Brain Computer Interfacing-Applications

Egyptian Informatics Journal (2015) 16, 213–230
Cairo University
Egyptian Informatics Journal

www.elsevier.com/locate/eij
www.sciencedirect.com
REVIEW
Brain computer interfacing: Applications

and challenges
Sarah N. Abdulkader *, Ayman Atia, Mostafa-Sami M. Mostafa
HCI-LAB, Department of Computer Science, Faculty of Computers and Information, Helwan University, Cairo, Egypt
Received 20 December 2014; accepted 4 June 2015

Available online 6 July 2015
KEYWORDS Abstract Brain computer interface technology represents a highly growing field of research with
Brain Computer Interfaces; application systems. Its contributions in medical fields range from prevention to neuronal rehabil-
Brain signal acquisition; itation for serious injuries. Mind reading and remote communication have their unique fingerprint
BCI applications; in numerous fields such as educational, self-regulation, production, marketing, security as well as
Mind commands; games and entertainment. It creates a mutual understanding between users and the surrounding sys-
Brain monitoring; tems. This paper shows the application areas that could benefit from brain waves in facilitating or
BCI challenges achieving their goals. We also discuss major usability and technical challenges that face brain sig-
nals utilization in various components of BCI system. Different solutions that aim to limit and
decrease their effects have also been reviewed.
Ó 2015 Production and hosting by Elsevier B.V. on behalf of Faculty of Computers and Information,
Cairo University. This is an open access article under the CC BY-NC-ND license (http://creativecommons.
org/licenses/by-nc-nd/4.0/).
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
2. BCI functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
2.1. Communication and control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
2.2. User state monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
3. BCI applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
3.1. Medical applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
* Corresponding author.
E-mail addresses: [email protected] (S.N. Abdulkader),
[email protected] (A. Atia), [email protected]
(M.-S.M. Mostafa).
Peer review under responsibility of Faculty of Computers and
Information, Cairo University.
Production and hosting by Elsevier
http://dx.doi.org/10.1016/j.eij.2015.06.002
1110-8665 Ó 2015 Production and hosting by Elsevier B.V. on behalf of Faculty of Computers and Information, Cairo University.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
214 S.N. Abdulkader et al.
3.1.1. Prevention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215

3.1.2. Detection and diagnosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
3.1.3. Rehabilitation and restoration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
3.2. Neuroergonomics and smart environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
3.3. Neuromarketing and advertisement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
3.4. Educational and self-regulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
3.5. Games and entertainment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
3.6. Security and authentication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
4. BCI system components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
5. Signal acquisition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
5.1. Invasive techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
5.1.1. Intracortical. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
5.1.2. Cortical surface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
5.2. Noninvasive techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
5.2.1. Magnetoencephalography (MEG) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
5.2.2. Functional magnetic resonance imaging (fMRI) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
5.2.3. Functional near-infrared spectroscopy (fNIRS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
5.2.4. Electroencephalogram (EEG) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
6. BCI electrical signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
6.1. Evoked potential or evoked response (EP) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
6.2. Event related desynchronizationnsynchronization (ERDnERS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
7. Challenges and proposed solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
7.1. Challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
7.1.1. Usability challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
7.1.2. Technical challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
7.2. Proposed solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
7.2.1. Noise removal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
7.2.2. Separability of multiple classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
8. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
1. Introduction On the other hand, some of BCI advantages for able-

bodied users have been enlightened in [5]. BCI could be helpful
Brain Computer Interface (BCI) technology is a powerful especially for safety applications or applications where it is
communication tool between users and systems. It does not instantaneously difficult to move and the response time is cru-
require any external devices or muscle intervention to issue cial. Besides they can also be used to increase the accuracy of
commands and complete the interaction [1]. The research com- the HCI systems, resulting in BCI contribution in various fields
munity has initially developed BCIs with biomedical applica- such as industry, educational, advertising, entertainment, and
tions in mind, leading to the generation of assistive devices smart transportation. Despite its expected success, Brain com-
[2]. They have facilitated restoring the movement ability for puter interfacing needs to overcome technical difficulties as
physically challenged or locked-in users and replacing lost well as challenges posed by user acceptance to deal with such
motor functionality [3]. The promising future predicted for newly discovered technology.
BCI has encouraged research community to study the involve- The next sections will provide more information about BCI
ment of BCI in the life of non-paralyzed humans through functions and associated applications. Various methods for
medical applications. acquiring brain signals are then explored along with the electri-
However, the scope of research has been further widened to cal changes reflected in the recorded brain waves. This paper
include non-medical applications. More recent studies have will also discuss the issues facing BCI systems and some found
targeted normal individuals by exploring the use of BCIs as solutions to their consequences in details.
a novel input device and investigating the generation of
hands-free applications [1,2]. The use of BCI interfaces for 2. BCI functions
healthy users has been subject to some doubts as discussed
in [4]. The problem of poor information transfer rate (ITR) Applications of Brain Computer Interface base its func-
of BCIs and its effect on reducing the commands user can give tionality on either observing the user state or allowing
has been addressed as one of those issues. It has been claimed the user to deliver hisnher ideas. BCI system records the
that this problem restricts BCI utilization for locked-in persons brain waves and sends them to the computer system to
as it will not be able to keep up with ordinary communication complete the intended task. The transmitted waves are
ways or even existing human computer interfaces. therefor used to express an idea or control an object.
Brain computer interfacing 215
Figure 1 BCI application fields.
Garcia-Molina et al. [13], the precise awareness of the current

emotional or cognitive state can affect the recognition of the
mental task associated with the recorded brain waves.
Another beneficial employment of such information is to
determine the state itself and use that knowledge for enhancing
various BCI systems. BCI User state monitoring function is
considered a helpful hand in Human Computer Interfaces
and adapts them according to the estimated user emotional
or cognitive state [14,13]. It participates in a shared control
environment and decides the best type of control that might
be used in certain situations.
It also contributes in the development of smart environ-
ments and emotion controlling applications [4,15]. Working
conditions’ assessment and educational methods’ evaluation
are examples of other fields that could benefit from measuring
user’s brain state [16,17]. The next section highlights some
Figure 2 Usage of BCI in medical field phases. applications that exploit brain computer interface.
The following subsections give a brief introduction to

