I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
pu
t
er
E
ng
ineering
(
I
J
E
CE
)
Vo
l.
15
,
No
.
3
,
J
u
n
e
20
25
,
p
p
.
3084
~
3
0
9
4
I
SS
N:
2088
-
8
7
0
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijece.
v
15
i
3
.
pp
3
0
8
4
-
3
0
9
4
3084
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
A
hy
brid
co
nv
o
lutiona
l neural ne
t
wo
rk
-
recu
rren
t
n
eura
l
network
a
pp
ro
a
ch f
o
r br
ea
st cance
r det
e
ction
throu
g
h
M
a
sk
R
-
CN
N an
d ARI
-
T
FMO
A
o
ptimiza
tion
K
esh
et
t
i Sre
ek
a
la
1
,
Srila
t
ha
Ya
la
m
a
t
i
2
,
Annem
needi La
k
s
hm
a
na
ra
o
3
,
G
ub
ba
la
K
um
a
ri
4
,
T
a
na
pa
neni
M
un
i K
um
a
ri
5
,
Venk
a
t
a
Su
bb
a
ia
h De
s
a
na
m
uk
ula
6
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
,
M
a
h
a
t
ma
G
a
n
d
h
i
I
n
st
i
t
u
t
e
o
f
T
e
c
h
n
o
l
o
g
y
(
A
)
,
H
y
d
e
r
a
b
a
d
,
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
,
G
u
n
t
u
r
,
I
n
d
i
a
3
D
e
p
a
r
t
me
n
t
o
f
I
n
f
o
r
mat
i
o
n
Te
c
h
n
o
l
o
g
y
,
A
d
i
t
y
a
U
n
i
v
e
r
si
t
y
,
S
u
r
a
m
p
a
l
e
m,
I
n
d
i
a
4
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
,
C
M
R
E
n
g
i
n
e
e
r
i
n
g
C
o
l
l
e
g
e
,
I
n
d
i
a
5
D
e
p
a
r
t
me
n
t
o
f
M
a
st
e
r
o
f
C
o
m
p
u
t
e
r
A
p
p
l
i
c
a
t
i
o
n
s
,
S
r
i
V
e
n
k
a
t
e
sw
a
r
a
C
o
l
l
e
g
e
o
f
E
n
g
i
n
e
e
r
i
n
g
,
T
i
r
u
p
a
t
i
,
I
n
d
i
a
6
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
,
L
a
k
i
r
e
d
d
y
B
a
l
i
R
e
d
d
y
C
o
l
l
e
g
e
o
f
E
n
g
i
n
e
e
r
i
n
g
(
A
u
t
o
n
o
m
o
u
s
)
,
M
y
l
a
v
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
Au
g
1
8
,
2
0
2
4
R
ev
is
ed
J
an
2
3
,
2
0
2
5
Acc
ep
ted
Ma
r
3
,
2
0
2
5
Th
is
p
a
p
e
r
p
re
se
n
ts
a
n
o
v
e
l
h
y
b
r
id
d
e
e
p
lea
rn
i
n
g
-
b
a
se
d
a
p
p
r
o
a
c
h
fo
r
b
re
a
st
c
a
n
c
e
r
d
e
tec
ti
o
n
,
a
d
d
re
ss
in
g
c
ri
ti
c
a
l
c
h
a
ll
e
n
g
e
s
s
u
c
h
a
s
o
v
e
rfi
tt
in
g
a
n
d
p
e
rfo
rm
a
n
c
e
d
e
g
ra
d
a
ti
o
n
in
v
a
ry
in
g
d
a
ta
c
o
n
d
it
i
o
n
s.
U
n
li
k
e
trad
it
io
n
a
l
m
e
th
o
d
s
th
a
t
stru
g
g
le
wit
h
d
e
tec
ti
o
n
a
c
c
u
ra
c
y
,
th
is
wo
r
k
i
n
teg
ra
te
s
a
u
n
i
q
u
e
c
o
m
b
in
a
ti
o
n
o
f
a
d
v
a
n
c
e
d
se
g
m
e
n
tatio
n
a
n
d
c
las
sifica
ti
o
n
tec
h
n
i
q
u
e
s.
T
h
e
se
g
m
e
n
tatio
n
p
h
a
se
lev
e
ra
g
e
s
M
a
sk
re
g
i
o
n
-
b
a
se
d
c
o
n
v
o
l
u
ti
o
n
a
l
n
e
u
ra
l
n
e
two
rk
(R
-
CNN
)
,
e
n
h
a
n
c
e
d
b
y
t
h
e
a
d
a
p
ti
v
e
ra
n
d
o
m
i
n
c
re
m
e
n
t
-
b
a
se
d
to
m
ti
t
flo
c
k
m
e
tah
e
u
risti
c
o
p
ti
m
iza
ti
o
n
a
lg
o
rit
h
m
(ARI
-
TF
M
OA
),
a
n
o
v
e
l
a
lg
o
rit
h
m
in
sp
ired
b
y
n
a
tu
ra
l
f
lo
c
k
i
n
g
b
e
h
a
v
io
r.
ARI
-
TF
M
OA
fi
n
e
-
tu
n
e
s
M
a
sk
R
-
CNN
p
a
ra
m
e
ters
,
a
c
h
iev
in
g
imp
ro
v
e
d
fe
a
tu
re
e
x
trac
ti
o
n
a
n
d
se
g
m
e
n
tatio
n
p
re
c
isio
n
w
h
il
e
e
n
su
ri
n
g
a
d
a
p
tab
i
li
ty
t
o
d
iv
e
rse
d
a
tas
e
ts.
F
o
r
c
las
si
fica
ti
o
n
,
a
h
y
b
rid
c
o
n
v
o
l
u
ti
o
n
a
l
n
e
u
ra
l
n
e
two
rk
-
re
c
u
rre
n
t
n
e
u
ra
l
n
e
two
r
k
(
C
NN
-
RNN
)
m
o
d
e
l
is
in
tro
d
u
c
e
d
,
c
o
m
b
i
n
i
n
g
sp
a
ti
a
l
fe
a
t
u
re
e
x
trac
ti
o
n
b
y
C
NN
s
with
tem
p
o
ra
l
p
a
tt
e
rn
re
c
o
g
n
it
i
o
n
b
y
RNN
s,
re
su
lt
in
g
in
a
m
o
re
n
u
a
n
c
e
d
a
n
d
c
o
m
p
re
h
e
n
si
v
e
a
n
a
l
y
sis
o
f
b
re
a
st
c
a
n
c
e
r
ima
g
e
s.
Th
e
p
ro
p
o
se
d
f
ra
m
e
wo
rk
a
c
h
iev
e
d
sig
n
if
ica
n
t
a
d
v
a
n
c
e
m
e
n
ts
o
v
e
r
e
x
isti
n
g
m
e
th
o
d
s,
d
e
m
o
n
stra
ti
n
g
imp
ro
v
e
d
p
e
rfo
rm
a
n
c
e
.
Th
is
h
y
b
r
id
in
te
g
ra
ti
o
n
o
f
ARI
-
TF
M
OA
a
n
d
Hy
b
ri
d
CNN
-
RNN
m
o
d
e
ls
re
p
re
se
n
ts
a
u
n
iq
u
e
c
o
n
tri
b
u
ti
o
n
,
e
n
a
b
li
n
g
r
o
b
u
st,
a
c
c
u
ra
te,
a
n
d
e
fficie
n
t
b
re
a
st ca
n
c
e
r
d
e
tec
ti
o
n
.
K
ey
w
o
r
d
s
:
Ad
ap
tiv
e
r
an
d
o
m
in
c
r
em
en
t
-
b
ased
to
m
tit f
lo
ck
m
etah
eu
r
is
t
ic
o
p
tim
izatio
n
alg
o
r
ith
m
B
r
ea
s
t c
an
ce
r
d
etec
tio
n
C
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
Ma
s
k
r
ec
u
r
r
en
t
n
eu
r
al
n
etwo
r
k
R
ec
u
r
r
en
t n
eu
r
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
:
Kesh
etti Sr
ee
k
ala
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
,
Ma
h
atm
a
Gan
d
h
i I
n
s
titu
te
o
f
T
ec
h
n
o
lo
g
y
(
A)
Hy
d
er
ab
ad
,
I
n
d
ia
E
m
ail:
k
s
r
ik
ala_
cse@
m
g
it.a
c.
in
1.
I
NT
RO
D
UCT
I
O
N
B
r
ea
s
t
ca
n
ce
r
is
o
n
e
o
f
th
e
m
o
s
t
life
-
th
r
ea
ten
in
g
d
is
ea
s
es
am
o
n
g
wo
m
e
n
g
lo
b
ally
.
Acc
o
r
d
in
g
to
r
ec
en
t
s
tatis
t
ics,
ea
r
ly
d
iag
n
o
s
is
is
cr
i
tical
in
im
p
r
o
v
in
g
p
atien
t
o
u
tco
m
es
an
d
s
u
r
v
i
v
al
ch
an
ce
s
.
H
o
wev
er
,
tr
ad
itio
n
al
m
eth
o
d
s
,
s
u
ch
as
m
a
m
m
o
g
r
ap
h
y
,
u
ltra
s
o
u
n
d
,
an
d
b
io
p
s
y
,
f
ac
e
ch
allen
g
es
lik
e
in
ter
p
r
etativ
e
v
ar
iab
ilit
y
,
h
u
m
an
er
r
o
r
,
an
d
in
v
asiv
e
n
ess
,
lim
itin
g
th
eir
ef
f
ec
tiv
en
ess
.
T
h
ese
li
m
itatio
n
s
h
av
e
m
o
tiv
ated
th
e
a
d
o
p
tio
n
o
f
ar
tific
ial
in
tellig
en
ce
(
AI
)
an
d
m
ac
h
in
e
lear
n
in
g
(
ML
)
tech
n
iq
u
es
in
m
ed
ical
im
ag
in
g
,
o
f
f
er
in
g
a
u
to
m
ated
,
ac
cu
r
ate,
a
n
d
non
-
in
v
asiv
e
d
iag
n
o
s
tic
s
o
lu
tio
n
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
A
h
yb
r
id
C
N
N
-
R
N
N
a
p
p
r
o
a
c
h
fo
r
b
r
ea
s
t c
a
n
ce
r
d
etec
tio
n
t
h
r
o
u
g
h
…
(
K
esh
etti S
r
ee
ka
la
)
3085
Dee
p
lear
n
in
g
,
esp
ec
ially
co
n
v
o
lu
tio
n
al
n
e
u
r
al
n
etwo
r
k
s
(
C
NNs
)
,
h
as
tr
an
s
f
o
r
m
e
d
th
e
f
ield
o
f
m
ed
ical
im
ag
e
an
aly
s
is
; its
ab
ilit
y
to
lear
n
h
ier
a
r
ch
ical
f
ea
tu
r
es f
r
o
m
co
m
p
lex
d
atasets
m
ad
e
it a
p
r
o
m
in
en
t p
i
o
n
ee
r
in
g
m
eth
o
d
o
l
o
g
y
.
U
n
f
o
r
t
u
n
ately
,
b
asic
C
NN
m
o
d
els
s
u
f
f
er
f
r
o
m
o
v
er
f
itti
n
g
an
d
p
er
f
o
r
m
an
c
e
d
r
o
p
w
h
en
tr
ain
e
d
on
n
ew
o
r
d
iv
e
r
s
e
d
atasets
,
in
s
p
ite
o
f
th
e
f
ac
t
th
at
it
i
s
v
er
y
p
o
wer
f
u
l.
E
x
is
tin
g
r
esear
ch
is
p
r
im
ar
ily
d
ir
ec
ted
to
war
d
s
m
am
m
o
g
r
a
p
h
y
o
r
v
i
s
ib
le
lig
h
t
im
ag
in
g
,
with
th
er
m
al
im
ag
in
g
b
ein
g
a
n
u
n
d
er
-
ex
p
lo
r
ed
tech
n
iq
u
e
.
T
r
ad
itio
n
al
ap
p
r
o
ac
h
es
to
m
o
d
el
p
r
o
ce
s
s
in
g
s
tr
u
g
g
le
with
is
s
u
es
o
f
lo
w
s
eg
m
e
n
tatio
n
ac
cu
r
ac
y
,
o
v
e
r
f
itti
n
g
a
n
d
d
em
an
d
d
if
f
icu
lty
m
er
g
in
g
s
p
a
tial
an
d
tem
p
o
r
al
f
ea
tu
r
es
f
o
r
e
x
ten
s
iv
e
an
aly
s
is
.
T
h
ese
co
n
s
t
r
ain
ts
d
im
in
is
h
th
e
r
o
b
u
s
tn
ess
o
f
a
u
to
m
atic
s
o
lu
ti
o
n
s
in
h
eter
o
g
en
e
o
u
s
clin
ical
e
n
v
ir
o
n
m
en
ts
,
wh
er
e
t
h
er
e
is
s
ig
n
if
ican
t
v
a
r
iab
ilit
y
in
im
ag
in
g
cir
c
u
m
s
tan
ce
s
an
d
p
atien
t d
ata.
