Indonesi
an
Journa
l
of El
ect
ri
cal Enginee
r
ing
an
d
Comp
ut
er
Scie
nce
Vo
l.
13
,
No.
2
,
Febr
uar
y
201
9
, pp.
721
~
72
8
IS
S
N: 25
02
-
4752, DO
I: 10
.11
591/ijeecs
.v1
3
.i
2
.pp
721
-
72
8
721
Journ
al h
om
e
page
:
http:
//
ia
es
core.c
om/j
ourn
als/i
ndex.
ph
p/ij
eecs
F
eatur
e anal
ysis for sta
ge identific
atio
n
of
Plasmo
diu
m
viva
x
based on
digital
micros
copic ima
ge
Ha
n
un
g Adi
Nugro
ho
1
,
I
Md. De
ndi
M
ay
s
anj
aya
2
,
N
oo
r
Akhm
ad
Setiaw
an
3
E. El
sa
Herdi
ana
Murhan
d
arwa
ti
4
, Widh
ia K.
Z
Ok
to
e
b
erz
a
5
1,
2,3,5
Depa
rtment
of
Elec
tr
ical and
Inform
at
ion
Tec
hnolog
y
,
Fa
cul
t
y
of
Eng
ineeri
ng
,
Univer
sita
s Gad
j
ah
Mada
,
Yog
y
a
kar
ta,
Indon
esia
2
Depa
rtment of I
nform
at
ic
s E
du
c
at
ion
,
Fa
cul
t
y
of
Engi
n
ee
ring
an
d
Voca
t
ion,
Uni
ver
sita
s Pend
idikan
Gane
sh
a
,
In
donesia
4
Depa
rtment of
Para
sitol
og
y
,
Fa
cul
t
y
of
Med
ic
in
e,
Univ
ersitas Gadj
ah
Mada
,
Yo
g
y
ak
arta, Indones
ia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Oct
9,
2018
Re
vised N
ov
19, 2
018
Accepte
d
Nov
28, 201
8
Pl
asm
odium
par
asit
e
is
ide
n
ti
fi
ed
to
conf
irm
m
al
a
ria
dise
ase
.
Para
m
edi
cs
nee
d
to
observe
the pr
ese
nce of
th
is parasit
e
pre
par
ed
o
n
thi
ck
and
thi
n
blood
fil
m
s
under
m
ic
roscop
e.
How
eve
r
,
fa
lse
ide
n
ti
fi
ca
t
ion
stil
l
o
cc
urs
which
is
ca
use
d
b
y
hum
an
f
ac
tor
durin
g
th
e
exa
m
i
nat
ion
.
Thus,
m
a
la
ri
a
ide
nt
ifi
c
atio
n
base
d
on
digi
tal
image
pr
oce
ss
ing
h
as
b
e
en
wid
ely
dev
eloped
to
ov
erc
o
m
e
the
err
or
poss
ibi
li
t
y
.
Thi
s
pape
r
proposes
a
sche
m
e
to
id
ent
i
f
y
and
class
if
y
th
e
stage
s
o
f
Pl
asm
odium
vi
v
ax
par
asi
te
on
d
i
git
al
m
i
cro
sco
pi
c
image
of
thi
n
blood
fil
m
s
base
d
on
f
ea
tu
r
e
an
aly
sis.
Sha
pe
and
te
x
ture
fea
tur
es
ar
e
ex
t
rac
t
ed
from
segm
ent
ed
par
asit
e
ob
je
c
ts.
Fea
tu
re
select
ion
base
d
on
wrappe
r
m
e
thod
is
the
n
conduc
t
ed
to
obt
ai
n
re
le
van
t
fe
atures
which
m
a
y
cont
ribute
in
improving
th
e
cl
assifi
ca
t
ion
res
ult
.
Th
e
class
ifi
c
at
ion
proc
ess
is
conduc
t
ed
b
ase
d
on
Naïv
e
Ba
y
es
cl
assifi
er.
Th
e
p
erf
orm
ance
of
proposed
m
et
hod
is
eva
lu
ate
d
using
73
digi
tal
m
ic
rosco
pic
images of
P
-
vi
va
x
p
ara
sit
e on
thi
n
blood
fil
m
s
comprisin
g
of
29
tropho
zoi
t
es,
10
sch
iz
onts
and
34
game
to
c
y
te
s
stage
s
.
B
y
using
six
sele
c
te
d
fe
at
ure
s
including
per
i
m
et
er,
dispersio
n,
m
ea
n
of
in
tens
ity
,
AS
M,
cont
rast
GLCM
and
en
trop
y
GL
CM,
the
proposed
sche
m
e
ac
h
ieves
the
be
s
t
cl
assifi
ca
t
ion
r
ate
with
th
e
ac
cu
r
acy
,
sensit
ivi
t
y
and
spec
ificity
of
97.
29%
,
97.
30%
and
97
.
3
0%,
r
espe
ctively.
Th
is
ind
ic
a
te
s
t
hat
the
proposed
sche
m
e
has
a
pot
ent
i
al
to
be
implement
ed
in
the
dev
el
opm
e
nt
of
a
compute
rised
a
ided
m
al
ari
a
di
agnosi
s s
y
stem
for
assi
sting
th
e
pa
rame
dic
s
.
Ke
yw
or
d
s
:
Feat
ur
e
an
al
ysi
s
Stages
of p
la
s
m
od
iu
m
v
iva
x
Thin bl
ood fil
m
Wr
a
pper
featu
r
e sele
ct
ion
Copyright
©
201
9
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
Hanu
ng Adi
N
ugr
oho,
Dep
a
rtm
ent o
f El
ect
rical
En
gi
neer
i
ng and
Inf
or
m
at
ion
Tec
hnol
og
y,
Faculty
of E
ngineerin
g, U
nive
rsita
s G
a
djah
Ma
da,
Jl. Grafi
ka 2
K
a
m
pu
s
UG
M
, Yo
gyaka
rta
55281, I
ndonesi
a.
Em
a
il
: adinu
gr
oho@u
gm
.ac.id
1.
INTROD
U
CTION
Ma
la
ria
is
a
disease
ca
us
e
d
by
Plas
modium
par
asi
te
w
hich
is
tran
sm
i
tt
ed
to
hum
ans
th
rough
the
bite
of
fem
al
e
Anopheles
m
os
quit
os
.
As
re
porte
d
by
Wor
ld
H
eal
th
Org
anisat
ion
(
W
HO),
this
dis
ease
are
transm
itted
in
m
or
e
than
90
c
ountries
a
nd
pu
t
about
3.2
bill
ion
pe
ople
at
ris
k
of
m
al
aria
wi
th
m
ai
nly
m
or
bid
it
y
occur
i
n
A
fr
ic
a,
S
ou
t
h
-
Ea
st
Asia,
Lat
in
A
m
erica
and
the
Mi
dd
le
Ea
st
[
1]
.
