Content uploaded by Shadiya Alingal Meethal
Author content
All content in this area was uploaded by Shadiya Alingal Meethal on Apr 28, 2023
Content may be subject to copyright.
Analysis of EEG Microstates During Execution of a
Nine Hole Peg Test
Shadiya Alingal Meethal
Robotic Research Group
University of Hertfordshire
Hatfield, United Kingdom
email: s.alingal-meethal@herts.ac.uk
Volker Steuber
Biocomputation Research Group
University of Hertfordshire
Hatfield, United Kingdom
email: v.steuber@herts.ac.uk
Farshid Amirabdollahian
Robotic Research Group
University of Hertfordshire
Hatfield, United Kingdom
email: f.amirabdollahian2@herts.ac.uk
Abstract—Electroencephalogram(EEG) microstates are brief
periods of time during which the brain’s electrical activity
remains stable. The analysis of EEG microstates can help to
identify the background neuronal activity at the millisecond
level. The utilization of haptic and robotic technologies can help
in evaluating human motor skills. A haptic device Geomagic
Touch is used in this study to recreate Nine Hole Peg Test
(NHPT) in an embedded reality setup. A preliminary study is
conducted to explore changes in neural assemblies related to
resting state and fine motor state EEG when fatigue sets in. Five
healthy participants are recruited to perform a haptic NHPT
under different physical conditions. Three distinct microstates
are observed during the resting state and a separate set of 3 states
are observed during the NHPT. Changes are assessed by utilising
microstate parameters such as occurrence, coverage, duration,
and global explained variance. It is found that the coverage of
microstate C for resting states decreases for all the participants
after the dumbbell exercise. During the fine-motor task, the
coverage of microstate MS3 decreases for all participants except
one. These results support the involvement of different neural
assemblies, but also highlight the potential that physical fatigue
can be observed and identified by assessing changes in microstate
features, in this case, a parameter such as coverage.
Index Terms—EEG; microstates; NHPT; Geomagic Touch.
I. INTRODUCTION
Human movements are controlled by the Central Nervous
System. Stroke is a condition where the blood supply to
the brain is disrupted, resulting in oxygen starvation, brain
damage and loss of function. One in six people worldwide
will have a stroke in their lifetime [1]. Over three-quarters
of stroke survivors report arm weakness, which makes their
daily living activities difficult [2]. Understanding the neural
mechanisms related to hand movements will help in effective
therapy designs for stroke patients. Therapy often benefits
from assessment to inform progress. The Nine Hole Peg
Test (NHPT) is one of the easiest and widely used tests for
measuring dexterity. A reliable outcome measure of NHPT
is the time taken to complete placing nine pegs into nine
holes [3]. Haptic devices have the potential to provide fur-
ther performance metrics to inform on the quality of the
fine motor task. We have designed haptic instruments for
simulating the NHPT as detailed in earlier work [4] [5].
These studies indicated that the addition of haptic and virtual
reality may introduce new cognitive demands while providing
more extensive performance metrics. To explore this further,
we decided to utilise the brain’s microstate recording before,
during, and after NHPT. Separately, we have also explored the
electromyographical impact of fatigue on gross motor muscles
[6] that highlighted needs to further explore neural correlates
at the brain level, to understand the complete chain of events
leading to mental and physical fatigue.
Human-computer interaction can induce fatigue. The state
of mental and physical fatigue can be attributed to the reduced
activity of the central nervous system, which is character-
ized by prolonged cognitive processing time and decreased
attention levels [7]. When people start to feel fatigued in
human-computer interaction, they tend to use the peripherals
in significantly different ways. Studies suggest methods to
classify levels of fatigue by taking interaction patterns as input
[8].