T
o
s
o
lv
e
th
ese
p
r
o
b
lem
s
,
th
is
p
ap
er
p
r
esen
ts
a
h
y
b
r
id
d
ee
p
l
ea
r
n
in
g
(
DL
)
f
r
am
ew
o
r
k
f
o
r
th
e
d
etec
tio
n
o
f
b
r
ea
s
t
ca
n
ce
r
.
I
t
ap
p
lies
ad
a
p
tiv
e
r
an
d
o
m
in
cr
em
e
n
t
-
b
ased
to
m
tit
f
lo
ck
m
etah
eu
r
is
tic
o
p
tim
izatio
n
alg
o
r
ith
m
(
AR
I
-
T
FMOA)
to
f
in
e
-
tu
n
e
p
ar
am
eter
s
o
f
th
e
m
o
d
el,
f
ir
s
tl
y
co
m
b
in
ed
with
Ma
s
k
ed
r
e
g
io
n
al
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
(
Ma
s
k
R
-
C
NN
)
,
wh
ich
is
th
e
m
o
s
t a
d
v
an
ce
d
in
s
eg
m
en
tatio
n
p
e
r
f
o
r
m
an
ce
.
I
t h
elp
s
t
o
in
c
r
ea
s
e
th
e
q
u
ality
o
f
s
eg
m
en
tatio
n
with
o
u
t
lead
in
g
to
o
v
er
f
itti
n
g
.
Ad
d
itio
n
ally
,
a
Hy
b
r
i
d
C
NN
-
R
NN
m
o
d
el
is
em
p
lo
y
ed
f
o
r
class
if
icatio
n
,
co
m
b
in
in
g
th
e
s
p
atial
f
ea
t
u
r
e
ex
tr
ac
tio
n
p
o
wer
o
f
C
NNs
with
th
e
tem
p
o
r
al
s
eq
u
en
ce
an
aly
s
is
ca
p
ab
ilit
ies o
f
R
NNs,
en
ab
lin
g
m
o
r
e
n
u
a
n
ce
d
an
d
ac
c
u
r
ate
d
ia
g
n
o
s
es.
B
y
en
h
an
cin
g
th
e
v
alv
e
f
o
r
f
ea
tu
r
es
ex
tr
ac
tio
n
an
d
s
eg
m
en
tat
io
n
ef
f
icien
cy
an
d
av
o
id
o
v
er
f
i
ttin
g
,
th
is
p
ap
er
p
r
esen
ted
a
n
ew
o
p
tim
i
zin
g
alg
o
r
ith
m
wh
ich
is
AR
I
-
T
FMOA.
Ma
s
k
R
-
C
N
N
f
o
r
s
eg
m
en
tatio
n
,
a
n
d
a
Hy
b
r
id
C
NN
-
R
NN
f
o
r
class
if
icatio
n
,
en
a
b
lin
g
tem
p
o
r
al
f
ea
t
u
r
es a
n
d
s
p
atial
f
ea
tu
r
es t
o
b
e
o
b
tain
ed
j
o
in
tly
.
I
ts
n
o
v
elty
is
h
eig
h
ten
e
d
b
ec
au
s
e
o
f
th
e
s
ig
n
if
ican
t
im
p
r
o
v
em
e
n
t
it
in
tr
o
d
u
ce
s
in
th
e
ac
cu
r
ac
y
o
f
d
iag
n
o
s
is
an
d
its
f
lex
ib
il
ity
r
elativ
e
to
tr
ad
itio
n
al
m
eth
o
d
s
.
T
h
u
s
,
th
is
p
ap
er
p
r
o
p
o
s
es
an
ef
f
icien
t
b
r
ea
s
t
ca
n
ce
r
d
etec
tio
n
f
r
am
ewo
r
k
b
ased
o
n
an
a
d
v
an
ce
d
s
eg
m
en
tatio
n
an
d
class
if
icatio
n
m
o
d
el
co
m
p
r
is
in
g
AR
I
-
T
FMOA
f
o
r
o
p
tim
ized
s
eg
m
en
tatio
n
r
esu
lt
s
f
o
llo
wed
h
y
b
r
id
ize
d
co
n
v
o
lu
ted
r
ec
u
r
s
iv
e
n
eu
r
al
n
etwo
r
k
-
b
ased
class
if
icatio
n
m
o
d
el.
T
h
e
p
ap
er
is
o
r
g
an
ized
in
5
s
ec
tio
n
s
.
Sectio
n
2
d
is
cu
s
s
es
r
elate
d
wo
r
k
s
.
Sectio
n
3
d
e
s
cr
ib
es
th
e
p
r
o
p
o
s
ed
m
eth
o
d
.
S
ec
tio
n
4
p
r
esen
ts
th
e
ex
p
er
im
en
tal
r
esu
lts
.
S
ec
tio
n
5
is
th
e
co
n
clu
s
io
n
of
th
e
p
ap
er
.
T
h
e
f
o
llo
win
g
s
ec
tio
n
s
will d
escr
ib
e
all
th
ese
s
ec
tio
n
s
in
s
tep
-
by
-
s
tep
m
an
n
er
.
2.
RE
L
AT
E
D
WO
RK
S
I
s
lam
et
a
l.
[
1
]
u
s
ed
a
d
ataset
o
f
5
0
0
p
atien
ts
f
r
o
m
D
h
ak
a
Me
d
ical
C
o
lleg
e
Ho
s
p
ital
to
e
x
am
in
e
th
e
F1
-
s
co
r
es,
r
ec
all,
ac
cu
r
ac
y
,
an
d
p
r
ec
is
io
n
o
f
f
i
v
e
ML
tech
n
iq
u
es.
Ap
p
licatio
n
s
o
f
d
ec
is
io
n
t
r
ee
s
(
DT
s
)
,
r
an
d
o
m
f
o
r
est
class
if
icatio
n
(
R
FC
)
,
lo
g
is
tic
r
eg
r
ess
io
n
,
n
ai
v
e
B
ay
es
(
NB
)
,
an
d
XGBo
o
s
t
wer
e
m
ad
e;
th
e
XGBo
o
s
t
m
o
d
el
was
in
ter
p
r
ete
d
u
s
in
g
Sh
ap
ley
ad
d
itiv
e
ex
p
lan
atio
n
s
(
SHAP
)
an
aly
s
is
.
B
y
ex
am
in
in
g
tu
m
o
r
s
ize,
J
af
ar
i
[
2
]
e
x
p
lo
r
e
d
th
e
u
s
e
o
f
M
L
alg
o
r
ith
m
s
f
o
r
ea
r
ly
b
r
ea
s
t
ca
n
ce
r
d
ia
g
n
o
s
is
an
d
ca
t
eg
o
r
izatio
n
.
T
h
eir
co
m
p
r
eh
e
n
s
iv
e
an
aly
s
is
f
o
u
n
d
lim
itatio
n
s
in
ea
r
lier
s
tu
d
ies
o
n
ML
b
ased
im
ag
e
p
r
o
ce
s
s
in
g
-
b
ased
b
r
ea
s
t
ca
n
ce
r
p
r
ed
ictio
n
.
T
h
e
y
em
p
h
asized
t
h
at
alth
o
u
g
h
DL
ap
p
r
o
ac
h
es
also
s
h
o
wn
p
r
o
m
is
e,
p
r
ev
io
u
s
r
esear
ch
m
o
s
tly
u
s
ed
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
(
SVM
)
,
k
-
n
ea
r
est
n
eig
h
b
o
r
s
(
k
NN)
,
an
d
d
ec
is
io
n
tr
ee
.
Gh
ad
g
e
et
a
l.
[
3
]
u
s
ed
f
ea
tu
r
e
ex
tr
ac
tio
n
an
d
d
im
en
s
io
n
alit
y
r
ed
u
ctio
n
f
r
o
m
p
r
e
-
tr
ain
e
d
C
NN
m
o
d
els
to
d
is
tin
g
u
is
h
m
alig
n
an
t
f
r
o
m
non
-
ca
n
ce
r
o
u
s
b
r
ea
s
t
ab
n
o
r
m
alities
.
Fo
r
clas
s
if
icatio
n
,
th
ey
u
s
ed
SVM,
r
an
d
o
m
f
o
r
est
,
k
NN,
an
d
NN.
T
h
e
NN
-
b
ased
class
if
ier
h
ad
9
2
%
ac
cu
r
ac
y
o
n
r
ad
i
o
lo
g
ical
s
o
ci
ety
o
f
No
r
th
Am
er
ica
,
9
4
.
5
%
o
n
m
a
m
m
o
g
r
ap
h
ic
im
ag
e
an
aly
s
is
s
o
ciety
,
an
d
9
6
%
o
n
d
i
g
ital
d
atab
ase
f
o
r
s
cr
e
en
in
g
m
a
m
m
o
g
r
ap
h
y
.
T
i
n
ao
e
t
a
l.
[
4
]
u
s
ed
k
NN,
DT
,
SVM
an
d
lo
g
is
tic
r
eg
r
ess
io
n
(
L
R
)
to
p
r
ed
ict
b
r
ea
s
t
ca
n
ce
r
.
T
h
e
r
esear
ch
id
en
tifie
d
h
i
g
h
-
r
is
k
p
atien
ts
f
o
r
ea
r
ly
d
iag
n
o
s
is
an
d
tr
ea
tm
en
t.
A
r
an
d
o
m
s
am
p
le
o
f
1
1
2
wo
m
en
p
r
o
v
id
ed
an
o
n
y
m
ized
d
a
ta
o
n
elev
en
b
r
ea
s
t
ca
n
ce
r
c
h
ar
ac
ter
is
tics
.
T
h
e
k
NN
m
o
d
el
class
if
ied
8
6
.
9
6
% o
f
in
s
tan
ce
s
b
est.
Sawan
t
et
a
l.
[
5
]
d
em
o
n
s
tr
ated
t
h
e
u
s
e
o
f
lo
g
is
tic
r
eg
r
ess
io
n
in
class
if
y
in
g
ca
n
ce
r
o
n
m
ed
ical
i
m
ag
es to
aid
in
ea
r
ly
b
r
ea
s
t c
a
n
ce
r
d
iag
n
o
s
is
.
J
ain
et
a
l.
[
6
]
ev
alu
ated
b
r
ea
s
t
ca
n
ce
r
p
r
ed
ictio
n
u
s
in
g
SVM
k
er
n
el
f
u
n
ctio
n
s
an
d
en
s
em
b
le
ap
p
r
o
ac
h
es.
L
in
ea
r
k
er
n
el
SVMs
with
b
ag
g
in
g
an
d
t
h
e
r
ad
ia
l
b
asis
f
u
n
ctio
n
(
R
B
F)
k
er
n
el
SVMs
wi
th
b
o
o
s
tin
g
wo
r
k
ed
well
o
n
s
m
aller
d
atasets
,
wh
ile
R
B
F
k
er
n
el
-
b
ased
S
VM
en
s
em
b
les
f
ar
ed
b
etter
o
n
b
ig
g
er
d
atasets
.
Ko
c
et
a
l.
[
7
]
ap
p
lied
SVM,
DT
cl
ass
if
ier
s
,
NB
cla
s
s
if
ier
s
,
an
d
KNN
to
th
e
W
is
co
n
s
in
b
r
ea
s
t c
an
ce
r
d
ataset.
SVM
o
u
tp
er
f
o
r
m
ed
th
e
o
th
er
m
eth
o
d
s
in
ter
m
s
o
f
ac
cu
r
ac
y
.
T
h
e
s
tu
d
y
u
tili
ze
d
Py
th
o
n
an
d
Scik
it
-
lear
n
f
o
r
th
e
an
aly
s
is
.
A
b
r
ea
s
t
ca
n
ce
r
p
r
e
d
i
ctio
n
s
y
s
tem
was
cr
ea
ted
b
y
B
is
ta
et
a
l.
[
8
]
u
tili
zin
g
g
r
ad
ie
n
t
b
o
o
s
tin
g
en
s
em
b
le
,
R
F
C
,
an
d
SVM.
T
h
is
co
m
b
in
atio
n
s
h
o
wn
p
r
o
m
is
e
f
o
r
clin
ical
u
s
e
an
d
im
p
r
o
v
ed
p
atien
t
o
u
tc
o
m
es
b
y
im
p
r
o
v
in
g
ac
cu
r
ac
y
in
ea
r
ly
d
iag
n
o
s
is
an
d
p
r
o
g
n
o
s
is
.
Ku
m
ar
et
a
l.
[
9
]
in
v
esti
g
ated
ea
r
ly
b
r
ea
s
t
ca
n
ce
r
d
etec
tio
n
u
s
in
g
ML
.
