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.
14
,
No
.
3
,
J
u
n
e
2
0
2
5
,
p
p
.
2220
~
2
2
2
8
I
SS
N:
2
2
5
2
-
8938
,
DOI
: 1
0
.
1
1
5
9
1
/i
j
ai.
v
14
.i
3
.
p
p
2
2
2
0
-
2
2
2
8
2220
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
H
y
per
-
pa
ra
m
e
ter
s
o
pti
m
i
z
ed
deep
feature
co
nca
tena
ted
netw
o
rk
for
ped
ia
tric
pn
eu
m
o
nia
d
etec
tion
M
a
ry
Sh
y
ni H
illa
ry
,
Chitr
a
E
k
a
m
ba
ra
m
D
e
p
a
r
t
me
n
t
of
El
e
c
t
r
o
n
i
c
s
a
n
d
C
o
mm
u
n
i
c
a
t
i
o
n
En
g
i
n
e
e
r
i
n
g
,
S
R
M
I
n
st
i
t
u
t
e
of
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
K
a
t
t
a
n
k
u
l
a
t
h
u
r
,
I
n
d
i
a
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Feb
2
,
2
0
2
4
R
ev
i
s
ed
No
v
2
8
,
2
0
2
4
A
cc
ep
ted
J
an
2
7
,
2
0
2
5
P
n
e
u
m
o
n
ia,
an
in
f
e
c
ti
o
n
th
a
t
f
il
ls
t
h
e
a
lv
e
o
li
of
th
e
l
u
n
g
re
g
io
n
w
it
h
pus
c
a
u
se
s
a
h
ig
h
ra
te
of
c
h
ro
n
ic
il
l
n
e
ss
a
n
d
f
a
talit
y
a
m
o
n
g
st
c
h
il
d
re
n
a
c
ro
ss
th
e
g
lo
b
e
.
T
h
e
m
o
st
u
ti
li
z
e
d
im
a
g
in
g
m
o
d
a
li
ty
f
o
r
p
e
d
iatric
p
n
e
u
m
o
n
ia
i
d
e
n
ti
f
ica
ti
o
n
is
c
h
e
st
X
-
ra
y
s,
w
h
o
se
fe
a
tu
re
s
a
re
not
a
lw
a
y
s
re
a
d
il
y
v
isib
le
to
th
e
n
a
k
e
d
e
y
e
,
m
a
k
in
g
it
c
h
a
ll
e
n
g
in
g
f
o
r
ra
d
io
l
o
g
ists
to
m
a
k
e
p
re
c
ise
p
re
d
ictio
n
s
a
n
d
sa
v
e
li
v
e
s.
Kn
o
w
in
g
how
e
ss
e
n
ti
a
l
it
is
to
h
a
v
e
an
e
a
rl
y
a
n
d
d
isti
n
c
t
d
i
a
g
n
o
sis
of
p
n
e
u
m
o
n
ia,
sp
e
e
d
i
n
g
up
or
a
u
t
o
m
a
ti
n
g
th
e
d
e
tec
ti
o
n
p
r
o
c
e
ss
is
h
ig
h
ly
se
n
sib
le.
T
h
is
a
rti
c
le
p
ro
v
id
e
s
a
s
m
a
rt,
a
u
to
m
a
ted
s
y
ste
m
th
a
t
o
p
e
ra
tes
on
c
h
e
st
X
-
ra
y
i
m
a
g
e
s
a
n
d
can
be
su
c
c
e
ss
f
u
ll
y
u
ti
l
ize
d
f
o
r
sp
o
tt
i
n
g
p
n
e
u
m
o
n
ia.
T
h
e
d
e
e
p
f
e
a
tu
re
c
o
n
c
a
ten
a
ti
o
n
m
e
th
o
d
u
se
d
by
th
is
d
e
tec
ti
o
n
sy
ste
m
in
ten
d
s
to
c
o
m
b
in
e
th
e
o
u
tco
m
e
s
of
th
re
e
e
ff
e
c
ti
v
e
p
re
-
train
e
d
m
o
d
e
ls
to
c
o
n
f
irm
th
e
re
li
a
b
il
it
y
of
th
e
su
g
g
e
ste
d
a
p
p
ro
a
c
h
.
To
o
b
tain
its
o
p
ti
m
a
l
p
e
rf
o
rm
a
n
c
e
,
th
e
h
y
p
e
r
-
p
a
ra
m
e
t
e
rs
a
re
d
e
m
o
n
stra
ted
u
sin
g
a
tri
a
l
-
a
n
d
-
e
rro
r
a
p
p
r
o
a
c
h
th
a
t
su
rp
a
ss
e
s
e
x
isti
n
g
m
o
d
e
ls
w
it
h
9
9
.
6
8
%
a
c
c
u
ra
c
y
f
o
r
th
e
e
a
rly
d
ia
g
n
o
sis
of
p
n
e
u
m
o
n
ia.
A
re
a
l
-
ti
m
e
d
a
ta
sa
m
p
le
tes
t
is
c
o
n
d
u
c
ted
on
t
h
e
p
ro
p
o
se
d
p
n
e
u
m
o
n
ia
d
e
tec
ti
o
n
m
o
d
e
l
to
e
v
a
lu
a
te
its
ro
b
u
st
n
e
ss
.
K
ey
w
o
r
d
s
:
Data
au
g
m
e
n
tatio
n
Dee
p
lear
n
in
g
Featu
r
e
co
n
ca
te
n
atio
n
H
y
p
er
-
p
ar
a
m
eter
o
p
ti
m
izatio
n
K
-
f
o
ld
cr
o
s
s
-
v
alid
atio
n
P
ed
iatr
ic
p
n
eu
m
o
n
ia
T
h
is
is
an
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r
th
e
CC
BY
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
C
h
i
tr
a
E
k
a
m
b
ar
a
m
Dep
ar
t
m
en
t
of
E
lectr
o
n
ics
a
n
d
C
o
m
m
u
n
icat
io
n
E
n
g
in
ee
r
i
n
g
,
S
R
M
I
n
s
tit
u
te
of
Scie
n
ce
an
d
T
ec
h
n
o
lo
g
y
Kattan
k
u
lath
u
r
,
I
n
d
ia
E
m
ail:
c
h
itra
e@
s
r
m
is
t.e
d
u
.
i
n
1.
I
NT
RO
D
UCT
I
O
N
P
n
eu
m
o
n
ia
is
a
s
i
g
n
if
ic
a
n
t
t
h
r
ea
t
to
ch
ild
r
en
u
n
d
er
5,
w
i
th
an
an
t
icip
ated
11
m
i
llio
n
d
ea
t
h
s
by
2030
w
it
h
o
u
t
in
ter
v
e
n
tio
n
.
T
h
e
co
n
d
itio
n
f
il
ls
al
v
eo
li
w
it
h
p
u
s
a
n
d
f
l
u
id
,
m
a
k
i
n
g
b
r
ea
th
i
n
g
d
if
f
icu
l
t
an
d
r
ed
u
cin
g
o
x
y
g
en
i
n
tak
e
[
1
]
.
Un
tr
ea
ted
p
n
eu
m
o
n
ia
can
lead
to
co
m
p
licatio
n
s
lik
e
b
r
ea
th
i
n
g
f
ail
u
r
e
an
d
s
ep
s
is
,
ev
en
ca
u
s
i
n
g
d
ea
th
[
2
]
.
An
in
d
i
v
id
u
al
is
b
o
u
n
d
to
g
et
p
n
eu
m
o
n
ia
as
a
k
id
,
k
n
o
w
n
as
p
ed
iatr
ic
p
n
eu
m
o
n
ia
th
a
n
th
e
y
ar
e
as
a
g
r
o
w
n
-
up
[
3
]
.
P
ed
iatr
ic
p
n
eu
m
o
n
ia,
m
o
r
e
co
m
m
o
n
in
ch
i
ld
r
en
th
a
n
ad
u
l
ts
,
ex
h
ib
it
s
s
y
m
p
to
m
s
i
n
f
l
u
e
n
ce
d
by
th
e
in
f
ec
tio
n
's
ca
u
s
e,
ag
e,
an
d
o
v
er
all
h
ea
lt
h
.
C
o
m
m
o
n
s
i
g
n
s
i
n
clu
d
e
r
ap
id
b
r
ea
th
in
g
,
lo
w
o
x
y
g
e
n
s
atu
r
atio
n
,
co
u
g
h
,
an
d
h
i
g
h
b
o
d
y
te
m
p
er
at
u
r
e
[
4
]
.
P
ath
o
g
en
s
in
a
c
h
ild
'
s
n
a
s
al
p
ass
a
g
e
can
h
ar
m
t
h
e
l
u
n
g
s
w
h
e
n
i
n
h
aled
.
A
d
d
itio
n
al
l
y
,
d
u
r
i
n
g
an
d
r
ig
h
t
af
ter
ch
ild
b
ir
t
h
,
p
n
eu
m
o
n
ia
can
s
p
r
ea
d
th
r
o
u
g
h
b
lo
o
d
[
5
]
.
C
h
e
s
t
r
ad
io
g
r
ap
h
s
ar
e
w
id
ely
u
s
ed
to
d
etec
t
p
n
eu
m
o
n
ia,
r
ev
ea
lin
g
in
f
il
tr
ates
as
w
h
it
e
s
p
o
ts
on
X
-
r
a
y
s
[
6
]
,
[
7
]
.
Ho
w
e
v
er
,
ex
a
m
in
i
n
g
ch
est
X
-
r
a
y
s
is
d
if
f
icu
l
t
an
d
v
u
l
n
er
ab
le
to
s
u
b
j
ec
tiv
it
y
[
8
]
.
Dee
p
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
et
w
o
r
k
s
(D
-
C
NN)
ar
e
f
r
eq
u
en
tl
y
e
m
p
lo
y
ed
f
o
r
th
e
a
n
al
y
s
i
s
of
i
m
a
g
es
w
h
o
s
e
p
r
i
m
ar
y
f
u
n
ctio
n
is
p
atter
n
r
ec
o
g
n
itio
n
an
d
as
a
r
esu
lt,
u
tili
ze
d
to
id
en
tify
t
h
e
ab
n
o
r
m
alit
ies
in
m
ed
ic
al
i
m
ag
e
s
[
9
]
,
[
1
0
]
.
C
h
a
g
as
et
al
.
[
1
1
]
em
p
lo
y
ed
12
I
m
ag
e
Net
-
tr
ain
ed
C
N
N
s
,
en
h
a
n
ci
n
g
r
ad
io
g
r
ap
h
s
f
o
r
i
n
f
ec
ted
lu
n
g
r
e
g
io
n
d
etec
tio
n
t
h
r
o
u
g
h
ad
ap
ti
v
e
h
is
to
g
r
a
m
eq
u
al
izatio
n
.
T
esti
n
g
7
cla
s
s
i
f
ier
s
w
it
h
p
r
e
-
tr
ai
n
ed
m
o
d
els,
v
is
u
al
g
eo
m
etr
y
g
r
o
u
p
(
VGG
-
19
)
co
m
b
in
ed
w
it
h
a
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
(
SVM
)
o
u
tp
er
f
o
r
m
ed
84
c
o
m
b
i
n
atio
n
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
A
r
ti
f
I
n
tell
I
SS
N:
2252
-
8938
Hyp
er
-
p
a
r
a
mete
r
s
o
p
timiz
ed
d
ee
p
fea
tu
r
e
co
n
c
a
ten
a
ted
n
etw
o
r
k
fo
r
p
ed
ia
tr
ic
…
(
Ma
r
y
S
h
yn
i H
illa
ry
)
2221
In
th
eir
s
t
u
d
y
[
1
2
]
,
Nallu
r
i
an
d
Sas
ik
ala
p
r
o
p
o
s
ed
a
p
n
eu
m
o
n
ia
s
cr
ee
n
i
n
g
m
et
h
o
d
o
lo
g
y
in
teg
r
at
in
g
d
y
n
a
m
i
c
h
is
to
g
r
a
m
eq
u
al
izatio
n
(
DHE
)
an
d
m
ed
ia
n
f
il
ter
in
g
f
o
r
i
m
a
g
e
en
h
a
n
ce
m
e
n
t,
f
o
llo
w
ed
by
s
eg
m
e
n
tat
io
n
u
s
in
g
i
m
p
r
o
v
ed
w
ater
s
h
ed
s
eg
m
e
n
ta
tio
n
.
T
h
e
r
esear
ch
er
s
d
ev
is
ed
a
m
e
th
o
d
to
ex
tr
ac
t
p
er
tin
en
t
f
ea
t
u
r
es
an
d
s
elec
t
o
p
tim
a
l
o
n
es
f
o
r
tr
ain
i
n
g
d
ee
p
lear
n
in
g
clas
s
i
f
ier
s
.
T
h
r
o
u
g
h
th
e
u
tili
za
tio
n
of
th
e
m
ea
n
o
u
tp
u
t
f
r
o
m
t
h
ese
class
i
f
ier
s
t
h
e
y
ac
h
ie
v
ed
a
n
o
t
ab
le
d
etec
tio
n
ac
cu
r
ac
y
of
9
3
.
2
3
%.
Fer
n
an
d
e
s
et
a
l
.
[
1
3
]
in
tr
o
d
u
ce
d
a
p
ed
iatr
ic
p
n
eu
m
o
n
ia
C
N
N,
o
p
tim
izin
g
h
y
p
er
p
ar
a
m
eter
s
th
r
o
u
g
h
B
ay
e
s
ian
o
p
ti
m
izatio
n
.
T
h
e
y
e
n
h
a
n
ce
d
VGG
-
16
w
it
h
a
s
p
ec
i
alize
d
C
NN
o
v
er
la
y
,
ac
h
iev
i
n
g
o
p
ti
m
al
F1
-
s
co
r
e
p
er
f
o
r
m
a
n
ce
d
esp
ite
u
s
i
n
g
p
r
e
-
p
r
o
ce
s
s
i
n
g
tec
h
n
iq
u
es.
A
cc
o
r
d
in
g
to
Mo
h
a
m
m
ed
et
a
l.
[
1
4
]
,
th
e
alg
o
r
ith
m
en
h
a
n
ce
s
p
n
eu
m
o
n
ia
d
etec
tio
n
u
s
i
n
g
f
o
u
r
cu
s
to
m
-
m
o
d
if
ied
a
d
v
an
ce
d
C
NN
ar
c
h
itect
u
r
es,
r
ed
u
cin
g
p
ar
a
m
eter
s
by
e
x
cl
u
s
i
v
el
y
e
m
p
lo
y
i
n
g
co
n
v
o
l
u
tio
n
a
l
co
m
p
o
n
e
n
ts
.
