I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
pu
t
er
Science
Vo
l.
3
7
,
No
.
3
,
Ma
r
ch
20
2
5
,
p
p
.
1
90
8
~
1
91
7
I
SS
N:
2
502
-
4
7
52
,
DOI
:
1
0
.
1
1
5
9
1
/ijee
cs
.v
3
7
.
i
3
.
pp
1
90
8
-
1
91
7
1908
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs
.
ia
esco
r
e.
co
m
M
edica
l
ima
g
e re
g
istra
tion a
nd cla
ss
ificatio
n using
s
mell a
g
ent
ra
t
swa
rm o
ptimi
za
tion
ba
sed
dee
p
M
a
x
o
ut
ne
twork
S
a
n
je
e
v
k
u
m
a
r
Ang
a
d
i
1
,
Mu
k
e
s
h
K
u
m
a
r
Tr
i
p
a
th
i
2
,
Chu
d
a
m
a
n
De
v
id
a
sr
a
o
S
u
k
te
3
,
S
h
iv
e
n
d
r
a
4
1
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
N
u
t
a
n
C
o
l
l
e
g
e
o
f
E
n
g
i
n
e
e
r
i
n
g
a
n
d
R
e
s
e
a
r
c
h
,
P
u
n
e
,
I
n
d
i
a
2
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
V
a
r
d
h
a
ma
n
C
o
l
l
e
g
e
o
f
En
g
i
n
e
e
r
i
n
g
,
H
y
d
e
r
a
b
a
d
,
I
n
d
i
a
3
D
e
p
a
r
t
me
n
t
o
f
I
n
f
o
r
mat
i
o
n
Te
c
h
n
o
l
o
g
y
,
V
i
s
h
w
a
k
a
r
m
a
I
n
st
i
t
u
t
e
o
f
I
n
f
o
r
ma
t
i
o
n
Te
c
h
n
o
l
o
g
y
,
P
u
n
e
,
I
n
d
i
a
4
D
e
p
a
r
t
me
n
t
o
f
B
C
A
,
D
.
K
C
o
l
l
e
g
e
,
D
u
mr
a
o
n
,
I
n
d
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
ul
10
,
2
0
2
4
R
ev
is
ed
Sep
30
,
2
0
2
4
Acc
ep
ted
Oct
7
,
2
0
2
4
M
e
d
ica
l
ima
g
e
re
g
istratio
n
(
M
IR)
is
a
c
ru
c
ial
tas
k
i
n
c
li
n
ic
a
l
ima
g
e
p
ro
c
e
ss
in
g
,
in
v
o
l
v
i
n
g
t
h
e
a
li
g
n
m
e
n
t
o
f
ima
g
e
s
fro
m
d
iffere
n
t
m
o
d
a
li
ti
e
s,
su
c
h
a
s
m
a
g
n
e
ti
c
re
so
n
a
n
c
e
ima
g
in
g
(M
RI)
a
n
d
c
o
m
p
u
te
d
to
m
o
g
ra
p
h
y
(CT)
,
a
c
ro
ss
v
a
rio
u
s
ti
m
e
p
o
in
ts
a
n
d
s
u
b
jec
ts.
De
sp
it
e
n
u
m
e
ro
u
s
a
d
v
a
n
c
e
m
e
n
ts
,
n
o
u
n
i
v
e
rsa
l
m
e
th
o
d
c
a
ters
to
a
ll
M
IR
a
p
p
li
c
a
ti
o
n
s.
Th
is
p
a
p
e
r
in
tr
o
d
u
c
e
s
th
e
sm
e
ll
a
g
e
n
t
ra
t
sw
a
rm
o
p
ti
m
iza
ti
o
n
b
a
se
d
d
e
e
p
M
a
x
o
u
t
n
e
two
r
k
(S
ARSO
-
DMN)
fo
r
M
IR
a
n
d
c
las
sifica
ti
o
n
.
T
h
is
wo
r
k
a
ims
to
e
n
h
a
n
c
e
th
e
a
c
c
u
ra
c
y
a
n
d
e
fficie
n
c
y
o
f
m
e
d
ica
l
ima
g
e
a
li
g
n
m
e
n
t
a
n
d
c
las
sifica
ti
o
n
,
a
d
d
r
e
ss
in
g
th
e
c
h
a
ll
e
n
g
e
s
p
o
se
d
b
y
d
i
v
e
rse
ima
g
in
g
m
o
d
a
li
ti
e
s
a
n
d
tem
p
o
ra
l
v
a
riatio
n
s.
Th
e
p
ro
b
lem
in
v
o
lv
e
s
e
ffe
c
ti
v
e
l
y
re
g
isterin
g
CT
a
n
d
M
RI
ima
g
e
s,
fo
ll
o
we
d
b
y
in
h
a
le
a
n
d
e
x
h
a
le
c
las
sifica
t
io
n
.
T
h
e
p
ro
p
o
se
d
a
p
p
r
o
a
c
h
b
e
g
in
s
with
fe
e
d
in
g
th
e
i
n
p
u
t
ima
g
e
s
in
t
o
a
c
o
n
v
o
l
u
ti
o
n
a
l
n
e
u
ra
l
n
e
two
rk
(CNN
),
fo
ll
o
we
d
b
y
a
p
p
l
y
in
g
a
d
e
fo
rm
a
ti
o
n
field
to
g
e
n
e
ra
te
a
n
i
n
term
e
d
i
a
te
o
u
t
p
u
t
(o
u
t
p
u
t
-
1
).
Th
is
o
u
tp
u
t,
a
l
o
n
g
with
th
e
i
n
p
u
t
M
RI
ima
g
e
s,
is
fu
rt
h
e
r
p
ro
c
e
ss
e
d
b
y
a
CNN
t
o
p
ro
d
u
c
e
o
u
tp
u
t
-
2
.
S
u
b
se
q
u
e
n
t
ly
,
o
u
t
p
u
t
-
2
a
n
d
th
e
in
p
u
t
M
RI
ima
g
e
a
re
su
b
jec
ted
to
a
n
o
t
h
e
r
CNN
,
re
su
lt
in
g
in
th
e
fin
a
l
re
g
istere
d
ima
g
e
.
Th
e
c
las
sifica
ti
o
n
p
h
a
se
u
ti
l
ize
s
a
DMN
o
p
ti
m
ize
d
b
y
th
e
S
ARSO
a
lg
o
rit
h
m
,
wh
ich
c
o
m
b
i
n
e
s
sm
e
ll
a
g
e
n
t
o
p
ti
m
iza
ti
o
n
(
S
A
O)
a
n
d
ra
t
sw
a
rm
o
p
ti
m
i
z
e
r
(RS
O).
T
h
e
re
su
lt
s
d
e
m
o
n
stra
te
th
a
t
S
A
RS
O
-
DMN
a
c
h
iev
e
s
a
m
a
x
imu
m
a
c
c
u
ra
c
y
o
f
9
0
.
7
%
,
a
m
in
imu
m
fa
lse
p
o
siti
v
e
ra
te
(F
P
R)
o
f
1
1
.
3
%
,
a
n
d
a
m
a
x
im
u
m
tru
e
p
o
siti
v
e
ra
te
(T
P
R)
o
f
9
1
.
2
%
.
Th
e
S
ARSO
-
DMN
a
p
p
r
o
a
c
h
p
ro
v
i
d
e
s
a
ro
b
u
st
s
o
lu
t
io
n
fo
r
M
IR
a
n
d
c
las
sifica
t
io
n
,
lev
e
ra
g
i
n
g
a
d
v
a
n
c
e
d
o
p
t
imiz
a
ti
o
n
tec
h
n
i
q
u
e
s
t
o
e
n
h
a
n
c
e
p
e
rfo
rm
a
n
c
e
.
K
ey
w
o
r
d
s
:
C
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
Dee
p
Ma
x
o
u
t
n
etwo
r
k
Me
d
ical
im
ag
e
r
eg
is
tr
atio
n
R
at
s
war
m
o
p
tim
izer
Sm
ell
ag
en
t o
p
tim
izatio
n
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
San
jeev
k
u
m
ar
A
n
g
ad
i
Dep
ar
tm
en
t o
f
C
o
m
p
u
ter
Scie
n
ce
an
d
E
n
g
in
ee
r
in
g
,
N
u
tan
C
o
lleg
e
o
f
E
n
g
in
ee
r
i
n
g
an
d
R
esear
ch
Pu
n
e,
I
n
d
ia
E
m
ail: a
n
g
ad
i.san
jeev
k
u
m
ar
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
I
n
clin
ical
im
ag
e
ass
es
s
m
en
t,
d
ef
o
r
m
ab
le
im
ag
e
r
e
g
is
tr
atio
n
(
DI
R
)
is
th
e
p
r
e
-
p
r
o
ce
s
s
in
g
m
eth
o
d
,
wh
ich
id
en
tifie
s
n
o
n
-
lin
ea
r
s
p
atial
tr
an
s
f
o
r
m
atio
n
to
alig
n
an
im
ag
e.
I
t
is
s
ig
n
if
ican
t
f
o
r
s
ev
er
al
m
ed
ical
ap
p
licatio
n
s
;
wh
er
ein
s
p
atial
alig
n
in
g
o
f
an
ato
m
ical
s
tr
u
ct
u
r
e
is
n
ec
ess
ar
y
.
Su
ch
m
o
d
alities
in
clu
d
e
im
ag
e
-
ass
is
ted
p
r
o
ce
s
s
f
o
r
d
iag
n
o
s
in
g
as
well
as
p
atien
t o
r
g
an
izati
o
n
,
wh
e
r
e
im
ag
es
a
r
e
o
b
tain
ed
at
d
iv
er
s
e
p
o
in
ts
in
a
tim
e
o
r
u
tili
zin
g
v
ar
io
u
s
m
o
d
alities
[
1
]
.
Par
ticu
lar
ly
in
ca
r
d
iac
im
ag
e
ass
ess
m
en
t,
DI
R
i
s
u
tili
ze
d
in
im
ag
e
-
ass
is
ted
in
ter
f
er
en
ce
s
,
wh
ich
n
ee
d
m
y
o
ca
r
d
ial
m
o
tio
n
tr
ac
k
i
n
g
o
r
else
in
m
y
o
ca
r
d
ial
p
er
f
u
s
io
n
r
esear
ch
es
[
2
]
.
W
ith
q
u
ick
p
r
o
g
r
ess
io
n
o
f
co
m
p
u
ter
an
d
c
lin
ical
tech
n
o
lo
g
ies,
clin
ical
im
ag
es
o
f
d
iv
er
s
e
m
o
d
alities
ca
n
b
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
Med
ica
l ima
g
e
r
eg
is
tr
a
tio
n
a
n
d
cla
s
s
ifica
tio
n
u
s
in
g
s
mell
a
g
en
t
…
(
S
a
n
jeev
ku
ma
r
A
n
g
a
d
i
)
1909
ac
q
u
ir
ed
t
h
r
o
u
g
h
co
m
p
u
ted
to
m
o
g
r
ap
h
y
(
CT
)
,
n
u
clea
r
m
a
g
n
etic
r
eso
n
an
ce
im
a
g
in
g
(
MRI
)
an
d
o
th
er
t
y
p
es
o
f
im
ag
in
g
t
o
o
ls
[
3
]
.
I
n
an
ar
e
a
o
f
clin
ical
im
ag
es,
th
er
e
a
r
e
f
ew
g
en
er
al
is
s
u
es
in
d
ee
p
lear
n
i
n
g
-
en
a
b
led
tech
n
iq
u
es
[
4
]
,
lik
e
lo
n
g
er
tr
ai
n
in
g
p
er
io
d
,
h
ig
h
er
n
ec
ess
ity
f
o
r
h
a
r
d
war
e
a
n
d
wo
r
s
t
p
o
r
ta
b
ilit
y
o
f
s
y
s
tem
.
A
n
in
d
is
tin
ctn
ess
o
f
co
n
v
en
tio
n
al
r
eg
is
tr
atio
n
tech
n
iq
u
e
f
r
eq
u
e
n
tly
em
er
g
es
o
win
g
t
o
en
o
r
m
o
u
s
v
ar
iatio
n
am
o
n
g
r
ef
er
en
ce
an
d
f
lo
ati
n
g
im
ag
es
[
4
]
.
A
r
e
g
is
tr
atio
n
q
u
ality
d
ir
ec
tly
in
f
lu
en
ce
s
t
h
e
co
n
s
eq
u
e
n
ce
o
f
its
ch
o
r
e
an
d
f
o
llo
w
-
u
p
task
s
,
th
u
s
th
e
p
r
o
b
e
o
n
MI
R
tech
n
o
lo
g
y
h
as
e
x
ten
s
iv
e
an
d
r
ea
s
o
n
ab
le
im
p
o
r
tan
ce
.
Dep
e
n
d
in
g
u
p
o
n
ty
p
es
o
f
s
p
atial
tr
an
s
f
o
r
m
atio
n
s
,
an
im
ag
e
r
e
g
is
tr
atio
n
in
c
lu
d
es
n
o
n
-
lin
ea
r
tr
a
n
s
f
o
r
m
atio
n
as
well
as
lin
ea
r
tr
an
s
f
o
r
m
atio
n
.
A
lin
ea
r
tr
an
s
f
o
r
m
atio
n
co
n
s
is
ts
o
f
a
f
f
in
e
r
e
g
is
tr
atio
n
a
n
d
r
ig
id
r
e
g
is
tr
atio
n
.
I
t
is
th
e
to
tal
tr
an
s
f
o
r
m
atio
n
f
o
r
g
lo
b
a
l
im
ag
e
an
d
is
f
r
eq
u
en
tly
u
tili
ze
d
as
p
r
e
-
r
eg
is
tr
atio
n
f
u
n
cti
o
n
f
o
r
co
m
p
licated
m
u
ltip
le
s
tag
e
r
eg
is
tr
atio
n
[
5
]
.