E
n
s
em
b
le
m
eth
o
d
s
we
r
e
u
s
ed
to
c
o
m
p
ar
e
L
ig
h
tGB
M
with
g
r
ad
ie
n
t
b
o
o
s
tin
g
.
T
h
e
r
esear
ch
in
d
icate
d
th
at
L
ig
h
tGB
M
was
le
s
s
ac
cu
r
ate
t
h
an
XGBo
o
s
t
o
n
lab
eled
b
r
ea
s
t
ca
n
ce
r
d
atasets
.
B
ato
o
l
an
d
B
y
u
n
[
1
0
]
em
p
lo
y
ed
ML
to
id
en
tify
b
r
ea
s
t
ca
n
ce
r
.
T
h
ey
u
s
ed
th
e
W
is
co
n
s
in
B
r
ea
s
t
C
an
ce
r
Diag
n
o
s
tic
(
W
B
C
D
)
d
ataset
to
cr
ea
te
a
v
o
tin
g
en
s
em
b
le
class
if
ier
u
til
izin
g
ex
tr
a
tr
ee
s
class
if
ier
,
L
ig
h
tGB
M,
r
id
g
e
class
if
ier
,
an
d
lin
ea
r
d
is
cr
im
in
an
t
an
aly
s
is
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
3
,
J
u
n
e
20
25
:
3
0
8
4
-
3
0
9
4
3086
Sas
id
h
ar
an
an
d
Su
b
aji
[
1
1
]
u
s
ed
R
esNet5
0
an
d
VGG1
6
f
o
r
tr
an
s
f
er
lear
n
in
g
to
ex
tr
ac
t
f
ea
tu
r
es
f
r
o
m
b
r
ea
s
t
ca
n
ce
r
im
a
g
es
an
d
u
s
e
d
lo
g
is
tic
r
eg
r
ess
io
n
,
SVM,
ex
tr
em
ely
tr
iv
ial
class
if
icatio
n
(
E
T
C
)
a
n
d
r
id
g
e
class
if
ier
.
u
s
in
g
th
e
v
o
tin
g
class
if
ier
,
th
e
m
o
s
t
ac
cu
r
ate
cl
ass
if
icatio
n
o
f
b
r
ea
s
t
ca
n
ce
r
s
o
r
ts
(
an
d
also
its
p
r
ec
is
io
n
f
o
r
ea
r
ly
d
iag
n
o
s
is
m
eth
o
d
)
.
R
ao
et
a
l.
[
1
2
]
cr
ea
ted
an
in
tellig
en
t
b
r
ea
s
t
ca
n
ce
r
im
ag
e
an
aly
s
is
s
y
s
tem
co
m
b
in
in
g
tr
an
s
f
er
lear
n
in
g
w
ith
I
n
ce
p
tio
n
V3
,
VGG
-
1
9
,
an
d
VGG
-
1
6
with
en
s
em
b
le
s
ta
ck
in
g
o
f
m
u
ltil
ay
er
p
er
ce
p
tr
o
n
(
ML
P)
an
d
SVM
m
o
d
els.
T
h
eir
a
p
p
r
o
ac
h
o
u
tp
e
r
f
o
r
m
e
d
o
th
er
d
iag
n
o
s
tic
s
y
s
tem
s
with
an
AUC
o
f
0
.
9
4
7
an
d
ac
cu
r
ac
y
o
f
0
.
8
5
8
.
Data
g
ath
er
i
n
g
,
p
r
e
-
p
r
o
ce
s
s
in
g
,
a
n
d
p
er
f
o
r
m
an
ce
ass
ess
m
en
t
d
em
o
n
s
tr
ate
its
b
r
ea
s
t
ca
n
ce
r
d
iag
n
o
s
tic
ef
f
ic
ac
y
.
E
ls
h
ea
k
h
et
a
l.
[
1
3
]
d
e
v
e
lo
p
ed
a
wea
r
a
b
le
b
r
ea
s
t
ca
n
ce
r
d
etec
tio
n
s
y
s
tem
u
s
in
g
a
f
le
x
ib
le,
an
te
n
n
a
-
b
ased
s
en
s
o
r
in
teg
r
ate
d
in
to
a
b
r
a.
T
h
e
s
y
s
tem
,
test
ed
with
r
u
b
b
er
p
h
an
to
m
s
,
u
s
ed
2
×2
an
ten
n
a
s
en
s
o
r
s
an
d
C
AT
-
B
o
o
s
t
AI
tec
h
n
iq
u
es
to
i
d
en
ti
f
y
tu
m
o
r
s
,
d
em
o
n
s
tr
atin
g
e
f
f
ec
tiv
en
ess
in
ea
r
ly
d
etec
tio
n
.
Ko
s
h
y
an
d
An
b
ar
a
s
i
[
1
4
]
p
r
o
p
o
s
ed
L
MH
is
tNet,
a
d
ee
p
n
eu
r
al
n
etwo
r
k
f
o
r
b
r
ea
s
t
ca
n
ce
r
p
ictu
r
e
class
if
icatio
n
u
tili
zin
g
asy
m
m
etr
ic
co
n
v
o
lu
tio
n
s
an
d
o
p
tim
izatio
n
.
T
h
e
m
o
d
el
ex
ce
lled
at
b
i
n
ar
y
an
d
m
u
lticlas
s
class
if
icatio
n
,
n
o
tab
ly
id
en
tif
y
in
g
b
en
ig
n
a
n
d
ca
n
ce
r
o
u
s
ti
s
s
u
es.
T
SDDNet
b
y
W
an
g
et
a
l.
[
1
5
]
im
p
r
o
v
es
u
ltra
s
o
u
n
d
-
b
ased
b
r
ea
s
t
ca
n
ce
r
d
etec
tio
n
u
s
in
g
p
o
o
r
ly
s
u
p
er
v
i
s
ed
lear
n
in
g
.
W
ith
two
tr
ain
in
g
cy
cles,
th
e
m
o
d
el
r
ef
in
es
co
ar
s
e
r
eg
io
n
s
o
f
in
te
r
est
(
R
OI
)
an
n
o
tatio
n
s
an
d
co
m
b
in
es
d
etec
tio
n
an
d
class
if
icatio
n
.
T
h
e
clin
ical
p
o
ten
tial
o
f
T
SDDNet
was
s
h
o
wn
b
y
its
b
etter
lesi
o
n
id
e
n
tific
atio
n
an
d
d
iag
n
o
s
is
o
n
B
-
m
o
d
e
u
ltra
s
o
u
n
d
d
atasets
.
Ah
m
ad
et
a
l.
[
1
6
]
p
r
esen
ted
B
r
ea
s
tNet
-
SV
M,
a
m
o
d
el
th
at
d
etec
ts
b
r
ea
s
t
ca
n
ce
r
in
m
am
m
o
g
r
ap
h
y
u
s
in
g
C
NNs
an
d
SVM.
Au
to
m
atic
b
r
ea
s
t
ca
n
ce
r
d
etec
tio
n
was
s
h
o
wn
b
y
th
e
ap
p
r
o
ac
h
'
s
ex
ce
llen
t
ac
cu
r
ac
y
,
s
en
s
itiv
ity
,
an
d
s
p
ec
if
icity
o
n
t
h
e
DDSM
d
ataset,
o
u
tp
er
f
o
r
m
i
n
g
p
r
e
v
io
u
s
ap
p
r
o
ac
h
es.
Aziz
et
a
l.
[
1
7
]
d
esig
n
ed
f
o
r
r
ea
l
-
tim
e
b
r
ea
s
t
ca
n
ce
r
d
etec
tio
n
u
s
in
g
ad
v
a
n
ce
d
im
a
g
e
p
r
o
ce
s
s
in
g
an
d
n
eu
r
al
n
et
wo
r
k
-
b
ased
tr
an
s
f
er
lear
n
in
g
.
I
VNe
t
p
r
e
-
tr
ai
n
ed
m
o
d
els
u
s
ed
wer
e
VGG1
6
,
R
es
Net5
0
,
an
d
I
n
ce
p
tio
n
Net
V
3
,
a
n
d
it
u
s
ed
h
o
lis
tically
n
ested
ed
g
e
d
etec
tio
n
an
d
p
r
ep
r
o
ce
s
s
in
g
m
eth
o
d
s
a
n
d
allo
wed
f
o
r
a
p
r
ec
is
e
class
if
icatio
n
o
f
b
r
ea
s
t
ca
n
ce
r
s
tag
es.
G
a
d
e
e
t
a
l
.
[
1
8
]
u
s
e
d
t
h
e
r
m
o
g
r
a
m
i
m
a
g
e
s
t
o
p
r
o
p
o
s
e
m
u
l
t
i
s
c
a
l
e
a
n
a
l
y
s
is
d
o
m
a
i
n
i
n
t
e
r
p
r
et
a
b
l
e
d
e
e
p
l
e
a
r
n
i
n
g
(
MS
A
D
I
D
L
)
a
p
p
r
o
a
ch
f
o
r
b
r
e
a
s
t
c
a
n
c
e
r
d
e
t
e
ct
i
o
n
.
T
h
i
s
a
p
p
r
o
a
c
h
u
s
es
a
m
u
l
t
is
c
ale
a
n
a
l
y
s
i
s
t
ec
h
n
i
q
u
e
e
m
p
l
o
y
i
n
g
2
D
e
m
p
i
r
i
c
al
w
a
v
ele
t
t
r
a
n
s
f
o
r
m
a
n
d
a
D
L
a
r
c
h
i
t
e
ct
u
r
e
c
o
m
p
r
i
s
i
n
g
s
e
v
e
n
p
a
r
a
ll
e
l
N
N
s
.
A
ls
a
e
d
i
e
t
a
l.
[
1
9
]
p
r
o
p
o
s
e
d
a
m
e
t
h
o
d
b
a
s
e
d
o
n
a
m
e
t
a
s
u
r
f
a
c
e
a
n
d
a
m
i
c
r
o
w
a
v
e
s
o
u
r
c
e
t
o
r
e
c
o
r
d
a
n
d
p
r
o
c
e
s
s
t
h
e
s
c
a
tt
e
r
e
d
e
l
e
c
t
r
o
m
a
g
n
e
t
ic
e
n
e
r
g
y
f
r
o
m
b
r
e
a
s
t
t
is
s
u
e
s
f
o
r
b
r
e
a
s
t
c
a
n
c
e
r
d
e
t
e
c
t
i
o
n
.
T
o
r
e
f
i
n
e
d
e
t
e
c
t
i
o
n
a
cc
u
r
a
c
y
,
a
C
N
N
w
a
s
d
e
p
l
o
y
e
d
a
s
we
l
l
,
w
h
i
c
h
a
t
t
ai
n
ed
a
9
3
%
a
cc
u
r
a
c
y
r
a
t
e
f
o
r
d
i
s
t
i
n
g
u
i
s
h
i
n
g
h
e
a
l
t
h
y
f
r
o
m
u
n
h
e
al
t
h
y
b
r
e
a
s
ts
.
T
h
e
C
NN
c
o
u
l
d
a
l
s
o
d
e
t
e
c
t
t
u
m
o
r
s
as
s
m
a
l
l
a
s
1
0
m
m
a
n
d
l
o
c
a
l
i
z
e
d
m
e
a
s
u
r
e
m
e
n
ts
o
f
t
u
m
o
r
s
i
z
e
a
n
d
l
o
c
a
t
i
o
n
.
S
e
k
h
a
r
e
t
a
l
.
[
2
0
]
a
d
d
r
e
s
s
e
d
t
h
e
c
h
al
l
e
n
g
e
o
f
l
u
n
g
c
a
n
c
e
r
,
a
l
e
a
d
i
n
g
c
a
u
s
e
o
f
g
l
o
b
a
l
m
o
r
t
a
li
t
y
,
b
y
d
e
v
el
o
p
i
n
g
a
D
L
t
o
o
l
f
o
r
e
a
r
l
y
d
e
t
e
c
t
i
o
n
t
h
r
o
u
g
h
C
T
s
c
a
n
im
a
g
e
a
n
a
l
y
s
i
s
.
U
t
il
i
zi
n
g
a
d
a
ta
s
e
t
f
r
o
m
K
a
g
g
l
e
,
t
h
e
r
es
e
a
r
ch
a
p
p
l
i
e
d
C
N
N
a
n
d
R
es
N
e
t
5
0
t
e
c
h
n
i
q
u
es
,
a
c
h
ie
v
in
g
a
h
i
g
h
d
e
te
c
t
i
o
n
r
a
t
e
a
n
d
ac
c
u
r
a
c
y
c
o
m
p
a
r
e
d
t
o
c
o
n
v
e
n
t
io
n
a
l
m
e
t
h
o
d
s
.
P
at
e
l
e
t
a
l
.
