ArticlePDF AvailableLiterature Review

Toward functioning and usable brain-computer interfaces (BCIs): A literature review

Abstract and Figures

The aim of this paper is to provide an exhaustive review of the literature about brain-computer interfaces (BCIs) that could be used with these paralysed patients. The electroencephalography (EEG) is the best candidate for the continuous use in the environment of patients' houses, due to its portability and ease of use. For this reason, the present paper will focus on this kind of BCI. Moreover, it is our aim to focus more on the patients, regarding their active role in the modulation of the brain activity. This leads to a differentiation between studies that use an active regulation and studies that use a non-active regulation. Relevant articles in the BCIs field were selected using MEDLINE and PsycINFO. Research through data banks produced 980 results, which were reduced to 127 after exclusion criteria selection. These references were divided in four categories, based on the use of active or non-active regulation, and on the event related potential used. In most of the examined works, the focus was on the development of systems and algorithms able to recognise and classify brain events. Although this kind of research is fundamental, a user-centred point of view was rarely adopted. [Box: see text].
Content may be subject to copyright.

Disability and Rehabilitation: Assistive Technology
2012
7
2
89
103
© 2012 Informa UK, Ltd.
10.3109/17483107.2011.589486
1748-3107
1748-3115
Disability and Rehabilitation: Assistive Technology, 2012; 7(2): 89–103
Copyright © 2012 Informa UK, Ltd.
ISSN 1748-3107 print/ISSN 1748-3115 online
DOI: 10.3109/17483107.2011.589486
Correspondence: Emanuele Pasqualotto, Institute of Medical Psychology and Behavioral Neurobiology, Eberhard-Karls-University, Gartenstr. 29,
D-72074 Tübingen, Germany. Tel: +49-7071-29-78388, Fax: +49-7071-29-5956. E-mail: pasqualotto.emanuele@gmail.com
(Accepted May 2011)
Purpose: The aim of this paper is to provide an exhaustive
review of the literature about brain–computer interfaces
(BCIs) that could be used with these paralysed patients. The
electroencephalography (EEG) is the best candidate for the
continuous use in the environment of patients’ houses, due
to its portability and ease of use. For this reason, the present
paper will focus on this kind of BCI. Moreover, it is our aim to
focus more on the patients, regarding their active role in the
modulation of the brain activity. This leads to a differentiation
between studies that use an active regulation and studies that
use a non-active regulation. Method: Relevant articles in the
BCIs field were selected using MEDLINE and PsycINFO. Results:
Research through data banks produced 980 results, which were
reduced to 127 after exclusion criteria selection. These references
were divided in four categories, based on the use of active or
non-active regulation, and on the event related potential used.
Conclusions: In most of the examined works, the focus was on
the development of systems and algorithms able to recognise
and classify brain events. Although this kind of research is
fundamental, a user-centred point of view was rarely adopted.
Keywords: Brain–computer interfaces, assistive technology,
ALS, P300, VEP, SSVEP, sensorimotor rhythm
Background
In 1929, Hans Berger published his rst report on the electro-
encephalography (EEG), a system that was supposedly able
to measure synaptic potential dierences from the scalp [1].
At that time he could have not imagined that 40 years later,
electroencephalography would be used to control machines
with brain impulses [2]. More than a century has passed since
the discovery of EEG [1], and more than 30 years have gone
by from the attempt to create an interface able to establish a
direct connection between brain and machine, that we actu-
ally call brain–computer interface (BCI). In the last 15 years,
a remarkable acceleration of studies occurred in this eld. A
BCI is a control and/or communication system in which the
user’s commands and messages are not dependent on common
brain-motor periphery communication channels: informa-
tion is not conveyed directly from nerves and muscles, and
neuromuscular activity is not necessary for the production of
the activity needed to convey the message [3]. anks to better
technologies, dierent systems enabling a connection between
brain and machine (e.g. a computer) have been developed.
According to literature, it is possible to identify two dier-
ent approaches: invasive and non-invasive [4].
e invasive approach is characterised by intracranial
recordings of electrical activity, performed directly on single
neurons or on neural assemblies. e aim of this approach
is to re-establish interrupted connections: for example, using
voluntary motor signals in order to control prostheses in pa-
tients with paralysed limbs.
REVIEW ARTICLE
Toward functioning and usable brain–computer interfaces (BCIs):
A literature review
Emanuele Pasqualotto1,2, Stefano Federici2,3 & Marta Olivetti Belardinelli2,4
1Institute of Medical Psychology and Behavioral Neurobiology, Eberhard-Karls-University, Tübingen, Germany, 2Interuniversity
Center for Research on Cognitive Processing in Natural and Articial Systems (ECoNA), Sapienza University of Rome, Rome, Italy,
3Department of Human and Education Sciences, University of Perugia, Perugia, Italy, and 4Department of Psychology, Sapienza
University of Rome, Rome, Italy
Locked-in syndrome (LIS) is a condition of severe or •
complete motor paralysis, where the intellect is still
intact.
A brain–computer interface (BCI) is a direct con-•
nection between the brain and a machine.
In the past 15 years, several dierent BCI prototypes •
have been developed for research purposes.
BCIs could be used for rehabilitation to restore com-•
munication and movement in patients with severe
and multiple disabilities.
Implications for Rehabilitation
Disabil Rehabil Assist Technol Downloaded from informahealthcare.com by Prof. Stefano Federici on 02/07/12
For personal use only.
90 E. Pasqualotto et al.
Disability and Rehabilitation: Assistive Technology
It is important to point out that progress in this research
area has slowed down due to both technical and ethical prob-
lems. Nonetheless, many studies have been carried out, also
by testing primates, leading up to relevant results related to
humans as well (for a review, see [3,4]).
e non-invasive approach exploits the brain activity re-
corded using mainly the EEG in order to allow the control of
a personal computer or of a peripheral device. is kind of
approach is especially used to enable paralysed patients to de-
velop a communication channel [5]. Although the EEG is the
main non-invasive technique, in the last years some attempts
were made using other techniques, as the real time functional
magnetic resonance imaging (rt-fMRI [6–9]), the magneto-
encephalography (MEG [10]), and the near-infrared spec-
troscopy (NIRS [11]). Compared to the invasive approach,
the non-invasive one does not expose patients to potentially
dangerous surgical operations to implant the electrodes.
According to Lebedev and Nicolelis [4], however, the non-
invasive approach has also the disadvantage of a limitation in
data transmission, which is inadequate for prosthesis control
(e.g. a mechanical arm or leg), since it would only allow the
representation of two alternative signals (of the yes/no, on/
o type). ese authors acknowledge that a non-invasive BCI
could allow other practical solutions (as the control of a cur-
sor on a personal computer or the activation of augmentative/
alternative communication systems, etc.). Other authors do
not recognise the limits of the non-invasive modality: in fact,
with the support of a wide share of the scientic community,
they encourage eort in this research eld [12].
Objectives
e technical nature of some critical points in the BCI eld
[5] has partially delayed the researches, but its continuous
evolution encouraged the scientic community to clarify the
state of the art several times, through congresses [3], reviews
on scientic journals [5,12–16] and special editions entirely
devoted to the subject. Many reviews undertake topics that
are related to BCIs, both from a historical and a technical
point of view, analysing details and solutions that are neces-
sary for the development of interfaces. As observed by various
authors, the human brain has an intense electrical and chemi-
cal activity, partially characterised by specic patterns, which
emerge at a specic time in specic brain areas: these patterns
are repeatable under well-dened environmental conditions
[17], and it is possible to develop very dierent communica-
tion and control systems. Indeed, in order to be implied in
BCIs, dierent brain patterns have been selected.
Considering the aim of BCIs to provide a communica-
tion channel for paralysed patients, the aim of this review is
to provide an exhaustive review of the literature about BCIs
that could be used with these patients. Although, as reported
previously, new imaging techniques were used, the EEG is the
best candidate for the continuous use in the environment of
patients’ houses, due to its portability and ease of use. More-
over, the dierences in dimensions and costs of these ma-
chines, and the recent development of the dry EEG [18–20],
lead to hoping in a future development of systems that could
better match the needs of paralysed patients. For this reason,
it is our aim to focus only on EEG-based BCIs.
Moreover, it is our aim to focus more on the patients, re-
garding their active role in the modulation of the brain ac-
tivity. is leads to a dierentiation between studies that use
an active regulation and studies that use a non-active regu-
lation. We will examine and describe the studies regarding
the basis of the cognitive and sensory components of human
EEG. Some authors have identied the components that have
been already used in literature 5,17: visual evoked potentials
(VEPs), slow cortical potentials (SCPs), the P300 component
of evoked potentials, and the sensorimotor rhythms (SMRs).
Method
Strategy for the bibliographical research
Relevant articles in the BCI eld were selected using MED-
LINE and PsycINFO, two search engines chosen from the data
bank list provided by EBSCO for the research and processing
of scientic literature information. ese databases were cho-
sen in order to nd references with a psychological and bio-
medical valence. Nevertheless, by using these two databases,
it is possible to nd articles covering from psychological to
more engineering and computer science aspects. e selec-
tion of keywords was carried out considering the terms used
in the most recent reviews: following the articles’ references,
the keywords adopted more frequently in this research eld
have been tracked: ‘brain–computer interface(s)’, ‘brain ma-
chine interface(s)’, ‘direct brain interface(s)’, ‘man machine
interface(s)’, ‘mental prosthesis’. Aerwards, the possibility to
increase the number of bibliographical references was exam-
ined, using a keyword generator. A very well known generator
is provided free of charge on Internet by the Google research
engine. e keywords obtained with this method, however,
did not bring any new results.
Article selection
We performed a research through the selected data banks
without using any timeframe limitation. is research
produced 980 results (see Figure 1). Since references were
taken from dierent databases, it became necessary to
standardise some of the information contained in the bib-
liographical les: keywords, authors and journals. From a
rst descriptive analysis, it was possible to point out that
27 authors (the ones with more than 15 publications) out
of more than 2000 covered 36% of the publications, while
39% of the references were published in 11 journals. ese
results highlight the fact that most studies were carried out
by a small number of research teams. ereby, we decided to
classify publications considering the dierent cognitive and
sensory components and dividing them according to dier-
ent research groups.
Following the classication tips used by Beverina and
colleagues [17], we selected all works containing explicit
references to electroencephalographic cognitive and sensory
components, such as beta and mu rhythms, P300, SCP, VEP
and SSVEP (Steady-State Visually Evoked Potentials) either in
the title or in the abstract.
Disabil Rehabil Assist Technol Downloaded from informahealthcare.com by Prof. Stefano Federici on 02/07/12
For personal use only.
Toward functioning and usable BCIs 91
Copyright ©  Informa UK Ltd.
We excluded from our search proceedings (36 references),
articles that were published in non-English speaking journals
(4 references) and articles that were not focusing on the BCI
topic (42 references). Hence, the number of publications was
reduced to 127 (see Tabl e I).
BCIs
A direct communication between brain and computer could
be very useful to people suering from neurodegenerative
and muscle disease, such as Guillain-Barré syndrome, amyo-
trophic lateral sclerosis (ALS), brainstem stroke, and spinal
cord or traumatic brain injury. ese diseases may lead to
severe or complete motor paralysis. ese patients gradually
lose the ability to use the common communication systems,
which usually require using speech or muscles (e.g. for key
or switch pressing). When this happens, people become
unable to communicate needs and thoughts to the surround-
ing environment. is condition is dened as locked-in syn-
drome (LIS [21]). While the disease leads to a de-eerentation
of the motor system [22] and a consequent condition of
paralysis of voluntary muscles, the intellect is still intact. is
condition imprisons patients in a body that does not react to
the brain’s nervous impulses, which are sometimes still pres-
ent. Since life quality is strongly based on the ability to com-
municate with the external world, great eort has been made
to provide these patients with systems that can interface the
brain activity with a computer.
e BCIs that have been developed so far dier sig-
nicantly. e main dierence concerns the invasivity and
Figure 1. e research produced 980 results, of which aer the exclusion criteria only 127 remained.
Disabil Rehabil Assist Technol Downloaded from informahealthcare.com by Prof. Stefano Federici on 02/07/12
For personal use only.
92 E. Pasqualotto et al.
Disability and Rehabilitation: Assistive Technology
non-invasivity criteria. Another criterion adopted to classify
interfaces considers the dependence and independence from
the brain communication pathway of muscles and peripheral
nerves [5].
A dependent BCI is an indirect and alternative method for
identifying messages that are conveyed by the brain through
common communication pathways. For example, gaze direc-
tion can be detected directly through the eyes, using an eye
Table I. e complete list of references divided into categories.
