sensors

Review
Past, Present, and Future of EEG-Based BCI Applications
Kaido Värbu 1, *, Naveed Muhammad 1 and Yar Muhammad 2

1 Institute of Computer Science, University of Tartu, 51009 Tartu, Estonia; [email protected]

2 Department of Computing & Games, School of Computing, Engineering & Digital Technologies,
Teesside University, Middlesbrough TS1 3BX, UK; [email protected]
* Correspondence: [email protected]

Abstract: An electroencephalography (EEG)-based brain–computer interface (BCI) is a system that

provides a pathway between the brain and external devices by interpreting EEG. EEG-based BCI
applications have initially been developed for medical purposes, with the aim of facilitating the
return of patients to normal life. In addition to the initial aim, EEG-based BCI applications have also
gained increasing significance in the non-medical domain, improving the life of healthy people, for
instance, by making it more efficient, collaborative and helping develop themselves. The objective of
this review is to give a systematic overview of the literature on EEG-based BCI applications from the
period of 2009 until 2019. The systematic literature review has been prepared based on three databases
PubMed, Web of Science and Scopus. This review was conducted following the PRISMA model. In
this review, 202 publications were selected based on specific eligibility criteria. The distribution of the
research between the medical and non-medical domain has been analyzed and further categorized
into fields of research within the reviewed domains. In this review, the equipment used for gathering
EEG data and signal processing methods have also been reviewed. Additionally, current challenges
in the field and possibilities for the future have been analyzed.

Keywords: brain–computer interface (BCI); electroencephalography (EEG); rehabilitation; systematic

literature review
Citation: Värbu, K.; Muhammad, N.;
Muhammad, Y. Past, Present, and
Future of EEG-Based BCI
Applications. Sensors 2022, 22, 3331. 1. Introduction
https://doi.org/10.3390/s22093331 The electroencephalography (EEG)-based brain–computer interface (BCI) is one of
Academic Editors: Sung-Phil Kim the most rapidly developing fields of BCI [1,2] and has potential to expand far beyond the
and Yvonne Tran domain of medical applications, in which they were initially most popular. The use of EEG
has become possible due to the work and discovery by Hans Berger who discovered in 1924
Received: 23 February 2022
that electrical signals of the human brain could be measured from the scalp [3]. The initial
Accepted: 25 April 2022
discovery was made using a simple galvanometer and confirmed the possibility that neural
Published: 26 April 2022
activity could be measured by this method [4]. EEG measures the brain electrical activity
Publisher’s Note: MDPI stays neutral caused by the flow of electric currents during synaptic excitations of neuronal dendrites
with regard to jurisdictional claims in and is measured via electrodes placed on the scalp [1,4]. After the initial discovery by Hans
published maps and institutional affil- Berger, additional brainwave types and mental states associated with the brainwave types
iations. have been determined. An EEG system can provide a pathway between the brain and
external device, making it possible to read biological signals and interpret certain aspects
of a person’s cognitive state [5]. BCI systems are EEG systems that allow users nearly
real-time control of an external actuator. EEG-based BCI applications could be used by a
Copyright: © 2022 by the authors.
person to control a computer or other device using one’s thoughts without using ordinary
Licensee MDPI, Basel, Switzerland.
methods of working with a computer (e.g., using hands) [6]. EEG-based BCI applications
This article is an open access article
distributed under the terms and
could be used also to monitor subject’s mental states such as emotions [7], concentration [8]
conditions of the Creative Commons
or drowsiness [9]. BCI adds to the monitoring processes, for example, via enabling systems
Attribution (CC BY) license (https://
to be adaptive or sending alert signals.
creativecommons.org/licenses/by/ The BCI applications have been initially developed to help people with disabilities to
4.0/). be able to communicate, control computers and use aiding equipment such as wheelchairs

Sensors 2022, 22, 3331. https://doi.org/10.3390/s22093331 https://www.mdpi.com/journal/sensors

Sensors 2022, 22, 3331 2 of 44

or robotic arms. One of the first BCI applications assisted individuals with speech anoma-
lies [5]. The possibility of using BCI to develop prosthetic arms was considered as early
as 1917 [1]. In the overview by Tariq et al., EEG-based BCI has also been used in order to
develop wearable lower-limb exoskeletons, as BCI has emerged as an alternative communi-
cation system between the human brain and output devices [10]. BCI-based prosthetics
are supporting patients in cases of paralysis, amputations and loss of central nervous
system functionality due to other reasons [2]. BCI could also be applied in order to enhance
neuroplasticity [11]. Neuroplasticity of the brain has been characterized as the capacity of
the brain to change or adapt its morphology and functioning in response to experience [12].
The applications have been used in the medical domain also by people suffering from
locked-in syndrome or amyotrophic lateral sclerosis (ALS) [1].
EEG-based BCI applications have been used in different areas stretching from the
medical domain to non-medical domain where EEG-based BCI applications have been
used for entertainment, art, as well as some other areas [5,6,13]. There have been studies
concerning the non-medical domain, for example, developing devices to monitor the
alertness level of employees [14]. Another aspect that has been investigated is the overall
car traffic safety and avoidance of driving fatigue and drowsiness during driving that
could result in fatal accidents [1]. BCI applications can be used for controlling smart
homes [15–18] or a car [19,20]. In the domain of non-medical applications, the BCI could
also be used for sport motor skills improvement, acting skills improvement or surgical
skills improvement [11]. The applications for entertainment could include games designed
for improving subject attention level or concentration level [13], but also control of drones
and humanoid robots [6].
During the current review, the literature on EEG-based BCI applications in the period
from 2009 to 2019 was analyzed. The current review gives an overview concerning articles
and conference proceedings per year, publications per region/continent, experimental
publications per year, publication distribution by domain, EEG devices used, number of
EEG channels used, techniques used to obtain EEG data, feature extraction and classification
methods used.
In the first part of the review, an overview concerning the published articles as well
as conference papers is also presented through the years. The regions/continents at the
forefront of the scientific progress in the field have been highlighted.
An overall review of EEG-based BCI applications has been created and the BCI ap-
plications have been categorized based on domain (medical or non-medical) and by field,
describing the current trends in BCI applications development in more detail. The BCI
applications have been included in the medical domain in case they have been designed to
provide assistance, monitor, assess or support rehabilitation and in the non-medical domain
in case the applications have been designed to entertain, control machines, authenticate,
monitor or assist without a medical purpose. The distinction between the domains has
been done in a way similar to the work by Al-Nafjan et al. [5].
This review introduces the current trends for the development of EEG-based BCI
applications. Although the initial need for BCI applications has been in the medical domain,
this review shows that there is currently a higher pace of development in the non-medical
domain. The applications meant for widespread use could be used for monitoring user
attention, supporting daily activities or for entertainment.
In this review, EEG signal acquisition and processing is also analyzed. The overview
concerning the prevalence of different EEG devices has been given together with the
details regarding the number of EEG channels used in previous studies. An overview is
given concerning the techniques used to obtain the EEG data. Regarding the EEG signal
processing, feature extraction and classification methods have been reviewed.
This review was conducted in connection to the first author’s master’s thesis [21]. This
review provides a comprehensive summary of recent works conducted in the field and an
overview of current obstacles that inhibit progress, both technically and in other aspects. In
addition to technical aspects there could be, for example, user-related aspects that limit the
Sensors 2022, 22, 3331 3 of 44

use of BCI applications due to difficulty for the user to learn to use the applications and
inability to generate or alter the required EEG signals.
The ethical, legal and safety considerations in BCI application development are as
important as the technical aspects. The synthesis on current challenges gives a mapping of
what to focus on in order to support BCI development and shows where the possible risks
exist that would need to be addressed. The trends and future possibilities give a better
understanding of what we could expect in years to come.

2. Background
BCI can be defined as a system that translates the brain activity patterns of a user
into messages or commands for an interactive application [22]. A BCI is a control and/or
communication system in which the user’s commands and messages are not dependent
on common brain-motor periphery communication channels [23]. The BCI user’s brain
activity is typically measured using EEG [22]. BCI generally functions through four distinct
processes consisting of recording neuronal activity, extracting features, gathering important
information and combining information for useful purposes [13]. The EEG signal analysis
has been further divided into four steps which are gathering of raw EEG data, signal
pre-processing, feature extraction and classification [2].
According to Padfield et al. [2], the signal analysis steps could be characterized as
follows. The collected raw EEG data comprises all the EEG data collected before pre-
processing and further analysis. After the collection of raw EEG data, pre-processing is
used to clean noise and enhance the quality of collected EEG data for further analysis.
During feature extraction, discriminative and non-redundant information is extracted
from the EEG data to form a set of features. Extracted features will capture distinct signal
characteristics which can be used as a basis for the differentiation between task-specific
brain states. According to Padfield et al. during classification step, the task carried out
by a subject is determined and corresponding actions are taken by the system. The use
of specific feature extraction and classification method depends on specific type of BCI
application under study.
The most common way to categorize BCI is on the basis of invasiveness. The BCI
could be invasive or non-invasive, depending on whether the device is connected physi-
cally to one’s brain and the electrodes are placed inside the brain or the device will read
brainwaves from the top of the scalp [1]. The use of invasive method of direct contact of
the electrodes with brain is more efficient as there are less interfering factors that influence
the signal quality [24], but the invasive method contains the risks of surgical procedures
and related complications.
More popular and easy is the non-invasive method, where the brainwaves are read
from the top of the scalp via electrodes located on the scalp. As this method also has some
drawbacks, including disturbance from external noise, effects from the posture and mood
of the subject, detecting a low strength signal resulting in a reduced signal quality [2], there
is a significant amount of on-going research to determine the best ways for signal detection
and analysis.
The most popular current non-invasive method for acquiring brainwaves for BCI is
detection through EEG. The popularity of EEG is supported by the inexpensiveness of
the equipment, reduced complications compared to invasive procedures, portability, easy
process to set-up and use and the possibility to directly measure neural activity [1,2]. EEG
reading is easy to use and has high potential to be applied by majority of the population
in order to use high variety of possibilities that BCI offers. Other non-invasive methods
include functional magnetic resonance imaging (fMRI), magnetoencephalography (MEG)
and near-infrared spectroscopy (NIRS) which can be used separately or combined [1]. The
advantages of EEG are high temporal resolution, good portability, high temporal resolution,
low cost, less invasive when compared to fMRI and independence from need of a complex
environment when compared to MEG according to Alsharif et al. The disadvantage of EEG
is low spatial resolution [25].
Sensors 2022, 22, 3331 4 of 44

BCI could be divided into separate subgroups based on the way to detect and convey
the signal from the brain to BCI application and principles of the functioning of the BCI. As
per Al-Nafjan et al. [5], BCIs could be categorized as active or passive, based on the control
of the BCI application. The categorization of BCIs based on active or passive control and
corresponding techniques used to obtain EEG data is presented in Table 1.

Table 1. Categorization of BCIs and corresponding techniques used to obtain EEG data according to
Al-Nafjan et al. [5] and Abiri et al. [1].

Category Technique Used to Obtain EEG Data Description

Imagining the movement of a specific body part for example
hands, feet or tongue. The intent will affect the brain activity and
Motor-imagery could be detected from the EEG. The imagination activates
the brain areas that are responsible for generating the
actual movement.
The brain activity is affected by external visual stimulation and
the corresponding altered EEG activity is registered. For example,
in case of steady-state visual evoked potential (SSVEP) there are
Visual evoked potential
different visual stimuli flickering at different frequencies and
depending on the direction of the gaze of the subject the EEG
pattern will be consistent with the specific flickering rate.
Auditory stimulation is generated and depending on the focus of
Auditory evoked potential
the subject corresponding EEG activity is registered.
Active
Physical vibrations at different frequencies are generated for
example on the hands and feet of the subject. Depending on the
Vibrotactile evoked potential focus of the subject, a corresponding EEG pattern to the specific
physical vibration is registered and could be used to control some
external device.
Imagined speech Imagination of words or sentences that are recognized from EEG.
The error-related potential is generated when there is a mismatch
between the subject’s intentions and response from the BCI
application. The technique can be used to correct tasks given by
Error-related potential the subject. For example, when the subject is controlling the
cursor, but the cursor is moving in the wrong direction, an
error-related potential is generated, and the course of the cursor
can be corrected.
Systems wherein brain signals yield outputs without any
voluntary control. For example, monitoring drowsiness,
Passive Analyzing EEG spectral changes
attention, mental workload, emotions, concentration and
other states of the mind.

The BCIs using active control react on conscious efforts to alter brainwave patterns
and the applications could be controlled via active efforts by the user. The BCIs applying
passive control react to the involuntary status of the brainwaves, for example on emotional
states such as meditation, excitement and stress. Different emotions could be elicited by,
for example, visual-based elicitation using images, prepared task or audio-visual elicitation
using short film video clips [5].
During active control of the BCI application, the signal could be detected via different
techniques. The range of these techniques is broad and covers motor imagery, external
stimulation (such as visual, auditory and vibrotactile), error-related potential, hybrid and
other techniques. During the motor imagery task, for example, the subject is imagining
the movement of a specific body part, and during error-related potential task, error-related
potential is generated when there is a mismatch between the subject’s intentions and
response from the BCI application [1].
As per Pasqualotto et al. [23] and Machado et al. [26] BCI could also be categorized
depending on whether BCI is dependent or independent of certain muscle movements.
Sensors 2022, 22, 3331 5 of 44

Padfield et al. [2] have also categorized BCI as evoked or spontaneous. As per Nicolas-
Alonso and Gomez-Gil [27] BCI could be categorized as synchronous or asynchronous.
An overview concerning additional different categorization of BCIs in the literature is
presented in Table 2.

Table 2. Additional different categorization of the BCIs in the literature by Pasqualotto et al. [23],
Machado et al. [26], Padfield et al. [2] and Nicolas-Alonso and Gomez-Gil [27].

Author BCI Categorization Description

Dependent on muscles and peripheral nerves. For example, in
Dependent case of visual evoked potential (VEP), gaze is directed by
Pasqualotto et al. [23] muscles to focus on different visual stimuli.
Machado et al. [26]
Muscle movement is not needed to control BCI. For example, in
Independent
case of P300 response is detected from EEG and analyzed.
Also named as exogenous. Some type of external stimulation is
required such as visual, auditory, or sensory. Can be further
divided to evoked potentials and event-related potentials. In
Evoked case of evoked potentials, changes in EEG can be detected due
to responses to external stimuli. In case of event-related
Padfield et al. [2] potentials, the changes in EEG are caused by sensory or
cognitive events.
Also named as endogenous. External stimulation is not
required. For example, motor-imagery technique, where
Spontaneous
subjects imagine movement of a limb and there is no additional
input from external stimuli.
The BCI analyzes signals during certain time windows, and the
Synchronous
subject is able to give commands after fixed time intervals.
Nicolas-Alonso and The brain waves of the subject are analyzed constantly, and the
Gomez-Gil [27] subject can give commands whenever the subject wants.
Asynchronous
Asynchronous BCI gives the subject more possibilities and
flexibility concerning controlling the BCI.

“Dependent BCIs” require muscle control, for example, via gaze control. “Independent
BCI” on the other hand detect signals only based on changes in the brainwaves without
required muscle movement [23,26]. According to Padfield et al. [2], another possibility
to categorize BCIs would be depending on whether external stimulation is required for
the functioning of the BCI or not, thus dividing the systems into “evoked” when external
stimulation is needed and to “spontaneous” in case external stimulation is not needed.
According to the aforementioned categorization by Padfield et al., evoked systems include
for example steady-state visual evoked potential (SSVEP), where visual stimulation is
received via flickering at unique frequencies that causes corresponding changes in EEG
when focusing on specific stimulus at a specific flickering frequency. According to this
categorization spontaneous systems also include for example motor-imagery technique,
where external stimulation is not needed and the changes in EEG patterns are generated
via imagining the movement of a limb. Padfield et al. have noted that the categorization
to evoked and spontaneous systems has also been named by some authors as exogenous
and endogenous.
An additional way to classify BCIs is based on the time when the signals from the
user are gathered by the BCI. The BCIs could therefore be divided as synchronous and
asynchronous [27]. BCI is considered synchronous, when the information concerning the
brainwave status is gathered during specific time intervals. It means that the user can give
commands only at distinct timing, and the brainwaves are not measured at other times. In
case of asynchronous analysis, the brain patterns are analyzed on an ongoing basis and the
user is more flexible when giving commands to the system.
Sensors 2022, 22, 3331 6 of 44

3. Objectives
The systematic literature review was prepared based on three databases, PubMed,
Web of Science and Scopus, in order to cover a wide range of reliable peer reviewed
publications on EEG-based BCI applications. In previous studies, either one [5] or multiple
databases [28,29] have been used to conduct the literature search, but in the current review
a selection of multiple databases was used in order to cover a broader range of publications
on the topic.
The time period for the publications from between 2009 and 2019 was selected in order
to give an overview concerning longer period of time and to be able to compare the results
with other studies. While 2019 might not seem truly up to date, we would like to highlight
that it is a systematic review. Thus, once such a review has begun, altering the chosen period
would mean either (i) compromising the systematic nature of the review, or (ii) conducting
a new systematic review from scratch, which are undesirable and impractical, respectively.
Similar lengths of the time interval have been selected in previous studies [5,28]. From the
time period above, 635 publications were selected according to the search criteria specified
in Appendix A, which were then further screened and assessed for eligibility. As this
review includes articles until 2019, in Appendix B we list additional publications which
have been published after the time period of the current review, i.e., 2009 to 2019.
After the identification of the publications, a screening of the publications was per-
formed. After the assessment of eligibility, 202 publications were included in the final
analysis. The process was conducted as per the PRISMA model [30]. Listed below are the
objectives that were set at the beginning for conducting our systematic literature review.
The objectives of the systematic literature review on EEG-based BCI applications:
1. Determine the trends concerning the publication of articles, conference proceed-
ings and overall number of publications on EEG-based BCI applications from 2009
until 2019.
2. Determine the techniques used for obtaining EEG data and give an overview con-
cerning the trends in EEG signal processing, which includes feature extraction and
classification methods.
3. Give an overview concerning the devices used for EEG signal collection and specify
the number of EEG channels used for obtaining the data.
4. Determine the proportion of scientific studies conducted on the topic in the medical
and non-medical domain and further analyze the distribution of the studies per
application field.
5. Analyze the literature by regions/continents, i.e., which regions/continents are at the
forefront of scientific progress in EEG-based BCI applications and highlight the most
contributing authors on the topic.

4. Methods
As per best practice, the PRISMA model [30,31] has been used to conduct systematic
literature reviews in many fields of research. In addition to the PRISMA model, guidance
from Cochrane Collaboration can be followed in order to prepare a systematic literature
review [32]. As the PRISMA model has been developed together with the Cochrane
Collaboration and with a large number of experts in the field, the PRISMA model was
selected as the best current collection of principles in order to conduct the current systematic
literature review on EEG-based BCI applications. The PRISMA method has been developed
by Moher et al. [30] and has been used widely in conducting well organized systematic
literature reviews such as [5,23,28,29].

4.1. Information Sources

The publications concerning EEG-based BCI applications can be found from a high
number of different sources. In the current review, three well-known databases, i.e.,
PubMed [33], Scopus [34] and Web of Science [35] were selected to conduct the publication
Sensors 2022, 22, 3331 7 of 44

Sensors 2022, 22, 3331 8 of 44

search [5,28,29]. The search terms have been included in Appendix A. The process of
including publications in the analysis is shown in Figure 1.

Figure 1.
Figure Theflow
1. The flow
ofof information
information during
during selection
selection of studies
of studies according
according to thetoPRISMA
the PRISMA
modelmodel
com-
comprising Identification, Screening, Eligibility and Inclusion
prising Identification, Screening, Eligibility and Inclusion phase.phase.

During the full review of the 202 publications included in the analysis, 25 separate
During the full review of the 202 publications included in the analysis, 25 separate
data items were extracted from each publication. The extraction of the data items follows
data items were extracted from each publication. The extraction of the data items follows
the PRISMA model for preparing systematic literature reviews, and categorization of the
the PRISMA model for preparing systematic literature reviews, and categorization of the
extracted data items has also been applied in the past by Roy et al. [28] in a similar way
extracted data items has also been applied in the past by Roy et al. [28] in a similar way
while following the PRISMA model.
while following the PRISMA model.
4.2. Eligibility Criteria
4.2. Eligibility Criteria
Selection of the publications for the current review was based on pre-determined
Selection
eligibility of theThe
criteria. publications
eligibility for the current
criteria review wastobased
were determined filteron pre-determined
relevant el-
publications
igibility
concerningcriteria. The eligibility
EEG-based criteria were
BCI applications determined
as per objectivestooffilter relevant
this review publications
for con-
further analysis.
cerning EEG-based BCI applications as per objectives of this review for further
The eligibility criteria were selected according to similar principles used in previous review analysis.
The eligibility
studies criteria
and studies were
that selected
follow according
the PRISMA to similar
model principles used in previous re-
[5,23,28–30].
viewThestudies and studies that follow the PRISMA model [5,23,28–30].
following eligibility criteria were applied during review of the publications during
The following
screening eligibility
and eligibility criteria were applied during review of the publications dur-
assessment:
ing screening and eligibility assessment:
• Publications needed to be relatively current. In the further analysis, publications were
• Publications
included fromneeded to be
the period relatively
between 2009current.
and 2019.In the further analysis, publications
• were includedmeeting
We excluded from the period between
abstracts, 2009 and
book chapters, 2019.and doctoral dissertations and
masters
• We excludedpublications.
non-English meeting abstracts, book chapters, masters and doctoral dissertations
• and non-English
We excluded publications.
non-peer-reviewed journal articles and non-peer-reviewed confer-
• We excluded
ence proceedings.non-peer-reviewed journal articles and non-peer-reviewed conference
proceedings.
Following the above-mentioned inclusion and exclusion steps, a manual scanning
Following the
was conducted above-mentioned
for the titles, keywords inclusion and exclusion
and abstracts steps, a manual
of the publications. scanning
The screening
was conducted
process for the titles,
was conducted by one keywords and abstractsthat
person. Publications of the
did publications.
not address the Thesubject,
screening
but
rather mentioned
process the subject
was conducted by onein passing
person. in general that
Publications weredidexcluded. During
not address the review,
the subject, but
402 publications
rather mentioned were excluded
the subject that did
in passing in not meetwere
general the excluded.
eligibility During
criteria.the
The scanning
review, 402
resulted in 233
publications werepublications
excluded that for did
further full text
not meet assessment
the eligibility of eligibility.
criteria. The overview
The scanning resulted
in 233 publications for further full text assessment of eligibility. The overview concerning
the process of identification, screening, assessing the eligibility and inclusion of the pub-
lications in the final analysis as per PRISMA model is shown in Figure 1.
Sensors 2022, 22, 3331 8 of 44
Sensors 2022, 22, 3331 9 of 44

concerning the process of identification, screening, assessing the eligibility and inclusion of
During the full text review of 233 publications, 31 publications which did not corre-
the publications in the final analysis as per PRISMA model is shown in Figure 1.
spondDuring
to the eligibility criteria
the full text reviewwere further
of 233 excluded. After
publications, all steps ofwhich
31 publications publication
did not selec-
cor-
tion per PRISMA model were completed, 202 publications were included
respond to the eligibility criteria were further excluded. After all steps of publication (i.e., left at the
completion
selection perofPRISMA
the process)
model for were
the final analysis,202
completed, which are [1–11,13–20,24,26,36–216].
publications were included (i.e., left at
the completion of the process) for the final analysis, which are [1–11,13–20,24,26,36–216].
5. Results
5. Results
The results have been divided into Sections 5.1–5.9. The first part of the results covers
the distribution
The results of articles
have been and conference
divided proceedings
into Sections 5.1–5.9.perThe year,
firstpublications
part of the by re-
results
gions/continents, experimental
covers the distribution publications
of articles per year
and conference and publication
proceedings distribution
per year, publicationsby do- by
main. The sections inexperimental
regions/continents, the second part focus on per
publications the year
EEGanddevices used, number
publication of EEG
distribution by
channels,
domain. Thesignal analysis,
sections techniques
in the usedfocus
second part to obtain EEG
on the EEGdata, feature
devices extraction
used, number andof clas-
EEG
sification.
channels, The results
signal reflecttechniques
analysis, the publications
used tomatching
obtain EEG the search
data, terms
featureused during and
extraction the
Identification
classification.phase.
The results reflect the publications matching the search terms used during
the Identification phase.
5.1. Articles and Conference Proceedings per Year
5.1. Articles and Conference Proceedings per Year
During the period from 2009 to 2019, the overall number of articles and conference
During published
proceedings the periodper from 2009
year hadtobeen
2019, the overall
gradually number
rising. In theofbeginning
articles and conference
of the period,
the total number of the publications had been 11–15 publications per year from the
proceedings published per year had been gradually rising. In the beginning of 2009period,
until
the total
2011. number
During of the publications
the period from 2012 to had 2016,been 11–15
there was publications
a slight increaseper in
year
thefrom
volume2009ofuntil
the
2011. During the period from 2012 to 2016, there was a slight increase
publications to 16–19 publications, and only in year 2014 was there a decrease to 10 pub- in the volume
of the publications
lications to 16–19 publications,
per year. Significant increases canand only infrom
be noted yearyear
20142017
was onwards
there a decrease
when the to
10 publications per year. Significant increases can be noted from year
number of publications per year had increased to up to 32 publications per year in 2017.2017 onwards when
the number
Please see theofoverview
publications per yearthe
concerning hadnumber
increased to up toand
of articles 32 publications per year in 2017.
conference proceedings per
Please see the overview concerning the number of articles and conference
year in Figure 2. As the search of the publications was conducted in the three databases proceedings per
year in Figure 2. As the search of the publications was conducted in the three databases
during the period from 20 October 2019 to 30 October 2019, the final number of publica-
during the period from 20 October 2019 to 30 October 2019, the final number of publications
tions for 2019 could be a slightly higher number.
for 2019 could be a slightly higher number.

