TUGraz OPEN Library

Verhelst, Werner

Herre, Jurgen

Kubin, Gernot

Hermansky, Hynek

Jensen, Soren Holdt

EURASIP Journal on Applied Signal Processing 2005:9, 1289–1291
c© 2005 Hindawi Publishing Corporation
Editorial
Werner Verhelst
Department of Electronics and Information Processing, Vrije Universiteit Brussel, 1050 Brussel, Belgium
Email: wverhels@etro.vub.ac.be
Ju¨rgen Herre
Fraunhofer Institute for Integrated Circuits (IIS), 91058 Erlangen, Germany
Email: hrr@iis.fhg.de
Gernot Kubin
Signal Processing and Speech Communication Laboratory, Graz University of Technology, 8010 Graz, Austria
Email: g.kubin@ieee.org
Hynek Hermansky
IDIAP Research Institute, 1920 Martigny, Switzerland
Email: hynek@idiap.ch
Søren Holdt Jensen
Department of Communication Technology, Institute of Electronic Systems, Aalborg University, Fredrik Bajers Vej 7A,
DK-9220 Aalborg, Denmark
Email: shj@kom.aau.dk
Anthropomorphic systems process signals “at the image of
man.” They are designed to solve a problem in signal process-
ing by imitation of the processes that accomplish the same
task in humans. In the area of audio and speech processing,
remarkable successes have been obtained by anthropomor-
phic systems: perceptual audio coding even caused a land-
slide in the music business.
At first sight, it could seem obvious that the performance
of audio processing systems should benefit from taking into
account the perceptual properties of human audition. For ex-
ample, front ends that extract perceptually meaningful fea-
tures currently show the best results in speech recognizers.
However, their features are typically used for a stochastic op-
timization that is itself not anthropomorphic at all. Thus, it is
not obvious why they should perform best, and perhaps the
truly optimal features have not yet been found because, after
all, “airplanes do not flap their wings.”
In general, we believe that there are several situations
when an anthropomorphic approach may not be the best so-
lution. First, its combination with nonanthropomorphic sys-
tems could result in a suboptimal overall performance (the
quantization noise that was cleverly concealed by a percep-
tual audio coder could become unmasked by subsequent lin-
ear or nonlinear processing). Second, other approaches that
are not anthropomorphic might be better adapted to the
technology that is chosen for the implementation (airplanes
do not flap their wings because it is technically much more
eﬃcient to use jet engines for propulsion). Nevertheless, a lot
can be learned from imitating natural systems that were opti-
mized through natural selection. As such, anthropomorphic
and, by extension, biomorphic systems can be considered to
play an important role in the process of developing new tech-
nologies.
This special issue brings together a dozen papers from
diﬀerent areas of audio and speech processing that deal with
aspects of anthropomorphic processing or in which an an-
thropomorphic or perceptual approach was taken.
The first of two papers on perceptual audio coding pro-
poses a perceptual model for the specific distortion that is
typically encountered in sinusoidal modelling, while the sec-
ond paper introduces a novel parametric stereo coding tech-
nique based on binaural psychoacoustics. While these papers
illustrate the use of human auditory perception for eﬃcient
audio coding, the three following papers present examples
of eﬀorts towards using diﬀerent levels of neurophysiologic
modelling directly for the representation and processing of
audio signals: from a model for the adaptation behaviour
in the chemical synapses between the inner hair cells and
the auditory neurons, to a signal processing model for the
early auditory system, and then a cortical audio representa-
1290 EURASIP Journal on Applied Signal Processing
tion for sound modification. In the last pair of audio papers,
signal features that are based on our knowledge of the audi-
tory system are used in conjunction with machine learning
techniques, such as neural networks, to achieve more cogni-
tive goals, such as audio source separation and classification.
A generally applicable technique that allows for discrim-
inative training of hidden Markov models is introduced and
applied on the confusable set of visemes for lip reading pur-
poses in the first of five papers on speech processing. The next
three of these papers all deal with the important problem
of finding objective distortion measures for speech, and the
last paper describes an articulatory speech synthesizer that,
among other things, brought a better understanding of the
Portuguese nasal vowels.
While the papers in this special issue can represent only
a small sampling of anthropomorphic techniques in audio
and speech processing, they are all very valuable in their own
right and together, if nothing else, they show that anthropo-
morphic sound processing systems are invaluable in the form
of computational models for human perception and that
they can fuel our quest for further understanding of human
nature and self-knowledge.
Werner Verhelst
Ju¨rgen Herre
Gernot Kubin
Hynek Hermansky
Søren Holdt Jensen
