Indonesi
an
Journa
l
of El
ect
ri
cal Engineer
ing
an
d
Comp
ut
er
Scie
nce
Vo
l.
1
4
,
No.
1
,
A
pr
il
201
9
, p
p.
471
~
477
IS
S
N: 25
02
-
4752, DO
I: 10
.11
591/ijeecs
.v1
4
.i
1
.pp
471
-
477
471
Journ
al h
om
e
page
:
http:
//
ia
es
core.c
om/j
ourn
als/i
ndex.
ph
p/ij
eecs
A symmetry b
ased an
om
aly det
ection in b
ra
in u
sin
g cellula
r
automat
a for
comp
uter aided di
ag
n
osis
Nisha V
M, L
Jegana
than
School
of
Com
p
uti
ng
Sci
ences and E
ngin
ee
ring
,
Vell
ore
Instit
u
te
of
Technol
og
y
,
Chenna
i
,
Ind
ia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
A
ug
1
4
, 201
8
Re
vised
N
ov
10, 2
018
Accepte
d
J
an
23
, 2
01
9
Com
pute
r
ai
ded
dia
gnosis
(CAD
)
is
an
adva
n
ci
ng
technolog
y
in
m
edi
cal
imaging.
CAD
a
ct
s
as
an
addi
t
io
nal
computing
p
ower
for
doct
ors
to
interpre
t
the
m
edica
l
im
age
s
which
l
eads
to
a
m
ore
ac
cur
ate
di
agno
sis
of
the
disea
se.
CAD
s
ystem
inc
re
ase
s
t
he
ch
anc
es
of
d
et
e
ct
ion
of
bra
in
le
sions
b
y
assisting
the
ph
y
sici
ans
in
dec
r
eas
ing
the
observa
ti
onal
ov
ersight
in
the
ear
l
y
stage
of
disea
s
es.
Thi
s
pap
er
foc
uses
on
the
deve
lopment
of
a
ce
ll
u
lar
aut
om
at
a
base
d
m
odel
to
fin
d
the
anomal
y
prone
are
as
in
hum
an
bra
ins.Beca
use
o
f
the
b
ilate
r
al
s
ym
m
et
ric
nat
ur
e
of
hum
an
bra
in
,
a
s
y
m
m
etr
y
base
d
c
el
lu
la
r
a
utomata
m
odel
i
s
proposed.
An
a
lgori
thm
is
d
esigne
d
base
d
on
the
proposed
m
odel
to
de
tect
the
anomal
y
pr
one
ar
ea
s
in
br
a
in
images.
The
proposed
m
odel
c
an
be
a
sta
ndal
one
m
odel
o
r
it
c
an
be
inc
or
pora
te
d
to
a
sophistic
ated
co
m
pute
r
ai
ded
di
agnosis
sy
st
em.
B
y
inc
orpor
at
in
g
as
y
m
m
et
r
y
informati
on
into
a
computer
ai
ded
dia
gno
sis
sy
stem,
en
hanc
es
it
s
per
form
anc
e
in
ide
nti
f
y
i
ng
the
anomali
es
exi
sts
in
bil
at
er
al
l
y
s
y
m
m
et
r
ical
bra
in im
age
s.
Ke
yw
or
d
s
:
Anom
al
y
Brai
n
im
ages
Ce
ll
ular
aut
oma
ta
Com
pu
te
r
ai
de
d diag
nosis
Sy
m
m
e
try
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
:
Nish
a
V
M
,
School
of Com
pu
ti
ng
Scie
nce
s and E
nginee
r
ing
,
Vell
or
e
Insti
tute o
f
Tec
hnolog
y,
Vandalu
r
-
Kel
a
m
bak
kam
Ro
ad,
C
he
nn
ai
-
6001
27, In
dia
.
Em
a
il
:
nish
av
m
@v
it
.ac.in
1.
INTROD
U
CTION
Me
dical
i
m
ages
play
a
key
r
ol
e
in
detect
io
n
and
dia
gnos
is
of
va
rio
us
diseases.
Adva
nces
in
m
edical
i
m
age
te
chn
iq
ues
ha
vef
aci
li
ta
te
d
to
i
m
pr
ov
e
accurate
diag
no
sis
.
Acc
ur
at
e
detect
ion
of
abno
rm
aliti
es
i
n
br
ai
n
is
ver
y
com
plex
ta
sk
.
Seve
ral
researc
hes
ha
ve
car
ried
ou
t
i
n
this
area
.
Th
e
nu
m
erous
res
earches
over
t
he
la
st
few
deca
des
in
analy
sin
g
m
edical
i
m
ag
es
de
velo
ped
auto
m
at
ed
te
chn
i
qu
e
s
[
1]
for
dia
gnos
i
s.
Th
e
perform
ance
of
the
tradit
io
na
l
autom
at
ed
com
pu
te
r
diagno
sis
syst
e
m
is
trivia
l
becau
se
it
has
it
s
own
ba
rr
ie
r
as
the
a
uto
m
ated
diag
nosis
s
yst
e
m
cann
ot
r
eplace
doct
or
s
to
detect
th
e
di
seases.
Un
li
ke
autom
at
ed
dia
gnos
is
syst
e
m
,
wh
ere
the
diagnosis
is
do
ne
by
the
m
achines,
com
pu
te
r
ai
de
d
diag
nosis
[2
]
-
[
7]
is
a
syste
m
by
consi
der
i
ng
th
e
ro
le
of
ra
diol
og
ist
s
or
phys
ic
ia
ns
in
diag
nosin
g
the
le
sio
ns
.
C
A
D
is
use
d
to
gi
ve
a
s
econd
op
i
nion
to
t
he
ph
ysi
ci
ans
to
detect
the
an
om
al
i
es
in
br
ai
n.
The
va
rio
us
t
echnolo
gies
de
velo
ped
in
CA
D
are
cat
al
ysi
ng
the en
ha
ncem
ent
of
le
sion
d
et
ect
ion
i
n
brai
n
im
a
ges.
C
AD
syst
e
m
can
ha
ve
m
ulti
ple
m
od
ules
su
c
h
as
an
om
al
y
detect
ion
,
diag
no
sis
an
d
ris
k
as
sessm
ent
et
c.
The
m
ai
n
obj
e
ct
ive
of
this
pa
per
i
s
to
de
ve
lop
a
cel
lular
automa
ta
(CA)
m
od
el
,
wh
ic
h
dete
ct
s
the
ano
m
a
ly
pr
one
areas
in
hu
m
an
brai
n,
w
hich
ca
n
be
integrate
d wit
h a s
ophisti
cat
ed
CA
D sy
ste
m
to enha
nce its
perform
ance.
Ra
dio
lo
gists
a
re
com
m
on
ly
us
in
g
Ma
gn
e
ti
c
Re
so
na
nce
Im
age
(MRI)
to
a
naly
se
the
inter
nal
structu
re
of
hum
an
brai
n.
