I
AE
S In
t
er
na
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI
)
Vo
l.
1
5
,
No
.
4
,
A
u
g
u
s
t 2
0
2
6
,
p
p
.
3
1
2
0
~
3
1
3
0
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
15
.i
4
.
p
p
3
1
2
0
-
3
1
3
0
3120
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
Deep learni
ng
f
o
r
lung ca
ncer
diag
no
sis
:
a
com
pa
ra
tive
a
rtif
icia
l in
tellige
nce study
P
ha
neendra
Va
rm
a
Chin
t
a
la
pa
t
i
1
,
P
ra
s
a
nth
Arucha
m
y
1
,
Alur
P
ra
neet
ha
2
1
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
V
e
l
T
e
c
h
R
a
n
g
a
r
a
j
a
n
D
r
.
S
a
g
u
n
t
h
a
l
a
R
&D
I
n
st
i
t
u
t
e
o
f
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
C
h
e
n
n
a
i
,
I
n
d
i
a
2
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
K
o
n
e
r
u
L
a
k
sh
m
a
i
a
h
E
d
u
c
a
t
i
o
n
F
o
u
n
d
a
t
i
o
n
,
V
a
d
d
e
sw
a
r
a
m
,
I
n
d
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Sep
1
1
,
2
0
2
5
R
ev
is
ed
Ma
y
2
0
,
2
0
2
6
Acc
ep
ted
J
u
n
1
9
,
2
0
2
6
Early
a
n
d
a
c
c
u
ra
te
lu
n
g
c
a
n
c
e
r
d
iag
n
o
sis
(LCD)
is
c
rit
ica
l,
y
e
t
trad
it
io
n
a
l
ima
g
in
g
m
e
th
o
d
s
su
c
h
a
s
X
-
ra
y
s
a
n
d
c
o
m
p
u
ted
t
o
m
o
g
ra
p
h
y
(CT)
sc
a
n
s
a
re
c
o
stly
,
in
v
a
siv
e
,
a
n
d
h
e
a
v
il
y
re
li
a
n
t
o
n
e
x
p
e
rt
in
terp
re
tati
o
n
.
T
h
is
stu
d
y
in
v
e
stig
a
tes
a
rti
ficia
l
in
tell
ig
e
n
c
e
(AI)
-
d
riv
e
n
d
ia
g
n
o
stics
b
y
c
o
m
p
a
rin
g
d
e
e
p
lea
rn
in
g
m
o
d
e
ls
(
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
two
r
k
(CNN
)
,
re
sid
u
a
l
n
e
two
rk
(Re
sN
e
t)
,
a
n
d
v
isu
a
l
g
e
o
m
e
try
g
r
o
u
p
16
(
VG
G
1
6
)
)
with
trad
i
ti
o
n
a
l
m
a
c
h
in
e
lea
rn
in
g
a
lg
o
rit
h
m
s
(
lo
g
isti
c
re
g
re
ss
io
n
(LR)
a
n
d
s
u
p
p
o
rt
v
e
c
to
r
m
a
c
h
in
e
(S
VM)
),
u
si
n
g
l
u
n
g
ima
g
e
d
a
tab
a
se
c
o
n
so
rti
u
m
a
n
d
ima
g
e
d
a
tab
a
se
re
so
u
rc
e
in
it
iati
v
e
(
LIDC
-
IDRI
)
a
n
d
Ka
g
g
le
lu
n
g
CT
sc
a
n
d
a
tas
e
ts.
P
e
rf
o
r
m
a
n
c
e
wa
s
e
v
a
lu
a
ted
a
c
ro
ss
m
u
lt
ip
le
m
e
tri
c
s
:
a
c
c
u
ra
c
y
,
se
n
siti
v
it
y
,
sp
e
c
ifi
c
it
y
,
a
n
d
a
re
a
u
n
d
e
r
th
e
c
u
r
v
e
-
re
c
e
iv
e
r
o
p
e
ra
ti
n
g
c
h
a
ra
c
teristic
(
AUC
-
ROC
)
.
Am
o
n
g
t
h
e
m
o
d
e
ls,
Re
sN
e
t
a
c
h
iev
e
d
t
h
e
h
i
g
h
e
st
p
e
rfo
rm
a
n
c
e
,
wit
h
a
n
a
c
c
u
ra
c
y
o
f
9
4
%
,
se
n
siti
v
it
y
o
f
9
5
%
,
sp
e
c
ifi
c
it
y
o
f
9
3
%
,
a
n
d
AU
C
-
ROC
c
lo
se
to
1
.
CNN
a
n
d
VG
G
1
6
a
lso
sh
o
we
d
su
p
e
ri
o
r
m
e
tri
c
s
c
o
m
p
a
re
d
to
LR
a
n
d
S
VM
,
h
ig
h
li
g
h
ti
n
g
t
h
e
ro
b
u
stn
e
ss
o
f
d
e
e
p
lea
rn
in
g
tec
h
n
i
q
u
e
s.
Th
e
se
re
su
lt
s
d
e
m
o
n
stra
te
t
h
a
t
d
e
e
p
lea
rn
i
n
g
m
o
d
e
ls
n
o
t
o
n
ly
a
c
h
iev
e
h
ig
h
e
r
d
iag
n
o
stic
a
c
c
u
ra
c
y
b
u
t
a
lso
sig
n
ifi
c
a
n
t
l
y
re
d
u
c
e
fa
lse
d
e
tec
ti
o
n
s
c
o
m
p
a
re
d
to
trad
it
io
n
a
l
a
p
p
ro
a
c
h
e
s.
Th
e
f
in
d
i
n
g
s
s
u
p
p
o
rt
th
e
p
o
ten
ti
a
l
o
f
AI
t
o
a
u
t
o
m
a
te
a
n
d
e
n
h
a
n
c
e
LCD
,
t
h
e
re
b
y
imp
r
o
v
i
n
g
a
c
c
e
ss
ib
il
it
y
,
c
o
n
siste
n
c
y
,
a
n
d
s
p
e
e
d
in
c
li
n
ica
l
se
tt
in
g
s.
F
u
t
u
re
wo
r
k
will
e
m
p
h
a
siz
e
c
li
n
ica
l
v
a
li
d
a
ti
o
n
,
a
d
d
re
ss
e
th
ica
l
c
h
a
ll
e
n
g
e
s,
a
n
d
f
o
c
u
s
o
n
in
teg
ra
ti
n
g
AI
m
o
d
e
ls
in
t
o
r
e
a
l
-
wo
rld
h
e
a
lt
h
c
a
re
wo
rk
fl
o
ws
to
imp
ro
v
e
p
a
ti
e
n
t
o
u
tc
o
m
e
s.
K
ey
w
o
r
d
s
:
AI
-
d
r
iv
en
d
iag
n
o
s
tics
C
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
s
Dee
p
lear
n
in
g
L
u
n
g
ca
n
ce
r
d
etec
tio
n
R
esid
u
al
n
etwo
r
k
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
:
Ph
an
ee
n
d
r
a
Var
m
a
C
h
in
talap
a
ti
Dep
ar
tm
en
t o
f
C
o
m
p
u
ter
Scie
n
ce
an
d
E
n
g
in
ee
r
in
g
Vel
T
ec
h
R
an
g
ar
ajan
Dr
.
Sag
u
n
th
ala
R
&
D
I
n
s
titu
te
o
f
Scie
n
ce
an
d
T
ec
h
n
o
l
o
g
y
Av
ad
i,
C
h
en
n
ai,
I
n
d
ia
E
m
ail:
v
td
1
1
0
6
@
v
eltec
h
.
ed
u
.
i
n
1.
I
NT
RO
D
UCT
I
O
N
Nea
r
ly
1
8
%
o
f
g
lo
b
al
ca
n
ce
r
f
atalities
ar
e
d
u
e
to
lu
n
g
ca
n
ce
r
.
Acc
u
r
ate
d
ia
g
n
o
s
is
is
cr
u
cial
f
o
r
ea
r
ly
in
ter
v
en
tio
n
s
to
en
h
an
ce
p
atien
t
s
u
r
v
iv
al.
C
o
m
p
u
ted
to
m
o
g
r
ap
h
y
(
C
T
)
s
ca
n
s
an
d
ch
est
X
-
r
ay
s
ar
e
u
tili
ze
d
,
b
u
t
th
ey
ar
e
in
v
asiv
e,
ex
p
en
s
iv
e,
an
d
r
eq
u
ir
e
p
r
o
f
ess
io
n
al
k
n
o
wled
g
e,
lim
itin
g
th
eir
av
ail
ab
ilit
y
an
d
ef
f
icac
y
[
1
]
.
R
ec
en
t
ad
v
a
n
ce
s
,
p
ar
tic
u
lar
ly
d
ee
p
lear
n
in
g
-
b
ased
o
n
es,
h
av
e
m
ad
e
lu
n
g
ca
n
ce
r
d
iag
n
o
s
is
(
L
C
D
)
non
-
in
v
asiv
e,
co
s
t
-
ef
f
ec
tiv
e,
an
d
ef
f
icien
t
[
2
]
.
Dee
p
le
ar
n
in
g
m
o
d
els,
esp
ec
ially
co
n
v
o
lu
ti
o
n
al
n
eu
r
al
n
etwo
r
k
s
(
C
NNs),
h
av
e
d
em
o
n
s
tr
ated
r
em
ar
k
a
b
le
ca
p
ab
ilit
ies
in
m
ed
ical
im
ag
e
an
aly
s
is
,
o
f
f
er
in
g
p
o
te
n
tial
f
o
r
im
p
r
o
v
e
d
s
en
s
itiv
ity
,
s
p
ec
if
icity
,
an
d
e
f
f
icien
cy
i
n
lu
n
g
c
an
ce
r
d
etec
tio
n
an
d
class
if
icatio
n
[
1
]
,
[
3
]
,
[
4
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
Dee
p
lea
r
n
in
g
fo
r
lu
n
g
ca
n
ce
r
d
ia
g
n
o
s
is
:
a
co
mp
a
r
a
tive
a
r
ti
ficia
l
…
(
P
h
a
n
ee
n
d
r
a
V
a
r
ma
C
h
in
ta
la
p
a
ti
)
3121
C
NNs
ca
n
id
en
tify
an
d
ca
teg
o
r
ize
m
alig
n
an
t
tu
m
o
r
s
in
m
ed
ical
im
a
g
in
g
d
atasets
.
T
h
ese
d
ev
elo
p
m
e
n
ts
in
cr
ea
s
e
d
iag
n
o
s
tic
ac
cu
r
ac
y
an
d
en
ab
le
s
e
am
less
in
teg
r
atio
n
o
f
au
to
m
a
ted
tech
n
o
lo
g
y
in
to
clin
ical
wo
r
k
f
lo
ws,
m
ak
in
g
e
ar
ly
d
etec
tio
n
ea
s
ier
an
d
m
o
r
e
r
eliab
le
[
5
]
.
Pre
v
io
u
s
d
iag
n
o
s
tic
ap
p
r
o
ac
h
es
ar
e
co
m
p
ar
ed
to
d
ee
p
lear
n
in
g
m
e
th
o
d
s
to
ex
p
lain
h
o
w
t
h
ey
m
i
g
h
t
o
v
er
co
m
e
th
ei
r
d
is
ad
v
an
ta
g
es.
B
y
co
m
p
ar
in
g
th
ese
m
eth
o
d
s
,
th
e
p
o
ten
tial
o
f
d
ee
p
lear
n
i
n
g
t
o
e
n
h
an
c
e
lu
n
g
ca
n
ce
r
d
etec
tio
n
a
n
d
o
u
tco
m
es
ca
n
b
e
u
n
d
er
s
to
o
d
[
6
]
.
R
ec
en
t
r
esear
ch
s
h
o
ws
d
ee
p
lear
n
in
g
s
y
s
tem
s
m
ay
d
etec
t
lu
n
g
ca
n
ce
r
.
Dee
p
lear
n
in
g
a
lg
o
r
ith
m
s
ca
n
au
to
m
atica
lly
class
if
y
lu
n
g
ca
n
ce
r
in
X
-
r
ay
s
,
wh
o
le
s
lid
e
im
ag
es
(
W
SI)
,
C
T
s
c
an
s
,
an
d
m
a
g
n
etic
r
eso
n
an
ce
im
ag
in
g
(
MRI
)
,
ac
co
r
d
in
g
to
c
o
m
p
r
e
h
en
s
iv
e
r
e
s
ea
r
ch
.
An
o
th
er
s
tu
d
y
f
o
u
n
d
th
at
d
ee
p
lear
n
in
g
alg
o
r
ith
m
s
ca
n
en
h
a
n
ce
p
o
s
itro
n
em
is
s
io
n
to
m
o
g
r
ap
h
y
(
PET
)
/C
T
lu
n
g
ca
n
ce
r
d
etec
tio
n
f
o
r
ea
r
ly
d
iag
n
o
s
is
an
d
tailo
r
ed
t
r
ea
tm
en
t
[
7
]
.
Dee
p
lear
n
in
g
allo
ws
m
o
d
e
ls
to
d
iag
n
o
s
e
an
d
ca
teg
o
r
i
ze
lu
n
g
ca
n
ce
r
s
in
d
e
p
en
d
e
n
tly
u
s
in
g
h
is
to
p
ath
o
lo
g
y
p
ictu
r
es.
T
h
is
s
tu
d
y
r
ev
ea
led
th
at
d
ee
p
lear
n
in
g
co
u
ld
b
e
u
s
ed
to
d
ia
g
n
o
s
e
m
alig
n
an
cies
an
d
p
r
ed
ict
p
atien
t
o
u
tc
o
m
es
b
ase
d
o
n
h
is
to
lo
g
ical
m
ar
k
e
r
s
.
De
ep
lear
n
in
g
m
a
y
e
n
h
an
ce
L
C
D
an
d
d
etec
tio
n
[
8
]
.
Dee
p
lear
n
in
g
tech
n
o
lo
g
y
is
in
tr
ig
u
in
g
,
b
u
t
clin
ical
u
s
e
p
r
esen
ts
ch
allen
g
es
th
at
ca
n
b
e
o
v
er
co
m
e
[
9
]
.
