I
AE
S In
t
er
na
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI
)
Vo
l.
1
4
,
No
.
6
,
Dec
em
b
er
2
0
2
5
,
p
p
.
4
5
4
1
~
4
5
5
1
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
14
.i
6
.
p
p
4
5
4
1
-
4
5
5
1
4541
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
Co
mbinin
g
conv
o
lutiona
l opera
tors
in
unsu
p
ervi
sed
networks
for kidney
abno
r
ma
lities
Aek
k
a
r
a
t
Su
k
s
uk
o
nt
1
,
Anur
uk
P
ro
mm
a
k
ho
t
2
,
J
a
k
k
re
e
Srino
ncha
t
3
1
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
En
g
i
n
e
e
r
i
n
g
,
F
a
c
u
l
t
y
o
f
I
n
d
u
s
t
r
i
a
l
Ed
u
c
a
t
i
o
n
,
R
a
j
a
ma
n
g
a
l
a
U
n
i
v
e
r
s
i
t
y
o
f
T
e
c
h
n
o
l
o
g
y
S
u
v
a
r
n
a
b
h
u
mi
,
S
u
p
h
a
n
b
u
r
i
,
Th
a
i
l
a
n
d
2
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
En
g
i
n
e
e
r
i
n
g
,
F
a
c
u
l
t
y
o
f
En
g
i
n
e
e
r
i
n
g
,
R
a
j
a
m
a
n
g
a
l
a
U
n
i
v
e
r
s
i
t
y
o
f
T
e
c
h
n
o
l
o
g
y
T
h
a
n
y
a
b
u
r
i
,
P
a
t
h
u
m
T
h
a
n
i
,
T
h
a
i
l
a
n
d
3
D
e
p
a
r
t
m
e
n
t
o
f
E
l
e
c
t
r
o
n
i
c
s
a
n
d
T
e
l
e
c
o
m
m
u
n
i
c
a
t
i
o
n
E
n
g
i
n
e
e
r
i
n
g
,
F
a
c
u
l
t
y
o
f
E
n
g
i
n
e
e
r
i
n
g
,
R
a
j
a
m
a
n
g
a
l
a
U
n
i
v
e
r
s
i
t
y
o
f
T
e
c
h
n
o
l
o
g
y
T
h
a
n
y
a
b
u
r
i
,
P
a
t
h
u
m
T
h
a
n
i
,
T
h
a
i
l
a
n
d
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Oct
3
1
,
2
0
2
4
R
ev
is
ed
Sep
1
2
,
2
0
2
5
Acc
ep
ted
Oct
1
8
,
2
0
2
5
De
e
p
lea
rn
in
g
p
la
y
s
a
p
iv
o
tal
ro
le
in
a
d
v
a
n
c
in
g
th
e
d
iag
n
o
sis
o
f
re
n
a
l
d
y
sfu
n
c
ti
o
n
,
a
c
h
iev
in
g
p
e
rfo
rm
a
n
c
e
lev
e
ls
c
o
m
p
a
ra
b
le
to
t
h
o
se
o
f
m
e
d
ica
l
e
x
p
e
rts.
Ho
we
v
e
r,
d
ise
a
se
d
o
m
a
in
v
a
riati
o
n
s
a
n
d
m
o
d
e
l
d
iffer
e
n
c
e
s
c
a
n
imp
a
c
t
lea
rn
in
g
q
u
a
li
t
y
.
To
a
d
d
re
ss
re
n
a
l
d
y
sf
u
n
c
ti
o
n
,
we
p
r
o
p
o
se
d
u
a
l
-
stre
a
m
c
o
n
v
o
l
u
ti
o
n
a
l
(DSC)
a
n
d
d
u
a
l
-
i
n
p
u
t
c
o
n
v
o
lu
t
io
n
a
l
(
DIC)
fo
r
u
n
su
p
e
rv
ise
d
lea
rn
i
n
g
.
Th
e
p
r
o
p
o
se
d
n
e
two
r
k
is
d
e
sig
n
e
d
to
p
ro
c
e
ss
m
u
lt
i
-
sc
a
le
d
a
ta
a
n
d
e
m
p
lo
y
s
p
a
ra
ll
e
l
d
a
ta
a
g
g
re
g
a
ti
o
n
t
o
e
n
h
a
n
c
e
lea
rn
in
g
c
a
p
a
b
il
it
ies
,
imp
ro
v
in
g
th
e
re
li
a
b
il
it
y
o
f
th
e
e
x
p
e
rime
n
tal
re
su
lt
s.
DSC
a
c
h
iev
e
d
train
in
g
l
o
ss
e
s
o
f
0
.
0
0
6
9
,
0
.
0
0
5
6
,
0
.
0
0
4
2
,
a
n
d
0
.
0
0
4
8
f
o
r
n
o
rm
a
l,
c
y
st,
sto
n
e
,
a
n
d
t
u
m
o
r
d
a
tas
e
ts,
re
sp
e
c
ti
v
e
ly
,
wh
il
e
DIC
a
c
h
iev
e
d
lo
ss
e
s
o
f
0
.
0
0
6
6
,
0
.
0
0
6
3
,
0
.
0
0
4
4
,
a
n
d
0
.
0
0
5
8
fo
r
th
e
sa
m
e
c
a
teg
o
ries
.
T
h
e
e
x
p
e
rime
n
tal
re
su
lt
s d
e
m
o
n
stra
te t
h
a
t
o
u
r
p
ro
p
o
se
d
m
o
d
e
ls
o
u
tp
e
rf
o
rm
sta
te
-
of
-
th
e
-
a
rt
a
p
p
ro
a
c
h
e
s,
m
a
k
i
n
g
t
h
e
m
we
ll
-
su
it
e
d
fo
r
b
r
o
a
d
a
p
p
l
ica
ti
o
n
i
n
c
li
n
ica
l
re
se
a
rc
h
stu
d
ies
.
K
ey
w
o
r
d
s
:
C
o
m
b
in
ed
co
n
v
o
lu
tio
n
al
o
p
er
ato
r
C
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
Dee
p
lear
n
in
g
Kid
n
ey
ab
n
o
r
m
ality
Un
s
u
p
er
v
is
ed
lear
n
in
g
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
:
J
ak
k
r
ee
Srin
o
n
c
h
at
Dep
ar
tm
en
t o
f
E
lectr
o
n
ics an
d
T
elec
o
m
m
u
n
icatio
n
E
n
g
in
ee
r
in
g
,
Facu
lty
o
f
E
n
g
in
ee
r
in
g
R
ajam
an
g
ala
Un
iv
er
s
ity
o
f
T
e
ch
n
o
lo
g
y
T
h
a
n
y
ab
u
r
i
Path
u
m
T
h
an
i,
T
h
ailan
d
E
m
ail: ja
k
k
r
ee
.
s
@
en
.
r
m
u
tt.a
c.
th
1.
I
NT
RO
D
UCT
I
O
N
D
ee
p
lear
n
in
g
h
as
g
ain
ed
wid
e
ac
ce
p
tan
ce
ac
r
o
s
s
v
ar
io
u
s
r
e
s
ea
r
ch
f
ield
s
,
p
ar
ticu
lar
ly
in
t
h
e
m
ed
ical
d
o
m
ain
,
d
u
e
to
its
ab
ilit
y
to
c
o
llect
d
ata
an
d
lear
n
co
m
p
lex
f
ea
tu
r
es
o
r
p
atter
n
s
f
o
r
d
etec
tio
n
,
s
eg
m
en
tatio
n
,
class
if
icatio
n
,
an
d
r
ec
o
g
n
itio
n
,
esp
ec
ially
in
d
iag
n
o
s
in
g
d
is
ea
s
es
th
r
o
u
g
h
m
ag
n
etic
r
eso
n
a
n
ce
im
ag
in
g
(
MRI
)
,
X
-
r
ay
,
o
r
m
icr
o
s
co
p
ic
im
ag
es
[
1
]
–
[
3
]
.
W
h
ile
d
ee
p
lear
n
in
g
y
ield
s
h
ig
h
ly
ef
f
icien
t
r
esu
lts
,
u
n
d
er
s
tan
d
in
g
th
e
ess
en
tial
co
m
p
o
n
en
ts
o
f
d
is
ea
s
e,
s
u
ch
as
r
eg
i
o
n
s
o
f
in
ter
est
o
r
d
is
ea
s
e
p
o
s
itio
n
s
in
m
e
d
ic
al
im
ag
es,
r
em
ain
s
cr
u
cial
f
o
r
e
x
p
lain
in
g
d
is
ea
s
e
h
allm
ar
k
s
an
d
th
e
n
etwo
r
k
'
s
ab
ilit
y
to
lear
n
th
e
p
at
h
o
g
e
n
esis
[
4
]
.
Ov
er
lap
p
in
g
f
ea
tu
r
es,
wh
ich
o
b
s
cu
r
e
cr
iti
ca
l
d
is
ea
s
e
co
m
p
o
n
en
ts
[
5
]
,
an
d
n
o
is
e
in
th
e
im
a
g
es
[
6
]
,
[
7
]
ar
e
s
ig
n
if
ica
n
t
f
ac
to
r
s
th
at
im
p
ac
t
lear
n
in
g
an
d
ca
n
lead
to
n
o
is
e
s
ig
n
als
b
ef
o
r
e
er
r
o
r
s
o
cc
u
r
.
T
h
ese
ch
alle
n
g
es,
m
ain
ly
wh
e
n
d
ea
lin
g
with
n
etwo
r
k
s
task
ed
with
lear
n
in
g
th
e
id
e
n
tity
o
r
p
atter
n
s
o
f
o
v
er
lap
p
i
n
g
d
is
ea
s
es,
s
u
ch
as
C
OVI
D
-
1
9
an
d
k
id
n
e
y
a
b
n
o
r
m
alities
in
X
-
r
ay
im
a
g
es
[
8
]
,
[
9
]
,
o
r
b
r
ain
tu
m
o
r
s
an
d
k
id
n
ey
d
is
ea
s
e
in
MRI
s
ca
n
s
[
1
0
]
,
[
1
1
]
,
p
o
s
e
lim
itatio
n
s
th
at
r
eq
u
ir
e
ca
r
ef
u
l a
tten
tio
n
.
I
n
r
ec
en
t
y
ea
r
s
,
d
ee
p
lear
n
in
g
tech
n
iq
u
es
h
av
e
b
ee
n
d
ev
elo
p
ed
to
en
h
an
ce
th
e
ef
f
icien
cy
o
f
lear
n
in
g
co
m
p
lex
d
is
ea
s
e
co
m
p
o
n
en
ts
an
d
h
an
d
lin
g
d
is
ea
s
e
f
ea
tu
r
es.
T
h
ese
ad
v
an
ce
m
en
ts
h
a
v
e
led
to
n
o
tab
le
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
4
,
No
.
6
,
Dec
em
b
er
2
0
2
5
:
4
5
4
1
-
4
5
5
1
4542
p
er
f
o
r
m
an
ce
im
p
r
o
v
em
en
ts
,
s
u
ch
as
en
h
an
cin
g
co
n
v
o
lu
ti
o
n
al
n
eu
r
al
n
etwo
r
k
(
C
NN
)
p
ar
alleliza
tio
n
with
r
esid
u
al
f
u
s
io
n
to
m
itig
ate
v
an
is
h
in
g
g
r
a
d
ien
t
is
s
u
es
an
d
ap
p
ly
in
g
Dr
o
p
B
lo
ck
tech
n
iq
u
es
t
o
b
etter
ex
tr
ac
t
k
e
y
f
ea
tu
r
es
f
r
o
m
co
m
p
lex
m
ed
i
ca
l
im
ag
es
[
1
2
]
.
T
h
e
in
teg
r
atio
n
o
f
v
is
u
al
g
e
o
m
etr
y
g
r
o
u
p
(
VGG
)
-
1
6
an
d
VGG
-
1
9
h
as
en
h
an
ce
d
th
e
a
b
ilit
y
to
ca
p
tu
r
e
b
o
th
s
h
allo
w
an
d
d
ee
p
f
ea
tu
r
es,
th
er
e
b
y
im
p
r
o
v
in
g
m
ed
ical
in
s
ig
h
t d
etec
tio
n
[
1
3
]
.
T
h
e
d
ev
elo
p
m
en
t
o
f
th
e
ad
a
p
tiv
e
h
y
b
r
id
ized
d
ee
p
C
NN
f
o
cu
s
es
o
n
r
esid
u
al
lear
n
in
g
a
n
d
d
ilated
co
n
v
o
l
u
tio
n
s
to
ad
d
r
ess
m
is
s
in
g
in
f
o
r
m
atio
n
,
allo
win
g
f
o
r
b
r
o
ad
e
r
an
d
m
o
r
e
d
etailed
f
ea
tu
r
e
ex
tr
ac
tio
n
.
I
t
in
co
r
p
o
r
ates
au
to
e
n
co
d
e
r
s
an
d
s
k
ip
co
n
n
ec
tio
n
s
to
p
a
s
s
in
f
o
r
m
atio
n
to
d
ee
p
e
r
lay
er
s
an
d
r
e
d
u
ce
th
e
d
im
en
s
io
n
ality
o
f
c
o
m
p
lex
im
ag
e
d
ata
f
o
r
m
o
r
e
ac
cu
r
ate
d
is
ea
s
e
r
ec
o
g
n
itio
n
in
lar
g
e
im
ag
es
[
1
4
]
.
A
m
o
d
if
ied
d
ee
p
b
elief
n
etwo
r
k
h
as
also
b
ee
n
d
esig
n
ed
as
a
class
if
icatio
n
alg
o
r
ith
m
f
o
r
p
r
e
d
ic
tin
g
k
id
n
e
y
-
r
elate
d
d
is
ea
s
es,
u
s
in
g
So
f
tm
ax
as
th
e
ac
tiv
atio
n
f
u
n
ctio
n
an
d
ca
teg
o
r
ical
cr
o
s
s
-
en
tr
o
p
y
as
th
e
lo
s
s
f
u
n
ctio
n
f
o
r
ea
r
ly
d
etec
tio
n
o
f
c
h
r
o
n
ic
d
is
ea
s
es
[
1
5
]
.