It
u
t
ilizes
a
So
f
tMa
x
ac
ti
v
atio
n
w
it
h
g
lo
b
al
av
er
a
g
e
p
o
o
lin
g
to
m
ap
to
th
e
o
u
tp
u
t
l
a
y
er
,
g
en
er
ati
n
g
a
h
ea
t
m
ap
f
o
r
s
ig
n
al
s
tr
en
g
t
h
as
s
es
s
m
e
n
t.
Ker
m
an
y
et
al
.
[
1
5
]
ac
h
ie
v
ed
9
2
.
8
%
ac
cu
r
ac
y
in
d
ia
g
n
o
s
in
g
r
etin
al
d
is
ea
s
es
a
n
d
p
ed
iatr
ic
p
n
eu
m
o
n
ia
w
it
h
I
n
ce
p
tio
n
V3
tr
an
s
f
er
lea
r
n
in
g
u
s
in
g
5
,
232
lab
ele
d
ch
est
X
-
r
a
y
i
m
a
g
es
s
u
b
j
ec
ted
to
q
u
alit
y
co
n
tr
o
l.
To
o
v
er
co
m
e
th
e
ch
al
len
g
e
of
s
ca
r
ce
lab
eled
p
n
eu
m
o
n
ia
d
ata,
Ath
ar
e
m
p
lo
y
ed
ad
v
er
s
ar
ial
tr
ai
n
in
g
,
a
m
et
h
o
d
th
at
en
tail
s
tr
ain
i
n
g
a
s
ec
o
n
d
ar
y
n
et
w
o
r
k
to
p
r
o
d
u
ce
s
y
n
t
h
etic
X
-
r
a
y
s
r
ese
m
b
li
n
g
r
ea
l
o
n
e
s
b
u
t
w
i
t
h
s
li
g
h
t
v
ar
iat
io
n
s
.
Usi
n
g
a
m
i
x
of
a
u
t
h
e
n
tic
a
n
d
s
y
n
th
et
ic
X
-
r
a
y
s
th
e
Alex
Net
m
o
d
el
w
as
tr
ain
ed
to
ac
h
iev
e
an
i
m
p
r
es
s
iv
e
v
alid
atio
n
ac
c
u
r
ac
y
of
9
8
.
2
8
%
[
1
6
]
.
A
cc
o
r
d
in
g
to
Ku
n
d
u
et
a
l.
[
1
7
]
,
a
p
n
eu
m
o
n
ia
d
iag
n
o
s
tic
s
y
s
te
m
w
as
i
n
tr
o
d
u
ce
d
u
s
in
g
w
eig
h
ted
en
s
e
m
b
le
lear
n
i
n
g
,
co
m
b
i
n
in
g
th
r
ee
cla
s
s
if
ier
s
w
it
h
w
ei
g
h
ts
d
eter
m
in
ed
by
f
o
u
r
a
s
s
es
s
m
e
n
t
cr
iter
ia
an
d
t
h
e
h
y
p
er
b
o
lic
tan
g
e
n
t
f
u
n
ctio
n
.
T
h
e
f
r
a
m
e
w
o
r
k
ac
h
ie
v
ed
o
p
tim
al
r
es
u
lt
s
w
h
e
n
all
la
y
er
s
wer
e
tr
ain
ed
on
t
w
o
o
p
en
-
ac
ce
s
s
p
n
eu
m
o
n
ia
X
-
r
a
y
d
atasets
.
Yi
et
al
.
[
1
8
]
p
r
o
p
o
s
ed
a
d
ee
p
C
NN
m
o
d
el
w
it
h
52
co
n
v
o
l
u
tio
n
al
la
y
er
s
f
o
r
f
ea
t
u
r
e
ex
tr
ac
t
io
n
a
n
d
t
wo
d
en
s
e
la
y
er
s
f
o
r
p
n
e
u
m
o
n
i
a
v
er
s
u
s
n
o
r
m
al
cla
s
s
i
f
icat
io
n
f
r
o
m
X
-
r
a
y
a
n
d
co
m
p
u
ted
to
m
o
g
r
ap
h
y
(
CT
)
i
m
a
g
es.
T
h
e
c
h
o
s
en
i
m
ag
e
s
ize
of
2
0
0
×2
0
0
×3
y
ield
ed
a
p
r
ec
is
e
v
al
id
atio
n
ac
cu
r
ac
y
of
9
6
.
0
9
%.
A
l
m
a
s
lu
k
h
[
19]
d
ev
elo
p
ed
a
lig
h
t
w
ei
g
h
t
p
n
eu
m
o
n
ia
d
etec
tio
n
m
o
d
el
b
ased
on
Den
s
eNe
t
-
121,
u
tili
zi
n
g
r
an
d
o
m
s
ea
r
ch
h
y
p
er
p
ar
am
eter
t
u
n
i
n
g
.
Den
s
e
co
n
n
ec
tiv
it
y
in
De
n
s
eNe
t
-
121
ad
d
r
ess
es
t
h
e
v
a
n
is
h
i
n
g
g
r
ad
ien
t
is
s
u
e
an
d
p
r
o
m
o
tes
f
e
atu
r
e
r
eu
s
e.
T
h
e
m
o
d
el's
r
ed
u
ce
d
p
a
r
am
eter
co
u
n
t
m
ak
e
s
it
en
er
g
y
-
e
f
f
icie
n
t
an
d
s
u
itab
le
f
o
r
r
ap
id
d
etec
tio
n
in
m
ed
ical
s
y
s
te
m
s
.
A
l
s
h
m
r
a
n
i
et
a
l.
[
2
0
]
em
p
lo
y
ed
th
e
p
r
e
-
tr
ain
ed
VGG
1
9
n
et
w
o
r
k
to
e
x
tr
ac
t
f
ea
t
u
r
es
f
r
o
m
ch
e
s
t
r
ad
io
g
r
ap
h
s
,
ca
teg
o
r
izin
g
s
i
x
d
is
t
in
ct
lu
n
g
d
is
ea
s
es.
T
o
au
g
m
e
n
t
t
h
e
f
ea
t
u
r
e
e
x
tr
ac
tio
n
ca
p
ab
ilit
y
o
f
VGG
1
9
,
th
e
r
esear
ch
er
s
in
t
eg
r
ated
th
r
ee
s
u
p
p
le
m
e
n
tar
y
c
o
n
v
o
l
u
tio
n
al
b
lo
ck
s
in
to
th
e
n
et
w
o
r
k
.
Fo
llo
w
in
g
tr
ain
i
n
g
o
n
8
0
,
0
0
0
s
am
p
les,
t
h
i
s
ad
ap
ted
n
et
w
o
r
k
ac
h
ie
v
ed
a
r
e
m
ar
k
ab
le
tes
tin
g
ac
cu
r
ac
y
o
f
9
6
.
4
8
% [
2
0
]
.
T
h
is
r
esear
ch
o
p
ti
m
ize
s
h
y
p
e
r
-
p
ar
a
m
eter
s
f
o
r
a
d
ee
p
f
ea
tu
r
e
co
n
ca
te
n
ated
m
o
d
el,
m
er
g
in
g
d
ee
p
f
ea
t
u
r
es
f
r
o
m
t
h
r
ee
s
u
cc
e
s
s
f
u
l
p
r
e
-
tr
ain
ed
m
o
d
els
I
n
ce
p
ti
o
n
V3
,
VGG
-
16
an
d
Den
s
e
Net
-
201
to
m
it
ig
ate
v
an
i
s
h
in
g
g
r
ad
ien
t
a
n
d
o
v
er
f
i
ttin
g
in
a
b
in
ar
y
d
etec
tio
n
s
y
s
te
m
.
T
h
e
p
r
o
p
o
s
ed
co
n
ca
ten
a
ted
ap
p
r
o
ac
h
w
a
s
d
em
o
n
s
tr
ated
w
it
h
m
a
n
u
al
h
y
p
er
-
p
ar
am
e
ter
tu
n
i
n
g
to
o
b
tain
an
o
p
tim
al
m
o
d
el
f
o
r
p
r
ec
is
e
esti
m
atio
n
s
.
Desp
ite
th
e
ted
io
u
s
n
atu
r
e
of
m
an
u
al
t
u
n
in
g
,
it
is
v
al
u
ab
le
f
o
r
y
o
u
n
g
r
esear
ch
er
s
to
g
r
asp
h
y
p
er
-
p
ar
a
m
eter
b
eh
a
v
io
r
an
d
its
i
m
p
ac
t
on
n
et
w
o
rk
w
e
ig
h
t
s
.
2.
P
RO
P
O
SE
D
F
I
N
E
-
T
UN
E
D
CO
NCAT
E
NA
T
I
O
N
M
E
T
H
O
D
Fig
u
r
e
1
p
r
esen
ts
th
e
v
i
s
u
al
lay
o
u
t
of
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
o
lo
g
y
.
T
h
e
f
r
am
e
w
o
r
k
co
n
s
is
ts
of
s
ix
m
ai
n
s
tag
e
s
:
d
ataset
co
llect
io
n
a
n
d
s
p
litt
i
n
g
,
d
ata
p
r
e
-
p
r
o
ce
s
s
i
n
g
,
tr
a
n
s
f
er
lear
n
i
n
g
,
m
o
d
el
co
n
ca
ten
ati
o
n
,
h
y
p
er
-
p
ar
a
m
eter
tu
n
i
n
g
a
n
d
f
i
n
all
y
m
o
d
el
tr
ain
i
n
g
an
d
p
r
ed
ictio
n
.
2
.
1
.
Da
t
a
s
et
des
cr
iptio
n
T
h
e
X
-
r
ay
im
ag
es
u
s
ed
f
o
r
th
e
d
ev
e
lo
p
m
en
t
of
th
e
p
r
o
p
o
s
e
d
m
o
d
el
a
r
e
f
r
o
m
a
m
ed
ica
l
im
ag
e
d
ir
ec
t
o
r
y
cr
ea
t
ed
by
Ke
r
m
an
y
et
al
.
[
2
1
]
w
h
ich
is
p
u
b
lic
ly
ac
ce
s
s
ib
l
e
f
r
o
m
th
e
Kag
g
le
d
a
ta
b
as
e.
T
h
e
d
ir
e
ct
o
r
y
c
o
n
s
is
ts
of
ch
est
X
-
r
ay
(
C
XR
)
im
ag
es
of
c
h
ild
r
en
ag
ed
f
r
o
m
1
to
5
y
ea
r
s
p
r
o
v
i
d
ed
by
th
e
Gu
an
g
zh
o
u
W
o
m
en
an
d
C
h
ild
r
en
’
s
Me
d
ic
al
C
en
te
r
.
Ou
t
of
th
e
to
t
al
5
,
856
C
XR
im
ag
es,
4
,
273
a
r
e
ass
o
ci
at
ed
w
ith
p
n
eu
m
o
n
ia,
an
d
th
e
r
em
ain
in
g
1
,
583
ar
e
h
ea
l
th
y
.
1
0
%
of
th
e
n
o
r
m
al
cl
ass
h
as
b
e
en
ch
o
s
en
f
r
o
m
each
c
lass
f
o
r
test
in
g
to
p
r
e
v
en
t
c
lass
im
b
alan
ce
ch
all
en
g
es.
Fu
r
th
er
m
o
r
e,
eth
ic
al
a
p
p
r
o
v
al
f
o
r
th
e
ex
ec
u
ti
o
n
of
th
is
s
tu
d
y
w
as
g
r
an
te
d
by
th
e
SR
M
Me
d
ic
al
C
o
ll
eg
e
Ho
s
p
i
tal
an
d
R
ese
a
r
ch
C
en
t
r
e
,
lo
ca
te
d
in
K
att
an
k
u
lath
u
r
,
I
n
d
i
a.
13
r
e
al
-
tim
e
ch
est
X
-
r
ay
im
ag
es
in
clu
d
in
g
6
p
e
d
i
at
r
i
c
p
n
eu
m
o
n
ia
s
p
ec
im
en
s
an
d
4
n
o
r
m
al
X
-
r
ay
s
h
av
e
b
e
en
co
lle
cte
d
w
h
ich
p
r
o
v
i
d
e
d
iv
e
r
s
e
s
et
of
c
ases
to
ass
ess
th
e
m
o
d
e
l'
s
p
e
r
f
o
r
m
an
ce
ac
r
o
s
s
d
if
f
er
en
t
c
o
n
d
iti
o
n
s
an
d
s
ce
n
ar
io
s
.
2
.
2
.
Da
t
a
s
et
pre
-
pro
ce
s
s
ing
No
r
m
a
lizatio
n
in
t
h
e
p
r
e
-
p
r
o
ce
s
s
i
n
g
s
tag
e
s
p
ee
d
s
up
co
n
v
er
g
en
ce
by
s
tab
ilizi
n
g
t
h
e
lear
n
i
n
g
p
r
o
ce
s
s
.
In
th
i
s
r
esear
ch
,
C
X
R
i
m
ag
es
ar
e
s
ca
led
f
r
o
m
0
-
255
to
0
-
1
by
d
iv
id
i
n
g
each
p
ix
el
by
255,
en
s
u
r
i
n
g
n
o
r
m
aliza
t
i
on
w
ith
in
th
e
r
an
g
e
of
0
-
1
[
2
2
]
.
Data
au
g
m
en
ta
ti
o
n
s
tr
ateg
y
is
i
m
p
le
m
e
n
ted
to
en
h
a
n
ce
v
ar
iab
ilit
y
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
I
n
t J
A
r
ti
f
I
n
tell
,
Vo
l.
1
4
,
No
.
3
,
J
u
n
e
2
0
2
5
:
2
2
2
0
-
2228
2222
an
d
ad
d
r
ess
class
i
m
b
ala
n
ce
ch
alle
n
g
e
s
in
t
h
e
tr
ain
i
n
g
d
atas
et.
Giv
en
t
h
e
s
u
b
s
ta
n
tial
n
u
m
b
er
of
p
n
eu
m
o
n
ia
-
af
f
ec
ted
i
m
a
g
es
an
d
a
li
m
i
ted
n
u
m
b
er
of
n
o
r
m
al
i
m
a
g
e
s,
p
o
t
en
tial
b
ias
to
w
ar
d
s
th
e
p
n
eu
m
o
n
ia
clas
s
e
x
is
t
s
[
2
3
]
.
