REVIEW ARTICLE

Recent Progress in Wearable

Brain–Computer Interface (BCI) Citation: Zhang J, Li J, Huang Z,
Devices Based on Electroencephalogram (EEG) Huang D, Yu H, Li Z. Recent Progress
in Wearable Brain–Computer

for Medical Applications: A Review Interface (BCI) Devices Based on

Electroencephalogram (EEG) for
Medical Applications: A Review.
Jiayan Zhang1,2, Junshi Li1,2, Zhe Huang1,2,3, Dong Huang1,2,4, Health Data Sci. 2023;3:Article 0096.
Huaiqiang Yu5, and Zhihong Li1,2* https://doi.org/10.34133/hds.0096

1 Submitted 24 February 2023

Beijing Advanced Innovation Center for Integrated Circuits, Beijing, China. 2National Key Laboratory
Accepted 19 October 2023
of Advanced Micro and Nano Manufacture Technology, School of Integrated Circuits, Peking University, Published 19 December 2023
Beijing, China. 3Shenzhen Graduate School, Peking University, Shenzhen, China. 4School of Electronics,
C​opy​rig​h​t © 2023 Jiayan Zhang
Peking University, Beijing, China. 5Sichuan Institute of Piezoelectric and Acousto-optic Technology, et al. ​Exc​lusive licensee P​eki​ng ​
Chongqing, China. University Health Science Center. No
claim to original U.S. Government

Downloaded from https://spj.science.org on June 22, 2024

Works. Distributed under a Creative
*Address correspondence to: [email protected] Commons Attribution License 4.0
(CC BY 4.0).
Importance: Brain–computer interface (BCI) decodes and converts brain signals into machine instructions
to interoperate with the external world. However, limited by the implantation risks of invasive BCIs and the
operational complexity of conventional noninvasive BCIs, applications of BCIs are mainly used in laboratory
or clinical environments, which are not conducive to the daily use of BCI devices. With the increasing
demand for intelligent medical care, the development of wearable BCI systems is necessary. Highlights: Based
on the scalp-electroencephalogram (EEG), forehead-EEG, and ear-EEG, the state-of-the-art wearable BCI
devices for disease management and patient assistance are reviewed. This paper focuses on the EEG
acquisition equipment of the novel wearable BCI devices and summarizes the development direction of
wearable EEG-based BCI devices. Conclusions: BCI devices play an essential role in the medical field.
This review briefly summarizes novel wearable EEG-based BCIs applied in the medical field and the latest
progress in related technologies, emphasizing its potential to help doctors, patients, and caregivers better
understand and utilize BCI devices.

Introduction degrading the performance and lifespan of invasive BCIs [2]. To

minimize the damage to the brain, semi-invasive BCIs have been
Brain–computer interface (BCI) establishes a connection bet­ proposed. The electrodes of semi-invasive BCIs are usually placed
ween the brain and the machine to replace, restore, supplement, under the skull or below the dura mate to record electrocorti­
or enhance brain functions. A BCI system has 4 functional mod­ cography (ECoG). Compared with invasive BCIs, semi-invasive
ules: signal acquisition, data pre-processing, feature extraction BCIs are less invasive to the brain. Invasive BCIs and semi-invasive
and classification, and output equipment [1]. According to the BCIs can be collectively referred to as invasive BCIs.
brain signal acquisition method, there are 3 main types of BCI: To avoid the risk of infection and brain damage caused by
invasive BCI, semi-invasive BCI, and noninvasive BCI [2]. electrode implantation, noninvasive BCIs have been proposed.
Invasive BCIs realize external devices controlled by decoding Brain signal acquisition of noninvasive BCIs does not require
neuronal activities inside the brain. Since the recording elec­ surgical implantation of electrodes, which has a low risk of
trodes are close to the brain signals’ source and unaffected by brain damage and can reduce the psychological and physical
the brain tissues’ attenuation and filtering, invasive BCIs can burden of the users. Standard noninvasive techniques include
record and analyze the information in neural signals with the electroencephalogram (EEG), magnetoencephalogram (MEG),
highest temporal–spatial resolution and accuracy. At present, functional near-infrared imaging (fNIR), and functional mag­
invasive BCIs have been applied in motion control [3–6], disease netic resonance imaging (fMRI) [2]. EEG waves are clusters of
diagnosis and treatment [7–9], communication assistance [10–13], electrical signals from neurons in the brain and are usually
cursor control [14–16], and other aspects. However, invasive obtained through electrodes placed on the scalp, forehead, and
BCIs require the surgical implantation of signal acquisition behind the ears [17], while MEG, fMRI, and fNIR need to be
devices into the brain, which may bring risks of damage or infec­ collected with bulky specialized acquisition equipment.
tion to the brain. In addition, the immune response surrounding For medical applications of BCIs, intermittent neurological
the implanted electrode will decrease signal quality over time, diseases (epilepsy, migraine, etc.) require long-term continuous

Zhang et al. 2023 | https://doi.org/10.34133/hds.0096 1

monitoring with BCI systems to diagnose and predict, and the
assistive devices (prosthetics, robotic arms, wheelchairs, etc.)
controlled by BCI systems must be used in real scenarios. Long-
term continuous signal monitoring and daily device control rely
on the wearability of BCIs. It is of great importance to make
BCIs wearable. Wearability is defined as a performance that can
be carried out in daily activities without notably obstructing the
subject and affecting the devices’ operation. For ease of use,
wearable devices are mainly placed outside the user’s body and
can be worn and taken off by users independently. Invasive BCIs
require surgical implantation of signal acquisition electrodes
and are normally defined as implantable devices. Although some
wearable BCI systems are composed of semi-invasive brain sig­
nal acquisition devices and wearable actuators [18–20], the rel­
evant work is limited. Noninvasive BCIs based on MEG, fMRI,
and fNIR are limited by large signal acquisition devices and can
only be used in hospitals and other specific places, which cannot
meet the requirements for daily wearable use. On the contrary,
EEG-based BCI devices do not need electrode implantation for
signal acquisition and the use is not restricted by the venue,
making EEG-based BCIs suitable for daily wearable applications.
Hence, EEG-based BCIs for wearable medical applications are
mainly focused in this review.

Downloaded from https://spj.science.org on June 22, 2024

Fig. 1. Categories of novel wearable EEG-based BCIs and the primary medical
Nowadays, wearable EEG-based BCIs mainly rely on the wet applications.
electrode EEG caps for stable and reliable EEG signal acquisition
and have been applied to the diagnosis and auxiliary treatment
of various diseases, such as disorders of consciousness [21–23],
Parkinson’s disease [24], paralysis [25–27], stroke [28–30], de­ applications) to 256 [2]. In the past 2 decades, many scalp-EEG-
pression [31], autism [32], and sleep disorders [33]. Besides, based BCIs with conventional EEG caps have been applied in
EEG-based BCIs are also used for wheelchair control [34–36]. medical, education, military, and other fields. However, due to
However, wet electrodes require skin preparation and the appli­ inconvenience of skin preparation and conductive gels, and the
cation of conductive gels for low electrode-skin impedances and complexity of wire connection, the EEG-based BCIs with con­
high-quality EEG signals are obtained. Because the conductivity ventional EEG caps cannot meet the requirements of daily use.
of the gel decreases with time, it is hard to realize long-term EEG Scalp-EEG recording has experienced a long development
signal acquisition with the wet electrode caps. Moreover, con­ history. To improve the wearability of the scalp-EEG-based BCIs,
ventional EEG-based BCI systems also have limitations such as scalp-EEG acquisition devices with unique structures and mate­
bulky size, complex wires, and dependence on large apparatus rials have been proposed. In this section, 3 commercial scalp-
(e.g., the EEG amplifier), which increase the discomfort of wear­ EEG headsets, Emotiv EPOC (Fig. 2A), DSI-24 (Fig. 2B), and
ing and limit the daily use of EEG-based BCIs. OpenBCI Ultracortex “Mark IV” (Fig. 2C), and their wearable
For wearable BCI systems, brain signal acquisition equip­ BCI applications in the medical field are introduced in detail.
ment is of great importance not only because the quality and
reliability of brain signal acquisition will affect the performance Emotiv EPOC
of the BCI systems but also the signal-acquiring methods will Emotiv EPOC is a typical consumer-oriented EEG equipment
influence the wearability of the BCIs to some extent. Therefore, released by Emotiv Systems Company of the United States.
this topic review introduces wearable EEG-based BCIs from Emotiv EPOC is an octopus structure that gives users good flex­
the perspective of EEG acquisition equipment and focuses on ibility and movement. The device has 14 saline-based recording
their applications in the medical field. According to the distri­ electrodes and 2 reference electrodes [37]. The headset can wire­
bution of recording electrodes, EEG-based BCIs reviewed in lessly connect to computer or mobile devices and continuously
this paper are divided into scalp-EEG-based BCIs, forehead- work for up to 12 h. Emotiv EPOC has been widely used in
EEG-based BCIs, and ear-EEG-based BCIs, as shown in Fig. 1. research, since its development [38].
The applications of wearable EEG-based BCIs in the medical Here, we review some of the representative wearable BCI
field are summarized as disease management (including disease work by using Emotiv EPOC. For the neuro-rehabilitation of
prevention and diagnosis), rehabilitation therapy, health mon­ stroke patients, Jure et al. [39] presented a functional electrical
itoring, communication assistance, and equipment control. stimulation (FES)-based BCI system (Fig. 2D), which was made
up of an Emotiv EPOC headset for EEG signal recording and
a 2-channel controlled stimulator for neuromuscular system
Scalp-EEG-Based BCI stimulation. When the BCI system detected cerebral activities
Brain signals of EEG-based BCIs are usually obtained from the related to motor imagery (MI), the electrical stimulator would
scalp. Compared with the forehead and behind-/in-ear areas, be activated to realize therapeutic intervention. Tabernig et al.
the scalp offers more sampling space, which is conducive to [40] used the proposed BCI-FES system to perform neuro­
arranging more EEG signal recording points. The number of rehabilitation therapy for patients with sequelae of ischemic
signal-recording electrodes ranges from a few (for targeted BCI stroke and evaluated the effects. The cerebral cortex activation