T
h
is
wo
r
k
s
ig
n
if
ican
tly
co
n
t
r
ib
u
tes
to
th
e
f
ield
o
f
clin
ical
im
ag
e
p
r
o
ce
s
s
in
g
,
o
f
f
er
in
g
a
m
eth
o
d
th
at
h
as
th
e
p
o
ten
tial
to
im
p
r
o
v
e
d
iag
n
o
s
tic
ac
cu
r
ac
y
an
d
p
atien
t
o
u
tc
o
m
es
in
r
ea
l
-
wo
r
l
d
clin
ical
s
ettin
g
s
.
Dee
p
lea
r
n
in
g
[
6
]
ca
n
lea
r
n
h
ier
ar
ch
ical
f
ea
tu
r
es
b
y
ass
u
m
i
n
g
s
im
p
ler
im
a
g
e
f
ea
tu
r
es
in
i
tially
an
d
th
er
ea
f
ter
in
cr
ea
s
in
g
ly
c
o
n
s
tr
u
ctin
g
u
p
m
an
y
d
ee
p
o
n
es
f
r
o
m
p
r
io
r
lev
els.
Few
d
ee
p
lear
n
in
g
tech
n
i
q
u
es
lik
e
s
tack
ed
au
to
-
e
n
co
d
e
r
(
SAE)
[
7
]
,
d
ee
p
b
elief
n
etwo
r
k
(
D
B
N)
an
d
C
NN
h
av
e
r
ev
ea
le
d
th
e
a
d
v
an
ta
g
e
in
cr
ea
tin
g
th
e
co
m
p
licate
f
ea
tu
r
e
in
f
o
r
m
atio
n
in
h
ig
h
er
-
d
i
m
en
s
io
n
al
d
ata
a
n
d
h
av
e
b
e
en
u
s
ed
in
im
ag
e
p
r
o
ce
s
s
in
g
as
well
as
ass
es
s
m
en
t
ch
o
r
es
lik
e
im
a
g
e
d
e
n
o
is
in
g
,
class
if
icatio
n
an
d
s
eg
m
en
t
atio
n
[
8
]
.
R
ec
en
tly
,
d
ee
p
lea
r
n
in
g
,
p
ar
ticu
lar
ly
C
NN
h
as
s
u
b
v
er
s
iv
e
im
p
ac
t
o
n
a
d
o
m
ain
o
f
co
m
p
u
tatio
n
v
is
io
n
th
at
in
clu
d
es
im
ag
e
s
eg
m
en
tatio
n
,
class
if
icatio
n
as
well
as
tar
g
et
d
etec
tio
n
.
A
d
ee
p
lea
r
n
in
g
-
en
ab
le
d
r
e
g
is
tr
atio
n
tech
n
iq
u
e
s
u
b
s
titu
tes
an
iter
ativ
e
o
p
tim
izatio
n
p
r
o
ce
d
u
r
e
o
f
co
n
v
o
lu
tio
n
al
ap
p
r
o
ac
h
u
tili
zin
g
th
e
tr
ain
ed
s
y
s
tem
f
o
r
ac
h
iev
in
g
q
u
ick
r
eg
is
tr
atio
n
o
f
u
n
o
b
s
er
v
ed
p
air
o
f
im
ag
e
v
o
lu
m
es.
T
h
is
h
as
b
ee
n
g
en
er
a
ted
q
u
ick
ly
an
d
h
as
b
ec
o
m
e
a
m
ajo
r
d
ir
ec
tio
n
o
f
MRI
p
r
o
b
e
n
o
wad
ay
s
b
y
g
o
o
d
q
u
ality
o
f
its
o
u
ts
tan
d
in
g
c
o
m
p
u
tatio
n
s
p
ee
d
as
well
as
ac
cu
r
ac
y
[
9
]
.
Dee
p
lea
r
n
in
g
tech
n
iq
u
es
th
at
in
d
icate
a
p
r
o
m
is
in
g
d
o
m
ain
o
f
p
r
o
b
e
h
av
e
d
ev
elo
p
e
d
f
o
r
im
ag
e
r
eg
is
tr
atio
n
lately
.
Ma
n
y
o
f
th
ese
a
p
p
r
o
a
ch
es
co
n
c
en
tr
ate
o
n
r
eg
i
o
n
s
o
f
in
ter
est
(
R
OI
s
)
in
h
ig
h
ly
“static”
p
ar
ts
lik
e
p
r
o
s
tate,
liv
er
o
r
b
r
ai
n
.
T
h
e
d
ee
p
lear
n
i
n
g
im
ag
e
r
eg
is
tr
atio
n
(
DL
I
R
)
m
o
d
el
[
1
0
]
ex
ec
u
te
d
u
n
s
u
p
er
v
is
ed
a
f
f
in
e
as
well
as
d
ef
o
r
m
ab
le
3
D
im
ag
e
r
eg
is
tr
atio
n
[
1
1
]
.
I
n
p
r
esen
t
y
ea
r
s
,
d
ee
p
lear
n
in
g
an
d
m
ac
h
in
e
lear
n
in
g
h
av
e
b
ee
n
p
av
ed
th
eir
p
ath
in
to
im
ag
e
r
eg
is
tr
atio
n
an
d
o
th
er
ass
o
ciate
d
ap
p
licatio
n
s
n
am
ely
co
m
p
u
ter
-
aid
ed
d
iag
n
o
s
is
[
1
2
]
,
o
b
tain
i
n
g
h
i
g
h
ly
p
r
ec
is
e
a
n
d
co
n
s
tan
t
s
o
lu
tio
n
s
.
T
h
e
r
e
is
th
e
s
h
o
r
tag
e
o
f
co
n
s
en
s
u
s
o
n
ca
teg
o
r
ical
m
et
h
o
d
,
wh
ich
b
e
n
ef
its
f
r
o
m
r
o
b
u
s
tn
ess
o
f
d
ee
p
lear
n
in
g
ag
ain
s
t
o
f
f
er
in
g
h
ig
h
ac
cu
r
ac
y
r
e
g
is
tr
atio
n
s
in
s
p
ite
o
f
o
b
tain
e
d
im
ag
e
p
air
cr
iter
io
n
[
1
3
]
.
B
o
v
e
ir
i
et
a
l.
[
1
4
]
in
tr
o
d
u
ce
d
d
ee
p
lear
n
i
n
g
f
o
r
th
e
n
o
n
-
r
i
g
id
m
u
lti
-
m
o
d
el
MRI
th
at
co
u
ld
d
ig
th
e
in
tr
in
s
ic
f
ea
tu
r
es
o
f
clin
ical
im
ag
es
m
o
r
e
ef
f
icien
tly
t
h
an
h
an
d
-
d
ev
is
ed
f
ea
tu
r
e
e
x
tr
ac
ti
o
n
tech
n
iq
u
es.
E
v
en
th
o
u
g
h
,
th
is
m
eth
o
d
was
n
o
t
ca
p
ab
le
to
o
f
f
e
r
ac
ce
p
tab
le
r
e
g
is
ter
ed
o
u
tco
m
es.
Yan
g
et
a
l
.
[
1
5
]
p
r
esen
ted
⊥
-
lo
s
s
f
o
r
MRI
im
ag
e
r
ec
o
n
s
tr
u
ctio
n
an
d
r
e
g
is
tr
atio
n
.
I
t
m
i
n
im
ized
en
d
-
p
o
in
t
m
is
tak
e
w
ith
less
er
r
esid
u
al
er
r
o
r
s
,
b
u
t
p
h
ase
er
r
o
r
was
in
cr
ea
s
ed
s
lo
wly
wh
ile
in
c
r
ea
s
in
g
ex
tr
a
n
o
is
e.
Z
h
u
et
a
l.
[
1
6
]
d
esig
n
ed
p
r
o
g
r
ess
iv
e
an
ato
m
ically
co
n
s
tr
ain
ed
d
ee
p
n
e
u
r
al
n
etwo
r
k
(
PAC
N)
f
o
r
3
D
-
d
ef
o
r
m
ab
le
MI
R
.
T
h
is
tech
n
iq
u
e
g
en
er
ated
h
ig
h
e
r
q
u
ality
o
f
d
e
f
o
r
m
atio
n
f
ield
,
th
o
u
g
h
it
d
id
n
o
t
in
clu
d
e
f
in
e
-
tu
n
ed
h
y
p
er
p
ar
am
eter
s
.
T
r
ip
ath
i
an
d
Sh
iv
en
d
r
a
[
1
7
]
d
ev
elo
p
ed
m
u
ltis
ca
le
h
i
er
ar
ch
ical
d
ef
o
r
m
ab
le
r
eg
is
tr
atio
n
n
etwo
r
k
(
MH
Net)
f
o
r
3
D
b
r
ain
MR
im
ag
e
r
eg
is
tr
atio
n
.
I
t
h
ad
p
r
o
b
a
b
le
f
o
r
r
e
g
is
ter
in
g
o
th
e
r
o
r
g
an
s
as
well
as
m
o
d
al
im
a
g
es,
b
u
t
s
till
it
was
o
n
ly
u
tili
ze
d
f
o
r
b
r
ain
MRI
r
e
g
is
tr
atio
n
.
I
t
was
ab
le
to
ex
ec
u
te
in
tr
a
-
p
atien
t
a
b
d
o
m
i
n
al
C
T
-
MRI
r
eg
is
tr
atio
n
s
u
cc
ess
f
u
lly
,
b
u
t it
f
ailed
to
p
r
ed
ict
h
elp
f
u
l sy
n
th
etic
tr
an
s
f
o
r
m
atio
n
s
f
o
r
im
p
r
o
v
in
g
d
etailed
alig
n
m
en
t.
Me
d
ical
im
ag
e
r
eg
is
tr
atio
n
(
MI
R
)
is
a
cr
itical
task
in
clin
ical
im
ag
e
p
r
o
ce
s
s
in
g
,
i
n
v
o
lv
in
g
t
h
e
alig
n
m
en
t
o
f
im
ag
es
f
r
o
m
v
ar
i
o
u
s
m
o
d
alities
,
s
u
ch
as
MRI
a
n
d
C
T
,
ac
r
o
s
s
d
if
f
er
en
t
tim
e
p
o
in
ts
an
d
s
u
b
jects.
Desp
ite
th
e
ex
is
tin
g
m
eth
o
d
s
,
m
o
r
e
u
n
iv
er
s
al
s
o
lu
tio
n
s
m
u
s
t
b
e
a
v
ailab
le
f
o
r
all
MI
R
task
s
.
T
h
is
p
ap
er
in
tr
o
d
u
ce
s
a
n
o
v
el
ap
p
r
o
ac
h
,
th
e
s
m
ell
ag
en
t
r
at
s
war
m
o
p
ti
m
izatio
n
d
ee
p
Ma
x
o
u
t
n
etwo
r
k
(
SAR
SO
-
DM
N)
,
to
im
p
r
o
v
e
th
e
ac
c
u
r
ac
y
a
n
d
e
f
f
icien
cy
o
f
MI
R
an
d
class
if
icatio
n
.
T
h
is
s
tu
d
y
aim
s
to
ad
d
r
ess
th
e
lim
itat
io
n
s
o
f
cu
r
r
en
t M
I
R
tech
n
iq
u
es b
y
d
ev
elo
p
in
g
a
m
o
r
e
r
o
b
u
s
t a
n
d
ef
f
ec
tiv
e
m
eth
o
d
.
T
h
e
p
r
o
b
le
m
s
tatem
en
t f
o
cu
s
es
o
n
th
e
ch
al
len
g
es
o
f
r
eg
is
ter
in
g
C
T
an
d
MRI
im
ag
es,
f
o
llo
wed
b
y
ac
cu
r
ate
i
n
h
ale
a
n
d
e
x
h
ale
class
if
icatio
n
.
T
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
in
v
o
lv
es
a
m
u
lti
-
s
tag
e
p
r
o
ce
s
s
s
tar
tin
g
with
t
h
e
in
p
u
t
im
ag
es
f
ed
i
n
to
a
c
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
(
C
NN)
.
A
d
ef
o
r
m
atio
n
f
ield
is
th
en
a
p
p
lied
to
p
r
o
d
u
ce
an
in
ter
m
e
d
iate
o
u
t
p
u
t (
o
u
t
p
u
t
-
1
)
.
T
h
is
o
u
tp
u
t
an
d
th
e
in
p
u
t
MRI
im
ag
es
ar
e
p
r
o
ce
s
s
ed
th
r
o
u
g
h
a
n
o
th
er
C
NN
to
g
e
n
er
ate
o
u
tp
u
t
-
2
.
Su
b
s
eq
u
en
tly
,
o
u
tp
u
t
-
2
an
d
th
e
in
p
u
t M
R
I
im
ag
e
ar
e
s
u
b
jecte
d
to
a
f
i
n
al
C
NN
s
tag
e,
r
esu
ltin
g
in
th
e
r
e
g
is
ter
ed
im
ag
e.
2.
P
RO
P
O
SE
D
M
E
T
H
O
D
T
h
e
s
tu
d
y
in
tr
o
d
u
ce
s
a
n
o
v
el
ap
p
r
o
ac
h
f
o
r
MI
R
an
d
class
if
icatio
n
b
y
c
o
m
b
in
i
n
g
SAR
SO
with
a
DM
N.
T
h
e
k
ey
f
in
d
i
n
g
s
h
ig
h
li
g
h
t th
at
th
is
ap
p
r
o
ac
h
s
ig
n
if
ic
an
tly
im
p
r
o
v
es th
e
ac
cu
r
ac
y
o
f
im
ag
e
r
eg
is
tr
atio
n
an
d
class
if
icatio
n
co
m
p
ar
ed
to
tr
ad
itio
n
al
m
eth
o
d
s
.