Thenumber
Figure2.2.The
Figure numberofofpublications
publications(conference
(conferenceproceedings
proceedingsand
andarticles)
articles)per
peryear
yearfrom
fromthe
theperiod
period
from January 2009 to October 2019.
from January 2009 to October 2019.

Thetrend
The trendconcerning
concerningincrease
increase
inin
thethe overall
overall number
number of publications
of publications per year
per year has
has also
also been noted by Al-Nafjan et al. [5] when analyzing the volume of EEG-based emotion
been noted by Al-Nafjan et al. [5] when analyzing the volume of EEG-based emotion
recognition publications. In the aforementioned article, there has been a rapid increase in the
recognition publications. In the aforementioned article, there has been a rapid increase in
overall number of publications on the topic from 2010 to 2016, and it was suggested that the
the overall number of publications on the topic from 2010 to 2016, and it was suggested
increase could be caused by increased knowledge of neurobiological processes, computers
that the increase could be caused by increased knowledge of neurobiological processes,
 number of conference proceedings published on EEG-based BCI ap
higher than the number of articles published. Out of the 202 publications
proceedings (58% of the publications) and 85 are articles (42% of the pub
Sensors 2022, 22, 3331
in the years from 2009 until 2016 the number of conference proceedin 9 of 44

than the number of published articles (except for year 2014), a trend ha
cent years
with faster concerning
computational the increase
processing, in the relative
greater availability volume
of devices of the
for recording publishe
brain
signalsTheand more
reason powerful
whysignal theprocessing
proportion and machine learning
of articles algorithms.
has increased Increase
over the
in the number of publications concerning EEG-based BCIs has been also demonstrated by
be
Hwangtheetoverall development
al. [99], where of thehastechnology
a significant increase been illustratedmaking
during theit easier
period fromand m
duct
2007 to the
2011.research on EEG-based BCI applications and inclusion of th
It can be noted from Figure 2 that during the period from 2009 to 2019, the overall
articles. In the beginning of the period under current focus, a higher v
number of conference proceedings published on EEG-based BCI applications has been
of publications
higher than the number were included
of articles published.in conference
Out proceedings.
of the 202 publications, This trend n
117 are conference
proceedings (58% of the publications) and 85 are articles (42% of the publications). Although
review illustrates the observation by Roy et al. [28] noting that ther
in the years from 2009 until 2016 the number of conference proceedings has been greater
variety of research
than the number ideas
of published within
articles (exceptdifferent repositories
for year 2014), andnoted
a trend has been among for diff
recent years concerning the increase in the relative volume of
cations. The effect highlights the need to include a higher variety of r the published articles.
The reason why the proportion of articles has increased over the recent years could be
ferent
the overalltype of publications
development of the technologyin making
the literature
it easier andreview in order
more efficient to conduct totheinclude
of the on
research research
EEG-basedideas on the topic
BCI applications and of
and inclusion avoid possible
the topic more in thepublication
articles. In the bias.
beginning of the period under current focus, a higher variety and number of publications
were included in conference proceedings. This trend noted in the current review illustrates
5.2. Publications
the observation by Royper
et al.Region/Continent
[28] noting that there would be a wide variety of research ideas
within different repositories and among different type of publications. The effect highlights
The
the need distribution
to include of the
a higher variety publications
of repositories concerning
and different EEG-based
type of publications in the BCI
further analyzed
literature review by
in order region/continent.
to include Out
objective coverage ofresearch
of the the 202 publications,
ideas on the topic the
and avoid possible publication bias.
publications on the topic has been published in Asia (111 publications
number of publications
5.2. Publications per Region/Continentper country has been published in China (39 p
The distribution of the publications
The other regions/continents concerning
where EEG-based
higherBCI applications
number of were
publicatio
further analyzed by region/continent. Out of the 202 publications, the highest number of
lished on on
publications thethetopic during
topic has the period
been published in Asiaof(111
2009 to 2019 are
publications) whereEurope (56 publ
the highest
America (27 publications).
number of publications per country has Inbeen
Europe, most
published productive
in China countries have
(39 publications).
The other regions/continents where higher number of publications have been pub-
Italy (both
lished on countries
the topic during thewith
period9of
publications), and (56
2009 to 2019 are Europe in North America,
publications) and Norththe high
lications
America (27 have been In
publications). published
Europe, mostin the USA
productive (23 publications).
countries have been Germany Similar
and res
Italy (both countries with 9 publications), and in North America, the highest number of
found by Hwang et al. [99] and Roy et al. [28] where USA and China hav
publications have been published in the USA (23 publications). Similar results have also
countries
been found bywith
Hwang the highest
et al. number
[99] and Roy ofwhere
et al. [28] publications on have
USA and China EEG-based
dominated BCI. A
as the countries with the highest number of publications
concerning the publications per region/continent is presented on EEG-based BCI. A detailed
in Figure
overview concerning the publications per region/continent is presented in Figure 3.

Figure 3. Number of publications per region/continent. The highest number of publications on the
Figure 3. Number
topic has been ofAsia.
published in publications per region/continent. The highest number o
topic has been published in Asia.

Altogether, 37 different countries have contributed to the publicat

BCI applications within the time period reviewed. The highest num
countries came from Europe, where the publications have been contrib
 Sensors 2022, 22, 3331 10 of 44

Sensors 2022, 22, 3331 11 of 44

Altogether, 37 different countries have contributed to the publications on EEG-based
BCI applications within the time period reviewed. The highest number of contributing
countries came from Europe, where the publications have been contributed by 18 different
but the average
countries. number
From of publications
Asia, the publicationsper country
have been is higher. From
contributed by North America,
14 different pub-
countries,
lications
but thehave beennumber
average contributed by three countries:
of publications USA, Canada
per country and From
is higher. Mexico. In Australia,
North America,
7 publications
publications were
have published, but from
been contributed bySouth
threeAmerica
countries:only one Canada
USA, publication
andfrom BrazilIn
Mexico.
has been published. It is interesting to note that no publications were contributed
Australia, 7 publications were published, but from South America only one publication from
from
Africa. It is important to note that the results reflect the publications matching the
Brazil has been published. It is interesting to note that no publications were contributed search
terms
fromused.
Africa. It is important to note that the results reflect the publications matching the
search terms used.
5.3. Experimental Publications per Year
5.3.Among
Experimental Publications per
the experimental Year
publications on EEG-based BCI applications, non-medical
Amonghave
publications the experimental
contributed the publications
majority ofontheEEG-based
experimental BCIpublications
applications,per
non-medical
year. Ex-
publications
ceptions to thishave
trendcontributed
have been seenthe majority of the
in 2009 and 2013 experimental publications
when more medical per year.
publications
Exceptions
were to this
published whentrend have been
compared seen
to the in 2009ofand
number 2013 whenpublications
non-medical more medical perpublications
year. Alt-
were in
hough published whenhas
general there compared
been higherto the number
number of non-medical
of non-medical publications
publications perdur-
per year year.
Although in general there has been higher number of non-medical publications
ing the period under review, during recent years the proportion of medical publications per year
during the period under review, during recent
has risen, reaching up to 36% of the publications. years the proportion of medical publications
hasDuring
risen, reaching up to the
recent years, 36%overall
of the publications.
number of experimental publications has increased
from 8–15 publications per yearoverall
During recent years, the duringnumber of experimental
2014–2016 publications
to 19–25 publications per has
yearincreased
during
from 8–15 publications per year during 2014–2016 to 19–25 publications
the period of 2017–2019. The trend of a rapid increase in the overall number of publica- per year during
the period of 2017–2019. The trend of a rapid increase in the overall
tions on EEG-based BCI applications can also be seen in Hwang et al. [99]. number of publications
on EEG-based
ThroughoutBCI theapplications
period that wascan also be seen
analyzed, in Hwang
there has beenet 99
al. publications
[99]. (58% of the
Throughout the period that was analyzed, there has been 99 publications (58% of
publications) published in the non-medical domain and 53 publications (31% of the pub-
the publications) published in the non-medical domain and 53 publications (31% of the
lications) in the medical domain. In 19 cases (11% of the publications) the publications
publications) in the medical domain. In 19 cases (11% of the publications) the publications
cover both domains. During interpretation of the results, it should be also noted that as
cover both domains. During interpretation of the results, it should be also noted that as
the database search was conducted during the period from 20 October 2019 to 30 October
the database search was conducted during the period from 20 October 2019 to 30 October
2019, the final number of publications for 2019 could have been slightly increased by the
2019, the final number of publications for 2019 could have been slightly increased by the
end of 2019. Please see the overview concerning the number of experimental publications
end of 2019. Please see the overview concerning the number of experimental publications
per year in Figure 4.
per year in Figure 4.

Figure
Figure 4.4. Experimental
Experimental publications
publications per
per year
year covering
covering the
the period
period from
from January
January 2009
2009 toto October
October 2019.
2019.
The publications have been
The publications have been further divided into medical or non-medical domain or both in case
divided into medical or non-medical domain or both in case both
both domains
domains have
have been
been covered.
covered.

Althoughinitially
Although initiallythe
theEEG-based
EEG-based BCI
BCI applications
applications were
were mainly
mainlydeveloped
developedfor
formedical
med-
ical reasons to help patients to communicate, grasp objects, move around and support inin
reasons to help patients to communicate, grasp objects, move around and support
otherdaily
other dailyactivities,
activities,the
thefocus
focushas
hasbeen
beenmoving
movingfrom
fromthe
themedical
medicaldomain
domaintotonon-medical
non-medical
applications.The
applications. Theshift
shiftininthe
thefocus
focus does
does not
not reduce
reduce the
the importance
importance ofofthese
theseapplications
applicationsin
in the medical domain but rather shows the wider potential of EEG-based BCI applica-
tions and opens doors for new possibilities.
2022, 22, 3331
Sensors 2022, 22, 3331 11 of 44

5.4. Publication Distribution

the medical domain by shows
but rather Domain
the wider potential of EEG-based BCI applications
and opensthe
Among doors for new possibilities.
publications within the period from 2009 to 2019, there were 58
publications (99 publications)
5.4. Publication from the non-medical domain. Among the overall
Distribution by Domain
of publications
Among thethatpublications
were analyzed,
within the31% offrom
period the 2009
publications (53
to 2019, there publications)
were 58% of the are
publications (99 publications) from the non-medical domain. Among
medical domain and in an additional 11% of the publications (19 publications) the overall number
of publications that were analyzed, 31% of the publications (53 publications) are from
mainsthehave been
medical included.
domain Please
and in an see11%
additional theofdistribution of(19
the publications the publications
publications) both per do
Figure 5. Similar
domains prevalence
have been of publications
included. Please in medical
see the distribution domainperhas
of the publications been
domain in determ
Figure 5. Similar prevalence of publications in medical domain has
Al-Nafjan et al. [5], where 23% of the publications belonged to the medical domabeen determined by
Al-Nafjan et al. [5], where 23% of the publications belonged to the medical domain.

Figure 5. Distribution of publications per domain. The figure illustrates the distribution of the
Figure 5. Distribution of publications per domain. The figure illustrates the distribution of th
publications among medical and non-medical domain and volume of the publications covering
tions among medical and non-medical domain and volume of the publications covering both
both domains.

The The publications in the medical domain could be further divided based on the type
publications in the medical domain could be further divided based on
of applications into fields such as assistive, monitoring, rehabilitation, assessment and
of applications into fields
others. Applications such field
in the assistive as assistive, monitoring,
help users with rehabilitation,
disorders or disabilities to performassessm
others. Applications
daily tasks and provide in assistance.
the assistive
Withinfield help users
the medical domain,with disorders
the field or disabilitie
covering assistive
applications is the largest contributing 74% of the publications in the domain. The assistive
form daily tasks and provide assistance. Within the medical domain, the field
field includes studies on robotic arm movement in a medical setting [78], lower-limb
assistive applications
prosthesis control [155] isand
theintelligent
largestwheelchair
contributingdriving74% of[111].
system the publications
The applications in the
The assistive field includes studies on robotic arm movementthat
in the rehabilitation field for example help to restore the physical functions in have
a medical
been sett
lost by the patients due to accident or disease. Among studies analyzing possibilities for
lower-limb prosthesis control [155] and intelligent wheelchair driving system [1
rehabilitation there is an important role for the studies supporting the rehabilitation of
applications in the
stroke patients rehabilitation
[70,185,213]. field for example
As per Chaudhary help toBCIs
et al., the assistive restore the physical
are intended to f
that have
enablebeen lost
patients to by the patients
communicate dueexternal
or control to accident
devicesoranddisease. Among
rehabilitative BCIs studies
are a
intended to facilitate patient recovery [217]. When compared to assistive BCIs, the amount
possibilities for rehabilitation there is an important role for the studies supportin
of rehabilitative BCIs is lower, covering 9% of the publications. An important field in the
habilitation of stroke
medical domain patientscovering
is monitoring, [70,185,213].
9% of theAs per Chaudhary
publications et al.,
in the medical the assistive
domain.
intendedThe to enable
publicationspatients to communicate
concerning monitoring include or for
control
exampleexternal devices
monitoring and reha
emotional
changes in patients [43]. Among studies analyzing possibilities for rehabilitation there is an
BCIs are intended to facilitate patient recovery [217]. When compared to assisti
important role for the studies supporting the rehabilitation of stroke patients [70,185,213].
the amount
The fieldsof
of rehabilitative
“assessment” and BCIs “other”iscover
lower, covering
a total 9%
of 8% of the of the publications.
publications in the medical An im
field in the medical
domain, domain
both contributing 4%isofmonitoring,
the publicationscovering 9% of the publications in the
in the domain.
domain. The prevalence of different fields in the medical domain differ when compared to
previous work of Al-Nafjan et al. [5], as the most popular fields in this domain then were
The publications concerning monitoring include for example monitoring em
changes in patients [43]. Among studies analyzing possibilities for rehabilitation
an important role for the studies supporting the rehabilitation of stroke
[70,185,213]. The fields of “assessment” and “other” cover a total of 8% of the pub
 Sensors 2022, 22, 3331 12 of 44

assessment and monitoring. The difference would be explained by the difference in the
scope of the studies where in previous case the focus was on emotion recognition based
on EEG-based BCI, but in the current review we focus on EEG- based BCI applications
Sensors 2022, 22, 3331 13 of 44
in general. An overview covering the publications in the medical domain is presented in
Figure 6.

Distributionofofpublications
Figure6.6.Distribution
Figure publicationsininmedical
medicaldomain.
domain.InIn the
the medicaldomain,
medical domain,the
thelargest
largestfield
fieldisis
assistive
assistivefollowed
followedbybymonitoring,
monitoring,rehabilitation,
rehabilitation,assessment
assessmentandandother.
other.

Thelargest
The largest field
field in in
thethe non-medical
non-medical domain
domain is monitoring
is monitoring contributing
contributing 50% 50% ofar-
of the the
articles
ticles in non-medical
in the the non-medical domain.
domain. Additionally,
Additionally, the twothelarge
two large
fieldsfields are control
are control machine
machine and
and entertainment
entertainment contributing
contributing 17% and 17% andof16%
16% of the publications,
the publications, respectively.
respectively. Smaller Smaller cov-
coverage
erage in the non-medical domain is by the publications in the field of
in the non-medical domain is by the publications in the field of assistive applications cover- assistive applications
covering
ing 4% of the 4%publications
of the publications and authentication
and authentication coveringcovering
2% of the 2% of the publications.
publications.
Other different types of publications cover the
Other different types of publications cover the rest of 11% of the rest of 11% of thepublications
publicationsininthethe
domain. When compared to previous similar studies, Al-Nafjan
domain. When compared to previous similar studies, Al-Nafjan et al. [5] has also noted et al. [5] has also noted
thatthe
that themajority
majorityofofthe thepublications
publicationsininthe thenon-medical
non-medicaldomain domainconcerning
concerningEEG-based
EEG-based
BCI for emotion recognition has contributed to the monitoring field. An overview ofofthe
BCI for emotion recognition has contributed to the monitoring field. An overview the
publications in the non-medical domain is presented
publications in the non-medical domain is presented in Figure 7. in Figure 7.
InInthe
themonitoring
monitoring field,
field, aa total
total of
of4949publications
publicationshave havebeen
beenincluded.
included. The publications
The publica-
in the monitoring field include studies where data are collected from
tions in the monitoring field include studies where data are collected from the subjects the subjects in orderinto
monitor
order and analyze
to monitor their mental
and analyze their activity. The studies
mental activity. Thecover
studiesforcover
example mental fatigue
for example mentales-
timation [182], emotion recognition [7] and detecting meditation [4].
fatigue estimation [182], emotion recognition [7] and detecting meditation [4]. In the con- In the control machine
field, the applications include those that enable people to control different machines in
trol machine field, the applications include those that enable people to control different
their daily environment, via EEG, to make their life easier. These applications can be used
machines in their daily environment, via EEG, to make their life easier. These applications
for example for home appliance and smart home control [15,16] or robotic systems [38].
can be used for example for home appliance and smart home control [15,16] or robotic
EEG-based BCI applications are also used for entertainment such as live brain-computer
systems [38]. EEG-based BCI applications are also used for entertainment such as live
cinema performance [216], driving in a virtual city [19] or recommendations for music
brain-computer cinema performance [216], driving in a virtual city [19] or recommenda-
based on a person’s mood [174]. Other applications include user authentication using
tions for music based on a person’s mood [174]. Other applications include user authenti-
visual stimuli of geometric shapes [37], age and gender prediction [109] and other various
cation using visual stimuli of geometric shapes [37], age and gender prediction [109] and
possibilities. As illustrated in Figure 7, the majority of applications in the non-medical
other various possibilities. As illustrated in Figure 7, the majority of applications in the
domain are related to monitoring. As monitoring applications have many practical applica-
non-medical domain are related to monitoring. As monitoring applications have many
tions such as monitoring mental fatigue at the workplace or during driving monitoring
practical applications such as monitoring mental fatigue at the workplace or during driv-
field has high potential for further development and practical applications for daily life.
ing monitoring field has high potential for further development and practical applications
for daily life.
Sensors 2022,22,
Sensors2022, 22,3331
3331 13 of
14 of44
44

Figure 7. Distribution of publications in non-medical domain. In the non-medical domain, the larg-
est field is monitoring followed by control machine, entertainment and other smaller fields.

In a smaller number of the publications, both medical and non-medical domains have
been covered. The domain has been further divided into the following fields: assistive,
framework, control machine, monitoring and others, as per type of publications. The de-
Figure
Figure
tails 7.7.Distribution
Distribution
concerning ofpublications
publicationsof
theofdistribution ininnon-medical
non-medical
publications domain.
domain. InIn
covering thethe non-medical
non-medical
both domains domain,
domain,
have thethe
been larg-
largest
illus-
est field
field is in
trated is monitoring
monitoring followed
Figure 8.followed
Amongbystudies by control machine,
control covering entertainment
machine, entertainment and
and other
both the medical other
andsmallersmaller fields.
fields. field, 42%
non-medical
of the cases publications cover the assistive category, such as detection of imagined speech
InInaasmaller
smallernumber
numberof ofthe
thepublications,
publications,bothbothmedical
medicalandandnon-medical
non-medicaldomains
domainshave have
and classification of unspoken words from EEG signals [85,86]. A total of 21% of the stud-
been
beencovered.
covered. The Thedomain
domainhas hasbeen
beenfurther
furtherdivided
dividedinto
intothethefollowing
followingfields:
fields:assistive,
assistive,
ies concern works such as developing a new framework for practical BCI communication
framework,
framework,control controlmachine,
machine, monitoring
monitoring and
andothers, as per
others, typetype
as per of publications.
of publications.The The
detailsde-
development [57].
concerning
tails concerningthe distribution of publications
the distribution covering
of publications both domains
covering have been
both domains haveillustrated
been illus-
Studies
intrated
Figure in the field of control machine aremedical
covering 16% of the studies.
field,They include
in 8. Among
Figure studies
8. Among covering
studies both the
covering and
both the medical non-medical
and non-medical 42% of 42%
field, the
research
cases for example on controlling a category,
car in an such
experimental environment outside labora-
of thepublications cover
cases publications the assistive
cover the assistive category,as detection
such of imagined
as detection speech
of imagined and
speech
tory conditions
classification of [20]. Some words
unspoken studiesfromare focused
EEGEEG on monitoring
signals [85,86]. suchof
A total as detection of kines-
and classification of unspoken words from signals [85,86]. A total21%
of 21%of the studies
of the stud-
thetic
concern attention
works [147].
such asOther fields
developing cover
a newfor example
framework the
for development
practical BCI of serious
communication games
ies concern works such as developing a new framework for practical BCI communication
that could be used in entertainment, e-learning or medical applications [186].
development
development[57]. [57].
Studies in the field of control machine are covering 16% of the studies. They include
research for example on controlling a car in an experimental environment outside labora-
tory conditions [20]. Some studies are focused on monitoring such as detection of kines-
thetic attention [147]. Other fields cover for example the development of serious games
that could be used in entertainment, e-learning or medical applications [186].

Figure
Figure 8.
8. Distribution
Distributionofofpublications
publicationscovering
coveringbothbothdomains.
domains.The publications
The covering
publications both
covering do-
both
mains
domainshave been
have focused
been focusedmainly on on
mainly assistive, butbut
assistive, other fields
other such
fields as framework
such as frameworkandand
control ma-
control
chine alsoalso
machine make up up
make a significant proportion
a significant proportionof of
thethe
publications.
publications.