The concept of EEG microstates was developed by Dietrich
Lehmann and his team in the late 1970s to quantify the
spatio-temporal dynamics of the brain [9]. They suggested
that the multichannel EEG recorded over the brain follows
a stable map configuration for a short period of time. EEG
microstates were called atoms of thought since they were
thought to reflect individual high-level aspects of cognition
and information processing [10]. Changes in the scalp electric
field configuration imply changes in the distribution of under-
lying neural generators. This means that different microstate
topographies at any time reflect the neural network activity
predominating at that time [11]. The effect of fatigue on
microstate intensity has been investigated and it was found that
the amplitude of microstates increases when going from alert
to the fatigued state [12]. Not many studies have investigated
changes in EEG microstates during physical fatigue, however,
changes in microstate parameters during mental fatigue are
described in [13]. Most of the studies of EEG microstates
deal with resting state microstates. Here, an attempt is made
to perform the analysis of microstates for EEG data acquired
from a person while performing the NHPT, before and after
a fatiguing condition that is induced using a wrist dumbbell
exercise.
The rest of this paper is organised as follows. Section
II explains the materials and methods used in the study. In
Section III, the results of the study are explained and further
167Copyright (c) IARIA, 2023. ISBN: 978-1-68558-078-0
ACHI 2023 : The Sixteenth International Conference on Advances in Computer-Human Interactions
discussion on the results is done in Section IV. Finally, Section
V concludes the study.
II. MATERIALS AND METH OD S
This section details the experimental setup and protocol
used in the study. Additionally, methodologies for finding EEG
microstates are also briefed here.
A. Experiment Set up
The experiment setup includes an EEG signal acquisi-
tion device (g.USBamp), g.GAMMAcap, the haptic device
Geomagic Touch, which recreated the NHPT in a virtual
environment and a physical rig for NHPT.
EEG signals from each participant are collected with the
help of biosignal amplifier g.USBamp at a sampling rate of
1200Hz. EEG signals are recorded from electrodes FP1, FP2,
F3, Fz, F4, FC3, FCz, FC4, C5, C3, C1, Cz, C2, C4, C6, and
CP3 by means of g.GAMMAcap. Virtual and physical NHPT
rig are localised in a way to provide an embedded reality task
setup. A C++ code running on a Windows 10 (64-bit) machine
using Visual Studio 2017 is used to configure the virtual reality
environment and the Geomagic Touch. NHPT is performed
using the stylus of the Touch device. The physical rig is kept
in front of the participant and mapped onto a virtual rig on
the LCD screen. There are nine pegs and a peg board on the
screen. A peg is attached to the end of the stylus with the help
of a rubber end cap. In each trial of NHPT, the participant has
to pick the pegs on the screen one by one and insert them into
one of the holes. The haptic feedback helps the participants
to feel the virtual pegs. The time at which each peg is picked
and released is recorded as peg status. At the beginning of
the experiment, participants are allowed a practice run to get
familiarised with the haptic device. Fig.1 shows a participant
performing the experiment.
B. Experiment Protocol
Five healthy right-handed participants with no previous
injuries to the upper limb or brain are recruited for the study.
Table I has details about participants’ physical characteristics.
The total duration of the experiment, including setup time, for
one participant, was 45-60 minutes. The ethics approval was
obtained from the University of Hertfordshire under approval
reference: ECS/PGR/UH/04035.
TABLE I
PHYSICAL CHARACTERISTICS OF PARTICIPANTS.
Subject Gender Age BMI
1 Male 25 22
2 Female 36 21
3 Male 36 28
4 Male 34 36
5 Male 28 25
Two 4-minute-long EEG recordings are taken with eyes
closed and eyes open at the beginning and end of the ex-
periment. The participants are instructed to stay focused and
try to minimise eye blinks, swallowing or any other motions
Figure. 1. A participant performing the NHPT experiment.
that alter EEG recordings. Subsequently, they are asked to do
two trials of NHPT followed by a fatiguing exercise for the
forearm. Once the participants report fatigue, they are asked
to do the next two trials of NHPT. There was no break given
between the end of the dumbbell exercise and the start of trial
3, however, 20 - 30 seconds elapsed between them during
the experiment. The experiment flow is given in Fig.2. The
experiment is designed in a way that each participant can be
their own control. Data is gathered during pre and post-fatigue
to allow comparing neural correlates of these phases.