T
h
e
y
o
u
o
n
ly
lo
o
k
o
n
ce
v
er
s
i
o
n
8
(
YOL
Ov
8
)
m
o
d
el
h
as
b
ee
n
em
p
lo
y
e
d
to
an
aly
ze
d
if
f
er
en
tiated
m
ed
ical
f
ea
tu
r
es,
d
em
o
n
s
tr
atin
g
its
ef
f
ec
tiv
e
n
ess
in
d
is
ea
s
e
d
etec
tio
n
an
d
d
iag
n
o
s
is
[
1
6
]
.
Ad
d
itio
n
ally
,
d
ee
p
lear
n
in
g
h
as
b
ee
n
ap
p
lied
to
ad
d
r
ess
th
e
ch
allen
g
e
o
f
d
is
ea
s
e
ch
ar
ac
ter
is
tics
th
at
ar
is
e
f
r
o
m
th
e
d
i
v
er
s
ity
o
f
im
ag
e
d
ata,
s
u
ch
as
u
s
in
g
m
u
lti
-
s
ca
le
lear
n
in
g
to
h
an
d
le
v
ar
y
in
g
d
ata
co
m
p
lex
ities
.
Netwo
r
k
s
ar
e
o
f
te
n
d
esig
n
ed
with
d
if
f
er
e
n
t
r
ec
e
p
tiv
e
f
ield
s
o
r
co
n
v
o
lu
tio
n
al
o
p
er
ato
r
s
(
m
u
lti
-
s
ca
le)
to
ca
p
tu
r
e
tar
g
et
p
atch
es
ef
f
ec
tiv
ely
[
1
7
]
,
[
1
8
]
,
p
r
o
v
id
in
g
s
ig
n
if
ican
t d
esig
n
a
d
v
an
ta
g
e
s
o
v
er
tr
ad
itio
n
al
m
ac
h
in
e
lear
n
in
g
m
eth
o
d
s
.
I
n
p
r
e
v
i
o
u
s
w
o
r
k
s
[
1
2
]
–
[
1
8
]
,
t
h
e
y
i
n
v
es
t
i
g
at
e
d
t
h
e
e
f
f
e
c
tiv
e
n
e
s
s
o
f
d
e
e
p
l
ea
r
n
i
n
g
a
p
p
r
o
a
c
h
e
s
i
n
a
d
d
r
e
s
s
i
n
g
c
h
al
l
e
n
g
es
r
e
l
at
e
d
t
o
i
m
a
g
i
n
g
a
n
d
i
m
p
r
o
v
i
n
g
l
ea
r
n
i
n
g
o
u
t
c
o
m
e
s
f
r
o
m
i
m
b
al
an
c
e
d
m
e
d
i
c
a
l
d
at
a
.
A
l
t
h
o
u
g
h
d
e
e
p
l
e
a
r
n
i
n
g
h
a
s
d
e
m
o
n
s
t
r
a
t
e
d
o
u
ts
t
a
n
d
i
n
g
p
o
t
e
n
t
i
a
l
f
o
r
m
e
d
i
c
al
i
m
a
g
e
a
p
p
li
c
at
i
o
n
s
,
s
o
m
e
c
r
i
ti
c
a
l
i
s
s
u
e
s
r
e
m
a
i
n
u
n
r
e
s
o
l
v
e
d
.
F
i
r
s
t
,
i
m
a
g
e
q
u
a
l
it
y
i
m
p
a
c
ts
o
v
er
a
l
l
l
ea
r
n
i
n
g
p
e
r
f
o
r
m
a
n
c
e
,
o
r
c
h
a
n
g
e
s
i
n
t
h
e
d
a
t
a
d
o
m
a
i
n
i
n
f
l
u
e
n
c
e
t
h
e
q
u
a
l
it
y
o
f
t
h
e
n
e
t
w
o
r
k
'
s
le
a
r
n
i
n
g
.
Se
c
o
n
d
,
d
e
s
i
g
n
i
n
g
C
NN
s
t
o
h
a
n
d
l
e
d
i
f
f
e
r
e
n
t
d
a
ta
d
i
m
e
n
s
i
o
n
s
a
f
f
ec
ts
t
h
e
n
e
tw
o
r
k
'
s
a
b
i
li
t
y
t
o
c
a
p
t
u
r
e
f
e
a
t
u
r
e
s
.
T
h
e
s
e
is
s
u
e
s
a
r
e
c
e
n
t
r
a
l
t
o
t
h
e
d
e
v
e
l
o
p
m
e
n
t
o
f
d
e
e
p
l
e
a
r
n
in
g
,
h
i
g
h
l
i
g
h
t
i
n
g
t
h
e
n
ee
d
f
o
r
c
o
n
t
i
n
u
e
d
i
m
p
r
o
v
e
m
e
n
t
i
n
d
e
e
p
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s
.
W
e
p
r
o
p
o
s
e
a
l
e
a
r
n
i
n
g
m
o
d
e
l
f
o
r
k
i
d
n
e
y
d
i
s
o
r
d
e
r
s
u
s
i
n
g
a
c
o
n
v
o
l
u
t
i
o
n
a
l
o
p
e
r
a
t
o
r
f
u
s
i
o
n
s
t
r
a
t
e
g
y
w
it
h
V
G
GN
e
t
,
w
h
i
c
h
i
s
d
i
v
i
d
e
d
i
n
t
o
tw
o
m
o
d
e
l
s
:
d
u
a
l
-
s
t
r
e
a
m
c
o
n
v
o
l
u
t
i
o
n
a
l
(
DS
C
)
,
d
e
s
i
g
n
e
d
t
o
c
a
p
t
u
r
e
a
d
d
i
t
i
o
n
a
l
f
e
a
t
u
r
e
s
a
n
d
a
d
d
r
es
s
i
n
f
o
r
m
a
t
i
o
n
lo
s
s
d
u
r
i
n
g
f
ea
t
u
r
e
e
x
t
r
a
c
t
i
o
n
a
n
d
d
u
al
-
in
p
u
t
co
n
v
o
lu
tio
n
al
(
D
I
C
)
,
w
h
i
c
h
i
s
d
e
s
i
g
n
e
d
t
o
l
e
a
r
n
m
u
lt
i
-
s
c
a
l
e
d
a
t
a
an
d
c
a
p
t
u
r
e
c
o
m
p
l
e
x
f
e
a
t
u
r
e
s
,
a
l
l
o
w
i
n
g
t
h
e
m
o
d
e
l
t
o
a
d
a
p
t
e
f
f
e
c
t
i
v
e
l
y
t
o
c
o
n
s
t
r
a
i
n
e
d
d
o
m
a
i
n
s
.
T
h
e
p
r
o
p
o
s
e
d
n
e
t
w
o
r
k
l
e
v
e
r
a
g
e
s
u
n
s
u
p
e
r
v
i
s
e
d
l
e
a
r
n
i
n
g
t
o
m
a
n
ag
e
c
o
m
p
l
e
x
d
a
t
a
a
n
d
r
e
d
u
c
e
t
h
e
d
i
m
e
n
s
i
o
n
a
li
t
y
o
f
m
e
d
i
c
al
i
m
a
g
e
s
.
2.
M
E
T
H
O
D
2
.
1
.
Co
nv
o
lutio
na
l o
pera
t
o
r
T
h
e
co
n
v
o
l
u
ti
o
n
al
o
p
er
at
o
r
[
1
9
]
is
a
n
ess
e
n
t
ial
m
at
h
e
m
at
ic
al
f
u
n
ct
io
n
t
h
at
e
x
t
r
a
cts
f
ea
t
u
r
es
f
r
o
m
i
n
p
u
t
d
at
a.
I
t h
i
g
h
li
g
h
ts
c
r
u
ci
al
p
att
e
r
n
s
,
s
u
ch
as
e
d
g
es
,
co
r
n
e
r
s
,
o
r
tex
tu
r
es.
T
h
e
o
p
er
at
o
r
w
o
r
k
s
wit
h
a
k
e
r
n
el
,
wh
ic
h
s
lid
es
ac
r
o
s
s
t
h
e
i
n
p
u
t
d
ata
,
p
e
r
f
o
r
m
in
g
c
o
n
v
o
lu
ti
o
n
s
at
e
ac
h
p
o
s
iti
o
n
t
o
d
ete
ct
f
ea
tu
r
es li
k
e
ed
g
es
an
d
t
ex
tu
r
es.
Stri
d
e
,
t
h
e
s
te
p
s
iz
e
o
f
th
e
f
i
lte
r
'
s
m
o
v
e
m
e
n
t
,
r
e
d
u
c
es
th
e
o
u
t
p
u
t
s
iz
e
to
m
in
im
ize
c
o
m
p
u
tati
o
n
al
c
o
m
p
le
x
it
y
.
T
h
e
s
am
e
p
ad
d
in
g
e
n
h
an
ce
s
th
is
p
r
o
ce
s
s
,
en
s
u
r
in
g
th
e
o
u
tp
u
t
s
ize
m
atc
h
es
th
e
i
n
p
u
t
s
ize.
T
h
e
r
ec
tifie
d
lin
ea
r
u
n
it
(
R
eL
U
)
a
ctiv
atio
n
f
u
n
ctio
n
is
also
a
p
p
l
ied
af
ter
co
n
v
o
lu
tio
n
t
o
r
e
m
o
v
e
n
e
g
ativ
e
v
alu
es,
allo
win
g
th
e
m
o
d
el
to
lear
n
m
o
r
e
c
o
m
p
lex
p
atter
n
s
.
As
m
en
tio
n
ed
,
th
e
co
n
v
o
lu
tio
n
al
o
p
e
r
ato
r
in
C
NNs
wo
r
k
s
b
y
m
u
ltip
ly
in
g
p
i
x
el
v
a
lu
es
f
r
o
m
th
e
in
p
u
t
with
th
e
f
i
lter
's
weig
h
ts
,
co
m
b
in
in
g
th
e
r
esu
lts
in
to
a
s
in
g
le
v
alu
e
at
ea
ch
p
o
s
itio
n
o
f
th
e
f
e
atu
r
e
m
ap
,
as in
(
1
)
.
(
,
)
=
(
∗
)
(
,
)
∑
∑
(
+
,
+
)
⋅
(
,
)
(
1
)
W
h
er
e
(
,
)
r
e
p
r
esen
ts
th
e
o
u
tp
u
t
at
p
o
s
itio
n
,
o
n
th
e
f
ea
tu
r
e
m
ap
,
o
r
th
e
r
esu
lt
o
f
th
e
c
o
n
v
o
lu
ti
o
n
,
(
,
)
is
th
e
in
p
u
t
v
alu
e
at
p
o
s
itio
n
,
o
n
th
e
in
p
u
t
im
ag
e,
a
n
d
(
,
)
d
en
o
te
s
th
e
weig
h
ts
in
th
e
f
ilter
o
r
k
er
n
el
at
p
o
s
itio
n
,
.
T
h
e
s
y
m
b
o
l *
r
ep
r
es
en
ts
th
e
co
n
v
o
lu
tio
n
o
p
er
atio
n
.
2
.
2
.
Dua
l
co
nv
o
lutio
na
l o
per
a
t
o
r
Du
al
c
o
n
v
o
lu
tio
n
al
[
2
0
]
o
p
e
r
ato
r
s
ar
e
co
n
v
o
l
u
tio
n
al
f
ilt
er
s
d
esig
n
ed
to
ex
tr
ac
t
f
ea
t
u
r
es
f
r
o
m
d
if
f
er
en
t
d
ata
v
iewp
o
in
ts
.
E
ac
h
f
ilter
is
tailo
r
e
d
to
d
etec
t
s
p
ec
if
ic
f
ea
tu
r
es,
s
u
c
h
as
h
o
r
izo
n
tal
ed
g
es,
v
er
tical
ed
g
es,
o
r
co
m
p
lex
s
u
r
f
ac
es,
a
n
d
th
ese
f
ea
tu
r
es
ar
e
th
en
co
m
b
in
ed
i
n
to
a
s
in
g
le
f
ea
tu
r
e
m
ap
.
T
h
is
ap
p
r
o
ac
h
en
ab
les
th
e
m
o
d
el
to
ca
p
tu
r
e
i
n
-
d
ep
th
in
f
o
r
m
atio
n
an
d
f
ea
tu
r
es
o
f
v
ar
y
in
g
s
izes.
I
n
th
is
s
t
u
d
y
,
b
o
th
f
ilter
s
ar
e
co
n
ca
ten
ated
in
t
o
a
v
ec
to
r
to
f
o
r
m
th
e
f
in
al
f
ea
tu
r
e
m
ap
,
as i
n
(
2
)
.
(
,
)
=
1
(
,
)
+
2
(
,
)
(
2
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
C
o
mb
in
in
g
co
n
v
o
lu
tio
n
a
l o
p
er
a
to
r
s
in
u
n
s
u
p
ervis
ed
n
etw
o
r
ks fo
r
kid
n
ey
a
b
n
o
r
ma
liti
es (
A
ek
ka
r
a
t S
u
ksu
ko
n
t
)
4543
W
h
er
e
1
(
,
)
is
th
e
o
u
tp
u
t
o
f
th
e
f
ir
s
t
co
n
v
o
lu
tio
n
at
p
o
s
itio
n
(
,
)
,
2
(
,
)
is
th
e
o
u
tp
u
t
o
f
th
e
s
ec
o
n
d
co
n
v
o
l
u
tio
n
o
p
er
atio
n
at
t
h
e
s
am
e
p
o
s
itio
n
(
,
)
,
an
d
(
,
)
is
th
e
s
u
m
o
f
th
ese
two
v
alu
es
at
e
a
ch
p
o
s
itio
n
(
,
)
.