E
th
ics,
v
alid
atio
n
ac
r
o
s
s
s
ev
er
al
p
atie
n
t
p
o
p
u
latio
n
s
,
an
d
f
air
ac
ce
s
s
to
ar
tific
ial
in
tellig
en
ce
(
AI
)
-
p
o
wer
ed
d
iag
n
o
s
tic
tech
n
o
lo
g
ies
ar
e
im
p
o
r
tan
t
ch
allen
g
es.
Dee
p
lear
n
in
g
alg
o
r
i
th
m
s
f
o
r
L
C
D
wer
e
e
v
alu
ated
an
d
s
h
o
w
n
to
n
ee
d
ad
d
itio
n
al
s
tu
d
y
to
s
o
lv
e
t
h
ese
ch
allen
g
es
an
d
b
o
o
s
t
AI
-
b
a
s
ed
clin
ical
d
iag
n
o
s
tics
[
1
0
]
.
Dee
p
lear
n
in
g
m
a
y
id
en
tify
lu
n
g
ca
n
ce
r
.
Dee
p
l
ea
r
n
in
g
al
g
o
r
ith
m
s
ca
n
im
p
r
o
v
e
h
ea
lth
ca
r
e,
d
iag
n
o
s
is
ac
cu
r
ac
y
,
a
n
d
p
atien
t
o
u
tco
m
es
b
y
p
r
e
d
ictin
g
d
iag
n
o
s
tic
co
n
s
tr
ain
ts
[
1
1
]
.
Fu
tu
r
e
r
esear
ch
m
u
s
t
o
v
er
co
m
e
cu
r
r
e
n
t
o
b
s
tacle
s
to
b
r
in
g
n
ew
tech
n
o
lo
g
y
in
t
o
clin
ical
p
r
ac
tice.
B
y
co
m
p
ar
in
g
t
h
e
p
er
f
o
r
m
a
n
ce
o
f
d
if
f
e
r
en
t
d
ee
p
lear
n
in
g
m
o
d
els,
th
is
co
m
p
ar
ativ
e
s
tu
d
y
aim
s
to
p
r
o
v
id
e
a
co
m
p
r
eh
en
s
iv
e
o
v
e
r
v
iew
o
f
t
h
e
cu
r
r
en
t
s
tate
-
of
-
th
e
-
ar
t
in
A
I
-
ass
is
ted
L
C
D
.
I
t
also
ex
am
in
es
th
eir
s
tr
en
g
th
s
an
d
wea
k
n
ess
es.
T
h
i
s
ex
p
lo
r
atio
n
will
co
n
tr
ib
u
te
t
o
th
e
o
n
g
o
in
g
ef
f
o
r
ts
to
im
p
r
o
v
e
ea
r
ly
d
etec
tio
n
,
p
er
s
o
n
alize
tr
ea
tm
e
n
t
s
tr
ateg
i
es,
an
d
u
ltima
tely
en
h
an
ce
th
e
s
u
r
v
iv
al
r
ates
o
f
in
d
iv
id
u
als
af
f
ec
ted
b
y
th
is
d
ev
astatin
g
d
is
ea
s
e
[
1
2
]
.
R
esear
ch
h
as
d
em
o
n
s
tr
ated
th
e
ef
f
icac
y
an
d
th
er
a
p
eu
tic
p
r
o
m
is
e
o
f
d
ee
p
lear
n
in
g
ap
p
r
o
ac
h
es
f
o
r
lu
n
g
ca
n
ce
r
s
cr
ee
n
in
g
.
A
co
m
p
r
eh
en
s
iv
e
a
n
aly
s
is
b
y
J
av
ed
et
a
l.
[
1
]
h
ig
h
lig
h
te
d
th
e
u
s
ef
u
ln
ess
o
f
d
ee
p
co
n
v
o
l
u
tio
n
al
n
e
u
r
al
n
etwo
r
k
s
(
DC
NNs)
in
au
to
m
atin
g
l
u
n
g
ca
n
ce
r
d
etec
tio
n
u
s
in
g
X
-
r
a
y
s
,
W
SI,
C
T
s
ca
n
s
,
an
d
MRI.
B
o
u
ch
af
f
r
a
et
a
l.
[
1
3
]
u
s
ed
d
ee
p
lear
n
i
n
g
in
l
o
w
-
d
o
s
e
C
T
s
cr
ee
n
in
g
p
r
o
g
r
am
s
to
id
e
n
tify
an
d
s
tr
atif
y
lu
n
g
ca
n
ce
r
in
h
ig
h
-
r
is
k
p
o
p
u
latio
n
s
.
Dee
p
lear
n
in
g
h
as
im
p
r
o
v
ed
h
is
to
p
ath
o
lo
g
ical
an
aly
s
is
,
with
Kalk
an
et
a
l.
[
1
4
]
s
h
o
win
g
au
to
m
ated
lu
n
g
tis
s
u
e
tu
m
o
r
id
en
tific
atio
n
an
d
c
h
ar
ac
ter
i
za
tio
n
f
o
r
ac
cu
r
ate
d
iag
n
o
s
is
an
d
p
r
o
g
n
o
s
is
.
I
n
a
th
o
r
o
u
g
h
ass
ess
m
en
t
o
f
d
ee
p
lear
n
in
g
m
o
d
els
f
o
r
ea
r
ly
-
s
tag
e
lu
n
g
ca
n
ce
r
d
etec
tio
n
,
Ho
s
s
ein
i
et
a
l
.
[
2
]
i
d
en
tifie
d
p
r
o
b
lem
s
an
d
p
o
te
n
tial a
p
p
r
o
ac
h
es.
L
C
D
u
s
in
g
C
T
im
ag
es
h
as
b
ee
n
s
ig
n
if
ican
tly
en
h
an
ce
d
th
r
o
u
g
h
d
ee
p
lear
n
in
g
tech
n
iq
u
es.
Ma
m
u
n
et
a
l.
[
1
5
]
p
r
o
p
o
s
ed
t
h
e
L
C
DctCN
N,
d
em
o
n
s
tr
atin
g
h
ig
h
ac
cu
r
ac
y
an
d
r
ec
all
f
o
r
L
C
D
.
L
iu
et
a
l.
[
1
6
]
r
ev
iewe
d
th
e
e
v
o
lu
tio
n
o
f
p
u
l
m
o
n
ar
y
n
o
d
u
le
d
etec
tio
n
o
v
er
th
e
p
ast
th
r
ee
d
ec
ad
es,
h
ig
h
li
g
h
tin
g
th
e
tr
an
s
itio
n
f
r
o
m
co
n
v
en
tio
n
al
im
ag
e
an
aly
s
is
to
d
ee
p
lear
n
in
g
-
b
ased
d
ec
is
io
n
s
u
p
p
o
r
t.
W
an
g
[
1
7
]
co
m
p
r
e
h
en
s
iv
el
y
d
is
cu
s
s
ed
r
ec
en
t
d
ee
p
lear
n
i
n
g
tech
n
iq
u
es
f
o
r
L
C
D
,
em
p
h
asizin
g
th
eir
p
o
te
n
tial
to
im
p
r
o
v
e
d
iag
n
o
s
tic
ac
cu
r
ac
y
,
ea
r
l
y
d
etec
tio
n
,
a
n
d
clin
ical
d
ec
is
io
n
-
m
ak
i
n
g
.
So
h
n
an
d
Field
s
[
1
8
]
d
e
m
o
n
s
tr
ated
th
at
th
e
in
teg
r
atio
n
o
f
r
ad
io
m
ics
an
d
d
ee
p
lear
n
in
g
e
n
ab
les
ac
cu
r
ate
p
r
ed
ictio
n
o
f
p
u
lm
o
n
ar
y
n
o
d
u
le
m
etastas
i
s
f
r
o
m
C
T
im
ag
es,
th
er
e
b
y
s
u
p
p
o
r
tin
g
p
r
ec
is
io
n
m
e
d
icin
e.
Hash
im
o
to
et
a
l
.
[
1
9
]
h
ig
h
lig
h
ted
th
e
o
p
p
o
r
tu
n
ities
an
d
ch
allen
g
es
o
f
AI
in
h
ea
lth
ca
r
e,
em
p
h
asizin
g
its
p
o
ten
tial
to
im
p
r
o
v
e
d
iag
n
o
s
tic
ef
f
icien
cy
an
d
clin
ical
o
u
tco
m
es.
L
e
e
et
a
l.
[
2
0
]
r
e
v
iewe
d
th
e
cu
r
r
en
t
s
tatu
s
a
n
d
f
u
tu
r
e
d
ir
ec
tio
n
s
o
f
r
ad
i
o
m
ics
in
lu
n
g
ca
n
ce
r
,
d
em
o
n
s
tr
atin
g
its
r
o
le
in
ex
tr
ac
tin
g
q
u
a
n
titativ
e
im
ag
in
g
b
io
m
ar
k
er
s
f
o
r
d
iag
n
o
s
is
an
d
p
r
o
g
n
o
s
is
.
Fu
r
th
er
m
o
r
e
,
Z
h
an
g
et
a
l.
[
2
1
]
p
e
r
f
o
r
m
ed
a
s
y
s
tem
atic
r
ev
iew
an
d
m
eta
-
a
n
aly
s
is
,
co
n
f
ir
m
in
g
th
e
h
i
g
h
d
iag
n
o
s
tic
p
er
f
o
r
m
an
ce
o
f
C
NN
s
in
p
u
lm
o
n
ar
y
n
o
d
u
le
d
etec
tio
n
an
d
r
ein
f
o
r
cin
g
th
ei
r
ef
f
ec
tiv
en
ess
in
co
m
p
u
ter
-
aid
ed
L
C
D
.
Dee
p
lear
n
in
g
h
as
b
ee
n
u
s
ed
with
g
en
o
m
es
an
d
clin
ical
d
ata
to
en
h
an
ce
lu
n
g
ca
n
ce
r
d
etec
tio
n
alg
o
r
ith
m
s
.
T
r
an
s
f
er
lear
n
in
g
h
as
also
b
ee
n
u
s
ed
to
id
en
tify
lu
n
g
ca
n
ce
r
u
s
in
g
p
r
e
-
t
r
ain
ed
m
o
d
els
with
o
u
t
b
ig
an
n
o
tated
d
atasets
.
E
x
p
lain
ab
le
ar
tific
ial
in
tellig
en
ce
(X
AI
)
m
o
d
els
ar
e
em
p
h
asized
in
d
ee
p
lear
n
in
g
-
b
ased
lu
n
g
ca
n
ce
r
d
etec
tio
n
s
y
s
tem
s
to
p
r
o
v
id
e
t
r
an
s
p
ar
en
c
y
a
n
d
tr
u
s
two
r
th
in
ess
.
Fo
r
f
aster
d
ec
is
io
n
-
m
ak
in
g
an
d
b
etter
p
atien
t o
u
tc
o
m
es,
clin
ic
al
d
ee
p
-
lear
n
i
n
g
m
o
d
els h
av
e
b
ee
n
test
ed
in
r
ea
l tim
e.
2.
CH
AL
L
E
NG
E
S AN
D
UN
A
DDRES
SE
D
I
S
SUE
S
C
las
s
ic
an
d
m
o
d
er
n
lu
n
g
ca
n
ce
r
s
cr
ee
n
in
g
m
eth
o
d
s
r
em
ai
n
u
n
d
er
s
tu
d
ie
d
d
esp
ite
m
ed
ical
im
ag
in
g
an
d
AI
b
r
ea
k
th
r
o
u
g
h
s
.
C
h
est
X
-
r
ay
s
,
C
T
s
ca
n
s
,
a
n
d
b
io
p
s
ie
s
ar
e
co
m
m
o
n
l
y
u
s
ed
,
b
u
t
th
ei
r
in
v
asiv
en
ess
,
h
ig
h
co
s
t,
an
d
r
elian
ce
o
n
ex
p
e
r
t
in
ter
p
r
etatio
n
lim
it
th
eir
u
s
e,
esp
ec
ially
in
r
eso
u
r
ce
-
co
n
s
tr
ain
ed
co
u
n
tr
ies.
Dee
p
lear
n
in
g
ap
p
r
o
ac
h
es
lik
e
C
NN
s
,
r
esid
u
al
n
etwo
r
k
(
R
esNet)
,
an
d
v
is
u
al
g
eo
m
etr
y
g
r
o
u
p
1
6
(
VGG1
6
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
5
,
No
.
4
,
Au
g
u
s
t 2
0
2
6
:
3
1
2
0
-
3
1
3
0
3122
en
h
an
ce
ac
cu
r
ac
y
an
d
e
f
f
ic
ien
cy
with
o
u
t
s
u
r
g
e
r
y
.
E
ar
l
y
s
tu
d
ies
s
eld
o
m
co
m
p
ar
e
th
ese
alg
o
r
ith
m
s
'
p
er
f
o
r
m
an
ce
o
n
m
an
y
r
ea
l
-
wo
r
ld
d
atasets
.
C
lin
ical
ac
cu
r
ac
y
,
s
en
s
itiv
ity
,
an
d
s
ca
lab
ilit
y
ar
e
r
ar
ely
ad
d
r
ess
ed
.
T
h
is
d
is
cr
ep
an
cy
h
ig
h
li
g
h
ts
th
e
n
ec
ess
ity
to
in
te
g
r
ate
o
l
d
an
d
m
o
d
er
n
ap
p
r
o
ac
h
es
f
u
lly
.
T
h
ese
is
s
u
es
ar
e
ad
d
r
ess
ed
v
ia
a
r
ig
o
r
o
u
s
s
tu
d
y
th
at
co
m
p
ar
es
ex
is
tin
g
d
iag
n
o
s
tic
m
eth
o
d
s
to
d
ee
p
lear
n
in
g
-
b
ased
lu
n
g
ca
n
ce
r
d
etec
tio
n
m
o
d
els.
Fo
r
ac
cu
r
a
cy
,
s
en
s
itiv
ity
,
an
d
s
ca
lab
ilit
y
,
C
NNs,
R
esNe
t
,
an
d
VGG1
6
will
b
e
test
ed
o
n
lu
n
g
im
ag
e
d
atab
ase
co
n
s
o
r
tiu
m
an
d
im
a
g
e
d
atab
ase
r
eso
u
r
c
e
in
itiativ
e
(
L
I
DC
-
I
DR
I
)
an
d
Kag
g
le
lu
n
g
ca
n
ce
r
d
atasets
is
il
lu
s
tr
ated
in
T
ab
le
1
.
T
ab
le
1
.
