Internati
o
nal
Journal of Ele
c
trical
and Computer
Engineering
(IJE
CE)
V
o
l.
5, N
o
. 2
,
A
p
r
il
201
5, p
p
.
30
4
~
31
0
I
S
SN
: 208
8-8
7
0
8
3
04
Jo
urn
a
l
h
o
me
pa
ge
: h
ttp
://iaesjo
u
r
na
l.com/
o
n
lin
e/ind
e
x.ph
p
/
IJECE
Col
o
rin
g
of
Cervi
c
al
Cancer’
s CT Images
t
o
L
o
cali
ze Cervi
c
al
Cancer
Erlinda
Ratnasa
r
i
Putri, Ama
r
Vi
jai Nasr
ulloh,
Ar
fan E
k
o
Fah
r
udin
Ph
y
s
ics Stud
y
Pr
ogram, University
of Lambung Mangkur
at, B
a
njar
masin, South
Kalimantan
,
Indon
esia
Article Info
A
B
STRAC
T
Article histo
r
y:
Received Nov 25, 2014
Rev
i
sed
Jan 12, 201
5
Accepte
d
Ja
n 26, 2015
Cervic
al
can
cer
is
the
m
o
s
t
com
m
on g
y
ne
cologi
c c
a
nc
er in
wom
e
n. C
e
rvi
cal
canc
e
r and
the
n
o
rm
al cerv
i
x us
u
a
ll
y h
a
ve s
i
m
i
l
a
r
att
e
nua
tions
on
CT im
ages
which are obtained
.
The no
rmal
cervix an
d the tumour cannot b
e
distinguished on
normal CT
images. CT
image o
f
cerv
i
cal ca
ncer
is used
b
y
the expe
rts
for t
h
e anal
ys
is
of di
s
eas
es
. In this
r
e
s
earch s
t
ud
y,
C
T
im
age of
cervical cancer is done with process
of image
segmentation an
d coloring
.
The process of image segmen
tation is
done after th
e image sharpening
proces
s
and th
e
determ
ina
tion of
cervi
ca
l c
a
nc
er
’s
area
. F
u
zz
y
C-M
eans
is
us
ed as
the
a
l
gorithm
for im
age s
e
gm
ent
a
ti
on. Th
e co
lors
of im
ag
e
segmentation r
e
sult are
chang
e
d b
y
program
module. The result is the color
s
of im
age segm
enta
tion unifor
m
with the other results. Th
e im
age is
overla
y
e
d
with
im
age res
u
lt of
im
age s
h
arpenin
g
proces
s
.
Colo
ring im
age
purposes are to
distinguish between
cer
v
i
cal can
cer’s
area
and
n
o
rmal organ
and to loc
a
li
ze
the exis
t
e
nc
e
of cervi
cal
can
cer.
B
a
sed on
the docto
r’s
observation
,
the
empirical r
a
te of
test
ing 20
samples on th
e progr
am is 100%.
Keyword:
Cervical ca
nce
r’s
area
Co
lo
ri
n
g
CT im
ages
Fuzzy C-Mea
n
s
Ove
r
lay
Copyright ©
201
5 Institut
e
o
f
Ad
vanced
Engin
eer
ing and S
c
i
e
nce.
All rights re
se
rve
d
.
Co
rresp
ond
i
ng
Autho
r
:
Erlin
d
a
Ratn
asari Pu
tri,
Physics St
udy
Program
,
Uni
v
ersi
t
y
o
f
Lam
bung
M
a
n
g
k
u
r
at
,
A.
Ya
ni
St
reet
km
36.
5, B
a
n
j
a
r
ba
ru
,
In
d
onesi
a
Em
a
il: erlin
d
a
p
u
t
ri17
03
@g
mail.co
m
1.
INTRODUCTION
C
e
rvi
cal
ca
nce
r
i
s
t
h
e m
o
st
com
m
on gy
nec
o
l
o
gi
c
ca
ncer
in wom
e
n. Wo
rldw
ide, cervic
a
l cancer is
com
m
on, and
rank
s seco
nd
am
ong al
l
m
a
l
i
gnanci
e
s f
o
r wom
e
n [1]
.
M
o
st
of t
h
es
e cancers st
e
m
from
in
fection
wit
h
th
e hu
m
a
n
p
a
pillo
m
a
v
i
ru
s, alt
h
oug
h
o
t
h
e
r
ho
st fact
o
r
s affect n
e
op
lastic prog
ression
fo
llo
wi
ng
in
itial in
fectio
n
.
Co
m
p
ared
with
o
t
h
e
r g
y
n
eco
l
o
g
i
c m
a
li
g
n
a
n
c
ies, cervical can
cer d
e
v
e
lop
s
in
a yo
ung
er
po
p
u
l
a
t
i
on o
f
wom
e
n. The
m
o
st co
m
m
onl
y
used st
agi
ng syste
m
for cervical cancer
was de
veloped by the
Féde
rat
i
on
Int
e
rnat
i
o
nal
e
de
Gy
néc
o
l
o
gi
e et
d’
Obst
ét
ri
que
(FI
G
O).