Sensorimotor P300 VEP SCP
Bai et al. (2008) Allison and Pineda (2003) Allison et al. (2008) Bostanov (2004)
Brunner et al. (2005) Bayliss 120031 Allison et al. (2010) Hinterberger et al. (2003)
Chen et al. (2009) Bayliss and Ballard (2000) Allison et al. (2010) Hinterberger et al. (2004)
Fatourechi et al. (2008) Bayliss et al. (2004) Bakardjian et al. (2010) Hinterberger et al. (2004)
Huang et al. (2009) Beverina et al. (2003) Beverina et al. (2003) Hinterberger et al. (2005)
Keinrath et al. (2006) Blanchard and Blankertz (2004) Bin et al. (2009) Hinterberger et al. (2005)
Krusienski et al. (2007) Bostanov (2004) Cecotti (2010) Iversen et al. (2008)
Lemm et al. (2004) Bostanov and Kotchoubey (2006) Cheng et al. (2002) Iversen et al. (2008)
Li et al. (2010) Brouwer and van Erp (2010) Friman et al. (2007) Kübler et al. (2001)
McFarland and Wolpaw (2003) Brunner et al. (2010) Gao et al. (2003) Kübler et al. (2004)
McFarland et al. (2003) Citi et al. (2008) Gollee et al. (2010) Mensh et al. (2004)
McFarland et al. (2006) Citi et al. (2010) Guo et al. (2008) Neumann and Birbaumer (2003)
Miner et al. (1998) Dal Seno et al. (2009) Gupta and Palaniappan (2007) Neumann et al. (2003)
Müller-Putz et al. (2007) Dal Seno et al. (2010) Hong et al. (2009) Neumann et al. (2004)
Pfurtscheller et al. (2006) Donchin et al. (2000) Jones et al. (2003) Pham et al. (2005)
Pineda et al. (2003) Farwell and Donchin (1988) Kelly et al. (2005)
Schalk et al. (2000) Finke et al. (2009) Kelly et al. (2005)
Wolpaw et al. (1991) Furdea et al. (2009) Lee et al. (2006)
Wolpaw et al. (2000) Glassman (2005) Lee et al. (2008)
Guger et al. (2009) Lee et al. (2010)
Halder et al. (2010) Lin et al. (2006)
Homann et al. (2008) Lin et al. (2007)
Jansen et al. (2004) Liu et al. (2010)
Kleih et al (2010) Luo and Sullivan (2010)
Klobassa et al. (2009) Martinez et al. (2007)
Krusienski et al. (2008) Middendorf et al. (2000)
Kübler et al. (2009) Müller et al. (2000)
Lenhardt et al. (2008) Müller-Putz and Pfurtscheller (2008)
Li et al. (2010) Müller-Putz et al. (2005)
Lu et al. (2009) Müller-Putz et al. (2006)
Martens et al. (2009) Müller-Putz et al. (2008)
Nam et al. (2010) Nielsen et al. (2006)
Nijboer et al. (2008) Parini et al. (2009)
Nijboer et al. (2010) Pfurtscheller et al. (2010)
Piccione et al. (2006) Pfurtscheller et al. (2010)
Piccione et al. (2008) Shyu et al. (2010)
Rakotomamonjy and Guigue (2008) Srihari Mukesh et al. (2006)
Rivet et al. (2009) Trejo et al. (2006)
Salimi-Khorshidi et al. (2008) Wang et al. (2006)
Salvaris and Sepulveda (2009) Wu and Yao (2008)
Salvaris and Sepulveda (2010) Wu et al. (2008)
Schreuder et al. (2010) Yoshimura and Itakura (2008)
Sellers and Donchin (2006) Yoshimura and Itakura (2009)
Sellers et al. (2006) Zhang et al. (2010)
Sellers et al. (2006)
Serby et al. (2005)
Silvoni et al. (2009)
Takano et al. (2009)
ulasidas et al. (2006)
Townsend et al. (2010)
Vaughan et al. (2006)
Zhang et al. (2008)
Disabil Rehabil Assist Technol Downloaded from informahealthcare.com by Prof. Stefano Federici on 02/07/12
For personal use only.
Toward functioning and usable BCIs 93
Copyright ©  Informa UK Ltd.
tracker, or indirectly through VEPs: in both cases, it is neces-
sary for muscles to aect the eyes, in order to shi gazes.
In an independent BCI, information is not generated in
order to be conveyed through these communication channels:
this is the case of the P300 response, elicited by brain circuits
and located therein.
e focus of these classications is never user-centred: in
both cases, the perspective is strictly medical. Relatively to in-
vasive interfaces, testing electrode implants in the human brain
requires specic knowledge, which is not always available even
in multi-disciplinary teams. Moreover, the risks related to such
procedures oen restrain their accomplishment. Relatively
to the dependence on muscles, the focus is on functionality,
which is absent in the case of degenerative diseases.
is review suggests a particular perspective, based on the
dependence or independence of the user from the conscious
regulation of the brain activity used by the interface.
Some systems require users to modulate intentionally some
aspects of their own EEG activity. Such systems are based
on voluntarily learning of the necessary operations and the
biofeedback procedures, in order to regulate dierent aspects
of brain signals [23]. is is what happens, for example, in
the system developed in the Wadsworth Center by Wolpaw’s
research team [5,24], which uses the mu rhythm coming from
the sensorimotor cortex, or in the one developed in Tübingen
by Birbaumer’s team [12,13], which uses SCPs. Other systems
use some properties with spontaneous occurrences of brain
electrical activity [23]. Such occurrences are recognised di-
rectly by real-time classication algorithms, without any in-
tentional learning of the procedures. e work of Donchin’s
team [25,26] can be considered as an example of this category,
with an interface system that exploits P300, which is an event-
related potential (ERP) evoked by a rare or signicant event.
e following review will therefore present dierent stud-
ies according to a specic reference frame: dierent works will
be presented distinguishing between the ones that require the
user’s voluntary regulation of the brain activity and the ones
in which modulation is not necessary. Each section will be
subdivided on the basis of the electroencephalographic com-
ponent involved. Lastly, all works will be grouped considering
the dierent research teams.
Systems depending on voluntary modulation
SCPs
SCPs are generally dened as negative or positive synaptic po-
larisations in EEG, or as magnetic eld variations in MEG [27].
Such changes usually last from a minimum of 300 ms to several
seconds. Polarisation variation originates from the depolari-
sation of the apical dendritic tree, located in the upper cortical
layer, which is provoked by a series of synchronous activations
mainly resulting from thalamus-cortical aerent nerves. is
system regulates the excitation (with negative potentials) and
the inhibition (with positive potentials) of cortical circuits, by
means of a threshold mechanism.
It has been demonstrated that people can learn to volun-
tarily regulate these potentials by using immediate feedbacks
as a positive reinforcement [28]. is evidence represented
the basis for the development of Tübingen’s system, called
‘thought translation device’ (TTD). e development of the
TTD required several years: building and testing a learning
protocol that enabled patients to learn how to regulate SCPs,
and consequently use them to communicate, required intense
work [29,30].
Moreover, it was necessary to develop a protocol for the
assessment of cognitive functions and communication abili-
ties in locked-in patients [31]. e TTD has been developed
mainly for clinical use and has been tested with a wide num-
ber of patients with ALS. e authors used a SCP-based BCI
as a tool to assess cognitive functions in paralysed patients
[32,33]. Moreover, the testing on clinical subjects showed
that the TTD enables people to retrieve basic communication
abilities [34].
e EEG is registered by placing the electrodes on the
vertex (Cz) and on both mastoids (A1, A2), according to the
international 10–20 system [35]. Aer being ltered and cor-
rected by the electro-oculographic (EOG) artefacts, EEG is
shown to the user on a screen. Despite the fact that the visual
feedback is the main kind of feedback used, it is not the only
one that was tested. In dierent studies, the research team
tried to use the same system with an auditory feedback, or
combining auditory and visual feedbacks, with a fairly good
eect [36,37].
e system registers the voltage level: by increasing or de-
creasing such level, the user can shi the cursor presented on
the screen through vertical movements, so that the upper and
lower bounds of the screen can be reached. In an EEG-based
system, it is necessary to have a specic device that classies
the dierent components of the EEG as precisely as possible,
in order to improve the subjects’ performance. For this rea-
son, the study and implementation of dierent algorithms for
the classication of the EEG was necessary [38].
is challenge has been taken up by many researchers, who
have tested dierent algorithms providing very interesting re-
sults. Bostanov [39] presented a method, called t-CWT, for
the detection of the characteristics of biological signals, based
on a mathematical procedure named Continuous Wavelet
Transform (CWT) and on Student’s t-distribution. Similarly,
Mensh, Werfel, and Seung [40] used a system that completed
the SCP measurement with an estimate of the high frequency
activity (Gamma waves), usually associated with intentional
states of attention: using a linear procedure, they obtained
better EEG classications, compared to the ones that only
used SCP activity.
By using the TTD, the training required to learn how to
regulate the brain activity can last several months, and is not
always successful. For this reason, Tübingen’s research team
studied patient’s dierent behaviours while using the TTD,
in order to understand which strategies people use for the
regulation of their own brain activity [34], to identify predict-
ability factors that could be useful during training [41], and
to understand if such mechanisms could become automatic
with long-lasting practice [42]. e best eects were obtained
in the study of performance predictors: with a narrow group
of ALS patients, Neumann and Birbaumer [43] demonstrated
that initial performance in the learning stage can be used to
Disabil Rehabil Assist Technol Downloaded from informahealthcare.com by Prof. Stefano Federici on 02/07/12
For personal use only.
94 E. Pasqualotto et al.
Disability and Rehabilitation: Assistive Technology
make predictions about future performance. Although this can
be very useful to improve and personalise the learning stage
with subjects that are not able to regulate their brain activity,
the authors pointed out possible ethical implications of this
factor, highlighting that such information should not be used
as a criterion to include or exclude patients in the following
learning stages [41]. Once users have reached a sucient level
of accuracy (estimated around 75%), there is another stage
in which a ‘Language Support Program’ (LSP) could be used.
e LSP is an application that enables users to communicate
with the surrounding environment. is soware adopts the
same strategy as the learning programme. e user selects a
letter or a combination of letters through a series of dichoto-
mic choices. In each phase, choice is between two groups of
one or more letters: beginning with two exact halves of the
alphabet, the user will nally get to select one single letter, by
halving in each phase the group of letters selected previously.
Selection speed, just as learning quickness, is not very high: in
several occasions, subjects had to be trained for months to be
able to control SCPs, not always with brilliant results. It has
been estimated, indeed, that the LSP allows writing up to 36
words per hour [44]. Although this is not an exciting result,
it could be considered as an important step in the develop-
ment of a communication system that can enable users with
diseases such as ALS to interact.
e studies performed by Birbaumer’s team are broad-
spectrum: not only do they aim to develop an interfacing
system between brain and computer, but they also investigate
the nature of SCPs. Several studies that used in vivo visualisa-
tion techniques such as functional magnetic resonance imag-
ing (fMRI) have investigated the brain structures underlying
the development of SCPs: a positive SCP correlates with an
increase of activity in the basal ganglia, which is visible as a
greater blood inux, while a negative potential correlates with
an increase of activity in the thalamus [7,22].
SMRs
Wolpaw’s research team at the Wadsworth Center in Albany
(NY, USA) and Pfurtscheller’s research team in Graz (Austria)
demonstrated through a series of experiments [24,45] that
both healthy subjects and paralysed patients could achieve a
voluntary control of SMR. Such control takes place in both
hemispheres, near the rolandic cortex, and can be developed
by imagining movements [44].
By registering the brain activity in adult subjects, it is pos-
sible to detect a cortical activity coming from the sensory and
motor areas: this activity varies between 8 and 12 Hz, and is
visible when subjects are not engaged in processing sensory
information or in the production of movement. Sensorimo-
tor rhythms are produced during inactivity, and are probably
caused by thalamus-cortical circuits. Such rhythms have dif-
ferent names, according to their origin: mu rhythm when
focused over the somatosensory or motor cortex, visual alpha
rhythm when focused over the visual cortex [46].
Dierent analyses demonstrate that mu activity is not uni-
tary, and can be divided according to dierent factors, such as
origin, frequency, and relationship with concomitant sensory
inputs or motor outputs. e whole variability of mu rhythm
is normally associated with beta rhythm (18–26 Hz): although
some beta rhythms are harmonics (multiple frequencies of the
basic frequency) of the mu rhythm, it is possible to consider
them dierently, according to the topography and/or latency,
making them as independent characteristics of EEG.
Several factors suggest that sensorimotor rhythms have
the ideal features to be used in EEG-based communication
systems. e SMRs are associated to brain areas that are con-
nected with the eerent motor pathways. Moreover, the in-
crease and decrease of the rhythms do not require a real body
movement: motor imagination is sucient for the developing
of such potentials [5].
Wolpaw and McFarland carried out two parallel research
projects: on one side, they worked on the development and
improvement of the BCI system, rening EEG recognition
and control; on the other side, they tried to provide users with
a system that could allow them both to communicate with
simple responses [47] and to use word-processor soware
[48]. By means of the BCI2000 system, developed mainly in
the Albany Wadsworth Center [49,50], people with or without
motor disabilities can learn to control the width of sensorimo-
tor rhythms, and to subsequently use this ability to control a
visible cursor on a computer screen, shiing in one or two
dimensions [51].
Compared to a system that uses SCPs, the learning stage
is shorter and in a few weeks users can acquire a signicant
control. During the learning stage, the importance of motor
imagination decreases: at the beginning, it is important for
users to imagine movements, in order to gain rhythm con-
trol, but in the end such mechanism becomes automatic, and
thinking about performance details becomes unnecessary.
For the coding of sensorimotor rhythm signals, a specic
mathematical algorithm is used: a linear equation transforms
the signal coming from one or more scalp locations into the
cursor signal, 10 times per second. To control the movement
of the cursor, it is necessary to place electrodes on one or
two scalp locations. Nonetheless, data from 64 locations is
registered, in order to carry out the following analyses that
increase the accuracy of operations [46]. Such analyses al-
low dening changes in the whole EEG topography, which
are associated to the target position on the screen; besides,
they enable the detection of artefacts related to the central
nervous system (CNS), electromyographic (EMG) activity
from scalp and facial muscles and EOG activity from eye
movements and eye blinks that could obscure the EEG activ-
ity used by a BCI system [52,53]. It has been shown that when
users are learning how to control their EEG activity, an initial
contamination by target-related EMG artefacts can remark-
ably aect performance. When users have not yet gained a
full EEG control, EMG artefacts can be exploited to move
the cursor toward the target. For this reason, careful and
complete topographic and spectral analyses are necessary
during the entire training, in order to detect possible EMG
contaminations [54].
By directing the cursor, users can answer simple questions
(with an estimated accuracy of 95%), to select items from a
menu and to learn how to use mechanical prostheses, which
allow them to grab objects [3]. Besides, it is possible to learn
Disabil Rehabil Assist Technol Downloaded from informahealthcare.com by Prof. Stefano Federici on 02/07/12
For personal use only.
Toward functioning and usable BCIs 95
Copyright ©  Informa UK Ltd.
how to move the cursor in two distinct dimensions, control-
ling two dierent channels with the sensorimotor rhythms.
A satisfactory communication requires an adequate bit
rate (and an adequate writing speed). In order to optimise
the bit rate, Wolpaw’s team monitored users’ performances.
By measuring the accuracy and the amount of transferred in-
formation (calculated as the number of operations performed
every minute by the user) they have veried that increasing
complexity corresponds to decreasing performance, identify-
ing the best parameters for a successful EEG-based commu-
nication [55,56]. Although this is not the only parameter [57],
it is widely used.
Research carried out with this BCI system has focused
mainly on the denition of sensorimotor features (topo-
graphic, spectral and temporal) and on the optimisation of
the user-interface interaction. Some studies have analysed
the possibility to add new features, which could provide new
information, to the EEG features that are already in use: target
selection errors are associated with a positive potential located
on the vertex, which could be used for the recognition and
deletion of selection errors [58].
e following step was to improve the selection of math-
ematical constants in the equations that transform the EEG
signal into the signal that controls the cursor [5].