C
o
m
p
a
r
ativ
e
an
aly
s
is
o
f
d
ee
p
lear
n
in
g
alg
o
r
ith
m
s
f
o
r
lu
n
g
ca
n
ce
r
d
etec
tio
n
R
e
f
e
r
e
n
c
e
A
l
g
o
r
i
t
h
m
Ex
t
r
a
c
t
e
d
f
e
a
t
u
r
e
s
A
c
c
u
r
a
c
y
p
e
r
f
o
r
m
a
n
c
e
Jav
e
d
e
t
a
l
.
[
1
]
D
C
N
N
s
A
u
t
o
ma
t
e
d
f
e
a
t
u
r
e
e
x
t
r
a
c
t
i
o
n
f
r
o
m X
-
r
a
y
s
,
W
S
I
,
C
T
sca
n
s,
a
n
d
M
R
I
H
i
g
h
a
c
c
u
r
a
c
y
a
c
r
o
ss m
u
l
t
i
p
l
e
i
ma
g
i
n
g
m
o
d
a
l
i
t
i
e
s
M
a
m
u
n
e
t
a
l
.
[
1
5
]
LC
D
c
t
C
N
N
F
e
a
t
u
r
e
s fr
o
m
C
T
s
c
a
n
i
ma
g
e
s
A
c
c
u
r
a
c
y
:
9
2
%
Zh
a
n
g
e
t
a
l
.
[
2
1
]
C
N
N
H
i
e
r
a
r
c
h
i
c
a
l
f
e
a
t
u
r
e
s
a
u
t
o
m
a
t
i
c
a
l
l
y
e
x
t
r
a
c
t
e
d
f
r
o
m
p
u
l
m
o
n
a
r
y
C
T
i
m
a
g
e
s
H
i
g
h
d
i
a
g
n
o
st
i
c
a
c
c
u
r
a
c
y
f
o
r
p
u
l
m
o
n
a
r
y
n
o
d
u
l
e
d
e
t
e
c
t
i
o
n
b
a
s
e
d
o
n
s
y
st
e
ma
t
i
c
r
e
v
i
e
w
a
n
d
met
a
-
a
n
a
l
y
si
s
.
S
u
z
u
k
i
[
2
2
]
M
a
ss
i
v
e
t
r
a
i
n
i
n
g
a
r
t
i
f
i
c
i
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
(
M
TA
N
N
)
F
e
a
t
u
r
e
s fr
o
m
l
o
w
-
d
o
s
e
C
T
sca
n
s
R
e
d
u
c
t
i
o
n
o
f
f
a
l
se
p
o
si
t
i
v
e
s
i
n
n
o
d
u
l
e
d
e
t
e
c
t
i
o
n
V
á
z
q
u
e
z
e
t
a
l
.
[
2
3
]
C
N
N
f
o
r
m
a
mm
o
g
r
a
p
h
i
c
mi
c
r
o
c
a
l
c
i
f
i
c
a
t
i
o
n
s
P
a
t
t
e
r
n
r
e
c
o
g
n
i
t
i
o
n
f
e
a
t
u
r
e
s
f
r
o
m ma
m
mo
g
r
a
ms
A
c
h
i
e
v
e
d
u
p
t
o
8
9
.
5
6
%
a
c
c
u
r
a
c
y
M
i
t
t
a
l
e
t
a
l
.
[
2
4
]
N
e
u
r
a
l
n
e
t
w
o
r
k
f
o
r
l
i
v
e
r
l
e
s
i
o
n
c
l
a
ss
i
f
i
c
a
t
i
o
n
F
e
a
t
u
r
e
s
o
f
C
T
f
o
c
a
l
l
i
v
e
r
l
e
si
o
n
s
I
mp
r
o
v
e
d
c
l
a
s
si
f
i
c
a
t
i
o
n
a
c
c
u
r
a
c
y
V
a
r
ma
e
t
a
l
.
[
2
5
]
C
N
N
f
o
r
c
a
r
o
t
i
d
a
t
h
e
r
o
sc
l
e
r
o
si
s
d
i
a
g
n
o
si
s
U
l
t
r
a
s
o
u
n
d
i
ma
g
e
st
a
t
i
s
t
i
c
s
a
n
d
t
e
x
t
u
r
e
f
e
a
t
u
r
e
s
En
h
a
n
c
e
d
d
i
a
g
n
o
s
t
i
c
a
c
c
u
r
a
c
y
O
r
o
z
c
o
e
t
a
l
.
[
2
6
]
S
u
p
p
o
r
t
v
e
c
t
o
r
c
l
a
ss
i
f
i
c
a
t
i
o
n
f
o
r
l
u
n
g
n
o
d
u
l
e
d
e
t
e
c
t
i
o
n
F
e
a
t
u
r
e
s fr
o
m
c
h
e
st
r
a
d
i
o
g
r
a
p
h
s
I
mp
r
o
v
e
d
d
e
t
e
c
t
i
o
n
a
c
c
u
r
a
c
y
E
ac
h
s
tr
ateg
y
'
s
p
r
o
s
an
d
co
n
s
will
h
ig
h
lig
h
t
AI
-
d
r
iv
en
d
ia
g
n
o
s
tics
'
tr
an
s
f
o
r
m
ativ
e
p
o
te
n
tial
wh
ile
ad
d
r
ess
in
g
s
ca
lab
ilit
y
a
n
d
clin
ical
in
teg
r
atio
n
is
s
u
es.
Fin
a
lly
,
th
is
in
itiativ
e
will
d
ev
el
o
p
a
n
d
im
p
lem
en
t
th
er
ap
ies
to
en
h
an
ce
ea
r
ly
L
C
D
to
ad
v
an
ce
th
e
f
ield
.
Dee
p
lear
n
in
g
h
as
s
ig
n
if
ican
tly
a
d
v
an
ce
d
l
u
n
g
ca
n
ce
r
d
etec
tio
n
ac
r
o
s
s
im
ag
in
g
m
o
d
alities
lik
e
C
T
,
MRI,
an
d
h
is
t
o
p
ath
o
lo
g
y
;
n
ev
er
t
h
eless
,
s
ev
er
al
cr
itical
r
esear
ch
g
ap
s
r
em
ain
.
Mo
s
t
s
tu
d
ies,
in
clu
d
in
g
th
o
s
e
b
y
au
th
o
r
s
[
1
]
,
[
2
]
,
h
ig
h
lig
h
t
lim
itatio
n
s
s
u
ch
as
d
ata
h
eter
o
g
en
eity
,
lack
o
f
s
tan
d
ar
d
ized
im
ag
in
g
p
r
o
to
co
ls
,
an
d
ch
allen
g
es
in
g
en
er
alizin
g
m
o
d
els
ac
r
o
s
s
d
iv
er
s
e
p
o
p
u
latio
n
s
.
T
h
er
e
is
a
s
h
o
r
ta
g
e
o
f
lar
g
e,
a
n
n
o
tate
d
d
atasets
n
ee
d
ed
f
o
r
r
o
b
u
s
t m
o
d
el
tr
ain
in
g
,
as d
is
cu
s
s
ed
b
y
L
iu
et
a
l.
[
1
6
]
.
Mo
r
eo
v
er
,
alth
o
u
g
h
X
AI
an
d
r
ea
l
-
tim
e
clin
ical
d
ep
lo
y
m
en
t
h
av
e
b
ee
n
ex
p
lo
r
ed
,
b
u
il
d
in
g
f
u
lly
tr
an
s
p
ar
en
t,
b
ias
-
f
r
ee
,
an
d
et
h
ically
s
o
u
n
d
AI
s
y
s
tem
s
s
til
l
r
eq
u
ir
es
m
o
r
e
f
o
cu
s
ed
r
ese
ar
ch
.
I
n
teg
r
atio
n
o
f
m
u
ltimo
d
al
d
ata
(
im
a
g
in
g
,
clin
ical,
an
d
m
o
lec
u
lar
)
f
o
r
co
m
p
r
eh
en
s
iv
e
d
iag
n
o
s
is
an
d
p
er
s
o
n
alize
d
th
er
ap
y
r
em
ain
s
u
n
d
er
d
e
v
elo
p
ed
.
Fu
r
th
er
,
v
alid
atio
n
s
tu
d
ies
in
r
e
al
-
wo
r
ld
clin
ical
e
n
v
ir
o
n
m
en
ts
ar
e
lim
ited
,
an
d
en
s
u
r
in
g
e
q
u
ity
an
d
f
air
n
ess
ac
r
o
s
s
d
if
f
er
en
t
d
em
o
g
r
ap
h
ic
g
r
o
u
p
s
in
AI
-
d
r
iv
en
lu
n
g
ca
n
ce
r
d
iag
n
o
s
tics
is
a
m
ajo
r
f
u
t
u
r
e
n
ee
d
.
3.
M
E
T
H
O
D
3
.
1
.
Da
t
a
c
o
llect
io
n
T
h
e
L
I
DC
-
I
DR
I
co
llectio
n
co
m
p
r
is
es
r
ad
io
lo
g
is
ts
’
n
o
d
u
les
ca
teg
o
r
ized
in
lu
n
g
C
T
im
ag
es,
wh
ich
in
clu
d
es
d
em
o
g
r
ap
h
ics
an
d
c
lin
ical
h
is
to
r
y
.
T
h
e
Kag
g
le
d
ataset
s
h
o
ws
lu
n
g
C
T
im
ag
e
s
with
o
r
with
o
u
t
ca
n
ce
r
.
I
n
m
o
d
el
co
n
s
tr
u
ctio
n
,
s
tan
d
ar
d
m
ac
h
in
e
lea
r
n
in
g
an
d
m
o
d
er
n
d
ee
p
lear
n
in
g
m
o
d
els
ar
e
u
s
ed
a
n
d
tr
ain
ed
.
E
ac
h
s
o
lu
tio
n
h
as
ad
v
an
tag
es
an
d
m
ee
ts
d
esig
n
g
o
als.
L
o
g
is
tic
r
eg
r
ess
io
n
(
L
R
)
an
d
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
(
SVM)
ar
e
f
u
n
d
am
e
n
tal
m
o
d
els.
Pre
p
r
o
ce
s
s
in
g
p
r
o
v
id
es
h
u
m
an
-
d
ef
in
e
d
lu
n
g
n
o
d
u
le
s
ize,
s
h
ap
e,
an
d
tex
tu
r
e
f
o
r
th
ese
alg
o
r
ith
m
s
.
Alth
o
u
g
h
co
m
p
u
tatio
n
all
y
ef
f
icien
t
an
d
in
ter
p
r
etab
le,
th
ese
m
o
d
els
s
tr
u
g
g
le
with
m
ed
ical
im
ag
in
g
'
s
co
m
p
l
ex
p
atter
n
s
.
T
h
is
s
tu
d
y
an
aly
ze
s
clin
ically
em
p
lo
y
e
d
d
iag
n
o
s
tic
m
eth
o
d
s
s
u
ch
as
ch
est
X
-
r
ay
s
,
C
T
s
ca
n
s
,
an
d
b
io
p
s
ies,
wh
ich
ar
e
in
v
asiv
e,
co
s
tly
,
an
d
n
ee
d
ex
p
er
t
in
ter
p
r
etatio
n
.
R
ec
en
tly
d
ev
elo
p
ed
d
ee
p
lear
n
in
g
alg
o
r
ith
m
s
,
s
u
ch
as
C
NNs,
R
esNet
,
an
d
VGG1
6
,
o
f
f
er
a
u
to
m
ated
,
n
o
n
-
in
v
asiv
e,
a
n
d
ef
f
icien
t
d
iag
n
o
s
tic
s
o
lu
tio
n
s
.
Acc
u
r
ac
y
,
s
en
s
itiv
ity
,
an
d
s
ca
lab
ilit
y
ar
e
em
p
lo
y
e
d
to
ass
ess
th
e
ad
v
a
n
tag
es
an
d
d
is
ad
v
a
n
tag
es
o
f
th
e
ap
p
r
o
ac
h
,
wh
ile
r
o
b
u
s
tn
ess
an
d
r
ea
l
-
wo
r
ld
r
elev
a
n
ce
ar
e
en
s
u
r
ed
th
r
o
u
g
h
v
alid
at
io
n
o
n
b
en
ch
m
a
r
k
d
atasets
s
u
ch
as
L
I
DC
-
I
DR
I
an
d
th
e
Kag
g
le
l
u
n
g
ca
n
ce
r
d
ataset.
T
h
is
s
tu
d
y
aim
s
to
e
lu
cid
ate
h
o
w
n
o
v
el
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
Dee
p
lea
r
n
in
g
fo
r
lu
n
g
ca
n
ce
r
d
ia
g
n
o
s
is
:
a
co
mp
a
r
a
tive
a
r
ti
ficia
l
…
(
P
h
a
n
ee
n
d
r
a
V
a
r
ma
C
h
in
ta
la
p
a
ti
)
3123
m
eth
o
d
o
l
o
g
ies
s
u
r
p
ass
tr
a
d
itio
n
al
o
n
es
b
y
ex
a
m
in
in
g
p
er
f
o
r
m
an
ce
d
is
p
ar
ities
an
d
id
en
tify
in
g
o
p
tim
al
clin
ica
l
alg
o
r
ith
m
s
f
o
r
th
e
ea
r
ly
id
e
n
tific
atio
n
an
d
d
iag
n
o
s
is
o
f
lu
n
g
ca
n
ce
r
.
T
h
is
s
tu
d
y
e
m
p
lo
y
s
a
f
o
u
r
-
s
tag
e
f
r
am
ewo
r
k
:
d
ata
co
llectio
n
,
p
r
e
-
p
r
o
ce
s
s
in
g
,
m
o
d
el
d
e
v
elo
p
m
en
t,
an
d
p
er
f
o
r
m
an
c
e
ev
alu
atio
n
.
T
h
is
f
r
am
ewo
r
k
en
s
u
r
es r
o
b
u
s
tn
ess
,
r
ep
r
o
d
u
cib
ilit
y
,
a
n
d
s
ca
lab
ilit
y
f
o
r
L
C
D
.
T
h
e
wo
r
k
f
lo
w
f
r
a
m
ewo
r
k
o
f
th
is
s
tu
d
y
is
p
r
esen
ted
in
Fig
u
r
e
1
,
p
r
o
v
id
in
g
a
clea
r
a
n
d
s
tr
u
ctu
r
e
d
r
ep
r
esen
tatio
n
o
f
th
e
co
m
p
lete
ex
p
er
im
en
tal
an
d
s
im
u
latio
n
p
ip
elin
e
to
e
n
s
u
r
e
r
e
p
r
o
d
u
cib
ilit
y
.
