I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
pu
t
er
Science
Vo
l.
3
9
,
No
.
1
,
Ju
ly
2
0
2
5
,
p
p
.
4
1
0
~
4
2
4
I
SS
N:
2
5
0
2
-
4
7
5
2
,
DOI
: 1
0
.
1
1
5
9
1
/ijeecs.v
3
9
.i
1
.
pp
4
1
0
-
4
2
4
410
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs.ia
esco
r
e.
co
m
Cla
ss
ificatio
n
mo
del
for i
nfec
tious
lung
disea
ses
usin
g
co
nv
o
lutiona
l neural networks
on
web and mo
bile
appli
ca
tions
K
ennedy
O
k
o
k
pu
j
ie
1,
2
,
Alv
i
n K
.
Ag
a
ma
h
1
,
Abid
em
i O
rim
o
g
un
j
e
3
,
I
j
eh
P
rincess
Ada
o
ra
4
,
O
lus
a
ny
a
O
la
m
ide O
mo
la
ra
5
,
Sa
m
uel A
deba
y
o
Da
ra
m
o
l
a
1
,
M
o
ra
y
o
E
m
it
ha
Awo
mo
y
i
6
1
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
i
c
a
l
a
n
d
I
n
f
o
r
mat
i
o
n
E
n
g
i
n
e
e
r
i
n
g
,
C
o
l
l
e
g
e
o
f
E
n
g
i
n
e
e
r
i
n
g
,
C
o
v
e
n
a
n
t
U
n
i
v
e
r
si
t
y
,
O
t
a
,
N
i
g
e
r
i
a
2
A
f
r
i
c
a
C
e
n
t
r
e
o
f
E
x
c
e
l
l
e
n
c
e
f
o
r
I
n
n
o
v
a
t
i
v
e
a
n
d
Tr
a
n
sf
o
r
m
a
t
i
v
e
S
TE
M
Ed
u
c
a
t
i
o
n
,
La
g
o
s S
t
a
t
e
U
n
i
v
e
r
s
i
t
y
,
O
j
o
,
N
i
g
e
r
i
a
3
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
i
c
a
l
a
n
d
El
e
c
t
r
o
n
i
c
E
n
g
i
n
e
e
r
i
n
g
,
R
e
d
e
e
m
e
r
’
s Un
i
v
e
r
si
t
y
Ed
e
,
Ed
e
,
N
i
g
e
r
i
a
4
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
a
n
d
I
n
f
o
r
m
a
t
i
o
n
S
c
i
e
n
c
e
,
C
o
l
l
e
g
e
o
f
S
c
i
e
n
c
e
a
n
d
Te
c
h
n
o
l
o
g
y
,
C
o
v
e
n
a
n
t
U
n
i
v
e
r
s
i
t
y
,
O
t
a
,
N
i
g
e
r
i
a
5
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
En
g
i
n
e
e
r
i
n
g
,
C
o
l
l
e
g
e
o
f
E
n
g
i
n
e
e
r
i
n
g
,
B
e
l
l
s U
n
i
v
e
r
si
t
y
o
f
T
e
c
h
n
o
l
o
g
y
,
O
t
a
,
N
i
g
e
r
i
a
6
S
c
h
o
o
l
o
f
I
n
t
e
r
n
a
t
i
o
n
a
l
S
e
r
v
i
c
e
,
A
m
e
r
i
c
a
n
U
n
i
v
e
r
si
t
y
,
W
a
sh
i
n
g
t
o
n
,
U
S
A
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
u
l
26
,
2
0
2
4
R
ev
is
ed
Feb
24
,
2
0
2
5
Acc
ep
ted
Mar
26
,
2
0
2
5
Ac
c
u
ra
te
lu
n
g
d
ise
a
se
d
iag
n
o
sis
in
in
fe
c
ted
p
a
ti
e
n
ts
is
c
rit
ica
l
fo
r
e
ffe
c
ti
v
e
trea
tme
n
t.
Tu
b
e
rc
u
lo
sis
,
COV
I
D
-
1
9
,
p
n
e
u
m
o
n
ia,
a
n
d
lu
n
g
o
p
a
c
it
y
a
re
in
fe
c
ti
o
u
s
lu
n
g
d
ise
a
se
s
with
v
isu
a
ll
y
sim
il
a
r
c
h
e
st
X
-
ra
y
p
re
s
e
n
tatio
n
s.
Hu
m
a
n
e
x
p
e
rti
se
c
a
n
b
e
su
sc
e
p
t
ib
le
t
o
e
rr
o
rs
d
u
e
to
fa
ti
g
u
e
o
r
e
m
o
ti
o
n
a
l
fa
c
to
rs.
Th
is
re
se
a
rc
h
p
ro
p
o
se
s
a
re
a
l
-
ti
m
e
d
e
e
p
lea
rn
in
g
-
b
a
se
d
c
la
ss
ifi
c
a
ti
o
n
sy
ste
m
fo
r
l
u
n
g
d
ise
a
se
s.
Th
re
e
m
o
d
e
ls
o
f
c
o
n
v
o
l
u
ti
o
n
a
l
n
e
u
ra
l
n
e
two
r
k
s
(CNN
s)
we
re
d
e
p
lo
y
e
d
to
c
las
sify
l
u
n
g
i
ll
n
e
ss
e
s
fro
m
c
h
e
st
X
-
ra
y
ima
g
e
s
:
M
o
b
i
leN
e
tV3
,
Re
sN
e
t
-
5
0
,
a
n
d
I
n
c
e
p
ti
o
n
V3
.
To
e
v
a
l
u
a
te
th
e
e
ffe
c
t
o
f
h
i
g
h
in
terc
las
s
sim
il
a
rit
y
,
th
e
m
o
d
e
ls
we
re
e
v
a
lu
a
ted
in
3
-
c
las
s
(Tu
b
e
rc
u
lo
sis,
COV
ID
-
1
9
,
n
o
rm
a
l),
4
-
c
las
s
(lu
n
g
o
p
a
c
it
y
,
t
u
b
e
rc
u
lo
sis,
C
OV
ID
-
1
9
,
n
o
rm
a
l),
a
n
d
5
-
c
las
s
(tu
b
e
rc
u
l
o
sis,
lu
n
g
o
p
a
c
it
y
,
p
n
e
u
m
o
n
ia,
C
OV
ID
-
1
9
,
n
o
rm
a
l)
m
o
d
e
s.
T
h
e
b
e
st
c
las
sifica
ti
o
n
a
c
c
u
ra
c
y
wa
s
a
tt
a
in
e
d
b
y
re
train
in
g
M
o
b
i
leN
e
tV3
,
wh
ich
o
b
tain
e
d
9
4
%
a
n
d
9
3
.
5
%
fo
r
5
-
c
las
s
a
n
d
4
-
c
las
s
,
re
sp
e
c
ti
v
e
ly
.
In
c
e
p
ti
o
n
V
3
h
a
d
th
e
lo
we
st
a
c
c
u
ra
c
y
(
9
0
%
,
8
9
%
,
9
3
%
fo
r
5
-
,
4
-
,
a
n
d
3
-
c
las
s),
wh
il
e
Re
sN
e
t
-
5
0
p
e
rfo
rm
e
d
b
e
st
fo
r
th
e
3
-
c
las
s
se
tt
in
g
.
Th
e
se
fin
d
i
n
g
s
su
g
g
e
st
M
o
b
il
e
N
e
tV3
'
s
p
o
ten
ti
a
l
fo
r
a
c
c
u
ra
te
lu
n
g
d
ise
a
se
d
iag
n
o
sis
fro
m
c
h
e
st
X
-
ra
y
s
d
e
s
p
it
e
t
h
e
in
terc
las
s
sim
il
a
rit
y
,
su
p
p
o
rti
n
g
th
e
a
d
o
p
t
io
n
o
f
c
o
m
p
u
ter
-
a
id
e
d
d
e
tec
ti
o
n
sy
ste
m
s fo
r
l
u
n
g
d
ise
a
se
c
las
sifica
ti
o
n
.
K
ey
w
o
r
d
s
:
C
h
est X
-
r
ay
C
las
s
if
icatio
n
C
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
s
Diag
n
o
s
is
L
u
n
g
d
is
ea
s
e
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
:
Ken
n
ed
y
Ok
o
k
p
u
jie
Dep
ar
tm
en
t o
f
E
lectr
ical
an
d
I
n
f
o
r
m
atio
n
E
n
g
in
ee
r
in
g
,
C
o
lleg
e
o
f
E
n
g
in
ee
r
i
n
g
,
C
o
v
e
n
an
t
Un
iv
er
s
ity
Ota,
Nig
er
ia
E
m
ail:
k
en
n
ed
y
.
o
k
o
k
p
u
jie@
co
v
en
an
t
u
n
iv
er
s
ity
.
ed
u
.
n
g
1.
I
NT
RO
D
UCT
I
O
N
L
u
n
g
d
is
ea
s
es
ar
e
also
k
n
o
wn
as
p
u
lm
o
n
ar
y
d
is
ea
s
es,
an
d
th
e
ter
m
is
u
s
ed
to
d
escr
ib
e
a
co
n
d
itio
n
o
r
a
g
r
o
u
p
o
f
c
o
n
d
itio
n
s
th
at
af
f
ec
ts
th
e
lu
n
g
s
an
d
m
a
k
es
th
e
m
m
o
r
e
v
u
ln
e
r
ab
le
to
m
ed
ical
in
ju
r
ies
[
1
]
.
I
t
also
af
f
ec
ts
th
e
ab
ilit
y
o
f
th
e
lu
n
g
s
to
f
u
n
ctio
n
p
r
o
p
er
ly
.
T
h
ese
co
n
d
itio
n
s
ca
n
af
f
ec
t
v
a
r
io
u
s
p
ar
ts
o
f
th
e
lu
n
g
s
,
s
u
ch
as
th
e
b
r
o
n
c
h
i,
b
r
o
n
ch
i
o
l
es,
alv
eo
li
an
d
lu
n
g
tis
s
u
e.
B
ec
au
s
e
o
f
h
o
w
c
o
m
p
licated
th
e
s
e
d
is
ea
s
es
ar
e
an
d
th
eir
s
im
ilar
ities
,
s
y
m
p
to
m
s
alo
n
e
ca
n
n
o
t
b
e
u
s
ed
to
d
iag
n
o
s
e
th
em
.
Var
io
u
s
m
et
h
o
d
s
,
s
u
ch
as
m
a
g
n
etic
im
ag
in
g
r
eso
n
a
n
ce
(
MI
R
)
,
co
m
p
u
ted
em
is
s
io
n
to
m
o
g
r
ap
h
y
(
C
E
T
)
,
p
o
s
itro
n
em
is
s
io
n
to
m
o
g
r
ap
h
y
,
a
n
d
ch
est
X
-
r
ay
s
,
ar
e
u
s
ed
to
an
aly
ze
l
u
n
g
d
is
ea
s
es
[
2
]
.
Ho
wev
er
,
C
h
est X
-
r
ay
s
ar
e
m
o
r
e
co
m
m
o
n
b
ec
au
s
e
th
ey
ar
e
less
ex
p
en
s
iv
e
a
n
d
r
ea
d
ily
av
ailab
l
e
in
m
o
s
t
h
ea
lth
ca
r
e
ce
n
ter
s
.
L
u
n
g
d
is
ea
s
es
h
av
e
b
ee
n
a
s
ig
n
i
f
ican
t
th
r
ea
t
to
th
e
h
ea
lth
o
f
h
u
m
a
n
s
f
o
r
as
lo
n
g
as
th
ey
h
av
e
ex
is
ted
.
T
h
ese
d
is
ea
s
es
af
f
ec
t
in
d
iv
id
u
als
o
f
all
ag
es;
h
en
ce
,
th
ey
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4
7
5
2
C
la
s
s
i
fica
tio
n
mo
d
el
fo
r
in
fectio
u
s
lu
n
g
d
is
ea
s
es u
s
in
g
…
(
K
en
n
ed
y
Oko
k
p
u
jie
)
411
ar
e
a
g
lo
b
al
c
o
n
ce
r
n
.