Another improvement was obtained by verifying that cur-
sor monitoring can be more eective when the EEG signal
is translated by means of a weighted combination of rhythm
widths, instead of using a simple linear equation. Moreover,
each of the factors’ weight is calculated using a regression algo-
rithm based on the user’s previous performance [59]. Finally,
a new method for the ltering of EEG data was proposed: rst
in order to have the exact topography of the mu rhythm, it is
necessary to record the EEG signal to extract its characteris-
tics; then, using these data, a parameterised model is created
and used to compare cortical signals in real time [60].
A similar system is the one based on the event-related
synchronisation (ERS) and de-synchronisation (ERD) of mu
and beta rhythms, developed by Pfurtscheller’s research team
in Graz. is kind of brain activity is dened as an increase
(ERS) or decrease (ERD) of intensity according to a reference
interval [61]. Research carried out by this group is mostly
focused on the attempt to dierentiate the EEG components
associated with motor imagery of simple movements [62]. In
the protocol used by Pfurtscheller, the user undertakes several
stages, beginning with the learning stage. e user is enabled
to nd the best paradigm for the motor imagery control. He/
she has to imagine one of four simple actions (right hand
movement, le hand movement, foot movement or tongue
movement), while the EEG coming from electrodes located
on the sensorimotor cortex are registered and analysed. e
analyses carried on the recorded data are used to identify the
signal’s characteristics, with a procedure that adopts dier-
ent classication methods [63]. Discrimination between the
four imagery tasks is based on the EEG classication during
the single trials. It has been shown that the classication can
improve if both ERD and ERS patterns are used in at least
one or two tasks [64]. In the next stage, the interface uses the
classication system to transform the user’s motor imagery in
an output shown on a screen as a feedback: such output can be
continuous, as the extension of a luminous bar or the move-
ment of a cursor, or discrete, as the selection of a letter. Aer
a few sessions, about 90% of the people are able to use the
system, with an accuracy level that reaches 90% [5].
Many of the present studies in Graz aim to investigate the
nature of event-related motor synchronisation and de-syn-
chronisation. Researchers are trying to understand how much
this type of brain activity is similar in active motor actions,
passive motor actions and imagined motor actions [65,66],
and which are the dierences between healthy subjects and
subjects with motor disability [67,68]. During the last years,
research focusing on mu rhythm increased. is interest is re-
lated both to its possible use in BCIs [69] and to the relevance
that this brain activity had in studies regarding the function-
ing of mirror-neurons [70].
For example, in the Cognitive Neuroscience Laboratory of
the University of California in San Diego, directed by Jaime
Pineda, research is addressed to the discovery of brain activity
regulation mechanisms: since sensorimotor rhythms can be
inuenced by an auto-produced movement, or by an observed
movement, or even by an imagined movement, dierences
between these factors have been investigated, showing that
rhythm intensity diers according to the condition in which
it is elicited [71]. Besides, regulation mechanisms for each
hemisphere have been examined, investigating their ability to
regulate rhythms separately, by creating a double communica-
tion channel [69].
Many researchers have faced the critical problem of devel-
oping an eective algorithm for the EEG signal classication.
is complication is due to the need to limit false positives
(Type I error) and false negatives (Type II errors) in the clas-
sication. When the system does not recognise the produced
EEG event, the interface does not respond to a command con-
veyed by the user, generating a false negative. When happen-
ing in the learning stages, this implies a time prolongation,
while in the advanced stages it could involve even frustration.
When the system recognises any EEG event as the desired one,
it responds to commands that were not actually conveyed,
generating a false positive: besides facing the same problems
as with false negatives, the user would have to correct pos-
sible mistakes. Schalk and colleagues [58] proposed to use an
error-evoked potential for detecting and avoiding errors. is
potential requires no additional time and could improve the
speed and the accuracy of a BCI system. For this reason, many
researchers have focused on the assessment of alternative
solutions: some have employed new systems to lter signals,
based on origins’ spatial locations [72]; others used Bayes-
ian probabilistic models that accelerate the algorithm deci-
sion time [73]; some authors used model analyses, through
mathematical models and neural networks [74,75]; nally,
some authors used features extracted from dierent neuro-
logical phenomena (movement-related potentials, changes in
the power of mu rhythms and changes in the power of beta
rhythms) to detect an intentional control command [76].
e idea of using dierent phenomena inspired the concept
of hybrid BCI proposed by Pfurtscheller [77]. A hybrid BCI is
composed of two BCIs, or one BCI and another system, and
Disabil Rehabil Assist Technol Downloaded from informahealthcare.com by Prof. Stefano Federici on 02/07/12
For personal use only.
96 E. Pasqualotto et al.
Disability and Rehabilitation: Assistive Technology
could hypothetically have a better performance. Relying on
more than one system, a hybrid BCI can operate two systems
simultaneously or sequentially, whereby the rst system can
act as a ‘brain switch’ or as a ‘selector’. In a brain switch system,
one system may be used to activate a device when needed for
control, and the other system to actually control it [78]. In a
selector system, both are used to control the device [79,80].
Systems depending on non-voluntary
modulation
VEPs
Using computer-generated visual stimulation and sophisti-
cated signal processing, Vidal [81] showed that single-trial
VEPs could provide a communication channel that allows a
human to control the movement of a cursor through a two-
dimensional maze.
A VEP is a brain response to a sensory stimulation in the
subject’s visual eld: in traditional experimental paradigms,
which are also used in the clinical domain to assess the condi-
tions of the visual system, the stimuli are lights or ashing
chessboards presented on a screen.
Sutter [82] developed a system called brain response in-
terface (BRI), which uses VEPs. Subjects were shown a video
screen with 64 symbols located on an 8 × 8 grid. In order to
identify the response to the stimulus in the subjects’ EEG, the
grid was divided into two groups, each with a dierent alter-
nation (40/70 times per second) of red and green. e groups
could have the same brightness for the same time, or have a
dierent one according to predetermined patterns. Although
this system led to a good performance with healthy subjects,
when subjects had a disability that caused an uncontrollable
head and neck muscular activity, the EMG activity prevented
a reliable measurement of VEPs, reducing the systems per-
formance.
Despite the fact that the VEP is not widely used, several
systems were tested, using ashing [83], in motion [84–86],
and transient [87,88] VEPs.
SSVEPs
A particular kind of VEPs aroused the interest of several
authors. SSVEPs are natural responses to visual stimulations
with stationary periodic oscillations. When the eye retina is
excited by a stimulus that varies between 3.5 and 75 Hz, the
brain generates an electrical activity of the same frequency as
the stimulus, or one of its multiples. SSVEPs are considered
very useful in research, both for their excellent signal-noise
ratio and for their resistance to artefacts. Furthermore, some
SSVEP-based BCI approaches may not depend on gaze con-
trol [89]. e general idea is to use ickering stimuli to convey
user commands in order to induce SSVEPs: when the user
wants to select one of the commands, he/she focuses on one of
the stimuli. en, by analysing the generated SSVEP, the BCI
system tries to infer which stimulus the user selected [90].
e air force research laboratory research group in Dayton
(OH, USA) proposed a system that determines gaze direction
with the support of SSVEPs [91]. Several virtual buttons are
presented on a computer screen. Each button radiates a light,
with a dierent ashing frequency. When the user looks at
one of these buttons, the system determines the visual cortex’s
response frequency: when this frequency coincides with the
frequency of a button, the system deduces what the user wants
to select. Although this kind of system highly depends on the
user’s ability to control gaze direction, it is independent from
his/her ability to control the brain response that takes place
naturally. Subsequently, the same research group assessed the
accuracy and velocity of the system while reaching targets of
dierent sizes and at dierent distances [23], according to
Fitts’s law model [92,93]. By comparing BCI data with data
obtained while subjects performed the same tasks with a
mouse, the authors have shown that the general performance
levels are lower and still quite distant from the ones that can
be obtained with more traditional methods.
e group of Shangkai Gao at Tsinghua University in
Beijing (China) developed a similar system. e system uses
SSVEPs to control a virtual telephone keypad formed by 12
buttons enlightened with dierent frequencies [94]. e users’
task was to dial a phone number: most users achieved this
goal with very high bit rates and aer a very brief learning
stage, although researchers complained that performance de-
creased when users reduced their attention. In order to prove
the validity of using SSVEPs, the authors adapted their system
to a computer-independent use [95]. Using a panel with light-
emitting diodes (LEDs) programmed to switch on at dierent
frequencies, they have shown that such a system could be used
for environmental control. e same functions that are mainly
used to manage word-processing applications can be used to
convey input signals to a remote control system at home. In
another attempt they tried to use covert attention shis in-
stead of changing gaze direction. By using two superimposed
surfaces with two sets of dots with dierent colours rotating
in opposite directions, they were able to elicit distinguishable
SSVEPs [96]. One of the advantages of using SSVEP-based
BCIs is that only a few electrodes are required for signal regis-
tration: as a matter of fact, one electrode located on the visual
cortex (Oz electrode) could be sucient for the classication
of SSVEP events [97]. In spite of the simplicity of this sys-
tem, the importance of having a good classication algorithm
encouraged the authors to test dierent solutions. Although
most subjects had a very high bit rate, a reliable number was
not able to use the system in a protable way. For this reason,
researchers of the Tsinghua University group assessed which
could be the best combination of communication channel lo-
cation, stimulus frequency and selection speed [98]. Lastly, the
use of the canonical correlation analysis (CCA) on the EEG
signals of SSVEPs was assessed. e CCA is a statistical mul-
tivariate method that is used when correlations between two
data sets are hypothesised. e group of Tsinghua University
developed an algorithm for frequency recognition, based on
this statistical method. e method was later compared with
a more traditional one, the Fast Fourier Transform (FFT): the
analysis showed that the CCA-based approach obtains signi-
cantly higher recognitions compared to the results obtained
with FFT [99,100]. Using this method it is possible to develop
a reliable online BCI [101]. Friman, Volosyak & Gräser [102]
developed another valid algorithm for signal detection. With
Disabil Rehabil Assist Technol Downloaded from informahealthcare.com by Prof. Stefano Federici on 02/07/12
For personal use only.
Toward functioning and usable BCIs 97
Copyright ©  Informa UK Ltd.
their system, preliminary procedures such as noise estimate
or extraction of the signal’s features are not necessary: by
automatically looking for the best electrode combination, the
algorithm uses very short segments of the EEG recordings,
getting to classify SSVEPs with an 84% accuracy.
Results obtained on visual potentials with multiple LEDs
[94] encouraged Pfurtscheller’s group to focus on SSVEP-
based BCIs [103]. e response to each stimulation frequency
is usually studied in the frequency domain: if in the EEG regis-
tration the response to this stimulation frequency (or pattern)
is present, then such frequency has been produced. In order
to decrease the variability between subjects, and according to
their potentiality to produce the same response, the stimula-
tion frequencies are selected. Although it is easier to produce
harmonically correlated frequencies, these are avoided, since
they produce similar responses [94]. For example, a stimula-
tion of 6 Hz produces peaks at 6 Hz (rst harmonic) and at 12
Hz (second harmonic), although also by using stimulation at
12 Hz produce the peak at 12 Hz. For this reason it cannot be
used alone.
Previous studies have been developed on the basis of the
recognition of the rst harmonic only [91], or of the rst two
harmonics [94,95]. e study which was carried out in Graz,
showed that using the rst three harmonics could produce
a signicant increase in the classication system, compared
to the use of only the rst, or the rst two harmonics [103].
Moreover, applying a screening procedure, which could be
used to nd the optimal electrode positions for bipolar deri-
vations, could increase the accuracy [104]. Srihari Mukesh,
Jaganathan, and Reddy [105] have obtained independently a
similar result: aiming to increase the number of possible se-
lections, they doubled the subjects’ stimulation. In this way, it
is possible to increase the number of commands using a com-
bination of stimuli with dierent frequencies. For example,
with a combination of only three frequencies, it is possible to
obtain six dierent stimuli that correspond to as many com-
mands. Another interesting alternative use of SSVEPs is the
one developed by Shyu, Lee, Liu, and Sie [106]. eir study
proposes a dual-frequency icker to induce SSVEPs: each
dual-frequency icker comprises two ashing LEDs with dif-
ferent ickering frequencies.
Pfurtscheller’s research group has studied the use of evoked
potentials at steady conditions, also assessing the utility of a
type of potential which diers from the visual potential: the
steady-state somatosensory evoked potential (SSSEP). Results
obtained with these potentials showed the possibility to under-
take a new path. As a matter of fact, subjects demonstrated
70–80% of accuracy with paradigms that were similar to the
ones used for SSVEPs [107]. Another important result was
to show that an SSVEP-based BCI could be used to control a
two-axes electrical hand prosthesis [108].
At University College in Dublin (Ireland), Reilly’s group
investigated the possibility to use a signal with an alpha band
frequency (8–14 Hz). e alpha band has a high signal-noise
ratio, which allows a better discrimination of the required
signal. According to an oine comparison between registra-
tions, the accuracy in target selection is better when alpha
band and SSVEPs are used together, compared to when they
are used individually [109]. is study shows that although
SSVEPs are a natural phenomenon, they are oen related to
visuospatial attention mechanisms. As highlighted by other
authors [94], SSVEPs are reliable involuntary responses, but
their modulation requires a control from spatial attention
mechanisms. Such mechanisms are normally used in every-
day life, but in the particular BCI contexts they can require a
conscious control [110]: an example of this is represented by
the ink Pointer, developed by NASA, which allows a user to
surf on a map [111].
e importance of attention is also evident in the systems
proposed by the group of San Camillo Hospital in Venice
(Italy). Aer having presented a system based on P300 and
one of the SSVEPs, the group noticed how the main eort
was involved in focusing on the user’s side of the interface,
making it more pleasant and less distracting. e importance
of classication algorithms diminished compared to interface
usability. As a matter of fact, the authors observe that it is not as
important as the user’s feeling toward the interfaces. e more
the user is aware of the task proposed by the BCI, the more the
brain activity will be recognisable for the machine. Since they
consider the user’s emotional involvement when he/she uses
the interface an important issue, they recommend, in future
research guidelines, a more careful use of the interface with a
greater involvement of the user in his/her activity [17].
Although promising, this approach still cannot be com-
pletely reliable and cannot be used with everyone, since some
approaches and parameters are less eective with some users
than with others [112,113]. Several aspects of current SSVEP-
based BCI systems require improvement, specically in rela-
tion to speed, user variability and ease of use [101]. is fact
leads several research groups to develop dierent methods
and approaches. Wu and Yao investigated the spectrum dier-
ences of three kinds of ickers and the dierences in SSVEPs
evoked by three dierent stimulators [114] and a new param-
eter named stability coecient (SC) to measure the stability
of a frequency [115]. Many authors tested and developed dif-
ferent methods to extract EEG features and classify them, as
the common spatial patterns (CSPs) method [116], and the
stimulus-locked inter-trace correlation (SLIC) method [20].
