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
8
,
No
.
3
,
J
u
n
e
2
0
2
5
,
p
p
.
199
0
~
200
0
I
SS
N:
2
5
0
2
-
4
7
5
2
,
DOI
: 1
0
.
1
1
5
9
1
/ijeecs.v
38
.i
3
.
p
p
1
9
9
0
-
2
0
0
0
1990
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs.ia
esco
r
e.
co
m
Clea
rNet
:
a
uto
-
e
nco
der bas
ed d
en
o
ising
mo
del f
o
r e
ndo
sco
py
ima
g
es
Vik
ra
nt
Sh
o
k
ee
n
1
,
Sa
nd
ee
p K
um
a
r
1
,
Vidh
u M
a
t
hu
r
1
,
A
m
it
Sh
a
rm
a
2
,
I
nd
ra
j
ee
t
G
up
t
a
3
,
P
a
rit
a
J
a
in
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
,
M
a
h
a
r
a
j
a
S
u
r
a
j
ma
l
I
n
st
i
t
u
t
e
o
f
T
e
c
h
n
o
l
o
g
y
,
D
e
l
h
i
,
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
,
I
M
S
E
n
g
i
n
e
e
r
i
n
g
C
o
l
l
e
g
e
,
G
h
a
z
i
a
b
a
d
,
I
n
d
i
a
3
D
e
p
a
r
t
me
n
t
S
c
h
o
o
l
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
A
I
,
S
R
U
n
i
v
e
r
s
i
t
y
,
W
a
r
a
n
g
a
l
,
I
n
d
i
a
4
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
,
K
I
ET
G
r
o
u
p
o
f
I
n
st
i
t
u
t
i
o
n
s
G
h
a
z
i
a
b
a
d
,
D
e
l
h
i
-
N
C
R
,
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
Ap
r
5
,
2
0
2
4
R
ev
is
ed
Dec
26,
2
0
2
4
Acc
ep
ted
Feb
27
,
2
0
2
5
G
a
stro
in
tes
ti
n
a
l
(G
I)
e
n
d
o
sc
o
p
y
ima
g
e
s
p
lay
a
c
ru
c
ial
ro
le
in
th
e
d
e
tec
ti
o
n
a
n
d
d
iag
n
o
sis
o
f
d
ise
a
se
s
wi
th
in
t
h
e
d
ig
e
stiv
e
trac
t.
Ho
w
e
v
e
r,
th
e
d
e
v
e
lo
p
m
e
n
t
o
f
e
ffe
c
ti
v
e
c
o
m
p
u
t
e
r
v
isio
n
m
o
d
e
ls f
o
r
a
u
to
m
a
ted
a
n
a
ly
sis a
n
d
d
e
n
o
isi
n
g
o
f
e
n
d
o
sc
o
p
y
ima
g
e
s
fa
c
e
s
c
h
a
ll
e
n
g
e
s
a
risin
g
fr
o
m
t
h
e
d
iv
e
rse
n
a
tu
re
o
f
a
b
n
o
rm
a
li
ti
e
s
a
n
d
t
h
e
p
r
e
se
n
c
e
o
f
ima
g
e
a
rt
e
fa
c
ts.
In
th
is
wo
rk
,
t
h
e
u
ti
li
z
a
ti
o
n
o
f
a
n
e
n
c
o
d
e
r
-
d
e
c
o
d
e
r
n
e
two
r
k
f
o
r
d
e
n
o
isin
g
GI
e
n
d
o
sc
o
p
y
ima
g
e
s
u
sin
g
t
h
e
Hy
p
e
rKv
a
sir
d
a
tas
e
t
h
a
s
b
e
e
n
a
n
a
ly
z
e
d
.
T
h
is
a
p
p
r
o
a
c
h
in
v
o
lv
e
s
trai
n
in
g
a
c
u
sto
m
e
n
c
o
d
e
r
-
d
e
c
o
d
e
r
m
o
d
e
l
o
n
th
is
e
x
ten
siv
e
m
u
lt
i
-
c
las
s
e
n
d
o
sc
o
p
y
ima
g
e
d
a
tas
e
t
a
n
d
a
ss
e
ss
in
g
it
s
p
e
rfo
rm
a
n
c
e
a
c
ro
ss
2
3
p
re
v
a
len
t
c
las
se
s
o
f
d
ig
e
sti
v
e
tra
c
t
issu
e
s.
He
re
e
x
p
e
rime
n
ts
sh
o
wc
a
se
th
e
m
o
d
e
l
’
s
a
b
i
li
ty
t
o
lea
rn
r
o
b
u
st
v
i
su
a
l
re
p
re
se
n
tati
o
n
s
fro
m
e
n
d
o
sc
o
p
ic
d
a
ta
,
e
n
a
b
li
n
g
a
c
c
u
ra
te
d
ise
a
se
p
re
d
ictio
n
.
Th
e
a
c
h
iev
e
d
p
r
o
m
isin
g
re
su
lt
s
h
ig
h
li
g
h
t
t
h
e
p
o
ten
ti
a
l
o
f
e
n
c
o
d
e
r
-
d
e
c
o
d
e
r
a
rc
h
it
e
c
tu
re
s
a
s
a
fo
u
n
d
a
t
io
n
a
l
fra
m
e
wo
rk
fo
r
c
o
m
p
u
ter
-
a
id
e
d
e
n
d
o
sc
o
p
y
a
n
a
l
y
sis
wit
h
a
s
p
e
c
ifi
c
fo
c
u
s
o
n
d
e
n
o
isi
n
g
a
p
p
li
c
a
ti
o
n
s.
O
u
r
m
o
d
e
l
m
a
n
a
g
e
s
to
in
c
re
a
se
th
e
p
e
a
k
sig
n
a
l
-
to
-
n
o
ise
ra
ti
o
(
P
S
NR
)
o
f
o
ri
g
in
a
l
-
n
o
isy
p
a
ir
fro
m
1
9
.
1
1
8
9
5
4
t
o
6
9
.
8
9
2
6
3
1
f
o
r
o
rig
i
n
a
l
-
re
c
o
n
str
u
c
ted
p
a
ir
sh
o
wc
a
sin
g
a
lmo
st
p
e
rfe
c
t
re
c
o
n
stru
c
ti
o
n
.
K
ey
w
o
r
d
s
:
Dis
ea
s
e
cla
s
s
if
icatio
n
E
n
co
d
er
-
d
ec
o
d
er
Me
d
ical
im
ag
in
g
MRI
T
r
an
s
f
o
r
m
e
r
s
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
d
ee
p
Ku
m
ar
Dep
ar
tm
en
t o
f
C
o
m
p
u
ter
a
n
d
E
n
g
in
ee
r
in
g
,
Ma
h
a
r
aja
Su
r
aj
m
al
I
n
s
titu
te
o
f
T
ec
h
n
o
lo
g
y
J
an
ak
p
u
r
i,
New
Delh
i,
I
n
d
ia
E
m
ail:
s
an
d
ee
p
.
jag
lan
@
m
s
it.in
,
s
an
.
jag
lan
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
G
astro
in
test
in
al
(
GI
)
en
d
o
s
co
p
y
is
a
v
ital
m
e
d
ical
p
r
o
ce
d
u
r
e
allo
win
g
f
o
r
th
e
d
ir
ec
t
v
is
u
a
lizatio
n
o
f
th
e
d
ig
esti
v
e
tr
ac
t
to
s
cr
ee
n
an
d
d
iag
n
o
s
e
d
is
ea
s
es.
Ho
wev
er
,
th
e
q
u
ality
o
f
en
d
o
s
co
p
y
im
ag
es
is
o
f
ten
h
am
p
er
ed
b
y
n
o
is
e
an
d
a
r
tef
ac
ts
,
p
r
esen
tin
g
a
ch
allen
g
e
f
o
r
ac
cu
r
ate
d
iag
n
o
s
is
[
1
]
.
Un
lik
e
tr
ad
itio
n
al
class
if
icatio
n
m
o
d
els,
th
is
ap
p
r
o
ac
h
f
o
c
u
s
es
o
n
d
en
o
is
in
g
en
d
o
s
co
p
y
im
ag
es
o
b
tain
e
d
f
r
o
m
th
e
Hy
p
er
Kv
asir
d
ataset
[
2
]
wh
ich
co
n
tain
s
a
d
iv
er
s
e
s
et
o
f
1
0
,
6
6
2
im
ag
es
with
2
3
d
is
ea
s
e
lab
els.
GI
d
is
ea
s
es,
in
clu
d
in
g
ca
n
ce
r
s
,
in
f
lam
m
atio
n
,
an
d
in
f
ec
tio
n
s
,
im
p
ac
t m
illi
o
n
s
g
lo
b
a
lly
ea
ch
y
ea
r
.
T
h
e
f
ield
o
f
m
e
d
ical
im
ag
e
a
n
aly
s
is
h
as
s
ee
n
r
ap
id
a
d
v
an
c
em
en
ts
with
th
e
ad
v
en
t
o
f
d
e
ep
lear
n
in
g
tech
n
iq
u
es,
p
ar
ticu
lar
ly
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
s
(
C
NNs).
T
h
ese
m
o
d
els
h
av
e
d
em
o
n
s
tr
ated
r
em
ar
k
ab
l
e
s
u
cc
ess
in
v
ar
io
u
s
en
d
o
s
co
p
y
im
ag
e
an
aly
s
is
task
s
,
f
r
o
m
l
esio
n
d
etec
tio
n
to
d
is
ea
s
e
cla
s
s
if
icatio
n
[
3
]
-
[
5
]
.
Ho
wev
er
,
th
e
m
ajo
r
ity
o
f
e
x
is
tin
g
r
esear
ch
h
as
f
o
cu
s
e
d
o
n
class
if
icatio
n
an
d
d
et
ec
tio
n
task
s
,
o
f
ten
o
v
er
lo
o
k
in
g
t
h
e
cr
itical
p
r
e
p
r
o
ce
s
s
in
g
s
tep
o
f
im
ag
e
d
e
n
o
is
in
g
.
R
ec
en
t
s
tu
d
ies
h
av
e
b
eg
u
n
t
o
ex
p
lo
r
e
t
h
e
p
o
te
n
tial
o
f
r
e
al
-
tim
e
ar
tific
ial
in
tellig
en
ce
(
AI
)
im
ag
e
p
r
o
ce
s
s
in
g
s
y
s
tem
s
f
o
r
t
h
e
d
iag
n
o
s
is
o
f
u
p
p
er
GI
m
alig
n
an
cies
Van
ia
et
a
l.
[
6
]
.
T
h
ese
ad
v
a
n
ce
m
en
ts
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:
2502
-
4
7
5
2
C
lea
r
N
e
t:
a
u
to
-
en
c
o
d
er b
a
s
ed
d
en
o
is
in
g
mo
d
el
fo
r
en
d
o
s
co
p
y
ima
g
es
(
V
ikra
n
t S
h
o
ke
en
)
1991
co
n
tr
ib
u
te
t
o
th
e
o
n
g
o
in
g
d
e
v
elo
p
m
en
t
o
f
c
o
m
p
u
te
r
-
aid
e
d
d
iag
n
o
s
is
(
C
AD)
s
y
s
tem
s
f
o
r
GI
e
n
d
o
s
co
p
y
,
p
r
o
m
is
in
g
to
en
h
a
n
ce
th
e
ac
cu
r
ac
y
an
d
ef
f
icien
cy
o
f
d
is
ea
s
e
d
etec
tio
n
an
d
class
if
icatio
n
.
T
h
e
p
r
i
m
a
r
y
c
o
n
t
r
i
b
u
t
i
o
n
o
f
t
h
i
s
p
a
p
e
r
i
s
t
h
e
i
n
t
r
o
d
u
c
t
i
o
n
o
f
C
l
e
a
r
N
e
t
,
a
n
i
n
n
o
v
a
t
i
v
e
a
u
t
o
-
e
n
c
o
d
e
r
b
a
s
e
d
d
e
n
o
i
s
i
n
g
m
o
d
e
l
s
p
e
c
i
f
ic
a
l
l
y
d
es
i
g
n
e
d
f
o
r
e
n
d
o
s
c
o
p
y
im
a
g
e
s
.
O
u
r
a
p
p
r
o
a
c
h
l
e
v
e
r
a
g
es
t
h
e
p
o
we
r
o
f
d
e
e
p
l
e
a
r
n
i
n
g
,
p
a
r
t
i
c
u
l
a
r
l
y
t
h
e
e
n
c
o
d
e
r
-
d
e
c
o
d
e
r
a
r
c
h
i
t
e
c
t
u
r
e
i
n
s
p
i
r
e
d
b
y
r
e
c
e
n
t
a
d
v
a
n
c
e
m
e
n
t
s
i
n
v
i
s
i
o
n
t
r
a
n
s
f
o
r
m
e
r
n
e
t
w
o
r
k
s
,
t
o
e
f
f
e
c
t
i
v
el
y
r
e
m
o
v
e
n
o
i
s
e
a
n
d
e
n
h
a
n
c
e
i
m
a
g
e
q
u
a
l
i
t
y
.