those BCI operations. 3. BCI applications
2.1. Communication and control Brain computer interfaces have contributed in various fields of
research. As briefed in Fig. 1, they are involved in medical,
Brain computer interface (BCI) systems build a communica- neuroergonomics and smart environment, neuromarketing
tion bridge between human brain and the external world elim- and advertisement, educational and self-regulation, games
inating the need for typical information delivery methods. and entertainment, and Security and authentication fields.
They manage the sending of messages from human brains
and decoding their silent thoughts. Thus they can help handi- 3.1. Medical applications
capped people to tell and write down their opinions and ideas
via variety of methods such as in spelling applications [6], Healthcare field has a variety of applications that could take
semantic categorization [7], or silent speech communication advantage of brain signals in all associated phases including
[8]. prevention, detection, diagnosis, rehabilitation and restoration
BCIs can also facilitate hands-free applications bringing the as shown in Fig. 2.
ease and comfort to human beings through mind-controlling
of machines. They only require incorporating brain signals in 3.1.1. Prevention
order to accomplish a set of commands and no muscles inter- Various consciousness level determination systems along with
vention is needed [9,10,3]. BCI assistive robots can offer sup- their brain-related studies have been developed. The attentive-
port for disabled users in daily and professional life, ness influences of smoking and alcohol on brain waves have
increasing their cooperation in building their community [11]. been enlightened in [18–22]. The importance of such studies
for medical prevention lies in the possible loss of function
2.2. User state monitoring and decrease of alertness level resulting from smoking andnor
alcohol drinking, while the authors of [23] have investigated
Early BCI applications have targeted disabled users who have the most responding brain parts to alcoholism.
mobility or speaking issues. Their aim was to provide an alterna- Traffic accidents are considered the main cause for death or
tive communication channel for those users. But later on, BCI some serious injuries as claimed in [24,25]. Analyzing their
enters the world of healthy people as well. It works as a physio- causes for later prevention has been a concern for researches
logical measuring tool that retrieves and uses information about in various fields. Thus concentration level for those suffer from
an individual’s emotional, cognitive or affectiveness state. The motion sickness, especially drivers, has been studied. Motion
target of brain signals utilization has been extended beyond sickness, which occurs as a result of sending conflicted sensory
controlling some object or offering a substitution for specific information generated from body, inner ear and eye to the
functions, in what is called passive BCI [12]. According to brain, is usually happening on moving transportation media.
It can cause traffic accidents as it declines in a person’s ability body can become paralyzed. Disabilities and brain strokes
to maintain self-control. And according to [26,27], a prediction have been subject for many studies interested in solutions
of motion sickness could contribute in a driver-state monitor- involving brain signals. It has been pointed out in [42] that
ing and alertness system using a set of EEG power indicators. brain structures associated with stroke injuries could be reor-
Its accompanying EEG signals from different five brain ganized and the damaged motor functions could be restored
regions have been examined in [26]. The human hearing level, via neuroplasticity [43,44].
as part of sensory information gathering process, has been Mobile robots can be used to help locked-in people com-
measured via auditory evoked potential BCI-based system in pleting daily life activities as discussed in [3,45]. For patients
[28]. In another study [29], a virtual reality-based motion- who cannot recover previous levels of mobility or communi-
sickness platform has been designed with a 32-channel EEG cation, BCI based prosthetic limbs, also called neuropros-
system and a joystick which is used to report the motion sick- thetic devices, can be used to regain normal functionality
ness level (MSL) in real time experiments. Consciousness level [46–49].
monitoring via brain waves has been expanded to include not Various reality approaches for BCI-based rehabilitation
only drivers but also stayed-alone sick people as suggested in training such as real, virtual, and augmented approaches have
[30]. been presented. Real rehabilitation approach exploits brain
signals generated from healthy people along with the decoded
3.1.2. Detection and diagnosis kinematic parameters [50]. It assists stroke patients modifying
Mental state monitoring function of BCI systems has also con- their thinking behavior to resemble the recorded signals and
tributed in forecasting and detecting health issues such as retraining healthy areas of the brain to take over. Another
abnormal brain structure (such as brain tumor), Seizure disor- approach for rehabilitation involves virtual reality through
der (such as epilepsy), Sleep disorder (such as narcolepsy), and monitoring and controlling avatar movement generated from
brain swelling (such as encephalitis). Tumor, which is gener- the outgoing brain waves [51]. Augmented reality represents
ated from uncontrolled self-dividing of cells, could be discov- the third approach in the reality based BCI treatment such
ered using EEG as a cheap secondary alternative for MRI as augmented mirror box system which appears as a develop-
and CT-SCAN. EEG-based Brain tumors detection systems ment for Mirror Box Therapy (MBT). MBT uses brain signals
have been the main subject of the researches in [31,32], while generated from symmetrical movements that incorporate
[33] has been concerned with identifying breast cancer using injured and healthy limbs [52].
EEG signals. Motor imagery signals also contribute in neurofeedback
Sharanreddy and Kulkarni have proposed a system in [34] systems for poststroke motor therapy [53,54]. Classification
that recognizes EEG abnormalities associated with brain of and Comparing the results of motor imageries and actions
tumors and epilepsy seizures. Early detection of epilepsy sei- are shown in [55,56].
zure, one of the most common neurological disorders, and
controlling its effects are presented in [35,36]. 3.2. Neuroergonomics and smart environment
Dyslexia, one of the brain disorders, can be diagnosed by
measuring brain behavior as described in [37]. It influences As previously mentioned, deploying brain signals is not exclu-
the reading and learning ability making its discovery at an sive to the medical field. Smart environments such as smart
early stage saves the children from self-esteem and self- houses, workplaces or transportations could also exploit brain
confidence issues and allows them to gain their basic skills computer interfaces in offering further safety, luxury and phys-
and knowledge. Sleep disorders can be detected with BCI assis- iological control to humans’ daily life. They are also expected
tance as well as claimed in [38,39]. They demonstrate some to witness cooperation between Internet Of Things (IOT) and
methods for deploying EEG signals in noticing Idiopathic BCI technologies as stated in [57].
rapid eye-movement (REM) sleep behavior disorder (iRBD) Lin et al. [58,59] have proposed a cognitive controller sys-
as iRBD has been found to be a strong early predictor for tem called Brain computer interface-based Smart Living
Parkinson’s disease (PD). Environmental Auto-adjustment Control System
Wei et al. [40] have experimentally confirmed the relation- (BSLEACS). It monitors user’s mental state and adapts the
ship between human gait cycle and EEG signals through the surrounding components accordingly. It has extended its func-
use of plantar pressure measuring system. This relationship tionality with the involvement of universal plug and play
helps predicting diseases such as dyskinesia, peripheral neu- (UPnP) home networking. On the other hand, the surrounding
ropathy, and musculoskeletal disease. environmental contribution in enhancing BCI based home
applications via context awareness has been considered.
3.1.3. Rehabilitation and restoration Navarro et al. [4] have developed such an application that
Mobility rehabilitation is a form of physical rehabilitation automatically changes the available options accessible by the
used with patients who have mobility issues, to restore their user according to the current context. Also integration of both
lost functions and regain previous levels of mobility or at least healthcare and smart house in gaining non-intrusive mental
help them adapt to their acquired disabilities [41]. People suffer health care has been an existing approach in brain computer
from serious injuries or events such as strokes may also be able applications as shown in [60].
to fully recover. Brain signals also assist in improving workplace conditions
Stroke is a condition in which the brain cells suddenly die by assessment of an operator’s cognitive state [16]. They also
because of the lack of oxygen. It can be caused by an obstruc- analyze the impact of workload mental fatigue and task time
tion in the blood flow. The patient may suddenly lose the abil- on EEG features [61]. Operating room as well represents a can-
ity to speak, there may be memory problems, or one side of the didate place for smart workplace BCI-based application as in
[62]. The system measures the stress level of a surgeon and alert 3.4. Educational and self-regulation
according to the response type.
The field of intelligent transportation has also been benefit- Neurofeedback is a promising approach for enhancing brain
ted from the cognitive state monitoring BCI function. Driver’s performance via targeting human brain activity modulation.
behavior has been studied in numerous studies. It has been It invades the educational systems, which utilizes brain elec-
found that distraction and fatigue are two main sources for trical signals to determine the degree of clearness of studied
driver’s inattention, which is considered as a strong cause information. Personalized interaction to each learner is
for most traffic accidents [63]. Various types of measures established according to the resultant response experienced
have contributed in determining the driver’s cognitive state [17].
[64–66]. Uses of EEG signals for fatigue detection have been Learning to self-regulate through noninvasive BCI has also
widely studied in [67], while [68] has discussed the utilization been studied. It provides a mean for improving cognitive thera-
of workload index to assess the driver’s mental state. Several peutic approaches. The research in [78] has analyzed the feasibil-
models for distinguishing distracted drivers have been exam- ity fMRI for the emotional regulation, while [79] has suggested
ined in [69]. Kim et al. [70] have presented multimodal context the use of hybrid rtfMRI–EEG BCI to fight the depression feel-
recognition for smart driving system to predict concentration ing as well as other neuropsychiatric disorders through training
and stress by analyzing both ECG and EEG signals and con- sessions. Furthermore, EEG based emotional intelligence has
trolling car speed by concentration value of brain signals. been applied in sport competitions to control the accompanying
Alcoholic drivers, as a contributor to road accidents, could stress as examined in [80]. In [43], BCI technology has been elab-
also be characterized through the use of EEG signals as men- orated in self-regulation and skill learning via functional
tioned in [71]. [72] has developed an audio-visual virtual envi- Magnetic Resonance Imaging (fMRI) neurofeedback.
ronment in order to evaluate and analyze the driving responses
along with the associated brain signals. [73] has suggested 3.5. Games and entertainment
some driving specific tasks to the simulated driving model
and explore the neural dynamics generated, while in [74],
Entertainment and gaming applications have opened the mar-
kawamura et al. have described the use of multiple stimulation
ket for nonmedical brain computer interfaces. Various games
methods when dealing with drowsy drivers to increase their
are presented like in [81] where helicopters are made to fly to
attention level. [24] has investigated the feasibility of using dri-
any point in either a 2D or 3D virtual world.
ver’s EEG signals to detect emergency conditions such as the
Combining the features of existing games with brain con-
sudden appearance of a pedestrian.
trolling capabilities has been subject to many researches such
as [82] which tend to provide a multi-brain entertainment expe-
3.3. Neuromarketing and advertisement rience. The video game is called BrainArena. The players can
join a collaborative or competitive football game by means
Marketing field has also been an interest for BCI researches. of two BCIs. They can score goals by imagining left or right
The research in [75] has explained the benefits of using EEG hand movements.
evaluation for TV advertisements related to both commercial On the other hand, some EEG serious games have been
and political fields. BCI based assessment measures the gener- employed for emotional control and/or neuroprosthetic reha-
ated attention accompanying watching activity [76]. On the bilitation. They are containing either a new game idea or a
other hand, the researchers of [77] have considered the impact modified one. In [5], Tan and Nijholt have described
of another cognitive function in neuromarketing field. They Brainball game which intends to drop the stress level. The
have been interested in estimating the memorization of TV users can only move the ball by relaxing; thus, the calmer
advertisements thus providing another method for advertising player is more likely to be the winner and thus they would
evaluation. learn to control their stress while being amused.
Figure 3 BCI system.

Figure 4 Signal acquisition methods.
Figure 5 Examples of different types of recorded signals: (a) blood-oxygen-level-dependent (BOLD) change, (b) magnetic signals and (c)
electrical signals.
3.6. Security and authentication biometric modality, could be used to send covert warning
when the authorized user is under external forcing conditions,
Security systems involve knowledge based, object based and/or as implemented in [87].
biometrics based authentication. They have shown to be vul- Several researches have considered authenticating the EEG
nerable to several drawbacks such as simple insecure pass- signal generated from driving behavior as part of smart driving
word, shoulder surfing, theft crime, and cancelable systems. In [88,89], the authors have used a simplified driving
biometrics [83]. Cognitive Biometrics or electrophysiology, simulator with mental-tasked condition to verify driver’s iden-
where only modalities using biosignals (such as brain signals) tity on demand. Unconscious driver authentication has taken
are used as sources of identity information, gives a solution place in [90].
for those vulnerabilities [84,85]. The motivation behind explor-
ing the feasibility of electrophysiology is that biosignals cannot 4. BCI system components
be casually acquired by external observers. They also can be of
great value for disabled patients or users missing the associated As shown in Fig. 3, BCI system consists of four basic compo-
physical trait [86]. This makes such signals difficult to synthe- nents. They include signal acquisition, signal preprocessing,
size and therefore improves the resistance of biometric systems feature extraction, and classification. Signal acquisition com-
to spoofing attacks. Besides electroencephalogram (EEG), as a ponent, described in details in the following section, is
Figure 6 Intracortical acquistion.
responsible for recording the brain waves and sending them to BCI application and the category of its intended users that
the preprocessing component for signal enhancement and decides the proper signal acquisition method and its measured
noise reduction. Feature extraction component generates the phenomena.
discriminative characteristics for the improved signal, decreas- As shown in Fig. 4, there are two general classes of brain
ing the size of the data applied to the classification component. acquisition methods: invasive and non-invasive methods. In
Classifiers are translating the produced features into device invasive technology, electrodes are neurosurgically implanted
commands [3,91]. either inside the user’s brain or over the surface of the brain,
while in non-invasive technologies, the brain activity is mea-
5. Signal acquisition sured using external sensors [5]. Examples of acquired signals
are presented in Fig. 5. Fig. 5.a [92] presents the blood-
Measuring brain generated oscillations is one of the main com- oxygen-level-dependent (BOLD) changes. Fig. 5.b [93], shows
ponents in any BCI based system. It reflects the voluntary neu- magnetic signals generated from the brain while electrical
ral actions generated by user’s current activity. Various brain signals are recorded in Fig. 5.c.
methods for signal acquisition have been studied. It is the
5.1. Invasive techniques
Invasive recording methods implant electrodes under the scalp.