The fatigue exercise involves flexion and extension of the
wrist using a dumbbell. Participants are asked to select one
dumbbell from the set of weights provided and they are asked
to perform 3 sets of 12, 10, and 8 repetitions with a 30 seconds
rest in between the sets. All participants reported fatigue after
the 3 sets.
A questionnaire is provided as part of the experiment in
order to assess their fatigue status. Participants are asked to
fill out parts of the questionnaire at the beginning of the
experiment and requested to update their fatigue status before
NHPT Trial1, after NHPT Trial2, before NHPT Trial3 and
after NHPT Trial4.
C. Methodology
MATLAB R2019A is used to develop the EEG process-
ing algorithms. The recorded EEG signals are segmented to
extract data corresponding to each phase of the experiment
and each NHPT trial. To remove high-frequency noise and
low-frequency drift all signals are filtered in the frequency
band 0.5-60 Hz using an FIR filter. Independent Component
Analysis (ICA) is used for removing EEG artefacts [14].
Microstates are found for the resting state data at the begin-
ning and the end. Also, microstates are found for pre-fatigue
and post-fatigue NHPT trials. The EEG microstate analysis is
performed with the help of the microstate EEGLAB toolbox
in MATLAB 2019a [15]. The main part of the microstate
analysis involved segmenting the EEG recordings into quasi-
stable states using a clustering method. Modified K means
clustering is used in this project to find the microstates. A
two-step clustering is used to find microstate maps. The first
168Copyright (c) IARIA, 2023. ISBN: 978-1-68558-078-0
ACHI 2023 : The Sixteenth International Conference on Advances in Computer-Human Interactions
Figure. 2. Experiment Flow and different phases.
clustering is performed on individual participants and in the
next level, the clustering is done across the subjects [16].
Eye close data at the beginning and end are used for finding
resting state microstates. Each of the recordings is segmented
into 20 sets of 2s data segments. The data is divided into
2s epochs. For the analysis of microstates, topographies at
maximal potential field strength are considered. The strength
of the scalp potential can be quantified using global field power
(GFP), calculated as
GF P (t) = sPk
iVi(t)−Vmean(t)2
k(1)
where Vi(t)is the voltage at electrode iat time t,Vmean(t)
is the mean voltage across all electrodes at time tand kis the
number of electrodes [17].
The optimal signal to noise ratio and stable topography are
obtained at the local maximum of GFP [18]. In the first step,
the GFP of all aggregated data sets is generated. EEG maps
that correspond to GFP peaks are submitted to modified K-
means clustering for generating microstate prototypes. During
the clustering, the polarity of the maps is ignored. Microstate
prototypes are sorted by decreasing global explained variance.
Most of the literature predefined the number of microstates
as four which previously has been reported as able to explain
more than 70% of the total topographic variance [17]. In
contrast, in this project, the number of microstates are se-
lected based on the evaluation of prototype topographies and
measures of fitness. The microstate clusters obtained at the
individual level are then again clustered to obtain the global
microstate maps [19]. Three microstates are observed for both
the resting state and NHPT trials EEG. To find the microstate
parameters, these global microstate maps are backfitted to the
original EEG data [16]. After backfitting, microstate labels are
smoothed temporally to remove small segments of unstable
topography. For each microstate class, different microstate
parameters are calculated.
The microstate parameters found here are the duration,
occurrence, coverage, and global explained variance. The
duration of a microstate is the average time for which a given
microstate remains stable whenever it appears. The coverage of
a microstate is the fraction of the total recording time when the
given microstate is dominant. The occurrence is the average
number of times per second a microstate is dominant [20].
III. RES ULTS
Initially, microstates are found for resting state data at the
beginning and end of the experiment. Furthermore, microstates
are found for the first peg transfer in trial 1 and trial 3. Both
of these sets of microstates are shown in Fig.3.