Acc
o
r
d
i
n
g t
o
F
I
G
O
,
onl
y
cl
i
n
i
cal
st
agi
n
g
fu
lfills th
ese criteria, an
d
th
erefo
r
e, th
e stagin
g
classi
ficatio
n
o
f
cerv
i
cal can
cer sh
ou
ld
b
e
en
tirely b
a
sed
on
t
h
e fi
ndi
ngs
o
b
t
a
i
n
ed
f
r
om
t
h
e p
r
et
reatm
e
nt clinical evaluation [1].
Th
e
resu
lts of CT, MRI,
o
r
PET
exam
inations and t
h
e s
u
rgic
al-pathologic findings m
a
y not
be
use
d
for staging classi
fication,
but they are
essent
i
a
l
f
o
r t
r
eatm
e
nt
pl
an
ni
ng
an
d m
a
y
pr
ovi
de
pr
o
g
n
o
st
i
c
i
n
f
o
rm
at
i
on
[2]
.
C
o
m
put
ed To
m
ograp
hy
(C
T) i
s
a non
-i
n
v
a
s
i
v
e t
echni
que
t
o
pr
o
v
i
d
e C
T
im
ages of eve
r
y
part
of t
h
e
hum
an b
ody
wi
t
h
o
u
t
su
pe
ri
m
posi
t
i
on of
adjace
nt
st
r
u
ct
ures
[3]
.
C
T
i
s
usef
ul
f
o
r
d
e
t
ect
i
ng t
h
i
s
c
a
ncer
,
esp
ecially in
m
o
n
ito
rin
g
p
a
tien
t
s fo
r recu
rren
ce. A
sse
ssm
ent of the stage of dis
ease is im
portant in
determ
ining whethe
r the
patient m
a
y benefit from
surg
ery or
will receive ra
dia
tion
therapy. The norm
a
l
ut
eri
n
e
cer
vi
x i
s
r
o
u
n
d
st
r
u
ct
u
r
e wi
t
h
h
o
m
o
g
e
no
us
so
ft
-
tiss
u
e attenuation
on CT im
ages. Cervical ca
nc
er a
n
d
t
h
e n
o
r
m
a
l
cervi
x
us
ual
l
y
ha
ve si
m
i
l
a
r at
t
e
nuat
i
o
ns
o
n
C
T
i
m
ages whi
c
h a
r
e
obt
ai
ne
d
[8]
.
The
n
o
rm
al
cervi
x
and t
h
e t
u
m
o
r cann
o
t
be
di
st
i
u
n
g
i
s
hed
on
n
o
rm
al
C
T
im
ages. T
h
e
det
ect
abl
e
fi
n
d
i
n
g ca
n be
use
d
a cl
u
s
t
e
ri
n
g
segm
ent
a
t
i
on m
e
t
hod wi
t
h
F
u
zzy
C
-
M
e
a
n
s al
go
ri
t
h
m
[4, 5
,
6
,
7, 8,
9]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Co
lo
ring
o
f
Cervica
l
Can
cer’s CT Ima
g
e
s to Lo
ca
lize Cervi
c
a
l
Can
cer
(Erlin
d
a
Ra
tna
s
ari Pu
tri)
30
5
As we known,
m
o
st of resea
r
ches a
b
out medical
im
age s
e
gm
entation usually use MR im
ages and
ul
t
r
aso
u
n
d
i
m
ages [4
,
5
,
6, 9,
10]
. I
n
this
res
earch, cervical
cancer data is t
a
ken from
the
CT-Sca
n im
age data.
We use
d
CT images becaus
e
CT-Scan is the m
a
in tool
whic
h use
d
for physical
examination in most of
h
o
s
p
itals in
th
is p
r
o
v
i
n
ce. Besid
e
s th
at, th
e u
s
e of CT-Scan
is a stan
d
a
rd an
alysis fo
r rad
i
o
l
og
ists in
so
m
e
hos
pi
t
a
l
s
. T
o
m
a
ke easi
e
r f
o
r ra
di
ol
ogi
st
s t
o
a
n
al
y
ze cer
vical cancer
data im
ages, it is im
portant t
o
proce
s
s
CT-Sca
n
data images of ce
rvi
cal cancer and
t
h
e n
o
r
m
a
l
cervi
x
so
i
t
can
be
di
st
i
n
gui
s
h
ed
.
The cl
ustering segm
entation
can
be s
u
ccess
f
ully use
d
with the
intention of disc
rim
a
tion
of
va
rious
tissues on CT im
ages. In
part
icular, borde
rs
betwee
n
tissues are not clearly define
d a
n
d m
e
m
b
erships in the
bo
u
nda
ry
r
e
gi
ons
are
i
n
st
ri
n
s
i
cal
l
y
fuzzy
.