Plas
m
od
i
um
is
div
ide
d
i
nto
five
sp
eci
es,
i.e.
Pl
as
m
odiu
m
f
alc
ipa
r
um
(
P.
fal
ci
pa
r
um
),
Pl
asmo
dium
vi
vax
(
P
.
vi
v
ax
)
,
Plasmo
dium
ov
al
e
(
P.
ova
le
),
Pla
sm
odiu
m
m
ala
ri
ae
(
P.
ma
l
ar
iae
)
and
Pl
asmo
di
um
k
nowl
esi
(
P.
kn
owlesi
).
The
gr
eat
est
th
reat
of
m
al
aria causes
com
es f
ro
m
P.
falci
pa
r
um
a
nd
P. viv
ax
.
[1]
.
The
Pla
sm
odium
unde
rgoes
two
ph
ase
s
du
rin
g
the
in
fecti
on
process
of
the
hum
an
body,
nam
el
y
exo
e
ryt
hrocyt
ic
phase
in
the
l
iver
a
nd
intrae
ryt
hrocyt
ic
ph
a
se
in
blood
str
ea
m
ci
rcu
la
ti
on
.
In
the
bloo
dst
ream
ci
rcu
la
ti
on,
it
will
go
th
r
ough
oth
e
r
t
hr
ee
sta
ges,
i.e.
t
rophoz
oite
s,
sc
hi
zon
ts
an
d
gam
et
ocyt
es
sta
ge
s
[
2]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
13
, N
o.
2
,
Fe
bru
ary 2
019
:
721
–
72
8
722
Figure
1
s
how
s
the
li
fe
cy
cl
e
of
m
al
aria.
When
Plas
m
od
i
um
infecti
on
is
s
us
pe
ct
ed,
thic
k
and
thi
n
bl
ood
film
s
pr
e
par
at
io
n
wi
ll
be
m
ade.
Ex
a
m
inati
on
on
t
hick
bl
ood
ai
m
s
to
detect
th
e
prese
nce
of
Plasmo
dium
pa
rasit
es
wh
il
e the
thi
n blo
od f
il
m
ex
am
inati
on
is to
iden
ti
fy
w
hat s
pecies
of
Plas
modium
ca
us
i
ng the
d
ise
ase
.
Figure
1. The
li
fe cycl
e of m
a
la
ria [2
]
A
false
diag
no
sis
on
thin
bl
ood
film
exa
m
i
nation
can
be
aff
ect
ed
by
s
om
e
factor
s
part
ic
ularly
the
exp
e
rtise
le
vel
of
par
am
edics,
the
blood
fil
m
pr
epara
ti
on
m
et
ho
d,
t
he
s
ta
ining
m
et
hod
an
d
the
qual
it
y
of
m
ic
ro
sco
pe
use
d.
Hen
ce
,
se
ve
ral
stud
ie
s
ha
ve
bee
n
c
ondu
ct
ed
to
dev
el
op
com
pu
te
r
-
a
i
ded
m
al
aria
diagnosis
base
d on di
gital
i
m
age p
r
oces
sing t
o red
uce t
he
e
rror p
os
si
bi
li
t
y.
Kh
a
n
et
al
.
[
3]
app
li
ed
k
-
m
ean
s
on
b
cha
nnel
of
the
L*a
*b
c
olour
m
od
el
to
segm
ent
P.
vi
v
ax
par
a
sit
e.
Howe
ver,
t
he
k
val
ue
wa
s
det
erm
ined
m
anu
al
ly
,
an
d
t
he
vi
su
al
qual
it
y
of
segm
entat
ion
r
esult
was
po
or
.
Na
sir
et
al
.
[
2]
em
pl
oyed
the
c
om
bin
at
ion
of
m
oving
k
-
m
eans
cl
us
te
rin
g
(M
K
M)
a
nd
see
ded
re
gion
gro
wing
a
rea
extracti
on
(S
R
GA
E
)
m
et
hods
to
i
den
ti
fy
P
.
v
iv
ax
.
T
heir
stu
dy
pr
ov
e
d
t
hat
the
use
of
sat
ur
at
ion
(S
)
band
of
H
SI
colo
ur
m
od
el
was
able
to
ob
t
ai
n
bette
r
segm
entat
ion
r
e
su
lt
t
han
th
at
of
inte
ns
it
y
(
I)
band.
Dian
et
al.
[
4]
de
te
ct
ed
the
bloo
d
cel
l
com
po
ne
nt
i
n
r
ed
t
hin
blood
s
m
ear
by
a
pp
ly
ing
global
t
hr
es
ho
l
ding
an
d
co
nn
ect
e
d
com
ponent
la
belli
ng
(CCL).
R
ub
e
rto
et
al
.
pro
posed
the
com
bin
at
ion
of
a
u
tom
atic
thres
ho
l
ding
a
nd
m
or
phol
og
ic
a
l
appr
oach
to
de
te
ct
and
cl
assif
y
m
al
aria
par
as
it
es
[5]
.
Furthe
rm
or
e,
A
kbar
e
t
al
.
[6]
i
ntrod
uced
com
bin
at
ion
of
k
-
m
eans
cl
us
te
rin
g
a
nd
m
or
phol
og
ic
al
oper
at
ion
m
et
hods
on
HSV
c
olour
m
od
el
to
se
gm
ent
P.
falci
parum
on
the
thin
bl
ood
fil
m
s.
The
n,
s
ever
al
s
ha
pe
a
nd
te
xture
featur
es
we
re
e
xtr
act
ed
an
d
cl
as
sifie
d
by
us
i
ng
ML
P
cl
assifi
er
to
cl
assify
P.
falci
parum
sta
ge
into
t
hr
ee
cl
ass
es,
i.e.
tr
opho
zoite
s,
sc
hizo
nt
s
an
d
gam
et
o
cy
te
s.
Howe
ver,
t
he
determ
inati
on
of
the
cl
u
ste
r
num
ber
in
k
-
m
e
an
was
sti
ll
m
a
nu
al
an
d
the
obta
ined
featu
re
s
wer
e
sti
ll
too
m
any.
To
c
om
plete
the
ide
ntific
at
ion
stu
dy
of
Plas
modium
pa
rasit
e,
this
pa
per
pr
opos
es
a
sc
he
m
e
to
cl
assify
P.
vi
vax
pa
rasit
e
on
dig
it
al
m
ic
ro
sc
op
ic
im
a
ge
of
t
hin
bloo
d
film
s.
The
cl
assifi
cat
ion
is
cat
egorised
i
nto
th
re
e
sta
ges,
i.e.
tr
opho
z
oites,
sc
hizon
ts
a
nd
gam
eto
cy
te
s.
