T
E
L
KO
MNIK
A
, V
ol
.
1
7
,
No.
6
,
Dec
e
mb
er
201
9
, p
p.
3
21
8
~
3225
IS
S
N: 1
69
3
-
6
93
0
,
accr
ed
ited
F
irst
Gr
ad
e b
y K
em
en
r
istekdikti,
Decr
ee
No: 2
1/E/
K
P
T
/20
18
DOI:
10.12928/TE
LK
OM
N
IK
A
.v
1
7
i
6
.
13220
◼
32
18
Rec
ei
v
ed
Ma
y
27
,
20
1
9
; R
ev
i
s
ed
J
u
l
y
2
, 2
01
9
; A
c
c
ep
t
ed
J
ul
y
18
,
20
1
9
Aut
om
a
tic
i
mag
e slice m
arkin
g
prop
a
gat
i
on
on
se
gm
e
nt
a
tio
n o
f
den
tal
C
BC
T
Ag
u
s Zainal
Ari
f
in
*
1
,
E
va
n
Tan
u
w
ija
ya
2
, Ba
sk
o
r
o
Nu
g
r
o
h
o
3
, Ar
if
Mud
i P
r
iy
atn
o
4
,
Rar
a
smay
a I
n
d
r
as
w
a
r
i
5
,
E
h
a Re
n
w
i A
stu
t
i
6
, Dini Ad
n
i Nav
a
star
a
7
1
,2
,3
,4
,5
,
7
Dep
a
r
tm
e
n
t
o
f
I
n
fo
r
m
a
t
i
c
s
,
F
a
c
u
l
t
y
o
f
In
f
o
rm
a
ti
o
n
a
n
d
Com
m
u
n
i
c
a
ti
o
n
T
e
c
h
n
o
l
o
g
y
,
In
s
ti
tu
t
T
e
k
n
o
l
o
g
i
Se
p
u
l
u
h
No
p
e
m
b
e
r
(I
TS),
Su
ra
b
a
y
a
,
In
d
o
n
e
s
i
a
6
Dep
a
rt
m
e
n
t
o
f
Den
to
m
a
x
i
l
l
o
fa
c
i
a
l
Ra
d
i
o
l
o
g
y
,
Fa
c
u
l
t
y
o
f
De
n
t
a
l
M
e
d
i
c
i
n
e
,
Uni
v
e
r
s
i
t
a
s
A
i
rl
a
n
g
g
a
,
Su
ra
b
a
y
a
,
I
n
d
o
n
e
s
i
a
*C
o
rre
s
p
o
n
d
i
n
g
a
u
th
o
r,
e
-
m
a
i
l
:
a
g
u
s
z
a
@
c
s
.
i
ts
.
a
c
.
i
d
*1
,
e
v
a
n
ta
n
u
wij
a
y
a
.
1
8
0
5
1
@m
h
s
.i
ts
.a
c
.i
d
2
,
b
a
s
k
o
ro
.
1
8
0
5
1
@m
h
s
.i
ts
.
a
c
.
i
d
3
,
a
ri
f.
1
8
0
5
1
@m
h
s
.i
ts
.
a
c
.i
d
4
,
r
a
ra
s
m
a
y
a
1
6
@m
h
s
.
i
f.
i
t
s
.a
c
.i
d
5
,
e
h
a
re
n
w
i
@gm
a
i
l
.
c
o
m
6
,
d
i
n
i
_
n
a
v
a
s
ta
ra
@
i
f.
i
t
s
.a
c
.i
d
7
Ab
strac
t
Con
e
Be
a
m
Com
p
u
te
d
To
m
o
g
ra
p
h
y
(CBC
T)
i
s
a
ra
d
i
o
g
ra
p
h
i
c
te
c
h
n
i
q
u
e
th
a
t
h
a
s
b
e
e
n
c
o
m
m
o
n
l
y
u
s
e
d
to
h
e
l
p
d
o
c
to
r
s
p
ro
v
i
d
e
m
o
re
d
e
ta
i
l
e
d
i
n
f
o
rm
a
ti
o
n
fo
r
fu
rth
e
r
e
x
a
m
i
n
a
t
i
o
n
.
Te
e
t
h
s
e
g
m
e
n
ta
t
i
o
n
o
n
CBC
T
i
m
a
g
e
h
a
s
m
a
n
y
c
h
a
l
l
e
n
g
e
s
s
u
c
h
a
s
l
o
w
c
o
n
tra
s
t,
b
l
u
rre
d
te
e
t
h
b
o
u
n
d
a
r
y
a
n
d
i
rre
g
u
l
a
r
c
o
n
to
u
r
o
f
th
e
t
e
e
th
.
I
n
a
d
d
i
t
i
o
n
,
b
e
c
a
u
s
e
th
e
CBC
T
p
ro
d
u
c
e
s
a
l
o
t
o
f
s
l
i
c
e
s
,
i
n
whi
c
h
th
e
n
e
i
g
h
b
o
ri
n
g
s
l
i
c
e
s
h
a
v
e
re
l
a
te
d
i
n
fo
rm
a
ti
o
n
,
t
h
e
s
e
m
i
-
a
u
to
m
a
ti
c
i
m
a
g
e
s
e
g
m
e
n
t
a
ti
o
n
m
e
th
o
d
,
th
a
t
n
e
e
d
s
m
a
n
u
a
l
m
a
rk
i
n
g
fro
m
th
e
u
s
e
r,
b
e
c
o
m
e
s
e
x
h
a
u
s
ti
v
e
a
n
d
i
n
e
ff
i
c
i
e
n
t.
In
t
h
i
s
re
s
e
a
rc
h
,
we
p
r
o
p
o
s
e
a
n
a
u
to
m
a
ti
c
i
m
a
g
e
s
l
i
c
e
m
a
rk
i
n
g
p
ro
p
a
g
a
ti
o
n
o
n
s
e
g
m
e
n
ta
ti
o
n
o
f
d
e
n
ta
l
CBC
T.
Th
e
s
e
g
m
e
n
ta
ti
o
n
re
s
u
l
t
o
f
t
h
e
f
i
rs
t
s
l
i
c
e
w
i
l
l
b
e
p
ro
p
a
g
a
te
d
a
s
th
e
m
a
rk
e
r
fo
r
th
e
s
e
g
m
e
n
ta
ti
o
n
o
f
th
e
n
e
x
t
s
l
i
c
e
s
.
T
h
e
e
x
p
e
ri
m
e
n
ta
l
r
e
s
u
l
ts
s
h
o
w
th
a
t
th
e
p
ro
p
o
s
e
d
m
e
th
o
d
i
s
s
u
c
c
e
s
s
f
u
l
i
n
s
e
g
m
e
n
t
i
n
g
t
h
e
te
e
th
o
n
CBC
T
i
m
a
g
e
s
wi
t
h
th
e
v
a
l
u
e
o
f
M
i
s
c
l
a
s
s
i
fi
c
a
t
i
o
n
Err
o
r (M
E) a
n
d
Re
l
a
t
i
v
e
F
o
re
g
ro
u
n
d
Are
a
Er
ro
r (RAE) o
f
0
.1
1
2
a
n
d
0
.4
7
8
,
re
s
p
e
c
ti
v
e
l
y
.
Key
w
ords
: a
u
to
m
a
ti
c
s
e
g
m
e
n
ta
ti
o
n
,
d
e
n
ta
l
CBCT,
h
i
e
ra
r
c
h
i
c
a
l
c
l
u
s
te
ri
n
g
,
m
e
a
n
-
s
h
i
ft
,
m
o
rp
h
o
l
o
g
y
.
Copy
righ
t
©
2
0
1
9
Uni
v
e
rsi
t
a
s
Ahm
a
d
D
a
hl
a
n.
All
rig
ht
s
r
e
s
e
rve
d
.
1.
Int
r
o
d
u
ctio
n
Cone
-
b
ea
m
c
o
mp
ute
d
to
mo
gra
ph
y
(
CB
CT
)
i
s
a
r
ad
i
og
r
ap
hi
c
tec
h
ni
qu
e
t
h
at
ha
s
be
en
c
o
mm
o
nl
y
us
ed
i
n
v
ario
us
ap
pl
i
c
at
i
on
s
[1
-
3]
.
CB
CT
prov
i
de
s
i
nf
ormat
i
o
n
i
n
th
e
form
of
three
-
d
i
me
ns
i
o
na
l
(
3D)
i
m
ag
es
.
T
h
i
s
i
s
a
n
ad
v
a
nta
g
e
ov
er
two
-
di
m
en
s
i
on
a
l
(
2
D)
pa
no
r
am
i
c
.
T
he
3D
i
m
ag
e
prov
i
de
s
m
ore
de
t
ai
l
e
d
i
nfo
r
m
ati
on
t
h
at
c
an
be
us
e
d
for
furth
er
an
a
l
y
s
i
s
an
d
ex
am
i
na
t
i
on
s
.
T
he
r
ef
ore,
CB
CT
ha
s
be
en
w
i
d
e
l
y
us
ed
f
or
ex
am
i
n
ati
on
r
ath
er
t
ha
n
pa
no
r
am
i
c
t
ee
t
h
[4
-
1
0]
.
CB
CT
c
an
be
us
e
d
t
o
he
l
p
orth
od
o
nti
c
s
urger
y
[
11
]
,
b
y
s
eg
me
nti
ng
th
e
bo
n
e
or
tee
th
.
