ISSN: 1693-6
930
73
A Novel Ap
proach for Co
nfigurin
g The S
t
im
ulat
or of a
BCI Fram
ewo
r
k…… (Ind
ar
Sugiarto
)
A NOVEL APPROACH FOR CONFIGURING THE
STIMULATOR OF A BCI FRAMEWORK USI
N
G XM
L
Indar Sugiar
to
Dep
a
rtme
nt of Electrical En
ginee
ring – P
e
tra Ch
ri
stian
University
Jl. Siwalan
k
e
r
to 121
-13
1
, phon
e: +62
-
3
1
-29
834
42
Email: indi@petra.a
c.id
Abs
t
rak
Dalam
BCI (Brain
-Com
puter Interfa
c
e),
setiap
asp
e
k h
a
ru
s diperh
a
tika
n dem
i
keb
e
rh
asil
an
operasio
nal
dari si
stem
BCI terse
b
u
t. Term
asu
k
didal
am
nya adala
h
prose
s
pem
buatan stim
ulator BCI yang h
and
al
dan fleksibel
, terutam
a
stim
ulator yan
g
berkaitan e
r
at
deng
an um
p
an bali
k
d
a
l
a
m
bentuk
aplikasi
dari
sistem
BCI
.
Makala
h i
n
i m
enjelaskan
pend
ekatan b
a
ru
untu
k
m
e
m
buat stim
ulator
vis
ual
ya
ng fle
ksi
bel d
enga
n m
e
m
anfaatka
n form
at
XML
(E
xten
sible
Ma
rkup
Lang
uage
) yang dap
at
di
terap
k
an
pa
d
a
sebu
ah
un
it sistem
BCI.
Den
gan m
e
n
ggun
akan form
at XML untuk m
engatu
r
konfigu
r
a
s
i dari stim
ulator
vi
sual seb
u
a
h
unit BCI, kit
a
dap
at m
engem
bang
ka
n apli
k
a
s
i B
C
I ya
ng m
a
m
pu m
engakom
odasi b
a
n
y
a
k
strategi
pe
rcobaa
n d
a
lam
pen
elitian te
ntang B
C
I. Unit BCI da
n p
l
atform
konfig
ura
s
in
ya
dibu
at
deng
an m
enggun
akan ba
hasa pem
rog
r
am
an C++ d
an m
e
m
anfaatkan XML p
a
rser d
a
ri Qt yan
g
bern
a
m
a
QX
m
l
Stream
. Dari h
a
sil im
pl
em
entasi d
a
n
pen
gujia
n t
e
rlihat
bah
wa file konfigu
r
asi
XML da
pat d
i
ekse
ku
si d
e
ngan
bai
k
ol
eh
sistem
B
C
I yang
dig
u
nakan. Selai
n
kem
a
m
puannya
dalam
m
enghasil
ka
n fre
k
uen
si ke
dipa
n yan
g
fl
eksibel se
rta pe
ngaturan form
at teks untuk
sistem
BCI berbasi
s SS
VEP, file konfigurator
tersebut juga
m
e
m
berikan
pilihan pem
a
kaian
hingg
a 3 ben
tuk bang
un, 16 wa
rna, da
n 5 indikato
r um
pan balik
yan
g
berbed
a. Metode ya
ng
dipap
arkan
d
a
lam
m
a
kala
h ini da
pat di
perg
una
ka
n
untuk m
enin
g
katkan
keg
u
n
aan d
a
ri u
n
it BC
I
yan
g
telah ad
a saat ini sep
e
rti BF++ Toys dan BCI 20
00.
Kata kunci
:
Stim
ulator BCI, file konfig
ura
s
i, XML
A
b
st
r
a
ct
In a
workin
g
BCI fram
ewo
r
k,
all a
s
pe
ct
s m
u
st
be
co
nsid
ere
d
a
s
an inte
gral
p
a
rt that
contri
bute
s
to
the succe
s
sful operation
of a BCI
system
. This also
incl
ud
es the
developm
ent of
robu
st but flexible
stim
ulator, esp
e
ci
ally the
one that closely relat
ed to the feedba
ck of a BCI
system
. This
pape
r d
e
scrib
es
a n
o
vel a
p
p
roa
c
h
in
pro
v
idin
g fle
x
ible
vi
sual
stim
ulator u
s
in
g X
M
L
whi
c
h ha
s be
en appli
ed fo
r a BCI (b
rai
n
-com
puter i
n
terface) fra
m
ework. Usi
n
g XML file form
a
t
for config
uri
n
g the visual st
im
ulator of a BCI
system
,
we can de
vel
op BCI appli
c
ations
whi
c
h can
accom
m
odate m
any
exp
e
rim
ent strategie
s
in B
C
I
re
sea
r
ch. T
he BCI f
r
am
ewo
r
k an
d it
s
config
uratio
n
platform
is
develop
ed using C
++ p
r
og
ram
m
i
ng languag
e whi
c
h
incorporate Qt’
s
m
o
st powerf
u
l XML pa
rser n
a
m
ed Q
X
m
l
Stream
.
The im
plem
entation an
d
exp
e
rim
ent shows
that
the
XML config
uratio
n file
can be well
exec
uted
within the
pro
posed BCI fram
ework. Be
side
its capability
in presenting flexibl
e
flickering
frequencies and text
form
atting for SSVEP-based
BCI, the
confi
guratio
n pl
atform
also p
r
o
v
ide
s
3
sha
p
e
s, 1
6
colo
rs, and
5
distin
ct feedba
ck b
a
rs.
It is not
ne
ce
ssary to
incre
a
se
the
num
ber of
sh
ape
s
nor
col
o
rs
sin
c
e th
ose p
a
ra
m
e
ters a
r
e l
e
ss
im
portant for
the BCI
stimulator. Th
e p
r
opo
se
d m
e
thod
can
then
be e
x
ten
ded
to enh
an
ce t
h
e
usa
b
ility of cu
rre
ntly e
x
iste
d BCI fram
ework
su
ch a
s
BF++ Toys an
d BCI 2000.