Fig
u
r
e
1
illu
s
tr
ates
d
ata
ac
q
u
is
itio
n
,
p
r
ep
r
o
ce
s
s
in
g
,
m
o
d
el
d
e
v
elo
p
m
en
t,
p
er
f
o
r
m
a
n
ce
ev
alu
atio
n
,
an
d
o
p
tim
al
m
o
d
el
s
elec
tio
n
f
o
r
L
C
D
.
At
th
e
o
u
ts
et,
C
T
im
ag
es
wer
e
o
b
tai
n
ed
f
r
o
m
t
h
e
L
I
DC
-
I
DR
I
d
ataset
with
ex
p
er
t
an
n
o
tatio
n
s
a
n
d
s
u
p
p
lem
en
te
d
with
clin
ical
d
ata
f
r
o
m
t
h
e
Kag
g
le
lu
n
g
ca
n
ce
r
d
ataset.
T
h
e
d
ata
wer
e
o
r
g
an
ized
an
d
s
p
lit
in
to
tr
ai
n
in
g
,
v
alid
atio
n
,
an
d
test
in
g
s
ets
to
en
s
u
r
e
u
n
b
iased
ev
alu
atio
n
.
T
h
e
d
ata
p
r
e
-
p
r
o
ce
s
s
in
g
s
tag
e
in
clu
d
e
s
a
p
r
e
-
p
r
o
ce
s
s
in
g
[
2
7
]
,
wh
ich
in
clu
d
es
c
o
n
tr
ast
en
h
a
n
ce
m
en
t,
h
is
to
g
r
a
m
eq
u
aliza
tio
n
,
Gau
s
s
ian
,
an
d
m
ed
ian
f
ilter
in
g
f
o
r
n
o
is
e
r
em
o
v
al,
an
d
p
ix
el
n
o
r
m
aliza
tio
n
to
s
tan
d
ar
d
ize
i
n
p
u
ts
ac
r
o
s
s
d
atasets
.
T
h
e
d
atasets
wer
e
s
y
s
tem
atica
lly
cu
r
ated
an
d
in
teg
r
ated
t
o
en
s
u
r
e
s
u
f
f
ic
ien
t
d
iv
er
s
ity
an
d
r
o
b
u
s
tn
ess
f
o
r
m
o
d
el
d
ev
el
o
p
m
en
t.
T
h
e
co
n
s
o
lid
ated
d
atase
t
was
s
u
b
s
eq
u
en
tly
p
ar
titi
o
n
ed
in
to
th
r
ee
s
u
b
s
ets,
co
m
p
r
is
in
g
7
0
%
f
o
r
tr
ai
n
in
g
,
1
5
%
f
o
r
v
alid
atio
n
,
a
n
d
1
5
%
f
o
r
test
in
g
.
T
h
is
s
tr
atif
ied
d
iv
is
io
n
f
ac
ilit
ates
ef
f
ec
tiv
e
m
o
d
el
lear
n
i
n
g
,
e
n
ab
les
h
y
p
er
p
ar
am
eter
tu
n
in
g
th
r
o
u
g
h
v
alid
atio
n
,
an
d
en
s
u
r
es
an
u
n
b
iased
ev
alu
atio
n
o
f
m
o
d
el
p
er
f
o
r
m
a
n
ce
o
n
u
n
s
ee
n
d
ata,
th
er
eb
y
m
in
im
izin
g
th
e
r
is
k
o
f
o
v
er
f
it
tin
g
an
d
en
h
an
ci
n
g
g
en
er
aliza
tio
n
ca
p
ab
ilit
y
.
Fig
u
r
e
1
.
C
o
m
p
r
eh
en
s
iv
e
wo
r
k
f
lo
w
o
f
t
h
e
p
r
o
p
o
s
ed
e
x
p
er
i
m
en
tal
an
d
s
im
u
latio
n
wo
r
k
f
l
o
w
I
n
a
b
r
o
ad
e
r
s
en
s
e,
in
m
ed
ical
im
ag
es
(
e.
g
.
,
C
T
s
ca
n
s
)
,
p
ix
el
v
alu
es
ca
n
v
ar
y
wid
e
ly
d
u
e
to
d
if
f
er
en
ce
s
in
s
ca
n
n
er
s
,
ac
q
u
is
itio
n
s
ettin
g
s
,
o
r
p
atien
t
co
n
d
itio
n
s
.
Pix
el
s
ca
lin
g
n
o
r
m
a
lizatio
n
tr
an
s
f
o
r
m
s
th
ese
v
alu
es
in
to
a
f
ix
e
d
r
an
g
e
-
co
m
m
o
n
ly
[
0
,
1
]
o
r
[
-
1
,
1
]
,
u
s
in
g
a
lin
ea
r
s
ca
lin
g
o
p
er
ati
o
n
[
2
8
]
.
T
h
is
h
elp
s
s
tab
ilize
tr
ain
in
g
im
p
r
o
v
es
co
n
v
er
g
e
n
ce
o
f
m
a
ch
in
e
lear
n
in
g
an
d
d
ee
p
lear
n
in
g
m
o
d
els.
B
y
d
o
in
g
th
is
,
it
is
o
n
ce
ag
ain
r
eiter
ated
th
at
p
ix
e
l
in
ten
s
ity
n
o
r
m
aliza
tio
n
was
p
er
f
o
r
m
ed
b
y
lin
ea
r
ly
s
ca
lin
g
im
ag
e
v
alu
es
to
th
e
r
an
g
e
[
0
,
1
]
to
en
s
u
r
e
u
n
if
o
r
m
in
p
u
t d
is
tr
ib
u
tio
n
ac
r
o
s
s
d
atasets
.
I
n
th
e
m
o
d
el
d
ev
elo
p
m
en
t
s
tag
e,
b
o
t
h
tr
ad
itio
n
al
m
ac
h
in
e
lear
n
in
g
m
o
d
els
(
L
R
an
d
SVM)
an
d
d
ee
p
lear
n
in
g
m
o
d
els
(
C
NNs
with
R
esNet
an
d
VGG1
6
ar
ch
itect
u
r
es)
wer
e
d
ev
elo
p
e
d
.
Dee
p
m
o
d
els
wer
e
tr
ain
e
d
u
s
in
g
b
ac
k
p
r
o
p
ag
atio
n
with
s
to
ch
asti
c
g
r
ad
ien
t
d
escen
t,
with
h
y
p
er
p
ar
a
m
eter
s
o
p
tim
ized
v
ia
v
alid
atio
n
.
Fin
ally
,
m
o
d
els
wer
e
ev
alu
ated
u
s
in
g
ac
cu
r
ac
y
,
s
en
s
itiv
ity
,
s
p
ec
if
icity
,
F1
-
s
co
r
e,
an
d
ar
ea
u
n
d
er
th
e
c
u
r
v
e
-
r
ec
eiv
er
o
p
er
atin
g
ch
a
r
ac
ter
is
tic
(
AUC
-
R
O
C
)
,
en
ab
lin
g
a
f
air
co
m
p
ar
is
o
n
o
f
d
iag
n
o
s
tic
p
er
f
o
r
m
an
ce
an
d
clin
ical
r
eliab
ilit
y
.
T
h
e
wo
r
k
f
l
o
w
f
o
llo
ws
a
lo
g
ical
p
r
o
g
r
ess
io
n
f
r
o
m
d
ata
ac
q
u
is
itio
n
to
f
in
al
m
o
d
el
s
elec
tio
n
,
with
clea
r
in
ter
co
n
n
ec
tio
n
s
b
e
twee
n
s
tag
es,
en
s
u
r
in
g
tr
an
s
p
ar
en
cy
an
d
ea
s
e
o
f
r
ep
licatio
n
.
T
h
e
in
teg
r
atio
n
o
f
th
is
v
is
u
al
r
ep
r
esen
tatio
n
with
th
e
d
etailed
m
eth
o
d
o
l
o
g
i
ca
l
d
escr
ip
tio
n
p
r
o
v
i
d
es
a
co
m
p
r
eh
e
n
s
iv
e
an
d
r
ep
r
o
d
u
cib
le
ex
p
er
im
en
tal
f
r
a
m
ewo
r
k
f
o
r
L
C
D
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
5
,
No
.
4
,
Au
g
u
s
t 2
0
2
6
:
3
1
2
0
-
3
1
3
0
3124
3
.
2
.
Arc
hite
ct
ure
dia
g
ra
m
f
o
r
lun
g
ca
ncer
det
ec
t
io
n
T
h
e
d
esig
n
em
p
h
asizes
d
ata
co
llected
f
r
o
m
r
eliab
le
s
o
u
r
ce
s
,
u
tili
zin
g
d
atasets
f
r
o
m
L
I
DC
-
I
DR
I
an
d
Kag
g
le.
Fig
u
r
e
2
s
h
o
ws
lu
n
g
C
T
im
ag
in
g
d
ata
f
lo
w
f
r
o
m
co
llectio
n
to
m
o
d
el
d
ep
lo
y
m
en
t
.
T
h
e
an
n
o
tated
lu
n
g
C
T
s
ca
n
s
f
r
o
m
s
k
illed
r
ad
io
l
o
g
is
ts
ar
e
in
L
I
DC
-
I
DR
I
.
I
t
i
n
co
r
p
o
r
ates
im
p
o
r
tan
t
m
etad
a
ta
s
u
ch
as
p
atien
t
d
em
o
g
r
a
p
h
ics,
clin
ical
h
is
to
r
y
,
an
d
n
o
d
u
le
ch
ar
ac
ter
is
tics
f
o
r
m
o
d
el
d
ev
el
o
p
m
e
n
t.
T
h
e
Kag
g
le
d
ataset
co
n
tain
s
ca
n
ce
r
-
a
n
n
o
tate
d
lu
n
g
C
T
im
ag
es.
T
h
e
d
ataset
is
u
s
ef
u
l
f
o
r
cr
ea
tin
g
an
d
test
in
g
d
iag
n
o
s
tic
alg
o
r
ith
m
s
.
Data
is
o
r
g
an
ized
an
d
k
ep
t
in
a
d
atab
ase
o
r
d
ata
lak
e
f
o
r
an
aly
s
is
af
ter
g
ath
er
in
g
.
Data
m
u
s
t
b
e
well
-
o
r
g
an
ized
f
o
r
d
e
v
elo
p
m
e
n
t,
u
s
ab
ilit
y
,
a
n
d
tr
ac
ea
b
ilit
y
.
R
aw
d
ata
m
u
s
t
b
e
p
r
ep
r
o
ce
s
s
ed
f
o
r
m
o
d
el
in
p
u
t
to
en
s
u
r
e
co
n
s
is
ten
cy
an
d
p
ic
tu
r
e
q
u
ality
.
T
h
is
p
h
ase
u
s
es
m
u
ltip
le
m
eth
o
d
s
to
h
ig
h
lig
h
t
lu
n
g
n
o
d
u
les
an
d
r
ed
u
ce
ar
tifa
cts,
wh
ich
in
clu
d
e
:
i)
I
m
ag
e
en
h
a
n
ce
m
en
t:
h
is
to
g
r
am
eq
u
aliza
tio
n
an
d
co
n
tr
ast
m
o
d
if
icatio
n
h
elp
id
en
tif
y
lu
n
g
n
o
d
u
les.
E
n
h
an
ce
d
p
h
o
to
s
h
elp
th
e
m
o
d
el
f
o
cu
s
o
n
k
ey
tr
aits
,
im
p
r
o
v
in
g
d
iag
n
o
s
is
ac
cu
r
ac
y
.
ii)
No
is
e
r
ed
u
ctio
n
:
g
au
s
s
ian
an
d
m
ed
ian
f
ilter
s
r
ed
u
ce
C
T
im
ag
e
n
o
is
e
wh
ile
p
r
eser
v
in
g
ed
g
es
an
d
tex
tu
r
in
g
.
T
h
is
en
s
u
r
es n
o
is
e
d
o
es n
o
t im
p
ed
e
th
e
m
o
d
el'
s
ab
ilit
y
to
d
etec
t m
ea
n
in
g
f
u
l p
atte
r
n
s
.
iii)
No
r
m
aliza
tio
n
:
all
p
h
o
to
s
a
r
e
n
o
r
m
alize
d
to
en
s
u
r
e
p
ix
el
v
alu
es
ar
e
with
in
a
r
an
g
e,
s
u
ch
as
0
to
1
.
Un
if
o
r
m
ly
s
ca
lin
g
in
p
u
t
d
ata
s
p
ee
d
s
u
p
an
d
s
tab
ilizes
m
o
d
el
tr
ain
in
g
co
n
v
er
g
en
ce
.
T
h
ese
p
r
ep
ar
atio
n
m
eth
o
d
s
p
r
o
v
id
e
d
ata
q
u
ality
f
o
r
f
ea
tu
r
e
ex
tr
ac
tio
n
an
d
m
o
d
e
l tr
ain
in
g
,
en
a
b
lin
g
ac
c
u
r
ate
p
r
ed
ictio
n
s
.
Fig
u
r
e
2
.
L
u
n
g
C
T
im
a
g
in
g
d
a
ta
f
lo
w:
f
r
o
m
c
o
llectio
n
to
m
o
d
el
d
ep
lo
y
m
en
t
3
.
3
.
M
o
del dev
elo
pm
ent
Featu
r
e
ex
tr
ac
tio
n
is
cr
u
cial
to
th
e
d
etec
tio
n
o
f
lu
n
g
ca
n
ce
r
b
ec
au
s
e
it
tr
an
s
f
o
r
m
s
r
aw
m
ed
ical
im
ag
in
g
d
ata
in
to
m
ea
n
i
n
g
f
u
l
n
u
m
e
r
ical
r
ep
r
esen
tatio
n
s
th
a
t
ca
n
b
e
an
aly
ze
d
b
y
m
ac
h
in
e
lear
n
in
g
an
d
d
ee
p
lear
n
in
g
alg
o
r
ith
m
s
.
W
h
ile
o
ld
er
ap
p
r
o
ac
h
es
n
ee
d
h
u
m
a
n
f
ea
tu
r
e
en
g
i
n
ee
r
in
g
b
ased
o
n
d
o
m
ain
ex
p
er
tis
e,
m
o
d
er
n
d
ee
p
lear
n
in
g
s
y
s
tem
s
au
to
m
atica
lly
id
en
tify
f
ea
tu
r
es
f
r
o
m
th
e
d
ata.