Studies in the field of control machine are covering 16% of the studies. They include
research
Figure 8.for example on
Distribution of controlling
publicationsacovering
car in anboth
experimental environment
domains. The outside
publications laboratory
covering both do-
conditions [20]. Some studies are focused on monitoring such as detection of kinesthetic
mains have been focused mainly on assistive, but other fields such as framework and control ma-
chine also make up a significant proportion of the publications.
Sensors 2022, 22, 3331 14 of 44

Sensors 2022, 22, 3331 15 of 44

attention [147]. Other fields cover for example the development of serious games that could
be used in entertainment, e-learning or medical applications [186].
5.5. EEG Devices Used
5.5. EEG Devices Used
Among the studies analyzing EEG-based BCI applications, the most commonly used
Among the studies analyzing EEG-based BCI applications, the most commonly used
EEG devices are Emotiv EPOC from Emotiv (San Francisco, USA), Quik-Cap from Com-
EEG devices are Emotiv EPOC from Emotiv (San Francisco, CA, USA), Quik-Cap from
pumedics Neuroscan (Charlotte, USA) and MindWave from NeuroSky (San Jose, USA).
Compumedics Neuroscan (Charlotte, NC, USA) and MindWave from NeuroSky (San Jose,
These three devices combined cover 57% of the publications on the topic. The most com-
CA, USA). These three devices combined cover 57% of the publications on the topic. The
mon EEG device, the Emotiv EPOC (Emotiv, San Francisco, USA), has been used in 40%
most common EEG device, the Emotiv EPOC (Emotiv, San Francisco, CA, USA), has been
of the publications analyzed while the prevalence of other popular EEG devices Quik-Cap
used in 40% of the publications analyzed while the prevalence of other popular EEG devices
(Compumedics Neuroscan, Charlotte, USA) and MindWave (NeuroSky, San Jose, USA)
Quik-Cap (Compumedics Neuroscan, Charlotte, NC, USA) and MindWave (NeuroSky, San
was 9% and 8%, respectively. The high prevalence of Emotiv EPOC (Emotiv, San Fran-
Jose, CA, USA) was 9% and 8%, respectively. The high prevalence of Emotiv EPOC (Emotiv,
cisco, USA) and Quik-Cap (Compumedics Neuroscan, Charlotte, USA) use among studies
San Francisco, CA, USA) and Quik-Cap (Compumedics Neuroscan, Charlotte, NC, USA)
analyzing
use amongEEG-based BCI applications
studies analyzing EEG-basedfor emotion
BCI recognition
applications has also
for emotion been noted
recognition byalso
has Al-
Nafjan
been et al.by[5]Al-Nafjan
noted and in general forand
et al. [5] EEG inequipment
general forby Roy
EEG et al. [28].by Roy et al. [28].
equipment
The most common EEG devices (Emotiv EPOC,
The most common EEG devices (Emotiv EPOC, Emotiv, San Emotiv, San Francisco,
Francisco, CA,
CA, USA;
USA;
Quik-Cap, Compumedics Neuroscan, Charlotte, USA; MindWave,
Quik-Cap, Compumedics Neuroscan, Charlotte, NC, USA; MindWave, NeuroSky, San Jose, NeuroSky, San Jose,
USA) are all manufactured in USA. Other EEG devices used in the studies
CA, USA) are all manufactured in USA. Other EEG devices used in the studies have been have been man-
ufactured also also
manufactured in Germany (Easycap,
in Germany Easycap
(Easycap, GmbH,
Easycap Wörthsee,
GmbH, Germany;
Wörthsee, actiCAP,
Germany; Brain
actiCAP,
Products
Brain GmbH,
Products Gliching,
GmbH, Germany),
Gliching, Germany),Netherlands
Netherlands(Active
(ActiveTwo,
Two,Biosemi,
Biosemi, Amsterdam,
Amsterdam,
The Netherlands), Spain (Enobio, Neuroelectrics, Barcelona, Spain)
The Netherlands), Spain (Enobio, Neuroelectrics, Barcelona, Spain) and UK (MyndPlay and UK (MyndPlay
BrainBandXL, MyndPlay, London, UK). The details of different EEG
BrainBandXL, MyndPlay, London, UK). The details of different EEG devices used, numberdevices used, num-
ber of publications and percentage of publications covered is presented
of publications and percentage of publications covered is presented in Figure 9. in Figure 9.

Figure9.9.EEG
Figure EEGdevices
devicesused
usedininthe
the publications
publications ordered
ordered by by
thethe number
number andand percentage
percentage of publica-
of publications
tions where the device has been
where the device has been used. used.

Emotiv
Emotiv EPOC
EPOC (Emotiv,
(Emotiv, San San Francisco, USA)
CA, USA) and MindWave
and MindWave (NeuroSky,
(NeuroSky, San
San Jose,
Jose, SC, USA) are considered to be low-cost consumer EEG devices,
SC, USA) are considered to be low-cost consumer EEG devices, while Quik-Cap (Com- while Quik-Cap
(Compumedics
pumedics Neuroscan,Neuroscan, Charlotte,
Charlotte, USA)NC, USA)expensive
is more is more expensive
to purchaseto purchase
[28]. While[28].
theWhile
Emo-
the Emotiv EPOC (Emotiv, San Francisco, CA, USA) EEG device
tiv EPOC (Emotiv, San Francisco, USA) EEG device has 14 channels and MindWave (Neu- has 14 channels and
MindWave (NeuroSky, San Jose, CA, USA) has 1 channel, the Quik-Cap
roSky, San Jose, USA) has 1 channel, the Quik-Cap (Compumedics Neuroscan, Charlotte, (Compumedics
Neuroscan,
USA) EEG deviceCharlotte,
usesNC, USA) EEGThe
32 channels. device uses
cause of 32
thechannels. The cause of
general popularity of the
the general
Emotiv
popularity of the Emotiv EPOC (Emotiv, San Francisco, CA, USA)
EPOC (Emotiv, San Francisco, USA) EEG device is due to the relative low cost of the EEG device is duede-
to
the relative low cost of the device, sufficient number of EEG channels and
vice, sufficient number of EEG channels and the device being considered easy to use. The the device being
considered
MindWave easy to use. San
(NeuroSky, The Jose,
MindWave
USA) EEG(NeuroSky,
device has Sanlimitations
Jose, CA, USA)
for useEEG
duedevice has
to the lim-
limitations for use due to the limited number of EEG channels, but the low
ited number of EEG channels, but the low cost and ease of use make the device still pop- cost and ease of
use make the device still popular. The device could be applied in specific
ular. The device could be applied in specific applications that do not require a higher applications that
do not require
number of EEG a higher number
channels. of EEG channels.
The Quik-Cap The Quik-Cap
(Compumedics (Compumedics
Neuroscan, Charlotte, Neuroscan,
USA) de-
Charlotte,
vice on the NC, USA)
other handdevice on the
is more other hand
expensive butishas
moretheexpensive
advantage butofhas the advantage
a higher number of of
a higher number of EEG channels.
EEG channels.
Dadebayev et al. [218] have found that low-cost consumer-grade EEG devices can
perform equally well as research-grade devices for EEG-based emotion recognition. The
Sensors 2022, 22, 3331 15 of 44

Dadebayev et al. [218] have found that low-cost consumer-grade EEG devices can
perform equally well as research-grade devices for EEG-based emotion recognition. The
advantages of the Emotiv EPOC (Emotiv, San Francisco, CA, USA) device are the high
quality of the signal, less artifacts, 14 sensors, ready to use assembly and modern design.
The disadvantages of the Emotiv EPOC (Emotiv, San Francisco, CA, USA) device are
being non-dry sensors, 10–15 min setup time and required license for raw data access. For
MindWave (NeuroSky, San Jose, CA, USA), the advantages are its low price, compactness
and practical wireless device. The disadvantages of MindWave (NeuroSky, San Jose, CA,
USA), as highlighted by Dadebayev et al., are the inclusion of only 1 sensor and a lower
quality for EEG experiments.
The low-cost EEG headsets can be more convenient for the user and the devices
such as Emotiv EPOC (Emotiv, San Francisco, CA, USA) and MindWave (NeuroSky, San
Jose, CA, USA) have been successfully used in the studies developing EEG-based BCI
applications [219]. The final decision concerning the use of a specific EEG device depends
on the type of EEG-based BCI application determining the need for a specific number of
EEG channels. The decision also depends on the cost planned for the study and devices
for the end users. The EEG equipment market shows rapid development and new devices
appear on the market continuously [220].
Concerning the design of the EEG devices it has been highlighted by Soufineyestani
et al. [221] that the EEG headsets with dry electrodes such as MindWave (NeuroSky, San
Jose, CA, USA) have been more robust to line noise when compared to EEG devices with
wet electrodes, such as Emotiv EPOC (Emotiv, San Francisco, CA, USA) or Quick-Cap.
According to Soufineyestani et al., the dry electrodes contain more artifacts. The dry EEG
devices may lose humidity during use that may lead to decline in signal quality. In case of
wet electrodes, the solution could evaporate over time, and it could be necessary to reapply
the solution to the electrodes. The most convenient for the user would be dry wireless
EEG devices that enable flexibility in movement and lower setup time. In case of dry
electrodes, it would need to be considered that dry electrodes present lower performance
when compared to wet electrodes [222]. It has been also highlighted by Soufineyestani
et al. that when using wireless connectivity, the data would need to be encrypted prior to
wireless transfer in order to avoid any security risk to the data of the user.
Similarly, Hinrichs et al. [223] have compared the use of dry and wet EEG systems
for clinical applications and have found that although the number of artifacts was slightly
higher for the dry EEG system the results were comparable between dry and wet EEG sys-
tems. The dry EEG system is more robust and less sensitive to electromagnetic interference
that the subject could encounter at the clinic or at home. It has been stated by Hinrichs
et al. that importantly both patients and healthy volunteers preferred the dry electrodes
and reported that the dry headset was more suitable for self-application and home use.
Ratti et al. [224] have shown that although medical EEG systems offer clear advantages
in data quality over consumer systems, EEG data can be collected from all systems tested
in both medical and consumer contexts. Consumer EEG systems are more convenient and
faster to set up. According to Ratti et al., consumer EEG systems would be more useful
for a quick assessment when time is limited, while the medical grade systems would be
preferred in research and clinical trial settings.
In the study by Zerafa et al. [225], a comparison was performed between a broad range
of EEG devices for SSVEP BCIs. Zerafa et al. showed that low-end research grade EEG
systems are comparable to the high-end research grade EEG systems in terms of signal
quality. Low-cost EEG headsets suffer more from poor EEG signal quality, and it could be
suggested that the best choice for developing BCI applications outside of the laboratory
setting would be affordable wireless research grade systems.
When comparing the advantages and disadvantages of low-cost and more expensive
EEG headsets LaRocco et al. [219] have found that traditional medical- and research-grade
EEG systems have been successfully used in various applications, but are less versatile
outside a laboratory environment. Low-cost EEG headsets show greater design convenience
 (Emotiv, San Francisco, USA) and MindWave (NeuroSky, San Jose, USA) have been used,
for example, as drowsiness detectors. LaRocco et al. have further suggested that the use
of open-source software and occupational refinement may further boost the capabilities
of the systems over time.
Sensors 2022, 22, 3331 16 of 44

5.6. Number of EEG Channels

Among the studies analyzed, the number of EEG channels used varies across a large
range. Theoccupational
for daily useof
highest number and
EEGmany of theused
channels low-cost devices
was 163 including
by Zhou Emotiv
et al. for EPOC
EEG-based
(Emotiv, San Francisco, CA, USA) and MindWave (NeuroSky, San Jose, CA,
classification of elbow versus shoulder torque intentions [213]. Altogether there were USA) have
5
been used, for example, as drowsiness detectors. LaRocco et al. have further
studies (3% of studies) where more than 100 EEG channels were used, indicating that the suggested
thatofthe
use thisuse of number
high open-source software
of EEG andisoccupational
channels refinement
rather exceptional. may further
The majority of theboost the
studies
capabilities of the systems over time.
(64% of studies) involve 1–20 EEG channels, where 34% of the studies employ 1–10 EEG
channels and 30% of the studies employ 11–20 EEG channels. There are also a significant
5.6. Number of EEG Channels
number of studies that use 21–40 EEG channels and 11 studies have been conducted with
61–70 Among the studies analyzed, the number of EEG channels used varies across a large
EEG channels. The detailed overview regarding the number of EEG channels used
range. The highest number of EEG channels used was 163 by Zhou et al. for EEG-based
in the studies is presented in Figure 10.
classification of elbow versus shoulder torque intentions [213]. Altogether there were
Previously, Roy et al. [28] found that for obtaining EEG signals, 1 to 256 electrodes
5 studies (3% of studies) where more than 100 EEG channels were used, indicating that the
have been used, with half of the studies using between 8 to 62 electrodes. It was thought
use of this high number of EEG channels is rather exceptional. The majority of the studies
that a very high number of electrodes would not give added value to the studies, more
(64% of studies) involve 1–20 EEG channels, where 34% of the studies employ 1–10 EEG
important is the exact location of smaller number of electrodes. In that article, it was found
channels and 30% of the studies employ 11–20 EEG channels. There are also a significant
that, concerning the number of electrodes, there is a significant increase in sensitivity and
number of studies that use 21–40 EEG channels and 11 studies have been conducted with
specificity when increasing the number of channels up to 22, but a further increase in the
61–70 EEG channels. The detailed overview regarding the number of EEG channels used in
number of EEG channels would not give a similar advantage.
the studies is presented in Figure 10.

Figure10.
Figure NumberofofEEG
10.Number EEGchannels
channelsused
usedininpublications.
publications.InInthe
themajority
majorityofofthe
thepublications,
publications,up
uptoto
40 EEG channels have been used.
40 EEG channels have been used.

Previously,
Al-Nafjan Roy
et al. [5]ethave
al. [28]
alsofound that for
concluded thatobtaining EEG
the majority ofsignals, 1 to up
studies use 256toelectrodes
64 chan-
have been used, with half of the studies using between 8 to 62 electrodes.
nels for obtaining EEG data. In the study by Al-Nafjan et al., it has been emphasizedIt was thought
that
that aplanning
when very high
thenumber
use of aofspecific
electrodes would
number not give
of EEG addeditvalue
channels, to the studies,
is important more
to also con-
important is the exact location of smaller number of electrodes. In that article, it was found
sider the time required for the system setup, the comfort level for the subject and the num-
that, concerning the number of electrodes, there is a significant increase in sensitivity and
ber of features to be processed. When planning studies or developing applications for us-
specificity when increasing the number of channels up to 22, but a further increase in the
ers, it is important to select the electrode positions carefully and limit the number of EEG
number of EEG channels would not give a similar advantage.
channels. A limited number of carefully selected electrode positions would make the fu-
Al-Nafjan et al. [5] have also concluded that the majority of studies use up to 64 channels
ture devices more user friendly and optimize the system performance.
for obtaining EEG data. In the study by Al-Nafjan et al., it has been emphasized that when
planning the use of a specific number of EEG channels, it is important to also consider the
time required for the system setup, the comfort level for the subject and the number of
features to be processed. When planning studies or developing applications for users, it is
important to select the electrode positions carefully and limit the number of EEG channels.
A limited number of carefully selected electrode positions would make the future devices
more user friendly and optimize the system performance.
 Among the publications applying motor-imagery paradigm, there were 62% of the
publications in the medical domain and 24% of the publications in the non-medical do-
main. In addition to the aforementioned, 14% of the publications covered both the do-
mains. Among the medical domain there are innovative applications such as EConHand
[163], neuro-rehabilitation using virtual reality feedback [106] and intelligent brain-con-
Sensors 2022, 22, 3331 17 of 44
trolled robotic limbs [158], applying the motor-imagery paradigm.
In the visual evoked potential paradigm, the non-medical domain is dominant as in
61% of the reviewed publications (where the technique used was stated) applied this tech-
5.7. Technique
nique. Used to Obtain
In the medical domain, EEGtheData
prevalence of this type of technique used was 23% and
bothIn domains
the publications, a variety of of
were included in 16% the cases.are
techniques Visual
usedevoked
for thepotential
interactionparadigm
between is
widespread among BCI applications as it has been long used and tested
brain and computer in BCI applications. The most popular techniques include the motor- in a high number
of previous
imagery studies.
paradigm The visual
(applied in 38evoked potential
publications), paradigm
visual evokedalso involves
potential the visual
paradigm P300
(applied
paradigm
in and the steady
31 publications) state visualdrowsiness/attention
and monitoring evoked potential paradigm
(applied[1].in Among the most
29 publications)
interesting
or applicationsstates
emotions/affective of the(applied
aforementioned technique inThe
in 15 publications). theoverview
non-medical domain the
concerning are
the deceit identification
prevalence test [74], use
of different techniques of EEG-based
is shown in FigureBCI
11. devices to subliminally
These techniques probe
have been for
used
information
in 73% of the[83] and authentication
reviewed based on
publications (where theemotionally significant
technique used images [166].
was stated).

Figure 11.
Figure 11. Techniques
Techniques used
used in
in the
the publications.
publications. The
The techniques
techniques have
have been
been represented
represented according
according to
to
the number of publications where the techniques have been used and prevalence among medical,
the number of publications where the techniques have been used and prevalence among medical,
non-medical or both domains.
non-medical or both domains.

There is the
Among significant interest
publications in monitoring
applying the mindparadigm,
motor-imagery for drowsiness/attention
there were 62%orofemo- the
tions/affective states. Both techniques use the analysis of changes
publications in the medical domain and 24% of the publications in the non-medical in the EEG spectrumdomain. to
determine the states and changes in one’s mind. The vast majority of
In addition to the aforementioned, 14% of the publications covered both the domains. the publications con-
cerning the
Among drowsiness/attention
medical domain there (88%)areor innovative
emotions/affective statessuch
applications (93%) as are categorized
EConHand in
[163],
the non-medical domain. The monitoring of drowsiness/attention based
neuro-rehabilitation using virtual reality feedback [106] and intelligent brain-controlled on EEG spectral
changeslimbs
robotic has high
[158],practical
applying value as there are a number
the motor-imagery of studies implementing the tech-
paradigm.
niqueIninthe
order to create helmets for on-site workers for drowsiness
visual evoked potential paradigm, the non-medical domain detection [14], predict-
is dominant as
in 61% of the reviewed publications (where the technique used was stated) order
ing driver fatigue [134] or implementing EEG-based attention feedback in appliedto this
im-
prove focusIninthe
technique. e-learning
medical [171].
domain, the prevalence of this type of technique used was 23%
Other
and both techniques
domains wereused in theinpublications
included are applied
16% of the cases. less frequently,
Visual evoked potentialbut there isisa
paradigm
general trend
widespread notedBCI
among in the overall use
applications as of various
it has beentechniques.
long used andAs in the case
tested of motor-im-
in a high number
agery,
of auditory
previous evoked
studies. The potential and vibrotactile
visual evoked evoked potential
potential paradigm technique,
also involves the main
the visual P300
paradigm and the steady state visual evoked potential paradigm [1]. Among the most
interesting applications of the aforementioned technique in the non-medical domain are
the deceit identification test [74], use of EEG-based BCI devices to subliminally probe for
information [83] and authentication based on emotionally significant images [166].
There is significant interest in monitoring the mind for drowsiness/attention or emo-
tions/affective states. Both techniques use the analysis of changes in the EEG spectrum
to determine the states and changes in one’s mind. The vast majority of the publications
concerning drowsiness/attention (88%) or emotions/affective states (93%) are categorized
in the non-medical domain. The monitoring of drowsiness/attention based on EEG spec-
tral changes has high practical value as there are a number of studies implementing the
technique in order to create helmets for on-site workers for drowsiness detection [14],
predicting driver fatigue [134] or implementing EEG-based attention feedback in order to
improve focus in e-learning [171].
Other techniques used in the publications are applied less frequently, but there is
a general trend noted in the overall use of various techniques. As in the case of motor-
imagery, auditory evoked potential and vibrotactile evoked potential technique, the main
 5.8. Feature Extraction
The feature extraction is an important processing step in order to
details from the wide range of signals collected during acquisition o
Sensors 2022, 22, 3331 previous studies, high number of different methods have been 18 ofused
44 for
According to Al-Nafjan et al. [5] the feature extraction step is one of th
in BCI systems
application and the
is in the medical technique
domain, is not
but the majority andoptimal across
larger variety different
of techniques are appl
view in
applied concerning
the non-medicalthe prevalence
domain. of different
The trend indicates the highmethods for feature ext
potential of non-medical
applications for the EEG-based BCI applications. When taking into consideration that the
studiesof investigating
majority general populationEEG-based BCI applications
apply for non-medical domain, vast numberis presented
of people couldin Figu
benefitAmong theof studies
from this type applications reviewed
in the future.the most frequent feature extracti
analysis
5.8. of power spectral density (used in 23 publications), Fourier tra
Feature Extraction
publications) or theisanalysis
The feature extraction an important ofprocessing
common stepspatial
in order topattern
extract the(used
relevantin 18 p
details
gether, fromthetheaforementioned
wide range of signals publications
collected during acquisition
contribute of the
to EEG
48%data. In revi
of the
previous studies, high number of different methods have been used for feature extraction.
These results
According areetsimilar
to Al-Nafjan to previous
al. [5] the feature extraction work byofAl-Nafjan
step is one et al.in[5], wh
the major challenges
spectral
BCI systemsdensity and Fourier
and the technique transform
is not optimal have been
across different mostThe
applications. frequently
overview used
concerning the prevalence of different methods for feature extraction among the studies
tion in the studies to classify emotional features from EEG.
investigating EEG-based BCI applications is presented in Figure 12.

Figure 12. Feature extraction used in the publications ordered by the number and percentage of
Figure 12. Feature extraction used in the publications ordered by the numb
publications where the extraction method has been used.
publications where the extraction method has been used.
Among the studies reviewed the most frequent feature extraction approaches are
analysis of power spectral density (used in 23 publications), Fourier transform (used in
According
20 publications) or theto Xie et
analysis al. [226]spatial
of common the pattern
advantage
(used inof18 power spectral
publications). Alto- dens
ity. Intheterms
gether, of disadvantages,
aforementioned the power
publications contribute to 48%spectral density
of the reviewed is not suitab
publications.
These results are similar to previous work by Al-Nafjan et al. [5], who state that power spec-
nals, and it would not be possible to analyze time-domain signals. The
tral density and Fourier transform have been most frequently used for feature extraction in
is, studies
the according to emotional
to classify Xie et al., applicable
features from EEG.for stationary signals and is appro
band According
signals,to Xie
butet al. [226] not
does the advantage
work for of power spectral densitysignals,
nonstationary is feature stability.
greatly sust
In terms of disadvantages, the power spectral density is not suitable for unstable signals,
ity itand
and does
would not benot havetoshorter
possible duration signals.
analyze time-domain data record.
The FourierCommon spatial
transform is,
effect when
according to Xie etapplied to EEG
al., applicable data signals
for stationary basedand onis appropriate
motor imagination
for narrowbandand do
signals, but does not work for nonstationary signals, greatly sustains noise sensitivity and
selection
does not haveofshorter
a specific
durationfrequency band inspatial
data record. Common advance, buta is
pattern has sensitive
good effect to
multiple
when appliedelectrodes
to EEG data based[226].on motor imagination and does not require the selection
of a specific frequency band in advance, but is sensitive to noise and requires multiple
Other methods for feature extraction are used less frequently, but
electrodes [226].
ferent feature extraction methods are used within one study. Similar
concerning the use of power spectral density and use of different featu
ods within studies has been noted by Al-Nafjan et al. [44], for analyzin
 5.9. Classification
For the purpose of classification, different machine learning
used in previous studies for EEG-based BCI applications. The m
Sensors 2022, 22, 3331 learning algorithms used are linear discriminant analysis 19 ofand
44 sup
which have been applied in 52% of the studies. Other machine lear
theOther
coverage of 2–5%. Among all of the methods, linear discrimin
methods for feature extraction are used less frequently, but in many cases
applied
different in 31%
feature of the
extraction cases
methods areand support
used within vector
one study. machine
Similar in 21% of the
general prevalence
concerning the use of power spectral density and use of different feature extraction methods
of the classification methods employed in the reviewed studies is p
within studies has been noted by Al-Nafjan et al. [44], for analyzing human emotions from
Linear
EEG, Padfield discriminant
et al. [2] when analyzing analysis
EEG-basedhas BCIbeen characterized
interfaces as a simp
using motor-imagery
technique and Sourina et al. [178] for real-time brain state recognition from EEG.
computation requirements and acceptable accuracy [122]. Linear di
Our overall results correlate with previous studies in the field highlighting the im-
easy toofuse
portance withpower
analyzing low spectral
computational complexity,
density, the use but requires
of Fourier transform and common a linear
spatial pattern for feature extraction and emphasizing the need to apply different types of
vector machine is a speedy classifier that supports binary and multi
feature extraction methods depending on the application under study.
perform linear and nonlinear modes [122]. Support vector machine
5.9. Classification
compared to other linear classifiers, but it has low computational co
For the purpose of classification, different machine learning algorithms have been
used inOther
previousmethods appliedBCI
studies for EEG-based in applications.
EEG-based TheBCI applications
most common machine are m
learning
rithmsalgorithms
such asused are linear discriminant
convolutional neuralanalysis
network,and support
naivevector
Bayesmachine,
and rand
which have been applied in 52% of the studies. Other machine learning algorithms have
forcoverage
the 4–5% of of2–5%.
the cases
Among and artificial
all of the methods,neural network,
linear discriminant deephasneural
analysis been netw
applied in 31% of the cases and support vector machine in
model and k-nearest neighbors that account for 2–3% of the cases. 21% of the studies. An overview
of the classification methods employed in the reviewed studies is presented in Figure 13.

