Correction of EOG Artifacts in Event-Related Potentials of the EEG: Aspects of Reliability and Validity

Correction of EOG artifacts using a regression approach is evaluated in terms of reliability and validity. Transmission rates are estimated for eight EEG channels in 67 subjects. The trimmed group means of these rates are shown to provide reliable measures. Eye artifact correction based on these group means is superior to the conventional rejection in terms of reducing correlation between EOG and EEG.     

EOG correction: Which regression should we use?

Electrooculogram (EOG) correction is used to remove eye-movement-related contamination from electroencephalograms (EEG). Correction is reliant on the regression procedure, although when multiple EOG channels are used in the correction, the appropriate type of regression to use is not known. In the present study, we aimed to resolve this matter. Computer simulations were used to compare the simultaneous, multiple-stage, and single-channel regression methods of correction. EOG propagation was modeled on prior findings, under conditions of varying vertical and horizontal EOG (VEOG/HEOG) correlation. The dependent variable was the correlation between the uncontaminated and the corrected EEG. The simultaneous regression procedure gave the best correction, with its advantage increasing as a function of VEOG/HEOG correlation. It is recommended that the simultaneous regression procedure be used for EOG correction of the EEG.     

Computerized processing of EEG-EOG-EMG artifacts for multi-centric studies in EEG oscillations and event-related potentials.

The aim of this study is to present a package including standard software for the electroencephalographic (EEG), electro-oculographic (EOG) and electromyographic (EMG) preliminary data analysis, which may be suitable to standardize the results of many EEG research centers studies (i.e. multi-centric studies) especially focused on event-related potentials. In particular, our software package includes (semi)automatic procedures for (i) EOG artifact detection and correction, (ii) EMG analysis, (iii) EEG artifact analysis, (iv) optimization of the ratio between artifact-free EEG channels and trials to be rejected. The performances of the software package on EOG-EEG-EMG data related to cognitive-motor tasks were evaluated with respect to the preliminary data analysis performed by two expert electroencephalographists (gold standard). Due to its extreme importance for multi-centric EEG studies, we compared the performances of two representative "regression" methods for the EOG correction in time and frequency domains. The aim was the selection of the most suitable method in the perspective of a multi-centric EEG study. The results showed an acceptable agreement of approximately 95% between the human and software behaviors, for the detection of vertical and horizontal EOG artifacts, the measurement of hand EMG responses for a cognitive-motor paradigm, the detection of involuntary mirror movements, and the detection of EEG artifacts. Furthermore, our results indicated a particular reliability of a 'regression' EOG correction method operating in time domain (i.e. ordinary least squares). These results suggest that such a software package could be used for multi-centric EEG studies.     

Temporal stability of regression‐based electrooculographic correction coefficients

A means of accounting for ocular artifact in the electroencephalograph (EEG) is to subtract portions (Bs) of ocular voltage measured by the electrooculograph (EOG) from the EEG. Some such EOG correction methods calculate Bs at one time and use these to correct data recorded at a different time; these require information about the temporal stability of the Bs. This study investigated the stability of Bs over a 2‐hr EEG recording session. Participants performed 5 eye movement tasks, each separated by 30 min. Four EOG correction methods were then used to calculate Bs from each of the 5 data sets, resulting in VEOG, HEOG, and REOG (where appropriate) Bs for each methods at each of the 5 time points. We did not find evidence that Bs changed over the 2‐hr period, nor of any difference in temporal stability between the methods. This study suggests that it is appropriate to employ Bs calculated from calibration trials to correct data recorded within at least a 2‐hr time window.     

