ArticlePDF Available

The Influence of Different EEG References on Scalp EEG Functional Network Analysis During Hand Movement Tasks


Abstract and Figures

Although scalp EEG functional networks have been applied to the study of motor tasks using electroencephalography (EEG), the selection of a suitable reference electrode has not been sufficiently researched. To investigate the effects of the original reference (REF-CZ), the common average reference (CAR), and the reference electrode standardization technique (REST) on scalp EEG functional network analysis during hand movement tasks, EEGs of 17 right-handed subjects performing self-paced hand movements were collected, and scalp functional networks [coherence (COH), phase-locking value (PLV), phase lag index (PLI)] with different references were constructed. Compared with the REF-CZ reference, the networks with CAR and REST references exhibited more significant increases in connectivity during the left-/right-hand movement preparation (MP) and movement execution (ME) stages. The node degree of the channel near the reference electrode was significantly reduced by the REF-CZ reference. CAR and REST both decreased this reference effect, REST more so than CAR. We confirmed that the choice of reference would affect the analysis of the functional network during hand movement tasks, and the REST reference can greatly reduce the effects of the online recording reference on the analysis of EEG connectivity.
Content may be subject to copyright.
published: 02 September 2020
doi: 10.3389/fnhum.2020.00367
Edited by:
Zhen Yuan,
University of Macau, China
Reviewed by:
Peng Xu,
University of Electronic Science and
Technology of China, China
Shiang Hu,
Joint China-Cuba Laboratory for
Frontier Research in Translational
Neurotechnology, China
Yuxia Hu
These authors have contributed
equally to this work and share first
Specialty section:
This article was submitted to
Brain Imaging and Stimulation,
a section of the journal
Frontiers in Human Neuroscience
Received: 19 March 2020
Accepted: 11 August 2020
Published: 02 September 2020
Zhang L, Wang P, Zhang R, Chen M,
Shi L, Gao J and Hu Y (2020) The
Influence of Different EEG References
on Scalp EEG Functional Network
Analysis During Hand
Movement Tasks.
Front. Hum. Neurosci. 14:367.
doi: 10.3389/fnhum.2020.00367
The Influence of Different EEG
References on Scalp EEG Functional
Network Analysis During Hand
Movement Tasks
Lipeng Zhang1,2† ,Peng Wang1,2† ,Rui Zhang1,2,Mingming Chen 1,2 ,Li Shi3,4,
Jinfeng Gao1,2 and Yuxia Hu1,2 *
1School of Electrical Engineering, Zhengzhou University, Zhengzhou, China, 2Henan Key Laboratory of Brain Science and
Brain-Computer Interface Technology, Zhengzhou, China, 3Department of Automation, Tsinghua University, Beijing, China,
4Beijing National Research Center for Information Science and Technology, Beijing, China
Although scalp EEG functional networks have been applied to the study of motor tasks
using electroencephalography (EEG), the selection of a suitable reference electrode has
not been sufficiently researched. To investigate the effects of the original reference (REF-
CZ), the common average reference (CAR), and the reference electrode standardization
technique (REST) on scalp EEG functional network analysis during hand movement
tasks, EEGs of 17 right-handed subjects performing self-paced hand movements
were collected, and scalp functional networks [coherence (COH), phase-locking value
(PLV), phase lag index (PLI)] with different references were constructed. Compared with
the REF-CZ reference, the networks with CAR and REST references exhibited more
significant increases in connectivity during the left-/right-hand movement preparation
(MP) and movement execution (ME) stages. The node degree of the channel near the
reference electrode was significantly reduced by the REF-CZ reference. CAR and REST
both decreased this reference effect, REST more so than CAR. We confirmed that the
choice of reference would affect the analysis of the functional network during hand
movement tasks, and the REST reference can greatly reduce the effects of the online
recording reference on the analysis of EEG connectivity.
Keywords: EEG, common average reference, reference electrode standardization technique, scalp EEG functional
network, hand movement tasks
Maintaining body movement ability is one of the most important functions of the brain, and
many researchers have devoted themselves to studying the mechanisms behind motor processes.
Electroencephalography (EEG) is a non-invasive approach with a high temporal resolution that is
widely used to study neural activity during motor tasks.
After Kornhuber and Deecke first discovered the ‘‘Bereitschaftspotential’’ (BP; Kornhuber
and Deecke, 1965; Shibasaki and Hallett, 2006), several studies on movement-related cortical
potentials (MRCP) have been reported (Tarkka and Hallett, 1991; Rong and Deecke, 1999;
Frontiers in Human Neuroscience | 1September 2020 | Volume 14 | Article 367
Zhang et al. Influence Electroencephalogram References Functional Network
Berchicci et al., 2016). For simple movements, the BP
initiates for 1–2 s and abruptly begins to increase at 0.5 s
before the movement onset over the frontal-central areas
of the scalp (Rong and Deecke). The power variation of
the relevant frequency bands during movement has been
extensively studied (Pfurtscheller, 1990; Pfurtscheller et al.,
2003; Bai et al., 2005). Pfurtscheller (1990) studied the
event-related desynchronization/synchronization (ERD/ERS) of
various frequency bands associated with voluntary movement.
Their research demonstrated ERD over contralateral and
ipsilateral motor cortices during the movement preparation
(MP) and movement execution (ME). Recent developments in
functional connectivity research provided a new method for
neuroimaging (Biswal et al., 1995). Complex brain networks
provide some quantitative indicators for the topology of brain
networks and have become important tools for describing
anatomical and functional brain connectivity (Bullmore and
Sporns, 2009; Rubinov and Sporns, 2010). Although some
questions remain to be solved, this new method will help us
understand the mechanisms behind motor processes (Sporns
et al., 2004). Recently, brain network methods have been
utilized to analyze changes in network connectivity between
brain regions (Popovych et al., 2016; Fleck et al., 2018). Brain
connectivity analysis has also been used as a feature extraction
method (Li et al., 2018b).
The choice of the reference electrode is an inevitable issue
in EEG research. But there is still debate about which is the
most suitable reference (Yao et al., 2019). All signals of EEG in
each electrode are obtained as the difference between the electric
potentials in its location and in the location of the reference
electrode. However, it is impossible to find a neutral reference
on the scalp or body. To solve this problem, many studies have
aimed to find a relatively non-active point on the body surface
(Offner, 1950; Yao, 2001; Hagemann et al., 2001). Common
online recording references include the FZ, CZ, PZ, OZ left/right
earlobe, and nose (Andrew and Pfurtscheller, 1996; Bruder et al.,
1997; Bas¸ar et al., 1998; Essl and Rappelsberger, 1998; Hu
et al., 2018a). Offline re-references mainly include the linked
ears/mastoids (Garneski and Steelman, 1958), the common
average reference (CAR; Offner, 1950), the reference electrode
standardization technique (REST; Yao, 2001) and the unified
referencing framework (rREST; Hu et al., 2018b). The family
of EEG unipolar references including REF-Cz, CAR, and REST,
etc. has the ‘‘no memory’’ property (Hu et al., 2019). Because of
this, we can always re-reference the dataset. Although the brain
network method has been applied to the study of motor tasks, the
problem of selecting an appropriate reference electrode in EEG
research has not been solved (Popovych et al., 2016; Storti et al.,
2016; Li et al., 2018b). Similarly, the problem of the selection
of the reference electrode also exists in the motor imagination
(MI) that have similar pattern of brain networks with the ME
(Zhang et al., 2015; Li et al., 2018a, 2019). Recently, some
studies investigated how different reference choices influence
scalp EEG functional connectivity using simulated EEG data
(Chella et al., 2016; Huang et al., 2017). These studies all pointed
out that different references significantly alter the topography
of EEG connectivity patterns. Although the influence of the
reference electrode on MRCP has been studied (Hu et al., 2017),
its effect on the brain network of motor tasks has not been
fully discussed.