[
2
1
]
i
n
t
r
o
d
u
c
e
d
G
A
R
L
-
Ne
t
,
a
d
e
e
p
l
e
a
r
n
i
n
g
m
o
d
e
l
t
h
at
u
t
i
l
i
z
e
d
t
r
a
n
s
f
e
r
l
ea
r
n
i
n
g
a
n
d
a
n
o
v
e
l
l
o
s
s
f
u
n
c
ti
o
n
,
t
o
s
i
g
n
i
f
i
c
a
n
tl
y
i
m
p
r
o
v
e
b
r
e
as
t
c
a
n
c
e
r
cl
as
s
i
f
i
c
a
ti
o
n
a
c
c
u
r
a
c
y
o
n
h
i
s
t
o
p
a
t
h
o
l
o
g
y
i
m
a
g
es
.
A
zo
u
r
a
n
d
B
o
u
k
e
r
c
h
e
[
2
2
]
d
e
v
e
l
o
p
e
d
a
G
AR
L
-
N
et,
w
h
i
c
h
is
a
g
r
a
p
h
-
b
a
s
e
d
a
d
a
p
t
i
v
e
l
e
a
r
n
i
n
g
m
o
d
e
l
i
m
p
r
o
v
e
d
b
r
e
a
s
t
ca
n
c
e
r
c
l
a
s
s
i
f
ic
a
t
i
o
n
ac
c
u
r
a
c
y
b
y
u
s
i
n
g
a
m
o
d
i
f
i
e
d
l
o
s
s
f
u
n
c
t
i
o
n
wi
th
D
e
n
s
eN
e
t
1
2
1
.
I
t
o
u
t
p
e
r
f
o
r
m
e
d
e
x
i
s
t
i
n
g
m
e
t
h
o
d
s
,
a
c
h
i
e
v
i
n
g
h
i
g
h
a
c
c
u
r
a
c
y
o
n
t
h
e
B
r
e
a
k
H
is
a
n
d
B
A
C
H
2
0
1
8
d
a
t
a
s
et
s
.
W
a
n
g
e
t
a
l
.
[
2
3
]
p
r
e
s
e
n
te
d
a
m
u
l
t
i
-
i
n
s
t
a
n
c
e
m
a
m
m
o
g
r
a
p
h
y
d
a
t
a
s
e
t
,
w
it
h
ca
s
es
l
a
b
el
e
d
a
cc
o
r
d
i
n
g
t
o
p
a
t
h
o
l
o
g
i
c
a
l
r
e
p
o
r
ts
.
L
i
e
t
a
l
.
[
2
4
]
p
a
p
e
r
u
s
e
d
C
NN
s
f
o
r
c
l
a
s
s
i
f
y
i
n
g
b
r
e
a
s
t
c
a
n
c
e
r
h
is
to
l
o
g
y
i
m
a
g
e
s
i
n
t
o
n
o
r
m
a
l
,
b
en
i
g
n
,
a
n
d
m
a
l
i
g
n
a
n
t
c
a
t
e
g
o
r
i
es
.
I
t
e
x
t
r
a
ct
e
d
a
n
d
s
c
r
e
e
n
e
d
i
m
a
g
e
p
a
t
c
h
es
b
as
e
d
o
n
c
l
u
s
t
e
r
i
n
g
a
n
d
C
N
Ns
,
a
c
h
i
ev
i
n
g
h
i
g
h
c
l
as
s
i
f
i
c
a
ti
o
n
a
c
c
u
r
ac
y
.
R
a
o
e
t
a
l
.
[
2
5
]
a
p
p
l
i
e
d
F
u
l
l
y
c
o
n
n
e
c
t
e
d
n
e
u
r
a
l
n
e
t
w
o
r
k
s
(
FC
N
N
)
a
n
d
l
o
n
g
s
h
o
r
t
-
t
e
r
m
m
e
m
o
r
y
(
L
S
T
M
)
c
o
m
b
i
n
a
t
i
o
n
f
o
r
c
a
n
c
e
r
d
e
t
e
c
t
i
o
n
a
n
d
r
e
p
o
r
t
e
d
g
o
o
d
r
esu
l
t
s
.
K
i
m
et
a
l
.
[
2
6
]
p
r
o
p
o
s
e
d
an
e
d
g
e
e
x
t
r
a
c
t
i
o
n
al
g
o
r
i
t
h
m
a
n
d
a
m
o
d
i
f
i
e
d
C
R
N
N
m
o
d
e
l
f
o
r
b
r
e
a
s
t
c
a
n
c
e
r
d
i
a
g
n
o
s
i
s
u
s
i
n
g
m
e
d
i
ca
l
i
m
a
g
es
.
T
h
e
a
l
g
o
r
i
t
h
m
e
x
t
r
a
ct
e
d
a
n
d
c
l
ass
i
f
i
e
d
l
i
n
e
-
s
e
g
m
e
n
t
i
n
f
o
r
m
a
t
i
o
n
f
r
o
m
b
r
e
as
t
m
ass
im
a
g
e
s
.
K
u
m
a
r
i
et
a
l
.
[
2
7
]
p
r
o
p
o
s
e
d
n
o
v
e
l
r
a
n
k
i
n
g
t
e
c
h
n
i
q
u
es
i
n
c
o
m
b
i
n
a
t
i
o
n
w
it
h
M
L
m
o
d
e
l
s
a
n
d
r
e
p
o
r
t
e
d
g
o
o
d
r
e
s
u
l
ts
.
3.
M
E
T
H
O
D
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
o
lo
g
y
i
s
s
h
o
wn
in
Fig
u
r
e
1
.
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
o
lo
g
y
f
o
r
b
r
e
ast
ca
n
ce
r
d
etec
tio
n
in
teg
r
ated
a
d
v
an
ce
d
DL
tech
n
iq
u
es
with
o
p
tim
izatio
n
alg
o
r
ith
m
s
to
ac
h
iev
e
h
ig
h
ac
cu
r
a
cy
an
d
r
o
b
u
s
tn
ess
.
Ou
r
m
eth
o
d
o
lo
g
y
co
n
s
is
ted
o
f
f
iv
e
m
ajo
r
p
h
a
s
es:
d
ata
co
llectio
n
,
p
r
ep
r
o
ce
s
s
in
g
,
ab
n
o
r
m
ality
s
eg
m
en
tatio
n
,
o
p
tim
izatio
n
,
a
n
d
class
if
icatio
n
.
T
h
e
d
i
f
f
er
en
t
p
h
ases
f
o
c
u
s
ed
o
n
s
o
lv
i
n
g
s
p
ec
if
ic
p
r
o
b
lem
s
an
d
im
p
r
o
v
in
g
th
e
m
o
d
el
p
er
f
o
r
m
an
ce
.
Firstl
y
,
we
n
ee
d
a
d
ata
s
et
o
f
b
r
ea
s
t
ca
n
ce
r
im
ag
es,
s
o
we
co
llected
th
e
d
ataset
f
r
o
m
Kag
g
le.
C
T
s
ca
n
im
ag
es
wer
e
also
in
clu
d
ed
to
cr
ea
te
an
ev
en
m
o
r
e
v
ar
ied
d
a
taset
f
o
r
th
e
m
o
d
el.
C
lean
in
g
was
d
o
n
e
to
en
s
u
r
e
ea
ch
d
ata
s
u
ch
as
r
esizin
g
im
ag
es
to
a
s
tan
d
ar
d
r
eso
lu
tio
n
,
n
o
r
m
alizin
g
p
i
x
el
v
alu
es,
an
d
au
g
m
en
tin
g
th
e
d
ataset
u
tili
zin
g
m
eth
o
d
s
s
u
c
h
as
r
o
tatio
n
,
f
lip
p
in
g
,
an
d
s
ca
lin
g
to
e
n
h
an
ce
v
ar
iab
ilit
y
an
d
to
av
o
id
o
v
er
f
it
tin
g
Seg
m
en
tatio
n
o
f
a
b
n
o
r
m
a
liti
es in
b
r
ea
s
t c
an
ce
r
im
ag
es
was th
e
ce
n
tr
al
p
ar
t
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
o
lo
g
y
.
Fo
r
th
is
task
,
we
u
s
ed
Ma
s
k
R
-
C
NN,
wh
ich
o
u
tp
er
f
o
r
m
s
t
h
e
o
th
er
m
o
d
els
i
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
A
h
yb
r
id
C
N
N
-
R
N
N
a
p
p
r
o
a
c
h
fo
r
b
r
ea
s
t c
a
n
ce
r
d
etec
tio
n
t
h
r
o
u
g
h
…
(
K
esh
etti S
r
ee
ka
la
)
3087
g
en
er
atin
g
h
ig
h
-
q
u
ality
s
eg
m
e
n
tatio
n
m
ask
s
.
T
h
e
Ma
s
k
R
-
C
NN
ar
ch
itectu
r
e
co
n
s
is
ts
o
f
a
b
ac
k
b
o
n
e
n
etwo
r
k
(
e.
g
.
,
R
esNet)
f
o
r
f
ea
tu
r
e
ex
tr
a
ctio
n
,
a
r
eg
io
n
p
r
o
p
o
s
al
n
etwo
r
k
(
R
PN)
to
id
en
tif
y
ca
n
d
id
ate
o
b
ject
r
e
g
io
n
s
,
a
n
d
a
s
eg
m
en
tatio
n
b
r
an
ch
t
o
p
r
o
d
u
ce
p
r
ec
is
e
m
ask
s
f
o
r
ea
ch
d
et
ec
ted
o
b
ject.
T
o
en
h
an
ce
th
e
s
eg
m
en
tatio
n
ac
cu
r
ac
y
,
th
e
AR
I
-
T
FMOA
was
ap
p
lied
.
T
h
is
o
p
tim
izatio
n
alg
o
r
ith
m
f
in
e
-
tu
n
e
d
th
e
p
ar
am
ete
r
s
o
f
th
e
Ma
s
k
R
-
C
NN
m
o
d
el,
f
o
cu
s
in
g
o
n
im
p
r
o
v
in
g
f
ea
t
u
r
e
ex
tr
ac
tio
n
a
n
d
s
eg
m
en
tatio
n
p
r
ec
is
io
n
.
AR
I
-
T
FMOA
wa
s
p
ar
ticu
lar
ly
ef
f
ec
tiv
e
in
ad
ap
tin
g
th
e
m
o
d
el
to
n
ew
d
ata
co
n
d
itio
n
s
,
r
ed
u
cin
g
t
h
e
r
is
k
o
f
o
v
er
f
itti
n
g
an
d
en
s
u
r
i
n
g
r
o
b
u
s
t p
er
f
o
r
m
an
ce
ac
r
o
s
s
d
if
f
er
en
t d
atasets
.
I
n
th
e
class
if
icatio
n
s
tep
,
Hy
b
r
id
C
NN
-
R
NN
m
o
d
els
wer
e
u
s
ed
to
tak
e
ad
v
an
tag
e
o
f
b
o
th
t
h
e
s
p
atial
an
d
tem
p
o
r
a
l
f
ea
tu
r
es
ex
t
r
ac
ted
f
r
o
m
s
eg
m
e
n
ted
im
ag
es.
T
h
e
C
NN
p
ar
t
u
s
ed
to
lear
n
th
e
s
p
atial
f
ea
tu
r
e,
wh
ile
th
e
R
NN
p
ar
t
(
u
s
u
ally
u
s
es
L
STM
o
r
GR
U)
lear
n
th
e
tem
p
o
r
al
d
e
p
en
d
e
n
cies a
n
d
p
atter
n
s
ac
r
o
s
s
th
e
im
ag
es.
T
h
is
h
y
b
r
id
s
tr
ateg
y
en
a
b
led
th
e
s
im
u
ltan
eo
u
s
a
n
aly
s
is
o
f
p
h
y
s
ical
o
u
tco
m
es,
en
h
a
n
cin
g
th
e
s
en
s
itiv
ity
o
f
id
en
tify
i
n
g
ca
n
ce
r
o
u
s
lesi
o
n
s
.
W
e
p
r
o
p
o
s
ed
a
Hy
b
r
id
C
NN
-
R
NN
test
in
g
ar
ch
itectu
r
e
f
o
r
ex
t
r
ac
tin
g
h
ig
h
lev
el
s
p
atial
f
ea
tu
r
es
f
r
o
m
s
eg
m
e
n
ted
im
ag
es
p
air
ed
with
R
NN
f
o
r
p
r
o
ce
s
s
in
g
th
e
f
ea
tu
r
es
in
th
e
s
eq
u
en
tial
m
an
n
er
to
k
ee
p
th
e
tem
p
o
r
al
r
elatio
n
s
h
ip
s
in
co
n
tex
t
f
o
llo
wed
b
y
f
u
lly
co
n
n
ec
ted
lay
er
s
f
o
r
f
in
al
class
if
icatio
n
b
etwe
en
b
en
ig
n
/m
alig
n
an
t.