Moreover, BCI systems were developed to control objects on
a screen as a small car [117], the movement of a cursor in four
directions [118], and a functional electrical stimulation (FES)
system to allow the user to control stimulation settings and
parameters [119].
P300
e P300 is the most studied ERP. It is usually evoked by a
rare or task-relevant event [120]. e P300 is a positive deec-
tion of the EEG, usually associated with the update of current
memory traces. Its latency varies between 200 and 700 ms:
the peak, at around 300 ms, implies the processing of a simple
stimulus while a greater latency seems to reect the time that
is necessary to make a decision about the stimulus. e P300
is usually elicited by using an oddball paradigm [121,122]:
low probability deviant stimuli in a series of standard higher
probability stimuli when the deviants have to be attended and
actively answered. Such oddball paradigms reliably yield P300
Disabil Rehabil Assist Technol Downloaded from informahealthcare.com by Prof. Stefano Federici on 02/07/12
For personal use only.
98 E. Pasqualotto et al.
Disability and Rehabilitation: Assistive Technology
responses with a parietocentral scalp distribution to target
compared to standard stimuli irrespective of stimulus (visual,
auditory, somatosensory) or response (button press, counting)
modality [123]. e central to parietal areas are the most ap-
propriate for recording P300 [122], although the anatomical
sources that elicit P300 are oen in the hippocampus, and are
related to the content of the memory trace and of the stimulus
[124]. e positive polarity of this potential shows an inhibi-
tory function: it probably blocks the competing information
process related to new stimuli. P300 is widely used in the
clinical eld: to assess disorders in attentional processing, in
cognitive disorders, in aging, and recently, in research on lie
detection [44]. A very debatable topic is habituation to P300.
Donchin exploited in his BCI system the reliability and valid-
ity of P300: even aer long time intervals, the P300 amplitude,
as a response to the stimulus during the BCI training, is not
subjected to habituation; nonetheless, P300 is absent only
in a small number of subjects [44]. In contrast, according to
other researchers, all ERPs change in response to condition-
ing: in the long term, P300 could be subjected to habituation,
decreasing the performance with the BCI, or could become
larger, increasing performance [5]. is problem is not insol-
uble. Indeed, an appropriate translation algorithm could keep
track of possible changes, minimising their eects.
Contrary to other interfaces, BCIs that use P300 do not
require users to learn and to auto-regulate brain responses.
High spelling accuracy can be achieved with the P300 BCI
system using approximately ve minutes of training data
[125]. Nonetheless, they do not require a feedback, and the
brief latency of P300 allows a faster letter selection, compared
to any other BCI system. Conversely, an eective P300-based
BCI requires some psychophysiological conditions that are
not easily found in completely paralysed patients, such as the
ability to spell internally and the integrity of the visual and
attentional systems [44,126,127]. Moreover, motivational fac-
tors could signicantly aect BCI performance [128,129].
Donchin was one of the rst researchers to use P300 in a
BCI, developing a system called P300 Speller [25,26,120]. In
the typical paradigm built in his studies, the user is located
in front of a 6 × 6 matrix of letters, numbers, symbols and
other commands. Every row and column ashes with a pre-
arranged inter-stimulus interval: in order to select the desired
choice, the user should count how many times the row and the
column that contain the choice ashes. During the trial, pari-
etal EEG is registered, and subsequently the mean response in
each row and column is calculated, obtaining the P300 width
for each of the possible choices. e P300 is prominent only in
responses that are evoked by the desired choice, and the BCI
uses this eect in order to determine the user’s intention. e
ecacy of the P300 Speller as a communication device was
successfully assessed on eight ALS patients [130].
Dierent algorithms for the recognition of the desired
choice have been assessed, testing analysis performed in real
time a posteriori: the Step-Wise Linear Discriminant Analy-
sis (SWLDA) and the discrete wavelet transform (DWT) are
the analyses that have been used most. e SWLDA and the
DWT suggest that the actual P300-based BCIs could lead to
the communication of one word (of about ve letters) per
minute. e possibility of further improvement related to
velocity is not excluded. Besides, systems that use auditory
stimuli to allow people with visual disorders are also being
tested, although such systems are slow and still need to be
improved [37,131–136].
Bayliss has developed a very original system. Aer verify-
ing with a virtual driving system the strength of the P300, a
virtual interface was developed to test an environment control
system [137,138]. By means of a monitor and a helmet for vir-
tual reality, subjects were presented with a virtual apartment,
where they could control some of the objects (a lamp, a stereo
and a television). Initial successes led the authors to test the
system with dierent classication algorithms. Following the
suggestion of Schalk, Wolpaw, McFarland, and Pfurtscheller
[58], they tried to use a system that exploited the error-related
negativity (ERN) to automatically correct errors, in order to
increase the system’s eectiveness [139].
In the attempt to develop a portable BCI system, Wolpaw’s
team studied the use of P300 [140]. e authors successfully
used the same presentation modality as the matrices used by
Farwell and Donchin [25]. Starting from a similar work by
Allison and Pineda [141], the authors investigated the size of
the matrices and the inter-stimulus interval eects, obtaining
similar results: the width of P300 increases with the decrease
of the target’s presentation probability, that is with larger ma-
trices. Besides, it has been demonstrated that a shorter inter-
stimulus interval allows greater response accuracy and higher
bit rates [142].
Although the P300 was used in one of the rst BCIs, ini-
tially it did not raise great interest, probably because of the
diculty in the development of a real-time classication algo-
rithm. Designing a reliable and accurate BCI is now one of the
most challenging elds in BCI. To achieve this goal, dierent
methods have been adopted.
In the last years, several researchers are developing math-
ematical algorithms that are probably more suitable for this
aim [143–148].
In the same work in which the eectiveness of t-CWT on
SCPs was assessed, Bostanov [39] tested its use on P300 as
well. Aer the preliminary examinations, already seen with
SCPs, the author broadened the study of the algorithm, test-
ing it both with the oddball paradigm (that elicits P300) and
the semantic priming paradigm (eliciting another cognitive
component: N400): the t-CWT, compared to DWT and to
principal component analysis (PCA), leads to better results
[149]. Other authors have developed a threshold algorithm
for real time processing [150]. eir system has 70% accuracy.
Such a percentage is low, if compared to other systems, but
the algorithm has the advantage of being resistant to dier-
ent stimuli characteristics: type of stimulus, inter-stimulus
interval, ltering method, and target occurrence probability.
is last factor contradicts what was oen claimed in litera-
ture: larger matrices should elicit a greater P300 [141,142].
In order to assess the algorithm’s functioning, Donchin’s
P300 Speller was used in two dierent studies, varying the
information-processing stage only. In the rst study, a family
of algorithms for supervised learning, named support vector
machine (SVM), was adopted [151]. In the second study the
Disabil Rehabil Assist Technol Downloaded from informahealthcare.com by Prof. Stefano Federici on 02/07/12
For personal use only.
Toward functioning and usable BCIs 99
Copyright ©  Informa UK Ltd.
independent component analysis (ICA) was used. e ICA is
a method for dividing a multivariate signal into some additive
components [152]. e rst study emphasises the possibility
to diminish the time required for training (which is actually
20 minutes); the second study highlights that although the re-
sults obtained with ICA are only slightly better than Donchin’s
oine analyses (ICA: 5.45 symbols per minute, 92.1% accu-
racy; Donchin: 4.8, 90%), compared to online analyses, they
are signicantly better (ICA: 4.5, 79.5%; Donchin: 4.8, 56%).
e group of San Camillo Hospital in Venice developed a
BCI that uses ICA for P300 detection [17]. In their system,
ICA was used for detecting the peak, which is later analysed
by a neural network and may subsequently be classied as
P300. is allows a rapid processing that can be used in real
time, guaranteeing accuracy in the control of a cursor on a
monitor [153]. Using a SWLDA classier, Krusienski, Sellers,
McFarland, Vaughan, and Wolpaw [154] compared among
spatial channel selection, reference, decimation, and number
of model features, to determine optimal settings for the P300
Speller data.
Dierent modality among the visual were explored, as were
the tactile [155]. e feasibility of an auditory BCI has been
investigated in several studies using dierent EEG input sig-
nals, e.g. P300 evoked potential or SMR. Sellers and Donchin
[120] tested healthy volunteers and patients with ALS with a
four-choice P300 BCI. Patients were presented either visually,
auditory, or both types of stimuli with the words ‘yes’, ‘no, ‘pass
and ‘end. e patients’ task was to focus their attention on
either ‘yes’ or ‘no’. e authors showed that a target probability
of 25% was low enough to reliably elicit a P300. Hill [131]
classied P300 evoked responses that occurred aer two audi-
tory stimuli streams presented simultaneously. To choose one
of two possible targets, the participant focused on either one
of the streams. In some recent studies the P300 Speller matrix
used in visual studies [120] was transferred to the auditory
modality. In two experiments the visual matrix was compared
to an auditory one where spoken numbers were used to iden-
tify each row and column [134,135]. In order to select an item,
the user focused on the corresponding numbers, which are
presented auditory. Users of the auditory speller achieved an
average accuracy of 65%, while the users presented with the
visual speller were able to select characters with an average
accuracy of 95%. Koblassa [156] used a similar design with
natural sounds instead of words. A group of healthy subjects
received only auditory stimuli while another group received
simultaneous auditory and visual stimuli in initial sessions
aer which the visual stimuli were systematically removed.
e rst group achieved an online accuracy of 59%, and the
second group achieved 68%. Halder and colleagues [136]
explored which physical properties (loudness, pitch or loca-
tion) of a tone could give the best discrimination between two
targets in an auditory BCI. Using a three-stimulus oddball
paradigm [157], they found that pitch had the best ecacy for
most participants, although for six patients out of 20, location
was the same or better. Localisation of sounds in space was
recently done using ve dierent loudspeakers as stimulus’
sources on the frontal plane testing three dierent inter-stim-
ulus intervals (1000 ms, 300 ms, 175 ms [158]). Moreover, in
another condition all stimuli were presented through a single
speaker (front), leaving the pitch properties of the stimulus
the only discriminating cue. e oine analysis showed that
all the healthy subjects in the spatial conditions were able to
reach a selection score over 70%, while in the non-spatial con-
dition most subjects showed no markedly dierent traces for
non-target and target stimuli.
e P300-based BCI was successfully assessed using healthy
[26,127,135,152,159] and ALS patients [120,129,134,153].
Several BCI applications were developed by using the P300:
an on-screen mouse [160], a MindGame [161], an Internet
browser [162,163], and a brain painter [164]. Moreover, in or-
der to better elicit the P300, dierent kinds of changes in the
stimulation were tested: the eect of matrix size [142], stimuli
based on apparent motion [165], changes to the dimensions
of the symbols, the distance between the symbols and the
colours used [166], the eects of dierent selective attention
tasks [167], the eectiveness of green/blue icker matrices as
visual stimuli [168], background noise and interface colour
contrast [169], dierent visual elicitation paradigms [170],
and a checkerboard paradigm [171].
Conclusions
e present review describes dierent systems that aim to
create a direct connection between brain and computer. First,
we described the systems that require subjects to intentionally
regulate specic characteristics of their own electrical brain
activity. e TTD, developed by Birbaumer and colleagues,
which exploits SCPs in order to generate binary responses,
was also described. en we described the system developed
by Wolpaw and colleagues and the system developed by
Pfurtscheller and colleagues, which use sensorimotor rhythm
regulation to control the movement of a cursor.
We then described systems that do not depend on inten-
tional regulation, which use natural brain responses to exter-
nal stimulation. is is the case of systems that use luminous
LEDs or monitors to convey impulses at dierent frequencies,
associated to dierent buttons, like the ones developed by
Middendorf and colleagues or by Gao and colleagues. Last, we
described the P300 Speller, created by Donchin and inspiring
several other researchers, that uses the P300 response in order
to select a letter located in a matrix.
BCIs are novel technology that needs further develop-
ment. BCIs could be used to restore some human functions,
as communication or environmental control. Nonetheless,
many of these systems allow an information transfer that
is comparable neither to verbal communication nor to a
keyboard.
In most of the examined works, this main problem led to a
research based on the development of systems and algorithms
that are able to recognise and classify brain events in the most
rapid and accurate way. Although this kind of research is
fundamental, a user-centred point of view was rarely adopted.
Only few researchers seemed to realise the importance of
maintaining attention on the task: very oen, tasks required
a slow and repetitious scanning of a virtual keyboard, which
can easily lead to an attentional breakdown. In most cases, the
Disabil Rehabil Assist Technol Downloaded from informahealthcare.com by Prof. Stefano Federici on 02/07/12
For personal use only.
100 E. Pasqualotto et al.
Disability and Rehabilitation: Assistive Technology
development of user-centred interfaces was not considered: a
few attempts were made by researchers who tried to develop
virtual reality systems, but without making an ergonomic as-
sessment of the interfaces. Some authors explored the eect
of motivation using BCI systems, nding that performance
might be aected by a lack of motivation.
Greater attention to the cognitive aspects of the interface
could lead to better results. By guaranteeing better structuring
of the information, it should be possible to increase both the
bit rate and the perception of a synchronous communication
with the surrounding environment. By involving the users
with more pleasant and eective interfaces, it should be also
possible to obtain more satisfying and motivating conditions.
Besides, using methodologies of usability assessment that
have been extensively studied in literature could provide a
new support for the improvement of BCIs.
Acknowledgments
e authors would like to thank G. Liberati for her helpful
support in reviewing the manuscript and the anonymous
reviewers for their suggestions, which helped to improve the
quality of this paper.
Declaration of interest
e authors report no conicts of interest.
References
1. Swartz BE, Goldensohn ES. Timeline of the history of EEG and associ-
ated elds. Electroencephalogr Clin Neurophysiol 1998;106:173–176.
2. Vidal JJ. Toward direct brain-computer communication. Annu Rev
Biophys Bioeng 1973;2:157–180.
3. Wolpaw JR, Birbaumer N, Heetderks WJ, McFarland DJ, Peckham
PH, Schalk G, Donchin E, et al. Brain-computer interface technology:
a review of the rst international meeting. IEEE Trans Rehabil Eng
2000;8:164–173.