T
h
er
ef
o
r
e,
f
o
r
class
0
p
n
eu
m
o
n
ia
i
m
a
g
e
s
,
au
g
m
e
n
tatio
n
in
v
o
lv
es
clo
ck
w
i
s
e/co
u
n
ter
clo
ck
w
i
s
e
r
o
tatio
n
by
15°
an
d
h
o
r
izo
n
tal
f
lip
p
i
n
g
.
To
b
alan
ce
d
ata
in
c
lass
1
(
n
o
r
m
al
i
m
ag
e
s
)
,
au
g
m
en
tatio
n
i
n
cl
u
d
e
s
r
o
tatio
n
,
h
o
r
izo
n
tal
f
lip
p
in
g
,
20°
s
h
ea
r
,
20%
zo
o
m
,
10%
lef
t/rig
h
t
s
h
i
f
t,
an
d
v
ar
y
in
g
b
r
i
g
h
t
n
es
s
f
r
o
m
20
to
9
0
%
.
Fig
u
r
e
1.
T
h
e
v
is
u
al
la
y
o
u
t
of
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
o
lo
g
y
2
.
3
.
T
ra
ns
f
er
lea
rning
T
r
an
s
f
er
lear
n
in
g
,
th
e
ap
p
licatio
n
of
k
n
o
w
led
g
e
g
ai
n
ed
f
r
o
m
a
p
r
ev
io
u
s
task
to
en
h
a
n
ce
lea
r
n
in
g
in
a
n
e
w
r
elate
d
tas
k
,
is
e
m
p
lo
y
ed
in
th
i
s
r
esear
ch
[
2
4
]
.
Du
e
to
li
m
ited
ac
ce
s
s
to
th
e
p
ed
iatr
ic
p
n
eu
m
o
n
ia
d
ataset,
s
ev
e
n
p
r
e
-
tr
ain
ed
m
o
d
els
n
a
m
el
y
VGG
-
16,
VGG
-
19,
I
n
ce
p
tio
n
V3
,
Xce
p
tio
n
,
M
o
b
ileNet
V2
,
Den
s
eNe
t
-
201,
an
d
R
esNe
t
-
50
th
a
t
ar
e
o
f
ten
e
m
p
lo
y
ed
in
m
ed
ical
ap
p
lica
tio
n
s
a
n
d
tr
ain
ed
w
it
h
t
h
e
I
m
ag
eNe
t
d
ataset
ar
e
u
tili
ze
d
.
T
h
e
f
u
l
l
y
co
n
n
ec
ted
l
a
y
er
,
w
h
ic
h
s
er
v
ed
as
t
h
e
f
i
n
a
l
la
y
er
in
t
h
ese
m
o
d
els
alo
n
e
is
r
etr
ain
ed
w
i
th
o
u
t
alter
in
g
th
e
w
ei
g
h
ts
of
th
e
i
n
it
ial
la
y
er
s
.
2
.
4
.
M
o
del
co
nca
t
ena
t
i
o
n
T
h
e
c
o
n
ca
ten
a
te
d
m
o
d
e
l
is
f
o
r
m
ed
by
c
o
m
b
in
in
g
f
e
atu
r
es
f
r
o
m
th
e
to
p
-
p
e
r
f
o
r
m
in
g
th
r
e
e
o
u
t
of
th
e
s
ev
en
p
r
e
-
tr
ain
e
d
m
o
d
els
ev
a
lu
ate
d
on
th
e
p
ed
iat
r
i
c
p
n
eu
m
o
n
ia
d
at
aset
.
T
h
e
c
o
n
v
o
lu
t
io
n
al
b
a
s
e
of
I
n
ce
p
ti
o
n
V3
,
VGG
-
1
6
,
an
d
Den
s
e
N
et
-
2
0
1
is
f
r
o
z
en
,
s
er
v
in
g
as
f
e
atu
r
e
ex
tr
a
ct
o
r
s
.
I
n
p
u
t
im
ag
es
p
r
o
p
ag
a
te
f
o
r
w
ar
d
th
r
o
u
g
h
th
ese
n
e
tw
o
r
k
s
,
an
d
o
p
t
im
al
f
ea
tu
r
es
a
r
e
ex
t
r
a
ct
ed
f
r
o
m
th
e
la
y
er
p
r
i
o
r
to
th
e
f
u
lly
co
n
n
e
ct
ed
lay
er
[
2
5
]
.
A
t
o
t
a
l
of
2
,
0
4
8
f
ea
tu
r
es
f
r
o
m
I
n
ce
p
ti
o
n
V
3
,
5
1
2
f
r
o
m
VGG
-
16
,
an
d
1
,
920
f
r
o
m
Den
s
eN
et
-
201
ar
e
ex
tr
ac
te
d
,
r
esu
ltin
g
in
a
co
n
ca
t
en
at
ed
m
o
d
el
w
ith
4
,
480
f
ea
tu
r
es.
T
h
es
e
f
ea
tu
r
e
s
ets
a
r
e
th
en
f
ed
in
t
o
a
r
esh
a
p
ed
f
u
lly
co
n
n
ec
t
e
d
lay
er
,
f
o
l
lo
w
ed
by
a
s
ig
m
o
id
c
lass
if
ie
r
f
o
r
p
n
eu
m
o
n
ia
an
d
n
o
r
m
al
C
XR
im
ag
e
c
lass
if
i
ca
t
io
n
.
2
.
5
.
H
y
per
-
pa
ra
m
et
er
t
un
in
g
H
y
p
er
-
p
ar
a
m
eter
t
u
n
i
n
g
is
an
i
m
p
er
at
iv
e
p
er
s
p
ec
tiv
e
t
h
at
o
u
t
co
m
e
s
in
t
h
e
b
est
ex
ec
u
tio
n
of
th
e
m
o
d
el
by
tr
ac
k
i
n
g
t
h
e
r
i
g
h
t
co
m
b
i
n
at
io
n
of
h
y
p
er
-
p
ar
a
m
eter
s
in
a
s
en
s
ib
le
m
ea
s
u
r
e
of
ti
m
e
[
2
6
]
.
Fo
r
a
m
o
r
e
r
eliab
le
an
d
o
p
tim
ized
mo
d
el,
o
p
tim
al
h
y
p
er
-
p
ar
a
m
eter
s
m
u
s
t
be
d
ef
i
n
ed
p
r
io
r
to
d
ata
f
itti
n
g
b
ec
au
s
e
t
h
e
y
v
ar
y
f
o
r
d
is
tin
ct
d
atase
ts
[
2
7
]
.
T
h
e
h
y
p
er
p
ar
am
eter
s
ar
e
d
e
m
o
n
s
tr
at
ed
u
s
in
g
a
tr
ial
-
an
d
-
er
r
o
r
ap
p
r
o
ac
h
to
o
b
tain
th
e
o
p
tim
a
l
p
er
f
o
r
m
a
n
ce
of
t
h
e
p
r
o
p
o
s
ed
co
n
ca
ten
ated
m
o
d
e
l,
an
d
t
h
e
o
p
ti
m
al
m
o
d
el
is
u
til
ized
to
d
etec
t
p
n
eu
m
o
n
ia
in
ch
ild
r
en
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
A
r
ti
f
I
n
tell
I
SS
N:
2252
-
8938
Hyp
er
-
p
a
r
a
mete
r
s
o
p
timiz
ed
d
ee
p
fea
tu
r
e
co
n
c
a
ten
a
ted
n
etw
o
r
k
fo
r
p
ed
ia
tr
ic
…
(
Ma
r
y
S
h
yn
i H
illa
ry
)
2223
T
h
e
p
r
elim
i
n
ar
y
s
tep
is
to
s
elec
t
th
e
o
p
ti
m
izer
,
w
h
ich
is
an
alg
o
r
it
h
m
u
s
ed
to
u
p
d
ate
th
e
v
ar
io
u
s
attr
ib
u
tes
of
th
e
m
o
d
el
to
m
i
n
i
m
ize
t
h
e
lo
s
s
es.
T
h
e
A
d
a
m
o
p
ti
m
izer
p
er
f
o
r
m
ed
w
ell
w
i
th
o
u
r
da
taset.
Up
o
n
s
elec
tio
n
of
t
h
e
o
p
ti
m
izer
,
t
h
e
m
o
d
el
is
v
alid
ated
f
o
r
o
p
ti
m
u
m
p
er
f
o
r
m
a
n
ce
u
s
i
n
g
a
r
an
g
e
of
lear
n
i
n
g
r
ates
a
n
d
b
atch
s
izes.
Af
ter
th
e
lear
n
i
n
g
r
ate
an
d
b
atch
s
ize
ar
e
s
et
to
th
e
d
esire
d
lev
els
th
e
m
o
d
e
l
is
ev
al
u
ated
w
it
h
d
if
f
er
e
n
t
v
al
u
es
of
m
o
m
e
n
t
u
m
to
ac
h
ie
v
e
it
s
o
p
ti
m
al
v
al
u
e.
T
h
e
m
o
d
el
is
a
s
s
es
s
ed
f
o
r
v
ar
io
u
s
w
ei
g
h
t
d
ec
a
y
an
d
ep
s
ilo
n
v
al
u
es
to
d
eter
m
in
e
its
id
ea
l
v
al
u
e.
Fin
a
ll
y
,
a
lear
n
in
g
r
ate
ad
j
u
s
t
m
en
t
h
as
b
ee
n
m
ad
e
w
h
ich
p
r
o
d
u
ce
d
m
ea
n
in
g
f
u
l
i
m
p
r
o
v
e
m
en
ts
.
2
.
6
.
M
o
del
t
ra
ini
ng
a
nd
predict
i
on
T
h
e
co
n
ca
ten
ated
m
o
d
el
is
tr
ain
ed
an
d
v
alid
ated
f
o
r
50
ep
o
ch
s
u
s
i
n
g
10
-
f
o
ld
cr
o
s
s
-
v
alid
atio
n
,
r
ed
u
cin
g
b
ias
an
d
i
m
p
r
o
v
i
n
g
g
e
n
er
aliza
tio
n
.
T
h
e
tr
ain
in
g
in
v
o
lv
ed
r
an
d
o
m
l
y
d
iv
id
i
n
g
th
e
d
ataset
i
n
to
10
f
o
ld
s
(
9
:1
s
p
lit
f
o
r
tr
ain
i
n
g
an
d
v
al
id
atio
n
)
ac
r
o
s
s
t
en
iter
a
tio
n
s
[
2
8
]
.
T
h
e
av
er
ag
e
ac
c
u
r
a
c
y
o
b
tain
ed
in
ea
c
h
iter
atio
n
is
t
h
e
f
in
al
ac
c
u
r
ac
y
of
th
e
m
o
d
el.
Star
tin
g
w
i
th
r
a
n
d
o
m
l
y
i
n
itial
ized
w
ei
g
h
ts
a
n
d
b
iase
s
,
th
e
m
o
d
el
's
p
r
ed
icted
o
u
tp
u
t
s
ar
e
co
m
p
ar
ed
w
it
h
ac
tu
al
o
u
tp
u
ts
.
W
ei
g
h
ts
an
d
b
iases
ar
e
th
e
n
u
p
d
at
ed
an
d
b
ac
k
p
r
o
p
ag
ated
th
r
o
u
g
h
i
n
itial
la
y
er
s
b
ased
on
th
e
lo
s
s
f
u
n
ctio
n
,
a
i
m
in
g
to
m
i
n
i
m
ize
lo
s
s
an
d
i
m
p
r
o
v
e
a
cc
u
r
ac
y
[
2
9
]
.
T
r
ain
in
g
co
n
tin
u
ed
u
n
til
p
ar
a
m
eter
u
p
d
ates
no
lo
n
g
er
e
n
h
a
n
ce
d
v
alid
atio
n
ac
c
u
r
ac
y
,
e
m
p
lo
y
in
g
ea
r
l
y
s
to
p
p
in
g
w
it
h
a
p
atien
ce
v
alu
e
of
5.
C
las
s
i
f
icatio
n
u
tili
ze
d
th
e
f
u
ll
y
co
n
n
ec
ted
la
y
er
w
it
h
a
s
ig
m
o
id
ac
tiv
at
io
n
f
u
n
ctio
n
f
o
r
b
in
ar
y
o
u
tp
u
t
co
r
r
esp
o
n
d
in
g
to
p
n
eu
m
o
n
ia
an
d
n
o
r
m
al
ca
s
e
s
.
3.
RE
SU
L
T
S
AND
D
I
SCU
SS
I
O
N
T
h
e
p
r
o
p
o
s
ed
co
n
ca
ten
ated
m
o
d
el
is
r
ef
in
ed
s
y
s
te
m
atica
l
l
y
to
en
h
an
ce
its
e
f
f
ec
t
iv
e
n
es
s
in
d
etec
tin
g
p
ed
iatr
ic
p
n
eu
m
o
n
ia
f
r
o
m
c
h
est
r
ad
io
g
r
ap
h
s
.
T
h
e
f
in
e
-
t
u
n
i
n
g
p
r
o
ce
s
s
b
eg
in
s
w
ith
p
r
e
-
tr
ain
i
n
g
o
n
a
d
iv
er
s
e
d
ataset,
allo
w
i
n
g
th
e
m
o
d
el
to
lear
n
f
u
n
d
a
m
e
n
tal
p
atte
r
n
s
an
d
f
ea
tu
r
es
r
ele
v
an
t
to
m
ed
ical
i
m
a
g
i
n
g
.
Su
b
s
eq
u
e
n
tl
y
,
th
e
m
o
d
el
u
n
d
er
g
o
es
iter
ativ
e
ad
j
u
s
t
m
e
n
t
s
,
in
clu
d
in
g
d
o
m
ain
-
s
p
ec
if
ic
tr
ain
i
n
g
,
h
y
p
er
p
ar
a
m
eter
o
p
tim
izatio
n
,
an
d
p
er
f
o
r
m
a
n
c
e
ev
alu
atio
n
,
to
en
s
u
r
e
it
ac
h
i
ev
es
h
i
g
h
ac
cu
r
ac
y
a
n
d
r
eliab
ilit
y
i
n
p
n
eu
m
o
n
ia
class
i
f
icatio
n
.
3.
1
.
P
er
f
o
rm
a
nce
m
et
rics
P
er
f
o
r
m
a
n
ce
m
etr
ics
e
v
al
u
at
e
th
e
p
er
f
o
r
m
an
ce
of
t
h
e
d
ee
p
lear
n
i
n
g
m
o
d
el
b
ased
on
its
ab
ilit
y
to
f
o
r
ec
ast
u
n
o
b
s
er
v
ed
d
ata.