Hand movement tasks rely on multiple brain regions working
together and consist of motor preparation and motor execution
processes. The present study aimed to investigate the influence of
different references (REF-CZ, CAR, and REST) on brain network
analysis for motor tasks. First, we divided motor tasks into
two phases based on ERD in the beta band: motor preparation
and motor execution. Functional brain networks were then
constructed with different re-referencing schemes in the beta
band. Finally, we compared connectivity variations and the
graph-theoretic measures of functional brain networks during
motor tasks using different re-referencing schemes.
A total of seventeen right-handed healthy subjects (one female;
mean age 26 years, range: 23–29 years) were recruited from
Zhengzhou University. They had a normal or corrected-
to-normal vision, and none of them had a history of
motor or neurological disease. Every subject was informed of
the experimental procedure and signed a letter of consent
before the experiment. The study was approved by the local
ethics committee for the Protection of Human Subjects at
Zhengzhou University.
Procedure and Task
Each subject was comfortably seated in front of a computer
screen and remained idle, with their hands, forearms, and elbows
resting on the armrest of the chair. During EEG recording, each
subject was asked to try to avoid eye movement, swallowing, and
unnecessary limb movements.
As shown in Figure 1, subjects performed self-paced hand
movement tasks that required intervals of more than 5 s between
each task. Subjects were free to choose their left or right
hands for each movement, but the number of left-/right-hand
movements was approximately the same. They were instructed
not to count the number of seconds in an interval, and we
emphasized the importance of movement intention immediately
before performing the movement. The recording was done in ten
4 min long blocks with intermittent 2 min breaks.
FIGURE 1 | The timeline of a run. Each trial began in the idle stage, in which
the subject rested his or her hands, forearms, and elbows on the armrest of a
chair and relaxed his or her hands. The subject performed each hand
movement task after taking more than 5 s of idle time. After the completion of
a task, the subject returned to the idle stage and prepared for the next task.
Frontiers in Human Neuroscience | 2September 2020 | Volume 14 | Article 367
Zhang et al. Influence Electroencephalogram References Functional Network
EEG Recording and Signal Preprocessing
EEG data were recorded by a Neuroscan NuAmps digital
amplifier system with 64 electrodes arranged in the standard
10–10 EEG configuration. All the brain regions were covered
by these electrodes. Two extended bipolar channels (BP3 and
BP4) were used to acquire the left and right arms’ EMG
(electromyography) signals. The EEG signals were acquired at a
sampling rate of 250 Hz with the REF-CZ as an online recording
reference, and the impedance of all electrodes was less than 5 k.
The REF-CZ was located between the CZ and CPZ electrodes.
Data preprocessing consisted of two parts: EMG data and EEG
data preprocessing. EMG data were used to obtain the onset time
of each hand movement for each trial. The EMG data were band
filtered by a basic finite impulse response filter with respective
cutoff frequencies of 6 and 50 Hz. We then calculated the energy
of the filtered data and set the proper threshold to detect the
onset time of movement. Next, we checked the records visually
to remove instances of false recognition. Finally, we recorded the
onset times in a TXT file (Hu et al., 2017).
The EEGLAB toolbox was used to preprocess and analyze the
EEG data (Delorme and Makeig, 2004). First, we imported the
time labels of movement onset into the data. Next, the raw EEG
data were visually inspected for eye-blinks and muscle artifacts.
Then, testing with CAR was conducted using the reref function
from the EEGLAB toolbox, and testing with REST was conducted
using the rest_refer function1.
To extract the frequency band of interest, the data were
band-pass filtered in 13–30 Hz. Then, taking the EMG onset as
0 ms, all EEG data of the three references were segmented into
5,000 ms epochs ranging from 3,000 ms before to 2,000 ms after
the onset of movement [i.e., (3,000, 2,000) ms].
For examining the ERD patterns in the beta band, we used
the FieldTrip toolbox for time-frequency analysis (Oostenveld
et al., 2011). Multitaper method (MTM) based Hanning tapers
were used to calculate the time-frequency representations (TFRs)
of power (Thomson, 1982). Time-frequency representations
(TFRs) were calculated using the data with the REST reference
due to the theoretical advantages of the REST reference
(Yao et al., 2005). As shown in Figure 2, our results
showed that ERD was observed in the central areas of
the contralateral hemisphere as early as about 1,000 ms
before movement onset and continued to enhance. ERD
was also observed soon after over the central area on the
ipsilateral hemisphere, and its distribution became bilaterally
symmetrical from 0 ms concerning movement onset. At about
1,000 ms, weak ERD was observed only on the ipsilateral
hemisphere, and ERS begins to appear in the contralateral
hemisphere. The results of the ERD distribution in the beta
band are consistent with previous research (Pfurtscheller and
Aranibar, 1977; Bai et al., 2005). To compute functional
network changes, the preprocessed signals were divided into
three stages according to the ERD change pattern: idle stage
[(3,000, 2,000) ms], MP stage [(1,000, 0) ms], and
ME stage [(0, 1,000) ms].
Brain Network Construction
This study used the following most commonly used connectivity
measurement methods.
(1) Coherence: The coherence, also referred to as the magnitude
squared coherence or coherence spectrum, between two
signals is their cross-spectral density function (Pereda et al.,
2005). Coherence is a widely-used measure for characterizing
linear dependence between a pair of stochastic processes, as
well as a quantitative measure of their phase consistency, and
may be viewed as the equivalent measure of cross-correlation
in the frequency domain. It is defined as
Sxx(f)Syy (f)(1)
where Sxy(f) is the cross-spectrum of the signals x(t) and
y(t), and Sxx(f) and Syy (f) are their respective self-spectra. The
COH ranges between 0 and 1: 0 COH 1.
(2) Phase-Locking Value: The phase-locking value (PLV) makes
use only of the relative phase difference (Lachaux et al.,
1999). It is defined as
PLV =1
where tis the time, and θ(tk) is the phase difference Ø1(tk) –
Ø2(tk) PLV measures the variability of this phase difference
at t. If the phase difference varies little across the trials, PLV
is close to 1; it is close to zero otherwise.
(3) Phase Lag Index: The phase lag index (PLI) was introduced
by Stam et al. (2007), aiming to deal with the problem of
volume conduction and active reference electrodes in the
assessment of functional connectivity. The central idea is to
discard phase differences that center around 0 mod π. It is
defined as
sign [1Ø(tk)]
The PLI ranges between 0 and 1: 0 PLI 1. A PLI of
0 indicates either no coupling or coupling with a phase
difference centered around 0 mod π. A PLI of 1 indicates
perfect phase locking at a value of Ø different from 0 mod
π. The stronger this non-zero phase locking, the larger
the PLI.
We adopted the above-mentioned three methods to construct
scalp functional networks for the idle stage, MP stage, and ME
stage of left-/right-hand movement tasks. For every method,
we calculated the connectivity value between per electrode
pairs and stored all the connectivity values in the adjacency
matrix in which the rows and columns are arranged in order
of channel number. Then, the adjacency matrix G of each
subject has been obtained by averaging the adjacency matrices
of all trials for left-/right-hand movement networks from
each reference and stage. Finally, 18 adjacency matrices of
functional networks of the left-/right-hand movement were
calculated from the data of three stages with three references for
each subject.