T
ee
th
s
e
gm
en
ta
ti
o
n
i
s
the
mo
s
t
i
mp
ortant
pa
r
t
of
t
he
proc
ed
ure
an
d
i
t
i
s
ai
de
d
b
y
the
c
o
mp
ut
er.
T
he
r
es
u
l
ts
of
th
e
c
om
p
ute
r
-
ai
de
d
s
eg
m
en
t
ati
on
c
an
prov
i
de
fea
t
ure
i
nf
ormati
on
ab
ou
t
the
di
ffe
r
e
nc
es
be
twe
en
de
nta
l
t
i
s
s
ue
s
an
d
ot
he
r
s
.
T
hi
s
c
an
be
us
ed
i
n
th
e
ap
pl
i
c
ati
o
n
of
de
nta
l
di
a
gn
os
i
s
,
h
um
an
i
de
n
ti
f
i
c
a
ti
on
,
de
n
tal
c
are
,
an
d
s
o
o
n.
Howev
er,
C
B
CT
i
ma
ge
us
ua
l
l
y
ha
s
l
o
w
c
on
tr
as
t,
bl
urr
ed
an
d
i
r
r
eg
ul
ar
to
oth
b
order
s
[12
-
1
4]
.
T
he
three
-
d
i
m
en
s
i
on
a
l
(
3D
)
CB
CT
i
ma
ge
c
an
be
s
l
i
c
e
d
i
nto
s
ev
eral
two
-
d
i
me
ns
i
o
na
l
(
2D)
i
ma
ge
s
.
E
ac
h
2
D
s
l
i
c
e
h
as
i
n
format
i
on
th
at
r
el
ate
d
to
the
n
ei
g
hb
or
i
ng
s
l
i
c
es
.
H
owev
er,
th
e
t
ee
th
top
o
l
og
y
i
n
e
ac
h
s
l
i
c
e
i
s
us
ua
l
l
y
d
i
ffe
r
en
t.
T
ho
s
e p
r
o
bl
e
ms
prov
i
de
c
h
al
l
en
g
es
fo
r
t
he
t
ee
th
s
eg
m
en
tat
i
o
n o
n
CB
CT
i
ma
g
e
[1
5
-
17]
.
Res
ea
r
c
h
on
CB
CT
i
ma
g
e
s
eg
me
n
tat
i
on
m
et
h
od
s
ha
s
be
en
c
arr
i
ed
o
ut.
B
r
oa
d
l
y
s
pe
ak
i
n
g,
the
s
eg
me
nta
t
i
on
me
t
ho
d
c
a
n
be
c
l
as
s
i
fi
e
d
i
nt
o
3
c
l
as
s
es
na
me
l
y
ma
nu
al
,
s
em
i
-
au
tom
ati
c
a
nd
a
uto
ma
ti
c
s
eg
me
n
tat
i
on
me
t
ho
d
[1
8]
.
In
ge
n
eral
,
t
he
au
t
om
ati
c
s
eg
me
n
tat
i
on
m
eth
od
s
c
a
n
b
e
c
l
as
s
i
fi
ed
furth
er
i
n
t
o
s
ev
eral
c
l
as
s
es
,
wh
i
c
h
are
e
dg
e
-
b
as
ed
,
thres
ho
l
d,
hy
brid
,
a
nd
s
o
on
.
W
a
ng
,
et
al
.
[
19
]
c
o
n
du
c
ted
r
es
ea
r
c
h
ab
o
ut
s
e
gm
en
tat
i
on
on
CB
CT
i
ma
g
e
us
i
n
g
an
o
pti
ma
l
t
hres
ho
l
d.
T
he
op
t
i
ma
l
thres
ho
l
d
i
s
ob
t
ai
ne
d
fr
o
m
t
he
i
n
forma
ti
o
n
of
f
i
r
s
t
gr
ay
s
c
al
e
s
l
i
c
e
an
d
the
m
ergi
ng
of
grad
i
e
nt
v
a
l
u
e
ac
r
os
s
th
e
s
l
i
c
es
.
Naum
ov
i
c
h,
et
al
.
[2
0]
s
eg
me
nt
th
e
te
eth
an
d
j
aws
o
n
th
e
C
B
CT
i
ma
ge
by
us
i
n
g
a
wate
r
s
h
ed
tr
an
s
format
i
o
n.
T
he
r
es
ea
r
c
h
c
ut
CB
CT
i
nto
s
ev
era
l
s
l
i
c
es
b
efo
r
e
s
e
gm
e
nta
t
i
on
proc
es
s
i
s
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
MNIK
A
IS
S
N: 1
69
3
-
6
93
0
◼
A
uto
ma
t
i
c
i
ma
g
e s
l
i
c
e m
ark
i
ng
prop
ag
at
i
o
n o
n
s
eg
m
en
tat
i
on
…
(
A
gu
s
Za
i
n
al
A
r
i
f
i
n
)
3219
c
arr
i
ed
o
ut.
How
ev
er,
t
he
r
es
e
arc
h
i
mp
os
es
t
h
e
s
am
e
tr
ea
tme
nt
i
n
e
ac
h
s
l
i
c
e
fo
r
the
s
eg
me
n
tat
i
on
proc
es
s
,
s
o
tha
t
i
t
do
es
no
t
pa
y
att
e
nti
o
n
to
t
he
i
nf
ormat
i
on
c
o
n
tai
n
ed
i
n
e
ac
h
s
l
i
c
e.
In
ad
di
t
i
o
n,
us
i
n
g
s
e
mi
-
au
tom
ati
c
i
ma
g
e
s
e
gm
e
nta
t
i
on
m
eth
o
d
be
c
o
me
s
e
x
ha
us
ti
v
e
an
d
i
ne
f
fi
c
i
en
t
b
ec
au
s
e
th
e
CB
CT
i
ma
ge
pro
du
c
es
a
l
ot
o
f
s
l
i
c
es
an
d
t
h
e
m
eth
od
n
e
ed
s
the
us
er
t
o
ma
r
k
th
e
tee
t
h o
n e
ac
h
s
l
i
c
e
to
prov
i
de
i
nfo
r
m
ati
on
f
or the
s
eg
me
n
tat
i
on
al
go
r
i
thm
.
In
thi
s
r
es
ea
r
c
h,
w
e
pro
po
s
e
an
a
uto
ma
t
i
c
i
m
ag
e
s
l
i
c
e
m
ark
i
ng
pro
pa
ga
t
i
on
o
n
s
eg
me
n
tat
i
on
of
de
nta
l
C
B
CT
.
Ma
r
k
er
fr
o
m
the
s
e
gm
en
tat
i
on
r
es
u
l
t
of
t
he
f
i
r
s
t
s
l
i
c
e
wi
l
l
be
propa
ga
te
d
as
the
i
n
form
a
ti
on
for
t
he
s
eg
m
en
t
ati
on
of
n
ex
t
s
l
i
c
es
,
the
r
efo
r
e
th
e
r
el
ati
on
s
h
i
p
be
twe
en
e
ac
h
s
l
i
c
e
i
s
no
t
i
gn
ored.
T
h
e
prop
os
ed
me
th
od
p
erfor
ms
s
em
i
-
au
t
om
at
i
c
i
m
ag
e
s
eg
me
n
tat
i
on
on
s
ev
er
al
s
l
i
c
es
,
i
n
w
hi
c
h
the
us
er
ne
e
ds
to
ma
r
k
the
s
eg
m
en
t
a
ti
on
ob
j
ec
t
t
o
prov
i
de
th
e
i
nfo
r
m
at
i
on
ne
e
de
d
for
t
he
s
e
gm
e
nta
t
i
o
n
a
l
go
r
i
th
m
an
d
us
e
the
s
eg
m
e
nta
t
i
on
r
es
ul
t
to
s
eg
me
nt
ot
he
r
s
l
i
c
es
au
tom
ati
c
a
l
l
y
.
Us
i
ng
th
e
propos
e
d
me
t
ho
d
,
the
CB
CT
i
m
ag
e
s
eg
me
n
tat
i
on
proc
es
s
c
an
be
m
ore
eff
ec
ti
v
e,
be
c
au
s
e
i
t
i
nte
grates
i
nfo
r
m
ati
on
fr
o
m
the
us
er
for
the
s
eg
m
en
t
ati
on
proc
es
s
,
an
d
be
c
o
me
s
mo
r
e
eff
i
c
i
en
t
be
c
au
s
e
i
t
on
l
y
ne
ed
s
to
ma
r
k
s
ev
eral
s
l
i
c
es
.
2.
Re
se
a
r
ch
Me
t
h
o
d
In
thi
s
r
es
ea
r
c
h,
w
e
pro
po
s
e
a
s
tr
ate
gy
for
s
eg
me
nt
i
ng
tee
t
h
on
C
B
CT
i
ma
ge
s
au
to
ma
t
i
c
al
l
y
ac
c
ordi
ng
t
o
F
i
gu
r
e
1.
E
v
ery
s
l
i
c
e
i
n
CB
CT
i
m
ag
es
w
i
l
l
be
i
ns
erte
d
s
tarti
n
g
fr
o
m
the
fi
r
s
t
t
o
ni
ne
ty
pi
ec
es
.