Werner Verhelst obtained the Engineer-
ing degree, BurgerlijkWerktuigkundig Elec-
trotechnisch Ingenieur, in 1980, and the
Ph.D. degree in 1985, both from the
Vrije Universiteit Brussel, Belgium. He spe-
cialised in digital speech and audio process-
ing in general, and in speech and audio sig-
nal modification in particular. Verhelst also
studied speech synthesis at the Institute for
Perception Research, and audio signal mod-
elling at the Katholieke Universiteit Leuven, Belgium. Since his
graduation, he has been with the Vrije Universiteit Brussel where
he is heading the Research Laboratory on Digital Speech and Audio
Processing (DSSP) and teaching courses on digital signal process-
ing and speech and audio processing.
Ju¨rgen Herre joined the Fraunhofer Insti-
tute for Integrated Circuits (IIS), Erlangen,
Germany, in 1989. Since then he has been
involved in the development of percep-
tual coding algorithms for high-quality au-
dio, including the well-known ISO/MPEG-
Audio Layer III coder (aka “MP3”). In 1995,
Dr. Herre joined Bell Laboratories for a
postdoc term working on the development
of MPEG-2 advanced audio coding (AAC).
Since the end of 1996, he has returned to Fraunhofer to work
on the development of advanced multimedia technologies includ-
ing MPEG-4, MPEG-7, and secure delivery of audiovisual con-
tent. Currently he is the Chief Scientist for the audio/multimedia
activities at Fraunhofer Institute for Integrated Circuits (IIS), Er-
langen. Dr. Herre is a Fellow of the Audio Engineering Society,
Cochair of the AES Technical Committee on Coding of Audio Sig-
nals, and Vice Chair of the AES Technical Council. He also served
as an Associate Editor of the IEEE Transactions on Speech and Au-
dio Processing and is an active member of the MPEG audio sub-
group.
Gernot Kubin was born in Vienna, Austria,
on June 24, 1960. He received the Dipl.-
Ing. degree in 1982, and Dr. Techn. degree
(sub auspiciis praesidentis) in 1990, both
in electrical engineering from TU Vienna.
He has been a Professor of nonlinear signal
processing and the Head of the Signal Pro-
cessing and Speech Communication Lab-
oratory (SPSC), Graz University of Tech-
nology, Austria, since September 2000. Ear-
lier international appointments include CERN, Geneva, Switzer-
land, 1980; TU Vienna, from 1983 to 2000; Erwin Schroedinger
Fellow at Philips Natuurkundig Laboratorium, Eindhoven, The
Netherlands, 1985; AT&T Bell Labs, Murray Hill, USA, from
1992 to 1993, and 1995; KTH, Stockholm, Sweden, 1998; Vi-
enna Telecommunications Research Centre (FTW) from 1999 up
to date as Key Researcher and Member of the Board; Global
IP Sound, Sweden and USA, from 2000 to 2001 as a Scien-
tific Consultant; Christian Doppler Laboratory for Nonlinear
Signal Processing from 2002 up to date as the Founding Di-
rector. Dr. Kubin is a Member of the Board of the Austrian
Acoustics Association and Vice Chair for the European COST
Action 277, Nonlinear Speech Processing. He has authored or
coauthored over ninety peer-reviewed publications and three
patents.
Hynek Hermansky works at the IDIAP Re-
search Institute, Martigny, Switzerland. He
has been working in speech processing for
over 30 years, previously as a Research Fel-
low at the University of Tokyo, a Research
Engineer at Panasonic Technologies, Santa
Barbara, California, a Senior Member of
the research staﬀ at US WEST Advanced
Technologies, and a Professor and Direc-
tor of the Center for Information Process-
ing, OHSU, Portland, Oregon. He is a Fellow of the IEEE for
the “invention and development of perceptually-based speech pro-
cessing methods,” a Member of the Board of the International
Speech Communication Association, and a Member of the Ed-
itorial Boards of Speech Communication and of Phonetica. He
holds 5 US patents and authored or coauthored over 130 papers
in reviewed journals and conference proceedings. He holds a Dr.
Eng. degree from the University of Tokyo, and Dipl.-Ing. degree
from Brno University of Technology, Czech Republic. His main re-
search interests are in acoustic processing for speech and speaker
recognition.
Editorial 1291
Søren Holdt Jensen received the M.S. de-
gree in electrical engineering from Aal-
borg University, Denmark, in 1988, and
the Ph.D. degree from the Technical Uni-
versity of Denmark, in 1995. He has been
with the Telecommunications Laboratory of
Telecom Denmark, the Electronics Institute
of the Technical University of Denmark, the
Scientific Computing Group of the Danish
Computing Center for Research and Educa-
tion (UNI-C), the Electrical Engineering Department of Katholieke
Universiteit Leuven, Belgium, the Center for PersonKommunika-
tion (CPK) of Aalborg University, and is currently an Associate Pro-
fessor in the Department of Communication Technology, Aalborg
University. His research activities are in digital signal processing,
communication signal processing, and speech and audio process-
ing. Dr. Jensen is a Member of the Editorial Board of the EURASIP
Journal on Applied Signal Processing, and a former Chairman of
the IEEE Denmark Section and the IEEE Denmark Section’s Signal
Processing Chapter.