T
h
is
s
tu
d
y
co
n
ca
ten
ates f
ea
tu
r
e
m
ap
s
in
to
a
v
ec
to
r
to
f
o
r
m
th
e
f
in
al
f
ea
tu
r
e
m
ap
.
2
.
3
.
Resid
ua
l la
y
er
s
R
esid
u
al
lay
er
s
[
2
1
]
a
r
e
d
esig
n
ed
to
ad
d
r
ess
th
e
p
r
o
b
lem
o
f
v
an
is
h
in
g
g
r
ad
ien
ts
,
wh
ic
h
o
f
ten
o
cc
u
r
s
as
n
etwo
r
k
s
d
ee
p
en
,
b
y
in
tr
o
d
u
cin
g
s
k
ip
co
n
n
ec
tio
n
s
.
T
h
ese
co
n
n
ec
tio
n
s
allo
w
th
e
in
p
u
t
o
f
a
g
iv
en
b
lo
c
k
to
b
y
p
ass
th
e
tr
an
s
f
o
r
m
atio
n
s
i
n
o
th
er
lay
er
s
an
d
b
e
d
ir
ec
tly
ad
d
ed
to
th
e
o
u
tp
u
t,
en
a
b
l
in
g
m
o
r
e
ef
f
ec
tiv
e
lear
n
in
g
o
f
n
ew
f
ea
t
u
r
es,
as sh
o
wn
in
(
3
)
.
=
(
)
+
(
3
)
W
h
er
e
is
th
e
in
p
u
t,
(
)
r
ep
r
esen
ts
th
e
p
r
o
ce
s
s
in
g
f
u
n
ctio
n
o
f
t
h
e
co
n
v
o
l
u
tio
n
al
lay
er
,
an
d
th
e
f
in
al
r
esu
lt
is
o
b
tain
ed
b
y
ad
d
i
n
g
th
e
v
alu
es
o
f
an
d
(
)
,
r
esid
u
al
lay
er
s
als
o
en
a
b
les
th
e
m
o
d
el
to
lear
n
m
o
r
e
c
o
m
p
lex
tr
an
s
f
o
r
m
atio
n
s
wh
ile
m
ain
tai
n
in
g
its
tr
ain
ab
ilit
y
.
2
.
4
.
Dee
p lea
rning
t
ec
hn
iqu
es
Dee
p
lear
n
in
g
in
v
o
lv
es
n
etwo
r
k
s
th
at
p
r
o
ce
s
s
d
ata
th
r
o
u
g
h
m
u
ltip
le
lay
er
s
,
with
ea
ch
lay
er
lear
n
in
g
p
r
o
g
r
ess
iv
ely
co
m
p
lex
f
ea
tu
r
e
s
.
T
h
is
r
esear
ch
em
p
lo
y
s
VGG
-
19
[
2
2
]
,
a
d
ee
p
lear
n
in
g
ar
ch
itectu
r
e
d
esig
n
ed
to
ca
p
tu
r
e
in
tr
icate
f
ea
t
u
r
es.
T
h
e
n
etwo
r
k
is
n
o
tab
le
f
o
r
its
u
s
e
o
f
s
m
all
f
ilter
s
an
d
a
co
n
s
is
ten
t
s
tr
u
ctu
r
e
ac
r
o
s
s
all
lay
er
s
,
a
k
ey
f
ea
tu
r
e
th
at
en
h
an
ce
s
its
ab
ilit
y
to
co
llect
f
ea
tu
r
es
f
r
o
m
d
iv
er
s
e
d
ata
an
d
im
p
r
o
v
e
o
v
er
all
p
er
f
o
r
m
a
n
ce
ef
f
icien
tl
y
.
3.
P
RO
P
O
SE
D
M
E
T
H
O
D
T
h
is
r
es
ea
r
c
h
ai
m
s
t
o
cl
ass
if
y
r
e
n
al
a
b
n
o
r
m
a
liti
es
i
n
co
m
p
u
ted
t
o
m
o
g
r
a
p
h
y
(
CT
)
s
c
an
i
m
ag
es.
O
u
r
ex
p
e
r
i
m
e
n
ts
u
t
ili
ze
t
h
e
wel
l
-
e
s
tab
lis
h
e
d
VG
GNe
t
,
i
n
co
r
p
o
r
a
tin
g
l
o
o
s
el
y
co
m
b
i
n
e
d
c
o
n
v
o
l
u
ti
o
n
al
o
p
e
r
a
to
r
s
t
o
en
h
a
n
ce
its
ab
ilit
y
t
o
c
ap
tu
r
e
v
a
r
i
a
n
ce
-
d
e
p
e
n
d
e
n
t
f
e
at
u
r
e
s
.
Un
li
k
e
t
h
e
o
r
i
g
i
n
al
d
e
e
p
-
s
tr
u
ct
u
r
e
d
e
x
p
a
n
d
e
d
n
et
wo
r
k
s
,
o
u
r
a
p
p
r
o
a
ch
i
n
t
r
o
d
u
ce
s
a
d
is
ti
n
c
t
e
x
p
e
r
i
m
e
n
t
al
f
r
a
m
ew
o
r
k
.
T
h
e
te
c
h
n
iq
u
es
an
d
e
x
p
er
im
en
ts
ar
e
d
i
v
i
d
e
d
i
n
t
o
t
h
r
e
e
p
a
r
ts
,
as
s
h
o
wn
i
n
F
ig
u
r
e
1
.
I
n
F
ig
u
r
e
1
(
a
)
,
t
h
is
n
etw
o
r
k
c
o
m
b
i
n
es
a
m
o
d
if
i
ed
c
o
n
v
o
l
u
ti
o
n
a
l
o
p
e
r
at
o
r
(
MCO
)
wi
th
VG
G
-
1
9
to
i
m
p
r
o
v
e
l
ea
r
n
i
n
g
f
r
o
m
m
e
d
ic
al
i
m
a
g
e
d
at
a.
I
n
p
u
t
i
m
ag
es
wit
h
a
s
iz
e
o
f
1
0
0
×1
0
0
×
3
p
i
x
e
ls
a
r
e
f
ed
i
n
t
o
b
o
t
h
b
r
an
c
h
es
.
O
n
t
h
e
MCO
s
id
e
,
t
h
e
n
etw
o
r
k
u
s
es
co
n
v
o
l
u
ti
o
n
al
k
er
n
els
w
it
h
3
2
,
6
4
,
1
2
8
,
a
n
d
2
5
6
f
il
te
r
s
,
e
a
ch
wi
th
a
3
×
3
k
er
n
e
l,
SAM
E
p
ad
d
i
n
g
,
a
n
d
a
R
eL
U
a
cti
v
a
ti
o
n
f
u
n
ct
io
n
.
T
h
ese
la
y
e
r
s
a
r
e
g
r
o
u
p
e
d
i
n
to
b
l
o
c
k
s
c
o
n
s
is
ti
n
g
o
f
th
r
e
e
s
ta
ck
e
d
l
ay
er
s
,
f
o
ll
o
w
ed
b
y
a
2
×2
m
a
x
-
p
o
o
li
n
g
l
a
y
e
r
a
t
t
h
e
e
n
d
o
f
ea
ch
b
lo
c
k
.
R
esi
d
u
al
c
o
n
n
e
cti
o
n
s
a
r
e
a
d
d
e
d
to
h
el
p
a
v
o
i
d
t
h
e
v
a
n
is
h
i
n
g
g
r
a
d
ie
n
t
p
r
o
b
le
m
as
t
h
e
n
et
w
o
r
k
d
ee
p
e
n
s
.
I
n
p
a
r
al
lel
,
th
e
VGG
-
1
9
b
r
a
n
c
h
a
p
p
lies
its
s
t
an
d
a
r
d
co
n
v
o
l
u
ti
o
n
al
ar
ch
ite
ct
u
r
e,
u
s
in
g
m
u
lti
p
l
e
c
o
n
v
o
lu
ti
o
n
s
a
n
d
m
ax
-
p
o
o
l
i
n
g
la
y
e
r
s
t
o
ex
tr
ac
t
d
ee
p
f
e
at
u
r
e
r
e
p
r
ese
n
t
ati
o
n
s
.
Af
te
r
p
r
o
ce
s
s
i
n
g
,
t
h
e
f
e
at
u
r
es
f
r
o
m
b
o
t
h
b
r
a
n
c
h
es
ar
e
f
u
s
e
d
,
f
l
att
e
n
e
d
i
n
t
o
a
v
ec
t
o
r
,
an
d
p
ass
e
d
th
r
o
u
g
h
th
r
ee
f
u
l
ly
co
n
n
e
cte
d
la
y
e
r
s
w
it
h
1
2
0
,
5
0
0
,
a
n
d
5
0
0
n
o
d
es
,
r
es
p
ec
ti
v
el
y
,
b
ef
o
r
e
r
ea
c
h
i
n
g
th
e
f
i
n
al
o
u
tp
u
t
la
y
e
r
.
T
h
is
m
o
d
e
l
d
esi
g
n
f
o
cu
s
es
o
n
i
m
p
r
o
v
i
n
g
f
ea
t
u
r
e
f
u
s
i
o
n
a
n
d
a
d
a
p
ta
b
ilit
y
to
c
o
m
p
le
x
d
ata
b
y
le
ar
n
i
n
g
d
i
v
e
r
s
e
f
e
at
u
r
es.
T
h
e
n
et
wo
r
k
was
t
r
ai
n
e
d
o
n
t
h
e
k
i
d
n
e
y
d
atas
et
,
as i
n
Fi
g
u
r
e
2
,
u
s
i
n
g
t
h
e
h
y
p
er
p
ar
am
ete
r
s
as
i
n
T
ab
l
e
1
.
T
ab
le
1
.
Par
am
eter
s
f
o
r
tr
ain
in
g
th
e
m
o
d
el
P
a
r
a
me
t
e
r
V
a
l
u
e
I
n
p
u
t
si
z
e
1
0
0
×
1
0
0
×
3
,
1
5
0
×
50
×
3
,
2
0
0
×
2
0
0
×
3
M
a
x
e
p
o
c
h
s
20
β1
a
n
d
β
2
(
A
d
a
m
o
p
t
i
mi
z
e
d
)
0
.
9
,
0
.
0
0
0
9
F
u
n
c
t
i
o
n
S
i
g
m
o
i
d
Ep
si
l
o
n
1e
-
8
B
a
t
c
h
si
z
e
2
Le
a
r
n
i
n
g
r
a
t
e
10
-
3
,
1
0
-
4
,
1
0
-
5
I
n
Fig
u
r
e
1
(
b
)
,
th
e
d
esig
n
tak
es
d
u
al
-
in
p
u
t
d
ata
with
d
if
f
er
en
t
c
h
ar
ac
ter
is
tics
.
I
m
ag
es
s
ized
1
5
0
×1
5
0
×3
p
ix
els
ar
e
f
ed
in
t
o
th
e
MCO
,
wh
ich
u
s
es
a
co
n
v
o
lu
tio
n
al
k
er
n
el
with
3
2
,
6
4
,
1
2
8
,
an
d
2
5
6
f
ilter
s
.
E
ac
h
lay
er
ap
p
lies
a
3
×3
k
er
n
el
with
SAME
p
ad
d
in
g
an
d
R
eL
U
ac
tiv
atio
n
,
ar
r
an
g
e
d
i
n
to
b
lo
c
k
s
o
f
th
r
e
e
s
tack
ed
lay
er
s
f
o
llo
wed
b
y
a
2
×2
m
ax
-
p
o
o
lin
g
lay
er
.
R
esid
u
al
co
n
n
ec
tio
n
s
also
s
u
p
p
o
r
t
s
tab
le
lear
n
in
g
an
d
r
ed
u
ce
th
e
r
is
k
o
f
v
a
n
is
h
in
g
g
r
ad
ien
ts
.
Me
an
wh
ile,
th
e
s
ec
o
n
d
b
r
an
c
h
r
ec
ei
v
es
2
0
0
×2
0
0
×
3
p
ix
el
im
ag
es
a
n
d
p
r
o
ce
s
s
es
th
em
th
r
o
u
g
h
th
e
s
tan
d
ar
d
VGG
-
1
9
ar
ch
itectu
r
e
,
wh
ich
ex
tr
ac
ts
d
ee
p
f
ea
tu
r
es
u
s
in
g
3
×
3
k
e
r
n
els,
2
×2
m
a
x
-
p
o
o
lin
g
lay
er
s
,
an
d
co
n
v
o
lu
tio
n
al
k
e
r
n
els
f
r
o
m
6
4
to
5
1
2
.
On
ce
b
o
th
s
tr
ea
m
s
co
m
p
lete
f
ea
tu
r
e
ex
tr
ac
tio
n
,
th
e
o
u
tp
u
ts
ar
e
f
u
s
ed
an
d
p
ass
ed
th
r
o
u
g
h
two
f
u
lly
co
n
n
ec
ted
lay
er
s
,
ea
ch
with
5
0
0
n
o
d
es,
b
e
f
o
r
e
r
ea
ch
in
g
th
e
f
in
al
o
u
tp
u
t
lay
er
.
T
h
is
ar
ch
itectu
r
e
is
d
esig
n
ed
to
p
r
o
ce
s
s
m
u
ltip
le
im
ag
e
f
ea
tu
r
es
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
4
,
No
.
6
,
Dec
em
b
er
2
0
2
5
:
4
5
4
1
-
4
5
5
1
4544
s
im
u
ltan
eo
u
s
ly
,
m
ak
in
g
it
h
i
g
h
ly
ef
f
icien
t
in
h
a
n
d
lin
g
co
m
p
lex
d
atasets
.
T
h
e
n
etwo
r
k
was
tr
ain
ed
o
n
th
e
k
id
n
ey
d
ataset
as in
Fig
u
r
e
2
,
u
s
in
g
th
e
p
a
r
am
eter
s
in
T
ab
le
1
.
(a
)
(
b
)
Fig
u
r
e
1
.