T
h
e
y
ar
e
a
s
ig
n
if
ican
t
ca
u
s
e
o
f
m
o
r
tal
ity
an
d
m
o
r
b
id
ity
i
n
th
e
wo
r
ld
,
af
f
ec
tin
g
m
ajo
r
ly
lo
w
-
in
co
m
e
a
n
d
m
id
d
le
-
i
n
co
m
e
co
u
n
tr
ies
[
3
]
.
T
h
e
s
ev
er
ity
o
f
lu
n
g
d
is
o
r
d
er
s
v
a
r
ies,
r
an
g
in
g
f
r
o
m
m
in
o
r
/
s
elf
-
lim
itin
g
s
y
m
p
to
m
s
,
s
u
ch
as
in
f
lu
en
za
an
d
th
e
c
o
m
m
o
n
co
ld
,
to
life
-
th
r
ea
ten
in
g
o
n
es,
s
u
ch
as
b
ac
ter
ial
p
n
eu
m
o
n
ia,
l
u
n
g
ca
n
ce
r
,
lu
n
g
o
p
ac
ity
,
asth
m
a
,
an
d
T
B
.
[
4
]
.
T
h
ese
co
n
d
itio
n
s
d
is
tu
r
b
th
e
t
is
s
u
es
an
d
air
way
s
o
f
th
e
lu
n
g
s
an
d
m
a
k
e
it
m
o
r
e
d
if
f
icu
lt
f
o
r
th
e
lu
n
g
s
to
f
u
n
ctio
n
n
o
r
m
ally
,
r
esu
ltin
g
in
s
er
io
u
s
h
ea
lth
is
s
u
es
[
5
]
,
[
6
]
.
I
n
th
e
co
n
te
x
t
o
f
g
lo
b
al
h
ea
lth
,
th
e
s
u
s
tain
ab
le
d
ev
elo
p
m
en
t
g
o
als
(
SD
Gs)
em
p
h
asized
th
e
im
p
o
r
tan
ce
o
f
ad
d
r
ess
in
g
h
ea
lth
-
r
elate
d
ch
allen
g
es.
So
m
e
o
f
th
e
SDGs
,
s
u
ch
as
h
ea
lth
an
d
well
-
b
ein
g
,
ar
e
d
ir
ec
tly
af
f
ec
ted
b
y
lu
n
g
d
is
e
ases
b
ec
au
s
e
th
ey
ar
e
a
m
ajo
r
ca
u
s
e
o
f
m
o
r
tality
an
d
m
o
r
b
id
ity
g
l
o
b
ally
,
an
d
th
ey
co
n
tr
ib
u
te
to
t
h
e
in
c
r
ea
s
in
g
b
u
r
d
e
n
o
f
n
o
n
-
co
m
m
u
n
ica
b
le
d
is
ea
s
es
wo
r
ld
wid
e.
H
o
wev
er
,
lu
n
g
d
is
ea
s
es
af
f
ec
t
m
o
r
e
th
a
n
j
u
s
t
a
n
i
n
d
iv
i
d
u
al'
s
h
ea
lth
;
th
ey
also
im
p
ac
t
h
ea
lth
ca
r
e
s
y
s
tem
s
,
ec
o
n
o
m
i
c
p
r
o
d
u
ctiv
ity
,
an
d
s
o
ciety
's
well
-
b
ein
g
.
T
h
e
n
atio
n
s
an
d
g
lo
b
al
o
r
g
an
izatio
n
s
m
u
s
t
wo
r
k
to
g
eth
er
to
p
r
e
v
en
t
,
d
iag
n
o
s
e,
an
d
tr
ea
t
lu
n
g
d
is
ea
s
es
to
m
ee
t
SDG
3
.
B
y
u
n
d
e
r
s
tan
d
in
g
th
e
r
ela
tio
n
s
h
ip
b
etwe
en
R
esp
ir
ato
r
y
h
e
alth
an
d
s
u
s
tain
ab
l
e
d
ev
elo
p
m
e
n
t
o
b
jectiv
es,
we
c
an
p
r
o
v
id
e
an
d
im
p
lem
en
t
s
y
s
tem
s
an
d
s
tr
u
ctu
r
es
to
d
ev
elo
p
a
h
ea
lth
ier
f
u
t
u
r
e
f
o
r
ev
e
r
y
b
o
d
y
.
R
ec
en
tly
,
co
m
p
u
ter
-
aid
ed
d
ia
g
n
o
s
is
(
C
AD)
s
y
s
tem
s
h
av
e
b
ec
o
m
e
h
an
d
y
i
n
d
etec
tin
g
an
d
m
an
ag
in
g
lu
n
g
d
is
ea
s
es
[
7
]
,
[
8
]
.
T
h
ey
h
av
e
r
ev
o
lu
tio
n
ized
th
e
p
r
o
ce
s
s
o
f
an
aly
zin
g
m
ed
ical
im
ag
e
s
s
u
ch
as
ch
e
s
t
X
-
r
ay
s
.
X
-
r
ay
s
ar
e
v
er
y
u
s
ef
u
l
i
n
m
ed
ical
im
ag
in
g
b
ec
au
s
e
t
h
ey
p
r
o
v
id
e
d
etailed
in
s
ig
h
ts
in
to
th
e
an
ato
m
ical
s
tr
u
ctu
r
es
o
f
th
e
lu
n
g
s
,
m
ak
i
n
g
it
ea
s
ier
to
o
b
tain
a
m
o
r
e
p
r
ec
is
e
id
e
n
tific
atio
n
o
f
a
b
n
o
r
m
alities
s
u
ch
as
n
o
d
u
les,
o
p
ac
ity
an
d
o
th
er
p
ath
o
lo
g
ical
ch
a
n
g
es.
B
y
u
s
in
g
d
ig
ital
s
ig
n
al
p
r
o
ce
s
s
in
g
tech
n
iq
u
es
f
o
r
im
ag
e
f
ilter
in
g
an
d
n
o
is
e
r
ed
u
ctio
n
,
in
co
n
ju
n
ctio
n
with
d
ig
ital
im
ag
e
p
r
o
ce
s
s
in
g
te
ch
n
iq
u
es,
th
ese
im
ag
es
ar
e
th
en
en
h
an
ce
d
an
d
an
aly
s
ed
to
p
r
o
v
id
e
b
etter
v
is
u
als
o
f
lu
n
g
t
is
s
u
es.
B
y
in
teg
r
atin
g
d
ig
ital
s
ig
n
als
an
d
d
ig
ital
im
ag
e
p
r
o
ce
s
s
in
g
in
to
C
AD
s
y
s
tem
s
,
r
ad
io
lo
g
is
ts
an
d
o
th
er
m
ed
ical
p
r
ac
titi
o
n
er
s
ca
n
b
e
n
ef
it
f
r
o
m
ac
c
u
r
ate
an
d
tim
ely
d
iag
n
o
s
is
.
Als
o
,
th
ey
ar
e
h
elp
in
id
e
n
tify
in
g
s
u
b
tle
ab
n
o
r
m
alities
an
d
th
is
r
esu
lts
in
ea
r
ly
d
iag
n
o
s
es a
n
d
im
p
r
o
v
em
en
t in
p
atien
t o
u
tco
m
es.
Me
d
ical
im
ag
in
g
tech
n
o
lo
g
ies
,
s
u
ch
as
X
-
r
ay
s
an
d
u
ltra
s
o
u
n
d
s
,
h
a
v
e
alter
e
d
th
e
p
r
o
ce
s
s
o
f
m
e
d
ical
d
iag
n
o
s
is
,
allo
wi
n
g
m
ed
ical
s
p
ec
ialis
ts
to
o
b
s
er
v
e
th
e
in
s
id
e
s
tr
u
ctu
r
es
o
f
th
e
h
u
m
an
b
o
d
y
n
o
n
-
in
v
asiv
ely
[
9
]
.
Fu
r
th
er
m
o
r
e
,
th
ese
p
h
o
to
g
r
a
p
h
s
g
iv
e
v
ital
in
s
ig
h
ts
in
to
th
e
in
ter
io
r
ar
c
h
itectu
r
e
o
f
t
h
e
h
u
m
a
n
b
o
d
y
b
y
p
h
y
s
ical
in
s
p
ec
tio
n
o
n
ly
[
1
0
]
.
I
t
g
en
er
ates
a
v
ast
v
o
lu
m
e
o
f
d
ata
f
o
r
co
m
p
u
ter
-
aid
e
d
d
iag
n
o
s
tic
s
y
s
tem
s
an
d
d
ee
p
lear
n
i
n
g
alg
o
r
ith
m
s
.
I
n
te
g
r
atin
g
d
ee
p
lear
n
in
g
with
m
e
d
ical
im
ag
in
g
y
ield
s
a
s
o
p
h
is
ticated
C
AD
s
y
s
tem
ca
p
ab
le
o
f
an
al
y
s
in
g
d
ata,
h
ig
h
lig
h
tin
g
r
e
g
io
n
s
with
an
o
m
alies
o
r
d
if
f
icu
lties
,
an
d
im
p
r
o
v
in
g
o
v
er
al
l
d
iag
n
o
s
is
ac
cu
r
ac
y
.
Sig
n
if
ica
n
t
p
r
o
g
r
ess
h
as
b
ee
n
r
ec
o
r
d
e
d
in
th
e
u
s
e
o
f
im
ag
in
g
tech
n
o
lo
g
y
in
m
ak
i
n
g
d
iag
n
o
s
is
an
d
o
th
e
r
clin
ical
d
ec
is
io
n
s
.
B
ec
au
s
e
th
ey
ar
e
wid
ely
av
ailab
le
an
d
n
o
n
-
in
v
asiv
e,
ch
est
X
-
r
ay
s
ar
e
v
ital
f
o
r
ass
ess
in
g
lu
n
g
-
r
elate
d
d
is
ea
s
es,
an
d
am
o
n
g
o
th
er
t
ec
h
n
o
lo
g
ies,
X
-
r
ay
s
ar
e
wid
el
y
p
o
p
u
lar
b
ec
a
u
s
e
th
ey
ar
e
r
ea
d
ily
av
ailab
le
ev
e
n
in
r
u
r
al
ar
ea
s
,
th
u
s
m
ak
in
g
i
t
ea
s
y
to
o
b
tain
s
ca
n
s
[
1
1
]
.
T
h
ese
s
ca
n
s
h
av
e
b
ee
n
u
s
ed
to
d
iag
n
o
s
e
s
ev
er
al
lu
n
g
-
r
elate
d
d
is
ea
s
es,
s
u
ch
as
C
OV
I
D
-
19
[
1
2
]
,
[
1
3
]
.
Dee
p
co
n
v
o
lu
tio
n
al
n
e
u
r
al
n
etwo
r
k
s
(
DC
NN)
ar
e
c
r
itical
f
o
r
au
to
m
atin
g
a
n
d
s
im
p
lify
in
g
th
e
d
iag
n
o
s
is
o
f
lu
n
g
illn
ess
es.
I
t
is
o
n
e
o
f
th
e
tactics
u
s
ed
in
m
ed
icin
e
to
s
o
lv
e
p
r
o
b
lem
s
.
A
d
ee
p
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
is
a
ty
p
e
o
f
d
ee
p
lear
n
i
n
g
m
o
d
el
th
at
ca
n
h
an
d
le
p
ictu
r
e
r
ec
o
g
n
itio
n
t
ask
s
well
d
u
e
to
its
ab
ilit
y
to
lear
n
.
T
h
is
m
o
d
el
ca
n
also
ex
tr
ac
t
th
e
h
ier
ar
ch
al
ch
ar
ac
ter
is
tics
f
r
o
m
r
aw
p
ictu
r
e
d
ata;
h
en
ce
,
DC
NN
s
h
av
e
b
ec
o
m
e
a
v
alu
ab
le
to
o
l
in
m
ed
ical
im
a
g
e
an
aly
s
is
.