T
he
existi
ng
te
c
hn
i
qu
e
s
us
ed
f
or
t
um
ou
r
detect
io
n
us
es
va
rio
us
m
et
ho
ds
us
i
ng
MR
I
i
m
ages.
Textur
e
ba
sed
feat
ur
e
analy
sis
[
8]
and
water
she
d
al
gorithm
[
9]
is
m
a
inly
u
sed
f
or
brai
n
tum
our
detect
ion.
Ex
pe
ct
at
ion
m
axi
m
iz
at
ion
m
et
h
od
[10]
is
us
e
d
f
or
brai
n
a
nom
alies
detection
ba
sed
on
bilat
eral
filt
er
[10].
A
t
wo
sta
ge
re
gion
of
interest
s
egm
entat
ion
[
11
]
ba
sed
on
m
ul
ti
le
vel
threshold
[
11]
an
d
ha
rd
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.
1
4
, N
o.
1
,
A
pr
il
201
9
:
471
–
477
472
thres
ho
l
ding
[
12
]
is
us
e
d
to
detect
the
area
of
tum
ou
r
s.
All
these
te
ch
niq
ues
m
entioned
a
bove
us
e
d
im
age
processi
ng
te
c
hn
i
qu
e
s
to
de
te
ct
the
an
oma
li
es.
In
orde
r
to
i
ncor
pora
te
intel
li
gen
ce
al
ong
with
br
ai
n
tum
ou
rd
et
ect
io
n
al
gorithm
s,
fu
zzy
lo
gics[
13
]
-
[
17]
are
in
cl
ud
e
d,
wh
ic
h
helps
to
th
e
detect
ion
of
m
or
e
accurate
le
sio
ns
in
br
a
i
n.
But
trai
nin
g
of
da
ta
is
req
uire
d
to
inco
rpo
rate
intel
li
gen
ce
in
an
om
al
y
detect
ion
al
gorithm
s.
The
an
om
al
y
p
arts
in
the
brai
n
m
ay
be
in
va
rio
us
siz
e
an
d
sh
a
pe,
so
t
he
s
egm
entat
ion
of
ano
m
al
y
is
a
chall
en
ging
ta
sk
.
In
te
gr
at
io
n
of
a
natom
ic
a
l
know
le
dgewi
th
an
om
al
y
det
ect
ion
te
ch
niques,
ai
ds
the
s
yst
e
m
to
unco
ve
r
the
sy
m
m
e
try
or
asym
m
e
try
in
the
brai
n
str
uc
ture
le
ads
t
o
e
nh
a
nce
the
syst
e
m
per
form
ance
in
com
pu
te
r
ai
de
d
dia
gnos
is
of
the
br
ai
n
an
om
al
ie
s.H
um
an
br
ai
n
st
ru
ct
ure
has
tw
o
ap
pa
ren
tl
y
si
m
il
ar
halves
tha
t
exh
ibit
hi
gh
le
vel
of
bilat
eral
sy
m
m
et
ry.
But
sy
m
m
et
ry
is
vio
la
te
d
in
the
presence
of
le
sion
s
in
brai
n.
S
o
the
ai
m
of
this
pa
per
is
to
e
xplo
re
the
de
gr
e
e
of
asy
m
m
et
ryoccurs
du
e
to
the
pr
ese
nce
of
le
si
on
s
inth
e
br
ai
n
and
propose
a
n
autom
at
ed
tech
ni
qu
e
us
i
ng
cel
lular
automa
ta
to
detect
th
e
ano
m
al
ie
s
pron
e
area
e
xists
in
the
hu
m
an
brai
n.
Most
of
the
im
age
proces
sing
te
ch
nique
use
d
for
a
no
m
al
y
detect
ion
ne
eds
im
age
reg
i
strat
ion
[18]
and
are
pro
ne
to
inter
-
i
nd
i
vidual
and
inter
-
e
qu
i
pm
ent
var
ia
ti
on
s
ev
en
unde
r
co
ntr
olled
ci
rcu
m
st
ances
,
le
ads
to
e
rro
ne
ou
s
sit
uatio
n
t
o
dr
a
w
in
fer
e
nc
es
base
d
on
a
bs
ol
ute
values
directl
y.
S
o,
a
ppr
oach
es
ba
s
ed
on
relat
ive
values
of
the
anato
m
ic
al
data
hav
e
high
im
pac
t
in
identify
ing
the
le
sion
r
egio
n
in
hum
a
n
br
a
i
n.
Sy
m
m
e
try
bas
ed
m
et
ho
d
is
base
d
on
sta
ti
sti
cal
ly
s
ign
ific
ant
relat
ive
values,
m
ay
pr
ovi
de
m
or
e
insig
ht
s
for
identify
in
g
an
d
qua
ntifyi
ng
the
br
ai
n
le
sions
in
com
pu
te
ri
zed
a
naly
sis.
T
her
e
a
re
te
ch
ni
qu
e
s
us
e
s
sym
m
et
ry
base
d
ap
proac
h
[
19
]
-
[
26]
f
or
detect
ing
brai
n
tum
ou
rs
a
nd
for
b
rain
tum
ou
r
se
gm
entat
ion
,
bilat
eral
sym
m
e
try
of
t
he
br
ai
n
str
uctu
re
is
e
xp
l
oi
te
d.
Since
the
sy
m
m
et
ry
based
a
ppr
oach
w
orks
on
a
nato
m
ic
al
info
rm
ation
of
the
br
ai
n,
it
does
not
require
trai
ning
of
the
data.
T
his
pa
pe
r
pro
po
ses
a
s
ymm
et
ry
base
d
m
et
ho
d
to
id
ent
ify
ano
m
al
y
pr
one
areas b
ase
d
on
cel
lular
a
uto
m
at
a
[27
]
,
[
28]
.
Since
cel
lular
au
tom
at
a
are
a p
arall
el
com
puta
ti
on
m
od
el
s,
differ
ent
va
riants
of
cel
lular
a
utom
at
a
is
us
ed
i
n
var
io
us
m
ed
ic
al
i
m
aging
a
pp
li
cat
io
ns
s
uc
h
as
cel
lular
aut
oma
ta
base
d
m
odel
for
pr
e
d
ic
ti
ng
patte
rn
of
de
ngue
fe
ver
[
29]
and
le
ar
ning
cel
lular
aut
oma
ta
[30]
for
tum
ou
r
det
ect
ion
in
m
a
mm
og
raphy.
So,
in
this
pap
e
r,
a
n
al
gorithm
based
on
CA
is
pro
po
se
d
to
fi
nd
ou
t
the
asym
m
et
ry
in
br
ai
n
im
ag
es
in
co
ns
ta
nt
tim
e.
In
order
to
fin
d
out
m
or
e
accu
ra
te
are
as,
that
are
pr
on
e
t
o
ano
m
al
y,
the
ne
ighbou
rho
od
inf
or
m
at
ion
a
bout
the
a
sym
metri
c
area
al
so
consi
der
e
d.