The
m
a
in
pur
po
se
of
t
his
st
ud
y
is
t
o
ob
ta
in
t
he si
gnific
ant
featur
e
s
f
or
im
pro
ving
the
cl
assifi
cat
ion
res
ul
t
based
on
w
ra
pp
e
r s
ub
set
e
va
luati
on
.
The s
tructu
re
of
this
pap
e
r
is
organ
ise
d
as
fo
ll
ows.
Sect
ion
I
I
il
lustrate
s
the
ex
pe
rim
en
ta
l
set
up
.
T
he
resu
lt
s
a
nd
d
is
cussion
a
re
pre
sente
d
in Secti
on
III foll
ow
e
d by c
oncl
us
i
on in Sec
ti
on
IV
2.
APP
ROAC
H
The
m
et
ho
dolog
y
c
onsist
s
of
five
m
ai
n
processes
,
nam
ely
pr
e
-
proce
ssing,
se
gm
entat
i
on,
feat
ure
extracti
on,
fea
ture
sel
ect
ion
and
cl
assifi
cat
ion
as
de
picte
d
i
n
Fi
gure
2.
The
fi
rst
tw
o
proce
sses,
i.e
.
pr
e
-
processi
ng
an
d
segm
entat
ion
,
are
c
onduct
ed
by
ad
opti
ng
the
pro
posed
sche
m
e
in
our
previ
ou
s
w
ork
[
7]
.
F
irstl
y,
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c
En
g
&
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Feature
an
alys
is f
or
st
ag
e
ide
ntif
ic
ation
of
pl
as
m
od
i
um viv
ax base
d o
n di
gital …
(
H
anung A
di Nu
groho
)
723
the
Ro
I
im
age
with
the
res
olut
ion
of
250x
250
pix
el
s
is
cr
op
ped
f
ro
m
the
ori
gin
al
im
age.
The
re
d
a
nd
sat
urat
io
n
bands
a
re
us
e
d
in
this
stud
y.
The
n,
c
on
tra
st
stret
chin
g,
a
nd
m
edian
filt
er
are
ap
plied
to
enh
a
nce
the
qu
al
it
y
of
RoI
im
age.
F
urt
her
m
or
e,
Otsu t
hr
es
holdin
g
a
nd m
or
phol
og
i
cal
o
pe
rati
ons
are c
onduct
ed
t
o
se
gm
ent
P. v
iv
ax.
T
h
i
n
b
l
o
o
d
f
i
l
m
i
m
a
g
e
P
r
e
-
p
r
o
c
e
s
s
i
n
g
S
e
g
m
e
n
t
a
t
i
o
n
F
e
a
t
u
r
e
e
x
t
r
a
c
t
i
o
n
C
l
a
s
s
i
f
i
c
a
t
i
o
n
A
c
c
u
r
a
c
y
,
s
e
n
s
i
t
i
v
i
t
y
,
s
p
e
c
i
f
i
c
i
t
y
I
n
p
u
t
P
r
o
c
e
s
s
O
u
t
p
u
t
F
e
a
t
u
r
e
s
e
l
e
c
t
i
o
n
Figure
2. Bl
oc
k diag
ram
o
f
th
e ap
proac
h
2.1.
Fe
ature
ext
r
act
i
on
The
se
gm
ented
im
age
su
bs
e
qu
e
ntly
unde
r
go
e
s
feat
ur
e
e
xtracti
on
pr
oc
ess
base
d
on
the
s
hap
e
an
d
te
xtu
re
featu
res
.
The
sh
a
pe
fea
ture
c
om
pr
isi
ng
the
c
onto
ur
-
ba
sed
a
nd
i
nvari
ant
m
o
m
ent
featur
es.
For
the
te
xture
featur
e
,
hist
ogr
a
m
-
based
an
d
GLCM
featu
re
s
are
ext
r
act
ed.
Ther
e
are
se
ve
n
co
ntou
r
-
base
d
feat
ur
es
i
nclud
i
ng
per
im
et
er,
area,
rou
ndness
,
sli
m
ness,
co
nvex
it
y,
so
li
dity
a
nd
dis
per
si
on.
P
erim
e
te
r
re
pr
e
s
ents
t
he
ed
ge
l
eng
t
h
of
a
n
ob
j
ect
as
form
ulate
d
in
(
1).
The
obj
ect
with
4
-
ad
j
ace
nc
y
obta
in
bette
r
res
ult
of
pe
ri
m
et
er
tha
n
t
hat
of
8
-
adj
ace
ncy.
He
r
e,
is
an
e
ven
num
ber
of
c
od
e
s
an
d
is
an
od
d
nu
m
ber
of
c
od
e
s.
Ar
ea
is
t
he
total
of
pi
xe
ls
obj
ect
as
cal
culat
ed
in
(
2).
The
nota
ti
on
of
an
d
re
present
the
ob
j
e
ct
area
a
nd
ed
ge
of
t
he
obj
e
ct
,
resp
ect
ively
.
=
+
√
2
(1)
=
∬
=
∫
(
)
(
)
−
∫
(
)
(
)
(2)
Roun
dn
es
s
is
t
he
rati
o
betwe
en
the
obj
ect
a
rea
a
nd
qua
dr
a
ti
c
per
im
et
er
wh
il
e
sli
m
ness
is
the
rati
o
betwee
n
the
width
a
nd
th
e
le
ng
th
of
the
obj
ect
.
R
oundne
ss
an
d
slim
ness
are
expresse
d
in
(
3)
and (
4),
resp
ec
ti
vely
.
=
4
2
(
3)
=
ℎ
ℎ
(4)
Conve
xity
is
t
he
rati
o
bet
we
en
c
onve
x
per
i
m
et
er
an
d
obj
e
ct
pe
rim
e
te
r
as
decla
red
in
(5)
a
nd
s
olidit
y
is
the
r
at
io
be
tween
t
he
obje
ct
and
c
onve
x
a
reas
as
f
orm
ula
te
d
in
(6)
.
Dis
persi
on
f
eat
ur
e
e
xpres
s
es
the
irre
gula
rity
of
the
obj
ect
wh
i
ch
is
cal
culat
ed
us
i
ng
(7)
as
the
rati
o
bet
we
en
the
le
ngths
of
m
ai
n
cord
to
th
e
obj
ect
a
rea.
=
(5)
=
(6)
(
)
=
max
(
√
(
−
)
2
+
(
−
̅
)
2
)
(
)
(7)
her
e
,
(
̅
,
̅
)
is t
he
c
e
ntre p
oin
t
of
t
he
m
ass area
(
)
w
hile
(
)
is t
he
ob
j
e
ct
area.