Figure. 3. Microstates during resting state and an NHPT trial. These
microstate maps are found by clustering EEG topographies at the local GFP
maxima of all the participants. Red colour shows positive and blue colour
shows negative potential areas.
A. Microstate analysis for resting state data
The participants are asked to keep their eyes closed for two
minutes at the beginning and end of the experiment. Microstate
analysis is performed on this EEG data to find any changes in
microstate parameters while a person undergoes a fatiguing
exercise. Three microstates A, B and C are found for the
resting state data. The microstate parameters derived from the
resting state data are shown in Table II. It can be seen that the
occurrence of microstate A increases for all subjects except
subject 4. For microstates B and C, the occurrence increases
for three subjects and decreases for two subjects. At the same
time, the duration of microstate A increases for all subjects
except subject 4. The duration of microstate B increases for
three subjects and decreases for one subject. The duration of
microstate C decreases for all subjects.
The Coverage of microstate A increases for all subjects
except subject 4. The coverage of microstate B increases for
three subjects and decreases for two subjects. The coverage
of microstate C decreases for all the subjects. Changes in
coverage for resting state microstates are shown in Fig.4. The
global explained variance of microstate A increases for all
subjects except subject 4. GEV of microstate B increases
for three subjects and decreases for two subjects. GEV of
microstate C decreases for four subjects and remains the same
for one subject.
B. Microstate analysis for NHPT trial data
The microstates are found when a person performs the
first peg transfer in trial 1 and trial 3. Other trials were
excluded because this study is comparing neurological changes
169Copyright (c) IARIA, 2023. ISBN: 978-1-68558-078-0
ACHI 2023 : The Sixteenth International Conference on Advances in Computer-Human Interactions
TABLE II
RESTING STATE MICROSTATE PARAMETERS.
Subject Microstates Occurrence Duration(ms) Coverage(%) GEV
Pre Post Change Pre Post Change Pre Post Change Pre Post Change
1
A 4.32 4.57 ↑78.34 150.32 ↑34 64 ↑0.17 0.36 ↑
B 4.12 3.10 ↓77.74 57.16 ↓32 18 ↓0.15 0.07 ↓
C 4.35 2.85 ↓80.18 61.47 ↓35 18 ↓0.19 0.08 ↓
2
A 2.00 2.42 ↑49.68 54.79 ↑10 13 ↑0.03 0.03
B 4.72 4.45 ↓101.05 140.23 ↑47 60 ↑0.18 0.21 ↑
C 4.77 3.55 ↓92.85 76.75 ↓43 27 ↓0.17 0.07 ↓
3
A 0.57 1.30 ↑40.00 44.26 ↑3 6 ↑0.00 0.01 ↑
B 4.20 4.60 ↑123.88 106.72 ↓48 46 ↓0.17 0.13 ↓
C 4.25 4.47 ↑119.89 119.71 ↓49 48 ↓0.18 0.15 ↓
4
A 1.15 0.62 ↓72.49 30.90 ↓11 3 ↓0.02 0.01 ↓
B 2.37 3.57 ↑116.21 116.41 ↑27 41 ↑0.07 0.14 ↑
C 2.45 3.67 ↑402.22 165.44 ↓62 55 ↓0.25 0.25
5
A 0.95 2.30 ↑47.18 64.40 ↑6 15 ↑0.01 0.03 ↑
B 2.75 3.75 ↑60.34 74.45 ↑17 28 ↑0.03 0.08 ↑
C 3.62 4.30 ↑233.95 145.44 ↓77 56 ↓0.32 0.26 ↓
↑and ↓indicates increase and decrease of microstate parameters respectively
happening to a person while performing NHPT when fatigue
sets in. Trial 1 is the first trial when the participant is not
fatigued yet. Trial 3 is the trial just after the fatiguing exercise
and can therefore be called the post-fatigue trial. Three mi-
crostates are observed in the task EEG and named MS1, MS2
and MS3 in order to distinguish them from the resting state
microstates. Just like for the resting state microstates here also
the microstate parameters, occurrence, coverage, duration and
GEV are calculated and tabulated. Table III shows the pre-
fatigue and post-fatigue values of the microstate parameters.
