A comparative study of different references for EEG default mode network: The use of the infinity reference
Introduction
The choice of electroencephalograph (EEG) reference greatly influences the delineation and analysis of EEG scalp recordings, and has attracted much attention in brain electrophysiology research (Hagemann et al., 2001, Nunez et al., 1999). In EEG scalp recording, only the potential difference between two points can be measured, meaning that the use of an appropriate reference is vital (Geselowitz, 1998). Several different types of reference, including the vertex reference (CZ), the linked mastoids reference (LM), the average reference (AR) and the left mastoid reference (L), are currently used for EEG measurement. However, all of these references may introduce an undesired temporal bias since no neutral point exists on the body surface. Thus, the reference signal itself may involve physiological dynamic processes that will inevitably influence the data. Previous studies have examined the effects of reference choice on EEG data using several methods, including the estimation of the effect of head surface on recordings using AR (Jugnhöfer et al., 1999), the examination of coherence and reference signals (Nunez et al., 1997, Essl and Rappelsberger, 1998, Nunez and Srinivasan, 2006) and the investigation of brain asymmetry (Hagemann et al., 2001).
To entirely resolve the problems involved in using body surface points for referencing, a reference with neutral potential is required. Theoretically, a point at infinity is far from brain sources, and has an ideally neutral potential. Therefore, a point at infinity constitutes an ideal reference (infinity reference, IR). In 2001, Yao (2001) proposed a ‘reference electrode standardisation technique (REST)’ to approximately transform EEG data recorded with a scalp point reference to recordings using an infinity reference (IR; the software for REST transformation can be downloaded at www.neuro.uestc.edu.cn/rest; a simplified MATLAB version can be found in Supplementary materials). In recent years, the REST has been quantitatively validated through simulation studies with assumed neural sources in both a concentric three-sphere head model (Yao, 2001) and a realistic head model (Zhai and Yao, 2004). These studies have shown that data referenced with REST are more consistent with physiology than data referenced using traditional scalp references. This has been shown using a variety of techniques, including EEG spectral imaging (Yao et al., 2005), EEG coherence (Marzetti et al., 2007), brain evoked potentials (EP) and spatiotemporal analysis (Yao et al., 2007).
However, previously reported studies on EEG electrode reference effects have predominantly focussed on power spectra or spatiotemporal analysis in certain frequency bands. Relatively few studies have investigated the effects of reference choice on other EEG bands. Recently, an increasing number of studies have examined functional connectivity networks in the brain. In particular, a great deal of research attention has focussed on connectivity during the resting state (Buckner and Vincent, 2007, Pawela et al., 2008, Honey et al., 2009), because brain activity in the resting state (in the absence of task stimuli) plays a fundamental role in both simple and complex cognitive processes. Many researchers regard the resting state as the ‘default mode’, in terms of neural network functioning (Raichle et al., 2001, Raichle and Snyder, 2007). In recent years, the default mode network (DMN) has been primarily investigated using functional magnetic resonance imaging (fMRI) (Greicius et al., 2003, Friston, 1994, Buckner and Vincent, 2007), while few studies of the DMN have used EEG measures. Chen et al. (2008) reported an EEG DMN study, examining the spatial characteristics of power spectra in different resting-state EEG frequency bands. Previous studies in both human and animal brains have revealed that the DMN is a kind of ‘small world network’ with high cluster coefficient and short path length (Sporns et al., 2004, Stam et al., 2007, Bassett and Bullmore, 2006). The choice of electrode reference is a particularly vital issue when EEG-related techniques are used to investigate neural connectivity. As such, the systematic investigation of the effects of reference choice on connectivity measurements is an important research goal.
In this study, we conducted a simulation experiment to test the utility of REST in recovering a true EEG-related connectivity network and to evaluate the relative error (RE) introduced by AR, LM and L references. The four references were then applied to real resting-state EEG to investigate the differences among them, and the deviations caused by AR, LM and L compared with REST in measuring power spectra, coherence and DMN configuration for conventional frequency bands were assessed.
Section snippets
Reference electrode standardisation technique
REST is a novel method that builds a bridge between the body reference and the theoretical neutral reference at an infinity point (Yao, 2001, Yao et al., 2005). For an IR, the forward EEG calculation is given bywhere G is the transfer matrix referenced at infinity, only dependent on the head model, source configuration and electrode montage; S is the distributed source; and V is the scalp EEG recording with a reference at infinity generated by S. Scalp noise is not considered in this model
Simulation results
It is obvious from Fig. 1(a) that the different references resulted in different REs. REST generated the smallest error, with the mean RE of the whole dipole-pair 0.062%, while the mean RE reached 16.61% with L, and 9.73% and 15.47% with AR and LM, respectively. A Tukey’s test revealed significant differences for all pair-wise comparisons among these references (P = 0.002 for LM vs. L; P < 0.001 for other pairs).
The relative error after reducing the electrode number is shown in Fig. 1(b). It can be
Simulation analysis
The simulation results showed that the relative error of coherence was greatly reduced by REST. Irrespective of the dipole-pair locations, the RE level of the four references was significantly different. Although the RE introduced by AR was smaller compared with mastoid references (LM and L), the network configuration was still markedly distorted by AR, which was in striking contrast to REST, lacking notable distortions. These results strongly suggest that the effects of using different EEG
Conclusions
In this study, we investigated the effect of reference choice on simulated data as well as real resting-state EEG data, as measured by spectra, coherence and DMN connectivity. The results of the simulation revealed that RE was the smallest when REST was used, while AR, LM and L references introduced a substantially higher level of error. All spectral properties, including power values, spatial distribution and AWC, were found to be distinctly different when alternative references were adopted.
Acknowledgements
This work was supported by NSFC #60736029, NSFC #30525030, #60701015 and the 863 Project 2009AA02Z301. We thank the two anonymous reviewers for their constructive comments that improved the manuscript considerably.
References (41)
- et al.
Unrest at rest: the importance of default activity and spontaneous network correlations
NeuroImage
(2007) - et al.
EEG default mode network in the human brain: spectral regional field powers
NeuroImage
(2008) - et al.
The use of standardized infinity reference in EEG coherency studies
NeuroImage
(2007) - et al.
A theoretical and experimental study of high resolution EEG based on surface Laplacian and cortical imaging
Electroenceph Clin Neurophysiol
(1994) - et al.
EEG coherency I: statistics, reference electrode, volume conduction, Laplacians, cortical imaging, and interpretation at multiple scales
Electroenceph Clin Neurophysiol
(1997) - et al.
EEG coherency II: experimental comparisons of multiple measures
Clin Neurophysiol
(1999) - et al.
Topographical maps, sources localization inference, and the reference electrode: comments on a paper by Desmedt et al
Electroenceph Clin Neurophysiol
(1993) - et al.
Nonlinear multivariate analysis of neurophysiological signals 2005
Prog Neurobiol
(2005) - et al.
Mapping of scalp potentials by surface spline interpolation
Electroenceph Clin Neurophysiol
(1987) - et al.
A default mode of brain function: a brief history of an evolving idea
NeuroImage
(2007)