Frontiers in Human Neuroscience | 3September 2020 | Volume 14 | Article 367
Zhang et al. Influence Electroencephalogram References Functional Network
FIGURE 2 | The ground-average event-related desynchronization (ERD) patterns in the beta band (13–30 Hz) during the left-/right-hand movement. The baseline
was 3 to 2 s. Blue indicates ERD, yellow indicates event-related synchronization (ERS). Left, Left-hand movement; Right, Right-hand movement.
Network Measures
Graph theory provides a quantitative analysis method
for us to analyze complex brain networks. Commonly
used network parameters can be roughly divided into
local parameters (node degree, clustering coefficient, local
efficiency, and so on) and global parameters (characteristic
path length, global efficiency, and so on). Node degree
is the most fundamental network measure, and most
other measures are ultimately linked to the node degree
(Bullmore and Sporns, 2009). The characteristic path length
is a commonly used global parameter, and most other
global parameters are related to it. Therefore, this article
selected the two parameters of node degree and characteristic
path length.
Frontiers in Human Neuroscience | 4September 2020 | Volume 14 | Article 367
Zhang et al. Influence Electroencephalogram References Functional Network
(1) Degree: the degree of a node is the number of connections
involving it (Sporns et al., 2004). It can be defined as follows:
Di=Xj6=iAij (4)
where Diis the degree of node i, and Aij represents the
connectivity between node iand node j.
(2) Characteristic path length: path length is the minimum
number of edges that must be traversed to go from one node
to another. The average length of the shortest paths between
per pairs of nodes is the characteristic path length. Thus,
it is also called the shortest path length. It can be defined
as follows:
N(N1)Xi,jN,i6=jdij (5)
Where dij is the shortest path (geodesic) between iand j. Note
that dij =for all disconnected pairs i,j.
Statistical Analysis
To find out which edges had significantly increased connectivity
during the MP and ME stages compared to the idle stage,
the three networks of the left-/right-hand movement were
constructed with different references. To get the average
adjacency matrix for the three stages, we averaged all subjects’
adjacency matrices at each stage. Then the averaged adjacency
matrices of the idle stage were subtracted from the averaged
adjacency matrices during MP and ME stage. All elements in
these difference matrices greater than 0.02 are considered to have
significantly increased. To avoid the consideration of spurious
interactions, a one-tailed paired t-test was used to test the
differences in connectivity of all the subjects between two stages.
The sample size of the t-test was 17 since we had 17 subjects.
To reduce the false-positive rate, statistical tests were corrected
with the BHFDR method (Benjamini and Hochberg, 1995). If the
BHFDR-corrected significance level of an edge was less than 0.05,
we considered the corresponding connectivity to have increased
significantly and retained the edge, otherwise discard this edge.
To compare the network parameters of functional networks with
different references, the two-sample paired t-test was used to test
the differences in degree and characteristic path length between
three references (REF-CZ vs. CAR, REF-CZ vs. REST, and CAR
vs. REST). The sample size of the t-test was 17.
Effect of Reference Choice on COH, PLV,
and PLI Networks Analysis in Hand
Movement Task
The number of edges with significantly increased connectivity
during the MP stage and ME stage is shown in Table 1 (P<0.05,
BHFDR). A significant increase in connectivity can be observed
in COH and PLV networks. The networks with the REF-CZ
reference has far fewer edges than the networks with CAR and
REST reference during the ME stage. The networks with REST
reference has slightly more edges than the networks with CAR
reference in both the stages of MP and ME. There was almost no
TABLE 1 | The number of edges with significantly increased connectivity during
the movement preparation (MP) stage and movement execution (ME) stage.
Connectivity Movement stage The number of edges with
significantly increased connectivity
(P<0.05, BHFDR)
COH MP (L) 21 6 17
ME (L) 5 60 86
MP (R) 11 16 25
ME (R) 7 19 46
PLV MP (L) 28 24 30
ME (L) 10 85 99
MP (R) 10 21 38
ME (R) 2 46 65
PLI MP (L) 0 0 0
ME (L) 0 0 0
MP (R) 0 0 1
ME (R) 0 0 0
significant increase in connectivity in the PLI networks, which
may reflect that non-zero lag phase synchronization between
almost all electrode pairs does not increase significantly during
MP and ME stages in the beta band.
Figure 3 shows the edges of connectivity significantly
increased compared to the idle stage in the PLV and COH
networks with different references during left-/right-hand
movements (P<0.05, BHFDR). For the REF-CZ reference,
the connections between the contralateral respective areas of
the motor cortex can be observed during the MP stage, but
there are few connections between these areas during the ME
stage. For the CAR and REST references, not only can be
observed the connections between the contralateral respective
areas of the motor cortex during the MP stage, but also the
connections between the bilateral respective areas of the motor
cortex and the somatosensory cortex area during the ME stage.
The connections between the respective areas of the motor cortex
areas of the networks with REST reference were more than the
CAR reference network. The networks with CAR reference show
some non-motor cortex activities. Also, the left-hand movement
had significantly more significant connectivity increases than the
right-hand movement, which may be related to the fact that
our subjects were all right-handed. Ipsilateral motor cortical
activation could be due to additional internal effort required
for left-handed movement, which is more complex than right-
handed movement (Dhamala et al., 2003). This phenomenon
may also be explained as the left-brain dominance for motor
planning in humans (Sabate et al., 2004).
Effect of Reference Choice on Network
To examine the differences in the local network properties
of different reference networks, we calculated the degree
distribution of the brain network of 17 subjects. The topographic
map of the average nodal degree of the 17 subjects during the
left-hand movement execution (ME) stage is shown in Figure 4.
For the COH and PLV networks, the degree distribution of
networks with REF-CZ reference is different from CAR reference
and REST reference. Especially the degrees of the parietal near
Frontiers in Human Neuroscience | 5September 2020 | Volume 14 | Article 367
Zhang et al. Influence Electroencephalogram References Functional Network
FIGURE 3 | The significantly increased edges (P<0.05, FDR corrected) compared to the idle stage during the movement preparation (MP) stage and movement
execution (ME) stage in the beta band (13–30 Hz) from 17 subjects. The thickness of the line indicates the increase in connectivity. (A) Coherence (COH) network
changes pattern during left hand movement. (B) COH network changes pattern during right hand movement. (C) Phase-locking value (PLV) network changes
pattern during left hand movement. (D) PLV network changes pattern during right hand movement.
the reference electrode is very small. The degree distribution
of networks with CAR and REST reference is similar, and the
degree is mainly higher in the frontal. For the PLI networks,
the degrees near the reference electrode of network with the
REF-CZ reference is only slightly lower than the CAR and REST
references. Similar results were obtained in the networks of the
MP stage and ME stage of both the left and right hand.
The number of nodes with significant differences is shown
in Table 2 (P<0.05, t-test). For the COH and PLV networks,
the degrees of many nodes of networks with different references
are significantly different. The networks with REF-CZ reference
differ greatly from networks with the CAR and REST reference,
while the difference between the CAR and REST reference
networks is relatively small. For the PLI network, only a
few nodes have significant differences between networks with
different references.
For global network properties, we calculated the characteristic
path length of the brain network of the 17 subjects. By
comparing the characteristic path lengths among different
reference networks, no significant differences were found. The
average characteristic path length of the left-hand ME stage is
shown in Figure 5.