The
in
va
riant
m
o
m
ent
know
n
as
Hu
m
o
m
e
nt
is
cal
culat
ed
base
d
on
nor
m
al
ise
d
centre
m
o
m
ents
[8]
.
The
m
o
m
ent
values
do
not
de
pend
on
tr
ansla
ti
on
,
scal
in
g
a
nd
r
otati
on.
The
re
are
se
ve
n
fe
at
ur
es
of
t
he
in
var
ia
nt
m
o
m
ent
bu
t
on
ly
three
feat
ur
e
s
us
e
d
in
this
s
tud
y,
i.e
.
m
o
m
ent
1,
m
o
m
ent
2
a
nd
m
o
m
ent
3
as
m
at
he
m
at
i
cal
ly
form
ulate
d
in
(
8) to
(10
).
N
orm
al
ise
d
m
o
m
e
nt is
declare
d b
y
ŋ
w
hile
is t
he
m
o
m
ent o
r
de
r.
∅
1
=
(
ŋ
20
+
ŋ
02
)
(8)
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
13
, N
o.
2
,
Fe
bru
ary 2
019
:
721
–
72
8
724
∅
2
=
(
ŋ
20
+
ŋ
02
)
2
+
(
2
ŋ
02
)
2
(9)
∅
3
=
(
ŋ
30
+
3ŋ
12
)
2
+
(
ŋ
03
−
3
ŋ
21
)
2
(10)
Textu
re
is
the b
asi
c feature
r
e
la
te
d
to roug
hness, g
ra
nula
ti
on
a
nd r
eg
ularit
y
of
pi
xels
st
ruct
ur
e
a
nd a
s
the
rep
et
it
io
n
of
basic
pi
xels
i
s
cal
le
d
as
te
xe
l
(text
ur
e
el
em
ent)
[
9]
.
The
t
wo
kinds
of
te
xture
featu
re
ba
sed
on
the
sta
ti
sti
cal
or
de
r
us
ed
incl
ude
histo
gr
am
-
ba
sed
a
nd
grey
l
evel
co
-
occurr
ence
m
at
rices
(G
LCM
)
featu
r
es.
T
he
histo
gr
am
-
base
d
feature
is
th
e
first
-
orde
r
st
at
ist
ic
al
wh
ic
h
com
pr
ise
s
of
six
feat
ur
es
,
i.
e.
m
ean
of
int
ensity
,
dev
ia
ti
on sta
ndard, s
kewness
, e
nergy, e
ntr
opy and sm
oo
th
ne
ss.
T
hey are
f
or
m
ulate
d
in
(
11)
t
o
(
16)
.
=
∑
.
(
)
−
1
=
0
(11)
=
√
∑
(
−
)
2
(
)
−
1
=
1
(12)
=
∑
(
−
)
3
(
)
−
1
=
1
(13)
=
∑
[
(
)
]
2
−
1
=
0
(14)
=
−
∑
(
)
−
1
=
0
log
2
(
(
)
)
(15)
=
1
−
1
1
+
2
(16)
The
sec
ond
-
order
sta
ti
sti
cal
m
e
tho
d
is
c
onduct
ed
by
cal
culat
ing
t
he
pro
ba
bili
ty
of
a
djace
ncy
relat
ion
s
hip
be
tween
tw
o
pix
e
ls
at
a
certai
n
di
sta
nce
a
nd
an
gu
la
r
ori
entat
ion
(
0,
45,
90
a
nd
135
de
gr
ee
s
)
[
10
]
.
Five
GLCM
fe
at
ur
es
e
xtracte
d
a
re
a
ngular
s
econd
m
om
ent
(A
SM
),
i
nv
e
rs
e
dif
fer
e
nce
m
om
ent
(I
DM
),
entr
op
y,
co
nt
rast an
d
c
orrelat
ion.
AS
M
is
us
e
d
to
cal
culat
e
the
ho
m
og
e
neity
of
im
age
us
ing
(17
)
with
t
he
num
ber
of
le
vels
for
com
pu
ta
ti
on
e
xpresse
d
as
L
.
T
he
m
easur
e
m
ent
va
riat
ion
of
grey
le
vel
pi
xels
im
age
known
as
c
on
t
rast
is
form
ulate
d
in (18).
Wh
il
st, ID
M i
s u
se
d
t
o
m
easur
e
hom
og
e
neity
as for
m
ulate
d
in
(
19)
.
2
11
(
,
)
LL
ij
A
S
M
G
L
C
M
i
j
(17)
2
1
|
|
(
,
)
L
n
i
j
n
c
o
n
t
r
a
s
t
n
G
L
C
M
i
j
(18)
2
11
(
,
)
1
(
)
LL
ij
G
L
C
M
i
j
I
D
M
ij
(19)
Entr
op
y
desc
ribes
th
e
ir
regul
arit
y
of
grey
le
vel
im
age.
If
e
lem
ents
of
GL
CM
are
relat
iv
e
the
sam
e,
high
e
ntr
opy
va
lue
would
be
ob
ta
ine
d.
L
ow
entr
op
y
val
ue
i
s
ac
hieve
d
i
f
t
he
el
em
ents
of
G
LCM
near
0
or
1.
Correl
at
ion
fea
tures
is
use
d
to
m
easur
e
t
he
li
near
dep
e
nde
nce
of
grey
le
vel
va
lue
of
th
e
im
age.
Entr
opy
an
d
correla
ti
on are den
oted
in (
20)
and
(21).
11
(
,
)
l
o
g
(
(
,
)
)
LL
ij
e
n
t
r
o
p
y
G
L
C
M
i
j
G
L
C
M
i
j
(20)
''
11
''
2
(
)
(
(
,
)
)
LL
ij
ij
i
i
j
G
L
C
M
i
j
c
o
r
r
e
l
a
t
i
o
n
(21)
2.2.
Fe
ature
sel
ection
Feat
ur
e
sel
ect
i
on
is
c
onduct
e
d
to o
btain
t
he
sign
ific
a
nt
extr
act
ed
featu
res
f
or
im
pr
ovin
g
the
accu
racy
and
re
duci
ng
th
e
com
pu
ta
ti
on
tim
e
du
ri
ng
cl
assifi
cat
ion
proc
ess
[
11
]
.
Wr
a
pper
subset
e
val
uation
-
based
m
et
hod
us
e
d
in
t
his
stud
y
si
nce
it
use
s
a
le
ar
ning
al
gorithm
and
k
fo
l
ds
c
ro
s
s
-
va
li
dation
as
pa
rt
of
the
e
valu
at
ion
functi
on
wh
il
e
searchi
ng
the
f
eat
ur
es
[
12
]
.
It
erati
vely
,
w
ra
pp
er
will
prese
r
ve
th
e
rele
va
nt
featur
e
s
a
nd
el
im
inate
the irr
el
e
va
nt fea
tures.