T
h
e
p
r
ed
icted
o
u
tco
m
es
of
t
h
e
m
o
d
el
s
ar
e
v
i
s
u
a
lized
in
t
h
e
f
o
r
m
of
a
co
n
f
u
s
io
n
m
atr
i
x
t
h
at
h
as
4
e
n
tr
ies:
i
)
tr
u
e
p
o
s
iti
v
es
(
T
P
)
:
co
r
r
ec
tly
p
r
ed
icted
p
n
eu
m
o
n
ia
c
a
s
es,
ii
)
t
r
u
e
n
e
g
ati
v
es
(
T
N)
:
co
r
r
ec
tly
p
r
ed
icted
n
o
r
m
al
ca
s
es,
iii
)
f
alse
p
o
s
iti
v
es
(
FP
)
:
in
co
r
r
ec
tl
y
p
r
ed
icted
n
o
r
m
al
ca
s
es,
an
d
iv
)
f
alse
n
eg
at
iv
e
s
(
FN)
:
in
co
r
r
ec
tl
y
p
r
ed
icted
p
n
eu
m
o
n
ia
ca
s
e
s
[
3
0
]
.
T
ab
le
1
d
is
p
lay
s
th
e
m
e
tr
ics
u
s
ed
to
ev
alu
ate
th
e
p
er
f
o
r
m
a
n
ce
of
t
h
e
p
r
o
p
o
s
ed
m
o
d
el.
T
ab
le
1.
Me
tr
ics
u
s
ed
to
ev
alu
ate
th
e
p
er
f
o
r
m
a
n
ce
of
t
h
e
m
o
d
el
M
e
t
r
i
c
s
F
o
r
mu
l
a
A
c
c
u
r
a
c
y
+
+
+
+
P
r
e
c
i
si
o
n
+
R
e
c
a
l
l
+
F1
-
sco
r
e
2
×
×
+
S
p
e
c
i
f
i
c
i
t
y
+
M
C
C
×
−
×
√
(
+
)
(
+
)
(
+
)
(
+
)
3.
2
.
Resul
t
s
T
h
e
co
n
ca
ten
atio
n
m
et
h
o
d
in
t
h
is
d
etec
tio
n
s
y
s
te
m
i
n
te
g
r
ates
f
ea
t
u
r
es
f
r
o
m
th
r
ee
p
r
e
-
tr
ain
e
d
m
o
d
els
:
I
n
ce
p
tio
n
V3
,
VGG
-
16,
an
d
Den
s
eNe
t
-
201,
ch
o
s
en
b
ased
on
th
eir
p
er
f
o
r
m
an
ce
m
etr
ics
a
m
o
n
g
s
e
v
e
n
ev
al
u
ated
m
o
d
el
s
w
id
el
y
u
s
e
d
in
m
ed
ica
l
d
iag
n
o
s
tics
.
E
v
a
lu
at
io
n
m
etr
i
cs
in
T
ab
le
2
in
d
icate
th
at
VG
G
-
16
an
d
I
n
ce
p
tio
n
V3
ac
h
ie
v
ed
th
e
h
i
g
h
est
ac
c
u
r
ac
y
s
co
r
e
of
9
4
.
9
4
%.
C
o
n
s
id
e
r
in
g
o
t
h
er
m
etr
ics
lik
e
p
r
ec
is
i
o
n
,
s
p
ec
if
icit
y
,
an
d
Ma
tth
e
w
co
r
r
elatio
n
co
ef
f
ici
en
t
(
MCC
)
,
I
n
ce
p
tio
n
V3
o
u
tp
er
f
o
r
m
ed
VGG
-
16
w
it
h
100%
p
r
ec
is
io
n
an
d
s
p
ec
if
icit
y
.
Si
m
ilar
l
y
,
Den
s
eN
et
-
201
o
u
tp
er
f
o
r
m
ed
VGG
-
19
w
it
h
an
ac
cu
r
ac
y
of
9
3
.
6
7
%,
d
e
m
o
n
s
tr
atin
g
100%
p
r
ec
is
io
n
an
d
s
p
ec
i
f
icit
y
.
Fi
g
u
r
e
s
2
(
a)
to
2
(
c
)
d
ep
icts
th
e
co
n
f
u
s
io
n
m
atr
i
x
f
o
r
th
e
t
h
r
ee
b
est
-
p
er
f
o
r
m
ed
m
o
d
el
s
on
th
e
tes
t
d
ata.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
I
n
t J
A
r
ti
f
I
n
tell
,
Vo
l.
1
4
,
No
.
3
,
J
u
n
e
2
0
2
5
:
2
2
2
0
-
2228
2224
T
ab
le
2
.
C
lass
if
icatio
n
r
es
u
lt
s
o
f
th
e
s
e
v
e
n
p
r
e
-
tr
ain
ed
m
o
d
el
s
M
o
d
e
l
TP
TN
FP
FN
P
e
r
f
o
r
man
c
e
(
%)
M
C
C
A
c
c
u
r
a
c
y
P
r
e
c
i
si
o
n
S
e
n
si
t
i
v
i
t
y
F1
-
s
c
o
r
e
S
p
e
c
i
f
i
c
i
t
y
VGG
-
16
1
4
3
1
5
7
1
15
9
4
.
9
4
9
9
.
3
1
9
0
.
5
1
9
4
.
7
0
9
9
.
3
7
0
.
9
0
2
3
VGG
-
19
1
3
9
1
5
7
1
19
9
3
.
6
7
9
9
.
2
9
8
7
.
9
7
9
3
.
2
9
9
9
.
3
7
0
.
8
7
9
1
I
n
c
e
p
t
i
o
n
V3
1
4
2
1
5
8
0
16
9
4
.
9
4
1
0
0
8
9
.
8
7
9
4
.
6
7
1
0
0
0
.
9
0
3
4
X
c
e
p
t
i
o
n
1
3
4
1
5
6
2
24
9
1
.
7
7
9
8
.
5
3
8
4
.
8
1
9
1
.
1
6
9
8
.
7
3
0
.
8
4
3
7
M
o
b
i
l
e
N
e
t
V2
1
3
6
1
5
6
2
22
9
2
.
4
1
9
8
.
5
5
8
6
.
0
8
9
1
.
8
9
9
8
.
7
3
0
.
8
5
4
9
D
e
n
se
N
e
t
-
201
1
3
8
1
5
8
0
20
9
3
.
6
7
1
0
0
8
7
.
3
4
9
3
.
2
4
1
0
0
0
.
8
8
0
5
R
e
sN
e
t
-
50
1
2
3
1
5
7
1
35
8
8
.
6
1
9
9
.
1
9
7
7
.
8
5
8
7
.
2
3
9
9
.
3
7
0
.
7
9
0
7
(
a)
(
b
)
(
c)
Fig
u
r
e
2
.
C
o
n
f
u
s
io
n
m
atr
ix
f
o
r
th
e
th
r
ee
b
est
-
p
er
f
o
r
m
ed
m
o
d
els o
f
(
a)
I
n
ce
p
tio
n
V3
,
(
b
)
V
GG
-
1
6
,
an
d
(
c)
Den
s
eNe
t
-
201
T
h
e
b
est
f
ea
tu
r
es
o
f
th
e
to
p
3
m
o
d
els
I
n
ce
p
tio
n
V3
,
VGG
-
1
6
,
an
d
Den
s
eNe
t
-
2
0
1
ar
e
c
o
m
b
in
ed
to
d
ev
elo
p
th
e
co
n
ca
te
n
ated
m
o
d
el.
T
a
b
le
3
h
ig
h
l
ig
h
t
s
th
e
p
er
f
o
r
m
an
ce
o
f
th
e
co
n
ca
ten
a
ted
m
o
d
el.
I
n
co
m
p
ar
is
o
n
to
th
e
i
n
d
iv
id
u
al
p
er
f
o
r
m
an
ce
s
o
f
t
h
e
t
h
r
ee
m
o
d
el
s
,
th
e
co
n
ca
ten
ated
m
o
d
el
p
er
f
o
r
m
s
b
etter
.
T
h
e
co
n
ca
ten
ated
m
o
d
el
ac
h
iev
ed
an
ac
cu
r
ac
y
o
f
9
8
.
7
3
%
w
h
ic
h
is
al
m
o
s
t
a
4
%
in
cr
ea
s
e
w
h
e
n
co
m
p
ar
ed
w
it
h
t
h
e
in
d
iv
id
u
a
l
p
er
f
o
r
m
a
n
ce
.
T
h
e
ac
cu
r
ac
y
o
f
th
e
co
n
ca
te
n
ated
m
o
d
el
in
cr
ea
s
ed
as
a
r
esu
lt
o
f
a
s
h
ar
p
in
cr
ea
s
e
in
t
h
e
tr
u
e
p
o
s
itiv
e
v
al
u
e.
T
h
e
tr
u
e
p
o
s
itiv
e
v
al
u
e
f
o
r
th
e
co
n
ca
ten
ated
m
o
d
el
is
1
5
5
,
w
h
ic
h
in
d
icate
s
th
at
o
u
t
o
f
th
e
1
5
8
p
n
eu
m
o
n
ia
ca
s
es
1
5
5
ar
e
co
r
r
e
ctl
y
cla
s
s
i
f
ied
as
p
n
e
u
m
o
n
ia
a
n
d
3
ar
e
m
is
cla
s
s
i
f
ied
as
n
o
r
m
al.
T
h
e
tr
u
e
n
eg
a
tiv
e
v
alu
e
i
s
1
5
7
,
w
h
ic
h
in
d
icate
s
t
h
at
o
u
t
o
f
t
h
e
1
5
8
n
o
r
m
al
ca
s
es,
1
5
7
a
r
e
co
r
r
ec
tly
clas
s
i
f
ied
as
n
o
r
m
al
w
h
ile
1
in
s
ta
n
ce
is
m
i
s
clas
s
i
f
ied
as p
n
eu
m
o
n
ia.
T
ab
le
3.
P
er
f
o
r
m
a
n
ce
m
etr
ic
s
of
th
e
co
n
ca
te
n
ated
m
o
d
el
M
o
d
e
l
TP
TN
FP
FN
P
e
r
f
o
r
man
c
e
(
%)
M
C
C
A
c
c
u
r
a
c
y
P
r
e
c
i
si
o
n
S
e
n
si
t
i
v
i
t
y
F1
s
c
o
r
e
S
p
e
c
i
f
i
c
i
t
y
In
c
e
p
t
i
o
n
V3
1
4
2
1
5
8
0
16
9
4
.
9
4
1
0
0
8
9
.
8
7
9
4
.
6
7
1
0
0
0
.
9
0
3
4
VGG
-
16
1
4
3
1
5
7
1
15
9
4
.
9
4
9
9
.
3
1
9
0
.
5
1
9
4
.
7
0
9
9
.
3
7
0
.
9
0
2
3
D
e
n
se
N
e
t
-
201
1
3
8
1
5
8
0
20
9
3
.
6
7
1
0
0
8
7
.
3
4
9
3
.
2
4
1
0
0
0
.
8
8
0
5
C
o
n
c
a
t
e
n
a
t
e
d
m
o
d
e
l
1
5
5
1
5
7
1
3
9
8
.
7
3
9
9
.
3
6
9
8
.
1
0
9
8
.
7
3
9
9
.
3
7
0
.
9
7
4
8
3
.
2
.
1
.
Op
t
i
m
izer
s
elec
t
io
n
T
h
e
co
n
ca
ten
ated
m
o
d
el,
d
ev
e
lo
p
ed
by
m
er
g
i
n
g
f
ea
t
u
r
es
f
r
o
m
t
h
r
ee
m
o
d
els,
is
tr
ain
ed
u
s
i
n
g
th
e
A
d
a
m
o
p
tim
izer
.
To
id
en
tify
th
e
m
o
s
t
ef
f
ec
ti
v
e
o
p
ti
m
izatio
n
ap
p
r
o
ac
h
,
th
e
m
o
d
el
is
also
ass
ess
ed
w
i
th
alter
n
ati
v
e
alg
o
r
ith
m
s
,
i
n
cl
u
d
in
g
R
MSP
r
o
p
,
s
to
ch
asti
c
g
r
ad
ien
t
d
escen
t
(
SGD
)
,
A
d
ad
elta,
an
d
A
d
ag
r
ad
.
R
esu
lt
s
in
T
ab
le
4
r
ev
ea
l
th
at
t
h
e
m
o
d
el
ex
ce
ls
w
it
h
A
d
a
m
o
p
ti
m
izatio
n
co
m
p
ar
ed
to
o
th
er
alg
o
r
ith
m
s
.
T
h
r
o
u
g
h
o
u
t
t
h
e
en
tire
f
i
n
e
-
t
u
n
i
n
g
p
r
o
ce
s
s
,
t
h
e
A
d
a
m
o
p
tim
izatio
n
al
g
o
r
ith
m
co
n
s
is
t
en
tl
y
m
a
in
tai
n
ed
a
h
i
g
h
ac
cu
r
a
c
y
of
9
8
.
7
3
%
w
it
h
th
e
p
ed
iatr
ic
p
n
eu
m
o
n
ia
d
atas
et.
3
.
2
.
2
.
O
pti
m
a
l
lea
rning
ra
t
e
s
elec
t
io
n
T
h
e
m
o
d
e
l
is
in
it
i
al
ly
t
r
ain
e
d
w
i
t
h
th
e
A
d
am
o
p
t
im
iz
e
r
'
s
d
ef
au
lt
l
e
a
r
n
in
g
r
at
e
of
0
.
0
0
1
.
E
x
p
e
r
im
en
t
at
i
o
n
w
ith
d
if
f
e
r
en
t
l
e
a
r
n
in
g
r
a
t
es
,
in
c
lu
d
in
g
1e
-
2,
1e
-
4,
a
n
d
1e
-
5,
is
c
o
n
d
u
c
t
e
d
.
T
a
b
l
e
4
p
r
e
s
en
ts
th
e
c
la
s
s
if
i
ca
t
i
o
n
r
e
s
u
l
ts
,
in
d
i
c
at
in
g
im
p
r
o
v
e
d
a
cc
u
r
a
cy
at
a
l
e
a
r
n
in
g
r
at
e
of
1e
-
4.
C
o
n
s
e
q
u
en
t
ly
,
t
h
e
l
e
a
r
n
in
g
r
a
t
e
is
f
ix
e
d
at
1e
-
4
f
o
r
f
u
r
th
e
r
m
o
d
e
l
t
u
n
i
n
g
.
T
h
e
c
o
n
c
at
en
a
t
e
d
m
o
d
el
a
ch
i
ev
e
d
an
ac
cu
r
a
cy
of
9
9
.