T
h
e
ar
tific
ial
n
eu
r
a
l
n
etwo
r
k
b
e
h
av
es
si
m
ilar
ly
to
a
h
u
m
an
in
th
at
i
t
m
im
ics
th
e
s
tr
u
ctu
r
e
an
d
f
u
n
ctio
n
o
f
th
e
h
u
m
an
co
r
tex
,
a
llo
win
g
it
to
lear
n
co
m
p
licated
p
atter
n
s
f
r
o
m
co
ll
ec
tio
n
s
o
f
m
e
d
ical
p
ictu
r
es.
T
h
e
in
itial
lay
er
s
o
f
th
e
DC
NN
d
etec
t
f
u
n
d
am
en
tal
elem
en
ts
s
u
ch
as
ed
g
es
an
d
f
o
r
m
s
.
Su
b
s
eq
u
en
t
lay
er
s
in
teg
r
ate
th
ese
tr
aits
to
r
ec
o
g
n
is
e
m
o
r
e
co
m
p
lex
p
atter
n
s
,
r
esu
ltin
g
in
o
b
ject
r
ec
o
g
n
itio
n
.
I
n
m
e
d
ical
im
ag
in
g
,
DC
NNs
ex
am
in
e
ch
est
X
-
r
ay
s
to
d
is
co
v
er
p
atter
n
s
an
d
an
o
m
alies
an
d
u
tili
s
e
th
i
s
to
d
iag
n
o
s
e
d
is
ea
s
e
s
s
u
ch
as
tu
b
er
cu
lo
s
is
an
d
lu
n
g
o
p
a
city
.
Sev
e
r
al
DC
NN
m
o
d
els
h
av
e
b
ee
n
b
u
i
lt
an
d
tr
ai
n
ed
t
o
r
ec
o
g
n
ize
lu
n
g
illn
ess
es
f
r
o
m
ch
est
X
-
r
a
y
s
.
C
h
ex
Net,
a
well
-
k
n
o
wn
m
o
d
el
d
ev
elo
p
ed
b
y
a
team
at
Stan
d
f
o
r
d
Un
iv
e
r
s
ity
,
was
tr
ain
ed
o
n
a
h
u
g
e
d
ataset
to
d
iag
n
o
s
e
s
ev
er
al
lu
n
g
d
is
ea
s
es
[
1
4
]
−
[
1
6
]
.
T
h
e
C
h
e
x
Net
m
o
d
el
em
p
lo
y
s
a
f
o
r
m
o
f
DC
NN
ar
ch
itectu
r
e
ca
lled
Den
s
eNe
t.
Sev
er
al
o
th
er
ex
am
p
les
o
f
DC
NN
m
o
d
els
h
av
e
ex
h
ib
ited
g
r
ea
t
ac
cu
r
ac
y
,
f
r
eq
u
e
n
tly
s
u
r
p
ass
in
g
s
tan
d
ar
d
d
iag
n
o
s
tic
m
eth
o
d
s
a
n
d
g
iv
in
g
a
v
ital
s
e
co
n
d
o
p
in
i
o
n
.
Desp
ite
th
e
ex
c
ellen
t
ac
cu
r
ac
y
g
ai
n
ed
in
t
h
e
s
tu
d
y
,
th
e
ch
allen
g
e
with
DC
NN
r
em
ain
s
th
at
it ta
k
es m
u
ltip
le
im
ag
es a
n
d
a
lo
n
g
tr
ain
in
g
tim
e
e
v
en
with
GP
U
s
u
p
p
o
r
t [
1
7
]
,
[
1
8
]
.
T
r
an
s
f
er
lear
n
in
g
is
a
m
ac
h
i
n
e
lear
n
in
g
tech
n
i
q
u
e
f
o
r
im
ag
e
class
if
icatio
n
th
at
u
s
es
th
e
i
n
f
o
r
m
atio
n
o
b
tain
ed
f
r
o
m
tr
ain
i
n
g
a
m
o
d
el
o
n
o
n
e
jo
b
to
e
n
h
an
ce
p
er
f
o
r
m
an
c
e
o
n
an
o
th
er
,
b
u
t
r
elate
d
task
.
I
t
is
ad
v
an
tag
e
o
u
s
wh
en
th
e
tar
g
et
task
h
as
lim
ited
tr
ain
in
g
d
ata,
s
in
ce
it
all
o
ws
th
e
m
o
d
el
to
b
en
ef
it
f
r
o
m
th
e
r
ic
h
f
ea
tu
r
e
r
ep
r
esen
tatio
n
s
lear
n
t o
n
a
b
i
g
g
er
d
ataset.
T
r
an
s
f
er
le
ar
n
in
g
h
as
e
m
er
g
ed
as
a
v
ital
tech
n
iq
u
e
in
m
ed
ical
im
ag
in
g
,
e
n
ab
lin
g
s
ig
n
if
ican
t
p
er
f
o
r
m
an
ce
in
cr
ea
s
es
in
task
s
lik
e
a
s
d
is
ea
s
e
d
etec
t
io
n
,
o
r
g
a
n
s
eg
m
en
tatio
n
,
a
n
d
m
ed
ical
im
ag
e
ca
teg
o
r
is
atio
n
.
T
r
a
n
s
f
er
lear
n
i
n
g
,
b
y
u
s
in
g
p
r
e
-
tr
ain
ed
m
o
d
e
ls
co
n
s
tr
u
cted
o
n
l
ar
g
e
-
s
ca
le
d
atasets
f
r
o
m
b
r
o
ad
d
o
m
ain
s
,
s
u
c
h
as
n
atu
r
al
p
h
o
t
o
s
o
r
o
th
e
r
m
e
d
ical
d
atasets
,
g
r
ea
tly
d
ec
r
ea
s
es
th
e
r
e
q
u
ir
em
e
n
t
f
o
r
b
ig
,
lab
elled
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
9
,
No
.
1
,
Ju
ly
20
25
:
4
1
0
-
4
2
4
412
m
ed
ical
d
atasets
,
wh
ich
ar
e
f
r
eq
u
en
tly
co
s
tly
an
d
tim
e
-
co
n
s
u
m
in
g
.
T
h
is
ap
p
r
o
ac
h
s
ig
n
if
ican
tly
ac
ce
ler
ates
th
e
tr
ai
n
in
g
p
r
o
ce
s
s
,
esp
ec
ially
f
o
r
task
s
with
lim
ited
d
ata.
T
r
an
s
f
er
lear
n
in
g
o
f
ten
lead
s
to
b
etter
r
esu
lts
o
n
th
e
n
ew
task
th
an
tr
ain
in
g
f
r
o
m
s
cr
atch
,
an
d
it
h
elp
s
m
in
im
ize
co
m
p
u
tatio
n
al
co
s
ts
,
as
tr
ain
in
g
f
r
o
m
s
cr
atch
ca
n
b
e
r
eso
u
r
ce
in
ten
s
iv
e.
I
n
a
r
ec
e
n
t
s
tu
d
y
L
o
ey
el
a
t.
[
1
9
]
p
r
o
p
o
s
ed
u
s
in
g
a
g
en
er
ati
v
e
ad
v
er
s
ar
ial
n
etwo
r
k
(
GAN)
with
d
ee
p
tr
an
s
f
er
lear
n
in
g
to
d
etec
t
C
OVI
D
-
1
9
in
ch
est
X
-
r
ay
im
a
g
es,
o
v
er
c
o
m
in
g
th
e
lack
o
f
d
atasets
b
y
s
y
n
th
esis
in
g
n
ew
im
ag
es.
T
h
e
d
ataset
co
n
tain
ed
3
0
7
p
h
o
to
g
r
ap
h
s
f
r
o
m
f
o
u
r
clas
s
es.
T
h
r
ee
d
ee
p
tr
an
s
f
er
s
m
o
d
els
-
Alex
Net,
Go
o
g
L
eNe
t,
a
n
d
R
esNet1
8
-
wer
e
in
v
esti
g
ated
t
o
m
i
n
im
is
e
co
m
p
lex
ity
an
d
ex
ec
u
tio
n
tim
e.
T
h
e
s
tu
d
y
in
v
esti
g
ated
th
r
ee
s
ce
n
ar
io
s
with
v
ar
y
in
g
class
n
u
m
b
e
r
s
.
T
esti
n
g
ac
cu
r
ac
y
was
th
e
k
ey
m
etr
ic,
with
Go
o
g
L
eNe
t
ac
h
iev
in
g
t
h
e
h
ig
h
est
ac
cu
r
ac
y
o
f
8
0
.
6
%
(
f
ir
s
t
s
ce
n
ar
io
)
,
8
5
.
2
%
(
s
ec
o
n
d
s
ce
n
ar
io
)
,
an
d
1
0
0
%
(
th
ir
d
s
ce
n
ar
io
)
.
T
h
is
r
esear
ch
Ma
lik
et
a
l.
[
2
0
]
h
ig
h
lig
h
ted
t
h
e
g
r
o
win
g
im
p
o
r
tan
ce
o
f
m
e
d
ical
im
ag
in
g
,
p
ar
ticu
lar
ly
ch
est
X
-
r
ay
s
,
i
n
h
ea
lth
ca
r
e.
R
esear
ch
er
s
u
tili
ze
d
c
h
est
X
-
r
ay
d
atasets
to
d
e
v
elo
p
AI
m
o
d
els
th
at
ca
n
ac
cu
r
ately
d
etec
t
C
OVI
D
-
1
9
,
p
n
eu
m
o
n
ia,
an
d
h
ea
lth
y
ca
s
es.
Ad
v
an
ce
m
en
ts
in
tech
n
o
lo
g
y
h
av
e
s
ig
n
if
ican
tly
im
p
r
o
v
e
d
th
e
ac
cu
r
ac
y
an
d
e
f
f
icien
cy
o
f
m
ed
ical
d
iag
n
o
s
es,
m
ak
in
g
th
ese
im
a
g
in
g
tec
h
n
iq
u
es
c
r
u
cial
f
o
r
d
is
ea
s
e
d
etec
tio
n
an
d
clin
ica
l
d
ec
is
io
n
-
m
ak
in
g
.
I
n
a
d
d
iti
o
n
,
Sh
ib
ly
et
a
l.
[
2
1
]
o
b
tain
ed
a
class
if
i
ca
tio
n
ac
cu
r
ac
y
o
f
9
7
.
3
6
%,
a
s
en
s
itiv
ity
o
f
9
7
.
6
5
%,
an
d
a
p
r
ec
is
i
o
n
o
f
9
9
.
2
8
%
in
id
e
n
tify
in
g
p
er
s
o
n
s
im
p
ac
ted
b
y
C
OVI
D
-
1
9
,
h
ig
h
lig
h
tin
g
th
e
p
o
ten
tial
o
f
th
ese
m
eth
o
d
o
l
o
g
ies
to
h
elp
h
ea
lth
ca
r
e
p
r
o
f
ess
io
n
als
v
alid
ate
th
eir
ev
alu
atio
n
s
.
T
h
e
c
o
n
tr
ib
u
tio
n
o
f
th
is
s
tu
d
y
in
clu
d
es th
e
f
o
llo
win
g
:
−
C
o
llected
h
y
b
r
id
ch
est
X
-
r
ay
im
ag
es
f
o
r
lu
n
g
o
p
ac
ity
,
C
OVI
D
-
1
9
,
v
ir
al
p
n
eu
m
o
n
ia,
b
ac
te
r
ial
p
n
eu
m
o
n
ia,
h
ea
lth
y
in
d
i
v
id
u
als an
d
t
u
b
er
c
u
lo
s
is
.
−
I
m
p
r
o
v
ed
c
h
est
r
ad
io
g
r
ap
h
y
a
n
aly
s
is
b
y
d
ev
elo
p
in
g
a
m
o
d
el
f
o
r
lu
n
g
illn
ess
es
u
s
in
g
s
ev
er
al
s
o
u
r
ce
s
o
f
X
-
r
ay
im
ag
es.