T
h
e
in
teg
r
atio
n
o
f
SAR
SO
o
p
tim
izes
th
e
alig
n
m
en
t
p
r
o
ce
s
s
ef
f
ec
tiv
ely
,
wh
ile
th
e
DM
N
en
h
an
ce
s
class
if
icati
o
n
ac
cu
r
ac
y
[
1
8
]
.
Ad
d
itio
n
al
ly
,
th
e
m
eth
o
d
is
co
m
p
u
tatio
n
ally
ef
f
icien
t,
m
ak
in
g
it
s
u
itab
le
f
o
r
lar
g
e
-
s
c
ale
m
ed
ical
im
ag
in
g
task
s
.
T
h
e
ap
p
r
o
ac
h
also
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
7
,
No
.
3
,
Ma
r
ch
20
2
5
:
1
90
8
-
1
91
7
1910
d
em
o
n
s
tr
ates
r
o
b
u
s
tn
ess
ag
ain
s
t
n
o
is
e
an
d
im
a
g
e
v
ar
iatio
n
s
,
in
d
icatin
g
its
p
o
te
n
tial
f
o
r
r
eliab
le
clin
ical
ap
p
licatio
n
s
in
d
ia
g
n
o
s
is
an
d
tr
ea
tm
en
t
p
lan
n
i
n
g
.
M
I
R
tr
an
s
f
o
r
m
s
two
o
r
s
ev
er
al
g
r
o
u
p
s
o
f
an
im
a
g
in
g
d
a
ta
in
to
s
in
g
le
co
o
r
d
in
ate
s
ch
em
e
.
I
t
h
as
v
ital
p
ar
t
in
d
iag
n
o
s
is
,
s
u
r
g
ical
p
r
ep
ar
atio
n
to
r
ea
l
-
tim
e
ass
i
s
tan
ce
an
d
p
o
s
t
-
p
r
o
ce
d
u
r
al
an
al
y
s
is
.
I
n
th
is
wo
r
k
,
SAR
SO
-
DM
N
i
s
d
esig
n
ed
f
o
r
MI
R
an
d
class
if
icatio
n
.
T
h
e
two
in
p
u
t
im
ag
es
co
n
s
id
er
ed
ar
e
C
T
an
d
MRI
im
ag
es.
T
h
ese
two
in
p
u
ts
p
er
f
o
r
m
im
ag
e
r
eg
is
tr
ati
o
n
u
s
in
g
C
NN
an
d
th
en
d
ef
o
r
m
atio
n
f
ield
is
ap
p
lied
,
th
u
s
o
u
tp
u
t
-
1
is
ac
q
u
ir
ed
.
T
h
er
ea
f
ter
,
o
u
tp
u
t
-
1
an
d
in
p
u
t
MRI
im
ag
e
is
f
ed
to
C
NN
to
o
b
tain
o
u
tp
u
t
-
2
.
T
h
en
,
o
u
tp
u
t
-
2
an
d
in
p
u
t
M
R
I
im
ag
e
ar
e
g
iv
e
n
t
o
C
NN.
Fig
u
r
e
1
p
r
esen
ts
d
iag
r
am
m
atica
l illu
s
tr
atio
n
o
f
SAR
S
O
-
DM
N
f
o
r
MI
R
an
d
cl
ass
if
icatio
n
.
Fig
u
r
e
1
.
Diag
r
a
m
m
atica
l illu
s
tr
atio
n
o
f
SAR
SO
-
DM
N
f
o
r
MI
R
an
d
class
if
icatio
n
2.
1
.
Acquis
it
io
n o
f
inp
ut
M
RI a
nd
CT
im
a
g
e
An
in
p
u
t
MRI
an
d
C
T
im
ag
es
co
n
s
id
er
ed
f
o
r
MI
R
an
d
c
lass
if
icatio
n
ar
e
o
b
tain
ed
f
r
o
m
d
ef
in
ite
d
atab
ase
[
18
]
,
wh
ich
is
s
p
ec
if
i
ed
b
y
,
=
{
1
,
2
}
(
1
)
Her
e,
R
in
d
icate
s
d
atab
ase
co
m
p
r
is
es o
f
C
T
an
d
MRI
im
ag
es wh
er
ea
s
1
R
an
d
2
R
ca
n
b
e
g
iv
e
n
as,
1
=
{
1
,
2
,
.
.
.
,
,
.
.
.
,
}
(
2
)
2
=
{
1
,
2
,
.
.
.
,
,
.
.
.
,
}
(
3
)
w
h
er
e,
an
d
r
ep
r
esen
ts
ℎ
in
p
u
t
MRI
im
ag
e
an
d
o
v
er
all
MRI
im
ag
es
in
d
atab
ase
wh
er
ea
s
a
n
d
in
d
icate
s
ℎ
C
T
im
ag
e
an
d
to
tal
C
T
im
ag
es in
d
atab
ase.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
Med
ica
l ima
g
e
r
eg
is
tr
a
tio
n
a
n
d
cla
s
s
ifica
tio
n
u
s
in
g
s
mell
a
g
en
t
…
(
S
a
n
jeev
ku
ma
r
A
n
g
a
d
i
)
1911
2
.2
.
I
ma
g
e
re
g
is
t
ra
t
io
n us
ing
CNN
MI
R
is
a
p
r
im
ar
y
p
r
o
ce
s
s
in
g
s
tag
e
in
th
e
clin
ical
im
ag
e
p
r
o
ce
s
s
in
g
.
I
t
is
a
p
r
o
ce
s
s
o
f
as
s
ig
n
in
g
two
o
r
s
ev
er
al
in
p
u
t
im
ag
es
in
to
o
n
e
im
ag
e
th
at
o
f
f
e
r
m
u
ch
i
n
f
o
r
m
atio
n
.
T
h
e
r
eg
is
tr
atio
n
p
r
o
ce
d
u
r
e
in
v
o
lv
es
an
ev
alu
atio
n
o
f
o
p
tim
u
m
tr
an
s
f
o
r
m
atio
n
s
,
w
h
ich
s
u
p
e
r
io
r
ly
allo
ca
tes
o
b
jects
o
f
i
n
ter
est
in
in
p
u
t
im
ag
es.
Her
e,
MRI
an
d
C
T
im
ag
es si
g
n
if
ied
as
an
d
ar
e
tak
en
as in
p
u
t,
wh
er
e
in
p
u
t M
R
I
im
ag
e
is
f
ix
ed
im
ag
e.
I
n
itially
,
b
o
th
i
n
p
u
t
im
a
g
es
ar
e
f
ed
to
C
NN
an
d
th
e
n
d
ef
o
r
m
atio
n
f
ield
is
ap
p
lied
t
o
g
et
o
u
tp
u
t
-
1
d
en
o
ted
as
.
T
h
e
d
ef
o
r
m
atio
n
f
ield
[
19
]
f
r
o
m
ev
e
r
y
p
r
ev
i
o
u
s
co
ar
s
e
le
v
el
ar
e
u
tili
ze
d
as
p
r
elim
i
n
ar
y
s
u
p
p
o
s
itio
n
b
y
in
co
r
p
o
r
atin
g
th
em
b
y
m
ea
n
s
o
f
f
u
n
ctio
n
al
co
m
p
o
s
itio
n
as
well
as
war
p
a
m
o
v
in
g
im
ag
e
o
n
h
ig
h
r
eso
lu
tio
n
.
T
h
er
ea
f
ter
,
in
p
u
t
MRI
im
ag
e
an
d
f
ir
s
t
o
u
tp
u
t
s
p
ec
if
ied
b
y
an
d
ar
e
p
ass
ed
to
C
N
N.
T
h
u
s
,
o
u
tp
u
t
-
2
r
ep
r
esen
ted
b
y
is
o
b
tain
ed
.
Af
ter
war
d
s
,
an
d
ar
e
s
u
b
jecte
d
t
o
C
NN
to
ac
h
iev
e
r
eg
is
ter
ed
o
u
tp
u
t.
3.
M
E
T
H
O
D
C
NN
[
20
]
is
d
ef
in
e
d
as
a
ca
t
eg
o
r
y
o
f
ar
tific
ial
n
eu
r
al
n
et
wo
r
k
(
ANN)
,
wh
ich
n
ee
d
s
co
n
v
o
lu
tio
n
al
lay
er
an
d
also
co
n
s
is
ts
o
f
o
t
h
er
lay
er
s
lik
e
c
o
n
v
o
lu
tio
n
al
l
ay
er
,
f
u
lly
co
n
n
ec
ted
(
FC
)
an
d
p
o
o
lin
g
lay
e
r
s
to
f
o
r
m
d
ee
p
C
NN.
−
C
o
n
v
o
lu
tio
n
al
la
y
er
A
co
n
v
o
lu
tio
n
al
lay
e
r
in
ten
d
s
to
lear
n
th
e
f
ea
tu
r
e
d
ep
ictio
n
s
o
f
in
p
u
ts
.
T
h
is
lay
er
is
co
m
p
r
is
ed
o
f
v
ar
io
u
s
co
n
v
o
l
u
tio
n
al
k
er
n
els
th
at
ar
e
u
tili
ze
d
f
o
r
co
m
p
u
tin
g
n
u
m
er
o
u
s
f
ea
tu
r
e
m
ap
s
.
Par
ticu
lar
ly
,
in
d
i
v
id
u
al
n
eu
r
o
n
o
f
f
ea
tu
r
e
m
a
p
is
jo
in
e
d
to
a
n
ar
ea
o
f
n
e
ig
h
b
o
r
h
o
o
d
n
eu
r
o
n
s
in
p
r
io
r
lay
er
.
An
en
ti
r
e
f
ea
tu
r
e
m
ap
s
a
r
e
ac
q
u
ir
ed
u
tili
zin
g
v
ar
io
u
s
d
iv
er
s
e
k
er
n
els.
A
f
ea
tu
r
e
v
alu
e
at
p
o
s
itio
n
(
,
)
in
ℎ
f
ea
tu
r
e
m
ap
o
f
ℎ
lay
er
ca
n
b
e
m
ath
e
m
atica
lly
ca
lcu
la
ted
b
y
,
,
,
=
,
+
(
4
)
h
er
e,
an
d
in
d
icate
s
weig
h
t
v
ec
to
r
as
well
as
b
ias
o
f
ℎ
f
ilter
o
f
ℎ
lay
er
wh
er
ea
s
,
s
p
ec
if
ies
in
p
u
t
p
atch
th
at
is
ce
n
ter
ed
at
p
o
s
itio
n
(
,
)
of
ℎ
lay
er
.
−
Activ
atio
n
f
u
n
ctio
n
Activ
atio
n
f
u
n
ctio
n
p
r
esen
ts
n
o
n
-
lin
ea
r
ity
to
C
NN
th
at
a
r
e
d
esira
b
le
f
o
r
th
e
m
u
lti
-
lay
er
n
et
wo
r
k
s
f
o
r
d
etec
tin
g
n
o
n
-
lin
ea
r
f
ea
tu
r
es.
I
t c
an
b
e
ca
lcu
lated
as
(
5
)
.
,
,
=
(
,
,
)
(
5
)
W
h
er
e,
(
.
)
im
p
lies
n
o
n
-
lin
ea
r
ac
t
iv
atio
n
f
u
n
ctio
n
.
A
g
en
er
a
l
ac
tiv
atio
n
f
u
n
ctio
n
s
ar
e
n
am
ely
r
ec
tifie
d
lin
ea
r
u
n
it (
R
eL
U)
,
tan
h
a
n
d
s
ig
m
o
i
d
.
−
Po
o
lin
g
lay
er
T
h
is
lay
er
in
ten
d
s
f
o
r
ac
h
ie
v
in
g
s
h
if
t
-
in
v
ar
ian
ce
b
y
d
ec
r
ea
s
in
g
f
ea
tu
r
e
m
ap
s
r
eso
lu
tio
n
.
I
t
is
n
o
r
m
ally
l
o
ca
ted
am
o
n
g
two
c
o
n
v
o
lu
ti
o
n
al
lay
er
s
.
I
n
d
i
v
id
u
a
l
f
ea
tu
r
e
m
a
p
o
f
p
o
o
lin
g
la
y
er
is
ass
o
ciate
d
to
its
r
elate
d
f
ea
tu
r
e
m
ap
o
f
p
r
e
v
io
u
s
co
n
v
o
lu
tio
n
al
lay
er
.
T
h
e
p
o
o
lin
g
f
u
n
ctio
n
ca
n
b
e
r
e
p
r
esen
t
ed
as
(
6
)
,
,
,
=
(
,
,
)
,
∀
(
,
)
∈
,
(
6
)
h
er
e,
p
o
o
lin
g
f
u
n
ctio
n
is
d
en
o
ted
b
y
(
.
)
a
nd
,
s
p
ec
if
ies
lo
ca
l
n
eig
h
b
o
r
ac
r
o
s
s
p
o
s
itio
n
(
,
)
.
An
av
er
ag
e
p
o
o
lin
g
an
d
m
ax
-
p
o
o
lin
g
ar
e
co
m
m
o
n
p
o
o
lin
g
f
u
n
ctio
n
s
.
−
FC
lay
er
Fo
llo
win
g
v
ar
io
u
s
co
n
v
o
lu
ti
o
n
al
lay
er
s
a
n
d
p
o
o
lin
g
lay
er
s
,
C
NN
s
tr
u
ctu
r
e
co
n
s
is
ts
o
f
o
n
e
o
r
n
u
m
er
o
u
s
FC
lay
er
s
th
at
in
ten
d
s
to
ex
ec
u
te
h
ig
h
er
-
le
v
el
r
ea
s
o
n
in
g
.
T
h
e
FC
lay
er
s
tak
e
ev
er
y
n
eu
r
o
n
in
p
r
io
r
lay
er
an
d
lin
k
th
em
t
o
all
s
in
g
l
e
n
eu
r
o
n
o
f
p
r
esen
t la
y
e
r
f
o
r
g
en
er
atin
g
g
l
o
b
al
s
em
an
tic
in
f
o
r
m
atio
n
.