Differential Contributions of Blinks and Vertical Eye Movements as Artifacts in EEG Recording

Blinks and vertical eye movements were studied as artifacts of EEG recording. The electro-oculogram (EOG) and vertex vs joined mastoids EEG were recorded in 13 college-aged subjects. Subjects were asked to blink “normally, without excessive effort,” and move their eyes through vertical visual arcs of 5°, 10°, 20°, 30°, and 60°. The ratio EEG/EOG, the fraction of the EOG potential transmitting to the scalp EEG electrode as artifact, was calculated for potentials generated during both blinks and eye movement. Vertical eye movement scalp EEG artifact was a constant percentage of the vertical eye movement EOG across visual arcs of 10° to 60°. Mean percentage eye blink EEG artifact (9.3%) was significantly (p < .001) less than the mean percentage vertical eye movement artifact (13.9%). Thus, blink and vertical eye movement artifact fields are quantitatively different in terms of their transmission to the scalp (C_z) EEG electrode. Subtraction of a single subject specific percentage of the EOG from the EEG would correct for either artifact source, but different subtraction percentages must be used for each.     

Automatic Removal of Eye-Movement and Blink Artifacts from EEG Signals

Frequent occurrence of electrooculography (EOG) artifacts leads to serious problems in interpreting and analyzing the electroencephalogram (EEG). In this paper, a robust method is presented to automatically eliminate eye-movement and eye-blink artifacts from EEG signals. Independent Component Analysis (ICA) is used to decompose EEG signals into independent components. Moreover, the features of topographies and power spectral densities of those components are extracted to identify eye-movement artifact components, and a support vector machine (SVM) classifier is adopted because it has higher performance than several other classifiers. The classification results show that feature-extraction methods are unsuitable for identifying eye-blink artifact components, and then a novel peak detection algorithm of independent component (PDAIC) is proposed to identify eye-blink artifact components. Finally, the artifact removal method proposed here is evaluated by the comparisons of EEG data before and after artifact removal. The results indicate that the method proposed could remove EOG artifacts effectively from EEG signals with little distortion of the underlying brain signals.     

Is ocular voltage propagation to the electroencephalogram frequency dependent?

Conventional eye correction methods subtract portions (propagation coefficients; Bs ) of electrooculogram (EOG) voltages from the electroencephalogram (EEG). The frequency domain approach (FDA) uses different B s for different frequencies whereas the time domain approach (TDA) uses the same B s. To determine whether measured B s are dependent on frequency and whether one should employ frequency-dependent methods, 20 min of EEG from eye movement (EM) and blink data (24 participants) were recorded, and B s were calculated for eye movement ERPs of differing signal-to-noise ratios for frequency bands ranging from 0 to 40 Hz and compared. At high signal to noise, EM B s for different frequency bands did not differ, for both vertical and horizontal EOG, at all scalp sites tested. There were small differences in blink B s for different bands, but smaller than the margin of error of this analysis. This indicates that TDA may be more appropriate than FDA.     

Automatic correction of ocular artifacts in the EEG: a comparison of regression-based and component-based methods.

A variety of procedures have been proposed to correct ocular artifacts in the electroencephalogram (EEG), including methods based on regression, principal components analysis (PCA) and independent component analysis (ICA). The current study compared these three methods, and it evaluated a modified regression approach using Bayesian adaptive regression splines to filter the electrooculogram (EOG) before computing correction factors. We applied each artifact correction procedure to real and simulated EEG data of varying epoch lengths and then quantified the impact of correction on spectral parameters of the EEG. We found that the adaptive filter improved regression-based artifact correction. An automated PCA method effectively reduced ocular artifacts and resulted in minimal spectral distortion, whereas ICA correction appeared to distort power between 5 and 20 Hz. In general, reducing the epoch length improved the accuracy of estimating spectral power in the alpha (7.5-12.5 Hz) and beta (12.5-19.5 Hz) bands, but it worsened the accuracy for power in the theta (3.5-7.5 Hz) band and distorted time domain features. Results supported the use of regression-based and PCA-based ocular artifact correction and suggested a need for further studies examining possible spectral distortion from ICA-based correction procedures.     