−
I
n
teg
r
ated
t
h
e
cr
ea
ted
m
o
d
el
i
n
to
th
e
T
eleg
r
am
co
n
v
er
s
atio
n
b
o
t a
n
d
web
ap
p
licatio
n
.
2.
M
E
T
H
O
D
T
im
ely
an
d
ac
cu
r
ate
id
en
tifi
ca
tio
n
o
f
in
f
ec
tio
u
s
lu
n
g
d
is
o
r
d
er
s
is
cr
u
cial
f
o
r
p
r
o
v
i
d
in
g
ef
f
ec
tiv
e
p
atien
t
ca
r
e
a
n
d
m
a
n
ag
in
g
t
h
ese
co
n
d
itio
n
s
.
W
h
ile
ch
est
X
-
r
ay
s
o
f
f
er
a
n
ac
ce
s
s
ib
le
an
d
co
s
t
-
ef
f
ec
tiv
e
ap
p
r
o
ac
h
f
o
r
i
n
itial
ev
alu
atio
n
,
co
n
v
en
tio
n
al
d
iag
n
o
s
tic
tech
n
iq
u
es
lar
g
ely
d
ep
en
d
o
n
th
e
ex
p
er
tis
e
o
f
r
ad
io
lo
g
is
ts
,
wh
ich
ca
n
b
e
s
u
b
jectiv
e
an
d
tim
e
-
co
n
s
u
m
in
g
.
Fig
u
r
e
1
g
iv
es
th
e
f
r
a
m
ewo
r
k
o
f
th
e
C
NN
m
o
d
el
d
ev
elo
p
e
d
f
r
o
m
d
ata
ac
q
u
is
itio
n
to
m
o
d
el
ev
al
u
atio
n
.
Fig
u
r
e
1
.
C
o
n
ce
p
tu
al
f
r
a
m
ewo
r
k
o
f
t
h
e
d
e
v
elo
p
ed
m
o
d
el
I
n
th
is
wo
r
k
,
s
am
p
le
im
ag
es
o
f
lu
n
g
d
is
ea
s
es
wer
e
tak
en
f
r
o
m
ex
is
tin
g
Kag
g
le
r
ep
o
s
ito
r
y
[
2
2
]
,
[
2
3
]
an
d
p
u
b
lis
h
ed
jo
u
r
n
als
[
2
4
]
−
[
2
6
]
a
n
d
th
e
d
atasets
ar
e
g
r
o
u
p
ed
in
to
th
r
ee
.
T
h
e
f
ir
s
t g
r
o
u
p
o
f
d
atasets
was
f
r
o
m
a
p
u
b
lis
h
ed
wo
r
k
[
2
1
]
wh
ic
h
was
also
m
ad
e
av
ailab
le
o
n
K
ag
g
le
[
2
3
]
an
d
it
h
as
3
,
6
1
6
X
r
ay
im
ag
es
o
f
c
h
ests
in
f
ec
ted
b
y
C
OVI
D
-
1
9
c
o
n
tai
n
in
g
6
,
0
1
2
im
ag
es
o
f
lu
n
g
o
p
ac
ity
an
d
X
r
ay
im
a
g
es
o
f
1
0
,
1
9
2
n
o
r
m
al
(
n
o
n
-
in
f
ec
ted
)
c
h
ests
.
T
h
e
s
ec
o
n
d
g
r
o
u
p
o
f
d
atasets
co
n
tain
im
ag
es
o
f
ch
ests
X
r
ay
in
f
ec
ted
b
y
tu
b
e
r
cu
lo
s
is
d
is
ea
s
es
with
3
,
5
0
0
ch
est
im
a
g
es
[
2
2
]
,
[
2
3
]
.
T
h
e
th
ir
d
g
r
o
u
p
o
f
d
atasets
u
s
ed
in
th
is
wo
r
k
was
o
b
tain
e
d
f
r
o
m
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4
7
5
2
C
la
s
s
i
fica
tio
n
mo
d
el
fo
r
in
fectio
u
s
lu
n
g
d
is
ea
s
es u
s
in
g
…
(
K
en
n
ed
y
Oko
k
p
u
jie
)
413
Kag
g
le
r
ep
o
s
ito
r
y
a
n
d
a
p
u
b
l
is
h
ed
wo
r
k
.
So
,
a
to
tal
o
f
1
9
,
0
0
0
im
a
g
e
d
atasets
wer
e
em
p
lo
y
ed
,
w
h
ich
wer
e
d
iv
id
ed
i
n
to
f
i
v
e
ch
est
p
ictu
r
e
g
r
o
u
p
s
:
C
OVI
D
-
1
9
,
l
u
n
g
o
p
ac
ity
,
T
B
,
p
n
eu
m
o
n
ia,
a
n
d
n
o
r
m
al
c
h
est.
E
ac
h
g
r
o
u
p
h
as
3
,
8
0
0
d
ata
s
am
p
les.
As
with
a
v
iab
le
m
ac
h
in
e
lear
n
in
g
tech
n
iq
u
e,
th
e
to
tal
d
at
asets
wer
e
d
iv
id
ed
in
to
7
5
%
f
o
r
th
e
tr
ain
in
g
,
2
0
%
f
o
r
th
e
test
in
g
an
d
5
%
f
o
r
th
e
v
alid
atio
n
o
f
th
e
n
ew
m
o
d
el.
T
h
e
d
is
tr
ib
u
tio
n
o
f
th
ese
d
atasets
f
u
r
th
er
s
h
o
wn
in
T
ab
le
1
.
T
ab
le
1
.
Descr
ip
tio
n
o
f
d
ataset
C
l
a
s
ses
To
t
a
l
n
u
m
b
e
r
o
f
C
X
I
Tr
a
i
n
i
n
g
se
t
V
a
l
i
d
a
t
i
o
n
se
t
Te
st
i
n
g
s
e
t
Tu
b
e
r
c
u
l
o
s
i
s
3
8
0
0
3
0
4
0
3
2
0
3
2
0
P
n
e
u
mo
n
i
a
3
8
0
0
3
0
4
0
3
2
0
3
2
0
Lu
n
g
o
p
a
c
i
t
y
3
8
0
0
3
0
4
0
3
2
0
3
2
0
C
O
V
I
D
-
19
3
8
0
0
3
0
4
0
3
2
0
3
2
0
N
o
r
mal
3
8
0
0
3
0
4
0
3
2
0
3
2
0
T
h
e
b
atch
n
o
r
m
alis
atio
n
alg
o
r
ith
m
f
r
o
m
k
er
as
T
en
s
o
r
Flo
w
was
u
s
ed
to
n
o
r
m
alis
e
th
e
i
n
p
u
t
d
ata
b
ef
o
r
e
p
r
ep
r
o
ce
s
s
in
g
th
e
d
atas
et.
E
ac
h
f
ea
tu
r
e
in
th
e
i
n
p
u
t
a
r
r
ay
is
n
o
r
m
alis
ed
to
a
r
an
g
e
o
f
v
alu
es
b
etwe
en
0
an
d
1
af
te
r
r
em
o
v
in
g
th
e
m
ea
n
v
alu
e
f
r
o
m
th
e
a
r
r
ay
a
n
d
d
i
v
id
in
g
th
e
s
tan
d
ar
d
d
ev
iatio
n
o
f
th
e
f
ea
tu
r
es
ac
r
o
s
s
th
e
b
atch
o
f
d
ata.
T
o
im
p
r
o
v
e
th
e
p
e
r
f
o
r
m
an
ce
a
n
d
r
eliab
ilit
y
o
f
th
e
m
o
d
el,
th
e
two
p
ar
am
eter
s
o
f
b
atch
n
o
r
m
aliza
tio
n
n
am
el
y
;
th
e
s
ca
le
f
ac
to
r
(γ)
an
d
s
h
if
t
f
ac
to
r
(
β)
p
ar
am
eter
s
wer
e
n
o
r
m
alize
d
d
u
r
in
g
th
e
d
a
ta
tr
ain
in
g
.
T
h
r
ee
c
u
ttin
g
-
ed
g
e
p
r
e
-
tr
ai
n
e
d
n
etwo
r
k
ar
ch
itectu
r
es,
Mo
b
ileNetV3
,
R
esNet5
0
,
an
d
I
n
c
ep
tio
n
V3
,
wer
e
em
p
lo
y
ed
to
e
x
tr
ac
t
d
ee
p
f
ea
tu
r
es
f
r
o
m
th
e
c
h
est
X
-
r
ay
d
ataset.
T
h
ese
n
etwo
r
k
m
o
d
els
wer
e
f
ir
s
t
p
r
e
-
tr
ain
ed
u
s
in
g
t
h
e
I
m
a
g
eNe
t
d
ataset,
wh
ich
co
m
p
r
is
es
o
n
e
m
illi
o
n
p
ictu
r
es
f
r
o
m
1
0
0
0
cl
ass
if
icatio
n
s
.
I
n
th
is
s
tu
d
y
,
th
e
c
h
est
X
-
r
ay
p
ictu
r
e
s
wer
e
in
p
u
t
s
ep
ar
atel
y
in
to
e
ac
h
o
f
th
e
p
r
e
-
tr
ain
e
d
n
etwo
r
k
s
to
ex
tr
ac
t
f
ea
tu
r
e
v
ec
to
r
s
f
r
o
m
th
e
f
u
lly
co
n
n
ec
t
ed
lay
er
s
.
All
tr
ai
n
in
g
a
n
d
im
p
lem
en
tatio
n
t
o
o
k
p
la
ce
in
th
e
J
u
p
y
ter
No
teb
o
o
k
en
v
ir
o
n
m
en
t,
u
s
in
g
Py
t
h
o
n
a
s
th
e
p
r
o
g
r
a
m
m
in
g
lan
g
u
ag
e.
T
h
e
Mo
b
ileNetV3
,
R
esNet5
0
,
an
d
I
n
ce
p
tio
n
V3
m
o
d
els
wer
e
d
ev
elo
p
ed
u
s
in
g
th
e
T
en
s
o
r
Flo
w
2
.
0
Ker
as
f
r
am
ewo
r
k
,
with
tr
ain
in
g
an
d
ex
p
er
im
en
tatio
n
d
o
n
e
o
n
th
e
Go
o
g
le
C
o
llab
p
latf
o
r
m.
I
n
ad
d
itio
n
,
th
e
v
a
r
io
u
s
d
atasets
u
s
ed
in
th
is
wo
r
k
wer
e
p
r
ep
r
o
ce
s
s
ed
to
a
co
m
m
o
n
s
ize
o
f
2
2
4
b
y
2
2
4
p
ix
el
f
o
r
Mo
b
ileNetV3
a
n
d
R
esNet5
0
m
o
d
els,
2
9
9
b
y
2
9
9
p
ix
els
f
o
r
t
h
e
I
n
ce
p
tio
n
V
3
Mo
d
el.
A
tr
an
s
f
er
lear
n
in
g
a
p
p
r
o
ac
h
was
em
p
lo
y
ed
with
t
h
e
Mo
b
ileNetV3
,
R
esNet5
0
an
d
I
n
ce
p
tio
n
V3
m
o
d
els
as
b
ase
m
o
d
els,
all
o
f
th
ese
wer
e
f
in
e
tu
n
e
d
to
ad
o
p
t
th
e
ch
est
X
r
ay
d
atasets
,
an
d
a
v
iew
o
f
t
h
is
co
n
ce
p
t
is
s
h
o
wn
in
Fig
u
r
e
2
.
T
h
e
f
ea
tu
r
e
ex
tr
ac
tio
n
s
f
r
o
m
th
e
d
atasets
an
d
its
f
in
e
tu
n
in
g
b
y
th
e
tr
an
s
f
er
lear
n
in
g
m
e
th
o
d
s
h
elp
s
in
th
e
p
er
f
o
r
m
an
ce
im
p
r
o
v
e
m
en
t.