T
he
i
n
pu
tt
ed
i
m
ag
e
w
i
l
l
be
s
pl
i
t
i
nt
o
r
e
gi
o
ns
,
an
d
th
e
m
ark
i
ng
proc
es
s
i
s
th
e
proc
es
s
of
l
ab
el
i
ng
th
e regi
on
s
as
an
o
bj
ec
t o
r
b
ac
k
ground
.
F
i
gu
r
e
1.
T
h
e a
l
go
r
i
th
m o
f
t
he
pro
po
s
ed
me
t
ho
d
Evaluation Warning : The document was created with Spire.PDF for Python.
◼
IS
S
N: 16
93
-
6
93
0
T
E
L
KO
MNIK
A
V
ol
.
1
7
,
No
.
6
,
D
ec
em
b
er
20
19
:
32
1
8
-
3225
3220
T
he
ma
r
k
i
ng
proc
es
s
c
on
s
i
s
ts
of
ba
c
k
groun
d
thres
ho
l
d,
mo
r
ph
o
l
og
i
c
a
l
proc
e
s
s
,
an
d
the
prop
erty
r
eg
i
on
s
of
mo
r
ph
ol
og
i
c
al
r
es
ul
ts
.
T
he
m
orpho
l
og
y
proc
es
s
w
i
l
l
be
c
arr
i
ed
ou
t
pe
r
group
where
o
ne
group
c
on
s
i
s
t
s
of
10
CB
CT
i
ma
ge
s
.
S
o
th
at
on
e
gro
up
o
nl
y
do
es
the
m
orpho
l
o
gi
c
a
l
proc
es
s
on
c
e
the
n
th
e
r
es
ul
t
of
th
e
r
eg
i
o
n
prop
erty
wi
l
l
be
us
ed
on
t
he
n
ex
t
ni
n
e
i
ma
g
es
.
T
he
r
eg
i
on
pr
op
erti
es
ob
tai
ne
d
wi
l
l
be
us
ed
as
t
he
ob
j
ec
t
m
ark
er.
A
f
ter
tha
t
,
r
eg
i
on
me
r
gi
ng
w
i
l
l
be
do
n
e
to
gr
ou
p
r
eg
i
o
ns
tha
t
h
av
e
no
t
be
en
l
a
be
l
ed
as
o
bj
ec
ts
or
ba
c
k
ground
s
.
T
he
r
es
ul
t
of
r
eg
i
o
n
me
r
g
i
n
g
i
s
the
i
m
ag
e
of
tee
t
h
tha
t
ha
v
e
be
e
n
s
eg
m
en
t
ed
.
F
or
thi
s
r
es
ea
r
c
h,
we u
s
ed
Ma
t
l
a
b s
oft
w
are i
n
th
e
proc
es
s
.
2.1
. D
ataset
T
he
da
t
a
us
ed
i
n
thi
s
r
es
e
a
r
c
h
i
s
Dent
a
l
CB
CT
(
Co
n
e
-
B
ea
m
Com
pu
t
ed
T
om
og
r
ap
hy
)
da
ta
tak
e
n
fr
o
m
t
he
s
c
an
s
of
th
e
hu
m
an
j
a
w.
T
h
i
s
d
at
a
was
ob
t
ai
n
ed
fr
om
th
e
D
en
ta
l
an
d
O
r
al
Hos
pi
t
al
,
A
i
r
l
an
gg
a
Un
i
v
ers
i
ty
(
RS
G
M
UNA
IR)
.
E
v
ery
da
ta
ob
t
ai
n
ed
ha
s
a
grou
n
d
tr
uth
tha
t
ha
s
be
en
c
on
f
i
r
me
d
by
r
a
d
i
o
l
o
gi
c
al
ex
p
erts
s
o
t
ha
t
th
e
a
c
c
urac
y
of
th
e
pro
po
s
e
d
me
th
od
c
an
be
c
al
c
ul
a
ted
.
T
h
i
s
d
ata
s
et
i
s
3D
d
ata
,
wh
i
c
h
i
s
t
he
n
s
l
i
c
ed
ac
c
ordi
n
g
t
o
t
he
ax
i
a
l
pl
an
e
an
d
r
es
u
l
t
i
n
2D
i
ma
g
es
as
s
ho
wn
i
n
F
i
gu
r
e
2.
T
h
e
r
es
ul
t
o
f
the
s
l
i
c
i
n
g
proc
es
s
i
s
20
0
i
m
ag
es
wi
th
th
e
s
i
z
e
of
26
6x
26
6
pi
x
e
l
s
e
ac
h
[2
1]
.
F
i
gu
r
e
2.
D
en
ta
l
C
B
CT
2.2
. R
egio
n
S
p
l
it
t
ing
T
he
fi
r
s
t
s
tag
e
i
n
th
i
s
s
eg
m
en
tat
i
o
n
proc
es
s
i
s
th
e
r
eg
i
on
s
pl
i
tt
i
ng
where
CB
CT
i
ma
ge
s
are
s
ep
arat
ed
i
nto
s
ev
era
l
r
eg
i
on
s
.
T
he
al
g
orit
hm
th
at
r
es
ea
r
c
he
r
s
us
e
to
d
o
th
i
s
r
eg
i
on
s
pl
i
tt
i
ng
i
s
me
an
-
s
h
i
ft
c
l
us
ter
i
ng
[2
2]
.
T
hi
s
al
go
r
i
th
m
di
v
i
de
s
CB
CT
i
ma
g
es
i
nto
c
l
us
ter
r
eg
i
on
s
ac
c
ordi
ng
to
c
o
l
or
s
i
mi
l
ar
i
ty
.
T
h
e
me
an
-
s
hi
ft
th
at
the
r
es
ea
r
c
he
r
s
us
ed
w
as
a
m
ea
n
-
s
h
i
ft
by
the
E
di
s
o
n
S
y
s
tem
[
23
]
.
The
me
an
-
s
hi
f
t u
s
ed
ha
s
s
ev
era
l
i
np
ut
p
a
r
am
ete
r
s
i
n t
he
form
of
s
p
at
i
al
ba
nd
w
i
dt
h,
ba
nd
wi
d
th
r
an
ge
,
a
nd
m
i
n
i
mu
m
ar
ea
r
eg
i
on
.
T
he
v
a
l
ue
s
of
ea
c
h
pa
r
a
me
ter
ar
e
7,
1
0
a
nd
1
1.
T
he
r
es
u
l
ts
o
f
t
he
s
p
l
i
tti
ng
r
eg
i
on
wi
th
the
me
an
-
s
hi
ft
a
l
go
r
i
th
m
ar
e
i
ma
ge
s
t
h
at
h
av
e
be
e
n
di
v
i
d
ed
i
nt
o c
l
us
ter r
e
gi
on
s
,
whi
c
h c
a
n b
e s
ee
n
i
n
Fig
ur
e 3
.
F
i
gu
r
e
3.
R
eg
i
on
s
p
l
i
t
ti
ng
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
MNIK
A
IS
S
N: 1
69
3
-
6
93
0
◼
A
uto
ma
t
i
c
i
ma
g
e s
l
i
c
e m
ark
i
ng
prop
ag
at
i
o
n o
n
s
eg
m
en
tat
i
on
…
(
A
gu
s
Za
i
n
al
A
r
i
f
i
n
)
3221
2.3
.
Ma
r
king
P
r
o
ce
ss
A
n
i
ma
ge
c
o
ns
i
s
ts
of
ob
j
ec
ts
an
d
no
n
-
ob
j
ec
ts
(
ba
c
k
ground
)
.
In
th
e
s
eg
me
nta
t
i
on
proc
es
s
,
i
t
i
s
n
ec
es
s
ary
to
k
no
w
t
he
o
bj
ec
t's
bo
un
da
r
i
es
to
s
ep
arate
ob
j
ec
ts
fr
om
the
ba
c
k
gro
un
d.
In
th
e
r
eg
i
on
s
p
l
i
tt
i
n
g
proc
es
s
,
CB
C
T
i
ma
g
es
are
d
i
v
i
d
ed
i
nto
s
ev
eral
c
l
us
ter
r
eg
i
o
ns
.
T
he
c
l
us
ter
w
i
l
l
l
at
er
be
l
ab
el
ed
as
an
ob
j
ec
t
or
ba
c
k
grou
nd
.
T
h
i
s
m
ark
i
n
g
proc
es
s
ai
ms
to
m
ark
the
s
e
c
l
us
ters
as
ob
j
ec
ts
or
ba
c
k
groun
ds
.
T
he
r
efo
r
e,
th
e
r
es
e
arc
he
r
c
on
c
l
ud
es
th
at
i
ma
ge
(
I
)
of
th
e
r
es
ul
ts
of
thi
s
ma
r
k
i
ng
proc
es
s
wi
l
l
c
on
s
i
s
t
of
th
e
ob
j
ec
t
(
O
)
,
b
ac
k
ground
(
B
)
,
an
d
no
n
-
m
ark
ed
c
l
us
ter (
C
)
ac
c
ordi
n
g
to
(
1
)
.
=
{
,
,
}
(
1)
T
o
ma
r
k
the
too
th
as
o
bj
e
c
ts
,
the
i
ni
t
i
a
l
i
ma
g
e
i
s
us
e
d
be
f
ore
th
e
r
eg
i
on
s
p
l
i
tti
n
g.