Editorial

Feldbauer, Christian

Kleijn, W Bastiaan

Anthropomorphic Coding of Speech and Audio: A Model Inversion Approach

Di Camillo, Barbara

Sanchez-Cabo, Fatima

Toffolo, Gianna

Nair, Sreekumaran

Trajanoski, Zlatko

A quantization method based on threshold optimization for microarray short time series

Hackl, Hubert

Burkard, Thomas

Sturn, Alexander

Rubio, Renee

Schleiffer, Alexander

Tian, Sun

Quackenbush, John

Eisenhaber, Frank

Molecular processes during fat cell development revealed by gene expression profiling and functional annotation

Schanz, Martin

Kielhorn, Lars

Poroelastodynamic Boundary Element Method in Time Domain: Numerical Aspects

Koller, Martin

Bona, Rodolfo

Braunegg, Gerhart

Hermann, Carmen

Horvat, Predrag

Kroutil, Markus

Martinz, Julia

Neto, José

Varila, Paula

Pereira, Luis

Production of Polyhydroxyalkanoates from Agricultural Waste and Surplus Materials

Schlögl, Alois

Lee, Felix Yuan-Tseng

Bischof, Horst

Pfurtscheller, Gert

Characterization of Four-Class Motor Imagery EEG Data for the BCI-Competition 2005