Pro
p
o
s
ed
m
et
h
o
d
f
o
r
k
id
n
ey
ab
n
o
r
m
ality
of
(
a)
DS
C
f
u
s
io
n
f
o
r
u
n
s
u
p
e
r
v
is
ed
lear
n
in
g
an
d
(
b
)
DI
C
f
u
s
io
n
f
o
r
u
n
s
u
p
er
v
is
ed
lear
n
in
g
4.
E
XP
E
R
I
M
E
N
T
A
L
AND
R
E
SUL
T
S
4
.
1
.
K
idn
ey
d
a
t
a
s
et
A
k
i
d
n
e
y
im
a
g
i
n
g
d
at
ase
t
w
as
co
m
p
il
ed
f
r
o
m
t
h
e
p
ic
tu
r
e
ar
ch
iv
in
g
a
n
d
c
o
m
m
u
n
ic
ati
o
n
s
y
s
t
em
(
PAC
S
)
o
f
v
ar
io
u
s
h
o
s
p
ita
ls
i
n
D
h
a
k
a,
B
an
g
l
a
d
es
h
[
1
3
]
,
[
2
3
]
.
T
h
e
d
a
tase
t
c
o
n
s
is
ts
o
f
i
m
ag
es
f
r
o
m
p
ati
e
n
ts
d
ia
g
n
o
s
e
d
wi
th
t
u
m
o
r
s
,
cy
s
ts
,
s
to
n
es
,
o
r
n
o
r
m
al
c
o
n
d
iti
o
n
s
.
Af
t
er
th
o
r
o
u
g
h
e
x
am
i
n
ati
o
n
,
1
2
,
4
4
6
u
n
i
q
u
e
im
a
g
es
wer
e
g
en
er
at
ed
,
in
cl
u
d
i
n
g
5
,
0
7
7
n
o
r
m
al
i
m
a
g
es
as
s
h
o
w
n
i
n
F
ig
u
r
e
2
(
a)
,
3
,
7
0
9
c
y
s
t
i
m
a
g
es
as
s
h
o
wn
i
n
Fig
u
r
e
2
(
b
)
,
1
,
3
7
7
s
to
n
e
im
a
g
es
as
s
h
o
w
n
i
n
Fi
g
u
r
e
2
(
c
)
,
a
n
d
2
,
2
8
3
t
u
m
o
r
i
m
a
g
es
as
s
h
o
w
n
i
n
Fi
g
u
r
e
2
(
d
)
.
T
h
e
d
atas
et
w
as u
p
l
o
a
d
ed
t
o
K
ag
g
l
e
to
e
n
c
o
u
r
ag
e
o
th
er
r
es
ea
r
c
h
e
r
s
t
o
d
e
v
el
o
p
i
n
n
o
v
a
ti
v
e
m
eth
o
d
s
.
I
t
w
as
s
p
lit
in
t
o
8
0
%
f
o
r
th
e
tr
ai
n
i
n
g
s
e
t,
1
0
%
f
o
r
t
h
e
v
al
id
ati
o
n
s
et
,
an
d
1
0
%
f
o
r
t
h
e
tes
t s
et
f
o
r
b
e
n
ch
m
a
r
k
in
g
p
u
r
p
o
s
es.
(
a)
(
b
)
(
c)
(
d
)
Fig
u
r
e
2
.
Kid
n
ey
ab
n
o
r
m
alitie
s
d
ataset
of
(
a)
n
o
r
m
al
,
(
b
)
c
y
s
t
,
(
c)
s
to
n
e
,
a
n
d
(
d
)
tu
m
o
r
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
C
o
mb
in
in
g
co
n
v
o
lu
tio
n
a
l o
p
er
a
to
r
s
in
u
n
s
u
p
ervis
ed
n
etw
o
r
ks fo
r
kid
n
ey
a
b
n
o
r
ma
liti
es (
A
ek
ka
r
a
t S
u
ksu
ko
n
t
)
4545
4
.
2
.
E
x
perim
ent
s
et
up
T
h
e
s
etu
p
co
n
s
is
ts
o
f
a
W
in
d
o
ws
s
y
s
tem
with
an
I
n
tel
C
o
r
e
i5
-
1
2
4
0
0
F
L
GA
1
7
0
0
C
PU
@
2
.
5
GHz
,
3
2
GB
o
f
R
AM
(
clo
ck
ed
at
5
,
6
0
0
MH
z)
,
an
d
an
NVI
DI
A
R
T
X
4
0
7
0
GPU
with
1
2
GB
o
f
VR
AM
an
d
5
,
8
8
8
C
UDA
co
r
es.
T
h
e
co
d
e
is
im
p
lem
en
ted
u
s
in
g
Nu
m
Py
an
d
T
en
s
o
r
Flo
w.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
an
d
d
ee
p
lear
n
in
g
n
etwo
r
k
ar
e
also
tr
ain
ed
u
s
in
g
s
tan
d
ar
d
ized
p
a
r
am
et
er
s
,
as su
m
m
ar
ized
in
T
a
b
le
1
.
4
.
3
.
E
v
a
lua
t
i
o
n
T
h
e
m
ea
n
s
q
u
ar
ed
e
r
r
o
r
(
M
SE)
ev
alu
ates
th
e
d
if
f
e
r
en
ce
b
etwe
en
p
r
e
d
icted
an
d
ac
tu
al
v
alu
es,
m
ea
s
u
r
in
g
th
eir
d
ev
iatio
n
a
n
d
alwa
y
s
y
ield
in
g
a
p
o
s
itiv
e
r
esu
lt.
I
n
th
is
s
tu
d
y
,
it is
u
s
ed
to
a
s
s
es
s
th
e
ca
p
ab
ilit
y
o
f
th
e
u
n
s
u
p
er
v
is
ed
n
etwo
r
k
,
as
s
h
o
wn
in
(
4
)
,
w
he
re
n
is
th
e
n
u
m
b
er
o
f
s
am
p
les
.
is
th
e
a
ctu
al
v
alu
e
o
f
th
e
s
am
p
le
,
∧
is
th
e
p
r
ed
icted
v
alu
e
o
f
th
e
s
am
p
le
,
(
−
∧
)
2
is
th
e
s
q
u
ar
ed
er
r
o
r
o
f
t
h
e
s
am
p
le
.
=
1
2
+
∑
(
−
∧
)
2
=
1
(
4
)
4
.
4
.
P
er
f
o
r
m
a
nce
o
f
t
ra
ini
ng
m
o
del
T
h
is
s
tu
d
y
p
r
esen
ts
a
s
y
s
tem
f
o
r
d
etec
tin
g
k
id
n
ey
a
b
n
o
r
m
a
liti
es,
f
o
cu
s
in
g
o
n
n
etwo
r
k
s
with
d
en
s
e
lay
er
s
an
d
d
u
al
i
n
p
u
t
s
ca
les
b
ased
o
n
u
n
s
u
p
er
v
is
ed
lear
n
i
n
g
.
T
h
e
ex
p
er
im
en
ts
ar
e
co
n
d
u
cted
o
n
a
d
ataset
co
n
tain
in
g
f
o
u
r
ty
p
es
o
f
ab
n
o
r
m
alities
:
n
o
r
m
al,
cy
s
t,
s
to
n
e,
an
d
tu
m
o
r
.
MSE
is
em
p
lo
y
ed
to
ev
al
u
ate
th
e
lear
n
in
g
p
e
r
f
o
r
m
an
ce
o
f
th
e
p
r
o
p
o
s
ed
ex
p
er
im
en
ts
.
F
i
g
u
r
e
3
s
h
o
w
s
e
x
p
e
r
i
m
e
n
t
al
r
e
s
u
l
t
s
u
s
i
n
g
d
u
a
l
d
e
n
s
e
l
a
y
e
r
s
w
i
t
h
d
i
f
f
e
r
e
n
t
h
y
p
e
r
p
a
r
a
m
e
te
r
s
e
t
t
i
n
g
s
.
O
b
s
e
r
v
i
n
g
t
h
e
l
e
a
r
n
i
n
g
p
r
o
c
e
s
s
a
t
e
a
c
h
e
p
o
c
h
w
i
t
h
l
e
a
r
n
i
n
g
r
at
e
s
o
f
1
0
-
3
,
1
0
-
4
,
1
0
-
5
,
w
e
n
o
t
i
c
e
r
a
p
i
d
l
e
a
r
n
i
n
g
t
h
a
t
p
o
i
n
t
s
t
o
s
t
a
b
l
e
p
e
r
f
o
r
m
a
n
c
e
.
I
n
F
i
g
u
r
e
3
(
a)
,
n
o
r
m
a
l
h
a
s
t
h
e
lo
w
e
s
t
MS
E
at
0
.
0
0
6
9
,
w
h
i
c
h
i
s
d
u
e
t
o
t
h
e
b
e
t
te
r
p
e
r
f
o
r
m
a
n
c
e
o
f
t
h
e
10
-
4
c
o
m
p
a
r
e
d
t
o
10
-
3
a
n
d
10
-
5
,
w
h
i
c
h
a
r
e
0
.
0
0
9
3
a
n
d
0
.
0
1
0
6
,
r
e
s
p
e
c
t
i
v
e
l
y
.
I
n
F
i
g
u
r
e
3
(
b
)
,
t
h
e
c
y
s
t
a
c
h
i
e
v
es
t
h
e
l
o
w
e
s
t
MS
E
o
f
0
.
0
0
5
6
w
i
t
h
10
-
4
,
w
h
i
c
h
i
s
h
i
g
h
e
r
t
h
a
n
10
-
3
a
n
d
10
-
5
b
y
0
.
0
0
2
8
a
n
d
0
.
0
0
6
,
i
n
d
i
c
a
t
i
n
g
t
h
at
10
-
4
p
r
o
v
i
d
e
s
b
et
t
e
r
l
e
a
r
n
i
n
g
s
t
a
b
i
l
i
t
y
.
S
i
m
i
l
a
r
ly
,
i
n
F
i
g
u
r
e
3
(
c
)
,
s
t
o
n
e
r
e
c
o
r
d
s
t
h
e
l
o
w
es
t
MS
E
at
0
.
0
0
4
2
u
s
i
n
g
t
h
e
10
-
4
,
r
e
f
l
e
c
t
i
n
g
i
t
s
e
f
f
e
c
t
i
v
e
c
a
p
a
b
il
i
t
y
i
n
r
e
d
u
c
i
n
g
l
o
s
s
.
Fi
n
a
l
l
y
,
i
n
F
i
g
u
r
e
3
(
d
)
,
t
h
e
t
u
m
o
r
a
c
h
i
e
v
e
s
t
h
e
l
o
we
s
t
MS
E
a
t
0
.
0
0
4
8
w
i
t
h
t
h
e
10
-
4
.
T
h
e
s
e
r
e
s
u
l
t
s
s
h
o
w
t
h
e
o
v
e
r
a
l
l
e
f
f
e
c
ti
v
e
n
e
s
s
a
n
d
c
o
n
s
is
t
e
n
cy
o
f
t
h
e
10
-
4
a
c
r
o
s
s
a
l
l d
a
t
a
t
y
p
e
s
.
F
u
r
t
h
e
r
m
o
r
e
,
a
c
r
o
s
s
a
l
l c
o
n
d
i
ti
o
n
s
(
n
o
r
m
a
l
,
c
y
s
t
,
s
t
o
n
e
,
a
n
d
tu
m
o
r
)
,
t
h
e
l
e
a
r
n
i
n
g
r
a
t
e
o
f
1
0
-
4
o
f
f
e
r
s
t
h
e
m
o
s
t
b
a
la
n
c
e
d
r
e
s
u
l
ts
,
a
c
h
i
e
v
i
n
g
t
h
e
l
o
w
e
s
t
f
i
n
al
l
o
s
s
i
n
M
S
E
wi
t
h
a
t
r
a
i
n
i
n
g
s
p
e
e
d
o
f
2
0
e
p
o
c
h
s
.
T
h
e
h
i
g
h
e
r
l
e
a
r
n
i
n
g
r
a
t
e
o
f
1
0
-
3
s
h
o
w
s
f
a
s
t
e
r
c
o
n
v
e
r
g
e
n
c
e
b
u
t
s
l
i
g
h
t
l
y
h
i
g
h
e
r
l
o
s
s
,
w
h
il
e
t
h
e
l
o
w
e
r
l
e
a
r
n
i
n
g
r
a
t
e
o
f
1
0
-
5
r
e
s
u
l
ts
i
n
s
l
o
w
e
r
l
e
a
r
n
i
n
g
a
n
d
t
h
e
h
i
g
h
es
t
l
o
s
s
i
n
a
l
l
e
x
p
e
r
i
m
e
n
t
s
.
(
a)
(
b
)
(
c)
(
d
)
Fig
u
r
e
3
.
L
o
s
s
o
f
DSC
f
u
s
io
n
f
o
r
u
n
s
u
p
er
v
is
ed
lear
n
in
g
of
(
a)
n
o
r
m
al
,
(
b
)
cy
s
t
,
(
c)
s
to
n
e
,
an
d
(
d
)
tu
m
o
r
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
4
,
No
.
6
,
Dec
em
b
er
2
0
2
5
:
4
5
4
1
-
4
5
5
1
4546
Fig
u
r
e
4
s
h
o
ws
ex
p
er
im
en
tal
r
esu
lt
s
u
s
in
g
d
u
al
in
p
u
t
s
ca
les
with
d
if
f
er
en
t
h
y
p
e
r
p
ar
a
m
eter
s
ettin
g
s
.
Ob
s
er
v
in
g
th
e
lear
n
in
g
at
ea
c
h
iter
atio
n
f
o
r
lear
n
in
g
r
ates
o
f
1
0
-
3
,
1
0
-
4
,
an
d
1
0
-
5
,
we
s
ee
f
ast
lear
n
in
g
ac
r
o
s
s
all
r
ates,
in
d
icatin
g
ef
f
ec
tiv
e
l
ea
r
n
in
g
.
I
n
Fig
u
r
e
4
(
a
)
,
n
o
r
m
al
h
as
th
e
lo
west
MSE
at
0
.