Frontiers in Human Neuroscience | 6September 2020 | Volume 14 | Article 367
Zhang et al. Influence Electroencephalogram References Functional Network
FIGURE 4 | The topographic map of node degree during ME. The black
crosses mark the location of the original reference REF-CZ.
To find a better reference electrode for hand movement task
studies based on EEG, we constructed the functional networks
of ten subjects with REF-CZ, CAR, and REST references. We
studied the influence of the reference choice on the results of
network analysis and graph theory indicators.
Many studies based on brain imaging technology had found
that the interactions of cortico-motor networks, including the
primary motor cortex (M1), the supplementary motor area
(SMA), the premotor cortex (PMC), and parietal cortex, are
strengthened during the MP stage and ME stage, and the
networks showed contralateral advantage (Jiang et al., 2004;
Wheaton et al., 2005; Wu et al., 2011; Popovych et al., 2016).
The interaction of these areas during the MP was related to the
planning of movement, and similar interaction could also be
observed in motor imagery (MI; Sabate et al., 2004; Hanakawa
et al., 2008). SMA and PMC were responsible for the selection,
planning and some aspects of motor control for movement or
the direct control of some movements and more involved in
planning or preparing for movement than M1 (Weinrich and
FIGURE 5 | The mean (±SD) values of the characteristic path length of
functional networks with different references.
TABLE 2 | The number of nodes with significant differences in node degree
during the MP stage and ME stage.
Connectivity Movement
The number of significantly
different nodes (P<0.05, t-test)
vs. CAR vs. REST vs. REST
COH MP (L) 43 43 25
ME (L) 45 42 26
MP (R) 42 43 21
ME (R) 45 43 27
PLV MP (L) 40 41 21
ME (L) 43 41 22
MP (R) 41 41 21
ME (R) 45 44 23
PLI MP (L) 2 1 2
ME (L) 1 0 1
MP (R) 1 1 1
ME (R) 0 0 1
Wise, 1982; Schluter et al., 1998; Picard and Strick, 2003; Nguyen
et al., 2014). The motor cortex showed stronger connectivity
during the ME stage than the MP stage. M1, which was
responsible for the execution of movement, played an important
role in movement networks during the ME stage (Jiang et al.,
2004). However, areas such as PMC and SMA also contributed to
the execution of movement. One study showed that the coupling
parameters among SMA, PMC and M1 increased with recovery
and predicted a better outcome in stroke patients suffering
from hand motor deficits (Rehme et al., 2011). Besides, the
parietal and sensory cortices were also activated during hand
movements (Okuda et al., 1995; Christoph et al., 2001), and
showed increased cortical coupling with the PMC and SMA
(Wheaton et al., 2005).
We confirmed that the choice of reference electrode
significantly influences the network analysis results. For the
REF-CZ reference, we could not observe obvious synchronized
activity between the respective areas of the motor cortex during
the ME stage where the M1 activity dominates, most network
connectivity increases are overwhelmed by reference effects. The
main reason for this phenomenon is that the M1 is close to
REF-CZ reference, and the electrical activity could be conducted
to the reference channel. The amplitude and phase information
of the channel signals in M1 are lost, resulting in low measured
synchronization and difficulty in distinguishing its changes.
Therefore, the original REF reference is not a good choice. Both
the CAR and REST references achieve much better results than
the REF-CZ reference. The significant increase in connectivity
between the respective areas of the motor cortex (i.e., PMC, SMA,
and M1) can be observed from these two reference networks,
and a clear contralateral advantage can be observed. Compared
with the CAR reference, the REST reference network shows more
pronounced activity in the motor cortex. The networks with CAR
reference show a lot of occipital activities. This phenomenon
has not been supported by relevant electrophysiological evidence
during voluntary movement and our experiments did not involve
any visual stimuli. It may show that the performance of the REST
reference is more similar to real brain activity than CAR on scalp
EEG functional network analysis.
Frontiers in Human Neuroscience | 7September 2020 | Volume 14 | Article 367
Zhang et al. Influence Electroencephalogram References Functional Network
For graph theory indicators, our results show that the local
properties of the network are also influenced by different
references. Beta rhythm is associated more with motor cortical
function, and tend to show maximum power in the frontal
and central areas (Hari and Salmelin, 1997). As mentioned
above, the synchrony between the respective areas of the motor
cortex will increase during the movement task. However, the
degrees of networks of these areas with REF-CZ reference is
small because the reference electrode is too close to the active
areas. Since the degree of any node is related to all nodes in
the network, the overall degree distribution will be affected by
the reference effect. This reference effect is greatly reduced in
the network with CAR and REST reference. As for the global
property of the network, the results among different references
are not significantly different. The characteristic path length,
as a global network attribute, reflects the overall information
transmission efficiency of the network (Bassett and Bullmore,
2006). Although different references may significantly affect the
connectivity between some nodes, they are still not enough to
greatly change the overall performance of the network.
For all three connectivity measures, different connectivity
calculation methods are affected by reference choice. In
this study, due to the lack of meaningful PLI results, the
connectivity results of PLI could not be compared. The degree
distribution of PLI networks is least affected by the reference.
This was expected because PLI only retains non-zero lag
correlation that is more likely to reflect the true correlations
of underlying sources and discards zero-lag correlation that
may be affected by volume conduction and reference effects
(Stam et al., 2007). Thus, the PLI is theoretically much less
affected by the influence of common sources and active reference
electrodes. However, there are the risks of missing functionally
meaningful correlations at zero lag in this approach. Compared
to COH, PLV uses only phase information and no amplitude
information. Therefore, PLV is slightly less affected by reference
effects than COH.
In previous studies of the effect of reference electrodes on RP,
the performance of the CAR reference and the REST reference
has similar results (Hu et al., 2017). In this study of the influence
of reference electrodes on motor functional brain networks,
REST references have shown superior performance to CAR. This
also illustrates that the phase information used in functional
network connectivity calculations is more sensitive to the impact
of reference effects than the amplitude information.
On the one hand, our study provides evidence for the reference
selection in the construction of the scalp EEG network during
movement tasks. On the other hand, the results further confirm
that the EEG reference plays an important role in data analysis in
neuroscience. For the study of movement Scalp EEG functional
networks, it is very important to choose the appropriate reference
electrode. Our results show that the reference has a great
impact on the connectivity changes of functional networks and
graph theory indicators in hand movement tasks. Researchers
may obtain different conclusions when functional networks
are constructed with different references. In particular, online
recording references located near the motor areas can lead to
large biases. According to our conclusion, the REST and CAR
reference can greatly reduce the effect of the online recording
reference location, while REST may be slightly better than CAR.
The raw data supporting the conclusions of this article will be
made available by the authors, without undue reservation.
The studies involving human participants were reviewed and
approved by the local ethics committee for the Protection of
Human Subjects at Zhengzhou University. Participants provided
their written informed consent to participate in the study.
LZ: conceptualization, methodology, software, and
writing—reviewing and editing. PW: conceptualization,
methodology, software, and writing—original draft preparation.
RZ: conceptualization, methodology, and data curation.
LS: investigation, funding acquisition, and supervision. JG:
investigation, supervision, and validation. MC: investigation,
and supervision. YH: conceptualization, project administration,
funding acquisition, investigation, and supervision.
This work was supported by the National Natural Science
Foundation of China #61603344 and 61803342, the Key Project
at Central Government Level #2060302, Program for Science and
Technology of Henan Province of China #182102210099, Science
and Technology Project of Henan Province #202102310210, and
Key Project of Discipline Construction of Zhengzhou University
We thank LetPub for its linguistic assistance
during the preparation of this manuscript.