T
hi
s
ma
r
k
i
ng
proc
es
s
i
s
c
arr
i
ed
ou
t
ev
ery
10
s
l
i
c
es
of
da
t
a
s
tarti
ng
da
ta
pi
ec
es
fr
om
the
10
th
,
20
th
,
an
d
u
p
to
t
he
9
0
th
s
l
i
c
e
i
n
to
s
pe
ed
up
t
he
s
eg
me
nt
ati
o
n
proc
es
s
wi
t
h
l
ar
ge
a
mo
un
ts
of
da
ta
s
l
i
c
es
.
T
he
ma
r
k
i
ng
proc
es
s
of
the
10
th
s
l
i
c
e
wi
l
l
be
u
s
ed
as
a
ma
r
k
er
for
the
ne
x
t
9
s
l
i
c
es
.
F
or
the
20
th
s
l
i
c
e,
t
he
r
e
-
ma
r
k
i
ng
pr
oc
es
s
wi
l
l
b
e
c
arr
i
ed
ou
t,
an
d
the
ne
x
t
9
s
l
i
c
es
w
i
l
l
us
e
the
m
ark
i
ng
i
n
forma
ti
o
n f
r
o
m t
he
20
th
s
l
i
c
e.
T
he
fi
r
s
t
s
ta
ge
of
th
e
m
ark
i
ng
proc
es
s
i
s
t
he
ba
c
k
grou
nd
thres
h
ol
d.
T
h
e
thres
h
ol
d
v
al
ue
us
ed
i
s
the
av
erag
e
thr
es
h
ol
d
v
a
l
u
e
of
a
l
l
CB
CT
i
m
a
ge
e
ntri
es
wi
t
h
th
e
a
l
go
r
i
th
m
[1
8]
s
o
t
ha
t
the
gray
c
o
l
or
be
l
ow
th
e
t
hres
ho
l
d
wi
l
l
be
c
on
s
i
de
r
e
d
as
ba
c
k
grou
nd
a
nd
om
i
tte
d
to
0.
F
or
the
v
al
ue
a
bo
v
e
t
he
thr
es
ho
l
d,
th
e
gray
c
ol
or
i
s
c
ha
ng
ed
to
1.
T
he
r
es
u
l
t
of
t
he
ba
c
k
gro
un
d
thres
ho
l
d
i
s
the
n
i
ns
ert
ed
i
nto
th
e
fi
l
ter
proc
es
s
to
r
em
ov
e
ba
c
k
grou
nd
no
i
s
e
tha
t
i
s
s
ti
l
l
r
em
ai
ni
ng
.
T
he
r
e
ma
i
n
i
ng
ba
c
k
groun
d
n
oi
s
e
l
i
k
e
s
ma
l
l
do
ts
s
ti
l
l
r
e
ma
i
ns
i
n
the
i
ma
g
e.
T
o remo
v
e
th
e n
oi
s
e,
me
di
an
f
i
l
t
erin
g i
s
pe
r
for
me
d
th
r
ee
t
i
me
s
.
In
F
i
gu
r
e 4
,
th
e
l
e
ft i
ma
g
e s
ho
ws
the
i
m
ag
e
be
f
or
e
the
m
ed
i
an
fi
l
teri
ng
proc
es
s
was
do
ne
,
an
d
the
r
i
g
ht
i
ma
g
e
s
ho
ws
the
r
es
ul
t
fr
om
th
e m
ed
i
an
f
i
l
teri
ng
.
F
i
gu
r
e
4.
T
h
e res
ul
t o
f
the
me
d
i
an
fi
l
teri
ng
T
he
n
ex
t
s
tep
i
s
t
he
mo
r
p
ho
l
og
i
c
a
l
proc
es
s
.
Mo
r
ph
o
l
og
y
us
ed
i
s
a
c
i
r
c
l
e.
T
h
e
c
i
r
c
l
e
mo
de
l
i
s
us
ed
be
c
au
s
e
t
h
e
s
ha
pe
of
hu
ma
n
te
eth
i
n
C
B
CT
i
m
ag
es
i
s
as
s
u
m
ed
t
o
be
l
i
k
e
a
c
i
r
c
l
e
.
T
h
e
r
es
u
l
ts
o
f
t
h
e
m
orpho
l
o
gy
are
tak
e
n
f
r
om
t
he
prop
erty
r
eg
i
on
of
e
ac
h
c
i
r
c
l
e
ob
ta
i
ne
d.
Reg
i
o
n
pr
op
erty
ob
ta
i
ne
d
i
s
a
c
en
tr
oi
d
,
ma
j
or
ax
i
s
l
e
ng
th
,
mi
n
or
ax
i
s
l
en
gt
h.
T
o
f
i
nd
the
d
i
a
me
ter
D
of
a c
i
r
c
l
e
, t
he
r
es
ea
r
c
he
r
us
es
th
e
fo
l
l
owi
n
g
(
2
)
.
D
=
(
major
ax
i
s
l
eng
th
2
⁄
)
×
(
mi
no
r
ax
i
s
l
eng
th
2
⁄
)
2
(
2)
T
he
ma
j
or
ax
i
s
l
en
g
th
a
n
d
the
m
i
n
or
ax
i
s
l
e
ng
th
a
r
e
v
al
ue
s
of
t
he
pro
pe
r
ty
r
eg
i
o
n.
T
he
v
a
l
ue
o
f
t
he
ma
j
or
ax
i
s
l
en
gth
a
nd
mi
no
r
ax
i
s
l
e
ng
th
i
s
di
v
i
de
d
by
2
to
s
hr
i
nk
t
he
ob
j
ec
t
ma
r
k
ers
s
o
tha
t
the
ba
c
k
gr
ou
nd
i
s
n
ot
m
ark
ed
.
T
h
e
r
e
s
ul
ts
of
t
he
ma
r
k
i
n
g
proc
es
s
c
an
be
s
e
en
i
n
F
i
gu
r
e
5
.
I
n
t
he
r
e
s
ul
ts
of
the
pro
pe
r
ty
r
e
gi
o
ns
,
t
he
r
es
e
arc
he
r
g
ets
c
i
r
c
l
es
tha
t
i
nd
i
c
ate
the
po
s
i
t
i
on
of
th
e
o
bj
ec
t
l
i
k
e
F
i
g
ure
5
(
b)
.
T
h
e
c
i
r
c
l
es
are
t
he
en
t
i
r
e
c
o
ordi
na
te
o
f
the
ob
j
ec
t
s
o
the
r
es
ea
r
c
h
er
tak
es
s
ev
eral
c
oo
r
d
i
n
ate
s
(
x
,
y
)
by
dra
wi
ng
a
s
tr
ai
gh
t
l
i
ne
i
n
th
e
mi
d
dl
e
of
e
ac
h
c
i
r
c
l
e
,
s
o
tha
t
a
l
l
t
he
c
oo
r
di
na
t
es
of
the
ob
j
ec
t
are
ob
ta
i
ne
d.
F
or
the
b
ac
k
ground
ma
r
k
i
n
g
proc
es
s
i
s
do
ne
by
tak
i
ng
a
nu
m
be
r
of
c
oo
r
di
n
ate
s
t
ha
t
are
f
ar
ou
ts
i
de
e
ac
h
c
i
r
c
l
e
.
S
o
tha
t
th
ere
Evaluation Warning : The document was created with Spire.PDF for Python.
◼
IS
S
N: 16
93
-
6
93
0
T
E
L
KO
MNIK
A
V
ol
.
1
7
,
No
.
6
,
D
ec
em
b
er
20
19
:
32
1
8
-
3225
3222
i
s
a
c
l
us
ter
r
e
gi
on
tha
t
h
as
be
e
n
l
a
be
l
ed
as
o
bj
ec
t
O
c
on
s
i
s
t
of
N
c
l
us
ter
o
bj
ec
t
O
i
l
i
k
e
(
3
)
a
nd
ba
c
k
groun
d
B
c
o
ns
i
s
t o
f
M
c
l
us
ter bac
k
grou
nd
B
i
l
i
k
e
(
4
)
.
O
=
{
O
i
}
i
=
1
,
…
,
N
(
3)
B
=
{
B
i
}
i
=
1
,
…
,
M
(
4)
(
a)
(
b)
F
i
gu
r
e
5.
(
a
)
R
es
ul
t
of
b
ac
k
gro
un
d t
hres
h
ol
d
an
d m
ed
i
an
f
i
l
t
er, an
d
(
b
)
i
ts
m
ark
er
2.4
. R
egio
n
M
er
g
ing
In
the
r
eg
i
on
me
r
g
i
ng
ph
as
e,
the
hi
erar
c
hi
c
a
l
c
l
us
t
erin
g
me
t
ho
d
we
us
e
was
m
o
di
f
i
ed
,
whi
c
h
was
m
od
i
fi
ed
at
th
e
di
s
ta
nc
e
m
ea
s
ure
me
n
t
[1
8]
.
D
i
s
tan
c
e
me
as
ureme
nt
i
n
hi
erar
c
hi
c
a
l
c
l
us
teri
n
g
i
s
mo
di
f
i
e
d
by
ob
s
erv
i
ng
i
nte
r
-
c
l
as
s
an
d
i
ntra
-
c
l
as
s
di
s
tan
c
es
.