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c
En
g
&
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Feature
an
alys
is f
or
st
ag
e
ide
ntif
ic
ation
of
pl
as
m
od
i
um viv
ax base
d o
n di
gital …
(
H
anung A
di Nu
groho
)
725
2.3.
Clas
si
ficat
i
on
The
cl
assifi
cat
ion
process
ai
m
s
to
determ
i
ne
in
depen
de
nt
var
ia
ble
(f
ea
tures
)
that
ha
s
the
hi
gh
est
correla
ti
on
t
o
dep
e
nde
nt
va
riable
(class
of
t
he
ob
j
ec
t).
Naï
ve
Ba
ye
s
cl
assifi
er
is
us
e
d
in
t
his
study
since
it
s
relat
ively
fast
in
tr
ai
nin
g,
able
t
o
handle
t
he
re
al
and
discrete
data
an
d
una
f
fected
by
irrel
evan
t
featur
e
s
[
13]
, [14
]
.
3.
RESU
LT
S
A
ND
DI
SCUS
S
ION
A
A
t
otal
of
73
dig
it
al
m
ic
ro
sco
pic
im
ages
of
P.
vi
vax
parasi
te
on
thi
n
bl
ood
film
s
ta
ken
from
the
Dep
a
rtm
ent
of
Parasit
ology,
Faculty
of
Me
dicine,
U
niv
e
r
sit
as
Gadja
h
Ma
da,
were
use
d
i
n
this
stu
dy
.
The
dataset
co
ns
ist
s
of
t
hr
ee
sta
ge
s
im
ages,
na
m
el
y
29
im
ages
of
tr
ophoz
oi
te
s,
10
im
ages
of
sc
hizo
nts
and
34
i
m
a
ges
of g
am
et
ocyt
es stages
, in
BM
P
for
m
at
w
it
h
the
r
e
sol
ution
of
1600
x1200 pi
xels.
Firstl
y,
ori
gin
a
l
i
m
age
is
cr
opped
into
2
50
x250
pi
xels
in
Ro
I
of
p
a
rasit
e
ar
ea
as d
epict
ed
in
Fig
ure 3
.
The
n,
c
ontrast
stret
chin
g
is
a
pp
li
ed
to
e
nha
nce
the
qual
it
y
of
Ro
I
im
age.
For
se
gm
entation
proces
s,
R
-
ba
nd
from
the
RGB
colo
ur
m
od
el
a
nd
S
-
ba
nd
fro
m
HS
V
c
olour
m
od
el
are
c
hosen
since
they
hav
e
the
best
qual
it
y
of
i
ntensity
.
A
f
te
rw
ar
d,
each
of
t
hem
is
filt
e
red
by
m
edian
filt
er
an
d
com
bin
e
d.
T
o
obta
in
the
pa
rasit
e
obj
ect
,
Ots
u
t
hr
e
shold
ing
f
ollow
e
d
by
m
or
phol
og
i
cal
op
e
rati
on
are
c
onduct
ed
to
filt
ered
im
age.
The
sam
ple
of
segm
entat
ion
re
su
lt
is prese
nted
in
Fig
ure
4.
Fo
r
the
d
et
ai
l p
ro
ces
s
has bee
n
e
xp
la
ine
d
i
n
[7]
.
(a)
(b)
Figure
3. (a
) O
rigin
al
im
age (b) Ro
I
im
age
[7]
(a)
(b)
(c)
Figure
4. The
s
egm
entat
ion
r
e
su
lt
of
(a)
t
rop
ho
z
oites (
b) sc
hizo
nts a
nd (
c
)
g
am
et
ocyt
es stages
[
7]
Hav
i
ng
ob
ta
in
ed
t
he
pa
rasit
e
ob
j
ect
,
t
he
sha
pe
-
base
d
a
nd
te
xture
-
base
d
featur
e
e
xtracti
on
a
re
t
he
n
cond
ucted.
A
t
otal
of
10
s
ha
pe
-
base
d
fe
at
ur
e
s
are
extracte
d
wh
ic
h
com
pr
is
es
of
se
ve
n
co
ntour
-
ba
sed
fe
at
ur
es
and
th
ree
feat
ures
of
i
nv
a
riant
m
o
m
ent.
The
r
e
are
se
ve
n
fea
tures
of
in
var
ia
nt
m
o
m
ent
but
only
th
ree
feat
ur
es
us
e
d
since
the f
our
ot
her
s
obt
ai
n
0
val
ue.
The
value
of
m
ome
nt
1
represe
nts
the
centre
of
gra
vity
,
the val
ue
of
m
o
m
ent
2
denotes
t
he
sm
oo
thn
e
ss
a
nd
the
3
-
m
om
ent
value
re
pr
ese
nts
the
a
sym
m
et
ry
of
inte
ns
it
y.
F
or
th
e
te
xtu
re
-
ba
sed
f
eat
ur
es,
a
total
of
11
featu
res
are
e
xtracted
c
on
sist
in
g
of
si
x
histogram
-
ba
sed
featu
res
a
nd
fi
ve
featur
e
s
of
G
LCM
.
The
s
um
m
ary
of
21
extracte
d
feat
ures
is
desc
ribe
d
in
Ta
ble
1.
Fu
rt
her
m
or
e,
f
eat
ur
e
sel
ect
ion
is
co
nducted
to
ob
t
ai
n
the
sig
nific
ant
featu
res
ba
sed
on
Wrap
pe
r
m
et
ho
d.
Si
x
sel
ect
ed
featu
r
es
are
per
im
et
er,
dispe
rsion,
m
ean
of
inte
ns
it
y,
A
S
M,
co
ntrast
G
LCM
an
d
e
ntr
op
y
GLCM.
T
hese
e
xtracte
d
featur
e
s
are the
n
cl
assi
f
ie
d
by
us
i
ng Naï
ve
Ba
ye
s cla
ssifie
r base
d o
n 10
-
f
olds cr
oss vali
datio
n.
To
e
valuate
the
pro
po
se
d
s
chem
e,
so
m
e
sta
ti
sti
cal
par
am
et
ers
are
in
volve
d
in
cl
udin
g
acc
ur
acy
,
sensiti
vity
and
sp
eci
fici
ty
w
hi
ch
are
m
at
hem
at
ic
ally
fo
rm
ulate
d
from
(22)
t
o
(24).
Ac
cur
acy
e
xpress
es
the
su
ccess
fu
l
rate
of
cl
as
sific
at
ion
process
.
Se
ns
it
ivit
y
is
a
c
apab
il
it
y
of
cl
assifi
er
t
o
pre
dict
posit
ive
c
la
ss
as
po
sit
ive
whil
e
sp
eci
fici
ty
is a
capab
il
it
y of cl
assifi
er to p
re
di
ct
n
egati
ve
class as
negat
ive.