B
y
f
o
c
u
s
i
n
g
o
n
t
h
e
c
r
u
c
i
a
l
p
r
e
p
r
o
c
e
s
s
i
n
g
s
t
e
p
o
f
d
e
n
o
i
s
i
n
g
,
C
l
ea
r
N
e
t
a
i
m
s
to
i
m
p
r
o
v
e
t
h
e
o
v
e
r
al
l
p
e
r
f
o
r
m
a
n
c
e
o
f
s
u
b
s
e
q
u
e
n
t
a
n
a
l
y
s
i
s
t
as
k
s
,
s
u
c
h
as
d
i
s
ea
s
e
c
l
a
s
s
i
f
ic
a
t
i
o
n
a
n
d
l
es
i
o
n
d
e
t
ec
t
io
n
.
2.
RE
L
AT
E
D
WO
RK
S
Nu
m
er
o
u
s
s
tu
d
ies
h
av
e
d
elv
e
d
in
to
lev
er
a
g
in
g
d
ee
p
lear
n
i
n
g
f
o
r
au
to
m
ated
an
aly
s
is
in
en
d
o
s
co
p
y
im
ag
es,
in
itially
co
n
ce
n
tr
atin
g
o
n
b
in
a
r
y
class
if
icatio
n
an
d
d
etec
tio
n
task
s
.
A
m
eth
o
d
p
r
o
p
o
s
ed
b
y
[
7
]
u
tili
ze
s
J
DP
C
A
f
o
r
lesi
o
n
d
etec
tio
n
in
GI
e
n
d
o
s
co
p
ic
im
ag
es,
ac
h
i
ev
in
g
h
ig
h
ac
cu
r
ac
y
r
ates
f
o
r
v
ar
io
u
s
c
o
n
d
itio
n
s
,
s
u
r
p
ass
in
g
tr
ad
itio
n
al
m
eth
o
d
s
.
He
et
a
l.
[
8
]
em
p
lo
y
s
m
u
ltip
le
p
r
e
-
tr
ain
ed
C
NNs
to
im
p
r
o
v
e
class
if
icatio
n
o
f
lesi
o
n
ty
p
es in
co
lo
n
o
s
co
p
y
i
m
ag
es.
Fo
r
p
o
l
y
p
d
etec
tio
n
,
[
9
]
in
d
icate
s
in
cr
ea
s
ed
ad
en
o
m
a
d
etec
tio
n
r
ates w
ith
AI
co
m
p
u
ter
-
aid
e
d
d
etec
tio
n
d
u
r
in
g
co
lo
n
o
s
co
p
y
b
u
t
also
h
ig
h
lig
h
ts
h
ig
h
er
r
ates
o
f
u
n
n
ec
ess
ar
y
p
o
ly
p
r
em
o
v
al.
A
th
em
atic
s
u
r
v
ey
[
1
0
]
o
n
m
ed
ical
im
a
g
e
s
eg
m
en
tatio
n
u
s
in
g
d
ee
p
lear
n
in
g
o
f
f
er
s
in
s
ig
h
ts
in
to
s
u
p
er
v
is
ed
an
d
wea
k
l
y
s
u
p
e
r
v
is
ed
lear
n
in
g
m
eth
o
d
s
.
T
an
wa
r
et
a
l.
[
1
1
]
in
tr
o
d
u
ce
a
d
ee
p
lear
n
in
g
a
p
p
r
o
ac
h
f
o
r
co
lo
r
ec
tal
p
o
ly
p
d
etec
tio
n
an
d
class
if
icatio
n
with
9
2
%
ac
cu
r
ac
y
.
B
o
r
g
li
et
a
l.
[
2
]
h
y
p
er
Kv
asir
d
ataset
f
ac
ilit
ates
m
u
lti
-
class
cla
s
s
if
i
ca
tio
n
,
an
d
[
1
2
]
r
e
v
iew
im
b
alan
ce
p
r
o
b
lem
s
in
o
b
ject
d
e
tectio
n
,
b
u
t
s
tu
d
ies
u
s
in
g
th
is
d
ataset
[
1
3
]
t
y
p
ically
f
o
cu
s
o
n
s
in
g
le
-
lab
el
class
if
icatio
n
.
T
h
is
wo
r
k
ad
d
r
ess
es
th
e
tas
k
o
f
d
e
n
o
is
in
g
en
d
o
s
co
p
y
i
m
ag
es,
co
n
s
id
er
in
g
v
ar
io
u
s
n
o
is
e
lev
els,
p
r
esen
tin
g
a
s
o
p
h
is
ticated
d
ee
p
lear
n
in
g
m
o
d
el
tailo
r
ed
f
o
r
t
h
is
p
r
o
b
lem
.
W
h
ile
tr
an
s
f
o
r
m
er
s
lik
e
DE
T
R
h
av
e
g
ain
ed
p
o
p
u
lar
ity
f
o
r
d
etec
tio
n
an
d
class
if
icatio
n
(
1
6
)
,
o
u
r
s
tu
d
y
p
io
n
ee
r
s
th
e
u
s
e
o
f
an
en
co
d
er
-
d
ec
o
d
e
r
tr
an
s
f
o
r
m
er
-
b
ased
m
o
d
el
f
o
r
m
u
lti
-
lab
el
en
d
o
s
co
p
y
im
ag
e
d
en
o
is
in
g
,
s
h
o
wca
s
in
g
s
tr
o
n
g
p
er
f
o
r
m
an
ce
an
d
g
en
er
aliza
b
ilit
y
.
I
m
a
g
e
d
en
o
is
in
g
r
em
ain
s
a
cr
u
cial
ch
a
llen
g
e
ac
r
o
s
s
m
ed
ical
im
ag
i
n
g
m
o
d
alities
,
as
h
ig
h
lig
h
ted
in
[
1
4
]
s
u
r
v
ey
o
n
n
o
is
e
r
e
d
u
ctio
n
tech
n
iq
u
es
i
n
lu
n
g
ca
n
ce
r
c
o
m
p
u
ted
t
o
m
o
g
r
ap
h
y
(
CT
)
s
ca
n
im
ag
es.
L
itjen
s
et
a
l.
[
1
5
]
p
r
o
v
id
e
a
c
o
m
p
r
e
h
en
s
iv
e
r
e
v
ie
w
o
f
d
ee
p
lear
n
in
g
ap
p
licatio
n
s
in
m
ed
ical
im
ag
e
an
aly
s
is
,
wh
ile
C
NNs,
s
u
ch
as
th
e
o
n
e
u
tili
ze
d
b
y
[
1
6
]
,
h
a
v
e
s
h
o
wn
p
r
o
m
is
e
f
o
r
d
e
n
o
is
in
g
m
ed
ical
im
ag
es.
T
h
e
s
tu
d
y
[
1
7
]
p
r
o
p
o
s
es
th
e
d
esig
n
o
f
d
ee
p
f
ee
d
-
f
o
r
war
d
d
e
n
o
is
in
g
C
NN
s
u
s
in
g
r
esid
u
al
lear
n
in
g
a
n
d
b
atch
n
o
r
m
aliza
tio
n
to
im
p
r
o
v
e
m
e
d
ical
im
ag
e
d
en
o
is
in
g
p
er
f
o
r
m
a
n
ce
,
p
ar
ticu
lar
l
y
f
o
r
s
m
all
s
am
p
le
s
izes.
E
n
co
d
er
-
d
ec
o
d
er
m
o
d
els,
ex
em
p
lifie
d
b
y
R
o
n
n
e
b
er
g
e
r
et
a
l.
U
-
n
et
[
1
8
]
an
d
T
ah
m
id
et
a
l
.
co
n
d
itio
n
al
ad
v
er
s
ar
ial
tr
ain
i
n
g
[
1
9
]
,
ar
e
c
o
m
m
o
n
ly
em
p
l
o
y
ed
f
o
r
im
ag
e
r
esto
r
atio
n
ta
s
k
s
.
L
eh
tin
en
et
a
l.
[
2
0
]
p
r
o
p
o
s
e
a
n
o
v
el
a
p
p
r
o
ac
h
to
s
ig
n
al
r
ec
o
n
s
tr
u
ctio
n
u
s
in
g
m
ac
h
in
e
lear
n
in
g
,
d
em
o
n
s
tr
atin
g
th
e
ab
ilit
y
t
o
r
esto
r
e
im
ag
es
f
r
o
m
co
r
r
u
p
te
d
ex
am
p
les
with
o
u
t
clea
n
d
ata
o
r
ex
p
licit
s
tatis
tica
l
m
o
d
els
o
f
co
r
r
u
p
tio
n
.
T
r
an
s
f
o
r
m
e
r
s
,
in
itially
d
ev
elo
p
ed
f
o
r
n
atu
r
al
la
n
g
u
a
g
e
p
r
o
ce
s
s
in
g
,
h
av
e
r
ec
en
tly
b
ee
n
ap
p
lied
to
m
ed
ical
im
ag
e
an
aly
s
is
,
d
e
m
o
n
s
tr
atin
g
th
eir
ca
p
ab
ilit
y
to
m
o
d
el
lo
n
g
-
r
an
g
e
d
ep
e
n
d
en
cies
in
im
ag
e
d
ata.
T
h
e
r
esear
ch
by
He
et
a
l.
[
2
1
]
h
ig
h
lig
h
ts
t
h
e
p
o
ten
tial
o
f
tr
a
n
s
f
o
r
m
er
s
in
im
p
r
o
v
in
g
v
ar
io
u
s
asp
ec
ts
o
f
m
e
d
ical
im
ag
i
n
g
,
in
clu
d
in
g
s
eg
m
en
tatio
n
,
class
if
icatio
n
,
a
n
d
d
en
o
is
in
g
task
s
.
T
h
ese
ad
v
a
n
ce
m
en
ts
c
o
u
ld
s
ig
n
if
ican
tly
en
h
a
n
ce
th
e
p
er
f
o
r
m
an
ce
o
f
o
u
r
au
t
o
-
e
n
co
d
er
b
ased
d
en
o
is
in
g
m
o
d
el
b
y
lev
er
ag
in
g
th
e
atten
tio
n
m
ec
h
an
is
m
s
in
h
er
en
t
in
tr
an
s
f
o
r
m
e
r
ar
ch
itectu
r
es.
Gen
er
ativ
e
ad
v
er
s
ar
ial
n
etwo
r
k
s
(
GANs)
h
av
e
s
h
o
wn
r
em
ar
k
ab
le
s
u
cc
ess
in
im
ag
e
d
en
o
i
s
in
g
task
s
d
u
e
to
th
eir
ab
ilit
y
to
lear
n
an
d
g
en
er
ate
h
i
g
h
-
q
u
ality
im
ag
es.
Als
aiar
i
et
a
l.
[
2
2
]
u
tili
ze
d
a
GAN
-
b
ased
ap
p
r
o
ac
h
f
o
r
im
a
g
e
d
en
o
is
in
g
,
ac
h
iev
in
g
s
ig
n
i
f
ican
t
im
p
r
o
v
em
en
ts
in
im
ag
e
clar
ity
a
n
d
n
o
is
e
r
ed
u
cti
o
n
.
Un
lik
e
th
eir
GAN
-
b
ased
m
eth
o
d
,
o
u
r
ap
p
r
o
ac
h
em
p
lo
y
s
an
au
to
e
n
co
d
er
-
s
ty
le
U
-
n
et
a
r
ch
itectu
r
e,
wh
ic
h
f
o
cu
s
es o
n
lear
n
i
n
g
a
n
ef
f
icien
t r
ep
r
esen
tatio
n
o
f
th
e
in
p
u
t d
a
ta
to
ac
h
iev
e
ef
f
ec
tiv
e
d
en
o
is
in
g
.
3.
M
E
T
H
O
D
I
n
th
is
s
ec
tio
n
th
e
m
eth
o
d
o
lo
g
y
o
f
o
u
r
p
ap
er
.
T
h
is
s
ec
tio
n
talk
s
ab
o
u
t
t
h
e
d
ataset
p
r
ep
r
atio
n
an
d
m
o
d
el
u
s
ed
in
d
etail.
3
.
1
.
Da
t
a
s
et
d
escript
io
n
Hy
p
er
Kv
asir
co
n
tain
s
1
0
6
6
2
GI
tr
ac
t
im
ag
es
lab
elled
wi
th
2
3
co
m
m
o
n
class
es.