The
ha
rd
cl
ust
e
ri
n
g
m
e
t
hods
base
d
o
n
cl
as
si
cal
set
t
h
eo
r
y
, an
d
requ
ire t
h
at an ob
j
ect eith
er
d
o
e
s or
d
o
e
s no
t b
e
l
o
ng
t
o
a
clu
s
ter.
Hard
clu
s
tering
m
ean
s are
p
a
rtitio
n
i
n
g
the
d
a
ta in
to
a sp
ecified
nu
m
b
er o
f
m
u
tu
ally ex
clu
s
iv
e sub
s
ets. Fu
zzy clu
s
terin
g
m
e
t
hods,
ho
we
ver
,
al
l
o
w t
h
e
ob
ject
s t
o
bel
o
n
g
t
o
seve
ra
l
cl
ust
e
rs sim
u
l
t
a
neo
u
sl
y
,
wi
t
h
di
f
f
ere
n
t
degr
ees of
m
e
m
b
ershi
p
.
I
n
m
a
n
y
si
t
u
at
i
ons,
fuz
z
y
cl
ust
e
ri
ng i
s
m
o
re nat
u
ral
t
h
an ha
rd cl
u
s
t
e
ri
ng
[1
1]
. T
h
ere
f
o
r
e, f
u
zz
y
cl
ust
e
ri
ng m
e
t
h
o
d
s
tu
rn
o
u
t
t
o
b
e
particu
l
arly su
it
ab
le fo
r th
e
segmen
tatio
n
o
f
C
T
im
ag
es.
In
g
e
n
e
ral, com
p
le
tely au
to
no
m
o
u
s
seg
m
en
tatio
n
is on
e
of th
e m
o
st d
i
fficu
lt task
s in
t
h
e
d
e
sign
o
f
com
puter vision syste
m
s and rem
a
ins
an active field of image processi
ng
and m
achine
vision researc
h
. The
b
a
sic
g
o
a
l
of seg
m
en
tatio
n
,
t
h
en, is t
o
p
a
rti
tio
n
the im
ag
e in
to m
u
tu
ally
ex
clu
s
i
v
e
reg
i
o
n
s
to wh
ich
we can
su
bqu
en
tly atta
ch
m
ean
in
g
f
u
l
lab
e
ls [12
]
. Seg
m
en
tatio
n
can
be
descri
bed
as t
h
e pr
ocess
rel
a
t
e
d t
o
cl
ust
e
rs, i
n
the m
u
ltim
odal feature s
p
ace
, whose
poi
nt
s are asso
ciated to
si
m
ilar sets
o
f
in
ten
s
ity v
a
lu
es in
th
e d
i
fferen
t
im
ages [12].
The clustering process is t
h
e m
a
in step
i
n
t
h
e se
gm
ent
a
ti
on
pr
oce
d
u
r
e
and
cl
ust
e
ri
ng
-
b
ase
d
t
echni
q
u
es
ha
v
e
been s
h
ow
n t
o
be m
o
re r
o
b
u
st
t
o
n
o
i
s
e i
n
di
scri
m
i
nat
i
on of
di
ffe
re
nt
t
i
ssues t
h
a
n
t
ech
ni
q
u
es
base
d on
ed
ge det
ect
i
o
n
[
4
, 1
3
]
.
The o
u
t
p
ut
i
m
age f
r
om
segm
ent
a
t
i
on
pr
ocess
i
s
t
h
en
colore
d aut
o
m
a
tically. Co
l
o
ri
ng as
a pu
rp
ose t
o
increase the
visual interest
o
f
o
u
t
p
ut
im
age and s
h
o
w
s di
ffe
rent
de
tails o
f
im
ag
e, certain
ly in
RGB
co
lor
m
odel
.
Fo
r th
e
pu
rp
ose o
f
pr
el
im
i
n
ary
t
r
eatm
e
nt
usi
n
g
ra
di
ot
he
ra
p
y
and
su
rge
r
y
[1
4]
, t
h
e
Fuzz
y
C
-
M
eans
clu
s
tering
alg
o
rith
m was in
tro
d
u
c
ed
for the d
i
ag
no
sis
of ev
ery p
a
tien
t
. Un
certain
ty is
m
a
in
ly
p
r
esen
t in
med
i
cal i
m
ag
es, b
e
cau
s
e of t
h
e
n
o
i
se i
n
acq
u
i
sition
an
d
o
f
t
h
e
p
a
rtial
v
o
l
u
m
e effects. Du
e t
o
th
is,
b
o
rd
ers
bet
w
ee
n t
i
ssue
s
are n
o
t
exact
l
y
defi
ne
d a
nd
m
e
m
b
ershi
p
s i
n
b
o
u
n
d
a
r
y
reg
i
ons a
r
e f
u
zzy
.