Muller-Putz, Gernot R

Pfurtscheller, Jorg

Rupp, Rudiger

EURASIP Journal on Applied Signal Processing 2005:19, 3152–3155
c© 2005 Hindawi Publishing Corporation
EEG-Based Asynchronous BCI Controls Functional
Electrical Stimulation in a Tetraplegic Patient
Gert Pfurtscheller
Laboratory of Brain-Computer Interfaces, Institute of Computer Graphics and Vision, and Ludwig Boltzmann-Institute for
Medical Informatics and Neuroinformatics, Graz University of Technology, Inﬀeldgasse 16a, 8010 Graz, Austria
Email: pfurtscheller@tugraz.at
Gernot R. Mu¨ller-Putz
Laboratory of Brain-Computer Interfaces, Institute of Computer Graphics and Vision, Graz University of Technology,
Inﬀeldgasse 16a, 8010 Graz, Austria
Email: gernot.mueller@tugraz.at
Jo¨rg Pfurtscheller
Department of Traumatology, Hospital Villach, Nikolaigasse 43, 9400 Villach, Austria
Email: j.pfurtscheller@tugraz.at
Ru¨diger Rupp
Department II, Orthopedic Hospital of Heidelberg University, Schlierbacher Landstraße 200a, 69118 Heidelberg, Germany
Email: ruediger.rupp@ok.uni-heidelberg.de
Received 29 January 2004
The present study reports on the use of an EEG-based asynchronous (uncued, user-driven) brain-computer interface (BCI) for
the control of functional electrical stimulation (FES). By the application of FES, noninvasive restoration of hand grasp function
in a tetraplegic patient was achieved. The patient was able to induce bursts of beta oscillations by imagination of foot movement.
These beta oscillations were recorded in a one EEG-channel configuration, bandpass filtered and squared. When this beta activity
exceeded a predefined threshold, a trigger for the FES was generated. Whenever the trigger was detected, a subsequent switching
of a grasp sequence composed of 4 phases occurred. The patient was able to grasp a glass with the paralyzed hand completely on
his own without additional help or other technical aids.
Keywords and phrases: beta oscillations, motor imagery, functional electrical stimulation, brain-computer interface, spinal cord
injury, neuroprosthesis.
1. INTRODUCTION
The idea of direct brain control of functional electrical stim-
ulation (FES) seems to be a realistic concept for restoration of
the hand grasp function in patients with a high spinal cord
injury. Today, electrical brain activity either recorded from
the intact scalp (EEG) or with subdural electrodes (ECoG)
can be classified and transferred into signals for control of
FES system (neuroprosthesis). Nowadays both implantable
systems [1, 2] and devices using surface electrodes [3] are
available for clinical use. For the transformation of mental
commands reflected in changes of the brain signal into con-
trol signals for FES devices, an asynchronous, user-driven
brain-computer interface (BCI) is necessary [4]. Such an
asynchronous BCI analyses the EEG (ECoG) continuously
and uses no cue stimuli.
For the realization of a reliable and easy to apply BCI,
only one signal channel (one recording with two electrodes)
should be used. Further, it is necessary to have a mental strat-
egy established to produce short increases or bursts in the
EEG (ECoG) amplitude and to detect the increase with a sim-
ple threshold comparator.
We report for the first time on restoration of hand grasp
function composed of 4 phases by electrical stimulation of
handmuscles with surface electrodes and control of the stim-
ulation by one-channel EEG recording.
2. MATERIALS ANDMETHODS
2.1. Subject
The tetraplegic patient we report on is a 29-year old man suf-
fering from a traumatic spinal cord injury since April 1998.
Asynchronous BCI Controls FES in a Tetraplegic Patient 3153
He is aﬀected by a complete motor and sensory lesion below
C5 and an incomplete lesion below C4. As a preparation for
the experiment, the patient performed an individual stimu-
lation program until he achieved a strong and fatigue resis-
tant contraction of the paralyzed muscles of the hand and
forearm. The residual volitional muscle activation of his left
upper extremity is as follows.
Shoulder: active abduction and flexion up to 90◦; grade
3/5 before - grade 4/5 after training, full rotational range of
motion (ROM); full passive ROM.
Elbow: active flexion grade 3/5 before / grade 4/5 af-
ter training, no active extension (triceps grade 0/5); pro-
and supination possible (partly trick movement); full passive
ROM.
Forearm, hand, and fingers: M. extensor carpi radialis
(ECR) showed a palpable active contraction (grade 1/5)
without change over training; all other muscles grade 0/5; al-
most full passive ROM in finger joints; full wrist, thumb, and
forearm ROM.
2.2. Functional electrical stimulation
Our aim was to find a functional grasp pattern that would