0
5
%
,
co
r
r
e
c
t
ly
c
l
as
s
if
y
in
g
156
p
n
e
u
m
o
n
i
a
ca
s
es
an
d
157
n
o
r
m
a
l
c
as
es
,
w
ith
o
n
ly
2
an
d
1
m
is
c
l
ass
if
i
c
at
i
o
n
s
,
r
es
p
e
c
tiv
ely
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
A
r
ti
f
I
n
tell
I
SS
N:
2252
-
8938
Hyp
er
-
p
a
r
a
mete
r
s
o
p
timiz
ed
d
ee
p
fea
tu
r
e
co
n
c
a
ten
a
ted
n
etw
o
r
k
fo
r
p
ed
ia
tr
ic
…
(
Ma
r
y
S
h
yn
i H
illa
ry
)
2225
T
ab
le
4.
Hy
p
er
-
p
ar
a
m
eter
o
p
ti
m
izatio
n
of
t
h
e
co
n
ca
ten
a
ted
m
o
d
el
H
y
p
e
r
-
p
a
r
a
me
t
e
r
s
TP
TN
FP
FN
P
e
r
f
o
r
man
c
e
(
%)
M
C
C
A
c
c
u
r
a
c
y
Pr
e
c
i
si
o
n
S
e
n
si
t
i
v
i
t
y
F1
s
c
o
r
e
S
p
e
c
i
f
i
c
i
t
y
O
p
t
i
mi
z
a
t
i
o
n
a
l
g
o
r
i
t
h
m
A
d
a
m
1
5
5
1
5
7
1
3
9
8
.
7
3
9
9
.
3
6
9
8
.
1
0
9
8
.
7
3
9
9
.
3
7
0
.
9
7
4
8
R
M
S
P
r
o
p
1
5
2
1
5
8
0
6
9
8
.
1
0
1
0
0
9
6
.
2
0
9
8
.
0
6
1
0
0
0
.
9
6
2
7
S
G
D
1
5
3
1
5
7
1
5
9
8
.
1
0
9
9
.
3
5
9
6
.
8
4
9
8
.
0
8
9
9
.
3
7
0
.
9
6
2
7
A
d
a
d
e
l
t
a
1
4
2
1
5
8
0
16
94
.
9
4
1
0
0
8
9
.
8
7
9
4
.
6
7
1
0
0
0
.
9
0
3
4
A
d
a
g
r
a
d
1
3
6
1
5
6
2
22
9
2
.
4
1
9
8
.
5
5
8
6
.
0
8
9
1
.
8
9
9
8
.
7
3
0
.
8
5
4
9
L
e
a
r
n
i
n
g
r
a
t
e
0
.
0
1
1
5
2
1
5
8
0
6
9
8
.
1
0
1
0
0
9
6
.
2
0
9
8
.
0
6
1
0
0
0
.
9
6
2
7
0
.
0
0
1
1
5
5
1
5
7
1
3
9
8
.
7
3
9
9
.
3
6
9
8
.
1
0
9
8
.
7
3
9
9
.
3
7
0
.
9
7
4
8
1e
-
4
1
5
6
1
5
7
1
2
9
9
.
0
5
9
9
.
3
6
9
8
.
7
3
9
9
.
0
5
9
9
.
3
7
0
.
9
8
1
0
1e
-
5
1
5
2
1
5
8
0
6
9
8
.
1
0
1
0
0
9
6
.
2
0
9
8
.
0
6
1
0
0
0
.
9
6
2
7
B
a
t
c
h
si
z
e
16
1
5
2
1
5
8
0
6
9
8
.
1
0
1
0
0
9
6
.
2
0
9
8
.
0
6
1
0
0
0
.
9
6
2
7
25
1
5
3
1
5
8
0
5
9
8
.
4
2
1
0
0
9
6
.
8
4
9
8
.
3
9
1
0
0
0
.
9
6
8
8
30
1
5
1
1
5
8
0
7
9
7
.
7
8
1
0
0
9
5
.
5
7
9
7
.
7
3
1
0
0
0
.
9
5
6
6
32
1
5
6
1
5
7
1
2
9
9
.
0
5
9
9
.
3
6
98
.
7
3
9
9
.
0
5
9
9
.
3
7
0
.
9
8
1
0
35
1
5
3
1
5
8
0
5
9
8
.
4
2
1
0
0
9
6
.
8
4
9
8
.
3
9
1
0
0
0
.
9
6
8
8
40
1
5
1
1
5
8
0
7
9
7
.
7
8
1
0
0
9
5
.
5
7
9
7
.
7
3
1
0
0
0
.
9
5
6
6
64
1
5
1
1
5
8
0
7
9
7
.
7
8
1
0
0
9
5
.
5
7
9
7
.
7
3
1
0
0
0
.
9
5
6
6
M
o
me
n
t
u
m
β
1
=
0
.
9
β
2
=
0
.
9
9
9
1
5
6
1
5
7
1
2
9
9
.
0
5
9
9
.
3
6
9
8
.
7
3
9
9
.
0
5
9
9
.
3
7
0
.
9
8
1
0
β
1
=
0
.
9
2
β
2
=
0
.
9
9
9
1
5
1
1
5
8
0
7
9
7
.
7
8
1
0
0
9
5
.
5
7
9
7
.
7
3
1
0
0
0
.
9
5
6
6
β
1
=
0
.
9
5
β
2
=
0
.
9
9
9
1
5
2
1
5
8
0
6
9
8
.
1
0
1
0
0
9
6
.
2
0
9
8
.
0
6
1
0
0
0
.
9
6
2
7
β
1
=
0
.
9
9
β
2
=
0
.
9
9
9
1
5
3
1
5
8
0
5
9
8
.
4
2
1
0
0
9
6
.
8
4
9
8
.
3
9
1
0
0
0
.
9
6
8
8
β
1
=
0
.
9
9
2
β
2
=
0
.
9
9
9
1
5
3
1
5
8
0
5
9
8
.
4
2
1
0
0
9
6
.
8
4
9
8
.
3
9
1
0
0
0
.
9
6
8
8
β
1
=
0
.
999
β
2
=
0
.
9
9
9
1
5
1
1
5
8
0
7
9
7
.
7
8
1
0
0
9
5
.
5
7
9
7
.
7
3
1
0
0
0
.
9
5
6
6
β
1
=
0
.
8
8
β
2
=
0
.
9
9
9
1
5
1
1
5
8
0
7
9
7
.
7
8
1
0
0
9
5
.
5
7
9
7
.
7
3
1
0
0
0
.
9
5
6
6
β
1
=
0
.
8
β
2
=
0
.
9
9
9
1
5
3
1
5
8
0
5
9
8
.
4
2
1
0
0
9
6
.
8
4
9
8
.
3
9
1
0
0
0
.
9
6
8
8
β
1
=
0
.
7
β
2
=
0
.
9
9
9
1
5
3
1
5
8
0
5
9
8
.
4
2
1
0
0
9
6
.
8
4
9
8
.
3
9
1
0
0
0
.
9
6
8
8
β
1
=
0
.
9
β
2
=
0
.
9
9
7
1
5
4
1
5
8
0
4
9
8
.
7
3
1
0
0
9
7
.
4
7
9
8
.
7
2
1
0
0
0
.
9
7
5
0
β
1
=
0
.
9
β
2
=
0
.
9
9
1
5
6
1
5
8
0
2
9
9
.
3
7
1
0
0
9
8
.
7
3
9
9
.
3
6
1
0
0
0
.
9
8
7
4
β
1
=
0
.
9
β
2
=
0
.
9
9
2
1
5
3
1
5
8
0
5
9
8
.
4
2
1
0
0
9
6
.
8
4
9
8
.
3
9
1
0
0
0
.
9
6
8
8
β
1
=
0
.
9
β
2
=
0
.
9
7
1
5
4
1
5
8
0
4
9
8
.
7
3
1
0
0
9
7
.
4
7
9
8
.
7
2
1
0
0
0
.
9
7
5
0
β
1
=
0
.
9
β
2
=
0
.
9
1
5
4
1
5
7
1
4
9
8
.
4
2
9
9
.
3
5
9
7
.
4
7
9
8
.
4
0
9
9
.
3
7
0
.
9
6
8
5
β
1
=
0
.
9
β
2
=
0
.
9
2
1
5
2
1
5
8
0
6
9
8
.
1
0
1
0
0
9
6
.
2
0
9
8
.
0
6
1
0
0
0
.
9
6
2
7
W
e
i
g
h
t
d
e
c
a
y
0
1
5
6
1
5
8
0
2
9
9
.
3
7
1
0
0
9
8
.
7
3
9
9
.
3
6
1
0
0
0
.
9
8
7
4
0
.
1
1
3
2
1
5
8
0
26
9
1
.
7
7
1
0
0
8
3
.
5
4
9
1
.
0
3
1
0
0
0
.
8
4
7
0
1e
-
2
1
4
7
1
5
8
0
11
9
6
.
5
2
1
0
0
9
3
.
0
4
9
6
.
3
9
1
0
0
0
.
9
3
2
6
1e
-
3
1
5
2
1
5
8
0
6
9
8
.
1
0
1
0
0
9
6
.
2
0
9
8
.
0
6
1
0
0
0
.
9
6
2
7
1e
-
4
1
5
2
1
5
8
0
6
9
8
.
1
0
1
0
0
9
6
.
2
0
9
8
.
0
6
1
0
0
0
.
9
6
2
7
1e
-
5
1
5
1
1
5
8
0
7
9
7
.
7
8
1
0
0
9
5
.
5
7
9
7
.
7
3
1
0
0
0
.
9
5
6
6
1e
-
6
1
5
1
1
5
8
0
7
9
7
.
7
8
1
0
0
9
5
.
5
7
9
7
.
7
3
1
0
0
0
.
9
5
6
6
Ep
si
l
o
n
1e
-
7
1
5
6
1
5
8
0
2
9
9
.
3
7
1
0
0
9
8
.
7
3
9
9
.
3
6
1
0
0
0
.
9
8
7
4
1e
-
8
1
5
2
1
5
7
1
6
9
7
.
7
8
9
9
.
3
5
9
6
.
2
0
9
7
.
7
5
9
9
.
3
7
0
.
9
5
6
2
1e
-
6
1
5
1
1
5
8
0
7
9
7
.
7
8
1
0
0
9
5
.
5
7
9
7
.
7
3
1
0
0
0
.
9
5
6
6
1e
-
3
1
5
1
1
5
8
0
7
9
7
.
7
8
1
0
0
9
5
.
5
7
9
7
.
7
3
1
0
0
0
.
9
5
6
6
0
.
1
1
5
1
1
5
8
0
7
9
7
.
7
8
1
0
0
9
5
.
5
7
9
7
.
7
3
1
0
0
0
.
9
5
6
6
1
1
4
7
1
5
8
0
11
9
6
.
5
2
1
0
0
9
3
.
0
4
9
6
.
3
9
1
0
0
0
.
9
3
2
6
3
.
2
.
3
.
O
pti
m
a
l
ba
t
ch
s
ize
s
el
ec
t
io
n
T
h
e
m
o
d
el
w
h
ic
h
is
alr
ea
d
y
tr
ain
ed
w
i
th
b
atch
s
ize
32
u
n
d
er
w
e
n
t
ad
d
itio
n
al
tr
ain
i
n
g
w
ith
a
r
an
g
e
of
b
atch
s
izes
to
o
b
tain
th
e
o
p
tim
al
v
al
u
e.
T
h
e
o
u
tco
m
e
s
in
T
ab
le
4
d
em
o
n
s
tr
ate
h
o
w
t
h
e
m
o
d
el
p
er
f
o
r
m
s
w
i
th
v
ar
io
u
s
b
atch
s
izes.
It
ca
n
be
s
ee
n
f
r
o
m
th
e
tab
le
th
a
t
h
ig
h
e
r
p
er
f
o
r
m
a
n
ce
is
o
b
tain
ed
w
i
t
h
a
b
atc
h
s
ize
of
32.
T
h
e
m
o
d
el
is
f
u
r
t
h
er
ad
j
u
s
ted
f
o
r
o
th
er
h
y
p
er
-
p
ar
a
m
eter
s
a
f
t
er
f
ix
i
n
g
th
e
b
atc
h
s
ize
to
32.
3
.
2
.
4
.
O
pti
m
a
l
m
o
m
ent
u
m
s
e
lect
io
n
T
h
e
co
n
ca
ten
ated
m
o
d
el,
in
iti
all
y
tr
ain
ed
w
it
h
t
h
e
d
ef
a
u
lt
m
o
m
e
n
t
u
m
v
al
u
e
s
of
th
e
A
d
a
m
o
p
tim
izer
β1
=0
.
9
an
d
β2
=0
.
9
9
9
,
is
f
u
r
t
h
er
ev
alu
ated
w
it
h
v
ar
io
u
s
m
o
m
e
n
t
u
m
v
al
u
es.
T
ab
le
4
d
is
p
la
y
s
th
e
m
o
d
el
's
p
er
f
o
r
m
a
n
ce
,
i
n
d
icati
n
g
th
a
t
t
h
e
co
m
b
in
atio
n
of
β1
=0
.
9
an
d
β
2
=0
.
9
9
ac
h
iev
ed
th
e
h
i
g
h
e
s
t
a
cc
u
r
ac
y
at
9
9
.
3
7
%.
W
ith
100%
p
r
ec
is
io
n
an
d
s
p
ec
if
icit
y
,
all
n
o
r
m
al
ca
s
es
ar
e
c
o
r
r
ec
tly
clas
s
i
f
ied
,
w
h
ile
2
o
u
t
of
158
p
n
eu
m
o
n
ia
ca
s
es
ar
e
m
is
c
lass
if
ied
as
n
o
r
m
al.
T
h
ese
m
o
m
e
n
t
u
m
v
al
u
es
β1
=0
.
9
an
d
β2
=0
.
9
9
ar
e
t
h
en
f
r
o
ze
n
b
ef
o
r
e
p
r
o
ce
ed
in
g
to
th
e
n
e
x
t
t
u
n
i
n
g
s
tag
e
s
.
3
.
2
.
5
.
O
pti
m
a
l
w
eig
ht
deca
y
s
elec
t
io
n
T
h
e
m
o
d
el
w
h
ich
is
in
i
tiall
y
tr
ain
ed
w
it
h
th
e
d
ef
a
u
lt
w
e
ig
h
t
d
ec
a
y
of
0
is
f
u
r
th
er
tr
ain
ed
w
it
h
d
if
f
er
e
n
t
v
alu
e
s
to
g
et
t
h
e
o
p
ti
m
a
l
o
n
e.