In t
h
i
s
pa
per
we us
e
Fuzzy C-Mea
n
s segm
entation techni
que
is
used
for tissu
e
differen
tiatio
n
i
n
CT im
ag
es.
2.
R
E
SEARC
H M
ETHOD
The m
e
dical image data are taken
fr
om
CT
im
ages of the c
e
rvical can
ce
r
patients. T
h
e coloring steps
usi
n
g se
gm
ent
a
t
i
on
wi
t
h
F
u
z
z
y
C
-
M
eans al
go
ri
t
h
m
can be
seen
o
n
bl
oc
k
di
ag
ram
i
n
fi
g
u
re
1
.
Fi
gu
re
1.
B
l
oc
k
di
ag
ram
of re
search
m
e
t
hod
a.
Inpu
t of CT
I
m
age
This resea
r
ch
work
use
d
cervical cancer’s
CT im
ag
es from so
m
e
patients. Cervical cancer
has four
Grayscaling a
n
d im
age sharpe
ning
Select ROI
Se
g
m
e
nt
at
i
on
p
rocess
Co
lo
ri
n
g
In
p
ut of CT im
a
g
e
Out
put im
age
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E V
o
l
.
5, No
. 2, A
p
ri
l
20
15
:
30
4 – 3
1
0
30
6
STAGES
, i.e.,
STAGE-1, S
T
AGE
-2, ST
AGE-3, and STAG
E 4. Fi
gure 2 shows CT images for eac
h stage
[
1
5
,
16
, 17
].
(a)
(b
)
(c)
(d
)
Figu
re
2.
Cer
v
ical cancer
’s C
T
Im
ages (a
) S
T
AG
E-
1,
(
b
)
S
T
AG
E-
2,
(c
) S
T
AG
E-
3,
(
d
)
S
T
AG
E-
4
b.
Graysc
aling
and Im
age S
h
arpening
The preprocess
i
ng of the coloring
steps is grayscaling and image sh
ar
peni
ng pr
ocess. Gr
ay
scal
i
ng
i
s
th
e process to
co
nv
ert th
e R
G
B of ce
rvical
cance
r im
age
into
grayscale. Im
age shar
p
e
n
i
ng
is th
e process to
sh
arp
e
n
th
e edg
e
s i
n
im
ag
e an
d in
crease the qu
ality o
f
im
a
g
e.
c.
Select ROI
This proce
ss locates the area
of
the ce
rvical
organ
which i
s
infect
ed s
u
s
p
i
c
i
on. T
h
e R
O
I
(R
egi
o
n
of
In
terest) pro
c
ess p
r
o
ceed
s
in
sid
e
th
e ellip
se lin
e th
at h
a
d
been
cho
s
en
. It can
b
e
seen
in
figu
re 3. Th
e
ou
tsid
e
had
bee
n
bl
ac
k
e
ned
t
o
m
a
ke s
e
gm
ent
a
t
i
on p
r
ocess ea
si
er.
d.
Segme
n
t
a
ti
on Process
Segm
entation process
uses the ROI im
age to distin
guis
h
t
h
e norm
al cells and ce
rvical
cancer cells.
This proc
ess uses Fuzzy C-Means al
gorithm
.
The
m
a
in adva
ntage
of fu
zzy c – m
eans
clustering, it allows
g
r
adu
a
l m
e
m
b
ersh
i
p
s of
d
a
ta p
o
i
n
t
s to
clu
s
ters m
easu
r
ed
as d
e
g
r
ees i
n
[0,1
]. Th
is g
i
ves th
e flex
i
b
ility to
express t
h
at da
ta poi
nts can
belong to
m
o
re t
h
an
one cluste
r [18]. T
h
e Fu
z
z
y C-Means cl
ustering al
gorithm
is
p
r
esen
ted
as follo
ws [18
,
19
]:
1.
Pu
t th
e d
a
ta in
t
o
cluster
X,
which
is a
x
b
m
a
t
r
ix
.
XX
=
⋯
⋯⋯
⋯
⋯
(1
)
a = d
a
ta t
o
tal
b
=
v
a
riab
le to
t
a
l
X
ij
= e
x
am
pl
e dat
a
o
f
i
(i
=
1,
2,
3
,
…
..a
) a
n
d
j
(j
=
1,
2,
3
,
….b
)
2.
Determ
in
e so
me v
a
riab
les su
ch
as:
Clu
s
ter to
tal =
c (
2)
Ex
po
ne
nt
=
w
M
a
xi
m
u
m
it
erat
i
on =
Max
I
te
r
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Co
lo
ring
o
f
Cervica
l
Can
cer’s CT Ima
g
e
s to Lo
ca
lize Cervi
c
a
l
Can
cer
(Erlin
d
a
Ra
tna
s
ari Pu
tri)
30
7
Er
ro
r v
a
lu
e =
Earliest ob
j
ecti
v
e
fun
c
tion
P
0
=
0
Earliest iteratio
n
= t = 1
3.