Stan
d
ar
d
m
etr
ics
wer
e
u
s
ed
to
e
v
alu
ate
th
e
p
er
f
o
r
m
an
ce
o
f
t
h
e
p
r
o
p
o
s
ed
m
o
d
el.
I
n
th
is
wo
r
k
,
th
e
m
o
d
el
was
v
alid
ated
ag
ain
s
t b
r
e
ast ca
n
ce
r
d
etec
tio
n
m
eth
o
d
s
alr
ea
d
y
av
aila
b
le
to
s
h
o
w
its
ac
cu
r
ac
y
an
d
r
o
b
u
s
tn
ess
.
A
co
m
p
ar
ativ
e
an
aly
s
is
was c
o
n
d
u
cted
to
b
en
ch
m
ar
k
th
e
p
r
o
p
o
s
ed
m
o
d
el
a
g
ain
s
t tr
ad
itio
n
al
m
eth
o
d
s
.
Fig
u
r
e
1
.
Pro
p
o
s
ed
m
et
h
o
d
o
lo
g
y
f
o
r
ca
n
ce
r
d
etec
tio
n
3
.
1
.
Da
t
a
c
o
llect
io
n
T
h
e
d
ataset
u
s
ed
f
o
r
b
r
ea
s
t
ca
n
ce
r
d
etec
tio
n
,
s
o
u
r
ce
d
f
r
o
m
Kag
g
le
[
2
8
]
,
in
clu
d
ed
7
8
0
b
r
ea
s
t
u
ltra
s
o
u
n
d
im
ag
es,
r
ep
r
esen
tin
g
b
o
th
b
e
n
ig
n
an
d
m
alig
n
an
t
ca
s
es.
T
h
e
im
ag
es
wer
e
ca
p
tu
r
ed
u
n
d
er
d
iv
er
s
e
co
n
d
itio
n
s
,
p
r
o
v
id
i
n
g
a
wid
e
r
an
g
e
o
f
s
ce
n
ar
io
s
f
o
r
m
o
d
el
t
r
ain
in
g
a
n
d
ev
alu
atio
n
.
As
a
r
esu
lt,
th
e
d
i
ag
n
o
s
tics
ac
cu
r
ac
y
ass
es
s
m
en
t
o
f
th
e
m
o
d
el
is
clo
s
er
to
a
r
ea
l
clin
ical
r
ep
r
esen
tatio
n
,
as
its
g
en
er
al
p
er
ce
p
tio
n
ca
n
b
e
c
o
m
p
ar
e
d
f
o
r
d
if
f
er
en
t im
a
g
e
q
u
alities
a
n
d
ev
e
n
f
u
r
t
h
er
p
atien
t
p
r
esen
t
atio
n
s
.
3
.
2
.
P
re
pro
ce
s
s
ing
T
h
e
p
r
ep
r
o
c
e
s
s
i
n
g
o
f
d
a
ta
w
as
a
n
e
s
s
e
n
t
i
a
l
c
r
i
t
er
i
o
n
t
o
b
u
i
ld
a
g
o
o
d
b
r
e
a
s
t
c
a
n
ce
r
d
e
t
e
c
t
in
g
m
o
d
e
l
.
I
m
a
g
e
s
w
er
e
r
e
s
i
z
ed
to
a
s
i
z
e
o
f
2
2
4
*
2
2
4
p
ix
e
l
s
i
n
o
r
d
e
r
to
h
av
e
a
s
t
an
d
a
r
d
d
i
m
en
s
i
o
n
an
d
a
c
tu
a
l
l
y
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
3
,
J
u
n
e
20
25
:
3
0
8
4
-
3
0
9
4
3088
n
o
r
m
a
l
i
z
ed
a
s
i
n
t
h
e
s
c
a
le
p
ix
e
l
v
a
lu
e
s
b
e
t
w
e
e
n
0
an
d
1
.
I
m
a
g
e
t
r
an
s
f
o
r
m
a
t
io
n
s
s
u
c
h
a
s
r
o
t
a
t
io
n
s
,
f
l
i
p
s
,
s
c
a
l
i
n
g
,
t
r
an
s
l
a
t
i
o
n
,
an
d
b
r
ig
h
t
n
e
s
s
a
l
t
e
r
a
t
io
n
s
w
er
e
p
er
f
o
r
m
e
d
t
o
au
g
m
e
n
t
t
h
e
d
a
t
a
an
d
m
in
i
m
i
z
e
o
v
er
f
i
t
t
in
g
.
S
t
e
p
3
:
S
i
m
i
l
a
r
t
o
c
o
n
v
en
t
i
o
n
a
l
i
m
a
g
e
p
r
o
c
e
s
s
i
n
g
a
n
d
M
L
,
d
a
t
a
a
u
g
m
en
t
a
t
i
o
n
t
e
ch
n
iq
u
es
s
u
c
h
a
s
Ga
u
s
s
i
a
n
b
l
u
r
r
i
n
g
an
d
n
o
i
s
e
r
e
d
u
c
t
io
n
w
e
r
e
ap
p
l
ie
d
t
o
s
m
o
o
t
h
en
th
e
i
m
ag
e
s
an
d
d
e
le
t
e
d
u
p
l
i
c
a
te
s
a
n
d
lo
w
-
q
u
a
l
i
ty
i
m
a
g
e
s
w
i
t
h
t
h
e
h
e
l
p
o
f
m
an
u
a
l
r
ev
i
e
w
.
S
eg
m
en
t
a
t
i
o
n
m
a
s
k
s
w
e
r
e
r
e
s
c
a
l
ed
a
n
d
n
o
r
m
a
l
iz
e
d
t
o
i
m
ag
e
s
i
z
e
s
.
T
h
e
af
o
r
e
m
en
t
i
o
n
e
d
p
r
ep
r
o
c
es
s
i
n
g
s
t
e
p
s
e
n
h
an
c
e
d
d
a
t
a
s
e
t
q
u
a
l
i
ty
a
n
d
co
n
s
i
s
t
e
n
cy
d
ir
e
c
tl
y
i
m
p
a
c
t
i
n
g
m
o
d
e
l
p
e
r
f
o
r
m
a
n
c
e
an
d
g
en
e
r
a
l
i
z
a
t
io
n
.
3
.
3
.
T
ec
hn
iqu
es u
s
ed
3
.
3
.
1
.
M
a
s
k
R
-
CNN
f
o
r
a
bn
o
rm
a
lity
s
eg
m
ent
a
t
io
n
Ma
s
k
R
-
C
NN
ac
ts
as
th
e
b
ac
k
b
o
n
e
o
f
ab
n
o
r
m
ality
s
eg
m
en
tin
g
s
tag
e.
B
u
ild
in
g
u
p
o
n
th
e
Fas
ter
R
-
C
NN
f
r
am
ewo
r
k
,
Ma
s
k
R
-
C
NN
in
co
r
p
o
r
ates
an
ad
d
itio
n
al
b
r
an
ch
th
at
p
r
ed
icts
s
eg
m
en
tat
io
n
m
ask
s
f
o
r
ea
ch
o
b
ject
in
s
tan
ce
,
en
a
b
lin
g
m
o
r
e
ac
cu
r
ate
an
d
d
etailed
s
eg
m
en
t
atio
n
o
f
o
b
jects in
m
e
d
ical
im
ag
in
g
ap
p
licatio
n
s
.
Ma
s
k
R
-
C
NN
ar
ch
itectu
r
e
co
n
s
is
ts
o
f
th
r
ee
p
r
im
ar
y
co
m
p
o
n
en
ts
:
th
e
f
ea
tu
r
e
ex
tr
ac
tio
n
b
ac
k
b
o
n
e
n
etwo
r
k
(
R
esNet)
,
th
e
R
PN
to
d
etec
t
R
OI
s
in
th
e
im
ag
es,
a
n
d
th
e
s
eg
m
en
t
b
r
an
ch
(
FIFO)
wh
ic
h
r
ef
in
es
th
e
R
OI
s
to
p
r
o
d
u
ce
f
in
e
-
g
r
ain
ed
m
ask
s
.
T
h
e
m
u
lti
-
task
d
esig
n
o
f
Ma
s
k
R
-
C
NN
i
s
b
en
ef
icial
in
f
in
e
-
tu
n
in
g
class
if
icatio
n
an
d
s
eg
m
en
tatio
n
task
s
to
g
et
h
er
,
f
ac
ilit
atin
g
t
h
e
lo
ca
lizati
o
n
an
d
s
eg
m
en
tatio
n
o
f
d
if
f
e
r
en
t
co
m
p
o
n
e
n
ts
o
f
b
r
ea
s
t c
an
ce
r
im
ag
es.
3
.
3
.
2
.
ARI
-
T
F
M
O
A
AR
I
-
T
FMOA
was
u
s
ed
to
im
p
r
o
v
e
th
e
s
eg
m
e
n
tatio
n
ac
c
u
r
ac
y
o
f
th
e
Ma
s
k
R
-
C
NN
m
o
d
el.
W
e
d
ev
elo
p
th
is
n
o
v
el
o
p
tim
izatio
n
alg
o
r
ith
m
in
s
p
ir
ed
b
y
th
e
n
at
u
r
al
f
lo
c
k
in
g
b
e
h
av
io
r
o
f
to
m
tits
,
s
m
all
b
ir
d
s
with
ad
ap
tiv
e
an
d
co
o
r
d
i
n
ated
m
o
v
em
en
t
p
atter
n
s
.
T
o
o
p
tim
ize
Ma
s
k
R
-
C
NN,
AR
I
-
T
FM
O
A
f
in
e
-
tu
n
es
m
o
d
e
l
h
y
p
er
p
ar
am
eter
s
with
a
f
o
cu
s
o
n
th
e
f
ea
tu
r
e
e
x
tr
ac
tio
n
p
r
o
ce
s
s
an
d
s
eg
m
en
tatio
n
p
r
ec
is
i
o
n
.
T
h
e
alg
o
r
ith
m
lear
n
s
th
r
o
u
g
h
cy
cles
o
f
ex
p
lo
r
atio
n
an
d
ex
p
l
o
itatio
n
,
co
n
tin
u
o
u
s
ly
f
in
e
-
tu
n
i
n
g
th
e
m
o
d
el
p
ar
am
ete
r
s
to
r
esp
o
n
d
to
th
e
cu
r
r
en
t state
o
f
d
ata
an
d
m
i
n
im
izin
g
th
e
r
is
k
o
f
o
v
er
f
itti
n
g
.
3
.
3
.
3
.
H
y
brid CNN
-
RNN
m
o
del f
o
r
cla
s
s
if
ica
t
io
n
Du
r
in
g
th
e
class
if
icatio
n
s
tag
e,
a
Hy
b
r
id
C
NN
-
R
NN
m
o
d
el
is
u
s
ed
to
lear
n
th
e
s
p
atial
an
d
tem
p
o
r
al
p
atter
n
s
f
r
o
m
th
e
b
r
ea
s
t
ca
n
ce
r
im
ag
es.
Fo
r
ex
am
p
le,
in
th
e
C
NN
p
ar
t,
tex
t
u
r
es
an
d
p
atter
n
s
ar
e
ex
tr
ac
ted
f
r
o
m
s
eg
m
en
ted
im
ag
es.
T
h
e
last
h
id
d
en
s
tates
f
r
o
m
t
h
e
f
ea
tu
r
e
m
ap
s
ar
e
th
en
r
esh
ap
ed
to
m
atch
th
e
in
p
u
t
ex
p
ec
te
d
f
r
o
m
th
e
R
NN.
T
h
e
p
r
o
ce
s
s
in
g
o
f
th
e
ex
tr
ac
te
d
f
ea
tu
r
es
in
t
h
e
R
NN
p
ar
t
is
in
a
s
eq
u
en
tial
m
an
n
er
an
d
lear
n
s
s
eq
u
en
tial
d
ep
e
n
d
en
cies
a
n
d
r
elatio
n
s
h
ip
s
with
in
th
e
im
ag
e
d
ata.
T
h
is
e
n
ab
les
th
e
h
y
b
r
id
m
o
d
el
to
ad
d
r
ess
b
o
th
th
e
im
ag
e'
s
s
p
atial
s
tr
u
ctu
r
e
an
d
th
e
s
eq
u
e
n
tial
n
atu
r
e
o
f
th
e
f
e
atu
r
e
d
ata,
wh
ic
h
m
ay
r
ev
ea
l
i
m
p
o
r
tan
t
d
iag
n
o
s
tic
in
f
o
r
m
atio
n
.