In
this
w
ork,
f
our
ty
pe
s
of
cl
a
ssific
at
ion
ba
sed
on
e
xtracte
d
featu
res
a
re
cond
ucted.
T
he
y
are
s
ha
pe
featur
e
s,
te
xtur
e
feat
ur
es
,
s
ha
pe
a
nd
te
xtu
re
f
eat
ur
e
s
a
nd
se
le
ct
ed
feat
ur
es
.
Table
2 pr
ese
nts
the
c
om
par
ison
of
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
13
, N
o.
2
,
Fe
bru
ary 2
019
:
721
–
72
8
7
26
cl
assifi
cat
ion
r
esult
of
these
f
eat
ur
es.
As
de
pi
ct
ed
in
Ta
ble
2,
el
ev
en
t
ext
ur
e
feat
ur
es
yi
el
d
the
l
ow
cl
assifi
c
at
ion
rate
with
the
accuracy,
se
nsi
ti
vity
and
s
pe
ci
fici
ty
of
94.
59%,
94.
6%
a
nd
94.6%,
res
pecti
vely
.
The
bette
r
evaluati
on
rate
is
gaine
d
by
usi
ng
10
sh
a
pe
featur
e
s
with
t
he
acc
uracy
of
97.
29%,
sen
sit
ivit
y
of
97.
30%
an
d
sp
eci
fici
ty
of
97.
30%.
T
he
sa
m
e
resu
lt
is
no
t
on
ly
achie
ve
d
by
t
he
21
fu
l
l
featur
es
of
co
m
bin
at
ion
sh
a
pe
an
d
te
xtu
re
f
eat
ur
e
s but also
is ac
hieve
d by six
s
el
ect
ed
featu
re
s.
Table
1.
T
he
r
e
su
lt
of
featur
e
extracti
on
Sh
ap
e f
eatu
res
Textu
re
f
eatu
res
Peri
m
e
ter
Mean o
f
inten
sity
Area
Co
n
trast
Ro
u
n
d
n
ess
Sk
ewn
ess
Sli
m
n
ess
Energy
Co
n
v
ex
ity
Entro
p
y
So
lid
ity
S
m
o
o
th
n
ess
Disp
ersio
n
ASM
Mo
m
en
t 1
IDM
Mo
m
en
t 2
Co
n
trast GL
CM
Mo
m
en
t 3
Entro
p
y
G
LCM
Co
rr
elatio
n
=
+
+
+
+
100%
(22)
=
+
100%
(23)
=
+
100%
(24)
Table
2.
T
he
c
om
par
ison eval
uation res
ult
of extracte
d feat
ur
es
Extracted f
eatu
res
Accurac
y
(
%)
Sen
sitiv
ity
(
%
)
Sp
ecif
icity
(%
)
Sh
ap
e f
eatu
res (
1
0
)
9
7
.29
9
7
.30
9
7
.30
Textu
re
f
eatu
res
(11
)
9
4
.59
9
4
.60
9
4
.60
Sh
ap
e and
textu
re
f
eatu
res (
2
1
)
9
7
.29
9
7
.30
9
7
.30
Selecte
d
featu
re
s
(
6
)
9
7
.29
9
7
.30
9
7
.30
Althou
gh
they
pro
du
ce
d
t
he
s
a
m
e
value,
the
evaluati
on
rate
by
us
i
ng
six
fe
at
ur
es
is
bette
r
than
t
hat
of
the
f
ull
feature
s.
It
in
dicat
es
that
not
al
l
of
t
he
21
f
ull
featu
r
es
m
a
y
sign
ific
antly
con
tri
bu
t
e
in
the
cl
assifi
cat
ion
process
.
M
or
e
ov
e
r,
by
usi
ng
a
sm
al
l
nu
m
ber
of
feat
ur
es
,
the
propose
d
s
chem
e
is
sti
ll
able
to
gain
th
e
hi
gh
accuracy,
sens
it
ivit
y
and
sp
e
ci
fici
ty
even
m
ay
red
uce
t
he
com
pu
ta
ti
on
tim
e.
This
re
su
lt
ind
ic
at
es
t
hat
the
pro
po
se
d
sche
m
e
su
ccessf
ully
ob
ta
in
s
the
s
ign
ific
a
nt
feat
ur
es
f
or
ide
ntif
y
ing
a
nd
cl
assi
fyi
ng
the
sta
ge
of
P.
vi
vax
pa
rasit
e
on the
dig
it
al
m
ic
ro
sco
pic i
m
age of
thi
n blood fil
m
s
.
4.
CONCL
US
I
O
N
AND
F
UT
U
RE W
ORK
This
stu
dy
pro
po
s
es
a
sc
hem
e
to
cl
assify
P.
vi
vax
pa
rasit
e
on
dig
it
al
m
ic
r
os
c
op
ic
im
age
of
t
hin
bloo
d
fil
m
s
into
th
ree
sta
ge
s,
nam
el
y
tr
ophozoite
s,
s
chizo
nts
a
nd
gam
et
ocyt
es.
A
total
of
10
s
hape
-
base
d
feat
ur
e
s
a
nd
11
te
xt
ur
e
-
ba
s
ed
featu
res
a
r
e
extra
ct
ed
t
o
facil
it
at
e
the
cl
assifi
cat
ion
process
.
Feat
ure
sel
ect
io
n
ba
sed
on
wr
a
pper
m
et
ho
d
is
c
onduct
e
d
to
gain
the
releva
nt
featu
r
es
w
hich
m
ay
co
ntribute
to
i
m
pr
ove
the
r
at
e
of
cl
assifi
cat
ion
r
esult.
Six
sel
ect
e
d
fe
at
ur
es
co
ns
ist
ing
of
pe
rim
et
er
,
disp
e
rsion,
m
ean
of
i
ntensity
,
AS
M,
co
ntras
t
GLCM
a
nd
entr
op
y
GLC
M
achieve
the
best
eval
uatio
n
rate
with
th
e
accu
racy
of
97.29%
,
se
ns
it
ivi
ty
of
97.
30
%
an
d
sp
eci
fici
ty
of
97.
30%.
T
he
propose
d
sc
hem
e
is
able
to
ide
nt
ify
and
cl
assi
f
y
the
sta
ge
of
P.
vi
vax
pa
rasi
te
by
us
in
g
only
sig
nificant
sel
ect
ed
featu
res
res
ul
ti
ng
in
t
he
m
or
e
e
ff
ic
ie
nt
co
m
pu
ta
ti
on
ti
m
e
du
rin
g
t
he
pr
ocess
.