Gene
rat
e
ra
n
d
o
m
num
ber µ
ik
whi
c
h i
=
1,
2, …
…
a;
k
=
1,
2,
3,
…..c;
as t
h
e el
em
ents of
t
h
e ea
rl
i
e
st
p
a
rtitio
n m
a
tri
x
U. Calcu
l
ate th
e to
tal
o
f
attrib
u
t
es
(co
l
u
m
n
)
:
(2
)
whi
c
h j
= 1,
2
,
……n
Calculate:
4.
C
a
l
c
ul
at
e t
h
e c
e
nt
r
o
i
d
of
k:
V
kj
, w
h
i
c
h
k =
1
,
2,
….c;
a
n
d
j
= 1,
2
,
…
…
b
.
∑
∑
(3
)
5.
Calcu
l
ate o
b
j
e
ctiv
e fu
n
c
tion
o
n
t iteratio
n,
(4
)
6.
Calcu
l
ate th
e ch
ang
e
of
p
a
rtitio
n m
a
trix
:
∑
∑
∑
(5
)
whi
c
h i
=
1,
2
,
…a;
an
d
k =
1,
2,
….c.
7.
Com
p
are
.
If
(
or (t > MaxIter), th
en
st
o
p
. Else, t = t +
1
,
retu
rn
to step
4
.
In
th
is research
, we determ
in
ed
v
a
lu
e for some v
a
riab
les. Clu
s
ter to
tal was 3
,
exp
o
n
e
n
t
was 2,
er
ro
r v
a
l
u
e
w
a
s 0
,
00
0001
an
d m
a
x
i
m
u
m
i
t
er
atio
n
w
a
s
10
0 ti
m
e
s.
e.
Col
o
ring
Th
is
p
r
o
cess giv
e
s th
e im
ag
e so
m
e
co
lors t
o
d
i
stin
guis
h
norm
al cells and ce
rvical ca
ncer cells.
In
th
is research,
we
u
s
e R
G
B mo
d
e
l. RGB m
o
d
e
l is
usually used t
o
re
present a static im
age [20].
3.
R
E
SU
LTS AN
D ANA
LY
SIS
Figure 3
(a) is
an origi
n
al image in RGB m
odel.
Fi
gure
3
(b) is the res
u
l
t
of grayscaling and im
age
shar
pe
nin
g
pr
o
cess.
It is in
g
r
ay
scale
m
odel.
Fig
u
re
3
(c
) s
h
o
w
s
the
res
u
lt o
f
R
O
I
p
r
oces
s. Fi
gu
re
3
(d
)
is th
e
resul
t
of segm
ent
a
t
i
on p
r
oces
s
usi
n
g
Fuzzy
C-Means al
gorith
m
.
The c
o
lors
on
Fig
u
re
3
(d
) ar
e c
h
an
ge
d a
n
d
the res
u
lt is Fi
gu
re
3
(e)
.
Fi
g
u
re
3 (f) is th
e
fin
a
l resu
lt im
a
g
e.
As we can see
from
the result of ROI process on Figu
re
3 (c), we use a
n
ellipse form
because m
o
st
o
f
cerv
i
cal cancer cells
wh
ich showed on
C
T
im
ag
es ar
e ellip
se. Seg
m
en
tatio
n
p
r
o
cess i
s
don
e in ellip
se area
to distinguish
norm
al cells a
n
d cervical ca
ncer cells
lik
e on
Fi
g
u
re
3
(d
). Th
e m
a
in
g
o
a
l
of seg
m
e
n
tatio
n
pr
ocess i
s
t
o
l
o
cal
i
ze t
h
e ex
i
s
t
e
nce of ce
r
v
i
cal
cancer
. We use cl
ust
e
ri
n
g
m
e
t
hod
wi
t
h
Fu
zzy
C
-
M
ean
s
algorithm
.
W
e
determ
ine specific values
for
so
m
e
v
a
riab
les, i.e., clu
s
ter total, ex
p
o
n
e
n
t
valu
e, error v
a
lue an
d
max
i
m
u
m
iteratio
n
.
Clu
s
ter total is 3
,
expo
nen
t
v
a
lu
e
i
s
2
,
err
o
r
val
u
e i
s
0,
00
0
0
0
1
an
d
m
a
xim
u
m
i
t
e
rat
i
on i
s
10
0 t
i
m
es. These val
u
e
s
p
r
od
uce re
sul
t
i
m
age s
u
ch
as Fi
g
u
re
3
(d
).