F
ig
u
r
e
1
s
h
o
ws
s
am
p
le
im
ag
es
f
r
o
m
th
e
d
ataset
d
em
o
n
s
tr
atin
g
t
h
e
d
iv
er
s
ity
o
f
v
is
u
al
p
atter
n
s
.
W
ith
a
p
r
i
m
ar
y
e
m
p
h
asis
o
n
GI
en
d
o
s
co
p
y
,
th
e
d
ataset
co
v
er
s
d
iv
er
s
e
s
eg
m
en
ts
o
f
th
e
d
ig
esti
v
e
tr
ac
t,
s
u
ch
as
th
e
o
eso
p
h
ag
u
s
,
s
to
m
ac
h
,
an
d
c
o
lo
n
.
T
h
e
d
is
tin
g
u
is
h
in
g
f
ea
tu
r
e
o
f
th
e
H
y
p
er
Kv
asir
d
ataset
is
its
in
clu
s
iv
en
ess
,
en
co
m
p
ass
in
g
n
o
t
o
n
ly
ty
p
ical
en
d
o
s
co
p
y
im
a
g
es
b
u
t
also
a
co
m
p
r
eh
en
s
iv
e
ar
r
ay
o
f
im
ag
es
d
is
p
lay
i
n
g
a
wid
e
r
an
g
e
o
f
an
o
m
alies
an
d
GI
d
is
o
r
d
er
s
,
Fig
u
r
e
2
s
h
o
ws th
e
d
is
tr
ib
u
tio
n
o
f
class
es i
n
th
e
d
ataset
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
38
,
No
.
3
,
J
u
n
e
20
25
:
1
9
9
0
-
200
0
1992
Fig
u
r
e
1
.
Sam
p
le
im
a
g
es f
r
o
m
d
ataset
Fig
u
r
e
2
.
Data
s
et
class
es d
escr
ip
tio
n
3
.
2
.
M
o
del descript
io
n
C
lear
Net
is
an
en
co
d
er
-
d
ec
o
d
er
ar
ch
itectu
r
e
in
s
p
ir
ed
b
y
r
ec
en
t
v
is
io
n
tr
a
n
s
f
o
r
m
er
n
et
wo
r
k
s
lik
e
DE
T
R
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
i
s
tailo
r
ed
f
o
r
d
en
o
is
in
g
GI
.
A
s
s
h
o
wn
in
Fig
u
r
e
3
,
th
e
p
r
o
p
o
s
ed
m
o
d
el
c
o
n
s
is
ts
o
f
an
I
n
ce
p
tio
n
-
v
3
en
co
d
er
th
at
ex
tr
ac
ts
v
is
u
al
f
ea
tu
r
es,
f
o
llo
wed
b
y
a
cu
s
to
m
d
ec
o
d
er
f
o
r
d
en
o
is
in
g
.
I
t
h
as
a
U
-
Net
-
s
ty
le
ar
ch
itectu
r
e
with
s
o
m
e
m
o
d
if
icatio
n
s
.
U
-
n
et
i
s
a
C
NN
ar
ch
itectu
r
e
th
at
is
co
m
m
o
n
l
y
u
s
ed
f
o
r
s
em
an
tic
s
eg
m
en
tatio
n
,
a
task
wh
er
e
th
e
m
o
d
el
m
u
s
t d
e
n
o
is
e
th
e
g
iv
en
im
ag
e.
Fig
u
r
e
3
.
I
n
ce
p
tio
n
V3
a
r
ch
itec
tu
r
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:
2502
-
4
7
5
2
C
lea
r
N
e
t:
a
u
to
-
en
c
o
d
er b
a
s
ed
d
en
o
is
in
g
mo
d
el
fo
r
en
d
o
s
co
p
y
ima
g
es
(
V
ikra
n
t S
h
o
ke
en
)
1993
DE
T
R
en
co
d
er
-
d
ec
o
d
er
:
DE
T
R
(
DE
tectio
n
T
R
an
s
-
f
o
r
m
er
)
p
r
esen
ted
b
y
[
1
3
]
is
a
r
e
ce
n
t
v
is
io
n
tr
an
s
f
o
r
m
er
m
o
d
el
f
o
r
o
b
ject
d
etec
tio
n
.
I
t
c
o
n
s
is
ts
o
f
a
C
NN
en
co
d
er
f
o
llo
wed
b
y
a
tr
an
s
f
o
r
m
er
d
ec
o
d
er
.
T
h
e
C
NN
en
co
d
er
ex
tr
ac
ts
f
ea
tu
r
e
r
ep
r
esen
tatio
n
s
f
r
o
m
in
p
u
t
im
ag
es.
T
h
e
t
r
an
s
f
o
r
m
e
r
d
ec
o
d
er
th
e
n
p
er
f
o
r
m
s
d
en
o
is
in
g
u
s
in
g
th
ese
en
co
d
ed
f
ea
tu
r
es.
A
b
asic
d
ep
ictio
n
o
f
an
au
to
e
n
co
d
er
m
o
d
el
is
s
h
o
wn
in
Fig
u
r
e
4
.
T
h
e
in
s
p
ir
atio
n
is
d
r
awn
f
r
o
m
DE
T
R
’
s
o
v
er
all
en
c
o
d
er
-
d
ec
o
d
er
ar
ch
itectu
r
e
f
o
r
o
u
r
en
d
o
s
co
p
y
im
a
g
e
d
e
n
o
is
in
g
m
o
d
el.
Ho
wev
e
r
,
a
cu
s
to
m
d
e
co
d
er
d
esig
n
tailo
r
ed
f
o
r
th
e
d
en
o
is
in
g
o
f
i
m
ag
es is
d
ev
elo
p
ed
.
I
n
ce
p
tio
n
V3
:
w
e
lev
er
ag
e
a
p
r
e
-
tr
ain
ed
I
n
ce
p
tio
n
-
v
3
C
NN
as
th
e
en
co
d
er
b
ac
k
b
o
n
e,
s
im
ilar
to
DE
T
R
’
s
u
s
e
o
f
a
R
es
Net
C
N
N
b
ac
k
b
o
n
e
.
I
n
ce
p
tio
n
-
v
3
is
an
in
n
o
v
ativ
e
C
NN
ar
ch
itectu
r
e
p
r
o
p
o
s
ed
b
y
[
2
3
]
.
T
h
e
k
ey
f
ea
tu
r
e
o
f
I
n
ce
p
tio
n
-
v
3
is
th
e
I
n
ce
p
tio
n
m
o
d
u
les.
T
h
ese
ap
p
ly
co
n
v
o
lu
tio
n
al
f
ilter
s
o
f
d
if
f
er
en
t
s
izes
in
p
ar
allel
an
d
co
n
ca
ten
ate
th
e
o
u
tp
u
ts
.
Sp
ec
if
ically
,
th
ey
u
s
e
1
×
1
,
3
×
3
,
an
d
5
×
5
co
n
v
o
l
u
tio
n
s
.
T
h
is
allo
ws
th
e
n
etwo
r
k
t
o
ca
p
tu
r
e
m
u
lti
-
s
ca
le
v
is
u
al
f
ea
tu
r
es f
r
o
m
f
in
e
-
g
r
ain
ed
p
atter
n
s
to
ab
s
tr
ac
t c
o
n
ce
p
ts
.
T
h
e
wo
r
k
h
y
p
o
t
h
esizes
th
at
s
u
b
tle
tex
tu
r
e
ch
an
g
es
in
e
n
d
o
s
co
p
y
im
a
g
es
ca
n
in
d
icate
p
ath
o
lo
g
y
,
s
o
th
is
ab
ilit
y
is
s
ig
n
if
ican
t.
T
h
e
ad
d
itio
n
o
f
1
×
1
c
o
n
v
o
lu
tio
n
s
also
r
ed
u
ce
s
d
im
e
n
s
io
n
ality
to
im
p
r
o
v
e
co
m
p
u
tatio
n
al
ef
f
icien
cy
co
m
p
ar
ed
to
n
aiv
e
s
tack
in
g
o
f
m
u
lti
-
s
ize
co
n
v
o
lu
tio
n
s
.
Ov
er
al
l,
th
e
I
n
ce
p
tio
n
-
v3
ar
ch
itectu
r
e
s
h
o
wed
h
ig
h
p
er
f
o
r
m
an
ce
o
n
m
e
d
ical
im
ag
in
g
task
s
[
2
4
]
wh
ile
b
ein
g
o
p
tim
ized
f
o
r
ef
f
icien
cy
.
Au
to
e
n
co
d
e
r
s
(U
-
n
et)
:
a
u
to
en
co
d
er
s
r
ep
r
esen
t
a
f
u
n
d
am
e
n
tal
co
n
ce
p
t
in
d
ee
p
lear
n
in
g
,
s
p
ec
if
ically
d
esig
n
ed
f
o
r
u
n
s
u
p
er
v
is
ed
lear
n
in
g
task
s
.
T
h
e
c
o
r
e
id
ea
b
eh
in
d
au
to
en
c
o
d
er
s
is
to
lear
n
ef
f
icien
t
r
ep
r
esen
tatio
n
s
o
f
d
ata,
ca
p
tu
r
in
g
ess
en
tial
f
ea
tu
r
es
with
o
u
t
ex
p
licit
lab
els.
Ma
th
em
a
tically
,
th
e
e
n
co
d
er
o
p
er
atio
n
h
=
f
decoder
(
x
)
tr
a
n
s
f
o
r
m
s
th
e
in
p
u
t
x
in
t
o
a
lo
we
r
-
d
im
en
s
io
n
al
r
e
p
r
esen
tatio
n
,
wh
ile
th
e
d
ec
o
d
er
o
p
er
atio
n
x
ˆ
=f
decoder
(
h
)
r
ec
o
n
s
tr
u
cts
th
e
in
p
u
t a
p
p
r
o
x
im
atio
n
x
ˆ
.
h
=
σ
(
W
_
e
n
c
ode
r
⋅
x
+
b
_
e
n
c
ode
r
)
(
1
)
x̂
=
σ
(
W
_
de
c
ode
r
⋅
h
+
b
_
de
c
ode
r
)
(
2
)
ℒ
(
x
,
x̂
)
=
M
SE
(
x
,
x̂
)
=
1
/
n
∑
(
x
_
i
−
x̂
_
i
)
^
2
(
3
)
3
.
3
.
Arc
hite
ct
ure
A
U
-
n
et
s
ty
le
au
to
en
co
d
er
n
et
wo
r
k
with
a
tr
an
s
f
o
r
m
er
-
in
s
p
i
r
ed
s
tr
u
ctu
r
e
with
a
C
NN
b
ac
k
b
o
n
e
f
o
r
b
lin
d
d
en
o
is
in
g
en
d
o
s
co
p
y
im
ag
es
is
p
r
o
p
o
s
ed
.
T
h
e
C
NN
b
ased
au
to
en
c
o
d
er
will
w
o
r
k
as
d
ep
icted
in
Fig
u
r
e
5
.
T
h
e
en
c
o
d
er
lev
er
a
g
es
a
p
r
e
-
tr
ain
ed
I
n
ce
p
tio
n
-
v
3
m
o
d
el
to
ex
tr
a
ct
h
ier
a
r
ch
ical
v
is
u
al
f
ea
tu
r
es
f
r
o
m
n
o
is
y
in
p
u
ts
(
2
4
)
.
He
r
e’
s
a
r
ep
r
esen
tatio
n
o
f
h
o
w
a
U
-
n
et
wo
r
k
s
:
Fig
u
r
e
4
.
B
asic a
u
to
en
co
d
er
m
o
d
el
Fig
u
r
e
5
.
C
o
n
v
o
lu
tio
n
al
au
to
e
n
co
d
er
m
o
d
el
3.
4
.
Alg
o
rit
hm
Her
e
’
s
a
r
ep
r
esen
tatio
n
o
f
h
o
w
th
e
U
-
n
et
in
o
u
r
p
r
o
p
o
s
ed
ar
ch
itectu
r
e
wo
r
k
s
:
1
.
I
n
p
u
t
l
ay
er
:
-
X
is
th
e
in
p
u
t im
a
g
e.
2
.
C
o
n
tr
ac
tin
g
p
ath
:
a.
Ap
p
ly
co
n
v
o
lu
tio
n
al
lay
er
wit
h
r
ec
tifie
d
lin
ea
r
u
n
it
(
R
eL
U
)
ac
tiv
atio
n
:
(
)
=
(
⋅
+
)
(
4
)
b.
Ap
p
ly
m
ax
-
p
o
o
lin
g
lay
e
r
:
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
38
,
No
.
3
,
J
u
n
e
20
25
:
1
9
9
0
-
200
0
1994
(
)
=
(
(
)
)
(
5
)
c.