−
Ou
tp
u
t la
y
er
T
h
e
last
C
NN
lay
er
is
o
u
tp
u
t la
y
er
an
d
So
f
tMa
x
f
u
n
ctio
n
is
g
en
er
ally
u
tili
ze
d
f
o
r
s
ev
e
r
al
task
s
.
3
.3
.
I
nh
a
le
a
nd
ex
ha
le
cla
s
s
if
ica
t
io
n ut
ilizin
g
DM
N
T
h
e
in
h
ale
an
d
ex
h
ale
clas
s
if
icatio
n
ar
e
p
er
f
o
r
m
ed
f
o
r
v
alid
atin
g
a
d
e
f
o
r
m
a
b
le
r
e
g
is
tr
atio
n
.
Her
e,
DM
N
is
u
tili
ze
d
t
o
ca
r
r
y
o
u
t
in
h
ale
an
d
ex
h
ale
p
e
r
f
o
r
m
an
ce
b
y
c
o
n
ce
r
n
in
g
as
in
p
u
t.
T
h
e
t
u
n
in
g
o
f
DM
N
is
ac
co
m
p
lis
h
ed
b
y
S
AR
SO,
wh
ich
is
f
o
r
m
ed
b
y
j
o
in
in
g
s
m
ell
ag
e
n
t
o
p
tim
izati
o
n
(
SAO
)
with
r
at
s
war
m
o
p
tim
izer
(
R
SO
)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
7
,
No
.
3
,
Ma
r
ch
20
2
5
:
1
90
8
-
1
91
7
1912
3
.
3
.
1
.
Arc
hite
ct
ure
o
f
DM
N
Ma
x
o
u
t
m
o
d
el
r
ef
e
r
s
to
f
ee
d
-
f
o
r
war
d
co
n
f
ig
u
r
atio
n
n
a
m
ely
d
ee
p
C
NN
o
r
else
m
u
ltil
ay
er
p
er
ce
p
tr
o
n
’
s
th
at
m
ak
e
u
s
e
o
f
n
ewe
r
v
ar
iety
o
f
an
ac
tiv
atio
n
f
u
n
ctio
n
s
p
ec
if
ied
as
Ma
x
o
u
t
u
n
it.
DM
N
[
21
]
,
[
22
]
c
o
n
tain
s
n
u
m
er
o
u
s
lay
er
s
th
at
g
en
er
ate
h
id
d
en
ac
tiv
atio
n
s
u
s
in
g
Ma
x
o
u
t
f
u
n
ctio
n
.
−
R
eL
U
I
n
itially
,
R
eL
U
is
en
g
ag
ed
in
t
h
e
r
estricte
d
B
o
ltzm
an
n
m
ac
h
in
es
(
R
B
M)
th
at
ca
n
b
e
illu
s
tr
ated
as,
=
{
,
≥
0
0
,
<
0
(
7
)
h
er
e,
s
y
m
b
o
lizes in
p
u
t su
b
jec
ted
to
n
eu
r
o
n
a
n
d
d
en
o
tes o
u
t
p
u
t.
−
Ma
x
o
u
t
I
t
is
an
ac
tu
al
s
o
r
t
o
f
R
eL
U
o
p
er
atio
n
t
h
at
o
b
tain
s
m
a
x
f
u
n
ctio
n
o
n
(
=
2
)
tr
ain
a
b
le
lin
e
ar
f
u
n
ctio
n
s
.
Ass
u
m
e
p
ar
ticu
lar
i
n
p
u
t
∈
,
wh
er
e
im
p
lies
s
tate
v
e
cto
r
o
r
else
r
aw
in
p
u
t
v
ec
t
o
r
o
f
a
h
id
d
en
lay
er
an
d
Ma
x
o
u
t
u
n
it o
u
tp
u
t
ca
n
b
e
f
o
r
m
u
lated
b
y
,
(
)
=
∈
[
1
,
]
(
8
)
w
h
er
e,
=
.
.
.
+
,
∈
×
×
s
ig
n
if
ies
tr
ain
ab
le
p
a
r
am
eter
s
wh
er
ea
s
s
ig
n
if
ies
co
u
n
t
o
f
lin
ea
r
s
u
b
-
h
id
d
e
n
u
n
its
.
A
Ma
x
o
u
t
u
n
it
ac
tiv
atio
n
b
alan
ce
s
u
tter
m
o
s
t
ac
r
o
s
s
f
ea
tu
r
e
m
ap
s
in
C
NN.
Fu
r
th
er
m
o
r
e,
Ma
x
o
u
t
u
n
it
is
an
alo
g
o
u
s
to
n
o
r
m
ally
e
m
p
lo
y
i
n
g
s
p
atial
m
ax
-
p
o
o
lin
g
i
n
C
NN
co
n
f
ig
u
r
atio
n
.
T
h
is
attain
s
h
ig
h
er
v
alu
e
m
o
r
e
th
an
eq
u
al
i
n
p
u
t,
wh
e
r
e
s
p
atial
m
ax
-
p
o
o
li
n
g
is
ass
o
ciate
d
to
n
u
m
e
r
o
u
s
i
n
p
u
ts
.
−
DM
N
T
h
is
is
th
e
s
o
r
t
o
f
tr
ain
a
b
le
ac
tiv
atio
n
f
u
n
ctio
n
with
n
u
m
er
o
u
s
lay
er
ar
r
an
g
em
en
ts
.
Fo
r
a
n
in
p
u
t
∈
,
h
id
d
e
n
u
n
it a
ctiv
atio
n
is
m
o
d
eled
as,
,
1
=
∈
[
1
,
1
]
.
.
.
+
(
9
)
,
2
=
∈
[
1
,
2
]
,
1
.
.
.
+
(
1
0
)
,
=
∈
[
1
,
]
,
−
1
.
.
.
+
(
1
1
)
,
=
∈
[
1
,
]
,
−
1
.
.
.
+
(
1
2
)
=
∈
[
1
,
]
,
(
1
3
)
h
er
e,
im
p
lies
to
tal
u
n
its
in
ℎ
lay
er
an
d
s
ig
n
if
ies en
tire
lay
er
s
c
o
n
s
is
ted
in
DM
N.
T
h
e
DM
N
a
ctiv
atio
n
is
m
u
ch
s
tr
o
n
g
er
f
o
r
e
v
alu
atin
g
an
ar
b
itra
r
y
r
ec
u
r
r
en
t
ac
t
iv
atio
n
f
u
n
ctio
n
e
x
ce
p
t
n
o
n
-
co
n
v
e
x
ac
tiv
atio
n
o
p
er
atio
n
s
.
A
co
n
v
en
tio
n
al
n
o
n
-
lin
ea
r
ac
tiv
atio
n
f
u
n
ctio
n
s
s
u
ch
as
r
ec
tifie
d
lin
ea
r
an
d
an
a
b
s
o
lu
te
v
alu
e
r
ec
tifie
r
ar
e
ass
ess
ed
b
y
DM
N
wh
en
is
n
o
t
lo
wer
th
an
2
.
Ho
wev
er
,
an
ef
f
ec
t
u
al
f
ea
tu
r
e
ex
t
r
ac
tio
n
r
e
q
u
ir
es a
lar
g
e
am
o
u
n
t
d
if
f
icu
lt
n
o
n
-
lin
ea
r
o
p
er
atio
n
.
DM
N
is
ca
p
ab
le
f
o
r
esti
m
atin
g
an
ar
b
itra
r
y
a
ctiv
atio
n
o
p
er
atio
n
d
u
r
in
g
p
ar
am
eter
au
g
m
e
n
tatio
n
.
3
.
3
.
2
.
T
ra
ini
ng
o
f
DM
N
utili
zing
SARSO
SAO
[
23
]
u
tili
ze
s
p
h
en
o
m
en
a
o
f
th
e
s
m
ell
an
d
an
in
s
tin
ctiv
e
tr
ailin
g
ch
ar
ac
ter
is
tic
o
f
ag
en
t
f
o
r
id
en
tify
in
g
a
s
m
ellin
g
s
o
u
r
c
e.
I
t
was
p
r
o
v
e
n
as
ef
f
ec
ti
v
e
ap
p
r
o
ac
h
i
n
r
eso
lv
i
n
g
u
n
i
m
o
d
al
as
well
as
m
u
ltimo
d
a
l
o
p
er
atio
n
s
.
R
SO
[
24
]
is
d
esig
n
ed
to
s
o
lv
e
c
h
al
len
g
in
g
o
p
tim
izatio
n
is
s
u
es.
A
m
ajo
r
s
tim
u
latio
n
o
f
th
is
ap
p
r
o
ac
h
is
attac
k
in
g
an
d
ch
asin
g
ch
ar
a
cter
is
tics
o
f
r
ats
n
atu
r
ally
.
T
h
is
alg
o
r
ith
m
h
as
h
ig
h
er
p
er
f
o
r
m
an
ce
ab
ilit
y
in
u
n
id
e
n
tifie
d
s
ea
r
ch
s
p
ac
es.
Her
e,
SA
O
an
d
R
SO
ar
e
in
co
r
p
o
r
ated
t
o
f
o
r
m
SAR
SO
th
at
is
an
ef
f
ec
tu
al
ap
p
r
o
ac
h
to
tu
n
e
DM
N
f
o
r
in
h
ale
a
n
d
e
x
h
ale
class
if
icatio
n
.
A
f
itn
ess
co
m
p
u
tatio
n
is
d
o
n
e
b
y
id
en
tif
y
in
g
a
v
a
r
iatio
n
am
id
tar
g
eted
o
u
tp
u
t
an
d
D
MN
o
u
tp
u
t,
wh
ich
is
ca
lcu
lated
as,
ƛ
=
1
∑
[
−
]
2
=
1
(
1
4
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
Med
ica
l ima
g
e
r
eg
is
tr
a
tio
n
a
n
d
cla
s
s
ifica
tio
n
u
s
in
g
s
mell
a
g
en
t
…
(
S
a
n
jeev
ku
ma
r
A
n
g
a
d
i
)
1913
w
h
er
e,
i
s
ig
n
if
ies o
v
er
all
C
T
an
d
MRI
im
ag
es wh
er
ea
s
an
d
r
ep
r
esen
ts
tar
g
eted
an
d
DM
N
o
u
tp
u
ts
.
SAR
S
O
p
er
f
o
r
m
s
b
en
ea
th
elu
cid
ated
s
tep
s
to
g
et
s
u
p
er
io
r
s
o
lu
tio
n
.
−
Step
1
: I
n
itialize
th
e
s
o
lu
tio
n
SAO
p
r
o
ce
s
s
is
in
itiated
b
y
g
e
n
er
atin
g
in
itial
lo
ca
tio
n
(
p
o
p
u
l
atio
n
)
o
f
th
e
s
m
ell
m
o
lecu
les
r
an
d
o
m
ly
.
T
h
e
in
itializatio
n
p
r
o
ce
s
s
ca
n
b
e
f
o
r
m
u
lated
as,
ℜ
=
{
ℜ
1
,
ℜ
2
,
.
.
.
,
ℜ
,
.
.
.
,
ℜ
}
(
1
5
)
w
h
er
e,
ℜ
r
ep
r
esen
ts
to
tal
v
ar
iab
les,
ℜ
d
en
o
tes
ℎ
ca
n
d
id
ate
s
o
lu
tio
n
an
d
ℜ
s
p
ec
if
ies p
o
p
u
latio
n
.
−
Step
2
: Calcu
latio
n
o
f
o
b
jectiv
e
f
u
n
ctio
n
T
h
e
ca
lcu
latio
n
o
f
o
b
jectiv
e
f
u
n
ctio
n
is
p
er
f
o
r
m
ed
b
y
c
o
m
p
u
tin
g
d
if
f
er
e
n
ce
am
id
tar
g
eted
an
d
DM
N
o
u
tp
u
ts
u
s
in
g
(
1
4
)
.
−
Step
3
: Sn
if
f
in
g
m
eth
o
d
Ass
u
m
e
s
m
ell
m
o
lecu
les
s
i
g
n
if
ied
as
an
d
h
y
p
er
s
p
ac
e,
w
h
er
ein
s
m
ell
m
o
lecu
les
ar
e
ev
a
p
o
r
atin
g
is
g
iv
en
b
y
.
T
h
e
s
m
ell
m
o
lecu
le
s
p
o
p
u
latio
n
is
allo
ca
ted
a
lo
c
atio
n
as,
ℜ
=
[
ℜ
,
1
,
ℜ
,
2
,
.
.
.
,
ℜ
,
]
(
1
6
)
w
h
er
e,
=
{
1
,
2
,
.
.
.
,
}
an
d
in
d
icate
s
s
m
ell
m
o
lecu
les
p
o
s
itio
n
.
A
v
elo
city
v
ec
to
r
T
is
co
n
s
id
er
ed
as
d
if
f
u
s
io
n
v
ec
to
r
th
at
is
th
e
d
is
p
lace
m
en
t
o
f
s
m
ell
m
o
lecu
les
f
r
o
m
s
m
ell
s
o
u
r
ce
o
r
o
r
ig
in
.
T
h
e
s
m
ell
m
o
lecu
les v
elo
city
ca
n
b
e
co
m
p
u
ted
b
y
(
1
7
)
.
=
[
,
1
,
,
2
,
.
.
.
,
,
]
(
1
7
)
T
h
e
lo
ca
tio
n
o
f
in
d
iv
i
d
u
al
ca
n
d
id
ate
s
o
lu
tio
n
ca
n
b
e
d
eter
m
in
ed
b
y
lo
ca
tio
n
v
ec
t
o
r
∈
an
d
m
o
lecu
le
v
elo
city
∈
.