They measure the neural activity of the brain either intracorti-
cally from within the motor cortex or on the cortical surface
(electrocorticography (ECoG)). Their greatest advantage is
that they provide high temporal and spatial resolution, increas-
ing the quality of the obtained signal and its signal to noise
ratio.
However, these techniques suffer from a lot of issues. Aside
from Usability issues rising from the involvement of surgical
procedure, problems related to the system’s output have
occurred. The small size of the monitored brain regions by
those implants is considered one of them. Once implanted,
Figure 7 ECoG acquisition. they cannot be shifted to measure brain activity in another
area. Besides, the body adaptation to the new object, which

may fail, can cause medical complications. Problems regarding
the stability of implants and protection from infection can
arise as well. Thus the usage of invasive recording in real world
has been usually restricted to the BCI based medical applica-
tions for a few disabled users [94,5].
According to [95], the invasive systems have mostly been
tried in BCI systems’ experiments that use monkeys. A few
patients with tetraplegia have used implanted electrodes. The
next subsection provides more details about these invasive
methods.
5.1.1. Intracortical
Intracortical acquisition technique represents the most inva- Figure 9 fMRI acquisition.
sive method shown in Fig. 6 [96]. It is planted under the cortex
surface of the brain. It can be achieved using single electrode,
or array of electrodes that measure the action signals out of movement prediction algorithm. Monkeys have been used to
individual neurons. Electrode tips are placed very close to move a brain-controlled robot arm in virtual reality.
the signal source and the arrays have to be stable over a long Researchers have also succeeded in assisting them to eat with
period of time. Due to its relatively high spatial resolution, its a real robot arm [94].
usage in source localization problem is extensively recom- Studies using invasive recording techniques with human
mended. But intracortical acquisition could encounter long subjects have been limited to some severely disabled peoples.
term signal variability. This could happen as a result of neu- Amyotrophic lateral sclerosis (ALS) is a progressive neurode-
ronal cell death or increased tissue resistance. Besides, if the generative disease that affects nerve cells in the brain and the
system involves a stimulus to activate the disabled limb, this spinal cord. Motor neurons reach from the brain to the spinal
additional stimulus might also generate a significant noise cord and from the spinal cord to the muscles throughout the
effect [95]. body. An ALS patient has been able to move a cursor on a
Monkeys and rats have been involved deeply in BCI computer screen to select presented items after implanting a
research studies that employ intracortical invasive acquisition. single electrode into the motor cortex [94]. Another study in
Movement has been analyzed with animals using implanted [97] has aimed to show that the classification accuracy
electrodes. Monkeys have learnt first to move a cursor into increases with the gradual rising of the number of electrodes.
eight targets located at the corners of an imaginary cube in a On the other hand, several researches concerned with reducing
research aiming to minimize the number of used electrodes. the number of electrodes to decrease the features’ size or
The extracted information has helped in the estimation of enhance user’s acceptability have taken place in [98–100].
the movement intention and has been used to train an adaptive
5.1.2. Cortical surface
Electrocorticography (ECoG) is a recording method that
brings a less invasive option while at the same time preserves
the advantages of invasive approach. It involves implanting
electrode grids or strips over the cortex surface through a sur-
gical operation shown in Fig. 7 [101]. It records the electrical
activity of neurons at the embracing area [102]. [94] has consid-
ered the number of electrodes as measurement for invasiveness
degree.
Figure 8 MEG acquisition. Figure 10 fNIRS acquisition.

skull layer compared to electric fields. But this advantage does

not lead to huge improvement either in performance or in train-
ing times over noninvasive electronic acquiring techniques [95].
5.2.2. Functional magnetic resonance imaging (fMRI)

fMRI detects the changes in blood flow which are related to
neural activity in the brain using the device shown in Fig. 9
[108]. Thus it helps mapping activities to the corresponding
used brain areas which is known as source localization prob-
Figure 11 EEG acquisition. lem. It depends on the fact that any usage of brain part
requires the increase of incoming blood flow. It uses blood-
ECoG recording is located in the middle between invasive- oxygen-level-dependent (BOLD) contrast, which is sensitive
ness accuracy and the safety of non-invasiveness. As a result of to the hemodynamic response [109]. The intensities of BOLD
its relative closure to signal sources, it offers a higher spatial contrast reflect the changes in the deoxyhemoglobin concentra-
resolution and signal amplitude than those provided by non- tion in the brain tissue. Although fMRI temporal resolution is
invasive techniques such as EEG. Proposing a better amplitude low, it provides a high spatial resolution and captures informa-
signal makes it less affected by the noise and artifacts gener- tion from deep parts of the brain that cannot be gathered by
ated from muscle engagement. These advantages make electrical or magnetic measuring [95].
ECoG a good candidate solution for seizure localization prob-
lem. Thus it has been used by epilepsy patients before surgery 5.2.3. Functional near-infrared spectroscopy (fNIRS)
[103]. fNIRS is a noninvasive technique that measures blood
Researchers have exploited the over-cortex recording in dynamic in the brain in order to detect the neuronal activity.
multiple studies concerned with motor tasks and speech pro- It uses light in the near-infrared range to determine the blood
cess classification. [104] has used ECoG signals for the predic- flow [110]. It has the advantage of providing high spatial reso-
tion of kinematic parameters for five-class finger flexion, while lution signals. But regarding the temporal resolution, fNIRS
[105] has distinguished multiple motor-imagery task pairs such recording is likely to be less effective than that based on elec-
as left versus right hand and finger versus tongue imagery tromagnetic signals. Compared to fMRI, fNIRS is portable as
movement with ECoG brain signals for both paralyzed and shown in Fig. 10 [111] and less expensive but provides less
non-paralyzed epileptic patients. It has been found that non- imaging capabilities. Its advantages present a viable alternative
paralyzed subjects’ activities could have been categorized and for clinical studies and possibly for practical use [95].
understood by the system. ECoG has also been very beneficial
especially in speech and language processing where animals 5.2.4. Electroencephalogram (EEG)
could not be helpful in assessment of brain signals related to
vocal actions and language semantics. The feasibility of decod- Electroencephalography (EEG) is the recording of electrical
ing semantic information, associated with various pictures cat- activity along the scalp through measuring voltage fluctuations
egories, has been analyzed in [7]. They have requested the accompanying neurotransmission activity within the brain.
subjects to apply different language related tasks, such as pic- The electrodes are attached in a cap-like device as shown in
ture naming, on the presented pictures. Fig. 11 [112]. It has unique usability advantages over other
types of brain signal recording that recommend it for commer-
5.2. Noninvasive techniques cial use. It is easy to use, portable and inexpensive. EEG
recording also provides high temporal resolution. However
its signal to noise ratio and spatial resolution represent a lim-
These recording methods follow the approach that does not itation compared to other methods.
require implanting of external objects into subject’s brain. Several solutions have been provided to enhance EEG spa-
Thus it avoids the surgical procedures or permanent device tial resolution issue and improve signal localization. The
attachment needed by invasive acquisition. Various assessment increased use of electrodes up to 256 has been suggested. An
methods for different types of measured signals such as func- international electrode positioning system has been revealed.
tional magnetic resonance imaging (fMRI), functional nearin- It makes the distance between adjacent pair of electrodes to
frared spectroscopy (fNIRS), magnetoencephalography be either 10% or 20% of the scalp diameter [3], according to
(MEG), and electroencephalogram (EEG) are presented next. the assignment demonstrated in Fig. 12 [113]. This configura-
tion has been commonly used across different EEG systems.
5.2.1. Magnetoencephalography (MEG) [114] has reviewed sensors that allow less obtrusiveness and
It is a non-invasive method that measures magnetic fields pro- high portability option for extensive market use such as
duced by electrical currents occurring naturally in the brain. NeuroSky and Emotiv. Another EEG acquiring method is
The magnetic signal outside of the head is currently acquired shown in [115]. It presents an unobtrusive in-the-ear EEG
only using the superconducting quantum interference device recording. It has been tested against on-scalp EEG and proven
(SQUID). MEG signals could interfere with other magnetic sig- feasibility. This approach’s gain appears in fixing electrode
nals such as the earth’s magnetic field so this recording method position, comforting user and robustness to electromagnetic
requires laboratory configuration with shields and specific interference. A Lot of researches regarding BCI based applica-
equipments [106,95] as shown in Fig. 8 [107]. Despite its porta- tions have begun concerning with reducing the number of used
bility and cost issues, MEG signals are less distorted by the electrodes, while maintaining signal to noise ratio.
Figure 12 The 10–20 international system.
Table 1 Summary of acquisition method characteristics.

Cortical surface Intracortical EEG MEG fMRI fNIRS
Invasiveness and medical issues Invasive Invasive Non-invasive Non-invasive Non-invasive Non-invasive
Spatial resolution High Very high Low Mediate High Mediate
Temporal resolution High High Mediate Mediate Low Low
Portability Portable Portable Portable Non-portable Non-portable Portable
Recorded signal Electrical Electrical Electrical Magnetic Metabolic Metabolic
using EEG and ECoG signals is presented in [105]. Also the

presence of speech in the classification of motor imagery has
been examined [122] and it has been found that speech exis-
tence does not notably affect the accuracy results.
Table 1 [123] gives a summary for brain acquisition meth-
ods along with their advantages and disadvantages.
6. BCI electrical signals