T
h
is
p
ap
er
p
r
o
v
id
es
a
d
etailed
ex
p
lan
atio
n
o
f
th
e
f
ea
tu
r
e
e
x
tr
ac
tio
n
p
r
o
ce
s
s
es a
n
d
r
elate
d
m
ath
em
atica
l f
o
r
m
u
las.
3
.
2
.
1
.
L
o
g
is
t
ic
re
g
re
s
s
io
n f
ea
t
ure
ex
t
ra
ct
io
n
LR
p
r
ed
icts
th
e
p
r
o
b
ab
ilit
y
o
f
o
u
tco
m
e
(
=
1
∣
)
b
ased
o
n
in
p
u
t
ch
ar
ac
ter
is
tics
in
b
in
ar
y
class
if
icatio
n
[
2
8
]
.
T
h
e
p
r
o
b
ab
ilit
y
o
f
a
b
in
a
r
y
o
u
tco
m
e
is
esti
m
ated
u
s
in
g
LR
,
as e
x
p
r
ess
ed
in
(
1
)
.
(
=
1
∣
)
=
1
1
+
−
(
0
+
1
1
+
2
2
+
⋯
+
)
(
1
)
W
h
er
e
is
m
o
d
el
co
e
f
f
icien
ts
d
eter
m
in
ed
d
u
r
in
g
tr
ain
in
g
an
d
is
in
p
u
t
f
ea
tu
r
es
ex
tr
ac
te
d
f
r
o
m
m
e
d
ical
im
ag
es.
E
x
p
er
ts
m
an
u
ally
cr
e
ate
LR
f
ea
tu
r
es
,
in
clu
d
in
g
i)
s
ize:
t
h
e
lu
n
g
C
T
s
ca
n
n
o
d
u
le
ar
ea
o
r
v
o
lu
m
e
a
n
d
ii)
tex
tu
r
e:
t
h
e
g
r
ay
lev
el
c
o
-
o
cc
u
r
r
e
n
ce
m
atr
i
x
(
GL
C
M)
m
ea
s
u
r
es
p
ix
el
i
n
ten
s
ity
ass
o
ciatio
n
s
.
T
ex
t
-
r
elate
d
f
ea
tu
r
es a
r
e
ex
tr
ac
te
d
u
s
in
g
th
e
GL
C
M,
wh
ich
is
d
ef
in
ed
in
(
2
)
.
(
,
∣
,
)
(
2
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
Dee
p
lea
r
n
in
g
fo
r
lu
n
g
ca
n
ce
r
d
ia
g
n
o
s
is
:
a
co
mp
a
r
a
tive
a
r
ti
ficia
l
…
(
P
h
a
n
ee
n
d
r
a
V
a
r
ma
C
h
in
ta
la
p
a
ti
)
3125
T
h
is
m
atr
ix
r
e
p
r
esen
ts
th
e
f
r
eq
u
en
cy
o
f
p
ix
el
-
i
n
ten
s
ity
p
a
ir
s
i
an
d
j
o
cc
u
r
r
in
g
at
d
is
ta
n
ce
d
a
n
d
an
g
le
θ.
Sh
ap
e:
d
ef
in
ed
b
y
m
etr
ics
lik
e
co
m
p
ac
tn
ess
as
(
3
)
.
T
h
e
s
h
a
p
e
f
ea
tu
r
es
ar
e
q
u
an
tifie
d
u
s
in
g
th
e
c
o
m
p
ac
tn
ess
m
etr
ic,
as g
iv
en
in
(
3
)
.
=
2
4
(
3
)
W
h
er
e
is
th
e
p
er
im
eter
an
d
is
th
e
ar
ea
.
3
.
2
.
2
.
Su
pp
o
rt
v
ec
t
o
r
m
a
chi
nes
f
ea
t
ure
ex
t
ra
ct
io
n
SVM
f
in
d
s
th
e
o
p
tim
al
h
y
p
e
r
p
lan
e
s
ep
ar
atin
g
two
class
es
.
T
h
e
ex
tr
ac
ted
f
ea
tu
r
es
ar
e
s
u
b
s
eq
u
en
tly
n
o
r
m
alize
d
ac
c
o
r
d
in
g
to
(
4
)
.
+
=
0
(
4
)
W
h
er
e
is
w
eig
h
t
v
ec
to
r
d
eter
m
in
in
g
th
e
o
r
ie
n
tatio
n
o
f
th
e
h
y
p
er
p
lan
e
;
is
b
ias
ad
ju
s
tin
g
th
e
h
y
p
er
p
lan
e'
s
p
o
s
itio
n
;
an
d
is
i
n
p
u
t
f
ea
tu
r
e
v
ec
to
r
r
ep
r
esen
tin
g
d
ata
p
o
in
ts
.
F
ea
tu
r
es
f
o
r
SVM
ar
e
d
er
iv
e
d
t
h
r
o
u
g
h
p
r
ep
r
o
ce
s
s
in
g
an
d
s
tatis
tical
an
aly
s
is
:
i)
I
n
ten
s
ity
h
is
to
g
r
am
s
:
ca
p
tu
r
es
p
ix
el
in
ten
s
ity
d
is
tr
ib
u
tio
n
as d
ef
in
e
in
(
5
)
.
(
)
=
ℎ
(
5
)
ii)
T
ex
tu
r
e
f
ea
tu
r
es:
th
ese
ar
e
d
er
iv
ed
f
r
o
m
s
tatis
tical
d
escr
i
p
to
r
s
lik
e
co
n
tr
ast,
co
r
r
elatio
n
,
an
d
e
n
tr
o
p
y
ca
lcu
lated
f
r
o
m
th
e
GL
C
M.
3
.
2
.
3
.
Co
nv
o
lutio
na
l
neura
l
net
wo
rk
s
f
ea
t
ure
ex
t
ra
ct
io
n
C
NNs
au
to
m
ate
th
e
f
ea
tu
r
e
ex
tr
ac
tio
n
p
r
o
ce
s
s
u
s
in
g
co
n
v
o
lu
tio
n
al
o
p
er
atio
n
s
.
E
ac
h
co
n
v
o
lu
tio
n
al
lay
er
ap
p
lies
a
k
er
n
el
to
th
e
in
p
u
t im
ag
e
[
2
]
as d
e
f
in
e
in
(
6
)
.
,
=
∑
∑
+
,
+
×
−
1
=
0
−
1
=
0
,
+
(
6
)
W
h
er
e
,
is
f
ea
tu
r
e
m
ap
v
alu
e
at
p
o
s
itio
n
(
,
)
f
o
r
k
er
n
el
k
;
+
,
+
is
p
ix
e
l
in
ten
s
ity
in
th
e
r
ec
ep
tiv
e
f
iel
d
;
,
is
w
eig
h
ts
o
f
th
e
co
n
v
o
lu
tio
n
k
er
n
el
; a
n
d
is
b
ias ter
m
.
C
NN
s
ex
tr
ac
t h
ier
ar
ch
ical
f
ea
t
u
r
es
,
in
clu
d
in
g
:
i)
L
o
w
-
lev
el
f
ea
tu
r
es:
ed
g
es
,
g
r
a
d
ien
ts
,
an
d
tex
tu
r
es.
ii)
Hig
h
-
lev
el
f
ea
tu
r
es:
s
h
ap
es
,
p
atter
n
s
,
an
d
s
em
an
tic
s
tr
u
ctu
r
e
s
o
f
lu
n
g
n
o
d
u
les.
3
.
2
.
4
.
ResNet
f
ea
t
ure
ex
t
ra
ct
io
n
R
esNet
ad
d
r
ess
es
th
e
v
an
is
h
i
n
g
g
r
ad
ie
n
t
p
r
o
b
le
m
b
y
in
tr
o
d
u
cin
g
r
esid
u
al
co
n
n
ec
tio
n
s
.
T
h
e
r
esid
u
al
m
ap
p
in
g
is
r
ep
r
esen
ted
as
(
7
)
.
=
(
,
{
}
)
+
(
7
)
W
h
er
e
is
r
esid
u
al
m
ap
p
in
g
an
d
is
i
n
p
u
t
f
ea
tu
r
es
p
ass
ed
t
o
th
e
n
ex
t
lay
er
.
R
esNet
f
o
cu
s
es
o
n
ex
tr
ac
tin
g
f
in
e
-
g
r
ain
e
d
d
etails,
s
u
ch
as
s
u
b
tle
tex
tu
r
e
v
ar
iatio
n
s
in
lu
n
g
n
o
d
u
les
an
d
ir
r
e
g
u
lar
ed
g
es,
wh
ich
ar
e
cr
itical
f
o
r
ea
r
ly
-
s
tag
e
d
etec
tio
n
.
3
.
2
.
5
.
VG
G
1
6
f
ea
t
ure
e
x
t
ra
c
t
io
n
VGG1
6
em
p
lo
y
s
f
ix
e
d
c
o
n
v
o
l
u
tio
n
al
f
ilter
s
o
f
s
ize
3
×3
wit
h
r
ec
tifie
d
lin
ea
r
u
n
it
(
R
eL
U
)
act
iv
atio
n
to
in
tr
o
d
u
ce
n
o
n
-
lin
ea
r
ity
as d
ef
in
e
in
(
8
)
.
(
)
=
(
0
,
)
(
8
)
Po
o
lin
g
lay
e
r
s
d
o
w
n
s
am
p
le
f
e
atu
r
e
m
ap
s
,
p
r
eser
v
i
n
g
k
ey
f
e
atu
r
es
wh
ile
r
ed
u
cin
g
co
m
p
u
t
atio
n
al
co
m
p
lex
ity
.
Hier
ar
ch
ical
f
ea
tu
r
e
e
x
tr
ac
tio
n
in
clu
d
es:
i)
L
o
w
-
lev
el
f
ea
tu
r
es:
tex
tu
r
es
an
d
ed
g
es.
ii)
Mid
-
lev
el
f
ea
tu
r
es:
s
h
ap
es
an
d
p
atter
n
s
.
iii)
Hig
h
-
lev
el
f
ea
tu
r
es:
n
o
d
u
le
s
tr
u
ctu
r
es a
n
d
c
o
n
tex
t.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
5
,
No
.
4
,
Au
g
u
s
t 2
0
2
6
:
3
1
2
0
-
3
1
3
0
3126
3
.
2
.
6
.
YO
L
O
v
5
f
e
a
t
ure
ex
t
r
a
ct
io
n
YOL
Ov
5
p
r
ed
icts
b
o
u
n
d
i
n
g
b
o
x
es a
n
d
class
p
r
o
b
ab
ilit
ies in
a
s
in
g
le
p
ass
as in
(
9
)
.
(
∣
)
⋅
⋅
⋅
⋅
ℎ
(
9
)
W
h
er
e
(
∣
)
is
c
las
s
p
r
o
b
ab
ilit
y
an
d
,
,
is
a
b
o
u
n
d
in
g
b
o
x
p
a
r
am
ete
r
(
co
o
r
d
i
n
ates
an
d
d
im
en
s
io
n
s
)
.
YOL
O
d
etec
ts
s
p
atial
f
ea
tu
r
es
an
d
p
r
ed
icts
b
o
u
n
d
in
g
b
o
x
es,
d
ir
ec
tly
id
en
tify
i
n
g
lu
n
g
n
o
d
u
l
e
lo
ca
tio
n
s
[
3
]
.
3
.
2
.
7
.
F
ea
t
ure
enha
ncem
ent
t
ec
hn
iqu
e
s
Featu
r
e
en
h
an
ce
m
e
n
t
tech
n
iq
u
es
im
p
r
o
v
e
C
T
im
ag
e
q
u
alit
y
b
y
en
h
an
cin
g
c
o
n
tr
ast,
r
ed
u
cin
g
n
o
is
e,
an
d
p
r
eser
v
in
g
im
p
o
r
tan
t
f
ea
tu
r
es f
o
r
ac
c
u
r
ate
lu
n
g
ca
n
ce
r
c
lass
if
icatio
n
.
i)
His
to
g
r
am
eq
u
aliza
tio
n
: im
p
r
o
v
es
co
n
tr
ast in
C
T
im
ag
es
as d
ef
in
e
in
(
1
0
)
.
′
(
,
)
=
(
(
,
)
)
−
(
)
(
)
−
(
)
(
1
0
)
ii)
Gau
s
s
ian
f
ilter
in
g
: r
ed
u
ce
s
n
o
is
e
wh
ile
p
r
eser
v
in
g
cr
itical
ed
g
es
as d
ef
in
e
in
(
1
1
)
.
(
,
)
=
1
2
2
(
−
2
+
2
2
2
)
(
1
1
)
T
h
is
ar
ticle
d
escr
ib
es
m
ath
em
atica
l
f
r
am
ewo
r
k
s
an
d
f
ea
t
u
r
e
ex
tr
ac
tio
n
m
et
h
o
d
s
f
o
r
lu
n
g
ca
n
ce
r
d
etec
tio
n
alg
o
r
ith
m
s
.
T
h
ese
tech
n
iq
u
es
r
a
n
g
e
f
r
o
m
L
R
t
o
R
esNet
an
d
YOL
Ov
5
.
T
h
ey
em
p
h
asize
th
e
r
elev
an
ce
o
f
f
ea
tu
r
e
ex
tr
ac
tio
n
in
im
p
r
o
v
in
g
d
iag
n
o
s
tic
ac
cu
r
ac
y
an
d
r
eliab
ilit
y
[
3
]
.
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
I
n
Fig
u
r
e
3
,
LR
,
SVM,
C
NN,
R
esNet,
an
d
VGG1
6
lu
n
g
ca
n
ce
r
d
iag
n
o
s
tic
ac
cu
r
ac
y
ar
e
g
r
ap
h
e
d
.
I
t
ca
n
b
e
u
n
d
er
s
to
o
d
th
at
d
ee
p
lear
n
in
g
is
m
o
r
e
ac
cu
r
ate
th
an
m
ac
h
in
e
lear
n
in
g
.
T
h
e
p
o
o
r
est m
eth
o
d
is
LR
,
wi
th
7
5
%
ac
cu
r
ac
y
.
L
in
ea
r
l
y
s
ep
ar
ab
le
d
ata
p
er
f
o
r
m
s
well
with
s
im
p
le
LR
.