O
n
F
i
gu
re 3
(
d
)
,
T
h
e bl
ue i
s
bac
k
g
r
o
u
n
d
,
the red is cervi
cal cancer’s a
r
ea and the
gre
e
n is an area
w
h
i
c
h n
o
t
co
nt
ai
ni
n
g
a t
u
m
o
r. But, we wa
nt chan
g
e
the color.
So,
we do c
h
anging col
o
r
pr
oces
s an
d i
t
pr
od
uc
es a res
u
l
t
im
ag
e like Figure
3 (e
) whe
r
e the red is
back
g
r
o
u
n
d
,
t
h
e gree
n i
s
a
n
ar
ea whi
c
h n
o
t
c
ont
ai
ni
ng
a
tum
o
r and the
bl
ue is cervical cancer’s are
a
. T
h
ere is
a diffe
re
nt color in cervical ca
ncer’s area
. It
cause
d by a
differe
n
t intens
ity of
grayscale
on that area
. It s
h
ows
that cervical cancer cells are growin
g
wi
t
h
o
u
t
f
o
rm
i
ng any
pat
t
e
r
n
s.
B
e
si
des t
h
at
, t
h
e
bl
ue
o
n
ce
rvi
cal
cancer’s a
r
ea i
s
rec
o
gnized as
cervical cance
r cells.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E V
o
l
.
5, No
. 2, A
p
ri
l
20
15
:
30
4 – 3
1
0
30
8
(a)
(b
)
(c)
(d
)
(e)
(
f)
Fi
gu
re
3.
(a
)
O
r
i
g
i
n
al
i
m
age, (
b
)
R
e
sul
t
of
g
r
ay
scal
i
ng a
n
d
i
m
age sha
r
pe
ni
ng
p
r
ocess,
(c
)
R
e
sul
t
o
f
R
O
I
pr
ocess
,
(
d
) R
e
sul
t
o
f
t
h
e
se
g
m
ent
a
t
i
on
pr
oc
ess, (
e
) R
e
su
lt
o
f
th
e ch
an
g
i
ng
co
lor
p
r
o
cess, (f) Fin
a
l
resu
lt
im
age
Fi
gu
re 3 (
f) i
s
t
h
e fi
nal
re
sul
t
im
age. It
i
s
pro
duce
d
by
ove
rl
ay
2 im
ages, i
.
e., t
h
e res
u
l
t
of
gray
scal
i
n
g a
n
d i
m
age shar
pe
ni
n
g
p
r
o
cess (Fig
ur
e
3
(b
))
an
d
th
e r
e
su
lt of
the cha
n
ging
color proc
ess (Figure
3 (e
))
. T
h
e
ov
erl
a
y
im
age i
s
use
f
ul
fo
r
rad
i
ol
ogi
st
s t
o
a
n
alyze cervical
cancer and
det
e
rm
ine the spread
of
cervical ca
ncer cells. Moreover,
radi
ol
og
ists can also d
e
termin
e th
e stag
e
of ce
rvical ca
ncer and e
x
pla
i
n to
pat
i
e
nt
s a
b
o
u
t
t
h
e
gr
o
w
t
h
of
c
e
rvi
cal
ca
ncer
i
n
t
h
ei
r
bo
dy
.
In t
h
is resea
r
c
h
,
we used
20 CT im
ages as obj
ects
for the coloring cervi
cal cancer
a
u
tom
a
tically.
Results of the
ove
rlay im
ages are ex
am
i
n
ed by
t
h
e doct
o
r.
B
a
sed on t
h
e
doct
o
r’s obse
rvation, the empirical
rate of testin
g
2
0
sam
p
les o
n
th
e pro
g
ram
i
s
10
0
%
.
It
m
e
ans that t
h
e a
u
tom
a
tical
coloring ce
rvical ca
ncer
program
can localize the exist
e
nce
of cervica
l cancer and
do coloring
of cervical cance
r
on CT im
ages.
4.
CO
NCL
USI
O
N
Im
age processi
ng
steps t
h
at are co
nve
r
t the
RGB of cervi
c
a
l cancer im
age into
grayscal
e, determ
ine
the ROI
of ce
rvical cancer’s
area, do
a segmen
tatio
n
pro
c
ess to
th
e ROI i
m
ag
e, d
o
co
lo
ri
n
g
to
t
h
e resu
lt o
f
segm
ent
a
t
i
on pr
ocess a
nd
d
o
im
age ove
rl
ay
i
ng p
r
oce
ss t
o
kn
o
w
t
h
e l
o
cat
i
on
of t
h
e
exi
s
t
e
nce o
f
cervi
cal
cancer on ori
g
inal
im
age.
Co
lo
ri
n
g
im
ag
e purpo
se is t
o
d
i
stin
gu
ish
bet
w
ee
n cervical canc
e
r’s a
r
ea a
n
d norm
al
o
r
g
a
n
.