4. Lebedev MA, Nicolelis MA. Brain-machine interfaces: past, present
and future. Trends Neurosci 2006;29:536–546.
5. Wolpaw JR, Birbaumer N, McFarland DJ, Pfurtscheller G, Vaughan
TM. Brain-computer interfaces for communication and control. Clin
Neurophysiol 2002;113:767–791.
6. Weiskopf N, Veit R, Erb M, Mathiak K, Grodd W, Goebel R, Birbaum-
er N. Physiological self-regulation of regional brain activity using
real-time functional magnetic resonance imaging (fMRI): methodol-
ogy and exemplary data. Neuroimage 2003;19:577–586.
7. Hinterberger T, Weiskopf N, Veit R, Wilhelm B, Betta E, Birbaumer
N. An EEG-driven brain-computer interface combined with func-
tional magnetic resonance imaging (fMRI). IEEE Trans Biomed Eng
2004;51:971–974.
8. Weiskopf N, Scharnowski F, Veit R, Goebel R, Birbaumer N, Mathiak
K. Self-regulation of local brain activity using real-time functional mag-
netic resonance imaging (fMRI). J Physiol Paris 2004;98:357–373.
9. Sitaram R, Caria A, Veit R, Gaber T, Rota G, Kuebler A, Birbaumer
N. FMRI brain-computer interface: a tool for neuroscientic research
and treatment. Comput Intell Neurosci 2007;25487.
10. Mellinger J, Schalk G, Braun C, Preissl H, Rosenstiel W, Birbaumer N,
Kübler A. An MEG-based brain-computer interface (BCI). Neuroim-
age 2007;36:581–593.
11. Sitaram R, Zhang H, Guan C, ulasidas M, Hoshi Y, Ishikawa A,
Shimizu K, Birbaumer N. Temporal classication of multichannel
near-infrared spectroscopy signals of motor imagery for developing a
brain-computer interface. Neuroimage 2007;34:1416–1427.
12. Birbaumer N. Brain-computer-interface research: coming of age. Clin
Neurophysiol 2006;117:479–483.
13. Birbaumer N. Breaking the silence: brain-computer interfaces (BCI) for
communication and motor control. Psychophysiology 2006;43:517–532.
14. Pfurtscheller G, Müller-Putz GR, Schlögl A, Graimann B, Scherer R,
Leeb R, Brunner C, et al. 15 years of BCI research at Graz University
of Technology: current projects. IEEE Trans Neural Syst Rehabil Eng
2006;14:205–210.
15. Daly JJ, Wolpaw JR. Brain-computer interfaces in neurological reha-
bilitation. Lancet Neurol 2008;7:1032–1043.
16. Mak JN, Wolpaw JR. Clinical Applications of Brain-Computer In-
terfaces: Current State and Future Prospects. IEEE Rev Biomed Eng
2009;2:187–199.
17. Beverina F, Palmas G, Silvoni S, Piccione F, Giove S. User adaptive
BCIs: SSVEP and P300 based interfaces. Psychn J 2003;1:331–354.
18. Popescu F, Fazli S, Badower Y, Blankertz B, Müller KR. Single trial
classication of motor imagination using 6 dry EEG electrodes. PLoS
ONE 2007;2:e637.
19. Gargiulo G, Calvo RA, Bifulco P, Cesarelli M, Jin C, Mohamed A, van
Schaik A. A new EEG recording system for passive dry electrodes.
Clin Neurophysiol 2010;121:686–693.
20. Luo A, Sullivan TJ. A user-friendly SSVEP-based brain-computer in-
terface using a time-domain classier. J Neural Eng 2010;7:26010.
21. Plum F, Posner JB. e diagnosis of stupor and coma. Philadelphia:
FA Davis; 1966.
22. Hinterberger T, Veit R, Wilhelm B, Weiskopf N, Vatine JJ, Birbaumer
N. Neuronal mechanisms underlying control of a brain-computer in-
terface. Eur J Neurosci 2005;21:3169–3181.
23. Jones KS, Middendorf M, McMillan GR, Calhoun G, Warm J. Com-
paring mouse and steady-state visual evoked response-based control.
Interact Comput 2003;15:603–621.
24. Wolpaw JR, McFarland DJ, Vaughan TM. Brain-computer inter-
face research at the Wadsworth Center. IEEE Trans Rehabil Eng
2000;8:222–226.
25. Farwell LA, Donchin E. Talking o the top of your head: toward a
mental prosthesis utilizing event-related brain potentials. Electroen-
cephalogr Clin Neurophysiol 1988;70:510–523.
26. Donchin E, Spencer KM, Wijesinghe R. e mental prosthesis: assess-
ing the speed of a P300-based brain-computer interface. IEEE Trans
Rehabil Eng 2000;8:174–179.
27. Birbaumer N. Slow cortical potentials: Plasticity, operant control, and
behavioral eects. Neuroscientist 1999;5:74.
28. Birbaumer N, Ghanayim N, Hinterberger T, Iversen I, Kotchoubey
B, Kübler A, Perelmouter J, et al. A spelling device for the paralysed.
Nature 1999;398:297–298.
29. Kübler A, Neumann N, Kaiser J, Kotchoubey B, Hinterberger T, Bir-
baumer NP. Brain-computer communication: self-regulation of slow
cortical potentials for verbal communication. Arch Phys Med Rehabil
2001;82:1533–1539.
30. Neumann N, Kübler A. Training locked-in patients: a challenge for
the use of brain-computer interfaces. IEEE Trans Neural Syst Rehabil
Eng 2003;11:169–172.
31. Hinterberger T, Birbaumer N, Flor H. Assessment of cognitive func-
tion and communication ability in a completely locked-in patient.
Neurology 2005;64:1307–1308.
32. Iversen I, Ghanayim N, Kübler A, Neumann N, Birbaumer N, Kaiser J.
Conditional associative learning examined in a paralyzed patient with
amyotrophic lateral sclerosis using brain-computer interface technol-
ogy. Behav Brain Funct 2008;4:53.
33. Iversen IH, Ghanayim N, Kübler A, Neumann N, Birbaumer N, Kai-
ser J. A brain-computer interface tool to assess cognitive functions in
completely paralyzed patients with amyotrophic lateral sclerosis. Clin
Neurophysiol 2008;119:2214–2223.
34. Neumann N, Kübler A, Kaiser J, Hinterberger T, Birbaumer N. Con-
scious perception of brain states: mental strategies for brain-computer
communication. Neuropsychologia 2003;41:1028–1036.
35. Klem GH, Lüders HO, Jasper HH, Elger C. e ten-twenty electrode
system of the International Federation. e International Federation
of Clinical Neurophysiology. Electroencephalogr Clin Neurophysiol
Suppl 1999;52:3–6.
36. Hinterberger T, Neumann N, Pham M, Kübler A, Grether A, Hof-
mayer N, Wilhelm B, et al. A multimodal brain-based feedback and
communication system. Exp Brain Res 2004;154:521–526.
37. Pham M, Hinterberger T, Neumann N, Kübler A, Hofmayer N, Gre-
ther A, Wilhelm B, et al. An auditory brain-computer interface based
on the self-regulation of slow cortical potentials. Neurorehabil Neural
Repair 2005;19:206–218.
38. Hinterberger T, Kübler A, Kaiser J, Neumann N, Birbaumer N. A
brain-computer interface (BCI) for the locked-in: comparison of
dierent EEG classications for the thought translation device. Clin
Neurophysiol 2003;114:416–425.
Disabil Rehabil Assist Technol Downloaded from informahealthcare.com by Prof. Stefano Federici on 02/07/12
For personal use only.
Toward functioning and usable BCIs 101
Copyright ©  Informa UK Ltd.
39. Bostanov V. BCI Competition 2003–Data sets Ib and IIb: feature
extraction from event-related brain potentials with the continuous
wavelet transform and the t-value scalogram. IEEE Trans Biomed Eng
2004;51:1057–1061.
40. Mensh BD, Werfel J, Seung HS. BCI Competition 2003–Data set Ia:
combining gamma-band power with slow cortical potentials to im-
prove single-trial classication of electroencephalographic signals.
IEEE Trans Biomed Eng 2004;51:1052–1056.
41. Kübler A, Neumann N, Wilhelm B, Hinterberger T, Birbaumer N.
Predictability of Brain-Computer Communication. J Psychophysiol
2004;18:121–129.
42. Neumann N, Hinterberger T, Kaiser J, Leins U, Birbaumer N, Kübler
A. Automatic processing of self-regulation of slow cortical potentials:
evidence from brain-computer communication in paralysed patients.
Clin Neurophysiol 2004;115:628–635.
43. Neumann N, Birbaumer N. Predictors of successful self control dur-
ing brain-computer communication. J Neurol Neurosurg Psychiatr
2003;74:1117–1121.
44. Birbaumer N, Cohen LG. Brain-computer interfaces: communica-
tion and restoration of movement in paralysis. J Physiol (Lond)
2007;579:621–636.
45. Pfurtscheller G, Guger C, Müller G, Krausz G, Neuper C. Brain
oscillations control hand orthosis in a tetraplegic. Neurosci Lett
2000;292:211–214.
46. Wolpaw JR, McFarland DJ, Vaughan TM, Schalk G. e Wadsworth
Center brain-computer interface (BCI) research and development
program. IEEE Trans Neural Syst Rehabil Eng 2003;11:204–207.
47. Wolpaw JR, McFarland DJ, Neat GW, Forneris CA. An EEG-based
brain-computer interface for cursor control. Electroencephalogr Clin
Neurophysiol 1991;78:252–259.
48. Miner LA, McFarland DJ, Wolpaw JR. Answering questions with an
electroencephalogram-based brain-computer interface. Arch Phys
Med Rehabil 1998;79:1029–1033.
49. Schalk G, McFarland DJ, Hinterberger T, Birbaumer N, Wolpaw JR.
BCI2000: a general-purpose brain-computer interface (BCI) system.
IEEE Trans Biomed Eng 2004;51:1034–1043.
50. Mellinger J, Schalk G. BCI 2000: A general-purpose soware platform
for BCI research. In: Dornhege G, del R. Millán J, Hinterberger T,
McFarland DJ, Müller KR, editors. Towards Brain-Computer Inter-
facing. Cambridge: MIT Press; 2007.
51. McFarland DJ, Krusienski DJ, Wolpaw JR. Brain-computer interface
signal processing at the Wadsworth Center: mu and sensorimotor
beta rhythms. Prog Brain Res 2006;159:411–419.
52. Goncharova II, McFarland DJ, Vaughan TM, Wolpaw JR. EMG con-
tamination of EEG: spectral and topographical characteristics. Clin
Neurophysiol 2003;114:1580–1593.
53. McFarland DJ, Leowicz AT, Wolpaw JR. Design and operation of
an EEG-based brain-computer interface with digital signal processing
technology. Behav Res Meth Instrum Comput 1997;29:337–345.
54. McFarland DJ, Sarnacki WA, Vaughan TM, Wolpaw JR. Brain-com-
puter interface (BCI) operation: signal and noise during early training
sessions. Clin Neurophysiol 2005;116:56–62.
55. McFarland DJ, Wolpaw JR. EEG-based communication and con-
trol: speed-accuracy relationships. Appl Psychophysiol Biofeedback
2003;28:217–231.
56. McFarland DJ, Sarnacki WA, Wolpaw JR. Brain-computer interface
(BCI) operation: optimizing information transfer rates. Biol Psychol
2003;63:237–251.
57. Dal Seno B, Matteucci M, Mainardi LT. e utility metric: a novel
method to assess the overall performance of discrete brain-computer
interfaces. IEEE Trans Neural Syst Rehabil Eng 2010;18:20–28.
58. Schalk G, Wolpaw JR, McFarland DJ, Pfurtscheller G. EEG-based
communication: presence of an error potential. Clin Neurophysiol
2000;111:2138–2144.
59. McFarland DJ, Wolpaw JR. Sensorimotor rhythm-based brain-com-
puter interface (BCI): feature selection by regression improves perfor-
mance. IEEE Trans Neural Syst Rehabil Eng 2005;13:372–379.
60. Krusienski DJ, Schalk G, McFarland DJ, Wolpaw JR. A mu-rhythm
matched lter for continuous control of a brain-computer interface.
IEEE Trans Biomed Eng 2007;54:273–280.
61. Brunner C, Graimann B, Huggins JE, Levine SP, Pfurtscheller G.
Phase relationships between dierent subdural electrode recordings
in man. Neurosci Lett 2005;375:69–74.
62. Pfurtscheller G, Müller GR, Pfurtscheller J, Gerner HJ, Rupp R.
‘ought’–control of functional electrical stimulation to restore hand
grasp in a patient with tetraplegia. Neurosci Lett 2003;351:33–36.
63. Pfurtscheller G, Neuper C, Guger C, Harkam W, Ramoser H,
Schlögl A, Obermaier B, Pregenzer M. Current trends in Graz
Brain-Computer Interface (BCI) research. IEEE Trans Rehabil Eng
2000;8:216–219.
64. Pfurtscheller G, Brunner C, Schlögl A, Lopes da Silva FH. Mu rhythm
(de)synchronization and EEG single-trial classication of dierent
motor imagery tasks. Neuroimage 2006;31:153–159.
65. Bai O, Lin P, Vorbach S, Floeter MK, Hattori N, Hallett M. A high
performance sensorimotor beta rhythm-based brain-computer in-
terface associated with human natural motor behavior. J Neural Eng
2008;5:24–35.
66. Huang D, Lin P, Fei DY, Chen X, Bai O. Decoding human motor ac-
tivity from EEG single trials for a discrete two-dimensional cursor
control. J Neural Eng 2009;6:046005.
67. Keinrath C, Wriessnegger S, Müller-Putz GR, Pfurtscheller G. Post-
movement beta synchronization aer kinesthetic illusion, active and
passive movements. Int J Psychophysiol 2006;62:321–327.
68. Müller-Putz GR, Zimmermann D, Graimann B, Nestinger K,
Korisek G, Pfurtscheller G. Event-related beta EEG-changes during
passive and attempted foot movements in paraplegic patients. Brain
Res 2007;1137:84–91.
69. Pineda JA, Silverman DS, Vankov A, Hestenes J. Learning to control
brain rhythms: making a brain-computer interface possible. IEEE
Trans Neural Syst Rehabil Eng 2003;11:181–184.
70. Rizzolatti G, Craighero L. e mirror-neuron system. Annu Rev
Neurosci 2004;27:169–192.