Key
w
ords
: B
C
I Stim
ulator, Configu
r
atio
n File, XML
1. INTRO
DUCTIO
N
One of the most pro
m
isi
n
g
applications
of
biomedi
cal
electro
n
ics i
n
the area of
human
-
ma
c
h
in
e in
te
ra
c
t
io
n is th
e
s
o
-c
a
l
le
d
BC
I (
b
r
a
in-c
o
m
pu
te
r
in
te
r
f
ac
e)
. T
h
e BC
I s
y
s
t
e
m
c
o
mp
r
i
se
s
of many
com
pone
nts in
clu
d
ing h
a
rdwares,
softwa
r
e
s
and
protocols. The
protocol in the
BCI
will
be u
s
ed
to d
r
ive the
syst
em in o
r
d
e
r t
o
en
su
re
that
the brain
elicite ne
ce
ssa
r
y signal
s to
be
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 7, No. 2, Agustus 2
009 : 73 - 82
74
acq
u
ire
d
by t
he sy
stem.
Dep
endin
g
o
n
the type
of
the BCI, the
pre
s
ent
of a
stimulato
r
a
s
a
mean
of the
proto
c
ol
impl
ementation
may be
man
dat
ory. Th
e
stimulator
play
s vital
role
in
BCI
s
y
s
t
em, es
pec
ially for B
C
Is
which operate ex
ogenous
ly. The
s
o
-c
all
ed
VEP (vis
ual evoked
potential) is a
sp
eci
a
l p
o
te
ntial which
can b
e
eli
c
ited
usi
ng vi
sual
stimulato
r
. S
o
me
re
sea
r
ches
sho
w
th
at ce
rtain p
a
ttern
on thi
s
stim
ul
ator m
a
y
pro
duce differen
t
results, thu
s
p
r
ovidin
g e
x
tra
flexibilities for the stimulator will enri
c
h t
he BCI re
search. It is also
preferabl
e that the number o
f
trials
in every
BCI system
must be set
as sm
a
ll a
s
possibl
e a
nd
it mean
s th
at the
stimulat
or,
esp
e
ci
ally the
one
which p
r
ovides the vi
sual
stimu
li,
mus
t
operate in the very
exac
t time
and in
the pred
etermined fashio
n.
These
con
d
itions,
whi
c
h
are rel
a
ted to t
he ma
nipul
ation of visual
stimulator p
a
ra
meters,
need
to b
e
well
defined
and
adju
s
ted
to fit the g
eneral
requi
rements in th
e BCI p
r
o
g
ram.
Unfortu
nately
,
most of the resea
r
ch gro
ups involv
e
d
with BCI “do
”
it in different ways a
c
cord
ing
to not only the system
s an
d platform
s u
s
ed b
u
t al
so
to data file formats. Mo
reo
v
er, these a
r
e
relative not
o
n
ly to the ele
c
trop
hysiol
ogi
cal p
r
o
c
e
s
se
d data
(e.g.
EEG, ERP si
gnal
s) b
u
t also to
config
uratio
n
setting
s (e.
g
. feedba
ck rule, stim
ul
ator be
havior,
etc). Thi
s
fa
ct rep
r
e
s
ent
s an
obsta
cle i
n
tools
exch
ang
e amon
g different la
bo
rat
o
rie
s
which
perfo
rm si
mil
a
r BCI resea
r
ch.
Also, it is practically imp
o
ssible to
si
mulate
the b
ehavior
of a system
a
nd t
o
optimize it by
assembli
ng
module
s
from
different gro
ups a
s
a uni
q
ue way for de
scribin
g
their
cha
r
a
c
teri
stics.
A typical
situ
ation i
s
a
group
whi
c
h
is involv
ed i
n
t
he impl
eme
n
tation of
spel
lers that
woul
d like to simul
a
te t
he pe
rform
a
nce
s
of
syst
ems
built by
assem
b
ling
their d
e
velo
ped
module
s
with
othe
r o
n
e
s
a
v
ailable from
the literat
ure.
A re
ce
nt
stu
d
y [2] ha
s
de
monst
r
ated
that
unde
r
ce
rtain
co
ndition
s it
is
po
ssibl
e
to relia
bl
y predict th
e be
havior
of
a system built by
assembli
ng
different mod
u
les if a wel
l
-define
d
and
relatively si
mple de
scrip
t
ion of them
is
available. Th
is will all
o
w
the optimization of
many
BCI system
s with
out the
need of
rea
lly
building the
m
: a way to b
r
ea
k the inte
rdep
end
en
ce
of the modules and thu
s
of the various
resea
r
ch g
r
o
ups. A
n
effo
rt of p
r
ovidin
g
su
ch
“b
ridg
e” in th
e B
C
I re
sea
r
ch
are
a
has be
en
sta
r
ted
by [5] which
describ
e file formats ba
sed on t
he XML techn
o
lo
gy for storin
g some e
n
tities
freque
ntly encou
ntere
d
a
m
ong the BCI community as
well as
so
me free tool
s (BF++ Toys) that
use th
em a
n
d
that were
d
e
velope
d to
optimize
the
perfo
rman
ce
s of com
p
lete
system
s a
nd
to
document th
em. Unfo
rtun
ately, they only focu
s
ed t
heir p
r
op
ose
d
method fo
r only stori
n
g
th
e
experim
ent data. Indeed, in orde
r to be fully access
i
b
le to a wide
audien
ce, h
o
weve
r, it is
also
necessa
ry to
provide
an ef
ficient an
d ex
tensibl
e
file format pl
atform for the
sti
m
ulator
part
of a
BCI pro
g
ram
in order to fulfill the nee
d
s
of virt
ually
any scienti
s
t
and to
provid
e tools that a
r
e
able to
han
dl
e the
situatio
n. Also, a
s
p
o
inted
out by
Wen
k
e
Burde
[3], most B
C
I re
sea
r
ch te
nds
to improve B
C
I pe
rform
a
n
c
e th
rou
gh
d
e
veloping
bet
ter si
gnal
p
r
o
c
e
ssi
ng
algo
rithms b
u
t onl
y a
few whi
c
h consi
der the
aforementio
ned bri
dging
mecha
n
ism
between in
depe
ndent B
C
I
resea
r
chers
as
chall
engi
n
g
opp
ortu
nity. Maybe it
is quite difficult for BCI researche
r
s to fi
nd
approp
riate
method
s o
r
p
r
otocols
which provide the
necessa
ry pl
atform for
bui
lding a
wo
rki
ng
BCI framework
.