As
th
e
s
m
ell
m
o
lecu
les
m
o
v
em
en
ts
ar
e
n
o
n
-
u
n
if
o
r
m
in
s
ix
-
d
im
en
s
io
n
al
co
o
r
d
in
ates,
a
v
elo
city
ca
n
b
e
illu
s
tr
ated
b
y
,
(
,
,
)
=
(
,
,
)
(
1
8
)
h
er
e,
(
,
,
)
im
p
lies
d
is
p
lace
m
en
t,
in
d
icate
s
ch
an
g
e
in
tim
e
an
d
(
,
,
)
s
p
ec
if
ies v
elo
city
.
(
,
,
)
=
(
,
,
)
×
(
1
9
)
T
h
u
s
,
a
g
en
er
alize
d
f
o
r
m
o
f
ab
o
v
e
eq
u
atio
n
ca
n
b
e
m
o
d
eled
as,
(
,
,
)
(
)
+
1
−
(
,
,
)
=
(
,
,
)
×
(
2
0
)
t
h
e
af
o
r
em
e
n
tio
n
ed
eq
u
atio
n
c
an
b
e
r
ea
r
r
an
g
e
d
as,
(
,
,
)
(
)
+
1
=
(
,
,
)
×
+
(
,
,
)
(
2
1
)
h
en
ce
,
in
ab
o
v
e
eq
u
atio
n
s
is
tak
en
as 1
.
No
w,
s
u
b
s
titu
te
in
ab
o
v
e
e
q
u
atio
n
a
n
d
th
u
s
eq
u
at
io
n
b
ec
o
m
es,
(
,
,
)
(
)
+
1
=
(
,
,
)
+
(
,
,
)
(
2
2
)
h
er
e,
(
,
,
)
an
d
(
,
,
)
ar
e
lo
ca
tio
n
a
n
d
v
el
o
city
co
o
r
d
in
ates o
f
s
m
ell
m
o
l
ec
u
les in
s
ea
r
ch
s
p
ac
e.
−
Step
4
: Sm
ell
v
elo
city
u
p
d
atin
g
an
d
l
o
ca
tio
n
All
s
m
ell
m
o
lecu
les
co
n
s
is
t
o
f
r
elat
ed
v
el
o
cities,
wh
er
ein
t
h
ey
m
o
v
e
an
d
u
p
d
ates
th
eir
l
o
ca
tio
n
in
s
ea
r
ch
s
p
ac
e.
As
g
as
m
o
lec
u
les
tr
av
el
in
n
o
n
-
u
n
if
o
r
m
it
y
m
an
n
er
,
th
e
g
as
v
el
o
city
is
ac
q
u
ir
ed
f
r
o
m
h
y
d
r
o
s
tatic
p
r
ess
u
r
e
o
f
g
as th
a
t c
an
b
e
g
i
v
en
b
y
,
=
1
2
(
2
)
=
3
2
(
)
(
2
3
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
7
,
No
.
3
,
Ma
r
ch
20
2
5
:
1
90
8
-
1
91
7
1914
h
er
e,
d
en
o
tes
q
u
an
tity
o
f
s
u
b
s
tan
ce
in
g
as.
T
h
u
s
,
v
elo
city
o
f
n
o
n
-
u
n
if
o
r
m
ity
g
as
m
o
lecu
les
in
s
ea
r
ch
s
p
ac
e
is
f
o
r
m
u
lated
b
y
,
(
+
1
)
=
√
3
(
2
4
)
f
r
o
m
(
2
4
)
,
s
m
ell
m
o
lecu
les v
e
lo
city
is
u
p
d
ated
as,
(
+
1
)
=
+
0
×
√
3
(
2
5
)
−
Step
5
: T
r
ailin
g
m
eth
o
d
Fo
r
an
ag
en
t
t
o
ef
f
ec
tiv
ely
tr
a
il
a
s
m
ell
p
ath
,
ag
en
t
m
u
s
t
b
e
ca
p
ab
le
to
k
ee
p
in
m
in
d
ab
o
u
t
m
o
lecu
le
p
o
s
itio
n
with
f
in
est
s
m
ell
atte
n
tiv
en
ess
an
d
lo
ca
tio
n
with
w
o
r
s
e
atten
tiv
en
ess
o
f
s
m
ell
[
2
5
]
.
An
ag
en
t
u
tili
ze
s
th
is
in
f
o
r
m
atio
n
wh
ile
tr
ain
in
g
to
e
v
ad
e
m
o
v
in
g
to
war
d
s
a
d
o
m
ain
with
p
o
o
r
f
itn
ess
.
A
tr
ailin
g
ch
ar
ac
ter
is
tic
o
f
an
a
g
en
t c
an
b
e
m
ath
e
m
atica
lly
m
o
d
eled
as
,
ℜ
(
+
1
)
=
ℜ
+
1
×
×
(
ℜ
(
)
−
ℜ
(
)
)
−
2
×
×
(
ℜ
(
)
−
ℜ
(
)
)
(
2
6
)
ℜ
(
+
1
)
=
ℜ
[
1
−
1
×
+
2
×
]
+
1
×
×
ℜ
(
)
−
2
×
×
ℜ
(
)
(
2
7
)
t
h
e
s
tan
d
ar
d
e
q
u
atio
n
o
f
R
SO c
an
b
e
m
o
d
eled
b
y
,
→
(
+
1
)
=
|
→
(
)
−
→
|
(
2
8
)
−
Step
6
: T
er
m
in
atio
n
T
h
e
p
r
o
ce
s
s
ab
o
v
e
m
e
n
tio
n
ed
f
o
r
SAR
SO p
er
f
o
r
m
s
co
n
ti
n
u
a
lly
u
n
til o
p
tim
u
m
s
o
lu
tio
n
is
o
b
tain
ed
.
4.
RE
SU
L
T
S
AND
D
I
SCU
SS
I
O
N
T
h
e
b
est
ac
h
iev
e
m
en
ts
attain
ed
b
y
SAR
SO
-
DM
N
f
o
r
MI
R
an
d
class
if
icatio
n
ar
e
ex
p
o
u
n
d
e
d
in
b
elo
w
s
u
b
-
s
ec
tio
n
s
.
4
.
1
.
E
x
perim
ent
a
t
io
n set
up
An
ex
p
er
im
en
tal
ex
ec
u
tio
n
o
f
th
e
SA
R
SO
-
DM
N
f
o
r
MI
R
an
d
class
if
icatio
n
was
ca
r
r
ied
o
u
t
u
s
in
g
Py
th
o
n
.
Mu
ltimo
d
al
g
r
o
u
n
d
tr
u
th
d
atasets
f
o
r
ab
d
o
m
in
al
MI
R
wer
e
u
tili
ze
d
in
th
is
s
tu
d
y
.
T
h
e
d
ataset
in
clu
d
es
T
1
-
weig
h
te
d
MRI,
C
T
,
an
d
co
n
e
b
e
am
(
C
B
)
C
T
im
ag
es
th
at
ar
e
in
tr
in
s
ically
co
-
r
e
g
is
ter
ed
.
T
h
e
MRI,
C
T
,
an
d
C
B
C
T
im
ag
es
wer
e
p
r
o
v
i
d
ed
in
.
zi
p
f
ile
f
o
r
m
at
an
d
wer
e
ex
tr
ac
te
d
f
o
r
u
s
e
in
th
e
ex
p
er
i
m
en
ts
.
T
h
ese
d
atasets
p
r
o
v
id
e
g
r
o
u
n
d
tr
u
th
f
o
r
im
a
g
e
r
e
g
is
tr
atio
n
an
d
s
eg
m
en
tatio
n
,
m
ak
i
n
g
th
em
i
d
ea
l
f
o
r
ev
alu
atin
g
th
e
p
er
f
o
r
m
an
ce
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
.
T
h
e
.
zip
f
iles
co
n
tain
in
g
th
e
MR
I
,
C
T
,
an
d
C
B
C
T
im
ag
es
wer
e
u
n
zip
p
ed
,
an
d
th
e
im
ag
es
wer
e
lo
ad
ed
in
to
th
e
Py
th
o
n
en
v
ir
o
n
m
e
n
t.
T
h
e
in
t
en
s
ity
v
alu
es
o
f
th
e
im
ag
es
wer
e
n
o
r
m
alize
d
to
e
n
s
u
r
e
u
n
if
o
r
m
it
y
an
d
to
en
h
a
n
ce
th
e
p
er
f
o
r
m
a
n
ce
o
f
th
e
n
eu
r
al
n
etwo
r
k
s
.
All
im
ag
es we
r
e
r
esized
to
a
u
n
if
o
r
m
d
im
en
s
io
n
s
u
itab
le
f
o
r
t
h
e
C
NN
in
p
u
t la
y
er
.
4.
2
.
Co
m
pa
ra
t
iv
e
a
na
ly
s
is
Dee
p
lear
n
in
g
[
1
]
,
PAC
N
[
3
]
,
d
ef
o
r
m
ab
le
d
ee
p
lear
n
i
n
g
[
5
]
,
an
d
u
n
s
u
p
er
v
is
ed
d
ee
p
lear
n
in
g
[
7
]
a
r
e
th
e
co
m
p
ar
ativ
e
m
et
h
o
d
s
ta
k
en
in
to
co
n
s
id
er
atio
n
to
ev
a
lu
ate
SAR
SO
-
DM
N
to
s
h
o
w
its
ef
f
ec
tiv
en
ess
.
SAR
S
O
-
DM
N
p
er
f
o
r
m
s
co
m
p
ar
ativ
e
ass
ess
m
en
t
b
y
v
ar
y
in
g
tr
ain
in
g
d
ata
a
n
d
k
-
v
alu
e
co
n
ce
r
n
in
g
e
v
alu
atio
n
m
ea
s
u
r
es.
Fig
u
r
e
2
ex
p
o
u
n
d
s
o
n
t
h
e
ev
al
u
atio
n
o
f
t
h
e
SAR
SO
-
DM
N
b
y
v
ar
y
in
g
th
e
p
e
r
ce
n
tag
e
o
f
tr
ain
i
n
g
d
ata
an
d
co
n
s
id
er
in
g
v
a
r
io
u
s
p
er
f
o
r
m
an
ce
m
ea
s
u
r
es.
T
h
e
v
alu
es
ac
q
u
ir
ed
b
y
th
e
d
e
v
is
ed
SAR
SO
-
DM
N
f
o
r
9
0
%
o
f
th
e
tr
ain
in
g
d
ata
ar
e
d
etailed
b
elo
w.
Fig
u
r
e
2
(
a
)
d
elin
ea
tes
th
e
ass
e
s
s
m
en
t
o
f
SAR
SO
-
DM
N
co
n
s
id
er
in
g
ac
c
u
r
ac
y
.
SAR
S
O
-
DM
N
attain
ed
an
ac
cu
r
a
cy
o
f
0
.
9
0
7
,
wh
er
ea
s
d
ee
p
lear
n
in
g
,
PAC
N,
d
ef
o
r
m
a
b
le
d
ee
p
lear
n
in
g
,
an
d
u
n
s
u
p
er
v
is
ed
d
ee
p
lear
n
in
g
ac
h
iev
ed
ac
cu
r
ac
ies
o
f
0
.
8
1
3
,
0
.
8
3
5
,
0
.
8
6
1
,
an
d
0
.
8
8
7
,
r
esp
ec
tiv
el
y
.
T
h
is
r
ev
e
als
a
p
er
f
o
r
m
a
n
ce
en
h
an
ce
m
e
n
t
o
f
1
0
.
3
7
2
%,
7
.
9
4
6
%,
5
.
0
0
2
%,
an
d
2
.
2
1
1
%
b
y
SAR
S
O
-
DM
N
o
v
er
th
ese
m
eth
o
d
s
.
T
h
e
h
ig
h
er
ac
c
u
r
ac
y
i
n
d
icate
s
th
at
SAR
SO
-
D
MN
i
s
m
o
r
e
ef
f
ec
tiv
e
i
n
co
r
r
ec
tly
class
if
y
in
g
th
e
im
ag
es
th
an
th
e
o
th
er
ap
p
r
o
ac
h
es.
E
v
alu
atio
n
o
f
SAR
SO
-
DM
N
ab
o
u
t
f
alse
p
o
s
itiv
e
r
ate
(
FPR
)
is
s
h
o
wn
in
Fig
u
r
e
2
(
b
)
.
T
h
e
FP
R
ac
h
iev
ed
b
y
SAR
SO
-
DM
N
is
0
.
1
1
3
,
wh
ile
t
h
e
FP
R
s
ac
q
u
ir
ed
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
Med
ica
l ima
g
e
r
eg
is
tr
a
tio
n
a
n
d
cla
s
s
ifica
tio
n
u
s
in
g
s
mell
a
g
en
t
…
(
S
a
n
jeev
ku
ma
r
A
n
g
a
d
i
)
1915
b
y
d
ee
p
lear
n
in
g
,
PAC
N,
d
ef
o
r
m
ab
le
d
ee
p
lear
n
in
g
,
a
n
d
u
n
s
u
p
er
v
is
ed
d
ee
p
lear
n
in
g
ar
e
0
.
2
2
3
,
0
.
2
0
9
,
0
.
1
7
9
,
an
d
0
.
1
3
1
,
r
esp
ec
tiv
el
y
.
T
h
is
lo
wer
FP
R
f
o
r
SA
R
SO
-
D
MN
s
ig
n
if
ies
a
r
ed
u
ctio
n
in
n
eg
ativ
e
in
s
tan
ce
s
in
co
r
r
ec
tly
class
if
ied
as
p
o
s
itiv
e,
h
ig
h
lig
h
ti
n
g
its
r
eliab
i
lity
in
m
in
im
izin
g
f
alse
alar
m
s
.
Fig
u
r
e
2
(
c
)
d
em
o
n
s
tr
ates
th
e
ass
es
s
m
en
t
o
f
SAR
S
O
-
DM
N
ab
o
u
t
tr
u
e
p
o
s
itiv
e
r
ate
(
T
P
R
)
.