R
ep
ea
t
th
e
ab
o
v
e
s
tep
s
to
cr
ea
te
a
co
n
tr
ac
tin
g
p
ath
with
m
u
ltip
le
co
n
v
o
lu
tio
n
al
b
lo
ck
s
an
d
p
o
o
lin
g
lay
er
s
.
Sav
e
th
e
in
ter
m
e
d
iate
f
ea
tu
r
e
m
ap
s
:
=
{
,
}
(
6
)
3
.
B
o
ttlen
ec
k
:
a.
Ap
p
ly
co
n
v
o
lu
tio
n
al
lay
er
wit
h
R
eL
U
ac
tiv
atio
n
:
(
)
=
(
⋅
+
)
(
7
)
b.
Per
f
o
r
m
u
p
-
s
am
p
lin
g
:
(
(
)
)
(
8
)
4
.
E
x
p
a
n
s
iv
e
p
ath
:
a.
C
o
n
ca
ten
ate
th
e
f
ea
tu
r
e
m
ap
s
f
r
o
m
th
e
co
n
tr
ac
tin
g
p
ath
:
=
(
,
(
(
)
)
)
(
9
)
b.
Ap
p
ly
co
n
v
o
lu
tio
n
al
lay
er
wit
h
R
eL
U
ac
tiv
atio
n
:
(
)
=
(
∗
+
)
(
1
0
)
c.
Ap
p
ly
u
p
s
am
p
lin
g
:
(
(
)
)
(
1
1
)
d.
R
ep
ea
ted
th
e
ab
o
v
e
s
tep
s
to
cr
ea
te
an
ex
p
a
n
s
iv
e
p
ath
w
ith
m
u
ltip
le
co
n
v
o
lu
tio
n
al
b
l
o
ck
s
an
d
u
p
s
am
p
lin
g
lay
er
s
:
=
{
,
}
(
1
2
)
5
.
Ou
tp
u
t
l
a
y
er
:
-
Ap
p
ly
c
o
n
v
o
lu
tio
n
al
lay
e
r
w
ith
So
f
tMa
x
ac
tiv
atio
n
to
o
b
tai
n
th
e
f
in
al
d
e
-
n
o
is
ed
im
ag
e:
=
(
∗
+
)
(
1
3
)
I
n
ce
p
tio
n
-
v
3
,
r
en
o
wn
ed
f
o
r
its
m
u
lti
-
s
ca
le
f
ea
tu
r
e
lea
r
n
in
g
,
em
p
lo
y
s
p
ar
allel
m
o
d
u
les
with
co
n
v
o
l
u
tio
n
s
o
f
v
a
r
y
in
g
r
ec
e
p
tiv
e
f
ield
s
izes,
cr
u
cial
f
o
r
ca
p
tu
r
in
g
b
o
th
f
in
e
d
etails
an
d
g
lo
b
al
c
o
n
tex
t
in
m
ed
ical
im
ag
in
g
task
s
.
I
ts
ef
f
ec
tiv
en
ess
o
n
task
s
with
lim
it
ed
d
ata
m
ak
es
it
s
u
itab
le
f
o
r
GI
im
ag
e
d
en
o
is
in
g
[
2
5
]
.
T
h
e
d
ec
o
d
er
m
o
d
u
le
u
tili
ze
s
tr
an
s
p
o
s
e
co
n
v
o
lu
tio
n
s
to
u
p
s
am
p
le
en
c
o
d
er
em
b
ed
d
in
g
s
f
o
r
im
ag
e
p
r
ed
ictio
n
,
aid
e
d
b
y
s
k
ip
co
n
n
ec
tio
n
s
f
r
o
m
en
c
o
d
er
lay
e
r
s
f
o
r
lo
ca
l
d
etail
r
ec
o
n
s
tr
u
ctio
n
.
Ou
r
h
y
b
r
id
tr
an
s
f
o
r
m
er
-
b
ased
U
-
n
et
c
o
m
b
in
es
g
lo
b
al
r
ea
s
o
n
i
n
g
with
lo
ca
lizatio
n
ab
ilit
y
,
b
en
e
f
itin
g
d
e
n
o
is
in
g
task
s
.
Op
tim
izatio
n
em
p
lo
y
s
r
m
s
p
r
o
p
o
p
tim
izer
an
d
m
ea
n
s
q
u
ar
e
d
er
r
o
r
lo
s
s
o
n
p
air
ed
r
ea
l
s
am
p
les
with
s
im
u
lated
n
o
is
e,
en
ab
lin
g
th
e
m
o
d
el
to
h
an
d
le
d
i
v
er
s
e
d
is
to
r
tio
n
t
y
p
e
s
with
o
u
t
ex
p
licit
n
o
is
e
m
o
d
elin
g
.
T
h
e
f
lex
ib
le
en
co
d
er
-
d
ec
o
d
er
s
ch
em
e,
c
o
u
p
led
with
lo
ca
lized
s
k
ip
c
o
n
n
e
ctio
n
s
,
f
ac
ilit
ates im
ag
e
r
esto
r
atio
n
.
Du
r
in
g
in
f
e
r
en
ce
,
th
e
tr
ain
ed
n
etwo
r
k
en
h
an
ce
s
n
o
is
y
e
n
d
o
s
co
p
y
im
ag
es
to
m
i
n
im
ize
d
is
to
r
tio
n
,
s
h
o
wca
s
in
g
th
e
s
y
n
er
g
y
o
f
C
NN
an
d
tr
an
s
f
o
r
m
er
ar
c
h
itectu
r
es
in
d
en
o
is
in
g
.
T
h
o
r
o
u
g
h
ev
alu
atio
n
d
em
o
n
s
tr
ates
s
ig
n
if
ican
t
im
p
r
o
v
em
en
t
in
tis
s
u
e
s
tr
u
ctu
r
e
a
n
d
lesi
o
n
v
is
u
aliza
tio
n
am
i
d
s
t
n
o
is
e.
T
h
e
m
o
d
el
(
Fig
u
r
e
6
)
is
tr
ain
ed
en
d
-
to
-
e
n
d
u
s
in
g
MSE
lo
s
s
.
Her
e
is
a
b
r
ea
k
d
o
wn
o
f
th
e
a
r
ch
itectu
r
e:
a.
I
n
p
u
t:
t
h
e
m
o
d
el
tak
es a
2
5
6
×
2
5
6
-
p
ix
el
in
p
u
t im
ag
e
with
3
co
lo
r
ch
a
n
n
els
r
ed
,
g
r
ee
n
,
an
d
b
lu
e
(
R
GB
)
.
b.
I
n
ce
p
tio
n
V3
:
t
h
is
m
o
d
el
is
p
r
e
-
tr
ain
ed
o
n
th
e
I
m
ag
eNe
t
d
at
aset,
wh
ich
co
n
tain
s
o
v
er
1
4
m
illi
o
n
im
ag
es.
T
h
e
m
o
d
el
h
as
1
0
lay
er
s
an
d
in
clu
d
es
a
v
ar
iety
o
f
la
y
er
s
li
k
e
co
n
v
o
lu
tio
n
al
lay
er
s
,
m
ax
-
p
o
o
lin
g
lay
er
s
,
an
d
b
atch
n
o
r
m
aliza
tio
n
lay
er
s
.
c.
C
o
n
v
o
lu
tio
n
al
a
n
d
b
atc
h
n
o
r
m
aliza
tio
n
lay
er
s
:
th
ese
l
ay
er
s
ar
e
u
s
ed
f
o
r
f
ea
tu
r
e
ex
tr
ac
tio
n
an
d
class
if
icatio
n
.
Af
ter
ea
ch
co
n
v
o
lu
tio
n
al
lay
er
,
a
b
atch
n
o
r
m
aliza
tio
n
lay
er
is
ap
p
lied
to
n
o
r
m
alize
th
e
o
u
tp
u
t.
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:
2502
-
4
7
5
2
C
lea
r
N
e
t:
a
u
to
-
en
c
o
d
er b
a
s
ed
d
en
o
is
in
g
mo
d
el
fo
r
en
d
o
s
co
p
y
ima
g
es
(
V
ikra
n
t S
h
o
ke
en
)
1995
(
)
=
(
∗
+
)
(
1
4
)
ℎ
(
)
=
(
–
)
√
(
²
+
)
⊙
+
(
1
5
)
d.
Up
-
s
am
p
lin
g
lay
e
r
s
:
th
e
up
-
s
am
p
lin
g
lay
e
r
s
ar
e
u
s
ed
t
o
u
p
s
ca
le
th
e
o
u
tp
u
t
o
f
th
e
p
r
ev
io
u
s
lay
er
b
y
a
f
ac
to
r
o
f
2
.
T
h
is
p
r
o
ce
s
s
h
elp
s
in
o
b
tain
in
g
th
e
o
u
tp
u
t
o
f
th
e
s
am
e
s
ize
as th
e
in
p
u
t im
ag
e.
(
)
=
(
,
=
2
)
(
1
6
)
e.
C
o
n
v
o
lu
tio
n
al
lay
e
r
s
:
th
ese
lay
er
s
ar
e
u
s
ed
f
o
r
r
ef
i
n
in
g
th
e
o
u
tp
u
t
f
r
o
m
th
e
p
r
ev
i
o
u
s
lay
e
r
.
T
h
ey
i
n
clu
d
e
2
D
co
n
v
o
lu
tio
n
al
lay
e
r
s
,
ea
ch
f
o
llo
wed
b
y
a
b
atch
n
o
r
m
aliza
tio
n
lay
er
.
2
(
)
=
(
∗
+
)
(
1
7
)
ℎ
(
)
=
(
(
–
)
)
/
√
(
^
2
+
)
⊙
+
(
1
8
)
Fig
u
r
e
6
.
Pro
p
o
s
ed
ar
c
h
itectu
r
e
(
a
-
lef
t to
b
-
r
ig
h
t
)
3.
5
.
Alg
o
rit
hm
ic
ex
press
io
n f
o
r
mo
del
T
h
is
en
co
d
er
-
d
ec
o
d
er
m
o
d
el
c
an
b
e
ex
p
lain
ed
in
th
e
f
o
ll
o
win
g
s
tep
s
:
Ov
er
all
m
o
d
el
ar
c
h
itectu
r
e
:
t
h
e
o
v
er
all
m
o
d
el
ar
ch
itectu
r
e
c
an
b
e
r
e
p
r
esen
ted
as:
=
(
)
+
(
)
(
1
9
)
wh
er
e
x
is
th
e
in
p
u
t im
ag
e,
z
i
s
th
e
f
ea
tu
r
e
v
ec
to
r
,
an
d
m
is
t
h
e
o
u
tp
u
t d
en
o
is
ed
im
ag
e.
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
e
C
lear
Net
is
ev
alu
ated
o
n
th
e
Hy
p
er
Kv
asir
d
ataset
[
2
]
w
ith
1
0
,
6
6
2
im
ag
es
f
r
o
m
2
3
cla
s
s
es,
s
p
li
t
8
0
/2
0
in
to
tr
ain
an
d
v
alid
atio
n
s
ets.
Peak
s
ig
n
al
-
to
-
n
o
is
e
r
a
tio
(
PS
NR
)
,
a
m
etr
ic
q
u
an
tify
in
g
r
ec
o
n
s
tr
u
ctio
n
q
u
ality
co
m
p
ar
e
d
to
th
e
o
r
ig
i
n
al,
is
u
s
ed
to
as
s
ess
d
en
o
is
in
g
f
id
elity
as
s
h
o
wn
in
Fig
u
r
e
7
.
T
o
r
ig
o
r
o
u
s
ly
ev
alu
ate
n
o
is
e
r
o
b
u
s
tn
ess
,
s
p
ec
ialized
tr
ain
in
g
an
d
v
alid
atio
n
d
atasets
ar
e
co
n
s
tr
u
cted
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
38
,
No
.
3
,
J
u
n
e
20
25
:
1
9
9
0
-
200
0
1996
T
h
e
av
er
ag
e
PS
NR
o
f
o
r
ig
in
al
an
d
n
o
is
y
im
ag
es
is
1
9
.
1
1
8
9
5
4
an
d
th
e
PS
NR
o
f
o
r
ig
in
al
an
d
r
ec
o
n
s
tr
u
cted
im
ag
e
is
6
9
.
8
9
2
6
3
1
.
T
h
e
in
c
r
ea
s
e
in
PS
N
R
f
r
o
m
1
9
.
1
2
d
B
to
6
9
.
8
9
d
B
s
h
o
ws
th
at
th
e
n
o
is
e
r
ed
u
ctio
n
p
r
o
ce
s
s
was
h
ig
h
ly
ef
f
ec
tiv
e.