Si
nce
C
AD
syst
em
can
hav
e
m
ulti
ple
m
od
ules,
the
pro
posed
te
c
hn
i
qu
e
ca
n
be
int
egr
at
e
d
to
a
s
ophisti
cat
ed
CA
D
syst
em
to
en
han
ce
it
s
per
f
or
m
ance
or
i
t
can
be
us
e
d
as
an
i
ndepende
nt
CA
D
syst
e
m
to
assis
t
the
doct
ors
for
the
br
ai
n
im
age
interp
retat
ion.
2.
RESEA
R
CH
METHO
DS
2.1
.
Cell
ular
Au
t
om
ata
Ce
ll
ular
autom
at
a
are
com
pu
t
at
ion
al
m
od
el
consi
sts
of
an
arr
ay
of
cel
ls.
Each
cel
l
in
a
CA
act
s
as
a
processi
ng
el
e
m
ent.
It
is
a
pa
rall
el
com
pu
ta
ti
on
m
od
el
.
T
he
nei
ghbo
urh
ood
of
a
C
A
i
nd
ic
at
es
the
gr
oup
of
cel
ls
in
wh
ic
h
the
CA
r
ules
act
upon
to
upda
te
a
cel
l
s
ta
te
at
un
it
tim
e
step
.
T
he
in
pu
t
da
ta
can
be
sto
r
ed
as
sta
te
s
in
a
C
A.
T
he
sta
te
of
eac
h
cel
l
c
an
updates
in
unit
tim
e
step
s
base
d
on
a
local
r
ule
w
hi
ch
ar
e
char
act
e
rised
by
the
neig
hbour
hood
in
synch
r
onous
fas
hion.
Th
e
ru
le
of
the
CA
is
a
transiti
on
f
un
ct
io
n,
wh
ic
h
ta
kes
ne
ighbou
rho
od
a
nd
t
he
sta
te
of
a
cel
l
at
a
time
ste
p
as
par
a
m
et
ers
and
ret
urns
the
new
s
t
at
e
of
that
cel
li
n
the
nex
t
ti
m
e
ste
p.
CA
is
use
d
f
or
m
od
el
li
ng
c
om
plex
syst
em
s.
Steph
en
Wo
l
fr
am
has
done
a
n
extensi
ve
stu
dy
on
tw
o
dim
ension
al
cel
lular
autom
at
a.
Tw
o
di
m
ension
al
C
A
can
ha
ve
var
io
us
neig
hbour
hood
s.
I
n
order
to
find
out
the
a
no
m
al
y
pr
on
e
ar
e
as
in
a
br
ai
n
i
m
age
in
c
on
sta
nt
tim
e
com
ple
xity
,
a
m
od
ifie
d
cel
lular a
uto
m
at
a
m
od
el
call
e
d
ra
di
us
bounda
ry C
A
(RBC
A)
is
prop
os
ed
.
2.1.1
.
R
ad
ius
Boundar
y CA
The
propose
d
RB
CA
is
a
two
dim
ension
al
cel
lular
autom
at
a
con
ta
ins
gr
id
of
cel
ls.
RB
CA
w
orks
base
d
on
a
re
fer
e
nce
cel
l,
wh
ic
h
is
the
m
idd
le
cel
l
in
the
tw
o
dim
ensio
nal
gri
d
te
rm
ed
as
,
.
Th
e
neig
hbour
hood
of
the
cel
l
,
var
ie
s
in
each
r
adius
.
The
ra
di
us
in
an
RB
C
A
re
fer
s
t
o
th
e
distance
of
t
he
neig
hbour
hood
cel
lwit
h
res
pe
ct
to
t
he
cel
l
,
i
n
al
l
directi
ons
i.e.,
the
nu
m
ber
of
cel
ls
tra
ve
rsed
from
,
to
the
neig
hbour
hoods
of
,
wh
ic
h
are
the
bounda
ry
cel
ls
in
eac
h
rad
i
us
.
Ta
ble
1
s
hows
t
he
r
adius
str
uctu
re
in
an
RB
CA
,
R
1
i
nd
ic
at
es
a
nei
ghbo
urh
ood
cel
l
in
rad
i
us
1
a
nd
R
2
in
dicat
e
s
neig
hbour
hood
cel
l
in
ra
diu
s
2
with
resp
ect
t
o
th
e c
el
l
,
. Base
d on
,
, t
he
cel
ls i
n ea
ch
r
a
diu
s i
nteract
in parall
el
.
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
A symmet
ry
ba
sed a
noma
ly
det
ect
ion
in
b
r
ai
n usin
g
cel
lul
ar a
utomat
a
f
or
comp
uter
aid
e
d.
.
.
(
Ni
sha V M
)
473
Table
1.
RB
C
A’
s
r
a
diu
s
str
uc
ture
R
2
R
2
R
2
R
2
R
2
R
2
R
1
R
1
R
1
R
2
R
2
R
1
,
R
1
R
2
R
2
R
1
R
1
R
1
R
2
R
2
R
2
R
2
R
2
R
2
2.2.
RB
CA
M
od
el
fo
r
A
nomaly
D
e
tecti
on
In
t
he
pro
po
se
d
RB
CA b
ase
d
CAD
m
od
el
f
or
d
et
ect
in
g
a
nom
aly
pron
e
ar
eas
in b
rai
n,
a
r
egi
on
base
d
appr
oach
is
use
d.
T
he
brai
n
im
age
is
con
ve
r
te
d
into
a
two
dim
ension
al
arr
ay
based
on
i
ntensity
.
The
intensit
y
of
eac
h
pi
xe
l
va
ries
from
0
to
255.
T
he
cent
r
oid
of
the
im
age
is
cal
culat
ed
in
the
m
idd
le
axis.
T
he
centr
oi
d
of
the
i
m
age
is
t
reated
as
the
m
idd
le
cel
l
in
RB
CA
and
e
ach
pi
xel
valu
e
in
the
two
dim
ension
al
arra
y
are
consi
der
e
d
as
cel
ls
in
the
R
BC
A.
The
nei
ghbour
hood
cel
ls
of
the
centr
oid
in
va
rio
us
rad
i
us
disti
ngui
sh
the
reg
i
on.
T
he
se
t
of
nei
ghbour
hood
for
the
c
entr
oid
va
ries
accor
ding
t
o
t
he
rad
i
us
.
Sin
ce
the
br
ai
n
i
m
age
is
bilat
erall
y symm
et
rical
, in
o
rd
er to
ensu
re th
e b
il
at
eral sy
m
m
e
try
, th
e cor
res
pondin
g
cel
ls i
n
e
ach r
adiusbase
d
on
t
he
ce
ntr
oid
are
com
pu
te
d.
Ba
sed
on
a
t
hresh
old,
asy
m
m
et
ry
is
cal
cul
at
ed
in
eac
h
re
gion.T
he
locat
i
on
of
the asym
m
et
ric
r
e
gion also
ca
n
ide
ntify
base
d on the
re
gion
. F
ig
ure
1
s
hows
a
sam
ple aff
ect
ed brai
n
im
age.