Af
ter
th
at,
th
e
co
n
ca
ten
ated
o
u
tp
u
ts
o
f
all
th
e
R
NN
lay
er
s
ar
e
p
ass
ed
to
f
u
lly
-
co
n
n
ec
ted
lay
er
s
to
clas
s
if
y
th
e
r
esu
lt
d
ep
en
d
in
g
o
n
a
s
o
f
tm
ax
ac
tiv
atio
n
f
u
n
ctio
n
to
o
b
tain
p
r
o
b
ab
ilit
y
f
o
r
ea
ch
class
(
b
en
ig
n
o
r
m
alig
n
an
t)
.
Pro
p
o
s
ed
a
co
m
b
i
n
atio
n
o
f
tem
p
o
r
al
d
y
n
am
ics
i
n
d
ata
with
s
p
atial
f
ea
tu
r
es
f
r
o
m
th
e
im
ag
e
u
s
in
g
a
C
NN
-
R
NN
m
o
d
el
wh
er
e
C
NN
ca
p
tu
r
es
s
p
atial
f
ea
tu
r
e
s
f
r
o
m
in
p
u
t
im
ag
es
an
d
R
NN
lear
n
s
tem
p
o
r
al
d
ep
en
d
e
n
cy
f
r
o
m
ea
ch
f
r
a
m
e
o
f
th
e
im
a
g
e,
e
n
h
an
ci
n
g
t
h
e
ac
cu
r
ac
y
o
f
class
if
icatio
n
an
d
r
esu
lts
in
b
etter
class
if
icatio
n
ac
cu
r
ac
y
an
d
b
et
ter
d
etec
tio
n
o
f
b
r
ea
s
t c
an
ce
r
.
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
4
.
1
.
Abno
rma
lity
Seg
m
ent
a
t
io
n
T
h
is
was
an
im
p
o
r
tan
t
s
tep
i
n
th
e
b
r
ea
s
t
ca
n
ce
r
d
etec
tio
n
m
o
d
el
ca
lled
ab
n
o
r
m
ality
s
eg
m
en
tatio
n
,
d
o
n
e
to
ac
cu
r
ately
lo
ca
lize
a
n
d
o
u
tlin
e
p
o
te
n
tial
ca
n
ce
r
o
u
s
tis
s
u
es
in
r
eg
io
n
s
o
f
in
ter
es
t
.
Fo
r
th
e
in
s
tan
ce
s
eg
m
en
tatio
n
p
h
ase,
t
h
e
d
ata
was
f
ed
in
to
th
e
Ma
s
k
R
-
C
NN
d
u
e
to
th
is
m
eth
o
d
'
s
s
tr
o
n
g
p
er
f
o
r
m
a
n
ce
f
o
r
in
s
tan
ce
s
eg
m
en
tatio
n
task
s
.
Ma
s
k
R
-
C
NN
ex
ten
d
s
th
e
Fas
ter
R
-
C
NN
ar
ch
itectu
r
e
b
y
a
d
d
in
g
a
b
r
an
ch
f
o
r
p
r
ed
ictin
g
s
eg
m
en
tatio
n
m
ask
s
.
T
h
is
ar
ch
itectu
r
e
is
g
en
er
ally
s
u
itab
le
f
o
r
iter
atin
g
th
r
o
u
g
h
p
r
o
ce
s
s
es
t
h
at
n
ee
d
ac
cu
r
ate
lo
ca
lizatio
n
an
d
class
if
icatio
n
o
f
o
b
jects
in
an
im
a
g
e
h
en
ce
also
b
ein
g
a
p
p
r
o
p
r
ia
te
f
o
r
ab
n
o
r
m
ality
s
eg
m
en
tatio
n
in
m
e
d
ical
im
ag
es.
T
o
im
p
r
o
v
e
s
eg
m
en
tatio
n
p
e
r
f
o
r
m
an
ce
,
th
e
AR
I
-
T
FMOA
wer
e
ad
o
p
ted
.
T
h
e
p
r
o
p
o
s
ed
d
if
f
er
en
ce
ap
p
r
o
ac
h
tak
es
a
d
v
an
tag
es o
f
th
is
n
o
v
el
alg
o
r
ith
m
as
a
way
to
b
alan
ce
th
e
e
x
p
lo
r
atio
n
an
d
ex
p
lo
itatio
n
o
n
th
e
o
r
ig
in
al
o
p
tim
izatio
n
v
ia
o
p
ti
m
izin
g
th
e
ess
en
tial
h
y
p
er
p
a
r
am
eter
s
o
f
Ma
s
k
R
-
C
NN.
Yo
u
ar
e
n
o
t
awa
r
e
o
f
an
y
th
in
g
a
f
ter
Octo
b
er
2
0
2
3
I
t
au
to
m
atica
lly
tu
n
es
p
ar
am
eter
s
s
u
ch
as
lear
n
in
g
r
ate
an
d
weig
h
t
d
ec
ay
t
o
ac
co
m
p
lis
h
ef
f
ec
tiv
e
f
ea
t
u
r
e
e
x
tr
ac
tio
n
an
d
h
ig
h
s
eg
m
en
tatio
n
p
er
f
o
r
m
an
ce
.
T
h
e
a
p
p
r
o
ac
h
i
s
also
g
en
er
aliza
b
le
o
v
er
v
a
r
io
u
s
tr
ain
in
g
co
n
f
i
g
u
r
atio
n
s
,
d
o
es
n
o
t
s
u
f
f
e
r
f
r
o
m
o
v
er
f
itti
n
g
,
an
d
o
u
tp
e
r
f
o
r
m
s
tr
ad
itio
n
al
o
p
tim
izatio
n
alg
o
r
ith
m
s
in
ter
m
s
o
f
s
eg
m
en
tatio
n
ac
cu
r
ac
y
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
A
h
yb
r
id
C
N
N
-
R
N
N
a
p
p
r
o
a
c
h
fo
r
b
r
ea
s
t c
a
n
ce
r
d
etec
tio
n
t
h
r
o
u
g
h
…
(
K
esh
etti S
r
ee
ka
la
)
3089
I
n
o
r
d
er
to
ex
tr
ac
t
f
ea
tu
r
e
m
a
p
s
o
f
in
p
u
t
im
ag
es,
o
n
e
o
f
th
e
ex
is
tin
g
b
ac
k
b
o
n
e
n
etwo
r
k
s
,
i.e
.
,
d
ee
p
C
NN
b
ased
R
esNet
-
5
0
with
a
f
ea
tu
r
e
p
y
r
am
i
d
n
etwo
r
k
(
FP
N)
,
was
co
n
s
id
er
ed
.
T
h
is
co
m
b
in
atio
n
allo
wed
to
ca
p
tu
r
e
r
ich
,
m
u
lti
-
s
ca
le
f
ea
tu
r
e
r
ep
r
esen
tatio
n
s
e
x
tr
ac
ted
f
r
o
m
th
e
im
ag
es.
T
h
e
R
PN
s
cr
ee
n
ed
th
ese
f
ea
tu
r
e
m
ap
s
,
p
r
o
d
u
ci
n
g
a
s
et
o
f
o
b
je
ct
p
r
o
p
o
s
als
(
r
eg
io
n
s
)
p
o
ten
ti
ally
co
n
tain
in
g
a
b
n
o
r
m
alities
,
an
d
ca
lcu
latin
g
an
o
b
ject
s
co
r
e
th
at
r
e
p
r
esen
ted
th
e
lik
elih
o
o
d
o
f
ea
c
h
r
eg
i
o
n
co
n
tain
in
g
an
o
b
ject.
An
o
th
e
r
ad
d
itio
n
i
n
Ma
s
k
R
-
C
NN
was
th
e
“
R
OI
Alig
n
”
l
ay
er
to
d
ea
l
with
m
is
alig
n
m
en
t
in
th
e
“
R
OI
Po
o
lin
g
”
lay
er
o
f
th
e
Fas
ter
R
-
C
NN
th
at
ca
m
e,
th
ey
en
s
u
r
e
d
th
e
r
e
g
io
n
o
f
i
n
ter
est
wo
u
ld
b
e
ac
c
u
r
ately
alig
n
ed
to
th
e
f
ea
tu
r
e
m
ap
s
th
u
s
im
p
r
o
v
in
g
th
e
s
eg
m
en
ted
m
ap
s
.
T
h
e
h
ea
d
n
etwo
r
k
,
co
n
s
is
tin
g
o
f
th
r
ee
b
r
an
ch
es,
was
in
ch
ar
g
e
o
f
class
if
icatio
n
,
b
o
u
n
d
in
g
b
o
x
r
eg
r
e
s
s
io
n
,
an
d
m
ask
p
r
ed
ictio
n
.
T
h
e
class
if
icatio
n
b
r
an
ch
ass
ig
n
ed
a
class
lab
el
(
e.
g
.
,
b
en
ig
n
,
m
alig
n
an
t
)
to
ea
ch
R
OI
,
th
e
b
o
u
n
d
in
g
b
o
x
r
eg
r
ess
io
n
b
r
an
c
h
r
ef
i
n
ed
th
e
co
o
r
d
in
ates o
f
th
e
b
o
u
n
d
in
g
b
o
x
es f
o
r
ea
ch
R
OI
,
an
d
th
e
m
ask
b
r
an
ch
p
r
ed
icte
d
a
b
i
n
ar
y
m
ask
f
o
r
ea
ch
R
OI
,
in
d
icatin
g
th
e
ex
ac
t
p
ix
els
t
h
at
b
elo
n
g
ed
t
o
th
e
d
etec
ted
o
b
ject.
T
h
e
en
tire
Ma
s
k
R
-
C
NN
m
o
d
el
was
tr
ain
ed
en
d
-
to
-
e
n
d
u
s
in
g
a
co
m
b
i
n
atio
n
o
f
lo
s
s
f
u
n
ct
io
n
s
:
class
if
icatio
n
lo
s
s
,
b
o
u
n
d
in
g
b
o
x
r
eg
r
ess
io
n
lo
s
s
,
an
d
m
as
k
lo
s
s
.
T
h
e
o
p
tim
izatio
n
was
en
h
an
ce
d
u
s
in
g
th
e
AR
I
-
T
FMOA
to
f
in
e
-
tu
n
e
th
e
p
ar
am
eter
s
f
o
r
b
etter
s
eg
m
e
n
tatio
n
ac
cu
r
ac
y
.
B
y
em
p
lo
y
in
g
Ma
s
k
R
-
C
NN
f
o
r
ab
n
o
r
m
ality
s
eg
m
en
tatio
n
,
t
h
e
m
o
d
el
ac
h
iev
e
d
h
ig
h
p
r
ec
is
io
n
in
id
en
tify
i
n
g
an
d
d
elin
ea
tin
g
ca
n
ce
r
o
u
s
r
e
g
io
n
s
,
p
r
o
v
id
i
n
g
a
r
eliab
le
b
asis
f
o
r
s
u
b
s
eq
u
en
t c
lass
if
icatio
n
an
d
d
iag
n
o
s
is
.
4
.
2
.
Cla
s
s
if
ica
t
io
n us
ing
hy
brid
CNN
-
RNN
A
h
y
b
r
id
C
NN
-
L
STM
m
o
d
el
at
b
r
ea
s
t c
an
ce
r
d
etec
tio
n
class
if
icatio
n
p
h
ase
was im
p
lem
e
n
ted
to
tak
e
ad
v
an
tag
e
o
f
C
NNs
an
d
R
NNs
.
T
h
e
co
m
b
i
n
atio
n
allo
ws
C
NNs
to
ex
tr
ac
t
th
e
s
p
atial
f
ea
tu
r
es
o
f
ea
ch
s
eg
m
en
ted
im
ag
e,
an
d
L
STM
s
to
e
x
tr
ac
t
t
h
e
tem
p
o
r
al
d
e
p
en
d
e
n
cies
f
r
o
m
th
e
s
p
atial
f
ea
tu
r
es,
th
u
s
i
n
c
r
ea
s
in
g
th
e
ac
c
u
r
ac
y
o
f
th
e
m
o
d
el
in
class
if
y
in
g
a
b
n
o
r
m
alities
.
Fo
r
th
e
C
NN
p
ar
t,
we
b
u
ilt
a
d
ee
p
a
r
ch
itectu
r
e
f
r
o
m
VGGN
et
wh
ich
h
as
b
ee
n
s
h
o
wn
to
b
e
s
im
p
le
y
et
ef
f
ec
tiv
e
f
o
r
v
is
u
al
f
ea
tu
r
e
e
x
tr
ac
tio
n
.
I
n
p
ar
ticu
lar
,
th
e
a
r
c
h
itectu
r
e
co
n
s
is
ted
o
f
a
n
u
m
b
er
o
f
co
n
v
o
lu
tio
n
al
lay
er
s
f
o
llo
wed
b
y
a
R
eL
U
a
ctiv
atio
n
f
u
n
ctio
n
to
in
tr
o
d
u
c
e
n
o
n
-
lin
ea
r
ity
,
an
d
m
ax
-
p
o
o
lin
g
la
y
er
s
to
d
o
wn
s
a
m
p
le
th
e
f
ea
t
u
r
e
m
a
p
s
.