0
0
6
6
,
wh
ich
is
d
u
e
to
th
e
b
etter
1
0
-
4
c
o
m
p
ar
e
d
to
1
0
-
3
an
d
1
0
-
5
,
wh
ich
ar
e
0
.
0
0
9
0
an
d
0
.
0
1
3
6
,
r
esp
ec
tiv
ely
.
I
n
F
ig
u
r
e
4
(
b
)
,
c
y
s
t
h
as
th
e
lo
west
MSE
f
r
o
m
1
0
-
4
with
a
v
alu
e
o
f
0
.
0
0
6
3
,
wh
ich
i
s
h
ig
h
er
th
an
1
0
-
3
an
d
1
0
-
5
b
y
0
.
0
0
1
6
an
d
0
.
0
0
7
,
in
d
icatin
g
th
at
1
0
-
4
h
as
b
etter
l
ea
r
n
in
g
s
tab
ilit
y
.
Similar
ly
,
in
Fig
u
r
e
4
(
c)
,
s
to
n
e
h
as
th
e
lo
w
est
MSE
at
0
.
0
0
4
4
f
r
o
m
t
h
e
ex
p
er
im
en
t
with
1
0
-
4
,
d
em
o
n
s
tr
atin
g
its
ef
f
ec
tiv
e
lo
s
s
r
ed
u
ctio
n
ca
p
ab
ilit
y
.
Fin
ally
,
in
Fig
u
r
e
4
(
d
)
,
t
u
m
o
r
h
as
th
e
lo
west
MSE
at
0
.
0
0
5
8
,
wh
ich
is
d
u
e
to
th
e
1
0
-
4
.
T
h
e
ex
p
er
im
en
tal
r
esu
lts
co
n
f
ir
m
th
e
o
v
er
al
l
ef
f
ec
tiv
en
ess
o
f
th
is
lear
n
in
g
r
ate
ac
r
o
s
s
all
d
ata
ty
p
es.
Ad
d
itio
n
ally
,
th
e
cu
r
v
e
f
o
r
1
0
-
4
h
a
s
a
h
ig
h
er
lear
n
in
g
r
ate
th
an
th
at
o
f
1
0
-
3
wh
ich
s
h
o
ws
f
aster
lear
n
in
g
b
u
t
with
a
s
lig
h
tly
h
ig
h
er
lo
s
s
ac
r
o
s
s
all
ex
p
er
im
en
ts
.
At
1
0
-
5
,
th
e
cu
r
v
e
p
r
o
g
r
ess
es m
o
r
e
s
lo
wly
b
u
t
ac
h
iev
es th
e
l
o
west lo
s
s
v
alu
es in
ev
er
y
ex
p
er
im
en
t.
(
a)
(
b
)
(
c)
(
d
)
Fig
u
r
e
4
.
L
o
s
s
o
f
DI
C
f
u
s
io
n
f
o
r
u
n
s
u
p
er
v
is
ed
lear
n
in
g
of
(
a
)
n
o
r
m
al
,
(
b
)
cy
s
t
,
(
c)
s
to
n
e
,
an
d
(
d
)
t
u
m
o
r
4.
5
.
Densi
t
y
a
na
ly
s
is
Prin
cip
al
co
m
p
o
n
e
n
t
an
al
y
s
is
(
PC
A
)
o
n
a
h
ea
tm
a
p
is
a
tech
n
iq
u
e
u
s
ed
in
T
en
s
o
r
Flo
w
to
d
e
s
cr
ib
e
th
e
d
en
s
ity
o
f
r
en
al
ab
n
o
r
m
alities
.
T
h
is
s
tu
d
y
an
aly
ze
s
th
e
d
if
f
er
en
ce
s
b
etwe
en
f
o
u
r
c
o
n
d
iti
o
n
s
:
n
o
r
m
al,
cy
s
ts
,
s
to
n
es,
an
d
tu
m
o
r
s
.
T
h
e
co
m
p
o
n
en
t
d
escr
ip
tio
n
u
tili
ze
s
th
e
co
n
ca
ten
ated
lay
e
r
r
esu
lts
to
d
em
o
n
s
tr
ate
th
e
p
r
o
p
o
s
ed
n
etwo
r
k
'
s
ca
p
ab
ilit
y
f
o
r
u
n
s
u
p
er
v
is
ed
f
ea
t
u
r
e
f
u
s
io
n
.
T
h
e
d
ata
d
en
s
ity
,
r
ep
r
ese
n
tin
g
ch
a
n
g
es
an
d
d
is
tr
ib
u
tio
n
in
ea
c
h
co
n
d
itio
n
,
is
v
is
u
alize
d
in
Fig
u
r
es 5
an
d
6
.
I
n
Fig
u
r
e
5
,
t
h
e
co
m
p
o
s
itio
n
o
f
th
e
n
o
r
m
al,
c
y
s
t,
s
to
n
e,
an
d
tu
m
o
r
d
ata
r
ev
ea
ls
s
ig
n
if
ican
t
d
if
f
er
en
ce
s
in
d
e
n
s
ity
d
is
tr
ib
u
tio
n
.
T
h
e
n
o
r
m
al
d
ata
,
as
i
n
Fig
u
r
e
5
(
a)
,
s
h
o
ws
two
d
is
tin
ct
d
en
s
ity
p
ea
k
s
,
clu
s
ter
ed
b
etwe
en
PC
1
:
-
45
a
n
d
45
,
with
m
o
d
e
r
ate
v
ar
iab
il
ity
alo
n
g
PC
2
an
d
a
wid
e
d
is
p
er
s
io
n
.
I
n
co
n
tr
ast,
th
e
c
y
s
t
d
ata
,
as
s
h
o
wn
in
Fig
u
r
e
5
(
b
)
,
p
r
esen
t
a
h
i
g
h
ly
c
o
n
ce
n
tr
ated
clu
s
ter
b
etwe
en
PC
1
:
-
4
5
an
d
6
0
,
with
a
r
elativ
ely
lo
w
-
d
e
n
s
ity
p
ea
k
,
in
d
icatin
g
u
n
if
o
r
m
ity
in
t
h
e
cy
s
t
p
r
o
f
ile.
T
h
e
s
to
n
e
d
ata
,
as
in
Fig
u
r
e
5
(
c
)
,
s
h
o
ws
two
d
is
tin
ct
p
ea
k
s
,
o
n
e
ce
n
ter
ed
ar
o
u
n
d
PC
1
:
-
60
an
d
8
0
,
s
u
g
g
esti
n
g
d
is
tin
ct
s
u
b
g
r
o
u
p
s
wi
th
in
th
e
s
to
n
e
d
ata.
L
astl
y
,
as
in
Fig
u
r
e
5
(
d
)
,
th
e
t
u
m
o
r
d
ata
ex
h
ib
it
th
e
h
ig
h
est
v
ar
iab
ilit
y
,
with
a
b
r
o
ad
d
is
p
er
s
io
n
alo
n
g
b
o
t
h
PC
1
an
d
PC
2
,
r
ef
lectin
g
th
e
c
o
m
p
lex
ity
o
f
tu
m
o
r
ch
a
r
ac
ter
is
tics
.
T
h
ese
d
if
f
er
en
ce
s
in
d
en
s
ity
an
d
d
is
tr
ib
u
tio
n
u
n
d
er
s
co
r
e
th
e
v
a
r
y
in
g
lev
els o
f
v
ar
iatio
n
,
with
cy
s
t b
ein
g
th
e
m
o
s
t u
n
if
o
r
m
an
d
tu
m
o
r
th
e
m
o
s
t d
iv
er
s
e.
Fig
u
r
e
6
s
h
o
ws
th
e
d
ata
d
en
s
ity
an
d
d
is
tr
ib
u
tio
n
d
if
f
er
en
ce
s
ac
r
o
s
s
th
e
p
r
in
cip
al
co
m
p
o
n
en
ts
.
T
h
e
n
o
r
m
al
d
ata,
as
in
Fig
u
r
e
6
(
a)
,
ex
h
ib
its
two
p
ea
k
s
s
p
r
ea
d
f
r
o
m
PC
1
:
-
1
5
to
3
0
,
r
ef
l
ec
tin
g
a
b
alan
ce
d
d
is
tr
ib
u
tio
n
.
As
in
Fig
u
r
e
6
(
b
)
,
th
e
c
y
s
t
d
ata
s
h
o
w
th
e
l
ea
s
t
s
p
r
ea
d
,
with
co
n
to
u
r
s
c
o
n
ce
n
tr
ated
ar
o
u
n
d
PC
1
:
-
2
5
to
4
0
an
d
a
lo
wer
d
en
s
ity
p
ea
k
,
in
d
icatin
g
h
o
m
o
g
en
eity
.
T
h
e
s
to
n
e
d
ata,
as
in
Fig
u
r
e
6
(
c
)
,
s
p
an
s
a
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
C
o
mb
in
in
g
co
n
v
o
lu
tio
n
a
l o
p
er
a
to
r
s
in
u
n
s
u
p
ervis
ed
n
etw
o
r
ks fo
r
kid
n
ey
a
b
n
o
r
ma
liti
es (
A
ek
ka
r
a
t S
u
ksu
ko
n
t
)
4547
b
r
o
ad
e
r
r
an
g
e
f
r
o
m
PC
1
:
-
6
0
to
8
0
an
d
s
h
o
ws
two
d
is
tin
ct
p
ea
k
s
,
s
u
g
g
esti
n
g
a
m
o
d
er
ate
lev
el
o
f
v
ar
iab
ilit
y
.
I
n
co
n
tr
ast,
as
in
Fig
u
r
e
6
(
d
)
,
th
e
t
u
m
o
r
d
ata
d
is
p
lay
s
th
e
wid
est
s
p
r
ea
d
,
r
an
g
in
g
f
r
o
m
PC
1
:
-
4
5
to
6
0
,
with
th
r
ee
h
i
g
h
-
d
e
n
s
ity
p
ea
k
s
in
d
icatin
g
h
ig
h
v
ar
iab
ilit
y
.
T
h
e
s
e
r
esu
lts
in
d
icate
th
at
n
u
m
er
ical
an
d
d
e
n
s
ity
d
is
tr
ib
u
tio
n
s
ar
e
k
e
y
in
d
icato
r
s
o
f
th
e
co
m
p
lex
ity
an
d
v
ar
ia
b
ilit
y
o
f
ea
ch
co
n
d
itio
n
.
(
a)
(
b
)
(
c)
(
d
)
F
i
g
u
r
e
5
.
F
e
at
u
r
e
m
a
p
v
i
s
u
al
i
za
t
i
o
n
o
f
D
SC
f
u
s
i
o
n
f
o
r
u
n
s
u
p
e
r
v
i
s
e
d
l
e
a
r
n
i
n
g
of
(
a
)
d
e
n
s
i
t
y
o
f
c
o
n
c
a
t
e
n
a
t
e
l
a
y
e
r
s
(
n
o
r
m
a
l
(
10
-
4
)
)
,
(
b
)
d
e
n
s
i
t
y
o
f
c
o
n
c
a
t
e
n
a
t
e
l
a
y
e
r
s
(
c
y
s
t
(
10
-
4
)
)
,
(
c
)
d
e
n
s
i
t
y
o
f
c
o
n
c
a
t
e
n
at
e
l
a
y
e
r
s
(
s
t
o
n
e
(
10
-
4
)
),
an
d
(
d
)
d
en
s
ity
o
f
co
n
ca
ten
ate
lay
er
s
(
t
u
m
o
r
(
10
-
4
))
(
a)
(
b
)
(
c)
(
d
)
F
i
g
u
r
e
6
.
T
h
e
d
e
n
s
i
t
y
o
f
c
o
n
c
a
t
e
n
a
t
e
l
a
y
e
r
s
f
r
o
m
D
I
C
f
u
s
i
o
n
f
o
r
u
n
s
u
p
e
r
v
i
s
e
d
l
e
a
r
n
i
n
g
of
(
a
)
d
e
n
s
i
t
y
o
f
c
o
n
c
a
t
e
n
a
t
e
l
a
y
e
r
s
(
n
o
r
m
a
l
(
10
-
4
))
,
(
b
)
d
e
n
s
i
t
y
o
f
c
o
n
c
a
t
e
n
a
t
e
l
a
y
e
r
s
(
c
y
s
t
(
10
-
4
)
)
(
c
)
d
e
n
s
i
t
y
o
f
c
o
n
c
a
t
e
n
a
t
e
l
a
y
e
r
s
(
s
t
o
n
e
(
10
-
4
))
,
a
n
d
(
d
)
d
e
n
s
i
t
y
o
f
c
o
n
c
a
t
e
n
a
t
e
l
a
y
e
r
s
(
t
u
m
o
r
(
10
-
4
))
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
4
,
No
.
6
,
Dec
em
b
er
2
0
2
5
:
4
5
4
1
-
4
5
5
1
4548
4.
6
.
Co
m
pa
riso
n m
o
del
I
n
th
is
s
ec
ti
o
n
,
w
e
s
el
ec
t
o
t
h
e
r
s
t
ate
-
of
-
t
h
e
-
ar
t
m
o
d
els
(
VGG
-
1
6
,
VGG
-
1
9
,
X
ce
p
ti
o
n
,
De
n
s
eNe
t
-
1
2
1
,
an
d
E
f
f
ici
en
tNe
tB
7
[
2
2
]
,
[
2
4
]
–
[
2
6
]
)
,
wh
ic
h
ar
e
b
as
e
d
o
n
u
n
s
u
p
er
v
is
ed
l
ea
r
n
i
n
g
a
n
d
ad
a
p
te
d
f
r
o
m
p
r
ev
io
u
s
s
tu
d
ies
o
n
k
id
n
e
y
a
b
n
o
r
m
ali
tie
s
,
t
o
co
m
p
a
r
e
t
h
ei
r
p
e
r
f
o
r
m
a
n
c
e
wi
th
t
h
e
DSC
a
n
d
D
I
C
m
o
d
e
ls
.
