I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
p
u
t
er
Science
Vo
l.
22
,
No
.
3
,
J
u
n
e
2
0
2
1
,
p
p
.
1540
~
154
7
I
SS
N:
2
5
02
-
4
7
5
2
,
DOI
: 1
0
.
1
1
5
9
1
/i
j
ee
cs.v
2
2
.i
3
.
p
p
1
5
4
0
-
154
7
1540
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs.ia
esco
r
e.
co
m
Pre
-
trained
deep
lea
rning
m
o
d
els i
n auto
m
a
tic
C
O
V
ID
-
19
dia
g
no
sis
Ah
m
ed
Wa
s
if
Rez
a
,
M
d
M
a
ha
m
ud
u
l H
a
s
a
n,
Na
zla
No
wrin
,
M
ir
M
o
y
nu
dd
in Ah
m
e
d
Sh
ibl
y
De
p
a
rtme
n
t
o
f
Co
m
p
u
ter S
c
ien
c
e
a
n
d
E
n
g
in
e
e
rin
g
,
Eas
t
W
e
st Un
iv
e
rsit
y
,
Dh
a
k
a
,
Ba
n
g
lad
e
sh
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Feb
2
,
2
0
2
1
R
ev
i
s
ed
Ma
r
2
1
,
2
0
2
1
A
cc
ep
ted
Ma
r
2
9
,
2
0
2
1
Co
ro
n
a
v
iru
s
d
ise
a
se
(COV
ID
-
1
9
)
is
a
d
e
v
a
sta
ti
n
g
p
a
n
d
e
m
ic
in
th
e
h
isto
ry
o
f
m
a
n
k
in
d
.
It
is
a
h
ig
h
ly
c
o
n
tag
io
u
s
f
lu
th
a
t
c
a
n
sp
re
a
d
f
ro
m
h
u
m
a
n
to
h
u
m
a
n
w
it
h
o
u
t
re
v
e
a
li
n
g
a
n
y
s
y
m
p
to
m
s.
F
o
r
b
e
i
n
g
so
c
o
n
tag
io
u
s,
d
e
tec
ti
n
g
p
a
ti
e
n
ts
w
it
h
it
a
n
d
iso
lati
n
g
t
h
e
m
h
a
s
b
e
c
o
m
e
th
e
p
r
im
a
r
y
c
o
n
c
e
rn
f
o
r
h
e
a
lt
h
c
a
re
p
ro
f
e
ss
io
n
a
ls.
T
h
is
st
u
d
y
p
re
se
n
t
e
d
a
n
a
lt
e
rn
a
ti
v
e
w
a
y
to
id
e
n
ti
fy
COV
ID
-
1
9
p
a
ti
e
n
ts
b
y
d
o
in
g
a
n
a
u
to
m
a
ti
c
e
x
a
m
in
a
ti
o
n
o
f
c
h
e
st
X
-
ra
y
s
o
f
th
e
p
a
ti
e
n
ts.
T
o
d
e
v
e
lo
p
su
c
h
a
n
e
f
f
i
c
ien
t
sy
ste
m
,
six
p
re
-
train
e
d
d
e
e
p
lea
rn
i
n
g
m
o
d
e
ls
w
e
r
e
u
se
d
.
T
h
o
se
m
o
d
e
ls
w
e
r
e
:
V
GG
1
6
,
In
c
e
p
ti
o
n
V3
,
X
c
e
p
ti
o
n
,
De
n
se
Ne
t2
0
1
,
I
n
c
e
p
ti
o
n
Re
sN
e
tV2
,
a
n
d
Ef
f
icie
n
tNe
tB4
.
T
h
o
se
m
o
d
e
ls
w
e
re
d
e
v
e
lo
p
e
d
o
n
tw
o
o
p
e
n
-
s
o
u
rc
e
d
a
tas
e
ts
th
a
t
h
a
v
e
c
h
e
st
X
-
ra
y
s
o
f
p
a
ti
e
n
ts
d
iag
n
o
se
d
w
it
h
COV
ID
-
1
9
.
A
m
o
n
g
t
h
e
m
o
d
e
l
s,
Ef
f
ici
e
n
tNe
tB4
a
c
h
iev
e
d
b
e
tt
e
r
p
e
rf
o
rm
a
n
c
e
s
o
n
b
o
th
d
a
ta
se
ts
w
it
h
9
6
%
a
n
d
9
7
%
o
f
a
c
c
u
ra
c
ies
.
T
h
e
e
m
p
iri
c
a
l
re
su
lt
s
we
re
a
lso
e
x
e
m
p
lar
y
.
T
h
is
t
y
p
e
o
f
a
u
to
m
a
t
e
d
s
y
ste
m
c
a
n
h
e
lp
u
s f
ig
h
t
th
is
d
a
n
g
e
ro
u
s v
iru
s
o
u
t
b
re
a
k
.
K
ey
w
o
r
d
s
:
B
io
m
ed
ical
i
m
a
g
e
class
if
icati
o
n
C
OVI
D
-
19
Dee
p
lear
n
in
g
E
f
f
icien
tNetB
4
T
r
an
s
f
er
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
:
Ah
m
ed
W
asif
R
ez
a
Dep
ar
t
m
en
t o
f
C
o
m
p
u
ter
Scie
n
ce
an
d
E
n
g
i
n
ee
r
in
g
E
ast W
es
t U
n
i
v
er
s
i
t
y
Dh
a
k
a,
B
an
g
lad
es
h
E
m
ail:
w
a
s
i
f
@
e
w
u
b
d
.
ed
u
1.
I
NT
RO
D
UCT
I
O
N
C
o
r
o
n
av
ir
u
s
d
is
ea
s
e
(
C
O
VI
D
-
1
9
)
h
as
b
ec
o
m
e
a
w
o
r
ld
w
id
e
o
u
tb
r
ea
k
an
d
i
s
an
o
n
g
o
i
n
g
p
an
d
e
m
ic.
I
t
ca
n
tr
an
s
m
it
f
r
o
m
h
u
m
an
to
h
u
m
a
n
w
it
h
o
u
t
r
e
v
ea
lin
g
a
n
y
s
y
m
p
to
m
s
.
T
h
e
C
OVI
D
-
1
9
’
s
ep
id
e
m
io
lo
g
ical
f
ea
t
u
r
es
ar
e
s
till
a
m
atter
o
f
s
cr
u
tin
y
.
T
h
e
s
tati
s
tics
s
o
f
ar
i
n
d
icate
th
at
ab
o
u
t
8
0
%
o
f
C
O
VI
D
-
1
9
p
atien
ts
ar
e
n
o
t
in
m
o
r
e
th
an
m
ild
co
n
d
itio
n
s
an
d
ab
o
u
t
2
0
%
a
r
e
in
s
e
v
er
e
co
n
d
itio
n
s
.
B
esid
es,
ap
p
r
o
x
i
m
atel
y
1
0
%
o
f
p
atien
ts
n
ee
d
m
ec
h
an
ical
v
en
tilato
r
s
.
T
h
e
r
ate
o
f
d
ea
th
is
ab
o
u
t
2
%,
alth
o
u
g
h
r
esear
ch
er
s
co
n
s
id
er
th
at
th
e
d
ea
th
r
ate
is
0
.
5
%
lo
w
er
[
1
]
.
T
o
tr
ea
t
th
e
C
o
v
id
-
a
f
f
ec
ted
p
atien
t
p
r
o
p
er
l
y
,
it
i
s
v
ital
to
s
ee
k
o
u
t
w
h
et
h
er
a
p
atien
t
is
b
ein
g
af
f
ec
ted
.
A
s
t
an
d
ar
d
p
o
ly
m
er
ase
ch
ai
n
r
ea
ctio
n
(
P
C
R
)
test
ca
n
b
e
p
er
f
o
r
m
ed
o
n
th
e
s
a
m
p
le
s
o
f
th
e
p
atie
n
ts
,
b
u
t
it
ca
n
b
e
c
o
s
tl
y
a
n
d
ti
m
e
-
co
n
s
u
m
in
g
.
A
n
alter
n
ati
v
e
to
th
e
s
tan
d
ar
d
C
OVI
D
-
1
9
d
etec
tio
n
test
s
ca
n
b
e
id
en
ti
f
y
in
g
th
e
af
f
ec
ted
p
atien
ts
f
r
o
m
c
h
est
X
-
r
a
y
s
.
T
h
e
C
OVI
D
-
1
9
p
atien
ts
w
h
o
ar
e
m
o
r
e
v
u
l
n
er
ab
le
to
ca
s
u
alt
y
ar
e
th
o
s
e
w
h
o
s
e
l
u
n
g
s
ar
e
af
f
ec
ted
b
y
th
e
v
ir
u
s
.
T
h
er
ef
o
r
e,
ex
a
m
in
in
g
ch
est
X
-
r
a
y
s
ca
n
b
e
a
v
er
y
ap
p
r
o
p
r
iate
m
et
h
o
d
to
m
a
k
e
a
p
r
ed
ictio
n
ab
o
u
t
a
p
atien
t
t
h
at
h
a
s
C
OVI
D
-
1
9
o
r
n
o
t.
Ho
w
e
v
er
,
o
n
l
y
s
p
ec
ialis
t
p
h
y
s
icia
n
s
ca
n
d
iag
n
o
s
e
C
OVI
D
-
19
b
y
e
x
a
m
in
i
n
g
th
e
ch
e
s
t
X
-
r
a
y
.
T
h
er
e
is
a
s
ca
r
cit
y
o
f
d
o
cto
r
s
w
h
o
ca
n
p
er
f
o
r
m
s
u
c
h
a
d
iag
n
o
s
is
.
I
n
th
e
s
e
cir
cu
m
s
ta
n
ce
s
,
an
au
to
m
ated
an
d
in
telli
g
e
n
t
p
r
ed
ictiv
e
s
y
s
te
m
th
at
ca
n
d
iag
n
o
s
e
p
atien
ts
w
it
h
C
OVI
D
-
1
9
ca
n
b
e
v
er
y
h
elp
f
u
l.
Su
ch
an
au
to
m
ated
d
etec
tio
n
m
ec
h
a
n
is
m
ca
n
r
ed
u
ce
th
e
s
tr
ess
o
f
t
h
e
h
ea
lt
h
ca
r
e
s
y
s
te
m
i
n
a
m
ea
n
in
g
f
u
l
w
a
y
b
y
id
en
tify
i
n
g
th
e
af
f
ec
ted
p
atien
ts
r
ap
id
ly
an
d
ac
cu
r
atel
y
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
P
r
e
-
tr
a
in
ed
d
ee
p
lea
r
n
in
g
mo
d
els in
a
u
to
ma
tic
C
OV
I
D
-
1
9
d
ia
g
n
o
s
is
(
A
h
med
W
a
s
i
f R
eza
)
1541
T
h
is
s
t
u
d
y
h
as
ai
m
ed
at
d
ev
el
o
p
in
g
ap
p
r
o
p
r
iate
p
r
ed
ictiv
e
m
o
d
el
s
t
h
at
ca
n
p
er
f
o
r
m
d
iag
n
o
s
is
o
n
t
h
e
ch
est
X
-
r
a
y
i
m
a
g
es
o
f
s
u
s
p
ec
ted
p
atien
ts
.
T
h
e
d
ee
p
lear
n
in
g
m
o
d
el
s
h
a
v
e
b
ee
n
u
tili
ze
d
to
cr
ea
te
th
o
s
e
p
r
ed
ictiv
e
m
o
d
els.
T
o
d
ev
elo
p
s
u
ch
p
r
ed
ictiv
e
m
o
d
el
s
w
it
h
h
i
g
h
e
f
f
icie
n
c
y
,
a
lar
g
e
d
ataset
is
n
ee
d
ed
.
Ho
w
e
v
er
,
th
er
e
ar
e
a
v
er
y
li
m
i
ted
n
u
m
b
er
o
f
ch
e
s
t
X
-
r
a
y
i
m
ag
es
o
f
C
OVI
D
-
1
9
af
f
ec
ted
p
atien
t
s
ar
e
av
ailab
le
p
u
b
licl
y
.
A
cq
u
ir
i
n
g
a
lar
g
e
n
u
m
b
er
o
f
ap
p
r
o
p
r
iate
im
ag
e
s
is
also
n
o
t
f
ea
s
ib
le.
I
n
th
is
s
it
u
a
tio
n
,
th
e
co
n
ce
p
t
o
f
tr
an
s
f
er
lear
n
i
n
g
ca
n
co
m
e
to
aid
.
T
h
er
e
ar
e
a
h
an
d
f
u
l
o
f
d
e
ep
lear
n
in
g
m
o
d
els
th
a
t
ar
e
p
r
e
-
tr
ain
ed
o
n
lar
g
e
d
atasets
.
T
h
eir
ac
q
u
ir
ed
k
n
o
w
led
g
e
ca
n
b
e
tr
an
s
f
er
r
ed
to
s
o
lv
e
t
h
e
tas
k
o
f
C
OVI
D
-
1
9
d
etec
tio
n
w
it
h
s
m
all
d
atasets
.
Si
n
ce
th
e
y
alr
ea
d
y
k
n
o
w
s
o
m
a
n
y
a
s
p
ec
ts
o
f
v
ar
io
u
s
i
m
a
g
es,
th
o
s
e
m
o
d
els
ca
n
s
lig
h
tl
y
ad
j
u
s
t
t
h
eir
w
ei
g
h
ts
b
ased
o
n
th
e
X
-
r
a
y
i
m
ag
e
s
a
n
d
m
a
k
e
d
ec
en
t
p
r
ed
i
ctio
n
s
.
T
h
is
s
tu
d
y
h
a
s
e
x
p
lo
r
ed
th
e
ap
p
licab
ilit
y
o
f
p
r
e
-
tr
ain
ed
d
ee
p
lear
n
in
g
m
o
d
els
in
th
e
C
OVI
D
-
1
9
d
etec
tio
n
task
.
T
o
ac
h
iev
e
th
i
s
g
o
a
l,
s
ix
p
o
p
u
lar
p
r
e
-
tr
ain
ed
d
ee
p
lear
n
in
g
m
o
d
els
h
av
e
b
ee
n
tr
ain
ed
s
p
ec
ial
l
y
f
o
r
th
e
C
OVI
D
-
1
9
r
ec
o
g
n
itio
n
task
.