This pa
pe
r d
eals
with this issue, i.e. prov
idin
g a platform for
config
urin
g the visual
stimulato
r
of a BCI system
in the form of XM
L (eXtensible Ma
rkup
Lang
uage
) file format. In our
prop
osed pl
atform, we
store the
sti
m
ulator
conf
iguratio
n dat
a (the num
bers of stim
uli,
freque
nci
e
s,
etc.) in XML files in order t
o
a
llow fa
st and simpl
e
interchan
ging of
them, even on
Web,
and
at
the same
time a
ro
bust
a
nd
cu
stomiza
b
le
way of fo
rmatting th
e
m
. This pa
pe
r i
s
written i
n
the
followin
g
sy
stemati
c
. After a
br
i
e
f introdu
ction a
b
o
u
t the ba
ckground
situatio
n, a
review
abo
ut the BCI fra
m
ewo
r
k will
be de
scrib
e
d
and the X
M
L sh
ort introdu
ction
will be
provide
d
pri
o
r to the expl
anation
abou
t the prop
os
e
d
platform. A
fterwa
r
d, issu
es o
n
softwa
r
e
impleme
n
tation and te
stin
g will be expl
ained. Finally
, the paper wi
ll be clo
s
ed
with con
c
lu
sion
s.
2. METHO
D
2.1. XML and the Dev
e
lopment of the
BCI Frame
w
ork
Extensible M
a
rkup
Lang
u
age
(XML) i
s
a g
ene
ral
-
purp
o
se spe
c
ificatio
n for creating
person
a
l m
a
rkup
lan
gua
ge
s, which m
a
kes
use of t
ags
. It d
e
s
c
r
i
bes
da
ta
an
d
d
i
s
t
r
i
b
u
t
es th
em in
a form
at, whi
c
h i
s
i
nde
pen
dent from the
platform.
Thi
s
in
dep
end
en
ce
co
me
s fro
m
the fa
ct th
at
XML do
es no
t use
a
sp
ecif
ic la
ngu
age: i
n
fact,
XML
t
ags are
not p
r
edefin
ed,
so
that eve
r
yon
e
can
write hi
s
own p
e
rson
al
tags. XML is very
similar
to HTML (Hy
per Text Ma
rkup L
ang
uag
e)
but it is not it
s sub
s
titute a
s
its ai
m is
di
fferent: XML
wa
s de
sig
n
e
d
for
storin
g
and ex
cha
ngi
n
g
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOMNI
KA
ISSN:
1693-6930
■
A Novel App
r
oa
ch for
Con
f
iguring T
he
Stim
ulator of
a BCI Fram
ework…
… (Ind
ar Sugia
r
to)
75
data, while
HTML
wa
s created fo
r sh
owin
g data
i
n
a format e
a
sily re
adabl
e even if ha
rdly
adju
s
table d
u
ring time. In this way, it is ve
ry comfortable
sh
aring d
a
ta a
c
ro
ss differe
nt
laboratori
e
s
as eve
r
y re
searche
r
i
s
fre
e
to ext
end t
hem with th
e
results of a
n
analysi
s
wit
hout
disturbing the
activity of th
e other o
n
e
s
that c
an recei
v
e the new d
a
ta and
conti
nue to wo
rk with
the previou
s
l
y
developed t
ools. The e
ssential ch
ara
c
t
e
risti
c
of XML
is that data are inde
pen
de
nt.
The
co
ntent
of a file
is ke
pt se
pa
rate f
r
om its
p
r
e
s
en
tation, so
that
one
can
sto
r
e the
content
in
an XML
file o
n
ly on
ce
and
then
extra
c
t
and vi
suali
z
e
it in th
e d
e
si
red
form
at (a
ccordi
ng to
t
h
e
final format, there i
s
an ap
prop
riate XM
L techn
o
logy
that allows its generation, called XSLT).
BCI syste
m
tries to
creat
e a di
re
ct b
r
idge
bet
wee
n
hum
an
CNS (ce
n
tral
nervou
s
system
) with
a com
puter
o
r
machine th
rough n
europ
hysiolo
g
ic
sig
nals g
ene
rate
d by the brain
.
It
then cre
a
tes
a ne
w path
w
ay for the
brain to
carry
its me
ssag
e. At some
extent, it is better to
descri
be a BCI system a
s
a hybrid sy
stem in whi
c
h all of its com
pone
nts coll
a
borate tog
e
th
er to
form an integrated
sy
stem, as written i
n
IEEE
Signal Processing
M
agazi
ne Volume 25 Number
1, Jan
uary 2
008: “A b
r
ain
comp
uter int
e
rface is
a sy
stem that in
cl
ude
s a me
an
s for me
asuri
ng
neural
sign
al
s fro
m
the
brain, a m
e
thod
/algorit
hm
for
decodin
g
the
s
e
sig
nals an
d a m
e
thod
ol
ogy
for mappi
ng this de
co
ding
to a behavior or actio
n
”.
T
h
is de
scriptio
n of BCI system sou
n
d
s
m
o
re
techni
cal
sin
c
e it emph
asize
s
the imp
o
rtan
ce
of m
easure
m
ent-decodin
g
-ma
pping inte
ra
ction
within the
system. The follo
wing di
ag
ram
sho
w
s
ho
w those thre
e co
mpone
nts
wo
rk to
con
s
truct
a BCI syste
m
[6]. As Fig 1 sho
w
, the
measurem
e
n
t task i
s
pe
rformed
within
the Source
sub
-
system, de
coding is p
e
rf
orme
d by Signal
Pro
c
e
s
sing sub
-
syst
em and ma
pping tasks are
perfo
rmed b
y
the User
Applicatio
n. Operator
mo
dule a
c
ts as a central relay for syst
em
config
uratio
n and onli
ne prese
n
tation
of results to the
investigato
r
.
Fig 1. Mech
a
n
ism of a BCI
system in ge
neral
. Duri
ng
operation, informatio
n (i.e., signal
s,
para
m
eters, or event markers) is
comm
unicated
from
source to
sig
nal pro
c
e
s
sin
g
, to user
appli
c
ation, a
nd ba
ck to so
urce.
Those thre
e core compo
n
ents of any BCI syst
em
re
quire
spe
c
ific amount of compute
r
pro
c
e
ssi
ng
re
sou
r
ce in
ord
e
r to
work
as
ideal a
s
p
o
ssi
b
le a
s
in
“dre
am” B
C
I. In orde
r to mai
n
tain
its com
p
lexity, it is necessary to develo
p
a fr
ame
w
o
r
k whi
c
h e
n
ca
psul
ates all t
hose three
core
comp
one
nts
and provide
s
standa
rd p
r
otocol
s for
their commu
n
i
cation. An importa
nt re
mark
about this fra
m
ewo
r
k is th
at, as we al
ready
kn
ow, t
he hum
an n
e
rvou
s sy
ste
m
(incl
udin
g
th
e
brain
)
i
s
the
most
compl
e
x and a
dap
tive system
i
n
the world.