Figure
1
.
Sam
ple
aff
ect
e
d br
ai
n
im
age
Table
2
. RBC
A
m
od
el
R
4
R
4
R
4
R
4
R
4
R
4
R
4
R
4
R
4
R
4
R
3
R
3
R
3
R
3
R
3
R
3
R
3
R
4
R
4
R
3
R
2
R
2
R
2
R
2
R
2
R
3
R
4
R
4
R
3
R
2
R
1
R
1
R
1
R
2
R
3
R
4
R
4
R
3
R
2
R
1
C
i,j
R
1
R
2
R
3
R
4
R
4
R
3
R
2
R
1
R
1
R
1
R
2
R
3
R
4
R
4
R
3
R
2
R
2
R
2
R
2
R
2
R
3
R
4
R
4
R
3
R
3
R
3
R
3
R
3
R
3
R
3
R
4
R
4
R
4
R
4
R
4
R
4
R
4
R
4
R
4
R
4
Unde
r
the
assu
m
pt
ion
that
th
e
br
ai
n
im
age
is
bilat
erall
y
s
ymm
et
ric
with
resp
ect
to
the
m
idd
le
axis,
the
cent
ro
i
d
is
cal
culat
ed.
B
oth
the
sides
of
t
he
a
xis
are
sy
m
m
e
tric
al
in
na
ture.
S
o,
i
n
order
to
c
hec
k
th
e
le
vel
of
a
sym
m
et
ry
exists
in
t
he
i
m
age,
ba
sed
on
the
cent
ro
i
d,
cal
culat
e
the
l
evel
of
asy
m
m
et
ry
by
com
pa
rin
g
the
values
of
t
he
r
especti
ve
cel
ls
in
both
the
si
des
of
the
a
xis
.
Table
2
sho
ws
the
RB
CA
m
od
el
,
w
he
re
,
is
treat
ed
as
cent
ro
i
d
ie
,
m
idd
le
cel
l
and
the
c
el
ls
in
the
colu
m
n
wh
ic
h
co
nt
ai
ns
,
is
treat
ed
as
the
axis
of
the
i
m
age.
Re
gion
based
c
om
pari
so
n
is
us
ed
i
n
this
te
chn
i
qu
e
.
The
reg
i
on
is
identifie
d
bas
ed
on
the
rad
i
us
.
In
each
ra
diu
s,
th
e
cel
ls
in
each
ro
w
m
ark
ed
with
sam
e
col
our
base
d
on
the
m
idd
le
cel
l
are
com
par
ed
fo
r
the
sy
m
m
e
try
.
Th
e
RB
CA
m
od
el
is
al
so
a
pa
rall
el
com
pu
ta
ti
on
m
od
el
be
cause
the
cel
ls
in
each
ra
di
us
are
ind
e
pende
nt
an
d
ba
sed
on
the
m
idd
le
cel
l,
pa
rall
el
acce
ss
to
cel
ls
in
each
r
adius
is
possib
le
.
In
orde
r
to
m
ake
the RB
CA m
o
del as a p
a
rall
el
co
m
pu
ta
ti
on
m
od
el
, th
e co
m
pu
ta
ti
on
p
r
oc
ess is d
isc
usse
d
bel
ow. A
s
sum
ing
the
co
-
ordinate
of
the cent
ro
i
d
as
(
,
)
.
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.
1
4
, N
o.
1
,
A
pr
il
201
9
:
471
–
477
474
Compari
son re
qu
ire
d
i
n
r
ad
i
us 1
:
The
m
idd
le
lev
el
cell
s:
(
,
−
1
)
,
(
,
+
1
)
To
p
le
vel cel
ls
:
(
(
−
1
)
(
−
1
)
(
−
1
)
(
+
1
)
)
Bott
om
level cel
ls:
(
(
+
1
)
(
−
1
)
(
+
1
)
(
+
1
)
)
Compari
son re
qu
ire
d
i
n
r
ad
i
us 2
:
The
m
idd
le
lev
el
cell
s:
(
(
,
−
2
)
(
,
+
2
)
)
,
(
(
−
1
)
(
−
2
)
(
−
1
)
(
+
2
)
)
,
(
(
+
1
)
(
−
2
)
(
+
1
)
(
+
2
)
)
.
To
p
le
vel cel
ls
:
(
(
−
2
)
(
−
1
)
)
(
−
2
)
(
+
1
)
)
,
(
(
−
2
)
(
−
2
)
(
−
2
)
(
+
2
)
)
.
Bott
om
level cel
ls:
(
(
+
2
)
(
−
1
)
(
+
2
)
(
+
1
)
)
,
(
(
+
2
)
(
−
2
)
(
+
2
)
(
+
2
)
)
So
,
total
com
par
iso
n
operati
ons
re
qu
ir
ed
in
this
RB
CA
m
od
el
var
y
acco
r
ding
to
the
radi
us
.
I
n
each
rad
i
us
,
the
t
otal
num
ber
of
c
om
par
ison
s
re
quire
d
is
cal
c
ula
te
d
based
on
th
e
ab
ove
c
om
pu
ta
ti
on
s.
Assum
e
that
is t
he radi
us
w
it
h
res
pect to
the ce
ntr
oid
,
To
p
le
vel
(
)
le
ve
l,t
he
num
ber
of com
par
iso
ns
r
equ
i
red
:
com
par
iso
ns
Bott
om
level
(
)
l
evel, the
num
ber
of
c
om
par
ison
s
r
e
quire
d:
com
par
isons
Mi
dd
le
level,
the
nu
m
ber
of c
om
par
isons re
quire
d:
(
2
∗
)
−
1
com
par
isons
The
cel
ls
in
th
e
top
an
d
bott
om
le
vels
of
eac
h
ra
dius
with
r
espect
to
t
he
c
entr
oid
,
in
th
e
asym
m
e
tric
reg
i
on are
c
om
par
e
d wit
h
it
s
neig
hbour
hood
cell
s to
c
heck
the level
of asy
m
m
e
try
to
en
s
ur
e
m
or
e accu
r
acy
.
2.2.1. Alg
orit
h
m 1: chec
k
the
a
s
ymme
try in a
br
ain im
age
Step1
:
Fin
d
the
centr
oid o
f
t
he
g
ive
n
im
age,
a
fter
or
ie
ntati
on of the
sym
m
etr
ic
p
la
ne
in
the
sp
ace
.
Step
2: Let
be
t
he
tw
o dim
ension
al
a
rr
ay
,
whi
ch
st
or
es
the i
nt
eger
values
of
the im
age,
,
bethe
ce
ntr
oid
of the im
age
Step
3: Let
be
t
he radi
us
of th
e i
m
age f
r
om
the ce
ntr
oid
Step
4:
F
or eac
h radi
us
,
=
1
to
bo
undary
do in
P
ara
ll
el
Step
5:
F
or eac
h row,
=
0
to
doin
Par
allel
Step
5.1: if
is no
t e
qual
to
Step
5.1.1: C
om
par
e the v
al
ue
s of
(
(
−
)
(
−
)
)
and
(
(
−
)
(
+
)
)
Step
5.1.1.