Figure 13. Classification method used in the publications ordered by the number and percentage of
Figure 13.
publications Classification
where the classificationmethod
method hasused in the publications ordered by
been used. the nu
publications where the classification method has been used.
Linear discriminant analysis has been characterized as a simple classifier with low
computation requirements and acceptable accuracy [122]. Linear discriminant analysis is
easy to The
use with results of the current
low computational complexity,review area in
but requires correlation
linear with previ
model [226]. Support
vector machine is a speedy classifier that supports binary and multi-class method and can
Nafjanlinear
perform et al.and[5] havemodes
nonlinear shown that
[122]. thevector
Support use of support
machine vector
performance machine i
is better
ular methods
compared used
to other linear for classification
classifiers, for EEG-based
but it has low computational complexity BCI
[226].systems for
Other methods applied in EEG-based BCI applications are machine learning algo-
According
rithms to the aforementioned
such as convolutional neural network, naive study, therandom
Bayes and choice forthat
forest classification
account
the4–5%
for type ofcases
of the brainandsignal
artificial being recorded
neural network, deep and
neuralthe typeGaussian
network, of application
mixed th
model and k-nearest neighbors that account for 2–3% of the cases.
Wide use of linear discriminant analysis and support vector machin
The results of the current review are in correlation with previous studies where Al-
also been
Nafjan shown
et al. [5] by Padfield
have shown that the useet of al. [2] for
support EEG-based
vector machine is one BCIs
of theusing
most moto
popular methods used for classification for EEG-based BCI systems for emotion recognition.
Dadabayev et al. [218], which pertains to feature extraction te
cation methods used for EEG-based emotion recognition, has stat
traction technique or classification method could be named exclusiv
Sensors 2022, 22, 3331 20 of 44

According to the aforementioned study, the choice for classification algorithm depends on
the type of brain signal being recorded and the type of application that is being controlled.
Wide use of linear discriminant analysis and support vector machine for classification has
also been shown by Padfield et al. [2] for EEG-based BCIs using motor-imagery.
Dadabayev et al. [218], which pertains to feature extraction techniques and classifica-
tion methods used for EEG-based emotion recognition, has stated that no feature extraction
technique or classification method could be named exclusively the best option for all cases,
but the right strategy relies on the specific system paradigm and objectives. It has been
suggested that it would be needed to consider many different machine learning algorithms
and compare the results before choosing the best model for the given objective [218]. Simi-
larly, during a comparison of machine learning methods for emotion recognition, including
linear discriminant analysis, support vector machine, naive Bayes and k-nearest neighbors,
Doma and Pirouz [227] did not find that any of the methods would outperform one another.
According to Nakagome et al. [228], when comparing neural networks and machine
learning algorithms for EEG gait decoding neural network-based decoders with downsam-
pling or a wide range of frequency band features could improve the decoder performance
and robustness for stable use of BCIs. Varszegi [229] has suggested that in order to improve
artificial neural networks it would be further needed to look inside them, develop methods
to analyze their internal activations, figure out the behavior of their architectural elements
and create knowledge basis for conscious artificial neural network design to handle EEG
signal processing tasks. According to Lotte et al. [22] the ideal classification method would
use features and classifiers that are invariant over time, over users and contexts. There is a
need for a new generation of BCI classification methods that consider the user in the loop
and could adapt to the user [22].

6. Discussion
In the field of EEG-based BCI applications, there is a number of challenges hindering
the development of the applications that would need to be addressed during future studies.
It is important to acknowledge these aspects to find solutions or alternatives as needed.
There are also many opportunities in the field to be used in the future and this is also an
important aspect to highlight and share ideas among researchers. The sharing of new ideas
and possibilities for the future facilitates the development of a large variety of applications
in the field of EEG-based BCI applications.
During the preparation of this systematic review, a strict process has been followed
including the use of planned search terms and databases before the analysis of the literature.
The time period for the publications was selected from 2009 until 2019 for conducting of the
analysis in order to give an overview concerning a longer period of time. The review was
designed in order to be able to compare the results with other previously conducted reviews
on similar topics. Similar time interval has been selected also in previous studies [5,28].
The additional publications published after the aforementioned time period and outside of
the previous process have been included in the Appendix B of this review.
The selected search terms also set specific limitations for the publications identified
and analyzed in the review. Using the term “EEG-based” could set specific limitations
for the search results and term “EEG” could be used in order to obtain a wider variety of
search results. The aforementioned use of specific search words could be the reason why
some of the publications, for example [222,230–233], have not been identified for screening
during the first phase of the review. As the publications have been identified for analysis
until 2019, this aspect would need to be also considered as a limitation for the review.

6.1. Challenges
The challenges for the BCI applications could be divided into technology-related
and user-related, where technology-related challenges comprise technical aspects and the
usability of the system and the user-related contain the aspects of a person to learn to use
the BCI application and subjectiveness of the person to interpret the cues that generate
Sensors 2022, 22, 3331 21 of 44

or alter the acquired EEG signals [5]. Padfield et al. [2] has categorized the possible
challenges also as challenges faced in the research and development, challenges impeding
commercialization, flawed testing process, issues with BCI use and ethical issues. The
categorizations could be also combined and distinguish different technology-related and
user-related challenges under the five categories proposed by Padfield et al. [2].
In terms of challenges, the major aspect would be the current reliability of the BCI
system in everyday noisy environments [87]. This hindrance affects both medical and
non-medical applications, but higher effect is for the non-medical applications due to the
wider use outside of a controlled environment. The effect on non-medical applications
is also significant as the possibilities for the use of the EEG-based BCI applications in the
non-medical domain is relatively wider and number of users is higher in comparison to the
applications used in the medical domain.
In the medical domain (in the assistive field), the challenges include low recognition
rate for mental commands [158], problems with the signal acquisition equipment reliability
and training process [117]. The character of the brain signals and the amplitude varies
between persons, which makes it more difficult to develop BCI applications suitable for
all patients [6]. The work on separating specific EEG signals from other signals [234] and
the hindrances concerning the aspect of cross-subject classification [235] and accuracy of
interpreting the commands [236] is ongoing. The authors agree that one of the important
aspects is also the low throughput of information which may be a limiting factor in some
applications [117,158]. The challenge of the low accuracy of the system has also been
experienced in the field of monitoring emotional responses [161]. In the rehabilitation field,
the complexity of the system setup and high cost of the devices has been noted [163]. In
neurorehabilitation, it has been highlighted that the success of the therapeutic methods is
hard to measure and the repetitiveness that is needed from the subject during the therapy
could be demotivating [106]. Further research needs to be conducted to overcome the
repetitions needed by the patient during the use of the BCI and possibilities to overcome
time-consuming calibration for the BCI users [237].
Concerning BCI applications in the non-medical domain, Wei et al. have highlighted
that in order to transition laboratory-oriented BCI applications into real-world environ-
ments, the convenience to use and long-term wearing comfort for the devices needs to be
improved [193]. There have been alternative devices created for this reason, to increase for
example the user-friendliness of the devices meant to measure drowsiness at work [14].
During detection of human emotions, the hinderance could be the ambiguity of human
emotions and the complexity of EEG signals [203]. There has also been works on the
closed-loop interactions of human emotions with emotional stimuli for example in the
case of music interface which complicates the system setup further [77]. In the field of
entertainment, it has been emphasized that the seamless interaction between user and the
device is of utmost importance and a crucial concern [110].
The challenges hindering the development of BCI applications in both medical and
non-medical settings involve poor portability of the hardware, low user-friendliness of
the device, low signal quality and individual differences among persons that result in
difficulties interpreting the EEG signals across users [6,193,238,239]. The accuracy of EEG
pattern decoding is one of the key aspects for developing reliable BCI applications [240]. In
current development of BCI applications the variability of EEG signals received from one
subject and inter-subject variability has remained one of the most important obstacles [241]
and the designing of subject-independent BCI systems has remained a challenge [242].
The use of current subject-dependent applications is time consuming due to training
and especially inconvenient for people with mental disabilities [243]. Therefore, in addition
to the BCI application functioning for one specific person, it would be important to develop
subject-independent BCI systems that can be used by various users without previous train-
ing [241]. The challenge in developing subject-independent BCI applications is to overcome
person-dependent scenarios where the training and sets come from the same person and
would require high-performance person-independent classification [238]. Overcoming the
Sensors 2022, 22, 3331 22 of 44

challenge would enable efficient use of subject-independent system in turn enabling wider
implementation of the BCI applications.
Both medical and non-medical domains are affected by potential ethical issues in
the EEG-based BCI applications filed of research. It would need to be determined for
example who would be liable in case of any accidents during the use of BCI applications or
could the applications affect, for example, user’s mood and therefore affect user’s decision
making in a broader sense [2]. It has been shown that under certain circumstances it would
be possible to probe subliminally private information from the users using EEG-based
BCI devices [83]. In case the ethical and security aspects of the BCI applications are not
controlled, an especially wider spread of non-medical devices could be used to exploit
user emotions to push targeted marketing and political agendas [2]. These aspects would
need to be considered and the users would need to be notified about the possible risks
and responsibilities. Ethical standards would need to be established in order to guide the
general development of BCI technology and prevent ethical issues preemptively [2].

6.2. Future Possibilities

It has been suggested that in order to make the EEG-based BCI applications more
accurate and efficient hybrid BCI systems should be developed that combine BCI systems
with another BCI or other kinds of interfaces [2]. In addition to the use of only EEG for
obtaining the biological signals, other methods could be used to support the strength and
quality of the signals such as fNIRS or fMRI [6]. Such other physiological measures could
also be heart rate or eye movements [1].
Increasing attention has been paid to fNIRS due to the advantages such as non-
invasiveness, user safety, affordability and portability, but fNIRS signals are highly subject-
specific and have low test-retest reliability [244]. Barrios et al. [245] have also suggested
that combining fMRI with EEG could overcome significant limitations that are present
in the use of EEG alone. Padfield et al. [2] have suggested to combine as well different
approaches using EEG, such as motor-imagery system together with steady-state visually
evoked potential as a training aid. In addition, combining different methods such as fNIRS
with motor-imagery EEG or motor-imagery EEG with sensory interface has been proposed
by Padfield et al. [2].

6.2.1. Possibilities in Medical Domain

In the medical domain, the BCI could be used to control a variety of assistive robotic
devices. EEG-based BCI applications have been developed for controlling wheelchairs [111,200],
and research has been conducted on the development of robotic limbs [158]. Recent studies
have shown additional success for the development of EEG-based BCI applications for
controlling robotic devices [246,247]. Further research could be pursued in order to develop
humanoid robots and drones for the support of daily activities for the patients [6].
It has been further suggested to use virtual reality in neuro rehabilitation, where motor-
imagery BCI virtual reality systems could be used for real time applications for stroke
rehabilitation, to increase motivation of the patient during the rehabilitation process [106].
The monitoring of EEG could be used to enhance speech therapy sessions for people who
stutter, providing real-time visualization of the brain and insights concerning the brain
activity during the sessions [43]. Jeunet et al. [11] have suggested BCI neurofeedback for
the reduction of hypokinetic activity in case of a stroke and reducing hyperkinetic activity
in case of attention deficit hyperactivity disorder (ADHD).

6.2.2. Possibilities in Non-Medical Domain

In the non-medical domain, there is a high potential for the EEG-based BCI applications
monitoring cognitive load, attention, drowsiness and other aspects of the mind. Monitoring
cognitive load has been studied by Friedman et al. for intelligence tests, but measuring
cognitive load could also be used to better design and conduct e-learning, psychometric
exams, military training and other trainings [84].
Sensors 2022, 22, 3331 23 of 44

Several studies have been published on monitoring attention via EEG-based BCI
applications [3,150,171]. The level of attention is essential, both in the learning process
and during tasks with high responsibility. Sethi et al. [171] have developed EEG-based
attention feedback to improve focus in e-learning, but the principles could also be applied
for drivers to test their reflexes and attentiveness and for driving instructors to assess the
capability of the drivers. The monitoring of attention level could also be used to correlate
stress with attention level and creativity with attention level. According to Sethi et al. [171]
for example, optimal stress level can boost attention, and with the help of monitoring,
attention optimal stress level could be determined, i.e., when the attention level for the
individual is the highest.
EEG-based BCI applications could be further developed for wider authentication of
persons in addition to currently available methods [37]. It has been suggested that EEG
data could also be used for deceit identification. As polygraphy test is not fully reliable
and the results could be altered in case of specific practice and training by the subject. EEG
data could be an important alternative for deceit identification in the future [74]. Punsawad
et al. [161] have suggested that the monitoring of human emotions via EEG-based BCI
application could be applied in neuromarketing for product branding and advertising
slogan design.
Entertainment is an ever-developing field and includes, for example, gaming. The
availability and user friendliness of the games using EEG-based BCI has been increasing
over time [110]. With further development of the technology and integration with currently
available technologies there are numerous possibilities for the use of EEG-based BCI for
games for both entertainment and serious games for training purposes [8].
The main reasons for the higher pace of development in the non-medical domain
are due to the high potential of the domain for a wider range of users when compared
to the medical domain. While development in medical domain is focusing on specific
patient groups the EEG-based BCI applications in non-medical domain could be used by
the general public, for example for smart home control, entertainment or gaming. The
range of use for the devices in the non-medical domain is continuously broadening as new
possibilities emerge for the use of the devices in new situations such as deceit identification
test [74] or authentication based on emotionally significant images [166].
EEG-based BCI applications have a wide variety of applications in both medical and
non-medical domains, with even higher potential recently in the non-medical domain
due to a high number of potential users around the world. With the development of the
technology, reduction in cost and increase in comfort of use the EEG-based BCI applications
can gain increasing attention in the non-medical domain.

6.2.3. Safety and User-Friendliness

During the daily use of EEG-based BCI applications, the safety of use is one of the
most important aspects to focus on. In case of synchronous BCI, the controlling of the BCI is
divided into time windows when the commands by the user could be given and the transfer
of commands from the user to the device is well defined. For the user, asynchronous BCI
applications would be preferred, as in case of asynchronous applications the commands
could be given to the device at any time, independent of specific time windows. From the
safety perspective, it is important to define, in case of asynchronous devices, when the
user actually wanted to give command and when the user was thinking other thoughts
not related to the use of the device. For this reason, the concept of “brain switch” would
need to be further developed in practice, enabling the user to mentally disconnect from the
device when the user is not intending to use the device [6,131].
Further possibilities for the EEG-based BCI applications include new, more convenient
methods for obtaining EEG signals. Wei et al. [193] have suggested measuring EEG from
non-hair-bearing scalp areas for the further ease of use during long-term usage of the
EEG-based BCI devices in everyday situations. It has also been shown that in some cases it
would be sufficient to measure EEG only near the ear that could make the use of EEG-based
Sensors 2022, 22, 3331 24 of 44

BCI applications more convenient [113]. In order to make the use of BCI applications
more efficient and further integrate the possibilities into our daily tasks, BCIs could be
integrated with augmented reality which would create new dimensions of user experience
and practicality [15].

7. Analysis/Synthesis
In the current review, both articles and conference proceedings have been included.
The inclusion of both these types in the analysis gives better representation of the ideas in
the field and helps reduce possible publication bias.
Although initially the EEG-based BCI applications were mainly developed for medical
purposes to help patients and support in daily activities, the focus has been moving
from the medical domain to non-medical applications. The shift in focus does not reduce
the importance of these applications in the medical domain but rather shows the wider
potential of EEG-based BCI applications and opens new doors for applying the possibilities
mode broadly.
As per region, 111 publications (55%) have been contributed by Asia being firmly in
the lead concerning number of studies published. From Asia most of the publications per
country have been contributed by China. This is a relevant finding as China has increasing
influence in general in high technology sector and EEG-based BCI could be one important
field of research resulting in a high variety of high technology applications designed for
many different fields of life, both medical and non-medical. The high potential of EEG-
based BCI applications and diverse possibilities for use could support the future economic
strength of the countries and regions investing in the research and development in the field.
The reasons of the general popularity of Emotiv EPOC (Emotiv, San Francisco, CA,
USA) EEG device are its relatively low cost, sufficient number of EEG channels and it being
considered easy to use. Emotiv EPOC (Emotiv, San Francisco, CA, USA) and MindWave
(NeuroSky, San Jose, CA, USA) are considered to be low-cost consumer EEG devices
whereas Quik-Cap (Compumedics Neuroscan, Charlotte, NC, USA) is higher in cost.
Emotiv EPOC (Emotiv, San Francisco, CA, USA) EEG device has 14 channels and MindWave
(NeuroSky, San Jose, CA, USA) 1 channel, whereas the Quik-Cap (Compumedics Neuroscan,
Charlotte, NC, USA) EEG device uses 32 channels. MindWave (NeuroSky, San Jose, CA,
USA) EEG device has limitations in use due to the limited number of EEG channels, but
the low cost and ease of use make the device popular. The device could be applied in
specific applications that do not require higher number of EEG channels. The Quik-Cap
(Compumedics Neuroscan, Charlotte, NC, USA) device on the other hand is more expensive
but has the advantage of higher number of EEG channels. The final decision concerning
the use of a specific EEG device depends on the type of EEG-based BCI application,
determining the need for specific number of EEG channels. The decision also depends on
the cost planned for the study and the end users.
When planning the selection of the EEG device and number of channels, it is important
to consider the end users for the EEG-based BCI application, which determines the technical
requirements and possible overall cost of the device. A smaller number of carefully
selected electrode positions would also make the device more user friendly and support
the performance of the system.
There is a general trend noted in the overall use of various techniques. As in case of
motor-imagery, auditory evoked potential and vibrotactile evoked potential technique the
main application is in the medical domain, but the majority and larger variety of techniques
are applied in the non-medical domain. As seen from the results of the current review, a
highly diverse selection of techniques has been applied in the non-medical field supporting
further the development of diverse applications and supporting the high potential of
non-medical applications among the EEG-based BCI applications.
In the current review, a trend of high prevalence of using power spectral density and
Fourier transform for feature extraction has been noted. The overall results correlate with
previous studies in the field, highlighting the importance of these methods for feature
Sensors 2022, 22, 3331 25 of 44

extraction and emphasizing the need to apply different types of feature extraction methods
depending on the application under study. The results of the current review also show the
importance of linear discriminant analysis and support vector machine for classification
in correlation with previous studies in the field. The choice for classification algorithm
depends on the type of brain signal being recorded and the type of application that is
being controlled.
With the widespread use of the devices, the safety on an individual and community
level would need to be further analyzed. Nowadays, it has been unfortunately common
that due to security breaches malicious software have infiltrated computer networks even
with a high security level. This needs to be taken into consideration, especially when
considering the sensitivity of the biomedical information obtained from the BCI device and
also the possibility to alter the brain signals for example via neurofeedback. As the field
of EEG-based BCI is fast developing, the ethical aspects would need to be analyzed and
safeguarded in parallel with the development of technology.
The BCI applications are developing rapidly and therefore it is important to have an
up-to-date overview on the EEG-based BCI applications together with possible challenges
and the way forward. The significance of the current review is to provide an overview
of the current work conducted in the field of EEG-based BCI applications, along with the
challenges and future possibilities.

Author Contributions: Conceptualization, Y.M., N.M. and K.V.; methodology, K.V., Y.M. and N.M.;
formal analysis, K.V., N.M. and Y.M.; investigation, K.V.; writing—original draft preparation, K.V.;
writing—review and editing, K.V., N.M. and Y.M.; visualization, K.V.; supervision, Y.M. and N.M.;
project administration, Y.M. and N.M. All authors have read and agreed to the published version of
the manuscript.
Funding: The research has been funded by the European Social Fund via the Estonian Education and
Youth Board and IT Academy Programme.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design
of the review; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or
in the decision to publish the results.

Appendix A. Search Terms Used

The search was conducted in the databases PubMed, Scopus and Web of Science using
basic search settings with the below-described search terms relevant for current review.
The search was conducted in PubMed database on 20 October 2019 with the search term
(“electroencephalography based” OR “EEG based”) AND (“brain computer interface” OR
“BCI” OR “brain machine interface” OR “BMI”) AND (“application” OR “applications”).
The search resulted in 185 publications that were then included for further screening.
The search in the Web of Science database was conducted on 28 October 2019 with the
search term (“electroencephalography based” OR “EEG based”) AND (“brain computer
interface” OR “BCI” OR “brain machine interface” OR “BMI”) AND (“application” OR
“applications”) and the search in Web of Science database resulted in 451 publications.
The search in the Scopus database was conducted on 30 October 2019 using search
term (“electroencephalography based” OR “EEG based”) AND (“brain computer interface”
OR “BCI” OR “brain machine interface” OR “BMI”) AND (“application” OR “applications”)
and 569 publications were identified through the search for inclusion to further screening.
Sensors 2022, 22, 3331 26 of 44

Appendix B

Table A1. Listing of Additional Publications from the Period of 31 October 2019 Until 26 June 2021.

Author Year Published Title

A multi-day and multi-band dataset for a steady-state visual-evoked
Choi et al. 2019
potential-based brain–computer interface
Comparative analysis of mother wavelet selection for eeg signal application to
Ganorkar and Raut 2019
motor imagery-based brain–computer interface
Combination of discrete wavelet packet transform with detrended fluctuation
Hekmatmanesh et al. 2019 analysis using customized mother wavelet with the aim of an imagery-motor
control interface for an exoskeleton
Multiclass EEG motor-imagery classification with sub-band common
Khan et al. 2019
spatial patterns
Recovery of event-related potential signals using compressive sensing and
Khoshnevis and Ghorshi 2019
kronecker technique
Fully Passive Flexible Wireless Neural Recorder for the Acquisition of
Liu et al. 2019
Neuropotentials from a Rat Model
Multi optimized SVM classifiers for motor imagery left and right hand
Mebarkia and Reffad 2019
movement identification
A comparative study of motor imagery-based BCI classifiers on EEG and
Nagabushan et al. 2019
iEEG data
Assessment of CSP-based two-stage channel selection approach and local
Onay and Kose 2019 transformation-based feature extraction for classification of motor
imagery/movement EEG data
Oralhan 2019 2 Stages-region-based P300 Speller in Brain–Computer Interface
Saikia and Paul 2019 EEG signal processing and its classification for rehabilitation device control
Motor imagery tasks-based EEG signals classification using tunable-Q
Taran and Bajaj 2019
wavelet transform
A Parallel Multiscale Filter Bank Convolutional Neural Networks for Motor
Wu et al. 2019
Imagery EEG Classification
Yao and Shoaran 2019 Enhanced Classification of Individual Finger Movements with ECoG
Yu Chen and Mehmood 2019 A critical review on state-of-the-art EEG-based emotion datasets
A Graph-Based Hierarchical Attention Model for Movement Intention
Zhang et al. 2019
Detection from EEG Signals
Deep Learning Decoding of Mental State in Non-invasive Brain
Zhang et al. 2019
Computer Interface
Aggarwal and Chugh 2020 A decade of EEG Analysis: Prospects & Challenges in Biometric System
Alakus and Turkoglu 2020 Emotion recognition with deep learning using GAMEEMO data set
Wheelchair Free Hands Navigation Using Robust DWT-AR Features
Alhakeem et al. 2020
Extraction Method with Muscle Brain Signals
Classification of Motor Imagery Task by Using Novel Ensemble
Ali et al. 2020
Pruning Approach
Lightweight Building of an Electroencephalogram-Based Emotion
Al-Nafjan et al. 2020
Detection System
Andrade et al. 2020 An EEG Brain–Computer Interface to Classify Motor Imagery Signals
Instrumentation for motor imagery-based brain computer interfaces relying on
Angrisani et al. 2020
dry electrodes: A functional analysis
Sensors 2022, 22, 3331 27 of 44

Table A1. Cont.