Figure 13 BCI electrical signals.
Researches considering electrical signals generated from brain
activities have revealed two main approaches [2] for studying
Further studies concerning EEG signal acquisition method these signals as shown in Fig. 13. The first approach is con-
have taken place. Discriminating numerous types of actions cerned with exploring the effect of various triggering condi-
are considered such as mental and motor actions [116], knee tions on them like in evoked potentials (EP), while the other
and ankle contractions [117], Hand grasping [118,119], and aimed to detect the brain oscillatory that is not necessarily
imagined writing [120,121]. Analyzing the classification results associated with external stimulation like in event related desyn-
of motor actions of both paralyzed and non-paralyzed people chronizationnsynchronization (ERDnERS).
6.1. Evoked potential or evoked response (EP) not high. They can be induced by performing mental tasks,
such as motor imagery, mental arithmetic, or mental rotation.
It is an electrical response recorded from the nervous system
after the presentation of a stimulus. It can be further divided 7. Challenges and proposed solutions
into steady state evoked potential (SSEP) and event-related
potential (ERP) [124]. EP components are labeled either exoge- Establishing the communication interface using brain signals
nous or endogenous [125]. Exogenous components are influ- has faced a lot of challenges. They can be categorized as tech-
enced by physical attributes of stimuli such as intensity, nical and usability. Technical challenges are concerned with
modality, and presentation rate. On the other hand, the psy- the system obstacles specially those regarding EEG features
chological or cognitive significance of the stimulus, that is, characteristics. Usability challenges describe the limitations
the psychological or cognitive demands of the situation deter- affecting the level of human acceptance [134].
mines the endogenous components. They are not influenced by
the physical attributes of the stimuli. Endogenous components 7.1. Challenges
vary in amplitude, latency, and scalp distribution with strate-
gies, expectancies, and other mental activities triggered by 7.1.1. Usability challenges
the event [126]. They express the limitations facing the user acceptance of BCI
Steady State Evoked Potentials (SSEP) [127] are evoked by technology utilization [1]. They include the issues related to the
a stimulus modulated at a fixed frequency and occur as an training process necessary for classes’ discrimination.
increase in EEG activity at the stimulation frequency. The Information transfer rate (ITR) is one of the system evaluation
stimulation could be either visual as in Steady State Visually metrics that combines both performance and acceptance
Evoked Potentials (SSVEP) [128], auditory as in steady-state aspects.
auditory evoked potentials (SSAEP) [129] or even somatosen-
sory as in Steady-state somatosensory evoked potential 7.1.1.1. Training process. Training the user is a time-consuming
(SSSEP) [130]. activity either in guiding the user through the process or in the
Event-related potential (ERP) is elicited by a stimulus number of recorded sessions. It takes place either in prelimi-
change. It is a time-locked deflection on the ongoing brain nary phase or in the classifier calibration phase [135]. The user
activity after exposed to the random occurrence of a desired is taught to deal with the system as well as to control hisnher
target event. The event could be a sensory stimulus, a cognitive brain feedback signals in the preliminary phase, while in the
event, or the execution of a motor response. ERPs include calibration phase, trained subject’s signal has been used to
somatosensory evoked potential (SSEP), visual evoked poten- learn the used classifier.
tials (VEP) [131,132], and auditory evoked potentials (AEP) One of the commonly investigated solutions to this time-
[28]. consumption problem is to employ single trial instead of
ERP amplitudes have values that range from less than a multi-trial analysis, which is used for enhancing signal to noise
microvolt to several microvolts, compared to tens of micro- ration [133], and placing the burden of small training size on
volts for EEG, millivolts for EMG, and often close to a volt subsequent BCI system components to handle. Various
for ECG. Different methods such as sampling, averaging, fil- Adaptive and zero training classifiers have been examined as
tering and eye-artifact removal are proposed to isolate the solutions as mentioned in [126].
ERP signal buried in the EEG recorded specifically during
the brain’s response to any internal or external events. 7.1.1.2. Information transfer rate. It is the widely used evalua-
The components of ERP such as P300, N100 and N400 are tion metric for command BCI systems. It depends on the num-
described with amplitude and latency. The amplitude is either ber of choices, the accuracy of target detection, and the
positive, represented by the letter P, or negative, represented average time for a selection. Thus compared to imagery BCI,
by the letter N, whereas the latency is indicated in terms of mil- selective attention strategies achieve higher ITR as their
liseconds after the stimuli. offered choices are larger [5].
6.2. Event related desynchronizationnsynchronization 7.1.2. Technical challenges

(ERDnERS) They are issues related to the recorded electrophysiological
properties of the brain signals which include non-linearity,
It has been found that changes caused by some events can non-stationarity and noise, small training sets and the compa-
block or decrease the power of the ongoing EEG signal. nying dimensionality curse.
They are time-locked to the event but not phase-locked, and
thus cannot be extracted by a simple linear method, such as 7.1.2.1. Non-linearity. The brain is a highly complex nonlinear
averaging. The decreases or increases of power in given fre- system in which chaotic behavior of neural ensembles can be
quency bands can be used to detect these changes. This may detected. Thus EEG signals can be better characterized by
be related to the synchrony level of the underlying neuronal nonlinear dynamic methods than linear methods.
populations. The power decrease is called event-related desyn-
chronization or ERD while the power increased is called event- 7.1.2.2. Nonstationarity and noise. Nonstationarity attribute of
related synchronization (ERS) [133]. Although the ERDnERS electrophysiological brain signals represents a major issue in
BCI systems do not need external stimulation, they require developing a BCI system [2,136]. It originates a continuous
extensive training that may take many weeks, their perfor- change of the used signals over time either between or within
mance is quite variable between users and their accuracy is the recording sessions. The mental and emotional state
Table 2 Examples of BCI systems.