Andrew, C., and Pfurtscheller, G. (1996). Dependence of coherence
measurements on EEG derivation type. Med. Biol. Eng. Comput. 34, 232–238.
doi: 10.1007/bf02520079
Bai, O., Mari, Z., Vorbach, S., and Hallett, M. (2005). Asymmetric
spatiotemporal patterns of event-related desynchronization preceding
voluntary sequential finger movements: a high-resolution EEG
study. Clin. Neurophysiol. 116, 1213–1221. doi: 10.1016/j.clinph.2005.
Frontiers in Human Neuroscience | 8September 2020 | Volume 14 | Article 367
Zhang et al. Influence Electroencephalogram References Functional Network
Bas¸ar, E., Rahn, E., Demiralp, T., and Schurmann, M. (1998). Spontaneous
EEG theta activity controls frontal visual evoked potential amplitudes.
Electroencephalogr. Clin. Neurophysiol. 108, 101–109. doi: 10.1016/s0168-
Bassett, D. S., and Bullmore, E. (2006). Small-world brain networks. Neuroscientist
12, 512–523. doi: 10.1177/1073858406293182
Benjamini, Y., and Hochberg, Y. (1995). Controlling the false discovery rate: a
practical and powerful approach to multiple testing. J. R. Stat. Soc. Series B 57,
289–300. doi: 10.1111/j.2517-6161.1995.tb02031.x
Berchicci, M., Spinelli, D., and Russo, F. D. (2016). New insights into old waves.
Matching stimulus- and response-locked ERPs on the same time-window. Biol.
Psychol. 117, 202–215. doi: 10.1016/j.biopsycho.2016.04.007
Biswal, B., Yetkin, F. Z., Haughton, V. M., and Hyde, J. S. (1995).
Functional connectivity in the motor cortex of resting human brain using
echo-planar mri. Magn. Reson. Med. 34, 537–541. doi: 10.1002/mrm.19103
Bruder, G. E., Fong, R., Tenke, C. E., Leite, P., Towey, J. P., Stewart, J. E., et al.
(1997). Regional brain asymmetries in major depression with or without an
anxiety disorder: a quantitative electroencephalographic study. Biol. Psychiatry
41, 939–948. doi: 10.1016/s0006-3223(96)00260-0
Bullmore, E., and Sporns, O. (2009). Complex brain networks: graph theoretical
analysis of structural and functional systems. Nat. Rev. Neurosci. 10, 186–198.
doi: 10.1038/nrn2618
Chella, F., Pizzella, V., Zappasodi, F., and Marzetti, L. (2016). Impact of
the reference choice on scalp EEG connectivity estimation. J. Neural Eng.
13:036016. doi: 10.1088/1741-2560/13/3/036016
Christoph, B., Udo, H., Renate, S., Katja, W., Niels, B., and Helge, T. (2001).
Dynamic organization of the somatosensory cortex induced by motor activity.
Brain 124, 2259–2267. doi: 10.1093/brain/124.11.2259
Delorme, A., and Makeig, S. (2004). EEGLAB: an open source toolbox for
analysis of single-trial EEG dynamics including independent component
analysis. J. Neurosci. Methods 134, 9–21. doi: 10.1016/j.jneumeth.2003.
Dhamala, M., Pagnoni, G., Wiesenfeld, K., Zink, C. F., Martin, M., and Berns, G. S.
(2003). Neural correlates of the complexity of rhythmic finger tapping.
NeuroImage 20, 918–926. doi: 10.1016/s1053-8119(03)00304-5
Essl, M., and Rappelsberger, P. (1998). EEG cohererence and reference signals:
experimental results and mathematical explanations. Med. Biol. Eng. Comput.
36, 399–406. doi: 10.1007/BF02523206
Fleck, J. I., Olsen, R., Tumminia, M., DePalma, F., Berroa, J., Vrabel, A., et al.
(2018). Changes in brain connectivity following exposure to bilateral eye
movements. Brain Cogn. 123, 142–153. doi: 10.1016/j.bandc.2018.03.009
Garneski, T. M., and Steelman, H. F. (1958). Equalizing ear reference resistance
in monopolar recording to eliminate artifactual temporal lobe asymmetry.
Electroencephalogr. Clin. Neurophysiol. 10, 736–738. doi: 10.1016/0013-
Hagemann, D., Naumann, E., and Thayer, J. F. (2001). The quest for the EEG
reference revisited: a glance from brain asymmetry research. Psychophysiology
38, 847–857. doi: 10.1111/1469-8986.3850847
Hanakawa, T., Dimyan, M. A., and Hallett, M. (2008). Motor planning,
imagery and execution in the distributed motor network: a time-course
study with functional MRI. Cereb. Cortex 18, 2775–2788. doi: 10.1093/cercor/
Hari, R., and Salmelin, R. (1997). Human cortical oscillations: a neuromagnetic
view through the skull. Trends Neurosci. 20, 44–49. doi: 10.1016/s0166-
Hu, S., Lai, Y., Valdes-Sosa, P. A., Bringas-Vega, M. L., and Yao, D. (2018a).
How do reference montage and electrodes setup affect the measured scalp EEG
potentials? J. Neural Eng. 15:026013. doi: 10.1088/1741-2552/aaa13f
Hu, S., Yao, D., and Valdes-Sosa, P. A. (2018b). Unified bayesian estimator of EEG
reference at infinity: rREST (regularized reference electrode standardization
technique). Front. Neurosci. 12:297. doi: 10.3389/fnins.2018.00297
Hu, S., Yao, D., Bringas-Vega, M. L., Qin, Y., and Valdes-Sosa, P. A. (2019). The
statistics of EEG unipolar references: derivations and properties. Brain Topogr.
32, 696–703. doi: 10.1007/s10548-019-00706-y
Hu, Y., Zhang, L., Chen, M., Li, X., and Shi, L. (2017). How electroencephalogram
reference influences the movement readiness potential? Front. Neurosci. 11,
1–7. doi: 10.3389/fnins.2017.00683
Huang, Y., Zhang, J., Cui, Y., Yang, G., He, L., Liu, Q., et al. (2017). How different
EEG references influence sensor level functional connectivity graphs. Front.
Neurosci. 11:368. doi: 10.3389/fnins.2017.00368
Jiang, T., He, Y., Zang, Y., and Weng, X. (2004). Modulation of functional
connectivity during the resting state and the motor task. Hum. Brain Mapp.
22, 63–71. doi: 10.1002/hbm.20012
Kornhuber, H. H., and Deecke, L. (1965). Changes in the brain potential in
voluntary movements and passive movements in man: readiness potential and
reafferent potentials. Pflugers Arch. Gesamte Physiol. Menschen Tiere 284, 1–17.
Lachaux, J. P., Rodriguez, E., Martinerie, J., and Varela, F. J. J. (1999).
Measuring phase synchrony in brain signals. Hum. Brain Mapp. 8, 194–208.
doi: 10.1002/(sici)1097-0193(1999)8:4<194::aid-hbm4>;2-c
Li, F., Peng, W., Jiang, Y., Song, L., Liao, Y., Yi, C., et al. (2019). The
dynamic brain networks of motor imagery: time-varying causality analysis
of scalp EEG. Int. J. Neural Syst. 29:1850016. doi: 10.1142/s01290657185
Li, F., Yi, C., Song, L., Jiang, Y., Peng, W., Si, Y., et al. (2018a). Brain network
reconfiguration during motor imagery revealed by a large-scale network
analysis of scalp EEG. Brain Topogr. 32, 304–314. doi: 10.1007/s10548-018-
Li, T., Xue, T., Wang, B., and Zhang, J. (2018b). Decoding voluntary movement of
single hand based on analysis of brain connectivity by using EEG signals. Front.