1: if
the
diff
e
re
nce
is gr
eat
e
r
tha
n t
hr
es
ho
l
d
Assign ze
r
o
to
the im
age lo
cat
ion
,
hold
s the
gr
eat
er
v
al
ue.
Step
5.1.2:
Co
m
par
e the v
al
ue
s of
(
(
+
)
(
−
)
)
and
(
(
+
)
(
+
)
)
Step
5.1.2.
1: i
f
the
d
if
fer
e
nce
is great
er t
han thr
es
hold,
Assign ze
r
o
to
the im
age lo
cat
ion
,
hold
s the
gr
eat
er
v
al
ue
Step
5.2: Else
Step
5.2.1: F
or each c
olu
m
n,
=
0
d
o
in
P
ara
ll
el
Step
5.2.1.
1:Com
par
e the
val
ues of
(
(
−
)
(
−
)
)
an
d
(
(
−
)
(
+
)
)
Step
5.2.1.
1.1:
if the
d
if
fere
nc
e is g
reater t
ha
n
th
res
ho
l
d,
Assign ze
r
o
to
the im
age lo
cat
ion
,
hold
s the
gr
eat
er
v
al
ue
Step
5.2.1.
2:
C
om
par
e the
val
ues of
(
(
+
)
(
−
)
)
an
d
(
(
+
)
(
+
)
)
Step
5.2.1.
2.1:
if the
dif
fer
e
nc
e is g
reater t
ha
n
th
res
ho
l
d,
Assign ze
r
o
to
the im
age lo
cat
ion
,
hold
s the
gr
eat
er
v
al
ue.
Step
5.2.2:
End
for
St
ep 5.
3:
Com
par
e the
v
al
ues of l
eft, r
i
gh
t
an
d m
id
dle cel
l for
zero,
Step
5.3.1: I
f
a
ll
the thr
ee
v
al
ues
a
re ze
ro,
Assign ze
r
o
to
m
idd
le
cell
, 255
oth
e
rw
ise
.
Step
6: E
nd of
ste
p 5
Step
7: E
nd
of step
4
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
A symmet
ry
ba
sed a
noma
ly
det
ect
ion
in
b
r
ai
n usin
g
cel
lul
ar a
utomat
a
f
or
comp
uter
aid
e
d.
.
.
(
Ni
sha V M
)
47
5
2.2.2. Descri
p
t
ion
of t
he
A
l
gori
th
m
The
al
gorithm
to
fin
d
out
the
asym
m
et
ry
exists
in
bilat
eral
ly
sy
m
m
et
ric
br
ai
n
im
ages
are
pr
opos
e
d
in
Algorit
hm
1
.
Assum
e
that
a
T2
-
weig
hted
br
ai
n
is
c
onve
rted
int
o
a
two
dim
ension
al
arr
ay
,
with
inte
ge
r
values
ra
ngin
g
fr
om
0
to
255.
The
intensit
y
of
eac
h
pix
el
is
conver
te
d
into
intege
r
value
s.
Step
1
fi
nd
out
the
centr
oid
of
the
i
m
age
and
col
um
n
of
the
arra
y
wh
ic
h
co
nta
ins
the
centr
oid
is
con
si
der
e
d
as
m
idd
le
axis
of
the
i
m
age.
Step 2
assign
t
he
im
a
ge
into a t
wo
dim
ension
al
ar
r
ay
. S
te
p
4
i
niti
al
iz
es the
rad
iu
s,
from
the cen
troi
d
to
t
he
boun
dary
of
t
he
im
age.
I
n
ste
p
5,
for
loopis
us
ed
to
c
heck
eac
h
r
ow
from
0
t
o
,
wh
e
re
is
the
rad
i
us.
In
each
ra
diu
s
,
th
ere
are
th
ree
le
vels
of
com
pari
so
ns
s
uc
h
as
t
op
le
vel,
bott
om
le
vel
and
m
i
dd
le
le
vel
is
re
qu
i
red.
Si
nce
the
num
ber
of
c
om
par
ison
s
require
d
in
each
le
vel
is
diff
e
ren
t,
st
ep
5
is
use
d
t
o
chec
k
w
heth
er
the
sel
ect
ed
r
ow
is
eq
uiv
al
ent
to
rad
i
us
le
vel
or
no
t.
I
f
it
is
not
in
ra
diu
s
le
vel
,
ste
ps
from
5.
1.1
to
5.1.2
a
re
us
e
d
for
m
idd
le
le
ve
l
com
par
iso
ns.
I
n
eac
h
ra
dius,
(
2
∗
)
−
1
com
par
iso
ns
are
re
qu
i
red
in
the
m
idd
le
le
vel
with
the
co
rr
es
pond
ing
cel
ls
in
eac
h
r
ow.
If
t
he
re
sp
ect
ive
values
are
no
t
e
qual
,
base
d
on
a
thre
sh
ol
d,
a
ssig
n
a
0
in
the
locat
ion
w
her
e
the
sym
m
et
ry
is
br
ea
king,
to
in
dic
at
e
the
asym
m
et
ry.
Step
5.2
is
us
ed
for
the
el
se
conditi
on,
ie
,
top
an
d
bott
om
le
vel
com
p
ariso
ns
i
n
eac
h
rad
i
us
.
T
he
top
an
d
bott
om
le
vel,
num
ber
of
com
par
isons
re
qu
i
red.
I
n
ste
p
5.2.1,
f
or
l
oop
is
us
e
d
to
acce
ss
each
col
um
n
value
s.In
ea
ch
case,
t
he
al
go
r
it
h
m
checks
f
or
th
e
as
ymm
et
ry.
Step
5.3
c
he
cks
the
le
ft
a
nd
rig
ht
neig
hbour
hood
values
of
e
ach
c
el
l
for
asym
m
e
try
,
if
the
neig
hbou
r
hood
cel
ls
are
al
so
asy
m
m
etr
ic
,
m
ark
the
cel
l
value
as
0
t
o
in
dicat
e
an
om
aly
pro
ne
area
.
S
o, with t
his alg
ori
th
m
, th
e asy
m
m
et
ric areas, w
hich
a
re
pr
one
to anom
al
y, can
be
ide
ntifie
d
.
3.
RESU
LT
S
A
ND
A
N
ALYSIS
In
or
der
to
pe
r
form
the
exp
er
i
m
ental
wo
rk
to
i
m
ple
m
ent
t
he
ab
ov
e
m
entione
d
al
gorith
m
to
identify
the
asym
m
e
try
exists
in
bilat
e
rall
y
sy
m
m
e
tric
i
m
ages;
T2
-
weig
hted
im
ag
es
are
us
ed
.Ma
tl
ab
-
ve
rsion
20
18
is
us
e
d
f
or
t
he
im
plem
entat
ion
of
the
a
bove
m
entione
d
al
gorithm
.Con
side
red
aff
ect
e
d
brai
n
i
m
ages.