T
hi
s
mo
d
i
f
i
ed
hi
erar
c
h
i
c
al
c
l
us
teri
ng
me
tho
d
r
eq
u
i
r
es
s
ev
era
l
i
np
ut
p
arame
ters
.
T
he
s
e
p
a
r
am
ete
r
s
are
the
l
oc
a
ti
o
n
of
th
e
ob
j
ec
t
pi
x
el
s
,
wh
i
c
h
i
s
ma
r
k
e
d
i
n
the
prev
i
o
us
proc
es
s
,
l
oc
ati
o
n
of
the
ba
c
k
groun
d
pi
x
e
l
s
,
i
ni
t
i
al
i
ma
g
e,
an
d
th
e
r
es
u
l
t
of
r
eg
i
o
n
s
p
l
i
tt
i
n
g
proc
es
s
.
T
he
proc
es
s
c
arr
i
ed
o
ut
t
o
t
he
no
n
-
l
a
be
l
ed
r
e
gi
on
s
.
T
h
es
e
n
o
n
-
l
ab
el
ed
r
e
gi
on
s
w
i
l
l
b
e
c
ou
nt
ed
w
i
th
the
r
eg
i
o
n
tha
t
ha
s
b
ee
n
l
a
be
l
ed
fr
o
m
th
e
r
es
ul
ts
o
f
th
e
ma
r
k
i
n
g
proc
es
s
.
Re
gi
on
s
tha
t
are
no
t
l
ab
el
e
d
w
i
l
l
b
e
c
al
c
u
l
at
ed
i
nte
r
-
c
l
as
s
an
d
i
ntra
-
c
l
as
s
wi
th
the
r
eg
i
on
of
th
e
c
l
us
ter
ob
j
ec
t
a
nd
ba
c
k
groun
d.
T
he
r
es
u
l
ts
of
c
al
c
u
l
at
i
ng
t
he
d
i
s
ta
nc
e
be
tw
ee
n
ob
j
ec
ts
an
d
b
ac
k
ground
w
i
l
l
be
c
om
pa
r
ed
.
T
he
l
owes
t
di
s
tan
c
e
,
whe
the
r
i
t's
an
ob
j
ec
t
or
ba
c
k
grou
nd
w
i
l
l
be
l
ab
e
l
ed
as
the
c
l
us
ter.
S
o,
th
e
r
es
u
l
t
s
of
t
he
r
e
gi
on
s
p
l
i
tti
ng
wi
l
l
me
r
ge
i
nto
c
l
us
ters
t
ha
t
ha
v
e
be
en
l
ab
el
e
d
an
d
prod
uc
e
a
s
e
gm
en
ted
i
ma
ge
.
F
i
gu
r
e
6
i
s
the
r
es
ul
t
of
r
eg
i
on
m
ergi
n
g,
wh
i
c
h
i
s
a s
eg
m
en
te
d
i
ma
ge
.
(
a)
(
b)
F
i
gu
r
e
6.
(
a
)
G
r
ou
nd
tru
th
i
ma
ge
s
1
st
–
4
th
s
l
i
c
es
an
d
(
b
)
s
eg
me
nta
t
i
o
n res
ul
t 1
st
–
4
th
s
l
i
c
es
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
MNIK
A
IS
S
N: 1
69
3
-
6
93
0
◼
A
uto
ma
t
i
c
i
ma
g
e s
l
i
c
e m
ark
i
ng
prop
ag
at
i
o
n o
n
s
eg
m
en
tat
i
on
…
(
A
gu
s
Za
i
n
al
A
r
i
f
i
n
)
3223
3.
Re
sult
a
nd
Di
sc
u
ss
ion
T
he
r
es
ul
ts
of
t
he
au
tom
ati
c
too
th
s
eg
m
en
t
ati
on
r
es
ea
r
c
h
on
C
B
CT
da
t
a
wi
l
l
be
c
om
pa
r
e
d
w
i
th
th
e
ex
i
s
t
i
n
g
au
t
om
a
ti
c
me
th
od
s
,
na
me
l
y
O
T
S
U
[
24
]
an
d
HC
A
[25]
.
T
o
ev
al
ua
te
thi
s
prop
os
ed
m
eth
o
d,
r
es
ea
r
c
he
r
s
us
ed
Mi
s
c
l
as
s
i
fi
c
ati
o
n
E
r
r
or
(
ME
)
an
d
Re
l
at
i
v
e
F
oregr
ou
n
d
A
r
ea
E
r
r
or
(
R
A
E
)
.
T
h
e
ME
v
al
u
e
me
as
ures
th
e
ob
j
ec
t
an
d
ba
c
k
grou
nd
pi
x
e
l
s
t
h
at
ar
e
wr
o
ng
l
y
c
l
as
s
i
fi
e
d.
RA
E
s
erv
es
to
me
as
ure
th
e
di
f
ferenc
e
i
n
area
be
twee
n
s
eg
m
en
t
ed
ob
j
ec
ts
an
d
groun
d t
r
uth
.
ME
h
as
a
c
al
c
ul
at
i
o
n e
q
ua
t
i
o
n l
i
k
e
(
5
)
.
ME
=
1
−
|
O
g
∩
O
r
|
+
|
B
g
∩
B
r
|
|
O
g
∪
B
g
|
(
5)
T
he
c
al
c
u
l
at
i
o
n
proc
es
s
of
Mi
s
c
l
as
s
i
f
i
c
at
i
on
err
or
(
M
E
)
us
es
th
e
to
tal
nu
mb
er
of
err
or
proba
bi
l
i
t
i
es
(
1)
m
i
nu
s
th
e
nu
mb
er
of
tr
uth
s
t
ha
t
wer
e
s
uc
c
es
s
ful
l
y
pe
r
for
me
d.
T
he
proc
es
s
of
c
al
c
ul
a
ti
n
g
t
he
am
o
un
t
of
tr
uth
i
s
d
on
e
by
d
i
v
i
di
n
g
the
nu
mb
er
of
pi
x
e
l
s
th
a
t
a
r
e
c
orr
ec
t
l
y
c
l
as
s
i
fi
e
d
d
i
v
i
de
d
by
th
e
t
ot
al
pi
x
e
l
i
ma
ge
.
T
he
tr
uth
of
the
tot
al
pi
x
e
l
i
s
c
al
c
u
l
at
e
d
by
s
u
mm
i
ng
the
tot
al
s
i
mi
l
arit
y
p
i
x
el
s
of
the
grou
nd
tr
uth
(
O
g)
ob
j
e
c
t
an
d
the
t
ota
l
pi
x
e
l
s
eg
m
en
tat
i
o
n
r
es
ul
t
(
O
r
)
an
d
the
tot
a
l
pi
x
e
l
s
i
m
i
l
arit
y
to
the
b
ac
k
ground
i
m
a
ge
(
B
g)
an
d
the
to
tal
pi
x
e
l
ba
c
k
groun
d
of
the
s
e
gm
en
ta
ti
o
n
(
B
r
)
.
T
h
e
s
ma
l
l
er
M
E
r
es
ul
ts
,
th
e
be
tte
r
s
eg
m
en
t
ati
on
r
es
ul
t
s
.
A
c
a
l
c
u
l
at
i
on
formu
l
a f
or RA
E
i
s
ex
pl
ai
n
e
d i
n
(
6).
R
A
E
=
{
A
g
−
A
r
A
g
if
A
r
<
A
g
A
r
−
A
g
A
r
if
A
r
≥
A
g
(
6)
W
he
r
e
A
g
i
s
an
area
of
the
i
m
ag
e
of
grou
nd
tr
uth
,
A
r
i
s
an
are
a
o
f
i
ma
ge
s
e
gm
e
nta
t
i
on
r
es
ul
ts
.
T
h
e
s
m
al
l
er
the
R
A
E
v
al
ue
,
th
e
mo
r
e
s
i
m
i
l
ar
the
r
es
u
l
ts
o
f
s
eg
me
nta
t
i
o
n
w
i
th
grou
nd
tr
uth
s
o
th
at
i
t
prod
uc
es
g
o
od
s
e
gm
e
nta
t
i
o
n
r
es
u
l
ts
.
I
n
T
ab
l
e
1
s
h
ows
t
he
r
es
ul
ts
of
the
tr
i
al
of
the
pro
po
s
ed
me
t
ho
d
wi
t
h
O
T
S
U
an
d
HC
A
.
T
he
r
es
ul
ts
ob
ta
i
ne
d
s
tat
e
d
th
at
th
e
ME
a
nd
MA
E
me
an
v
a
l
ue
s
of
the
pro
po
s
ed
me
t
ho
d
wer
e
0.
11
2
a
nd
0.4
78
.
T
he
O
T
S
U
m
eth
od
ob
t
ai
ne
d
M
E
an
d
M
A
E
av
erag
es
of
0.4
1
0
an
d
0.6
5
7
an
d
the
HCA
me
th
od
av
era
ge
d
ME
an
d
MA
E
of
0.
39
8
an
d
0.
71
9
.
It
c
an
b
e
c
on
c
l
ud
e
d
tha
t
t
he
pr
op
os
e
d
me
tho
d
ha
s
a
be
tt
er
l
ev
el
of
ac
c
urac
y
c
om
pa
r
e
d t
o
au
t
om
a
ti
c
s
eg
me
nt
ati
on
me
th
od
s
s
uc
h a
s
O
T
S
U a
nd
HC
A
.