For MS1, the occurrence increases with fatigue for two par-
ticipants and decreases with fatigue for three participants. The
duration of MS1 increases with fatigue for three subjects and
decreases for the other two. The coverage of MS1 increases
for all subjects except one. The global explained variance of
MS1 increases with fatigue for all subjects.
For MS2 all the parameters increase with fatigue for three
subjects and decrease with fatigue for two subjects. No MS3 is
present for subject 2. In the other four subjects, duration, cov-
erage and GEV decrease with fatigue for MS3. The changes
in coverage for trial microstates are shown in Fig.5. The
Occurrence of MS3 increases for one subject and decreases
for three subjects.
C. Assessment of performance time during NHPT task for
different trials
Table IV shows the time taken for each trial which is
recorded with the help of Geomagic Touch API. A paired
sample t-test is done between trial 1 and trial 3 and it is found
that the time taken for post fatigue trial is not statistically
different from the time taken for pre-fatigue trial(p-value .608).
However, given the small number of samples, looking at the
individual values for trial 1 versus trial 3, two participants
(Subject 1 and Subject 2) show an increase in completion time
while the remaining participants show an improvement in the
peg-placement and task completion.
Figure. 4. Changes in coverage with fatigue for resting state microstates A,
B and C.
D. Questionnaire assessment of fatigue during experiment
progression
The forearm fatigue status of each participant is recorded
during the experiment. Participants are asked to update their
fatigue status on a scale of 1 indicating not fatigued, to 10
indicating extremely fatigued, before trial 1, after trial 2,
before trial 3 and after trial 4. The fatigue score of each
participant is shown in Table V. This table indicates that all
participants report fatigue after the Dumbbell exercise, but the
reduction of scores from before trial 3, to after trial 4, for
participants 1, 3 and 5 could indicate an adjustment to the
task complexity.
IV. DISCUSSIONS
The present study investigates the changes in EEG mi-
crostates when performing an embedded reality NHPT, pre
and post fatigue. To the best of our knowledge, this is the
170Copyright (c) IARIA, 2023. ISBN: 978-1-68558-078-0
ACHI 2023 : The Sixteenth International Conference on Advances in Computer-Human Interactions
TABLE III
TASK M ICR OSTATE PAR AM ETE RS .
Subject Microstates Occurrence Duration(ms) Coverage(%) GEV
Pre Post Change Pre Post Change Pre Post Change Pre Post Change
1
MS1 0.81 1.52 ↑78.89 65.56 ↓6.40 10.00 ↑0.0023 0.02 ↑
MS2 1.08 1.28 ↑48.96 61.98 ↑5.30 8.82 ↑0.0035 0.01 ↑
MS3 2.16 3.31 ↑409.69 266.01 ↓88.30 81.19 ↓0.63 0.38 ↓
2
MS1 0.29 0.67 ↑3.46∗1.76∗↓100 98.29 ↓0.69 0.81 ↑
MS2 0 0.33 ↑0 25.69 ↑0 1.72 ↑0 0.0011 ↑
MS3 0 0 0 0 0 0 0 0
3
MS1 1.86 1.00 ↓492.02 975.00 ↑91.68 97.79 ↑0.81 0.87 ↑
MS2 0.53 0.50 ↓75.42 44.17 ↓4.02 2.21 ↓0.0023 0.0015 ↓
MS3 1.06 0 ↓40.42 0 ↓4.30 0 ↓0.0054 0 ↓
4
MS1 2.67 2.62 ↓154.72 399.78 ↑42.45 73.67 ↑0.30 0.61 ↑
MS2 2.67 2.24 ↓258.64 102.69 ↓49.31 24.67 ↓0.33 0.07 ↓
MS3 1.00 0.25 ↓69.21 33.13 ↓8.23 1.66 ↓0.02 0.0007 ↓
5
MS1 3.92 1.31 ↓189.08 703.75 ↑74.13 92.45 ↑0.41 0.75 ↑
MS2 1.18 1.31 ↑39.72 57.50 ↑4.67 7.55 ↑0.01 0.01 ↑
MS3 3.92 0 ↓54.08 0 ↓21.20 0 ↓0.05 0 ↓
↑and ↓indicates increase and decrease of microstate parameters respectively
*Duration of MS1 for subject 2 is in seconds
first study to identify microstates during the performance of a
NHPT, and to reflect on changes to these microstates after a
fatiguing exercise.