Af
ter
t
h
at,
a
co
m
p
ar
ati
v
e
a
n
al
y
s
i
s
o
f
t
h
e
m
o
d
els
h
as
b
ee
n
p
er
f
o
r
m
e
d
.
T
h
e
m
aj
o
r
co
n
tr
ib
u
tio
n
s
o
f
th
is
s
tu
d
y
ar
e:
i)
Dev
elo
p
in
g
au
to
m
atic
d
ia
g
n
o
s
tic
s
y
s
te
m
b
ased
o
n
p
r
e
-
tr
ain
ed
d
ee
p
lear
n
in
g
m
o
d
els
f
o
r
C
OVI
D
-
19
d
etec
tio
n
f
r
o
m
ch
es
t
X
-
r
a
y
i
m
ag
e
s
;
ii)
A
c
h
ie
v
i
n
g
b
etter
p
er
f
o
r
m
a
n
ce
th
a
n
ex
i
s
ti
n
g
wo
r
k
s
in
C
OVI
D
-
19
d
etec
tio
n
tas
k
u
s
i
n
g
E
f
f
icie
n
t
NetB
4
p
r
e
-
tr
ain
ed
m
o
d
el
;
iii)
I
n
th
e
f
o
llo
w
i
n
g
s
ec
tio
n
o
f
th
i
s
ar
ticle,
th
e
r
elate
d
w
o
r
k
s
o
f
th
i
s
s
t
u
d
y
ar
e
p
r
esen
ted
.
T
h
en
,
th
e
m
et
h
o
d
s
an
d
m
ater
ials
o
f
t
h
e
v
ar
io
u
s
ex
p
e
r
i
m
en
ts
h
a
v
e
b
ee
n
d
escr
ib
ed
f
o
llo
w
ed
b
y
t
h
e
r
esu
lts
o
f
t
h
is
s
tu
d
y
w
ith
ap
p
r
o
p
r
iate
d
is
cu
s
s
io
n
.
I
n
th
e
las
t
s
e
ctio
n
o
f
th
i
s
ar
ticle,
f
e
w
co
n
cl
u
d
in
g
r
e
m
ar
k
s
h
av
e
b
ee
n
m
e
n
tio
n
ed
.
2.
RE
L
AT
E
D
WO
RK
S
Ma
j
ee
d
et
a
l.
[
2
]
h
av
e
p
r
o
v
id
ed
an
an
al
y
s
i
s
o
f
1
2
r
eg
u
lar
co
n
v
o
lu
tio
n
al
n
e
u
r
al
n
et
w
o
r
k
m
o
d
els
to
h
elp
r
ad
io
lo
g
is
ts
d
i
s
cr
i
m
i
n
ate
ag
ain
s
t
C
OVI
D
-
1
9
b
ased
o
n
ch
est
X
-
r
a
y
s
,
an
d
al
s
o
h
a
v
e
in
tr
o
d
u
ce
d
a
C
N
N
m
o
d
el
th
at
co
u
ld
g
iv
e
e
f
f
icie
n
t
f
i
n
al
p
r
ed
ictio
n
r
es
u
lts
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
h
as
b
ee
n
d
esig
n
ed
to
p
er
f
o
r
m
r
eliab
le
d
iag
n
o
s
tic
s
f
o
r
C
O
VI
D
v
s
.
No
r
m
al
c
lass
if
ica
t
io
n
an
d
C
O
VI
D
v
s
.
No
r
m
al
v
s
.
P
n
e
u
m
o
n
ia
class
i
f
icatio
n
.
T
h
eir
class
i
f
ier
h
as
p
r
o
v
id
ed
9
8
.
0
8
%
ac
cu
r
ac
y
f
o
r
b
in
ar
y
clas
s
i
f
icatio
n
an
d
8
7
.
0
2
%
ac
cu
r
ac
y
f
o
r
m
u
lticla
s
s
cla
s
s
i
f
ica
tio
n
.
T
h
e
Dar
k
Net
h
a
s
b
ee
n
p
r
o
p
o
s
ed
in
t
h
is
s
tu
d
y
[
3
]
as
a
clas
s
i
f
ier
f
o
r
C
OVI
D
-
19
d
iag
n
o
s
i
s
.
I
n
an
o
t
h
er
s
t
u
d
y
[
4
]
,
th
e
ex
p
er
i
m
e
n
tal
f
i
n
d
in
g
s
h
a
v
e
s
h
o
w
n
t
h
e
ab
ilit
y
o
f
th
e
De
T
r
aC
(
Dec
o
m
p
o
s
e,
T
r
an
s
f
er
,
an
d
C
o
m
p
o
s
e)
m
o
d
el
to
id
en
tify
C
O
VI
D
-
1
9
ca
s
e
s
f
r
o
m
a
lar
g
e
d
ata
co
llectio
n
o
f
i
m
a
g
es
o
b
tain
ed
f
r
o
m
h
o
s
p
itals
ar
o
u
n
d
t
h
e
wo
r
ld
.
Hig
h
ac
c
u
r
ac
y
h
as
b
ee
n
o
b
tain
ed
b
y
DeT
r
aC
f
o
r
t
h
e
id
en
t
if
icatio
n
o
f
C
OVI
D
-
1
9
f
r
o
m
ch
e
s
t
X
-
r
a
y
s
.
C
o
h
e
n
et
a
l.
[
5
]
h
av
e
g
at
h
er
ed
m
ed
ical
i
m
ag
e
s
f
r
o
m
w
eb
s
ites
an
d
p
u
b
licatio
n
s
,
an
d
th
e
d
ataset
c
u
r
r
en
tl
y
co
n
tai
n
s
1
2
3
f
r
o
n
t
-
v
i
e
w
X
-
r
a
y
s
o
f
C
OVI
D
-
1
9
p
ati
en
ts
.
T
h
e
au
th
o
r
s
o
f
f
e
w
w
o
r
k
s
[6
]
,
[
7]
h
av
e
u
s
ed
p
r
e
-
tr
ain
ed
d
ee
p
lear
n
in
g
clas
s
if
ier
s
f
o
r
C
OVI
D
-
1
9
d
iag
n
o
s
is
.
I
n
an
o
t
h
er
w
o
r
k
,
it
h
as
b
ee
n
p
r
o
v
ed
th
at
th
e
au
to
m
ated
s
y
s
te
m
ca
n
d
etec
t
p
n
eu
m
o
n
ia
f
r
o
m
c
h
est
X
-
r
a
y
s
w
ith
h
i
g
h
er
ac
cu
r
ac
y
in
les
s
ti
m
e
t
h
an
h
u
m
a
n
e
x
p
er
ts
[
8
]
.
T
h
e
au
th
o
r
s
o
f
th
e
s
e
w
o
r
k
s
[9
]
-
[
14]
h
av
e
u
s
ed
v
ar
io
u
s
d
ee
p
lear
n
i
n
g
m
et
h
o
d
s
f
o
r
th
e
au
to
m
atic
C
OVI
D
-
1
9
d
iag
n
o
s
i
s
tas
k
.
An
o
th
er
ar
ticle
[
1
5
]
h
as
in
tr
o
d
u
ce
d
a
f
r
am
e
w
o
r
k
f
o
cu
s
ed
o
n
C
ap
s
u
le
Net
w
o
r
k
s
w
h
ic
h
h
a
s
b
ee
n
n
a
m
ed
C
OV
I
D
-
C
A
P
S,
w
h
ic
h
ca
n
m
a
n
ag
e
s
m
al
l
d
atasets
th
a
t
ar
e
o
f
m
aj
o
r
s
ig
n
i
f
ica
n
ce
.
C
O
VI
D
-
C
A
P
S
h
a
s
r
ea
ch
ed
a
n
ac
cu
r
ac
y
o
f
9
5
.
7
%
w
it
h
a
m
u
ch
s
m
a
ller
n
u
m
b
er
o
f
tr
ain
i
n
g
p
ar
a
m
eter
s
.
A
f
a
s
ter
R
-
C
NN
m
o
d
el
h
as a
l
s
o
b
ee
n
i
n
tr
o
d
u
ce
d
to
d
etec
t COVI
D
-
1
9
p
atien
ts
f
r
o
m
ch
e
s
t
X
-
r
a
y
i
m
a
g
es
[
1
6
]
.
I
n
T
a
b
le
1
(
s
ee
ap
p
en
d
ix
)
,
a
s
u
m
m
a
r
y
o
f
f
e
w
s
tate
-
of
-
t
h
e
-
ar
t
a
u
to
m
ated
w
o
r
k
s
o
n
C
OVI
D
-
1
9
d
iag
n
o
s
i
s
h
a
s
b
ee
n
p
r
esen
ted
.
3.
M
E
T
H
O
DS A
ND
M
AT
E
RI
AL
S
T
h
is
s
tu
d
y
ai
m
s
at
d
etec
tin
g
C
OVI
D
-
1
9
f
r
o
m
b
io
m
ed
ical
i
m
ag
e
s
s
o
th
at
ea
r
l
y
an
d
ap
p
r
o
p
r
iate
tr
ea
t
m
e
n
t
o
f
t
h
e
a
f
f
ec
ted
p
ati
en
ts
is
p
o
s
s
ib
le.
T
h
e
o
v
er
all
w
o
r
k
i
n
g
p
r
o
ce
d
u
r
e
o
f
t
h
i
s
s
t
u
d
y
i
s
p
r
ese
n
ted
i
n
Fig
u
r
e
1
.
I
n
t
h
i
s
s
ec
tio
n
o
f
t
h
e
ar
ticle,
th
e
m
e
th
o
d
s
a
n
d
t
h
e
m
ater
ial
s
o
f
th
is
s
t
u
d
y
ar
e
p
r
esen
ted
.
F
ir
s
t,
t
h
e
d
ataset
th
at
h
as
b
ee
n
u
s
ed
is
d
escr
ib
ed
.
A
f
ter
t
h
at,
th
e
m
et
h
o
d
s
o
f
id
en
tify
i
n
g
C
OVI
D
-
1
9
f
r
o
m
X
-
r
a
y
i
m
a
g
es
ar
e
p
r
esen
ted
.
3
.
1
.
Da
t
a
s
et
T
h
er
e
ar
e
s
ev
er
al
p
u
b
licl
y
a
v
ailab
le
d
ataset
s
o
f
C
OVI
D
-
1
9
X
-
r
a
y
i
m
a
g
e
s
.
A
C
OVI
D
-
1
9
X
-
r
a
y
i
m
a
g
e
d
atab
ase
h
a
s
b
ee
n
d
e
v
elo
p
ed
b
y
C
o
h
e
n
et
a
l.
[
5
]
an
d
an
o
th
er
o
n
e
is
av
aila
b
le
at
th
e
Ka
g
g
le
r
ep
o
s
ito
r
y
[
2
1
]
.
Af
ter
co
llecti
n
g
th
e
X
-
r
a
y
i
m
ag
e
s
f
r
o
m
t
h
ese
t
w
o
s
o
u
r
ce
s
,
b
o
t
h
d
ataset
s
ar
e
co
m
b
in
ed
a
n
d
r
ef
in
ed
u
s
in
g
d
o
m
ai
n
k
n
o
w
le
d
g
e.
B
o
th
d
atasets
h
av
e
th
r
ee
class
es
o
f
C
OVI
D
-
1
9
,
No
r
m
al,
an
d
P
n
eu
m
o
n
ia
.
Data
s
et
-
1
co
n
tai
n
s
a
to
tal
o
f
1
,
7
8
0
X
-
r
a
y
i
m
ag
e
s
w
it
h
4
6
0
i
m
a
g
es
o
f
C
OVI
D
-
1
9
in
f
ec
te
d
,
6
6
2
n
o
r
m
al,
a
n
d
6
6
5
P
n
eu
m
o
n
ia.
Data
s
et
-
2
co
n
tai
n
s
a
to
tal
o
f
2
,
9
9
0
X
-
r
ay
i
m
ag
e
s
,
in
cl
u
d
i
n
g
9
9
0
C
OVI
D
-
19
-
i
n
f
ec
ted
i
m
a
g
es,
1
,
0
0
0
No
r
m
al
an
d
1
,
0
0
0
P
n
eu
m
o
n
ia
i
m
a
g
es.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2502
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
22
,
No
.
3
,
J
u
n
e
2
0
2
1
:
1
5
4
0
-
1
5
4
7
1542
Fig
u
r
e
1
.
P
r
o
p
o
s
ed
m
eth
o
d
s
o
f
th
is
s
t
u
d
y
3
.
2
.
Da
t
a
pre
-
pro
ce
s
s
ing
a
nd
i
m
a
g
e
a
ug
m
ent
a
t
io
n
T
o
tr
ain
a
m
ac
h
in
e
lear
n
i
n
g
cl
ass
i
f
ier
,
t
h
e
p
ix
e
l
-
v
al
u
e
r
ep
r
es
en
tatio
n
o
f
i
m
a
g
e
s
is
r
eq
u
ir
ed
.
T
h
e
p
ix
el
v
alu
e
s
o
f
a
n
i
m
a
g
e
ar
e
b
et
wee
n
0
an
d
2
5
5
.
T
h
e
p
ix
el
v
al
u
e
-
w
is
e
r
ep
r
esen
tatio
n
o
f
t
h
e
i
m
a
g
es
i
s
r
escaled
b
et
w
ee
n
0
an
d
1
s
o
t
h
at
t
h
e
m
ac
h
in
e
lear
n
i
n
g
m
o
d
el
ca
n
w
o
r
k
p
r
o
p
er
l
y
o
n
t
h
e
i
n
p
u
t
s
.
W
e
h
av
e
u
s
ed
t
h
is
n
o
r
m
aliza
t
io
n
f
u
n
ctio
n
i
n
b
o
th
tes
t
an
d
tr
ai
n
i
n
g
d
atasets
.
T
h
e
i
m
ag
e
s
ca
p
tu
r
ed
b
y
a
ca
m
er
a
h
a
v
e
v
ar
io
u
s
s
izes.
B
u
t
th
e
d
ee
p
lear
n
in
g
ar
ch
itect
u
r
e
s
h
o
u
ld
b
e
p
r
o
v
id
ed
w
ith
f
i
x
ed
-
s
ized
i
m
a
g
es.