The
i
m
age
is
co
nv
e
rted
in
to
a
t
wo
dim
en
sion
al
a
rr
ay
,
th
e
val
ues
of
the
ar
ray
va
ries
from
0
to
25
5
ba
sed
on
t
he
inte
ns
it
y
of
eac
h
pix
el
i
n
the
im
age.
E
xp
e
rim
ents
hav
e
do
ne
on
dif
fer
e
nt
sta
ndar
d
br
ai
n
im
ages.
Figure
2
an
d
F
igure
3
sh
ow
n
th
e sam
ple in
pu
t a
nd
outp
ut.
(a)
(b)
(c)
(d)
Figure
2
.
(a
)
A
ff
ect
ed
brain
,
(
b)
A
no
m
al
y pron
e
areas
d
et
ec
te
d
,
(c)
A
ff
ect
e
d br
ai
n,
(d)
Anom
al
y pr
one
a
reas
dete
ct
ed
(a)
(b)
(c)
(d)
Figure
3
.
(
a
)
A
ff
ect
ed
brain
,
(
b)
A
no
m
al
y pron
e
areas
d
et
ec
te
d
,
(c)
A
ff
ect
e
d br
ai
n,
(d)
Anom
al
y pr
one a
reas
dete
ct
ed
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.
1
4
, N
o.
1
,
A
pr
il
201
9
:
471
–
477
476
The
RB
CA
m
od
el
pro
pose
d
he
re
is
a
pa
rall
el
com
pu
ta
ti
on
al
m
od
el
.
The
c
om
pu
ta
ti
on
s
in each
ra
dius
perform
in
paral
le
l
and
the
c
om
par
ison
ope
rati
on
s
i
n
the
t
op
le
vel,
bott
om
le
vel
and
m
idd
le
le
vel
in
each
rad
i
us
al
so
pe
rfor
m
s
in
par
al
l
el
.
So,
t
he
ti
m
e
com
plexi
ty
of
the
a
bove
m
entioned
al
gorithm
is
co
nst
ant.
In
each
reg
i
on,
ie
,
ra
diu
s
,
t
he
al
gorithm
checks
f
or
the
im
m
e
diate
nei
ghbourh
ood
f
or
the
a
sy
m
m
e
try
resul
te
d
in
detect
ion
of
m
or
e
accurate
ano
m
al
y
pr
one
area
.
With
this
pr
op
os
e
d
m
od
el
,
eve
n
ver
y
m
ini
m
al
le
vel
of
asym
m
e
try
al
so
ca
n
be
detect
ed,
w
hich
hel
ps
the
physi
ci
an
s
to
i
den
ti
fy
t
he
area
in
w
hich
m
or
e
f
ocu
s
ne
eds
t
o
be
ca
rr
ie
d
ou
t.
So
do
ct
or
sca
n
i
nterpret
the
brai
n
im
ages
pro
per
ly
,
t
hat
resu
lt
i
n
a
pro
per
dia
gnos
is
of
th
e
disease
by r
e
duci
ng
obser
vational
o
ver
si
gh
t
of the
doct
ors.
3.1.
C
ompari
so
n wi
t
h
Exis
t
ing Tech
nique
s
Since
the
pro
pose
d
m
et
ho
d
is
a
sy
m
m
e
try
base
d
m
et
ho
d,
un
li
ke
t
he
qu
al
it
at
ive
analysis
on
brai
n
ano
m
al
y detec
ti
on
,
im
age r
egi
strat
ion
,
traini
ng the
d
at
a is
no
t require
d.
Sym
m
e
try
b
ased
app
r
oach is ba
sed on
relat
ive
data,
so
the
de
pe
ndency
of
the
pi
xel
va
riat
ion
s
can
be
m
ini
m
i
zed
.
Most
of
the
sym
m
et
ry
bas
e
d
te
chn
iq
ues
u
se
d
f
or a
no
m
al
y
detect
ion
us
e
d extensi
ve
c
omparis
ons, whic
h
le
ads
to hig
h t
i
m
e co
m
plexity
. The
works
w
hich
use
s
boun
ding
box
te
ch
nique
s,
[
18
]
-
[
22]
to
detect
ano
m
aly
based
on
sy
m
m
e
try
need
s
(
2
)
tim
e,
wh
ere
is
the
input
siz
e
.
W
he
reas
RB
C
A
m
od
el
requi
res
only
co
ns
ta
nt
tim
e
to
find
the
an
om
al
y
pron
e
areas.
S
o
the
pro
posed
RB
C
A
m
od
el
is
a
m
or
e
eff
ic
ie
nt
m
od
el
based
on
sym
m
et
ry
to
detect
br
ai
n
le
sion
s
.
Figure
4
s
how
s
the
tim
e
com
plexit
y
co
m
par
is
on
of
othe
r
te
chn
i
ques
and
RB
CA
m
od
el
.
T
he
gr
a
ph
in
the
gr
ee
n
c
olour
ind
ic
at
e
the
(
2
)
tim
e
and
t
he
gr
aph
in
t
he
re
d
colo
ur
in
dicat
e
the
c
on
sta
nt
ie
,
(
1
)
tim
e
com
plexity
.
Figure
4
.
Tim
e
co
m
plexity
com
par
ison
4.
CONCL
US
I
O
N
CAD
syst
em
is
con
side
re
d
as
a
com
ple
m
entary
com
pu
ti
ng
powe
r
to
diag
nose
the
disease
with
m
or
e
accuracy.
C
A
D
inc
reases
t
he
cha
nces
of
de
te
ct
ion
of
dis
eases
by
assist
ing
t
he
physi
ci
ans
in
dec
reasi
ng
the
ob
s
er
vational
ov
e
rsight
in
t
he
early
sta
ge
of
disease
s.
CA
D
ca
n
be
asse
m
bled
as
packages
a
nd
im
ple
m
ented.
The
RB
CA
m
o
del
pro
po
se
d
in
this
pa
per
he
lps
the
physi
ci
ans
to
ide
ntify
the
area
in
w
hich
m
or
e
fo
cu
s
nee
ds
to
be
car
ried
ou
t.
M
os
t
of
the
cases
the
s
m
al
l
le
sion
s
m
ay
be
m
issed
by
the
do
ct
or
s
,
in
co
r
por
at
ion
of
asym
m
e
tric
info
rm
at
ion
al
ong
with
the
CA
D
syst
e
m
enh
an
c
e
the
perform
ance
of
the
sys
tem
by
red
ucin
g
the
ob
s
er
vational
ov
e
rsight.
T
he
propose
d
RB
CA
m
od
el
can
be
integrate
d
with
a
soph
ist
i
cat
ed
CAD
sy
stem
or
m
ay
wo
rk
as
a
sta
nd
al
one
C
AD
syst
em
to
fin
d
the
an
om
a
ly
pr
one
areas
in
hum
an
br
ai
n.