This title is
reflected
by the
phen
omen
on
that eve
r
y h
u
man
brain
may re
act
differently to th
e same
BCI
system. T
h
e
r
e a
r
e
subj
ect
s
who
have go
od resp
on
se o
n
certain type
of
BCI system
but the othe
r
one
s do
n’t ha
ve
it. For
som
e
subj
ect
oscill
ations in th
e
alpha
an
d b
e
t
a ban
d d
u
rin
g
moto
r im
ag
ery
work
better
than the
evo
k
ed p
o
tential
a
ppro
a
ch a
nd
vice ve
rsa.
T
h
is i
n
ter-subj
ect va
riability co
rrespon
ds to
spatial
patte
rns and
spe
c
trotemp
o
ral
chara
c
te
rist
i
c
s of b
r
ain
si
g
nals. A
go
od
BCI frame
w
ork
sho
u
ld also address this
issue. An example of
excellent app
roa
c
h for this m
a
tter is given
by
BCI2000 p
r
o
g
ram [6], whi
c
h can be vie
w
ed a
s
a BCI framewo
r
k
as well. The BCI2000
syst
em
tries to
ma
ke
a flexible B
C
I fram
ework by co
ll
abo
ra
tion of ma
ny user
define
d
su
b-p
r
og
ram
s
.
Unfortu
nately
,
this sy
stem
requi
re
s eve
r
y su
b
-
prog
ra
m to be
hard-cod
ed a
nd e
m
bedd
ed into
the
main progra
m
using lo
w-level
inter-p
roce
ss com
m
unication (IP
C
) pa
ram
e
ters before one
can
use the BCI2
000 in the re
al application
.
This
“not so use
r-fri
endl
y” appro
a
ch will become
an
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 7, No. 2, Agustus 2
009 : 73 - 82
76
obsta
cle i
n
b
r
idging i
ndep
e
ndent
system
s of ea
ch
BC
I laborat
orie
s.
Thu
s
in thi
s
pape
r, in o
r
d
e
r
to acco
mmod
a
te the futu
re
onlin
e-b
a
sed
BCI
system,
we
develo
p
t
he BCI
fram
e
w
ork usi
ng th
e
s
a
me approac
h
as
[5].
(a)
(b)
Fig 2. Block d
i
agra
m
of the BCI frame
w
ork. (a) in
syste
m
setup
scen
ario, (b
) for d
a
ta abst
r
actio
n
Since th
e dev
elopme
n
t of
BCI frame
w
o
r
k requi
re
s ex
ce
ssive
pa
rt of pro
g
ra
m code
s, the
explanation
i
n
this p
ape
r
will be fo
cu
sed on th
e vi
sual
stimulat
or p
a
rt of th
e BCI fram
e
w
ork.
Based
on the human
capability to res
pond to external
stimuli, we
c
an expect that BCI could
gi
ve
feedba
ck in
many ways.
Sometimes it is quite
difficult to determine which type of feedb
ack i
s
best for
ce
rta
i
n appli
c
ation
of a BCI system.
Even p
r
esenting m
o
re than
one t
y
pe of stimul
us
woul
d help
subj
ect to improve hi
s/her ability
to control his/
her own EE
G. For example,
combi
n
ing b
o
t
h rando
m pa
ttern and flickering p
a
ttern
will provid
e a
multimodal a
ppro
a
ch for the
BCI system
e
nhan
cem
ent. Fig 2 sho
w
s
our p
r
op
os
ed
BCI frame
w
ork i
n
sy
stem
setup
scena
rio
and the data
abstractio
n
le
vel. As it can be se
en in
Fi
g 2, there a
r
e
at least three
comp
onent
s
of
sub
-
sy
stem
s in the BCI framework system se
tup scenari
o
whi
c
h
work simult
aneo
usly: da
ta
recorder,
sig
nals vi
suali
z
ation (analy
s
is tool
s), a
n
d
ope
rato
r i
n
terface. But
in the d
a
ta
abstractio
n
l
a
yer, there
are at lea
s
t
seven maj
o
r com
pon
ent
s whi
c
h
are
requi
red to
be
synchro
n
ized
: data acqui
sition, signal
s databa
se/s
t
o
rag
e
, feature pro
c
e
ssi
ng
(extractio
n and
cla
ssifi
cation
), visuali
z
ation
(in
c
ludi
ng
sti
m
ul
ator), co
mmand
ge
ne
ration fo
r
act
uator,
com
m
and
databa
se, an
d the feedba
ck acqui
sition.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOMNI
KA
ISSN:
1693-6930
■
A Novel App
r
oa
ch for
Con
f
iguring T
he
Stim
ulator of
a BCI Fram
ework…
… (Ind
ar Sugia
r
to)
77
This
re
sea
r
ch modifie
s
th
e [7] and [5]
method
s in o
r
de
r to compl
y
with our
propo
sed
method
and
the e
n
tire
syst
em works a
s
follows. Sinc
e
it is p
o
ssibl
e
to pre
d
ict
a
collectio
n of
CIs
(co
n
trol inte
rface
s
), given
a colle
ction o
f
TRs (t
ra
ns
d
u
ce
rs
), the p
e
rform
a
n
c
e
s
of all the system
that is p
o
ssibl
e
to buil
d
by
combi
n
ing
all
the T
R
s with
all the
CIs ca
n be
cle
a
rly u
s
ed
to optimi
z
e
BCI system
s
becau
se it wi
ll be suffi
cien
t to choo
se t
he mo
st effici
ent com
b
inati
on of TRs a
n
d
CIs. All the
entities u
s
ed
for
the de
scri
ption of TRs
and
CIs w
ill
be written in
XML format
both
for storage
necessity and for eas
ily managi
ng th
e simulatio
n
s for the computation of
the
perfo
rman
ce
s indicato
rs.