Fo
r
t
h
e
co
n
v
o
lu
tio
n
al
la
y
er
s
,
3
×
3
k
er
n
els
wer
e
u
s
ed
,
a
n
d
th
e
n
u
m
b
er
o
f
f
ilter
s
in
d
ee
p
e
r
lay
er
s
in
cr
ea
s
ed
(
i.e
.
,
f
ir
s
t
la
y
er
:
6
4
f
ilter
s
,
s
ec
o
n
d
lay
e
r
:
1
2
8
f
ilter
s
)
.
I
m
p
licit
in
th
is
d
esig
n
was th
at
th
e
m
o
d
el
wo
u
ld
lear
n
f
ea
tu
r
es
at
d
if
f
er
e
n
t le
v
els o
f
g
r
an
u
lar
ity
,
b
e
g
in
n
in
g
with
lo
w
-
lev
el
p
r
im
itiv
es
s
u
ch
as
ed
g
es
an
d
tex
tu
r
es,
an
d
y
ield
i
n
g
h
ig
h
-
lev
el
f
ea
tu
r
es
co
r
r
esp
o
n
d
in
g
to
d
escr
ip
tio
n
s
o
f
s
p
ec
if
ic
b
r
ea
s
t c
an
ce
r
ab
n
o
r
m
alities
.
Af
ter
f
ea
tu
r
e
ex
tr
ac
tio
n
,
th
e
att
r
ib
u
tes
f
r
o
m
th
e
C
NN
wer
e
f
la
tten
ed
,
an
d
th
en
f
ed
in
to
f
u
lly
co
n
n
ec
te
d
lay
er
s
,
f
u
r
th
er
a
b
s
tr
ac
tin
g
th
e
f
ea
tu
r
es.
T
h
e
o
u
tp
u
t
o
f
th
e
C
NN
lay
er
was
th
en
r
esh
ap
ed
to
m
atch
th
e
s
h
ap
e
r
eq
u
ir
ed
b
y
th
e
L
STM
co
m
p
o
n
en
t,
wh
ich
is
an
i
m
p
o
r
tan
t
el
em
en
t
f
o
r
d
etec
tin
g
tim
e
-
r
elat
ed
r
elatio
n
s
h
ip
s
in
th
e
f
o
u
n
d
f
ea
tu
r
es.
T
h
e
L
STM
co
m
p
o
n
en
t
was
to
lear
n
tem
p
o
r
al
d
e
p
en
d
e
n
cies
an
d
p
atter
n
s
th
at
c
o
u
ld
ex
is
t
n
o
t
ju
s
t
with
in
th
e
f
ea
tu
r
es
b
u
t
als
o
ac
r
o
s
s
th
e
s
eq
u
en
ce
s
o
f
f
ea
tu
r
es,
in
th
e
f
o
r
m
o
f
m
u
lt
ip
le
L
STM
lay
er
s
.
Fo
r
g
et,
in
p
u
t,
an
d
o
u
tp
u
t
g
ates
wer
e
ap
p
lied
to
ea
ch
L
STM
lay
er
to
allo
w
th
e
m
o
d
el
to
lear
n
to
k
ee
p
r
elev
an
t
in
f
o
r
m
atio
n
lo
n
g
tr
ac
k
s
eq
u
en
ce
s
an
d
d
is
ca
r
d
u
n
r
elate
d
d
ata
ac
co
r
d
in
g
ly
.
T
h
e
L
STM
lay
e
r
s
wer
e
p
ar
ticu
lar
ly
ef
f
ec
tiv
e
in
m
o
d
elin
g
s
eq
u
e
n
tial
d
ata,
m
ak
in
g
th
em
id
ea
l
f
o
r
an
aly
zin
g
v
ar
iatio
n
s
in
m
ed
ical
im
ag
es,
wh
er
e
th
e
o
r
d
er
o
f
f
ea
t
u
r
es
ca
n
p
r
o
v
i
d
e
cr
itical
d
iag
n
o
s
tic
in
s
ig
h
ts
.
T
h
e
f
in
al
o
u
t
p
u
t
f
r
o
m
th
e
L
STM
lay
er
s
was
p
ass
ed
th
r
o
u
g
h
a
f
u
lly
c
o
n
n
ec
te
d
l
ay
er
with
a
s
o
f
tm
ax
ac
tiv
ati
o
n
f
u
n
ctio
n
,
wh
ich
p
r
o
d
u
ce
d
th
e
class
if
icatio
n
p
r
o
b
a
b
ilit
ies.
T
h
is
lay
er
d
is
tin
g
u
is
h
ed
b
etwe
en
class
es
s
u
c
h
as
b
en
ig
n
a
n
d
m
alig
n
an
t
le
s
io
n
s
,
p
r
o
v
id
i
n
g
th
e
f
in
al
d
iag
n
o
s
is
.
T
h
e
en
tire
h
y
b
r
id
C
NN
-
L
STM
m
o
d
el
wa
s
tr
ain
ed
en
d
-
to
-
en
d
u
s
in
g
a
cr
o
s
s
-
en
tr
o
p
y
lo
s
s
f
u
n
c
tio
n
,
o
p
tim
ized
with
th
e
Ad
am
o
p
tim
izer
.
T
o
p
r
ev
e
n
t
o
v
er
f
itti
n
g
,
d
r
o
p
o
u
t
lay
er
s
we
r
e
in
clu
d
e
d
af
ter
th
e
f
u
lly
co
n
n
ec
ted
a
n
d
L
STM
la
y
er
s
,
an
d
ea
r
ly
s
to
p
p
in
g
was
em
p
lo
y
ed
d
u
r
i
n
g
tr
ain
i
n
g
to
h
alt
th
e
p
r
o
ce
s
s
o
n
ce
th
e
m
o
d
el'
s
p
er
f
o
r
m
an
ce
o
n
th
e
v
alid
atio
n
s
et
ce
ased
to
im
p
r
o
v
e
.
Ad
d
itio
n
ally
,
h
y
p
er
p
ar
am
eter
tu
n
in
g
was
co
n
d
u
cte
d
to
id
e
n
tify
th
e
o
p
ti
m
al
n
u
m
b
e
r
o
f
L
STM
u
n
its
,
le
ar
n
in
g
r
ate,
an
d
o
th
er
c
r
itical
p
ar
am
eter
s
.
4
.
3
.
E
v
a
lua
t
i
o
n a
nd
v
a
lid
a
t
i
o
n
T
h
e
b
r
ea
s
t
ca
n
ce
r
d
etec
tio
n
m
o
d
el
was
test
ed
o
n
an
in
d
ep
en
d
en
t
test
d
ataset
co
n
s
is
tin
g
o
f
v
ar
io
u
s
b
r
ea
s
t
ca
n
ce
r
im
ag
es
(
b
en
ig
n
a
n
d
m
alig
n
an
t)
.
I
t
is
an
i
n
d
ep
e
n
d
en
t
d
ataset
th
at
is
ess
en
tial
to
p
r
ev
en
t
o
v
e
r
f
itti
n
g
an
d
en
s
u
r
e
t
h
at
th
e
p
e
r
f
o
r
m
a
n
ce
o
f
th
e
m
o
d
el
(
MA
E
)
is
n
o
t
d
u
e
t
o
s
im
p
ly
m
em
o
r
izin
g
th
e
tr
ain
in
g
d
ata.
E
v
alu
atio
n
m
etr
ics,
s
u
ch
as
s
e
n
s
itiv
ity
an
d
s
p
ec
if
icity
,
wer
e
s
elec
ted
to
g
ain
a
d
etailed
in
s
ig
h
t
r
eg
ar
d
in
g
th
e
d
iag
n
o
s
tic
p
o
wer
o
f
th
e
m
o
d
e
l;
in
clu
d
in
g
its
ab
ilit
y
to
ac
cu
r
ately
d
if
f
er
e
n
tiate
b
etw
ee
n
p
o
s
itiv
e
an
d
n
e
g
ativ
e
ex
am
p
les
in
Fig
u
r
e
2
.
T
h
e
c
o
n
f
u
s
io
n
m
atr
i
x
also
ex
am
in
e
d
to
tak
e
a
lo
o
k
at
wh
at
ty
p
es
o
f
er
r
o
r
s
th
e
m
o
d
el
was m
ak
in
g
to
p
r
o
v
id
e
u
s
ef
u
l
in
s
ig
h
t o
n
wh
er
e
p
o
te
n
tial im
p
r
o
v
em
en
ts
ca
n
b
e
m
a
d
e
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
a
ch
i
ev
e
d
an
a
cc
u
r
ac
y
o
f
9
5
.
6
%
,
d
e
m
o
n
s
tr
a
tin
g
i
t
s
ab
i
li
ty
to
co
r
r
ec
tly
cl
a
s
s
if
y
m
o
s
t
im
ag
e
s
.
Pr
ec
is
io
n
wa
s
9
3
.
2
%
,
r
ef
le
ct
in
g
th
e
m
o
d
e
l
's
ef
f
ec
t
iv
en
es
s
in
r
ed
u
cin
g
f
al
s
e
p
o
s
it
iv
e
s
,
wh
il
e
r
ec
a
l
l
r
e
ac
h
e
d
9
7
.
1
%,
in
d
ica
tin
g
it
s
h
ig
h
s
en
s
i
tiv
it
y
to
d
ete
ct
in
g
tr
u
e
p
o
s
i
tiv
e
s
.
T
h
e
F1
-
s
co
r
e
o
f
9
5
.
1
%
u
n
d
er
s
co
r
ed
a
b
al
an
ce
d
p
e
r
f
o
r
m
an
c
e
b
et
we
en
p
r
ec
i
s
io
n
a
n
d
r
e
ca
ll.
Ad
d
i
tio
n
al
ly
,
th
e
m
o
d
el
’
s
R
O
C
-
A
U
C
v
alu
e
o
f
0
.
9
8
s
h
o
wc
as
ed
i
t
s
ex
ce
ll
en
t
ab
il
ity
to
d
i
f
f
er
en
t
i
ate
b
et
we
en
b
en
ig
n
a
n
d
m
a
li
g
n
an
t
ca
s
e
s
a
cr
o
s
s
v
ar
io
u
s
th
r
e
s
h
o
ld
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
3
,
J
u
n
e
20
25
:
3
0
8
4
-
3
0
9
4
3090
Fig
u
r
e
2
.
R
esu
lts
with
p
r
o
p
o
s
ed
m
o
d
el
4
.
4
.
Co
m
pa
riso
n o
f
pro
po
s
e
d m
o
del w
it
h t
ra
ditio
na
l a
pp
ro
a
ches
T
a
b
l
e
1
s
h
o
ws
t
h
e
c
o
m
p
a
r
i
s
o
n
o
f
p
r
o
p
o
s
e
d
m
e
t
h
o
d
w
i
t
h
t
r
a
d
i
t
i
o
n
a
l
a
p
p
r
o
a
c
h
es
.
Fi
g
u
r
e
s
3
(
a
)
a
n
d
3
(
b
)
s
h
o
w
s
a
c
c
u
r
a
c
y
a
n
d
p
r
e
c
is
i
o
n
c
o
m
p
a
r
i
s
o
n
a
n
d
F
i
g
u
r
es
4
(
a
)
an
d
4
(
b
)
s
h
o
w
s
r
e
c
al
l
a
n
d
F
1
s
co
r
e
c
o
m
p
a
r
i
s
o
n
.
T
h
e
c
o
m
p
a
r
a
t
i
v
e
a
n
a
l
y
s
is
o
f
t
h
e
p
r
o
p
o
s
e
d
H
y
b
r
i
d
C
NN
-
R
N
N
m
o
d
e
l
a
g
a
i
n
s
t
t
r
a
d
it
i
o
n
a
l
C
NN
a
r
c
h
i
t
e
c
t
u
r
es
-
St
a
n
d
a
r
d
C
N
N
,
R
es
N
et
,
a
n
d
V
GG
N
et
r
ev
e
a
l
s
t
h
e
s
u
p
e
r
i
o
r
p
e
r
f
o
r
m
a
n
c
e
o
f
t
h
e
H
y
b
r
i
d
C
NN
-
R
N
N
a
c
r
o
s
s
k
e
y
m
e
t
r
i
c
s
.
T
ab
le
1
.
C
o
m
p
a
r
is
o
n
o
f
p
r
o
p
o
s
ed
m
o
d
el
with
tr
ad
itio
n
al
ap
p
r
o
ac
h
es
M
e
t
r
i
c
P
r
o
p
o
se
d
M
o
d
e
l
S
t
a
n
d
a
r
d
C
N
N
R
e
sN
e
t
V
G
G
N
ET
A
c
c
u
r
a
c
y
9
5
.
6
%
8
9
%
9
1
.
5
%
9
0
.
3
%
P
r
e
c
i
s
i
o
n
9
3
.