Zhang et al. 2023 | https://doi.org/10.34133/hds.0096 2

 Downloaded from https://spj.science.org on June 22, 2024

Fig. 2. Wearable scalp-EEG-based BCIs and their medical applications. (A) Emotiv EPOC headset [43]. ©2016 EDP Sciences. Reprinted with permission from Swee et al.
[43]. (B) DSI-24 headset (https://www.neurospec.com/Products/Details/1079/dsi-24). (C) OpenBCI Ultracortex “Mark IV” headset (https://docs.openbci.com/AddOns/
Headwear/MarkIV/). (D) Block diagram of the BCI-FES system [39]. ©2016 IOPscience. Reprinted with permission from Jure et al. [39]. (E) Scores of the quality of movement
(left) and quality of life (right) were measured for each stroke patient [40]. Used with permission from Tabernig et al. [40]; permission conveyed through Copyright Clearance
Center Inc. (F) BCI-NFB system (left) and power spectrum density in one representative participant (right) [41]. ©2019 BMC. Reprinted with permission from Al-Taleb et al.
[41]. (G) User with Emotiv EPOC headset, BCI program, and FES electrodes on the arm [42]. ©2021 BMC. Reprinted with permission from Zulauf-Czaja et al. [42]. (H) The
percentage of true positive activation (left) and time to activate FES (right) out of all attempted trials [42]. ©2021 BMC. Reprinted with permission from Zulauf-Czaja et al.
[42]. (I) Illustration of mu suppression in affected and unaffected hemispheres [52]. ©2019 MDPI. Reprinted with permission from Choi et al. [52].

during the presence of MI and the sensory feedback produced Fugl–Meyer score, and marked posttreatment improvement
by the movement were used to facilitate neuroplasticity. Before was detected (Fig. 2E), suggesting that the proposed therapy
and after the intervention, the upper limb was assessed by the could benefit stroke individuals’ neuro-rehabilitation.

Zhang et al. 2023 | https://doi.org/10.34133/hds.0096 3

 Besides the neurological rehabilitation of stroke patients, Eldeeb et al. [50] developed an EEG-based BCI system to
wearable BCI with the Emotiv EPOC can also be used to detect analyze the distress effect on the brain activity of autism spec­
central neuropathic pain (CNP), which is a frequent chronic trum disorder (ADS) individuals [51]. Based on the affective
condition in spinal cord injury (SCI) patients. Al-Taleb et al. Posner task, the proposed BCI identified the patterns associated
[41] designed a BCI system based on the Emotiv EPOC device with emotion regulation. The EEG signals were obtained from
(Fig. 2F, left) for self-managed neurofeedback (NFB) treatment 21 ADS with DSI-24 EEG headsets. Choi et al. [52] designed an
of people with chronic SCI. According to the visual feedback action observation BCI system based on the DSI-24 EEG head­
of selected frequency EEG band power, users had to self-regu­ set and detected the participants’ attention level by analyzing
late their primary motor cortex brain activity with the BCI-NFB Mu rhythm (i.e., alpha wave, 8 to 13 Hz) power when watching
system. Results showed that users had successfully regulated a video of repetitive grasping actions. The system provided the
their brainwaves in a frequency-specific manner (Fig. 2F, right). steady-state visual evoked potentials (SSVEPs) as the feedback.
The reduction in pain experienced was clinically obvious Results showed that, compared with conventional action obser­
(greater than 30%) in 8 participants, demonstrating that the vation (AO), the proposed BCI-AO suppressed stroke patients’
BCI-NFB system could reduce CNP in people with SCI. In Mu rhythm more (Fig. 2I), suggesting that the proposed para­
2021, the same research team presented a BCI-FES system digm was an effective tool for stroke patients’ rehabilitation. Kim
based on an Emotiv EPOC headset (Fig. 2G) for hand function et al. [53] combined BCI-AO with peripheral electrical stimu­
rehabilitation and evaluated its usability [42]. Hand therapy was lation (PES) and assessed the effect of the system on corticos­
performed by producing the attempted movement of one hand pinal plasticity for motor recovery. In this work, participants
to lower the 8- to 12-Hz frequency band power and activate watched a video of repeated gripping movements under 4 dif­
FES to induce wrist flexion and extension. The system obtained ferent tasks, and visual feedback was provided by BCI. Motor-
an accuracy of 70 to 90%, and the median activation time of evoked potentials (MEPs) were measured during the task.
FES remained constant across sessions (Fig. 2H). According to the results, 4 tasks all realized that the MEP latency
BCIs based on the Emotiv EPOC headset also play an impor­ decreased and had the potential in promoting corticospinal

Downloaded from https://spj.science.org on June 22, 2024

tant role in patient assistance. To meet the needs of some para­ plasticity in stroke patients.
lyzed patients for wheelchair control without joysticks, different Zhang et al. [54] designed an SSVEP-BCI system for robotic
BCI-controlled wheelchairs have been developed. In 2016, Swee arm control in patient assistance. To improve the system oper­
et al. [43] developed a brainwave-controlled wheelchair with ation effectiveness, an adaptive decoding FBCCA algorithm
an Emotiv EPOC headset. Based on the EEG signal, the electric was adopted, which could adapt to individual differences. In
wheelchair performed the desired movement and achieved up this work, an average recognition success rate of 95.5% was
to 90% accuracy [44]. Voznenko et al. [45] developed a robotic obtained, proving that the proposed system allowed the hand­
wheelchair controlled by the onboard computer that received icapped to grasp objects by controlling the mechanical arm
commands from the extended Emotiv EPOC BCI. To improve through the brain in daily life.
system control accuracy, the robotic wheelchair could also be
controlled by voice and gestures. Zgallai et al. [46] designed a OpenBCI Ultracortex “Mark IV”
smart wheelchair for paralyzed people who are unable to con­ Ultracortex “Mark IV” is an EEG headset designed to work
trol their bodies. The wheelchair BCI system completed 4 kinds with OpenBCI boards. The headset can record research-grade
of command recognition and achieved an accuracy rate of up EEG signals and sample up to 16 channels from all 35 recording
to 96%. Shahin et al. [47] proposed a wheelchair control BCI locations [55]. Spikey (for the scalp with hair) and nonspikey
system that could switch between the automatic control mode (for the scalp without hair) dry electrodes could be screwed at
and the manual control mode. Three types of input, EEG sig­ the locations. Since the Ultracortex “Mark IV” headset could
nals, head gestures, and facial expressions, were collected and be 3D-printed in different sizes, it is adjustable for different
translated into 4 control instructions. head shapes and sizes and enables all electrodes to contact the
Besides direction control, Bousseta et al. [48] designed a scalp closely. Compared with other consumer-grade EEG head­
novel BCI system with an Emotiv EPOC headset for robotic sets, the OpenBCI “Mark IV” is cost-effective [56]. The BCIs
arm control. Participants were instructed to imagine the exe­ based on the Ultracortex “Mark IV” EEG headset are mainly
cution of hand or foot movements. After command translation, used for emotion recognition [57–59] and assistance technol­
subjects achieved control of the robotic arm in 4 directions ogy [60–63].
and obtained an average accuracy of 85.45%, which had the To help disabled persons control a robotic arm, Lim and Quan
potential to provide a helpful aid for the disabled. [60] presented an EEG-based BCI system for robotic arm con­
trol. EEG signals were classified into 8 mental commands with
DSI-24 convolutional neural network (CNN) model to realize 6-degree-
DSI-24 is a wireless EEG headset developed by the Wearable of-freedom operations. The recall rate and precision of the sys­
Sensing Company of the United States. DSI-24 has 21 signal-­ tem were 91.9% and 92%. Saragih et al. [61] applied CNN and
recording dry electrodes, including 19 electrodes on the head long short-term memory (LSTM) networks in an EEG-based
for full head coverage, 2 ear clip electrodes, and 3 auxiliary sen­ BCI system for effectively controlling the artificial hand. In the
sors [49]. The electrodes are spring-loaded to provide constant hand operations classification, the accuracy of the CNN model
and comfortable pressure, which enhances the contact between was 95.45%, while that of the LSTM model was 93.64%. Bolaños
electrodes and the skin and, at the same time, reduces motion et al. [62] created a room prototype that allowed people with
artifacts. With the unique structure, DSI-24 can obtain the EEG motor disabilities to realize the control of the light and bed, and
signal with a quality comparable to that of wet electrodes even request assistance with a buzzer through an EEG-based BCI.
with dry electrodes. The electrodes use active/passive shielding EEG data recorded with Ultracortex “Mark IV” headset were
technology to prevent electromagnetic interference. filtered in the alpha band to train a one-dimensional (1D) CNN