Application (and/or) Preprocessing Feature extraction Classifier Results
stimulation type (%)
Subjects identification with Autoregressive coefficients (AR) SVM 99.76
imagined speech [142]
Subjects identification with Autoregressive coefficients (AR) SVM 98.96
VEPs [142]
Subjects identification with Autoregressive coefficients (AR) KNN 99.41
imagined speech [142]
Person identification [143,144] Autoregressive coefficients (AR) + Power Spectrum Density (PSD) KNN + Fisher’s linear 97.5
discriminant
analysis(FDA)
Autoregressive coefficients (AR) + Power Spectrum Density (PSD) KNN 70.7
Autoregressive coefficients (AR) + Power Spectrum Density (PSD) SVM 79.6
Effect of image stimulus type on Bandpass filtering SVM 95.1
neural control of a smart TV
[145]
Effect of video stimulus type on Bandpass filtering SVM 93.3
neural control of a smart TV
[145]
Classification of Go/NoGo Bandpass filtering Wavelet transform and Short Time Fourier SVM 91
tasks [146] Transform (STFT)
Recognize familiar objects [147] Independent component analysis (ICA) SVM 87
Detecting reading intention Bandpass filtering PCA KNN 86.49
[148]
BCI arousal detection [149] Band pass filtering Asymmetric spatial pattern (ASP) KNN 82.25
Band pass filtering common spatial pattern (CSP) KNN 76.98
Band pass filtering Asymmetric features (AF) KNN 62.52
BCI arousal detection [149] Band pass filtering Asymmetric spatial pattern (ASP) SVM 82.03
Band pass filtering Common spatial pattern (CSP) SVM 77.72
Band pass filtering Asymmetric features (AF) SVM 69.42
Choice/no choice task [150] Bandpass filtering CSP LDA 80
Imagined movement of left and Stationary Subspace CSP LDA 79.9
right hand [151] Analysis (SSA)
Moving sound P300 speller [6] Bandpass filter SVM 71.4
Bandpass filter LDA 28.6
Motor imagery Bandpass filtering CSP+ Common Spatial Patterns Patches LDA 70
classification [152] (CSPP)
Detect emergency situations Power-line notch Mean power spectrum LDA 70
from driver’s mental states [24] filter+ Bandpass
filtering
Intended movement direction ICA Canonical correlation analysis (CCA) + EEG SVM 69.7
[153] spectral power modulations in alpha and theta
bands
ICA Canonical correlation analysis (CCA) SVM 65.6
ICA EEG spectral power modulations in alpha and SVM 65.4
theta bands
BCI valence detection [149] Bandpass filtering Asymmetric spatial pattern (ASP) KNN 66.51
Band pass filtering Asymmetric features (AF) KNN 62.01
Band pass filtering Common spatial pattern (CSP) KNN 58.23
BCI valence detection [149] Band pass filtering Asymmetric spatial pattern (ASP) SVM 65.39
Band pass filtering Asymmetric features (AF) SVM 61.89
Band pass filtering Common spatial pattern (CSP) SVM 57.54
Motor and mental activity Low pass filter Power spectral density (PSD) SVM 64.18
discrimination [116] Low pass filter Power spectral density (PSD) LDA 52.76
Static sound P300 speller [6] Bandpass filter SVM 62.9
Bandpass filter LDA 20
background through different sessions can contribute in EEG Temporal based preprocessing can contribute in removing
signals variability. Also fatigue and concentration levels are artifacts [140] from the signal using linear combination of
considered part of internal nonstationarity factors. Noise is the EOG-contaminated EEG signal and the EOG signal
also a big contributor in the challenges facing the BCI technol- recorded using eye movement recording electrodes. The combi-
ogy and causing the nonstationarity issue. It includes nation factors are determined via linear regression methods.
unwanted signals caused by alterations in electrode placement, Although it represents the most common technique for remov-
and environmental noise [106]. A combination of movement ing ocular artifacts from EEG signals, it does not meet the
artifacts, such as electrical activity produced by skeletal mus- same success with removing EMG signals because of the diffi-
cles electromyogram (EMG) and signals created by eye move- culty of placing muscular detection electrodes.
ments and blinking Electrooculogram (EOG) [131], is also Frequency-band filtering assists in removing noise and arti-
reflected in the acquired signals resulting in difficulties in dis- fact. It can also provide a significant help with handling the
tinguishing the underlying pattern. internal nonstationarity factors. The task related frequencies
can be selected to be used for further analysis in BCI systems.
7.1.2.3. Small training sets. The training sets are relatively This filtering type does not require extra electrodes to detect
small, since the training process is influenced by usability the eye or muscle movement. The advantage of using filtering
issues. Although heavily training sessions are considered time is its simplicity. However the effect for this method degrades if
consuming and demanding for the subjects, they provide the the unrelated signal overlaps or lies in the same frequency
user with necessary experience to deal with the system and band as the signal of interest [131,132].
learn to control hisnher neurophysiological signals. Thus a sig-
nificant challenge in designing a BCI is to balance the trade-off 7.2.2. Separability of multiple classes
between the technological complexity of interpreting the user’s Machine learning techniques are employed to translate user’s
brain signals and the amount of training needed for successful intent into a valid choice. They discriminate and identify the
operation of the interface [124]. selected class. They have been used, for example, to overcome
some limitations associated with small training sets, single
7.1.2.4. High dimensionality curse. In BCI systems, the signals trial, and also the variability between sessions and within indi-
are recorded from multiple channels to preserve high spatial vidual sessions. They also aim to achieve higher performance
accuracy. As the amount of data needed to properly describe and accordingly, higher ITR results. Next we demonstrate
different signals increases exponentially with the dimensional- three different machine learning algorithms such as linear dis-
ity of the vectors, various feature extraction methods have criminant analysis (LDA), support vector machine (SVM), and
been proposed. They play an important role in identifying dis- k nearest neighbors (KNN).
tinguishing characteristics. Thus the classifier performance will
be affected only by the small number of distinctive traits 7.2.2.1. Linear discriminant analysis. LDA is deployed to find
instead of the whole recorded signals that may contain the linear combinations of feature vectors which describe the
redundancy. characteristics of the corresponding signal. LDA seeks to sep-
Generally, it is recommended to use, at least, five to ten arate two or more classes of objects or events representing dif-
times as many training samples per class as the number of ferent classes. It utilizes hyperplanes to accomplish this
dimensions [134]. But this solution cannot be sustained in a mission. Separating hyperplane is obtained by searching for
highly dimensional environment as the BCI system, causing the projection that maximizes the distance between the classes’
the expanding of the dimensionality curse [137]. means and minimizes the interclass variance.
This technique has a very low computational requirement
7.2. Proposed solutions and it is simple to use. LDA has been used with success in var-
ious types of BCI systems such as motor imagery based BCI,
Several solutions have been investigated to confront and limit P300 speller, multiclass or asynchronous BCI. But although
the influence of the previously mentioned technical issues. it generally provides good results due to its immunity to non-
They are spread over various BCI system components. The fol- stationarity problem, its linearity can cause performance
lowing sections explain some employed methods for improving degradation in some situations having complex nonlinear
the performance of BCI based systems. EEG data.
7.2.1. Noise removal 7.2.2.2. Support vector machine. SVM is an algorithm that
Preprocessing in either spatial, time or frequency domains has belongs to a category of classification methods which use
contributed in enhancing the signal caused especially by exter- supervised learning to separate two different classes of data.
nal factors. Improving the signal to noise ration (SNR) [138] of It exploits a discriminant hyperplane to identify classes like
EEG signals is done by increasing the signal level and/or does LDA. However in SVM case, the selected hyperplane is
decreasing the noise level. the one that maximizes the distance from the nearest training
A widely used spatial filtering method is Independent points. This optimal hyperplane is described by the vectors
Component Analysis (ICA) [139]. ICA accomplishes spatial fil- which lie on the margin which are called support vectors.
tering in an unsupervised manner by decomposing the SVM has several advantages. It is known to have good gen-
observed EEG into statistically independent components eralization properties, to be insensitive to overtraining and to
(ICs). It aims at increasing the SNR of EEG signals via sepa- the curse-of-dimensionality [141]. Finally, SVM shows good
rating task-related EEG components from the task-irrelevant performance results in both Evoked potential and ERD/ERS
EEG components and the artifactual components. BCI.
7.2.2.3. K nearest neighbors. It is labeled as one of the unsuper- [8] Brumberg JS, Nieto-Castanon A, Kennedy PR, Guenther FH.
vised nearest neighbor classifiers in which the feature vector is Brain-computer interfaces for speech communication. Speech
assigned to its closest class among k neighbors. KNN algo- Commun 2010;52(4):367–79.
rithms’ main advantage is simplicity. However their sensitivity [9] Vourvopoulos A, Liarokapis F. Robot navigation using brain-
computer interfaces. In: Trust, Security and Privacy in
to the curse-of-dimensionality is considered the main draw-
Computing and Communications (TrustCom). In: 2012 IEEE
back that affects their performance regarding BCI systems. 11th International Conference on. IEEE; 2012. p. 1785–92.
Thus it can provide good results with efficient feature selection [10] Van de Laar B, Gurkok H, Plass-Oude Bos D, Poel M, Nijholt
and reduction algorithms. A. Experiencing BCI control in a popular computer game
Some examples for using these classifiers for increasing sep- computational intelligence and AI in Games. IEEE Trans
arability between classes in BCI systems along with the 2013;5(2):176–84.
deployed preprocessing and feature extraction methods are [11] Prataksita N, Lin Y-T, Chou H-C, Kuo C-H. Brain-robot
shown in Table 2. control interface: Development and application. In:
Bioelectronics and Bioinformatics (ISBB), 2014 IEEE
International Symposium on. IEEE; 2014. p. 1–4.
8. Conclusion [12] Brouwer A-M, van Erp J, Heylen D, Jensen O, Poel M.
Effortless passive BCIs for healthy users in Universal Access in
Brain signals reflect the handled activities and controlling Human-Computer Interaction Design Methods. Tools and
behavior of the brain or the influence of the received informa- Interaction Techniques for eInclusion. Springer; 2013.
tion from other body parts either sensing or internal organs. [13] Garcia-Molina G, Tsoneva T, Nijholt A. Emotional brain-
computer interfaces. Int J Auton Adapt Commun Syst
Brain Computer Interfacing provides a channeling facility
2013;6(1):9–25.
between brain and external equipment.
[14] Tan D, Nijholt A. Brain-computer interfaces and human–
BCI applications have attracted the research community. computer interaction. Brain-computer interfaces. Springer; 2010.
Several studies have been presented in this paper regarding [15] Fukushima M, Inoue A, Niwa T. Emotional evaluation of tv-cm
the growing interest in BCI application fields such as medical, using the fractal dimension and the largest lyapunov exponent.
organizational, transportation, games and entertainment, and In: Systems Man and Cybernetics (SMC), 2010 IEEE
security and authentication fields. It also demonstrates the var- International Conference on. IEEE; 2010. p. 1473–76.
ious devices used for capturing brain signals. [16] Venthur B, Blankertz B, Gugler MF, Curio G. Novel applica-
These recording devices are divided into two main cate- tions of bci technology: psychophysiological optimization of
gories: invasive and non-invasive. Invasive category, which working conditions in industry. In: Systems Man and
Cybernetics (SMC), 2010 IEEE International Conference on.
requires implanting surgery, is usually needed for critical par-
IEEE; 2010. p. 417–21.
alyzed situations because of their higher accuracy rates
[17] Sorudeykin KA. An educative brain-computer interface. arXiv
achieved either spatially or temporally. On the other hand, preprint arXiv:1003.2660; 2010.
the non-invasive category, as mentioned previously, has been [18] Hanafiah ZM, Taib MN, Hamid N. Eeg pattern of smokers for
widely spread in other application fields due to its advantages theta, alpha and beta band frequencies. In: Research and
over the invasive one. Other challenges and issues posed as a Development (SCOReD), 2010 IEEE Student Conference on.
result of utilizing brain signals have also been discussed along IEEE; 2010. p. 320–23.
with some solutions offered by different algorithms at various [19] Di D, Zhihua C, Ruifang F, Guangyu L, Tian L. Study on
BCI processing components. human brain after consuming alcohol based on eeg signal. In:
Computer Science and Information Technology (ICCSIT), 2010
3rd IEEE International Conference on, vol. 5. IEEE; 2010. p.
References 406–09.
[20] Eksi Z, Akgül A, Recep Bozkurt M. The classification of eeg
[1] van Erp J, Lotte F, Tangermann M. Brain-computer interfaces: signals recorded in drunk and non-drunk people. Int J Comput
beyond medical applications. Computer 2012;45(4):26–34. Appl 2013;68.
[2] Rao R, Scherer R. Brain-computer interfacing [in the spotlight]. [21] Shri P, Sriraam N. Eeg based detection of alcoholics using
Signal Process Mag, IEEE 2010;27(4), 152-150. spectral entropy with neural network classifiers. In: Biomedical
[3] Bi L, Fan X-A, Liu Y. Eeg-based brain-controlled mobile Engineering (ICoBE), 2012 International Conference on. IEEE;