Based
on
th
e do
ctor
’s
o
b
s
erv
a
tio
n, the em
p
i
r
i
cal r
a
te of
testin
g 20
sam
p
les o
n
th
e pr
og
r
a
m
is 100
%.
ACKNOWLE
DGE
M
ENTS
Th
is research
was fu
lly su
ppo
rted
b
y
Ulin
Ho
sp
ita
l
B
a
nja
r
m
a
si
n. W
e
t
h
ank
Arl
a
vi
n
d
a
A. L
ubi
s as
a
Radiologist and Alfia
n
Rizani as an
ope
rato
r o
f
CT-Sca
n
at Ulin Ho
spit
al Banjarm
a
sin, fo
r s
u
p
p
o
rtin
g an
d
al
l
o
wi
n
g
us t
o
w
o
r
k
o
n
t
h
e
R
a
di
ot
he
rap
h
y
I
n
st
al
at
i
on
as
well as t
h
e
use of CT
im
a
g
es
data
of ce
rvical
cancer’s patients.
REFERE
NC
ES
[1]
Parkin DM, Br
ay
F
,
Ferlay
J
.
Global
can
cer
statistics, 2002
.
C
A
Cancer
J
Clin
. 2
005; 55: 74.
[2]
Schorge JO, Sch
a
ffer JI, Halvorson LM,
Hoffma
n
BL, Bradshaw KD, Cunningham
FG.
Williams Ginecolog
y
. Th
e
McGraw-Hills C
o
m
p
an
y
.
Unit
ed
States of
Am
erica. 2008
.
[3]
We
bst
e
r JG.
Medical Instrumentation: App
lica
t
ion and Design
. Fourth
Edition. John
W
iley
& S
ons, Ltd
., United
States of
America. 2010
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Co
lo
ring
o
f
Cervica
l
Can
cer’s CT Ima
g
e
s to Lo
ca
lize Cervi
c
a
l
Can
cer
(Erlin
d
a
Ra
tna
s
ari Pu
tri)
30
9
[4]
Phillips WE, V
e
lthui
zen
RP, Phuphanich S
,
H
a
ll
LO, Cl
arke
LP, Silbig
er M
L
, Appl
ic
ation of
fuzzy
c
-m
eans
segmentation technique for diff
erentiation in MR
im
ages of
a hem
o
rrhagic
glioblastom
a
m
u
l
tiform
e
.
Magnetic
Resonance Imag
ing
. 1995
; 13
: 2
77-290.
[5]
Suckling J, Sig
mundsson
T, Gr
eenwood K, Bullmore ET.
A modified fuzzy
cluster
i
ng algorithm for operator
independ
ent brain tissue classifi
cation of dual echo MR images.
Magnetic resonance Imaging
.
1999; 17: 1065-
1076.
[6]
Rose RJ, Allwin S. Ultrasound Cervic
al Can
cer
Based Abnormality
Segmen
tation Using Adaptive Fuzzy
C-Mean
Clustering
.
Acad
emic Journal
o
f
Cancer
Research
. 2013; 6: 01-0
7
.
[7]
Masuli F, Schen
one A. A fuzzy
cluste
r
i
ng based
segmentation
sy
stem as suppor
t to diagnosis in
medical imagin
g.
Artifi
cial
Int
e
ll
ig
ence
in
Medi
cin
e
. 1999
; 16
: 129
-147.
[8]
Zheru C, Hong Y, Tuan P. Fuzzy
algorithms: with appl
ications to image processi
ng and pattern r
ecognition.
World
Scien
tifi
c Publis
hing
. 1996
; 10
:
57-58, 86-89
.
[9]
Vasuda P, Sateesh S. Impr
ove
d fuzz
y
c-m
e
an
s
for m
r
brain
im
age s
e
gm
entation
.
In
ternational Journal o
n
Computer Scien
ce and
Eng
i
neer
ing
. 2010
; 2
:
17
13-1715.
[10]
Sharma M, Mukherjee S. Fuzzy
c-mean
s, anf
i
s and genetic
algor
ithm for segm
enting astro
c
y
t
oma-a ty
pe of
brain
tum
o
r.
IA
ES Int
e
rnational
Journal of
Arti
ficial
Int
e
llig
ence
. 2014;
3: 16-23.
[11]
Babuska R, Ver
b
ruggen HB.
Fu
zzy Logic Contr
o
l: Ad
vances in
Applica
tions
. World Scientif
ic,
New Jersey
. 199
9.
[12]
Solomon C, Breckon T.
Funda
mental of Digita
l Image Process
i
ng
. John Wiley & Sons,
Ltd., United
Kingdo
m.
2011.
[13]
Lin JS, Cheng
KS, Mao CW
. Segm
entation of
m
u
ltispect
r
a
l
m
a
gnetic r
e
sona
nce im
age usin
g penal
i
zed fu
z
z
y
com
p
etitiv
e le
ar
ning
ne
twork.