The
T
R
i
s
co
mposed by
th
e
a
c
qui
sition stage, whi
c
h
deal
s with
th
e
ele
c
tro
physiologi
cal
sign
als, a
nd
by the Cl
assif
i
er, form
ed
b
y
the F
eatu
r
e
Extraction
a
nd the
Featu
r
e Cla
s
sificati
on,
w
h
os
e ta
sk
is
to e
x
tr
ac
t th
e
fe
a
t
ur
es
of in
te
r
e
s
t
fro
m
the
sig
nal
s a
nd to
trans
late them into a
logical sym
b
o
l
(LS) whi
c
h
b
e
long
s to th
e
cla
ssifie
r
lo
gi
cal al
pha
bet (LA). The
LS,
in gen
eral, ha
s
no sem
antic
meanin
g
but is just a map
p
i
ng with
som
e
subj
ect’
s p
e
rform
a
n
c
e (e.g. the character
sele
ction in t
he sp
elling p
r
ogra
m
[8]), the sele
ct
ion of
a row in P3
00 virtual keyboard task [4
]).
The
CI en
cod
e
s
seq
uen
ce
s of LS
s a
nd
turns them i
n
to sem
antic
symbols
(SS) t
hat belo
ng to
a
sema
ntic
alp
habet
(SA) b
y
the Actuato
r
pa
rt that
ca
n be u
s
e
d
to
drive a
n
external
pe
riph
e
r
al
(e.g. virtual
keyboards,
ro
bots, n
e
u
r
op
rosthe
se
s).
As an
exampl
e, if we
have
a
logi
cal
alph
a
bet
of four symb
ols
α
,
β
,
γ
,
δ
the CI
ca
n i
m
pleme
n
t an
en
codi
ng fo
r a virtu
a
l
key
board a
ppli
c
a
t
ion
that asso
ciat
es the
seque
nce
αα
α
to th
e se
manti
c
sy
mbol A, then
ααβ
to the
se
mantic
symb
ol B
and so on (u
sually one of the LSs
is re
served
for
the UNDO key).
In this way 27
semanti
c
sy
mbols
can be
encod
ed wit
h
3 LSs-lo
ng seq
uen
ce
s so that the
26 cha
r
a
c
ters of the English alphab
et plus t
he sp
ace
one can b
e
mappe
d. In addition to these
symboli
c
parameters,
the
so-cal
l
e
d
Extende
d
Confu
s
ion
Matrix
(ECM) metri
c
for the
evalu
a
tion
and
optimization of th
e p
e
rforman
c
e
s
of
BCI sy
st
ems
whi
c
h
com
b
in
es th
e info
rm
ation a
bout th
e
entities p
r
evi
ously d
e
scri
b
ed (LA, SA, Encod
e
rs
) wit
h
those de
riv
ed from
a n
e
w
comp
onent
, is
use
d
. We u
s
e
d
XML for sto
r
ing LA
s, SAs,
ECMs and
Encod
e
rs in the followi
ng format:
<?xml version="1.0" encoding="UTF-16"?>
<DOC>
<Type Name="Subject_01" SubType="Session_01"/>
<ECM>
<Row>
<In>
α
</In>
<C1>393</C1>
<C2>4</C2>
<C3>4</C3>
<C4>4</C4>
<Abstentions>45</Abstentions>
</Row>
<Row>
<In>
β
</In>
<C1>8</C1>
<C2>381</C2>
<C3>8</C3>
<C4>8</C4>
<Abstentions>45</Abstentions>
</Row>
<Row>
<In>
γ
</In>
<C1>16</C1>
<C2>16</C2>
<C3>357</C3>
<C4>16</C4>
<Abstentions>45</Abstentions>
</Row>
<Row>
<In>
δ
</In>
<C1>32</C1>
<C2>32</C2>
<C3>32</C3>
<C4>309</C4>
<Abstentions>45</Abstentions>
</Row>
</ECM>
</DOC>
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 7, No. 2, Agustus 2
009 : 73 - 82
78
In the visuali
z
ation
pa
rt, the EEG
sig
n
a
ls
ca
n b
e
d
i
splaye
d
spo
n
taneo
usly
(i
n time
domain
)
o
r
b
e
ing inte
rpol
a
t
ed for spatia
l visualizat
ion
.
After the feature
s
a
r
e ex
tracted,
ce
rta
i
n
transl
a
tion al
gorithm i
s
re
quire
d to de
code the fe
atu
r
e into exe
c
u
t
able comma
nds fo
r a
c
tua
t
or.
In order to
re
cord th
e a
c
tivity of the a
c
tu
at
or, a
comm
and
datab
ase
progr
am i
s
required t
o
ke
ep
track an
d store the comm
a
nds. After a certain me
nt
al task h
a
s b
e
e
n
given, the subje
c
t brain
will
rea
c
t in a certain manne
r whi
c
h will pro
duce spe
c
ific rhythmic sig
nals.
The
s
e feedb
ack sig
n
a
ls
are then b
e
in
g re
corded a
nd/or p
r
o
c
e
s
sed for furt
he
r
analysi
s
. In orde
r to synch
r
oni
ze all works
in the fram
e
w
ork, we de
velop a
synchroni
za
tio
n
p
r
otocol which
gene
rate ti
ming si
gnal
s and
requ
est/a
ckn
o
wle
dge
sign
als for data in
terch
ang
e bet
wee
n
su
b-systems.
2.2. The Proposed Platfo
rm for Stimulator Co
nfig
uration
Whe
n
wo
rki
n
g with exoge
nou
s BCI system, an ex
ternal
stimulat
or is re
qui
red
to elicit
certai
n EEG
pattern. In
a visu
al-b
ased BCI
stim
ulator, the
r
e
are
ge
ne
ral
req
u
irement
s for
displ
a
ying the cues. F
o
r
example i
n
SSVEP-based
BCI, the general param
e
ters will
be the
flicke
ring
ani
mation fre
q
u
enci
e
s,
sizes,
colo
rs
, sh
ap
es, po
sition
s,
feedba
ck
style, and p
a
tterns
(soli
d
or textu
r
ed
). To acco
mmodate the
s
e, we
d
e
velo
p the XML file format as foll
ows.
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE ssvep_p300>
<ssvep_p300 version="2.0">
<define
shape=
user_defined_shape
>
<self
user_defined_parameters
></self>
<bar
user_defined_parameters
></bar>
<label
user_defined_parameters
></label>
<texture
user_defined_parameters
></texture>
</define>
</ssvep_p300>
As the lab
e
l
name im
plies (i.e. ssvep_
p300
), we
de
velop the
stimulator
so th
at it can
work for SSV
EP- and P30
0
-ba
s
e
d
BCI or
co
mbin
atio
n of both BCI types.