robots: a survey. Human-Machine Syst, IEEE Trans 2012. p. 89–93.
2013;43(2):161–76. [22] Malar E. A novel approach for the detection of drunken driving
[4] Navarro AA, Ceccaroni L, Velickovski F, Torrellas S, Miralles using the power spectral density analysis of EEG. Int J Comput
F, Allison BZ, R.Scherer R, Faller J. Context-awareness as an Appl 2011;21.
enhancement of brain-computer interfaces. Ambient assisted [23] Shooshtari MA, Setarehdan SK. Selection of optimal eeg
living. Springer; 2011. channels for classification of signals correlated with alcohol
[5] Tan DS, Nijholt A. Brain-computer interfaces: applying our abusers. In: Signal Processing (ICSP), 2010 IEEE 10th
minds to human-computer interaction. Springer; 2010. International Conference on. IEEE; 2010. p. 1–4.
[6] Lelievre Y, Washizawa Y, Rutkowski TM. Single trial BCI [24] Fan X, Bi L, Wang Z. Detecting emergency situations by
classification accuracy improvement for the novel virtual sound monitoring drivers’ states from eeg. In: Complex Medical
movement-based spatial auditory paradigm. In: Signal and Engineering (CME), 2012 ICME International Conference on.
information processing association annual summit and confer- IEEE; 2012. p. 245–48.
ence (APSIPA), 2013 Asia-Pacific. IEEE; 2013. p. 1–6. [25] <http://www.eip.gov.eg/Documents/StudiesDetails.aspx?id=
[7] Wang W, Degenhart AD, Sudre GP, Pomerleau DA, Tyler- 522>, [accessed 05.10.14].
Kabara EC. Decoding semantic information from human [26] Lin C-T, Tsai S-F, Ko L-W. Eeg-based learning system for
electrocorticographic (ecog) signals. In: Engineering in online motion sickness level estimation in a dynamic vehicle
Medicine and Biology Society, EMBC, 2011 Annual environment. Neural Networks Learn Syst, IEEE Trans
International Conference of the IEEE. IEEE; 2011. p. 6294–98. 2013;24(10):1689–700.
[27] Wei C-S, Chuang S-W, Wang W-R, Ko L-W, Jung T-P, Lin C- (EMBC), 2010 Annual International Conference of the IEEE.
T. Implementation of a motion sickness evaluation system IEEE; 2010. p. 4234–37.
based on eeg spectrum analysis. In: Circuits and Systems [43] Birbaumer N, Ruiz S, Sitaram R. Learned regulation of brain
(ISCAS), 2011 IEEE International Symposium on. IEEE; 2011. metabolism. Trends Cognitive Sci 2013;17(6):295–302.
p. 1081–84. [44] Ruiz S, Buyukturkoglu K, Rana M, Birbaumer N, Sitaram R.
[28] Paulraj M, Bin Yaccob S, Adom B, Hamid A, Hema C, Real-time fmri brain computer interfaces: self-regulation of
Subramaniam K. Eeg based hearing perception level estimation single brain regions to networks. Biol Psychol 2014;95:4–20.
for normal hearing persons. In: Control, Systems & Industrial [45] Barbosa AO, Achanccaray DR, Meggiolaro MA. Activation of a
Informatics (ICCSII), 2012 IEEE Conference on. IEEE; 2012. p. mobile robot through a brain computer interface. In: Robotics
160–62. and Automation (ICRA), 2010 IEEE International Conference
[29] Ko L-W, Lee H-C, Tsai S-F, Shih T-C, Chuang Y-T, Huang H- on. IEEE; 2010. p. 4815–21.
L, Ho S-Y, Lin C-T. Eeg-based motion sickness classification [46] Ang KK, Guan C, Sui Geok Chua K, Ang BT, Kuah C, Wang
system with genetic feature selection. In: Computational C, Phua KS, Chin ZY, Zhang H. Clinical study of neuroreha-
Intelligence, Cognitive Algorithms, Mind, and Brain (CCMB), bilitation in stroke using eeg-based motor imagery brain-
2013 IEEE Symposium on. IEEE; 2013. p. 158–64. computer interface with robotic feedback. In: Engineering in
[30] Bagchi S, Chattopadhyay M. An easy-to-adopt approach for Medicine and Biology Society (EMBC), 2010 Annual
regular and routine monitoring of the consciousness level of International Conference of the IEEE. IEEE; 2010. p. 5549–52.
human brain of stayed alone sick person. In: Sensing Technology [47] King CE, Wang PT, Mizuta M, Reinkensmeyer DJ, Do AH,
(ICST), 2012 Sixth International Conference on. IEEE; 2012. p. Moromugi S, Nenadic Z. Noninvasive brain-computer interface
698–703. driven hand orthosis. In: Engineering in Medicine and Biology
[31] Selvam VS, Shenbagadevi S. Brain tumor detection using scalp Society, EMBC, 2011 Annual International Conference of the
eeg with modified wavelet-ica and multi layer feed forward IEEE. IEEE; 2011. p. 5786–89.
neural network. In: Engineering in Medicine and Biology [48] Meyer T, Peters J, Brtz D, Zander TO, Scholkopf B, Soekadar
Society, EMBC, 2011 Annual International Conference of the SR, Grosse-Wentrup M. A brain–robot interface for studying
IEEE. IEEE; 2011. p. 6104–09. motor learning after stroke. In: Intelligent Robots and Systems
[32] Sharanreddy M, Kulkarni P. Detection of primary brain tumor (IROS), 2012 IEEE/RSJ International Conference on. IEEE;
present in eeg signal using wavelet transform and neural 2012. p. 4078–83.
network. Int J Biol Med Res 2013;4(1). [49] Jones CL, Wang F, Morrison R, N.Sarkar N, Kamper DG.
[33] Poulos M, Felekis T, Evangelou A. Is it possible to extract a Design and development of the cable actuated finger exoskeleton
fingerprint for early breast cancer via eeg analysis? Med for hand rehabilitation following stroke. IEEE Syst J 2014.
Hypotheses 2012;78(6):711–6. [50] Presacco A, Forrester L, Contreras-Vidal JL. Towards a non-
[34] Sharanreddy M, Kulkarni P. Automated eeg signal analysis for invasive brain-machine interface system to restore gait function
identification of epilepsy seizures and brain tumour. J Med Eng in humans. In: Engineering in Medicine and Biology Society,
Technol 2013;37(8):511–9. EMBC, 2011 Annual International Conference of the IEEE.
[35] Liang S-F, Shaw F-Z, Young C-P, Chang D-W, Liao Y-C. A IEEE; 2011. p. 4588–91.
closed-loop brain computer interface for real-time seizure [51] Contreras-Vidal JL, Presacco A, Agashe H, Paek A. Restoration
detection and control. In: Engineering in Medicine and Biology of whole body movement: toward a noninvasive brain–machine
Society (EMBC), 2010 Annual International Conference of the interface system. Pulse, IEEE 2012;3(1):34–7.
IEEE. IEEE; 2010. p. 4950–53. [52] Regenbrecht H, Hoermann S, Ott C, Muller L, Franz E.
[36] Helini Kulasuriya K, Perera M. Forecasting epileptic seizures Manipulating the experience of reality for rehabilitation appli-
using eeg signals, wavelet transform and artificial neural cations. Proc IEEE 2014;102(2):170–84.
networks. In: IT in Medicine and Education (ITME), 2011 [53] Ang KK, Guan C, Phua KS, Wang C, Teh I, Chen CW, Chew E.
International Symposium on, vol. 1. IEEE; 2011. p. 557–62. Transcranial direct current stimulation and EEG-based motor
[37] Fadzal C, Mansor W, Khuan L. Review of brain computer imagery bci for upper limb stroke rehabilitation. In: Engineering
interface application in diagnosing dyslexia. In: Control and in Medicine and Biology Society (EMBC), 2012 Annual
System Graduate Research Colloquium (ICSGRC), 2011 IEEE. International Conference of the IEEE. IEEE; 2012. p. 4128–31.
IEEE; 2011. p. 124–28. [54] Tung SW, Guan C, Ang KK, Phua KS, Wang C, Zhao L, Teo
[38] Koch H, Christensen JA, Frandsen R, Arvastson L, Christensen WP, Chew E. Motor imagery BCI for upper limb stroke
SR, Sorensen HB, Jennum P, Classification of irbd and rehabilitation: an evaluation of the eeg recordings using coher-
parkinson’s patients using a general data-driven sleep staging ence analysis. In: Engineering in Medicine and Biology Society
model built on eeg. In: Engineering in Medicine and Biology (EMBC), 2013 35th Annual International Conference of the
Society (EMBC), 2013 35th Annual International Conference of IEEE. IEEE; 2013. p. 261–64.
the IEEE. IEEE; 2013. p. 4275–78. [55] Höller Y, Bergmann J, Kronbichler M, Crone JS, Schmid EV,
[39] Hansen IH, Marcussen M, Christensen JA, Jennum P, Sorensen Thomschewski A, Butz K, Schütze V, Höller P, Trinka E. Real
HB. Detection of a sleep disorder predicting parkinson’s disease. movement vs. motor imagery in healthy subjects. Int J
In: Engineering in Medicine and Biology Society (EMBC), 2013 Psychophysiol 2013;87(1):35–41.
35th Annual International Conference of the IEEE. IEEE; 2013. [56] Kang H, Park W, Kang J-H, Kwon G-H, Kim S-P, Kim L. A
p. 5793–96. neural analysis on motor imagery and passive movement using a
[40] Wei L, Hong Q, Yue H, Xi C. The research in a plantar pressure haptic device. In: Control, Automation and Systems (ICCAS),
measuring system connected with eeg’’. In: Signal Processing 2012 12th International Conference on. IEEE; 2012. p. 1536–41.
(ICSP), 2010 IEEE 10th International Conference on. IEEE; [57] Domingo MC. An overview of the internet of things for people
2010. p. 434–37. with disabilities. J Netw Comput Appl 2012;35(2):584–96.
[41] Ang CS, Sakel M, Pepper M, Phillips M. Use of brain computer [58] Lin C-T, Lin B-S, Lin F-C, Chang C-J. Brain computer
interfaces in neurological rehabilitation. Brit J Neurosci Nurs interface-based smart living environmental auto-adjustment
2011;7(3):523–8. control system in UpnP home networking. IEEE Syst J 2012.
[42] Tan H, Kong K, Shee C, Wang C, Guan C, Ang W. Post-acute [59] Ou C-Z, Lin B-S, Chang C-J, Lin C-T. Brain computer interface-
stroke patients use brain-computer interface to activate electrical based smart environmental control system. In: Intelligent
stimulation. In: Engineering in Medicine and Biology Society Information Hiding and Multimedia Signal Processing (IIH-
MSP), 2012 Eighth International Conference on. IEEE; 2012. p. [75] Vecchiato G, Astolfi L, De Vico Fallani F, Salinari S, Cincotti F,
281–84. Aloise F, Mattia D, Marciani MG, Bianchi L, Soranzo R et al.
[60] Peng H, Hu B, Qi Y, Zhao Q, Ratcliffe M. An improved EEG The study of brain activity during the observation of commercial
de-noising approach in electroencephalogram (EEG) for home advertising by using high resolution EEG techniques. In:
care. In: Pervasive Computing Technologies for Healthcare Engineering in Medicine and Biology Society, 2009. EMBC
(PervasiveHealth), 2011 5th International Conference on. IEEE; 2009. Annual International Conference of the IEEE. IEEE; 2009.
2011. p. 469–74. p. 57–60.
[61] Roy RN, Bonnet S, Charbonnier S, Campagne A. Mental [76] Yoshioka M, Inoue T, Ozawa J. Brain signal pattern of
fatigue and working memory load estimation: Interaction and engrossed subjects using near infrared spectroscopy (nirs) and
implications for eeg-based passive bci. In: Engineering in its application to tv commercial evaluation. In: Neural Networks
Medicine and Biology Society (EMBC), 2013 35th Annual (IJCNN), The 2012 International Joint Conference on. IEEE;
International Conference of the IEEE. IEEE; 2013. p. 6607–10. 2012. p. 1–6.
[62] Duru DG, Deniz Duru A, Barkana DE, Sanli O, Ozkan M. [77] Vecchiato G, Babiloni F, Astolfi L, Toppi J, Cherubino P, Dai J,
Assessment of surgeon’s stress level and alertness using eeg Kong W, Wei D. Enhance of theta eeg spectral activity related to
during laparoscopic simple nephrectomy. In: Neural Engineering the memorization of commercial advertisings in chinese and
(NER), 2013 6th International IEEE/EMBS Conference on. italian subjects. In: Biomedical Engineering and Informatics
IEEE; 2013. p. 452–55. (BMEI), 2011 4th International Conference on, vol. 3. IEEE;
[63] Dong Y, Hu Z, Uchimura K, Murayama N. Driver inattention 2011. p. 1491–94.
monitoring system for intelligent vehicles: a review. Intell [78] Johnston S, Boehm S, Healy D, Goebel R, Linden D.
Transport Syst, IEEE Trans 2011;12(2):596–614. Neurofeedback: a promising tool for the self-regulation of
[64] Coetzer R, Hancke G. Driver fatigue detection: A survey. In: emotion networks. Neuroimage 2010;49(1):1066–72.
AFRICON, 2009. AFRICON’09. IEEE; 2009. p. 1–6. [79] Zotev V, Phillips R, Yuan H, Misaki M, Bodurka J. Self-
[65] Pritchett S, Zilberg E, Xu ZM, Karrar M, Burton D, Lal S. regulation of human brain activity using simultaneous real-time
Comparing accuracy of two algorithms for detecting driver fmri and eeg neurofeedback. NeuroImage 2014;85:985–95.
drowsiness––single source (EEG) and hybrid (eeg and body [80] Marquez BY, Alanis A, Lopez MA, Magdaleno-Palencia JS.
movement). In: Broadband and Biomedical Communications Sport education based technology: Stress measurement in
(IB2Com), 2011 6th International Conference on. IEEE; 2011. p. competence. In: e-Learning and e-Technologies in Education
179–84. (ICEEE), 2012 International Conference on. IEEE; 2012. p. 247–
[66] Dahal N, Nandagopal N, Nafalski A, Nedic Z. Modeling of 52.
cognition using EEG: a review and a new approach. In: [81] Royer AS, Doud AJ, Rose ML, He B. Eeg control of a virtual
TENCON 2011–2011 IEEE Region 10 Conference. IEEE; helicopter in 3-dimensional space using intelligent control
2011. p. 1045–49, just the heirarichy of a model with no serious strategies. Neural Syst Rehabilitation Eng, IEEE Trans
results but good abstract introduction for BCI. 2010;18(6):581–9.
[67] Wang Q, Wang H, Zhao C, Yang J. Driver fatigue detection [82] Bonnet L, Lotte F, Lécuyer A. Two brains one game: design and
technology in active safety systems. In: Remote Sensing, evaluation of a multi-user bci video game based on motor
Environment and Transportation Engineering (RSETE), 2011 imagery 2013.
International Conference on. IEEE; 2011. p. 3097–3100. [83] Khalifa W, Salem A, Roushdy M, Revett K. A survey of EEG
[68] Borghini G, Vecchiato G, Toppi J, Astolfi L, Maglione A, based user authentication schemes. In: Informatics and Systems
Isabella R, Caltagirone C, Kong W, Wei D, Zhou Z et al. (INFOS), 2012 8th International Conference on. IEEE; 2012. p.
Assessment of mental fatigue during car driving by using high BIO–55.
resolution eeg activity and neurophysiologic indices. In: [84] Karthikeyan DT, Sabarigiri B. Enhancement of multi-modal
Engineering in Medicine and Biology Society (EMBC), 2012 biometric authentication based on iris and brain neuro image
Annual International Conference of the IEEE. IEEE; 2012. p. coding. Int J Biometrics Bioinform (IJBB) 2011;5(5):249–56.
6442–45. [85] Svogor I, Kisasondi T. Two factor authentication using eeg
[69] Wang Y-K, Chen S-A, Lin C-T. An EEG-based brain-computer augmented passwords. In: Information Technology Interfaces