T
h
e
m
o
d
el
was
tr
ain
ed
f
o
r
ea
ch
m
o
d
el'
s
f
iv
e
s
u
b
class
c
lass
if
icatio
n
s
(
tu
b
er
cu
lo
s
is
,
C
OVI
D
-
1
9
,
p
n
eu
m
o
n
ia,
an
d
lu
n
g
o
p
ac
ity
)
an
d
f
o
u
r
s
u
b
class
class
if
ica
t
io
n
s
(
C
OVI
D
-
1
9
,
p
n
e
u
m
o
n
ia,
lu
n
g
o
p
ac
ity
,
an
d
T
B
)
.
E
ac
h
m
o
d
el
h
as
a
b
atch
s
ize
o
f
6
4
,
a
d
r
o
p
o
u
t
o
f
0
.
6
,
a
lear
n
in
g
r
ate
o
f
0
.
0
0
1
,
g
lo
b
al
av
er
ag
e
p
o
o
lin
g
,
an
d
p
r
ec
is
e
class
if
icatio
n
u
s
in
g
th
e
Ad
am
o
p
tim
is
er
.
Fig
u
r
e
2
s
h
o
ws th
e
tr
an
s
f
er
lear
n
in
g
p
ath
way
.
Fig
u
r
e
2
.
T
r
an
s
f
er
lear
n
in
g
p
ip
elin
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
9
,
No
.
1
,
Ju
ly
20
25
:
4
1
0
-
4
2
4
414
T
h
e
m
o
d
els
wer
e
ev
alu
ated
b
a
s
ed
o
n
f
o
u
r
cr
iter
ia:
ac
cu
r
ac
y
,
r
ec
all,
p
r
ec
is
io
n
,
s
p
ec
if
icity
,
an
d
th
e
F1
s
co
r
e.
T
h
ey
a
r
e
d
escr
ib
e
d
:
=
+
+
+
+
(
1
)
=
+
+
(
2
)
=
+
(
3
)
1
=
2
∗
∗
+
(
4
)
wh
er
e:
T
P is
tr
u
e
p
o
s
itiv
e
,
T
N
is
tr
u
e
n
eg
ativ
e
,
FN is
f
alse n
eg
ativ
e
,
an
d
FP
is
f
alse p
o
s
itiv
e
.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
3
.
1
.
Resul
t
s
T
h
e
h
ig
h
e
r
th
e
d
iag
o
n
al
v
alu
es
in
th
e
co
n
f
u
s
io
n
m
atr
ix
,
th
e
m
o
r
e
ac
cu
r
ate
t
h
e
m
o
d
el'
s
p
r
ed
ictio
n
s
wer
e;
co
n
v
er
s
ely
,
t
h
e
lo
wer
th
e
o
f
f
-
d
iag
o
n
al
v
alu
es,
th
e
less
f
r
eq
u
en
tly
th
e
m
o
d
el
m
is
class
if
ied
th
e
d
ata.
T
h
e
co
n
f
u
s
io
n
m
atr
ix
c
o
m
p
r
e
h
en
s
iv
ely
ev
alu
ates
th
e
m
o
d
el'
s
class
if
icatio
n
p
er
f
o
r
m
a
n
ce
in
all
class
es.
T
h
is
s
h
o
ws
th
e
s
tr
en
g
th
s
an
d
wea
k
n
ess
es o
f
th
e
class
if
icatio
n
m
o
d
el
w
h
ile
p
r
ed
ictin
g
lu
n
g
d
is
ea
s
e.
3
.
1
.
1
.
3
-
s
ub
cla
s
s
cla
s
s
if
ica
t
io
n m
o
dels
T
h
is
s
ec
tio
n
ex
p
lo
r
es
th
e
I
n
ce
p
tio
n
V3
,
R
esNet
-
5
0
an
d
Mo
b
ileNetV3
p
er
f
o
r
m
an
ce
s
f
o
r
tr
a
in
in
g
an
d
v
alid
atio
n
o
f
lo
s
s
es
,
ac
cu
r
ac
ies
an
d
as
wel
l
as
co
n
f
u
s
io
n
m
atr
ix
v
is
u
ali
s
in
g
o
n
3
-
s
u
b
cl
ass
ef
f
ec
tiv
en
ess
in
class
if
y
in
g
p
u
lm
o
n
a
r
y
d
is
ea
s
es f
r
o
m
X
-
r
ay
s
in
a
3
-
s
u
b
class
s
ce
n
ar
io
(
C
OVI
D
-
1
9
,
n
o
r
m
al
an
d
tu
b
e
r
cu
lo
s
is
)
as
d
ep
icted
F
ig
u
r
es
3
-
5.
As
s
h
o
wn
in
Fig
u
r
es
3
(
a
)
,
4
(
a)
,
an
d
5
(
a)
,
I
n
ce
p
tio
n
V3
ac
h
iev
es
an
im
p
r
ess
iv
e
o
v
er
all
ac
cu
r
ac
y
o
f
9
3
%,
in
d
icatin
g
it
co
r
r
ec
tly
class
if
ied
9
3
%
o
f
th
e
s
am
p
les.
Ma
c
r
o
a
n
d
weig
h
ted
a
v
er
ag
es
f
o
r
p
r
ec
is
io
n
,
r
ec
all,
an
d
F1
-
s
co
r
e
all
r
ea
ch
0
.
9
3
,
s
u
g
g
esti
n
g
s
tr
o
n
g
p
er
f
o
r
m
a
n
ce
ac
r
o
s
s
all
clas
s
es,
ev
en
ac
co
u
n
tin
g
f
o
r
p
o
te
n
tial c
lass
im
b
alan
ce
s
.
Firstl
y
,
th
e
C
O
VI
D
class
h
as th
e
lo
west
s
co
r
e
(
8
8
% p
r
ec
is
io
n
,
9
5
%
r
ec
all,
an
d
9
1
%
F1
-
s
co
r
e)
;
i
n
d
icatin
g
th
at
th
e
m
o
d
el
m
a
y
h
av
e
s
o
m
e
d
if
f
icu
lty
in
ac
cu
r
ately
class
if
y
in
g
C
OVI
D
ca
s
e
s
co
m
p
ar
ed
to
th
e
o
th
er
class
es.
T
h
en
,
th
e
n
o
r
m
al
class
ac
h
iev
ed
m
o
d
er
atel
y
h
ig
h
s
co
r
es
(
9
3
%
p
r
ec
is
io
n
,
8
9
%
r
ec
all,
an
d
9
1
%
F1
s
co
r
e)
,
in
d
icatin
g
th
at
t
h
e
m
o
d
el
ca
n
r
ea
s
o
n
ab
ly
id
e
n
tify
n
o
r
m
al
ca
s
es.
Fin
ally
,
th
e
T
u
b
er
c
u
lo
s
is
class
h
as th
e
h
i
g
h
est (
9
9
% p
r
ec
is
io
n
,
9
5
% r
ec
all,
an
d
9
7
% f
1
-
s
co
r
e)
,
s
u
g
g
esti
n
g
th
at
th
e
m
o
d
el
m
ay
h
a
v
e
s
o
m
e
ch
a
llen
g
es
ac
cu
r
ately
class
if
y
in
g
n
o
r
m
al
ca
s
es
d
u
e
to
s
u
b
tle
v
ar
iatio
n
s
with
in
th
is
ca
teg
o
r
y
.
Fig
u
r
es
3
(
b
)
,
4
(
b
)
,
an
d
5
(
b
)
s
h
o
w
th
at
o
n
th
e
3
-
s
u
b
class
class
if
icatio
n
task
,
R
esNet5
0
p
er
f
o
r
m
s
im
p
r
ess
iv
ely
.
W
ith
an
o
v
e
r
all
ac
cu
r
ac
y
o
f
9
1
%,
th
e
m
o
d
el
co
r
r
ec
tly
i
d
en
tifie
s
9
1
%
o
f
t
h
e
s
am
p
les
in
th
e
d
ataset.
T
h
e
p
r
ec
is
io
n
,
r
ec
all,
an
d
F1
-
s
co
r
e
m
ac
r
o
an
d
weig
h
ted
av
er
a
g
es
all
s
u
r
p
ass
0
.
9
1
,
in
d
icatin
g
s
tr
o
n
g
p
er
f
o
r
m
an
ce
in
al
l
class
es.
T
h
e
weig
h
ted
a
v
er
ag
e
also
co
r
r
e
cts
f
o
r
an
y
p
o
ten
tial
im
b
alan
c
es
in
class
.
C
OVI
D
-
1
9
:
w
ith
s
co
r
es
o
f
9
2
%
p
r
ec
is
io
n
,
9
7
%
r
ec
all,
an
d
9
4
%
F1
-
s
co
r
e,
th
e
m
o
d
el
p
e
r
f
o
r
m
s
ex
ce
p
tio
n
ally
well
i
n
id
en
tify
in
g
C
OVI
D
-
1
9
in
s
tan
c
es.
lu
n
g
o
p
ac
ity
:
w
ith
8
9
% p
r
e
ci
s
io
n
,
9
0
% r
ec
all,
an
d
9
0
% F1
-
s
co
r
e,
th
e
m
o
d
el
p
er
f
o
r
m
s
well
in
lu
n
g
o
p
ac
ity
s
itu
atio
n
s
.
n
o
r
m
al:
w
ith
9
2
%
p
r
ec
is
io
n
,
8
7
%
r
ec
all,
an
d
8
9
%
F1
-
s
co
r
e,
t
h
e
m
o
d
el
p
er
f
o
r
m
s
th
e
s
ec
o
n
d
wo
r
s
t
o
n
n
o
r
m
al
ca
s
es.
T
h
is
s
u
g
g
ests
s
o
m
e
ch
allen
g
es
in
d
is
tin
g
u
is
h
in
g
s
u
b
tle
va
r
iatio
n
s
with
in
th
is
ca
teg
o
r
y
.
Mo
b
ileNetV3
r
esu
lts
f
o
r
3
-
s
u
b
class
class
if
icatio
n
ar
e
s
h
o
wn
in
Fig
u
r
es
3
(
c
)
,
4
(
c)
,
an
d
5
(
c)
.
T
h
e
o
v
er
all
ac
c
u
r
ac
y
o
f
9
0
%
in
d
icate
s
th
at
th
e
m
o
d
el
co
r
r
ec
tly
c
lass
if
ied
9
0
%
o
f
th
e
s
am
p
les
i
n
th
e
d
ataset.
B
o
th
m
ac
r
o
an
d
weig
h
te
d
av
er
ag
e
p
r
ec
is
io
n
,
r
ec
all,
an
d
F1
-
s
co
r
e
ar
e
all
0
.
9
0
,
s
u
g
g
esti
n
g
g
o
o
d
av
er
a
g
e
p
er
f
o
r
m
an
ce
ac
r
o
s
s
all
class
es,
with
th
e
weig
h
ted
av
e
r
a
g
e
ad
d
itio
n
ally
ac
co
u
n
tin
g
f
o
r
p
o
ten
tial
class
im
b
alan
ce
s
.
Firstl
y
,
th
e
m
o
d
e
l
p
er
f
o
r
m
ed
well
o
n
C
OVI
D
-
1
9
class
if
ic
atio
n
with
s
co
r
es
o
f
9
5
%
p
r
ec
is
io
n
,
9
2
%
r
ec
all,
an
d
9
3
%
F1
-
s
co
r
e
.
T
h
is
s
u
g
g
ests
a
g
o
o
d
ab
ilit
y
to
id
en
tify
C
OVI
D
-
1
9
ca
s
es.
Seco
n
d
ly
,
t
h
e
lu
n
g
o
p
ac
ity
class
ac
h
iev
e
d
m
o
d
e
r
a
tely
h
ig
h
s
co
r
es (
8
6
%
p
r
ec
is
io
n
,
8
9
%
r
ec
all,
an
d
8
8
%
F1
s
co
r
e
)
,
d
em
o
n
s
tr
atin
g
th
e
m
o
d
el'
s
ca
p
ab
ilit
y
to
id
en
tify
lu
n
g
o
p
ac
ity
ca
s
es
with
r
ea
s
o
n
ab
le
ac
c
u
r
ac
y
.