Author Year Published Title

Wireless Monitoring Using a Stretchable and Transparent Sensor Sheet
Araki et al. 2020
Containing Metal Nanowires
Arico et al. 2020 Brain–Computer Interfaces: Toward a Daily Life Employment
Cognitive-Driven Binaural Beamforming Using EEG-Based Auditory
Aroudi and Doclo 2020
Attention Decoding
A multi stage EEG data classification using k-means and feed forward
Bablani et al. 2020
neural network
Emotion-Inducing Imagery Versus Motor Imagery for a
Bigirimana et al. 2020
Brain–Computer Interface
Interpretable and lightweight convolutional neural network for EEG decoding:
Borra et al. 2020
Application to movement execution and imagination
Convolutional Neural Network for a P300 Brain–Computer Interface to
Borra et al. 2020
Improve Social Attention in Autistic Spectrum Disorder
STIMULUS: Noninvasive Dynamic Patterns of Neurostimulation Using
Cao and Grover 2020
Spatio-Temporal Interference
Development of a Deep Learning-Based Brain–Computer Interface for Visual
Castro et al. 2020
Imagery Recognition
Prediction of individual user’s dynamic ranges of EEG features from
Cha et al. 2020 resting-state EEG data for evaluating their suitability for passive
brain–computer interface applications
Chamola et al. 2020 Brain–computer interface-based humanoid control: A review
Chen et al. 2020 EEG-based biometric identification with convolutional neural network
Emotion recognition from spatiotemporal EEG representations with hybrid
Chen et al. 2020
convolutional recurrent neural networks via wearable multi-channel headset
Cheng et al. 2020 Motion Imagery-BCI Based on EEG and Eye Movement Data Fusion
A Novel Approach to Classify Natural Grasp Actions by Estimating Muscle
Cho et al. 2020
Activity Patterns from EEG Signals
Decoding of Grasp Motions from EEG Signals Based on a Novel Data
Cho et al. 2020
Augmentation Strategy
Improving Speller BCI performance using a cluster-based
Cortez et al. 2020
under-sampling method
Under-sampling and Classification of P300 Single-Trials using Self-Organized
Cortez et al. 2020
Maps and Deep Neural Networks for a Speller BCI
Cozza et al. 2020 Dimension Reduction Techniques in a Brain–Computer Interface Application
Reading into the mind’s eye: Boosting automatic visual recognition with
Cudlenco et al. 2020
EEG signals
de Melo et al. 2020 EEG Analysis in Coincident Timing Task Towards Motor Rehabilitation
Delvigne et al. 2020 Attention Estimation in Virtual Reality with EEG-based Image Regression
Self-adaptive shared control with brain state evaluation network for
Deng et al. 2020
human-wheelchair cooperation
A Bayesian Shared Control Approach for Wheelchair Robot with Brain
Deng et al. 2020
Machine Interface
Increasing the Classification Accuracy of EEG-based Brain–computer
Dimitrov et al. 2020
Interface Signals
Sensors 2022, 22, 3331 28 of 44

Table A1. Cont.

Author Year Published Title

An extensive analysis on deep neural architecture for classification of
Dutta and Nandy 2020
subject-independent cognitive states
Development of a BCI-based gaming application to enhance cognitive control
Dutta et al. 2020
in psychiatric disorders
A long short-term memory autoencoder approach for EEG motor
Elessawy et al. 2020
imagery classification
Proposed model for thought-based animation based on classifying EEG signals
Elkafrawy et al. 2020
using estimated parameters and multi-SVM
Fathima and Kore 2020 Enhanced Differential Evolution-Based EEG Channel Selection
Decoding of voluntary and involuntary upper-limb motor imagery based on
Feng et al. 2020
graph fourier transform and cross-frequency coupling coefficients
Filipp et al. 2020 Application of brain–computer interfaces in assistive technologies
Beyond Technologies of Electroencephalography-Based Brain–Computer
Fontanillo et al. 2020
Interfaces: A Systematic Review From Commercial and Ethical Aspects
Five Shades of Grey: Exploring Quintary m-Sequences for More User-Friendly
Gembler et al. 2020
c-VEP-Based BCIs
Bi-directional Long Short-Term Memory model to analyze psychological
Ghosh et al. 2020
effects on gamers
Artificial intelligence within the interplay between natural and artificial
Gorriz et al. 2020
computation: Advances in data science, trends and applications
Context Sensitivity of EEG-Based Workload Classification Under Different
Grissmann et al. 2020
Affective Valence
The effects of varying levels of mental workload on motor imagery-based
Gu et al. 2020
brain computer interface
The performance impact of data augmentation in CSP-based motor-imagery
Gubert et al. 2020
systems for BCI applications
Gurve et al. 2020 Trends in Compressive Sensing for EEG Signal Processing Applications
A comparison of ECG and EEG metrics for in-flight monitoring of helicopter
Haira et al. 2020
pilot workload
Neurophysiological Closed-Loop Control for Competitive Multi-brain Robot
Hernandez-Cuevas et al. 2020
Interaction
Hussain and Park 2020 HealthSOS: Real-Time Health Monitoring System for Stroke Prognostics
Efficient Classification of Motor Imagery using Particle Swarm
Idowu et al. 2020
Optimization-based Neural Network for IoT Applications
A novel multi-modal machine learning-based approach for automatic
Ieracitano et al. 2020
classification of EEG recordings in dementia
Low-Dimensional Subject Representation-based Transfer Learning in
Jeng et al. 2020
EEG Decoding
EEG classification using sparse Bayesian extreme learning machine for
Jin et al. 2020
brain–computer interface
Decoding Visual Recognition of Objects from EEG Signals based on
Kalafatovich et al. 2020
Attention-Driven Convolutional Neural Network
EEG-Based Prediction of Successful Memory Formation During
Kang et al. 2020
Vocabulary Learning
Sensors 2022, 22, 3331 29 of 44

Table A1. Cont.

Author Year Published Title

An Ear-EEG-based Brain–Computer Interface using Concentration Level
Kaongoen and Jo 2020
for Control
Kaur et al. 2020 A study of EEG for enterprise multimedia security
Khan et al. 2020 High performance multi-class motor imagery EEG classification
Indexing and Image Search by the Content According to the Biological Base of
Kouddad et al. 2020
the Cognitive Processing of Information using a Neural Sensor
A Hybrid Approach for Extracting EMG signals by Filtering EEG Data for IoT
Kurapa et al. 2020
Applications for Immobile Persons
Mental state space visualization for interactive modeling of personalized BCI
Kuzovkin et al. 2020
control strategies
Decoding of Intuitive Visual Motion Imagery Using Convolutional Neural
Kwon et al. 2020
Network under 3D-BCI Training Environment
Mind Your Mind: EEG-Based Brain–Computer Interfaces and Their Security in
Landau et al. 2020
Cyber Space
Complex Motor Imagery-based Brain–Computer Interface System: A
Lee et al. 2020
Comparison between Different Classifiers
Classification of Upper Limb Movements Using Convolutional Neural
Lee et al. 2020
Network with 3D Inception Block
Deep learning for EEG-based Motor Imagery classification: Accuracy-cost
Leon et al. 2020
trade-off
Enhancing BCI-Based Emotion Recognition Using an Improved Particle
Li et al. 2020
Swarm Optimization for Feature Selection
EEG-Based EMG Estimation of Shoulder Joint for the Power Augmentation
Liang et al. 2020
System of Upper Limbs
A Multi-Scale Activity Transition Network for Data Translation in EEG
Lin et al. 2020
Signals Decoding
EEG Signal Reconstruction Using a Generative Adversarial Network With
Luo et al. 2020
Wasserstein Distance and Temporal-Spatial-Frequency Loss
Luo et al. 2020 Estimation of Motor Imagination Based on Consumer-Grade EEG Device
Online learning using projections onto shrinkage closed balls for adaptive
Ma et al. 2020
brain computer interface
The Promotoer, a brain–computer interface-assisted intervention to promote
upper limb functional motor recovery after stroke: a study protocol for a
Mattia et al. 2020
randomized controlled trial to test early and long-term efficacy and to identify
determinants of response
Spatial-Frequency Feature Learning and Classification of Motor Imagery EEG
Miao et al. 2020
Based on Deep Convolution Neural Network
Driver Fatigue Detection Based on Multi-scale Wavelet Log Energy Entropy of
Min and Cai 2020
Frontal EEG
Effect of hand grip actions on object recognition process: a machine
Mishra et al. 2020
learning-based approach for improved motor rehabilitation
Continuous low-frequency EEG decoding of arm movement for closed-loop,
Mondini et al. 2020
natural control of a robotic arm
An empirical comparison of neural networks and machine learning algorithms
Nakagome et al. 2020
for EEG gait decoding
Sensors 2022, 22, 3331 30 of 44

Table A1. Cont.

Author Year Published Title

Netzer et al. 2020 Real-time EEG classification via coresets for BCI applications
Reducing Sensors in Mental Imagery-Based Cognitive Task for Brain
Nisar et al. 2020
Computer Interface
Pa Aung and New 2020 Regions of Interest (ROI) Analysis for Upper Limbs EEG Neuroimaging Schemes
A novel deep learning classifier and genetic algorithm-based feature selection
Padmavathy et al. 2020
for hybrid EEG-FNIRS brain–computer interface
Towards a Portable Magnetoencephalography-Based Brain Computer Interface
Paek et al. 2020
with Optically-Pumped Magnetometers
Prognosis for patients with cognitive motor dissociation identified by
Pan et al. 2020
brain–computer interface
EEG-Based Emotion Recognition Using Logistic Regression with Gaussian
Pan et al. 2020
Kernel and Laplacian Prior and Investigation of Critical Frequency Bands
Quadcopter Control in Three-Dimensional Space Using SSVEP and Motor
Parikh and George 2020
Imagery-Based Brain–Computer Interface
Petukhov et al. 2020 Being present in a real or virtual world: A EEG study
Visual P300 Mind-Speller Brain–Computer Interfaces: A Walk Through the
Philip and George 2020
Recent Developments With Special Focus on Classification Algorithms
Qin et al. 2020 Smart Home Control for Disabled Using Brain Computer Interface
A Hybrid Brain–Computer Interface for Closed-Loop Position Control of a
Rakshit et al. 2020
Robot Arm
Current Status, Challenges, and Possible Solutions of EEG-Based
Rashid et al. 2020
Brain–Computer Interface: A Comprehensive Review
Five-Class SSVEP Response Detection using Common-Spatial Pattern
Rashid et al. 2020
(CSP)-SVM Approach
Riyad et al. 2020 Incep-eegnet: A convnet for motor imagery decoding
Roy et al. 2020 A hybrid classifier combination for home automation using EEG signals
Identification of motor and mental imagery EEG in two and multiclass
Sadiq et al. 2020
subject-dependent tasks using successive decomposition index
EEG signal analysis and classification on P300 speller-based BCI performance
Sahu et al. 2020
in ALS patients
The Effect That Auditory Distractions Have on a Visual P300 Speller While
Schembri et al. 2020
Utilizing Low Cost Off-the-Shelf Equipment
Real-time EEG Feedback on Alpha Power Lateralization Leads to Behavioral
Schneider et al. 2020
Improvements in a Covert Attention Task
EEG-Controlled Wall-Crawling Cleaning Robot Using SSVEP-Based
Shao et al. 2020
Brain–Computer Interface
Multi-class motor imagery EEG classification using collaborative
She et al. 2020
representation-based semi-supervised extreme learning machine
Feature Extraction of Brain–Computer Interface Electroencephalogram Based
Shi et al. 2020
on Motor Imagery
On assessing driver awareness of situational criticalities: Multi-modal
Siddharth and Trivedi 2020
bio-sensing and vision-based analysis, evaluations, and insights
Realising transfer learning through convolutional neural network and support
Singh and Singh 2020
vector machine for mental task classification
A Practical EEG-Based Human-Machine Interface to Online Control an
Song et al. 2020
Upper-Limb Assist Robot
Sensors 2022, 22, 3331 31 of 44

Table A1. Cont.

Author Year Published Title

Suma et al. 2020 Spatial-temporal aspects of continuous EEG-based neurorobotic control
Multimodal affective state assessment using fNIRS+ EEG and spontaneous
Sun et al. 2020
facial expression
Adaptive feature extraction in EEG-based motor imagery BCI: tracking
Talukdar et al. 2020
mental fatigue
Spiking Neural Networks: Background, Recent Development and the
Tan et al. 2020
NeuCube Architecture
Classification-Oriented Fuzzy-Rough Feature Selection for the EEG-Based
Tao 2020
Brain Computer Interfaces
Machine Learning approach for the classification of EEG signals of multiple
Tiwari et al. 2020
imagery tasks
Prediction of Motor Imagery Performance based on Pre-Trial Spatio-Spectral
Torkamani-Azar et al. 2020
Alertness Features
Torres et al. 2020 EEG-Based BCI Emotion Recognition: A Survey
Assessing the Relevance of Neurophysiological Patterns to Predict Motor
Tzdaka et al. 2020
Imagery-based BCI Users’ Performance
Can the occipital alpha-phase speed up visual detection through a real-time
Vigue-Guix et al. 2020
EEG-based brain–computer interface (BCI)?
A Novel EEG Classification Technique Based on Particle Swarm Optimization
Wafeek et al. 2020
for Hand and Finger Movements
Enhancing gesture decoding performance using signals from posterior parietal
Wang et al. 2020
cortex: a stereo-electroencephalograhy (SEEG) study
Wang et al. 2020 P300 Recognition Based on Ensemble of SVMs
William et al. 2020 ERP Template Matching for EEG Single Trial Classification
Wolpaw et al. 2020 Brain–computer interfaces: Definitions and principles
Implementing Over 100 Command Codes for a High-Speed Hybrid
Xu et al. 2020
Brain–Computer Interface Using Concurrent P300 and SSVEP Features
Motor Imagery-Based Continuous Teleoperation Robot Control with
Xu et al. 2020
Tactile Feedback
Two-level multi-domain feature extraction on sparse representation for motor
Xu et al. 2020
imagery classification
An improve d common spatial pattern combine d with channel-selection
Yan et al. 2020
strategy for electroencephalography-based emotion recognition
MI3DNet: A Compact CNN for Motor Imagery EEG Classification with
Yang et al. 2020
Visualizable Dense Layer Parameters
Yao et al. 2020 Information-preserving feature filter for short-term EEG signals
Efficient machine learning algorithm for electroencephalogram modeling in
Yi 2020
brain–computer interfaces
InstanceEasyTL: An Improved Transfer-Learning Method for EEG-Based
Zeng et al. 2020
Cross-Subject Fatigue Detection
Zhang et al. 2020 Pain Control by Co-adaptive Learning in a Brain–Machine Interface
Application of transfer learning in eeg decoding based on brain–computer
Zhang et al. 2020
interfaces: A review
A Hybrid Asynchronous Brain–Computer Interface Combining SSVEP and
Zhou et al. 2020
EOG Signals
Sensors 2022, 22, 3331 32 of 44

Table A1. Cont.

Author Year Published Title

State-of-the-art non-invasive brain–computer interface for neural
Zhuang et al. 2020
rehabilitation: A review
Consumers’ Preference Recognition Based on Brain–Computer Interfaces:
Aldayel et al. 2021
Advances, Trends, and Applications
An effective classification framework for brain–computer interface system
Alhudhaif 2021
design based on combining of fNIRS and EEG signals
Al-Saegh et al. 2021 Deep learning for motor imagery EEG-based classification: A review
Hybrid deep learning (Hdl)-based brain–computer interface (bci) systems: A
Alzahab et al. 2021
systematic review
A linearly extendible multi-artifact removal approach for improved upper
Asogbon et al. 2021
extremity EEG-based motor imagery decoding.
Aydarkhanov et al. 2021 Closed-loop EEG study on visual recognition during driving
EEG-Based Auditory Attention Detection and Its Possible Future Applications
Belo et al. 2021
for Passive BCI
Long-Term BCI Training of a Tetraplegic User: Adaptive Riemannian
Benaroch et al. 2021
Classifiers and User Training
Bhattacharyya et al. 2021 Neuro-feedback system for real-time BCI decision prediction
Phase/Amplitude Synchronization of Brain Signals During Motor Imagery
Cattai et al. 2021
BCI Tasks.
Neuropsychological and neurophysiological aspects of brain–computer
Chaudhary et al. 2021
interface (BCI) control in paralysis
EEG-Based Anxious States Classification Using Affective BCI-Based Closed
Chen et al. 2021
Neurofeedback System
Chen et al. 2021 Implementing a calibration-free SSVEP-based BCI system with 160 targets.
Recognizing single-trial motor imagery EEG based on interpretable
Fu et al. 2021
clustering method
Georgiev et al. 2021 Virtual reality for neurorehabilitation and cognitive enhancement
EEG-based Brain–Computer Interfaces (BCIs): A Survey of Recent Studies on
Gu et al. 2021 Signal Sensing Technologies and Computational Intelligence Approaches and
Their Applications
Guner and Erkmen 2021 A Low-Cost Real-Time BCI Integration for Automated Door Opening System
Gupta et al. 2021 Brain computer interface controlled automatic electric drive for neuro-aid system
Hong et al. 2021 Dynamic Joint Domain Adaptation Network for Motor Imagery Classification
Auditory Evoked Potential (AEP)-Based Brain–Computer Interface (BCI)
Islam et al. 2021
Technology: A Short Review
Probability mapping-based artifact detection and removal from single-channel
Islam et al. 2021
EEG signals for brain–computer interface applications.
Low-Dimensional Subject Representation-Based Transfer Learning in EEG
Jeng et al. 2021
Decoding
Hybrid classification model for eye state detection using
Ketu et al. 2021
electroencephalogram signals
Analysis of Human Gait Using Hybrid EEG-fNIRS-Based BCI System:
Khan et al. 2021
A Review
Influence of Signal Preprocessing When Highlighting Steady-State Visual
Kharchenko et al. 2021
Evoked Potentials Based on a Multivariate Synchronization Index
Sensors 2022, 22, 3331 33 of 44

Table A1. Cont.

Author Year Published Title

Implementation of robot–human control bio-interface when highlighting
Kharchenko et al. 2021
visual-evoked potentials based on multivariate synchronization index
The classification of EEG-based winking signals: a transfer learning and
Kumar et al. 2021
random forest pipeline
P300 event-related potential detection using one dimensional convolutional
Liu et al. 2021
capsule networks
Multiscale space-time-frequency feature-guided multitask learning CNN for
Liu et al. 2021
motor imagery EEG classification
Liu et al. 2021 A Utility Human Machine Interface Using Low Cost EEG Cap and Eye Tracker
Effect of power feature covariance shift on BCI spatial-filtering techniques: A
Miladinovic et al. 2021
comparative study
Effect of hand grip actions on object recognition process: a machine
Mishra et al. 2021
learning-based approach for improved motor rehabilitation
Spatiotemporal-Filtering-Based Channel Selection for Single-Trial
Qi et al. 2021
EEG Classification
Wielding and evaluating the removal composition of common artefacts in EEG
Qi et al. 2021
signals for driving behaviour analysis.
Sequence-to-sequence deep neural network with spatio-spectro and temporal
Rammy et al. 2021
features for motor imagery classification
Ravirahul et al. 2021 Mind Wave Controlled Assistive Robot
LSTM-based brain–machine interface tool for text generation through eyes
Reyes et al. 2021
blinking detection
A novel multi-scale convolutional neural network for motor imagery
Riyad et al. 2021
classification
Decoding of semantic categories of imagined concepts of animals and tools
Rybar et al. 2021
in fNIRS
Elucidation of EEG Characteristics of Fuzzy Reasoning-Based Heuristic BCI
Saga et al. 2021
and Its Application to Patient With Brain Infarction
Santos et al. 2021 Comparison of LORETA and CSP for Brain–Computer Interface Applications
Shaban et al. 2021 Classification of Lactate Level Using Resting-State EEG Measurements
VME-DWT: An Efficient Algorithm for Detection and Elimination of Eye Blink
Shahbakhti et al. 2021
from Short Segments of Single EEG Channel
A binary harmony search algorithm as channel selection method for motor
Shi et al. 2021
imagery-based BCI
Frequency Domain CSP for Foot Motor Imagery Classification Using SVM for
Somadder et al. 2021
BCI Application
Connectivity modeling meets machine learning: The next generation of
Stival et al. 2021
eeg-based brain computer interfaces
Sulaiman et al. 2021 Offline eeg-based dc motor control for wheelchair application
WLnet: Towards an Approach for Robust Workload Estimation Based on
Sun et al. 2021
Shallow Neural Networks
EEG-based auditory attention decoding using speech-level-based segmented
Wang et al. 2021
computational models
Review of brain encoding and decoding mechanisms for EEG-based
Xu et al. 2021
brain–computer interface
Sensors 2022, 22, 3331 34 of 44

Table A1. Cont.