T
h
e
r
ec
o
n
s
tr
u
cte
d
im
ag
e
h
as
f
ar
less
n
o
is
e
an
d
is
m
u
ch
clo
s
er
in
q
u
ality
to
th
e
o
r
ig
in
al
im
ag
e
.
A
PS
N
R
ab
o
v
e
5
0
d
B
g
en
er
ally
in
d
icate
s
n
ea
r
-
p
er
f
ec
t
r
e
co
n
s
tr
u
ctio
n
o
r
a
n
alm
o
s
t
im
p
er
ce
p
tib
le
d
if
f
er
en
ce
b
etwe
en
th
e
o
r
ig
in
al
an
d
th
e
r
ec
o
n
s
tr
u
cted
im
ag
e.
T
h
e
cl
o
s
er
th
e
PS
N
R
i
s
to
0
d
B
,
th
e
m
o
r
e
d
is
s
im
ilar
th
e
n
o
is
y
im
ag
e
is
f
r
o
m
th
e
o
r
ig
in
al,
m
ea
n
in
g
m
o
r
e
n
o
is
e
o
r
d
is
to
r
tio
n
is
p
r
esen
t.
Fig
u
r
e
7
.
PS
NR
v
alu
es
4
.
1
.
G
ener
a
t
io
n
o
f
no
is
y
i
m
a
g
es
T
o
en
a
b
le
th
e
t
r
ain
in
g
an
d
ev
alu
atio
n
o
f
im
ag
e
-
d
en
o
is
in
g
m
o
d
els,
cu
s
to
m
d
atasets
is
g
en
er
ated
b
y
ar
tific
ially
co
r
r
u
p
tin
g
im
ag
es
f
r
o
m
th
e
Hy
p
er
K
v
asir
GI
en
d
o
s
co
p
y
d
ataset.
Sp
ec
if
ically
,
r
an
d
o
m
g
au
s
s
ian
n
o
is
e
is
ad
d
ed
to
th
e
o
r
ig
i
n
al
im
ag
es
to
s
im
u
late
n
o
is
y
ac
q
u
is
itio
n
co
n
d
itio
n
s
.
Gau
s
s
ian
n
o
is
e
is
s
ta
tis
t
ically
g
en
er
ated
f
r
o
m
a
n
o
r
m
al
d
is
tr
i
b
u
tio
n
with
m
ea
n
μ
an
d
v
ar
ia
n
ce
σ
².
I
t
is
co
m
m
o
n
l
y
u
s
ed
t
o
m
o
d
el
n
o
is
e
f
r
o
m
n
atu
r
al
s
o
u
r
ce
s
lik
e
s
en
s
o
r
n
o
i
s
e.
T
h
e
f
o
r
m
u
la
f
o
r
a
g
a
u
s
s
ian
n
o
is
e
d
is
tr
ib
u
tio
n
is
:
(
)
=
1
√
2
2
−
2
(
−
)
2
2
2
(
2
0
)
T
o
r
ig
o
r
o
u
s
ly
ev
alu
ate
n
o
is
e
r
o
b
u
s
tn
ess
,
s
p
ec
ialized
tr
ain
in
g
an
d
v
alid
atio
n
d
atasets
ar
e
co
n
s
tr
u
cted
.
Fo
r
th
e
tr
ain
in
g
s
et,
g
au
s
s
ian
n
o
is
e
with
ze
r
o
m
ea
n
a
n
d
2
0
%
n
o
is
e
lev
el
is
ad
d
ed
to
ea
ch
im
a
g
e.
T
h
e
v
alid
atio
n
s
et
u
s
es
a
h
ig
h
er
3
0
%
n
o
is
e
lev
el,
1
0
%
h
ig
h
er
th
an
tr
ain
in
g
,
to
ev
alu
ate
m
o
d
el
g
en
er
aliza
tio
n
.
T
h
is
s
im
u
lated
n
o
is
e
g
en
er
ati
o
n
p
r
o
v
i
d
es
a
co
n
tr
o
llab
le
w
ay
to
c
r
ea
te
p
air
e
d
n
o
is
y
an
d
clea
n
im
ag
es
f
o
r
tr
ain
in
g
d
e
n
o
is
in
g
m
o
d
els.
E
x
p
o
s
in
g
m
o
d
els
to
v
ar
y
i
n
g
n
o
is
e
lev
els
d
u
r
in
g
tr
ain
i
n
g
h
elp
s
lear
n
r
o
b
u
s
t
r
ep
r
esen
tatio
n
s
tr
an
s
f
er
ab
le
t
o
h
an
d
lin
g
r
ea
l
-
wo
r
ld
n
o
is
e
an
d
d
is
to
r
tio
n
s
,
en
ab
lin
g
r
i
g
o
r
o
u
s
ev
alu
atio
n
an
d
ad
v
an
ce
m
e
n
t o
f
e
n
d
o
s
co
p
y
im
ag
e
d
en
o
is
in
g
tech
n
iq
u
es.
Usi
n
g
a
n
o
is
ier
v
alid
atio
n
s
et
r
ig
o
r
o
u
s
ly
test
s
th
e
m
o
d
el
’
s
g
e
n
er
aliza
tio
n
a
n
d
r
o
b
u
s
tn
ess
ca
p
ab
ilit
ies.
I
f
th
e
m
o
d
el
lear
n
s
n
o
is
e
-
in
v
ar
ian
t
r
ep
r
esen
tatio
n
s
d
u
r
in
g
tr
ain
in
g
,
it
s
h
o
u
ld
h
a
n
d
le
h
ig
h
er
n
o
is
e
lev
els
at
in
f
er
en
ce
,
p
r
ev
en
tin
g
o
v
er
f
itti
n
g
to
th
e
s
p
ec
if
ic
tr
ain
in
g
n
o
is
e
lev
el.
Fig
u
r
e
s
8
an
d
9
s
h
o
w
ex
am
p
les
f
r
o
m
th
e
n
o
is
y
tr
ain
in
g
(
2
0
%
n
o
is
e)
a
n
d
v
alid
atio
n
(
3
0
%
n
o
is
e)
s
ets,
r
esp
ec
tiv
ely
,
with
n
o
is
e
m
a
n
if
esti
n
g
as
g
r
ain
y
s
p
e
c
k
l
es
o
b
s
c
u
r
i
n
g
d
e
t
a
i
ls
a
n
d
e
d
g
e
s
,
p
r
e
s
e
n
t
i
n
g
a
m
a
j
o
r
c
h
a
ll
e
n
g
e
f
o
r
i
m
a
g
e
p
r
o
c
e
s
s
i
n
g
a
l
g
o
r
i
t
h
m
s
.
F
i
g
u
r
e
1
0
d
is
p
lay
s
C
lear
Net
’
s
d
en
o
is
ed
r
ec
o
n
s
tr
u
ctio
n
s
f
o
r
th
e
n
o
is
y
v
alid
atio
n
in
p
u
ts
f
r
o
m
Fig
u
r
e
9
,
ef
f
ec
tiv
ely
f
ilter
in
g
o
u
t
n
o
is
e
an
d
r
ec
o
v
er
in
g
clea
n
er
i
m
ag
es
with
r
esto
r
ed
k
ey
f
ea
tu
r
es
an
d
s
tr
u
ctu
r
es
wh
ile
s
u
p
p
r
ess
in
g
s
p
u
r
io
u
s
n
o
is
e.
Qu
alitativ
ely
,
th
e
m
o
d
el
p
r
o
d
u
ce
s
n
atu
r
al
-
lo
o
k
in
g
r
ec
o
n
s
tr
u
ctio
n
s
clo
s
e
to
th
e
o
r
ig
in
al
g
r
o
u
n
d
tr
u
th
im
a
g
es,
h
ig
h
lig
h
tin
g
its
d
en
o
is
in
g
ca
p
ab
ilit
i
es
an
d
ab
ilit
y
to
lear
n
r
o
b
u
s
t
r
ep
r
esen
tatio
n
s
ca
p
tu
r
in
g
r
ea
l
-
wo
r
ld
im
a
g
e
s
t
r
u
ctu
r
es,
d
is
tin
g
u
is
h
in
g
s
ig
n
al
f
r
o
m
ar
b
itra
r
ily
c
o
r
r
u
p
ted
n
o
is
e
d
u
r
in
g
test
in
g
.
Qu
an
titativ
e
an
aly
s
is
u
s
es
PS
NR
,
a
m
etr
ic
q
u
a
n
tify
in
g
th
e
q
u
ality
o
f
a
r
ec
o
n
s
tr
u
cted
s
ig
n
al
co
m
p
a
r
ed
t
o
its
o
r
ig
in
al
v
er
s
io
n
as
s
h
o
wn
i
n
Fig
u
r
e
7
,
t
o
ass
ess
th
e
f
id
e
lity
o
f
d
en
o
is
ed
im
ag
es.
PS
NR
an
d
p
er
ce
p
tu
al
s
im
ilar
ity
to
g
r
o
u
n
d
tr
u
th
v
ali
d
ate
th
e
q
u
alitativ
e
o
b
s
er
v
atio
n
s
.
4
.
2
.
G
ener
a
t
io
n o
f
no
is
y
i
m
a
g
es
Fo
r
th
e
t
r
ain
in
g
s
et,
g
a
u
s
s
ian
n
o
is
e
with
ze
r
o
m
ea
n
an
d
2
0
%
n
o
is
e
lev
el
is
ad
d
e
d
to
e
ac
h
im
ag
e.
T
h
e
v
alid
atio
n
s
et
u
s
es
a
h
ig
h
er
3
0
%
n
o
is
e
le
v
el
to
ev
alu
ate
m
o
d
el
g
e
n
er
aliza
tio
n
,
1
0
%
h
i
g
h
er
th
an
tr
ain
in
g
,
en
ab
lin
g
r
ig
o
r
o
u
s
test
in
g
o
f
r
o
b
u
s
tn
ess
an
d
p
r
ev
en
tin
g
o
v
er
f
itti
n
g
to
th
e
tr
ain
in
g
n
o
is
e
lev
el.
Fig
u
r
e
s
8
an
d
9
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:
2502
-
4
7
5
2
C
lea
r
N
e
t:
a
u
to
-
en
c
o
d
er b
a
s
ed
d
en
o
is
in
g
mo
d
el
fo
r
en
d
o
s
co
p
y
ima
g
es
(
V
ikra
n
t S
h
o
ke
en
)
1997
s
h
o
w
ex
am
p
les
f
r
o
m
th
e
n
o
is
y
tr
ain
in
g
(
2
0
%
n
o
is
e)
a
n
d
v
a
lid
atio
n
(
3
0
%
n
o
is
e)
s
ets,
r
es
p
ec
tiv
ely
,
with
th
e
n
o
is
e
m
an
if
esti
n
g
as
g
r
ain
y
s
p
ec
k
les
o
b
s
cu
r
in
g
d
etails
an
d
ed
g
es,
p
r
esen
tin
g
a
m
ajo
r
ch
allen
g
e.
Fig
u
r
e
1
0
d
is
p
lay
s
th
e
d
en
o
is
ed
r
ec
o
n
s
t
r
u
ctio
n
s
b
y
C
lear
Net
f
o
r
th
e
n
o
is
y
v
alid
atio
n
in
p
u
ts
f
r
o
m
Fig
u
r
e
9
.
C
lear
Net
ef
f
ec
tiv
ely
f
ilter
s
o
u
t
th
e
n
o
is
e,
r
ec
o
v
er
in
g
clea
n
er
im
ag
es
with
r
esto
r
ed
k
ey
f
ea
tu
r
es
a
n
d
s
tr
u
ctu
r
es
wh
ile
s
u
p
p
r
ess
in
g
s
p
u
r
io
u
s
n
o
is
e.
Qu
alitativ
ely
,
th
e
m
o
d
el
p
r
o
d
u
ce
s
n
atu
r
al
-
l
o
o
k
in
g
r
ec
o
n
s
tr
u
ctio
n
s
clo
s
e
to
th
e
o
r
ig
in
al
g
r
o
u
n
d
tr
u
th
im
ag
es,
h
ig
h
lig
h
tin
g
its
d
e
n
o
is
in
g
ca
p
ab
ilit
ies
an
d
a
b
ilit
y
to
lear
n
r
o
b
u
s
t
r
ep
r
esen
tatio
n
s
th
at
ca
p
tu
r
e
r
ea
l
-
wo
r
ld
im
ag
e
s
tr
u
ctu
r
es,
d
is
tin
g
u
is
h
in
g
s
ig
n
al
f
r
o
m
ar
b
itra
r
ily
co
r
r
u
p
ted
n
o
is
e
d
u
r
in
g
test
in
g
.