A. Resting state microstates
Several studies in the field of microstates found that mi-
crostate maps generally fall into four categories. However,
in our study, we used the polarity invariant measures of fit
Figure. 5. Changes in coverage with fatigue for trial microstates MS1, MS2
and MS3.
TABLE IV
TIM E TAKE N FOR E ACH T RIA L OF NHPT IN SECONDS.
Subject Trial 1 Trial 2 Trial 3 Trial 4
1 67 58 78 67
2 90 118 127 102
3 53 49 46 37
4 114 86 79 69
5 96 62 47 55
Mean 84 74.6 75.4 66
(SD) (21.59) (27.85) (32.99) (23.81)
global explained variance and cross validation to determine the
optimum number of microstates and found three microstates
for resting state data. Microstates B and C found in our study
resembled resting state microstates B and A in the literature.
It was interesting to investigate how microstate parameters
like occurrence, coverage, duration and GEV changed when a
person became fatigued physically and mentally. The dumbbell
exercise in this experiment contributed to physical fatigue for
the participants whereas performing NHPT using the haptic
device Geomagic Touch provided a cognitive load for the
participants. It was found that with fatigue all the parameters
of microstate A increased for all subjects except one. The
coverage of microstate C decreased for all subjects, which
implies that fatigue made microstate C less dominant and
increased the occurrence of other microstates. It could be seen
that less coverage of microstate C was compensated by the
increase in the occurrence of microstate B.
B. NHPT trial microstates
Three microstates were observed in the EEG while perform-
ing NHPT trials. The microstate parameters were determined
for transferring the first peg in pre-fatigue and post-fatigue
trials (trial 1 and trial 3). MS3 was not present for one subject
and, for all the other subjects, the coverage of microstate 3
decreased. This implies that new neural correlates were intro-
duced with fatigue which reduced the coverage of microstate 3.
While the coverage of MS3 decreased, the coverage of MS1
TABLE V
SEL F-RE PO RTED FATI GUE S TATUS.
Subject Before Trial1 After Trial2 Before Trial3 After Trial 4
1 1 2 8 7
2 1 1 8 8
3 1 1 8 6
4 1 4 9 9
5 1 2 8 7
171Copyright (c) IARIA, 2023. ISBN: 978-1-68558-078-0
ACHI 2023 : The Sixteenth International Conference on Advances in Computer-Human Interactions
increased, which means that the fatigue caused changes in
brain signals which created more MS1. A similar trend could
also be seen in the GEV values. For MS1, GEV increased
which was compensated by a decrease in GEV of MS3.
The duration of MS1 was very high compared to the other
microstates and it could be seen that its duration sometimes
reached seconds. For subject 2 the duration of MS1 was in the
order of seconds because only that particular microstate was
present in the pre-fatigue trial, and in the post-fatigue trial,
MS2 was present for a very short duration.
C. Performance time and fatigue status
Subject-score for fatigue for participants 1, 3 and 5 were
reduced after trial 4, also there is an improvement in NHPT
performance from trial 3 to trial 4. From Table V it can be
seen that NHPT alone induces fatigue for subjects 1,4 and
5 after trial 2. But looking at the trial time there was a
reduction in trial time for these subjects which indicates NHPT
performance improvement. This could indicate that while
participants worked harder, and perceived to work harder, they
actually improved their performance score as indicated by the
reduction in peg placement time.