Fin
ally
,
th
e
m
o
d
el'
s
p
er
f
o
r
m
an
ce
in
th
e
n
o
r
m
al
cla
s
s
wa
s
th
e
s
ec
o
n
d
lo
west
(
8
9
%
p
r
ec
is
io
n
,
8
8
%
r
ec
all,
an
d
8
8
%
F1
s
co
r
e
)
.
T
h
is
in
d
icate
s
s
o
m
e
ch
allen
g
es
in
ac
cu
r
ately
class
if
y
in
g
n
o
r
m
al
ca
s
es,
p
o
ten
tially
d
u
e
to
s
u
b
tl
e
v
ar
iatio
n
s
with
in
th
is
ca
teg
o
r
y
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4
7
5
2
C
la
s
s
i
fica
tio
n
mo
d
el
fo
r
in
fectio
u
s
lu
n
g
d
is
ea
s
es u
s
in
g
…
(
K
en
n
ed
y
Oko
k
p
u
jie
)
415
(
a)
(
b
)
(
c)
Fig
u
r
e
3
.
C
h
ar
t
v
is
u
alizin
g
:
(
a
)
I
n
ce
p
tio
n
V3
,
(
b
)
R
esNet
-
50
,
an
d
(
c)
Mo
b
ileNetV3
p
er
f
o
r
m
an
ce
s
f
o
r
tr
ain
i
n
g
an
d
v
alid
atio
n
lo
s
s
es o
n
3
Su
b
class
(
a)
(
b
)
(
c)
Fig
u
r
e
4
.
C
h
ar
t
v
is
u
alizin
g
: (
a
)
I
n
ce
p
tio
n
V3
,
(
b
)
R
esNet
-
50
,
an
d
(
c)
Mo
b
ileNetV3
p
er
f
o
r
m
an
ce
s
f
o
r
tr
ain
i
n
g
an
d
v
alid
atio
n
ac
cu
r
ac
ies o
n
3
Su
b
class
(
a)
(
b
)
(
c)
Fig
u
r
e
5
.
C
o
n
f
u
s
io
n
m
atr
i
x
v
i
s
u
alizin
g
: (
a)
Mo
b
ileNetV3
,
(
b
)
R
esNet
-
5
0
an
d
(
c)
I
n
ce
p
tio
n
V3
p
er
f
o
r
m
a
n
ce
s
o
n
C
lass
if
y
in
g
3
Su
b
class
es
3
.
1
.
2
.
4
-
Su
bcla
s
s
cla
s
s
if
ica
t
io
n m
o
dels
T
h
is
s
ec
tio
n
ex
p
lo
r
es
th
e
I
n
ce
p
tio
n
V3
,
R
esNet
-
5
0
an
d
Mo
b
ileNetV3
p
er
f
o
r
m
an
ce
s
f
o
r
tr
a
in
in
g
an
d
v
alid
atio
n
o
f
lo
s
s
es
,
ac
cu
r
ac
i
es
an
d
as
well
as
co
n
f
u
s
io
n
m
atr
ix
v
is
u
alizin
g
o
n
4
Su
b
cl
ass
ef
f
ec
tiv
en
ess
in
class
if
y
in
g
p
u
lm
o
n
ar
y
d
is
ea
s
es
f
r
o
m
X
-
r
ay
s
i
n
a
4
-
s
u
b
class
s
ce
n
ar
io
(
C
OVI
D
-
1
9
,
l
u
n
g
o
p
ac
ity
,
n
o
r
m
al
an
d
tu
b
er
cu
lo
s
is
)
as
d
ep
icted
Fig
u
r
es
6
-
8.
As
s
h
o
wn
in
Fig
u
r
es
6
(
a)
,
7
(
a)
,
an
d
8
(
a)
,
t
h
e
ev
alu
atio
n
r
esu
lts
r
ev
ea
l
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
9
,
No
.
1
,
Ju
ly
20
25
:
4
1
0
-
4
2
4
416
p
r
o
m
is
in
g
p
e
r
f
o
r
m
an
ce
f
r
o
m
th
e
I
n
ce
p
tio
n
V
3
m
o
d
el
o
n
4
-
s
u
b
class
clas
s
if
icatio
n
.
T
h
e
o
v
er
all
ac
cu
r
ac
y
o
f
0
.
8
9
in
d
icate
s
th
at
th
e
m
o
d
el
co
r
r
ec
tly
class
if
ied
8
9
%
o
f
th
e
s
am
p
les.
B
o
th
m
ac
r
o
a
n
d
weig
h
ted
a
v
er
ag
es
f
o
r
p
r
ec
is
io
n
,
r
ec
all,
an
d
F1
-
s
co
r
e
ar
e
all
0
.
8
9
,
wh
ich
s
u
g
g
ests
t
h
at
th
e
m
o
d
el
p
er
f
o
r
m
s
well
o
n
av
er
a
g
e
ac
r
o
s
s
all
class
es.
Firstl
y
,
th
e
C
OVI
D
class
h
as
th
e
lo
we
s
t
s
co
r
e
(
8
9
%
p
r
ec
is
io
n
,
8
3
%
r
ec
all,
an
d
8
6
%
F1
-
s
co
r
e)
;
in
d
icatin
g
th
at
th
e
m
o
d
el
m
a
y
h
av
e
s
o
m
e
d
if
f
icu
lty
in
ac
c
u
r
ately
class
if
y
in
g
C
o
v
id
ca
s
es
co
m
p
ar
ed
to
t
h
e
o
th
er
class
es.
Seco
n
d
ly
,
th
e
lu
n
g
o
p
ac
ity
class
h
ad
th
e
f
o
llo
win
g
s
co
r
es:
8
9
%
p
r
ec
is
io
n
,
8
3
%
r
ec
all
an
d
8
6
%
F1
-
s
co
r
e.
T
h
en
,
th
e
n
o
r
m
al
cl
ass
ac
h
iev
ed
m
o
d
er
ately
h
ig
h
s
co
r
es
(
9
3
%
p
r
ec
is
io
n
,
8
9
%
r
ec
all,
an
d
9
1
%
F1
s
co
r
e)
,
in
d
icatin
g
t
h
at
th
e
m
o
d
el
ca
n
r
ea
s
o
n
ab
ly
id
en
tif
y
n
o
r
m
al
ca
s
es.
Fin
ally
,
th
e
T
u
b
er
c
u
lo
s
is
class
h
as
th
e
h
ig
h
est
(
9
7
%
p
r
ec
is
io
n
,
9
6
%
r
ec
all,
an
d
9
6
%
f
1
-
s
co
r
e)
,
s
u
g
g
esti
n
g
th
at
th
e
m
o
d
el
m
ay
h
av
e
s
o
m
e
ch
allen
g
es
ac
cu
r
ately
class
if
y
in
g
n
o
r
m
al
ca
s
es d
u
e
to
s
u
b
tle
v
ar
iatio
n
s
with
in
th
is
ca
teg
o
r
y
.
Fig
u
r
es
6
(
b
)
,
7
(
b
)
,
a
n
d
8
(
b
)
s
h
o
w
th
e
4
-
s
u
b
class
class
if
icatio
n
R
esNet
-
5
0
task
,
with
a
n
o
v
er
all
a
cc
u
r
ac
y
o
f
9
1
%,
wh
ich
in
d
ic
ates
th
at
th
e
m
o
d
el
co
r
r
ec
tly
class
if
ied
a
s
ig
n
if
ican
t
p
o
r
tio
n
o
f
th
e
s
am
p
les
in
th
e
d
ataset.
B
o
th
m
ac
r
o
an
d
w
eig
h
ted
av
er
a
g
es
f
o
r
p
r
ec
is
io
n
,
r
ec
all
an
d
F1
-
s
co
r
e
r
ea
ch
0
.
9
3
,
s
u
g
g
esti
n
g
g
o
o
d
av
er
ag
e
p
er
f
o
r
m
a
n
ce
ac
r
o
s
s
all
class
e
s
.
T
h
e
weig
h
ted
av
er
ag
e
ad
d
itio
n
ally
co
n
f
ir
m
s
th
is
with
p
o
ten
tial
class
im
b
alan
ce
s
co
n
s
id
er
ed
.
Firstl
y
,
th
e
C
OVI
D
-
1
9
class
if
icati
o
n
s
co
r
es
wer
e
9
8
%
p
r
ec
is
io
n
,
9
2
%
r
ec
all,
an
d
9
5
%
F1
-
s
co
r
e.
T
h
is
s
u
g
g
ests
a
g
o
o
d
a
b
ilit
y
to
id
en
tify
C
OVI
D
-
1
9
ca
s
es.
T
h
ir
d
ly
,
th
e
lu
n
g
o
p
ac
ity
class
ac
h
iev
ed
m
o
d
er
ately
h
ig
h
s
co
r
es
(
9
2
%
p
r
ec
is
io
n
,
9
0
%
r
ec
all,
an
d
9
1
%
F1
-
s
co
r
e)
,
d
em
o
n
s
tr
atin
g
th
e
m
o
d
el'
s
ca
p
ab
ilit
y
to
id
en
tify
lu
n
g
o
p
a
city
ca
s
es
w
ith
r
ea
s
o
n
ab
le
ac
cu
r
ac
y
.
Fo
u
r
t
h
ly
,
th
e
n
o
r
m
al
class
wa
s
th
e
s
ec
o
n
d
lo
west
(
8
8
%
p
r
ec
is
io
n
,
9
2
%
r
ec
all,
an
d
9
0
%
F1
-
s
co
r
e)
.
T
h
is
in
d
icate
s
s
o
m
e
ch
alle
n
g
es
in
ac
cu
r
atel
y
class
if
y
in
g
n
o
r
m
al
ca
s
es,
p
o
t
en
tially
d
u
e
to
s
u
b
tle
v
a
r
iatio
n
s
with
in
th
is
ca
teg
o
r
y
.
Fin
al
ly
,
th
e
tu
b
er
c
u
lo
s
is
class
ac
h
iev
ed
th
e
h
ig
h
est
s
c
o
r
es
(
9
7
%
p
r
ec
is
io
n
,
1
0
0
%
r
ec
all,
an
d
9
8
%
F1
-
s
co
r
e)
,
d
is
p
lay
in
g
t
h
e
m
o
d
el'
s
ef
f
ec
tiv
en
ess
in
id
en
tif
y
in
g
tu
b
er
cu
lo
s
is
ca
s
es.
T
h
e
e
v
alu
atio
n
r
esu
lts
d
em
o
n
s
tr
ate
p
r
o
m
is
in
g
p
er
f
o
r
m
an
ce
f
r
o
m
th
e
p
r
o
p
o
s
ed
m
o
d
el.
T
h
e
o
v
er
all
ac
c
u
r
ac
y
o
f
9
2
%
in
d
icate
s
th
at
th
e
m
o
d
el
co
r
r
ec
tly
class
if
ied
a
s
ig
n
if
ican
t
p
o
r
tio
n
o
f
th
e
s
am
p
les
in
th
e
d
ataset.
B
o
th
m
ac
r
o
an
d
weig
h
ted
av
er
a
g
es
f
o
r
p
r
ec
is
io
n
,
r
ec
all
an
d
F1
-
s
co
r
e
r
ea
ch
0
.
9
2
,
s
u
g
g
es
tin
g
g
o
o
d
av
er
a
g
e
p
e
r
f
o
r
m
a
n
ce
ac
r
o
s
s
all
class
es.
T
h
e
weig
h
ted
av
er
ag
e
ad
d
itio
n
ally
co
n
f
ir
m
s
th
is
with
p
o
ten
tial c
lass
im
b
alan
ce
s
c
o
n
s
id
er
ed
.
Mo
b
ileNetV3
r
esu
lts
f
o
r
4
-
s
u
b
class
class
if
icatio
n
ar
e
s
h
o
wn
in
Fig
u
r
es
6
(
c)
,
7
(
c)
,
an
d
8
(
c)
.