De
sc
ription
:
The XM
L fil
e
mu
st h
a
ve do
cum
ent
ID a
s
ssve
p_p3
00
and
it mu
st co
ntain attrib
ute
versio
n=”2.0
”. The first
line, which con
s
i
s
ts of statement <?xml
versio
n=”1.
0
”
encodin
g
=”UTF-8
”
?
>
may be omitted bu
t it is not reco
mmend
ed.
Every ‘define’
element will
start a ne
w flicke
rin
g
anim
a
tion definitio
n and its sha
pe is mu
st
be spe
c
ified i
n
the ‘
s
hap
e’
attribute. Th
e followi
ng p
o
ssible val
u
e
s
for ‘
s
h
ape’
attribute a
r
e:
recta
ngle, tria
ngle, and
circle. Note that t
he attribute’
s value is case inse
nsitive.
After defining
flickeri
ng ani
mation’s
sha
pe, the
next step is creatin
g four elem
e
n
ts within the
‘define’ elem
ent. These four elem
ents a
r
e
‘self’, ‘ba
r
’, ‘label’, and ‘texture’.
‘self’ elem
ent
will be
used
to descri
be t
he flic
keri
ng
animation it
self and it con
t
ains
several
attributes. Th
ese attri
bute
s
, whi
c
h resi
de in an ele
m
ent definition, may be spe
c
ified i
n
rand
om orde
r. The followin
g
attributes a
r
e valid for ‘sel
f’ element:
o
siz
e
will determine the physical si
ze of the animation in pixels;
o
posx
and
po
sy
will
det
ermin
e
the
to
p-left
corner f
o
r
squ
a
re a
n
d
ci
rcle flicke
ring
animation a
n
d
determi
ne the first point f
o
r trian
g
le ani
mation;
o
posx2, po
sy2
,
posx3, posy
3
will be used ex
clusively for determi
ning the second
and the third
points of a tri
angle flicke
rin
g
animation;
o
freq
will de
termine the freque
ncy of flickerin
g
anim
a
tion;
o
colo
r
will determi
ne
flickeri
ng animat
i
on’s col
o
r;
possible
val
u
es
are: white, red,
gree
n, blu
e
,
cyan, ma
gent
a, yellow,
gra
y
, darkre
d
, d
a
rkgreen,
darkblu
e, da
rkcy
an,
darkma
genta,
darkyell
o
w,
d
a
rkgray, and lightgray
o
texture
will
determin
e
if the flickeri
n
g
animation h
a
s
texture or n
o
t;
‘bar’ elem
ent will be used to spe
c
ify the appe
ara
n
ce and beh
avior of feedback bar for ea
ch
flicke
ring a
n
i
m
ation. The followin
g
attrib
utes are valid
for ‘bar’ elem
ent:
o
type
will
determin
e
the style of feedba
ck ba
r; possi
ble v
a
lue
s
are: none,
individual, hal
fleft, halfright, and com
b
ination
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOMNI
KA
ISSN:
1693-6930
■
A Novel App
r
oa
ch for
Con
f
iguring T
he
Stim
ulator of
a BCI Fram
ework…
… (Ind
ar Sugia
r
to)
79
o
colo
r
will
determi
ne th
e color of fe
edba
ck b
a
r;
possibl
e valu
es
are: white
,
red,
gree
n, blu
e
,
cyan, ma
gent
a, yellow,
gra
y
, darkre
d
, d
a
rkgreen,
darkblu
e, da
rkcy
an,
darkma
genta,
darkyell
o
w,
d
a
rkgray, and lightgray
o
bru
s
h
will
determi
ne the filling rule for the bar
; possible values are solid and line
o
siz
e
will determine the width of t
he bar; the height of the bar i
s
f
o
llows the
size of
the
flicke
ring animation;
o
frame
will
determi
ne if the feedba
ck bar is di
splay
ed with a fra
m
e or not, so
the
possibl
e values are yes o
r
no
o
frame
c
olo
r
if the
feedba
ck b
a
r ha
s a frame, t
hen this attribute
will determi
n
e
its
colo
r; the possible value
s
are the same
with
anothe
r colo
r definitio
n; if this attribute
is ignored, then the fram
e
color will be
set the same
color with its parent
o
blinki
ng
will determin
e
wheth
e
r the f
eedb
ack ba
r
sho
u
ld al
so
blinki
ng with
the
same f
r
eq
uen
cy as th
e flickering
animati
on itself; ho
wever, it is recommen
ded to
set
this value a
s
no to avoid p
e
rform
a
n
c
e d
egra
dation
when u
s
ing ma
ny animation
s
o
threshold
will determine
the threshol
d
value of feedback ba
r;
Fig 3. Severa
l possible fee
dba
cks di
spla
ys
: (a) NO_FEEDBACK_B
A
R, (b) INDIVIDUAL, (c
)
HALF_
R
IG
HT, (d) HAL
F
_
L
EFT, (e)
CO
MBINATIO
N, and (f) a
re
su
lt for combini
ng seve
ral
feedba
ck types.
(a)
(b)
Fig 4. Two po
ssi
ble ap
pea
rances
of feed
back ba
r: soli
d (left) or di
screte (right).
‘label’ elem
e
n
t will be used to spe
c
ify flicke
ri
ng a
n
i
m
ation’s la
be
l and it will b
e
displ
a
yed
anywh
ere o
n
the scree
n
, depen
ding on t
he sp
ecifie
d value
s
belo
w
:
o
text
any ASCII text in th
is
attribute is
valid
as a flickeri
ng a
n
imat
ion’s la
bel; if this
attribute is NULL, then the
flickeri
ng ani
mation will be
displaye
d wit
hout label
o
siz
e
will d
e
t
ermine the fo
nt size of the
label;
o
colo
r
will specify the col
o
r of the text; the
possible
values are the same
with the
other colo
r’s
attribute abov
e
o
posx
a
nd
po
sy
will
dete
r
mine the
exa
c
t po
sition
of
the text; if these
attribute
s
are
negative, the
n
the text will
be displayed
on
the ce
nter
of the flicke
ring animatio
n
;
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 7, No. 2, Agustus 2
009 : 73 - 82
80
‘texture’ ele
m
ent will
be
use
d
to
spe
c
i
f
y fli
ckeri
ng
a
n
imation’
s te
xture. If textures a
r
e g
o
ing
to be u
s
e
d
fo
r flicke
ring
an
imation a
n
im
ations, th
e te
xture attrib
ute
in the ‘
s
elf’
el
ement m
u
st
be s
e
t as
‘on’ firs
t. The following a
ttribut
es
are valid for this
element:
o
firs
t
sp
ecify the full-p
a
th lo
cation
of the first t
e
xture im
ag
e filena
me;
any
recogni
za
ble
image by Qt may be use
d
o
se
con
d
sp
ecify the full
-path lo
cation
of t
he
se
con
d
texture i
m
a
ge filen
a
me;
any
recogni
za
ble
image by Qt may be use
d
Every eleme
n
t must be
clo
s
ed
with
approp
ri
ate
clo
s
ing tag.