When comparing the time between trial 1 and trial 3, it can
be seen that participants 1 and 2 took more time to complete
NHPT when fatigued. Participants 3, 4 and 5 completed NHPT
trial 3 in less time compared to trial 1, while all reported
fatigue after the dumbbell exercise. These observations can
indicate that physical fatigue alone is not impacting NHPT
performance. This could be because these participants did not
find the NHPT task challenging and haptic/visual assistance
given to these tasks provided a good medium for reducing
their completion time despite fatigue in their wrist. It could be
also possible that the NHPT task involves a different neural
assembly compared to the assemblies needed for the wrist
exercise. Furthermore, it is possible that participants’ fitness
level could impact their recovery from fatigue. Participant 4
who had a higher BMI increased his fatigue level by doing
NHPT alone compared to others. Comparing Table I and Table
V it can be seen that participants with better BMI recovered
soon from fatigue except for participant 2. No MS3 is present
for participant 2 while NHPT trial time is more for each trial
compared to others. Also, this participant struggled a lot to
place the pegs in the hole during the experiment. This suggests
that cognitive fatigue might also have an impact on MS3.
Research on human-computer interaction supports the notion
that performing mental tasks using a video display terminal
can lead to cognitive fatigue [21]. Coverage of MS3 for all
other participants decreased after fatiguing exercise which
shows the impact of physical fatigue on microstates.
V. CONCLUSION
This is a preliminary study which investigated the changes
in EEG microstates while a person performed a widely used
manual dexterity test, the Nine Hole Peg Test, before and
after fatigue conditions. The main goal of the study was
to observe differences in resting state and task performance
microstates and to observe the changes due to a physically
fatiguing dumbbell exercise. We observed these differences, as
highlighted by the topological maps, but also observed changes
in the microstate parameters. With resting state microstates it
was found that two of the microstates observed resembled the
microstates already established in the literature [22]. It was
also found that the coverage of some microstates decreased
with fatigue for both resting state and trial data which was also
backed up by a reduction in global explained variance. This
suggests that assessing the microstate parameter coverage can
help to identify physical fatigue. All participants reported fa-
tigue on the forearm after the dumbbell exercise. We found that
physical fatigue did not affect the NHPT performance as some
participants reporting physical fatigue improved performance
during the NHPT trials. However, alterations in microstate
parameters were observed after the physical fatigue. We intend
to further explore this by comparing the microstates during
NHPT trials, with microstates observed during the dumbbell
exercise. This can convey further information regarding the
similarity or dissimilarity of the neural assemblies present
during motor tasks performed in this experiment. Furthermore,
expanding the number of participants will enhance the likeli-
hood of conducting a statistically sound analysis of the results.
REFERENCES
[1] “Learn about stroke,” March. 15, 2023. [Online].
Available: https://www.world-stroke.org/component/content/article/16-
forpatients/84-facts-and-figures-about-stroke
[2] E. S. Lawrence et al., “Estimates of the prevalence of acute stroke
impairments and disability in a multiethnic population,” Stroke, vol. 32,
no. 6, pp. 1279–1284, 2001.
[3] D. T. Wade, “Measurement in neurological rehabilitation,” Current
Opinion in Neurology, vol. 5, no. 5, pp. 682–686, 1992.
[4] M. E. Bowler, “Interactive haptics for remote and on-site assessment
of arm function following a stroke,” Ph.D. dissertation, University of
Hertfordshire, 2019.
[5] F. Amirabdollahian and G. Johnson, “Analysis of the results from use of
haptic peg-in-hole task for assessment in neurorehabilitation,” Applied
Bionics and Biomechanics, vol. 8, no. 1, pp. 1–11, 2011.
[6] A. Thacham Poyil, V. Steuber, and F. Amirabdollahian, “Adaptive
robot mediated upper limb training using electromyogram-based muscle
fatigue indicators,” Plos one, vol. 15, no. 5, p. e0233545, 2020.