T
h
er
ef
o
r
e,
w
e
h
a
v
e
r
esized
all
th
e
i
m
ag
e
s
to
2
9
9
×
2
9
9
b
ef
o
r
e
t
r
ain
in
g
.
I
m
ag
e
a
u
g
m
e
n
tatio
n
is
a
p
o
w
er
f
u
l
m
et
h
o
d
in
b
u
ild
i
n
g
c
o
n
v
o
l
u
tio
n
al
n
eu
r
al
n
et
w
o
r
k
s
th
at
ca
n
in
cr
ea
s
e
th
e
s
ize
o
f
th
e
tr
ai
n
i
n
g
s
et
w
it
h
o
u
t
ac
q
u
ir
i
n
g
n
e
w
i
m
ag
e
s
.
A
s
p
ar
t
o
f
th
e
d
ata
au
g
m
en
tatio
n
,
w
e
h
a
v
e
r
o
tated
,
cr
o
p
p
ed
,
an
d
f
lip
p
ed
th
e
i
m
a
g
es
to
f
it
t
h
e
m
o
d
el
as
n
ee
d
ed
.
B
y
f
lip
p
in
g
th
e
i
m
ag
e
h
o
r
izo
n
tal
l
y
,
v
er
ticall
y
,
an
d
r
o
tati
n
g
w
e
c
an
co
m
p
letel
y
r
ea
r
r
an
g
e
t
h
e
p
ix
els.
T
h
e
Ker
as
d
ee
p
lear
n
in
g
n
e
u
r
al
n
et
w
o
r
k
lib
r
ar
y
p
r
o
v
id
es
th
e
ca
p
ab
ilit
y
to
f
it
m
o
d
els
u
s
i
n
g
i
m
a
g
e
d
ata
au
g
m
en
ta
tio
n
v
ia
th
e
I
m
ag
e
Data
Ge
n
er
ato
r
class
.
I
m
ag
e
d
ata
au
g
m
e
n
tatio
n
is
u
s
ed
to
ex
p
an
d
th
e
tr
ain
in
g
d
ataset
to
im
p
r
o
v
e
th
e
p
er
f
o
r
m
an
ce
an
d
ab
ilit
y
o
f
th
e
m
o
d
el
to
g
e
n
er
alize
.
3
.
3
.
P
re
-
t
ra
ined
m
o
dels
T
h
e
b
ig
g
est
p
r
o
b
le
m
in
t
h
e
an
al
y
s
is
o
f
m
ed
ical
d
ata
is
t
h
e
lack
o
f
ad
eq
u
ate
d
atasets
.
T
h
e
d
ee
p
lear
n
in
g
class
if
ier
s
n
ee
d
a
lar
g
e
am
o
u
n
t
o
f
d
ata
to
b
e
tr
ain
ed
o
n
.
C
o
llectin
g
m
o
r
e
d
ata
f
o
r
t
h
e
d
o
m
ai
n
-
s
p
ec
i
f
ic
task
s
a
n
d
an
n
o
tati
n
g
t
h
e
m
is
c
o
s
tl
y
a
n
d
ti
m
e
-
co
n
s
u
m
i
n
g
.
T
o
o
v
er
co
m
e
t
h
e
li
m
ita
tio
n
o
f
h
a
v
in
g
a
lar
g
e
d
ataset
in
th
e
tas
k
o
f
C
OVI
D
-
1
9
d
etec
tio
n
,
v
ar
io
u
s
d
ee
p
tr
an
s
f
er
lear
n
in
g
m
o
d
els
h
a
v
e
b
ee
n
em
p
lo
y
ed
in
th
is
s
t
u
d
y
.
T
h
e
b
ig
g
est
ad
v
an
ta
g
e
o
f
u
s
in
g
th
e
tr
an
s
f
er
lear
n
in
g
m
eth
o
d
is
th
at
it
ca
n
tr
ai
n
d
ata
w
it
h
s
m
aller
d
ataset
s
an
d
r
eq
u
ir
es
less
tr
ain
i
n
g
ti
m
e.
T
r
an
s
f
er
lear
n
in
g
is
r
eg
u
lar
l
y
u
s
ed
in
d
ee
p
lear
n
in
g
a
n
d
th
e
w
is
d
o
m
o
b
tain
ed
b
y
th
e
p
r
e
-
tr
ai
n
ed
m
o
d
el
is
co
n
v
e
y
ed
to
t
h
e
cla
s
s
i
f
ier
to
b
e
tr
ain
ed
.
Si
x
p
r
e
-
tr
ai
n
ed
m
o
d
els
h
a
v
e
b
ee
n
u
s
ed
to
d
etec
t
C
OVI
D
-
19
f
r
o
m
X
-
r
a
y
s
a
n
d
t
h
o
s
e
ar
e
VGG1
6
,
Den
s
eNe
t2
0
1
,
I
n
ce
p
tio
n
V3
,
I
n
ce
p
tio
n
R
e
s
NetV2
,
Xce
p
tio
n
,
an
d
E
f
f
icie
n
tNet
.
T
h
o
s
e
m
o
d
els
h
a
v
e
b
ee
n
p
r
e
-
tr
ain
ed
o
n
th
e
I
m
a
g
eNe
t
d
ataset
th
at
h
a
s
1
0
0
0
class
es
o
f
i
m
a
g
es.
Fo
r
b
ein
g
tr
ain
ed
o
n
s
u
c
h
a
h
u
g
e
d
ata
s
et,
th
e
lear
n
ed
w
ei
g
h
ts
o
f
t
h
o
s
e
m
o
d
els
ca
n
b
e
lev
er
ag
ed
to
class
if
y
t
h
e
i
m
a
g
es o
f
th
e
b
io
m
ed
ical
d
o
m
a
in
.
T
h
e
f
ir
s
t
ar
ch
itectu
r
e
th
at
h
as
b
ee
n
u
s
ed
f
o
r
C
OVI
D
-
1
9
d
e
tectio
n
is
VGG1
6
.
I
t
is
a
m
o
d
el
o
f
th
e
d
ee
p
n
eu
r
al
n
et
w
o
r
k
.
T
h
i
s
n
et
w
o
r
k
h
a
s
b
ee
n
i
n
tr
o
d
u
ce
d
b
y
Si
m
o
n
y
a
n
a
n
d
Z
is
s
er
m
a
n
[
2
2
]
.
VGG1
6
s
tar
ts
w
it
h
t
w
o
co
n
v
o
l
u
tio
n
al
la
y
er
s
.
A
f
t
er
th
at,
t
h
er
e
is
a
m
a
x
-
p
o
o
li
n
g
la
y
er
i
n
t
h
e
n
et
w
o
r
k
.
I
t
i
s
f
o
llo
w
ed
b
y
t
w
o
co
n
v
o
lu
tio
n
al
la
y
er
s
a
n
d
a
m
ax
-
p
o
o
lin
g
la
y
er
.
Af
ter
t
h
at,
t
h
er
e
ar
e
th
r
ee
co
n
v
o
l
u
tio
n
al
l
a
y
er
s
f
o
llo
w
ed
b
y
a
p
o
o
lin
g
la
y
e
r
in
th
e
n
et
w
o
r
k
.
A
g
ai
n
,
th
er
e
ar
e
th
r
ee
co
n
v
o
lu
tio
n
a
l
la
y
er
s
in
t
h
e
ar
ch
itec
tu
r
e
f
o
llo
w
ed
b
y
a
m
ax
-
p
o
o
lin
g
la
y
er
w
h
ich
r
ep
e
ats
t
w
o
ti
m
es.
T
h
e
ar
ch
itectu
r
e
en
d
s
w
it
h
t
h
r
ee
d
en
s
e
la
y
er
s
.
T
h
e
last
o
f
th
e
s
e
d
en
s
e
la
y
er
s
is
t
h
e
o
u
tp
u
t la
y
e
r
.
An
o
th
er
ar
ch
i
tect
u
r
e
t
h
at
h
a
s
b
ee
n
u
s
ed
is
I
n
ce
p
tio
n
V3
[
2
3
]
w
h
ic
h
is
an
e
x
te
n
d
ed
n
et
w
o
r
k
o
f
Go
o
g
leNe
t.
I
t
h
a
s
ac
h
iev
ed
s
tr
o
n
g
c
lass
if
ica
tio
n
e
f
f
icie
n
c
y
u
s
i
n
g
tr
an
s
f
er
lear
n
in
g
i
n
s
ev
er
al
b
io
m
ed
ical
ap
p
licatio
n
s
.
T
h
e
Xce
p
tio
n
[
2
4
]
ar
ch
itectu
r
e
is
a
lin
ea
r
s
tac
k
o
f
d
ep
th
w
i
s
e
s
ep
ar
ab
le
co
n
v
o
lu
tio
n
la
y
er
s
w
it
h
r
esid
u
al
co
n
n
ec
t
io
n
s
w
h
ic
h
im
p
l
y
t
h
at
s
p
atial
co
n
v
o
l
u
tio
n
an
d
cr
o
s
s
ch
an
n
el
co
n
v
o
lu
tio
n
is
co
m
p
letel
y
s
ep
ar
ated
.
Fo
llo
w
i
n
g
Go
o
g
L
eNe
t,
th
e
I
n
ce
p
tio
n
V3
m
o
d
el
h
as
b
ee
n
p
r
o
p
o
s
ed
to
co
n
ca
ten
ate
s
e
v
er
al
co
n
v
o
lu
tio
n
al
f
ilter
s
o
f
v
ar
io
u
s
s
izes
i
n
to
a
n
e
w
f
il
ter
.
Su
ch
a
d
esig
n
m
i
n
i
m
izes
t
h
e
n
u
m
b
er
o
f
p
ar
am
eter
s
to
b
e
lear
n
ed
,
an
d
th
e
co
m
p
u
tat
i
o
n
al
co
m
p
lex
it
y
is
t
h
er
eb
y
r
ed
u
ce
d
.
I
t
is
a
4
8
-
la
y
er
lar
g
e
co
n
v
o
l
u
tio
n
al
n
eu
r
al
n
et
w
o
r
k
.
I
n
ce
p
tio
n
V3
is
a
co
n
v
o
lu
tio
n
al
n
e
u
r
al
n
et
w
o
r
k
d
esi
g
n
o
f
th
e
I
n
ce
p
tio
n
f
a
m
il
y
.
Xc
ep
tio
n
ar
ch
itectu
r
e
f
o
cu
s
ed
en
tire
l
y
o
n
s
ep
ar
ab
le
co
n
v
o
lu
tio
n
al
la
y
er
s
i
n
d
ep
th
.
Xce
p
tio
n
n
et
w
o
r
k
h
as
b
ee
n
b
u
ild
b
ased
o
n
th
e
I
n
ce
p
tio
n
n
et
w
o
r
k
.
I
n
Xce
p
ti
o
n
,
th
e
co
n
v
o
lu
tio
n
al
la
y
er
s
o
f
in
ce
p
tio
n
b
lo
ck
s
o
f
t
h
e
I
n
ce
p
tio
n
n
et
w
o
r
k
h
a
v
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
P
r
e
-
tr
a
in
ed
d
ee
p
lea
r
n
in
g
mo
d
els in
a
u
to
ma
tic
C
OV
I
D
-
1
9
d
ia
g
n
o
s
is
(
A
h
med
W
a
s
i
f R
eza
)
1543
b
ee
n
r
ep
lace
d
w
it
h
s
ep
ar
ab
le
co
n
v
o
lu
tio
n
al
la
y
er
s
th
at
g
iv
e
th
e
s
a
m
e
f
u
n
ctio
n
al
ities
as
c
o
n
v
o
l
u
tio
n
al
la
y
er
s
b
u
t
w
ith
le
s
s
co
m
p
u
tatio
n
an
d
w
it
h
m
o
r
e
e
f
f
ic
ien
c
y
.
I
t
is
an
ef
f
ic
ien
t
ar
c
h
itect
u
r
e
th
at
d
ep
en
d
s
o
n
d
ep
th
w
i
s
e
s
ep
ar
ab
le
co
n
v
o
lu
tio
n
an
d
s
h
o
r
tc
u
ts
b
et
w
ee
n
co
n
v
o
l
u
tio
n
b
lo
ck
s
as i
n
R
e
s
Net.
Den
s
eNe
t2
0
1
[
2
5
]
is
a
d
ee
p
co
n
v
o
lu
t
io
n
al
n
e
u
r
al
n
e
t
w
o
r
k
th
at
is
2
0
1
la
y
er
s
d
ee
p
.
Den
s
eNe
t
co
n
tain
s
f
e
w
d
en
s
e
b
lo
ck
s
.
E
ac
h
d
en
s
e
b
lo
ck
h
as
f
e
w
c
o
n
v
o
l
u
tio
n
al
la
y
er
s
.
W
ith
i
n
a
d
en
s
e
b
lo
ck
,
th
e
co
n
v
o
lu
tio
n
al
la
y
er
s
ar
e
co
n
n
ec
ted
to
ea
ch
p
r
ev
io
u
s
la
y
e
r
.
T
h
e
d
en
s
e
b
lo
ck
s
ar
e
also
co
n
n
ec
ted
to
ea
ch
p
r
ev
io
u
s
d
en
s
e
b
lo
ck
.
A
d
ee
p
er
d
en
s
e
b
lo
ck
n
o
t
o
n
l
y
p
r
o
ce
s
s
es
t
h
e
i
n
f
o
r
m
atio
n
r
ec
eiv
ed
f
r
o
m
t
h
e
p
r
ev
io
u
s
la
y
er
s
b
u
t
also
co
n
s
id
er
s
th
e
o
u
tp
u
ts
o
f
all
p
r
ev
io
u
s
b
lo
ck
s
.
Den
s
eNe
t
s
h
a
v
e
s
o
m
e
p
er
s
u
asi
v
e
ad
v
a
n
tag
e
s
:
th
e
y
m
itig
a
te
t
h
e
v
a
n
is
h
i
n
g
g
r
ad
ien
t
i
s
s
u
e,
p
r
o
m
o
te
th
e
p
r
o
p
ag
atio
n
o
f
f
ea
t
u
r
es,
f
ac
i
litate
th
e
r
eu
s
e
o
f
f
ea
t
u
r
es,
an
d
s
i
g
n
i
f
ica
n
tl
y
m
i
n
i
m
ize
t
h
e
n
u
m
b
er
o
f
p
ar
a
m
e
ter
s
.