T
h
e
p
a
r
a
m
e
te
r
s
u
s
e
d
f
o
r
t
h
e
ex
p
er
i
m
e
n
ts
a
r
e
s
h
o
wn
i
n
T
ab
le
1
.
T
h
e
e
x
p
e
r
i
m
en
t
al
r
es
u
lts
a
r
e
s
u
m
m
a
r
iz
e
d
i
n
F
ig
u
r
e
7
.
Fig
u
r
e
7
c
o
m
p
ar
es
t
h
e
lo
s
s
v
a
lu
es
ac
r
o
s
s
d
if
f
er
en
t
m
o
d
els:
VGG
-
1
6
,
VGG
-
1
9
,
Xce
p
tio
n
,
Den
s
eNe
t
-
1
2
1
,
a
n
d
E
f
f
icien
tNetB
7
o
n
th
e
n
o
r
m
al,
cy
s
t,
s
to
n
e,
a
n
d
tu
m
o
r
d
atasets
.
T
h
e
r
esu
lts
in
d
icate
th
at
E
f
f
icien
tNetB
7
d
em
o
n
s
tr
ates
th
e
b
est
tr
ain
i
n
g
p
er
f
o
r
m
an
ce
ac
r
o
s
s
all
g
r
o
u
p
s
,
ac
h
iev
in
g
l
o
s
s
es
o
f
4
.
3
4
3
3
e
-
04
,
0
.
0
0
1
0
,
0
.
0
0
3
3
,
an
d
0
.
0
0
1
5
f
o
r
th
e
n
o
r
m
al,
cy
s
t,
s
to
n
e
,
an
d
tu
m
o
r
d
atasets
,
as
in
Fig
u
r
e
s
7
(
a
)
to
7
(
d
)
.
Den
s
eNe
t
-
1
2
1
an
d
Xce
p
tio
n
s
h
o
w
s
lig
h
tly
h
ig
h
er
lo
s
s
es
th
an
E
f
f
icien
tNetB
7
,
as
in
Fig
u
r
es
7
(
a)
to
7
(
d
)
,
b
u
t
th
ey
s
till
ef
f
icien
tly
lea
r
n
t
h
e
d
ata.
I
n
c
o
n
tr
ast,
VGG
-
1
6
a
n
d
VGG
-
19
,
as
in
Fig
u
r
es
7
(
a)
to
7
(
d
)
,
ex
h
ib
it
h
ig
h
er
lo
s
s
v
alu
es
th
an
t
h
e
o
t
h
er
m
o
d
els,
with
s
ig
n
if
ican
tly
s
lo
wer
lear
n
in
g
r
ates.
T
h
is
s
u
g
g
ests
th
at
th
ese
n
etwo
r
k
s
s
tr
u
g
g
le
with
m
o
r
e
c
o
m
p
lex
task
s
an
d
r
eq
u
ir
e
f
u
r
t
h
er
tu
n
in
g
.
C
o
m
p
a
r
ed
to
th
e
p
r
o
p
o
s
ed
m
eth
o
d
s
in
Fig
u
r
es
3
an
d
4
,
it
is
ev
id
en
t
th
at
wh
ile
DSC
an
d
DI
C
h
av
e
lo
wer
p
er
f
o
r
m
an
ce
th
an
E
f
f
icien
tNetB
7
,
th
ey
ar
e
co
m
p
ar
ab
le
t
o
Den
s
eNe
t
-
1
2
1
an
d
Xce
p
tio
n
,
an
d
o
u
tp
e
r
f
o
r
m
VGG
-
1
6
an
d
VGG
-
19.
(
a)
(
b
)
(
c)
(
d
)
Fig
u
r
e
7
.
L
o
s
s
co
m
p
ar
is
o
n
ac
r
o
s
s
v
ar
io
u
s
m
o
d
els f
o
r
k
id
n
ey
ab
n
o
r
m
ality
o
f
(
a
)
n
o
r
m
al
(
1
0
-
5
)
,
(
b
)
c
y
s
t (
1
0
-
5
)
,
(
c)
s
to
n
e
(
1
0
-
5
)
,
a
n
d
(
d
)
tu
m
o
r
(
1
0
-
5
)
5.
DIS
CU
SS
I
O
N
T
h
is
s
tu
d
y
p
r
o
p
o
s
es
a
c
o
m
b
i
n
atio
n
o
f
a
C
NN
an
d
th
e
VGG
-
19
i
n
a
DSC
n
etwo
r
k
to
e
n
h
an
ce
its
p
er
f
o
r
m
an
ce
a
n
d
ca
p
ab
ilit
y
i
n
lear
n
in
g
k
id
n
ey
-
m
ed
ical
f
e
atu
r
e
s
.
T
h
e
m
o
d
el
d
esig
n
in
Fig
u
r
e
1
(
a
)
em
p
lo
y
s
id
en
tical
in
p
u
t
d
im
e
n
s
io
n
s
(
1
0
0
×1
0
0
×3
)
f
o
r
a
two
-
s
tr
ea
m
n
etwo
r
k
[
2
7
]
.
I
t
in
te
g
r
ates
d
if
f
er
en
t
3
2
,
6
4
,
1
2
8
,
an
d
2
5
6
co
n
v
o
l
u
tio
n
al
k
er
n
els
an
d
2
×
2
k
er
n
el
s
izes
to
ca
p
tu
r
e
m
u
lti
-
s
ca
le
s
p
atial
f
ea
tu
r
es
an
d
th
eir
d
ep
en
d
e
n
cies
ef
f
ec
tiv
ely
.
Ad
d
itio
n
ally
,
r
esid
u
al
co
n
n
ec
tio
n
s
b
etwe
en
p
r
o
ce
s
s
in
g
b
lo
ck
s
h
elp
m
itig
ate
th
e
v
an
is
h
in
g
g
r
ad
ie
n
t
p
r
o
b
lem
a
n
d
f
ac
ilit
ate
th
e
lear
n
i
n
g
o
f
d
ee
p
n
etwo
r
k
s
.
I
n
c
o
n
tr
ast,
Fig
u
r
e
1
(
b
)
s
h
o
ws
th
e
two
d
if
f
er
e
n
t
in
p
u
t
s
izes
(
1
5
0
×1
5
0
×3
an
d
2
0
0
×2
0
0
×3
)
to
i
m
p
r
o
v
e
th
e
n
etwo
r
k
’
s
ab
ilit
y
to
p
r
o
ce
s
s
im
ag
es
with
v
ar
y
in
g
ch
a
r
ac
ter
is
tics
.
T
h
is
DI
C
ap
p
r
o
ac
h
[
2
8
]
,
[
2
9
]
en
ab
les
th
e
m
o
d
el
to
lear
n
f
r
o
m
m
u
ltip
le
d
ata
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
C
o
mb
in
in
g
co
n
v
o
lu
tio
n
a
l o
p
er
a
to
r
s
in
u
n
s
u
p
ervis
ed
n
etw
o
r
ks fo
r
kid
n
ey
a
b
n
o
r
ma
liti
es (
A
ek
ka
r
a
t S
u
ksu
ko
n
t
)
4549
s
o
u
r
ce
s
,
th
er
e
b
y
e
n
h
an
ci
n
g
its
d
ee
p
f
ea
t
u
r
e
lea
r
n
in
g
ca
p
ac
ity
an
d
f
ac
ilit
atin
g
a
co
m
p
ar
ativ
e
an
aly
s
is
o
f
r
esp
o
n
s
es a
cr
o
s
s
in
p
u
t v
ar
iatio
n
s
.
Ou
r
ex
p
e
r
im
en
ts
with
th
e
DS
C
m
o
d
el
s
u
g
g
est
it
ac
h
ie
v
es
a
r
ea
s
o
n
ab
le
b
alan
ce
b
etwe
e
n
co
m
p
lex
ity
an
d
p
er
f
o
r
m
an
ce
,
as
m
ea
s
u
r
ed
b
y
MSE
.
I
n
th
e
ex
p
e
r
im
en
ts
,
a
lear
n
in
g
r
ate
o
f
1
0
⁻⁴
y
ie
ld
s
th
e
lo
west
lo
s
s
ac
r
o
s
s
all
class
e
s
s
h
o
wn
in
Fig
u
r
e
s
3
(
a)
to
3
(
d
)
,
n
a
m
ely
n
o
r
m
al,
c
y
s
t,
s
to
n
e,
an
d
tu
m
o
r
.
T
h
e
o
b
tain
e
d
ac
cu
r
ac
ies
r
ef
lect
th
e
m
o
d
el'
s
ca
p
ab
ilit
y
to
lear
n
co
m
p
lex
f
e
atu
r
es
f
r
o
m
m
ed
ical
d
ata
ef
f
e
ctiv
ely
.
Alth
o
u
g
h
a
h
ig
h
er
lear
n
in
g
r
ate
o
f
1
0
⁻³
lead
s
to
f
aster
co
n
v
e
r
g
en
ce
,
it
r
esu
lts
in
a
h
ig
h
er
lo
s
s
.
C
o
n
v
er
s
ely
,
a
lo
wer
lear
n
in
g
r
ate
o
f
1
0
-
5
s
lo
ws
d
o
wn
lear
n
in
g
an
d
f
ails
to
r
ed
u
c
e
th
e
lo
s
s
as
ef
f
ec
tiv
el
y
o
f
1
0
⁻⁴
with
in
th
e
g
iv
en
n
u
m
b
er
o
f
e
p
o
ch
s
.
T
h
ese
r
esu
lts
in
d
icate
th
at
th
e
p
r
o
p
o
s
e
d
n
etwo
r
k
d
esig
n
en
h
an
ce
s
t
h
e
m
o
d
el'
s
lear
n
in
g
ca
p
ab
ilit
y
an
d
is
s
u
itab
le
f
o
r
a
n
aly
zin
g
v
ar
io
u
s
ty
p
es o
f
m
e
d
ical
im
ag
e
d
ata.
T
h
e
ex
p
e
r
im
en
tal
r
esu
lts
an
d
s
tu
d
ies
in
th
is
r
esear
ch
s
h
o
w
th
at
th
e
p
r
o
p
o
s
ed
DSC
an
d
DI
C
m
o
d
els,
wh
ich
co
m
b
in
e
th
e
f
u
n
ctio
n
al
ity
o
f
co
n
v
o
lu
tio
n
al
o
p
e
r
ato
r
s
an
d
VGG
-
1
9
n
etwo
r
k
s
,
ar
e
e
f
f
ec
tiv
e
in
lear
n
i
n
g
co
m
p
lex
m
e
d
ical
im
ag
e
f
ea
tu
r
es,
esp
ec
ially
wh
en
u
s
in
g
a
n
a
p
p
r
o
p
r
iate
lear
n
in
g
r
ate
o
f
1
0
⁻
⁴,
wh
ich
y
ield
s
th
e
lo
west
lo
s
s
ac
r
o
s
s
all
d
ata
g
r
o
u
p
s
.
Alth
o
u
g
h
th
e
p
r
o
p
o
s
ed
m
o
d
els
ar
e
i
n
f
er
io
r
to
E
f
f
icien
t
NetB7
[
2
6
]
in
ter
m
s
o
f
ac
h
iev
in
g
th
e
lo
west
lo
s
s
,
t
h
ey
d
em
o
n
s
tr
ate
lear
n
in
g
p
er
f
o
r
m
an
ce
co
m
p
ar
a
b
le
to
ad
v
an
ce
d
m
o
d
els
s
u
ch
as
Xce
p
tio
n
[
2
4
]
an
d
De
n
s
eNe
t
-
121
[
2
5
]
,
a
n
d
o
u
tp
er
f
o
r
m
V
GG
-
1
6
an
d
VGG
-
19
[
2
2
]
.
I
n
ad
d
itio
n
,
th
e
d
u
al
-
in
p
u
t
s
tr
ea
m
d
esig
n
im
p
r
o
v
es
th
e
ab
ilit
y
to
p
r
o
ce
s
s
d
ata
with
d
if
f
e
r
en
t
c
h
ar
ac
ter
is
tics
.
Ho
wev
er
,
th
is
r
esear
c
h
also
id
en
tifie
s
s
o
m
e
lim
itati
o
n
s
th
at
s
h
o
u
ld
b
e
ad
d
r
ess
ed
in
f
u
tu
r
e
wo
r
k
,
s
u
ch
as
r
ed
u
cin
g
n
etwo
r
k
co
m
p
lex
ity
,
ap
p
ly
i
n
g
r
eg
u
lar
izatio
n
tech
n
iq
u
es
to
p
r
ev
en
t
o
v
er
f
itti
n
g
,
in
c
o
r
p
o
r
atin
g
a
v
ar
iety
o
f
clin
ical
in
d
icato
r
s
,
ex
p
lo
r
i
n
g
tr
an
s
f
er
lear
n
in
g
[
3
0
]
ap
p
r
o
ac
h
es,
an
d
d
esig
n
in
g
lig
h
tweig
h
t
m
o
d
el
s
[
3
1
]
f
o
r
ef
f
ec
tiv
e
ap
p
licatio
n
in
r
e
al
-
wo
r
ld
m
ed
i
ca
l d
ev
ices.
6.
CO
NCLU
SI
O
N
T
h
is
p
ap
er
p
r
esen
ts
a
m
eth
o
d
f
o
r
c
o
m
b
in
in
g
co
n
v
o
lu
tio
n
al
o
p
er
ato
r
s
in
u
n
s
u
p
er
v
is
ed
n
e
two
r
k
s
f
o
r
k
id
n
ey
ab
n
o
r
m
alities
.
T
h
e
p
r
o
p
o
s
ed
n
etwo
r
k
in
teg
r
ates
two
ty
p
es
o
f
o
p
e
r
ato
r
s
:
i
)
D
S
C
,
w
h
i
c
h
e
x
t
r
a
c
ts
f
e
a
t
u
r
e
s
f
r
o
m
a
s
i
n
g
l
e
i
n
p
u
t
t
h
r
o
u
g
h
p
a
r
a
l
l
e
l
p
a
t
h
w
a
y
s
an
d
ii
)
D
I
C
,
w
h
i
c
h
p
r
o
c
es
s
es
i
m
a
g
e
s
o
f
v
a
r
y
i
n
g
r
e
s
o
l
u
t
i
o
n
s
t
o
e
n
h
a
n
c
e
m
u
l
t
i
-
s
c
a
l
e
l
e
a
r
n
i
n
g
.