T
h
e
o
v
er
all
ac
cu
r
ac
y
o
f
9
1
%
in
d
i
ca
tes
th
at
th
e
m
o
d
el
co
r
r
ec
tly
class
if
ied
a
s
ig
n
if
ican
t
p
o
r
tio
n
o
f
th
e
s
am
p
les
in
th
e
d
ataset.
B
o
th
m
ac
r
o
an
d
w
eig
h
ted
av
er
a
g
es
f
o
r
p
r
ec
is
io
n
,
r
ec
all
an
d
F1
-
s
co
r
e
r
ea
ch
0
.
9
3
,
s
u
g
g
esti
n
g
g
o
o
d
av
er
ag
e
p
er
f
o
r
m
a
n
ce
ac
r
o
s
s
all
class
e
s
.
T
h
e
weig
h
ted
av
er
ag
e
ad
d
itio
n
ally
co
n
f
ir
m
s
th
is
with
p
o
ten
tial
class
im
b
alan
ce
s
co
n
s
id
er
ed
.
T
h
e
C
OVI
D
-
1
9
class
if
icatio
n
s
co
r
es
wer
e
9
8
%
p
r
ec
is
io
n
,
9
2
%
r
ec
all,
an
d
9
5
%
F1
-
s
co
r
e.
T
h
is
s
u
g
g
ests
a
g
o
o
d
ab
ilit
y
to
id
en
tify
C
OVI
D
-
1
9
ca
s
es.
T
h
ir
d
ly
,
th
e
lu
n
g
o
p
a
city
class
ac
h
iev
ed
m
o
d
er
ately
h
ig
h
s
co
r
es
(
9
2
%
p
r
ec
is
io
n
,
9
0
%
r
ec
all,
an
d
9
1
%
F1
-
s
co
r
e)
,
d
em
o
n
s
tr
atin
g
th
e
m
o
d
el'
s
ca
p
ab
ilit
y
to
id
en
tify
lu
n
g
o
p
ac
ity
ca
s
es
with
r
ea
s
o
n
ab
le
ac
cu
r
ac
y
.
F
o
u
r
th
ly
,
th
e
n
o
r
m
al
class
was
th
e
s
ec
o
n
d
lo
west
(
8
8
%
p
r
ec
is
io
n
,
9
2
%
r
ec
all,
an
d
9
0
%
F1
-
s
co
r
e)
.
T
h
is
in
d
icate
s
s
o
m
e
ch
allen
g
es
in
ac
cu
r
ately
class
if
y
in
g
n
o
r
m
al
ca
s
es,
p
o
ten
tially
d
u
e
t
o
s
u
b
tle
v
ar
iatio
n
s
with
in
th
is
c
ateg
o
r
y
.
Fin
ally
,
th
e
t
u
b
er
cu
l
o
s
is
class
ac
h
iev
ed
th
e
h
ig
h
est
s
co
r
es
(
9
7
%
p
r
ec
i
s
io
n
,
1
0
0
%
r
ec
all,
an
d
9
8
%
F1
s
co
r
e)
,
s
h
o
wca
s
in
g
th
e
m
o
d
el's
ef
f
ec
tiv
en
ess
in
id
en
tify
in
g
T
u
b
er
c
u
lo
s
is
ca
s
es
.
(
a)
(
b
)
(
c)
Fig
u
r
e
6
.
C
h
ar
t
v
is
u
alizin
g
: (
a
)
I
n
ce
p
tio
n
V3
,
(
b
)
R
esNet
-
50
,
an
d
(
c)
Mo
b
ileNetV3
p
er
f
o
r
m
an
ce
s
f
o
r
tr
ain
in
g
an
d
v
alid
atio
n
lo
s
s
es o
n
4
-
Su
b
class
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4
7
5
2
C
la
s
s
i
fica
tio
n
mo
d
el
fo
r
in
fectio
u
s
lu
n
g
d
is
ea
s
es u
s
in
g
…
(
K
en
n
ed
y
Oko
k
p
u
jie
)
417
(
a)
(
b
)
(
c)
Fig
u
r
e
7
.
C
h
ar
t V
is
u
alizin
g
: (
a
)
I
n
ce
p
tio
n
V3
,
(
b
)
R
esNet
-
5
0
an
d
(
c)
Mo
b
ileNetV3
p
er
f
o
r
m
an
ce
s
f
o
r
tr
ain
i
n
g
an
d
v
alid
atio
n
ac
cu
r
ac
ies o
n
4
-
Su
b
class
(
a)
(
b
)
(
c)
Fig
u
r
e
8
.
C
o
n
f
u
s
io
n
m
atr
i
x
v
i
s
u
alizin
g
: (
a)
Mo
b
ileNetV3
,
(
b
)
R
esNet
-
50
,
an
d
(
c)
I
n
ce
p
tio
n
V3
p
er
f
o
r
m
a
n
ce
s
on
c
lass
if
y
in
g
4
s
u
b
class
es
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
9
,
No
.
1
,
Ju
ly
20
25
:
4
1
0
-
4
2
4
418
3
.
1
.
3
.
5
-
Su
bcla
s
s
cla
s
s
if
ica
t
io
n m
o
dels
T
h
is
s
ec
tio
n
ex
p
lo
r
es
th
e
I
n
ce
p
tio
n
V3
,
R
esNet
-
5
0
an
d
Mo
b
ileNetV3
p
er
f
o
r
m
an
ce
s
f
o
r
tr
a
in
in
g
an
d
v
alid
atio
n
o
f
lo
s
s
es,
ac
cu
r
ac
ies
as
well
as
co
n
f
u
s
io
n
m
a
tr
ix
v
is
u
alizin
g
o
n
5
-
Su
b
clas
s
ef
f
ec
tiv
en
ess
in
class
if
y
in
g
p
u
lm
o
n
ar
y
d
is
ea
s
es
f
r
o
m
X
-
r
ay
s
in
a
5
-
s
u
b
class
s
ce
n
ar
io
(
b
ac
ter
ial
p
n
eu
m
o
n
ia
,
C
OVI
D
-
1
9
,
lu
n
g
o
p
ac
ity
,
n
o
r
m
al
an
d
t
u
b
er
cu
l
o
s
is
)
as
d
ep
ic
ted
Fig
u
r
es
9
-
11.
As
s
h
o
wn
in
Fig
u
r
es
9
(
a
)
,
1
0
(
a
)
,
an
d
1
1
(
a
)
,
th
e
ev
alu
atio
n
r
esu
lts
r
ev
ea
l
p
r
o
m
is
in
g
p
er
f
o
r
m
a
n
ce
f
r
o
m
th
e
I
n
ce
p
tio
n
V3
m
o
d
el
o
n
5
-
s
u
b
class
cla
s
s
if
icat
io
n
.
T
h
e
ev
al
u
atio
n
'
s
f
in
d
in
g
s
s
h
o
w
th
at
th
e
I
n
ce
p
tio
n
V3
m
o
d
el
p
e
r
f
o
r
m
s
ad
m
ir
a
b
ly
.
T
h
e
m
o
d
e
l
p
r
o
p
e
r
ly
id
en
tifie
d
th
e
s
am
p
les
with
a
n
o
v
er
all
ac
cu
r
ac
y
o
f
9
0
%.
T
h
e
m
o
d
el
p
er
f
o
r
m
s
well
a
cr
o
s
s
a
ll
class
es,
with
th
e
weig
h
ted
av
er
a
g
e
co
n
s
id
er
in
g
class
im
b
alan
ce
s
.
T
h
e
m
ac
r
o
an
d
weig
h
ted
av
e
r
ag
e
p
r
ec
is
io
n
,
r
ec
all,
an
d
F1
-
s
co
r
e
ar
e
all
0
.
9
0
.
Firstl
y
,
th
e
Pn
eu
m
o
n
ia
class
ac
h
iev
ed
th
e
h
ig
h
est
s
co
r
es
(
9
8
%
p
r
ec
is
io
n
,
9
8
%
r
ec
all
an
d
9
8
%
F1
-
s
co
r
e)
,
in
d
icatin
g
th
at
th
e
m
o
d
el
ca
n
ac
c
u
r
ately
i
d
en
tify
Pn
eu
m
o
n
ia
ca
s
es.
Seco
n
d
ly
,
t
h
e
C
OVI
D
class
h
as
a
s
lig
h
tly
lo
wer
s
co
r
e
(
9
1
% p
r
ec
is
io
n
,
8
4
%
r
ec
all
an
d
8
8
%
F1
-
s
co
r
e)
,
in
d
icatin
g
th
at
th
e
m
o
d
el
m
ay
h
av
e
d
if
f
icu
lty
ac
c
u
r
ately
class
if
y
in
g
C
OVI
D
ca
s
e
s
co
m
p
ar
ed
to
th
e
o
th
er
class
es.
T
h
ir
d
ly
,
th
e
lu
n
g
o
p
ac
ity
class
ac
h
iev
ed
m
o
d
er
ately
h
i
g
h
s
co
r
es
(
8
4
%
p
r
ec
is
io
n
,
8
3
%
r
ec
all
an
d
8
3
%
F1
-
s
co
r
e)
,
in
d
icatin
g
th
at
t
h
e
m
o
d
el
ca
n
r
ea
s
o
n
ab
ly
id
en
tif
y
lu
n
g
o
p
ac
ity
ca
s
es.
Fo
u
r
t
h
ly
,
th
e
n
o
r
m
al
class
h
as
th
e
s
ec
o
n
d
lo
west
(
8
2
%
p
r
ec
is
io
n
,
8
6
%
r
ec
all
a
n
d
8
4
%
f
1
-
s
co
r
e)
,
s
u
g
g
esti
n
g
th
at
th
e
m
o
d
el
m
a
y
h
a
v
e
s
o
m
e
ch
allen
g
es
ac
cu
r
atel
y
class
if
y
in
g
n
o
r
m
al
ca
s
es
d
u
e
t
o
s
u
b
tle
v
ar
iatio
n
s
with
in
th
is
ca
teg
o
r
y
.
Fin
ally
,
th
e
T
u
b
e
r
c
u
lo
s
is
class
h
as
th
e
s
ec
o
n
d
h
ig
h
est
s
co
r
es
(
9
5
%
Pre
cisi
o
n
,
9
8
%
r
ec
all,
an
d
9
7
%
f
1
-
s
co
r
e)
,
in
d
icatin
g
th
at
th
e
m
o
d
el
ca
n
ef
f
ec
tiv
ely
id
en
tif
y
T
u
b
er
cu
lo
s
is
ca
s
e
s
.
Fig
u
r
es
9
(
b
)
,
1
0
(
b
)
,
an
d
1
1
(
b
)
s
h
o
w
th
e
5
-
s
u
b
class
class
i
f
icatio
n
R
esNet
-
5
0
task
;
th
e
ev
alu
atio
n
r
esu
lts
r
ev
ea
l
p
r
o
m
is
in
g
p
er
f
o
r
m
an
ce
f
r
o
m
th
e
p
r
o
p
o
s
ed
m
o
d
el.
T
h
e
o
v
er
all
ac
c
u
r
ac
y
o
f
0
.
9
3
i
n
d
icate
s
th
at
th
e
m
o
d
el
co
r
r
ec
tly
class
if
ied
9
3
%
o
f
th
e
s
am
p
les.
B
o
th
m
ac
r
o
an
d
weig
h
ted
av
er
a
g
e
p
r
ec
is
io
n
,
r
ec
all,
an
d
F1
-
s
co
r
e
ar
e
all
0
.
9
3
,
an
d
th
is
s
u
g
g
ests
th
at
th
e
m
o
d
el
p
er
f
o
r
m
s
well
o
n
av
er
ag
e
ac
r
o
s
s
all
class
es,
with
th
e
weig
h
ted
a
v
er
ag
e
ad
d
itio
n
ally
ac
co
u
n
tin
g
f
o
r
p
o
ten
tial
c
lass
im
b
alan
ce
s
.