T
h
e
f
i
f
t
h
d
ee
p
lear
n
i
n
g
a
r
ch
itect
u
r
e
th
at
h
a
s
b
ee
n
e
m
p
lo
y
ed
in
th
is
w
o
r
k
is
I
n
ce
p
tio
n
R
e
s
NetV2
[
2
6
]
.
I
t
is
a
co
n
v
o
lu
tio
n
a
l
n
eu
r
al
ar
ch
itectu
r
e
th
at
b
u
ild
s
o
n
th
e
ar
ch
itect
u
r
e
f
a
m
il
y
o
f
I
n
ce
p
tio
n
b
u
t
in
te
g
r
ates
r
esid
u
al
co
n
n
ec
tio
n
s
,
r
ep
lacin
g
t
h
e
I
n
ce
p
tio
n
ar
ch
itect
u
r
e's
f
ilter
co
n
ca
ten
a
tio
n
le
v
el.
I
t
h
as
1
6
4
la
y
er
s
.
T
h
is
n
et
w
o
r
k
h
as
b
ee
n
d
ev
el
o
p
ed
b
ased
o
n
th
e
I
n
ce
p
tio
n
an
d
t
h
e
r
esid
u
al
c
o
n
n
ec
tio
n
.
Mu
lt
ip
le
-
s
ized
co
n
v
o
l
u
tio
n
a
l
f
i
lter
s
ar
e
co
m
b
i
n
ed
w
ith
r
es
id
u
al
co
n
n
ec
tio
n
s
i
n
th
e
I
n
ce
p
tio
n
-
R
es
n
et
b
lo
ck
.
E
ac
h
o
f
t
h
e
C
NN
ar
c
h
itect
u
r
es
h
a
s
s
o
m
e
u
n
iq
u
e
ch
ar
ac
ter
i
s
tics
.
E
f
f
icie
n
tNe
ts
ar
e
a
f
a
m
il
y
o
f
i
m
a
g
e
class
i
f
icatio
n
m
o
d
els
th
at
ac
h
iev
ed
s
tate
-
of
-
t
h
e
-
ar
t
ac
c
u
r
a
c
y
y
e
t
b
ein
g
an
o
r
d
er
-
of
-
m
a
g
n
i
tu
d
e
s
m
al
ler
an
d
f
aster
th
a
n
o
th
er
d
ee
p
lear
n
in
g
m
o
d
els
[
2
7
]
.
T
h
e
c
o
r
e
id
ea
o
f
E
f
f
ic
ien
t
Net
ar
ch
itect
u
r
e
is
ab
o
u
t
s
tr
ateg
ic
all
y
s
ca
lin
g
d
ee
p
n
eu
r
al
n
et
w
o
r
k
s
.
T
h
e
s
ca
lin
g
m
et
h
o
d
in
tr
o
d
u
ce
d
in
E
f
f
icien
tNets
i
s
n
a
m
ed
co
m
p
o
u
n
d
s
ca
li
n
g
an
d
s
u
g
g
e
s
ts
t
h
at
in
s
tead
o
f
s
c
alin
g
o
n
l
y
o
n
e
m
o
d
el
attr
ib
u
te
o
u
t
o
f
d
ep
th
,
w
id
th
,
an
d
r
eso
lu
tio
n
;
s
tr
ateg
ica
ll
y
s
ca
lin
g
all
t
h
r
ee
o
f
t
h
e
m
to
g
et
h
er
d
eliv
er
s
b
etter
r
esu
lts
.
Mo
d
el
s
ca
lin
g
is
ab
o
u
t
s
ca
lin
g
t
h
e
ex
is
ti
n
g
m
o
d
el
i
n
ter
m
s
o
f
m
o
d
el
d
ep
th
,
m
o
d
el
w
id
t
h
,
an
d
les
s
p
o
p
u
lar
in
p
u
t
i
m
a
g
e
r
eso
lu
t
io
n
to
i
m
p
r
o
v
e
t
h
e
p
er
f
o
r
m
a
n
ce
o
f
th
e
m
o
d
el.
Scalin
g
u
p
an
y
d
i
m
e
n
s
io
n
o
f
n
et
w
o
r
k
w
id
th
,
d
ep
th
,
o
r
r
eso
lu
tio
n
i
m
p
r
o
v
e
s
ac
cu
r
ac
y
,
b
u
t
th
e
ac
cu
r
ac
y
g
a
in
d
i
m
in
i
s
h
es
f
o
r
b
ig
g
er
m
o
d
els.
I
t
is
cr
itical
to
b
alan
ce
all
d
im
e
n
s
io
n
s
o
f
n
et
w
o
r
k
w
id
t
h
,
d
ep
th
,
an
d
r
eso
lu
tio
n
d
u
r
in
g
s
ca
li
n
g
.
E
f
f
icien
tNet
is
o
n
e
o
f
th
e
m
o
s
t e
f
f
icien
t a
r
ch
itect
u
r
es
f
o
r
i
m
ag
e
class
if
icatio
n
.
4.
E
XP
E
R
I
M
E
NT
A
L
SE
T
UP
T
h
is
w
o
r
k
ai
m
s
at
d
etec
ti
n
g
C
OVI
D
-
1
9
f
r
o
m
x
-
r
a
y
s
.
Fo
r
d
ev
elo
p
in
g
t
h
e
tr
a
n
s
f
er
lear
n
i
n
g
m
o
d
els,
Ker
as
o
n
to
p
o
f
T
en
s
o
r
Flo
w
-
a
p
y
th
o
n
lib
r
ar
y
h
a
s
b
ee
n
u
tili
z
ed
.
T
h
e
m
o
d
els
h
a
v
e
b
ee
n
tr
ai
n
ed
o
n
a
co
m
p
u
ter
h
av
i
n
g
I
n
te
l
C
o
r
e
i5
7
2
0
0
U
C
P
U
an
d
8
GB
R
A
M
,
w
it
h
W
in
d
o
w
s
1
0
o
p
er
atin
g
s
y
s
te
m
.
T
h
e
d
atasets
h
av
e
b
ee
n
s
p
lit
i
n
to
th
r
ee
s
et
s
-
tr
ain
,
v
alid
atio
n
,
a
n
d
test
s
e
ts
.
T
h
e
tr
ain
in
g
,
v
alid
atio
n
,
a
n
d
test
s
e
ts
h
a
v
e
7
5
%,
1
0
%,
an
d
1
5
%
X
-
r
a
y
i
m
a
g
es,
r
esp
ec
tiv
el
y
.
Du
r
i
n
g
tr
ai
n
in
g
th
e
m
o
d
el
s
,
af
ter
ea
ch
ep
o
ch
,
th
e
m
o
d
els
h
a
v
e
b
ee
n
v
alid
ated
w
it
h
a
v
alid
atio
n
s
et.
Fo
r
h
av
in
g
th
r
ee
clas
s
e
s
(
No
r
m
al,
P
n
eu
m
o
n
ia,
an
d
C
OVI
D
-
1
9
)
in
th
e
d
atasets
,
th
e
clas
s
if
icatio
n
p
r
o
b
lem
o
f
th
is
s
t
u
d
y
is
a
m
u
lti
-
clas
s
class
i
f
icat
io
n
p
r
o
b
le
m
.
T
h
er
ef
o
r
e,
a
ca
teg
o
r
ical
cr
o
s
s
-
en
tr
o
p
y
lo
s
s
f
u
n
ctio
n
h
as
b
ee
n
u
s
ed
.
A
d
d
itio
n
all
y
,
ad
a
m
o
p
ti
m
izer
h
a
s
b
ee
n
u
s
ed
f
o
r
all
t
h
e
m
o
d
el
s
d
u
r
in
g
tr
ain
i
n
g
.
Af
ter
th
e
co
m
p
letio
n
o
f
tr
ain
in
g
,
th
e
m
o
d
els
h
a
v
e
b
ee
n
test
ed
ag
ai
n
s
t
t
h
e
test
s
et,
an
d
b
ased
o
n
f
e
w
ev
al
u
atio
n
m
etr
i
cs,
th
e
p
er
f
o
r
m
a
n
ce
o
f
t
h
e
tr
ai
n
ed
m
o
d
els is
r
ep
o
r
ted
.
4
.
1
.
E
v
a
lua
t
i
o
n
m
et
rics
T
h
e
d
e
v
e
l
o
p
e
d
m
o
d
e
l
s
h
a
v
e
b
e
e
n
e
v
a
l
u
a
t
e
d
u
s
i
n
g
a
m
u
l
t
i
-
c
l
a
s
s
c
o
n
f
u
s
i
o
n
m
a
t
r
i
x
.
P
r
e
c
i
s
i
o
n
,
r
e
c
a
l
l
,
f1
-
s
c
o
r
e
,
a
c
c
u
r
a
c
y
,
a
n
d
s
p
e
c
i
f
i
c
i
t
y
h
a
v
e
b
e
e
n
u
s
e
d
t
o
e
v
a
l
u
a
t
e
t
h
e
m
o
d
e
l
s
.
P
r
e
c
i
s
i
o
n
i
s
t
h
e
p
o
r
t
i
o
n
o
f
t
h
e
c
o
r
r
e
c
t
p
r
e
d
i
c
t
i
o
n
s
m
a
d
e
b
y
t
h
e
c
l
a
s
s
i
f
i
e
r
w
i
t
h
r
e
s
p
e
c
t
t
o
t
o
t
a
l
p
r
e
d
i
c
t
e
d
c
l
a
s
s
e
s
.
T
h
e
r
e
c
a
l
l
i
s
t
h
e
p
o
r
t
i
o
n
o
f
t
h
e
c
o
r
r
e
c
t
p
r
e
d
i
c
t
i
o
n
s
m
a
d
e
b
y
t
h
e
c
l
a
s
s
i
f
i
e
r
r
e
g
a
r
d
i
n
g
t
h
e
t
o
t
a
l
e
x
i
s
t
i
n
g
a
c
c
u
r
a
t
e
c
l
a
s
s
e
s
.
F
1
-
s
c
o
r
e
i
s
t
h
e
h
a
r
m
o
n
i
c
m
e
a
n
o
f
p
r
e
c
i
s
i
o
n
a
n
d
r
e
c
a
l
l
i
.
e
.
,
b
o
t
h
p
r
e
c
i
s
i
o
n
a
n
d
r
e
c
a
l
l
a
r
e
g
i
v
e
n
e
q
u
a
l
i
m
p
o
r
t
a
n
c
e
w
h
i
l
e
e
v
a
l
u
a
t
i
n
g
t
h
e
p
e
r
f
o
r
m
a
n
c
e
o
f
a
c
l
a
s
s
i
f
i
e
r
.
A
c
c
u
r
a
c
y
i
s
k
n
o
w
n
a
s
t
h
e
o
v
e
r
a
l
l
r
e
c
o
g
n
i
t
i
o
n
r
a
t
e
.
A
n
d
S
p
e
c
i
f
i
c
i
t
y
i
s
t
h
e
t
r
u
e
n
e
g
a
t
i
v
e
r
a
t
e
.
I
f
,
,
,
,
,
a
n
d
a
r
e
t
r
u
e
p
o
s
i
t
i
v
e
,
t
r
u
e
n
e
g
a
t
i
v
e
,
f
a
l
s
e
p
o
s
i
t
i
v
e
,
f
a
l
s
e
n
e
g
a
t
i
v
e
,
t
h
e
t
o
t
a
l
n
u
m
b
e
r
o
f
p
o
s
i
t
i
v
e
i
n
s
t
a
n
c
e
s
,
a
n
d
t
h
e
t
o
t
a
l
n
u
m
b
e
r
o
f
n
e
g
a
t
i
v
e
i
n
s
t
a
n
c
e
s
,
r
e
s
p
e
c
t
i
v
e
l
y
,
t
h
e
n
pr
e
c
isio
n
=
TP
TP
+
FP
(
1
)
e
c
a
l
l
=
TP
P
(
2
)
f1
_
s
c
or
e
=
2
×
p
r
ec
i
s
i
o
n
×
r
eca
ll
p
r
ec
i
s
i
o
n
+
r
ecal
l
(
3
)
a
c
c
ura
c
y
=
TP
+
TN
P
+
N
(
4
)
s
pe
c
ifi
c
ity
=
TN
N
(
5
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2502
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
22
,
No
.
3
,
J
u
n
e
2
0
2
1
:
1
5
4
0
-
1
5
4
7
1544
5.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
e
r
esu
lts
o
f
v
ar
io
u
s
ex
p
er
i
m
en
ts
o
f
th
is
s
t
u
d
y
h
a
v
e
b
ee
n
o
u
ts
ta
n
d
i
n
g
.
B
etter
p
er
f
o
r
m
i
n
g
au
to
m
ated
m
o
d
el
s
th
at
ca
n
d
etec
t
a
C
OVI
D
-
1
9
p
atien
t
f
r
o
m
an
x
-
r
ay
i
m
a
g
e
ca
n
b
e
v
er
y
h
elp
f
u
l
to
f
ig
h
t
th
e
o
n
g
o
i
n
g
p
an
d
e
m
ic.
T
h
e
s
ix
p
r
e
-
tr
ai
n
ed
d
ee
p
lear
n
in
g
m
o
d
els
h
av
e
d
e
m
o
n
s
tr
ated
o
v
er
all
g
o
o
d
r
esu
lts
in
t
h
e
d
etec
tio
n
task
.
I
n
t
h
i
s
s
ec
tio
n
o
f
th
e
ar
ti
cle,
th
e
r
esu
l
ts
o
f
t
h
i
s
s
t
u
d
y
w
i
th
v
ar
io
u
s
asp
ec
ts
ar
e
p
r
esen
te
d
.
A
ll
t
h
e
m
o
d
els
h
av
e
b
ee
n
tr
ain
ed
f
o
r
2
0
ep
o
ch
s
an
d
th
ei
r
tr
ain
in
g
an
d
v
alid
atio
n
ac
cu
r
ac
ies
ar
e
o
b
s
er
v
ed
to
g
et
to
k
n
o
w
w
h
eth
er
th
e
lear
n
i
n
g
o
f
th
e
m
o
d
els
h
as
b
ee
n
ap
p
r
o
p
r
iate
o
r
n
o
t.