W
h
en
d
ata
p
atter
n
s
ar
e
h
i
g
h
ly
non
-
lin
ea
r
an
d
co
m
p
lex
(
e
.
g
.
,
n
o
d
u
les'
tex
tu
r
e
an
d
s
h
ap
e)
,
LR
f
ails
to
id
en
tif
y
s
ig
n
if
ican
t
co
n
n
ec
tio
n
s
b
etwe
en
v
ar
iab
les
in
L
C
D
.
T
h
e
s
y
s
tem
ca
n
n
o
t
d
is
tin
g
u
is
h
m
alig
n
an
t
f
r
o
m
n
o
n
-
ca
n
ce
r
o
u
s
ev
en
ts
b
ased
o
n
lim
ited
h
u
m
an
ch
ar
ac
ter
is
tics
,
lo
wer
in
g
ac
cu
r
ac
y
.
SVM
b
o
o
s
ts
ac
cu
r
ac
y
to
8
3
%.
SVM
m
ay
d
is
tin
g
u
is
h
m
alig
n
an
t
f
r
o
m
n
o
n
-
ca
n
ce
r
o
u
s
n
o
d
u
les
b
etter
th
a
n
LR
b
y
tr
an
s
f
er
r
in
g
i
n
p
u
t
d
ata
to
h
ig
h
er
-
d
im
en
s
io
n
al
s
p
ac
es.
SVM
ca
n
n
o
t
f
u
lly
r
ep
r
esen
t
lu
n
g
n
o
d
u
le
co
m
p
le
x
ity
s
in
ce
it
u
s
es
m
an
u
ally
g
en
er
ated
in
ten
s
ity
h
is
to
g
r
am
s
an
d
tex
tu
r
e
m
etr
ic
s
.
SVM
also
h
as
i
s
s
u
es
wi
th
m
ass
iv
e
d
atasets
,
lim
it
in
g
its
p
r
ac
ticality
.
Fro
m
r
aw
C
T
s
ca
n
p
ictu
r
es,
C
NN
s
l
ea
r
n
h
ier
ar
ch
ical
attr
ib
u
tes
u
s
in
g
co
n
v
o
lu
tio
n
al,
p
o
o
lin
g
,
an
d
ac
tiv
atio
n
lay
er
s
.
SVMs
an
d
LR
n
ee
d
f
ea
tu
r
e
en
g
in
ee
r
in
g
,
wh
e
r
ea
s
C
NNs
ca
n
ca
p
tu
r
e
ed
g
es,
tex
tu
r
es,
s
h
ap
e
s
,
an
d
co
m
p
licated
n
o
d
u
lar
s
tr
u
ctu
r
es.
C
NNs id
en
tify
lu
n
g
ca
n
ce
r
well
d
u
e
to
im
ag
e
-
b
ased
d
ata
f
lex
ib
ilit
y
.
Fig
u
r
e
3
.
C
o
m
p
a
r
ativ
e
g
r
ap
h
o
f
th
e
alg
o
r
ith
m
s
: a
d
ef
i
n
itiv
e
an
aly
s
is
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
Dee
p
lea
r
n
in
g
fo
r
lu
n
g
ca
n
ce
r
d
ia
g
n
o
s
is
:
a
co
mp
a
r
a
tive
a
r
ti
ficia
l
…
(
P
h
a
n
ee
n
d
r
a
V
a
r
ma
C
h
in
ta
la
p
a
ti
)
3127
R
esNet
h
as
th
e
b
est
ac
cu
r
ac
y
(
9
4
%),
in
d
icatin
g
it
ca
n
h
an
d
le
d
ee
p
er
n
etwo
r
k
c
o
n
f
ig
u
r
atio
n
s
ef
f
icien
tly
.
R
esNet
'
s
m
ain
in
n
o
v
atio
n
is
r
esid
u
al
co
n
n
ec
ti
o
n
s
to
o
v
er
c
o
m
e
th
e
d
ee
p
n
etwo
r
k
'
s
v
an
is
h
in
g
g
r
ad
ien
t
p
r
o
b
lem
.
T
h
ese
co
n
n
ec
tio
n
s
allo
w
th
e
m
o
d
el
to
tr
an
s
m
it
g
r
a
d
ien
ts
,
p
r
eser
v
in
g
lu
n
g
n
o
d
u
le
f
in
e
-
g
r
ain
e
d
f
ea
tu
r
es
lik
e
tex
t
u
r
e
v
a
r
iatio
n
s
an
d
u
n
ev
en
ed
g
es
th
r
o
u
g
h
o
u
t
f
ea
tu
r
e
ex
tr
ac
ti
o
n
.
M
o
r
e
p
r
o
f
o
u
n
d
ar
ch
itectu
r
e
h
elp
s
R
esNet
r
ec
o
g
n
ize
c
o
m
p
licated
p
atter
n
s
an
d
en
h
an
ce
class
if
icatio
n
in
co
m
p
le
x
s
ce
n
ar
io
s
with
s
m
all
o
r
i
n
d
is
tin
ct
n
o
d
u
l
es.
C
o
n
v
o
lu
tio
n
al
d
ee
p
n
etwo
r
k
VGG1
6
is
less
ac
cu
r
ate
th
a
n
R
esNet
,
b
u
t,
lik
e
C
NN,
at
9
2
.
5
%.
VGG1
6
ca
p
tu
r
es
h
ier
ar
ch
ical
in
f
o
r
m
atio
n
u
s
in
g
1
6
co
n
v
o
lu
tio
n
al
lay
er
s
b
u
t
lack
s
r
esid
u
al
co
n
n
ec
tio
n
s
,
m
a
k
in
g
R
esNet
m
o
r
e
d
u
r
ab
le
in
d
ee
p
er
n
etwo
r
k
s
.
C
o
m
p
lex
d
at
a
p
a
tt
er
n
s
an
d
d
ee
p
n
e
two
r
k
s
b
ar
e
ly
a
f
f
e
ct
VG
G1
6
.
I
t
s
r
e
li
ab
l
e
d
e
s
ig
n
a
n
d
f
ea
tu
r
e
ex
tr
ac
tio
n
m
ak
e
it
a
to
p
lu
n
g
ca
n
c
er
d
et
ec
to
r
.
T
h
i
s
s
tu
d
y
r
e
v
ea
l
s
th
at
ad
v
a
n
ce
d
d
e
ep
lea
r
n
in
g
a
p
p
r
o
a
ch
e
s
ca
n
im
p
r
o
v
e
L
C
D
u
s
in
g
m
a
s
s
iv
e
m
ed
i
ca
l
im
ag
in
g
d
at
a
s
e
t
s
.
F
ig
u
r
e
4
an
d
T
a
b
l
e
2
co
m
p
a
r
e
a
c
cu
r
ac
y
,
s
en
s
i
t
iv
i
ty
,
s
p
e
cif
ic
i
ty
,
an
d
AU
C
-
R
O
C
f
o
r
LR
,
S
VM
,
C
NN
,
R
e
s
N
et,
an
d
VG
G1
6
.
T
h
e
tr
en
d
in
F
ig
u
r
e
4
s
h
o
ws
alg
o
r
it
h
m
i
c
p
er
f
o
r
m
an
c
e
f
r
o
m
c
la
s
s
i
ca
l
t
o
d
e
ep
le
ar
n
i
n
g
m
o
d
e
l
s
b
a
s
ed
o
n
ta
b
l
e
a
cc
u
r
ac
y
n
u
m
b
er
s
.
T
h
e
lo
w
e
s
t
ac
cu
r
a
cy
i
s
LR
a
t
7
5
.
2
%.
I
t
s
s
p
ec
if
i
ci
ty
(
7
8
.
4
%)
f
o
r
n
o
n
-
c
an
c
er
o
u
s
p
at
ien
t
s
ex
ce
ed
s
i
ts
s
en
s
i
ti
v
i
ty
(
7
2
.
1
%).
M
an
u
a
l
f
e
atu
r
e
en
g
i
n
ee
r
in
g
an
d
t
h
e
in
ab
i
l
ity
to
f
i
n
d
co
m
p
l
ic
at
ed
,
n
o
n
-
lin
ea
r
C
T
s
c
an
co
r
r
el
at
io
n
s
d
ec
r
e
a
s
e
i
ts
A
U
C
-
R
O
C
t
o
0
.
8
1
.
T
h
e
s
e
l
im
it
at
io
n
s
m
ak
e
i
t
i
n
ap
p
r
o
p
r
ia
te
f
o
r
a
d
v
an
ce
d
lu
n
g
ca
n
ce
r
d
e
tec
t
io
n
.
Fig
u
r
e
4
.
C
o
m
p
a
r
is
o
n
o
f
p
er
f
o
r
m
an
ce
m
etr
ics
T
ab
le
2
.
Per
f
o
r
m
an
ce
co
m
p
a
r
is
o
n
o
f
v
a
r
io
u
s
alg
o
r
ith
m
s
A
l
g
o
r
i
t
h
m
A
c
c
u
r
a
c
y
(
%)
S
e
n
s
i
t
i
v
i
t
y
(
%)
S
p
e
c
i
f
i
c
i
t
y
(
%)
AUC
-
R
O
C
LR
7
5
.
2
7
2
.
1
7
8
.
4
0
.
8
1
S
V
M
8
2
.
3
7
9
.
6
8
4
.
9
0
.
8
7
C
N
N
9
1
.
8
9
0
.
2
9
3
.
4
0
.
9
4
R
e
sN
e
t
9
4
.
2
9
2
.
7
9
5
.
8
0
.
9
6
V
G
G
1
6
9
2
.
6
9
1
.
0
9
4
.
2
0
.
9
5
An
a
d
v
a
n
c
ed
c
lass
i
ca
l
m
et
h
o
d
,
SV
M,
h
as
8
2
.
3
%
ac
c
u
r
a
cy
.
I
t
ca
n
i
d
e
n
t
if
y
ca
n
ce
r
f
r
o
m
n
o
n
-
ca
n
ce
r
o
u
s
in
s
tan
ce
s
with
7
9
.
6
%
s
en
s
itiv
ity
an
d
8
4
.
9
%
s
p
ec
if
icity
.
Ad
d
itio
n
ally
,
its
0
.
8
7
AUC
-
R
OC
h
elp
s
class
if
icatio
n
.
SVMs
'
m
an
u
ally
co
llected
f
e
atu
r
es
an
d
s
ca
lab
ilit
y
is
s
u
es
lim
it
th
eir
p
o
ten
tial
c
o
m
p
ar
e
d
to
d
ee
p
lear
n
in
g
.
Dee
p
lear
n
in
g
s
y
s
tem
s
n
o
w
d
iag
n
o
s
e
b
etter
.
C
NN
h
as
9
1
.
8
%
ac
c
u
r
ac
y
,
9
0
.
2
%
s
en
s
itiv
ity
,
an
d
9
3
.
4
%
s
p
ec
if
icity
.
R
ec
o
g
n
izin
g
h
ier
ar
ch
ical
C
T
s
ca
n
f
ea
tu
r
es
li
k
e
ed
g
es,
tex
tu
r
es,
an
d
co
m
p
licated
p
atter
n
s
.
T
h
e
0
.
9
4
AUC
-
R
OC
s
h
o
ws
s
ig
n
if
ican
t
d
is
cr
im
in
atio
n
.
T
h
e
b
est
alg
o
r
ith
m
is
R
esNet
(
9
4
.
2
%
ac
cu
r
ac
y
)
.
I
t’
s
AUC
-
R
O
C
o
f
0
.
9
6
,
s
en
s
itiv
ity
(
9
2
.
7
%),
an
d
s
p
ec
if
icity
(
9
5
.
8
%)
d
is
cr
im
in
ate
m
alig
n
an
t
f
r
o
m
n
o
n
-
ca
n
ce
r
o
u
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
5
,
No
.
4
,
Au
g
u
s
t 2
0
2
6
:
3
1
2
0
-
3
1
3
0
3128
p
atien
ts
.
R
esNet
o
v
er
co
m
es
v
an
is
h
in
g
g
r
a
d
ien
ts
an
d
o
b
tain
s
d
elica
te
d
ee
p
n
etwo
r
k
in
f
o
r
m
atio
n
f
o
r
lu
n
g
n
o
d
u
le
i
d
en
tific
atio
n
u
s
in
g
r
es
id
u
al
co
n
n
ec
tio
n
s
.
At
9
2
.
6
%
ac
cu
r
ac
y
,
VGG1
6
is
b
eh
in
d
R
esNet
y
et
p
er
f
o
r
m
s
well.
I
t
h
an
d
les
co
m
p
licate
d
im
ag
in
g
d
ata
with
9
1
.
0
a
n
d
9
4
.
2
%
s
en
s
itiv
ity
an
d
s
p
ec
if
icity
.
AUC
-
R
O
C
0
.
9
5
m
atc
h
es
R
esNet
class
d
if
f
er
en
tiatin
g
r
eliab
ilit
y
.
VGG1
6
lack
s
R
esNet's
r
esid
u
al
co
n
n
ec
tio
n
s
,
r
es
tr
ictin
g
d
ee
p
n
etwo
r
k
p
er
f
o
r
m
an
ce
.
Dee
p
lear
n
in
g
im
p
r
o
v
es
L
C
D
with
L
R
,
SV
M,
C
NN,
R
e
s
Net,
an
d
VGG1
6
.
T
h
e
s
u
p
er
i
o
r
p
er
f
o
r
m
a
n
ce
o
f
R
esNet
an
d
C
NN
em
p
h
asizes
th
e
n
ee
d
f
o
r
a
u
to
m
ated
f
ea
tu
r
e
e
x
tr
ac
tio
n
a
n
d
d
ee
p
ar
ch
itectu
r
es
in
th
e
co
m
p
lex
ity
o
f
m
ed
ical
im
ag
in
g
d
ata
.
T
h
ese
f
in
d
in
g
s
s
u
g
g
est
th
at
s
o
p
h
is
ticated
alg
o
r
ith
m
s
im
p
r
o
v
e
L
C
D
.