bring the most benefit for our patient, and to find a practical
way to generate it by use of surface stimulation electrodes. A
kind of fine manipulating grasp, providing the ability to pick
up objects from a table, for example, food or a glass, seemed
to be most suitable. This grasp is generated by flexion in the
metacarpophalangeal (MCP) joints of the extended fingers
against the thumb, so that small objects are held between the
ball of the end phalanx of the fingers and the thumb, while
larger objects are held between the palmar side of the whole
fingers and the thumb.
As a precondition for a functional hand grasp pattern, the
wrist needs to be dorsal flexed and held stable in this position
during flexion of the fingers. Due to the lack of an adequate
active wrist extension and a partial denervation (lesion of pe-
ripheral nerve fibers) of the wrist extensor muscle (M. exten-
sor carpi radialis muscle, grade 1/5) in our patient, it was not
possible to get a stable dorsal flexion of the wrist by stimula-
tion, forcing us to use a mechanical orthosis fixing the wrist
in a dorsal flexed position.
An opening of the hand (phases 1 and 4, Figure 1) by ex-
tension of all fingers joints and the thumb could be achieved
by stimulation of the finger extensors (M. extensor digito-
rum communis) and the thumb extensor muscle (M. exten-
sor pollicis longus) with electrodes on the radial side of the
proximal forearm.
For the actual grasping (phase 2, Figure 1), we simultane-
ously stimulated the finger flexors (M. flexor digitorum su-
perficialis, less the M. flexor digitorum profundus) by one
pair of electrodes on the ulnar side of the proximal forearm
and the intrinsic hand muscles with two further electrodes
on the dorsal side of the hand. The application of the orthosis
for dorsal flexion of the wrist leads to a light flexed position
of the thumb suﬃcient for serving as a stable counterpart to
the flexing fingers. Therefore, no additional stabilization of
the thumb via surface stimulation was necessary.
Motor
imagery
Motor
imagery
Motor
imagery
Motor
imagery
10 s
Threshold
Trigger
for FES Relax
50454035302520151050
−50
0
50
E
E
G
(µ
V
)
50454035302520151050
−25
0
25
B
et
a
(µ
V
)
50454035302520151050
−8
−6
−4
B
et
a
po
w
er
1 2 3 0
Figure 1: Example of bipolar EEG recording from the vertex (upper
trace) bandpass filtered (15–19Hz) EEG signal (middle) and band
power time course (lower trace, arbitrary units) over a time inter-
val of 50 seconds. Threshold and trigger pulse generation after FES
operation and grasp phases are indicated. Shots of the grasping are
shown in the lower part.
For the external stimulation, we used a stimulator
(Microstim8, Krauth & Timmermann, Germany) with bi-
phasic, rectangular constant current pulses. The stimulation
frequency was set to 18Hz; the current was set for each pair
of electrodes on an individual level. Due to the integrated
microcontroller, we were able to implement diﬀerent stimu-
lation patterns for a grasp sequence directly into the device.
The output of the BCI was then used as a trigger signal
for switching between the diﬀerent grasp phases (phase 0 -
no stimulation, phase 1 - opening hand, phase 2 - grasping,
phase 3 - releasing, phase 4 = phase 0, see Figure 1).
2.3. EEG recording and processing
The EEG was recorded bipolarly from 2 gold-electrodes fixed
in a distance of 5 cm in an anterior-posterior position on the
vertex (Cz according to the international 10–20 system). The
EEG signal was amplified (sensitivity was 50 µV) between 0.5
and 30Hz with a bipolar EEG-amplifier (Raich, Graz) and
sampled with 128Hz. The signal was online processed by
bandpass filtering (15–19Hz), squaring, averaging over 128
samples, and logarithmizing. After passing a threshold de-
3154 EURASIP Journal on Applied Signal Processing
tector, a trigger pulse was generated followed by a refractory
period of 3 seconds. The threshold was empirically selected
by comparing the band power values obtained from resting
and imagery periods.
2.4. Mental strategy
Our patient participated in a number of BCI training ses-
sions with the goal of developing a mental strategy to induce
movement-specific EEG patterns and to transform these pat-
terns into a binary control signal. During the training, sev-
eral types of imaginations were used in order to increase
the classification accuracy. Imaginations of left versus right
handmovements were carried out first. Then single-footmo-
tor imageries versus relaxing or handmovement imagination
could increase the accuracy. Finally, after 55 training sessions,
best results were achieved by the imagination of both feet
versus right hand imagery. These two patterns were discrim-