I
n
Fig
u
r
e
2
,
a
co
m
p
ar
is
o
n
o
f
tr
ain
in
g
ac
cu
r
ac
ies
o
f
th
e
s
ix
m
o
d
el
s
f
o
r
b
o
th
d
ataset
s
is
p
r
esen
ted
.
T
h
e
co
m
p
ar
is
o
n
o
f
v
alid
atio
n
ac
c
u
r
ac
ies
is
also
g
iv
e
n
i
n
th
e
f
ig
u
r
e.
Fro
m
th
e
f
ig
u
r
e,
it
is
o
b
s
er
v
ed
th
at
f
o
r
th
e
f
ir
s
t
d
ataset,
Xce
p
tio
n
an
d
E
f
f
ec
ie
n
tNe
tB
4
m
o
d
els
h
a
v
e
d
e
m
o
n
s
tr
ated
b
etter
ac
cu
r
ac
ies
th
a
n
o
th
er
m
o
d
els.
T
h
e
Xce
p
ti
o
n
h
as
tak
e
n
f
ir
s
t
p
lace
in
ter
m
s
o
f
tr
ain
in
g
ac
cu
r
ac
y
w
h
ile
th
e
E
f
f
ec
ien
tNetB
4
h
as
tak
e
n
th
e
f
ir
s
t
p
lace
o
f
v
alid
atio
n
ac
c
u
r
ac
y
.
T
h
e
E
f
f
e
cien
tNetB
4
m
o
d
el
h
as
co
n
t
in
u
ed
its
b
etter
p
er
f
o
r
m
a
n
ce
f
o
r
th
e
s
ec
o
n
d
d
ataset
ac
h
iev
in
g
f
ir
s
t
p
lace
f
o
r
b
o
th
tr
ain
in
g
an
d
v
alid
ati
o
n
ac
cu
r
ac
y
w
h
ile
t
h
e
I
n
ce
p
tio
n
V3
m
o
d
el
h
as
ac
h
iev
ed
s
ec
o
n
d
p
lace
in
b
o
th
e
v
al
u
atio
n
m
e
tr
ics.
T
h
e
o
th
er
m
o
d
els
h
av
e
s
h
o
w
n
g
o
o
d
p
er
f
o
r
m
a
n
ce
s
b
u
t
n
o
t
as
g
o
o
d
as
th
e
d
escr
ib
ed
o
n
es.
I
n
Fi
g
u
r
e
3
,
th
e
co
m
p
ar
is
o
n
o
f
lo
s
s
es
f
o
r
th
e
m
o
d
els
is
p
r
ese
n
ted
.
J
u
s
t
li
k
e
tr
ai
n
in
g
a
n
d
v
alid
atio
n
ac
c
u
r
ac
ies,
E
f
f
ec
ie
n
tNetB
4
,
Xce
p
tio
n
,
an
d
I
n
ce
p
tio
n
V3
m
o
d
els
h
av
e
b
ee
n
th
e
to
p
p
er
f
o
r
m
er
s
i
n
ter
m
s
o
f
tr
ain
in
g
a
n
d
v
alid
atio
n
lo
s
s
e
s
.
O
n
th
e
o
t
h
er
h
a
n
d
,
V
GG1
6
an
d
I
n
ce
p
tio
n
R
e
s
NetV
2
m
o
d
els h
a
v
e
b
ee
n
w
o
r
s
e
in
ter
m
s
o
f
tr
ain
i
n
g
lo
s
s
f
o
r
b
o
th
d
atasets
.
A
n
d
f
o
r
v
alid
atio
n
lo
s
s
,
Den
s
e
Net2
0
1
,
I
n
ce
p
tio
n
R
esNetV2
,
an
d
VGG1
6
h
av
e
b
ee
n
p
o
o
r
p
er
f
o
r
m
er
s
ac
r
o
s
s
t
h
e
d
atasets
.
T
r
ain
in
g
ti
m
e
is
a
n
i
m
p
o
r
tan
t
asp
ec
t
o
f
tr
ain
i
n
g
d
ee
p
lear
n
in
g
m
o
d
els.
T
h
e
r
ea
s
o
n
f
o
r
u
s
i
n
g
t
h
e
tr
an
s
f
er
lear
n
i
n
g
m
o
d
el
s
i
s
to
r
ed
u
ce
th
e
co
s
t
o
f
tr
ain
i
n
g
.
T
h
e
tr
ai
n
in
g
t
i
m
e
o
f
t
h
e
m
o
d
el
s
u
s
ed
in
th
is
s
t
u
d
y
h
as
b
ee
n
o
b
s
er
v
ed
.
I
n
T
ab
le
2
,
th
e
av
er
a
g
e
late
n
c
y
p
er
s
tep
o
f
th
e
m
o
d
els
o
n
t
w
o
d
ata
s
ets
is
g
i
v
e
n
.
Fro
m
t
h
e
tab
le,
it
is
ap
p
ar
en
t
t
h
at
VGG
1
6
,
Den
s
eNe
t2
0
1
,
an
d
E
f
f
ec
ie
n
tNetB
4
h
a
v
e
ta
k
e
n
les
s
er
ti
m
e
to
tr
ain
t
h
a
n
o
th
er
m
o
d
el
s
.
Xce
p
tio
n
h
as
ta
k
en
al
m
o
s
t
1
5
s
ec
o
n
d
s
p
er
s
tep
t
o
tr
ain
w
h
ich
is
t
h
e
s
lo
w
est
a
m
o
n
g
th
e
s
i
x
p
r
e
-
tr
ain
ed
m
o
d
els.
Fig
u
r
e
2
.
T
r
ain
in
g
an
d
v
a
lid
ati
o
n
ac
cu
r
ac
y
co
m
p
ar
is
o
n
f
o
r
t
w
o
d
atasets
Fig
u
r
e
3
.
T
r
ain
in
g
an
d
v
a
lid
ati
o
n
lo
s
s
co
m
p
ar
is
o
n
f
o
r
t
w
o
d
ataset
s
T
ab
le
2
.
T
r
ain
in
g
ti
m
e
co
m
p
ar
is
o
n
o
f
t
h
e
d
ee
p
lear
n
in
g
m
o
d
els
A
r
c
h
i
t
e
c
t
u
r
e
L
a
t
e
n
c
y
P
e
r
S
t
e
p
(
mi
l
l
i
se
c
o
n
d
)
Ef
f
e
c
i
e
n
t
N
e
t
B
4
6
5
8
I
n
c
e
p
t
i
o
n
V
3
9
,
0
0
0
V
g
g
1
6
5
5
4
I
n
c
e
p
t
i
o
n
R
e
sN
e
t
V
2
8
,
0
0
0
D
e
n
se
N
e
t
2
0
1
5
9
9
X
c
e
p
t
i
o
n
1
5
,
0
0
0
I
n
T
ab
le
3
,
th
e
class
-
w
i
s
e
p
er
f
o
r
m
a
n
ce
s
o
f
t
h
e
m
o
d
els
ac
r
o
s
s
t
w
o
d
ata
s
ets
ar
e
p
r
ese
n
ted
.
P
r
ec
is
io
n
an
d
r
ec
all
o
f
th
e
clas
s
i
f
ier
s
h
a
v
e
also
b
ee
n
p
r
esen
ted
.
B
ec
au
s
e
j
u
s
t
k
n
o
w
i
n
g
t
h
e
ac
c
u
r
ac
i
es
o
f
th
e
c
lass
if
ier
s
f
o
r
th
e
class
i
f
icatio
n
tas
k
s
is
n
o
t
g
o
o
d
en
o
u
g
h
in
th
e
m
ed
ical
d
o
m
ai
n
s
.
I
n
th
i
s
ca
s
e,
id
en
t
if
y
in
g
a
C
OVI
D
-
19
p
atien
t
p
r
ec
is
el
y
i
s
m
o
r
e
i
m
p
o
r
tan
t
th
an
id
en
tify
i
n
g
a
n
o
r
m
al
p
atie
n
t
ac
cu
r
atel
y
.
T
h
er
ef
o
r
e,
alo
n
g
w
ith
ac
cu
r
ac
ies,
o
t
h
er
ev
al
u
atio
n
m
etr
ics
li
k
e
p
r
ec
is
io
n
,
r
ec
all
,
an
d
s
p
ec
i
f
icit
y
ar
e
al
s
o
n
e
e
d
ed
to
p
r
o
v
e
th
e
ap
p
licab
ilit
y
o
f
a
clas
s
i
f
ier
i
n
a
r
ea
l
-
w
o
r
ld
en
v
ir
o
n
m
e
n
t.
Mo
r
eo
v
er
,
class
-
w
is
e
p
er
f
o
r
m
a
n
c
es
ar
e
p
r
esen
ted
in
T
ab
le
3
to
g
iv
e
m
o
r
e
clar
it
y
o
f
th
e
m
o
d
els
’
ap
p
r
o
p
r
iaten
ess
.
Fro
m
t
h
e
tab
le,
it
ca
n
b
e
s
ee
n
th
at
w
h
e
n
it
co
m
e
s
to
p
er
f
o
r
m
in
g
b
ett
er
o
n
th
e
test
s
et
f
o
r
C
OVI
D
-
1
9
d
etec
tio
n
,
th
e
E
f
f
ec
ien
t
NetB
4
m
o
d
el
h
as
ac
h
ie
v
ed
b
etter
p
er
f
o
r
m
a
n
ce
s
f
o
r
b
o
th
d
atase
ts
.
I
t
h
as
ac
h
ie
v
ed
0
.
9
6
an
d
0
.
9
7
ac
cu
r
ac
ies
f
o
r
t
w
o
d
atasets
,
r
esp
ec
ti
v
el
y
.
A
d
d
itio
n
al
l
y
,
f
o
r
t
h
e
cla
s
s
C
OVI
D
-
1
9
,
it
h
a
s
d
e
m
o
n
s
tr
ated
a
0
.
9
9
o
f
f
1
-
s
co
r
e
o
n
b
o
th
test
s
ets.
T
h
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
P
r
e
-
tr
a
in
ed
d
ee
p
lea
r
n
in
g
mo
d
els in
a
u
to
ma
tic
C
OV
I
D
-
1
9
d
ia
g
n
o
s
is
(
A
h
med
W
a
s
i
f R
eza
)
1545
E
f
f
ec
ien
t
NetB
4
h
as
also
y
iel
d
ed
0
.
9
9
o
f
p
r
ec
is
io
n
o
n
th
e
f
ir
s
t
d
ataset
an
d
1
.
0
0
o
f
p
r
ec
i
s
io
n
o
n
th
e
s
ec
o
n
d
d
ataset.
T
h
is
m
o
d
el
h
as
a
ls
o
d
em
o
n
s
tr
ated
t
h
e
h
i
g
h
est
s
p
ec
if
icit
y
o
f
0
.
9
5
o
n
th
e
f
ir
s
t
d
at
aset
an
d
t
h
e
h
i
g
h
est
s
p
ec
if
icit
y
o
f
0
.
9
6
o
n
th
e
s
e
co
n
d
d
ataset
f
o
r
th
e
C
OVI
D
-
1
9
class
.
Ho
w
e
v
er
,
am
o
n
g
th
e
s
ix
m
o
d
el
s
,
f
o
r
d
etec
tin
g
C
OVI
D
-
1
9
,
th
e
Xce
p
tio
n
m
o
d
el
h
as
th
e
h
i
g
h
es
t
r
ec
all
o
f
1
.
0
0
an
d
0
.
9
9
f
o
r
th
e
t
w
o
d
atasets
,
r
esp
ec
tiv
el
y
.
Fro
m
T
ab
le
3
,
it
is
also
ap
p
ar
en
t
th
at
t
h
e
o
th
er
m
o
d
els
ex
ce
p
t
E
f
f
ec
ie
n
tNetB
4
h
a
v
e
also
d
em
o
n
s
tr
ated
g
o
o
d
p
er
f
o
r
m
a
n
ce
s
.
T
h
e
o
v
er
all
g
o
o
d
r
esu
lts
also
in
d
icate
th
e
u
s
ab
ilit
y
o
f
s
u
c
h
d
ee
p
lear
n
in
g
m
o
d
el
s
in
d
etec
ti
n
g
C
OVI
D
-
1
9
p
atien
ts
f
r
o
m
co
r
r
esp
o
n
d
in
g
ch
est
x
-
r
a
y
i
m
ag
e
s
.
T
ab
le
3
.
C
lass
-
w
i
s
e
p
e
r
f
o
r
m
a
n
ce
s
o
f
th
e
d
ee
p
lear
n
i
n
g
m
o
d
el
s
D
a
t
a
se
t
M
o
d
e
l
L
a
b
e
l
R
e
c
a
l
l
S
p
e
c
i
f
i
c
i
t
y
P
r
e
c
i
si
o
n
F1
-
S
c
o
r
e
A
c
c
u
r
a
c
y
D
a
t
a
se
t
1
Ef
f
e
c
i
e
n
t
N
e
t
B
4
C
O
V
I
D
-
19
0
.
9
9
0
.
9
5
0
.
9
9
0
.
9
9
0
.
9
6
N
o
r
mal
1
.
0
0
0
.
9
5
0
.
9
1
0
.
9
5
P
n
e
u
mo
n
i
a
0
.
9
0
0
.
9
9
1
.
0
0
0
.
9
5
I
n
c
e
p
t
i
o
n
V
3
C
O
V
I
D
-
19
0
.
9
8
0
.
9
4
0
.
9
9
0
.
9
8
0
.
9
5
N
o
r
mal
1
.
0
0
0
.
9
2
0
.
8
9
0
.
9
4
P
n
e
u
mo
n
i
a
0
.
8
8
0
.
9
8
0
.
9
9
0
.
9
3
V
g
g
1
6
C
O
V
I
D
-
19
0
.
9
9
0
.
9
2
0
.
9
5
0
.
9
7
0
.
9
4
N
o
r
mal
0
.
9
1
0
.
9
6
0
.
9
5
0
.
9
3
P
n
e
u
mo
n
i
a
0
.
9
3
0
.
9
4
0
.
9
1
0
.
9
2
I
n
c
e
p
t
i
o
n
R
e
sN
e
t
V
2
C
O
V
I
D
-
19
0
.
9
4
0
.
9
3
0
.
9
9
0
.
9
6
0
.
9
3
N
o
r
mal
0
.
9
5
0
.
9
2
0
.