T
h
e
class
i
f
icatio
n
r
esu
lts
of
th
e
co
n
ca
te
n
ated
m
o
d
el
w
it
h
d
if
f
er
e
n
t
v
al
u
es
of
w
ei
g
h
t
d
ec
a
y
ar
e
s
h
o
w
n
in
T
a
b
le
4.
Ho
w
e
v
er
,
th
e
p
er
f
o
r
m
a
n
ce
of
t
h
e
m
o
d
el
h
a
s
s
h
o
w
n
no
im
p
r
o
v
e
m
e
n
t
w
it
h
v
alu
e
s
g
r
ea
ter
t
h
an
0
an
d
t
h
er
ef
o
r
e
we
h
av
e
co
n
s
id
er
ed
u
s
i
n
g
th
e
d
ef
a
u
lt
w
ei
g
h
t
d
ec
a
y
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
I
n
t J
A
r
ti
f
I
n
tell
,
Vo
l.
1
4
,
No
.
3
,
J
u
n
e
2
0
2
5
:
2
2
2
0
-
2228
2226
3
.
2
.
6
.
O
pti
m
a
l
e
ps
i
lo
n
s
elec
t
i
o
n
T
h
e
m
o
d
el
is
tr
ai
n
ed
w
it
h
d
i
f
f
er
en
t
v
a
lu
e
s
of
ep
s
ilo
n
w
h
o
s
e
p
er
f
o
r
m
a
n
ce
is
p
r
ese
n
ted
in
T
ab
le
4.
T
h
e
m
o
d
el
h
a
s
s
h
o
w
n
b
etter
p
er
f
o
r
m
a
n
ce
w
ith
t
h
e
d
ef
au
l
t
ep
s
ilo
n
v
al
u
e
1e
-
7
in
co
m
p
ar
is
o
n
w
i
th
o
th
er
v
al
u
es.
Hen
ce
th
e
d
e
f
au
lt
ep
s
ilo
n
v
al
u
e
ha
s
b
ee
n
c
h
o
s
en
f
o
r
th
e
m
o
d
el.
3
.
2
.
7
.
F
ine
-
t
un
ing
t
he
lea
rning
ra
t
e
Af
ter
t
u
n
i
n
g
all
h
y
p
er
-
p
ar
a
m
et
er
s
,
th
e
f
in
al
ad
j
u
s
t
m
e
n
t
of
t
h
e
lear
n
in
g
r
ate
s
i
g
n
if
ican
t
l
y
i
m
p
r
o
v
ed
th
e
m
o
d
el
's
ac
cu
r
ac
y
to
9
9
.
6
8
%.
W
ith
a
lear
n
in
g
r
ate
of
2e
-
4
(
0
.
0
0
0
2
)
,
th
e
m
o
d
el
co
r
r
ec
t
ly
class
i
f
ies
al
l
n
o
r
m
a
l
ca
s
es
an
d
m
is
c
lass
if
ies
o
n
l
y
o
n
e
p
n
e
u
m
o
n
ia
ca
s
e
as
s
h
o
w
n
in
T
ab
le
5.
T
h
e
o
p
tim
al
h
y
p
er
-
p
ar
am
eter
s
o
b
tain
ed
ar
e:
l
ea
r
n
in
g
r
ate=
2
e
-
4,
b
atch
s
ize=
3
2
,
m
o
m
e
n
tu
m
:
β1
=0
.
9
an
d
β2
=0
.
9
9
,
w
ei
g
h
t
d
ec
a
y
=0
,
an
d
ep
s
ilo
n
=1
e
-
7.
T
h
e
co
n
f
u
s
io
n
m
atr
i
x
an
d
r
ec
eiv
er
o
p
er
atin
g
c
h
ar
ac
ter
is
tic
(
R
OC
)
c
u
r
v
e
of
t
h
e
p
r
o
p
o
s
ed
m
o
d
el
ar
e
p
r
esen
ted
in
Fi
g
u
r
e
s
3
(
a)
an
d
3
(
b
)
.
T
ab
le
5.
P
er
f
o
r
m
a
n
ce
of
t
h
e
co
n
ca
ten
ated
m
o
d
el
w
it
h
f
in
e
-
t
u
n
ed
lear
n
in
g
r
ate
L
e
a
r
n
i
n
g
r
a
t
e
TP
TN
FP
FN
P
e
r
f
o
r
man
c
e
(
%)
M
C
C
A
c
c
u
r
a
c
y
Pre
c
i
si
o
n
S
e
n
si
t
i
v
i
t
y
F1
s
c
o
r
e
S
p
e
c
i
f
i
c
i
t
y
1e
-
4
1
5
6
1
5
8
0
2
9
9
.
3
7
1
0
0
9
8
.
7
3
9
9
.
3
6
1
0
0
0
.
9
8
7
4
0
.
0
0
8
1
5
6
1
5
7
1
2
9
9
.
0
5
9
9
.
3
6
9
8
.
7
3
9
9
.
0
5
9
9
.
3
7
0
.
9
8
1
0
0
.
0
0
5
1
5
2
1
5
8
0
6
9
8
.
1
0
1
0
0
9
6
.
2
0
9
8
.
0
6
1
0
0
0
.
9
6
2
7
2e
-
4
1
5
7
1
5
8
0
1
9
9
.
6
8
1
0
0
9
9
.
3
7
9
9
.
6
8
1
0
0
0
.
9
9
3
7
3e
-
4
1
5
2
1
5
8
0
6
9
8
.
1
0
1
0
0
9
6
.
2
0
9
8
.
0
6
1
0
0
0
.
9
6
2
7
(
a)
(
b
)
Fig
u
r
e
3
.
C
o
n
f
u
s
io
n
m
atr
ix
a
n
d
R
OC
cu
r
v
e
o
f
t
h
e
p
r
o
p
o
s
ed
m
o
d
el
o
f
(
a)
co
n
f
u
s
io
n
m
atr
i
x
an
d
(
b
)
R
OC
cu
r
v
e
3.
3
.
Rea
l
-
t
i
m
e
t
esting
1
3
r
ea
l
-
ti
m
e
s
a
m
p
le
s
o
b
tain
ed
f
r
o
m
SR
M
Me
d
ical
C
o
lle
g
e
Ho
s
p
it
al
an
d
R
esear
ch
C
en
tr
e
ar
e
u
s
ed
to
ass
es
s
th
e
r
o
b
u
s
t
n
es
s
o
f
t
h
e
m
o
d
el.
T
h
e
d
atab
ase
in
clu
d
e
s
6
p
ed
iatr
ic
p
n
eu
m
o
n
ia
s
p
ec
i
m
en
s
an
d
4
n
o
r
m
al
X
-
r
a
y
s
.
T
h
e
m
o
d
el
ac
cu
r
atel
y
ca
teg
o
r
izes e
v
er
y
s
a
m
p
le,
y
iel
d
in
g
a
1
0
0
% a
cc
u
r
ac
y
r
ate.
4.
CO
NCLU
SI
O
N
P
n
eu
m
o
n
ia
ca
u
s
es
p
le
u
r
al
ef
f
u
s
io
n
th
at
lead
s
to
a
f
atalit
y
r
ate
of
15%
in
ch
ild
r
e
n
b
elo
w
t
h
e
ag
e
of
5.
An
ea
r
l
y
d
iag
n
o
s
i
s
of
th
e
d
i
s
e
ase
an
d
p
r
o
m
p
t
m
ed
ical
i
n
ter
v
en
tio
n
can
l
i
m
it
t
h
e
r
a
m
if
icati
o
n
s
a
n
d
p
o
ten
tiall
y
s
av
e
t
h
e
liv
e
s
of
t
h
o
u
s
a
n
d
s
of
ch
ild
r
en
.
T
h
e
co
n
tr
ib
u
tio
n
of
th
is
r
esear
c
h
is
t
h
e
p
r
o
p
o
s
al
of
a
d
iag
n
o
s
tic
f
r
a
m
e
w
o
r
k
f
o
r
p
ed
iatr
ic
p
n
eu
m
o
n
ia
b
ased
on
co
n
ca
te
n
atio
n
an
d
o
p
ti
m
izatio
n
.
T
h
e
co
n
ca
te
n
atio
n
m
et
h
o
d
o
lo
g
y
e
m
p
lo
y
ed
in
th
i
s
d
etec
tio
n
s
y
s
te
m
in
te
g
r
ates
th
e
p
er
f
o
r
m
a
n
c
e
of
th
r
ee
s
u
cc
ess
f
u
l
p
r
e
-
tr
ain
e
d
m
o
d
els
I
n
ce
p
tio
n
V3
,
VGG
-
16
an
d
De
n
s
eNe
t
-
2
0
1
.
T
h
e
ai
m
is
to
en
h
an
ce
t
h
e
ac
cu
r
ac
y
of
th
e
co
n
ca
ten
a
ted
m
o
d
el
by
o
p
ti
m
izi
n
g
th
e
h
y
p
er
-
p
ar
a
m
eter
s
to
p
r
o
v
id
e
m
o
r
e
p
r
ec
is
e
esti
m
a
tes.
A
s
tep
-
by
-
s
tep
o
p
tim
izatio
n
is
ca
r
r
ied
o
u
t
w
i
th
t
h
e
A
d
a
m
o
p
ti
m
izer
an
d
th
e
f
o
llo
w
i
n
g
o
p
tim
a
l
h
y
p
er
-
p
ar
a
m
eter
s
ar
e
o
b
tain
ed
:
l
ea
r
n
in
g
r
ate=
2
e
-
4,
b
atch
s
ize=
3
2
,
m
o
m
e
n
t
u
m
:
β1
=0
.
9
an
d
β2
=0
.
9
9
,
w
ei
g
h
t
d
ec
a
y
=0
an
d
ep
s
i
lo
n
=1
e
-
7.
T
h
e
p
r
o
p
o
s
ed
f
in
e
-
t
u
n
ed
co
n
ca
te
n
ated
m
o
d
el
o
u
tp
er
f
o
r
m
ed
o
t
h
er
ex
is
ti
n
g
m
o
d
els
w
i
th
an
ac
c
u
r
a
c
y
of
9
9
.
6
8
%.
We
h
av
e
r
ea
c
h
ed
an
i
m
p
o
r
tan
t
co
n
clu
s
io
n
t
h
at
h
y
p
er
-
p
ar
a
m
e
ter
o
p
tim
izat
io
n
is
an
ess
e
n
ti
al
p
r
o
ce
d
u
re
to
o
b
tain
th
e
b
est
r
esu
l
ts
f
r
o
m
th
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
A
r
ti
f
I
n
tell
I
SS
N:
2252
-
8938
Hyp
er
-
p
a
r
a
mete
r
s
o
p
timiz
ed
d
ee
p
fea
tu
r
e
co
n
c
a
ten
a
ted
n
etw
o
r
k
fo
r
p
ed
ia
tr
ic
…
(
Ma
r
y
S
h
yn
i H
illa
ry
)
2227
m
o
d
el.
Ho
w
e
v
er
,
t
h
e
s
u
g
g
e
s
te
d
m
o
d
el
en
tail
s
co
n
s
id
er
ab
le
co
m
p
u
tatio
n
al
e
x
p
en
s
e
s
d
u
e
to
th
e
co
n
ca
te
n
atio
n
an
d
m
a
n
u
al
o
p
ti
m
iza
tio
n
m
et
h
o
d
s
e
m
p
lo
y
ed
.
In
t
h
e
f
u
tu
r
e,
we
w
o
u
ld
ex
te
n
d
o
u
r
r
esear
c
h
to
d
ev
elo
p
an
o
p
ti
m
al
m
o
d
e
l
f
o
r
m
u
lticla
s
s
l
u
n
g
d
is
e
ase
class
i
f
icat
io
n
.
F
UNDIN
G
I
NF
O
RM
AT
I
O
N
T
h
e
au
th
o
r
s
d
ec
lar
e
th
at
n
o
f
u
n
d
in
g
w
as r
ec
ei
v
ed
f
o
r
co
n
d
u
c
tin
g
th
i
s
r
esear
ch
.
AUTHO
R
CO
NT
RIB
UT
I
O
NS ST
A
T
E
M
E
NT
T
h
is
j
o
u
r
n
al
u
s
e
s
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
i
v
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
t
h
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
lla
b
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
Ma
r
y
Sh
y
n
i H
illar
y
C
h
i
tr
a
E
k
a
m
b
ar
a
m
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
si
s
I
:
I
n
v
e
st
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
si
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
Au
t
h
o
r
s
s
tate
n
o
co
n
f
lic
t o
f
i
n
t
er
est.
I
NF
O
RM
E
D
CO
NSE
N
T
W
e
h
av
e
o
b
tain
ed
in
f
o
r
m
ed
c
o
n
s
en
t f
r
o
m
al
l in
d
i
v
id
u
al
s
in
c
lu
d
ed
in
th
is
s
t
u
d
y
.
DATA AV
AI
L
AB
I
L
I
T
Y
D
a
t
a
av
a
il
a
b
i
li
ty
is
n
o
t
a
p
p
l
i
ca
b
l
e
t
o
th
is
p
a
p
e
r
a
s
n
o
n
ew
d
a
t
a
w
e
r
e
c
r
e
a
t
e
d
o
r
an
aly
ze
d
in
th
is
s
t
u
d
y
.
RE
F
E
R
E
NC
E
S
[
1
]
C.
J.
A
b
e
j
a
,
V.
N
i
y
o
n
z
i
ma,
J.
P.
B
y
a
g
a
my
,
a
n
d
C.
O
b
u
a
,
“
A
n
t
i
b
i
o
t
i
c
p
r
e
scri
p
t
i
o
n
r
a
t
i
o
n
a
l
i
t
y
a
n
d
a
s
so
c
i
a
t
e
d
in
-
p
a
t
i
e
n
t
t
r
e
a
t
me
n
t
o
u
t
c
o
me
s
in
c
h
i
l
d
r
e
n
u
n
d
e
r
-
f
i
v
e
w
i
t
h
se
v
e
r
e
p
n
e
u
mo
n
i
a
at
B
w
i
z
i
b
w
e
r
a
H
e
a
l
t
h
C
e
n
t
e
r
I
V
,
M
b
a
r
a
r
a
D
i
st
r
i
c
t
,
S
o
u
t
h
-
W
e
st
e
r
n
U
g
a
n
d
a
,
”
P
n
e
u
m
o
n
i
a
,
v
o
l
.
1
4
,
n
o
.
1,
2
0
2
2
,
d
o
i
:
1
0
.
1
1
8
6
/
s4
1
4
7
9
-
0
2
2
-
0
0
0
9
5
-
0
.