Qu
an
tita
tiv
e
an
aly
s
is
o
f
r
ec
o
n
s
tr
u
ctio
n
er
r
o
r
an
d
p
er
ce
p
t
u
al
s
im
ilar
ity
to
th
e
g
r
o
u
n
d
tr
u
th
v
alid
ates th
ese
q
u
alitativ
e
o
b
s
er
v
atio
n
s
.
Fig
u
r
e
8
.
2
0
% n
o
is
e
im
ag
es
Fig
u
r
e
9
.
3
0
% n
o
is
e
im
ag
es
Fig
u
r
e
1
0
.
R
ec
o
n
s
tr
u
cted
im
a
g
es
4
.
3
.
Dis
cus
s
io
n
a
nd
lim
it
a
t
io
ns
o
f
re
s
ea
rc
h
Ou
r
m
o
d
el
s
u
cc
ess
f
u
lly
in
cr
e
ases
th
e
av
er
a
g
e
PS
NR
v
alu
e
o
f
th
e
s
am
p
le
im
ag
es
f
r
o
m
1
9
.
1
2
d
B
to
6
9
.
8
9
d
B
,
wh
ich
s
h
o
wca
s
es
its
ef
f
ec
tiv
en
ess
in
d
en
o
is
in
g
t
h
e
im
ag
e.
T
h
e
PS
NR
v
alu
e
o
f
6
9
.
8
9
d
B
s
h
o
ws
th
at
th
e
r
ec
o
n
s
tr
u
cted
im
ag
es
ar
e
v
er
y
s
im
ilar
to
th
e
o
r
ig
in
al
im
ag
es.
T
h
is
d
en
o
is
in
g
h
as
p
o
ten
tial
to
b
ec
o
m
e
a
v
er
y
im
p
o
r
tan
t
p
a
r
t
o
f
m
e
d
ical
d
iag
n
o
s
is
o
f
GI
d
is
ea
s
es.
T
h
is
r
esear
ch
also
p
av
es
way
f
o
r
au
t
o
m
ate
d
s
y
s
tem
s
wh
ich
ca
n
b
e
cr
ea
ted
to
h
elp
d
o
cto
r
s
in
class
if
y
in
g
d
is
ea
s
ed
b
etter
an
d
f
aster
.
I
n
co
m
p
ar
is
o
n
to
ex
is
tin
g
m
eth
o
d
s
,
o
u
r
r
esear
c
h
in
tr
o
d
u
ce
s
a
cu
s
to
m
en
co
d
er
-
d
ec
o
d
er
m
o
d
el
f
o
r
d
en
o
is
in
g
GI
en
d
o
s
co
p
y
im
ag
es,
wh
ich
s
ig
n
if
ican
tly
o
u
tp
er
f
o
r
m
s
o
th
er
ap
p
r
o
ac
h
es
in
th
is
d
o
m
ai
n
.
J
if
ar
a
et
a
l.
[
1
7
]
u
tili
ze
d
a
C
NN
with
r
esid
u
al
lear
n
in
g
f
o
r
m
ed
ical
im
ag
e
d
en
o
is
in
g
,
ac
h
iev
in
g
a
PS
NR
o
f
4
1
.
6
8
4
3
,
wh
ile
C
h
en
et
a
l.
[
1
6
]
ap
p
lied
a
C
NN
to
l
o
w
-
d
o
s
e
C
T
s
ca
n
s
,
r
ep
o
r
tin
g
a
PS
NR
o
f
4
2
.
1
5
1
4
.
T
h
ese
s
tu
d
ies
p
r
im
ar
ily
f
o
cu
s
ed
o
n
d
if
f
er
en
t
im
ag
in
g
m
o
d
alities
lik
e
C
T
,
with
n
o
tab
le
s
u
cc
ess
i
n
im
p
r
o
v
in
g
im
ag
e
q
u
ality
b
u
t
n
o
t
s
p
ec
if
ic
to
en
d
o
s
co
p
y
.
Ou
r
m
o
d
el,
tailo
r
e
d
f
o
r
th
e
Hy
p
e
r
Kv
asir
d
ataset,
en
h
an
ce
s
th
e
PS
NR
f
r
o
m
1
9
.
1
1
8
9
5
4
to
6
9
.
8
9
2
6
3
1
,
s
h
o
wca
s
in
g
n
ea
r
ly
p
er
f
ec
t
r
ec
o
n
s
tr
u
ctio
n
a
n
d
d
em
o
n
s
tr
atin
g
s
u
p
e
r
io
r
d
en
o
is
in
g
p
er
f
o
r
m
an
ce
in
th
e
co
n
tex
t o
f
e
n
d
o
s
co
p
ic
im
ag
i
n
g
.
T
h
is
h
ig
h
lig
h
ts
th
e
r
o
b
u
s
tn
ess
an
d
ap
p
licab
ilit
y
o
f
o
u
r
ap
p
r
o
ac
h
in
s
ce
n
ar
i
o
s
wh
er
e
p
r
ec
is
e
im
ag
e
r
ec
o
n
s
tr
u
ctio
n
is
cr
itical,
th
u
s
o
f
f
e
r
in
g
a
s
u
b
s
tan
tial
im
p
r
o
v
em
e
n
t
o
v
e
r
tr
ad
itio
n
al
C
NN
-
b
ased
m
o
d
el
in
b
o
t
h
th
e
en
d
o
s
co
p
ic
d
o
m
ain
a
n
d
g
en
e
r
al
m
ed
ical
im
ag
e
d
en
o
is
in
g
.
T
h
is
s
tu
d
y
h
as
s
ev
er
al
lim
itatio
n
s
th
at
s
h
o
u
ld
b
e
ac
k
n
o
wl
ed
g
ed
.
First,
d
u
e
to
co
m
p
u
tat
io
n
al
an
d
r
eso
u
r
ce
c
o
n
s
tr
ain
ts
,
th
e
m
o
d
e
l
co
u
ld
n
o
t
b
e
test
ed
in
r
ea
l
-
w
o
r
ld
clin
ical
en
v
ir
o
n
m
e
n
ts
,
wh
ich
m
ay
af
f
ec
t
th
e
g
en
er
aliza
b
ilit
y
o
f
t
h
e
r
esu
lts
.
All
ev
alu
atio
n
s
wer
e
co
n
d
u
ct
ed
u
s
in
g
th
e
s
am
e
d
ataset,
wh
ich
m
ay
n
o
t
f
u
lly
r
ep
r
esen
t
th
e
v
ar
ia
b
ilit
y
en
c
o
u
n
ter
ed
i
n
p
r
ac
tical
m
ed
ica
l
im
ag
in
g
s
ce
n
ar
io
s
.
Ad
d
itio
n
ally
,
th
e
f
ield
o
f
AI
-
b
ased
d
en
o
is
in
g
f
o
r
m
ed
ic
al
im
ag
e
en
h
an
ce
m
en
t
is
r
elativ
ely
n
ew,
r
esu
ltin
g
in
a
lim
it
ed
am
o
u
n
t
o
f
d
ata
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
38
,
No
.
3
,
J
u
n
e
20
25
:
1
9
9
0
-
200
0
1998
an
d
estab
lis
h
ed
b
en
ch
m
a
r
k
s
f
o
r
co
m
p
ar
is
o
n
.
C
o
n
s
eq
u
e
n
tly
,
wh
ile
o
u
r
ap
p
r
o
ac
h
s
h
o
w
s
p
r
o
m
is
e,
f
u
r
th
er
v
alid
atio
n
with
d
iv
e
r
s
e
d
atasets
an
d
r
ea
l
-
wo
r
ld
ap
p
licatio
n
s
i
s
n
ec
ess
ar
y
to
f
u
lly
ass
ess
its
e
f
f
icac
y
.
5.
CO
NCLU
SI
O
N
C
lear
Net,
a
s
o
p
h
is
ticated
d
e
n
o
is
in
g
m
o
d
el
with
an
I
n
ce
p
tio
n
V3
e
n
co
d
e
r
-
d
ec
o
d
er
ar
ch
itectu
r
e,
d
em
o
n
s
tr
ated
s
ig
n
if
ica
n
t
n
o
is
e
r
ed
u
ctio
n
an
d
im
p
r
o
v
ed
tis
s
u
e
v
is
u
aliza
tio
n
o
n
th
e
d
i
v
er
s
e
Hy
p
er
Kv
asir
en
d
o
s
co
p
y
d
ataset.
T
r
ain
e
d
o
n
im
ag
es
with
2
0
%
ad
d
ed
G
au
s
s
ian
n
o
is
e,
it
s
u
cc
ess
f
u
lly
r
em
o
v
ed
ev
en
3
0
%
n
o
is
e
d
u
r
in
g
test
in
g
,
h
ig
h
lig
h
tin
g
its
r
o
b
u
s
tn
ess
to
h
an
d
le
n
o
is
e
le
v
els
b
ey
o
n
d
tr
ain
in
g
.
Qu
an
titativ
e
an
d
q
u
alitativ
e
r
esu
lts
co
n
f
ir
m
ed
e
n
h
an
ce
d
im
a
g
e
q
u
ality
an
d
r
es
to
r
ed
an
ato
m
ical
d
etails
o
b
s
cu
r
ed
b
y
n
o
is
e.
T
h
is
p
r
e
-
p
r
o
ce
s
s
in
g
s
tep
aid
s
g
ast
r
o
en
ter
o
l
o
g
is
ts
in
d
is
ea
s
e
d
et
ec
tio
n
b
y
im
p
r
o
v
in
g
tis
s
u
e
v
is
u
aliza
tio
n
.
Fu
tu
r
e
wo
r
k
in
clu
d
es
ex
p
an
d
in
g
e
v
a
lu
atio
n
ac
r
o
s
s
n
o
is
e
ty
p
es
an
d
en
d
o
s
co
p
y
m
o
d
alities
,
clin
ical
v
alid
atio
n
,
an
d
ex
ten
d
in
g
th
e
m
o
d
el
f
o
r
j
o
in
t
d
en
o
is
in
g
an
d
an
o
m
aly
s
eg
m
en
tatio
n
.
T
h
e
s
tu
d
y
d
em
o
n
s
tr
ates
th
e
p
r
o
m
is
e
o
f
tr
an
s
f
o
r
m
er
-
b
ased
en
co
d
er
-
d
e
co
d
er
n
etwo
r
k
s
f
o
r
b
lin
d
d
e
n
o
is
in
g
o
f
en
d
o
s
co
p
ic
d
ata,
p
o
ten
tially
en
h
an
cin
g
wo
r
k
f
lo
ws an
d
o
u
tco
m
es f
o
r
GI
d
is
ea
s
e
s
cr
ee
n
in
g
.
ACK
NO
WL
E
DG
M
E
N
T
S
Au
th
o
r
s
ar
e
g
r
atef
u
l to
t
h
eir
r
e
s
p
ec
tiv
e
in
s
titu
tio
n
s
f
o
r
all
f
o
r
m
s
o
f
s
u
p
p
o
r
t.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
T
h
is
r
esear
ch
r
ec
eiv
ed
n
o
s
p
ec
if
ic
g
r
an
t f
r
o
m
a
n
y
f
u
n
d
in
g
ag
en
cy
.
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
Vik
r
an
t Sh
o
k
ee
n
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
San
d
ee
p
Ku
m
ar
✓
✓
✓
✓
✓
✓
✓
✓
✓
Vid
h
u
Ma
th
u
r
✓
✓
✓
✓
✓
✓
✓
✓
✓
Am
it Sh
ar
m
a
✓
✓
✓
✓
✓
✓
I
n
d
r
ajee
t Sin
g
h
✓
✓
✓
✓
✓
✓
Par
ita
J
ain
✓
✓
✓
✓
✓
C
:
C
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
M
:
M
e
t
h
o
d
o
l
o
g
y
So
:
So
f
t
w
a
r
e
Va
:
Va
l
i
d
a
t
i
o
n
Fo
:
Fo
r
mal
a
n
a
l
y
s
i
s
I
:
I
n
v
e
s
t
i
g
a
t
i
o
n
R
:
R
e
so
u
r
c
e
s
D
:
D
a
t
a
C
u
r
a
t
i
o
n
O
:
W
r
i
t
i
n
g
-
O
r
i
g
i
n
a
l
D
r
a
f
t
E
:
W
r
i
t
i
n
g
-
R
e
v
i
e
w
&
E
d
i
t
i
n
g
Vi
:
Vi
su
a
l
i
z
a
t
i
o
n
Su
:
Su
p
e
r
v
i
s
i
o
n
P
:
P
r
o
j
e
c
t
a
d
mi
n
i
st
r
a
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
si
t
i
o
n
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
All a
u
th
o
r
s
d
ec
lar
e
th
at
th
ey
h
av
e
n
o
c
o
n
f
licts
o
f
in
te
r
est.