Author Year Published Title

Electroencephalogram-based neurofeedback training in persons with stroke: A
Yoo 2021
scoping review in occupational therapy
Cross-correlation-based discriminant criterion for channel selection in motor
Yu et al. 2021
imagery BCI systems
An EEG-Based Transfer Learning Method for Cross-Subject Fatigue Mental
Zeng et al. 2021
State Prediction
Zhang et al. 2021 Improving EEG Decoding via Clustering-Based Multitask Feature Learning
EEG-inception: an accurate and robust end-to-end neural network for
Zhang et al. 2021
EEG-based motor imagery classification
Tiny noise, big mistakes: adversarial perturbations induce errors in
Zhang et al. 2021
brain–computer interface spellers
Using convolution neural networks pattern for classification of motor imagery
Zolfaghari et al. 2021
in bci system

References
1. Abiri, R.; Borhani, S.; Sellers, E.W.; Jiang, Y.; Zhao, X. A comprehensive review of EEG-based brain–computer interface paradigms.
J. Neural. Eng. 2018, 16, 011001. [CrossRef] [PubMed]
2. Padfield, N.; Zabalza, J.; Zhao, H.; Masero, V.; Ren, J. EEG-Based Brain-Computer Interfaces Using Motor-Imagery: Techniques
and Challenges. Sensors 2019, 19, 1423. [CrossRef] [PubMed]
3. Ramakuri, S.K.; Peddi, A.; Rao, K.S.N.; Gupta, B.; Ghosh, S. Performance and Analysis of Human Attention Using Single-Channel
Wireless EEG Sensor for Medical Application’. In First International Conference on Artificial Intelligence and Cognitive Computing;
Springer: Singapore, 2019; pp. 355–362. [CrossRef]
4. Anwar, D.; Gupta, A.; Naik, V.; Sharma, S.K. Detecting meditation using a dry mono-electrode EEG sensor. In Proceedings of the
2017 9th International Conference on Communication Systems and Networks (COMSNETS), Bengaluru, India, 4–8 January 2017;
pp. 508–513. [CrossRef]
5. Al-Nafjan, A.; Hosny, M.; Al-Ohali, Y.; Al-Wabil, A. Review and Classification of Emotion Recognition Based on EEG Brain-
Computer Interface System Research: A Systematic Review. Appl. Sci. 2017, 7, 1239. [CrossRef]
6. Mao, X.; Li, M.; Li, W.; Niu, L.; Xian, B.; Zeng, M.; Chen, G. Progress in EEG-Based Brain Robot Interaction Systems. Comput.
Intell. Neurosci. 2017, 2017, 1742862. [CrossRef]
7. Yang, Y.; Wu, Q.; Fu, Y.; Chen, X. Continuous convolutional neural network with 3D input for EEG-based emotion recognition. In
Proceedings of the Neural Information Processing, Siem Reap, Cambodia, 13–16 December 2018; pp. 433–443.
8. Perales, F.J.; Amengual, E. A Serious Game Application using EEG-based Brain Computer Interface. In Proceedings of the
NEUROTECHNIX 2013—International Congress on Neurotechnology, Electronics and Informatics, Algarve, Portugal, 19–21
September 2013; pp. 249–255. [CrossRef]
9. Lin, C.-T.; Chang, C.-J.; Lin, B.-S.; Hung, S.-H.; Chao, C.-F.; Wang, I.-J. A Real-Time Wireless Brain–Computer Interface System for
Drowsiness Detection. IEEE Trans. Biomed. Circuits Syst. 2010, 4, 214–222. [CrossRef]
10. Tariq, M.; Trivailo, P.M.; Simic, M. EEG-Based BCI Control Schemes for Lower-Limb Assistive-Robots. Front. Hum. Neurosci. 2018,
12, 312. [CrossRef]
11. Jeunet, C.; Glize, B.; McGonigal, A.; Batail, J.-M.; Micoulaud-Franchi, J.-A. Using EEG-based brain computer interface and
neurofeedback targeting sensorimotor rhythms to improve motor skills: Theoretical background, applications and prospects.
Neurophysiol. Clin. Neurophysiol. 2018, 49, 125–136. [CrossRef] [PubMed]
12. Molina-Rueda, F.; Navarro-Fernández, C.; Cuesta-Gómez, A.; Alguacil-Diego, I.M.; Molero-Sánchez, A.; Carratalá-Tejada, M.
Neuroplasticity Modifications Following a Lower-Limb Amputation: A Systematic Review. PMR 2019, 11, 1326–1334. [CrossRef]
13. Ramakuri, S.K.; Ghosh, S.; Gupta, B. Behaviour state analysis through brain computer interface using wearable EEG devices: A
review. Electron. Gov. Int. J. 2017, 13, 377. [CrossRef]
14. Dhole, S.R.; Kashyap, A.; Dangwal, A.N.; Mohan, R. A novel helmet design and implementation for drowsiness and fall detection
of workers on-site using EEG and Random-Forest Classifier. Procedia Comput. Sci. 2019, 151, 947–952. [CrossRef]
15. Kim, M.; Kim, M.-K.; Hwang, M.; Kim, H.-Y.; Cho, J.; Kim, S.-P. Online Home Appliance Control Using EEG-Based Brain–
Computer Interfaces. Electronics 2019, 8, 1101. [CrossRef]
16. Gao, Q.; Zhao, X.; Yu, X.; Song, Y.; Wang, Z. Controlling of smart home system based on brain-computer interface. Technol. Health
Care 2018, 26, 769–783. [CrossRef]
Sensors 2022, 22, 3331 35 of 44

17. Holzner, C.; Guger, C.; Grönegress, C.; Edlinger, G.; Slater, M. Using a P300 Brain Computer Interface for Smart Home Control. In
World Congress on Medical Physics and Biomedical Engineering, Munich, Germany, 7–12 September 2009; Springer: Berlin/Heidelberg,
Germany, 2009; pp. 174–177. [CrossRef]
18. Holzner, C.; Guger, C.; Edlinger, G.; Gronegress, C.; Slater, M. Virtual Smart Home Controlled by Thoughts. In Proceedings of the
2009 18th IEEE International Workshops on Enabling Technologies: Infrastructures for Collaborative Enterprises, Groningen, The
Netherlands, 29 June–1 July 2009; pp. 236–239. [CrossRef]
19. Pan, X.; Zhang, Z.; Qu, J.; Zhao, C.; Li, Y. Enjoy driving from thought in a virtual city. In Proceedings of the 2017 36th Chinese
Control Conference (CCC), Dalian, China, 26–28 July 2017; pp. 11034–11040. [CrossRef]
20. Yu, Y.; Zhou, Z.; Yin, E.; Jiang, J.; Tang, J.; Liu, Y.; Hu, D. Toward brain-actuated car applications: Self-paced control with a motor
imagery-based brain-computer interface. Comput. Biol. Med. 2016, 77, 148–155. [CrossRef]
21. Arvutiteaduse instituut-Lõputööderegister’. Available online: https://comserv.cs.ut.ee/ati_thesis/datasheet.php?language=en
(accessed on 9 January 2021).
22. Lotte, F.; Bougrain, L.; Cichocki, A.; Clerc, M.; Congedo, M.; Rakotomamonjy, A.; Yger, F. A review of classification algorithms for
EEG-based brain–computer interfaces: A 10 year update. J. Neural Eng. 2018, 15, 031005. [CrossRef]
23. Pasqualotto, E.; Federici, S.; Belardinelli, M.O. Toward functioning and usable brain–computer interfaces (BCIs): A literature
review. Disabil. Rehabil. Assist. Technol. 2011, 7, 89–103. [CrossRef] [PubMed]
24. Kosmyna, N.; Lécuyer, A. A conceptual space for EEG-based brain-computer interfaces. PLoS ONE 2019, 14, e0210145. [CrossRef]
25. Alsharif, A.; Salleh, N.; Baharun, R.; Safaei, M. Neuromarketing Approach: An Overview and Future Research Directions. J. Theor.
Appl. Inf. Technol. 2020, 98, 7.
26. Machado, S.; Araújo, F.; Paes, F.; Velasques, B.; Cunha, M.; Budde, H.; Basile, L.F.; Anghinah, R.; Arias-Carrión, O.; Cagy, M.; et al.
EEG-based Brain-Computer Interfaces: An Overview of Basic Concepts and Clinical Applications in Neurorehabilitation. Rev.
Neurosci. 2010, 21, 451–468. [CrossRef]
27. Nicolas-Alonso, L.F.; Gomez-Gil, J. Brain Computer Interfaces, a Review. Sensors 2012, 12, 1211–1279. [CrossRef]
28. Roy, Y.; Banville, H.; Albuquerque, I.; Gramfort, A.; Falk, T.H.; Faubert, J. Deep learning-based electroencephalography analysis:
A systematic review. J. Neural Eng. 2019, 16, 051001. [CrossRef]
29. Craik, A.; He, Y.; Contreras-Vidal, J.L. Deep learning for electroencephalogram (EEG) classification tasks: A review. J. Neural Eng.
2019, 16, 031001. [CrossRef] [PubMed]
30. Moher, D.; Liberati, A.; Tetzlaff, J.; Altman, D.G.; PRISMA Group. Preferred reporting items for systematic reviews and
meta-analyses: The PRISMA statement. PLoS Med. 2009, 6, e1000097. [CrossRef] [PubMed]
31. PRISMA’. Available online: http://www.prisma-statement.org/PRISMAStatement/HistoryAndDevelopment (accessed on
9 November 2020).
32. Madhok, R.; Handoll, H.; Gillespie, W.J.; Gillespie, L.D. The Cochrane Collaboration: A leading role in producing reliable
evidence to inform healthcare decisions in musculoskeletal trauma and disorders. Indian J. Orthop. 2008, 42, 247–251. [CrossRef]
33. PubMed’, PubMed. Available online: https://pubmed.ncbi.nlm.nih.gov/ (accessed on 5 January 2021).
34. SCOPUS. Available online: https://www.scopus.com/home.uri (accessed on 30 October 2019).
35. Web of Science. Available online: https://login.webofknowledge.com/error/Error?Error=IPError&PathInfo=%2F&RouterURL=
https%3A%2F%2Ffanyv88.com%3A443%2Fhttps%2Fwww.webofknowledge.com%2F&Domain=.webofknowledge.com&Src=IP&Alias=WOK5 (accessed on
5 January 2021).
36. Kumar, S.V.K.; Julian, A. Oscitancy tracking using Brain-Computer Interface. In Proceedings of the 2012 IEEE Students’ Conference
on Electrical, Electronics and Computer Science, Bhopal, India, 1–2 March 2012; pp. 1–4. [CrossRef]
37. Mustafa, I.; Farooq, H.; Khatri, T.K. EEG based user authentication using visual stimuli of geometric shapes. In Proceedings of the
2019 2nd International Conference on Communication, Computing and Digital systems (C-CODE). IEEE, Islamabad, Pakistan,
6–7 March 2019; pp. 247–251. [CrossRef]
38. Abu-Alqumsan, M.; Ebert, F.; Peer, A. Goal-recognition-based adaptive brain-computer interface for navigating immersive robotic
systems. J. Neural Eng. 2017, 14, 036024. [CrossRef] [PubMed]
39. Adha, M.S.; Safri, N.; Othman, M.A. Real-time target selection based on electroencephalogram (EEG) signal. ARPN J. Eng. Appl.
Sci. 2015, 10, 8757–8761.
40. Ahn, J.; Ku, Y.; Kim, D.; Sohn, J.; Kim, J.; Kim, H. Wearable in-the-ear EEG system for SSVEP-based brain–computer interface.
Electron. Lett. 2018, 54, 413–414. [CrossRef]
41. Sun, A.; Fan, B.; Jia, C. Motor imagery EEG-based online control system for upper artificial limb. In Proceedings of the 2011
International Conference on Transportation, Mechanical, and Electrical Engineering (TMEE), Changchun, China, 16–18 December
2011; pp. 1646–1649. [CrossRef]
42. Al-Ghamdi, N.; Al-Hudhud, G.; Alzamel, M.; Al-Wabil, A. Trials and tribulations of BCI control applications. In Proceedings of
the 2013 Science and Information Conference, London, UK, 7–9 October 2013; pp. 212–217.
43. Al-Nafjan, A.; Al-Wabil, A.; AlMudhi, A.; Hosny, M. Measuring and monitoring emotional changes in children who stutter.
Comput. Biol. Med. 2018, 102, 138–150. [CrossRef]
44. Alnafjan, A.; Hosny, M.; Al-Wabil, A.; Al-Ohali, Y. Classification of Human Emotions from Electroencephalogram (EEG) Signal
using Deep Neural Network. Int. J. Adv. Comput. Sci. Appl. 2017, 8, 419–425. [CrossRef]
Sensors 2022, 22, 3331 36 of 44

45. Aloise, F.; Schettini, F.; Aricò, P.; Bianchi, L.; Riccio, A.; Mecella, M.; Babiloni, F.; Mattia, D.; Cincotti, F. Advanced brain computer
interface for communication and control. In Proceedings of the International Conference on Advanced Visual Interfaces, Roma,
Italy, 26–28 May 2010; pp. 399–400. [CrossRef]
46. AlQattan, D.; Sepulveda, F. Towards sign language recognition using EEG-based motor imagery brain computer interface.
In Proceedings of the 2017 5th International Winter Conference on Brain-Computer Interface (BCI), Gangwon, Korea, 9–11
January 2017; pp. 5–8. [CrossRef]
47. Amarasinghe, K.; Sivils, P.; Manic, M. EEG feature selection for thought driven robots using evolutionary Algorithms. In Proceed-
ings of the 2016 9th International Conference on Human System Interactions (HSI), Portsmouth, UK, 6–8 July 2016; pp. 355–361.
[CrossRef]
48. An, X.; Wei, J.; Liu, S.; Ming, N. A sLORETA study for gaze-independent BCI speller. In Proceedings of the 2017 39th Annual
International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), Jeju Island, Korea, 11–15 July 2017;
pp. 994–997. [CrossRef]
49. Anil, D.G.; Pelayo, P.; Mistry, K.S.; George, K. A tactile P300 based brain computer interface system for communication in iOS
devices. In Proceedings of the 2018 IEEE International Instrumentation and Measurement Technology Conference (I2MTC),
Houston, TX, USA, 14–17 May 2018; pp. 1–6. [CrossRef]
50. Arfaras, G.; Athanasiou, A.; Pandria, N.; Kavazidi, K.; Kartsidis, P.; Astaras, A.; Bamidis, P.D.; Niki, P.; Kyriaki, R.K.; Panagiotis,
D.B. Visual Versus Kinesthetic Motor Imagery for BCI Control of Robotic Arms (Mercury 2.0). In Proceedings of the 2017 IEEE
30th International Symposium on Computer-Based Medical Systems (CBMS), Thessaloniki, Greece, 22–24 June 2017; pp. 440–445.
[CrossRef]
51. Aricò, P.; Borghini, G.; Di Flumeri, G.; Colosimo, A.; Bonelli, S.; Golfetti, A.; Pozzi, S.; Imbert, J.-P.; Granger, G.; Benhacene, R.;
et al. Adaptive Automation Triggered by EEG-Based Mental Workload Index: A Passive Brain-Computer Interface Application in
Realistic Air Traffic Control Environment. Front. Hum. Neurosci. 2016, 10, 539. [CrossRef]
52. Aricò, P.; Borghini, G.; Di Flumeri, G.; Sciaraffa, N.; Colosimo, A.; Babiloni, F. Passive BCI in Operational Environments: Insights,
Recent Advances, and Future Trends. IEEE Trans. Biomed. Eng. 2017, 64, 1431–1436. [CrossRef]
53. Bai, O.; Lin, P.; Huang, D.; Fei, D.-Y.; Floeter, M.K. Towards a user-friendly brain–computer interface: Initial tests in ALS and PLS
patients. Clin. Neurophysiol. 2010, 121, 1293–1303. [CrossRef] [PubMed]
54. Bhattacharyya, S.; Konar, A.; Tibarewala, D.N. Motor imagery, P300 and error-related EEG-based robot arm movement control for
rehabilitation purpose. Med. Biol. Eng. Comput. 2014, 52, 1007–1017. [CrossRef] [PubMed]
55. Bhattacharyya, S.; Sengupta, A.; Chakraborti, T.; Banerjee, D.; Khasnobish, A.; Konar, A.; Tibarewala, D.N.; Janarthanan, R.;
Bhattacharyya, S.; Sengupta, A.; et al. EEG controlled remote robotic system from motor imagery classification. In Proceedings of
the 2012 Third International Conference on Computing, Communication and Networking Technologies (ICCCNT’12), Karur,
India, 26–28 July 2012; pp. 1–8. [CrossRef]
56. Boubakeur, M.R.; Wang, G.; Zhang, C.; Liu, K. EEG-Based Person Recognition Analysis and Criticism. In Proceedings of the 2017
IEEE International Conference on Big Knowledge (ICBK), Hefei, China, 9–10 August 2017; pp. 155–160. [CrossRef]
57. Brumberg, J.S.; Lorenz, S.D.; Galbraith, B.V.; Guenther, F.H. The unlock project: A Python-based framework for practical brain-
computer interface communication “app” development. In Proceedings of the 2012 Annual International Conference of the IEEE
Engineering in Medicine and Biology Society, San Diego, CA, USA, 28 August–1 September 2012; pp. 2505–2508. [CrossRef]
58. Brummer, N.B.; Rawson, J.A.; Grimm, M.W.; Tupper, P.; Miller, A.J.L.; Billings, C.J.; Jacobs, P.G. Electrode selection and other
design considerations for a modular, portable single-channel EEG-augmented hearing aid. In Proceedings of the 2013 6th
International IEEE/EMBS Conference on Neural Engineering (NER), San Diego, CA, USA, 6–8 November 2013; pp. 431–434.
[CrossRef]
59. Bryan, M.; Green, J.; Chung, M.; Chang, L.; Scherer, R.; Smith, J.; Rao, R.P.N. An adaptive brain-computer interface for humanoid
robot control. In Proceedings of the 2011 11th IEEE-RAS International Conference on Humanoid Robots, Bled, Slovenia,
26–28 October 2011; pp. 199–204. [CrossRef]
60. Chacko, S.G.; Tayade, P.; Kaur, S.; Sharma, R. Creation of a high resolution EEG based Brain Computer Interface for classifying
motor imagery of daily life activities. In Proceedings of the 2019 7th International Winter Conference on Brain-Computer Interface
(BCI), Gangwon, Korea, 18–20 February 2019; pp. 1–5. [CrossRef]
61. Chai, R.; Naik, G.R.; Nguyen, T.N.; Ling, S.H.; Tran, Y.; Craig, A.; Nguyen, H.T. Driver Fatigue Classification With Independent
Component by Entropy Rate Bound Minimization Analysis in an EEG-Based System. IEEE J. Biomed. Health Inform. 2017, 21,
715–724. [CrossRef]
62. Chakraborti, T.; Sengupta, A.; Banerjee, D.; Konar, A.; Bhattacharyya, S.; Khasnobish, A.; Janarthanan, R. Implementation of EEG
based control of remote robotic systems. In Proceedings of the 2011 International Conference on Recent Trends in Information
Systems, Kolkata, India, 21–23 December 2011; pp. 203–208. [CrossRef]
63. Chen, J.X.; Mao, Z.J.; Yao, W.X.; Huang, Y.F. EEG-based biometric identification with convolutional neural network. Multimed.
Tools Appl. 2019, 79, 10655–10675. [CrossRef]
64. Chen, S.-A.; Chen, C.-H.; Lin, J.-W.; Ko, L.-W.; Lin, C.-T. Gaming controlling via brain-computer interface using multiple
physiological signals. In Proceedings of the 2014 IEEE International Conference on Systems, Man, and Cybernetics (SMC), San
Diego, CA, USA, 5–8 October 2014; pp. 3156–3159. [CrossRef]
Sensors 2022, 22, 3331 37 of 44

65. Chennu, S.; Alsufyani, A.; Filetti, M.; Owen, A.M.; Bowman, H. The cost of space independence in P300-BCI spellers. J. Neuroeng.
Rehabil. 2013, 10, 82. [CrossRef]
66. Chiesi, M.; Guermandi, M.; Placati, S.; Scarselli, E.F.; Guerrieri, R. Creamino: A Cost-Effective, Open-Source EEG-Based BCI
System. IEEE Trans. Biomed. Eng. 2018, 66, 900–909. [CrossRef] [PubMed]
67. Chowdhury, A.; Raza, H.; Dutta, A.; Prasad, G. EEG-EMG based Hybrid Brain Computer Interface for Triggering Hand
Exoskeleton for Neuro-Rehabilitation. In Proceedings of the Advances in Robotics, New Delhi, India, 28 June–2 July 2017; pp.
1–6. [CrossRef]
68. Chumerin, N.; Manyakov, N.; Combaz, A.; Suykens, J.; Yazicioglu, R.F.; Torfs, T.; Merken, P.; Neves, H.; Van Hoof, C.; Van Hulle,
M. P300 Detection Based on Feature Extraction in On-line Brain-Computer Interface. In Proceedings of the KI 2009: Advances in
Artificial Intelligence, Berlin/Heidelberg, Germany, 15–18 September 2009; pp. 339–346. [CrossRef]
69. Chumerin, N.; Manyakov, N.V.; Combaz, A.; Suykens, J.; Van Hulle, M.M. An application of feature selection to on-line P300
detection in brain-computer interface. In Proceedings of the 2009 IEEE International Workshop on Machine Learning for Signal
Processing, Southampton, UK, 22–25 September 2013; pp. 1–6. [CrossRef]
70. Cincotti, F.; Pichiorri, F.; Arico, P.; Aloise, F.; Leotta, F.; Fallani, F.D.V.; Millan, J.D.R.; Molinari, M.; Mattia, D. EEG-based
Brain-Computer Interface to support post-stroke motor rehabilitation of the upper limb. In Proceedings of the 2012 Annual
International Conference of the IEEE Engineering in Medicine and Biology Society, San Diego, CA, USA, 28 August–1 September
2012; pp. 4112–4115. [CrossRef]
71. Cincotti, F.; Quitadamo, L.R.; Aloise, F.; Bianchi, L.; Babiloni, F.; Mattia, D. Interacting with the Environment through Non-invasive
Brain-Computer Interfaces. In Universal Access in Human-Computer Interaction. Intelligent and Ubiquitous Interaction Environments;
Springer: Berlin/Heidelberg, Germany, 2009; pp. 483–492. [CrossRef]
72. Cui, Y.; Wu, D. EEG-Based Driver Drowsiness Estimation Using Convolutional Neural Networks. In Neural Information Processing;
Springer: Cham, Switzerland, 2017; pp. 822–832. [CrossRef]
73. Djamal, E.C.; Lodaya, P. EEG based emotion monitoring using wavelet and learning vector quantization. In Proceedings of
the 2017 4th International Conference on Electrical Engineering, Computer Science and Informatics (EECSI), IEEE, Yogyakarta,
Indonesia, 19–21 September 2017; pp. 1–6. [CrossRef]
74. Dodia, S.; Edla, D.R.; Bablani, A.; Ramesh, D.; Kuppili, V. An efficient EEG based deceit identification test using wavelet packet
transform and linear discriminant analysis. J. Neurosci. Methods 2019, 314, 31–40. [CrossRef]
75. Dongwei, C.; Fang, W.; Zhen, W.; Haifang, L.; Junjie, C. Eeg-based emotion recognition with brain network using indepen-
dent components analysis and granger causality. In Proceedings of the 2013 International Conference on Computer Medical
Applications (ICCMA), IEEE, Sousse, Tunisia, 20–22 January 2013; pp. 1–6. [CrossRef]
76. Duvinage, M.; Castermans, T.; Dutoit, T.; Petieau, M.; Hoellinger, T.; De Saedeleer, C.; Seetharaman, K.; Cheron, G. A P300-based
Quantitative Comparison between the Emotiv Epoc Headset and a Medical EEG Device. Biomed. Eng. 2012, 765, 2012–2764.
[CrossRef]
77. Ehrlich, S.; Guan, C.; Cheng, G. A closed-loop brain-computer music interface for continuous affective interaction. In Proceedings
of the 2017 International Conference on Orange Technologies (ICOT), IEEE, Singapore, 8–10 December 2017; pp. 176–179.
[CrossRef]
78. Fatima, M.; Amjad, N.; Shafique, M. Analysis of Electroencephalographic Signal Acquisition and Processing for Use in Robotic
Arm Movement. In Proceedings of the 2018 IEEE-EMBS Conference on Biomedical Engineering and Sciences (IECBES), IEEE,
Sarawak, Malaysia, 3–6 December 2018; pp. 117–121. [CrossRef]
79. Fatima, M.; Shafique, M.; Khan, Z.H. Towards a low cost Brain-computer Interface for real time control of a 2 DOF robotic arm. In
Proceedings of the 2015 International Conference on Emerging Technologies (ICET), IEEE, Peshawa, Pakistan, 19–20 December
2015; pp. 1–6. [CrossRef]
80. Fazli, S.; Dahne, S.; Samek, W.; Bieszmann, F.; Mueller, K.-R. Learning From More Than One Data Source: Data Fusion Techniques
for Sensorimotor Rhythm-Based Brain–Computer Interfaces. Proc. IEEE 2015, 103, 891–906. [CrossRef]
81. Finke, A.; Hachmeister, N.; Riechmann, H.; Ritter, H. Thought-controlled robots-Systems, studies and future challenges. In
Proceedings of the 2013 IEEE International Conference on Robotics and Automation, IEEE, Karlsruhe, Germany, 6–10 May 2013;
pp. 3403–3408. [CrossRef]
82. Folgieri, R.; Zampolini, R. BCI Promises in Emotional Involvement in Music and Games. Comput. Entertain. 2014, 12, 1–10.
[CrossRef]
83. Frank, M.; Hwu, T.; Jain, S.; Knight, R.T.; Martinovic, I.; Mittal, P.; Perito, D.; Sluganovic, I.; Song, D. Using EEG-Based BCI
Devices to Subliminally Probe for Private Information. In Proceedings of the 2017 on Workshop on Privacy in the Electronic
Society, Dallas, TX, USA, 30 October 2017; pp. 133–136. [CrossRef]
84. Friedman, N.; Fekete, T.; Gal, K.; Shriki, O. EEG-Based Prediction of Cognitive Load in Intelligence Tests. Front. Hum. Neurosci.
2019, 13, 191. [CrossRef] [PubMed]
85. García-Salinas, J.S.; Villaseñor-Pineda, L.; Reyes-García, C.A.; Torres-Garcia, A.A. Transfer learning in imagined speech EEG-based
BCIs. Biomed. Signal Process. Control 2019, 50, 151–157. [CrossRef]
86. González-Castañeda, E.F.; Torres-García, A.A.; Reyes-García, C.A.; Villasenor-Pineda, L. Sonification and textification: Proposing
methods for classifying unspoken words from EEG signals. Biomed. Signal Process. Control 2017, 37, 82–91. [CrossRef]
Sensors 2022, 22, 3331 38 of 44