9
1
0
.
9
3
P
n
e
u
mo
n
i
a
0
.
9
1
0
.
9
4
0
.
9
2
0
.
9
2
D
e
n
se
N
e
t
2
0
1
C
O
V
I
D
-
19
0
.
9
7
0
.
9
5
0
.
9
9
0
.
9
8
0
.
9
5
N
o
r
mal
0
.
9
6
0
.
9
4
0
.
9
3
0
.
9
4
P
n
e
u
mo
n
i
a
0
.
9
2
0
.
9
6
0
.
9
4
0
.
9
3
X
c
e
p
t
i
o
n
C
O
V
I
D
-
19
1
.
0
0
0
.
9
5
0
.
9
4
0
.
9
7
0
.
9
6
N
o
r
mal
0
.
9
8
0
.
9
5
0
.
9
6
0
.
9
7
P
n
e
u
mo
n
i
a
0
.
9
2
0
.
9
8
0
.
9
8
0
.
9
5
D
a
t
a
se
t
2
Ef
f
e
c
i
e
n
t
N
e
t
B
4
C
O
V
I
D
-
19
0
.
9
8
0
.
9
6
1
.
0
0
0
.
9
9
0
.
9
7
N
o
r
mal
0
.
9
9
0
.
9
6
0
.
9
3
0
.
9
6
P
n
e
u
mo
n
i
a
0
.
9
4
0
.
9
8
0
.
9
7
0
.
9
6
I
n
c
e
p
t
i
o
n
V
3
C
O
V
I
D
-
19
0
.
9
9
0
.
9
5
0
.
9
9
0
.
9
9
0
.
9
6
N
o
r
mal
0
.
9
8
0
.
9
4
0
.
9
1
0
.
9
5
P
n
e
u
mo
n
i
a
0
.
9
0
0
.
9
8
0
.
9
7
0
.
9
4
V
g
g
1
6
C
O
V
I
D
-
19
0
.
9
9
0
.
9
3
0
.
9
7
0
.
9
8
0
.
9
5
N
o
r
mal
0
.
9
2
0
.
9
6
0
.
9
4
0
.
9
3
P
n
e
u
mo
n
i
a
0
.
9
4
0
.
9
6
0
.
9
4
0
.
9
4
I
n
c
e
p
t
i
o
n
R
e
sN
e
t
V
2
C
O
V
I
D
-
19
0
.
9
6
0
.
9
3
0
.
9
9
0
.
9
8
0
.
9
4
N
o
r
mal
0
.
9
5
0
.
9
4
0
.
9
0
0
.
9
2
P
n
e
u
mo
n
i
a
0
.
9
0
0
.
9
6
0
.
9
3
0
.
9
1
D
e
n
se
N
e
t
2
0
1
C
O
V
I
D
-
19
0
.
9
7
0
.
9
4
0
.
9
9
0
.
9
8
0
.
9
5
N
o
r
mal
0
.
9
5
0
.
9
4
0
.
9
2
0
.
9
4
P
n
e
u
mo
n
i
a
0
.
9
2
0
.
9
6
0
.
9
2
0
.
9
2
X
c
e
p
t
i
o
n
C
O
V
I
D
-
19
0
.
9
9
0
.
9
2
0
.
9
9
0
.
9
9
0
.
9
4
N
o
r
mal
0
.
9
8
0
.
9
2
0
.
8
6
0
.
9
3
P
n
e
u
mo
n
i
a
0
.
8
4
0
.
9
8
0
.
9
8
0
.
9
1
E
f
f
ec
ien
t
NetB
4
h
as
b
ee
n
p
r
o
v
ed
to
b
e
th
e
m
o
s
t
o
u
ts
ta
n
d
i
n
g
p
er
f
o
r
m
er
in
t
h
e
task
o
f
au
to
m
at
ic
C
OVI
D
-
1
9
d
iag
n
o
s
is
f
r
o
m
c
h
est
X
-
r
a
y
s
.
T
h
e
co
m
p
ar
is
o
n
w
it
h
th
e
p
er
f
o
r
m
a
n
ce
s
o
f
o
th
er
d
ee
p
lea
r
n
in
g
m
o
d
el
s
h
as
al
s
o
p
r
o
v
ed
th
e
s
u
p
er
io
r
it
y
o
f
E
f
f
ec
ie
n
tNe
tB
4
in
t
h
is
b
io
m
ed
ical
i
m
a
g
e
c
lass
i
f
icatio
n
.
Usi
n
g
E
f
f
ec
ien
t
NetB
4
f
o
r
C
OVI
D
-
1
9
d
etec
tio
n
in
t
h
i
s
w
o
r
k
i
s
also
o
n
e
o
f
t
h
e
v
er
y
f
ir
s
t
w
o
r
k
s
to
b
e
r
ep
o
r
te
d
.
T
o
th
e
b
est
o
f
o
u
r
k
n
o
w
led
g
e,
th
er
e
h
as
n
o
t
b
ee
n
an
y
r
ep
o
r
ted
w
o
r
k
w
it
h
E
f
f
ec
ie
n
tNetB
4
f
o
r
C
OVI
D
-
1
9
d
etec
tio
n
.
T
h
e
m
ai
n
r
ea
s
o
n
b
eh
i
n
d
th
e
b
etter
p
e
r
f
o
r
m
a
n
ce
o
f
t
h
is
d
ee
p
lear
n
in
g
ar
ch
itect
u
r
e
is
its
s
ca
lab
ilit
y
.
T
h
is
ar
ch
itect
u
r
e
h
as
b
ee
n
i
n
tr
o
d
u
ce
d
w
it
h
t
h
e
m
o
s
t
ap
p
r
o
p
r
iate
n
et
w
o
r
k
d
i
m
e
n
s
io
n
w
h
ich
a
llo
w
s
t
h
e
m
o
d
el
to
lear
n
th
e
m
o
s
t e
f
f
icien
t a
s
p
ec
t
s
o
f
th
e
i
m
a
g
es a
n
d
to
u
s
e
t
h
o
s
e
to
class
if
y
t
h
e
i
m
ag
e
s
ef
f
icie
n
tl
y
.
T
r
an
s
f
er
lear
n
i
n
g
h
as
p
la
y
ed
a
s
ig
n
i
f
ican
t
r
o
le
in
t
h
e
g
o
o
d
p
er
f
o
r
m
an
ce
s
o
f
t
h
is
s
t
u
d
y
.
A
ll
t
h
e
m
o
d
el
s
t
h
at
h
av
e
b
ee
n
u
s
ed
i
n
th
e
e
x
p
er
i
m
e
n
t
s
ar
e
p
r
e
-
tr
ai
n
ed
o
n
th
e
I
m
a
g
eNe
t
d
atase
t.
T
h
o
s
e
lear
n
in
g
s
h
a
v
e
h
elp
ed
o
u
r
m
o
d
els
to
w
o
r
k
b
etter
o
n
th
e
s
p
ec
if
ic
class
i
f
ica
ti
o
n
task
.
I
m
a
g
e
au
g
m
e
n
tatio
n
h
as
also
h
elp
ed
th
e
m
o
d
el
s
to
ac
h
ie
v
e
b
etter
p
er
f
o
r
m
an
ce
.
As t
h
e
d
atasets
u
s
ed
in
t
h
is
s
tu
d
y
ar
e
r
elativ
e
l
y
s
m
a
l
ler
an
d
C
OVI
D
-
19
r
elate
d
ch
est
x
-
r
a
y
i
m
ag
e
s
ar
e
s
ca
r
ce
,
v
ar
io
u
s
au
g
m
e
n
tat
io
n
tech
n
iq
u
es
ap
p
lied
to
th
e
e
x
is
t
in
g
i
m
a
g
es
h
a
v
e
led
th
e
m
o
d
els
to
ac
h
ie
v
e
b
et
ter
g
e
n
er
aliza
b
ilit
y
.
B
etter
b
alan
ce
d
in
d
i
v
id
u
a
l
clas
s
-
w
is
e
te
s
t
p
er
f
o
r
m
a
n
ce
s
i
n
v
ar
io
u
s
e
v
al
u
atio
n
m
etr
ic
s
i
n
d
icate
th
e
r
o
b
u
s
tn
e
s
s
a
n
d
s
u
s
tai
n
ab
ilit
y
o
f
t
h
e
m
o
d
el
s
.
Ho
w
ev
er
,
d
u
e
to
a
lack
o
f
co
m
p
u
tatio
n
al
r
eso
u
r
ce
s
,
th
e
m
o
d
el
s
h
a
v
e
n
o
t
b
ee
n
tr
ain
e
d
f
o
r
m
o
r
e
th
a
n
2
0
ep
o
ch
s
.
T
h
e
lo
n
g
er
tr
ain
in
g
p
h
ase
co
u
ld
h
a
v
e
i
m
p
r
o
v
ed
t
h
e
m
o
d
els
’
p
er
f
o
r
m
a
n
ce
s
.
An
o
th
er
li
m
ita
tio
n
o
f
t
h
i
s
w
o
r
k
i
s
n
o
t
to
ex
p
lo
r
e
th
e
v
ar
io
u
s
h
y
p
er
p
ar
a
m
eter
s
o
f
t
h
e
d
ee
p
lear
n
in
g
ar
ch
itect
u
r
es.
T
u
n
in
g
t
h
o
s
e
h
y
p
er
p
ar
a
m
e
ter
s
w
o
u
ld
s
u
r
el
y
allo
w
t
h
e
m
o
d
els to
p
er
f
o
r
m
b
etter
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2502
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
22
,
No
.
3
,
J
u
n
e
2
0
2
1
:
1
5
4
0
-
1
5
4
7
1546
T
h
is
s
tu
d
y
h
a
s
d
e
m
o
n
s
tr
ated
e
x
ce
p
tio
n
al
p
er
f
o
r
m
a
n
ce
i
n
t
h
e
au
to
m
at
ic
d
iag
n
o
s
is
o
f
C
OVI
D
-
1
9
f
r
o
m
b
io
m
ed
ical
i
m
a
g
es.
I
n
d
o
i
n
g
s
o
,
th
is
w
o
r
k
h
a
s
also
b
ea
ten
f
e
w
s
ta
te
-
of
-
th
e
-
ar
t
w
o
r
k
s
r
el
ated
to
th
e
d
o
m
a
in
-
s
p
ec
if
ic
clas
s
i
f
icatio
n
t
a
s
k
.
I
n
T
ab
le
4
,
a
co
m
p
ar
is
o
n
o
f
th
i
s
w
o
r
k
w
it
h
a
f
e
w
o
f
t
h
e
o
th
er
w
o
r
k
s
is
g
i
v
e
n
.
T
h
e
E
f
f
ec
ien
t
NetB
4
m
et
h
o
d
is
th
e
s
tan
d
o
u
t
p
er
f
o
r
m
er
a
m
o
n
g
th
e
m
eth
o
d
s
i
m
p
le
m
e
n
ted
in
o
u
r
w
o
r
k
.
O
n
t
w
o
d
atasets
,
t
h
is
m
e
th
o
d
h
as
9
6
%
an
d
9
7
%
te
s
t
ac
c
u
r
ac
y
,
r
esp
ec
tiv
el
y
.
T
h
ese
p
er
f
o
r
m
an
ce
s
ar
e
s
u
p
er
io
r
in
co
m
p
ar
is
o
n
to
o
th
er
w
o
r
k
s
l
ik
e
8
9
.
6
0
%
test
ac
cu
r
ac
y
ac
h
iev
ed
b
y
K
h
an
et
a
l.
[
2
8
]
,
9
5
%
test
ac
cu
r
ac
y
ac
h
iev
ed
b
y
W
ah
ee
d
[
2
9
]
an
d
8
7
.
0
2
% test ac
cu
r
ac
y
ac
h
ie
v
e
d
b
y
Ma
j
ee
d
et
a
l.
[
2
]
.
T
ab
le
4
.
T
est ac
cu
r
ac
y
co
m
p
a
r
is
o
n
w
it
h
o
th
er
w
o
r
k
s
W
o
r
k
M
e
t
h
o
d
T
e
st
a
c
c
u
r
a
c
y
K
h
a
n
e
t
a
l
.
[
2
8
]
C
o
r
o
N
e
t
8
9
.
6
0
%
W
a
h
e
e
d
[
2
9
]
C
o
v
i
d
G
A
N
9
5
%
A
b
b
a
s
e
t
a
l
.
[
4
]
A
d
e
e
p
C
N
N
9
2
.
5
%
A
f
sh
a
r
[
1
5
]
C
O
V
I
D
-
C
A
P
S
9
5
.
7
%
M
a
j
e
e
d
e
t
a
l
.
[
2
]
1
2
o
f
f
-
t
h
e
-
sh
e
l
f
C
N
N
a
r
c
h
i
t
e
c
t
u
r
e
s
8
7
.
0
2
%
P
r
o
p
o
se
d
w
o
r
k
(
o
n
d
a
t
a
se
t
1
)
Ef
f
e
c
i
e
n
t
N
e
t
B
4
9
6
%
P
r
o
p
o
se
d
w
o
r
k
(
o
n
d
a
t
a
se
t
2
)
Ef
f
e
c
i
e
n
t
N
e
t
B
4
9
7
%
6.
CO
NCLU
SI
O
N
An
in
te
lli
g
en
t
s
y
s
te
m
t
h
at
id
en
ti
f
ies
C
OVI
D
-
1
9
p
atien
ts
f
r
o
m
th
eir
ch
e
s
t
X
-
r
a
y
s
ca
n
b
e
v
er
y
h
elp
f
u
l
to
f
ig
h
t
t
h
e
o
n
g
o
in
g
p
an
d
e
m
ic
.
T
h
e
ef
f
icie
n
c
y
o
f
th
o
s
e
in
te
l
lig
e
n
t
s
y
s
te
m
s
is
t
h
e
h
ig
h
es
t
co
n
ce
r
n
.
T
h
is
s
t
u
d
y
is
an
atte
m
p
t
to
d
ev
elo
p
s
u
ch
ef
f
icien
t
in
telli
g
en
t
a
n
d
d
ata
-
d
r
iv
en
s
y
s
te
m
s
.
T
h
is
s
t
u
d
y
d
e
v
elo
p
s
s
i
x
tr
an
s
f
er
lear
n
in
g
m
o
d
els
f
o
r
th
e
d
etec
t
io
n
tas
k
.