interface for dual task driving detection. Neurocomputing 2013. (ITI), Proceedings of the ITI 2012 34th International Conference
[70] Kim T, Kim S, Shin D, Shin D. Design and implementation of on. IEEE; 2012. p. 373–78.
smart driving system using context recognition system. In: [86] Revett K, Deravi F, Sirlantzis K. Biosignals for user authenti-
Computers & Informatics (ISCI), 2011 IEEE Symposium on. cation-towards cognitive biometrics? In: Emerging Security
IEEE; 2011. p. 84–89. Technologies (EST), 2010 International Conference on. IEEE;
[71] Murata K, Fujita E, Kojima S, Maeda S, Ogura Y, Kamei T, 2010. p. 71–76.
Tsuji T, Kaneko S, Yoshizumi M, Suzuki N. Noninvasive [87] Su F, Zhou H, Feng Z, Ma J. A biometric-based covert warning
biological sensor system for detection of drunk driving. Inform system using EEG. In: Biometrics (ICB), 2012 5th IAPR
Technol Biomed, IEEE Tran 2011;15(1):19–25. International Conference on. IEEE; 2012. p. 342–47.
[72] Jingkun L, Guizhi X, Jianrong G, Feng Z, Xiuyan F. Analysis of [88] Nakanishi I, Baba S, Li S. Evaluation of brain waves as
virtual simulation environment for driving behavior based on biometrics for driver authentication using simplified driving
imaging movements. In: Networking and Digital Society simulator. In: Biometrics and Kansei Engineering (ICBAKE),
(ICNDS), 2010 2nd International Conference on, vol. 1. IEEE; 2011 International Conference on. IEEE; 2011. p. 71–76.
2010. p. 246–48. [89] Nakanishi I, Baba S, Ozaki K, Li S. Using brain waves as
[73] Lin C-T, Chen S-A, Ko L-W, Wang Y-K. EEG-based brain transparent biometrics for on–demand driver authentication. Int
dynamics of driving distraction. In: Neural Networks (IJCNN), J Biometrics 2013;5(3):288–305.
The 2011 International Joint Conference on. IEEE; 2011. p. [90] Nakanishi I, Ozaki K, Li S. Evaluation of the brain wave as
1497–500. biometrics in a simulated driving environment. In: Biometrics
[74] Kawamura R, Bhuiyan MS, Kawanaka H, Oguri K. Special Interest Group (BIOSIG), 2012 BIOSIG-Proceedings of
Simultaneous stimuli of vibration and audio for in-vehicle driver the International Conference of the. IEEE; 2012. p. 1–5.
activation. In: Intelligent Transportation Systems (ITSC), 2011 [91] Thorpe J, Van Oorschot P, Somayaji A. Pass-thoughts: authen-
14th International IEEE Conference on. IEEE; 2011. p. 1710–715. ticating with our minds. In: New Security Paradigms Workshop:
Proceedings of the 2005 workshop on New security paradigms, [112] <http://www.emotiv.com>, [accessed 08.07.14] .
vol. 20, no. 23; 2005. p. 45–56. [113] <http://www.bci2000.org/wiki/index.php/User_Tutorial:EEG_
[92] <http://fmri.ucsd.edu/images/BOLDSignal.jpg>, [accessed Measurement_Setup>, [accessed 02.10.14].
11.10.14]. [114] Hondrou C, Caridakis G. Affective, natural interaction using
[93] <http://vbmeg.atr.jp/docs/manual/manual_e.html>, [accessed EEG: sensors, application and future directions. In: Artificial
11.10.14]. Intelligence: Theories and Applications. Springer; 2012. p. 331–38.
[94] MULLER K-R, Kubler A. Toward brain computer interfacing. [115] Looney D, Kidmose P, Park C, Ungstrup M, Rank ML,
Massachusetts Institute of Technology 2007:1–25. Rosenkranz K, Mandic DP. The in-the-ear recording concept:
[95] He B, Gao S, Yuan H, Wolpaw JR. Brain-computer interfaces. user-centered and wearable brain monitoring. Pulse, IEEE
Neural Engineering. Springer; 2013. 2012;3(6):32–42.
[96] Hochberg LR, Serruya MD, Friehs GM, Mukand JA, Saleh M, [116] Ozmen N, Ktu L. Discrimination between mental and motor
Caplan AH, Branner A, Chen D, Penn RD, Donoghue JP. tasks of EEG signals using different classification methods. In:
Neuronal ensemble control of prosthetic devices by a human Innovations in Intelligent Systems and Applications (INISTA),
with tetraplegia. Nature 2006;442(7099):164–71. 2011 International Symposium on. IEEE; 2011. p. 143–47.
[97] Shan H, Yuan H, Zhu S, He B. Eeg-based motor imagery [117] Gwin JT, Ferris D. High-density eeg and independent compo-
classification accuracy improves with gradually increased chan- nent analysis mixture models distinguish knee contractions from
nel number. In: Engineering in Medicine and Biology Society ankle contractions. In: Engineering in Medicine and Biology
(EMBC), 2012 Annual International Conference of the IEEE. Society, EMBC, 2011 Annual International Conference of the
IEEE; 2012. p. 1695–98. IEEE. IEEE; 2011. p. 4195–198.
[98] Mikhail M, El-Ayat K, Coan JA, Allen JJ. Using minimal [118] Park W, Jeong W, Kwon G-H, Kim Y-H, Kim L. A rehabil-
number of electrodes for emotion detection using brain signals itation device to improve the hand grasp function of stroke
produced from a new elicitation technique. Int J Auton Adapt patients using a patient-driven approach. In: Rehabilitation
Commun Syst 2013;6(1):80–97. Robotics (ICORR), 2013 IEEE International Conference on.
[99] Jatupaiboon N, Pan-ngum S, Israsena P. Emotion classification IEEE; 2013. p. 1–4.
using minimal eeg channels and frequency bands. In: Computer [119] Che Wan Fadzal C, Mansor W, Khuan L. An analysis of EEG
Science and Software Engineering (JCSSE), 2013 10th signal generated from grasping and writing. In: Computer
International Joint Conference on. IEEE; 2013. p. 21–24. Applications and Industrial Electronics (ICCAIE), 2011 IEEE
[100] Wei Q, Wang Y, Lu Z. Channel reduction by cultural-based International Conference on. IEEE; 2011. p. 535–37.
multi-objective particle swarm optimization based on filter bank [120] Ismail K, Mansor W, Khuan L, Che Wan Fadzal C. Spectral
in brain-computer interfaces. Unifying Electrical Engineering analysis of eeg signals generated from imagined writing. In:
and Electronics Engineering. Springer; 2014. Signal Processing and its Applications (CSPA), 2012 IEEE 8th
[101] Roland JL, Hacker CD, Breshears JD, Gaona CM, Hogan RE, International Colloquium on. IEEE; 2012. p. 510–13.
Burton H, Corbetta M, Leuthardt EC. Brain mapping in a [121] Zabidi A, Mansor W, Lee Y, Che Wan Fadzal C. Short-time
patient with congenital blindness–a case for multimodal fourier transform analysis of EEG signal generated
approaches. Front Human Neurosci 2013;7. during imagined writing. In: System Engineering and
[102] Wolpaw JR, Boulay CB. Brain signals for brain-computer Technology (ICSET), 2012 International Conference on. IEEE;
interfaces. Brain-computer interfaces. Springer; 2010. 2012. p. 1–4.
[103] Graimann B, Allison B, Pfurtscheller G. Brain-computer inter- [122] Gurkok H, Poel M, Zwiers J. Classifying motor imagery in
faces: a gentle introduction. Brain-computer interfaces. Springer; presence of speech. In: Neural Networks (IJCNN), The 2010
2010. International Joint Conference on. IEEE; 2010. p. 1–8.
[104] Elghrabawy A, Wahed MA. Prediction of five-class finger flexion [123] Nicolas-Alonso LF, Gomez-Gil J. Brain computer interfaces, a
using ecog signals. In: Biomedical Engineering Conference review. Sensors 2012;12(2):1211–79.
(CIBEC), 2012 Cairo International. IEEE; 2012. p. 1–5. [124] Allison BZ, Dunne S, Leeb R. Towards practical brain-computer
[105] Hill NJ, Lal TN, Schroder M, Hinterberger T, Wilhelm B, interfaces: bridging the gap from research to real-world appli-
Nijboer F, Mochty U, Widman G, Elger C, Scholkopf B, et al. cations. Springer; 2012.
Classifying eeg and ecog signals without subject training for fast [125] Gupta CN, Palaniappan R, Paramesran R. Exploiting the p300
bci implementation: comparison of nonparalyzed and completely paradigm for cognitive biometrics. Int J Cognitive Biometrics
paralyzed subjects. Neural Syst Rehabilitation Eng, IEEE Trans 2012;1(1):26–38.
2006;14(2):183–6. [126] Gao S, Wang Y, Gao X, Hong B. Visual and auditory brain-
[106] Sosa OP, Quijano Y, Doniz M, Chong-Quero J. BCI: A historical computer interfaces 2014.
analysis and technology comparison. In: Health Care Exchanges [127] Maye A, Zhang D, Wang Y, Gao S, Engel AK. Multimodal
(PAHCE), 2011 Pan American. IEEE; 2011. p. 205–09. brain-computer interfaces. Tsinghua Sci Technol 2011;16(2):
[107] <http://www.cyberknifestlouis.com/index.php?page=magnetoen- 133–9.
cephalography>, [accessed 24.09.14]. [128] Valbuena D, Volosyak I, Graser A. SBCI: Fast detection of
[108] <http://www.zmescience.com/research/studies/vegetative-state- steady-state visual evoked potentials. In: Engineering in
patients-communication-possible-fmri-eeg-042343/>, [accessed Medicine and Biology Society (EMBC), 2010 Annual
02.10.14]. International Conference of the IEEE. IEEE; 2010.
[109] Ayaz H, Shewokis PA, Bunce S, Onaral B. An optical brain p. 3966–69.
computer interface for environmental control. In: Engineering in [129] Fairclough SH, Gilleade K. Advances in physiological comput-
Medicine and Biology Society, EMBC, 2011 Annual ing. Springer; 2014.
International Conference of the IEEE. IEEE; 2011. p. 6327–330. [130] Muller-Putz G, Scherer R, Neuper C, Pfurtscheller G. Steady-
[110] George L, Lécuyer A et al. An overview of research on’pas- state somatosensory evoked potentials: suitable brain signals for
sive’brain-computer interfaces for implicit human-computer brain-computer interfaces? Neural Syst Rehabilitation Eng,
interaction. In: International Conference on Applied Bionics IEEE Trans 2006;14(1):30–7.
and Biomechanics ICABB 2010-Workshop W1’Brain-Computer [131] Zúquete A, Quintela B, Cunha JPS. Biometric authentication
Interfacing and Virtual Reality’; 2010. using brain responses to visual stimuli. In: Proceedings of the
[111] <http://www.biopac.com/fNIR200A-Data-Collection-Solution>, International Conference on Bio-inspired Systems and Signal
[accessed 02.10.14] . Processing; 2010. p. 103–12.
[132] Neela TK, Kahlon KS. A framework for authentication using implementations. In: Systems Man and Cybernetics (SMC),
fingerprint and electroencephalogram as biometrics modalities. 2010 IEEE International Conference on. IEEE; 2010, p.
Int J Comput Sci Manage Res 2012;1(1). 3861–66.
[133] Pfurtscheller G, Lopes da Silva FH. Event-related eeg/meg [145] Kim JJ, Hwang T, Kim M, Oh E, Hwangbo M, Kim M-K, Kim
synchronization and desynchronization: basic principles. Clin S-P. The effect of stimulus type and distance on neural control of
Neurophysiol 1999;110(11):1842–57. a smart tv. In: Neural Engineering (NER), 2013 6th
[134] Lotte F, Congedo M, Lécuyer A, Lamarche F, Arnaldi B, et al. International IEEE/EMBS Conference on. IEEE; 2013. p.
A review of classification algorithms for eeg-based brain- 1343–45.
computer interfaces. J Neural Eng 2007;4. [146] Ahmadi A, Jafari R, Hart J. Light-weight single trial eeg signal
[135] Panoulas KJ, Hadjileontiadis LJ, Panas SM. Brain–computer processing algorithms: Computational profiling for low power
interface (BCI): Types, processing perspectives and applications. design. In: Engineering in Medicine and Biology Society, EMBC,
In: Multimedia Services in Intelligent Environments. Springer; 2011 Annual International Conference of the IEEE. IEEE; 2011.
2010. p. 299–321. p. 4426–30.
[136] Samek W, Muller K-R, Kawanabe M, Vidaurre C. Brain- [147] Stewart AX, Nuthmann A, Sanguinetti G. Single-trial classifi-
computer interfacing in discriminative and stationary subspaces. cation of EEG in a visual object task using ica and machine
In: Engineering in Medicine and Biology Society (EMBC), 2012 learning. J Neurosci Meth 2014;228:1–14.
Annual International Conference of the IEEE. IEEE; 2012. p. [148] Oliveira I, Grigore O, Guimarães NM, Duarte C. Experiences in
2873–876. reading detection with EEG signals. In: Proceedings of the 2010
[137] Soria-Frisch A. A critical review on the usage of ensembles for ACM Symposium on Applied Computing. ACM; 2010. p. 1236–
bci. Towards practical brain-computer interfaces. Springer; 2013. 37.
[138] Xing S, McCardle R, Xie S. Reading the mind: The potential of [149] Huang D, Guan C, Ang KK, Zhang H, Pan Y. Asymmetric
electroencephalography in brain computer interfaces. In: spatial pattern for eeg-based emotion detection. In: Neural
Mechatronics and Machine Vision in Practice (M2VIP), 2012 Networks (IJCNN), The 2012 International Joint Conference on.
19th International Conference. IEEE; 2012. p. 275–80. IEEE; 2012. p. 1–7.
[139] Wang Y, Jung T-P. Improving brain-computer interfaces using [150] Jain A, Kim I, Gluckman BJ. Low cost electroencephalographic
independent component analysis. Towards practical brain–com- acquisition amplifier to serve as teaching and research tool. In:
puter interfaces. Springer; 2013. Engineering in Medicine and Biology Society, EMBC, 2011
[140] Makeig S, Kothe C, Mullen T, Bigdely-Shamlo N, Zhang Z, Annual International Conference of the IEEE. IEEE; 2011. p.
Kreutz-Delgado K. Evolving signal processing for brain-com- 1888–91.
puter interfaces. Proc IEEE 2012;100(13):1567–84. [151] von Bunau P, Meinecke FC, Scholler S, Mu?ller K-R. Finding
[141] Wu W, Gao X, Hong B, Gao S. Classifying single-trial eeg stationary brain sources in EEG data. In: Engineering in
during motor imagery by iterative spatio-spectral patterns Medicine and Biology Society (EMBC), 2010 Annual
learning (isspl). Biomed Eng, IEEE Trans 2008;55(6):1733–43. International Conference of the IEEE. IEEE; 2010. p. 2810–13.
[142] Brigham K, Kumar BV. Subject identification from electroen- [152] Sannelli C, Vidaurre C, Muller K-R, Blankertz B. Common
cephalogram (EEG) signals during imagined speech. In: P roc. spatial pattern patches: online evaluation on bci-naive users. In:
IEEE Fourth International Conference on Biometrics: Theory, Engineering in Medicine and Biology Society (EMBC), 2012
Applications and Systems (BTAS’lO); 2010. Annual International Conference of the IEEE. IEEE; 2012. p.
[143] Su F, Xia L, Cai A, Wu Y, Ma J. EEG-based personal 4744–47.
identification: from proof-of-concept to a practical system. In: [153] Li J, Wang Y, Zhang L, Jung T-P. Combining ERPs and EEG
Pattern Recognition (ICPR), 2010 20th International Conference spectral features for decoding intended movement direction. In:
on. IEEE; 2010. p. 3728–731. Engineering in Medicine and Biology Society (EMBC), 2012
[144] Su F, Xia L, Cai A, Ma J. Evaluation of recording factors in Annual International Conference of the IEEE. IEEE; 2012. p.
eeg-based personal identification: A vital step in real 1769–72.