Ocular artifacts in recording EEGs and event-related potentials II: Source dipoles and source components

Summary The source dipoles for blinks point radially whereas the source dipoles for saccades point tangentially, in the direction of the eye movement. This indicates that blink potentials are not generated by eye movements but by the eyelid sliding down over the positively charged cornea. Dipole source dipole analysis shows that the rider artifact at the onset of upward and lateral saccades is caused by the eyelid as it lags a little behind the eyes at the beginning of the movement. Dipole source analysis allows both the EEG and the EOG to be modeled simultaneously and EOG generators to be distinguished from nearby EEG generators. Ocular source components can be calculated from a principal component analysis of EEG and EOG recordings during blinks and saccades. The effectiveness of propagation factors, source dipoles and source components in removing ocular artifacts from EEG samples was assessed. The most effective correction procedure uses source components.This research was supported by a grant from the Medical Research Council of Canada (MA5465) and a NATO collaborative research grant (0330/88). The programming for the analysis of ocular source components was supported by a grant from the James S. McDonnell Foundation (Grant 90-174, principal investigator Edgar Zurif). Dick Mowrey provided programming assistance.     

家庭远程医疗监护数据的传输及处理

介绍了家庭远程医疗监护系统，利用VB编程通过Intemet实现了监护数据的远程传输，医疗中心对接收到的监护数据可以进行去噪、去干扰、特征提取等预处理。在此基础上，实现脑电信号的远程传输，并对医疗中心接收到的脑电信号进行独立成分分析(ICA)分离眼动干扰的处理，取得了较好的效果。     

EOG correction: a comparison of four methods

EOG correction is a class of techniques that account for ocular artifact in the electroencephalogram (EEG) by subtracting electrooculographic data from the EEG. The purpose of this study was to evaluate four of these correction techniques (Verleger, Gasser, & Mocks, 1982 [VGM]; Gratton, Coles, & Donchin, 1983 [GCD]; Semlitsch, Presslich, Schuster, & Anderer, 1986 [SPSA]; Croft & Barry, 2000 [CB]). Blinks, vertical eye movements (VEM), and horizontal eye movements (HEM) from 26 subjects were corrected using these techniques, and eye movement event-related potentials computed to aid validation. HEMs were corrected better by CB, VGM/GCD then SPSA, VEMs by CB, VGM/GCD then SPSA, and blinks by CB, SPSA, GCD and then VGM, with the advantage of CB substantial for blinks (eta2>.72), VEMs (eta2>.60), and HEMs (eta2>.27). It is argued that the CB procedure adequately accounts for ocular artifact in the EEG. Reasons for the limitations of the other procedures are discussed.     

Removal of ocular artifacts from the EEG: a comparison between time-domain regression method and adaptive filtering method using simulated data

We recently proposed an adaptive filtering (AF) method for removing ocular artifacts from EEG recordings. The method employs two parameters: the forgetting factor λ and the filter length M. In this paper, we first show that when λ = M = 1, the adaptive filtering method becomes equivalent to the widely used time-domain regression method. The role of λ (when less than one) is to deal with the possible non-stationary relationship between the reference EOG and the EOG component in the EEG. To demonstrate the role of M, a simulation study is carried out that quantitatively evaluates the accuracy of the adaptive filtering method under different conditions and comparing with the accuracy of the regression method. The results show that when there is a shape difference or a misalignment between the reference EOG and the EOG artifact in the EEG, the adaptive filtering method can be more accurate in recovering the true EEG by using an M larger than one (e.g. M = 2 or 3).     