Zhang et al. 2023 | https://doi.org/10.34133/hds.0096 4

model. The proposed system achieved a real-time classification accuracy based on the support vector machine with radial basis
accuracy of 78.75%. function (SVMRBF) predictor. Compared with the previous
To enhance cognitive control, Dutta et al. [63] developed a work, the accuracy improved.
BCI-based application with interactive media. Corresponding Chiu et al. [67] built an SSVEP-based BCI eating assistive
alpha and beta waves were obtained from the recorded EEG system (Fig. 3D). In this system, different frequencies of visual
signals and regarded as parameters of attention to move the stimulation caused EEG waves with specific characteristics so
object in the proposed game. This work provided doctors with as to achieve different functional selections. Users can also use
a new choice for psychological disorder treatment. this system for subjective training to optimize the current user’s
Although the wearable BCIs based on scalp-EEG have been model. The SSVEP-based BCI enabled the disabled to have meals
widely used, the influence of hair on EEG signal acquisition by themselves, obtaining 91.35% average accuracy and 20.69 bits
cannot be ignored. Obtaining high-quality EEG signals on the per minute information transfer rate. To further improve the
scalp is challenging without the aid of conductive agents. Even practicability of the system, Lin et al. [65] integrated more enter­
if the spring pressurized structure is applied in scalp-EEG head­ tainment and interaction functions into the system and gained
sets to improve the contact between the electrodes and scalp, a 90.91% average accuracy.
the contact impedance is still high, which cannot guarantee
long-term stable EEG recording and is easily affected by motion
artifacts. BrainLink Lite
BrainLink Lite is a head-mounted EEG sensor developed by
Shenzhen Macrotellect Company for iOS and Android systems.
Forehead-EEG-Based BCI It has 3 gold-plated copper dry electrodes, including an EEG
Forehead-EEG-based BCI converts the EEG signals recorded recording electrode, a ground electrode, and a reference electrode
from the forehead area into commands to realize actuators’ con­ [68], as the Fig. 3B shows. BrainLink Lite can be easily worn on
trol. The forehead is an ideal place for electrodes attaching as a the forehead. Because the sensor connects to the smart terminals

Downloaded from https://spj.science.org on June 22, 2024

non-hair-bearing area. Recording EEG on the forehead requires via Bluetooth, it avoids complicated leads and is convenient to
no complex preparation work prior to signal acquisition and acquire EEG signals in nonlaboratory environments.
avoids the interference from hair. Besides, forehead-EEG includes Japaridze et al. [69] used the BCI based on the BrainLink Lite
rich information associated with cognitive abilities and dysfunc­ to detect absence seizures. The BrainLink Lite was used for fore­
tions [64]. Given this unique feature, the research on wearable head-EEG acquisition, and a predefined algorithm was used for
forehead-EEG-based BCIs has gained wide attention in recent the automated detection of absence seizures in real time. This
years, and forehead-EEG BCIs are often used for attention mon­ work obtained an average sensitivity of absence seizure detection
itoring and then to realize functions such as sleep monitoring, of 78.83% with the proposed BCI system, which showed the
cursor control, and external device control. potential to detect absence seizures with wearable BCIs in every­
Nowadays, many existing consumer-grade wearable BCIs day life. For children with autism NFB training, Mercado et al.
and clinical applications rely on forehead-EEG. This section [70] designed a BCI video game named FarmerKeeper. The
introduces wearable BCI systems’ development and medical attention of children with autism was read by BrainLink Lite
application based on forehead-EEG acquisition devices. Three and used to control a runner in the BCI video game. The results
forehead-EEG acquisition devices introduced in this section showed that the proposed BCI video game improved attention
are shown in Fig. 3. and reduced the anxiety of children with autism.
To restore motor function through robot-assisted rehabilita­
tion therapy, Li et al. [71] introduced a BCI system for wrist
Mindo EEG device rehabilitation (Fig. 3E). In the system, the attention level was
The Mindo 4S EEG device developed by the Brain Research
measured to activate a flexible wrist exoskeleton for wrist reha­
Center of National Chiao Tung University consists of 4-channel
bilitation training. The overall actuation success rate was 95%,
dry recording electrodes and 2 extra reference channels (Fig.
proving the feasibility of attention-based control.
3A). To realize appropriate electrode-skin contact impedance,
spring-loaded dry electrodes were adopted in the Mindo 4S
EEG device, whose wearing position can be adjusted [64]. The MindWave Mobile
flexible substrate acting as a buffer can eliminate pain when a MindWave Mobile developed by the NeuroSky Company is
force is applied, which is conducive to wear comfort and long- made up of a headset, an ear clip, and a sensor arm (Fig. 3C).
term EEG acquisition. The sampling rate of Mindo 4S is up to The only EEG recording electrode of the MindWave Mobile
512 Hz and can be adjusted according to the system’s require­ is located on the forehead above the eye (FP1 position accord­
ments. The system uses Bluetooth to realize wireless data trans­ ing to the 10-20 International System) with a reference elec­
mission and can work continuously for 20 h [65]. trode and grounding electrode inside the ear clip. Therefore,
In 2017, Lin et al. [64] proposed a silicon-based dry sensor the MindWave Mobile EEG headset is commonly used for eye
for forehead-EEG acquisition, the Mindo 4S EEG device pro­ blink detection [72–75]. The device outputs 12-bit 3- to 100-Hz
totype. The proposed system successfully realized 5 sleep stages original brainwaves with a sampling rate of 512 Hz and can
identification, headache prevention, and a rapid antidepressant achieve an 8-h battery run time. With only one recording elec­
agent assessment. In 2018, Cao et al. [66] from the same research trode, MindWave Mobile is more portable and more accessible
team used the Mindo 4S EEG device to assess the ketamine to wear than multi-channel EEG sensors.
effect in patients with treatment-resistant depression. The band For neurotherapy, Mercado et al. [76] used MindWave
power and asymmetry of the alpha band as the feedback of the Mobile for attention detection and used FarmerKeeper [70] as
BCI system were detected. Results showed that the BCI system an assistant tool. Results showed that participants who used
classified the responders and nonresponders with 81.3 ± 9.5% FarmerKeeper were more focused during NFB sessions. The

Zhang et al. 2023 | https://doi.org/10.34133/hds.0096 5

 Downloaded from https://spj.science.org on June 22, 2024
Fig. 3. Wearable forehead-EEG-based BCIs and their medical applications. (A) Mindo 4S EEG device [64]. ©2017 IEEE. Reprinted with permission from Lin et al. [64].
(B) BrainLink Lite [69] (https://www.epihunter.com/brainlink-headset). (C) MindWave Mobile [80]. ©2020 IEEE. Reprinted with permission from Tiwari et al. [80].
(D) Eating assistive BCI system based on SSVEP [67]. ©2017 IEEE. Reprinted with permission from Chiu et al. [67]. (E) Diagrammatic sketch of brain-controlled wrist
rehabilitation BCI [71]. Used with permission from Li et al. [71]; permission conveyed through Copyright Clearance Center Inc. (F) Text error rate and entry speed of
QWERTY virtual keyboard (left) and ABC virtual keyboard (right) [77]. ©2020 UAD. Reprinted with permission from Salih et al. [77].

pre- and post-­assessment indicated that all autistic children EEG signals. With 2 designed visual keyboards, an average text
improved their attention. This technique helps researchers real­ entry speed of about 1.55 to 1.8 words per minute and an error
ize real-time attention monitoring and regulation through rate of 5 to 5.25% were obtained (Fig. 3F).
wearable BCIs. However, because only a single recording electrode is placed
To help the disabled communicate with the outside world, on the forehead, the richness and accuracy of the EEG signals
Salih and Abdal [77] designed a BCI-based visual keyboard using recorded with the MindWave Mobile headset are limited, and
the MindWave Mobile headset. Participants were demanded to less electrode recording may be more obvious interference by
write “Help” words for 9 sessions on visual keyboards. For print­ artifacts. The accuracy of the wheelchair control studies [78–80]
ing proposes, voluntary blinks and attention were detected using that relied only on EEG signals collected with the MindWave

Zhang et al. 2023 | https://doi.org/10.34133/hds.0096 6

Mobile headset was lower than that of other BCI-controlled responses. Zibrandtsen et al. [87] compared the in-ear EEG and
wheelchair studies. To improve the control accuracy, Girase and scalp-EEG recorded from patients with suspected temporal lobe
Deshmukh [81] used not only the EEG signals but also the epilepsy. Results suggested that ear-EEG was a reliable signal
blinking eye signal as a control reference in the BCI-controlled source to detect EEG patterns associated with focal temporal
wheelchair system, and a control accuracy of about 95% was lobe seizures.
obtained. Although there are some feasibility studies on in-ear EEG
Although the acquisition of forehead-EEG is relatively sim­ for BCI applications [88–90], in-ear EEG has not been widely
ple because there is no need for complex skin processing, the applied in wearable medical BCIs.
application of forehead-EEG in wearable BCI may be proble­
matic since the forehead-EEG is easy to be affected by ocular Behind the ear EEG-based BCI
artifacts and facial muscle artifacts, which will influence the In 2015, based on funding from TSMi (Oldenzaal, The Netherlands),
quality of signal acquisition and the accuracy of the whole BCI Debener et al. [91] from the University of Oldenburg designed a
system. As the forehead region is far from the occipital region flexible c-shape ear-EEG acquisition sensor (cEEGrid), as shown
and temporal lobe, the characteristic EEG signals of the relevant in Fig. 4C, which is made up of 10 printed Ag/AgCl electrodes and
brain regions are difficult to collect. Moreover, for everyday placed behind the ear. To realize signal recording, the electrode gel
use, most forehead-EEG acquisition devices are conspicuous was applied to the cEEGrid’s electrodes. Researchers proved that
and unsuitable for daily wear. the cEEGrid electrodes array could record reliable EEG data for
the first time. One year later, the same research team found clear
Ear-EEG-Based BCI attention-modulated event-related potential (ERP) effects in EEG
signals recorded with cEEGrid sensors, which agreed with the sig­
EEG signal recorded by ear-EEG-based BCIs comes from the nals recorded with classical EEG caps in morphology and effect
ear canals or the area behind the ear. The areas in and behind size. The discovery demonstrated that the cEEGrid sensor could
the ears are favorable positions for EEG signal acquisition for measure well-described ERPs and be expected to replace classical