I
NF
O
RM
E
D
CO
NS
E
N
T
T
h
e
Hy
p
er
Kv
asir
d
ataset
u
s
ed
h
e
r
e,
is
t
h
e
lar
g
est
p
u
b
licly
r
elea
s
ed
GI
tr
ac
t
im
ag
e
d
ata
s
et.
I
t
was
co
llected
with
s
tr
ict
ad
h
er
en
ce
to
eth
ical
g
u
id
elin
es,
i
n
clu
d
in
g
o
b
tain
in
g
in
f
o
r
m
ed
co
n
s
en
t
f
r
o
m
p
atien
ts
.
E
T
H
I
CAL AP
P
RO
V
AL
T
h
e
d
ata
co
llectio
n
in
th
e
d
at
aset
m
en
tio
n
ed
was
co
n
d
u
cte
d
f
o
llo
win
g
eth
ical
g
u
i
d
elin
e
s
an
d
was
ap
p
r
o
v
ed
b
y
th
e
r
elev
a
n
t
in
s
titu
tio
n
al
r
ev
iew
b
o
ar
d
o
r
eth
ics
co
m
m
ittee.
T
h
is
en
s
u
r
es
th
at
th
e
r
esear
c
h
ad
h
er
ed
t
o
n
atio
n
al
r
eg
u
latio
n
s
an
d
in
s
titu
tio
n
al
p
o
licies.
DATA AV
AI
L
AB
I
L
I
T
Y
T
h
e
d
ata
t
h
at
s
u
p
p
o
r
t
th
e
f
i
n
d
in
g
s
o
f
th
is
s
tu
d
y
ar
e
av
a
ilab
le
f
r
o
m
th
e
c
o
r
r
esp
o
n
d
in
g
au
th
o
r
,
[
San
d
ee
p
Ku
m
a
r
]
,
u
p
o
n
r
ea
s
o
n
ab
le
r
eq
u
est.
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:
2502
-
4
7
5
2
C
lea
r
N
e
t:
a
u
to
-
en
c
o
d
er b
a
s
ed
d
en
o
is
in
g
mo
d
el
fo
r
en
d
o
s
co
p
y
ima
g
es
(
V
ikra
n
t S
h
o
ke
en
)
1999
RE
F
E
R
E
NC
E
S
[
1
]
P
.
F
.
N
i
e
d
e
r
e
r
,
J.
H
a
e
f
l
i
g
e
r
,
P
.
B
l
e
ssi
n
g
,
Y
.
Le
h
a
r
e
i
n
g
e
r
,
D
.
D
o
sw
a
l
d
,
a
n
d
N
.
F
e
l
b
e
r
,
“
I
mag
e
q
u
a
l
i
t
y
o
f
e
n
d
o
s
c
o
p
e
s,”
i
n
Bi
o
m
o
n
i
t
o
ri
n
g
a
n
d
E
n
d
o
sc
o
p
y
T
e
c
h
n
o
l
o
g
i
e
s
,
J
a
n
.
2
0
0
1
,
v
o
l
.
4
1
5
8
,
p
p
.
1
–
1
0
,
d
o
i
:
1
0
.
1
1
1
7
/
1
2
.
4
1
3
7
7
8
.
[
2
]
H
.
B
o
r
g
l
i
e
t
a
l
.
,
“
H
y
p
e
r
K
v
a
s
i
r
,
a
c
o
m
p
r
e
h
e
n
si
v
e
m
u
l
t
i
-
c
l
a
ss
i
ma
g
e
a
n
d
v
i
d
e
o
d
a
t
a
se
t
f
o
r
g
a
st
r
o
i
n
t
e
s
t
i
n
a
l
e
n
d
o
sc
o
p
y
,
”
S
c
i
e
n
t
i
f
i
c
D
a
t
a
,
v
o
l
.
7
,
n
o
.
1
,
p
.
2
8
3
,
A
u
g
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
3
8
/
s4
1
5
9
7
-
020
-
0
0
6
2
2
-
y.
[
3
]
M
.
F
.
B
y
r
n
e
e
t
a
l
.
,
“
R
e
a
l
-
t
i
me
d
i
f
f
e
r
e
n
t
i
a
t
i
o
n
o
f
a
d
e
n
o
m
a
t
o
u
s
a
n
d
h
y
p
e
r
p
l
a
st
i
c
d
i
mi
n
u
t
i
v
e
c
o
l
o
r
e
c
t
a
l
p
o
l
y
p
s
d
u
r
i
n
g
a
n
a
l
y
s
i
s
o
f
u
n
a
l
t
e
r
e
d
v
i
d
e
o
s
o
f
st
a
n
d
a
r
d
c
o
l
o
n
o
sco
p
y
u
si
n
g
a
d
e
e
p
l
e
a
r
n
i
n
g
mo
d
e
l
,
”
G
u
t
,
v
o
l
.
6
8
,
n
o
.
1
,
p
p
.
9
4
–
1
0
0
,
Ja
n
.
2
0
1
9
,
d
o
i
:
1
0
.
1
1
3
6
/
g
u
t
j
n
l
-
2
0
1
7
-
3
1
4
5
4
7
.
[
4
]
A
.
K
r
i
z
h
e
v
s
k
y
,
I
.
S
u
t
s
k
e
v
e
r
,
a
n
d
G
.
E.
H
i
n
t
o
n
,
“
I
mag
e
N
e
t
c
l
a
ssi
f
i
c
a
t
i
o
n
w
i
t
h
d
e
e
p
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
s
,
”
Ad
v
a
n
c
e
s i
n
N
e
u
r
a
l
I
n
f
o
rm
a
t
i
o
n
Pro
c
e
ssi
n
g
S
y
st
e
m
s
,
v
o
l
.
2
,
p
p
.
1
0
9
7
–
1
1
0
5
,
2
0
1
2
.
[
5
]
K
.
H
e
,
X
.
Z
h
a
n
g
,
S
.
R
e
n
,
a
n
d
J
.
S
u
n
,
“
D
e
e
p
r
e
si
d
u
a
l
l
e
a
r
n
i
n
g
f
o
r
i
ma
g
e
r
e
c
o
g
n
i
t
i
o
n
,
”
i
n
2
0
1
6
I
EEE
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
p
u
t
e
r
Vi
si
o
n
a
n
d
Pa
t
t
e
rn
Re
c
o
g
n
i
t
i
o
n
(
C
V
P
R)
,
Ju
n
.
2
0
1
6
,
v
o
l
.
2
0
1
6
-
D
e
c
e
m
,
p
p
.
7
7
0
–
7
7
8
,
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
.
2
0
1
6
.
9
0
.
[
6
]
M
.
V
a
n
i
a
,
B
.
A
.
Ta
ma,
H
.
M
a
u
l
a
h
e
l
a
,
a
n
d
S
.
L
i
m,
“
R
e
c
e
n
t
a
d
v
a
n
c
e
s
i
n
a
p
p
l
y
i
n
g
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
a
n
d
d
e
e
p
l
e
a
r
n
i
n
g
t
o
d
e
t
e
c
t
u
p
p
e
r
g
a
st
r
o
i
n
t
e
st
i
n
a
l
t
r
a
c
t
l
e
s
i
o
n
s
,
”
I
EEE
A
c
c
e
ss
,
v
o
l
.
1
1
,
p
p
.
6
6
5
4
4
–
6
6
5
6
7
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
3
.
3
2
9
0
9
9
7
.
[
7
]
D.
-
Y
.
L
i
u
e
t
a
l
.
,
“
I
d
e
n
t
i
f
i
c
a
t
i
o
n
o
f
l
e
s
i
o
n
i
m
a
g
e
s
f
r
o
m
g
a
s
t
r
o
i
n
t
e
s
t
i
n
a
l
e
n
d
o
s
c
o
p
e
b
a
s
e
d
o
n
f
e
a
t
u
r
e
e
x
t
r
a
c
t
i
o
n
o
f
c
o
m
b
i
n
a
t
i
o
n
a
l
m
e
t
h
o
d
s
w
i
t
h
a
n
d
w
i
t
h
o
u
t
l
e
a
r
n
i
n
g
p
r
o
c
e
s
s
,
”
M
e
d
i
c
a
l
I
m
a
g
e
A
n
a
l
y
s
i
s
,
v
o
l
.
3
2
,
p
p
.
2
8
1
–
2
9
4
,
A
u
g
.
2
0
1
6
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
m
e
d
i
a
.
2
0
1
6
.
0
4
.
0
0
7
.
[
8
]
Q
.
V
.
L
e
,
J.
N
g
i
a
m
,
A
.
C
o
a
t
e
s,
A
.
L
a
h
i
r
i
,
B
.
P
r
o
c
h
n
o
w
,
a
n
d
A
.
Y
.
N
g
,
“
O
n
o
p
t
i
mi
z
a
t
i
o
n
m
e
t
h
o
d
s
f
o
r
d
e
e
p
l
e
a
r
n
i
n
g
,
”
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
2
8
t
h
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
M
a
c
h
i
n
e
L
e
a
rn
i
n
g
,
I
C
ML 2
0
1
1
,
p
p
.
2
6
5
–
2
7
2
,
2
0
1
1
.
[
9
]
C
.
H
a
ss
a
n
e
t
a
l
.
,
“
R
e
a
l
-
t
i
me
c
o
m
p
u
t
e
r
-
a
i
d
e
d
d
e
t
e
c
t
i
o
n
o
f
c
o
l
o
r
e
c
t
a
l
n
e
o
p
l
a
si
a
d
u
r
i
n
g
c
o
l
o
n
o
sc
o
p
y
,
”
A
n
n
a
l
s
o
f
I
n
t
e
r
n
a
l
M
e
d
i
c
i
n
e
,
v
o
l
.
1
7
6
,
n
o
.
9
,
p
p
.
1
2
0
9
–
1
2
2
0
,
2
0
2
3
,
d
o
i
:
1
0
.
7
3
2
6
/
m2
2
-
3
6
7
8
.
[
1
0
]
A
.
O
u
b
a
a
l
l
a
,
H
.
El
M
o
u
b
t
a
h
i
j
,
a
n
d
N
.
El
A
k
k
a
d
,
“
M
e
d
i
c
a
l
i
ma
g
e
se
g
m
e
n
t
a
t
i
o
n
u
si
n
g
d
e
e
p
l
e
a
r
n
i
n
g
:
a
s
u
r
v
e
y
,
”
i
n
L
e
c
t
u
re
N
o
t
e
s
i
n
N
e
t
w
o
rks
a
n
d
S
y
st
e
m
s
,
v
o
l
.
6
6
9
LN
N
S
,
2
0
2
3
,
p
p
.
9
7
4
–
9
8
3
.
[
1
1
]
S
.
Ta
n
w
a
r
,
S
.
V
i
j
a
y
a
l
a
k
sh
m
i
,
M
.
S
a
b
h
a
r
w
a
l
,
M
.
K
a
u
r
,
A
.
A
.
A
l
Z
u
b
i
,
a
n
d
H
.
-
N
.
Le
e
,
“
[
R
e
t
r
a
c
t
e
d
]
d
e
t
e
c
t
i
o
n
a
n
d
c
l
a
ssi
f
i
c
a
t
i
o
n
o
f
c
o
l
o
r
e
c
t
a
l
p
o
l
y
p
u
si
n
g
d
e
e
p
l
e
a
r
n
i
n
g
,
”
Bi
o
M
e
d
Re
s
e
a
r
c
h
I
n
t
e
r
n
a
t
i
o
n
a
l
,
v
o
l
.
2
0
2
2
,
n
o
.
1
,
Ja
n
.
2
0
2
2
,
d
o
i
:
1
0
.
1
1
5
5
/
2
0
2
2
/
2
8
0
5
6
0
7
.
[
1
2
]
K
.
O
k
su
z
,
B
.
C
.
C
a
m,
S
.
K
a
l
k
a
n
,
a
n
d
E.
A
k
b
a
s
,
“
I
mb
a
l
a
n
c
e
p
r
o
b
l
e
ms
i
n
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
:
a
r
e
v
i
e
w
,
”
I
E
EE
T
ra
n
s
a
c
t
i
o
n
s
o
n
Pa
t
t
e
r
n
An
a
l
y
si
s
a
n
d
Ma
c
h
i
n
e
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
4
3
,
n
o
.
1
0
,
p
p
.
3
3
8
8
–
3
4
1
5
,
O
c
t
.
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
TPA
M
I
.
2
0
2
0
.
2
9
8
1
8
9
0
.
[
1
3
]
S
.
Z.
N
.
C
a
r
i
o
n
,
F
.
M
a
ssa
,
G
.
S
y
n
n
a
e
v
e
,
N
.