A NEW METHOD FOR DETECTING EYE MOVEMENT IN SLEEP

A new method of detecting eye movements (EMs) during sleep is described. The method consists of an electromechanical measurement using micro-miniaturized silver cup electrodes. These electrodes, when placed on the eyelid, produce electro-oculographic (EOG) recordings similar to the usual electrical method. The eyelid method offers the advantage of a relatively “clean” recording showing only EMs and movement artifact, with no intermingling of EEG and EOG. Furthermore, the method is at least one and one half times as sensitive as the usual EOG technique. In addition to these two special advantages, it also offers the features of conventional methods, convenience of DC coupling, independence from signal converters, ease of analyzing EM directionality, and durability despite the small size of the electrodes. Fabrication of the electrodes, recording configurations, and simultaneous comparisons to both the usual EOG technique and to a strain gauge method are described.     

Automatic detection and operant reinforcement of slow potential shifts

A technique for the automatic detection and operant reinforcement of slow potential (SP) changes is described. The SP shift detection device contains 3 inhibit channels to control sources of potential artifact including: vertical EOG, horizontal EOG and high voltage EEG transients. Two EEG SP shift detection circuits allow the simultaneous analysis of positive and negative shifts. The operation and potential uses of the device are discussed.     

Removing electroencephalographic artifacts by blind source separation

Eye movements, eye blinks, cardiac signals, muscle noise, and line noise present serious problems for electroencephalographic (EEG) interpretation and analysis when rejecting contaminated EEG segments results in an unacceptable data loss. Many methods have been proposed to remove artifacts from EEG recordings, especially those arising from eye movements and blinks. Often regression in the time or frequency domain is performed on parallel EEG and electrooculographic (EOG) recordings to derive parameters characterizing the appearance and spread of EOG artifacts in the EEG channels. Because EEG and ocular activity mix bidirectionally, regressing out eye artifacts inevitably involves subtracting relevant EEG signals from each record as well. Regression methods become even more problematic when a good regressing channel is not available for each artifact source, as in the case of muscle artifacts. Use of principal component analysis (PCA) has been proposed to remove eye artifacts from multichannel EEG. However, PCA cannot completely separate eye artifacts from brain signals, especially when they have comparable amplitudes. Here, we propose a new and generally applicable method for removing a wide variety of artifacts from EEG records based on blind source separation by independent component analysis (ICA). Our results on EEG data collected from normal and autistic subjects show that ICA can effectively detect, separate, and remove contamination from a wide variety of artifactual sources in EEG records with results comparing favorably with those obtained using regression and PCA methods. ICA can also be used to analyze blink-related brain activity.     

Prevalence and Methods of Control of the Cephalic Skin Potential EEG Artifact

Prevalence of the cephalic skin potential (CSP) artifact was studied in 21 Ss during EEG recording of the Contingent Negative Variation (CNV), averaged evoked response (AER), and verbal free association test. Skin potential response and electro-oculogram (EOG) were also recorded. Subdermal pin electrodes and local anesthesia infusion were employed to eliminate the CSP artifact in the EEG. Results indicated that EEG recorded from subdermal pin electrodes or from locally anesthetized scalp was free of CSP artifact. The EEG recorded from subdermal pin electrodes demonstrated spontaneous potential shifts but appeared adequate for EEG recording of the CNV or the AER. Significant CSP artifact was demonstrated in the EEG of 10 of 21 Ss, both evoked by stimuli (10 Ss) and spontaneous (3 Ss). CSP artifact significantly increased CNV amplitude. CSP artifact significantly increased the AER late positive wave (P3) to infrequent tones. Studies of CNV and AER can be confounded by CSP artifact. Above techniques appear promising for recording EEG free of CSP artifact.     

The Correction of EOG Artifacts by Frequency Dependent and Frequency Independent Methods