Si
nce
the
pr
opose
d
m
od
e
l
is
base
d
on
cel
lular
a
ut
om
a
ta
,
in
c
onsta
nt
tim
e
com
plexity
,
the
RB
CA
m
od
el
de
te
ct
s
ano
m
al
y
pro
ne
areas in
hum
an
brai
ns
.
REFERE
NCE
S
[1]
Doi
,
K
.
,
“
Diagn
ostic
imaging
o
ver
the
la
st
50
y
e
ars:
re
se
arc
h
and
deve
lopmen
t
in
m
edi
ca
l
im
agi
ng
scie
n
ce
a
nd
te
chno
log
y
”,
Ph
ysic
s in
Me
d
ic
in
e
&
Bi
ology
,
20
06
.
[2]
Giger
,
M.
L
.
,
et
al
.
“
Com
pute
r
-
aided
dia
gnosis
in
m
edi
ca
l
imagin
g
”,
I
EE
E
trans
act
ions
on
med
i
cal
imaging
.
vol
.
20
,
pp
.
1205
-
12
08
,
2001
.
[3]
Doi,
K.
,
“
Com
p
ute
r
-
a
ide
d
d
ia
gn
osis
in
m
edi
cal
imaging:
histor
i
ca
l
r
eview,
cu
rr
ent
sta
tus
and
f
uture
pot
ent
i
al
”,
Computerized
m
edi
ca
l
imaging
a
nd
graphics
.
Vol
.
31
,
no
.
4
,
pp
.
1
98
-
211
.
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
A symmet
ry
ba
sed a
noma
ly
det
ect
ion
in
b
r
ai
n usin
g
cel
lul
ar a
utomat
a
f
or
comp
uter
aid
e
d.
.
.
(
Ni
sha V M
)
477
[4]
Doi,
K.
,
“
Curre
n
t
stat
us
and
futu
re
pote
nt
ial
of
c
om
pute
r
-
ai
ded
d
ia
gnosis
in
m
edica
l
imagi
ng
”,
Th
e
Brit
ish
journal
of
radiolog
y.
Vol.
78
,
pp
.
s3
-
s
19
,
2005
.
[5]
Van
,
G.
,
Bram,
et
.
a
l
.
“
Com
pute
r
-
ai
ded
dia
gnosis
in
ch
est
ra
d
iogr
aph
y
:
a
surve
y
”,
IEE
E
Tr
ansactions
on
medic
al
imaging
.
Vol
.
20
,
no
.
12
,
pp
.
122
8
-
1241
,
2001
.
[6]
Duggira
la,
B
.
,
et.al
.
“
Com
pute
r
a
ide
d
dia
gnost
ic
assistanc
e
for
m
edi
c
al
imaging
”,
Sie
mens
Me
dical
Solut
ions
US
A
Inc
,
U
.
S.
Pa
te
nt
,
vol.
17
,
pp
.
244
-
230
,
2007
.
[7]
Ham
ia
ne,
M
.,
a
nd
Saee
d
,
F.
,
“
SV
M
Cla
ss
ifi
cation
of
MRI
Br
ai
n
Im
age
s
for
Com
pute
r
-
As
sist
ed
Diagnosis
”,
Inte
rnational
Jo
urnal
of El
e
ct
ri
c
al
and
Comput
er
Engi
n
ee
ring
(
IJ
ECE
)
.
Vol.
7
,
no
.
5
,
pp
.
2555
-
25
64
,
2017
.
[8]
Marshkole
,
e
t.
a
l
,
.
“
Te
x
ture
and
shape
base
d
class
ifi
cation
of
bra
int
um
ours
using
li
ne
ar
ve
ct
o
r
quant
i
za
t
ion
”
,
I
nte
rnational
Jo
urnal
of
Comput
er
Applications.
v
ol
.
30
,
no
.
11
,
p
p.
21
-
23
,
2011
.
[9]
Bhim
a
,
K
.,
and
Jaga
n,
A.
,
“
An
eff
icient
m
et
hod
for
ide
n
ti
fi
ca
t
io
n
of
anomalies
in
bra
in
MR
images.
Signa
l
and
Inform
at
ion
Pro
ce
ss
ing
(IConSI
P)
”
,
Int
ernati
on
al
Conf
ere
nce o
n
IEEE.
Pp.1
-
5,
2016
.
[10]
Charut
ha
,
S
.,
an
d
Ja
y
ashre
e
,
M.
J.
,
“
An
eff
i
cie
nt
bra
in
tumour
det
e
ct
ion
b
y
in
te
gra
ti
ng
m
o
dified
te
x
ture
b
ase
d
re
gion
growing
and
c
el
lu
la
r
aut
om
at
a
edg
e
detec
t
ion
”,
Control,
Instru
mentat
ion
,
Co
mm
unic
ati
on
a
nd
Computati
onal
T
ec
hnolog
ie
s (
ICCICCT
)
,
Inte
rnational
Con
fe
ren
c
e
on.
IE
EE,
pp
.
1193
-
1199,
201
4
.
[11]
Bane
rj
ee,
et
al
.
,
“
Single
see
d
del
in
ea
t
ion
of
bra
int
um
our
u
sing
m
ult
i
-
thre
s
holdi
ng
”,
Infor
mation
Scienc
e
s
,
pp.
330:
88
-
103
,
2016
.
[12]
Ham
ia
ne
,
M
.,
a
nd
Fate
m
a
,
S.
,
“
SVM
Cla
ss
ifi
ca
ti
on
of
MRI
Brai
n
Im
age
s
for
Com
pute
r
-
As
si
sted
Diagnosis
”,
Inte
rnational
Jo
urnal
of El
e
ct
ri
c
al
and
C
omput
er
Engi
n
ee
ring
(
IJ
ECE
)
,
vol
.
7
,
no
.
5
,
pp.
2555
-
256
4
,
2017
.
[13]
Khotanl
ou
,
H
.
,
et
a
l.
“
3D
bra
in
tumor
segm
ent
at
ion
in
MRI
u
sing
fuz
z
y
cl
ass
ifi
c
at
ion
,
s
y
m
m
et
r
y
an
aly
sis
an
d
spati
all
y
constra
i
ned
def
orm
ab
le
m
odel
s
”,
Fuzzy
sets
and
syst
ems
,
vol.
160
,
no
.
10
,
2009
.
[14]
Selva
kum
ar,
J
.
,
et.
al.
,
“
Brai
n
t
um
or
segm
ent
at
ion
and
it
s
ar
ea
ca
lc
u
lation
in
bra
in
MR
images
using
K
-
m
ea
n
cl
uster
ing
and
Fuzz
y
C
-
m
ea
n
al
gorit
hm
”,
A
dvanc
es
in
En
gine
ering
,
Scie
nce
and
Mana
geme
nt
(
ICAE
S
M)
,
Inte
rnational
Co
nfe
renc
e
on
.