Neurosci. 12:381. doi: 10.3389/fnhum.2018.00381
Picard, N., and Strick, P. L. (2003). Activation of the supplementary motor area
(SMA) during performance of visually guided movements. Cereb. Cortex 13,
977–986. doi: 10.1093/cercor/13.9.977
Nguyen, V. T., Breakspear, M., and Cunnington, R. (2014). Reciprocal interactions
of the SMA and cingulate cortex sustain premovement activity for voluntary
actions. J. Neurosci. 34, 16397–16407. doi: 10.1523/JNEUROSCI.2571-14.2014
Offner, F. F. (1950). The EEG as potential mapping: the value of the average
monopolar reference. Electroencephalogr. Clin. Neurophysiol. 2, 213–214.
doi: 10.1016/0013-4694(50)90040-x
Okuda, B., Tanaka, H., Tomino, Y., Kawabata, K., Tachibana, H., and Sugita, M.
(1995). The role of the left somatosensory cortex in human hand movement.
Exp. Brain Res. 106, 493–498. doi: 10.1007/bf00231073
Oostenveld, R., Fries, P., Maris, E., and Schoffelen, J. M. (2011). FieldTrip:
open source software for advanced analysis of MEG, EEG and invasive
electrophysiological data. Comput. Intell. Neurosci. 2011:156869.
doi: 10.1155/2011/156869
Pereda, E., Quiroga, R. Q., and Bhattacharya, J. (2005). Nonlinear multivariate
analysis of neurophysiological signals. Prog. Neurobiol. 77, 1–37. doi: 10.1016/j.
Pfurtscheller, G. (1990). Event-related EEG desynchronization.
Electroencephalogr. Clin. Neurophysiol. 75, S117–S118.
Pfurtscheller, G., and Aranibar, A. (1977). Event-related cortical
desynchronization detected by power measurements of scalp EEG.
Electroencephalogr. Clin. Neurophysiol. 42, 817–826. doi: 10.1016/0013-
Pfurtscheller, G., Graimann, B., Huggins, J. E., Levine, S. P., and Schuh, L. A.
(2003). Spatiotemporal patterns of beta desynchronization and gamma
synchronization in corticographic data during self-paced movement.
Clin. Neurophysiol. 114, 1226–1236. doi: 10.1016/s1388-2457(03)
Popovych, S., Rosjat, N., T’Oth, T. I., Wang, B. A., Liu, L., Abdollahi, R. O., et al.
(2016). Movement-related phase locking in the delta-theta frequency band.
NeuroImage 139, 439–449. doi: 10.1016/j.neuroimage.2016.06.052
Rehme, A. K., Eickhoff, S. B., Wang, L. E., Fink, G. R., and Grefkes, C. (2011).
Dynamic causal modeling of cortical activity from the acute to the chronic
stage after stroke. NeuroImage 55, 1147–1158. doi: 10.1016/j.neuroimage.2011.
Rong, Q. C., and Deecke, L. (1999). High resolution DC-eeg analysis of the
bereitschaftspotential and post movement onset potentials accompanying uni-
or bilateral voluntary finger movements. Brain Topogr. 11, 233–249.
Rubinov, M., and Sporns, O. (2010). Complex network measures of brain
connectivity: uses and interpretations. NeuroImage 52, 1059–1069.
doi: 10.1016/j.neuroimage.2009.10.003
Sabate, M., Gonzalez, B., and Rodriguez, M. (2004). Brain lateralization of motor
imagery: motor planning asymmetry as a cause of movement lateralization.
Frontiers in Human Neuroscience | 9September 2020 | Volume 14 | Article 367
Zhang et al. Influence Electroencephalogram References Functional Network
Neuropsychologia 42, 1041–1049. doi: 10.1016/j.neuropsychologia.2003.
Schluter, N. D., Rushworth, M. F., Passingham, R. E., and Mills, K. R.
(1998). Temporary interference in human lateral premotor cortex suggests
dominance for the selection of movements. A study using transcranial magnetic
stimulation. Brain 121, 785–799. doi: 10.1093/brain/121.5.785
Shibasaki, H., and Hallett, M. (2006). What is the Bereitschaftspotential? Clin.
Neurophysiol. 117, 2341–2356. doi: 10.1016/j.clinph.2006.04.025
Sporns, O., Chialvo, D. R., Kaiser, M., and Hilgetag, C. C. (2004). Organization,
development and function of complex brain networks. Trends Cogn. Sci. 8,
418–425. doi: 10.1016/j.tics.2004.07.008
Stam, C. J., Nolte, G., and Daffertshofer, A. (2007). Phase lag index:
assessment of functional connectivity from multi channel EEG and MEG with
diminished bias from common sources. Hum. Brain Mapp. 28, 1178–1193.
doi: 10.1002/hbm.20346
Storti, S. F., Formaggio, E., Manganotti, P., and Menegaz, G. (2016). Brain network
connectivity and topological analysis during voluntary arm movements. Clin.
EEG Neurosci. 47, 276–290. doi: 10.1177/1550059415598905
Tarkka, I. M., and Hallett, M. (1991). Topography of scalp-recorded motor
potentials in human finger movements. J. Clin. Neurophysiol. 8, 331–341.
doi: 10.1097/00004691-199107010-00009
Thomson, D. J. (1982). Spectrum estimation and harmonic analysis. Proc. IEEE 70,
1055–1096. doi: 10.1109/proc.1982.12433
Weinrich, M., and Wise, S. P. (1982). The premotor cortex of the
monkey. J. Neurosci. 2, 1329–1345. doi: 10.1523/JNEUROSCI.02-09-01
Wheaton, L. A., Nolte, G., Bohlhalter, S., Fridman, E., and Hallett, M. (2005).
Synchronization of parietal and premotor areas during preparation and
execution of praxis hand movements. Clin. Neurophysiol. 116, 1382–1390.
doi: 10.1016/j.clinph.2005.01.008
Wu, T., Wang, L., Hallett, M., Chen, Y., Li, K., and Chan, P. (2011). Effective
connectivity of brain networks during self-initiated movement in Parkinson’s
disease. NeuroImage 55, 204–215. doi: 10.1016/j.neuroimage.2010.11.074
Yao, D. (2001). A method to standardize a reference of scalp EEG recordings to a
point at infinity. Physiol. Meas. 22, 693–711. doi: 10.1088/0967-3334/22/4/305
Yao, D., Li, W., Oostenveld, R., Nielsen, K. D., Arendt-Nielsen, L., and
Chen, A. C. N. (2005). A comparative study of different references for EEG
spectral mapping: the issue of the neutral reference and the use of the infinity
reference. Physiol. Meas. 26, 173–184. doi: 10.1088/0967-3334/26/3/003
Yao, D., Qin, Y., Hu, S., Dong, L., Bringas Vega, M. L., and Valdes Sosa, P. A.
(2019). Which reference should we use for EEG and ERP practice? Brain
Topogr. 32, 530–549. doi: 10.1007/s10548-019-00707-x
Zhang, R., Yao, D., Valdes-Sosa, P. A., Li, F., Li, P., Zhang, T., et al. (2015). Efficient
resting-state EEG network facilitates motor imagery performance. J. Neural
Eng. 12:066024. doi: 10.1088/1741-2560/12/6/066024
Conflict of Interest: The authors declare that the research was conducted in the
absence of any commercial or financial relationships that could be construed as a
potential conflict of interest.