[
2
]
G.
L
a
b
h
a
n
e
,
R.
P
a
n
sare
,
S.
M
a
h
e
sh
w
a
r
i
,
R.
T
i
w
a
r
i
,
a
n
d
A.
S
h
u
k
l
a
,
“
D
e
t
e
c
t
i
o
n
of
p
e
d
i
a
t
r
i
c
p
n
e
u
mo
n
i
a
f
r
o
m
c
h
e
s
t
X
-
r
a
y
i
mag
e
s
u
si
n
g
C
N
N
a
n
d
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
,
”
2
0
2
0
3
r
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
Em
e
r
g
i
n
g
T
e
c
h
n
o
l
o
g
i
e
s
i
n
C
o
m
p
u
t
e
r
En
g
i
n
e
e
ri
n
g
:
Ma
c
h
i
n
e
L
e
a
r
n
i
n
g
a
n
d
I
n
t
e
rn
e
t
o
f
T
h
i
n
g
s
(
I
C
ET
C
E)
,
J
a
i
p
u
r
,
I
n
d
i
a
,
2
0
2
0
,
p
p
.
8
5
-
9
2
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
ET
C
E4
8
1
9
9
.
2
0
2
0
.
9
0
9
1
7
5
5
.
[
3
]
E.
T.
S
a
l
a
h
,
S.
H.
A
l
g
a
si
m,
A.
S.
M
h
a
mo
u
d
,
a
n
d
N.
E.
O.
S.
A.
H
u
si
a
n
,
“
P
r
e
v
a
l
e
n
c
e
of
h
y
p
o
x
e
mi
a
in
u
n
d
e
r
-
f
i
v
e
c
h
i
l
d
r
e
n
w
i
t
h
p
n
e
u
mo
n
i
a
in
an
e
me
r
g
e
n
c
y
p
e
d
i
a
t
r
i
c
s
h
o
sp
i
t
a
l
in
S
u
d
a
n
,
”
I
n
d
i
a
n
J
o
u
r
n
a
l
o
f
C
ri
t
i
c
a
l
C
a
re
Me
d
i
c
i
n
e
,
v
o
l
.
1
9
,
n
o
.
4,
p
p
.
2
0
3
–
2
0
7
,
2
0
1
5
,
d
o
i
:
1
0
.
4
1
0
3
/
0
9
7
2
-
5
2
2
9
.
1
5
4
5
4
9
.
[
4
]
D
.
K
.
S
mi
t
h
,
D
.
P
.
K
u
c
k
e
l
,
a
n
d
A
.
M
.
R
e
c
i
d
o
r
o
,
“
C
o
mm
u
n
i
t
y
-
a
c
q
u
i
r
e
d
p
n
e
u
mo
n
i
a
i
n
c
h
i
l
d
r
e
n
:
r
a
p
i
d
e
v
i
d
e
n
c
e
r
e
v
i
e
w
,
”
Am
e
r
i
c
a
n
Fa
m
i
l
y
Ph
y
si
c
i
a
n
,
v
o
l
.
104
,
n
o
.
6
,
p
p
.
6
1
8
-
625
,
2
0
2
1
.
[
5
]
P.
D
e
a
n
a
n
d
T.
A.
F
l
o
r
i
n
,
“
F
a
c
t
o
r
s
a
ss
o
c
i
a
t
e
d
w
i
t
h
p
n
e
u
mo
n
i
a
se
v
e
r
i
t
y
in
c
h
i
l
d
r
e
n
:
a
sy
st
e
mat
i
c
r
e
v
i
e
w
,
”
J
o
u
r
n
a
l
o
f
t
h
e
P
e
d
i
a
t
ri
c
I
n
f
e
c
t
i
o
u
s D
i
se
a
ses
S
o
c
i
e
t
y
,
v
o
l
.
7,
n
o
.
4,
p
p
.
3
2
3
–
3
3
4
,
2
0
1
8
,
d
o
i
:
1
0
.
1
0
9
3
/
j
p
i
d
s/
p
i
y
0
4
6
.
[
6
]
T.
P.
H
t
u
n
,
Y.
S
u
n
,
H.
L.
C
h
u
a
,
a
n
d
J.
P
a
n
g
,
“
C
l
i
n
i
c
a
l
f
e
a
t
u
r
e
s
f
o
r
d
i
a
g
n
o
si
s
of
p
n
e
u
mo
n
i
a
a
mo
n
g
a
d
u
l
t
s
in
p
r
i
ma
r
y
c
a
r
e
se
t
t
i
n
g
:
A
s
y
st
e
mat
i
c
a
n
d
me
t
a
-
r
e
v
i
e
w
,
”
S
c
i
e
n
t
i
f
i
c
Re
p
o
rt
s
,
v
o
l
.
9,
n
o
.
1,
2
0
1
9
,
d
o
i
:
1
0
.
1
0
3
8
/
s
4
1
5
9
8
-
0
1
9
-
4
4
1
4
5
-
y
.
[
7
]
S.
S
i
n
g
h
,
“
Ef
f
i
c
i
e
n
t
p
n
e
u
mo
n
i
a
d
e
t
e
c
t
i
o
n
u
si
n
g
v
i
si
o
n
t
r
a
n
sf
o
r
me
r
s
on
c
h
e
st
X
-
r
a
y
s,
”
S
c
i
e
n
t
i
f
i
c
Re
p
o
r
t
s
,
v
o
l
.
14,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
3
8
/
s
4
1
5
9
8
-
024
-
5
2
7
0
3
-
2
.
[
8
]
M.
M
u
j
a
h
i
d
,
F.
R
u
s
t
a
m
,
R.
Á
l
v
a
r
e
z
,
J.
L
.
V
.
M
a
z
ó
n
,
I.
de
la
T.
D
í
e
z
,
a
n
d
I.
A
sh
r
a
f
,
“
P
n
e
u
mo
n
i
a
c
l
a
ssi
f
i
c
a
t
i
o
n
f
r
o
m
X
-
r
a
y
i
mag
e
s
w
i
t
h
I
n
c
e
p
t
i
o
n
-
V3
a
n
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
,
”
D
i
a
g
n
o
s
t
i
c
s
,
v
o
l
.
1
2
,
n
o
.
5,
2
0
2
2
,
d
o
i
:
1
0
.
3
3
9
0
/
d
i
a
g
n
o
st
i
c
s
1
2
0
5
1
2
8
0
.
[
9
]
E.
Ç
a
l
l
ı
,
E.
S
o
g
a
n
c
i
o
g
l
u
,
B.
V.
G
i
n
n
e
k
e
n
,
K.
G.
V.
L
e
e
u
w
e
n
,
a
n
d
K.
M
u
r
p
h
y
,
“
D
e
e
p
l
e
a
r
n
i
n
g
f
o
r
c
h
e
st
X
-
r
a
y
a
n
a
l
y
si
s:
A
su
r
v
e
y
,
”
Me
d
i
c
a
l
I
m
a
g
e
A
n
a
l
y
s
i
s
,
v
o
l
.
7
2
,
2021
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
me
d
i
a
.
2
0
2
1
.
1
0
2
1
2
5
.
[
1
0
]
S.
I
q
b
a
l
,
A.
N
.
Q
u
r
e
sh
i
,
J.
L
i
,
a
n
d
T.
M
a
h
mo
o
d
,
“
On
th
e
a
n
a
l
y
se
s
of
me
d
i
c
a
l
i
m
a
g
e
s
u
si
n
g
t
r
a
d
i
t
i
o
n
a
l
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s
a
n
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,
”
Arc
h
i
v
e
s
o
f
C
o
m
p
u
t
a
t
i
o
n
a
l
M
e
t
h
o
d
s
i
n
E
n
g
i
n
e
e
r
i
n
g
,
v
o
l
.
30,
n
o
.
5,
p
p
.
3
1
7
3
–
3
2
3
3
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
1
8
3
1
-
023
-
0
9
8
9
9
-
9
.
[
1
1
]
J.
V.
S.
D.
C
h
a
g
a
s,
D.
D.
A
R
o
d
r
i
g
u
e
s,
R.
F.
I
v
o
,
M.
M.
H
a
ss
a
n
,
V.
H.
C.
de
A
l
b
u
q
u
e
r
q
u
e
,
a
n
d
P.
P.
R.
F
i
l
h
o
,
“
A
n
e
w
a
p
p
r
o
a
c
h
f
o
r
t
h
e
d
e
t
e
c
t
i
o
n
of
p
n
e
u
mo
n
i
a
in
c
h
i
l
d
r
e
n
u
s
i
n
g
C
X
R
i
mag
e
s
b
a
se
d
on
an
r
e
a
l
-
t
i
me
I
o
T
sy
st
e
m,
”
J
o
u
r
n
a
l
o
f
R
e
a
l
-
T
i
m
e
I
m
a
g
e
Pro
c
e
ssi
n
g
,
v
o
l
.
1
8
,
n
o
.
4,
p
p.
1
0
9
9
–
1
1
1
4
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
1
5
5
4
-
021
-
0
1
0
8
6
-
y
.
[
1
2
]
S.
N
a
l
l
u
r
i
a
n
d
R.
S
a
si
k
a
l
a
,
“
P
n
e
u
m
o
n
i
a
scre
e
n
i
n
g
on
c
h
e
st
X
-
r
a
y
s
w
i
t
h
o
p
t
i
mi
z
e
d
e
n
se
mb
l
e
mo
d
e
l
,
”
E
x
p
e
r
t
S
y
st
e
m
s
w
i
t
h
Ap
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
2
4
2
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
e
sw
a
.
2
0
2
3
.
1
2
2
7
0
5
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
I
n
t J
A
r
ti
f
I
n
tell
,
Vo
l.
1
4
,
No
.
3
,
J
u
n
e
2
0
2
5
:
2
2
2
0
-
2228
2228
[
1
3
]
V.
F
e
r
n
a
n
d
e
s,
G.
B.
J
u
n
i
o
r
,
A.
C.
de
P
a
i
v
a
,
A.
C.
S
i
l
v
a
,
a
n
d
M.
G
a
t
t
a
ss
,
“
B
a
y
e
si
a
n
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
e
st
i
mat
i
o
n
f
o
r
p
e
d
i
a
t
r
i
c
p
n
e
u
mo
n
i
a
d
e
t
e
c
t
i
o
n
a
n
d
d
i
a
g
n
o
si
s
,
”
C
o
m
p
u
t
e
r
Me
t
h
o
d
s
a
n
d
Pro
g
r
a
m
s
i
n
B
i
o
m
e
d
i
c
i
n
e
,
v
o
l
.
2
0
8
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
mp
b
.
2
0
2
1
.
1
0
6
2
5
9
.
[
1
4
]
I.
M
o
h
a
mm
e
d
,
N.
S
i
n
g
h
,
a
n
d
M.
V
e
n
k
a
t
a
su
b
r
a
m
a
n
i
a
n
,
C
o
m
p
u
t
e
r
-
a
ssi
s
t
e
d
d
e
t
e
c
t
i
o
n
a
n
d
d
i
a
g
n
o
s
i
s
o
f
p
e
d
i
a
t
r
i
c
p
n
e
u
m
o
n
i
a
i
n
c
h
e
st
x
-
ra
y
i
m
a
g
e
s
,
F
r
i
d
a
y
H
a
r
b
o
r
,
U
S
A
:
P
a
t
t
e
r
n
C
o
mp
u
t
e
r
2
0
1
9
.
[
1
5
]
D.
S.
K
e
r
man
y
et
a
l
.
,
“
I
d
e
n
t
i
f
y
i
n
g
me
d
i
c
a
l
d
i
a
g
n
o
se
s
a
n
d
t
r
e
a
t
a
b
l
e
d
i
se
a
se
s
by
i
mag
e
-
b
a
se
d
d
e
e
p
l
e
a
r
n
i
n
g
,
”
C
e
l
l
,
v
o
l
.
1
7
2
,
n
o
.
5,
p
p
.
1
1
2
2
-
1
1
3
1
,
2
0
1
8
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
e
l
l
.
2
0
1
8
.
0
2
.
0
1
0
.
[
1
6
]
A.
A
t
h
a
r
,
“
I
mp
r
o
v
i
n
g
p
n
e
u
mo
n
i
a
d
e
t
e
c
t
i
o
n
i
n
c
h
e
st
x
-
r
a
y
s
u
si
n
g
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
(
A
l
e
x
N
e
t
)
a
n
d
a
d
v
e
r
sari
a
l
t
r
a
i
n
i
n
g,
”
2
0
2
3
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
Bu
s
i
n
e
ss
A
n
a
l
y
t
i
c
s
f
o
r
T
e
c
h
n
o
l
o
g
y
a
n
d
S
e
c
u
r
i
t
y
(
I
C
BAT
S
)
,
D
u
b
a
i
,
U
n
i
t
e
d
A
r
a
b
Emi
r
a
t
e
s,
2
0
2
3
,
p
p
.
1
-
7
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
B
A
T
S
5
7
7
9
2
.
2
0
2
3
.
1
0
1
1
1
1
9
3
.
[
1
7
]
R.
K
u
n
d
u
,
R.
D
a
s,
Z.
W.
G
e
e
m,
G.
-
T.
H
a
n
,
a
n
d
R.
S
a
r
k
a
r
,
“
P
n
e
u
mo
n
i
a
d
e
t
e
c
t
i
o
n
in
ch
e
st
X
-
r
a
y
i
mag
e
s
u
si
n
g
an
e
n
se
mb
l
e
of
d
e
e
p
l
e
a
r
n
i
n
g
mo
d
e
l
s,
”
PL
o
S
O
n
e
,
v
o
l
.
1
6
,
n
o
.
9,
2
0
2
1
,
d
o
i
:
1
0
.
1
3
7
1
/
j
o
u
r
n
a
l
.
p
o
n
e
.
0
2
5
6
6
3
0
.
[
1
8
]
R.
Y
i
,
L.
T
a
n
g
,
Y.
T
i
a
n
,
J.
L
i
u
,
a
n
d
Z.
W
u
,
“
I
d
e
n
t
i
f
i
c
a
t
i
o
n
a
n
d
c
l
a
s
s
i
f
i
c
a
t
i
o
n
of
p
n
e
u
m
o
n
i
a
d
i
s
e
a
s
e
u
s
i
n
g
a
d
e
e
p
l
e
a
r
n
i
n
g
-
b
a
s
e
d
i
n
t
e
l
l
i
g
e
n
t
c
o
m
p
u
t
a
t
i
o
n
a
l
f
r
a
m
e
w
o
r
k
,
”
N
e
u
r
a
l
C
o
m
p
u
t
i
n
g
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
3
5
,
n
o
.