inable online 100%.
3. RESULTS
As a first result of the BCI training, the patient was able
to control the opening and closing of an electromechani-
cal hand orthosis by 2-channel EEG recording [5]. Inspec-
tion of EEG patterns induced by motor imagery has shown
that hand motor imagery was accompanied by a weak EEG
desynchronization [6] whereas foot motor imagery induced
large bursts of beta oscillations with frequencies of 17Hz. It is
therefore quite logical to use only one mental state, namely,
the state inducing beta oscillations for control purposes. At
the end of the training, the patient has learned to voluntar-
ily induce beta bursts. An example of an EEG signal recorded
bipolarly on the vertex close to the foot representation area
is shown in Figure 1. The EEG signal is disturbed by large ar-
tifacts from eye movements, because the patient watched his
hand. Bandpass filtering of the EEG in the beta band (15–
19Hz) reveals 4 bursts of beta activity with a duration of
about five seconds within the 50 seconds of recording period.
The beta power increase was used for generation of a trig-
ger pulse, whenever power exceeded the predefined thresh-
old. Applying a refractory period of 3 seconds a maximum of
20 switches can theoretically be achieved per minute.
The use of only 2 electrodes placed close to the vertex
and the recording of one bipolar EEG channel minimizes the
eﬀects brought about by using muscle activity for control.
Calculating the power spectra and computing the power in
the 20–60Hz band (part of the EMG activity band) showed
a band power close to zero.
The patient was able to trigger the FES grasp phases by
the induction of beta burst on his own. Using this setting,
our patient was able, for the first time after the accident, to
drink from a glass without any help and without the use of a
straw.
4. DISCUSSION
It is interesting to note that the motor imagery induced beta
burst is a relative stable phenomenon in our patient with a
constant frequency around 17Hz. Since this time about 3
years ago, foot motor imagery was always able to induce beta
bursts with constant frequency components. The generation
network of these beta bursts is very likely in the foot repre-
sentation area and/or the supplementary motor area (SMA).
In a foot motor imagery task, both primary sensorimotor
area and SMA play an important role, whereby the SMA is
located in the medial portion of Brodman’s area 6 in front of
the foot representation area. From scalp recordings, we can-
not expect, however, to diﬀerentiate between both sources,
because of the proximity of SMA and foot representation
area [7].
There is strong evidence from EEG, MEG, and ECoG
recordings that diﬀerent motor tasks including imagery can
generate beta oscillations between 20–35Hz in the SMA
and/or the foot representation area of able-bodied subjects
[8, 9, 10, 11, 12]. Common for all these reports on induced
beta oscillations close to the vertex are its strict localization
to the midcentral area and its dominant frequency between
20–35Hz.
There is, however, one important diﬀerence between all
the observed beta oscillations associated with a motor task
in able-bodied subjects and the induced beta oscillations in
our tetraplegic patient: The former are generated after ter-
mination of the motor task, the latter during execution of
the motor task. Whether in both cases the same or similar
networks in the SMA and/or foot representation area are in-
volved needs further research. Important is that the beta os-
cillations on the vertex induced by the reported patient are
a robust and reliable phenomenon that can be generated at
“will”.
For an EEG-based control of a neuroprosthesis in all-day
life under real-world conditions, the performance of the BCI
has to be maximized by using a minimum number of elec-
trodes. Using more than one single bipolar derivation, it is
likely to help identifying more than a binary switch conclud-
ing in the realization of a more complex EEG-based control
for the future.
ACKNOWLEDGEMENT
This project was supported by the Austrian Federal Min-
istry of Transport, Innovation and Technology, project
GZ140.587/2, the “Lorenz-Boehler Gesellschaft,” and the
“Allgemeine Unfallversicherungsanstalt” (AUVA).
REFERENCES
[1] B. Fromm, R. Rupp, and H. J. Gerner, “The freehand system:
an implantable neuroprosthesis for functional electrostimu-
lation of the upper extremity,”Handchirurgie, Mikrochirurgie,
Plastische Chirurgie, vol. 33, no. 3, pp. 149–152, 2001.
[2] P. H. Peckham, M. W. Keith, K. L. Kilgore, et al., “Eﬃcacy
of an implanted neuroprosthesis for restoring hand grasp in
tetraplegia: A multicenter study,” Archives of Physical Medicine