T
he
c
om
p
ared
me
th
od
s
pe
r
form
i
ma
ge
s
eg
me
nta
t
i
on
us
i
ng
thres
ho
l
d
v
al
ue
(
au
tom
ati
c
a
l
l
y
)
a
nd
do
no
t
pa
y
att
en
t
i
on
to
the
ne
i
gh
b
orhoo
d
v
al
ue
an
d
3D
i
nf
ormati
on
of
C
B
CT
i
ma
ge
.
T
hi
s
c
au
s
es
t
he
b
ac
k
ground
t
ha
t
ha
s
a
n
i
nt
en
s
i
ty
v
al
ue
ab
ov
e
the
t
h
r
es
ho
l
d
to
b
e
c
l
as
s
i
fi
e
d
as
an
ob
j
ec
t
an
d
the
ob
j
ec
ts
tha
t
h
as
an
i
nte
ns
i
ty
v
al
u
e
be
l
ow
th
e
t
hres
ho
l
d
to
be
c
l
as
s
i
fi
e
d
as
ba
c
k
groun
d.
T
he
err
or
c
om
pa
r
i
s
o
n
r
es
ul
t
on
T
a
bl
e
1
s
ho
ws
tha
t
the
propos
e
d
me
th
od
gi
v
es
mo
r
e
e
ffe
c
ti
v
e
s
eg
m
en
t
ati
on
r
es
ul
ts
,
p
r
ov
en
by
the
s
ma
l
l
er
err
or
v
al
u
es
,
t
ha
n
the
au
t
om
ati
c
s
eg
m
en
t
ati
o
n
me
th
od
s
be
c
au
s
e
the
pro
po
s
ed
m
eth
o
d
i
nt
eg
r
at
es
i
n
format
i
on
fr
o
m
the
us
er f
or th
e s
eg
m
en
t
ati
on
proc
es
s
.
Ma
r
k
i
ng
ob
j
ec
ts
au
t
om
at
i
c
al
l
y
us
i
ng
mo
r
ph
ol
og
y
an
d
property
r
eg
i
on
s
c
an
de
tec
t
tee
th
on
CB
CT
i
ma
ge
s
tha
t
are
s
ha
pe
d
l
i
k
e
c
i
r
c
l
es
.
How
ev
er,
as
s
ho
wn
i
n
F
i
gu
r
e
6,
the
r
eg
i
on
me
r
gi
ng
proc
es
s
c
an
n
ot
d
ete
c
t
al
l
th
e
tee
t
h
be
c
a
us
e
s
om
e
te
eth
r
eg
i
o
n
ha
s
a
mo
r
e
s
i
mi
l
ar
i
nte
ns
i
ty
t
o
t
he
ba
c
k
grou
n
d.
T
h
i
s
c
au
s
e
d
s
e
gm
e
nta
t
i
on
err
or
an
d
f
urther
an
al
y
s
i
s
ab
o
ut
mo
r
e
ac
c
urate reg
i
on
me
r
g
i
n
g m
eth
o
d c
an
be
p
erfor
m
ed
.
T
ab
l
e
1.
C
om
pa
r
i
s
o
n Re
s
u
l
ts
of
the
P
r
o
po
s
ed
Me
t
ho
d
No
C
B
C
T
D
a
t
a
P
r
o
p
o
s
e
d
Met
h
o
d
OTS
U
HCA
ME
MA
E
ME
MA
E
ME
MA
E
1
A
0
.
1
1
2
0
.
3
0
9
0
.
3
4
0
0
.
5
4
4
0
.
3
6
0
0
.
6
2
1
2
B
0
.
0
9
7
0
.
4
2
3
0
.
4
3
4
0
.
7
2
9
0
.
3
8
4
0
.
7
2
9
3
C
0
.
1
2
2
0
.
6
6
2
0
.
5
2
1
0
.
9
3
4
0
.
4
1
8
0
.
8
1
0
4
D
0
.
0
7
4
0
.
4
5
8
0
.
3
1
3
0
.
5
3
0
0
.
3
6
5
0
.
7
7
8
5
E
0
.
0
8
8
0
.
5
7
2
0
.
4
9
1
0
.
7
9
0
0
.
4
8
3
0
.
8
4
3
6
F
0
.
1
3
0
0
.
5
2
9
0
.
4
1
6
0
.
6
1
9
0
.
3
8
4
0
.
6
6
1
7
G
0
.
1
6
0
0
.
3
9
2
0
.
3
5
3
0
.
4
4
9
0
.
3
9
3
0
.
5
9
0
A
v
e
r
a
g
e
0
.
1
1
2
0
.
4
7
8
0
.
4
1
0
0
.
6
5
7
0
.
3
9
8
0
.
7
1
9
Evaluation Warning : The document was created with Spire.PDF for Python.
◼
IS
S
N: 16
93
-
6
93
0
T
E
L
KO
MNIK
A
V
ol
.
1
7
,
No
.
6
,
D
ec
em
b
er
20
19
:
32
1
8
-
3225
3224
4.
Co
n
clus
ion
In
t
hi
s
r
es
ea
r
c
h,
we
pro
po
s
e
a
s
tr
at
eg
y
i
n
s
eg
m
en
ti
ng
te
eth
au
t
om
ati
c
a
l
l
y
.
T
h
i
s
r
es
ea
r
c
h
us
ed
de
nt
al
CB
C
T
da
ta
t
ak
en
fr
om
s
ev
era
l
pa
ti
en
ts
fr
o
m
th
e
RS
G
M
U
NA
IR
ho
s
pi
tal
.
T
he
prop
os
ed
me
t
ho
d
ha
s
s
ev
eral
s
tag
es
,
w
hi
c
h
a
r
e
r
eg
i
o
n
s
pl
i
tt
i
n
g
us
i
n
g
t
he
m
ea
n
-
s
h
i
ft
al
g
orit
h
m,
ma
r
k
i
ng
us
i
n
g
mo
r
ph
o
l
og
y
an
d
prop
erty
r
eg
i
on
s
,
an
d
r
e
gi
on
me
r
gi
ng
us
i
ng
hi
erar
c
h
i
c
al
c
l
us
ter
i
ng
al
go
r
i
thm
.
T
he
r
es
ul
ts
ob
t
ai
ne
d
fr
om
th
e
pro
po
s
e
d
m
eth
od
h
av
e
l
ow
er
err
or
v
al
ue
s
t
ha
n
th
e
ex
i
s
t
i
ng
au
t
om
a
ti
c
s
eg
me
n
tat
i
o
n
me
t
ho
d,
wh
i
c
h
ar
e
th
e
ME
v
a
l
ue
wi
t
h
an
av
era
ge
of
0.1
1
2
a
nd
t
he
M
A
E
v
al
ue
wi
t
h
a
n
av
erage
of
0.4
7
8.
T
he
ex
pe
r
i
me
nta
l
r
es
u
l
ts
s
ho
w
t
ha
t
th
e
propos
e
d
m
eth
o
d
c
a
n
au
to
ma
t
i
c
al
l
y
s
e
gm
en
t
t
he
C
B
CT
da
ta
by
p
ay
i
ng
at
ten
t
i
o
n
to
i
ts
3
D
i
nfo
r
m
ati
on
.
T
h
e
eff
ec
ti
v
e
ne
s
s
of
t
he
pro
po
s
ed
me
t
ho
d
i
s
i
nd
i
c
ate
d
b
y
the
l
ow
err
or
r
ate
c
om
pa
r
ed
to
the
oth
er
me
th
od
s
.
F
urt
he
r
r
es
e
arc
h
ab
ou
t
t
he
pa
r
a
me
ters
or
al
g
orit
hm
s
for
c
on
du
c
ti
ng
r
eg
i
on
s
pl
i
tt
i
ng
an
d
r
eg
i
on
me
r
gi
n
g
proc
es
s
i
s
n
ee
d
ed
to
prod
uc
e
a
mo
r
e
eff
ec
t
i
v
e
an
d
eff
i
c
i
en
t
r
es
ul
t
.
Ref
er
en
ce
s
[1
]
Stu
d
e
b
a
k
e
r
B,
Ho
l
l
e
n
d
e
r
L
,
M
a
n
c
l
L
,
J
o
h
n
s
o
n
J
D,
Pa
r
a
n
j
p
e
A
,
Th
e
In
c
i
d
e
n
c
e
o
f
Se
c
o
n
d
M
e
s
i
o
b
u
c
c
a
l
Can
a
l
s
L
o
c
a
te
d
i
n
M
a
x
i
l
l
a
r
y
M
o
l
a
r
s
wit
h
th
e
Ai
d
o
f
Con
e
-
b
e
a
m
Com
p
u
te
d
To
m
o
g
ra
p
h
y
.
J
o
u
r
n
a
l
o
f
e
n
d
o
d
o
n
ti
c
s
.
2
0
1
8
;
4
4
(4
):
5
6
5
–
570.
[2
]
Pa
te
l
S
,
Dawo
o
d
A,
Fo
r
d
T
P,
Wh
a
i
te
s
E.
Th
e
p
o
t
e
n
t
i
a
l
a
p
p
l
i
c
a
ti
o
n
s
o
f
c
o
n
e
b
e
a
m
c
o
m
p
u
t
e
d
to
m
o
g
ra
p
h
y
i
n
th
e
m
a
n
a
g
e
m
e
n
t
o
f
e
n
d
o
d
o
n
t
i
c
p
ro
b
l
e
m
s
.