U
su
n
i
e
r
,
A
.
K
i
r
i
l
l
o
v
,
“
E
n
d
-
to
-
e
n
d
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
w
i
t
h
t
r
a
n
sf
o
r
me
r
s,”
i
n
Eu
r
o
p
e
a
n
c
o
n
f
e
re
n
c
e
o
n
c
o
m
p
u
t
e
r v
i
si
o
n
,
2
0
1
3
,
p
p
.
2
1
3
–
2
2
9
,
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s:
/
/
g
i
t
h
u
b
.
c
o
m
/
f
a
c
e
b
o
o
k
r
e
se
a
r
c
h
/
d
e
t
r
.
[
1
4
]
A
.
R
a
v
i
s
h
a
n
k
a
r
,
S
.
A
n
u
sh
a
,
H
.
K
.
A
k
sh
a
t
h
a
,
A
.
R
a
j
,
S
.
Ja
h
n
a
v
i
,
a
n
d
J.
M
a
d
h
u
r
a
,
“
A
su
r
v
e
y
o
n
n
o
i
se
r
e
d
u
c
t
i
o
n
t
e
c
h
n
i
q
u
e
s
i
n
med
i
c
a
l
i
ma
g
e
s,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
El
e
c
t
r
o
n
i
c
s
,
C
o
m
m
u
n
i
c
a
t
i
o
n
a
n
d
Ae
r
o
s
p
a
c
e
T
e
c
h
n
o
l
o
g
y
,
I
C
EC
A
2
0
1
7
,
A
p
r
.
2
0
1
7
,
v
o
l
.
2
0
1
7
-
J
a
n
u
a
r
y
,
p
p
.
3
8
5
–
3
8
9
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
EC
A
.
2
0
1
7
.
8
2
0
3
7
1
1
.
[
1
5
]
G
.
Li
t
j
e
n
s
e
t
a
l
.
,
“
A
s
u
r
v
e
y
o
n
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
a
n
a
l
y
si
s
,
”
Me
d
i
c
a
l
I
m
a
g
e
An
a
l
y
si
s
,
v
o
l
.
4
2
,
p
p
.
6
0
–
8
8
,
D
e
c
.
2
0
1
7
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
m
e
d
i
a
.
2
0
1
7
.
0
7
.
0
0
5
.
[
1
6
]
H
.
C
h
e
n
e
t
a
l
.
,
“
L
o
w
-
d
o
s
e
C
T
d
e
n
o
i
si
n
g
w
i
t
h
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
,
”
i
n
2
0
1
7
I
EE
E
1
4
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
S
y
m
p
o
si
u
m
o
n
Bi
o
m
e
d
i
c
a
l
I
m
a
g
i
n
g
(
I
S
BI
2
0
1
7
)
,
A
p
r
.
2
0
1
7
,
p
p
.
1
4
3
–
1
4
6
,
d
o
i
:
1
0
.
1
1
0
9
/
I
S
B
I
.
2
0
1
7
.
7
9
5
0
4
8
8
.
[
1
7
]
W
.
Ji
f
a
r
a
,
F
.
Ji
a
n
g
,
S
.
R
h
o
,
M
.
C
h
e
n
g
,
a
n
d
S
.
Li
u
,
“
M
e
d
i
c
a
l
i
m
a
g
e
d
e
n
o
i
s
i
n
g
u
si
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
:
a
r
e
s
i
d
u
a
l
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
,
”
T
h
e
J
o
u
rn
a
l
o
f
S
u
p
e
r
c
o
m
p
u
t
i
n
g
,
v
o
l
.
7
5
,
n
o
.
2
,
p
p
.
7
0
4
–
7
1
8
,
F
e
b
.
2
0
1
9
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
1
2
2
7
-
0
1
7
-
2
0
8
0
-
0.
[
1
8
]
O
.
R
o
n
n
e
b
e
r
g
e
r
,
P
.
F
i
sc
h
e
r
,
a
n
d
T.
B
r
o
x
,
“
U
-
N
e
t
:
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
t
w
o
r
k
s
f
o
r
b
i
o
m
e
d
i
c
a
l
i
m
a
g
e
se
g
m
e
n
t
a
t
i
o
n
,
”
i
n
L
e
c
t
u
r
e
N
o
t
e
s
i
n
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
(
i
n
c
l
u
d
i
n
g
s
u
b
ser
i
e
s
L
e
c
t
u
re
N
o
t
e
s
i
n
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
L
e
c
t
u
re
N
o
t
e
s
i
n
B
i
o
i
n
f
o
rm
a
t
i
c
s)
,
v
o
l
.
9
3
5
1
,
2
0
1
5
,
p
p
.
2
3
4
–
2
4
1
.
[
1
9
]
M
.
Ta
h
mi
d
,
M
.
S
.
A
l
a
m,
N
.
R
a
o
,
a
n
d
K
.
M
.
A
.
A
sh
r
a
f
i
,
“
I
mag
e
-
to
-
i
ma
g
e
t
r
a
n
sl
a
t
i
o
n
w
i
t
h
c
o
n
d
i
t
i
o
n
a
l
a
d
v
e
r
sar
i
a
l
n
e
t
w
o
r
k
s,
”
i
n
2
0
2
3
I
EEE
9
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
Wo
m
e
n
i
n
E
n
g
i
n
e
e
r
i
n
g
(
WIE)
C
o
n
f
e
re
n
c
e
o
n
El
e
c
t
r
i
c
a
l
a
n
d
C
o
m
p
u
t
e
r
E
n
g
i
n
e
e
ri
n
g
(
WIE
C
O
N
-
EC
E)
,
N
o
v
.
2
0
2
3
,
p
p
.
1
–
5
,
d
o
i
:
1
0
.
1
1
0
9
/
W
I
EC
O
N
-
EC
E
6
0
3
9
2
.
2
0
2
3
.
1
0
4
5
6
4
4
7
.
[
2
0
]
J.
Le
h
t
i
n
e
n
e
t
a
l
.
,
“
N
o
i
se
2
N
o
i
s
e
:
l
e
a
r
n
i
n
g
i
m
a
g
e
r
e
s
t
o
r
a
t
i
o
n
w
i
t
h
o
u
t
c
l
e
a
n
d
a
t
a
,
”
3
5
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
Ma
c
h
i
n
e
L
e
a
r
n
i
n
g
,
I
C
ML
2
0
1
8
,
v
o
l
.
7
,
p
p
.
4
6
2
0
–
4
6
3
1
,
2
0
1
8
.
[
2
1
]
K
.
H
e
e
t
a
l
.
,
“
Tr
a
n
sf
o
r
mers
i
n
med
i
c
a
l
i
ma
g
e
a
n
a
l
y
si
s,
”
I
n
t
e
l
l
i
g
e
n
t
Me
d
i
c
i
n
e
,
v
o
l
.
3
,
n
o
.
1
,
p
p
.
5
9
–
7
8
,
F
e
b
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
i
me
d
.
2
0
2
2
.
0
7
.
0
0
2
.
[
2
2
]
A
.
A
l
s
a
i
a
r
i
,
R
.
R
u
st
a
g
i
,
A
.
A
l
h
a
k
a
my
,
M
.
M
.
Th
o
m
a
s,
a
n
d
A
.
G
.
F
o
r
b
e
s,
“
I
mag
e
d
e
n
o
i
si
n
g
u
s
i
n
g
a
g
e
n
e
r
a
t
i
v
e
a
d
v
e
r
sar
i
a
l
n
e
t
w
o
r
k
,
”
2
0
1
9
I
EE
E
2
n
d
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
I
n
f
o
rm
a
t
i
o
n
a
n
d
C
o
m
p
u
t
e
r Te
c
h
n
o
l
o
g
i
e
s,
I
C
I
C
T
2
0
1
9
,
p
p
.
1
2
6
–
1
3
2
,
2
0
1
9
,
d
o
i
:
1
0
.
1
1
0
9
/
I
N
F
O
C
T.
2
0
1
9
.
8
7
1
0
8
9
3
.
[
2
3
]
C
.
S
z
e
g
e
d
y
,
V
.
V
a
n
h
o
u
c
k
e
,
S
.
I
o
f
f
e
,
J.
S
h
l
e
n
s,
a
n
d
Z
.
W
o
j
n
a
,
“
R
e
t
h
i
n
k
i
n
g
t
h
e
i
n
c
e
p
t
i
o
n
a
r
c
h
i
t
e
c
t
u
r
e
f
o
r
c
o
mp
u
t
e
r
v
i
si
o
n
,
”
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
I
E
EE
C
o
m
p
u
t
e
r
S
o
c
i
e
t
y
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
V
i
si
o
n
a
n
d
P
a
t
t
e
r
n
R
e
c
o
g
n
i
t
i
o
n
,
v
o
l
.
2
0
1
6
-
D
e
c
e
mb
e
r
,
p
p
.
2
8
1
8
–
2
8
2
6
,
2
0
1
6
,
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
.
2
0
1
6
.
3
0
8
.
[
2
4
]
Q
.
D
o
u
e
t
a
l
.
,
“
A
u
t
o
m
a
t
i
c
d
e
t
e
c
t
i
o
n
o
f
c
e
r
e
b
r
a
l
mi
c
r
o
b
l
e
e
d
s
f
r
o
m
M
R
i
ma
g
e
s
v
i
a
3
D
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
s,
”
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
.
3
5
,
n
o
.
5
,
p
p
.
1
1
8
2
–
1
1
9
5
,
2
0
1
6
,
d
o
i
:
1
0
.
1
1
0
9
/
TM
I
.
2
0
1
6
.
2
5
2
8
1
2
9
.
[
2
5
]
K
.
D
.
G
u
p
t
a
,
D
.
K
.
S
h
a
r
m
a
,
S
.
A
h
me
d
,
H
.
G
u
p
t
a
,
D
.
G
u
p
t
a
,
a
n
d
C
.
H
.
H
su
,
“
A
n
o
v
e
l
l
i
g
h
t
w
e
i
g
h
t
d
e
e
p
l
e
a
r
n
i
n
g
-
b
a
s
e
d
h
i
s
t
o
p
a
t
h
o
l
o
g
i
c
a
l
i
ma
g
e
c
l
a
ss
i
f
i
c
a
t
i
o
n
m
o
d
e
l
f
o
r
I
o
M
T,
”
N
e
u
ra
l
Pro
c
e
ssi
n
g
L
e
t
t
e
rs
,
v
o
l
.
5
5
,
n
o
.
1
,
p
p
.
2
0
5
–
2
2
8
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
1
0
6
3
-
021
-
1
0
5
5
5
-
1.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Dr
.
Vik
r
a
n
t
S
h
o
k
e
e
n
wo
rk
s
o
n
t
h
e
se
g
m
e
n
tatio
n
,
fe
a
tu
re
e
x
tr
a
c
ti
o
n
,
c
las
sifica
ti
o
n
o
f
ima
g
e
s.
He
u
se
s
v
a
rio
u
s
tec
h
n
iq
u
e
s
li
k
e
P
CA,
S
IF
T
,
a
n
d
S
U
RF
fo
r
fe
a
tu
re
e
x
trac
ti
o
n
a
n
d
S
VM,
LDA,
KN
N fo
r
c
las
sifica
ti
o
n
.
He
is p
a
rti
c
u
larl
y
in
tere
ste
d
in
h
y
b
ri
d
a
p
p
ro
a
c
h
e
s li
k
e
S
V
M
in
teg
ra
ted
wit
h
fe
e
d
f
o
rwa
rd
b
a
c
k
p
r
o
p
a
g
a
ti
o
n
n
e
u
ra
l
n
e
two
rk
t
o
c
re
a
te
a
h
y
b
ri
d
a
l
g
o
rit
h
m
t
h
a
t
fu
rth
e
r
h
e
lp
s
in
re
d
u
c
i
n
g
th
e
c
o
m
p
u
tati
o
n
c
o
m
p
lex
i
ty
o
f
th
e
c
las
sif
ica
ti
o
n
.
His
p
re
se
n
t
re
se
a
rc
h
a
re
a
is
fo
c
u
se
d
o
n
th
e
fa
c
tu
a
l
fi
n
d
in
g
s
o
f
t
h
e
p
o
te
n
ti
a
l
u
sa
g
e
o
f
t
h
e
c
o
m
b
in
a
ti
o
n
a
l
fe
e
d
-
fo
rwa
rd
b
a
c
k
p
r
o
p
a
g
a
ti
o
n
n
e
u
ra
l
n
e
two
r
k
a
s
a
ju
d
g
e
m
e
n
t
m
a
k
in
g
fo
r
d
ise
a
se
d
e
tec
ti
o
n
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
sh
o
k
e
e
n
1
8
@g
m
a
il
.
c
o
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.