T
h
e
m
o
d
els
h
a
v
e
b
e
en
p
r
o
v
ed
to
b
e
v
er
y
e
f
f
icie
n
t
in
t
h
e
C
OVI
D
-
19
id
en
ti
f
icatio
n
tas
k
.
Am
o
n
g
th
e
d
ev
elo
p
ed
m
o
d
els,
E
f
f
ec
ie
n
tNetB
4
h
as
y
ield
ed
th
e
b
est
p
er
f
o
r
m
an
ce
.
T
h
er
e
ar
e
f
e
w
li
m
itatio
n
s
o
f
t
h
is
s
t
u
d
y
li
k
e
n
o
t
tu
n
i
n
g
h
y
p
er
p
ar
am
ete
r
s
,
tr
ain
i
n
g
th
e
m
o
d
els
f
o
r
f
e
w
er
ep
o
ch
s
,
n
o
t
u
s
i
n
g
o
th
er
p
r
e
-
tr
ain
ed
m
o
d
els
an
d
n
o
t
u
s
in
g
th
e
e
n
s
e
m
b
le
m
et
h
o
d
s
f
o
r
b
etter
p
e
r
f
o
r
m
an
ce
s
.
T
h
o
s
e
ex
p
er
i
m
e
n
ts
ca
n
b
e
co
n
d
u
cte
d
to
i
m
p
r
o
v
e
t
h
e
ef
f
icie
n
c
y
o
f
t
h
e
p
r
ed
ictiv
e
s
y
s
te
m
s
.
Mo
r
eo
v
er
,
th
e
w
o
r
k
i
n
g
m
ec
h
a
n
i
s
m
o
f
d
ee
p
lear
n
i
n
g
m
o
d
el
s
i
s
n
o
t
u
n
d
er
s
ta
n
d
ab
le
to
th
e
co
m
m
o
n
s
ta
k
eh
o
ld
er
s
.
R
esear
c
h
ca
n
b
e
co
n
d
u
cted
to
en
h
a
n
ce
t
h
e
e
x
p
lain
ab
ilit
y
o
f
d
ee
p
lear
n
i
n
g
m
o
d
els
in
b
io
m
ed
ical
d
o
m
ai
n
s
.
Ho
w
e
v
er
,
d
esp
ite
h
av
i
n
g
t
h
o
s
e
l
i
m
itatio
n
s
,
t
h
e
au
to
m
atic
d
etec
tio
n
s
y
s
te
m
d
ev
elo
p
ed
in
t
h
i
s
s
t
u
d
y
ca
n
s
u
r
el
y
h
elp
u
s
to
o
v
er
co
m
e
th
i
s
ap
p
allin
g
s
it
u
at
io
n
an
d
f
i
g
h
t t
h
is
co
r
o
n
av
ir
u
s
o
u
tb
r
ea
k
.
AP
P
E
NDI
X
T
ab
le
1
.
A
s
u
m
m
ar
y
o
f
f
e
w
wo
r
k
s
in
a
u
to
m
ated
C
OVI
D
-
1
9
d
iag
n
o
s
i
s
f
r
o
m
ch
e
s
t X
-
r
a
y
s
W
o
r
k
M
e
t
h
o
d
D
a
t
a
se
t
F
i
n
d
i
n
g
L
i
mi
t
a
t
i
o
n
/
G
a
p
M
a
j
e
e
d
e
t
a
l
.
[
2
]
C
N
N
-
X
a
r
c
h
i
t
e
c
t
u
r
e
C
o
mp
o
si
t
i
o
n
o
f
t
h
r
e
e
p
u
b
l
i
c
l
y
a
v
a
i
l
a
b
l
e
d
a
t
a
se
t
s
C
N
N
-
X
w
o
r
k
s b
e
t
t
e
r
t
h
a
n
X
c
e
p
t
i
o
n
a
n
d
D
e
n
se
N
e
t
C
o
n
t
a
i
n
s
t
e
x
t
a
n
d
me
d
i
c
a
l
d
e
v
i
c
e
t
r
a
c
e
s o
n
c
h
e
st
X
-
r
a
y
i
mag
e
s
M
a
k
r
i
s
a
n
d
T
se
r
p
e
s
[
1
7
]
N
i
n
e
d
e
e
p
l
e
a
r
n
i
n
g
a
r
c
h
i
t
e
c
t
u
r
e
s
P
a
u
l
C
o
h
e
n
e
t
a
l
.
[
5
]
a
n
d
K
a
g
g
l
e
V
G
G
1
6
a
n
d
V
G
G
1
9
sh
o
w
t
h
e
b
e
st
r
e
su
l
t
s
t
h
a
n
o
t
h
e
r
mo
d
e
l
s.
L
a
c
k
o
f
d
a
t
a
c
o
u
l
d
b
e
e
v
a
l
u
a
t
e
d
w
i
t
h
mo
r
e
a
r
c
h
i
t
e
c
t
u
r
e
s.
Jai
sw
a
l
e
t
a
l
.
[
1
8
]
T
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
mo
d
e
l
w
i
t
h
D
e
n
se
N
e
t
2
0
1
K
a
g
g
l
e
M
a
x
i
m
u
m
A
c
c
u
r
a
c
y
o
f
9
9
.
8
2
%
L
a
c
k
o
f
so
p
h
i
s
t
i
c
a
t
e
d
f
e
a
t
u
r
e
e
x
t
r
a
c
t
i
o
n
t
e
c
h
n
i
q
u
e
s.
N
a
r
i
n
e
t
a
l
.
[
1
9
]
R
e
sN
e
t
5
0
,
R
e
sN
e
t
1
0
1
,
R
e
sN
e
t
1
5
2
,
I
n
c
e
p
t
i
o
n
-
V
3
a
n
d
I
n
c
e
p
t
i
o
n
R
e
sN
e
t
V
2
O
n
e
d
a
t
a
se
t
c
o
n
t
a
i
n
i
n
g
C
O
V
I
D
-
1
9
X
-
r
a
y
s a
n
d
t
w
o
d
a
t
a
se
t
s
o
f
g
e
n
e
r
a
l
c
h
e
st
X
-
r
a
y
s.
R
e
sN
e
t
5
0
y
i
e
l
d
e
d
t
h
e
h
i
g
h
e
st
a
c
c
u
r
a
c
y
F
e
a
t
u
r
e
s a
r
e
n
o
t
e
x
t
r
a
c
t
e
d
u
s
i
n
g
i
mag
e
p
r
o
c
e
ssi
n
g
me
t
h
o
d
s.
A
b
b
a
s
e
t
a
l
.
[
4
]
D
e
e
p
C
N
N
C
o
h
e
n
e
t
a
l
.
[
5
]
R
e
sN
e
t
1
8
p
e
r
f
o
r
me
d
b
e
t
t
e
r
.
L
a
c
k
o
f
u
si
n
g
o
t
h
e
r
d
e
e
p
l
e
a
r
n
i
n
g
mo
d
e
l
s.
O
z
t
u
r
k
e
t
a
l
.
[
3
]
D
a
r
k
N
e
t
C
o
h
e
n
e
t
a
l
.
[
5
]
B
i
n
a
r
y
-
c
l
a
ss a
c
h
i
e
v
e
d
b
e
t
t
e
r
a
c
c
u
r
a
c
y
t
h
a
n
mu
l
t
i
-
c
l
a
ss
c
l
a
ss
i
f
i
c
a
t
i
o
n
.
S
c
a
r
c
i
t
y
o
f
C
O
V
I
D
-
1
9
X
-
r
a
y
s.
S
a
l
i
h
[
2
0
]
M
o
d
i
f
i
e
d
A
l
e
x
N
e
t
A
d
a
t
a
se
t
f
r
o
m K
a
g
g
l
e
M
o
d
i
f
i
e
d
A
l
e
x
N
e
t
p
e
r
f
o
r
me
d
b
e
t
t
e
r
t
h
a
n
o
r
i
g
i
n
a
l
A
l
e
x
N
e
t
L
a
c
k
o
f
B
a
l
a
n
c
e
d
d
a
t
a
se
t
a
n
d
n
o
t
a
p
p
l
y
i
n
g
t
h
e
mo
d
i
f
i
c
a
t
i
o
n
s
t
o
o
t
h
e
r
d
e
e
p
l
e
a
r
n
i
n
g
mo
d
e
l
s.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
P
r
e
-
tr
a
in
ed
d
ee
p
lea
r
n
in
g
mo
d
els in
a
u
to
ma
tic
C
OV
I
D
-
1
9
d
ia
g
n
o
s
is
(
A
h
med
W
a
s
i
f R
eza
)
1547
RE
F
E
R
E
NC
E
S
[1
]
R.
M
.
P
e
re
ira,
D.
Be
rto
li
n
i,
L
.
O.
T
e
ix
e
ira,
C.
N.
S
il
la,
a
n
d
Y.
M
.
G
.
Co
sta
,
“
COV
ID
-
1
9
i
d
e
n
ti
f
ica
ti
o
n
i
n
c
h
e
st
X
-
ra
y
i
m
a
g
e
s
o
n
f
lat
a
n
d
h
iera
rc
h
ica
l
c
las
si
f
ica
ti
o
n
sc
e
n
a
rio
s,”
Co
mp
u
t.
M
e
th
o
d
s
Pro
g
ra
ms
Bi
o
me
d
.
,
2
0
2
0
,
d
o
i:
1
0
.
1
0
1
6
/j
.
c
m
p
b
.
2
0
2
0
.
1
0
5
5
3
2
.
[2
]
T
.
M
a
jee
d
,
R.
Ra
sh
id
,
D.
A
li
,
a
n
d
A
.
A
s
a
a
d
,
“
Iss
u
e
s
a
s
so
c
iate
d
w
it
h
d
e
p
lo
y
in
g
CNN
tran
s
f
e
r
lea
rn
in
g
to
d
e
tec
t
COV
ID
-
1
9
f
ro
m
c
h
e
st X
-
ra
y
s,”
Ph
y
s.
En
g
.
S
c
i.
M
e
d
.
,
2
0
2
0
,
d
o
i:
1
0
.
1
0
0
7
/s
1
3
2
4
6
-
0
2
0
-
0
0
9
3
4
-
8.
[3
]
T
.
Oz
tu
rk
,
M
.
T
a
lo
,
E.
A
.
Yild
iri
m
,
U.
B.
Ba
lo
g
lu
,
O.
Yild
iri
m
,
a
n
d
U.
Ra
jen
d
ra
A
c
h
a
r
y
a
,
“
A
u
to
m
a
ted
d
e
tec
ti
o
n
o
f
COV
ID
-
1
9
c
a
se
s
u
sin
g
d
e
e
p
n
e
u
ra
l
n
e
tw
o
rk
s
w
it
h
X
-
ra
y
i
m
a
g
e
s,”
Co
mp
u
t.
Bi
o
l.
M
e
d
.
,
2
0
2
0
,
d
o
i:
1
0
.
1
0
1
6
/j
.
c
o
m
p
b
io
m
e
d
.
2
0
2
0
.
1
0
3
7
9
2
.
[4
]
A
.
A
b
b
a
s,
M
.
M
.
A
b
d
e
lsa
m
e
a
,
a
n
d
M
.
M
.
G
a
b
e
r,
“
Clas
sif
ica
ti
o
n
o
f
COV
ID
-
1
9
in
c
h
e
st
X
-
ra
y
i
m
a
g
e
s
u
sin
g
De
T
r
a
C
d
e
e
p
c
o
n
v
o
l
u
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
,
”
Ap
p
l.
In
tell
.
,
2
0
2
1
,
d
o
i:
1
0
.
1
0
0
7
/s1
0
4
8
9
-
0
2
0
-
0
1
8
2
9
-
7.
[5
]
J.
P
a
u
l
C
o
h
e
n
,
P
.
M
o
r
riso
n
,
a
n
d
L
.
Da
o
,
“
COV
ID
-
1
9
Im
a
g
e
Da
ta C
o
ll
e
c
ti
o
n
,
”
a
rXiv
.
2
0
2
0
.
[6
]
I.
D.
A
p
o
sto
l
o
p
o
u
l
o
s
a
n
d
T
.
A
.
M
p
e
sia
n
a
,
“
Co
v
id
-
1
9
:
a
u
t
o
m
a
ti
c
d
e
tec
ti
o
n
f
ro
m
X
-
ra
y
i
m
a
g
e
s
u
ti
li
z
in
g
tran
sf
e
r
lea
rn
in
g
w
it
h
c
o
n
v
o
l
u
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
s,”
Ph
y
s.
En
g
.
S
c
i.
M
e
d
.
,
2
0
2
0
,
d
o
i:
1
0
.
1
0
0
7
/s1
3
2
4
6
-
0
2
0
-
0
0
8
6
5
-
4.
[7
]
T
.
Ra
h
m
a
n
,
M
.
E.
H
.
Ch
o
w
d
h
u
r
y
,
A
.
Kh
a
n
d
a
k
a
r,
a
n
d
A
.
Kh
a
n
d
a
k
a
r
,
“
T
ra
n
s
f
e
r
lea
rn
in
g
w
it
h
d
e
e
p
Co
n
v
o
lu
ti
o
n
a
l
Ne
u
ra
l
Ne
tw
o
rk
(CNN
)
f
o
r
p
n
e
u
m
o
n
ia d
e
tec
ti
o
n
u
sin
g
c
h
e
st
X
-
ra
y
,
”
Ap
p
l.
S
c
i.
,
2
0
2
0
,
d
o
i:
1
0
.
3
3
9
0
/
a
p
p
1
0
0
9
3
2
3
3
.
[8
]
H.
S
h
a
rm
a
,
J.
S
.
Ja
in
,
P
.
Ba
n
sa
l,
a
n
d
S
.
G
u
p
ta,
“
F
e
a
tu
re
e
x
trac
ti
o
n
a
n
d
c
las
sif
ica
ti
o
n
o
f
c
h
e
st
X
-
ra
y
i
m
a
g
e
s
u
sin
g
CNN
to
d
e
tec
t
p
n
e
u
m
o
n
ia,” 2
0
2
0
,
d
o
i:
1
0
.
1
1
0
9
/C
o
n
f
lu
e
n
c
e
4
7
6
1
7
.
2
0
2
0
.