Firstl
y
,
th
e
Pn
eu
m
o
n
ia
class
ac
h
iev
ed
t
h
e
h
ig
h
est
s
co
r
es
(
9
9
%
p
r
ec
is
io
n
,
9
8
%
r
ec
all
an
d
9
8
%
F1
-
s
co
r
e)
,
in
d
icatin
g
t
h
at
th
e
m
o
d
el
ca
n
ac
cu
r
ately
id
en
tif
y
Pn
eu
m
o
n
i
a
ca
s
es.
Seco
n
d
ly
,
th
e
C
OVI
D
class
h
as
a
s
lig
h
tly
lo
wer
s
co
r
e
(
8
9
%
p
r
ec
is
io
n
,
9
3
%
r
ec
all
an
d
9
1
%
F1
-
s
co
r
e
)
,
in
d
icatin
g
th
at
th
e
m
o
d
el
m
ay
h
av
e
d
i
f
f
icu
lty
ac
cu
r
ately
class
if
y
in
g
C
OVI
D
ca
s
es
co
m
p
ar
ed
to
th
e
o
th
e
r
c
lass
es.
T
h
ir
d
ly
,
th
e
lu
n
g
o
p
ac
ity
class
ac
h
iev
ed
m
o
d
e
r
ately
h
ig
h
s
co
r
es
(
8
6
%
p
r
ec
is
io
n
,
8
9
%
r
ec
all
an
d
8
8
%
F1
-
s
co
r
e)
;
in
d
icatin
g
th
at
th
e
m
o
d
el
ca
n
r
ea
s
o
n
ab
ly
id
en
tify
lu
n
g
o
p
ac
ity
ca
s
es.
F
o
u
r
th
ly
,
th
e
n
o
r
m
al
class
h
as
th
e
s
ec
o
n
d
lo
west
(
8
9
%
p
r
ec
is
io
n
,
8
4
%
r
ec
all
an
d
8
6
%
f
1
-
s
co
r
e)
,
s
u
g
g
esti
n
g
th
at
th
e
m
o
d
el
m
ay
h
av
e
s
o
m
e
ch
allen
g
es
ac
cu
r
ately
class
if
y
in
g
n
o
r
m
al
c
ases
d
u
e
to
s
u
b
tle
v
ar
iatio
n
s
with
in
th
is
ca
teg
o
r
y
.
Fin
ally
,
th
e
T
u
b
er
c
u
lo
s
is
cl
ass
h
as
th
e
s
ec
o
n
d
h
ig
h
est
s
co
r
es
(
9
9
%
Pre
cisi
o
n
,
1
0
0
% r
ec
all,
an
d
9
9
% f
1
-
s
co
r
e)
,
in
d
icatin
g
t
h
at
th
e
m
o
d
el
c
an
ef
f
ec
tiv
ely
i
d
en
tify
T
u
b
er
c
u
lo
s
is
ca
s
es.
Mo
b
ileNetV3
r
esu
lts
f
o
r
5
-
s
u
b
class
C
las
s
if
icatio
n
ar
e
as
s
h
o
wn
in
Fig
u
r
es
9
(
c
)
,
1
0
(
c
)
,
an
d
1
1
(
c
)
.
T
h
e
o
v
er
all
ac
cu
r
ac
y
o
f
0
.
9
4
in
d
icate
s
th
at
th
e
m
o
d
el
co
r
r
e
ctly
class
if
ied
9
4
%
o
f
th
e
s
am
p
les.
B
o
th
m
ac
r
o
an
d
weig
h
ted
a
v
er
ag
e
p
r
ec
is
io
n
,
r
ec
all,
a
n
d
F1
-
s
co
r
e
a
r
e
al
l
0
.
9
4
,
a
n
d
th
is
s
u
g
g
ests
th
at
t
h
e
m
o
d
el
p
er
f
o
r
m
s
v
er
y
well
o
n
av
e
r
ag
e
ac
r
o
s
s
all
class
es,
with
th
e
weig
h
ted
a
v
er
ag
e
a
d
d
itio
n
ally
a
cc
o
u
n
tin
g
f
o
r
p
o
te
n
tial
class
im
b
alan
ce
s
.
Firstl
y
,
th
e
Pn
eu
m
o
n
ia
class
ac
h
iev
ed
t
h
e
h
ig
h
est
s
co
r
es
(
9
8
%
p
r
ec
is
io
n
,
9
9
%
r
ec
all,
a
n
d
9
9
%
F1
-
s
co
r
e)
,
in
d
icatin
g
th
at
th
e
m
o
d
el
ca
n
ac
cu
r
ately
id
en
tify
Pn
eu
m
o
n
ia
ca
s
es.
Seco
n
d
l
y
,
th
e
C
OVI
D
clas
s
h
as
th
e
s
ec
o
n
d
-
h
i
g
h
est
s
co
r
es
(
9
4
%
p
r
ec
is
io
n
,
9
6
%
r
ec
all,
a
n
d
9
5
%
F1
-
s
co
r
e)
,
in
d
icatin
g
th
at
th
e
m
o
d
el
ca
n
ef
f
ec
tiv
ely
class
if
y
C
OVI
D
c
ases
.
T
h
e
lu
n
g
o
p
ac
ity
class
ac
h
iev
ed
m
o
d
er
ately
h
ig
h
s
co
r
es
(
9
1
%
p
r
ec
is
io
n
,
9
0
%
r
ec
all,
an
d
9
0
%
F1
-
sc
o
r
e)
,
in
d
icatin
g
th
at
th
e
m
o
d
el
ca
n
r
ea
s
o
n
ab
ly
id
e
n
tify
lu
n
g
o
p
ac
ity
ca
s
es.
T
h
e
n
o
r
m
al
class
h
as
th
e
f
o
u
r
th
-
h
ig
h
est
s
co
r
es
(
9
0
%
p
r
ec
is
io
n
,
8
8
%
r
ec
all,
an
d
8
9
%
F1
-
s
co
r
e)
,
s
u
g
g
esti
n
g
th
at
th
e
m
o
d
el
ca
n
also
ac
cu
r
atel
y
class
if
y
n
o
r
m
al
ca
s
es.
T
h
e
T
u
b
er
cu
l
o
s
is
class
h
as
th
e
h
ig
h
est
s
co
r
es
(
9
9
%
p
r
ec
is
io
n
,
9
9
%
r
ec
all,
an
d
9
9
%
F1
-
s
co
r
e)
,
al
o
n
g
with
th
e
Pn
eu
m
o
n
ia
class
,
in
d
icatin
g
th
at
th
e
m
o
d
el
ca
n
ex
ce
p
tio
n
ally
i
d
en
tify
T
u
b
er
cu
lo
s
is
ca
s
es.
T
h
e
m
o
d
el
d
e
m
o
n
s
tr
ates
ex
ce
llen
t
o
v
e
r
all
p
er
f
o
r
m
an
ce
,
with
Pn
eu
m
o
n
ia
an
d
T
u
b
er
cu
l
o
s
is
class
if
icat
io
n
s
b
ein
g
th
e
m
o
s
t
s
u
cc
ess
f
u
l.
T
h
e
m
o
d
el
c
an
also
ef
f
ec
tiv
ely
class
if
y
C
O
VI
D
,
lu
n
g
o
p
ac
ity
,
an
d
n
o
r
m
al
ca
s
es,
with
o
n
ly
m
in
o
r
r
o
o
m
f
o
r
im
p
r
o
v
em
en
t.
Fu
r
th
er
in
v
esti
g
atio
n
m
ig
h
t
b
e
b
en
e
f
ic
ial
to
en
h
an
ce
t
h
e
m
o
d
el'
s
ab
ilit
y
to
h
a
n
d
le
t
h
e
m
o
r
e
c
h
allen
g
in
g
class
es,
s
u
ch
as n
o
r
m
al,
alth
o
u
g
h
its
p
er
f
o
r
m
an
ce
in
th
is
class
is
s
til
l v
er
y
h
ig
h
.
Fig
u
r
e
1
1
(
a
)
d
ep
icts
th
e
c
o
n
f
u
s
io
n
m
atr
ix
f
o
r
th
e
Mo
b
ileNetV3
m
o
d
el'
s
f
iv
e
-
s
u
b
class
ca
teg
o
r
izatio
n
o
f
lu
n
g
illn
ess
es.
Similar
ly
,
Fi
g
u
r
e
1
1
(
b
)
e
p
icts
th
e
co
n
f
u
s
io
n
m
atr
ix
f
o
r
t
h
e
R
esNet5
0
m
o
d
el'
s
f
iv
e
-
s
u
b
class
ca
teg
o
r
izatio
n
o
f
lu
n
g
illn
es
s
es.
Fin
ally
,
Fig
u
r
e
1
1
(
c
)
s
h
o
ws
th
e
co
n
f
u
s
io
n
m
atr
i
x
f
o
r
th
e
I
n
ce
p
tio
n
V3
m
o
d
el'
s
f
iv
e
-
s
u
b
class
ca
teg
o
r
izatio
n
s
o
f
lu
n
g
illn
ess
es.
T
h
ese
co
n
f
u
s
io
n
m
atr
ices
th
o
r
o
u
g
h
ly
s
u
m
m
ar
is
e
th
e
class
if
icatio
n
p
er
f
o
r
m
an
ce
f
o
r
ea
ch
lu
n
g
-
r
elate
d
illn
ess
s
u
b
ty
p
e
ac
r
o
s
s
all
th
r
ee
m
o
d
el
ar
ch
i
tectu
r
es.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4
7
5
2
C
la
s
s
i
fica
tio
n
mo
d
el
fo
r
in
fectio
u
s
lu
n
g
d
is
ea
s
es u
s
in
g
…
(
K
en
n
ed
y
Oko
k
p
u
jie
)
419
(
a)
(
b
)
(
c)
Fig
u
r
e
9
.
C
h
ar
t V
is
u
alizin
g
: (
a
)
I
n
ce
p
tio
n
V3
,
(
b
)
R
esNet
-
5
0
an
d
(
c)
Mo
b
ileNetV3
p
er
f
o
r
m
an
ce
s
f
o
r
tr
ain
i
n
g
an
d
v
alid
atio
n
lo
s
s
es o
n
5
-
Su
b
class
(
a)
(
b
)
(
c)
Fig
u
r
e
1
0
.
C
h
ar
t V
is
u
alizin
g
: (
a)
I
n
ce
p
tio
n
V3
,
(
b
)
R
esNet
-
5
0
an
d
(
c)
Mo
b
ileNetV3
p
er
f
o
r
m
an
ce
s
f
o
r
tr
ai
n
in
g
an
d
v
alid
atio
n
ac
cu
r
ac
ies o
n
5
Su
b
class
(
a)
(
b
)
(
c)
Fig
u
r
e
1
1
.
C
o
n
f
u
s
io
n
m
atr
ix
v
is
u
alizin
g
: (
a)
Mo
b
ileNetV3
,
(
b
)
R
esNet
-
50
,
an
d
(
c)
I
n
ce
p
tio
n
V3
p
er
f
o
r
m
a
n
ce
s
on
c
lass
if
y
in
g
5
s
u
b
class
es
3
.
2
Dis
cus
s
io
n
I
n
th
e
3
-
s
u
b
class
s
ce
n
ar
io
,
R
esNet
-
5
0
ac
h
iev
ed
th
e
h
ig
h
est
p
r
ec
is
io
n
,
r
ec
all,
F1
-
s
co
r
e,
an
d
ac
cu
r
ac
y
at
9
6
%,
9
6
%,
an
d
9
6
%,
r
esp
ec
tiv
ely
.
Mo
b
ileNetV3
an
d
I
n
ce
p
tio
n
V3
also
p
er
f
o
r
m
ed
well,
with
9
0
% a
n
d
9
3
%
ac
r
o
s
s
th
e
v
ar
io
u
s
m
etr
ics.
I
n
th
e
4
-
s
u
b
class
s
ce
n
ar
io
,
Mo
b
ileNetV3
o
u
tp
er
f
o
r
m
e
d
th
e
o
th
er
m
o
d
els,
Evaluation Warning : The document was created with Spire.PDF for Python.