[7] A. Murata and A. Uetake, “Evaluation of mental fatigue in human-
computer interaction-analysis using feature parameters extracted from
event-related potential,” in Proceedings 10th IEEE International Work-
shop on Robot and Human Interactive Communication. ROMAN 2001
(Cat. No. 01TH8591). IEEE, 2001, pp. 630–635.
[8] A. Pimenta, D. Carneiro, J. Neves, and P. Novais, “A neural network
to classify fatigue from human–computer interaction,” Neurocomputing,
vol. 172, pp. 413–426, 2016.
[9] T. Koenig, M. I. Tomescu, T. A. Rihs, and M. Koukkou, “Eeg indices
of cortical network formation and their relevance for studying variance
in subjective experience and behavior,” In Vivo Neuropharmacology and
Neurophysiology, pp. 17–35, 2017.
[10] D. Lehmann, “Brain electric microstates and cognition: the atoms of
thought,” in Machinery of the Mind. Springer, 1990, pp. 209–224.
[11] A. P. Zanesco, B. G. King, A. C. Skwara, and C. D. Saron, “Within
and between-person correlates of the temporal dynamics of resting eeg
microstates,” Neuroimage, vol. 211, p. 116631, 2020.
[12] R. A. Thuraisingham, Y. Tran, A. Craig, N. Wijesuriya, and H. Nguyen,
“Using microstate intensity for the analysis of spontaneous eeg: Tracking
changes from alert to the fatigue state,” in 2009 Annual International
Conference of the IEEE Engineering in Medicine and Biology Society.
IEEE, 2009, pp. 4982–4985.
172Copyright (c) IARIA, 2023. ISBN: 978-1-68558-078-0
ACHI 2023 : The Sixteenth International Conference on Advances in Computer-Human Interactions
[13] W. Li, S. Cheng, H. Wang, and Y. Chang, “Eeg microstate changes
according to mental fatigue induced by aircraft piloting simulation: An
exploratory study,” Behavioural Brain Research, vol. 438, p. 114203,
2023.
[14] T.-P. Jung et al., “Removing electroencephalographic artifacts by blind
source separation,” Psychophysiology, vol. 37, no. 2, pp. 163–178, 2000.
[15] A. T. Poulsen, A. Pedroni, N. Langer, and L. K. Hansen, “Microstate
eeglab toolbox: An introductory guide,” BioRxiv, p. 289850, 2018.
[16] D. F. D’Croz-Baron, M. Baker, C. M. Michel, and T. Karp, “Eeg
microstates analysis in young adults with autism spectrum disorder
during resting-state,” Frontiers in human neuroscience, vol. 13, p. 173,
2019.
[17] A. Khanna, A. Pascual-Leone, and F. Farzan, “Reliability of resting-
state microstate features in electroencephalography,” PLoS ONE, vol. 9,
no. 12, pp. 1–21, 2014.
[18] K. Zhang et al., “Reliability of eeg microstate analysis at different
electrode densities during propofol-induced transitions of brain states,”
NeuroImage, vol. 231, p. 117861, 2021.
[19] M. Baradits, I. Bitter, and P. Czobor, “Multivariate patterns of eeg
microstate parameters and their role in the discrimination of patients
with schizophrenia from healthy controls,” Psychiatry Research, vol.
288, p. 112938, 2020.
[20] A. Khanna, A. Pascual-Leone, C. M. Michel, and F. Farzan, “Microstates
in resting-state eeg: current status and future directions,” Neuroscience
& Biobehavioral Reviews, vol. 49, pp. 105–113, 2015.
[21] M. J. Smith, F. T. Conway, and B.-T. Karsh, “Occupational stress in
human computer interaction,” Industrial health, vol. 37, no. 2, pp. 157–
173, 1999.
[22] C. M. Michel and T. Koenig, “Eeg microstates as a tool for studying
the temporal dynamics of whole-brain neuronal networks: a review,”
Neuroimage, vol. 180, pp. 577–593, 2018.
173Copyright (c) IARIA, 2023. ISBN: 978-1-68558-078-0
ACHI 2023 : The Sixteenth International Conference on Advances in Computer-Human Interactions