The reviewer PX declared a past co-authorship with one of the authors RZ to the
handling Editor.
Copyright © 2020 Zhang, Wang, Zhang, Chen, Shi, Gao and Hu. This is an
open-access article distributed under the terms of the Creative Commons Attribution
License (CC BY). The use, distribution or reproduction in other forums is permitted,
provided the original author(s) and the copyright owner(s) are credited and that the
original publication in this journal is cited, in accordance with accepted academic
practice. No use, distribution or reproduction is permitted which does not comply
with these terms.
Frontiers in Human Neuroscience | 10 September 2020 | Volume 14 | Article 367
... Patients in the MCS group had a higher degree of nodal value in the right-frontal area. The results were consistent with the enhancement of the COH brain network connection in the motor-related brain area during the motor preparation process reported by our previous study (Zhang et al. 2020). The statistical results showed that the nodes with significant differences between the three groups were mainly distributed in the motor-related brain areas. ...
Full-text available
Recent achievements in evaluating the residual consciousness of patients with disorders of consciousness (DOCs) have demonstrated that spontaneous or evoked electroencephalography (EEG) could be used to improve consciousness state diagnostic classification. Recent studies showed that the EEG signal of the task-state could better characterize the conscious state and cognitive ability of the brain, but it has rarely been used in consciousness assessment. A cue-guide motor task experiment was designed, and task-state EEG were collected from 18 patients with unresponsive wakefulness syndrome (UWS), 29 patients in a minimally conscious state (MCS), and 19 healthy controls. To obtain the markers of residual motor function in patients with DOC, the event-related potential (ERP), scalp topography, and time-frequency maps were analyzed. Then the coherence (COH) and debiased weighted phase lag index (dwPLI) networks in the delta, theta, alpha, beta, and gamma bands were constructed, and the correlations of network properties and JFK Coma Recovery Scale-Revised (CRS-R) motor function scores were calculated. The results showed that there was an obvious readiness potential (RP) at the Cz position during the motor preparation process in the MCS group, but no RP was observed in the UWS group. Moreover, the node degree properties of the COH network in the theta and alpha bands and the global efficiency properties of the dwPLI network in the theta band were significantly greater in the MCS group compared to the UWS group. The above network properties and CRS-R motor function scores showed a strong linear correlation. These findings demonstrated that the brain network properties of task-state EEG could be markers of residual motor function of DOC patients. Supplementary information: The online version contains supplementary material available at 10.1007/s11571-021-09741-7.
... Overall absolute reliability of graph measures remained mainly below 5%, which is interpreted as a good value in exercise science contexts of repeated measures 23 . The different modulations of brain graph reliability depending on FC estimator appear reasonable when the nature of the respective FC estimation method is considered and are in line with previous network analysis studies observing differences between Coh and PLI networks in different populations 24,25 . Whereas the wPLI detects non-zero phase lags and is less prone to spurious connections based on volume conduction and common sources, Coh-based functional networks rely on signal amplitude and volume conduction and therefore even resemble zero-phase lags in the signal 26 . ...
Full-text available
The interaction of acute exercise and the central nervous system evokes increasing interest in interdisciplinary research fields of neuroscience. Novel approaches allow to monitor large‑scale brain networks from mobile electroencephalography (EEG) applying graph theory, but it is yet uncertain whether brain graphs extracted after exercise are reliable. We therefore aimed to investigate brain graph reliability extracted from resting state EEG data before and after submaximal exercise twice within one week in male participants. To obtain graph measures, we extracted global small‑world‑index (SWI), clustering coefficient (CC) and characteristic path length (PL) based on weighted phase leg index (wPLI) and spectral coherence (Coh) calculation. For reliability analysis, Intraclass‑Correlation‑Coefficient (ICC) and Coefficient of Variation (CoV) were computed for graph measures before (REST) and after POST) exercise. Overall results revealed poor to excellent measures at PRE and good to excellent ICCs at POST in the theta, alpha‑1 and alpha‑2, beta‑1 and beta‑2 frequency band. Based on bootstrap‑analysis, a positive effect of exercise on reliability of wPLI based measures was observed, while exercise induced a negative effect on reliability of Coh‑based graph measures. Findings indicate that brain graphs are a reliable tool to analyze brain networks in exercise contexts, which might be related to the neuroregulating effect of exercise inducing functional connections within the connectome. Relative and absolute reliability demonstrated good to excellent reliability after exercise. Chosen graph measures may not only allow analysis of acute, but also longitudinal studies in exercise‑scientific contexts.
Full-text available
Which reference is appropriate for the scalp ERP and EEG studies? This unsettled problem still inspires unceasing debate. The ideal reference should be the one with zero or constant potential but unfortunately it is well known that no point on the body fulfills this condition. Consequently, more than ten references are used in the present EEG-ERP studies. This diversity seriously undermines the reproducibility and comparability of results across laboratories. A comprehensive review accompanied by a brief communication with rigorous derivations and notable properties (Hu et al. Brain Topogr, 2019. ) is thus necessary to provide application-oriented principled recommendations. In this paper current popular references are classified into two categories: (1) unipolar references that construct a neutral reference, including both online unipolar references and offline re-references. Examples of unipolar references are the reference electrode standardization technique (REST), average reference (AR), and linked-mastoids/ears reference (LM); (2) non-unipolar references that include the bipolar reference and the Laplacian reference. We show that each reference is derived with a different assumption and serves different aims. We also note from (Hu et al. 2019) that there is a general form for the reference problem, the 'no memory' property of the unipolar references, and a unified estimator for the potentials at infinity termed as the regularized REST (rREST) which has more advantageous statistical evidence than AR. A thorough discussion of the advantages and limitations of references is provided with recommendations in the hope to clarify the role of each reference in the ERP and EEG practice.
Full-text available
In this brief communication, which complements the EEG reference review (Yao et al. in Brain Topogr, 2019), we provide the mathematical derivations that show: (1) any EEG reference admits the general form of a linear transformation of the ideal multichannel EEG potentials with reference to infinity; (2) the average reference (AR), the reference electrode standardization technique (REST), and its regularized version (rREST) are solving the linear inverse problems that can be derived from both the maximum likelihood estimate (MLE) and the Bayesian theory; however, REST is based on more informative prior/constraint of volume conduction than that of AR; (3) we show for the first time that REST is also a unipolar reference (UR), allowing us to define a general family of URs with unified notations; (4) some notable properties of URs are ‘no memory’, ‘rank deficient by 1’, and ‘orthogonal projector centering’; (5) we also point out here, for the first time, that rREST provides the optimal interpolating function that can be used when the reference channel is missing or the ‘bad’ channels are rejected. The derivations and properties imply that: (a) any two URs can transform to each other and referencing with URs multiple times will not accumulate artifacts; (b) whatever URs the EEG data was previously transformed with, the minimum norm solution to the reference problem will be REST and AR with and without modeling volume conduction, respectively; (c) the MLE and the Bayesian theory show the theoretical optimality of REST. The advantages and limitations of AR and REST are discussed to guide readers for their proper use.