Otherwise, it will not be
recogni
ze
d a
s
a
n
ele
m
ent
in the XML
d
o
cum
ent fo
rm
at. This XM
L
document
mu
st be
save
d
in stand
ard A
S
CII format (with extensi
o
n .xml).
In our proposed m
e
thod,
we al
so
devel
op several t
y
pe of f
eedback bars for
SSVEP-
based BCI.
Usi
ng the
co
nfiguratio
n file explain
ed
above, the fo
llowing
feed
b
a
ck ba
rs can
be
ins
e
rted toget
her with the flic
ke
ring
s
t
imulator for SS
VEP-based B
C
I (c
an be
s
e
en in Fig 3
an
d
Fig 4).
3.
IMPLEMENTA
TION A
ND DISCUS
SION
The pro
g
ram
is develope
d using
C++ in Qt
’s ope
n sou
r
ce fra
m
ewo
r
k. The
current
versio
n of
Qt
bein
g
u
s
e
d
i
n
this pa
pe
r i
s
Qt
4.3.3. Si
nce
the f
r
ee
-l
icen
se
Qt i
s
use
d
, Mi
cro
s
oft
Visual Studi
o
can
not be
u
s
ed
as an I
D
E (Integr
ated
Develo
pmen
t Environme
n
t
) for d
e
velop
i
ng
the prog
ram,
inste
ad, Ecli
pse
version
3.3.1.1 (E
u
r
o
pa versio
n)
with Qt-Inte
g
r
ation
plug
-in
is
use
d
. As a
n
open
-sou
rce
mandato
r
y of
Qt and E
c
lip
se, g
+
+ versi
on 3.4.5
of M
i
nGW i
s
used
as
the compil
er.
Whe
n
usi
ng g
++ a
s
compil
er, gdb
can b
e
use
d
as the
debug
ger.
Qt supports t
w
o m
odel
s of
XML implem
entat
ion: SAX (Simple AP
I for XML) and DOM
(Document Object Model).
The main
different
of SAX and DOM is
SAX report
s
"parsing
event
s"
dire
ctly to the application throu
gh virtual
functi
on
s, wh
ile DOM conv
erts an XML
document int
o
a
tree st
ru
cture
,
which the a
pplication
ca
n then n
a
viga
te. Qt also p
r
ovides it
s o
w
n way to rea
d
or
write XML which i
s
calle
d QXml module (whi
ch
con
s
i
s
ts of several u
s
eful
class such as
QXmlAttributes, QXml
StreamRead
er, QXml
S
t
ream
Write
r, etc). I
n
this p
aper,
QXmlStream
Rea
der i
s
used for re
adin
g
config
uratio
n file.
After compili
ng the BCI framework p
r
o
g
ram,
we
p
r
o
v
ide the XML
configu
r
atio
n
file for
config
urin
g th
e stimulato
r
.
For exam
ple,
the
followin
g
XML script
will define
a
BCI appli
c
ati
o
n
whi
c
h con
s
ist
s
of four flickering a
n
imati
ons: tw
o
squ
a
re
-shap
e (e.
g
. the first one with texture
and
feedba
ck bar and the second one
with
neither text
ure nor fee
d
b
a
ck bar), one
textured circle-
sha
pe with fe
edba
ck bar, a
nd one textured triangl
e sh
ape with fee
d
back ba
r.
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE ssvep_p300>
<ssvep_p300 version="2.0">
<define shape="rectangle">
<self size="100" posx="0" posy="0" posx2="" posy2="" posx3="" posy3=""
freq="13.5" color="white" texture="no"></self>
<bar type="none" color="white" brush="solid" size="20" frame="no"
framecolor="ignore" blinking="no" threshold=""></bar>
<label text="" size="" color="" posx="" posy=""></label>
<texture first="" second=""></texture>
</define>
<define shape="rectangle">
<self size="100" posx="200" posy="0" posx2="" posy2="" posx3=""
posy3="" freq="15.0" color="yellow" texture="yes"></self>
<bar type="halfleft" color="white" brush="line" size="20" frame="yes"
framecolor="red" blinking="yes" threshold="0"></bar>
<label text="two" size="20" color="green" posx="200" posy="50"></label>
<texture first="default_texture1.bmp" second="default_texture2.bmp">
</texture>
</define>
<define shape="triangle">
<self size="100" posx="0" posy="200" posx2="100" posy2="200"
posx3="50" posy3="300" freq="17.5" color="blue" texture="yes">
</self>
<bar type="individual" color="red" brush="solid" size="20" frame="yes"
framecolor="red" blinking="no" threshold="200"></bar>
<label text="three" size="20" color="green" posx="0" posy="250"></label>
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOMNI
KA
ISSN:
1693-6930
■
A Novel App
r
oa
ch for
Con
f
iguring T
he
Stim
ulator of
a BCI Fram
ework…
… (Ind
ar Sugia
r
to)
81
<texture first="default_texture1.bmp" second="default_texture2.bmp">
</texture>
</define>
<define shape="circle">
<self size="100" posx="200" posy="200" posx2="" posy2="" posx3=""
posy3="" freq="10" color="red" texture="yes"></self>
<bar type="halfright" color="red" brush="solid" size="20" frame="yes"
framecolor="red" blinking="no" threshold="200"></bar>
<label text="this is a circle LED" size="20" color="yellow" posx="100"
posy="200"></label>
<texture first="default_texture1.bmp" second="default_texture2.bmp">
</texture>
</define>
</ssvep_p300>
Fig 5 sho
w
s the re
sult for t
he
above con
f
iguration scri
pt.
Fig 5. Displ
a
y result fro
m
e
x
perime
n
t wi
th the given XML config
uration script.