Hen
ce
,
the
pro
po
s
ed
sc
hem
e
has
a
po
te
ntial
to
be
im
ple
m
ented
as
pa
rt
of
t
he
com
pu
t
erised
ai
ded
m
al
aria
diag
nosis sy
ste
m
f
or
assist
in
g t
he
pa
ram
edics.
In
the
ne
xt
in
ve
sti
gation,
the
auth
or
s
c
on
si
de
r
m
or
e
data
w
it
h
the
bala
nce
d
pro
portio
n
in
each
cl
ass
and
feat
ur
e
us
a
ge
in
or
d
er
t
o
ga
in
the
highe
r
a
ccur
acy
.
Th
us,
the
pe
rfo
rm
ance
of
pr
opos
e
d
s
chem
e
can
be
m
or
e
convinci
ng.
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c
En
g
&
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Feature
an
alys
is f
or
st
ag
e
ide
ntif
ic
ation
of
pl
as
m
od
i
um viv
ax base
d o
n di
gital …
(
H
anung A
di Nu
groho
)
727
ACKN
OWLE
DGE
MENTS
This
proj
ect
is
fun
ded
by
Di
re
ct
or
at
e
Ge
ne
ral
of
Highe
r
E
ducat
ion
,
Mi
nistr
y
of
Re
sea
rch,
Tech
no
l
ogy
and
Higher
E
du
cat
io
n,
Re
public
of
I
ndon
esi
a.
The
a
uthors
woul
d
li
ke
to
than
k
the
In
te
ll
igence
S
yst
e
m
s
researc
h gro
up for
s
ha
rin
g
m
e
anin
gful
knowle
dg
e
and i
nspir
ing
disc
us
sio
n.
REFERE
NCE
S
[1]
W
.
H.
Organ
isation
.
"
Mal
aria:
Ke
y
Fac
ts
,"
2017
.
Avail
ab
le
:
htt
p
:/
/
ww
w.who.i
nt/
m
edi
a
ce
ntr
e/
f
ac
tsh
ee
ts/
fs094/
en/
[2]
A.
A.
Nasir
,
M.
Mashor,
and
Z.
Moham
ed,
"S
egmenta
ti
on
base
d
appr
oa
ch
for
de
te
c
ti
on
of
m
al
ar
i
a
par
asit
es
using
m
oving
k
-
m
ea
ns
cl
uster
ing,
"
in
2
012
IEE
E
EM
BS
Confe
renc
e
on
Bi
omedi
cal
Enginee
ring
and
S
cie
nce
s
(
IECB
ES)
,
pp.
653
-
658
,
20
12.
[3]
N.
A.
Kh
an,
H.
Perva
z
,
A.
K.
La
tif,
and
A.
M
usharra
f,
"U
nsupervise
d
ide
nt
ifica
t
ion
of
m
al
ar
i
a
par
asites
usin
g
computer
vision
,
"
in
2014
11th
Inte
rnational
Jo
int
Conf
ere
nc
e
on
Computer
Sc
ie
nc
e
and
So
ft
w
are
Engi
n
ee
rin
g
(
JCSSE
)
,
pp.
26
3
-
267,
2014
.
[4]
D.
Anggrai
ni
,
A.
S.
Nugroho,
C.
Prata
m
a
,
I.
E.
Roz
i,
V.
P
rag
esjva
r
a,
and
M.
Gunawan,
"
Autom
at
ed
status
ide
nti
f
icati
on
of
m
ic
roscopic
ima
ges
obta
in
ed
fro
m
m
al
ari
a
th
in
b
lood
sm
ea
rs
using
Ba
y
es
de
ci
sion
:
a
stud
y
ca
se
in
Plasm
odium
fal
c
ipa
rum
,
"
in
2011
Inte
rn
at
ional
C
onf
s
ere
nce
on
A
dvanc
ed
Comput
er
Scienc
e
and
In
formation
Syst
em
(
ICACSIS
)
,
pp.
3
47
-
352,
2011
.
[5]
C.
Di
Ruber
to,
A.
Dem
pster
,
S.
K
han,
and
B.
Jarra
,
"M
orphological
image
proc
essing
for
evalua
t
ing
m
al
ari
a
dise
ase
,
"
in
Visual
Form
2001,
ed
:
Spring
e
r
,
pp.
739
-
748
,
2
001
.
[6]
H.
A.
Nugroho,
S.
A.
Akb
ar,
and
E.
E
.
H.
Murhan
dar
wati,
"F
e
at
ur
e
ex
tracti
on
and
cl
assifi
ca
t
ion
for
detec
ti
on
m
a
la
r
i
a
par
asites
in
th
in
blood
sm
ea
r,
"
i
n
2015
2nd
In
ter
nati
onal
Con
fer
enc
e
on
In
formation
Te
chnol
og
y,
Computer,
an
d
El
e
ct
rica
l
En
g
in
ee
ring (
ICITACEE
)
,
pp.
197
-
20
1
,
2015
.
[7]
I.
M.
D.
Ma
y
sa
njay
a
,
H
.
A
.
Nu
groho,
N.
A.
Set
ia
wan,
E
.
E
.
H
.
Murhanda
rwati,
T.
R.
Nuringt
y
a
s,
R.
Roto
,
et
al
.
,
"S
egmenta
ti
on
o
f
Plasm
odium
viva
x
phase
on
d
i
git
al
m
ic
roscopi
c
images
of
th
in
blood
fil
m
s
using
col
our
c
h
ann
el
combinat
ion
and
Otsu m
et
hod,
"
i
n
AIP Conf
ere
nc
e
Proc
ee
d
ings
, p
p
.
160002
,
2016
.
[8]
M.
-
K.
Hu,
"V
is
ual
p
at
t
ern
r
ec
o
gnit
ion
b
y
m
om
ent
inva
ri
ant
s,"
I
RE
transacti
ons
on
inf
orm
ati
on
t
heory
,
vo
l.
8,
pp
.
179
-
187,
1962
.
[9]
B.
Chaudhur
i,
N.
Sarka
r,
and
P.
Kundu,
"Im
prove
d
fra
ctal
g
eometr
y
b
ase
d
te
xtu
re
segm
entati
on
te
chn
ique
,
"
IE
E
Proce
ed
ings E (
Computers and D
igi
tal
Te
chni
qu
es)
,
vol.
140
,
pp
.
233
-
242,
1993.
[10]
R.
M.
Har
al
i
ck
a
nd
K.
Shanm
uga
m
,
"Te
x
tura
l
f
ea
t
ure
s
for
image
c
l
assific
a
ti
on,
"
IE
EE
Tr
ansacti
ons
on
systems,
m
an,
and
cy
b
erne
tics
,
vol.
SM
C
-
3,
pp.
610
-
621,
1973
.
[11]
M.
A.