87. Gorjup, G.; Vrabič, R.; Stoyanov, S.P.; Andersen, M.; Manoonpong, P. Development of a Real-Time Motor-Imagery-Based EEG
Brain-Machine Interface. In Proceedings of the International Conference on Neural Information Processing, Siem Reap, Cambodia,
13–16 December 2018; Springer: Cham, Switzerland, 2018; pp. 610–622. [CrossRef]
88. Gu, J.-N.; Lu, H.-T.; Lu, B.-L. An integrated Gaussian mixture model to estimate vigilance level based on EEG recordings.
Neurocomputing 2013, 129, 107–113. [CrossRef]
89. Guger, C.; Daban, S.; Sellers, E.; Holzner, C.; Krausz, G.; Carabalona, R.; Gramatica, F.; Edlinger, G. How many people are able to
control a P300-based brain–computer interface (BCI)? Neurosci. Lett. 2009, 462, 94–98. [CrossRef]
90. Guger, C.; Holzner, C.; Groenegress, C.; Edlinger, G.; Slater, M. Brain Computer Interface for Virtual Reality Control. Cyberpsychol-
ogy Behav. 2009, 12, 84.
91. Haghighi, M.; Akcakaya, M.; Orhan, U.; Erdogmus, D.; Oken, B.; Fried-Oken, M.; Haghighi, M.; Akçakaya, M. Initial assessment
of artifact filtering for RSVP Keyboard. In Proceedings of the 2013 IEEE Signal Processing in Medicine and Biology Symposium
(SPMB), Brooklyn, NY, USA, 7 December 2013; pp. 1–5. [CrossRef]
92. Han, C.-H.; Im, C.-H. EEG-based brain-computer interface for real-time communication of patients in completely locked-in state.
In Proceedings of the 2018 6th International Conference on Brain-Computer Interface (BCI), Gangwon, Korea, 15–17 January 2018;
pp. 1–2. [CrossRef]
93. Heger, D.; Putze, F.; Schultz, T. Online Recognition of Facial Actions for Natural EEG-Based BCI Applications. In International
Conference on Affective Computing and Intelligent Interaction, Memphis, TN, USA, 9–12 October 2011; Springer: Berlin, Germany, 2011;
pp. 436–446. [CrossRef]
94. Hsieh, Z.-H.; Fang, W.-C.; Jung, T.-P. A brain-computer interface with real-time independent component analysis for biomedical
applications. In Proceedings of the 2012 IEEE International Conference on Consumer Electronics (ICCE), IEEE, Las Vegas, NV,
USA, 12–15 January 2012; pp. 339–340. [CrossRef]
95. Huang, D.; Zhang, S.; Zhang, Y. EEG-based emotion recognition using empirical wavelet transform. In Proceedings of the 2017
4th International Conference on Systems and Informatics (ICSAI), IEEE, Hangzhou, China, 11–13 November 2017; pp. 1444–1449.
[CrossRef]
96. Huang, D.; Zhang, H.; Ang, K.; Guan, C.; Pan, Y.; Wang, C.; Yu, J. Fast emotion detection from EEG using asymmetric spatial
filtering. In Proceedings of the 2012 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Kyoto,
Japan, 25–30 March 2012; pp. 589–592. [CrossRef]
97. Huster, R.J.; Mokom, Z.N.; Enriquez-Geppert, S.; Herrmann, C.S. Brain–computer interfaces for EEG neurofeedback: Peculiarities
and solutions. Int. J. Psychophysiol. 2014, 91, 36–45. [CrossRef]
98. Hwang, H.-J.; Kim, K.-H.; Jung, Y.-J.; Kim, D.-W.; Lee, Y.-H.; Im, C.-H. An EEG-based real-time cortical functional connectivity
imaging system. Med Biol. Eng. Comput. 2011, 49, 985–995. [CrossRef]
99. Hwang, H.-J.; Kim, S.; Choi, S.; Im, C.-H. EEG-Based Brain-Computer Interfaces: A Thorough Literature Survey. Int. J.
Hum.-Comput. Interact. 2013, 29, 814–826. [CrossRef]
100. Ibáñez, J.; Serrano, J.; del Castillo, M.; Gallego, J.; Rocon, E. Online detector of movement intention based on EEG—Application in
tremor patients. Biomed. Signal Process. Control 2013, 8, 822–829. [CrossRef]
101. Irimia, D.; Ortner, R.; Krausz, G.; Guger, C.; Poboroniuc, M. BCI Application in Robotics Control. IFAC Proc. Vol. 2012, 45,
1869–1874. [CrossRef]
102. Iturrate, I.; Montesano, L.; Minguez, J. Shared-control brain-computer interface for a two dimensional reaching task using EEG
error-related potentials. In Proceedings of the 2013 35th Annual International Conference of the IEEE Engineering in Medicine
and Biology Society (EMBC), IEEE, Osaka, Japan, 3–7 July 2013; pp. 5258–5262. [CrossRef]
103. Jayabhavani, G.N.; Raajan, N.R.; Abirami, J. A speller based mechanized messenger for smart phones using brain mobile interface
(BMI). In Proceedings of the 2013 International Conference on Emerging Trends in VLSI, Embedded System, Nano Electronics
and Telecommunication System (ICEVENT), Tiruvannamalai, India, 7–9 January 2013; pp. 1–5. [CrossRef]
104. Ji, H.; Li, J.; Cao, L.; Wang, D. A EEG-Based Brain Computer Interface System towards Applicable Vigilance Monitoring. In
Foundations of Intelligent Systems; Springer: Berlin/Heidelberg, Germany, 2012; pp. 743–749. [CrossRef]
105. Kapeller, C.; Hintermüller, C.; Guger, C. Augmented control of an avatar using an SSVEP based BCI. In Proceedings of the 3rd
augmented human international conference, Megève, France, 8–9 March 2012; pp. 1–2. [CrossRef]
106. Karácsony, T.; Hansen, J.P.; Iversen, H.K.; Puthusserypady, S. Brain Computer Interface for Neuro-rehabilitation With Deep
Learning Classification and Virtual Reality Feedback. In Proceedings of the 10th Augmented Human International Conference
2019, Reims, France, 11–12 March 2019; ACM: New York, NY, USA, 2019. [CrossRef]
107. Kartsch, V.; Benatti, S.; Rossi, D.; Benini, L. A wearable EEG-based drowsiness detection system with blink duration and alpha
waves analysis. In Proceedings of the 2017 8th International IEEE/EMBS Conference on Neural Engineering (NER), IEEE,
Shanghai, China, 25–28 May 2017; pp. 251–254. [CrossRef]
108. Katona, J.; Kővári, A. The Evaluation of BCI and PEBL-based Attention Tests. Acta Polytech. Hung. 2018, 15, 225–249. [CrossRef]
109. Kaushik, P.; Gupta, A.; Roy, P.P.; Dogra, D.P. EEG-Based Age and Gender Prediction Using Deep BLSTM-LSTM Network Model.
IEEE Sens. J. 2018, 19, 2634–2641. [CrossRef]
110. Kerous, B.; Skola, F.; Liarokapis, F. EEG-based BCI and video games: A progress report. Virtual Real. 2017, 22, 119–135. [CrossRef]
Sensors 2022, 22, 3331 39 of 44

111. Kim, K.-T.; Lee, S.-W. Towards an EEG-based intelligent wheelchair driving system with vibro-tactile stimuli. In Proceedings of
the 2016 IEEE International Conference on Systems, Man, and Cybernetics (SMC), IEEE, Budapest, Hungary, 9–12 October 2016;
pp. 002382–002385. [CrossRef]
112. Kim, M.; Kim, B.H.; Jo, S. Quantitative Evaluation of a Low-Cost Noninvasive Hybrid Interface Based on EEG and Eye Movement.
IEEE Trans. Neural Syst. Rehabil. Eng. 2014, 23, 159–168. [CrossRef]
113. Kim, Y.-J.; Kwak, N.-S.; Lee, S.-W. Classification of motor imagery for Ear-EEG based brain-computer interface. In Proceedings of
the 2018 6th International Conference on Brain-Computer Interface (BCI), IEEE, Gangwon, Korea, 15–17 January 2018; pp. 1–2.
[CrossRef]
114. Kinney-Lang, E.; Auyeung, B.; Escudero, J. Expanding the (kaleido)scope: Exploring current literature trends for translating
electroencephalography (EEG) based brain–computer interfaces for motor rehabilitation in children. J. Neural Eng. 2016, 13,
061002. [CrossRef]
115. Kovacevic, N.; Ritter, P.; Tays, W.; Moreno, S.; McIntosh, A. ‘My Virtual Dream’: Collective Neurofeedback in an Immersive Art
Environment. PLoS ONE 2015, 10, e0130129. [CrossRef]
116. Lan, Z.; Liu, Y.; Sourina, O.; Wang, L. Real-time EEG-based user’s valence monitoring. In Proceedings of the 2015 10th
International Conference on Information, Communications and Signal Processing (ICICS), IEEE, Singapore, 2–4 December 2015;
pp. 1–5. [CrossRef]
117. Landavazo, B.; Nandikolla, V.K. Brain-Computer Interface Application in Robotic Gripper Control. In Proceedings of thepresented
at the ASME 2018 International Mechanical Engineering Congress and Exposition, Pittsburgh, PA, USA, 9–15 November 2018.
[CrossRef]
118. Lee, Y.-C.; Lin, W.-C.; Cherng, F.-Y.; Ko, L.-W. A Visual Attention Monitor Based on Steady-State Visual Evoked Potential. IEEE
Trans. Neural Syst. Rehabil. Eng. 2015, 24, 399–408. [CrossRef]
119. Li, J.; Ji, H.; Cao, L.; Gu, R.; Xia, B.; Huang, Y. Wheelchair Control Based on Multimodal Brain-Computer Interfaces. In Neural
Information Processing; Springer: Berlin/Heidelberg, Germany, 2013; pp. 434–441. [CrossRef]
120. Li, J.; Liang, J.; Zhao, Q.; Li, J.; Hong, K.; Zhang, L. Design of Assistive Wheelchair System Directly Steered by Human Thoughts.
Int. J. Neural Syst. 2013, 23, 1350013. [CrossRef]
121. Li, M.; Zhang, Y.; Zhang, H.; Hu, H.S. An EEG Based Control System for Intelligent Wheelchair. Appl. Mech. Mater. 2013, 300–301,
1540–1545. [CrossRef]
122. Li, S.; Leider, A.; Qiu, M.; Gai, K.; Liu, M. Brain-Based Computer Interfaces in Virtual Reality. In Proceedings of the 2017 IEEE 4th
International Conference on Cyber Security and Cloud Computing (CSCloud), IEEE, New York, NY, USA, 26–28 June 2017; pp.
300–305. [CrossRef]
123. Li, Y.; Yu, T. EEG-based hybrid BCIs and their applications. In Proceedings of the 3rd International Winter Conference on
Brain-Computer Interface, Gangwon-Do, Korea, 12–14 January 2015; pp. 1–4. [CrossRef]
124. Liao, L.-D.; Chen, C.-Y.; Wang, I.-J.; Chen, S.-F.; Li, S.-Y.; Chen, B.-W.; Chang, J.-Y.; Lin, C.-T. Gaming control using a wearable
and wireless EEG-based brain-computer interface device with novel dry foam-based sensors. J. Neuroeng. Rehabil. 2012, 9, 5.
[CrossRef] [PubMed]
125. Liao, L.-D.; Wang, I.-J.; Chang, C.-J.; Lin, B.-S.; Lin, C.-T.; Tseng, K.C. Human cognitive application by using Wearable Mobile
Brain Computer Interface. In Proceedings of the TENCON 2010-2010 IEEE Region 10 Conference, IEEE, Fukuoka, Japan, 21–24
November 2010; pp. 346–351. [CrossRef]
126. Lin, C.-T.; Lin, F.-C.; Chen, S.-A.; Lu, S.-W.; Chen, T.-C.; Ko, L.-W. EEG-based Brain-computer Interface for Smart Living
Environment Auto-adjustment. J. Med. Biol. Eng. 2010, 30, 237–245. [CrossRef]
127. Lin, J.S.; Huang, S.M. An FPGA-Based Brain-Computer Interface for Wireless Electric Wheelchairs. Appl. Mech. Mater. 2013,
284–287, 1616–1621. [CrossRef]
128. Lin, J.-S.; Yang, W.-C. Wireless brain-computer interface for electric wheelchairs with EEG and eye-blinking signals. Int. J. Innov.
Comput. Inf. Control 2012, 8, 6011–6024.
129. Lin, Y.-P.; Wang, Y.; Wei, C.-S.; Jung, T.-P. A Mobile Brain-Computer Interface for Freely Moving Humans. In Human-Computer
Interaction. Towards Intelligent and Implicit Interaction; Springer: Berlin/Heidelberg, Germany, 2013; pp. 448–453. [CrossRef]
130. Liu, D.; Chen, W.; Lee, K.; Chavarriaga, R.; Bouri, M.; Pei, Z.; Millán, J.D.R. Brain-actuated gait trainer with visual and
proprioceptive feedback. J. Neural Eng. 2017, 14, 056017. [CrossRef]
131. Liu, D.; Chen, W.; Lee, K.; Chavarriaga, R.; Iwane, F.; Bouri, M.; Pei, Z.; Millan, J.D.R. EEG-Based Lower-Limb Movement Onset
Decoding: Continuous Classification and Asynchronous Detection. IEEE Trans. Neural Syst. Rehabil. Eng. 2018, 26, 1626–1635.
[CrossRef]
132. Liu, Y.; Ayaz, H.; Shewokis, P.A. Multisubject “Learning” for Mental Workload Classification Using Concurrent EEG, fNIRS, and
Physiological Measures. Front. Hum. Neurosci. 2017, 11, 389. [CrossRef]
133. Liu, Y.; Sourina, O.; Nguyen, M.K. Real-Time EEG-Based Human Emotion Recognition and Visualization. In Proceedings of the
2010 International Conference on Cyberworlds, Singapore, 20–22 October 2010; pp. 262–269. [CrossRef]
134. Liu, Y.-T.; Lin, Y.-Y.; Wu, S.-L.; Chuang, C.-H.; Lin, C.-T. Brain Dynamics in Predicting Driving Fatigue Using a Recurrent
Self-Evolving Fuzzy Neural Network. IEEE Trans. Neural Networks Learn. Syst. 2015, 27, 347–360. [CrossRef]
135. Lopez-Gordo, M.A.; Ron-Angevin, R.; Pelayo, F. Authentication of Brain-Computer Interface Users in Network Applications. In
Proceedings of the Advances in Computational Intelligence, Palma de Mallorca, Spain, 10–12 June 2015; pp. 124–132. [CrossRef]
Sensors 2022, 22, 3331 40 of 44

136. Lotte, F.; Fujisawa, J.; Touyama, H.; Ito, R.; Hirose, M.; Lecuyer, A. Towards ambulatory brain-computer Interfaces: A Pilot
Study with P300 Signals’, Book Ambul. Brain-Comput. Interfaces Pilot Study P300 Signals. In Proceedings of the International
Conference on Advances in Computer Enterntainment Technology, Athens, Greece, 29–31 October 2009. [CrossRef]
137. Lu, Y.; Hu, Y.; Liu, R.; Wang, H.; Asama, H.; Duan, F. The design of simulation vehicle system controlled by multichannel EEG
based on imaginary movements. In Proceedings of the 2016 35th Chinese Control Conference (CCC), IEEE, Chengdu, China, 27
July 2016; pp. 4976–4981. [CrossRef]
138. Ma, T.; Li, H.; Deng, L.; Yang, H.; Lv, X.; Li, P.; Li, F.; Zhang, R.; Liu, T.; Yao, D.; et al. The hybrid BCI system for movement control
by combining motor imagery and moving onset visual evoked potential. J. Neural Eng. 2017, 14, 026015. [CrossRef]
139. Mahajan, R.; Bansal, D. Real Time EEG Based Cognitive Brain Computer Interface for Control Applications via Arduino
Interfacing. Procedia Comput. Sci. 2017, 115, 812–820. [CrossRef]
140. Maiorana, E.; La Rocca, D.; Campisi, P. EEG-based biometric recognition using EigenBrains. In Proceedings of the 2015 IEEE
International Conference on Multimedia Expo Workshops (ICMEW), IEEE, Turin, Italy, 29 June–3 July 2015; pp. 1–6. [CrossRef]
141. Mandel, C.; Luth, T.; Laue, T.; Rofer, T.; Graser, A.; Krieg-Bruckner, B. Navigating a smart wheelchair with a brain-computer
interface interpreting steady-state visual evoked potentials. In Proceedings of the 2009 IEEE/RSJ International Conference on
Intelligent Robots and Systems, IEEE, Louis, MO, USA, 10–15 October 2009; pp. 1118–1125. [CrossRef]
142. Manolova, A.; Tsenov, G.; Lazarova, V.; Neshov, N. Combined EEG and EMG fatigue measurement framework with application
to hybrid brain-computer interface. In Proceedings of the 2016 IEEE International Black Sea Conference on Communications and
Networking (BlackSeaCom), IEEE, Bucharest, Romania, 6–9 June 2016; pp. 1–5. [CrossRef]
143. Mao, Z.; Yao, W.X.; Huang, Y. EEG-based biometric identification with deep learning. In Proceedings of the 2017 8th International
IEEE/EMBS Conference on Neural Engineering (NER), IEEE, Shanghai, China, 25–28 May 2017; pp. 609–612. [CrossRef]
144. Marchesotti, S.; Martuzzi, R.; Schurger, A.; Blefari, M.L.; Millan, J.D.R.; Bleuler, H.; Blanke, O. Cortical and subcortical mechanisms
of brain-machine interfaces. Hum. Brain Mapp. 2017, 38, 2971–2989. [CrossRef]
145. McFarland, D.; Wolpaw, J. EEG-based brain–computer interfaces. Curr. Opin. Biomed. Eng. 2017, 4, 194–200. [CrossRef]
146. Melinščak, F.; Jerbić, A.B.; Cifrek, M. A P300-based brain-computer interface: Towards a simpler approach. In Proceedings of the
2013 36th International Convention on Information and Communication Technology, Electronics and Microelectronics (MIPRO),
IEEE, Opatija, Croatia, 20–24 May 2013; pp. 1049–1052.
147. Melinscak, F.; Montesano, L.; Minguez, J. Asynchronous detection of kinesthetic attention during mobilization of lower limbs
using EEG measurements. J. Neural Eng. 2016, 13, 016018. [CrossRef]
148. Mikołajewska, E.; Mikołajewski, D. Non-invasive EEG-based brain-computer interfaces in patients with disorders of consciousness.
Mil. Med Res. 2014, 1, 14. [CrossRef]
149. Millán, J.D.R.; Rupp, R.; Mueller-Putz, G.; Murray-Smith, R.; Giugliemma, C.; Tangermann, M.; Vidaurre, C.; Cincotti, F.; Kubler,
A.; Leeb, R.; et al. Combining brain-computer interfaces and assistive technologies: State-of-the-art and challenges. Front.
Neurosci. 2010, 4, 161. [CrossRef]
150. Mohammadpour, M.; Mozaffari, S. Classification of EEG-based attention for brain computer interface. In Proceedings of the 2017
3rd Iranian Conference on Intelligent Systems and Signal Processing (ICSPIS), Shahrood, Iran, 20–21 December 2017; pp. 34–37.
[CrossRef]
151. Mohanchandra, K. Criminal forensic: An application to EEG. In Proceedings of the 2015 Recent and Emerging trends in Computer
and Computational Sciences (RETCOMP), IEEE, Bangalore, India, 8–10 January 2015; pp. 18–21. [CrossRef]
152. Molina, G.G.; Tsoneva, T.; Nijholt, A. Emotional brain-computer interfaces. In Proceedings of the 2009 3rd International
Conference on Affective Computing and Intelligent Interaction and Workshops, IEEE, Amsterdam, The Netherlands, 10–12
September 2009; pp. 1–9. [CrossRef]
153. Application of a brain-computer interface for person authentication using EEG responses to photo stimuli. J. Integr. Neurosci.
2018, 17, 53–60. [CrossRef]
154. Müller-Putz, G.R.; Rupp, R. Achievements and challenges of translational research in non-invasive SMR-BCI-controlled upper
extremity neuroprosthesis in spinal cord injury. In Proceedings of the 2015 7th International IEEE/EMBS Conference on Neural
Engineering (NER), Montpellier, France, 22–24 April 2015; pp. 158–161. [CrossRef]
155. Murphy, D.P.; Bai, O.; Gorgey, A.S.; Fox, J.; Lovegreen, W.T.; Burkhardt, B.W.; Atri, R.; Marquez, J.S.; Li, Q.; Fei, D.-Y.
Electroencephalogram-Based Brain–Computer Interface and Lower-Limb Prosthesis Control: A Case Study. Front. Neurol.
2017, 8, 696. [CrossRef]
156. Pan, J.; Xie, Q.; Huang, H.; He, Y.; Sun, Y.; Yu, R.; Li, Y. Emotion-Related Consciousness Detection in Patients With Disorders of
Consciousness Through an EEG-Based BCI System. Front. Hum. Neurosci. 2018, 12, 198. [CrossRef]
157. Park, J.; Woo, I.; Park, S. Application of EEG for multimodal human-machine interface. In Proceedings of the 2012 12th
International Conference on Control, Automation and Systems, Jeju Island, Korea, 17–21 October 2012; pp. 1869–1873.
158. Penaloza, C.; Hernandez-Carmona, D.; Nishio, S. Towards Intelligent Brain-Controlled Body Augmentation Robotic Limbs. In
Proceedings of the 2018 IEEE International Conference on Systems, Man, and Cybernetics (SMC), Miyazaki, Japan, 7–10 October
2018; pp. 1011–1015. [CrossRef]
159. Penaloza, C.I.; Alimardani, M.; Nishio, S. Android Feedback-Based Training Modulates Sensorimotor Rhythms During Motor
Imagery. IEEE Trans. Neural Syst. Rehabil. Eng. 2018, 26, 666–674. [CrossRef]
Sensors 2022, 22, 3331 41 of 44