71. Pineda JA, Allison BZ, Vankov A. e eects of self-movement, ob-
servation, and imagination on mu rhythms and readiness potentials
(RP’s): toward a brain-computer interface (BCI). IEEE Trans Rehabil
Eng 2000;8:219–222.
72. Cheng M, Jia W, Gao X, Gao S, Yang F. Mu rhythm-based cursor con-
trol: an oine analysis. Clin Neurophysiol 2004;115:745–751.
73. Lemm S, Schäfer C, Curio G. BCI Competition 2003 – Data set
III: probabilistic modeling of sensorimotor mu rhythms for clas-
sication of imaginary hand movements. IEEE Trans Biomed Eng
2004;51:1077–1080.
74. Chen CW, Lin CC, Ju MS. Detecting movement-related EEG change
by wavelet decomposition-based neural networks trained with single
thumb movement. Clin Neurophysiol 2007;118:802–814.
75. Chen CW, Ju MS, Sun YN, Lin CC. Model analyses of visual biofeed-
back training for EEG-based brain-computer interface. J Comput
Neurosci 2009;27:357–368.
76. Fatourechi M, Ward RK, Birch GE. A self-paced brain-computer inter-
face system with a low false positive rate. J Neural Eng 2008;5:9–23.
77. Pfurtscheller G, Allison BZ, Brunner C, Bauernfeind G, Solis-Escal-
ante T, Scherer R, Zander TO, et al. e hybrid BCI. Front Neurosci
2010;4:30.
78. Pfurtscheller G, Solis-Escalante T, Ortner R, Linortner P, Müller-Putz
GR. Self-paced operation of an SSVEP-Based orthosis with and with-
out an imagery-based “brain switch:” a feasibility study towards a hy-
brid BCI. IEEE Trans Neural Syst Rehabil Eng 2010;18:409–414.
79. Allison BZ, Brunner C, Kaiser V, Müller-Putz GR, Neuper C, Pfurt-
scheller G. Toward a hybrid brain-computer interface based on imag-
ined movement and visual attention. J Neural Eng 2010;7:26007.
80. Li Y, Long J, Yu T, Yu Z, Wang C, Zhang H, Guan C. An EEG-based
BCI system for 2-D cursor control by combining Mu/Beta rhythm and
P300 potential. IEEE Trans Biomed Eng 2010;57:2495–2505.
81. Vidal JJ. Real-Time Detection of Brain Events in EEG. Proc IEEE
1977;65:633.
82. Sutter EE. e brain response interface: communication through visual-
ly-induced electrical brain responses. J Microcomp App 1992;15:31–45.
83. Lee PL, Hsieh JC, Wu CH, Shyu KK, Wu YT. Brain–computer inter-
face using ash onset and oset visual evoked potentials. Clin Neuro-
physiol 2008;119:605–616.
84. Guo F, Hong B, Gao X, Gao S. A brain-computer interface using mo-
tion-onset visual evoked potential. J Neural Eng 2008;5:477–485.
85. Hong B, Guo F, Liu T, Gao X, Gao S. N200-speller using motion-onset
visual response. Clin Neurophysiol 2009;120:1658–1666.
86. Liu T, Goldberg L, Gao S, Hong B. An online brain-computer interface us-
ing non-ashing visual evoked potentials. J Neural Eng 2010;7:036003.
87. Yoshimura N, Itakura N. Study on transient VEP-based brain-com-
puter interface using non-direct gazed visual stimuli. Electromyogr
Clin Neurophysiol 2008;48:43–51.
88. Yoshimura N, Itakura N. A transient VEP-based real-time brain-com-
puter interface using non-direct gazed visual stimuli. Electromyogr
Clin Neurophysiol 2009;49:323–335.
89. Allison BZ, McFarland DJ, Schalk G, Zheng SD, Jackson MM, Wol-
paw JR. Towards an independent brain-computer interface using
steady state visual evoked potentials. Clin Neurophysiol 2008;119:
399–408.
Disabil Rehabil Assist Technol Downloaded from informahealthcare.com by Prof. Stefano Federici on 02/07/12
For personal use only.
102 E. Pasqualotto et al.
Disability and Rehabilitation: Assistive Technology
90. Vialatte FB, Maurice M, Dauwels J, Cichocki A. Steady-state visually
evoked potentials: focus on essential paradigms and future perspec-
tives. Prog Neurobiol 2010;90:418–438.
91. Middendorf M, McMillan G, Calhoun G, Jones KS. Brain-computer
interfaces based on the steady-state visual-evoked response. IEEE
Trans Rehabil Eng 2000;8:211–214.
92. Fitts PM. e information capacity of the human motor system in con-
trolling the amplitude of movement. J Exp Psychol 1954;47:381–391.
93. Fitts PM, Peterson JR. Information capacity of discrete motor re-
sponses. J Exp Psychol 1964;67:103–112.
94. Cheng M, Gao X, Gao S, Xu D. Design and implementation of a brain-
computer interface with high transfer rates. IEEE Trans Biomed Eng
2002;49:1181–1186.
95. Gao X, Xu D, Cheng M, Gao S. A BCI-based environmental con-
troller for the motion-disabled. IEEE Trans Neural Syst Rehabil Eng
2003;11:137–140.
96. Zhang D, Maye A, Gao X, Hong B, Engel AK, Gao S. An independent
brain-computer interface using covert non-spatial visual selective at-
tention. J Neural Eng 2010;7:16010.
97. Nielsen KD, Cabrera AF, do Nascimento OF. EEG based BCI-towards
a better control. Brain-computer interface research at Aalborg Univer-
sity. IEEE Trans Neural Syst Rehabil Eng 2006;14:202–204.
98. Wang Y, Wang R, Gao X, Hong B, Gao S. A practical VEP-based
brain-computer interface. IEEE Trans Neural Syst Rehabil Eng
2006;14:234–239.
99. Lin Z, Zhang C, Wu W, Gao X. Frequency recognition based on
canonical correlation analysis for SSVEP-based BCIs. IEEE Trans
Biomed Eng 2006;53:2610–2614.
100. Lin Z, Zhang C, Wu W, Gao X. Frequency recognition based on
canonical correlation analysis for SSVEP-based BCIs. IEEE Trans
Biomed Eng 2007;54:1172–1176.
101. Bin G, Gao X, Yan Z, Hong B, Gao S. An online multi-channel SSVEP-
based brain-computer interface using a canonical correlation analysis
method. J Neural Eng 2009;6:046002.
102. Friman O, Volosyak I, Gräser A. Multiple channel detection of steady-
state visual evoked potentials for brain-computer interfaces. IEEE
Trans Biomed Eng 2007;54:742–750.
103. Müller-Putz GR, Scherer R, Brauneis C, Pfurtscheller G. Steady-state
visual evoked potential (SSVEP)-based communication: impact of
harmonic frequency components. J Neural Eng 2005;2:123–130.
104. Müller-Putz GR, Eder E, Wriessnegger SC, Pfurtscheller G. Com-
parison of DFT and lock-in amplier features and search for opti-
mal electrode positions in SSVEP-based BCI. J Neurosci Methods
2008;168:174–181.
105. Srihari Mukesh TM, Jaganathan V, Reddy MR. A novel multiple
frequency stimulation method for steady state VEP based brain–
computer interfaces. Physiol Meas 2006;27:61–71.
106. Shyu KK, Lee PL, Liu YJ, Sie JJ. Dual-frequency steady-state vi-
sual evoked potential for brain–computer interface. Neurosci Lett
2010;483:28–31.
107. Müller-Putz GR, Scherer R, Neuper C, Pfurtscheller G. Steady-state
somatosensory evoked potentials: suitable brain signals for brain-com-
puter interfaces? IEEE Trans Neural Syst Rehabil Eng 2006;14:30–37.
108. Müller-Putz GR, Pfurtscheller G. Control of an electrical prosthesis
with an SSVEP-based BCI. IEEE Trans Biomed Eng 2008;55:361–364.
109. Kelly SP, Lalor EC, Reilly RB, Foxe JJ. Visual spatial attention track-
ing using high-density SSVEP data for independent brain-computer
communication. IEEE Trans Neural Syst Rehabil Eng 2005;13:
172–178.
110. Kelly SP, Lalor EC, Finucane C, McDarby G, Reilly RB. Visual spatial
attention control in an independent brain-computer interface. IEEE
Trans Biomed Eng 2005;52:1588–1596.
111. Trejo LJ, Rosipal R, Matthews B. Brain-computer interfaces for 1-D
and 2-D cursor control: designs using volitional control of the EEG
spectrum or steady-state visual evoked potentials. IEEE Trans Neural
Syst Rehabil Eng 2006;14:225–229.
112. Allison B, Luth T, Valbuena D, Teymourian A, Volosyak I, Graser A.
BCI demographics: how many (and what kinds of) people can use an
SSVEP BCI? IEEE Trans Neural Syst Rehabil Eng 2010;18:107–116.
113. Bakardjian H, Tanaka T, Cichocki A. Optimization of SSVEP brain
responses with application to eight-command Brain-Computer Inter-
face. Neurosci Lett 2010;469:34–38.
114. Wu Z, Lai Y, Xia Y, Wu D, Yao D. Stimulator selection in SSVEP-based
BCI. Med Eng Phys 2008;30:1079–1088.
115. Wu Z, Yao D. Frequency detection with stability coecient for steady-
state visual evoked potential (SSVEP)-based BCIs. J Neural Eng
2008;5:36–43.
116. Parini S, Maggi L, Turconi AC, Andreoni G. A robust and self-paced
BCI system based on a four class SSVEP paradigm: algorithms and
protocols for a high-transfer-rate direct brain communication. Com-
put Intell Neurosci 2009;Volume is missing864564.
117. Martinez P, Bakardjian H, Cichocki A. Fully online multicommand
brain-computer interface with visual neurofeedback using SSVEP
paradigm. Comput Intell Neurosci 2007;Volume is missing94561.
118. Lee PL, Sie JJ, Liu YJ, Wu CH, Lee MH, Shu CH, Li PH, et al. An SS-
VEP-actuated brain–computer interface using phase-tagged ick-
ering sequences: a cursor system. Ann Biomed Eng 2010;38:2383–
2397.
119. Gollee H, Volosyak I, McLachlan AJ, Hunt KJ, Gräser A. An SSVEP-
based brain-computer interface for the control of functional electrical
stimulation. IEEE Trans Biomed Eng 2010;57:1847–1855.
120. Sellers EW, Donchin E. A P300-based brain-computer interface: ini-
tial tests by ALS patients. Clin Neurophysiol 2006;117:538–548.
121. Fabiani M, Gratton G, Karis D, Donchin E. Denition, identication,
and reliability of measurement of the P300 component of the event-
related brain potential. Adv Psychophysiol 1987;2:78.
122. Polich J, Alexander JE, Bauer LO, Kuperman S, Morzorati S, O’Connor
SJ, Porjesz B, et al. P300 topography of amplitude/latency correlations.
Brain Topogr 1997;9:275–282.
123. Linden DE. e p300: where in the brain is it produced and what does
it tell us? Neuroscientist 2005;11:563–576.
124. Ludowig E, Bien CG, Elger CE, Rosburg T. Two P300 generators in the
hippocampal formation. Hippocampus 2010;20:186–195.
125. Guger C, Daban S, Sellers E, Holzner C, Krausz G, Carabalona
R, Gramatica F, Edlinger G. How many people are able to con-
trol a P300-based brain-computer interface (BCI)? Neurosci Lett
2009;462:94–98.
126. Silvoni S, Volpato C, Cavinato M, Marchetti M, Priis K, Merico A,
Tonin P, et al. P300-Based Brain-Computer Interface Communica-
tion: Evaluation and Follow-up in Amyotrophic Lateral Sclerosis.
Front Neurosci 2009;3:60.
127. Brunner P, Joshi S, Briskin S, Wolpaw JR, Bischof H, Schalk G. Does
the ‘P300 speller depend on eye gaze? J Neural Eng 2010;7:056013.
128. Kleih SC, Nijboer F, Halder S, Kübler A. Motivation modulates the
P300 amplitude during brain-computer interface use. Clin Neuro-
physiol 2010;121:1023–1031.
129. Nijboer F, Birbaumer N, Kübler A. e inuence of psychological
state and motivation on brain-computer interface performance in pa-
tients with amyotrophic lateral sclerosis - a longitudinal study. Front
Neurosci 2010;4.
130. Nijboer F, Sellers EW, Mellinger J, Jordan MA, Matuz T, Furdea A,
Halder S, et al. A P300-based brain-computer interface for people with
amyotrophic lateral sclerosis. Clin Neurophysiol 2008;119:1909–1916.
131. Hill NJ, Lal TN, Bierig K, Birbaumer N, Schölkopf B. An auditory
paradigm for brain–computer interfaces. In: Saul L, Weiss Y, Bottou
L, editors. Advances in Neural Information Processing Systems. Cam-
bridge MA: e MIT Press; 2005. p. 569–576.
132. Sellers EW, Kübler A, Donchin E. Brain-computer interface research at
the University of South Florida Cognitive Psychophysiology Laboratory:
the P300 Speller. IEEE Trans Neural Syst Rehabil Eng 2006;14:221–224.
133. Nijboer F, Furdea A, Gunst I, Mellinger J, McFarland DJ, Birbaumer
N, Kübler A. An auditory brain-computer interface (BCI). J Neurosci
Methods 2008;167:43–50.
134. Kübler A, Furdea A, Halder S, Hammer EM, Nijboer F, Kotchoubey B.
A brain-computer interface controlled auditory event-related poten-
tial (p300) spelling system for locked-in patients. Ann N Y Acad Sci
2009;1157:90–100.
135. Furdea A, Halder S, Krusienski DJ, Bross D, Nijboer F, Birbaumer N,
Kübler A. An auditory oddball (P300) spelling system for brain-com-
puter interfaces. Psychophysiology 2009;46:617–625.
136. Halder S, Rea M, Andreoni R, Nijboer F, Hammer EM, Kleih SC, Bir-
baumer N, Kübler A. An auditory oddball brain-computer interface
for binary choices. Clin Neurophysiol 2010;121:516–523.
137. Bayliss JD, Ballard DH. A virtual reality testbed for brain-computer
interface research. IEEE Trans Rehabil Eng 2000;8:188–190.
138. Bayliss JD. Use of the evoked potential P3 component for con-
trol in a virtual apartment. IEEE Trans Neural Syst Rehabil Eng
2003;11:113–116.