C
o
mputers and Biomedical Res
e
arch
. 1996
; 29
: 31
4-326.
[14]
Kalet IJ, Sey
m
o
u
r MMA. The use of medical
image in
plann
i
n
g
and deliv
er
y
o
f
radiation th
erap
y
.
Journal of the
American M
e
di
c
a
l Informati
c
s A
ssociation
. 1997
; 4: 327-339.
[15]
Benedet JL, Odicino F, Maisonn
euve P, et
al.
Ca
rcinom
a of
the
c
e
rvix u
t
eri
.
J Epi
d
emiol Biostat
. 2
001; 6: 7-43.
[16]
Janicek MF, Av
erette HE. Ce
rv
ical
can
cer: p
r
ev
ention
,
di
agnosis, and th
erapeu
tics.
CA Cancer J Clin
. 2001
; 51:
92-114.
[17]
Sm
ith RA, Mettlin CJ, Davis KJ,
E
y
r
e
H. Am
erican Cancer
Soci
et
y guide
lin
es for the ear
l
y
d
e
t
ect
i
on of cancer
.
CA
Cancer
J
Clin
. 2
000; 50: 34-49.
[18]
M
a
khalova
E.
F
u
zz
y
c-m
ean
s cl
ustering in
m
a
tl
ab.
The
7
th
International Days o
f
Statistics and
Economics
. 2
0
13;
905-914.
[19]
Forsy
t
h DA,
Ponc
e
J.
Computer
Vision: A
Mod
e
rn
Approach
. Second Edition. Prentic
e Hall, New
Jersey
. 2011
.
[20]
Zhu MF, Du JQ. A new method of color ton
gue image seg
m
entation b
a
sed
on random walk.
TE
LKOMNIKA
Indonesian Jour
nal of El
ectrical Engineering
. 20
14; 12: 4512-45
20.
BIOGRAP
HI
ES OF
AUTH
ORS
Erlinda Ratnasari Putr
i
was b
o
rn in Ban
j
armasin, Kaliman
ta
n
Selatan,
in 1993
. She r
eceived
an undergradu
ate degree in Ph
y
s
ics (2015) fro
m
University
of Lambung Mangkur
at (UNLAM),
Banjarb
a
ru Indo
nes
i
a. Her
curre
nt res
ear
ch int
e
r
e
s
t
s
are Digit
a
l
Im
age P
r
oces
s
i
ng and M
e
dica
l
Instrumentation
.
Amar Vijai Nasrulloh
was born in Surabay
a
, in 1978. He re
ceived an undergr
a
duate d
e
gree in
Ph
y
s
ics (2004)
from
Institut T
e
knologi Sepulu
h
N
opem
b
er (I
TS) and a grad
uate d
e
gree in
Electri
cal
Engi
neering (2010
)
specialt
y
in
Biom
edical En
gineer
ing Program
from
Institut
Teknologi Band
ung (ITB), Ind
onesia. He receiv
ed th
e Academic Achiev
ement Directorate
General High
er
Education (DIKTI) Scholarship
(BPPS) for his
graduate degree. In 2005, h
e
joined
the Ph
y
s
ics Stud
y
Program of University
of Lambung Mangkurat
as a
lecturer. Curr
ently
he is member of
HFI (Indonesian
Ph
y
s
ical Society
/
IPS) and IEEE.
His cu
rrent research
inter
e
sts
are Electronics
and Instrumentation of Ph
ys
ics, Video Processing, Imaging and Image
P
r
oces
s
i
ng, Bio
m
echanics
and
Medical Instrumentation.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E V
o
l
.
5, No
. 2, A
p
ri
l
20
15
:
30
4 – 3
1
0
31
0
Ar
fan Eko F
a
hr
udin
was born in Kedir
i
,
in 1
979. He r
eceived an undergr
a
d
u
ate d
e
gree in
Ph
y
s
ics (2004)
from
Institut T
e
knologi Sepulu
h
N
opem
b
er (I
TS) and a grad
uate d
e
gree in
Electrical
Engin
eering (2010) fr
om Universita
s Gadjah Mada (
UGM),
Indonesia. In 2005, h
e
joined th
e Ph
y
s
ics Stud
y
Program of Universita
s Lambung Mangkurat as a l
ectu
r
er. He received
the
Aca
d
e
m
ic Ac
hie
v
e
m
e
n
t Direc
t
orate
Ge
ne
ra
l
Highe
r E
d
uc
a
t
ion (DIKT
I) Sc
hola
r
ship (BPPS)
for his graduate degree. Curr
en
tly
h
e
is memb
er of HF
I (Indones
i
an P
h
y
s
ic
al
S
o
ciet
y). His
current r
e
sear
ch interests ar
e Image Pr
ocessing, Pattern
R
ecognition and
Medical
Instrumentation
.
Evaluation Warning : The document was created with Spire.PDF for Python.