented. The results obtained in the context of simulations and real data show that EMD outperformed the three other algorithms for the denoising of data highly contaminated by muscular activity. For less noisy data, and when spikes arose from a single cortical source, the myogenic artifact was best corrected with CCA and ICA. Otherwise when spikes originated from two distinct sources, either EMD or ICA offered the most reliable denoising result for highly noisy data, while WT offered the better denoising result for less noisy data. These results suggest that the performance of muscle artifact correction methods strongly depend on the level of data contamination, and of the source configuration underlying EEG signals. Eventually, some insights into the numerical complexity of these four algorithms are given.

Muscle Artifact Removal in Ictal Scalp-EEG Based on Blind Source Separation

“…The quantitative performance of two ICA methods and CCA algorithm has been evaluated by computing the Normalized Mean-Squared Error (NMSE) between the original EEG observed on the surface electrodes (EEG without muscle activity) and the reconstructed surface EEG after denoising using one of the BSS method, as described in [6,7]. Note that the selection of sources of interest is based on a visual inspection of components extracted by each algorithm in the time-frequency domain.…”

“…In the case of EEG, BSS assumes that electrical activities recorded at the level of surface electrodes can be considered as an instantaneous linear mixture of elementary sources [3,6,7] so that the linear observation model below holds:…”

“…Minimizing the disturbances due to muscular activity in EEG signals can be considered as a Blind Source Separation (BSS) problem [3], which consists in estimating the original sources underlying the multichannel EEG signals, without a priori (or very little) knowledge about the sources themselves and about the mixing process. BSS techniques was successfully applied to remove muscle artifacts from epileptic scalp-EEG [4][5][6][7]. However to date, no study address the challenging problem of extracting the low voltage rapid ictal discharges occurring during seizures.…”

Symbolic Entropy of the Amplitude rather than the Instantaneous Frequency of EEG Varies in Dementia

“…We used a linear filter to remove power line and used a nonlinear and non-stationary filter, EMD, to remove nonlinear trending or mixed non-stationarities from artifacts such as eye and muscle movements. Compared with independent component analysis [50,51], EMD works better for eliminating muscle-artifact in EEG [52,53]. Changes in the EEG dynamics of dementia were reported in varied frequency ranges such as alpha1 (8-10.5 Hz) [54,55], theta [55,56], delta [57], all frequencies [58] and all bands except delta [59].…”

A comparative study of denoising sEMG signals

“…Both EMD and wavelets are popular in denoising biomedical signals. There are some studies that compared EMD-based denoising and wavelet-based denoising on biomedical signals like ECG [20] and EEG [21]. However, they are not directly intended to investigate denoising sEMG signals while comparing the effect of median filtering.…”