139. Bayliss JD, Inverso SA, Tentler A. Changing the P300 brain–computer
interface. Cyberpsychol Behav 2004;7:694–704.
140. Vaughan TM, McFarland DJ, Schalk G, Sarnacki WA, Krusienski DJ,
Sellers EW, Wolpaw JR. e Wadsworth BCI Research and Develop-
ment Program: at home with BCI. IEEE Trans Neural Syst Rehabil
Eng 2006;14:229–233.
Disabil Rehabil Assist Technol Downloaded from informahealthcare.com by Prof. Stefano Federici on 02/07/12
For personal use only.
Toward functioning and usable BCIs 103
Copyright ©  Informa UK Ltd.
141. Allison BZ, Pineda JA. ERPs evoked by dierent matrix sizes: implica-
tions for a brain–computer interface (BCI) system. IEEE Trans Neural
Syst Rehabil Eng 2003;11:110–113.
142. Sellers EW, Krusienski DJ, McFarland DJ, Vaughan TM, Wolpaw JR.
A P300 event-related potential brain-computer interface (BCI): the
eects of matrix size and inter stimulus interval on performance. Biol
Psychol 2006;73:242–252.
143. Lenhardt A, Kaper M, Ritter HJ. An adaptive P300-based online
brain-computer interface. IEEE Trans Neural Syst Rehabil Eng
2008;16:121–130.
144. Salimi-Khorshidi G, Nasrabadi AM, Golpayegani MH. Fusion of clas-
sic P300 detection methods’ inferences in a framework of fuzzy labels.
Artif Intell Med 2008;44:247–259.
145. Zhang H, Guan C, Wang C. Asynchronous P300-based brain-comput-
er interfaces: a computational approach with statistical models. IEEE
Trans Biomed Eng 2008;55:1754–1763.
146. Lu S, Guan C, Zhang H. Unsupervised brain–computer interface
based on intersubject information and online adaptation. IEEE Trans
Neural Syst Rehabil Eng 2009;17:135–145.
147. Rivet B, Souloumiac A, Attina V, Gibert G. xDAWN algorithm to
enhance evoked potentials: application to brain-computer interface.
IEEE Trans Biomed Eng 2009;56:2035–2043.
148. Dal Seno B, Matteucci M, Mainardi L. Online detection of P300 and er-
ror potentials in a BCI speller. Comput Intell Neurosci 2010;307254.
149. Bostanov V, Kotchoubey B. e t-CWT: a new ERP detection and
quantication method based on the continuous wavelet transform
and Student’s t-statistics. Clin Neurophysiol 2006;117:2627–2644.
150. Jansen BH, Allam A, Kota P, Lachance K, Osho A, Sundaresan K. An
exploratory study of factors aecting single trial P300 detection. IEEE
Trans Biomed Eng 2004;51:975–978.
151. ulasidas M, Guan C, Wu J. Robust classication of EEG signal
for brain-computer interface. IEEE Trans Neural Syst Rehabil Eng
2006;14:24–29.
152. Serby H, Yom-Tov E, Inbar GF. An improved P300-based brain-com-
puter interface. IEEE Trans Neural Syst Rehabil Eng 2005;13:89–98.
153. Piccione F, Giorgi F, Tonin P, Priis K, Giove S, Silvoni S, Palmas G,
Beverina F. P300-based brain–computer interface: reliability and per-
formance in healthy and paralysed participants. Clin Neurophysiol
2006;117:531–537.
154. Krusienski DJ, Sellers EW, McFarland DJ, Vaughan TM, Wolpaw JR.
Toward enhanced P300 speller performance. J Neurosci Methods
2008;167:15–21.
155. Brouwer AM, van Erp JB. A tactile P300 brain-computer interface.
Front Neurosci 2010;4:19.
156. Klobassa DS, Vaughan TM, Brunner P, Schwartz NE, Wolpaw JR, Ne-
uper C, Sellers EW. Toward a high-throughput auditory P300-based
brain-computer interface. Clin Neurophysiol 2009;120:1252–1261.
157. Katayama J, Polich J. P300 from one-, two-, and three-stimulus audi-
tory paradigms. Int J Psychophysiol 1996;23:33–40.
158. Schreuder M, Blankertz B, Tangermann M. A new auditory multi-
class brain-computer interface paradigm: spatial hearing as an infor-
mative cue. PLoS ONE 2010;5:e9813.
159. Homann U, Vesin JM, Ebrahimi T, Diserens K. An ecient P300-
based brain-computer interface for disabled subjects. J Neurosci
Methods 2008;167:115–125.
160. Citi L, Poli R, Cinel C, Sepulveda F. P300-based BCI mouse with
genetically-optimized analogue control. IEEE Trans Neural Syst
Rehabil Eng 2008;16:51–61.
161. Finke A, Lenhardt A, Ritter H. e MindGame: a P300-based brain-
computer interface game. Neural Netw 2009;22:1329–1333.
162. Mugler E, Bensch M, Halder S, et al. Control of an Internet Browser
Using the P300 Signal. Int J Bioelectrom 2008;10:56–63.
163. Mugler EM, Ruf CA, Halder S, Bensch M, Kubler A. Design and imple-
mentation of a P300-based brain-computer interface for controlling an
internet browser. IEEE Trans Neural Syst Rehabil Eng 2010;18:599–609.
164. Muenssinger JI, Halder S, Kleih SC, et al. Brain Painting: evaluation
of a new brain-computer interface application with ALS patients and
healthy volunteers. Front Neurosci 2010;4.
165. Martens SM, Hill NJ, Farquhar J, Schölkopf B. Overlap and refrac-
tory eects in a brain-computer interface speller based on the visual
P300 event-related potential. J Neural Eng 2009;6:026003.
166. Salvaris M, Sepulveda F. Visual modications on the P300 speller BCI
paradigm. J Neural Eng 2009;6:046011.
167. Salvaris M, Sepulveda F. Classication eects of real and imaginary
movement selective attention tasks on a P300-based brain-computer
interface. J Neural Eng 2010;7:056004.
168. Takano K, Komatsu T, Hata N, Nakajima Y, Kansaku K. Visual stimuli
for the P300 brain-computer interface: a comparison of white/gray and
green/blue icker matrices. Clin Neurophysiol 2009;120:1562–1566.
169. Nam CS, Li Y, Johnson S. Evaluation of P300-based brain-com-
puter interface in real-world contexts. Int J Hum Comput Interact
2010;26:621–637.
170. Piccione F, Priis K, Tonin P, et al. Task and stimulation paradigm
eects in a P300 brain–computer interface exploitable in a Virtual En-
vironment: A pilot study. Psychn J 2008;6:99–108.
171. Townsend G, LaPallo BK, Boulay CB, et al. A novel P300-based
brain–computer interface stimulus presentation paradigm: Moving
beyond rows and columns. Clin Neurophysiol 2010;121:1109–1120.
Disabil Rehabil Assist Technol Downloaded from informahealthcare.com by Prof. Stefano Federici on 02/07/12
For personal use only.
... Further limiting current uses for BCIs, a majority of research studying applications in domains like communication or mobility assistance has looked only at animal models or use of a device by an able-bodied person, often excluding people with diseases like amyotrophic lateral sclerosis (ALS). In fact, only a strikingly small number of studies looking at BCIs for therapeutic applications have included would-be end-users (like people with ALS) at all [3,5,8,9]. The lack of research on BCIs using relevant individuals means that successful research outcomes are less likely to be indicative of BCI viability for potential human end-users. ...
... While we observe some exceptions, a majority of quotes concerning ethics and risk concern less pertinent threats like mind control by the government or private companies, elements that are unlikely to be problems related to current embodiments of BCIs, which are mostly for rehabilitation. Discussion of these issues takes place more often than problems currently relevant to BCIs, such as impact on personhood, justice, normalcy [11][12][13][14][15][16][17][18][19], or their technical limitations [3][4][5][7][8][9]. Increasing depictions of BCIs as well as the trend toward discussion about mind control appears to be a response to the increased presence of private BCI projects led by people like Elon Musk and Bryan Johnson. ...
... In particular, the questions that need to be asked include "do people first learn of BCIs through news media?" and "did what they learn in news media influence them to enrol in clinical trials or other potentially problematic activities?" Even beyond this context, BCI research needs to increase efforts to include potential end-users in trials, as well as their perspectives on issues related to therapies [3,5,6,8,9]. What is clear, however, is that the media bears significant responsibility to publish appropriately balanced and accurate information for its potential impact on BCI end-users. ...
Article
Full-text available
Discussion of brain-computer interfaces (BCIs) in news media is increasing. We use FACTIVA to review content of English-speaking news media portrayal of BCIs, between 01 January 2000 and 31 December 2017. Out of 3866 articles, we found 70.6% of this sample (n =2729) discussed BCIs in a future-focused and speculative manner; 51.6% (n =1996) depict BCIs positively, and 25.3% (n =977) with overly positive narratives. By comparison, we observed only 2.7% discussing ethical issues explicitly (n =106) and 12.7% of articles discussing risks associated with BCIs (n =489). A one-way ANOVA analysis yielded no statistical difference between populations of articles depicting BCIs positively, speculatively, and negligently (neglecting ethics and risk) on a significance standard of p < 0.05, suggesting that if an article depicts the technology positively, it will also be speculative and neglect risk analysis. Given news media is identified as having significant impact on influencing public perception and acceptance of medical technologies, we hypothesize that positively biased narratives surrounding BCIs, which make speculative promises about their uses while failing to address risks and ethical issues, can create serious problems related to informed consent, among other things. We argue it is imperative for scientists to accurately represent the benefits and limitations of their research, and for the news media to balance its discussion of BCIs, to make it more likely that the public is aware and considerate of BCI-associated risks – a goal which may be complicated by increasingly frequent press coverage of private BCI projects.
... As per Pasqualotto et al. [23] and Machado et al. [26] BCI could also be categorized depending on whether BCI is dependent or independent of certain muscle movements. Padfield et al. [2] have also categorized BCI as evoked or spontaneous. ...
... An overview concerning additional different categorization of BCIs in the literature is presented in Table 2. Table 2. Additional different categorization of the BCIs in the literature by Pasqualotto et al. [23], Machado et al. [26], Padfield et al. [2] and Nicolas-Alonso and Gomez-Gil [27]. ...
... Pasqualotto et al. [23] Machado et al. [26] Dependent Dependent on muscles and peripheral nerves. For example, in case of visual evoked potential (VEP), gaze is directed by muscles to focus on different visual stimuli. ...
Article
Full-text available
An electroencephalography (EEG)-based brain–computer interface (BCI) is a system that provides a pathway between the brain and external devices by interpreting EEG. EEG-based BCI applications have initially been developed for medical purposes, with the aim of facilitating the return of patients to normal life. In addition to the initial aim, EEG-based BCI applications have also gained increasing significance in the non-medical domain, improving the life of healthy people, for instance, by making it more efficient, collaborative and helping develop themselves. The objective of this review is to give a systematic overview of the literature on EEG-based BCI applications from the period of 2009 until 2019. The systematic literature review has been prepared based on three databases PubMed, Web of Science and Scopus. This review was conducted following the PRISMA model. In this review, 202 publications were selected based on specific eligibility criteria. The distribution of the research between the medical and non-medical domain has been analyzed and further categorized into fields of research within the reviewed domains. In this review, the equipment used for gathering EEG data and signal processing methods have also been reviewed. Additionally, current challenges in the field and possibilities for the future have been analyzed.
... A brain-computer interface (BCI) based on electroencephalography (EEG) enables a user to control external devices by decoding brain activities that reflect the user's thoughts [1]. For example, a user's motor imagery (MI) can be translated into external device control by an MI-BCI. ...
Article
Full-text available
A brain-computer interface (BCI) translates a user’s thoughts such as motor imagery (MI) into the control of external devices. However, some people, who are defined as BCI illiteracy, cannot control BCI effectively. The main characteristics of BCI illiterate subjects are low classification rates and poor repeatability. To address the problem of MI-BCI illiteracy, we propose a distribution adaptation method based on multi-kernel learning to make the distribution of features between the source domain and target domain become even closer to each other, while the divisibility of categories is maximized. Inspired by the kernel trick, we adopted a multiple-kernel-based extreme learning machine to train the labeled source-domain data to find a new high-dimensional subspace that maximizes data divisibility, and then use multiple-kernel-based maximum mean discrepancy to conduct distribution adaptation to eliminate the difference in feature distribution between domains in the new subspace. In light of the high dimension of features of MI-BCI illiteracy, random forest, which can effectively handle high-dimensional features without additional cross-validation, was employed as a classifier. The proposed method was validated on an open dataset. The experimental results show that that the method we proposed suits MI-BCI illiteracy and can reduce the inter-domain differences, resulting in a reduction in the performance degradation of both cross-subjects and cross-sessions.
... Researchers will need to ensure that study participants are representative of potential end-users, and that their methods and results are clearly reported. Although there have been multiple review chapters and articles on the current state of the science for AAC-BCI systems over the past 10 years, most have been non-systematic overviews of system types, algorithms, and applications, often describing results from studies that did not include participants with disabilities (Pasqualotto et al., 2012;Akcakaya et al., 2013;Alamdari et al., 2016;Chaudhary et al., 2016Chaudhary et al., , 2021Rezeika et al., 2018;Wang et al., 2019;Vansteensel and Jarosiewicz, 2020). Existing systematic reviews of AAC-BCI literature have focused on specific system types, such as non-visual interfaces (Riccio et al., 2012), or on the performance of specific end-user populations, such as people with ALS (Marchetti and Priftis, 2014) or cerebral palsy (Orlandi et al., 2021). ...
Article
Full-text available
Augmentative and alternative communication brain-computer interface (AAC-BCI) systems are intended to offer communication access to people with severe speech and physical impairment (SSPI) without requiring volitional movement. As the field moves toward clinical implementation of AAC-BCI systems, research involving participants with SSPI is essential. Research has demonstrated variability in AAC-BCI system performance across users, and mixed results for comparisons of performance for users with and without disabilities. The aims of this systematic review were to (1) describe study, system, and participant characteristics reported in BCI research, (2) summarize the communication task performance of participants with disabilities using AAC-BCI systems, and (3) explore any differences in performance for participants with and without disabilities. Electronic databases were searched in May, 2018, and March, 2021, identifying 6065 records, of which 73 met inclusion criteria. Non-experimental study designs were common and sample sizes were typically small, with approximately half of studies involving five or fewer participants with disabilities. There was considerable variability in participant characteristics, and in how those characteristics were reported. Over 60% of studies reported an average selection accuracy ≤70% for participants with disabilities in at least one tested condition. However, some studies excluded participants who did not reach a specific system performance criterion, and others did not state whether any participants were excluded based on performance. Twenty-nine studies included participants both with and without disabilities, but few reported statistical analyses comparing performance between the two groups. Results suggest that AAC-BCI systems show promise for supporting communication for people with SSPI, but they remain ineffective for some individuals. The lack of standards in reporting outcome measures makes it difficult to synthesize data across studies. Further research is needed to demonstrate efficacy of AAC-BCI systems for people who experience SSPI of varying etiologies and severity levels, and these individuals should be included in system design and testing. Consensus in terminology and consistent participant, protocol, and performance description will facilitate the exploration of user and system characteristics that positively or negatively affect AAC-BCI use, and support innovations that will make this technology more useful to a broader group of people. Clinical trial registration https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42018095345 , PROSPERO: CRD42018095345.