T
h
is
en
ab
les
th
e
n
etwo
r
k
to
lear
n
m
u
lti
-
v
iew
f
ea
tu
r
es
an
d
h
elp
s
r
ed
u
ce
o
v
er
f
itti
n
g
d
u
r
in
g
tr
ai
n
in
g
.
T
h
e
tech
n
iq
u
e
is
test
ed
o
n
th
e
k
id
n
e
y
ab
n
o
r
m
alities
d
ataset,
a
wid
ely
ac
ce
p
ted
an
d
p
u
b
licly
av
ailab
le
r
eso
u
r
ce
f
o
r
r
esear
c
h
an
d
d
ev
elo
p
m
e
n
t.
E
x
p
e
r
im
en
tal
r
e
s
u
lts
s
h
o
w
th
at
th
e
p
r
o
p
o
s
ed
n
etwo
r
k
o
u
tp
er
f
o
r
m
s
o
th
er
s
with
a
lo
wer
tr
ai
n
in
g
lo
s
s
r
ate.
T
h
is
s
u
cc
ess
d
em
o
n
s
tr
ates
th
at
th
e
m
eth
o
d
is
ef
f
ec
tiv
e
an
d
o
p
e
n
s
u
p
o
p
p
o
r
tu
n
ities
f
o
r
f
u
tu
r
e
wo
r
k
,
p
ar
ticu
lar
ly
in
ap
p
ly
in
g
m
u
lti
-
s
ca
le
an
d
m
u
lti
-
in
p
u
t m
o
d
els to
en
h
an
ce
3
D
d
is
ea
s
e
d
etec
tio
n
in
m
e
d
ical
im
ag
in
g
.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
T
h
is
wo
r
k
was
f
u
n
d
ed
b
y
t
h
e
Natio
n
al
Scien
ce
,
R
esear
ch
an
d
I
n
n
o
v
atio
n
Fu
n
d
,
T
h
ailan
d
Scien
ce
R
esear
ch
an
d
I
n
n
o
v
atio
n
(
T
SR
I
)
,
th
r
o
u
g
h
R
ajam
an
g
ala
Un
iv
er
s
ity
o
f
T
ec
h
n
o
lo
g
y
T
h
an
y
ab
u
r
i
(
FR
B
6
8
E
0
7
1
0
)
(
G
r
an
t N
o
.
: F
R
B
6
8
0
0
4
5
/0
1
6
8
)
.
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
T
h
is
jo
u
r
n
al
u
s
es
th
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT)
to
r
ec
o
g
n
ize
in
d
iv
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
th
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
llab
o
r
atio
n
.
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
Aek
k
ar
at
Su
k
s
u
k
o
n
t
✓
✓
✓
✓
✓
✓
✓
✓
An
u
r
u
k
Pro
m
m
ak
h
o
t
✓
✓
✓
✓
✓
✓
J
ak
k
r
ee
Srin
o
n
c
h
at
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
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
T
h
e
au
th
o
r
s
d
ec
lar
e
n
o
co
n
f
lict o
f
in
ter
est.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
4
,
No
.
6
,
Dec
em
b
er
2
0
2
5
:
4
5
4
1
-
4
5
5
1
4550
DATA AV
AI
L
AB
I
L
I
T
Y
T
h
e
d
ata
th
at
s
u
p
p
o
r
t
th
e
f
in
d
in
g
s
o
f
th
is
s
tu
d
y
ar
e
o
p
en
ly
av
ailab
le
in
th
e
C
T
KI
D
N
E
Y
d
ataset:
n
o
r
m
al
-
c
y
s
t
-
tu
m
o
r
an
d
s
to
n
e
at
h
ttp
s
://www.
k
ag
g
le.
co
m
/d
atasets
/n
az
m
u
l0
0
8
7
/ct
-
k
id
n
ey
-
d
ataset
-
n
o
r
m
al
-
cy
s
t
-
tu
m
o
r
-
a
n
d
-
s
to
n
e/d
ata.
RE
F
E
R
E
NC
E
S
[
1
]
Z.
X
u
,
Y
.
Li
u
,
G
.
X
u
,
a
n
d
T.
Lu
k
a
s
i
e
w
i
c
z
,
“
S
e
l
f
-
su
p
e
r
v
i
s
e
d
me
d
i
c
a
l
i
ma
g
e
se
g
me
n
t
a
t
i
o
n
u
si
n
g
d
e
e
p
r
e
i
n
f
o
r
c
e
d
a
d
a
p
t
i
v
e
mask
i
n
g
,
”
I
EE
E
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
4
,
n
o
.
1
,
p
p
.
1
8
0
–
1
9
3
,
2
0
2
5
,
d
o
i
:
1
0
.
1
1
0
9
/
TM
I
.
2
0
2
4
.
3
4
3
6
6
0
8
.
[
2
]
L.
F
a
n
,
M
.
L
i
,
Z
.
H
u
,
Y
.
H
o
n
g
,
a
n
d
D
.
K
o
n
g
,
“
D
N
G
G
:
M
e
d
i
c
a
l
i
ma
g
e
l
o
ssl
e
ss
e
n
c
r
y
p
t
i
o
n
v
i
a
d
e
e
p
n
e
t
w
o
r
k
g
u
i
d
e
d
g
e
n
e
r
a
t
i
v
e
,
”
I
EEE
S
i
g
n
a
l
Pr
o
c
e
ssi
n
g
L
e
t
t
e
rs
,
v
o
l
.
3
2
,
p
p
.
1
3
3
1
–
1
3
3
5
,
2
0
2
5
,
d
o
i
:
1
0
.
1
1
0
9
/
LSP
.
2
0
2
5
.
3
5
5
2
5
2
8
.
[
3
]
T.
D
h
a
r
,
N
.
D
e
y
,
S
.
B
o
r
r
a
,
a
n
d
R
.
S
.
S
h
e
r
r
a
t
t
,
“
C
h
a
l
l
e
n
g
e
s
o
f
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
—
i
m
p
r
o
v
i
n
g
e
x
p
l
a
i
n
a
b
i
l
i
t
y
a
n
d
t
r
u
s
t
,
”
I
EEE
T
ra
n
s
a
c
t
i
o
n
s
o
n
T
e
c
h
n
o
l
o
g
y
a
n
d
S
o
c
i
e
t
y
,
v
o
l
.
4
,
n
o
.
1
,
p
p
.
6
8
–
7
5
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
t
t
s.
2
0
2
3
.
3
2
3
4
2
0
3
.
[
4
]
J.
M
a
,
Y
.
H
e
,
F
.
L
i
,
L
.
H
a
n
,
C
.
Y
o
u
,
a
n
d
B
.
W
a
n
g
,
“
S
e
g
me
n
t
a
n
y
t
h
i
n
g
i
n
me
d
i
c
a
l
i
m
a
g
e
s,
”
N
a
t
u
re
C
o
m
m
u
n
i
c
a
t
i
o
n
s
,
v
o
l
.
1
5
,
n
o
.
1
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
3
8
/
s
4
1
4
6
7
-
0
2
4
-
4
4
8
2
4
-
z.
[
5
]
X
.
Li
e
t
a
l
.
,
“
D
e
e
p
l
e
a
r
n
i
n
g
b
a
se
d
d
e
-
o
v
e
r
l
a
p
p
i
n
g
c
o
r
r
e
c
t
i
o
n
o
f
p
r
o
j
e
c
t
i
o
n
s fr
o
m
a
f
l
a
t
-
p
a
n
e
l
m
i
c
r
o
a
r
r
a
y
X
-
r
a
y
s
o
u
r
c
e
:
s
i
m
u
l
a
t
i
o
n
st
u
d
y
,
”
P
h
y
si
c
a
M
e
d
i
c
a
,
v
o
l
.
1
1
1
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
e
j
mp
.
2
0
2
3
.
1
0
2
6
0
7
.
[
6
]
S
.
H
.
A
b
b
o
o
d
,
H
.
N
.
A
.
H
a
me
d
,
M
.
S
.
M
.
R
a
h
i
m,
A
.
R
e
h
ma
n
,
T
.
S
a
b
a
,
a
n
d
S
.
A
.
B
a
h
a
j
,
“
H
y
b
r
i
d
r
e
t
i
n
a
l
i
m
a
g
e
e
n
h
a
n
c
e
me
n
t
a
l
g
o
r
i
t
h
m
f
o
r
d
i
a
b
e
t
i
c
r
e
t
i
n
o
p
a
t
h
y
d
i
a
g
n
o
st
i
c
u
si
n
g
d
e
e
p
l
e
a
r
n
i
n
g
m
o
d
e
l
,
”
I
EE
E
Ac
c
e
ss
,
v
o
l
.
1
0
,
p
p
.
7
3
0
7
9
–
7
3
0
8
6
,
2
0
2
2
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
2
.
3
1
8
9
3
7
4
.
[
7
]
Y
.
Lu
o
e
t
a
l
.
,
“
D
e
h
a
z
e
o
f
c
a
t
a
r
a
c
t
o
u
s
r
e
t
i
n
a
l
i
m
a
g
e
s
u
si
n
g
a
n
u
n
p
a
i
r
e
d
g
e
n
e
r
a
t
i
v
e
a
d
v
e
r
sari
a
l
n
e
t
w
o
r
k
,
”
I
EEE
J
o
u
r
n
a
l
o
f
Bi
o
m
e
d
i
c
a
l
a
n
d
H
e
a
l
t
h
I
n
f
o
rm
a
t
i
c
s
,
v
o
l
.
2
4
,
n
o
.
1
2
,
p
p
.
3
3
7
4
–
3
3
8
3
,
2
0
2
0
,
d
o
i
:
1
0
.
1
1
0
9
/
JB
H
I
.
2
0
2
0
.
2
9
9
9
0
7
7
.
[
8
]
M
.
M
.
I
sl
a
m
,
F
.
K
a
r
r
a
y
,
R
.
A
l
h
a
j
j
,
a
n
d
J.
Z
e
n
g
,
“
A
r
e
v
i
e
w
o
n
d
e
e
p
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s
f
o
r
t
h
e
d
i
a
g
n
o
si
s
o
f
n
o
v
e
l
c
o
r
o
n
a
v
i
r
u
s
(
C
O
V
I
D
-
1
9
)
,
”
I
EEE
Ac
c
e
ss
,
v
o
l
.
9
,
p
p
.
3
0
5
5
1
–
3
0
5
7
2
,
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
1
.
3
0
5
8
5
3
7
.
[
9
]
J.
C
h
a
k
i
a
n
d
A
.
U
c
a
r
,
“
A
n
e
f
f
i
c
i
e
n
t
a
n
d
r
o
b
u
st
a
p
p
r
o
a
c
h
u
s
i
n
g
i
n
d
u
c
t
i
v
e
t
r
a
n
sf
e
r
-
b
a
se
d
e
n
sem
b
l
e
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
s
f
o
r
k
i
d
n
e
y
st
o
n
e
d
e
t
e
c
t
i
o
n
,
”
I
E
EE
Ac
c
e
ss
,
v
o
l
.
1
2
,
p
p
.
3
2
8
9
4
–
3
2
9
1
0
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
4
.
3
3
7
0
6
7
2
.
[
1
0
]
S
.
A
si
f
,
W
.
Y
i
,
Q
.
U
.
A
i
n
,
J
.
H
o
u
,
T.
Y
i
,
a
n
d
J.
S
i
,
“
I
mp
r
o
v
i
n
g
e
f
f
e
c
t
i
v
e
n
e
ss
o
f
d
i
f
f
e
r
e
n
t
d
e
e
p
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
-
b
a
s
e
d
m
o
d
e
l
s
f
o
r
d
e
t
e
c
t
i
n
g
b
r
a
i
n
t
u
mo
r
s fr
o
m
M
R
i
m
a
g
e
s,”
I
EEE
Ac
c
e
ss
,
v
o
l
.
1
0
,
p
p
.
3
4
7
1
6
–
3
4
7
3
0
,
2
0
2
2
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
2
.
3
1
5
3
3
0
6
.
[
1
1
]
H
.
C
u
i
e
t
a
l
.
,
“
A
u
t
o
m
a
t
i
c
s
e
g
me
n
t
a
t
i
o
n
o
f
k
i
d
n
e
y
v
o
l
u
m
e
u
si
n
g
m
u
l
t
i
-
m
o
d
u
l
e
h
y
b
r
i
d
b
a
se
d
u
-
sh
a
p
e
i
n
p
o
l
y
c
y
s
t
i
c
k
i
d
n
e
y
d
i
s
e
a
se
,
”
I
EEE
A
c
c
e
ss
,
v
o
l
.
1
1
,
p
p
.
5
8
1
1
3
–
5
8
1
2
4
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
3
.
3
2
8
4
0
2
9
.
[
1
2
]
A
.
P
r
o
m
ma
k
h
o
t
a
n
d
J.
S
r
i
n
o
n
c
h
a
t
,
“
C
o
m
b
i
n
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
s
f
o
r
f
u
n
g
i
c
l
a
ss
i
f
i
c
a
t
i
o
n
,
”
I
E
E
E
A
c
c
e
ss
,
v
o
l
.
1
2
,
p
p
.
5
8
0
2
1
–
5
8
0
3
0
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
4
.
3
3
9
1
6
3
0
.
[
1
3
]
A
.
P
r
o
mm
a
k
h
o
t
a
n
d
J.