In
te
r
n
a
ti
o
n
a
l
e
n
d
o
d
o
n
t
i
c
j
o
u
rn
a
l
.
2
0
0
7
;
4
0
(1
0
):
8
1
8
–
8
3
0
.
[3
]
J
o
h
n
G
P,
J
o
y
TE,
M
a
th
e
w
J
,
Ku
m
a
r
VR.
A
p
p
l
i
c
a
t
i
o
n
s
o
f
c
o
n
e
b
e
a
m
c
o
m
p
u
te
d
to
m
o
g
r
a
p
h
y
fo
r
a
p
ro
s
t
h
o
d
o
n
ti
s
t.
T
h
e
J
o
u
rn
a
l
o
f
th
e
I
n
d
i
a
n
Pro
s
th
o
d
o
n
t
i
c
So
c
i
e
ty
.
2
0
1
6
;
1
6
(1
)
:
3
.
[4
]
L
e
e
R
J
,
P
i
S,
Pa
rk
J
,
Nel
s
o
n
G
,
Hat
c
h
e
r
D,
O
b
e
ro
i
S.
Th
re
e
-
d
i
m
e
n
s
i
o
n
a
l
e
v
a
l
u
a
ti
o
n
o
f
r
o
o
t
p
o
s
i
ti
o
n
a
t
th
e
re
s
e
t
a
p
p
o
i
n
tm
e
n
t
wi
t
h
o
u
t
r
a
d
i
o
g
ra
p
h
s
:
a
p
ro
o
f
-
of
-
c
o
n
c
e
p
t
s
tu
d
y
.
Pro
g
re
s
s
i
n
o
rth
o
d
o
n
ti
c
s
.
2
0
1
8
;
1
9
(1
)
:
1
5
.
[5
]
Ku
m
a
r
M
,
Sh
a
n
a
v
a
s
M
,
Si
d
a
p
p
a
A
,
Ki
r
a
n
M
.
Con
e
b
e
a
m
c
o
m
p
u
te
d
to
m
o
g
ra
p
h
y
-
k
n
o
w
i
t
s
s
e
c
r
e
ts
.
J
o
u
rn
a
l
o
f
i
n
te
rn
a
ti
o
n
a
l
o
ra
l
h
e
a
l
th
:
J
IOH
.
2
0
1
5
;
7
(2
):
6
4
–
6
8.
[6
]
Ke
ta
b
i
AR,
Ke
t
a
b
i
S,
Na
b
l
i
M
B,
L
a
u
e
r
HC
,
Bre
n
n
e
r
M
.
Det
e
c
ti
o
n
a
n
d
m
e
a
s
u
r
e
m
e
n
ts
o
f
a
p
i
c
a
l
l
e
s
i
o
n
s
in
th
e
u
p
p
e
r
j
a
w
b
y
c
o
n
e
b
e
a
m
c
o
m
p
u
te
d
t
o
m
o
g
ra
p
h
y
a
n
d
p
a
n
o
ra
m
i
c
ra
d
i
o
g
r
a
p
h
y
a
s
a
fu
n
c
t
i
o
n
o
f
c
o
rti
c
a
l
b
o
n
e
t
h
i
c
k
n
e
s
s
.
Cli
n
i
c
a
l
o
ra
l
i
n
v
e
s
ti
g
a
ti
o
n
s
.
2
0
1
9
;
2
2
:
1
-
7.
[7
]
Am
i
n
o
s
h
a
r
i
a
e
A,
K
u
l
i
l
d
J
C,
Sy
e
d
A.
Con
e
-
b
e
a
m
Co
m
p
u
t
e
d
To
m
o
g
ra
p
h
y
Co
m
p
a
re
d
w
i
th
In
tr
a
o
ra
l
Rad
i
o
g
r
a
p
h
i
c
L
e
s
i
o
n
s
i
n
En
d
o
d
o
n
ti
c
O
u
tc
o
m
e
St
u
d
i
e
s
:
A
Sy
s
te
m
a
ti
c
Rev
i
e
w.
J
o
u
rn
a
l
o
f
e
n
d
o
d
o
n
ti
c
s
.
2
0
1
8
;
4
4
(1
1
):
1
6
2
6
–
1
6
3
1
.
[8
]
Pa
te
l
S,
Dura
c
k
C,
Ab
e
l
l
a
F,
Sh
e
m
e
s
h
H,
Roi
g
M
,
L
e
m
b
e
r
g
K.
Con
e
b
e
a
m
c
o
m
p
u
te
d
to
m
o
g
ra
p
h
y
i
n
En
d
o
d
o
n
ti
c
s
–
a
re
v
i
e
w.
In
te
r
n
a
ti
o
n
a
l
e
n
d
o
d
o
n
ti
c
j
o
u
rn
a
l
.
2
0
1
5
;
4
8
(1
):
3
-
1
5
.
[9
]
Rod
ríg
u
e
z
G
,
Ab
e
l
l
a
F,
Dur
á
n
-
Si
n
d
r
e
u
F,
Pa
te
l
S,
Roi
g
M
.
In
f
l
u
e
n
c
e
o
f
Con
e
-
b
e
a
m
Com
p
u
te
d
To
m
o
g
ra
p
h
y
i
n
Cli
n
i
c
a
l
De
c
i
s
i
o
n
M
a
k
i
n
g
a
m
o
n
g
Sp
e
c
i
a
l
i
s
ts
.
J
o
u
r
n
a
l
o
f
e
n
d
o
d
o
n
t
i
c
s
.
2
0
1
7
;
4
3
(2
):
1
9
4
–
199.
[1
0
]
Rod
ríg
u
e
z
G
,
Pa
te
l
S,
Durá
n
-
Si
n
d
re
u
F,
Roi
g
M
,
Ab
e
l
l
a
F.
In
fl
u
e
n
c
e
o
f
Con
e
-
b
e
a
m
Com
p
u
te
d
To
m
o
g
ra
p
h
y
o
n
En
d
o
d
o
n
ti
c
Ret
re
a
t
m
e
n
t
Stra
te
g
i
e
s
a
m
o
n
g
G
e
n
e
r
a
l
De
n
ta
l
Pra
c
t
i
ti
o
n
e
rs
a
n
d
En
d
o
d
o
n
ti
s
ts
.
J
o
u
rn
a
l
o
f
e
n
d
o
d
o
n
ti
c
s
.
2
0
1
7
;
4
3
(9
):
1
4
3
3
–
1
4
3
7
.
[1
1
]
Nil
s
s
o
n
J
,
Ric
h
a
rd
s
RG
,
Th
o
r
A,
Ka
m
e
r
L
.
Vi
rtu
a
l
b
i
te
re
g
i
s
t
ra
ti
o
n
u
s
i
n
g
i
n
tra
o
r
a
l
d
i
g
i
t
a
l
s
c
a
n
n
i
n
g
,
C
T
a
n
d
CBC
T:
In
v
i
tr
o
e
v
a
l
u
a
ti
o
n
o
f
a
n
e
w
m
e
th
o
d
a
n
d
i
t
s
i
m
p
l
i
c
a
ti
o
n
fo
r
o
rth
o
g
n
a
t
h
i
c
s
u
r
g
e
ry
.
J
o
u
r
n
a
l
o
f
Cran
i
o
-
M
a
x
i
l
l
o
fa
c
i
a
l
Su
rg
e
ry
.
2
0
1
6
;
4
4
(9
):
1
1
9
4
–
1
2
0
0
.
[1
2
]
Ka
k
e
h
b
a
ra
e
i
S,
Se
y
e
d
a
r
a
b
i
H,
Z
e
n
o
u
z
A
T.
De
n
ta
l
S
e
g
m
e
n
ta
t
i
o
n
i
n
Con
e
-
b
e
a
m
Com
p
u
te
d
To
m
o
g
ra
p
h
y
I
m
a
g
e
s
Us
i
n
g
Wa
te
rs
h
e
d
a
n
d
M
o
rp
h
o
l
o
g
y
O
p
e
ra
to
r
s
.
J
o
u
rn
a
l
o
f
m
e
d
i
c
a
l
s
i
g
n
a
l
s
a
n
d
s
e
n
s
o
rs
.
2
0
1
8
;
8
(
2
):
1
1
9
–
124.
[1
3
]
L
v
J
,
Wa
n
g
F,
X
u
L
,
M
a
Z,
Y
a
n
g
B.
A
s
e
g
m
e
n
ta
ti
o
n
m
e
th
o
d
o
f
b
a
g
g
e
d
g
re
e
n
a
p
p
l
e
i
m
a
g
e
.
Sc
i
e
n
t
i
a
h
o
rti
c
u
l
t
u
ra
e
.
2
0
1
9
;
2
4
6
:
4
1
1
–
417.