Ha
ll,
"Cor
rel
a
ti
on
-
base
d
fe
at
ure
sel
ec
t
ion
f
or
m
ac
hine l
ea
rn
ing
,
" T
h
e
Univ
er
sit
y
of
W
a
ika
to
,
1999.
[12]
G.
Naqvi,
"A
H
y
brid
Filt
er
-
W
ra
pper
Approac
h
f
or
Feat
ure
Selec
ti
on,
"
Mast
er
Th
esis,
Departmen
t
of
Techno
logy
,
Örebro
Univer
sit
y
,
Sw
ede
n
,
2012
.
[13]
C.
M.
Bishop,
"
Patt
ern
re
cogni
t
i
on
and
m
a
chi
ne
le
arn
ing
,"
New
York:
Springer
,
2006.
[14]
R.
O.
Duda
,
P.
E.
Hart
,
and
D.
G.
Stork,
"P
at
t
er
n
Cla
ss
ifica
ti
on
and
Sc
ene
Anal
y
sis
Par
t
1
:
Pa
tt
e
rn
Cla
ss
ifica
ti
on
,
"
Wil
e
y
,
Chi
ch
ester,
2000
.
BIOGR
AP
HI
ES OF
A
UTH
ORS
H.A.
Nugroho
complet
ed
h
is
bac
he
lor’s
degr
ee
in
elec
tri
c
al
engi
ne
eri
ng
(2
001)
from
the
Univer
sita
s
Gadj
ah
Mad
a,
Yog
y
a
kar
ta,
Indone
si
a
and
m
ast
er’
s
d
eg
ree
in
biomed
ic
a
l
engi
ne
eri
ng
(2005)
from
The
Univ.
of
Que
ensla
nd,
Br
isba
ne,
Aus
tra
l
ia.
In
2012,
he
r
ecei
ved
his
PhD
in
El
e
ct
ri
ca
l
and
E
l
ec
tron
ic
Engi
n
eering
from
th
e
U
nive
rsiti
T
eknologi
Petron
as
(UTP),
Mal
a
y
s
ia.
Curre
ntly
,
he
is
an
As
sistant
Profess
or
and
a
lso
a
Vi
ce
H
ea
d
of
Depa
r
tm
ent
of
E
le
c
trica
l
Engi
ne
eri
ng
and
Inform
at
ion
Technol
og
y
,
Facult
y
of
Eng
ineeri
ng,
Univer
si
ta
s
Gadja
h
Mad
a
(UG
M).
His
cu
rre
nt
rese
arc
h
i
nte
rests
in
cl
ude
biomedi
cal
sig
nal
and
imag
e
proc
essing
and
ana
l
y
sis,
comput
er
vision
,
m
edic
al
instrumenta
t
io
n
and
p
atter
n
recognit
ion.
I
Md.
Dend
i
Ma
y
sanj
a
y
a
co
m
ple
te
d
h
is
Ba
che
lor
degr
ee
i
n
Educat
ion
al
of
informatics
engi
ne
eri
ng,
Uni
ver
sita
s
Pendidikan
Gan
esha
(20
12)
and
h
is
Mast
er
d
egr
e
e
in
th
e
Depa
rtment
of
El
e
ct
ri
ca
l
Eng
in
ee
ring
and
Infor
m
at
ion
T
ec
hno
l
og
y
(2015)
from
the
Univer
si
ta
s
Gadja
h
Mad
a,
Yog
y
ak
art
a
,
Ind
onesia
.
Curre
nt
l
y
,
he
is
a
lectur
er
a
t
th
e
Dep
artm
ent
of
Inform
a
ti
cs
Edu
cation,
Facul
t
y
of Engin
ee
r
ing
and
Voca
ti
on,
Univer
sit
as
Pendidi
k
an
Gan
esha
,
Indone
sia
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
13
, N
o.
2
,
Fe
bru
ary 2
019
:
721
–
72
8
728
N.A.
Setiawan
complet
ed
his
b
ac
he
lor’s
degr
ee
and
m
aste
r
’s
d
egr
ee
in
elec
tr
ical
engi
ne
eri
ng
(1998)
from
the
Univer
sita
s
Gadj
ah
Mad
a
,
Yog
y
a
kar
ta,
Indone
si
a.
In
2009,
h
e
re
c
ei
ved
his
PhD
in
E
le
c
trica
l
and
El
e
ct
roni
c
Engi
n
ee
ring
from
the
Univer
siti
Te
kno
logi
Pe
trona
s
(U
TP),
Mal
a
y
s
ia.
Curre
ntly
,
he
i
s
an
As
sistant
Profess
or
at
th
e
Dep
art
m
ent
of
E
lectr
i
ca
l
E
ngine
er
ing
and
Inform
at
ion
T
echn
olog
y
,
Fa
cul
t
y
of
Engi
n
ee
r
ing
,
Univer
si
ta
s
Ga
dja
h
Mad
a
(UG
M).
His
cur
ren
t
rese
arc
h
int
er
est
s
include
compu
ta
ti
on
al
in
te
l
li
g
e
nce
,
soft
computing
m
ac
h
ine
l
ea
r
ning,
elec
tri
c
al
engi
ne
eri
ng,
and
biomedical
enginee
ring
.
E.
E
.
H.
Murhand
arwa
ti
comple
ted
her
b
a
che
lor’s
degr
ee
(1993)
and
m
aste
r’s
de
gre
e
(1996)
in
Facul
t
y
of
Medi
ci
ne
,
Univer
sit
as
Gadj
ah
Mada
.
I
n
2011,
she
r
ec
e
i
ved
h
er
PhD
in
Medic
a
l
Sc
ie
n
ce
Monash
Univer
s
ity
.
Curre
n
tly
,
s
he
is
a
L
ecture
r
in
th
e
Dep
art
m
ent
o
f
Par
asit
olo
g
y
,
Fa
cul
t
y
of
Medic
in
e,
Unive
rsita
s Gadj
ah
M
ada
(UG
M).
H
er
rese
a
rch
intere
st
is t
rop
ical
m
edicine.
W
.
K.Z
.
Okto
e
ber
za
comple
t
ed
her
ba
chelor’s
degr
e
e
i
n
informatics
(2012)
from
the
Univer
si
ta
s
Bengkul
u
and
h
er
m
aste
r’s
deg
ree
in
elec
tr
ical
engi
ne
eri
ng
(2
015)
from
the
Univer
sita
s
Gad
ja
h
Mada
,
Yog
y
ak
arta,
Indone
s
ia
.
Curr
ent
l
y
,
s
he
works
as
a
Le
c
ture
r
a
t
the
Depa
rtment
o
f
I
nform
at
ic
s,
Facu
lty
of
Engi
n
ee
r
in
g,
Univer
si
ta
s B
engkul
u,
Indone
sia.
Evaluation Warning : The document was created with Spire.PDF for Python.