Full-text available
Mentally imagining rather physically executing the motor behaviors is defined as motor imagery (MI). During MI, the mu rhythmical oscillation of cortical neurons is the event-related desynchronization (ERD) subserving the physiological basis of MI-based brain-computer interface. In our work, we investigated the specific brain network reconfiguration from rest idle to MI task states, and also probed the underlying relationship between the brain network reconfiguration and MI related ERD. Findings revealed that comparing to rest state, the MI showed the enhanced motor area related linkages and the deactivated activity of default mode network. In addition, the reconfigured network index was closely related to the ERDs, i.e., the higher the reconfigured network index was, the more obvious the ERDs were. These findings consistently implied that the reconfiguration from rest to task states underlaid the reallocation of related brain resources, and the efficient brain reconfiguration corresponded to a better MI performance, which provided the new insights into understanding the mechanism of MI as well as the potential biomarker to evaluate the rehabilitation quality for those patients with deficits of motor function.
Full-text available
Research about decoding neurophysiological signals mainly aims to elucidate the details of human motion control from the perspective of neural activity. We performed brain connectivity analysis with EEG to propose a brain functional network (BFN) and used a feature extraction algorithm for decoding the voluntary hand movement of a subject. By analyzing the characteristic parameters obtained from the BFN, we extracted the most important electrode nodes and frequencies for identifying the direction of movement of a hand. The results demonstrated that the most sensitive EEG components were for frequencies delta, theta, and gamma1 from electrodes F4, F8, C3, Cz, C4, CP4, T3, and T4. Finally, we proposed a model for decoding voluntary movement of the right hand by using a hierarchical linear model (HLM). Through a voluntary hand movement experiment in a spiral trajectory, the Poisson coefficient between the measurement trajectory and the decoding trajectory was used as a test standard to compare the HLM with the traditional multiple linear regression model. It was found that the decoding model based on the HLM obtained superior results. This paper contributes a feature extraction method based on brain connectivity analysis that can mine more comprehensive feature information related to a specific mental state of a subject. The decoding model based on the HLM possesses a strong structure for data manipulation that facilitates precise decoding.
Full-text available
The choice of reference for the electroencephalogram (EEG) is a long-lasting unsolved issue resulting in inconsistent usages and endless debates. Currently, both the average reference (AR) and the reference electrode standardization technique (REST) are two primary, apparently irreconcilable contenders. We propose a theoretical framework to resolve this reference issue by formulating both (a) estimation of potentials at infinity, and (b) determination of the reference, as a unified Bayesian linear inverse problem, which can be solved by maximum a posterior estimation. We find that AR and REST are very particular cases of this unified framework: AR results from biophysically non-informative prior; while REST utilizes the prior based on the EEG generative model. To allow for simultaneous denoising and reference estimation, we develop the regularized versions of AR and REST, named rAR and rREST, respectively. Both depend on a regularization parameter that is the noise to signal variance ratio. Traditional and new estimators are evaluated with this framework, by both simulations and analysis of real resting EEGs. Toward this end, we leverage the MRI and EEG data from 89 subjects which participated in the Cuban Human Brain Mapping Project. Generated artificial EEGs—with a known ground truth, show that relative error in estimating the EEG potentials at infinity is lowest for rREST. It also reveals that realistic volume conductor models improve the performances of REST and rREST. Importantly, for practical applications, it is shown that an average lead field gives the results comparable to the individual lead field. Finally, it is shown that the selection of the regularization parameter with Generalized Cross-Validation (GCV) is close to the “oracle” choice based on the ground truth. When evaluated with the real 89 resting state EEGs, rREST consistently yields the lowest GCV. This study provides a novel perspective to the EEG reference problem by means of a unified inverse solution framework. It may allow additional principled theoretical formulations and numerical evaluation of performance.
Full-text available
Objective: Human scalp electroencephalogram (EEG) is widely applied in cognitive neuroscience and clinical studies due to its non-invasiveness and ultra-high time resolution. However, the representativeness of the measured EEG potentials for the underneath neural activities is still a problem under debate. This study aims to investigate systematically how both reference montage and electrodes setup affect the accuracy of EEG potentials. Approach: First, the standard EEG potentials are generated by the forward calculation with a single dipole in the neural source space, for eleven channel numbers (10, 16, 21, 32, 64, 85, 96, 128, 129, 257, 335). Here, the reference is the ideal infinity implicitly determined by forward theory. Then, the standard EEG potentials are transformed to recordings with different references including five monopolar references (Left earlobe, Fz, Pz, Oz, Cz), and three re-references (Linked Mastoids (LM), Average Reference (AR) and Reference Electrode Standardization Technique (REST)). Finally, the relative errors between the standard EEG potentials and the transformed ones are evaluated in terms of channel number, scalp regions, electrodes layout, dipole source position and orientation, as well as sensor noise and head model. Main results: Mono-polar reference recordings are usually of large distortions; thus, a re-reference after online mono-polar recording should be adopted in general to mitigate this effect. Among the three re-references, REST is generally superior to AR for all factors compared, and LM performs worst. REST is insensitive to head model perturbation. AR is subject to electrodes coverage and dipole orientation but no close relation with channel number. Significance: These results indicate that REST would be the first choice of re-reference and AR may be an alternative option for high level sensor noise case. Our findings may provide the helpful suggestions on how to obtain the EEG potentials as accurately as possible for cognitive neuroscientists and clinicians.
Full-text available
Readiness potential (RP) based on electroencephalograms (EEG) has been studied extensively in recent years, but no studies have investigated the influence of the reference electrode on RP. In order to investigate the reference effect, 10 subjects were recruited and the original vertex reference (Cz) was used to record the raw EEG signal when the subjects performed a motor preparation task. The EEG was then transformed to the common average reference (CAR) and reference electrode standardization technique (REST) reference, and we analyzed the RP waveform and voltage topographies and calculated the classification accuracy of idle and RP EEG segments. Our results showed that the RP waveform and voltage topographies were greatly influenced by the reference, but the classification accuracy was less affected if proper channels were selected as features. Since the Cz channel is near the primary motor cortex, where the source of RP is located, using the REST and CAR references is recommended to get accurate RP waveforms and voltage topographies.
The common approach to the multiplicity problem calls for controlling the familywise error rate (FWER). This approach, though, has faults, and we point out a few. A different approach to problems of multiple significance testing is presented. It calls for controlling the expected proportion of falsely rejected hypotheses — the false discovery rate. This error rate is equivalent to the FWER when all hypotheses are true but is smaller otherwise. Therefore, in problems where the control of the false discovery rate rather than that of the FWER is desired, there is potential for a gain in power. A simple sequential Bonferronitype procedure is proved to control the false discovery rate for independent test statistics, and a simulation study shows that the gain in power is substantial. The use of the new procedure and the appropriateness of the criterion are illustrated with examples.
Motor imagery (MI) requires subjects to visualize the requested motor behaviors, which involves a large-scale network that spans multiple brain areas. The corresponding cortical activity reflected on the scalp is characterized by event-related desynchronization (ERD) and then by event-related synchronization (ERS). However, the network mechanisms that account for the dynamic information processing of MI during the ERD and ERS periods remain unknown. Here, we combined ERD/ERS analysis with the dynamic networks in different MI stages (i.e. motor preparation, ERD and ERS) to probe the dynamic processing of MI information. Our results show that specific dynamic network structures correspond to the ERD/ERS evolution patterns. Specifically, ERD mainly shows the contralateral networks, while ERS has the symmetric networks. Moreover, different dynamic network patterns are also revealed between the two types of MIs, in which the left-hand MIs exhibit a relatively less sustained contralateral network, which may be the network mechanism that accounts for the bilateral ERD/ERS observed for the left-hand MIs. Similar to the network topologies, the three MI stages also appear to be characterized by different network properties. The above findings all demonstrate that different MI stages that involve specific brain networks for dynamically processing the MI information.