T
h
is
f
in
d
in
g
is
co
n
s
is
ten
t
with
ea
r
lier
s
tu
d
ies
d
em
o
n
s
tr
atin
g
th
at
d
ee
p
lea
r
n
in
g
m
o
d
els
p
r
o
v
id
e
s
ca
lab
le
a
n
d
r
el
iab
le
s
o
lu
tio
n
s
f
o
r
lar
g
e
-
s
ca
le
clin
ical
im
ag
e
an
aly
s
is
an
d
f
ac
ilit
ate
th
eir
ad
o
p
tio
n
in
m
o
d
er
n
h
ea
lth
ca
r
e
s
y
s
tem
s
[
2
9
]
.
T
h
ese
f
in
d
in
g
s
s
u
g
g
est th
at
s
o
p
h
is
ticated
alg
o
r
ith
m
s
im
p
r
o
v
e
L
C
D
.
5.
CO
NCLU
SI
O
N
T
h
is
s
tu
d
y
d
e
m
o
n
s
tr
ates
th
a
t
d
ee
p
lear
n
in
g
m
o
d
els
m
ar
k
ed
ly
im
p
r
o
v
e
lu
n
g
ca
n
ce
r
d
iag
n
o
s
tic
p
er
f
o
r
m
an
ce
co
m
p
ar
ed
to
tr
a
d
itio
n
al
m
ac
h
in
e
lear
n
in
g
ap
p
r
o
ac
h
es.
Am
o
n
g
th
e
ev
alu
at
ed
m
o
d
els,
R
esNet
ac
h
iev
ed
th
e
h
ig
h
est d
iag
n
o
s
tic
ac
cu
r
ac
y
(
9
4
%),
s
en
s
itiv
ity
(
9
5
%),
s
p
ec
if
icity
(
9
3
%),
an
d
AUC
-
R
O
C
clo
s
e
to
1
.
0
,
f
o
llo
wed
clo
s
ely
b
y
C
NN,
wh
ich
attain
ed
an
ac
cu
r
ac
y
o
f
9
2
%,
s
en
s
itiv
ity
o
f
9
3
%,
an
d
s
p
ec
if
icity
o
f
9
1
%.
T
h
ese
r
esu
lts
h
ig
h
lig
h
t
th
e
s
u
p
er
i
o
r
a
b
ilit
y
o
f
d
ee
p
n
eu
r
al
n
etwo
r
k
s
to
au
to
m
atic
ally
lear
n
c
o
m
p
lex
im
ag
in
g
f
ea
tu
r
es,
with
R
es
Net's
d
ee
p
er
ar
ch
itectu
r
e
an
d
s
k
ip
co
n
n
ec
tio
n
s
en
ab
lin
g
m
o
r
e
ef
f
ec
tiv
e
in
ter
p
r
etatio
n
o
f
s
u
b
tle,
ea
r
ly
-
s
tag
e
m
alig
n
an
t
n
o
d
u
les.
T
h
e
in
teg
r
atio
n
o
f
d
ee
p
lear
n
in
g
m
o
d
els
in
to
clin
ical
wo
r
k
f
lo
ws
h
as
th
e
p
o
ten
tial
to
tr
an
s
f
o
r
m
lu
n
g
ca
n
ce
r
d
ete
ctio
n
an
d
tr
ea
tm
en
t.
T
h
ese
m
o
d
els
ca
n
au
to
m
ate
d
iag
n
o
s
tic
task
s
,
r
ed
u
ce
er
r
o
r
r
ates,
an
d
ex
p
e
d
ite
d
iag
n
o
s
is
tim
es,
wh
ich
ar
e
cr
itical
f
ac
to
r
s
in
im
p
r
o
v
in
g
p
atien
t
p
r
o
g
n
o
s
is
.
Fu
r
th
er
m
o
r
e,
th
eir
s
ca
lab
ilit
y
allo
ws
f
o
r
ef
f
icien
t
p
r
o
ce
s
s
in
g
o
f
la
r
g
e
-
s
ca
le
m
ed
ical
im
ag
in
g
d
atasets
,
alig
n
in
g
we
ll
with
th
e
n
ee
d
s
o
f
c
o
n
tem
p
o
r
ar
y
h
ea
lth
ca
r
e
s
y
s
tem
s
.
T
h
e
ad
o
p
tio
n
o
f
d
ee
p
lear
n
in
g
tec
h
n
iq
u
es,
p
ar
ticu
l
ar
ly
R
esNet
-
b
ased
m
o
d
els,
i
s
an
ticip
ated
to
en
h
a
n
ce
d
i
ag
n
o
s
tic
p
r
ec
is
io
n
,
f
ac
ilit
ate
ea
r
lier
in
ter
v
en
tio
n
s
,
an
d
co
n
tr
ib
u
te
to
m
o
r
e
p
er
s
o
n
alize
d
an
d
ef
f
ec
tiv
e
p
atien
t c
ar
e
p
ath
way
s
.
ACK
NO
WL
E
DG
M
E
N
T
S
T
h
e
au
t
h
o
r
s
w
o
u
ld
lik
e
t
o
s
i
n
ce
r
ely
t
h
an
k
Vel
T
ec
h
R
an
g
ar
ajan
Dr
.
Sag
u
n
th
ala
R
&
D
I
n
s
titu
te
o
f
Scien
ce
an
d
T
ec
h
n
o
lo
g
y
,
C
h
e
n
n
ai,
I
n
d
ia,
f
o
r
p
r
o
v
id
i
n
g
s
tu
d
y
f
ac
ilit
ies.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
Au
th
o
r
s
s
tate
n
o
f
u
n
d
in
g
in
v
o
lv
ed
.
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
T
h
is
jo
u
r
n
al
u
s
es
th
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT
)
to
r
ec
o
g
n
ize
in
d
iv
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
th
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
llab
o
r
atio
n
.
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
Ph
an
ee
n
d
r
a
Var
m
a
C
h
in
talap
ati
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
Pra
s
an
th
Ar
u
ch
am
y
✓
✓
✓
✓
✓
✓
✓
✓
Alu
r
Pra
n
ee
th
a
✓
✓
✓
✓
✓
✓
C
:
C
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
M
:
M
e
t
h
o
d
o
l
o
g
y
So
:
So
f
t
w
a
r
e
Va
:
Va
l
i
d
a
t
i
o
n
Fo
:
Fo
r
mal
a
n
a
l
y
s
i
s
I
:
I
n
v
e
s
t
i
g
a
t
i
o
n
R
:
R
e
so
u
r
c
e
s
D
:
D
a
t
a
C
u
r
a
t
i
o
n
O
:
W
r
i
t
i
n
g
-
O
r
i
g
i
n
a
l
D
r
a
f
t
E
:
W
r
i
t
i
n
g
-
R
e
v
i
e
w
&
E
d
i
t
i
n
g
Vi
:
Vi
su
a
l
i
z
a
t
i
o
n
Su
:
Su
p
e
r
v
i
s
i
o
n
P
:
P
r
o
j
e
c
t
a
d
mi
n
i
st
r
a
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
si
t
i
o
n
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
T
h
e
au
th
o
r
s
s
tate
n
o
co
n
f
lict o
f
in
ter
est.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
Dee
p
lea
r
n
in
g
fo
r
lu
n
g
ca
n
ce
r
d
ia
g
n
o
s
is
:
a
co
mp
a
r
a
tive
a
r
ti
ficia
l
…
(
P
h
a
n
ee
n
d
r
a
V
a
r
ma
C
h
in
ta
la
p
a
ti
)
3129
I
NF
O
RM
E
D
CO
NS
E
N
T
No
t
ap
p
licab
le.
T
h
is
s
tu
d
y
d
id
n
o
t
in
v
o
l
v
e
h
u
m
an
p
ar
tici
p
an
ts
,
h
u
m
an
d
ata,
o
r
a
n
y
p
er
s
o
n
ally
id
en
tifia
b
le
in
f
o
r
m
atio
n
.
All
d
ata
u
s
ed
wer
e
eith
er
p
u
b
licl
y
av
ailab
le,
f
u
lly
an
o
n
y
m
ize
d
,
o
r
d
er
i
v
ed
f
r
o
m
n
o
n
‑
h
u
m
an
s
o
u
r
ce
s
,
an
d
th
er
e
f
o
r
e
n
o
in
f
o
r
m
ed
co
n
s
en
t w
as
r
eq
u
ir
ed
f
r
o
m
in
d
iv
i
d
u
als.
E
T
H
I
CAL AP
P
RO
V
AL
No
t
ap
p
licab
le.
T
h
is
r
esear
ch
d
id
n
o
t
in
v
o
lv
e
h
u
m
an
s
u
b
jects,
h
u
m
an
b
io
lo
g
ical
m
ater
ials
,
o
r
ex
p
er
im
en
tal
p
r
o
ce
d
u
r
es
o
n
an
im
als.
T
h
e
wo
r
k
was
co
n
d
u
cted
s
o
lely
o
n
c
o
m
p
u
tatio
n
al
m
o
d
els,
p
u
b
licly
av
ailab
le
d
atasets
,
o
r
n
o
n
‑
s
en
s
itiv
e
d
ata
th
at
d
id
n
o
t
r
eq
u
ir
e
i
n
ter
v
en
tio
n
with
liv
in
g
o
r
g
an
is
m
s
.
T
h
er
ef
o
r
e,
eth
ical
ap
p
r
o
v
al
f
r
o
m
an
in
s
titu
tio
n
al
r
ev
iew
b
o
ar
d
o
r
a
n
im
al
eth
ics
c
o
m
m
itte
e
was
n
o
t
n
ec
ess
ar
y
f
o
r
th
is
s
tu
d
y
.
DATA AV
AI
L
AB
I
L
I
T
Y
Data
av
ailab
ilit
y
d
o
es n
o
t a
p
p
l
y
to
th
is
p
ap
e
r
as n
o
n
ew
d
ata
wer
e
cr
ea
ted
o
r
an
aly
ze
d
in
th
is
s
tu
d
y
.
RE
F
E
R
E
NC
E
S
[
1
]
R
.
Ja
v
e
d
,
T.
A
b
b
a
s
,
A
.
H
.
K
h
a
n
,
A
.
D
a
u
d
,
A
.
B
u
k
h
a
r
i
,
a
n
d
R
.
A
l
h
a
r
b
e
y
,
“
D
e
e
p
l
e
a
r
n
i
n
g
f
o
r
l
u
n
g
s
c
a
n
c
e
r
d
e
t
e
c
t
i
o
n
:
a
r
e
v
i
e
w
,
”
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
Re
v
i
e
w
,
v
o
l
.
5
7
,
n
o
.
8
,
Ju
l
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
0
4
6
2
-
0
2
4
-
1
0
8
0
7
-
1.
[
2
]
S
.
H
.
H
o
sse
i
n
i
,
R
.
M
o
n
sef
i
,
a
n
d
S
.
S
h
a
d
r
o
o
,
“
D
e
e
p
l
e
a
r
n
i
n
g
a
p
p
l
i
c
a
t
i
o
n
s
f
o
r
l
u
n
g
c
a
n
c
e
r
d
i
a
g
n
o
si
s
:
a
s
y
st
e
mat
i
c
r
e
v
i
e
w
,
”
Mu
l
t
i
m
e
d
i
a
T
o
o
l
s
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
8
3
,
n
o
.
5
,
p
p
.
1
4
3
0
5
–
1
4
3
3
5
,
J
u
l
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
1
0
4
2
-
0
2
3
-
1
6
0
4
6
-
w.
[
3
]
G
.
S
i
n
g
h
,
A
.
K
a
m
a
l
j
a
,
R
.
P
a
t
i
l
,
A
.
K
a
r
w
a
,
A
.
Tr
i
p
a
t
h
i
,
a
n
d
P
.
C
h
a
v
a
n
,
“
A
c
o
mp
r
e
h
e
n
s
i
v
e
a
ssessm
e
n
t
o
f
a
r
t
i
f
i
c
i
a
l
i
n
t
e
l
l
i
g
e
n
c
e
a
p
p
l
i
c
a
t
i
o
n
s f
o
r
c
a
n
c
e
r
d
i
a
g
n
o
si
s
,
”
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
R
e
v
i
e
w
,
v
o
l
.
5
7
,
n
o
.
7
,
Ju
n
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
0
4
6
2
-
0
2
4
-
1
0
7
8
3
-
6.
[
4
]
M
.
S
.
A
l
sh
u
h
r
i
e
t
a
l
.
,
“
A
r
t
i
f
i
c
i
a
l
i
n
t
e
l
l
i
g
e
n
c
e
i
n
c
a
n
c
e
r
d
i
a
g
n
o
si
s
:
o
p
p
o
r
t
u
n
i
t
i
e
s
a
n
d
c
h
a
l
l
e
n
g
e
s
,
”
Pa
t
h
o
l
o
g
y
-
Re
s
e
a
r
c
h
a
n
d
Pra
c
t
i
c
e
,
v
o
l
.
2
5
3
,
Ja
n
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
r
p
.
2
0
2
3
.
1
5
4
9
9
6
.
[
5
]
H
.
T.
G
a
y
a
p
a
n
d
M
.
A
.
A
k
h
l
o
u
f
i
,
“
D
e
e
p
mac
h
i
n
e
l
e
a
r
n
i
n
g
f
o
r
me
d
i
c
a
l
d
i
a
g
n
o
s
i
s,
a
p
p
l
i
c
a
t
i
o
n
t
o
l
u
n
g
c
a
n
c
e
r
d
e
t
e
c
t
i
o
n
:
a
r
e
v
i
e
w
,
”
Bi
o
Me
d
I
n
f
o
rm
a
t
i
c
s
,
v
o
l
.
4
,
n
o
.
1
,
p
p
.
2
3
6
–
2
8
4
,
J
a
n
.
2
0
2
4
,
d
o
i
:
1
0
.
3
3
9
0
/
b
i
o
med
i
n
f
o
r
ma
t
i
c
s4
0
1
0
0
1
5
.
[
6
]
H
.
N
a
si
e
f
e
t
a
l
.
,
“
I
mp
r
o
v
i
n
g
t
r
e
a
t
me
n
t
r
e
sp
o
n
s
e
p
r
e
d
i
c
t
i
o
n
f
o
r
c
h
e
mo
r
a
d
i
a
t
i
o
n
t
h
e
r
a
p
y
o
f
p
a
n
c
r
e
a
t
i
c
c
a
n
c
e
r
u
s
i
n
g
a
c
o
mb
i
n
a
t
i
o
n
o
f
d
e
l
t
a
-
r
a
d
i
o
m
i
c
s
a
n
d
t
h
e
c
l
i
n
i
c
a
l
b
i
o
ma
r
k
e
r
C
A
1
9
-
9
,
”
Fro
n
t
i
e
rs
i
n
O
n
c
o
l
o
g
y
,
v
o
l
.