[1
4
]
Fa
n
Y,
Be
a
re
R,
M
a
tt
h
e
ws
H,
Sc
h
n
e
i
d
e
r
P,
Ki
l
p
a
tr
i
c
k
N,
Cle
m
e
n
t
J
,
Cla
e
s
P,
Pe
n
i
n
g
t
o
n
A,
Ad
a
m
s
o
n
C. M
a
rk
e
r
-
b
a
s
e
d
wa
t
e
rs
h
e
d
t
r
a
n
s
f
o
rm
m
e
th
o
d
f
o
r f
u
l
l
y
a
u
t
o
m
a
ti
c
m
a
n
d
i
b
u
l
a
r s
e
g
m
e
n
ta
t
i
o
n
f
ro
m
CBC
T
i
m
a
g
e
s
.
Den
to
m
a
x
i
l
l
o
fa
c
i
a
l
Ra
d
i
o
l
o
g
y
.
2
0
1
9
;
4
8
(2
):
2
0
1
8
0
2
6
1
.
[1
5
]
G
a
o
H,
Cha
e
O
.
In
d
i
v
i
d
u
a
l
to
o
th
s
e
g
m
e
n
ta
ti
o
n
fro
m
CT
i
m
a
g
e
s
u
s
i
n
g
l
e
v
e
l
s
e
t
m
e
th
o
d
wit
h
s
h
a
p
e
a
n
d
i
n
te
n
s
i
t
y
p
r
i
o
r.
P
a
tt
e
rn
Re
c
o
g
n
i
ti
o
n
.
2
0
1
0
;
4
3
(7
):
2
4
0
6
-
2
4
1
7
.
[1
6
]
Xi
a
Z,
G
a
n
Y,
Cha
n
g
L
,
Xi
o
n
g
J
,
Zh
a
o
Q
.
I
n
d
i
v
i
d
u
a
l
to
o
t
h
s
e
g
m
e
n
ta
ti
o
n
fr
o
m
CT
i
m
a
g
e
s
s
c
a
n
n
e
d
wit
h
c
o
n
ta
c
t
s
o
f
m
a
x
i
l
l
a
r
y
a
n
d
m
a
n
d
i
b
l
e
te
e
t
h
.
Com
p
u
t
e
r
m
e
th
o
d
s
a
n
d
p
ro
g
ra
m
s
i
n
b
i
o
m
e
d
i
c
i
n
e
.
2
0
1
7
;
1
3
8
:
1
-
2.
[1
7
]
G
a
n
Y,
Xi
a
Z,
Xi
o
n
g
J
,
L
i
G
,
Z
h
a
o
Q
.
To
o
th
a
n
d
Al
v
e
o
l
a
r
Bo
n
e
Se
g
m
e
n
ta
t
i
o
n
Fro
m
Den
ta
l
Com
p
u
te
d
To
m
o
g
ra
p
h
y
I
m
a
g
e
s
.
IEEE
j
o
u
rn
a
l
o
f
b
i
o
m
e
d
i
c
a
l
a
n
d
h
e
a
l
th
i
n
fo
rm
a
ti
c
s
.
2
0
1
7
;
2
2
(
1
):
1
9
6
-
2
0
4
.
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
MNIK
A
IS
S
N: 1
69
3
-
6
93
0
◼
A
uto
ma
t
i
c
i
ma
g
e s
l
i
c
e m
ark
i
ng
prop
ag
at
i
o
n o
n
s
eg
m
en
tat
i
on
…
(
A
gu
s
Za
i
n
al
A
r
i
f
i
n
)
3225
[1
8
]
Ari
fi
n
AZ,
Ar
i
fi
a
n
i
S
,
Fa
r
i
z
a
A,
Nav
a
s
ta
ra
DA,
In
d
ra
s
w
a
ri
R.
Hie
ra
rc
h
i
c
a
l
C
l
u
s
te
ri
n
g
L
i
n
k
a
g
e
fo
r
Reg
i
o
n
M
e
rg
i
n
g
i
n
In
t
e
ra
c
ti
v
e
Im
a
g
e
Se
g
m
e
n
t
a
ti
o
n
o
n
De
n
ta
l
Co
n
e
Be
a
m
Com
p
u
te
d
T
o
m
o
g
r
a
p
h
y
.
IEE
E
In
te
rn
a
ti
o
n
a
l
Co
n
fe
re
n
c
e
o
n
A
p
p
l
i
e
d
I
n
fo
rm
a
ti
o
n
T
e
c
h
n
o
l
o
g
y
a
n
d
I
n
n
o
v
a
ti
o
n
(I
CAI
TI
).
2
0
1
8
:
1
2
4
–
128.
[1
9
]
Wa
n
g
L
,
L
i
J
P,
G
e
ZP,
L
i
G
.
CBC
T
i
m
a
g
e
b
a
s
e
d
s
e
g
m
e
n
t
a
ti
o
n
m
e
th
o
d
fo
r
to
o
th
p
u
l
p
c
a
v
i
t
y
re
g
i
o
n
e
x
tra
c
ti
o
n
.
De
n
to
m
a
x
i
l
l
o
fa
c
i
a
l
Rad
i
o
l
o
g
y
.
2
0
1
8
;
4
8
(2
):
2
0
1
8
0
2
3
6
.
[2
0
]
Nau
m
o
v
i
c
h
SS
,
Na
u
m
o
v
i
c
h
SA,
G
o
n
c
h
a
re
n
k
o
VG
.
Th
re
e
-
d
i
m
e
n
s
i
o
n
a
l
re
c
o
n
s
tru
c
ti
o
n
o
f
t
e
e
th
a
n
d
j
a
w
s
b
a
s
e
d
o
n
s
e
g
m
e
n
ta
t
i
o
n
o
f
CT
i
m
a
g
e
s
u
s
i
n
g
wat
e
rs
h
e
d
tra
n
s
fo
rm
a
ti
o
n
.
Den
t
o
m
a
x
i
l
l
o
f
a
c
i
a
l
Rad
i
o
l
o
g
y
.
2
0
1
5
;
4
4
(4
)
:
1
–
6.
[2
1
]
In
d
ra
s
wari
R,
Ku
ri
ta
T,
Ari
fi
n
AZ,
Su
c
i
a
ti
N,
A
s
tu
t
i
ER,
Nav
a
s
t
a
ra
DA.
3
D
Reg
i
o
n
M
e
rg
i
n
g
fo
r
Se
g
m
e
n
ta
t
i
o
n
o
f
Te
e
t
h
o
n
Co
n
e
-
Be
a
m
Com
p
u
t
e
d
To
m
o
g
ra
p
h
y
Im
a
g
e
s
.
IEEE
J
o
i
n
t
1
0
t
h
I
n
te
rn
a
t
i
o
n
a
l
Con
fe
re
n
c
e
o
n
So
ft
Com
p
u
ti
n
g
a
n
d
In
te
l
l
i
g
e
n
t
S
y
s
t
e
m
s
(SCIS)
a
n
d
1
9
th
In
te
rn
a
ti
o
n
a
l
Sy
m
p
o
s
i
u
m
o
n
Ad
v
a
n
c
e
d
I
n
te
l
l
i
g
e
n
t
Sy
s
te
m
s
(ISI
S).
2
0
1
8
:
3
4
1
-
3
4
5
.
[2
2
]
Che
n
g
Y.
M
e
a
n
Sh
i
ft
,
M
o
d
e
Se
e
k
i
n
g
a
n
d
Cl
u
s
t
e
ri
n
g
.
IE
EE
tra
n
s
a
c
ti
o
n
s
o
n
p
a
tt
e
rn
a
n
a
l
y
s
i
s
a
n
d
m
a
c
h
i
n
e
i
n
t
e
l
l
i
g
e
n
c
e
.
1
9
9
5
;
1
7
(8
):
7
9
0
–
7
9
9
.
[2
3
]
Com
a
n
i
c
i
u
D,
M
e
e
r
P.
M
e
a
n
Sh
i
ft
:
A
Rob
u
s
t
A
p
p
ro
a
c
h
To
ward
F
e
a
tu
re
Sp
a
c
e
A
n
a
l
y
s
i
s
.
2
0
0
2
;
2
4
(5
):
6
0
3
–
619.
[2
4
]
O
ts
u
N.
A
Th
re
s
h
o
l
d
Se
l
e
c
ti
o
n
M
e
th
o
d
fro
m
G
ra
y
-
L
e
v
e
l
H
i
s
to
g
r
a
m
s
.
IEEE
t
ra
n
s
a
c
ti
o
n
s
o
n
s
y
s
t
e
m
s
,
m
a
n
,
a
n
d
c
y
b
e
rn
e
ti
c
s
.
1
9
7
9
;
2
0
(1
):
6
2
–
6
6
.
[2
5
]
Pu
s
p
a
n
i
n
g
ru
m
A,
Nur
N,
Riz
a
O
S,
Ari
fi
n
AZ.
Im
a
g
e
th
re
s
h
o
l
d
i
n
g
b
a
s
e
d
o
n
h
i
e
ra
r
c
h
i
c
a
l
c
l
u
s
te
r
i
n
g
a
n
a
l
y
s
i
s
a
n
d
p
e
r
c
e
n
t
i
l
e
m
e
th
o
d
fo
r
tu
n
a
i
m
a
g
e
s
e
g
m
e
n
ta
t
i
o
n
.
Nus
a
n
t
a
ra
J
o
u
rn
a
l
o
f
Com
p
u
te
rs
a
n
d
i
ts
Ap
p
l
i
c
a
ti
o
n
s
.
2
0
1
8
;
2
(1
):
1
–
8.
Evaluation Warning : The document was created with Spire.PDF for Python.