S
r
i
n
o
n
c
h
a
t
,
“
V
G
G
N
e
t
i
n
t
e
g
r
a
t
i
o
n
f
o
r
k
i
d
n
e
y
t
u
m
o
r
c
l
a
ssi
f
i
c
a
t
i
o
n
,
”
1
2
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
E
l
e
c
t
ri
c
a
l
En
g
i
n
e
e
ri
n
g
C
o
n
g
ress:
S
m
a
rt
F
a
c
t
o
ry
a
n
d
I
n
t
e
l
l
i
g
e
n
t
T
e
c
h
n
o
l
o
g
y
f
o
r
T
o
m
o
rro
w
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
i
EE
C
O
N
6
0
6
7
7
.
2
0
2
4
.
1
0
5
3
7
9
0
4
.
[
1
4
]
G
.
C
h
e
n
e
t
a
l
.
,
“
P
r
e
d
i
c
t
i
o
n
o
f
c
h
r
o
n
i
c
k
i
d
n
e
y
d
i
s
e
a
s
e
u
s
i
n
g
a
d
a
p
t
i
v
e
h
y
b
r
i
d
i
z
e
d
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
o
n
t
h
e
i
n
t
e
r
n
e
t
o
f
me
d
i
c
a
l
t
h
i
n
g
s
p
l
a
t
f
o
r
m,
”
I
EE
E
A
c
c
e
ss
,
v
o
l
.
8
,
p
p
.
1
0
0
4
9
7
–
1
0
0
5
0
8
,
2
0
2
0
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
0
.
2
9
9
5
3
1
0
.
[
1
5
]
S
.
M
.
M
.
El
k
h
o
l
y
,
A
.
R
e
z
k
,
a
n
d
A
.
A
.
E
.
F
.
S
a
l
e
h
,
“
Ea
r
l
y
p
r
e
d
i
c
t
i
o
n
o
f
c
h
r
o
n
i
c
k
i
d
n
e
y
d
i
s
e
a
s
e
u
si
n
g
d
e
e
p
b
e
l
i
e
f
n
e
t
w
o
r
k
,
”
I
EEE
Ac
c
e
ss
,
v
o
l
.
9
,
p
p
.
1
3
5
5
4
2
–
1
3
5
5
4
9
,
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
1
.
3
1
1
4
3
0
6
.
[
1
6
]
S
.
D
.
P
a
n
d
e
a
n
d
R
.
A
g
a
r
w
a
l
,
“
M
u
l
t
i
-
c
l
a
ss
k
i
d
n
e
y
a
b
n
o
r
mal
i
t
i
e
s
d
e
t
e
c
t
i
n
g
n
o
v
e
l
s
y
s
t
e
m
t
h
r
o
u
g
h
c
o
m
p
u
t
e
d
t
o
mo
g
r
a
p
h
y
,
”
I
EE
E
Ac
c
e
ss
,
v
o
l
.
1
2
,
p
p
.
2
1
1
4
7
–
2
1
1
5
5
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
4
.
3
3
5
1
1
8
1
.
[
1
7
]
S
.
L
y
u
,
X
.
L
u
o
,
a
n
d
J.
L
i
,
“
H
y
Tr
a
n
sM
A
:
a
h
y
b
r
i
d
m
o
d
e
l
u
s
i
n
g
t
r
a
n
sf
o
r
mer
w
i
t
h
m
u
l
t
i
-
sc
a
l
e
a
t
t
e
n
t
i
o
n
s
f
o
r
me
d
i
c
a
l
i
ma
g
e
seg
m
e
n
t
a
t
i
o
n
,
”
4
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
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
I
n
t
e
r
n
e
t
a
n
d
D
i
g
i
t
a
l
E
c
o
n
o
m
y
,
p
p
.
2
4
6
–
2
4
9
,
2
0
2
5
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
A
I
D
6
5
2
7
5
.
2
0
2
5
.
1
1
0
3
4
4
1
7
.
[
1
8
]
X
.
Q
i
n
,
F
.
M
.
B
u
i
,
H
.
H
.
N
g
u
y
e
n
,
a
n
d
Z.
H
a
n
,
“
Le
a
r
n
i
n
g
f
r
o
m
l
i
mi
t
e
d
a
n
d
i
mb
a
l
a
n
c
e
d
m
e
d
i
c
a
l
i
ma
g
e
s wit
h
f
i
n
e
r
sy
n
t
h
e
t
i
c
i
ma
g
e
s
f
r
o
m G
A
N
s,”
I
E
EE
Ac
c
e
ss
,
v
o
l
.
1
0
,
p
p
.
9
1
6
6
3
–
9
1
6
7
7
,
2
0
2
2
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
2
.
3
2
0
2
5
6
0
.
[
1
9
]
H
.
I
.
P
e
y
a
l
e
t
a
l
.
,
“
P
l
a
n
t
d
i
s
e
a
s
e
c
l
a
ss
i
f
i
e
r
:
d
e
t
e
c
t
i
o
n
o
f
d
u
a
l
-
c
r
o
p
d
i
se
a
ses
u
si
n
g
l
i
g
h
t
w
e
i
g
h
t
2
D
C
N
N
a
r
c
h
i
t
e
c
t
u
r
e
,
”
I
EEE
Ac
c
e
ss
,
v
o
l
.
1
1
,
p
p
.
1
1
0
6
2
7
–
1
1
0
6
4
3
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
3
.
3
3
2
0
6
8
6
.
[
2
0
]
Y
.
Q
i
a
n
,
H
.
Ta
n
g
,
Y
.
R
a
n
,
a
n
d
B
.
Li
,
“
T
a
r
g
e
t
c
l
a
ssi
f
i
c
a
t
i
o
n
i
n
u
n
a
t
t
e
n
d
e
d
g
r
o
u
n
d
se
n
so
r
s
w
i
t
h
a
t
w
o
-
s
t
r
e
a
m
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
t
w
o
r
k
,
”
I
EEE
S
e
n
s
o
rs
J
o
u
r
n
a
l
,
v
o
l
.
2
3
,
n
o
.
4
,
p
p
.
3
7
4
7
–
3
7
5
5
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
JS
EN
.
2
0
2
2
.
3
2
2
6
4
6
6
.
[
2
1
]
J.
N
.
-
A
l
c
a
z
a
r
,
S
.
P
.
-
C
a
st
a
n
o
s
,
I
.
M
.
-
M
o
r
a
t
ó
,
P
.
Z
u
c
c
a
r
e
l
l
o
,
F
.
J
.
F
e
r
r
i
,
a
n
d
M
.
C
o
b
o
s
,
“
A
c
o
mp
a
r
a
t
i
v
e
a
n
a
l
y
si
s o
f
r
e
si
d
u
a
l
b
l
o
c
k
a
l
t
e
r
n
a
t
i
v
e
s
f
o
r
e
n
d
-
to
-
e
n
d
a
u
d
i
o
c
l
a
ss
i
f
i
c
a
t
i
o
n
,
”
I
EEE
A
c
c
e
ss
,
v
o
l
.
8
,
p
p
.
1
8
8
8
7
5
–
1
8
8
8
8
2
,
2
0
2
0
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
0
.
3
0
3
1
6
8
5
.
[
2
2
]
A
.
K
.
R
a
j
e
n
d
r
a
n
a
n
d
S
.
C
.
S
e
t
h
u
r
a
m
a
n
,
“
Y
o
g
i
C
o
m
b
i
n
e
D
e
e
p
:
e
n
h
a
n
c
e
d
y
o
g
i
c
p
o
s
t
u
r
e
c
l
a
ssi
f
i
c
a
t
i
o
n
u
si
n
g
c
o
m
b
i
n
e
d
d
e
e
p
f
u
si
o
n
o
f
V
G
G
1
6
a
n
d
V
G
G
1
9
f
e
a
t
u
r
e
s
,
”
I
E
E
E
Ac
c
e
ss
,
v
o
l
.
1
2
,
p
p
.
1
3
9
1
6
5
–
1
3
9
1
8
0
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
4
.
3
4
1
4
6
5
4
.
[
2
3
]
M
.
N
.
I
sl
a
m
,
M
.
H
a
s
a
n
,
M
.
K
.
H
o
ss
a
i
n
,
M
.
G
.
R
.
A
l
a
m
,
M
.
Z.
U
d
d
i
n
,
a
n
d
A
.
S
o
y
l
u
,
“
V
i
si
o
n
t
r
a
n
sf
o
r
mer
a
n
d
e
x
p
l
a
i
n
a
b
l
e
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
mo
d
e
l
s
f
o
r
a
u
t
o
d
e
t
e
c
t
i
o
n
o
f
k
i
d
n
e
y
c
y
st
,
st
o
n
e
,
a
n
d
t
u
m
o
r
f
r
o
m
C
T
-
r
a
d
i
o
g
r
a
p
h
y
,
”
S
c
i
e
n
t
i
f
i
c
Re
p
o
r
t
s
,
v
o
l
.
1
2
,
n
o
.
1
,
2
0
2
2
,
d
o
i
:
1
0
.
1
0
3
8
/
s4
1
5
9
8
-
0
2
2
-
1
5
6
3
4
-
4.
[
2
4
]
H
.
M
.
S
h
a
h
z
a
d
,
S
.
M
.
B
h
a
t
t
i
,
A
.
Jaff
a
r
,
M
.
R
a
s
h
i
d
,
a
n
d
S
.
A
k
r
a
m,
“
M
u
l
t
i
-
mo
d
a
l
C
N
N
f
e
a
t
u
r
e
s
f
u
si
o
n
f
o
r
e
mo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
:
a
mo
d
i
f
i
e
d
X
c
e
p
t
i
o
n
mo
d
e
l
,
”
I
EE
E
A
c
c
e
ss
,
v
o
l
.
1
1
,
p
p
.
9
4
2
8
1
–
9
4
2
8
9
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
3
.
3
3
1
0
4
2
8
.
[
2
5
]
D
.
A
n
i
t
h
a
,
T.
S
o
u
j
a
n
y
a
,
S
.
C
h
a
k
r
a
b
o
r
t
y
,
A
.
A
l
k
h
a
y
y
a
t
,
a
n
d
R
.
R
e
v
a
t
h
i
,
“
O
r
a
l
c
a
n
c
e
r
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
u
si
n
g
d
e
e
p
l
e
a
r
n
i
n
g
w
i
t
h
D
e
n
s
e
N
e
t
1
2
1
-
C
a
t
B
o
o
st
c
l
a
s
si
f
i
e
r
,
”
2
n
d
I
EEE
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
N
e
t
w
o
rks
,
M
u
l
t
i
m
e
d
i
a
a
n
d
I
n
f
o
rm
a
t
i
o
n
T
e
c
h
n
o
l
o
g
y
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
N
M
I
TC
O
N
6
2
0
7
5
.
2
0
2
4
.
1
0
6
9
8
8
3
6
.
[
2
6
]
H
.
M
.
T.
K
h
u
sh
i
,
T.
M
a
s
o
o
d
,
A
.
Jaff
a
r
,
M
.
R
a
s
h
i
d
,
a
n
d
S
.
A
k
r
a
m,
“
I
mp
r
o
v
e
d
m
u
l
t
i
c
l
a
ss
b
r
a
i
n
t
u
mo
r
d
e
t
e
c
t
i
o
n
v
i
a
c
u
st
o
m
i
z
e
d
p
r
e
t
r
a
i
n
e
d
e
f
f
i
c
i
e
n
t
N
e
t
B
7
m
o
d
e
l
,
”
I
E
EE
A
c
c
e
ss
,
v
o
l
.
1
1
,
p
p
.
1
1
7
2
1
0
–
1
1
7
2
3
0
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
3
.
3
3
2
5
8
8
3
.
[
2
7
]
W
.
W
a
n
g
e
t
a
l
.
,
“
Le
a
r
n
i
n
g
t
w
o
-
st
r
e
a
m
C
N
N
f
o
r
m
u
l
t
i
-
mo
d
a
l
a
g
e
-
r
e
l
a
t
e
d
mac
u
l
a
r
d
e
g
e
n
e
r
a
t
i
o
n
c
a
t
e
g
o
r
i
z
a
t
i
o
n
,
”
I
EEE
J
o
u
r
n
a
l
o
f
Bi
o
m
e
d
i
c
a
l
a
n
d
H
e
a
l
t
h
I
n
f
o
rm
a
t
i
c
s
,
v
o
l
.
2
6
,
n
o
.
8
,
p
p
.
4
1
1
1
–
4
1
2
2
,
2
0
2
2
,
d
o
i
:
1
0
.
1
1
0
9
/
J
B
H
I
.
2
0
2
2
.
3
1
7
1
5
2
3
.
[
2
8
]
Q
.
C
a
i
,
Z.
G
a
o
,
J
.
A
n
,
S
.
G
a
o
,
a
n
d
C
.
G
r
e
b
o
g
i
,
“
A
g
r
a
p
h
-
t
e
m
p
o
r
a
l
f
u
s
e
d
d
u
a
l
-
i
n
p
u
t
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
f
o
r
d
e
t
e
c
t
i
n
g
sl
e
e
p
s
t
a
g
e
s
f
r
o
m
EEG
s
i
g
n
a
l
s,
”
I
EE
E
T
ra
n
sa
c
t
i
o
n
s o
n
C
i
r
c
u
i
t
s
a
n
d
S
y
s
t
e
m
s
I
I
:
Ex
p
ress
Bri
e
f
s
,
v
o
l
.
6
8
,
n
o
.
2
,
p
p
.
7
7
7
–
7
8
1
,
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
TC
S
I
I
.
2
0
2
0
.
3
0
1
4
5
1
4
.
[
2
9
]
C
.
R
e
n
e
t
a
l
.
,
“
P
r
o
s
t
a
t
e
se
g
me
n
t
a
t
i
o
n
i
n
M
R
I
u
si
n
g
t
r
a
n
sf
o
r
mer
e
n
c
o
d
e
r
a
n
d
d
e
c
o
d
e
r
f
r
a
mew
o
r
k
,
”
I
EE
E
Ac
c
e
ss
,
v
o
l
.
1
1
,
p
p
.
1
0
1
6
3
0
–
1
0
1
6
4
3
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
3
.
3
3
1
3
4
2
0
.
Evaluation Warning : The document was created with Spire.PDF for Python.