9
,
J
a
n
.
2
0
2
0
,
d
o
i
:
1
0
.
3
3
8
9
/
f
o
n
c
.
2
0
1
9
.
0
1
4
6
4
.
[
7
]
V
.
S
.
P
a
r
e
k
h
a
n
d
M
.
A
.
J
a
c
o
b
s,
“
M
P
R
A
D
:
a
m
u
l
t
i
p
a
r
a
me
t
r
i
c
r
a
d
i
o
mi
c
s fr
a
mew
o
r
k
,
”
2
0
1
8
,
a
rXi
v
:
1
8
0
9
.
0
9
9
7
3
.
[
8
]
L.
P
e
n
g
e
t
a
l
.
,
“
D
i
s
t
i
n
g
u
i
s
h
i
n
g
t
r
u
e
p
r
o
g
r
e
ss
i
o
n
f
r
o
m
r
a
d
i
o
n
e
c
r
o
si
s
a
f
t
e
r
st
e
r
e
o
t
a
c
t
i
c
r
a
d
i
a
t
i
o
n
t
h
e
r
a
p
y
f
o
r
b
r
a
i
n
met
a
s
t
a
s
e
s
w
i
t
h
mac
h
i
n
e
l
e
a
r
n
i
n
g
a
n
d
r
a
d
i
o
mi
c
s,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
Ra
d
i
a
t
i
o
n
O
n
c
o
l
o
g
y
*
B
i
o
l
o
g
y
*
P
h
y
si
c
s
,
v
o
l
.
1
0
2
,
n
o
.
4
,
p
p
.
1
2
3
6
–
1
2
4
3
,
N
o
v
.
2
0
1
8
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
i
j
r
o
b
p
.
2
0
1
8
.
0
5
.
0
4
1
.
[
9
]
H
.
J.
W
.
L.
A
e
r
t
s
e
t
a
l
.
,
“
D
e
c
o
d
i
n
g
t
u
mo
u
r
p
h
e
n
o
t
y
p
e
b
y
n
o
n
i
n
v
a
s
i
v
e
i
m
a
g
i
n
g
u
si
n
g
a
q
u
a
n
t
i
t
a
t
i
v
e
r
a
d
i
o
mi
c
s
a
p
p
r
o
a
c
h
,
”
N
a
t
u
re
C
o
m
m
u
n
i
c
a
t
i
o
n
s
,
v
o
l
.
5
,
n
o
.
1
,
J
u
n
.
2
0
1
4
,
d
o
i
:
1
0
.
1
0
3
8
/
n
c
o
mm
s
5
0
0
6
.
[
1
0
]
T.
P
.
C
o
r
o
l
l
e
r
e
t
a
l
.
,
“
C
T
-
b
a
s
e
d
r
a
d
i
o
mi
c
si
g
n
a
t
u
r
e
p
r
e
d
i
c
t
s
d
i
st
a
n
t
m
e
t
a
s
t
a
s
i
s
i
n
l
u
n
g
a
d
e
n
o
c
a
r
c
i
n
o
ma,
”
Ra
d
i
o
t
h
e
r
a
p
y
a
n
d
O
n
c
o
l
o
g
y
,
v
o
l
.
1
1
4
,
n
o
.
3
,
p
p
.
3
4
5
–
3
5
0
,
M
a
r
.
2
0
1
5
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
r
a
d
o
n
c
.
2
0
1
5
.
0
2
.
0
1
5
.
[
1
1
]
R
.
S
u
n
e
t
a
l
.
,
“
I
n
r
e
g
a
r
d
t
o
M
a
t
t
o
n
e
n
e
t
a
l
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
Ra
d
i
a
t
i
o
n
O
n
c
o
l
o
g
y
,
Bi
o
l
o
g
y
,
P
h
y
s
i
c
s
,
v
o
l
.
9
5
,
n
o
.
5
,
p
p
.
1
5
4
4
–
1
5
4
5
,
A
u
g
.
2
0
1
6
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
i
j
r
o
b
p
.
2
0
1
6
.
0
3
.
0
3
8
.
[
1
2
]
J.
S
h
a
o
,
J.
F
e
n
g
,
J.
Li
,
S
.
Li
a
n
g
,
W
.
Li
,
a
n
d
C
.
W
a
n
g
,
“
N
o
v
e
l
t
o
o
l
s
f
o
r
e
a
r
l
y
d
i
a
g
n
o
s
i
s
a
n
d
p
r
e
c
i
si
o
n
t
r
e
a
t
me
n
t
b
a
s
e
d
o
n
a
r
t
i
f
i
c
i
a
l
i
n
t
e
l
l
i
g
e
n
c
e
,
”
C
h
i
n
e
se
M
e
d
i
c
a
l
J
o
u
rn
a
l
Pu
l
m
o
n
a
ry
a
n
d
C
ri
t
i
c
a
l
C
a
r
e
Me
d
i
c
i
n
e
,
v
o
l
.
1
,
n
o
.
3
,
p
p
.
1
4
8
–
1
6
0
,
S
e
p
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
c
c
m.
2
0
2
3
.
0
5
.
0
0
1
.
[
1
3
]
D
.
B
o
u
c
h
a
f
f
r
a
,
F
.
Y
k
h
l
e
f
,
a
n
d
S
.
B
e
n
b
e
l
k
a
c
e
m,
“
A
d
v
a
n
c
e
me
n
t
i
n
l
u
n
g
c
a
n
c
e
r
d
i
a
g
n
o
s
i
s:
a
c
o
mp
r
e
h
e
n
s
i
v
e
r
e
v
i
e
w
o
f
d
e
e
p
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
e
s,
”
i
n
L
u
n
g
C
a
n
c
e
r
D
i
a
g
n
o
si
s
a
n
d
T
r
e
a
t
m
e
n
t
:
An
I
n
t
e
rd
i
sc
i
p
l
i
n
a
r
y
A
p
p
ro
a
c
h
,
C
h
a
m,
S
w
i
t
z
e
r
l
a
n
d
:
S
p
r
i
n
g
e
r
,
2
0
2
4
,
p
p
.
1
5
–
24
,
d
o
i
:
1
0
.
1
0
0
7
/
1
6
8
3
3
_
2
0
2
4
_
3
0
2
.
[
1
4
]
M
.
K
a
l
k
a
n
,
M
.
S
.
G
u
z
e
l
,
F
.
E
k
i
n
c
i
,
E
.
A
k
c
a
p
i
n
a
r
S
e
z
e
r
,
a
n
d
T.
A
su
r
o
g
l
u
,
“
C
o
m
p
a
r
a
t
i
v
e
a
n
a
l
y
s
i
s
o
f
d
e
e
p
l
e
a
r
n
i
n
g
met
h
o
d
s
o
n
C
T
i
ma
g
e
s f
o
r
l
u
n
g
c
a
n
c
e
r
s
p
e
c
i
f
i
c
a
t
i
o
n
,
”
C
a
n
c
e
rs
,
v
o
l
.
1
6
,
n
o
.
1
9
,
S
e
p
.
2
0
2
4
,
d
o
i
:
1
0
.
3
3
9
0
/
c
a
n
c
e
r
s1
6
1
9
3
3
2
1
.
[
1
5
]
M
.
M
a
m
u
n
,
M
.
I
.
M
a
h
m
u
d
,
M
.
M
e
h
e
r
i
n
,
a
n
d
A
.
A
b
d
e
l
g
a
w
a
d
,
“
LC
D
c
t
C
N
N
:
l
u
n
g
c
a
n
c
e
r
d
i
a
g
n
o
s
i
s o
f
C
T
sc
a
n
i
mag
e
s u
si
n
g
C
N
N
b
a
s
e
d
m
o
d
e
l
,
”
i
n
2
0
2
3
1
0
t
h
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
S
i
g
n
a
l
Pr
o
c
e
ssi
n
g
a
n
d
I
n
t
e
g
ra
t
e
d
N
e
t
w
o
r
k
s
(
S
PI
N
)
,
M
a
r
.
2
0
2
3
,
p
p
.
2
0
5
–
212
,
d
o
i
:
1
0
.
1
1
0
9
/
S
P
I
N
5
7
0
0
1
.
2
0
2
3
.
1
0
1
1
6
0
7
5
.
[
1
6
]
B
.
L
i
u
e
t
a
l
.
,
“
Ev
o
l
v
i
n
g
t
h
e
p
u
l
m
o
n
a
r
y
n
o
d
u
l
e
s
d
i
a
g
n
o
s
i
s
f
r
o
m
c
l
a
ss
i
c
a
l
a
p
p
r
o
a
c
h
e
s t
o
d
e
e
p
l
e
a
r
n
i
n
g
-
a
i
d
e
d
d
e
c
i
s
i
o
n
s
u
p
p
o
r
t
:
t
h
r
e
e
d
e
c
a
d
e
s’
d
e
v
e
l
o
p
me
n
t
c
o
u
r
s
e
a
n
d
f
u
t
u
r
e
p
r
o
s
p
e
c
t
,
”
J
o
u
rn
a
l
o
f
C
a
n
c
e
r
Re
se
a
rc
h
a
n
d
C
l
i
n
i
c
a
l
O
n
c
o
l
o
g
y
,
v
o
l
.
1
4
6
,
n
o
.
1
,
p
p
.
1
5
3
–
1
8
5
,
J
a
n
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
0
7
/
s0
0
4
3
2
-
0
1
9
-
0
3
0
9
8
-
5.
[
1
7
]
L.
W
a
n
g
,
“
D
e
e
p
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s
t
o
d
i
a
g
n
o
se
l
u
n
g
c
a
n
c
e
r
,
”
C
a
n
c
e
rs
,
v
o
l
.
1
4
,
n
o
.
2
2
,
N
o
v
.
2
0
2
2
,
d
o
i
:
1
0
.
3
3
9
0
/
c
a
n
c
e
r
s1
4
2
2
5
5
6
9
.
[
1
8
]
J.
H
.
S
o
h
n
a
n
d
B
.
K
.
K
.
F
i
e
l
d
s,
“
R
a
d
i
o
mi
c
s
a
n
d
d
e
e
p
l
e
a
r
n
i
n
g
t
o
p
r
e
d
i
c
t
p
u
l
m
o
n
a
r
y
n
o
d
u
l
e
m
e
t
a
st
a
si
s
a
t
C
T,
”
R
a
d
i
o
l
o
g
y
,
v
o
l
.
3
1
1
,
n
o
.
1
,
p
p
.
1
8
1
6
–
1
8
2
6
,
A
p
r
.
2
0
2
4
,
d
o
i
:
1
0
.
1
1
4
8
/
r
a
d
i
o
l
.
2
3
3
3
5
6
.
[
1
9
]
D
.
A
.
H
a
sh
i
m
o
t
o
,
G
.
R
o
sma
n
,
D
.
R
u
s,
a
n
d
O
.
R
.
M
e
i
r
e
l
e
s,
“
A
r
t
i
f
i
c
i
a
l
i
n
t
e
l
l
i
g
e
n
c
e
i
n
s
u
r
g
e
r
y
:
p
r
o
m
i
ses
a
n
d
p
e
r
i
l
s
,
”
An
n
a
l
s
o
f
S
u
r
g
e
ry
,
v
o
l
.
2
6
8
,
n
o
.
1
,
p
p
.
7
0
–
7
6
,
J
u
l
.
2
0
1
8
,
d
o
i
:
1
0
.
1
0
9
7
/
S
LA
.
0
0
0
0
0
0
0
0
0
0
0
0
2
6
9
3
.
[
2
0
]
G
.
Le
e
,
H
.
P
a
r
k
,
S
.
H
.
B
a
k
,
a
n
d
H
.
Y
.
Le
e
,
“
R
a
d
i
o
mi
c
s
i
n
l
u
n
g
c
a
n
c
e
r
f
r
o
m
b
a
s
i
c
t
o
a
d
v
a
n
c
e
d
:
c
u
r
r
e
n
t
s
t
a
t
u
s
a
n
d
f
u
t
u
r
e
d
i
r
e
c
t
i
o
n
s,
”
K
o
r
e
a
n
J
o
u
r
n
a
l
o
f
R
a
d
i
o
l
o
g
y
,
v
o
l
.
2
1
,
n
o
.
2
,
2
0
2
0
,
d
o
i
:
1
0
.
3
3
4
8
/
k
j
r
.
2
0
1
9
.
0
6
3
0
.
[
2
1
]
X
.
Z
h
a
n
g
,
B
.
L
i
u
,
K
.
L
i
u
,
a
n
d
L.
W
a
n
g
,
“
T
h
e
d
i
a
g
n
o
si
s
p
e
r
f
o
r
m
a
n
c
e
o
f
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
i
n
t
h
e
d
e
t
e
c
t
i
o
n
o
f
p
u
l
m
o
n
a
r
y
n
o
d
u
l
e
s:
a
s
y
st
e
ma
t
i
c
r
e
v
i
e
w
a
n
d
met
a
-
a
n
a
l
y
s
i
s,
”
Ac
t
a
R
a
d
i
o
l
o
g
i
c
a
,
v
o
l
.
6
4
,
n
o
.
1
2
,
p
p
.
2
9
8
7
–
2
9
9
8
,
D
e
c
.
2
0
2
3
,
d
o
i
:
1
0
.
1
1
7
7
/
0
2
8
4
1
8
5
1
2
3
1
2
0
1
5
1
4
.
[
2
2
]
K
.
S
u
z
u
k
i
,
“
S
e
g
me
n
t
a
t
i
o
n
o
f
l
e
si
o
n
s
w
i
t
h
i
m
p
r
o
v
e
d
s
p
e
c
i
f
i
c
i
t
y
i
n
c
o
m
p
u
t
e
r
-
a
i
d
e
d
d
i
a
g
n
o
si
s
u
s
i
n
g
a
mass
i
v
e
-
t
r
a
i
n
i
n
g
a
r
t
i
f
i
c
i
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
(
M
TA
N
N
)
,
”
i
n
2
0
0
8
S
e
v
e
n
t
h
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
Ma
c
h
i
n
e
L
e
a
rn
i
n
g
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s
,
2
0
0
8
,
p
p
.
5
2
3
–
527
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
M
LA
.
2
0
0
8
.
1
1
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.