Downloaded from https://spj.science.org on June 22, 2024

non-hair bearing. Compared to other EEG acquisition devices, EEG-cap in auditory attention monitoring [92]. In a further study,
ear-EEG acquisition devices are more miniaturized, and only a the visual and cognitive ERPs (N1, P1, P300) and event-related
small part of the head is covered with electrodes, which is more lateralizations (ERLs) were recorded with cEEGrid, while motor-­
comfortable to wear and is suitable for daily use. In addition, related cortical potentials were not well measured [93,94].
ear-EEG can map more brain regions than the forehead-EEG. For the hearing impaired, the function of target speaker iden­
Recently, many research teams have developed different ear- tification in assistive devices is of great importance. In 2016, the
EEG sensors, studied the information contained in ear-EEG, and research team used EEG signals recorded by the cEEGrid sensor
designed corresponding wearable BCI systems. According to the and EEG caps to identify the attended speaker [95]. Based on
position where the ear-EEG sensor is worn, there are 2 categories the ear-EEG, the positive correlation of the performance scores
ear-EEG-based BCIs: (a) in-ear EEG-based BCIs with recording was evident (Fig. 4E). The decoding accuracy of ear-EEG was
electrodes placed in the ear canals, and (b) behind the ear EEG- 69.3%, suggesting that the cEEGrid sensor has the potential to
based BCIs with recording electrodes placed behind the ear. The apply for BCI control of hearing aids. To evaluate the cEEGrid
corresponding 2 types of ear-EEG acquisition devices are shown in attention selection, normal-hearing and cochlear implant
in Fig. 4. participants were recruited [96]. For the cEEGrid data, only half
of both the cochlear implant and normal-hearing users obtained
In-ear EEG-based BCI decoding accuracies above the chance level (Fig. 4F).
In 2011, Looney et al. [82] from Imperial College London Other research teams have also carried out a series of stud­
designed an in-the-ear (ITE) electrode (Fig. 4A) for wearable ies on wearable BCIs based on the behind the ear EEG. Millard
ear-EEG recording. The proposed ITE system used 2 or more et al. [97] recorded individual peak alpha frequency (PAF)
electrodes embedded into an earplug to record EEG signals. with cEEGrid sensors to predict future pain severity in patients
The earplug was produced based on the 3D printing of the ear undergoing thoracotomy. Segaert et al. [98] took use of the
canal, and the mounted electrodes were made from silver/silver cEEGrid setups for early language comprehension impairment
chloride (Ag/AgCl). ITE electrode showed excellent correlation detection in mild cognitive impairment (MCI). The proposed
and coherence with on-scalp electrodes and was proven to system showed outstanding classification ability when detect­
extract several key EEG features, including the auditory steady- ing patients with MCI from the healthy controls.
state response (ASSR), alpha attenuation response (AAR), and In addition to pain prediction and language comprehension
P300 paradigms, which illustrated the potential of ear-EEG in impairment detection, behind the ear EEG signal is also used
BCI application [83,84]. A further study demonstrated that the for seizure detection. However, most studies rely on experi­
signal-to-noise ratio of the ear-EEG signal was comparable to enced epileptologists to analyze and annotate the ear-EEG data,
that of the EEG recorded in the temporal region [85]. rather than using a complete BCI system [99,100]. Swinnen
Since ITE earphones need custom earpieces, the cost of time et al. [100] developed a semiautomatic absence seizure detec­
and money is relatively high. In 2015, Goverdovsky et al. from tion algorithm to label ear-EEG, which could reduce the con­
the same laboratory proposed a novel in-ear sensor (Fig. 4B) for sumption of the recording review time and the workload of
high-quality long-term EEG monitoring [86]. The in-ear sensor epileptologists to some extent.
used viscoelastic substrates and conductive clothes to realize For home appliance control, Kaongoen et al. [101] designed
stable electrical contact between the sensors and the ear canals. a novel online BCI system. The system used speech imagery
To establish a low-impedance contact, only saline solution was (SI) and ear-EEG recorded with a custom-made wearable BCI
required. The in-ear sensor was proved to be useful in capturing headphone (Fig. 4D). Each side of the headphone has 4 record­
a wide frequency range of EEG and all of the standard EEG ing semi-dry electrodes distributed behind the ear. During the

Zhang et al. 2023 | https://doi.org/10.34133/hds.0096 7

 Downloaded from https://spj.science.org on June 22, 2024

Fig. 4. Wearable ear-EEG-based BCIs and their medical applications. (A and B) In-the-ear electrodes [82,86]. (C and D) Behind the ear electrodes [101,124]. (E) Correlation
between performance scores [95]. (F) Selective attention accuracies for normal hearing (NH, left) and cochlear implants (CI, right) across lags [96]. Used with permission
from Looney et al. [82]; permission conveyed through Copyright Clearance Center Inc. (A). Used with permission from Goverdovsky et al. [86]; permission conveyed through
Copyright Clearance Center Inc. (B). ©2021 Taylor & Francis Ltd. Reprinted with permission from Knierim et al. [124] (C). ©2022 Elsevier Ireland Ltd. Reprinted with permission
from Kaongoen et al. [101] (D). ©2016 Frontiers. Reprinted with permission from Mirkovic et al. [95] (E). ©2019 Frontiers. Reprinted with permission from Nogueira et al. [96] (F).

Zhang et al. 2023 | https://doi.org/10.34133/hds.0096 8

online experiments, a few users could control the home appli­ electrodes with soft circuits [110] were proposed for EEG signal
ance freely with the ear-EEG SI-based BCI system. acquisition in the hairy scalp. Nevertheless, without the aid of
Compared with scalp-EEG-based BCIs and forehead-EEG- conductive agents, the contact impedance between the dry elec­
based BCIs, the wearable ear-EEG-based BCI is still in the stage trodes and hairy scalp was generally higher than that between
of practical exploration and improvement and the medical the wet electrode and hairy scalp. The proposal of the semi-dry
applications are relatively insufficient. The possible limitations electrodes [111,112] has made a compromise between wearing
and challenges of ear-EEG-based BCIs are listed as follows. comfort and contact impedance.
a. Ear-EEG signal is weak and can be easily affected by muscle For non-hair-bearing areas like the scalp without hair, fore­
artifacts when the jaw moves, such as speaking and swallowing. head, and behind-/in-the-ear, there are more kinds of electrodes
For wearable BCI applications, the hardware and software of to choose from. Increasing the number of electrodes is beneficial
signal processing need to be further improved. to obtain a richer EEG signal, so G.tec Company designed a
b. The signals from different brain regions overlap at the ear high-density pangolin wet electrode system, g.PANGOLIN, and
area, resulting in the difficulty of extracting the characteristic up to 1024 EEG channels were recorded on the head to achieve
signals. The current methods and technologies for feature extrac­ high-resolution signal acquisition. For improving the quality of
tion and classification are still not efficient or accurate enough. signal acquisition, various flexible electrodes have also been
c. Compared with the scalp-EEG, the ear-EEG mapping applied for wearable EEG-based BCIs [113–115]. In addition,
model to the brain region is not perfect and the algorithm is not flexible electrodes can also reduce motion artifacts and provide
yet mature. Perfecting the model and algorithm is the key to better wearing comfort. For wearable BCIs based on ear-EEG,
expanding the wearable application of the ear-EEG-based BCIs. the morphology of the ear canal and auricle should also be con­
sidered for electrode designing [116,117].
Limitations and Challenges of Wearable Artifact is another key factor affecting EEG signal quality.
For wearable EEG-based BCIs, random noise and unexpected
EEG-Based BCIs signal artifacts, such as muscle artifacts, ocular artifacts, cardiac

Downloaded from https://spj.science.org on June 22, 2024

Currently, wearable BCIs based on scalp-EEG, forehead-EEG, artifacts, and extrinsic artifacts, will affect EEG signal acquisition
and ear-EEG have played an important role in medical applica­ during daily activities [118]. The main artifact elimination meth­
tions such as disease diagnosis and prediction, neurological ods can be divided into 2 categories. One approach is to opti­
rehabilitation, health monitoring, and auxiliary equipment con­ mizing the EEG electrodes to realize low electrode-skin contact
trol, providing a new form of medical treatment. However, wear­ impedance, which can apparently reduce the artifact coupling.
able BCI technology is still in the research stage. At present, the The other way is setting the electrocardiogram (ECG), electrooc­
volume of BCI is large and brain signal processing relies on back- ulogram (EOG), and electromyogram (EMG) leads to record
end equipment such as a computer. Due to the individual differ­ the 3 main sources of the artifacts synchronously with EEG and
ences in EEG signal, for each new user, a lot of pre-training and then remove the artifacts in EEG signals with the algorithm for
professional guidance are indispensable to achieve a relatively artifact elimination. Different artifact removal algorithms have
stable BCI performance. Besides, current wearable BCI systems been proposed to minimize the influence of artifacts in recent
are not robust enough to overcome the interference caused by years. For instance, Lee et al. [119] proposed a method called
users and the environment. The high cost of time and money constrained independent component analysis with online learn­
also makes wearable BCIs hard to be widely used. To meet the ing (cIOL) to search and reject the movement artifacts in EEG
needs of further application, wearable BCI devices still have a signals. Chang et al. [120] optimized the parameters of the arti­
lot of room for improvement in high-quality signal acquisition, fact subspace reconstruction (ASR) artifacts removal approach,
wireless transmission, information security, wearing comfort, demonstrating that ASR successfully removed ocular and muscle
and so on. Various research teams are active in the frontier artifact components. Egambaram et al. [121] proposed unsu­
research of wearable EEG-based BCIs, and the latest research pervised eye blink artifact detection algorithms and achieved
progress is briefly introduced in this section. an over 90% average artifact removal accuracy.
Acquisition of EEG signals is an important part of BCI. The The complicated lead connection increases the installation
quality of signal acquisition directly influences the performance complexity and system weight, which brings inconvenience to
of the BCI system. Since wearable BCIs are commonly used for the wearer’s daily activities. In order to improve the wearability
a relatively long period, long-term stable and reliable EEG signal of BCI devices, it is important to realize low power consump­
recording is required. To realize high-quality EEG signal acqui­ tion, high bandwidth, and high-precision wireless data trans­
sition, different kinds of electrodes have been proposed. For mission. At present, because of the mature technology, stable
EEG signal recording on a hairy scalp, hair is a natural barrier. performance, and low cost, Bluetooth and WiFi modules are
Standard Ag/AgCl wet electrode uses conductive gel to enhance commonly used wireless signal transmission methods. Recently,
the electrical contact between the electrodes and the scalp. novel wireless transmission methods have been proposed. For
However, the properties of the conductive gel decline with time, instance, Qiu et al. [122] designed a low-power data acquisi­
which is difficult to meet the requirement of long-term EEG tion front-end solution based on the field programmable gate
recording. To overcome this shortage, novel hydrogel electrodes array (FPGA), and realized scalp-EEG signal real-time wireless
for long-term stable EEG recording were designed [102–104]. transmission.
The downside of hydrogel electrodes is that hair washing is In addition to the use of wireless transmission, improving
required after the use of hydrogel electrodes, which is not con­ the flexibility, minimization, and lightweight of the wearable
ducive to the daily use of wearable EEG-based BCIs. Therefore, BCI system is also conducive to increasing the wearing comfort
dry electrode research has gained more and more attention. of users. With the development of technology, multi-source
Comb-shaped active dry electrodes [105–108], annular-shaped acquisition and multi-function integrated wearable BCI sys­
dry microneedle array electrodes [109], and dry microneedle tems have become a new development trend [123]. However,