SA
R
SO
-
D
MN
o
b
tain
ed
a
T
PR
o
f
0
.
9
1
2
,
wh
e
r
ea
s
d
ee
p
lear
n
i
n
g
,
PAC
N,
d
ef
o
r
m
ab
le
d
ee
p
l
ea
r
n
in
g
,
a
n
d
u
n
s
u
p
er
v
is
ed
d
ee
p
lear
n
in
g
ac
h
iev
ed
T
PR
s
o
f
0
.
8
1
7
,
0
.
8
3
7
,
0
.
8
6
7
,
an
d
0
.
8
8
4
,
r
esp
ec
tiv
ely
.
T
h
is
r
ep
r
esen
ts
a
p
er
f
o
r
m
an
ce
im
p
r
o
v
em
e
n
t
o
f
1
0
.
4
8
4
%,
8
.
2
9
2
%,
5
.
0
0
4
%,
a
n
d
3
.
1
2
9
%
b
y
SAR
SO
-
DM
N.
T
h
e
h
ig
h
er
T
PR
in
d
icate
s
th
at
SAR
SO
-
DM
N
i
s
m
o
r
e
ef
f
ec
tiv
e
in
co
r
r
ec
tly
id
en
tify
in
g
p
o
s
itiv
e
in
s
tan
ce
s
,
e
n
h
an
cin
g
its
u
tili
ty
in
cli
n
ical
ap
p
licatio
n
s
wh
e
r
e
ac
cu
r
ate
p
o
s
itiv
e
d
etec
tio
n
is
cr
u
cial.
T
h
e
r
esu
lts
d
ep
icted
in
Fig
u
r
e
2
illu
s
t
r
ate
th
e
s
u
p
er
io
r
p
er
f
o
r
m
an
ce
o
f
SAR
SO
-
DM
N
in
ter
m
s
o
f
ac
cu
r
ac
y
,
FP
R
,
an
d
T
P
R
co
m
p
ar
ed
to
o
th
er
s
tate
-
of
-
th
e
-
a
r
t
m
eth
o
d
s
.
T
h
e
s
ig
n
if
ican
t
im
p
r
o
v
em
e
n
ts
in
th
ese
m
etr
ics
d
em
o
n
s
tr
ate
th
e
ef
f
ec
tiv
en
ess
o
f
SAR
SO
-
DM
N
in
MI
R
an
d
class
if
icatio
n
task
s
.
T
h
e
class
i
f
icatio
n
p
h
ase
em
p
l
o
y
s
a
DM
N
tu
n
ed
b
y
th
e
SAR
SO
alg
o
r
ith
m
,
wh
ich
is
a
h
y
b
r
id
o
f
SAO
an
d
R
SO.
T
h
is
n
o
v
el
o
p
tim
iz
atio
n
ap
p
r
o
ac
h
en
h
an
ce
s
th
e
DM
N
’
s
p
er
f
o
r
m
an
ce
in
class
i
f
icatio
n
task
s
.
T
h
e
k
ey
f
i
n
d
in
g
s
o
f
th
is
wo
r
k
u
n
d
er
s
co
r
e
th
e
s
u
b
s
tan
tial
im
p
r
o
v
em
en
t
in
ac
cu
r
ac
y
an
d
r
eliab
il
ity
ac
h
iev
ed
b
y
th
e
SAR
S
O
-
DM
N
ap
p
r
o
ac
h
.
I
t
att
ain
s
a
m
ax
im
u
m
ac
cu
r
ac
y
o
f
9
0
.
7
%,
a
FP
R
o
f
1
1
.
3
%,
an
d
a
m
ax
im
u
m
T
PR
o
f
9
1
.
2
%.
T
h
ese
r
esu
lts
clea
r
ly
o
u
tp
er
f
o
r
m
e
x
is
tin
g
m
eth
o
d
s
,
d
em
o
n
s
tr
atin
g
th
e
p
o
ten
tial
o
f
th
e
SAR
SO
-
DM
N
ap
p
r
o
ac
h
in
th
e
f
ield
o
f
m
ed
ic
al
im
ag
e
p
r
o
ce
s
s
in
g
an
d
class
if
icatio
n
.
T
h
e
SAR
SO
-
DM
N
ap
p
r
o
ac
h
p
r
esen
ts
a
s
u
b
s
tan
tial
ad
v
an
ce
m
e
n
t
in
th
e
f
ield
o
f
MI
R
an
d
class
if
icatio
n
.
I
n
teg
r
atin
g
a
d
v
an
ce
d
o
p
ti
m
izatio
n
tech
n
iq
u
es
an
d
d
ee
p
lear
n
in
g
m
o
d
els
o
f
f
er
s
a
p
r
o
m
is
in
g
s
o
lu
tio
n
f
o
r
en
h
an
cin
g
th
e
p
r
ec
is
io
n
a
n
d
ef
f
ec
tiv
en
ess
o
f
m
ed
ical
im
ag
e
p
r
o
ce
s
s
in
g
,
p
o
t
en
tially
lead
in
g
to
b
etter
d
iag
n
o
s
tic
ac
cu
r
ac
y
an
d
p
atien
t o
u
t
co
m
es.
(
a)
(
b
)
(
c)
Fig
u
r
e
2
.
C
o
m
p
a
r
ativ
e
ev
alu
at
io
n
co
n
ce
r
n
in
g
tr
ain
in
g
d
ata
;
(
a)
a
cc
u
r
ac
y
,
(
b
)
FP
R
,
an
d
(
c)
T
PR
5.
CO
NCLU
SI
O
N
I
n
r
ec
en
t d
ay
s
,
MI
R
h
as b
ec
o
m
e
a
v
alu
ab
le
ass
is
tan
t f
o
r
cli
n
ical
s
p
ec
ialis
t
s
.
I
ts
s
ig
n
if
ican
ce
lies
in
it
s
ab
ilit
y
to
aid
in
d
iag
n
o
s
is
,
m
o
n
ito
r
th
e
p
r
o
g
r
ess
io
n
o
f
d
is
ea
s
es,
an
d
s
elec
t
ap
p
r
o
p
r
iate
th
er
ap
ies
b
ased
o
n
th
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
7
,
No
.
3
,
Ma
r
ch
20
2
5
:
1
90
8
-
1
91
7
1916
p
atien
t
’
s
co
n
d
itio
n
.
I
m
a
g
e
-
b
as
ed
r
eg
is
tr
atio
n
ap
p
r
o
ac
h
es th
at
ag
g
r
eg
ate
d
ata
f
r
o
m
m
o
r
e
s
ig
n
if
ican
t p
ar
ts
o
f
an
im
ag
e
to
en
h
a
n
ce
ac
c
u
r
ac
y
ar
e
co
m
p
u
tatio
n
ally
in
ten
s
iv
e
a
n
d
o
f
ten
s
u
f
f
er
p
er
f
o
r
m
a
n
ce
d
eg
r
ad
atio
n
.
I
n
th
is
co
n
tex
t,
th
e
SS
AR
SO
-
DM
N
is
in
tr
o
d
u
ce
d
f
o
r
MI
R
an
d
cl
ass
if
icatio
n
.
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
in
v
o
lv
es
two
in
p
u
t
im
ag
es,
s
u
ch
as
C
T
an
d
MRI
im
ag
es,
f
r
o
m
th
e
d
at
aset.
T
h
e
im
ag
e
r
eg
is
tr
atio
n
p
r
o
ce
s
s
b
eg
in
s
b
y
p
ass
in
g
th
ese
in
p
u
t im
ag
es in
to
a
C
NN
an
d
th
en
ap
p
ly
i
n
g
a
d
ef
o
r
m
atio
n
f
ield
to
o
b
tain
an
in
ter
m
e
d
iate
o
u
tp
u
t
(
o
u
tp
u
t
-
1
)
.
T
h
is
o
u
tp
u
t
an
d
t
h
e
in
p
u
t
MRI
im
ag
es
ar
e
th
en
f
ed
b
ac
k
in
to
a
C
NN
to
p
r
o
d
u
ce
o
u
tp
u
t
-
2
.
Fin
ally
,
o
u
tp
u
t
-
2
an
d
th
e
in
p
u
t
MRI
im
ag
e
ar
e
s
u
b
jecte
d
to
an
o
th
er
C
NN
to
o
b
tain
th
e
r
eg
is
ter
ed
im
ag
e.
T
h
is
r
eg
is
ter
ed
im
ag
e
is
u
s
ed
f
o
r
in
h
al
e
a
n
d
ex
h
ale
class
if
icatio
n
u
tili
zin
g
a
DM
N
.
T
h
e
c
o
m
b
in
atio
n
o
f
SAO
with
R
SO,
ter
m
ed
SAR
S
O,
is
em
p
lo
y
ed
to
tr
ain
th
e
DM
N.
T
h
e
SAR
SO
-
DM
N
ac
h
iev
ed
m
ax
im
u
m
ac
c
u
r
ac
y
,
m
in
im
u
m
FP
R
,
an
d
m
ax
im
u
m
T
PR
o
f
ap
p
r
o
x
im
ately
9
0
.
7
%,
1
1
.
3
%,
an
d
9
1
.
2
%,
r
esp
ec
tiv
ely
,
wh
e
n
co
n
s
id
er
in
g
9
0
%
o
f
th
e
tr
ain
i
n
g
d
ata.
T
h
ese
f
in
d
in
g
s
d
em
o
n
s
tr
ate
th
e
ef
f
ec
tiv
e
n
ess
o
f
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
in
im
p
r
o
v
in
g
t
h
e
ac
cu
r
ac
y
an
d
r
eliab
ilit
y
o
f
MI
R
an
d
class
if
icatio
n
co
m
p
ar
ed
t
o
ex
is
tin
g
m
eth
o
d
s
.
T
h
e
SAR
SO
-
DM
N
m
eth
o
d
c
an
b
e
f
u
r
t
h
er
o
p
tim
ized
an
d
a
d
a
p
ted
f
o
r
o
th
er
m
ed
ical
im
a
g
in
g
task
s
,
s
u
c
h
as
tu
m
o
r
d
etec
tio
n
,
o
r
g
an
s
eg
m
e
n
tatio
n
,
an
d
tr
ea
tm
en
t
p
lan
n
in
g
.
RE
F
E
R
E
NC
E
S
[
1
]
F
.
Zh
u
,
X
.
Z
h
u
,
Z
.
H
u
a
n
g
,
M
.
D
i
n
g
,
Q
.
Li
,
a
n
d
X
.
Z
h
a
n
g
,
“
D
e
e
p
l
e
a
r
n
i
n
g
b
a
se
d
d
a
t
a
-
a
d
a
p
t
i
v
e
d
e
s
c
r
i
p
t
o
r
f
o
r
n
o
n
-
r
i
g
i
d
m
u
l
t
i
-
mo
d
a
l
me
d
i
c
a
l
i
ma
g
e
r
e
g
i
st
r
a
t
i
o
n
,
”
S
i
g
n
a
l
Pr
o
c
e
ssi
n
g
,
v
o
l
.
1
8
3
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
si
g
p
r
o
.
2
0
2
1
.
1
0
8
0
2
3
.
[
2
]
M
.
L.
Te
r
p
st
r
a
,
M
.
M
a
sp
e
r
o
,
A
.
S
b
r
i
z
z
i
,
a
n
d
C
.
A
.
T
.
V.
D.
B
e
r
g
,
“
⊥
-
l
o
s
s:
a
sy
mm
e
t
r
i
c
l
o
ss
f
u
n
c
t
i
o
n
f
o
r
m
a
g
n
e
t
i
c
r
e
s
o
n
a
n
c
e
i
ma
g
i
n
g
r
e
c
o
n
st
r
u
c
t
i
o
n
a
n
d
i
ma
g
e
r
e
g
i
s
t
r
a
t
i
o
n
w
i
t
h
d
e
e
p
l
e
a
r
n
i
n
g
,
”
Me
d
i
c
a
l
I
m
a
g
e
A
n
a
l
y
si
s
,
v
o
l
.
8
0
,
p
.
1
0
2
5
0
9
,
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
m
e
d
i
a
.
2
0
2
2
.
1
0
2
5
0
9
.
[
3
]
Z.
Z
h
e
n
g
,
W
.
C
a
o
,
Z
.
H
e
,
a
n
d
Y
.
Lu
o
,
“
P
r
o
g
r
e
ss
i
v
e
a
n
a
t
o
m
i
c
a
l
l
y
c
o
n
s
t
r
a
i
n
e
d
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
f
o
r
3
D
d
e
f
o
r
ma
b
l
e
m
e
d
i
c
a
l
i
ma
g
e
r
e
g
i
st
r
a
t
i
o
n
,
”
N
e
u
ro
c
o
m
p
u
t
i
n
g
,
v
o
l
.
4
6
5
,
p
p
.
4
1
7
–
4
2
7
,
N
o
v
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
n
e
u
c
o
m.
2
0
2
1
.
0
8
.
0
9
7
.
[
4
]
M
.
H
u
a
n
g
,
G
.
R
e
n
,
S
.
Z
h
a
n
g
,
Q
.
Z
h
e
n
g
,
a
n
d
H
.
N
i
u
,
“
A
n
u
n
s
u
p
e
r
v
i
se
d
3
D
i
ma
g
e
r
e
g
i
s
t
r
a
t
i
o
n
n
e
t
w
o
r
k
f
o
r
b
r
a
i
n
M
R
I
d
e
f
o
r
ma
b
l
e
r
e
g
i
s
t
r
a
t
i
o
n
,
”
C
o
m
p
u
t
a
t
i
o
n
a
l
a
n
d
Ma
t
h
e
m
a
t
i
c
a
l
Me
t
h
o
d
s
i
n
Me
d
i
c
i
n
e
,
v
o
l
.
2
0
2
2
,
p
p
.