Methods for correcting EEG which is contaminated by EOG artifacts, which are frequency dependent or frequency independent, were compared. EEG activity contaminated mainly by eye movements was treated separately from activity contaminated mainly by blinks. The statistical comparison was based on representative, real data incorporating differential effects over frequency. The influence of EOG activity on different derivations and bands, and the resulting need for correction of EOG artifacts were of interest too. Correction using a constant gain function proved to be consistently inferior to using a gain depending on frequency. For eye movements, differences were, however, not very large in practical terms, and this was true for all frequency bands. The correction of blinks by the time domain method (i.e., constant gain) may become misleading in the sub-delta band, which might have implications for very slow activity such as slow cortical potentials. Transfer functions not corrected for coherent EEG activity at the EOG electrodes overcompensated in the alpha and beta bands, in line with earlier results showing the need to incorporate coherent EEG activity into the determination of EOG-EEG transfer. It was found to be preferable to use sample average coefficients and gain functions, rather than individual ones, leading also to simpler computing. EOG correction is indispensable at frontal derivations, in particular for the delta and theta bands. It is advisable to correct at central and even at parietal derivations.     

Issues relating to the subtraction phase in EOG artefact correction of the EEG.

An important method for removing the effect of ocular artefact from the EEG is 'EOG correction'. This method estimates the proportion of ocular artefact that is in the EEG, and removes it by subtraction. To date, EOG correction research has focused on problems relating to the estimation of the correction coefficients. Using both mathematical rationale and empirical data, this paper addresses issues relating to the subtraction phase, such as the magnitude of error that can be expected due to EOG correction. Using ERP methodology, principal component and regression analyses, it is shown that the N1P2 complex propagates forward to the horizontal and radial (but not vertical) electrooculograms (EOG), and it is shown mathematically that this will result in EOG-correction error. Assuming an accurate estimate of ocular contamination of the EEG, maximal subtraction-phase error of the N1P2 complex was found to be a prefrontal attenuation of 15-22%, decreasing to central and occipital enhancements of 3-4% and 13-14%, respectively. The magnitude of this subtraction-phase error is compared to between-subject ERP variability and to error associated with EOG rejection (omitting data contaminated by ocular artefact). It is argued that such EOG correction error is small relative to both artefact rejection procedures and to normal variability found in ERP studies, and that it is less pernicious than artefact rejection procedures.     

Off-line removal of ocular artifacts from event-related potentials using a multiple linear regression model

A method for correction of event-related potentials (ERP) and cortical DC shifts, disturbed by eyeblink and eye movement potentials, is described. The correcting algorithm employs a multiple linear regression model with random regressors which prevents an incorrect calculation of the propagation factor when both ocular potentials and event-related cortical potentials occur together. This propagation factor is calculated for each event-related EEG record. Segmentation of the record into 2.56 s time intervals guarantees, moreover, calculation of different propagation factors for eyeblinks and eye movements within a single trial. The correcting algorithm is executed off-line with the propagation factors calculated from the experimental data proper. A correction is carried out only when the EOG has a significant influence on ERP. The application of the procedure is illustrated by individual examples.     

Removal of the Ocular Artifact from the EEG: A Comparison of Time and Frequency Domain Methods with Simulated and Real Data

Frequency-dependent transfer from EOG to EEG may be insufficiently accounted for by simple time domain regression methods (Gasser, Sroka, & Möcks, 1986; Woestenburg, Verbaten, & Slangen, 1983). In contrast, a multiple-lag time domain regression analysis, using lagged regression of EEG on EOG, must theoretically account for both frequency dependence and independence. Two data sets were constructed, in which the transfer from EOG to EEG was either frequency-independent (constant gain) or frequency-dependent. Subsequently, three different correction methods were applied: 1) a simple regression analysis in the time domain; 2) a multiple-lag regression analysis in the time domain; and 3) a regression analysis in the frequency domain. The major results were that, for data set 1, the three methods constructed the original EEG equally well. With data set 2, reconstruction of the original EEG was achieved reasonably well with the frequency domain method and the time domain multiple-lag method, but not with simple time domain regression. These three correction procedures were also applied to real data, consisting of concomitantly recorded EEG and high-variance EOG series. No appreciable differences in outcome of the three methods were observed, and estimated transfer parameters suggested that these data were marked by weak frequency dependence only, which can be accounted for by simple time domain regression (and also by the other two methods).