9
0
5
7
8
0
9
.
[9
]
X
.
X
u
,
e
t
a
l.
,
“
A
d
e
e
p
lea
rn
in
g
s
y
ste
m
to
sc
r
e
e
n
n
o
v
e
l
c
o
ro
n
a
v
iru
s
d
ise
a
se
2019
p
n
e
u
m
o
n
ia
,
”
En
g
i
n
e
e
rin
g
,
2
0
2
0
,
d
o
i:
1
0
.
1
0
1
6
/j
.
e
n
g
.
2
0
2
0
.
0
4
.
0
1
0
.
[1
0
]
O.
S
tep
h
e
n
,
M
.
S
a
in
,
U.
J.
M
a
d
u
h
,
a
n
d
D.
U.
Je
o
n
g
,
“
A
n
e
ff
icie
n
t
d
e
e
p
lea
rn
in
g
a
p
p
ro
a
c
h
to
p
n
e
u
m
o
n
ia
c
las
si
f
ica
ti
o
n
in
h
e
a
lt
h
c
a
re
,
”
J
.
H
e
a
lt
h
c
.
E
n
g
.
,
2
0
1
9
,
d
o
i:
1
0
.
1
1
5
5
/2
0
1
9
/4
1
8
0
9
4
9
.
[1
1
]
G
.
Ja
in
,
D.
M
it
tal,
D.
T
h
a
k
u
r,
a
n
d
M
.
K.
M
it
tal,
“
A
d
e
e
p
lea
rn
i
n
g
a
p
p
ro
a
c
h
t
o
d
e
tec
t
Co
v
id
-
1
9
c
o
ro
n
a
v
iru
s
w
it
h
X
-
Ra
y
i
m
a
g
e
s,”
Bi
o
c
y
b
e
rn
.
Bi
o
m
e
d
.
En
g
.
,
2
0
2
0
,
d
o
i:
1
0
.
1
0
1
6
/j
.
b
b
e
.
2
0
2
0
.
0
8
.
0
0
8
.
[1
2
]
M
.
T
o
ğ
a
ç
a
r,
B.
Erg
e
n
,
a
n
d
Z.
Cö
m
e
rt,
“
COV
ID
-
1
9
d
e
tec
ti
o
n
u
si
n
g
d
e
e
p
lea
rn
in
g
m
o
d
e
ls
to
e
x
p
lo
it
S
o
c
ial
M
im
ic
Op
ti
m
iza
ti
o
n
a
n
d
stru
c
t
u
re
d
c
h
e
st
X
-
ra
y
i
m
a
g
e
s
u
sin
g
f
u
z
z
y
c
o
lo
r
a
n
d
sta
c
k
in
g
a
p
p
ro
a
c
h
e
s,”
Co
mp
u
t.
Bi
o
l.
M
e
d
.
,
v
o
l.
1
2
1
,
p
p
.
1
0
3
8
0
5
,
Ju
n
.
2
0
2
0
,
d
o
i:
1
0
.
1
0
1
6
/j
.
c
o
m
p
b
io
m
e
d
.
2
0
2
0
.
1
0
3
8
0
5
.
[1
3
]
J.
P
.
Co
h
e
n
,
e
t
a
l
.
,
“
P
re
d
ict
in
g
c
o
v
id
-
1
9
p
n
e
u
m
o
n
ia
se
v
e
rit
y
o
n
c
h
e
st
x
-
ra
y
w
it
h
d
e
e
p
lea
rn
i
n
g
,
”
a
r
Xi
v
,
v
o
l.
1
2
,
n
o
.
7
,
2
0
2
0
,
d
o
i:
1
0
.
7
7
5
9
/c
u
re
u
s.9
4
4
8
.
[1
4
]
A
.
M
.
Is
m
a
e
l
a
n
d
A
.
Ş
e
n
g
ü
r,
“
De
e
p
lea
rn
in
g
a
p
p
ro
a
c
h
e
s
f
o
r
COV
ID
-
1
9
d
e
tec
ti
o
n
b
a
se
d
o
n
c
h
e
st
X
-
ra
y
i
m
a
g
e
s,”
Exp
e
rt S
y
st.
A
p
p
l
.
,
v
o
l.
1
6
4
,
p
.
1
1
4
0
5
4
,
F
e
b
.
2
0
2
1
,
d
o
i:
1
0
.
1
0
1
6
/j
.
e
sw
a
.
2
0
2
0
.
1
1
4
0
5
4
.
[1
5
]
P
.
A
f
sh
a
r,
S
.
He
id
a
rian
,
F
.
Na
d
e
r
k
h
a
n
i,
A
.
Oik
o
n
o
m
o
u
,
K.
N.
P
lat
a
n
io
ti
s,
a
n
d
A
.
M
o
h
a
m
m
a
d
i,
“
COV
ID
-
CA
P
S
:
A
c
a
p
su
le
n
e
tw
o
rk
-
b
a
se
d
f
ra
m
e
w
o
r
k
f
o
r
id
e
n
ti
f
ica
ti
o
n
o
f
COV
ID
-
1
9
c
a
se
s
f
ro
m
X
-
ra
y
ima
g
e
s,”
Pa
tt
e
rn
Rec
o
g
n
it
i
o
n
L
e
tt
e
rs
.
2
0
2
0
,
d
o
i:
1
0
.
1
0
1
6
/j
.
p
a
tre
c
.
2
0
2
0
.
0
9
.
0
1
0
.
[1
6
]
K.
H.
S
h
i
b
ly
,
S
.
K.
De
y
,
M
.
T
.
U.
Isla
m
,
a
n
d
M
.
M
.
Ra
h
m
a
n
,
“
COV
ID
f
a
ste
r
R
–
CNN
:
A
n
o
v
e
l
f
ra
m
e
w
o
rk
to
d
iag
n
o
se
n
o
v
e
l
C
o
ro
n
a
v
iru
s
D
ise
a
se
(COV
ID
-
1
9
)
i
n
X
-
Ra
y
im
a
g
e
s,”
In
fo
rm
a
t
ics
M
e
d
.
Un
lo
c
k
e
d
,
2
0
2
0
,
d
o
i:
1
0
.
1
0
1
6
/j
.
im
u
.
2
0
2
0
.
1
0
0
4
0
5
.
[1
7
]
A
.
M
a
k
ris,
I.
Ko
n
to
p
o
u
lo
s,
a
n
d
K
.
T
se
rp
e
s,
“
COV
ID
-
1
9
d
e
tec
ti
o
n
f
ro
m
c
h
e
st X
-
ra
y
i
m
a
g
e
s u
sin
g
d
e
e
p
lea
rn
in
g
a
n
d
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
s,”
2
0
2
0
,
d
o
i:
1
0
.
1
1
4
5
/3
4
1
1
4
0
8
.
3
4
1
1
4
1
6
.
[1
8
]
A
.
J
a
is
w
a
l,
N.
G
ian
c
h
a
n
d
a
n
i,
D.
S
in
g
h
,
V
.
Ku
m
a
r,
a
n
d
M
.
Ka
u
r,
“
Clas
sif
ic
a
ti
o
n
o
f
th
e
COV
ID
-
1
9
i
n
f
e
c
ted
p
a
ti
e
n
ts
u
sin
g
De
n
se
Ne
t2
0
1
b
a
se
d
d
e
e
p
tran
sf
e
r
lea
rn
in
g
,
”
J
.
Bi
o
mo
l
.
S
tru
c
t.
Dy
n
.
,
2
0
2
0
,
d
o
i:
1
0
.
1
0
8
0
/0
7
3
9
1
1
0
2
.
2
0
2
0
.
1
7
8
8
6
4
2
.
[1
9
]
A
.
Na
rin
,
C.
Ka
y
a
,
a
n
d
Z.
P
a
m
u
k
,
“
A
u
to
m
a
ti
c
d
e
tec
ti
o
n
o
f
c
o
r
o
n
a
v
iru
s
d
ise
a
se
(COV
ID
-
1
9
)
u
sin
g
X
-
ra
y
ima
g
e
s
a
n
d
d
e
e
p
c
o
n
v
o
l
u
ti
o
n
a
l
n
e
u
ra
l
n
e
t
w
o
rk
s,”
a
rXiv
.
2
0
2
0
.
[2
0
]
S
.
Q.
S
a
li
h
,
H.
K.
A
b
d
u
ll
a
,
Z.
S
.
A
h
m
e
d
,
N.
M
.
S
.
S
u
ra
m
e
e
r
y
,
a
n
d
R.
D.
Ra
sh
i
d
,
“
M
o
d
if
ied
A
lex
Ne
t
c
o
n
v
o
l
u
ti
o
n
n
e
u
ra
l
n
e
tw
o
rk
f
o
r
Co
v
id
-
1
9
d
e
tec
ti
o
n
u
si
n
g
c
h
e
st
X
-
ra
y
im
a
g
e
s
,
”
Ku
rd
ist
a
n
J
.
Ap
p
l
.
Res
.
,
2
0
2
0
,
d
o
i:
1
0
.
2
4
0
1
7
/c
o
v
id
.
1
4
.
[2
1
]
P
.
P
a
t
e
l
,
“
C
h
e
s
t
X
-
r
a
y
(
C
o
v
i
d
-
1
9
&
P
n
e
u
m
o
n
i
a
)
,
”
2
0
2
0
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
.
h
t
t
p
s
:
/
/
w
w
w
.
k
a
g
g
l
e
.
c
o
m
/
p
r
a
s
h
a
n
t
2
6
8
/
c
h
e
s
t
-
x
r
a
y
-
c
o
v
i
d
1
9
-
p
n
e
u
m
o
n
i
a
[2
2
]
K.
S
im
o
n
y
a
n
a
n
d
A
.
Zi
ss
e
r
m
a
n
,
“
V
GG
-
1
6
,
”
a
rXiv
Pre
p
r.
,
2
0
1
4
.
[2
3
]
C.
S
z
e
g
e
d
y
,
V
.
V
a
n
h
o
u
c
k
e
,
S
.
Io
ff
e
,
J.
S
h
len
s,
a
n
d
Z.
W
o
jn
a
,
“
Re
th
in
k
in
g
t
h
e
in
c
e
p
ti
o
n
a
rc
h
it
e
c
tu
r
e
f
o
r
c
o
m
p
u
ter
v
isio
n
,
”
Co
m
p
u
ter
Vi
si
o
n
a
n
d
Pa
t
ter
n
Rec
o
g
n
it
io
n
,
2
0
1
6
,
d
o
i:
1
0
.
1
1
0
9
/CVP
R.
2
0
1
6
.
3
0
8
.
[2
4
]
F
.
Ch
o
ll
e
t,
“
X
c
e
p
ti
o
n
:
De
e
p
lea
rn
in
g
w
it
h
d
e
p
t
h
w
ise
se
p
a
ra
b
le
c
o
n
v
o
lu
ti
o
n
s,”
IEE
E
Co
n
fer
e
n
c
e
o
n
Co
m
p
u
te
r
Vi
sio
n
a
n
d
P
a
tt
e
rn
Rec
o
g
n
it
i
o
n
(
CVP
R)
,
2
0
1
7
,
d
o
i:
1
0
.
1
1
0
9
/C
VP
R.
2
0
1
7
.
1
9
5
.
[2
5
]
G
.
Hu
a
n
g
,
Z.
L
iu
,
L
.
V
a
n
De
r
M
a
a
ten
,
a
n
d
K.
Q.
W
e
in
b
e
rg
e
r,
“
De
n
se
l
y
c
o
n
n
e
c
ted
c
o
n
v
o
lu
ti
o
n
a
l
n
e
tw
o
rk
s,
”
2
0
1
7
,
d
o
i:
1
0
.
1
1
0
9
/C
VP
R.
2
0
1
7
.
2
4
3
.
[2
6
]
C.
S
z
e
g
e
d
y
,
S
.
Io
f
f
e
,
V
.
V
a
n
h
o
u
c
k
e
,
a
n
d
A
.
A
.
A
le
m
i,
“
In
c
e
p
ti
o
n
-
v
4
,
in
c
e
p
t
io
n
-
Re
sN
e
t
a
n
d
th
e
imp
a
c
t
o
f
re
sid
u
a
l
c
o
n
n
e
c
ti
o
n
s o
n
lea
rn
i
n
g
,
”
Co
mp
u
ter
Vi
sio
n
a
n
d
P
a
tt
e
rn
Rec
o
g
n
it
i
o
n
,
2
0
1
7
.
[2
7
]
M
.
T
a
n
a
n
d
Q.
V
.
L
e
,
“
Eff
icie
n
tNe
t:
Re
th
in
k
in
g
m
o
d
e
l
sc
a
li
n
g
f
o
r
c
o
n
v
o
l
u
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
s,”
a
rXiv
,
2
0
1
9
.
[2
8
]
A
.
I.
Kh
a
n
,
J.
L
.
S
h
a
h
,
a
n
d
M
.
M
.
Bh
a
t
,
“
Co
ro
Ne
t:
A
d
e
e
p
n
e
u
ra
l
n
e
tw
o
rk
f
o
r
d
e
tec
ti
o
n
a
n
d
d
iag
n
o
sis
o
f
COV
ID
-
1
9
f
ro
m
c
h
e
st
x
-
ra
y
ima
g
e
s,”
Co
mp
u
t.
M
e
th
o
d
s
Pr
o
g
ra
ms
Bi
o
me
d
.
,
v
o
l
.
1
9
6
,
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j.
c
m
p
b
.
2
0
2
0
.
1
0
5
5
8
1
.
[2
9
]
A
.
W
a
h
e
e
d
,
M
.
G
o
y
a
l,
D.
G
u
p
ta,
A
.
Kh
a
n
n
a
,
F
.
A
l
-
T
u
rjm
a
n
,
a
n
d
P
.
R.
P
in
h
e
iro
,
“
Co
v
id
G
A
N:
Da
t
a
A
u
g
m
e
n
tatio
n
Us
in
g
A
u
x
il
iar
y
Cla
ss
i
f
ier
GA
N
f
o
r
I
m
p
ro
v
e
d
Co
v
id
-
1
9
De
tec
ti
o
n
,
”
IEE
E
Acc
e
ss
,
2
0
2
0
,
d
o
i:
1
0
.
1
1
0
9
/A
CCES
S
.
2
0
2
0
.
2
9
9
4
7
6
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.