Zhang et al. 2023 | https://doi.org/10.34133/hds.0096 9

with the use of wireless transmission technology and the wide­ As can be seen from Tables 1 and 2, the wearable BCIs based on
spread application of wearable BCIs, a large amount of user scalp-EEG and forehead-EEG are relatively mature and have
information has been collected, and data security needs to be been successfully commercialized, while the ear-EEG-based
paid more attention to. wearable BCI has not been sufficiently applied in the medical
field yet.
Conclusion The characteristics of the 3 types of EEG are discussed in
the corresponding sections, and each type of EEG has its own
EEG acquisition is an important component of wearable BCI strengths and weaknesses for wearable BCI applications. Scalp-
systems. In this paper, novel wearable EEG-based BCIs designed EEG contains relatively rich brain information compared with
for medical applications are reviewed from the perspective of the forehead-EEG and ear-EEG. As the mainstream wearable
EEG signal recording. According to the recording electrode BCI solution, both hardware and software of scalp-EEG-based
position, wearable EEG-based BCIs are divided into 3 categories: BCI technology are the most mature. Nevertheless, the scalp-
scalp-EEG-based BCI, forehead-EEG-based BCI, and ear-EEG- EEG acquisition equipment is complicated to wear and signal
based BCI. Table 1 summarizes the wearable EEG-based BCI recording can be easily interfered with hair. Both forehead-EEG
devices reviewed in this paper and their related applications in and ear-EEG are recorded from hairless or less hairy areas,
the medical field. Table 2 lists the work of wearable EEG-based which reduces for the inconvenience of wearing to some extent.
BCIs for equipment control and the corresponding accuracies. However, forehead-EEG is easily affected by artifacts, and hard

Table 1. Summary of wearable EEG devices and corresponding BCI applications in medical field

Reference EEG device EEG recorded Electrode type Recording channels BCI applications

Downloaded from https://spj.science.org on June 22, 2024

Emotiv Systems Emotiv EPOC Scalp-EEG Semi-dry electrode 14 Rehabilitation therapy
Company, USA [40–42]
Equipment control
[43–48]
Wearable Sensing DSI-24 Scalp-EEG Dry electrode 21 Rehabilitation therapy
Company, USA [50,52,53]
Equipment control
[54]
OpenBCI OpenBCI Ultracortex Scalp-EEG Dry electrode 16 Equipment control
Company, USA “Mark IV” [60–62]
Rehabilitation therapy
[63]
National Chiao Tung Mindo 4S EEG Forehead-EEG Dry electrode 4 Health monitoring
University, CHN device [64]
Disease management
[64]
Rehabilitation therapy
[64,66]
Equipment control
[65,67]
Macrotellect BrainLink Lite Forehead-EEG Dry electrode 3 Disease management
Company, CHN [69]
Rehabilitation therapy
[70,71]
NeuroSky MindWave Forehead-EEG Dry electrode 1 Rehabilitation therapy
Company, USA Mobile [76]
Communication
assistance [77]
Equipment control
[78–81]
Debener et al. [91] and cEEGrid Ear-EEG Wet electrode 10 Communication
TSMi, NLD assistance [95,96]
Disease management
[97,98]
Kaongoen et al. [101] Custom-made Ear-EEG Semi-dry electrode 8 Equipment control
headphone [101]

Zhang et al. 2023 | https://doi.org/10.34133/hds.0096 10

 Table 2. Summary of applications of wearable EEG-based BCIs in equipment control

Reference EEG recorded EEG acquisition device Equipment Accuracy

Swee et al. [43,44] Scalp-EEG Emotiv EPOC Wheelchair 90%
Voznenko et al. [45] Scalp-EEG Emotiv EPOC Wheelchair /
Zgallai et al. [46] Scalp-EEG Emotiv EPOC Wheelchair 70% (raw EEG data)
96% (spectrum of EEG)
Shahin et al. [47] Scalp-EEG Emotiv EPOC Wheelchair /
Bousseta et al. [48] Scalp-EEG Emotiv EPOC Robotic arm 85.45%
Zhang et al. [54] Scalp-EEG DSI-24 Robotic arm 95.5%
Lim and Quan [60] Scalp-EEG Ultracortex “Mark IV” Robotic arm 92%
Saragih et al. [61] Scalp-EEG Ultracortex “Mark IV” Artificial hand 95.45% (CNN)
93.64% (LSTM)
Bolaños et al. [62] Scalp-EEG Ultracortex “Mark IV” Light, bed, and buzzer 78.75%
for requesting
Chiu et al. [67] Forehead-EEG Mindo 4S Robotic arm 91.35%
Lin et al. [65] Forehead-EEG Mindo 4S Eating assistance, video 90.91%
entertainment, video
calling, and active inter-

Downloaded from https://spj.science.org on June 22, 2024

action functions
Permana et al. [78] Forehead-EEG MindWave Mobile Wheelchair 52.78–55.56%
Devasia et al. [79] Forehead-EEG MindWave Mobile Wheelchair, light, TV, fan, 78%
door
Tiwari et al. [80] Forehead-EEG MindWave Mobile Wheelchair /
Girase and Deshmukh [81] Forehead-EEG MindWave Mobile Wheelchair 95%
Kaongoen et al. [101] Ear-EEG Custom-made headphone Home appliances 85%

to record the brain signals produced by the regions far from Author contributions: All authors contributed and approved
the frontal head, while the ear-EEG is hard to accurately extract the final manuscript. J.Z. and J.L. conceived and contributed to
and classify the characteristic signal because of the low intensity the structure of the manuscript. J.Z. performed the literature
and high signal overlap. With the gradual improvement of the search and drafted and edited the manuscript. Z.H. performed
ear-EEG signal mapping model and the data processing algo­ the data collection. J.L., D.H., and H.Y. contributed to the edit­
rithm, wearable ear-EEG-based BCI is expected to become the ing of the manuscript. Z.L. supervised the work, revised the
mainstream technology in the next generation. manuscript, and critically edited the manuscript.
In order to meet the needs of daily use, wearable EEG-based Competing interests: The authors declare that they have no
BCI still needs to be further improved to meet the requirements competing interests.
of high-quality brain signal acquisition, efficient and stable
wireless data transmission, long-term wearing comfort, and References
high control accuracy. In the foreseeable future, owing to the
advanced electrode technology, the optimized montage con­ 1. Zhuang M, Wu Q, Wan F, Hu Y. State-of-the-art non-invasive
figuration, and the powerful algorithm, wearable EEG-based brain–computer interface for neural rehabilitation: A review.
BCI devices and systems will continue to develop in the form J Neurorestoratol. 2020;8(1):12–25.
of simplicity, comfort, high performance, and automation— 2. Rao RP. Brain-computer interfacing: An introduction.
reasons to believe that wearable EEG-based BCIs will bring an Cambridge, UK: Cambridge University Press; 2013.
innovation of medical technology, fully employed in disease 3. Bouton CE, Shaikhouni A, Annetta NV, Bockbrader MA,
prediction, diagnosis, treatment, auxiliary equipment control, Friedenberg DA, Nielson DM, Sharma G, Sederberg PB,
and other broad medical applications. Glenn BC, Mysiw WJ, et al. Restoring cortical control of
functional movement in a human with quadriplegia. Nature.
2016;533(7602):247–250.
Acknowledgments 4. Ajiboye AB, Willett FR, Young DR, Memberg WD,
Funding: This work was supported by the National Key Research Murphy BA, Miller JP, Walter BL, Sweet JA, Hoyen HA,
and Development Program of China (nos. 2022YFF1202301 Keith MW, et al. Restoration of reaching and grasping
and 2022YFF1502100), the National Natural Science Foundation movements through brain-controlled muscle stimulation in
of China (no. 62004007), and the China Postdoctoral Science a person with tetraplegia: A proof-of-concept demonstration.
Foundation (no. 2021M700258). Lancet. 2017;389(10081):1821–1830.