... The main goal of BCI is to support the daily life of individuals with severe disabilities. There are already numerous BCI studies reported in the literature focusing on patients with locked-in syndrome (LIS) [3]. Lately, BCI studies focusing on patients with completely locked-in syndrome (CLIS) have started to appear in the literature as well [4,5]. ...
Article
Full-text available
The aim of brain–computer interface (BCI) is to support the daily life of individuals with severe disabilities. For practical BCI, ease of use is one of the most important factors, which is enhanced when fewer electrodes are used. However, using fewer electrode affect the performance of BCI negatively. In this study, a novel single-channel steady-state visual evoked potential (SSVEP) detection method with subject-specific sinusoids approach (SSSA) was developed to enhance the performance of single channel SSVEP based BCI, therefore, to assist the ease of use. For the SSSA, subject-specific sinusoids were defined from training data based on SSVEP frequency and phase features. To detect the SSVEP response, defined sinusoids were used as reference. To evaluate the detection performance of the developed method, it was compared with the well-known power spectral density analysis (PSDA), least absolute shrinkage and selection operator (LASSO) and advanced canonical correlation analysis (CCA) methods on a benchmark dataset. The experimental results showed significantly greater detection accuracy and information transfer rate (ITR) with the SSSA method compared to the PSDA, LASSO and advanced CCA methods. And it is worth to noting that subject-specific sinusoids better represent SSVEP response than template signals that used in advanced CCA. Also proposed method reached one of the highest ITRs reported with max 125 and average 81 bits/min ITRs for single-channel SSVEP based BCI.
... In recent years, the great potential of using EEG signals in BCI systems to support the rehabilitation process of human limbs has been demonstrated in proposals such as those evidenced in the systematic review. The evidence shows that the use of EEG signals within a BCI system is optimal and suitable when said system has the following elements: the incorporation of non-invasive and inexpensive equipment, good resolution, ease of use, and portability [77][78][79]. Also, the use of these signals for the development of BCI systems that support rehabilitation processes is natural and intuitive [80] since the process directly extracts information about the user's motor intention. ...
Article
Full-text available
In recent years, various studies have demonstrated the potential of electroencephalographic (EEG) signals for the development of brain-computer interfaces (BCIs) in the rehabilitation of human limbs. This article is a systematic review of the state of the art and opportunities in the development of BCIs for the rehabilitation of upper and lower limbs of the human body. The systematic review was conducted in databases considering using EEG signals, interface proposals to rehabilitate upper/lower limbs using motor intention or movement assistance and utilizing virtual environments in feedback. Studies that did not specify which processing system was used were excluded. Analyses of the design processing or reviews were excluded as well. It was identified that 11 corresponded to applications to rehabilitate upper limbs, six to lower limbs, and one to both. Likewise, six combined visual/auditory feedback, two haptic/visual, and two visual/auditory/haptic. In addition, four had fully immersive virtual reality (VR), three semi-immersive VR, and 11 non-immersive VR. In summary, the studies have demonstrated that using EEG signals, and user feedback offer benefits including cost, effectiveness, better training, user motivation and there is a need to continue developing interfaces that are accessible to users, and that integrate feedback techniques.
... Finally, for EEG to become satisfactory end-products, they should also focus on design for usability. As some authors have stated, most of the current EEG prototypes are evaluated on the basis of speed and accuracy, rather than on usability (Moghimi et al., 2012), and they have argued that EEG engineers should integrate ergonomic factors and human-computer interaction principles into the design of their products (Bos et al., 2010;Pasqualotto et al., 2012). Here, one of the most critical concerns for EEG wearables is the inconvenience to wear them in largescale samples for extended periods of time (Xu and Zhong, 2018). ...
Article
Full-text available
The deployment of electroencephalographic techniques for commercial applications has undergone a rapid growth in recent decades. As they continue to expand in the consumer markets as suitable techniques for monitoring the brain activity, their transformative potential necessitates equally significant ethical inquiries. One of the main questions, which arises then when evaluating these kinds of applications, is whether they should be aligned or not with the main ethical concerns reported by scholars and experts. Thus, the present work attempts to unify these disciplines of knowledge by performing a comprehensive scan of the major electroencephalographic market applications as well as their most relevant ethical concerns arising from the existing literature. In this literature review, different databases were consulted, which presented conceptual and empirical discussions and findings about commercial and ethical aspects of electroencephalography. Subsequently, the content was extracted from the articles and the main conclusions were presented. Finally, an external assessment of the outcomes was conducted in consultation with an expert panel in some of the topic areas such as biomedical engineering, biomechatronics, and neuroscience. The ultimate purpose of this review is to provide a genuine insight into the cutting-edge practical attempts at electroencephalography. By the same token, it seeks to highlight the overlap between the market needs and the ethical standards that should govern the deployment of electroencephalographic consumer-grade solutions, providing a practical approach that overcomes the engineering myopia of certain ethical discussions.
... Electroencephalogram (EEG) signals are one of the most widely used types of biomedical signals for BCIs, owing to their portability, high time resolution, ease of acquisition and implementation, and cost-effectiveness (affordable) as compared to other brain activity monitoring techniques (Sayilgan et al. 2019, Sayilgan et al. 2020). There are four typical EEG-based BCI paradigms: steady-state visualevoked potentials (SSVEP), slow cortical potentials (SCP), the P300 component of evoked potentials, and sensorymotor rhythms (SMR) (Pasqualotto et al. 2012). The SSVEP signal is a periodic response to a visual stimulator modulated at a frequency greater than 6 Hz (Wang et al. 2006) (or higher than 4 Hz (Regan 1990)). ...
Article
Full-text available
Brain-Computer Interface (BCI) is a communication system that enables individuals who lack control and use of their existing muscular and nervous systems to interact with the outside world because of various reasons. A BCI enables its user to communicate with some electronic devices by processing signals generated during brain activities. This study attempts to detect and collect gaze data within Electroencephalogram (EEG) signals through classification. To this purpose, three datasets comprised of EEG signals recorded by researchers from the Autonomous University were adopted. The EEG signals in these datasets were collected in a setting where subjects' gaze into five boxes shown on a computer screen was recognized through Steady-State Visually Evoked Potential based BCI. The classification was performed using algorithms of Naive Bayes, Extreme Learning Machine, and Support Vector Machines. Three feature sets; Autoregressive, Hjorth, and Power Spectral Density, were extracted from EEG signals. As a result, using Autoregressive features, classifiers performed between 45.67% and 78.34%, whereas for Hjorth their classification performance was within 43.34-75.25%, and finally, by using Power Spectral Density their classification performance was between 57.36% and 83.42% Furthermore, classifier performances using Naive Bayes varied between 52.23% and 79.15% for Naive Bayes, 56.32-83.42% for Extreme Learning Machine, and 43.34-72.27% for Support Vector Machines by regarding classification algorithms. Among achieved accuracy performances, the best accuracy is 83.42%, achieved by the Power Spectral Density features and Extreme Learning Machine algorithm pair.
... Previous work has analyzed ergonomic issues in BCI design [2] [3] [4], approaching the usability assessment [5] [6] [7] and proposing user-centered designs [8] [9] [10]. Nevertheless, there is still a lack of research on standardized usability methodologies for BCI system with medical purposes, which implies compliance with a highly regulated process for the entire R&D, especially regarding risks management. ...
Article
Over past two decades, steady-state evoked potentials (SSVEP)-based brain computer interface (BCI) systems have been extensively developed. As we all know, signal processing algorithms play an important role in this BCI. However, there is no comprehensive review of the latest development of signal processing algorithms for SSVEP-based BCI. By analyzing the papers published in authoritative journals in nearly five years, signal processing algorithms of preprocessing, feature extraction and classification modules are discussed in detail. In addition, other aspects existed in this BCI are mentioned. The following key problems are solved. (1) In recent years, which signal processing algorithms are frequently used in each module? (2) Which signal processing algorithms attract more attention in recent years? (3) Which modules are the key to signal processing in BCI field? This information is very important for choosing the appropriate algorithms, and can also be considered as a reference for further research. Simultaneously, we hope that this work can provide relevant BCI researchers with valuable information about the latest trends of signal processing algorithms for SSVEP-based BCI systems.
Article
Full-text available
An electroencephalographic (EEG) brain-computer interface (BCI) internet browser was designed and evaluated with 10 healthy volunteers and three individuals with advanced amyotrophic lateral sclerosis (ALS), all of whom were given tasks to execute on the internet using the browser. Participants with ALS achieved an average accuracy of 73% and a subsequent information transfer rate (ITR) of 8.6 bits/min and healthy participants with no prior BCI experience over 90% accuracy and an ITR of 14.4 bits/min. We define additional criteria for unrestricted internet access for evaluation of the presented and future internet browsers, and we provide a review of the existing browsers in the literature. The P300-based browser provides unrestricted access and enables free web surfing for individuals with paralysis.
Article
Full-text available
Since 1996 we have been teaching more than 18 severely or totally paralyzed patients to successfully control the movements of a cursor on a computer screen by means of systematic changes in the amplitudes of their slow cortical potentials (SCPs; Birbaumer, Ghanayim, Hinterberger, Iversen, Kotchoubey et al., 1999). Patients learned regulation of their SCP amplitudes by means of a brain-computer interface (BCI) and on-line feedback about the time course of SCP amplitude shifts, represented by cursor movements on a computer screen. When patients were able to successfully regulate their SCP amplitude, they were trained to use this ability to communicate with friends and caregivers by means of a Language Support Program (Perelmouter, Kotchoubey, Kübler, Taub, & Birbaumer, 1999). Having a reliable predictor of progress in training would be particularly helpful because training patients at their homes requires substantial effort and a positive outcome is desirable given limited personal and financial resources. In this study we present data from healthy participants (n = 10) and a sample of patients (n = 10), diagnosed with amyotrophic lateral sclerosis, who participated in six BCI training sessions; six patients continued training for another six sessions. All participants except one achieved stable significant cursor control. The number of sessions needed to achieve significant cursor control (initial training phase) correlated moderately with the number of sessions needed to achieve a correct response rate of 70% (advanced training phase). Individual differences in performance remained stable within the six training sessions. After six sessions both groups had achieved significant cursor control, but patients' performance was poorer than that of healthy participants. The patients, however, were trained once a week only, and for some patients longer breaks in training occurred. We conclude that learning during the initial training phase indicates the duration of training that will be necessary to achieve 70% correct responses. A higher frequency of training sessions per week seems necessary to achieve faster progress. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Article
Full-text available
Brain computer interfaces (BCI) enable users to transform their thoughts into actions without involving voluntary motor movement. This relatively new user interface technology not only offers those with severe neuromotor disabilities a new means to interact with the environment, but can also provides effective and engaging rehabilitation to restore motor or cognitive functions that have been impaired due to disease or trauma. This article summarizes the use of BCI in neurorehabilitation settings based on a review of relevant literature. It also identifies the challenges of BCI and provides recommendations for practical use and future directions for research. BCI could be a promising tool to assist patients undergoing neurological rehabilitation by providing access to environmental control and communication, as well as neurofeedback rehabilitation.
Article
Canonical correlation analysis (CCA) is applied to analyze the frequency components of steady-state visual evoked potentials (SSVEP) in electroencephalogram (EEG). The essence of this method is to extract a narrowband frequency component of SSVEP in EEG. A recognition approach is proposed based on the extracted frequency features for an SSVEP-based brain computer interface (BCI). Recognition Results of the approach were higher than those using a widely used FFT (fast Fourier transform)-based spectrum estimation method
Article
Slow cortical potentials are negative or positive polarizations of the electroencephalo gram (EEG) or magnetic field changes in the magnetoencephalogram (MEG) that last from 300 ms to several seconds. They originate in depolarizations of the apical dendritic tree in the upper cortical layers that are caused by synchronous firing, mainly from thal amocortical afferents. Functionally they constitute a threshold regulation mechanism for local excitatory mobilization (negative slow potentials) or inhibition (positive slow poten tials) of cortical networks. Humans can learn to voluntarily regulate these potentials after operant training using immediate feedback and positive reinforcement for self-generated slow potentials shifts. After learned self-regulation of negative slow cortical potentials, motor and cognitive performance of various tasks improves, depending upon the specific cortical location of the learned response. Learned reduction of cortical negativity in creases seizure threshold and improves drug-resistant epilepsies. The learned self-reg ulation of slow cortical potentials is based on a redistribution of attentional resources and depends cntically on a prefrontal and thalamic attention system. Finally, a thought trans lation device uses the slow potential self-regulation skill in totally paralyzed locked-in patients for communication with a language-shaping computer system. NEURO SCIENTIST 5:74-78, 1999
Article
We present a Brain–Computer Interface (BCI) game, the MindGame, based on the P300 event-related potential. In the MindGame interface P300 events are translated into movements of a character on a three-dimensional game board. A linear feature selection and classification scheme is applied to identify P300 events and calculate gradual feedback features from a scalp electrode array. The classification during the online run of the game is computed on a single-trial basis without averaging over subtrials. We achieve classification rates of 0.65 on single-trials during the online operation of the system while providing gradual feedback to the player.
Article
We are developing an electroencephalographic (EEG)-based brain-computer interface (BCI) system that could provide an alternative communication channel for those who are totally paralyzed or have other severe motor impairments. The essential features of this system are as follows: (1) EEG analysis in real time, (2) real-time conversion of that analysis into device control, and (3) appropriate adaptation to the EEG of each user. Digital signal processing technology provides the speed and flexibility needed to satisfy these requirements. It also supports evaluation of alternative analysis and control algorithms, and thereby facilitates further BCI development.