and Rehabilitation, vol. 82, no. 10, pp. 1380–1388, 2001.
[3] M. R. Popovic, D. B. Popovic, and T. Keller, “Neuroprostheses
for grasping,” Neurological Research, vol. 24, no. 5, pp. 443–
452, 2002.
[4] G. Pfurtscheller and C. Neuper, “Motor imagery and direct
Asynchronous BCI Controls FES in a Tetraplegic Patient 3155
brain-computer communication,” Proc. IEEE, vol. 89, no. 7,
pp. 1123–1134, 2001.
[5] G. Pfurtscheller, C. Guger, G. R. Mu¨ller, G. Krausz, and
C. Neuper, “Brain oscillations control hand orthosis in a
tetraplegic,” Neuroscience Letters, vol. 292, no. 3, pp. 211–214,
2000.
[6] G. Pfurtscheller and F. H. Lopes da Silva, “Event-related
EEG/MEG synchronization and desynchronization: basic
principles,” Clinical Neurophysiology, vol. 110, no. 11,
pp. 1842–1857, 1999.
[7] A. Ikeda, H. O. Lu¨ders, R. C. Burgess, and H. Shibasaki,
“Movement-related potentials recorded from supplementary
motor area and primary motor area. Role of supplementary
motor area in voluntary movements,” Brain, vol. 115, no. 4,
pp. 1017–1043, 1992.
[8] C. Neuper and G. Pfurtscheller, “Motor imagery and ERD,” in
Event-Related Desynchronization. Handbook of Electroenceph.
and Clin. Neurophysiol. Revised Edition, G. Pfurtscheller and
F. H. Lopes da Silva, Eds., vol. 6, pp. 303–325, Elsevier, Ams-
terdam, the Netherlands, 1999.
[9] G. Pfurtscheller, M. Woertz, G. Supp, and F. H. Lopes da
Silva, “Early onset of post-movement beta electroencephalo-
gram synchronization in the supplementary motor area dur-
ing self-paced finger movement in man,”Neuroscience Letters,
vol. 339, no. 2, pp. 111–114, 2003.
[10] S. Ohara, A. Ikeda, T. Kunieda, et al., “Movement-related
change of electrocorticographic activity in human supple-
mentary motor area proper,” Brain, vol. 123, no. 6, pp. 1203–
1215, 2000.
[11] J. Kaiser, W. Lutzenberger, H. Preissl, D. Mosshammer, and
N. Birbaumer, “Statistical probability mapping reveals high-
frequency magnetoencephalographic activity in supplemen-
tary motor area during self-paced finger movements,” Neuro-
science Letters, vol. 283, no. 1, pp. 81–84, 2000.
[12] R. Salmelin, M. Ha¨ma¨la¨inen, M. Kajola, and R. Hari, “Func-
tional segregation of movement-related rhythmic activity in
the human brain,” NeuroImage, vol. 2, no. 4, pp. 237–243,
1995.
Gert Pfurtscheller received the M.S. and
Ph.D. degrees in electrical engineering from
the Graz University of Technology, Graz,
Austria. He is a Professor of medical in-
formatics, Director of the Laboratory of
Brain-Computer Interfaces, Graz University
of Technology, and Director of the Lud-
wig Boltzmann-Institute for Medical Infor-
matics and Neuroinformatics. His research
interests include functional brain topogra-
phy, the design of brain-computer communication systems, and
navigation in virtual environments by a brain-computer inter-
face.
Gernot R. Mu¨ller-Putz received the M.S.
degree in biomedical engineering from the
Graz University of Technology, Graz, Aus-
tria, in May 2000. He received the Ph.D. de-
gree in electrical engineering in August 2004
from the same university. His research in-
terests include brain-computer communi-
cation systems, the human somatosensory
system, rehabilitation engineering, and as-
sistive technology.
Jo¨rg Pfurtscheller received the M.D. degree
from the University of Graz in 1995. Cur-
rently he is with the Department of Trau-
matology, Hospital Villach, in the final stage
to become a trauma surgeon. His research
interests include rehabilitation after spinal
cord injury and applications of functional
electrical stimulation.
Ru¨diger Rupp received the M.S. degree
in electrical engineering with a focus on
biomedical engineering from the Technical
University of Karlsruhe, Germany, in 1994.
After working at the Institute for Biomedi-
cal Engineering and Biocybernetics (Profes-
sor G. Vossius), since 1996, he is currently
with the Orthopaedic University Hospital II
in Heidelberg (Professor H. J. Gerner), Ger-
many, where he holds the position as a Re-
search Group Manager. He is now finishing his Ph.D. work. His
main research interests are in the field of rehabilitation engineer-
ing especially for spinal cord injured patients. This includes neuro-
prosthetics mainly of the upper extremity, application of functional
electrical stimulation for therapeutic purposes, development and
clinical validation of novel methods, devices for locomotion ther-
apy, gait analysis in incomplete spinal cord injury, and realization
of software projects for standardized documentation of rehabilita-
tion outcome. He is a Member of IEEE, IFESS, and VDE.


EEG-Based Asynchronous BCI Controls Functional Electrical Stimulation in a Tetraplegic Patient

Graz University of Technology

Not a member yet