Zhang et al. 2023 | https://doi.org/10.34133/hds.0096 11

 5. Schwemmer MA, Skomrock ND, Sederberg PB, Ting JE, sensing movement intention and closed-loop deep brain
Sharma G, Bockbrader MA, Friedenberg DA. Meeting stimulation with wearable sensors in an essential tremor
brain–computer interface user performance expectations patient. J Neurosurg. 2016;127(3):580–587.
using a deep neural network decoding framework. Nat Med. 19. Herron JA, Thompson MC, Brown T, Chizeck HJ,
2018;24(11):1669–1676. Ojemann JG, Ko AL. Cortical brain–computer interface for
6. Flesher SN, Downey JE, Weiss JM, Hughes CL, Herrera AJ, closed-loop deep brain stimulation. IEEE Trans Neural Syst
Tyler-Kabara EC, Boninger ML, Collinger JL, Gaunt RA. Rehabil Eng. 2017;25(11):2180–2187.
A brain-computer interface that evokes tactile sensations 20. Laiwalla F, Lee J, Lee AH, Mok E, Leung V, Shellhammer S,
improves robotic arm control. Science. 2021;372(6544): Song YK, Larson L, Nurmikko A. A distributed wireless
831–836. network of implantable sub-mm cortical microstimulators
7. Gilja V, Pandarinath C, Blabe CH, Nuyujukian P, for brain-computer interfaces. Paper presented at: 2019 41st
Simeral JD, Sarma AA, Sorice BL, Perge JA, Jarosiewicz B, Annual International Conference of the IEEE Engineering in
Hochberg LR, et al. Clinical translation of a high-performance Medicine and Biology Society (EMBC); Berlin, Germany; 2019.
neural prosthesis. Nat Med. 2015;21(10):1142–1145. 21. Xie Q, Pan J, Chen Y, He Y, Ni X, Zhang J, Wang F, Li Y,
8. Bronte-Stewart HM, Petrucci MN, O’day JJ, Afzal MF, Yu R. A gaze-independent audiovisual brain-computer
Parker JE, Kehnemouyi YM, Wilkins KB, Orthlieb GC, interface for detecting awareness of patients with disorders of
Hoffman SL. Perspective: Evolution of control variables and consciousness. BMC Neurol. 2018;18(144):1–12.
policies for closed-loop deep brain stimulation for Parkinson’s 22. Pan J, Xie Q, Huang H, He Y, Sun Y, Yu R, Li Y. Emotion-
disease using bidirectional deep-brain-computer interfaces. related consciousness detection in patients with disorders of
Front Hum Neurosci. 2020;14:353. consciousness through an EEG-based BCI system. Front Hum
9. Yang Y, Sani OG, Lee MB, Dawes HE, Chang EF, Shanechi MM. Neurosci. 2018;12:198.
Developing a closed-loop brain-computer interface for treatment 23. Huang H, Xie Q, Pan J, He Y, Wen Z, Yu R, Li Y. An EEG-
of neuropsychiatric disorders using electrical brain stimulation. based brain computer interface for emotion recognition and

Downloaded from https://spj.science.org on June 22, 2024

Brain-Computer Interface Research. Cham, Switzerland: its application in patients with disorder of consciousness.
Springer; 2021. p. 103–106. IEEE Trans Affect Comput. 2019;12(4):832–842.
10. Jarosiewicz B, Sarma AA, Bacher D, Masse NY, Simeral JD, 24. Kasahara K, Hoshino H, Furusawa Y, Sayo Dasalla C,
Sorice B, Oakley EM, Blabe C, Pandarinath C, Gilja V, et al. Honda M, Murata M, Hanakawa T. Initial experience with
Virtual typing by people with tetraplegia using a self-calibrating a sensorimotor rhythm-based brain-computer interface
intracortical brain-computer interface. Sci Transl Med. in a Parkinson’s disease patient. Brain-Comput Interfaces.
2015;7(313):313ra179. 2018;5(2-3):88–96.
11. Pandarinath C, Nuyujukian P, Blabe CH, Sorice BL, Saab J, 25. Han CH, Kim YW, Kim DY, Kim SH, Nenadic Z, Im CH.
Willett FR, Hochberg LR, Shenoy KV, Henderson JM. High Electroencephalography-based endogenous brain–computer
performance communication by people with paralysis interface for online communication with a completely locked-
using an intracortical brain-computer interface. elife. in patient. J Neuroeng Rehabil. 2019;16(1):18.
2017;6:Article e18554. 26. Verbaarschot C, Tump D, Lutu A, Borhanazad M, Thielen J,
12. Milekovic T, Sarma AA, Bacher D, Simeral JD, Saab J, Van Den Broek P, Farquhar J, Weikamp J, Raaphorst J,
Pandarinath C, Sorice BL, Blabe C, Oakley EM, Tringale KR, Groothuis JT, et al. A visual brain-computer interface as
et al. Stable long-term BCI-enabled communication in ALS communication aid for patients with amyotrophic lateral
and locked-in syndrome using LFP signals. J Neurophysiol. sclerosis. Clin Neurophysiol. 2021;132(10):2404–2415.
2018;120(7):343–360. 27. Nicolelis MA, Alho EJ, Donati AR, Yonamine S, Aratanha MA,
13. Moses DA, Metzger SL, Liu JR, Anumanchipalli GK, Bao G, Campos DS, Almeida S, Fischer D, Shokur S. Training
Makin JG, Sun PF, Chartier J, Dougherty ME, Liu PM, with noninvasive brain–machine interface, tactile feedback, and
Abrams GM, et al. Neuroprosthesis for decoding speech locomotion to enhance neurological recovery in individuals
in a paralyzed person with anarthria. N Engl J Med. with complete paraplegia: A randomized pilot study. Sci Rep.
2021;385(3):217–227. 2022;12(1):20545.
14. Hochberg LR, Serruya MD, Friehs GM, Mukand JA, 28. Miao Y, Chen S, Zhang X, Jin J, Xu R, Daly I, Jia J, Wang X,
Saleh M, Caplan AH, Branner A, Chen D, Penn RD, Cichocki A, Jung TP. BCI-based rehabilitation on the stroke
Donoghue JP. Neuronal ensemble control of prosthetic in sequela stage. Neural Plast. 2020;2020:8882764.
devices by a human with tetraplegia. Nature. 29. Lu RR, Zheng MX, Li J, Gao TH, Hua XY, Liu G, Huang SH,
2006;442(7099):164–171. Xu JG, Wu Y. Motor imagery based brain-computer
15. Willett FR, Avansino DT, Hochberg LR, Henderson JM, interface control of continuous passive motion for wrist
Shenoy KV. High-performance brain-to-text communication extension recovery in chronic stroke patients. Neurosci Lett.
via handwriting. Nature. 2021;593(7858):249–254. 2020;718:Article 134727.
16. Simeral JD, Hosman T, Saab J, Flesher SN, Vilela M, 30. Sebastián-Romagosa M, Cho W, Ortner R, Murovec N,
Franco B, Kelemen JN, Brandman DM, Ciancibello JG, Von Oertzen T, Kamada K, Allison BZ, Guger C. Brain
Rezaii PG, et al. Home use of a percutaneous wireless computer interface treatment for motor rehabilitation of
intracortical brain-computer interface by individuals with upper extremity of stroke patients—A feasibility study. Front
tetraplegia. IEEE Trans Biomed Eng. 2021;68(7):2313–2325. Neurosci. 2020;14:Article 591435.
17. Teplan M. Fundamentals of EEG measurement. Sci Rev. 31. Acharya UR, Oh SL, Hagiwara Y, Tan JH, Adeli H, Subha DP.
2002;2(2):1–11. Automated EEG-based screening of depression using deep
18. Herron JA, Thompson MC, Brown T, Chizeck HJ, convolutional neural network. Comput Methods Prog Biomed.
Ojemann JG, Ko AL. Chronic electrocorticography for 2018;161:103–113.

Zhang et al. 2023 | https://doi.org/10.34133/hds.0096 12

 32. Amaral C, Mouga S, Simões M, Pereira HC, Bernardino I, 48. Bousseta R, El Ouakouak I, Gharbi M, Regragui F. EEG
Quental H, Playle R, Mcnamara R, Oliveira G, based brain computer interface for controlling a robot arm
Castelo-Branco M. A feasibility clinical trial to improve social movement through thought. IRBM. 2018;39(2):129–135.
attention in autistic spectrum disorder (ASD) using a brain 49. Son JE, Choi H, Lim H, Ku J. Development of a flickering
computer interface. Front Neurosci. 2018;12:477. action video based steady state visual evoked potential
33. Runnova A, Selskii A, Kiselev A, Shamionov R, Parsamyan R, triggered brain computer interface-functional electrical
Zhuravlev M. Changes in EEG alpha activity during attention stimulation for a rehabilitative action observation game.
control in patients: Association with sleep disorders. J Person Technol Health Care. 2020;28(S1):509–519.
Med. 2021;11(7):601. 50. Eldeeb S, Susam BT, Akcakaya M, Conner CM, White SW,
34. Lin JS, Hsieh CH. A wireless BCI-controlled integration Mazefsky CA. Trial by trial EEG based BCI for distress versus
system in smart living space for patients. Wirel Pers Commun. non distress classification in individuals with ASD. Sci Rep.
2016;88(2):395–412. 2021;11(1):6000.
35. Wang F, Xu Z, Zhang W, Wu S, Zhang Y, Coleman S. An 51. Mazefsky CA, Herrington J, Siegel M, Scarpa A, Maddox BB,
adaptive control approach for intelligent wheelchair based on Scahill L, White SW. The role of emotion regulation in autism
BCI combining with QoO. Int Joint Conf Neur Netw (IJCNN). spectrum disorder. J Am Acad Child Adolesc Psychiatry.
2020;2020:1–8. 2013;52(7):679–688.
36. Antoniou E, Bozios P, Christou V, Tzimourta KD, 52. Choi H, Lim H, Kim JW, Kang YJ, Ku J. Brain computer
Kalafatakis K, G. Tsipouras M, Giannakeas N and interface-based action observation game enhances
Tzallas AT. EEG-based eye movement recognition using mu suppression in patients with stroke. Electronics.
brain–computer interface and random forests. Sensors. 2019;8(12):1466.
2021;21(7):2339. 53. Kim MG, Lim H, Lee HS, Han IJ, Ku J, Kang YJ. Brain–
37. Vokorokos L, Mados B, Adám N, Baláz A. Data acquisition in computer interface-based action observation combined with
non-invasive brain-computer interface using Emotiv EPOC peripheral electrical stimulation enhances corticospinal