Brain Computer Interfacing-Applications

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Brain Computer Interfacing-Applications

Uploaded by

Copyright:

Available Formats

Egyptian Informatics Journal (2015) 16, 213–230

Egyptian Informatics Journal

Brain computer interfacing: Applications

Received 20 December 2014; accepted 4 June 2015

Production and hosting by Elsevier

3.1.1. Prevention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215

1. Introduction On the other hand, some of BCI advantages for able-

Figure 1 BCI application ﬁelds.

Garcia-Molina et al. [13], the precise awareness of the current

The following subsections give a brief introduction to

Figure 3 BCI system.

Figure 4 Signal acquisition methods.

Figure 6 Intracortical acquistion.

Invasive recording methods implant electrodes under the scalp.

area. Besides, the body adaptation to the new object, which

Figure 8 MEG acquisition. Figure 10 fNIRS acquisition.

skull layer compared to electric ﬁelds. But this advantage does

5.2.2. Functional magnetic resonance imaging (fMRI)

Figure 12 The 10–20 international system.

Table 1 Summary of acquisition method characteristics.

using EEG and ECoG signals is presented in [105]. Also the

6. BCI electrical signals

6.2. Event related desynchronizationnsynchronization 7.1.2. Technical challenges

Table 2 Examples of BCI systems.

You might also like