I
EEE,
2012
[15]
Murugava
lli,
S
.,
and
Raj
amani,
V.
,
“
An
I
m
prov
ed
Im
ple
m
ent
at
i
on
of
Brai
n
Tumor
Dete
ct
ion
Us
ing
Segm
ent
at
ion
Based
on
N
eur
o
Fuzz
y
T
ec
hni
qu
e”
,
Journal
of
C
omputer
Scienc
e
.
vol
.
3
,
no
11
,
p
p.
841
-
846
,
200
7
.
[16]
Hara
ti,
V
.,
e
t.
al.
,
“
Full
y
aut
om
at
e
d
tumour segm
ent
at
ion
base
d
on
improved
fuz
z
y
conne
c
te
dness al
gorit
hm
in
bra
in
MR image
s
”,
Co
mputers i
n
bio
lo
gy
and
m
edi
c
ine.
vol
.
4
,
no.
7
,
pp.
483
-
492
,
2011
.
[17]
Rana
,
R
.
,
e
t.
al.
,
“
Com
par
at
ive
Stud
y
of
Segm
entati
on
T
ec
hniqu
e
s
for
Ext
ra
cting
Brai
n
Tumor
from
MRI
I
m
age
.
In
Proc.
o
f the
sec
o
nd
Intl.
Con
f. on
Adv
an
ce
s
in El
e
ct
ronics,
Elec
tri
cal
and
Comput
er
Engi
n
ee
ring
–
EE
C,
2013.
[18]
Aks
ha
y
a
,
R
.,
a
nd
Hem
a,
P.
M.
,
“
A
Revi
ew
on
Regi
strat
ion
of
Medic
al
Im
age
s
U
sing
Graph
The
ore
tic
Approac
hes
”
,
In
donesian
Journa
l
of
Elec
tri
cal
E
ngine
ering
and
Computer
Scien
ce
.
vol.
12
,
no.
1
3,
pp.
974
-
983
,
2018.
[19]
Li
u,
S
.
X.
,
“
S
y
m
m
et
ry
and
as
ym
m
et
ry
anal
y
s
is
and
it
s
implicat
ions
to
compute
r
-
ai
ded
d
ia
gnosi
s:
A
re
vie
w
of
t
he
li
te
r
at
ur
e
”,
Journal
of
Bi
omed
ical
Informat
i
cs
.
Vol.
42
,
no.
6
,
p
p.
1056
-
1064
,
2
009
.
[20]
Ruppert
,
et
a
l.
“
A
new
sy
m
m
et
r
y
-
b
ase
d
m
et
ho
d
for
m
id
-
sagit
t
al
pla
n
e
ex
tra
c
tion
in
neur
oima
ges
”,
Biom
edica
l
Im
agi
ng:
From
Nano
to
Ma
cro
,
IEE
E
Inte
rnat
io
nal
Symposium.
pp.
285
-
288
,
20
11
.
[21]
Khotanl
ou
,
H
.
,
et.
al.
,
“
Autom
at
i
c
bra
in
tumour
segm
ent
at
ion
us
ing
s
y
m
m
et
r
y
a
naly
s
is
and
def
o
rm
abl
e
m
odel
s
”,
Adv
anc
es
in
Patt
ern
Recogni
t
ion.
pp
.
198
-
202
,
20
07
.
[22]
Sachi
n
,
N
.
,
et
a
l.
,
“
Brai
n
Tumo
r
Dete
ction
Based
on
Bil
atera
l
S
y
m
m
et
r
y
Infor
m
at
ion
”,
arXi
v
preprint
arXiv
:
pp.
1412
-
3009
,
2014.
[23]
Saddique
,
M
.
,
et
.
al.
,
“
A
H
y
br
id
appr
oa
ch
of
using
s
y
m
m
et
r
y
te
chn
ique
fo
r
bra
in
tumour
segm
ent
at
io
n
”,
Computati
onal
a
nd
mathematical
methods
in
medi
ci
ne
.
2014
[24]
Ra
y
,
Ni
la
nj
an
.
et
.
al.
,
“
Us
ing
s
y
m
m
et
r
y
to
detec
t
abnor
m
alitie
s
in
bra
in
MR
I
”,
Comput
er
S
oci
e
ty
o
f
Ind
ia
Comm
unic
ati
ons
.
Vol
.
31
,
no
.
19
,
pp.
7
-
10
,
2008
.
[25]
Ra
y
,
Nil
anj
an
.
et
.
al.,
“
Locat
ing
bra
in
tumours
from
MR
ima
ger
y
using
s
y
m
m
et
r
y
”,
Signa
ls,
Syste
ms
an
d
Computers,
ACS
SC 20
07
Conf
ere
nce Rec
ord of
t
he
Fort
y
-
Fi
rs
t
A
silomar Confe
re
nce
on
.
I
EEE,
20
07
.
[26]
Sim
,
et
.
al
.
,
“
Anal
y
s
is
of
an
a
xia
l
T2
weigh
t
ed
bra
in
MRI
”,
Asia
-
Pacific
J
ournal
of
Conv
erge
nt
R
ese
arc
h
Inte
rchange
,
vo
l
.
3,
no
.
1
pp.
45
-
55
,
2017
.
[27]
V
on
,
N.
J
.,
and
Arthur
,
W.
,
“
Burks.
The
or
y
of
s
elf
-
re
produc
ing
au
tomata
”
,
I
EE
E
Tr
ansacti
ons
on
Neural
Ne
tworks
.
Vol.
5
,
no
.
1
,
pp
.
3
-
14
,
1966
.
[28]
Packa
rd
,
N
.,
an
d
W
olfra
m
,
S.
,
“
Two
-
dimensional
ce
l
lul
ar
au
t
om
at
a”
,
Journal
of
Sta
ti
sti
cal
p
hysic
s
.
Vol
.
38
,
pp.
901
-
946
,
19
85
.
[29]
Hos
en,
P.
E
.
,
et.a
l
.
,
“
A
ce
l
lul
a
r
aut
om
at
a
m
ode
li
ng
for
visualiz
ing
and
pre
di
cti
ng
sprea
ding
pa
tt
ern
s
of
dengu
e
fe
ver
”
,
TEL
KOMNIKA
(
Tele
communic
ati
on
Computing
El
e
ct
r
onic
s
and
Control)
,
vol.
14
,
no.
1
,
pp.
228
-
2
37
,
2016
.
[30]
Chagha
ri
,
E
.,
an
d
Kari
m
i,
A.
A
.,
“
Novel
Approac
h
for
Tumor
Dete
c
ti
on
in
Ma
m
m
ogra
phy
Im
a
ges
”,
Indone
sia
n
Journal
of
Elec
t
rical
Engi
ne
erin
g
and
Computer
Sci
en
ce
.
vol
.
12
,
no.
8
,
pp.
6211
-
6216
,
2014
.
Evaluation Warning : The document was created with Spire.PDF for Python.