2
0
,
2023
,
d
o
i
:
1
0
.
1
0
0
7
/
s
0
0
5
2
1
-
021
-
06102
-
7
.
[
1
9
]
B.
A
l
masl
u
k
h
,
“
A
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
se
d
p
n
e
u
mo
n
i
a
d
e
t
e
c
t
i
o
n
a
p
p
r
o
a
c
h
f
o
r
e
n
e
r
g
y
-
e
f
f
i
c
i
e
n
t
me
d
i
c
a
l
sy
st
e
ms,
”
Wi
r
e
l
e
ss
C
o
m
m
u
n
i
c
a
t
i
o
n
s
a
n
d
M
o
b
i
l
e
C
o
m
p
u
t
i
n
g
,
v
o
l
.
2
0
2
1
,
n
o
.
1
.
2
0
2
1
,
d
o
i
:
1
0
.
1
1
5
5
/
2
0
2
1
/
5
5
5
6
6
3
5
.
[
2
0
]
G
.
M
.
M
.
A
l
s
h
m
r
a
n
i
,
Q
.
N
i
,
R
.
J
i
a
n
g
,
H
.
P
e
r
v
a
i
z
,
a
n
d
N
.
M
.
E
l
s
h
e
n
n
a
w
y
,
“
A
d
e
e
p
l
e
a
r
n
i
n
g
a
r
c
h
i
t
e
c
t
u
r
e
f
o
r
m
u
l
t
i
-
c
l
a
s
s
l
u
n
g
d
i
s
e
a
s
e
s
c
l
a
s
s
i
f
i
c
a
t
i
o
n
u
s
i
n
g
c
h
e
s
t
X
-
r
a
y
(
C
X
R
)
i
m
a
g
e
s
,
”
A
l
e
x
a
n
d
r
i
a
E
n
g
i
n
e
e
r
i
n
g
J
o
u
r
n
a
l
,
v
o
l
.
6
4
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
a
e
j
.
2
0
2
2
.
1
0
.
0
5
3
.
[
2
1
]
D.
K
e
r
man
y
,
K.
Z
h
a
n
g
,
a
n
d
M.
G
o
l
d
b
a
u
m,
“
L
a
b
e
l
e
d
o
p
t
i
c
a
l
c
o
h
e
r
e
n
c
e
t
o
mo
g
r
a
p
h
y
(
O
C
T
)
a
n
d
c
h
e
st
x
-
r
a
y
i
mag
e
s
f
o
r
c
l
a
ssi
f
i
c
a
t
i
o
n
,
”
Me
n
d
e
l
e
y
d
a
t
a
,
2
0
1
8
,
d
o
i
:
1
0
.
1
7
6
3
2
/
r
sc
b
j
b
r
9
sj
.
2
.
[
2
2
]
S.
R.
N
a
y
a
k
,
D.
R.
N
a
y
a
k
,
U.
S
i
n
h
a
,
V.
A
r
o
r
a
,
a
n
d
R.
B.
P
a
c
h
o
r
i
,
“
A
p
p
l
i
c
a
t
i
o
n
of
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
d
e
t
e
c
t
i
o
n
of
C
O
V
I
D
-
19
c
a
se
s
u
si
n
g
c
h
e
st
X
-
r
a
y
i
mag
e
s:
a
c
o
mp
r
e
h
e
n
s
i
v
e
st
u
d
y
,
”
Bi
o
m
e
d
i
c
a
l
S
i
g
n
a
l
Pro
c
e
ss
i
n
g
a
n
d
C
o
n
t
r
o
l
,
v
o
l
.
6
4
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
b
s
p
c
.
2
0
2
0
.
1
0
2
3
6
5
.
[
2
3
]
M.
M.
R
a
h
ma
n
a
n
d
D.
N.
D
a
v
i
s,
“
A
d
d
r
e
ssi
n
g
t
h
e
c
l
a
ss
i
m
b
a
l
a
n
c
e
p
r
o
b
l
e
m
in
me
d
i
c
a
l
d
a
t
a
se
t
s,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
Ma
c
h
i
n
e
L
e
a
r
n
i
n
g
a
n
d
C
o
m
p
u
t
i
n
g
,
v
o
l
.
3
2
,
n
o
.
2
,
p
p
.
224
–
2
2
8
,
2
0
1
3
.
[
2
4
]
F
.
Z
h
u
a
n
g
e
t
a
l
.
,
“
A
c
o
mp
r
e
h
e
n
s
i
v
e
s
u
r
v
e
y
on
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
,”
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
I
E
EE,
v
o
l
.
1
0
9
,
n
o
.
1
,
p
p
.
4
3
-
7
6
,
Ja
n
.
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
JP
R
O
C
.
2
0
2
0
.
3
0
0
4
5
5
5
.
[
2
5
]
M.
R
a
h
i
mz
a
d
e
h
a
n
d
A.
A
t
t
a
r
,
“
A
mo
d
i
f
i
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
f
o
r
d
e
t
e
c
t
i
n
g
C
O
V
I
D
-
19
a
n
d
p
n
e
u
m
o
n
i
a
f
r
o
m
c
h
e
st
X
-
r
a
y
i
mag
e
s
b
a
se
d
on
t
h
e
c
o
n
c
a
t
e
n
a
t
i
o
n
of
X
c
e
p
t
i
o
n
a
n
d
R
e
sN
e
t
5
0
V
2
,
”
I
n
f
o
rm
a
t
i
c
s
i
n
Me
d
i
c
i
n
e
U
n
l
o
c
k
e
d
,
v
o
l
.
1
9
,
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
i
mu
.
2
0
2
0
.
1
0
0
3
6
0
.
[
2
6
]
A.
M
o
r
a
l
e
s
-
H
e
r
n
á
n
d
e
z
,
I.
V.
N
i
e
u
w
e
n
h
u
y
s
e
,
a
n
d
S.
R
.
G
o
n
z
a
l
e
z
,
“
A
s
u
r
v
e
y
on
m
u
l
t
i
-
o
b
j
e
c
t
i
v
e
h
y
p
e
r
p
a
r
a
m
e
t
e
r
o
p
t
i
m
i
z
a
t
i
o
n
a
l
g
o
r
i
t
h
m
s
f
o
r
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
,
”
A
r
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
R
e
v
i
e
w
,
v
o
l
.
56,
n
o
.
8,
pp.
8043
–
8
0
9
3
,
2023
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
0
4
6
2
-
022
-
10359
-
2
.
[
2
7
]
S.
Y.
S
e
n
a
n
d
N.
O
z
k
u
r
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
h
y
p
e
r
p
a
r
a
m
e
t
e
r
t
u
n
i
n
g
w
i
t
h
A
d
a
m
o
p
t
i
m
i
z
e
r
f
o
r
E
C
G
c
l
a
s
si
f
i
c
a
t
i
o
n
,
”
2
0
2
0
I
n
n
o
v
a
t
i
o
n
s
i
n
I
n
t
e
l
l
i
g
e
n
t
S
y
s
t
e
m
s
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
C
o
n
f
e
r
e
n
c
e
(
A
S
Y
U
)
,
2
0
2
0
,
p
p
.
1
-
6
,
d
o
i
:
1
0
.
1
1
0
9
/
A
S
Y
U
5
0
7
1
7
.
2
0
2
0
.
9
2
5
9
8
9
6
.
[
2
8
]
S.
Y
a
d
a
v
a
n
d
S.
S
h
u
k
l
a
,
“
A
n
a
l
y
s
i
s
of
k
-
f
o
l
d
c
r
o
s
s
-
v
a
l
i
d
a
t
i
o
n
ov
er
h
o
l
d
-
o
u
t
v
a
l
i
d
a
t
i
o
n
on
c
o
l
o
ss
a
l
d
a
t
a
se
t
s
f
o
r
q
u
a
l
i
t
y
c
l
a
ss
i
f
i
c
a
t
i
o
n
,
”
in
2
0
1
6
I
EE
E
6
t
h
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
A
d
v
a
n
c
e
d
C
o
m
p
u
t
i
n
g
(
I
A
C
C
)
,
2
0
1
6
,
p
p
.
7
8
-
8
3
,
d
o
i
:
1
0
.
1
1
0
9
/
I
A
C
C
.
2
0
1
6
.
2
5
.
[
2
9
]
M.
C.
A
r
e
l
l
a
n
o
a
n
d
O.
E.
R
a
mo
s
,
“
D
e
e
p
l
e
a
r
n
i
n
g
mo
d
e
l
to
i
d
e
n
t
i
fy
C
O
V
I
D
-
19
c
a
se
s
f
r
o
m
c
h
e
st
r
a
d
i
o
g
r
a
p
h
s,
”
2
0
2
0
I
EEE
XXVII
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
ro
n
i
c
s,
El
e
c
t
ri
c
a
l
E
n
g
i
n
e
e
ri
n
g
a
n
d
C
o
m
p
u
t
i
n
g
(
I
N
T
E
RC
O
N
)
,
L
i
ma,
P
e
r
u
,
2
0
2
0
,
p
p
.
1
-
4
,
d
o
i
:
1
0
.
1
1
0
9
/
I
N
T
ER
C
O
N
5
0
3
1
5
.
2
0
2
0
.
9
2
2
0
2
3
7
.
[
3
0
]
D.
C
h
i
c
c
o
a
n
d
G.
Ju
r
ma
n
,
“
A
st
a
t
i
s
t
i
c
a
l
c
o
mp
a
r
i
so
n
b
e
t
w
e
e
n
M
a
t
t
h
e
w
s
c
o
r
r
e
l
a
t
i
o
n
c
o
e
f
f
i
c
i
e
n
t
(
M
C
C
)
,
p
r
e
v
a
l
e
n
c
e
t
h
r
e
sh
o
l
d
,
a
n
d
F
o
w
l
k
e
s
-
M
a
l
l
o
w
s
i
n
d
e
x
,
”
J
o
u
rn
a
l
o
f
Bi
o
m
e
d
i
c
a
l
I
n
f
o
rm
a
t
i
c
s
,
v
o
l
.
1
4
4
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
j
b
i
.
2
0
2
3
.
1
0
4
4
2
6
.
B
I
O
G
RAP
H
I
E
S
OF
AUTH
O
RS
M
a
r
y
S
h
y
n
i
H
il
la
r
y
re
c
e
iv
e
d
h
e
r
B.
E
.
in
e
lec
tro
n
ics
a
n
d
c
o
m
m
u
n
ica
ti
o
n
e
n
g
in
e
e
rin
g
f
ro
m
DMI
Co
ll
e
g
e
of
En
g
in
e
e
rin
g
,
Ch
e
n
n
a
i,
I
n
d
ia
a
n
d
h
e
r
M
.
T
e
c
h
.
d
e
g
re
e
in
las
e
r
a
n
d
e
lec
tro
-
o
p
ti
c
a
l
e
n
g
in
e
e
ri
n
g
f
ro
m
Co
ll
e
g
e
of
En
g
in
e
e
ri
n
g
G
u
in
d
y
,
A
n
n
a
Un
iv
e
rsity
,
Ch
e
n
n
a
i
,
In
d
ia.
S
h
e
is
p
u
rsu
i
n
g
h
e
r
P
h
.
D.
as
a
f
u
ll
-
ti
m
e
re
se
a
rc
h
sc
h
o
lar
in
th
e
De
p
a
rt
m
e
n
t
of
El
e
c
tro
n
ics
a
n
d
Co
m
m
u
n
ica
ti
o
n
En
g
in
e
e
rin
g
at
S
RM
In
stit
u
te
of
S
c
ien
c
e
a
n
d
T
e
c
h
n
o
lo
g
y
,
Ka
tt
a
n
k
u
lath
u
r,
T
a
m
il
Na
d
u
,
In
d
ia.
S
h
e
h
a
s
6
y
e
a
r
s
of
a
c
a
d
e
m
ic
tea
c
h
in
g
e
x
p
e
rien
c
e
.
He
r
re
se
a
rc
h
a
re
a
in
c
lu
d
e
s
im
a
g
e
p
ro
c
e
ss
in
g
,
m
a
c
h
in
e
lea
r
n
in
g
,
a
n
d
d
e
e
p
lea
rn
in
g
.
S
h
e
c
a
n
be
c
o
n
tac
ted
at
e
m
a
il
:
m
h
3
0
0
6
@s
rm
ist.
e
d
u
.
i
n
.
Dr
.
Ch
itr
a
E
k
a
m
b
a
r
a
m
is
an
A
s
sista
n
t
P
r
o
f
e
ss
o
r
in
th
e
De
p
a
rtme
n
t
of
El
e
c
tro
n
ic
s
a
n
d
Co
m
m
u
n
ica
ti
o
n
En
g
in
e
e
rin
g
at
S
RM
In
stit
u
te
o
f
S
c
ien
c
e
&
T
e
c
h
n
o
lo
g
y
,
Ka
tt
a
n
k
u
lath
u
r,
T
a
m
il
N
a
d
u
,
In
d
ia
sin
c
e
2
0
0
6
.
S
h
e
o
b
tain
e
d
h
e
r
P
h
.
D.
d
e
g
re
e
f
ro
m
S
RM
In
stit
u
te
of
S
c
ien
c
e
&
T
e
c
h
n
o
lo
g
y
,
K
a
tt
a
n
k
u
lath
u
r.
S
h
e
h
a
s
22
y
e
a
rs
of
e
x
p
e
rien
c
e
in
m
a
n
a
g
in
g
u
n
d
e
rg
ra
d
u
a
te,
a
n
d
p
o
stg
ra
d
u
a
te
p
r
o
g
ra
m
s
a
n
d
su
p
e
r
v
is
in
g
re
se
a
r
c
h
p
ro
jec
ts.
He
r
re
se
a
rc
h
in
tere
sts
in
c
lu
d
e
V
L
S
I
lo
w
p
o
w
e
r
h
ig
h
-
sp
e
e
d
d
e
sig
n
,
DS
P
stru
c
tu
re
s
a
n
d
V
L
S
I
d
e
sig
n
a
u
to
m
a
ti
o
n
,
im
a
g
e
p
ro
c
e
ss
in
g
,
m
a
c
h
in
e
lea
rn
in
g
,
a
n
d
d
e
e
p
lea
rn
i
n
g
.
S
h
e
can
be
c
o
n
tac
ted
at
e
m
a
il
:
c
h
it
ra
e
@s
rm
ist.
e
d
u
.
in
.
Evaluation Warning : The document was created with Spire.PDF for Python.