Downloaded from https://spj.science.org on June 22, 2024

neuroheadset. Acta Electrot Inform. 2012;12(1):5–8. excitability in healthy subjects and stroke patients. J Neural
38. Williams NS, Mcarthur GM, Badcock NA. 10 years of EPOC: Eng. 2022;19(3):Article 036039.
A scoping review of Emotiv’s portable EEG device. BioRxiv. 54. Zhang D, Yang B, Gao S, Gu X. Brain-controlled robotic arm
2020. https://doi.org/10.1101/2020.07.14.202085. based on adaptive FBCCA. Int Worksh Human Brain Artif
39. Jure FA, Carrere LC, Gentiletti GG, Tabernig CB. BCI-FES Intell. 2021;1369:102–112.
system for neuro-rehabilitation of stroke patients. J Phys Conf 55. Larocco J, Le MD, Paeng DG. A systemic review of available
Ser. 2016;705(1):Article 012058. low-cost EEG headsets used for drowsiness detection. Front
40. Tabernig CB, Lopez CA, Carrere LC, Spaich EG, Ballario CH. Neuroinform. 2020;14:Article 553352.
Neurorehabilitation therapy of patients with severe stroke 56. Aldridge A, Barnes E, Bethel CL, Carruth DW, Kocturova M,
based on functional electrical stimulation commanded by Pleva M, Juhar J. Accessible electroencephalograms (EEGs):
a brain computer interface. J Rehabil Assist Technol Eng. A comparative review with openbci’s ultracortex mark
2018;5:1–12. IV headset. Paper presented at: 2019 29th International
41. Al-Taleb M, Purcell M, Fraser M, Petric-Gray N, Conference Radioelektronika (RADIOELEKTRONIKA);
Vuckovic A. Home used, patient self-managed, brain- Pardubice, Czech Republic; 2019:1–6.
computer interface for the management of central 57. Torres EP, Torres EA, Hernández-Álvarez M, Yoo SG.
neuropathic pain post spinal cord injury: Usability study. Emotion recognition related to stock trading using machine
J Neuroeng Rehabil. 2019;16(1):128. learning algorithms with feature selection. IEEE Access.
42. Zulauf-Czaja A, Al-Taleb MK, Purcell M, Petric-Gray N, 2020;8:199719–199732.
Cloughley J, Vuckovic A. On the way home: A BCI-FES 58. Jiang L, Siriaraya P, Choi D, Kuwahara N. Emotion
hand therapy self-managed by sub-acute SCI participants recognition using electroencephalography signals of
and their caregivers: A usability study. J Neuroeng Rehabil. older people for reminiscence therapy. Front Physiol.
2021;18(1):44. 2022;12:Article 823013.
43. Swee SK, Kiang KDT, You LZ. EEG controlled wheelchair. 59. Suhaimi NS, Mountstephens J, Teo J. EEG-based emotion
MATEC Web Conf. 2016;51:02011. recognition: A state-of-the-art review of current trends and
44. Swee SK, You LZ. Fast fourier analysis and EEG classification opportunities. Comput Intell Neurosci. 2020;2020:8875426.
brainwave controlled wheelchair. Paper presented at: 2016 60. Lim ZY, Quan NY. Convolutional neural network based
2nd International Conference on Control Science and electroencephalogram controlled robotic arm. Paper
Systems Engineering (ICCSSE); Singapore; 2016:20–23. presented at: 2021 IEEE International Conference on
45. Voznenko TI, Chepin EV, Urvanov GA. The control system Automatic Control & Intelligent Systems (I2CACIS); Shah
based on extended BCI for a robotic wheelchair. Proced Alam, Malaysia; 2021:26–31.
Comput Sci. 2018;123:522–527. 61. Saragih AS, Basyiri HN, Raihan MY. Analysis of motor
46. Zgallai W, Brown JT, Ibrahim A, Mahmood F, Mohammad K, imagery data from EEG device to move prosthetic hands
Khalfan M, Mohammed M, Salem M, Hamood N. Deep by using deep learning classification. AIP Conf Proc.
learning AI application to an EEG driven BCI smart wheelchair. 2022;2537(1):Article 050009.
Adv Sci Eng Technol Int Conf (ASET). 2019;2019:1–5. 62. Bolaños CDS, Rodriguez N, CaP C. Non-invasive control of
47. Shahin MK, Tharwat A, Gaber T, Hassanien AE. A a intelligent room using EEG signals. IFAC PapersOnLine.
wheelchair control system using human-machine interaction: 2021;54(4):25–30.
Single-modal and multimodal approaches. J Intell Syst. 63. Dutta S, Banerjee T, Roy ND, Chowdhury B, Biswas A.
2019;28(1):115–132. Development of a BCI-based gaming application to enhance

Zhang et al. 2023 | https://doi.org/10.34133/hds.0096 13

 cognitive control in psychiatric disorders. Innov Syst Softw 79. Devasia D, Roshini T, Jacob NS, Jose SM, Joseph S. Assistance
Eng. 2021;17(2):99–107. for quadriplegic with BCI enabled wheelchair and IoT.
64. Lin CT, Chuang CH, Cao Z, Singh AK, Hung CS, Yu YH, Paper presented at: 2020 3rd International Conference on
Nascimben M, Liu YT, King JT, Su TP, et al. Forehead EEG Intelligent Sustainable Systems (ICISS); Thoothukudi, India;
in support of future feasible personal healthcare solutions: 2020:1220–1226.
Sleep management, headache prevention, and depression 80. Tiwari PK, Choudhary A, Gupta S, Dhar J, Chanak P.
treatment. IEEE Access. 2017;5:10612–10621. Sensitive brain-computer interface to help manoeuvre a
65. Lin CT, Chiu CY, Singh AK, King JT, Ko LW, Lu YC, miniature wheelchair using electroencephalography. Paper
Wang YK. A wireless multifunctional SSVEP-based brain– presented at: 2020 IEEE International Students’ Conference
computer interface assistive system. IEEE Trans Cogn Dev on Electrical, Electronics and Computer Science (SCEECS);
Syst. 2018;11(3):375–383. Bhopal, India; 2020:1–6.
66. Cao Z, Lin CT, Ding W, Chen MH, Li CT, Su TP. Identifying 81. Girase PD, Deshmukh M. MindWave device wheelchair
ketamine responses in treatment-resistant depression control. Int J Sci Res. 2016;5(6):2172–2176.
using a wearable forehead EEG. IEEE Trans Biomed Eng. 82. Looney D, Park C, Kidmose P, Rank ML, Ungstrup M,
2018;66(6):1668–1679. Rosenkranz K, Mandic DP. An in-the-ear platform for
67. Chiu CY, Singh AK, Wang YK, King JT, Lin CT. A wireless recording electroencephalogram. Ann Int Conf IEEE Eng Med
steady state visually evoked potential-based BCI eating Biol Soc. 2011;2011:6882–6885.
assistive system. Int Joint Conf Neur Netw (IJCNN). 83. Looney D, Kidmose P, Park C, Ungstrup M, Rank ML,
2017;2017:3003–3007. Rosenkranz K, Mandic DP. The in-the-ear recording concept:
68. Tarrant J, Jackson R, Viczko J. A feasibility test of a User-centered and wearable brain monitoring. IEEE Pulse.
brief mobile virtual reality meditation for frontline 2012;3(6):32–42.
healthcare workers in a hospital setting. Front Virtual Real. 84. Looney D, Kidmose P, Morrell MJ, Mandic DP. Ear-EEG:
2022;3:Article 764745. Continuous brain monitoring. Brain-Computer Interface

Downloaded from https://spj.science.org on June 22, 2024

69. Japaridze G, Loeckx D, Buckinx T, Armand Larsen S, Research. Cham, Switzerland: Springer. 2014;63–71.
Proost R, Jansen K, Macmullin P, Paiva N, Kasradze S, 85. Kidmose P, Looney D, Ungstrup M, Rank ML, Mandic DP.
Rotenberg A, et al. Automated detection of absence seizures A study of evoked potentials from ear-EEG. IEEE Trans
using a wearable electroencephalographic device: A phase Biomed Eng. 2013;60(10):2824–2830.
3 validation study and feasibility of automated behavioral 86. Goverdovsky V, Looney D, Kidmose P, Mandic DP.
testing. Epilepsia. 2022;1:1–7. In-ear EEG from viscoelastic generic earpieces: Robust
70. Mercado J, Espinosa-Curiel I, Escobedo L, Tentori M. and unobtrusive 24/7 monitoring. IEEE Sensors J.
Developing and evaluating a BCI video game for neurofeedback 2015;16(1):271–277.
training: The case of autism. Multimed Tools Appl. 87. Zibrandtsen IC, Kidmose P, Christensen CB, Kjaer TW.
2019;78(10):13675–13712. Ear-EEG detects ictal and interictal abnormalities in focal
71. Li M, Liang Z, He B, Zhao CG, Yao W, Xu G, Xie J, Cui L. and generalized epilepsy - a comparison with scalp EEG
Attention-controlled assistive wrist rehabilitation using a low- monitoring. Clin Neurophysiol. 2017;128(12):2454–2461.
cost EEG sensor. IEEE Sensors J. 2019;19(15):6497–6507. 88. Ahn JW, Ku Y, Kim DY, Sohn J, Kim JK, Kim HC. Wearable
72. Ruşanu OA, Cristea L, Luculescu MC. LabVIEW and in-the-ear EEG system for SSVEP-based brain–computer
Android BCI chat app controlled by voluntary eye-blinks interface. Electron Lett. 2018;54(7):413–414.
using NeuroSky MindWave mobile EEG headset. Int Conf 89. Athavipach C, Pan-Ngum S, Israsena P. A wearable
e-Health Bioeng (EHB). 2020;2020:1–4. in-ear EEG device for emotion monitoring. Sensors.
73. Sahu M, Shukla P, Chandel A, Jain S, Verma S. Eye blinking 2019;19(18):4014.
classification through NeuroSky MindWave headset using EegID 90. Kaveh R, Doong J, Zhou A, Schwendeman C, Gopalan K,
tool. Int Conf Innov Comput Commun. 2021;1165:789–799. Burghardt FL, Arias AC, Maharbiz MM, Muller R. Wireless
74. Prem S, Wilson J, Varghese SM, Pradeep M. BCI integrated user-generic ear EEG. IEEE Trans Biomed Circ Syst.
wheelchair controlled via eye blinks and brain waves. 2020;14(4):727–737.
Techno-Societal 2020. Cham, Switzerland: Springer; 91. Debener S, Emkes R, De Vos M, Bleichner M. Unobtrusive
2021:321–331. ambulatory EEG using a smartphone and flexible printed
75. Reyes AF, Camacho EC, Armando M, Calderón JM. LSTM electrodes around the ear. Sci Rep. 2015;5(1):16743.
based brain-machine interface tool for text generation 92. Bleichner MG, Mirkovic B, Debener S. Identifying auditory
through eyes blinking detection. Paper presented at: attention with ear-EEG: cEEGrid versus high-density cap-
2021 IEEE 18th Annual Consumer Communications EEG comparison. J Neural Eng. 2016;13(6):Article 066004.
& Networking Conference (CCNC); Las Vegas, NV, 93. Pacharra M, Debener S, Wascher E. Concealed around-the-
USA;2021:1–6. ear EEG captures cognitive processing in a visual simon task.
76. Mercado J, Escobedo L, Tentori M. A BCI video game using Front Hum Neurosci. 2017;11:290.
neurofeedback improves the attention of children with 94. Bleichner MG, Debener S. Concealed, unobtrusive ear-
autism. J Mult User Interfaces. 2021;15(3):273–281. centered EEG acquisition: cEEGrids for transparent EEG.
77. Salih TA, Abdal YM. Brain computer interface based smart Front Hum Neurosci. 2017;11:163.
keyboard using neurosky mindwave headset. TELKOMNIKA. 95. Mirkovic B, Bleichner MG, De Vos M, Debener S. Target
2020;18(2):919–927. speaker detection with concealed EEG around the ear. Front
78. Permana K, Wijaya S, Prajitno P. Controlled wheelchair Neurosci. 2016;10:349.
based on brain computer interface using Neurosky Mindwave 96. Nogueira W, Dolhopiatenko H, Schierholz I, Büchner A,
Mobile 2. AIP Conf Proc. 2019;2168(1):Article 020022. Mirkovic B, Bleichner MG, Debener S. Decoding selective