160. Prashant, P.; Joshi, A.; Gandhi, V. Brain computer interface: A review. In Proceedings of the 2015 5th Nirma University
International Conference on Engineering (NUiCONE), Ahmedabad, India, 26–28 November 2015; pp. 1–6. [CrossRef]
161. Punsawad, Y.; Wongsawat, Y. Self-paced Emotional Imagery-Based Brain Computer Interface System. In Proceedings of the 6th
International Conference on the Development of Biomedical Engineering in Vietnam (BME6), Singapore, 24 September 2017;
pp. 567–571. [CrossRef]
162. Qidwai, U.A.; Shakir, M. Fuzzy Classification-Based Control of Wheelchair Using EEG Data to Assist People with Disabilities. In
Neural Information Processing; Springer: Berlin/Heidelberg, Germany, 2012; pp. 458–467. [CrossRef]
163. Qin, Z.; Xu, Y.; Shu, X.; Hua, L.; Sheng, X.; Zhu, X. eConHand: A Wearable Brain-Computer Interface System for Stroke
Rehabilitation. In Proceedings of the 2019 9th International IEEE/EMBS Conference on Neural Engineering (NER), IEEE, San
Francisco, CA, USA, 20–23 March 2019; pp. 734–737. [CrossRef]
164. Ramakuri, S.K.; Chakraborty, C.; Ghosh, S.; Gupta, B. Performance analysis of eye-state charecterization through single electrode
EEG device for medical application. In Proceedings of the 2017 Global Wireless Summit (GWS), IEEE, Cape Town, South Africa,
15–18 October 2017; pp. 1–6. [CrossRef]
165. Risangtuni, A.G.; Suprijanto; Widyotriatmo, A. Towards online application of wireless EEG-based open platform Brain Com-
puter Interface. In Proceedings of the 2012 IEEE Conference on Control, Systems Industrial Informatics, Bandung, Indonesia,
23–26 September 2012; pp. 141–144. [CrossRef]
166. Rodriguez, R.J. Electroencephalogram (EEG) based authentication leveraging visual evoked potentials (VEP) resulting from
exposure to emotionally significant images. In Proceedings of the 2016 IEEE Symposium on Technologies for Homeland Security
(HST), Waltham, MA, USA, 10–12 May 2016; pp. 1–6. [CrossRef]
167. Rutkowski, T.; Tanaka, T.; Zhao, Q.; Cichocki, A. Spatial Auditory BCI/BMI Paradigm-Multichannel EMD Approach to Brain
Responses Estimation. In Proceedings of the APSIPA annual summit and conference, Auckland, New Zealand, 7–10 December 2010;
pp. 197–202.
168. Sanchez, A.M.; Gaume, A.; Dreyfus, G.; Vialatte, F.-B. A cognitive brain-computer interface prototype for the continuous
monitoring of visual working memory load. In Proceedings of the 2015 IEEE 25th International Workshop on Machine Learning
for Signal Processing (MLSP), Boston, MA, USA, 17–20 September 2015; pp. 1–5. [CrossRef]
169. Scherer, R.; Müller-Putz, G.R.; Pfurtscheller, G. Chapter 9 Flexibility and Practicality. Int. Rev. Neurobiol. 2009, 86, 119–131.
[CrossRef]
170. Seoane, L.F.; Gabler, S.; Blankertz, B. Images from the mind: BCI image evolution based on rapid serial visual presentation of
polygon primitives. Brain-Comput. Interfac. 2015, 2, 40–56. [CrossRef]
171. Sethi, C.; Dabas, H.; Dua, C.; Dalawat, M.; Sethia, D. EEG-Based Attention Feedback to Improve Focus in E-Learning. In
Proceedings of the 2018 2nd International Conference on Computer Science and Artificial Intelligence, Shenzhen, China,
8–10 December 2018; pp. 321–326. [CrossRef]
172. Shakir, M.; Malik, A.S.; Kamel, N.; Qidwai, U. Detection of partial seizure: An application of fuzzy rule system for wearable
ambulatory systems. In Proceedings of the 2014 5th International Conference on Intelligent and Advanced Systems (ICIAS),
Kuala Lumpur, Malaysia, 3–5 June 2014; pp. 1–5. [CrossRef]
173. Shi, M.-H.; Zhou, C.-L.; Xie, J.; Li, S.-Z.; Hong, Q.-Y.; Jiang, M.; Chao, F.; Ren, W.-F.; Liu, X.-Q.; Zhou, D.-J.; et al.
Electroencephalogram-based brain-computer interface for the Chinese spelling system: A survey. Front. Inf. Technol. Electron. Eng.
2018, 19, 423–436. [CrossRef]
174. Shin, S.; Jang, D.; Lee, J.J.; Jang, S.-J.; Kim, J.-H. MyMusicShuffler: Mood-based music recommendation with the practical usage of
brainwave signals. In Proceedings of the 2014 IEEE International Conference on Consumer Electronics (ICCE), Las Vegas, NV,
USA, 10–13 January 2014; pp. 355–356. [CrossRef]
175. Simonyi, D.; Kovács, T. Brain-Computer Interface-Based Feasibility of Entering Customer Code on Ticket Vending Machines.
Interdiscip. Descr. Complex Syst. 2018, 16, 350–359. [CrossRef]
176. Sourina, O.; Liu, Y.; Hou, X.; Lim, W.L.; Mueller-Wittig, W.; Wang, L.; Konovessis, D.; Chen, C.-H.; Ang, W.T. Neuroscience Based
Design: Fundamentals and Applications. In Proceedings of the 2016 International Conference on Cyberworlds (CW), Chongqing,
China, 28–30 September 2016; pp. 250–257. [CrossRef]
177. Sourina, O.; Liu, Y.; Wang, Q.; Nguyen, M.K. EEG-Based Personalized Digital Experience. In Universal Access in Human-Computer
Interaction. Users Diversity; Springer: Berlin/Heidelberg, Germany, 2011; pp. 591–599. [CrossRef]
178. Sourina, O.; Wang, Q.; Liu, Y.; Nguyen, M.K. A Real-time Fractal-based Brain State Recognition from EEG and its Applications. In
Proceedings of the BIOSIGNALS 2011—International Conference on Bio-Inspired Systems and Signal Processing, Rome, Italy,
26–29 January 2011.
179. Spuler, M. A Brain-Computer Interface (BCI) system to use arbitrary Windows applications by directly controlling mouse and
keyboard. In Proceedings of the 2015 37th Annual International Conference of the IEEE Engineering in Medicine and Biology
Society (EMBC), Milan, Italy, 25–29 August 2015; pp. 1087–1090. [CrossRef]
180. Tan, H.G.; Kong, K.H.; Shee, C.Y.; Wang, C.C.; Guan, C.T.; Ang, W.T. Post-acute stroke patients use brain-computer interface to
activate electrical stimulation. In Proceedings of the 2010 Annual International Conference of the IEEE Engineering in Medicine
and Biology, Buenos Aires, Argentina, 31 August–4 September 2010; pp. 4234–4237. [CrossRef]
181. Thomas, K.P.; Vinod, A.P.; Guan, C. Design of an online EEG based neurofeedback game for enhancing attention and memory.
Conf. Proc. Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. IEEE Eng. Med. Biol. Soc. Annu. Conf. 2013, 2013, 433–436. [CrossRef]
Sensors 2022, 22, 3331 42 of 44

182. Tian, S.; Wang, Y.; Dong, G.; Pei, W.; Chen, H. Mental Fatigue Estimation Using EEG in a Vigilance Task and Resting States. In
Proceedings of the 2018 40th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC),
Honolulu, HI, USA, 17–21 July 2018; pp. 1980–1983. [CrossRef]
183. Tseng, K.C.; Lin, B.-S.; Han, C.-M.; Wang, P.-S. Emotion recognition of EEG underlying favourite music by support vector machine.
In Proceedings of the 2013 1st International Conference on Orange Technologies (ICOT), Tainan, Taiwan, 12–16 March 2013; pp.
155–158. [CrossRef]
184. Ullah, K.; Ali, M.; Rizwan, M.; Imran, M. Low-cost single-channel EEG based communication system for people with lock-in
syndrome. In Proceedings of the 2011 IEEE 14th International Multitopic Conference, Karachi, Pakistan, 22–24 December 2011;
pp. 120–125. [CrossRef]
185. Wang, C.; Phua, K.S.; Ang, K.K.; Guan, C.; Zhang, H.; Lin, R.; Chua, K.S.G.; Ang, B.T.; Kuah, C.W.K. A feasibility study of
non-invasive motor-imagery BCI-based robotic rehabilitation for Stroke patients. In Proceedings of the 2009 4th International
IEEE/EMBS Conference on Neural Engineering, Antalya, Turkey, 29 April–2 May 2009; pp. 271–274. [CrossRef]
186. Wang, Q.; Sourina, O.; Nguyen, M.K. EEG-Based “Serious” Games Design for Medical Applications. In Proceedings of the 2010
International Conference on Cyberworlds, Singapore, 20–22 October 2010; pp. 270–276. [CrossRef]
187. Wang, Q.; Sourina, O.; Nguyen, M.K. Fractal dimension based neurofeedback in serious games. Vis. Comput. 2011, 27, 299–309.
[CrossRef]
188. Wang, Y.-K.; Chen, S.-A.; Lin, C.-T. An EEG-based brain–computer interface for dual task driving detection. Neurocomputing 2013,
129, 85–93. [CrossRef]
189. Wang, Y.; Hidaka, T.; Kawai, Y.; Okuda, J. Why Cannot Control Your Smartphones by Thinking? Hands-Free Information Control
System Based on EEG. In Proceedings of the Innovation in Medicine and Healthcare 2015, Kyoto, Japan, 11–12 September 2015;
pp. 553–564. [CrossRef]
190. Watanabe, H.; Tanaka, H.; Sakti, S.; Nakamura, S. Neural Oscillation-Based Classification of Japanese Spoken Sentences During
Speech Perception. IEICE Trans. Inf. Syst. 2019, 102, 383–391. [CrossRef]
191. Waytowich, N.; Henderson, A.; Krusienski, D.; Cox, D. Robot application of a brain computer interface to staubli TX40 robots-early
stages. In Proceedings of the 2010 World Automation Congress, Kobe, Japan, 19–23 September 2010; pp. 1–6.
192. Wei, C.-S.; Lin, Y.-P.; Wang, Y.-T.; Jung, T.-P.; Bigdely-Shamlo, N.; Lin, C.-T. Selective Transfer Learning for EEG-Based Drowsiness
Detection. In Proceedings of the 2015 IEEE International Conference on Systems, Man, and Cybernetics, Kowloon, China, 9–12
October 2015; pp. 3229–3232.
193. Wei, C.-S.; Wang, Y.-T.; Lin, C.-T.; Jung, T.-P. Toward Drowsiness Detection Using Non-hair-Bearing EEG-Based Brain-Computer
Interfaces. IEEE Trans. Neural Syst. Rehabil. Eng. 2018, 26, 400–406. [CrossRef]
194. Wei, P. An virtual vehicle control game for online brain computer interface feedback training. In Proceedings of the 2010 IEEE
International Conference on Mechatronics and Automation, Xi’an, China, 4–7 August 2010; pp. 1942–1944. [CrossRef]
195. Wolpaw, J.R. Brain–Computer Interface Research Comes of Age: Traditional Assumptions Meet Emerging Realities. J. Mot. Behav.
2010, 42, 351–353. [CrossRef]
196. Wriessnegger, S.C.; Hackhofer, D.; Muller-Putz, G.R. Classification of unconscious like/dislike decisions: First results towards a
novel application for BCI technology. Conf. Proc. Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. IEEE Eng. Med. Biol. Soc. Annu. Conf.
2015, 2015, 2331–2334. [CrossRef]
197. Wu, S.-L.; Liu, Y.-T.; Chou, K.-P.; Lin, Y.-Y.; Lu, J.; Zhang, G.; Chuang, C.-H.; Lin, W.-C.; Lin, C.-T. A motor imagery based
brain-computer interface system via swarm-optimized fuzzy integral and its application. In Proceedings of the 2016 IEEE
International Conference on Fuzzy Systems (FUZZ-IEEE), Vancouver, BC, Canada, 24–29 July 2016; pp. 2495–2500. [CrossRef]
198. Xiao, J.; Lin, Q.; Yu, T.; Xie, Q.; Yu, R.; Li, Y. A BCI system for assisting visual fixation assessment in behavioral evaluation
of patients with disorders of consciousness. In Proceedings of the 2017 8th International IEEE/EMBS Conference on Neural
Engineering (NER), Shanghai, China, 25–28 May 2017; pp. 399–402. [CrossRef]
199. Xiao, J.; Xie, Q.; He, Y.; Yu, T.; Lu, S.; Huang, N.; Yu, R.; Li, Y. An Auditory BCI System for Assisting CRS-R Behavioral Assessment
in Patients with Disorders of Consciousness. Sci. Rep. 2016, 6, 32917. [CrossRef] [PubMed]
200. Xie, Y.; Li, X. A brain controlled wheelchair based on common spatial pattern. In Proceedings of the 2015 International Symposium
on Bioelectronics and Bioinformatics (ISBB), Beijing, China, 14–17 October 2015; pp. 19–22. [CrossRef]
201. Xu, H.; Plataniotis, K.N. EEG-based affect states classification using Deep Belief Networks. In Proceedings of the 2016 Digital
Media Industry Academic Forum (DMIAF), Santorini, Greece, 4–6 July 2016; pp. 148–153. [CrossRef]
202. Xu, R.; Heilinger, A.; Murovec, N.; Spataro, R.; Cho, W.; Cao, F.; Allison, B.Z.; Guger, C. Effects of Repeating a Tactile Brain-
Computer Interface on Patients with Disorder of Consciousness: A Hint of Recovery? In Proceedings of the 2019 9th International
IEEE/EMBS Conference on Neural Engineering (NER), San Francisco, CA, USA, 20–23 March 2019; pp. 29–32. [CrossRef]
203. Yan, J.; Chen, S.; Deng, S. A EEG-based emotion recognition model with rhythm and time characteristics. Brain Inform. 2019, 6, 7.
[CrossRef] [PubMed]
204. Yin, J.; Hu, J. Design and Implement of Pocket PC Game Based on Brain-Computer Interface. In Computing and Intelligent Systems;
Springer: Berlin/Heidelberg, Germany, 2011; pp. 456–463. [CrossRef]
205. Yin, J.; Hu, J.; Mu, Z. Design and Implement of BCI System Based on Android Platform. In Information Engineering and Applications;
Springer: London, UK, 2012; pp. 522–529. [CrossRef]
Sensors 2022, 22, 3331 43 of 44

206. Yu, Y.; Liu, Y.; Yin, E.; Jiang, J.; Zhou, Z.; Hu, D. An Asynchronous Hybrid Spelling Approach Based on EEG–EOG Signals for
Chinese Character Input. IEEE Trans. Neural Syst. Rehabil. Eng. 2019, 27, 1292–1302. [CrossRef]
207. Yue, J.; Zhou, Z.; Jiang, J.; Liu, Y.; Hu, D. Balancing a simulated inverted pendulum through motor imagery: An EEG-based
real-time control paradigm. Neurosci. Lett. 2012, 524, 95–100. [CrossRef]
208. Zao, J.K.; Gan, T.-T.; You, C.-K.; Chung, C.-E.; Wang, Y.-T.; Méndez, S.J.R.; Mullen, T.; Yu, C.; Kothe, C.; Hsiao, C.-T.; et al.
Pervasive brain monitoring and data sharing based on multi-tier distributed computing and linked data technology. Front. Hum.
Neurosci. 2014, 8, 370. [CrossRef] [PubMed]
209. Zarjam, P.; Epps, J.; Lovell, N. Beyond Subjective Self-Rating: EEG Signal Classification of Cognitive Workload. IEEE Trans. Auton.
Ment. Dev. 2015, 7, 301–310. [CrossRef]
210. Zeng, H.; Yang, C.; Zhang, H.; Wu, Z.; Zhang, J.; Dai, G.; Babiloni, F.; Kong, W. A LightGBM-Based EEG Analysis Method for
Driver Mental States Classification. Comput. Intell. Neurosci. 2019, 2019, 3761203. [CrossRef]
211. Zhang, H.; Chavarriaga, R.; Khaliliardali, Z.; Gheorghe, L.A.; Iturrate, I.; Millan, J.D.R. EEG-based decoding of error-related brain
activity in a real-world driving task. J. Neural Eng. 2015, 12, 066028. [CrossRef]
212. Zhang, X.; Yao, L.; Sheng, Q.Z.; Kanhere, S.S.; Gu, T.; Zhang, D. Converting Your Thoughts to Texts: Enabling Brain Typing via
Deep Feature Learning of EEG Signals. In Proceedings of the 2018 IEEE International Conference on Pervasive Computing and
Communications (PerCom) Converting, Athens, Greece, 19–23 March 2018; pp. 1–10.
213. Zhou, J.; Yao, J.; Deng, J.; Dewald, J.P. EEG-based classification for elbow versus shoulder torque intentions involving stroke
subjects. Comput. Biol. Med. 2009, 39, 443–452. [CrossRef]
214. Zhou, Z.; Li, P.; Liu, J.; Dong, W. A Novel Real-Time EEG Based Eye State Recognition System. In Communications and Networking;
Springer: Cham, Switzerland, 2019; pp. 175–183. [CrossRef]
215. Zioga, P.; Chapman, P.; Ma, M.; Pollick, F. A Hypothesis of Brain-to-Brain Coupling in Interactive New Media Art and Games
Using Brain-Computer Interfaces. In Serious Games; Springer: Cham, Switzerland, 2015; pp. 103–113. [CrossRef]
216. Zioga, P.; Pollick, F.; Ma, M.; Chapman, P.; Stefanov, K. “Enheduanna—A Manifesto of Falling” Live Brain-Computer Cinema
Performance: Performer and Audience Participation, Cognition and Emotional Engagement Using Multi-Brain BCI Interaction.
Front. Neurosci. 2018, 12, 191. [CrossRef]
217. Chaudhary, U.; Birbaumer, N.; Ramos-Murguialday, A. Brain–computer interfaces for communication and rehabilitation. Nat.
Rev. Neurol. 2016, 12, 513–525. [CrossRef]
218. Dadebayev, D.; Goh, W.W.; Tan, E.X. EEG-based emotion recognition: Review of commercial EEG devices and machine learning
techniques. J. King Saud Univ.-Comput. Inf. Sci. 2021. [CrossRef]
219. Larocco, J.; Le, M.D.; Paeng, D.-G. A Systemic Review of Available Low-Cost EEG Headsets Used for Drowsiness Detection.
Front. Neuroinformatics 2020, 14, 553352. [CrossRef]
220. Radüntz, T.; Meffert, B. User Experience of 7 Mobile Electroencephalography Devices: Comparative Study. JMIR mHealth uHealth
2019, 7, e14474. [CrossRef]
221. Soufineyestani, M.; Dowling, D.; Khan, A. Electroencephalography (EEG) Technology Applications and Available Devices. Appl.
Sci. 2020, 10, 7453. [CrossRef]
222. Amaral, C.; Simões, M.; Mouga, S.; Andrade, J.; Castelo-Branco, M. A novel Brain Computer Interface for classification of social
joint attention in autism and comparison of 3 experimental setups: A feasibility study. J. Neurosci. Methods 2017, 290, 105–115.
[CrossRef]
223. Hinrichs, H.; Scholz, M.; Baum, A.K.; Kam, J.W.Y.; Knight, R.T.; Heinze, H.-J. Comparison between a wireless dry electrode EEG
system with a conventional wired wet electrode EEG system for clinical applications. Sci. Rep. 2020, 10, 5218. [CrossRef]
224. Ratti, E.; Waninger, S.; Berka, C.; Ruffini, G.; Verma, A. Comparison of Medical and Consumer Wireless EEG Systems for Use in
Clinical Trials. Front. Hum. Neurosci. 2017, 11, 398. [CrossRef]
225. Zerafa, R.; Camilleri, T.; Falzon, O.; Camilleri, K.P. A comparison of a broad range of EEG acquisition devices–is there any
difference for SSVEP BCIs? Brain-Comput. Interfaces 2018, 5, 121–131. [CrossRef]
226. Xie, Y.; Oniga, S. A Review of Processing Methods and Classification Algorithm for EEG Signal. Carpath. J. Electron. Comput. Eng.
2020, 13, 23–29. [CrossRef]
227. Doma, V.; Pirouz, M. A comparative analysis of machine learning methods for emotion recognition using EEG and peripheral
physiological signals. J. Big Data 2020, 7, 1–21. [CrossRef]
228. Nakagome, S.; Luu, T.P.; He, Y.; Ravindran, A.S.; Contreras-Vidal, J.L. An empirical comparison of neural networks and machine
learning algorithms for EEG gait decoding. Sci. Rep. 2020, 10, 4372. [CrossRef] [PubMed]
229. Varszegi, K. Comparison of algorithms for detecting hand movement from EEG signals. In Proceedings of the 2016 IEEE
International Conference on Systems, Man, and Cybernetics (SMC), Budapest, Hungary, 9–12 October 2016; pp. 002208–002213.
[CrossRef]
230. Pires, G.; Torres, M.; Casaleiro, N.; Nunes, U.; Castelo-Branco, M. Playing Tetris with non-invasive BCI. In Proceedings of the 2011
IEEE 1st International Conference on Serious Games and Applications for Health (SeGAH), Braga, Portugal, 16–18 November
2011; pp. 1–6. [CrossRef]
231. Pires, G.; Nunes, U.; Castelo-Branco, M. Evaluation of Brain-computer Interfaces in Accessing Computer and other Devices by
People with Severe Motor Impairments. Procedia Comput. Sci. 2012, 14, 283–292. [CrossRef]
Sensors 2022, 22, 3331 44 of 44

232. Cruz, A.; Pires, G.; Nunes, U.J. Double ErrP Detection for Automatic Error Correction in an ERP-Based BCI Speller. IEEE Trans.
Neural Syst. Rehabil. Eng. 2017, 26, 26–36. [CrossRef]
233. Cruz, A.; Pires, G.; Lopes, A.; Carona, C.; Nunes, U.J. A Self-Paced BCI With a Collaborative Controller for Highly Reliable
Wheelchair Driving: Experimental Tests With Physically Disabled Individuals. IEEE Trans. Hum.-Mach. Syst. 2021, 51, 109–119.
[CrossRef]
234. Muhammad, Y.; Vaino, D. Controlling Electronic Devices with Brain Rhythms/Electrical Activity Using Artificial Neural Network
(ANN). Bioengineering 2019, 6, 46. [CrossRef]
235. Roots, K.; Muhammad, Y.; Muhammad, N. Fusion Convolutional Neural Network for Cross-Subject EEG Motor Imagery
Classification. Computers 2020, 9, 72. [CrossRef]
236. Tamm, M.-O.; Muhammad, Y.; Muhammad, N. Classification of Vowels from Imagined Speech with Convolutional Neural
Networks. Computers 2020, 9, 46. [CrossRef]
237. Krauledat, M.; Tangermann, M.; Blankertz, B.; Müller, K.-R. Towards Zero Training for Brain-Computer Interfacing. PLoS ONE
2008, 3, e2967. [CrossRef]
238. Zhang, X.; Yao, L.; Wang, X.; Monaghan, J.; McAlpine, D.; Zhang, Y. A survey on deep learning-based non-invasive brain signals:
Recent advances and new frontiers. J. Neural Eng. 2021, 18, 031002. [CrossRef]
239. Jiao, Y.; Zhou, T.; Yao, L.; Zhou, G.; Wang, X.; Zhang, Y. Multi-View Multi-Scale Optimization of Feature Representation for EEG
Classification Improvement. IEEE Trans. Neural Syst. Rehabil. Eng. 2020, 28, 2589–2597. [CrossRef]
240. Zhang, Y.; Zhou, T.; Wu, W.; Xie, H.; Zhu, H.; Zhou, G.; Cichocki, A. Improving EEG Decoding via Clustering-Based Multitask
Feature Learning. IEEE Trans. Neural Netw. Learn. Syst. 2021, 1–11. [CrossRef]
241. Kwon, O.-Y.; Lee, M.-H.; Guan, C.; Lee, S.-W. Subject-Independent Brain–Computer Interfaces Based on Deep Convolutional
Neural Networks. IEEE Trans. Neural Netw. Learn. Syst. 2019, 31, 3839–3852. [CrossRef]
242. Mussabayeva, A.; Jamwal, P.K.; Akhtar, M.T. Comparison of Generic and Subject-Specific Training for Features Classification
in P300 Speller. In Proceedings of the 2020 Asia-Pacific Signal and Information Processing Association Annual Summit and
Conference (APSIPA ASC), Tokyo, Japan, 14–17 December 2020; pp. 222–227.
243. Joadder, A.; Siuly, S.; Kabir, E.; Wang, H.; Zhang, Y. A New Design of Mental State Classification for Subject Independent BCI
Systems. IRBM 2019, 40, 297–305. [CrossRef]
244. Kwon, J.; Im, C.-H. Subject-Independent Functional Near-Infrared Spectroscopy-Based Brain–Computer Interfaces Based on
Convolutional Neural Networks. Front. Hum. Neurosci. 2021, 15, 121. [CrossRef]
245. Barrios, L.J.; del Castillo, M.D.; Serrano, J.I.; Pons, J.L. A review of fMRI as a tool for enhancing EEG-based brain-machine
interfaces. Appl. Bionics Biomech. 2012, 9, 125–133. [CrossRef]
246. Edelman, B.J.; Meng, J.; Suma, D.; Zurn, C.; Nagarajan, E.; Baxter, B.S.; Cline, C.C.; He, B. Noninvasive neuroimaging enhances
continuous neural tracking for robotic device control. Sci. Robot. 2019, 4, eaaw6844. [CrossRef]
247. Meng, J.; Zhang, S.; Bekyo, A.; Olsoe, J.; Baxter, B.; He, B. Noninvasive Electroencephalogram Based Control of a Robotic Arm for
Reach and Grasp Tasks. Sci. Rep. 2016, 6, 38565. [CrossRef]