1
–
1
0
,
O
c
t
.
2
0
2
2
,
d
o
i
:
1
0
.
1
1
5
5
/
2
0
2
2
/
9
2
4
6
3
7
8
.
[
5
]
A
.
L
a
r
a
-
H
e
r
n
á
n
d
e
z
e
t
a
l
.
,
“
D
e
e
p
l
e
a
r
n
i
n
g
-
b
a
s
e
d
i
ma
g
e
r
e
g
i
s
t
r
a
t
i
o
n
i
n
d
y
n
a
m
i
c
my
o
c
a
r
d
i
a
l
p
e
r
f
u
si
o
n
C
T
i
m
a
g
i
n
g
,
”
I
EEE
T
r
a
n
s
a
c
t
i
o
n
s
o
n
M
e
d
i
c
a
l
I
m
a
g
i
n
g
,
v
o
l
.
4
2
,
n
o
.
3
,
p
p
.
6
8
4
–
6
9
6
,
M
a
r
.
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
T
M
I
.
2
0
2
2
.
3
2
1
4
3
8
0
.
[
6
]
Q
.
Z
h
e
n
g
,
Q
.
W
a
n
g
,
X
.
B
a
,
S
.
Li
u
,
J
.
N
a
n
,
a
n
d
S
.
Z
h
a
n
g
,
“
A
me
d
i
c
a
l
i
m
a
g
e
r
e
g
i
s
t
r
a
t
i
o
n
me
t
h
o
d
b
a
s
e
d
o
n
p
r
o
g
r
e
ssi
v
e
i
m
a
g
e
s,”
C
o
m
p
u
t
a
t
i
o
n
a
l
a
n
d
M
a
t
h
e
m
a
t
i
c
a
l
Me
t
h
o
d
s i
n
Me
d
i
c
i
n
e
,
v
o
l
.
2
0
2
1
,
p
p
.
1
–
9
,
Ju
l
.
2
0
2
1
,
d
o
i
:
1
0
.
1
1
5
5
/
2
0
2
1
/
4
5
0
4
3
0
6
.
[
7
]
G
.
A
.
B
e
n
v
e
n
u
t
o
,
M
.
C
o
l
n
a
g
o
,
M
.
A
.
D
i
a
s
,
R
.
G
.
N
e
g
r
i
,
E
.
A
.
S
i
l
v
a
,
a
n
d
W
.
C
a
sa
c
a
,
“
A
f
u
l
l
y
u
n
su
p
e
r
v
i
s
e
d
d
e
e
p
l
e
a
r
n
i
n
g
f
r
a
mew
o
r
k
f
o
r
n
o
n
-
r
i
g
i
d
f
u
n
d
u
s
i
m
a
g
e
r
e
g
i
s
t
r
a
t
i
o
n
,
”
Bi
o
e
n
g
i
n
e
e
ri
n
g
,
v
o
l
.
9
,
n
o
.
8
,
p
.
3
6
9
,
A
u
g
.
2
0
2
2
,
d
o
i
:
1
0
.
3
3
9
0
/
b
i
o
e
n
g
i
n
e
e
r
i
n
g
9
0
8
0
3
6
9
.
[
8
]
H
.
S
i
e
b
e
r
t
,
L
.
H
a
n
se
n
,
a
n
d
M
.
P
.
H
e
i
n
r
i
c
h
,
“
Le
a
r
n
i
n
g
a
me
t
r
i
c
f
o
r
mu
l
t
i
mo
d
a
l
me
d
i
c
a
l
i
ma
g
e
r
e
g
i
st
r
a
t
i
o
n
w
i
t
h
o
u
t
s
u
p
e
r
v
i
si
o
n
b
a
s
e
d
o
n
c
y
c
l
e
c
o
n
st
r
a
i
n
t
s,
”
S
e
n
s
o
rs
,
v
o
l
.
2
2
,
n
o
.
3
,
p
.
1
1
0
7
,
F
e
b
.
2
0
2
2
,
d
o
i
:
1
0
.
3
3
9
0
/
s
2
2
0
3
1
1
0
7
.
[
9
]
Y
.
F
u
,
Y
.
Le
i
,
T.
W
a
n
g
,
W
.
J.
C
u
r
r
a
n
,
T.
L
i
u
,
a
n
d
X
.
Y
a
n
g
,
“
D
e
e
p
l
e
a
r
n
i
n
g
i
n
me
d
i
c
a
l
i
m
a
g
e
r
e
g
i
st
r
a
t
i
o
n
:
a
r
e
v
i
e
w
,
”
P
h
y
s
i
c
s
i
n
Me
d
i
c
i
n
e
&
B
i
o
l
o
g
y
,
v
o
l
.
6
5
,
n
o
.
2
0
,
p
.
2
0
TR
0
1
,
O
c
t
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
8
8
/
1
3
6
1
-
6
5
6
0
/
a
b
8
4
3
e
.
[
1
0
]
M
.
K
.
Tr
i
p
a
t
h
i
,
M
.
N
e
e
l
a
k
a
n
t
a
p
p
a
,
P
.
N
.
M
a
h
a
l
l
e
,
S
.
V
.
C
h
a
n
n
a
p
a
t
t
a
n
a
,
G
.
D
e
s
h
m
u
k
h
,
a
n
d
G
.
P
r
a
sh
a
n
t
,
“
D
e
t
e
c
t
i
o
n
a
n
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
o
f
m
a
n
g
o
f
r
u
i
t
-
b
a
se
d
o
n
f
e
a
t
u
r
e
e
x
t
r
a
c
t
i
o
n
a
p
p
l
y
i
n
g
o
p
t
i
m
i
z
e
d
h
y
b
r
i
d
LA
-
FF
a
l
g
o
r
i
t
h
ms,
”
i
n
D
a
t
a
-
C
e
n
t
r
i
c
A
rt
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
f
o
r
Mu
l
t
i
d
i
sc
i
p
l
i
n
a
ry
Ap
p
l
i
c
a
t
i
o
n
s
,
B
o
c
a
R
a
t
o
n
:
C
h
a
p
m
a
n
a
n
d
H
a
l
l
/
C
R
C
,
2
0
2
4
,
p
p
.
1
7
7
–
1
8
5
.
[
1
1
]
M
.
K
.
Tr
i
p
a
t
h
i
,
P
.
K
.
R
e
d
d
y
,
M
.
N
e
e
l
a
k
a
n
t
a
p
p
a
,
C
.
V
.
A
n
d
h
a
r
e
,
a
n
d
S
.
S
h
i
v
e
n
d
r
a
,
“
I
d
e
n
t
i
f
i
c
a
t
i
o
n
o
f
m
a
n
g
o
v
a
r
i
e
t
y
u
si
n
g
n
e
a
r
i
n
f
r
a
r
e
d
s
p
e
c
t
r
o
s
c
o
p
y
,
”
I
n
d
o
n
e
s
i
a
n
J
o
u
r
n
a
l
o
f
E
l
e
c
t
r
i
c
a
l
E
n
g
i
n
e
e
ri
n
g
a
n
d
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
3
1
,
n
o
.
3
,
p
.
1
7
7
6
,
S
e
p
.
2
0
2
3
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
e
e
c
s
.
v
3
1
.
i
3
.
p
p
1
7
7
6
-
1
7
8
3
.
[
1
2
]
C
.
P
a
p
a
d
a
k
i
s,
Y
.
Y
i
n
,
M
.
D
a
n
i
k
a
s
,
a
n
d
C
.
C
h
a
r
a
l
a
m
b
o
u
s
,
“
S
u
r
f
a
c
e
d
i
sc
h
a
r
g
e
s
a
n
d
f
l
a
s
h
o
v
e
r
v
o
l
t
a
g
e
s
i
n
N
a
n
o
c
o
m
p
o
si
t
e
X
LPE
s
a
mp
l
e
s
,
”
E
n
g
i
n
e
e
ri
n
g
,
T
e
c
h
n
o
l
o
g
y
&
A
p
p
l
i
e
d
S
c
i
e
n
c
e
Re
s
e
a
r
c
h
,
v
o
l
.
8
,
n
o
.
6
,
p
p
.
3
5
0
2
–
3
5
0
4
,
D
e
c
.
2
0
1
8
,
d
o
i
:
1
0
.
4
8
0
8
4
/
e
t
a
sr
.
2
1
8
6
.
[
1
3
]
S
.
W
a
n
g
,
X
.
S
h
i
,
M
.
.
Z
h
o
u
,
M
.
.
X
u
,
Y
.
.
Z
h
a
n
g
,
a
n
d
H
.
Zh
a
n
g
,
“
A
s
u
r
v
e
y
o
n
d
e
e
p
l
e
a
r
n
i
n
g
i
n
m
e
d
i
c
a
l
i
ma
g
e
a
n
a
l
y
si
s,
”
Me
d
i
c
a
l
I
m
a
g
e
A
n
a
l
y
s
i
s
,
v
o
l
.
6
0
,
p
.
1
0
1
6
4
1
,
2
0
2
0
.
[
1
4
]
H
.
R
.
B
o
v
e
i
r
i
,
R
.
K
h
a
y
a
m
i
,
R
.
Ja
v
i
d
a
n
,
a
n
d
A
.
M
e
h
d
i
z
a
d
e
h
,
“
M
e
d
i
c
a
l
i
ma
g
e
r
e
g
i
st
r
a
t
i
o
n
u
s
i
n
g
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
s
:
a
c
o
m
p
r
e
h
e
n
si
v
e
r
e
v
i
e
w
,
”
C
o
m
p
u
t
e
rs
a
n
d
El
e
c
t
ri
c
a
l
E
n
g
i
n
e
e
ri
n
g
,
v
o
l
.
8
7
,
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
m
p
e
l
e
c
e
n
g
.
2
0
2
0
.
1
0
6
7
6
7
.
[
1
5
]
X
.
Y
a
n
g
,
R
.
K
w
i
t
t
,
M
.
S
t
y
n
e
r
,
a
n
d
M
.
N
i
e
t
h
a
mm
e
r
,
“
Q
u
i
c
k
s
i
l
v
e
r
:
f
a
st
p
r
e
d
i
c
t
i
v
e
i
ma
g
e
r
e
g
i
st
r
a
t
i
o
n
a
d
e
e
p
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
,
”
N
e
u
ro
I
m
a
g
e
,
v
o
l
.
1
5
8
,
p
p
.
3
7
8
–
3
9
6
,
2
0
1
7
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
n
e
u
r
o
i
ma
g
e
.
2
0
1
7
.
0
7
.
0
0
8
.
[
1
6
]
F
.
Zh
u
,
M
.
D
i
n
g
,
a
n
d
X
.
Z
h
a
n
g
,
“
S
e
l
f
-
si
m
i
l
a
r
i
t
y
i
n
sp
i
r
e
d
l
o
c
a
l
d
e
s
c
r
i
p
t
o
r
f
o
r
n
o
n
-
r
i
g
i
d
mu
l
t
i
-
mo
d
a
l
i
m
a
g
e
r
e
g
i
st
r
a
t
i
o
n
,
”
I
n
f
o
rm
a
t
i
o
n
S
c
i
e
n
c
e
s
,
v
o
l
.
3
7
2
,
p
p
.
1
6
–
3
1
,
D
e
c
.
2
0
1
6
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
i
n
s.
2
0
1
6
.
0
8
.
0
3
1
.
[
1
7
]
M
.
K
.
Tr
i
p
a
t
h
i
a
n
d
S
h
i
v
e
n
d
r
a
,
“
N
e
u
t
r
o
s
o
p
h
i
c
a
p
p
r
o
a
c
h
b
a
s
e
d
i
n
t
e
l
l
i
g
e
n
t
sy
s
t
e
m
f
o
r
a
u
t
o
ma
t
i
c
m
a
n
g
o
d
e
t
e
c
t
i
o
n
,
”
M
u
l
t
i
m
e
d
i
a
T
o
o
l
s
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
8
3
,
n
o
.
1
4
,
p
p
.
4
1
7
6
1
–
4
1
7
8
3
,
O
c
t
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
1
0
4
2
-
0
2
3
-
1
7
0
3
7
-
7.
[
1
8
]
B
.
D
.
d
e
V
o
s,
F
.
F
.
B
e
r
e
n
d
s
e
n
,
M
.
A
.
V
i
e
r
g
e
v
e
r
,
H
.
S
o
k
o
o
t
i
,
M
.
S
t
a
r
i
n
g
,
a
n
d
I
.
I
šg
u
m
,
“
A
d
e
e
p
l
e
a
r
n
i
n
g
f
r
a
mew
o
r
k
f
o
r
u
n
s
u
p
e
r
v
i
s
e
d
a
f
f
i
n
e
a
n
d
d
e
f
o
r
m
a
b
l
e
i
m
a
g
e
r
e
g
i
st
r
a
t
i
o
n
,
”
M
e
d
i
c
a
l
I
m
a
g
e
A
n
a
l
y
si
s
,
v
o
l
.
5
2
,
p
p
.
1
2
8
–
1
4
3
,
F
e
b
.
2
0
1
9
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
m
e
d
i
a
.
2
0
1
8
.
1
1
.
0
1
0
.
[
1
9
]
S
.
D
a
sari
r
a
j
u
,
M
.
H
u
o
,
a
n
d
S
.
M
c
C
a
l
l
a
,
“
D
e
t
e
c
t
i
o
n
a
n
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
o
f
i
mm
a
t
u
r
e
l
e
u
k
o
c
y
t
e
s
f
o
r
d
i
a
g
n
o
s
i
s
o
f
a
c
u
t
e
m
y
e
l
o
i
d
l
e
u
k
e
mi
a
u
si
n
g
r
a
n
d
o
m
f
o
r
e
s
t
a
l
g
o
r
i
t
h
m,”
B
i
o
e
n
g
i
n
e
e
r
i
n
g
,
v
o
l
.
7
,
n
o
.