Zhang et al. 2023 | https://doi.org/10.34133/hds.0096 14

 attention in normal hearing listeners and bilateral cochlear soft scalp electronics and virtual reality system for
implant users with concealed ear EEG. Front Neurosci. motor imagery-based brain-machine interfaces. Adv Sci.
2019;13:720. 2021;8(19):2101129.
97. Millard SK, Furman AJ, Kerr A, Seminowicz DA, Gao F, 111. Hua H, Tang W, Xu X, Feng DD, Shu L. Flexible multi-layer
Naidu BV, Mazaheri A. Predicting postoperative pain in lung semi-dry electrode for scalp EEG measurements at hairy
cancer patients using preoperative peak alpha frequency. Br J sites. Micromachines. 2019;10(8):518.
Anaesth. 2022;128(6):e346–e348. 112. Li G, Wang S, Li M, Duan YY. Towards real-life EEG
98. Segaert K, Poulisse C, Markiewicz R, Wheeldon L, applications: Novel superporous hydrogel-based semi-dry
Marchment D, Adler Z, Howett D, Chan D, Mazaheri A. EEG electrodes enabling automatically ‘charge–discharge’
Detecting impaired language processing in patients with electrolyte. J Neural Eng. 2021;18(4):Article 046016.
mild cognitive impairment using around-the-ear cEEgrid 113. Li Z, Guo W, Huang Y, Zhu K, Yi H, Wu H. On-skin graphene
electrodes. Psychophysiology. 2022;59(5):Article e13964. electrodes for large area electrophysiological monitoring and
99. Frankel MA, Lehmkuhle MJ, Watson M, Fetrow K, Frey L, human-machine interfaces. Carbon. 2020;164:164–170.
Drees C, Spitz MC. Electrographic seizure monitoring with a 114. Blum S, Emkes R, Minow F, Anlauff J, Finke A, Debener S.
novel, wireless, single-channel EEG sensor. Clin Neurophysiol Flex-printed forehead EEG sensors (fEEGrid) for long-term
Pract. 2021;6:172–178. EEG acquisition. J Neural Eng. 2020;17(3):Article 034003.
100. Swinnen L, Chatzichristos C, Jansen K, Lagae L, Depondt C, 115. Li J, Ma Y, Huang D, Wang Z, Zhang Z, Ren Y, Hong
Seynaeve L, Vancaester E, Dycke AV, Macea J, Vandecasteele K, M, Chen Y, Li T, Shi X, et al. High-performance flexible
et al. Accurate detection of typical absence seizures in adults and microneedle array as a low-impedance surface biopotential
children using a two-channel electroencephalographic wearable dry electrode for wearable electrophysiological recording and
behind the ears. Epilepsia. 2021;62(11):2741–2752. polysomnography. Nano-Micro Letters. 2022;14(1):132.
101. Kaongoen N, Choi J, Jo S. A novel online BCI system using 116. Valentin O, Viallet G, Delnavaz A, Cretot-Richert G,
speech imagery and ear-EEG for home appliances control. Ducharme M, Monsarat-Chanon H, Voix J. Custom-fitted

Downloaded from https://spj.science.org on June 22, 2024

Comput Methods Prog Biomed. 2022;224:Article 107022. in-and around-the-ear sensors for unobtrusive and on-the-
102. Hsieh J-C, Alawieh H, Li Y, Iwane F, Zhao L, Anderson R, go EEG acquisitions: Development and validation. Sensors.
Abdullah SI, Tang KWK, Wang W, Pyatnitskiy I, et al. A 2021;21(9):2953.
highly stable electrode with low electrode-skin impedance 117. Das A, Basu S, Adarsh A, Gubbi J, Muralidharan K, Meghana S,
for wearable brain-computer interface. Biosens Bioelectron. Mahendiran S, Biradar A, Pradhan U, Chakravarty T, et al.
2022;218:Article 114756. Surface potential simulation and electrode design for in-ear
103. Shen G, Gao K, Zhao N, Yi Z, Jiang C, Yang B, Liu J. A novel EEG measurement. Paper presented at: 2022 44th Annual
flexible hydrogel electrode with a strong moisturizing ability International Conference of the IEEE Engineering in Medicine &
for long-term EEG recording. J Neural Eng. 2021;18(6): Biology Society (EMBC); Glasgow, Scotland, UK; 2022:937–940.
Article 066047. 118. Jiang X, Bian GB, Tian Z. Removal of artifacts from EEG
104. Alba NA, Sclabassi RJ, Sun M, Cui XT. Novel hydrogel-based signals: A review. Sensors. 2019;19(5):987.
preparation-free EEG electrode. IEEE Trans Neural Syst 119. Lee YE, Kwak NS, Lee SW. A real-time movement artifact
Rehabil Eng. 2010;18(4):415–423. removal method for ambulatory brain-computer interfaces.
105. Lo CC, Chien TY, Chen YC, Tsai SH, Fang WC, Lin BS. A IEEE Trans Neural Syst Rehabil Eng. 2020;28(12):2660–2670.
wearable channel selection-based brain-computer interface 120. Chang CY, Hsu SH, Pion-Tonachini L, Jung TP. Evaluation
for motor imagery detection. Sensors. 2016;16(2):213. of artifact subspace reconstruction for automatic artifact
106. Lin BS, Pan JS, Chu TY, Lin BS. Development of a wearable components removal in multi-channel EEG recordings. IEEE
motor-imagery-based brain–computer interface. J Med Syst. Trans Biomed Eng. 2019;67(4):1114–1121.
2016;40(3):71. 121. Egambaram A, Badruddin N, Asirvadam VS, Begum T,
107. Lin BS, Lin BS, Yen TH, Hsu CC, Wang YC. Design Fauvet E, Stolz C. Online detection and removal of eye blink
of wearable headset with steady state visually evoked artifacts from electroencephalogram. Biomed Signal Proc
potential-based brain computer interface. Micromachines. Control. 2021;69:Article 102887.
2019;10(10):681. 122. Qiu F, Zhang H, Fan X, Yang Z. A FPGA-based integrated
108. Pourahmad A, Mahnam A. Evaluation of a low-cost and DAQ system towards to BCI signal transmission. IEEE Int
low-noise active dry electrode for long-term biopotential Conf Cyborg Bionic Syst (CBS). 2022;2023:311–314.
recording. J Med Signal Sensors. 2016;6(4):197–202. 123. Wang YT, Nakanishi M, Wang Y, Wei CS, Cheng CK,
109. Li J, Zhang Z, Wang Z, Ren Y, Huang D, Wang Q, Li Z. Jung TP. An online brain-computer interface based on
Annular micro-needle array as a minimally invasive flexible SSVEPs measured from non-hair-bearing areas. IEEE Trans
dry electrode for on-Hair EEG recording. Paper presented Neural Syst Rehabil Eng. 2016;25(1):14–21.
at: 2021 IEEE 34th International Conference on Micro 124. Knierim MT, Berger C, Reali P. Open-source concealed
Electro Mechanical Systems (MEMS); Gainesville, FL, USA; EEG data collection for brain-computer-interfaces-neural
2021:270–273. observation through OpenBCI amplifiers with around-
110. Mahmood M, Kwon S, Kim H, Kim YS, Siriaraya P, Choi J, the-ear cEEGrid electrodes. Brain-Comput Interfaces.
Otkhmezuri B, Kang K, Yu KJ, Jang YC, et al. Wireless 2021;8(4):161–179.

Zhang et al. 2023 | https://doi.org/10.34133/hds.0096 15