Fig 6. The co
mplete BCI framework in a
c
tion. T
he BCI framewo
r
k u
s
e
s
XML file config
uratio
n to
control all wid
gets on the
scre
en.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 7, No. 2, Agustus 2
009 : 73 - 82
82
The p
r
op
ose
d
method
ha
s be
en p
r
e
-
e
v
aluated a
n
d
impleme
n
ted
in real
appli
c
ation
of
BCI system f
o
r spellin
g progra
m
ap
plication. It
has b
een d
e
mon
s
t
r
ated
su
ccessfully as
rep
o
rted
by [8] and [1]
as
a result
of publi
c
exp
e
rime
nt
in th
e event of
CeBIT 2008
in
Han
nove
r
. F
i
g 6
sho
w
s the screen
shot re
sul
t
of the propo
sed
BCI fra
m
ewo
r
k a
nd its configu
r
ation
platform.
4. CONCL
U
S
I
ON
A novel approach in provi
d
ing flexible
vis
ual stimul
a
t
or usin
g XM
L has b
een p
r
esented
and a
pplied f
o
r devel
opm
ent of a BCI
(brain-co
m
p
u
ter inte
rface
)
frame
w
o
r
k. Usi
ng XML
file
format
fo
r co
nfigurin
g
the
visual stimula
t
or
of
a B
C
I
system,
we
can d
e
velop
B
C
I ap
plicatio
ns
whi
c
h
ca
n a
c
comm
odate
many expe
ri
ment st
rat
egi
es i
n
BCI
re
search. T
he B
C
I fram
ework and
its co
nfigu
r
ati
on platfo
rm i
s
develop
ed u
s
ing
C++ pro
g
rammi
ng la
n
guag
e which
inco
rpo
r
ate
Q
t
’s
most p
o
werf
ul XML pa
rser n
a
med
Q
X
mlStr
eam.
The impl
eme
n
tation an
d
experim
ent shows
that
the
XML config
uratio
n file
can be well
exec
uted within
the pro
posed
BCI framework. Be
side
its capability in presenting flexible flickering
frequencies and text
formatting for SSVEP-based
BCI, the con
f
iguration
pla
tform only li
mited to p
r
o
v
iding 3
sh
a
pes,
16
colo
rs, an
d 5
dist
inct
feedba
ck ba
rs. It is not n
e
ce
ssary to i
n
crea
se
the
numbe
r of shape
s no
r co
lors
sin
c
e th
ose
para
m
eters
are le
ss imp
o
rtant for th
e BCI
stimul
ator. The p
r
opo
sed m
e
th
od can then
be
extended to
enha
nce the
usa
b
ility of
currently exist
ed BCI fram
e
w
ork su
ch as
BF++
Toys and
BCI 2000.
REFERE
NC
ES
[1]
Allison, BZ
et
al.,
“BCI De
mographics
I: Ho
w
ma
n
y
(an
d
w
h
a
t
ki
nds o
f
) pe
op
le can
us
e
an SSVEP BCI?”
, Procee
ding of
4
th
Internatio
nal B
r
ain-Com
put
er Interface
Wo
rksho
p
a
nd
Traini
ng Course 20
08, Verl
ag der T
e
chni
sc
hen
Universität Gra
z
– Austria, 20
08.
[2]
Bianchi
L,
Quitadam
o
L
R
, Ga
rreffa
G, Ca
rd
arilli
GC,
Ma
rci
a
ni MG.,
“Pe
r
forman
ces
ev
aluation and optimiza
t
ion of Brain
-
Compu
t
er In
terface sy
stems in a cop
y
spellin
g
task
”
,
IEEE Transactions on Neural
Sy
stem
s
and Rehabilitation Engi
neering, 15(2):
207-
216, 200
7.
[3]
Burde
W, an
d Blankert
z
B.,
“Is the locus o
f
con
t
rol of r
e
inforcemen
t a
predictor o
f
brain-c
o
mpu
t
er inte
rfa
c
e
performa
nc
e?”
, Pro
c
e
e
d
ing
s
of the 3rd Internatio
nal Brain-
Computer Interfac
e
Works
h
op and Training Cours
e
2006, Verlag
der Tec
h
nisc
hen
Universität Graz: 76-77, 20
06.
[4] Millán,
J
.,
“Adap
t
iv
e Brain Inter
f
a
ces
for
Co
mmunicatio
n and
Con
t
rol”
, 10th
Internation
a
l Confe
r
en
ce o
n
Huma
n-
Co
mputer Inte
ra
ction, Gree
ce
, 2003.
[5]
Quitadam
o L
R
, Marciani
MG, Bianchi
L.,
“Optimizatio
n of Br
ain Computer Interface
s
y
stems by
means of XML a
nd BF
++
To
y
s
”
, Internation
a
l Jo
urnal
of
Bioelect
r
oma
gnetism, 9
(
3):
172- 18
4, 20
07.
[6]
Schal
k G,
M
c
Farla
nd DJ, Hinterbe
rge
r
T,
Birba
u
me
r N, Wolpa
w
JR.,
“
BCI2
000
: a
gen
e
ral-
purpos
e bra
i
n–computer
interface
(BCI) s
y
stem”
, IEEE Trans
a
c
t
ion
on
Biomedic
a
l
Enginee
ring, 51(6
)
:10
34–
4
3
,
2004.
[7]
Sugiarto I, and Handoyo
I.,
“Applicati
on of Distributed
Sy
stem in Neuroscience, A
Cas
e
Stud
y
of BCI Fra
m
e
w
o
r
k
”
,
Proceedi
ng of
The First Internatio
nal
Seminar
on
Scien
c
e an
d Tech
nolo
g
y (ISSTEC 2009
), Universita
s Islam Indon
e
s
ia, Jo
gja
k
art
a
, 2009.
[8]
Sugiarto I, Allison BZ, Gräser A.,
“
O
ptimiz
ation
Strateg
y
for SSVEP-Based BCI in
Spelling Program Applic
ation
”
, Proceedin
g
of Internatio
nal Confere
n
ce on
Compute
r
Enginee
ring
and T
e
chnol
ogy 200
9 (I
CCET 2
009
),
Internatio
nal
Asso
ciatio
n o
f
Comp
uter
Scien
c
e an
d Informatio
n Tech
nolo
g
y (IACSIT), Singa
pore, 20
09.
Evaluation Warning : The document was created with Spire.PDF for Python.