4
,
p
.
1
2
0
,
O
c
t
.
2
0
2
0
,
d
o
i
:
1
0
.
3
3
9
0
/
b
i
o
e
n
g
i
n
e
e
r
i
n
g
7
0
4
0
1
2
0
.
[
2
0
]
V
.
La
v
a
n
y
a
P
,
M
.
K
.
Tr
i
p
a
t
h
i
,
E
.
P
.
H
e
man
d
,
S
.
K
,
a
n
d
J
.
V
.
N
.
R
a
mes
h
,
“
H
y
p
e
r
s
p
e
c
t
r
a
l
c
r
o
p
i
m
a
g
e
c
l
a
ss
i
f
i
c
a
t
i
o
n
v
i
a
e
n
s
e
mb
l
e
o
f
c
l
a
ss
i
f
i
c
a
t
i
o
n
mo
d
e
l
w
i
t
h
o
p
t
i
m
a
l
t
r
a
i
n
i
n
g
,
”
We
b
I
n
t
e
l
l
i
g
e
n
c
e
,
p
p
.
1
–
3
1
,
2
0
2
4
,
d
o
i
:
1
0
.
3
2
3
3
/
w
e
b
-
2
3
0
2
0
9
.
[2
1
]
A
.
T.
S
a
l
a
w
u
d
e
e
n
e
t
a
l
.
,
“
o
n
t
h
e
d
e
v
e
l
o
p
m
e
n
t
o
f
a
n
o
v
e
l
sm
e
l
l
a
g
e
n
t
o
p
t
i
m
i
z
a
t
i
o
n
(
S
AO
)
f
o
r
o
p
t
i
mi
z
a
t
i
o
n
p
r
o
b
l
e
ms,”
i
-
m
a
n
a
g
e
r
’
s
J
o
u
rn
a
l
o
n
P
a
t
t
e
r
n
R
e
c
o
g
n
i
t
i
o
n
,
v
o
l
.
5
,
n
o
.
4
,
p
.
1
3
,
2
0
1
9
,
d
o
i
:
1
0
.
2
6
6
3
4
/
j
p
r
.
5
.
4
.
1
5
6
7
7
.
[2
2
]
G
.
D
h
i
ma
n
,
M
.
G
a
r
g
,
A
.
N
a
g
a
r
,
V
.
K
u
mar,
a
n
d
M
.
D
e
h
g
h
a
n
i
,
“
A
n
o
v
e
l
a
l
g
o
r
i
t
h
m
f
o
r
g
l
o
b
a
l
o
p
t
i
mi
z
a
t
i
o
n
:
r
a
t
s
w
a
r
m
o
p
t
i
m
i
z
e
r
,
”
J
o
u
rn
a
l
o
f
Am
b
i
e
n
t
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
H
u
m
a
n
i
ze
d
C
o
m
p
u
t
i
n
g
,
v
o
l
.
1
2
,
n
o
.
8
,
p
p
.
8
4
5
7
–
8
4
8
2
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
2
6
5
2
-
020
-
0
2
5
8
0
-
0.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
Med
ica
l ima
g
e
r
eg
is
tr
a
tio
n
a
n
d
cla
s
s
ifica
tio
n
u
s
in
g
s
mell
a
g
en
t
…
(
S
a
n
jeev
ku
ma
r
A
n
g
a
d
i
)
1917
[2
3
]
F
.
Z
ö
l
l
n
e
r
,
“
M
u
l
t
i
m
o
d
a
l
g
r
o
u
n
d
t
r
u
t
h
d
a
t
a
se
t
s
f
o
r
a
b
d
o
mi
n
a
l
m
e
d
i
c
a
l
i
ma
g
e
r
e
g
i
s
t
r
a
t
i
o
n
[
d
a
t
a
]
,
”
h
e
i
D
ATA
,
2
0
2
2
.
h
t
t
p
s
:
/
/
h
e
i
d
a
t
a
.
u
n
i
-
h
e
i
d
e
l
b
e
r
g
.
d
e
/
d
a
t
a
s
e
t
.
x
h
t
m
l
?
p
e
r
si
s
t
e
n
t
I
d
=
d
o
i
:
1
0
.
1
1
5
8
8
/
d
a
t
a
/
I
C
S
F
U
S
.
[2
4
]
M
.
K
.
Tr
i
p
a
t
h
i
a
n
d
S
h
i
v
e
n
d
r
a
,
“
I
mp
r
o
v
e
d
d
e
e
p
b
e
l
i
e
f
n
e
t
w
o
r
k
f
o
r
e
s
t
i
m
a
t
i
n
g
m
a
n
g
o
q
u
a
l
i
t
y
i
n
d
i
c
e
s
a
n
d
g
r
a
d
i
n
g
:
a
c
o
mp
u
t
e
r
v
i
s
i
o
n
-
b
a
se
d
n
e
u
t
r
o
so
p
h
i
c
a
p
p
r
o
a
c
h
,
”
N
e
t
w
o
rk:
C
o
m
p
u
t
a
t
i
o
n
i
n
N
e
u
r
a
l
S
y
s
t
e
m
s
,
v
o
l
.
3
5
,
n
o
.
3
,
p
p
.
2
4
9
–
2
7
7
,
J
u
l
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
8
0
/
0
9
5
4
8
9
8
X
.
2
0
2
3
.
2
2
9
9
8
5
1
.
[2
5
]
S
.
A
l
b
a
w
i
,
O
.
B
a
y
a
t
,
S
.
A
l
-
Az
a
w
i
,
a
n
d
O
.
N
.
U
c
a
n
,
“
S
o
c
i
a
l
t
o
u
c
h
g
e
st
u
r
e
r
e
c
o
g
n
i
t
i
o
n
u
s
i
n
g
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
,
”
C
o
m
p
u
t
a
t
i
o
n
a
l
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
N
e
u
r
o
sc
i
e
n
c
e
,
v
o
l
.
2
0
1
8
,
2
0
1
8
,
d
o
i
:
1
0
.
1
1
5
5
/
2
0
1
8
/
6
9
7
3
1
0
3
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Dr
.
S
a
n
je
e
v
k
u
m
a
r
Ang
a
d
i
re
c
e
iv
e
d
P
h
.
D.
fro
m
Vis
v
e
sv
a
ra
y
a
Tec
h
n
o
lo
g
ica
l
Un
iv
e
rsity
,
Ka
rn
a
tak
a
,
I
n
d
ia
in
2
0
2
3
a
n
d
M
.
Tec
h
i
n
C
o
m
p
u
ter
S
c
ien
c
e
a
n
d
E
n
g
in
e
e
rin
g
fr
o
m
Visv
e
sv
a
ra
y
a
Tec
h
n
o
lo
g
ica
l
Un
i
v
e
rsity
,
Ka
rn
a
tak
a
,
i
n
2
0
1
2
.
At
p
re
se
n
t
h
e
is
wo
rk
in
g
a
s
a
ss
o
c
iate
p
ro
fe
ss
o
r
h
e
a
d
o
f
D
e
p
a
rtme
n
t
in
t
h
e
De
p
a
rtme
n
t
o
f
C
o
m
p
u
ter
S
c
ien
c
e
a
n
d
En
g
i
n
e
e
rin
g
a
t
P
C
ET
’
s
NMVP
M
Nu
tan
Co
l
leg
e
o
f
E
n
g
i
n
e
e
rin
g
a
n
d
Re
se
a
rc
h
,
Tale
g
a
o
n
Da
b
h
a
d
e
,
P
u
n
e
,
I
n
d
ia.
His
re
se
a
rc
h
in
tere
st
in
c
lu
d
e
s
c
o
m
p
u
ter
v
isio
n
,
ima
g
e
p
r
o
c
e
ss
in
g
,
m
a
c
h
in
e
lea
rn
in
g
,
d
a
ta
m
in
in
g
,
a
n
d
n
a
tu
ra
l
lan
g
u
a
g
e
p
r
o
c
e
s
sin
g
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
a
n
g
a
d
i.
sa
n
jee
v
k
u
m
a
r@g
m
a
il
.
c
o
m
.
Dr
.
Mu
k
e
sh
K
u
m
a
r
Tr
ip
a
th
i
re
c
e
iv
e
d
a
P
h
.
D.
in
c
o
m
p
u
ter
sc
ien
c
e
a
n
d
e
n
g
in
e
e
rin
g
fr
o
m
VTU,
Be
lag
a
v
i
.
He
h
a
s
1
5
y
e
a
rs
o
f
tea
c
h
in
g
a
n
d
a
d
m
in
istrativ
e
e
x
p
e
rien
c
e
.
He
h
a
s
su
p
e
r
v
ise
d
a
n
d
c
o
-
su
p
e
rv
ise
d
m
o
re
th
a
n
fiv
e
m
a
ste
rs
a
n
d
2
0
B.
E.
st
u
d
e
n
ts.
He
is
wo
r
k
in
g
a
s
a
n
a
ss
istan
t
p
ro
fe
ss
o
r
wi
th
th
e
De
p
a
rtme
n
t
o
f
C
o
m
p
u
ter
S
c
ien
c
e
a
n
d
En
g
i
n
e
e
rin
g
,
Va
rd
h
a
m
a
n
Co
ll
e
g
e
o
f
En
g
i
n
e
e
rin
g
,
Hy
d
e
ra
b
a
d
,
In
d
ia.
He
is
a
u
t
h
o
re
d
o
r
co
-
a
u
th
o
re
d
m
o
re
t
h
a
n
twe
n
ty
p
u
b
li
c
a
ti
o
n
s
a
n
d
m
o
re
t
h
a
n
4
0
0
c
it
a
ti
o
n
s.
His
re
se
a
rc
h
in
tere
sts
in
c
lu
d
e
so
ft
c
o
m
p
u
ti
n
g
,
m
a
c
h
i
n
e
lea
r
n
in
g
,
i
n
telli
g
e
n
t
s
y
ste
m
s,
ima
g
e
p
ro
c
e
ss
in
g
,
a
n
d
h
y
p
e
rs
p
e
c
tral.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
m
u
k
e
sh
tri
p
a
t
h
i0
1
6
@
g
m
a
il
.
c
o
m
.
Dr
.
Chu
d
a
m
a
n
De
v
id
a
sr
a
o
S
u
k
te
re
c
e
iv
e
d
P
h
.
D
.
i
n
C
o
m
p
u
ter
S
c
ien
c
e
a
n
d
En
g
i
n
e
e
rin
g
fr
o
m
Dr
Ba
b
a
sa
h
e
b
Am
b
e
d
k
a
r
M
a
ra
th
wa
d
a
Un
iv
e
rsit
y
Au
ra
n
g
a
b
a
d
,
M
.
Tec
h
.
in
Co
m
p
u
ter
En
g
in
e
e
rin
g
fr
o
m
Dr
Ba
b
a
sa
h
e
b
Am
b
e
d
k
a
r
Tec
h
n
o
l
o
g
ica
l
Un
iv
e
rsity
L
o
n
e
re
-
Ra
ig
a
d
a
n
d
B
.
E
.
fro
m
S
GG
S
In
stit
u
te
o
f
E
n
g
in
e
e
rin
g
a
n
d
Tec
h
n
o
l
o
g
y
,
Na
n
d
e
d
.
He
h
a
s
1
9
y
e
a
rs
o
f
tea
c
h
in
g
e
x
p
e
rien
c
e
.
He
h
a
s
p
u
b
li
sh
e
d
4
p
a
ten
ts
a
n
d
m
o
re
th
a
n
2
0
re
se
a
rc
h
p
a
p
e
rs
in
Jo
u
rn
a
ls
a
n
d
c
o
n
fe
re
n
c
e
s.
His
a
re
a
o
f
re
se
a
r
c
h
in
c
lu
d
e
s
c
lo
u
d
c
o
m
p
u
ti
n
g
,
c
o
m
p
u
ter
n
e
two
r
k
a
n
d
se
c
u
rit
y
,
a
n
d
m
a
c
h
in
e
lea
rn
i
n
g
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
r
a
jes
h
su
k
te@
g
m
a
il
.
c
o
m
.
Dr
.
S
h
iv
e
n
d
r
a
re
c
e
iv
e
d
a
P
h
.
D.
in
El
e
c
tro
n
ics
a
n
d
Co
m
m
u
n
ica
ti
o
n
E
n
g
i
n
e
e
rin
g
fro
m
JJ
T
Un
iv
e
rsit
y
,
Ra
jas
th
a
n
,
I
n
d
ia,
i
n
2
0
2
3
.
He
re
c
e
iv
e
d
h
is
M
.
Tec
h
.
d
e
g
re
e
i
n
El
e
c
tro
n
ics
a
n
d
Co
m
m
u
n
ica
ti
o
n
fr
o
m
Th
ri
v
a
ll
u
a
v
a
r
Un
i
v
e
sity
Ve
ll
o
re
Ta
m
il
n
a
d
u
,
I
n
d
ia
2
0
1
5
.
He
re
c
e
iv
e
d
h
is
B.
E
.
d
e
g
re
e
i
n
El
e
c
tro
n
ics
a
n
d
Co
m
m
u
n
ica
ti
o
n
fr
o
m
G
N
DEC,
Bid
a
r,
In
d
ia
2
0
1
1
.
He
is
wo
r
k
in
g
a
s
a
n
a
ss
istan
t
p
ro
fe
ss
o
r
in
t
h
e
De
p
a
rtme
n
t
o
f
BCA
a
t
D.K
.
Co
ll
e
g
e
,
Du
m
ra
o
n
,
Bu
x
a
r.
H
e
h
a
s
su
p
e
r
v
ise
d
a
n
d
c
o
-
su
p
e
r
v
ise
d
m
o
re
t
h
a
n
2
0
stu
d
e
n
ts.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
sriv
a
sta
v
a
sh
iv
e
n
d
ra
2
9
@
g
m
a
il
.
c
o
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.