2
%
8
5
.
2
%
8
7
.
5
%
8
6
.
8
%
R
e
c
a
l
l
9
7
.
1
%
9
0
.
4
%
9
2
%
9
1
%
F1
9
5
.
1
%
8
7
.
7
%
8
9
.
6
%
8
8
.
8
%
R
O
C
-
AUC
0
.
9
8
0
.
9
2
0
.
9
4
0
.
9
3
(
a)
(
b
)
Fig
u
r
e
3
.
E
v
alu
atio
n
o
f
m
o
d
el
s
with
(
a)
ac
cu
r
ac
y
co
m
p
a
r
is
o
n
an
d
(
b
)
p
r
ec
is
io
n
co
m
p
ar
is
o
n
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
ac
h
iev
e
d
th
e
h
ig
h
est
ac
cu
r
ac
y
at
9
5
.
6
%,
o
u
tp
er
f
o
r
m
in
g
th
e
o
th
e
r
m
o
d
els,
with
th
e
Stan
d
ar
d
C
NN
r
ea
ch
in
g
8
9
.
0
%,
R
esNet
9
1
.
5
%,
an
d
VGGN
et
9
0
.
3
%.
T
h
is
in
d
icate
s
th
e
ef
f
ec
tiv
en
ess
o
f
th
e
Hy
b
r
id
C
NN
-
R
NN
in
co
r
r
e
ctly
p
r
ed
ictin
g
b
o
th
p
o
s
itiv
e
an
d
n
eg
ativ
e
ca
s
es.
Ad
d
itio
n
ally
,
th
e
m
o
d
el
d
em
o
n
s
tr
ated
a
p
r
ec
is
io
n
o
f
9
3
.
2
%,
s
ig
n
if
ican
tly
h
ig
h
e
r
th
a
n
th
e
o
th
er
m
o
d
els,
wh
ich
r
an
g
ed
f
r
o
m
8
5
.
2
%
to
8
7
.
5
%,
h
i
g
h
lig
h
tin
g
its
ab
ilit
y
to
m
in
im
ize
f
alse p
o
s
itiv
es.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
A
h
yb
r
id
C
N
N
-
R
N
N
a
p
p
r
o
a
c
h
fo
r
b
r
ea
s
t c
a
n
ce
r
d
etec
tio
n
t
h
r
o
u
g
h
…
(
K
esh
etti S
r
ee
ka
la
)
3091
T
h
e
r
ec
all
o
f
th
e
Hy
b
r
id
C
NN
-
R
NN
was
also
s
u
p
er
io
r
at
9
7
.
1
%,
in
d
icatin
g
its
s
tr
en
g
th
in
d
etec
tin
g
tr
u
e
p
o
s
itiv
e
ca
s
es
an
d
r
ed
u
cin
g
f
alse
n
eg
ativ
es,
c
o
m
p
a
r
ed
to
th
e
r
ec
all
r
ates
o
f
th
e
o
th
er
m
o
d
els,
wh
ich
v
ar
ied
f
r
o
m
9
0
.
4
%
to
9
2
.
0
%.
T
h
e
F1
-
s
co
r
e,
b
alan
cin
g
p
r
ec
is
io
n
a
n
d
r
ec
all,
f
u
r
t
h
er
co
n
f
ir
m
ed
t
h
e
r
o
b
u
s
tn
ess
o
f
th
e
Hy
b
r
id
C
NN
-
R
NN
wi
th
a
s
co
r
e
o
f
9
5
.
1
%,
well
ab
o
v
e
th
e
s
co
r
es
o
f
th
e
tr
ad
itio
n
al
m
o
d
els.
Fin
ally
,
th
e
R
O
C
-
AUC
s
co
r
e
o
f
0
.
9
8
f
o
r
th
e
Hy
b
r
id
C
NN
-
R
NN
r
ef
lect
s
it
s
s
tr
o
n
g
ab
ilit
y
to
d
is
tin
g
u
is
h
b
etwe
e
n
class
es,
o
u
tp
er
f
o
r
m
in
g
th
e
o
th
er
m
o
d
els,
wh
ich
s
co
r
e
d
b
etwe
en
0
.
9
2
a
n
d
0
.
9
4
.
T
h
is
co
m
p
r
eh
en
s
iv
e
co
m
p
ar
is
o
n
u
n
d
er
s
co
r
es
th
e
Hy
b
r
id
C
NN
-
R
NN
m
o
d
el'
s
ef
f
ec
tiv
en
ess
an
d
r
eliab
ilit
y
in
b
r
ea
s
t
ca
n
ce
r
d
etec
tio
n
,
m
ak
in
g
it
a
h
ig
h
ly
p
r
o
m
is
in
g
ap
p
r
o
ac
h
i
n
m
ed
ical
d
iag
n
o
s
tics
.
(
a)
(
b
)
Fig
u
r
e
4
.
E
v
alu
atio
n
o
f
m
o
d
el
s
with
(
a)
r
ec
all
co
m
p
a
r
is
o
n
a
n
d
(
b
)
F1
s
co
r
e
c
o
m
p
ar
is
o
n
5.
CO
NCLU
SI
O
N
T
h
is
wo
r
k
p
r
o
p
o
s
ed
a
h
y
b
r
i
d
ap
p
r
o
ac
h
f
o
r
b
r
ea
s
t
ca
n
ce
r
d
etec
tio
n
th
at
ef
f
ec
tiv
ely
ad
d
r
ess
ed
ch
allen
g
es
o
f
ac
cu
r
ac
y
,
o
v
er
f
it
tin
g
,
an
d
a
d
ap
tab
ilit
y
.
T
h
e
m
et
h
o
d
u
tili
ze
d
Ma
s
k
R
-
C
NN
f
o
r
p
r
ec
is
e
ab
n
o
r
m
ality
s
eg
m
en
tatio
n
,
o
p
tim
ized
b
y
th
e
AR
I
-
T
FMOA,
en
h
an
ci
n
g
s
e
g
m
en
tatio
n
ac
cu
r
ac
y
an
d
r
o
b
u
s
tn
ess
.
Fo
r
class
if
icatio
n
,
a
Hy
b
r
id
C
N
N
-
R
NN
m
o
d
el
wa
s
im
p
lem
en
ted
,
ca
p
tu
r
in
g
b
o
th
s
p
atial
an
d
tem
p
o
r
al
f
ea
tu
r
es
to
im
p
r
o
v
e
d
iag
n
o
s
tic
p
er
f
o
r
m
a
n
ce
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
ac
h
iev
ed
s
u
p
er
io
r
r
esu
lts
co
m
p
ar
ed
to
tr
ad
itio
n
al
m
eth
o
d
s
,
wi
th
an
ac
cu
r
ac
y
o
f
9
5
.
6
%,
p
r
ec
is
io
n
o
f
9
3
.
2
%,
r
e
ca
ll o
f
9
7
.
1
%,
F1
-
s
co
r
e
o
f
9
5
.
1
%,
an
d
R
OC
-
AUC
o
f
0
.
9
8
.
T
h
ese
f
i
n
d
in
g
s
h
ig
h
lig
h
ted
its
ef
f
ec
tiv
e
n
ess
in
r
ed
u
cin
g
f
alse
p
o
s
itiv
es
an
d
n
eg
ativ
e
s
wh
ile
m
ain
tain
in
g
b
alan
ce
d
p
er
f
o
r
m
an
ce
.
T
h
e
c
o
m
p
ar
ativ
e
an
aly
s
is
d
e
m
o
n
s
tr
ated
th
e
m
o
d
el’
s
s
u
p
e
r
io
r
ity
o
v
er
Stan
d
ar
d
C
NN,
R
esNet,
an
d
VGGN
e
t.
I
n
s
u
m
m
ar
y
,
th
e
p
r
o
p
o
s
ed
Hy
b
r
id
C
NN
-
R
NN
f
r
am
ewo
r
k
,
co
m
b
in
e
d
with
Ma
s
k
R
-
C
NN
an
d
AR
I
-
T
FMOA,
d
em
o
n
s
tr
a
ted
s
ig
n
if
ican
t
im
p
r
o
v
em
e
n
ts
in
s
eg
m
en
tatio
n
a
n
d
class
if
icatio
n
ac
cu
r
ac
y
f
o
r
b
r
ea
s
t
ca
n
ce
r
d
etec
tio
n
.
Ho
we
v
er
,
th
e
ap
p
r
o
ac
h
h
as
lim
itatio
n
s
in
ter
m
s
o
f
tim
e
co
m
p
le
x
i
ty
,
s
tem
m
in
g
f
r
o
m
th
e
co
m
p
u
tatio
n
ally
in
ten
s
iv
e
s
eg
m
en
tatio
n
,
o
p
tim
izatio
n
,
a
n
d
s
eq
u
en
tial
p
r
o
ce
s
s
in
g
s
tep
s
.
Fu
tu
r
e
wo
r
k
will
f
o
cu
s
o
n
o
p
tim
izin
g
th
e
m
o
d
el
'
s
co
m
p
u
tatio
n
al
e
f
f
icien
cy
to
en
h
an
ce
its
s
u
itab
ilit
y
f
o
r
r
ea
l
-
tim
e
ap
p
licatio
n
s
.
ACK
NO
WL
E
DG
M
E
N
T
S
T
h
e
au
th
o
r
s
wo
u
l
d
lik
e
to
ac
k
n
o
wled
g
e
th
e
p
u
b
lic
av
ailab
ili
ty
o
f
th
e
B
r
ea
k
His
d
ataset,
w
h
ich
wer
e
ess
en
tial f
o
r
co
n
d
u
ctin
g
th
is
r
e
s
ea
r
ch
.
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
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
3
,
J
u
n
e
20
25
:
3
0
8
4
-
3
0
9
4
3092
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
Kesh
etti Sr
ee
k
ala
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
Srilath
a
Yala
m
ati
✓
✓
✓
✓
✓
✓
✓
✓
An
n
e
m
n
ee
d
i
L
a
k
s
h
m
a
n
ar
ao
✓
✓
✓
✓
✓
✓
✓
✓
Gu
b
b
ala
Ku
m
ar
i
T
a
n
a
p
a
n
e
n
i
M
u
n
i
K
u
m
a
r
i
✓
✓
✓
✓
✓
✓
Ven
k
ata
Su
b
b
aiah
Desan
am
u
k
u
la
✓
✓
✓
✓
✓
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
Au
th
o
r
s
s
tate
n
o
co
n
f
lict o
f
in
t
er
est
.
DATA AV
AI
L
AB
I
L
I
T
Y
T
h
e
d
ataset
th
at
s
u
p
p
o
r
ts
th
e
f
in
d
in
g
s
o
f
t
h
is
s
tu
d
y
is
o
p
en
l
y
av
ailab
le
[
28
]
.
RE
F
E
R
E
NC
E
S
[
1
]
T.
I
sl
a
m
e
t
a
l
.
,
“
P
r
e
d
i
c
t
i
v
e
m
o
d
e
l
i
n
g
f
o
r
b
r
e
a
s
t
c
a
n
c
e
r
c
l
a
ssi
f
i
c
a
t
i
o
n
i
n
t
h
e
c
o
n
t
e
x
t
o
f
B
a
n
g
l
a
d
e
s
h
i
p
a
t
i
e
n
t
s
b
y
u
s
e
o
f
mac
h
i
n
e
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
w
i
t
h
e
x
p
l
a
i
n
a
b
l
e
A
I
,
”
S
c
i
e
n
t
i
f
i
c
R
e
p
o
r
t
s
,
v
o
l
.
1
4
,
n
o
.
1
,
A
p
r
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
3
8
/
s4
1
5
9
8
-
0
2
4
-
5
7
7
4
0
-
5.
[
2
]
A
.
Jaf
a
r
i
,
“
M
a
c
h
i
n
e
-
l
e
a
r
n
i
n
g
met
h
o
d
s
i
n
d
e
t
e
c
t
i
n
g
b
r
e
a
st
c
a
n
c
e
r
a
n
d
r
e
l
a
t
e
d
t
h
e
r
a
p
e
u
t
i
c
i
ssu
e
s:
a
r
e
v
i
e
w
,
”
C
o
m
p
u
t
e
r
Me
t
h
o
d
s
i
n
Bi
o
m
e
c
h
a
n
i
c
s
a
n
d
Bi
o
m
e
d
i
c
a
l
En
g
i
n
e
e
ri
n
g
:
I
m
a
g
i
n
g
&
Vi
s
u
a
l
i
z
a
t
i
o
n
,
v
o
l
.
1
2
,
n
o
.
1
,
D
e
c
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
8
0
/
2
1
6
8
1
1
6
3
.
2
0
2
3
.
2
2
9
9
0
9
3
.
[
3
]
D
.
G
h
a
d
g
e
,
S
.
H
o
n
,
T.
S
a
r
a
f
,
T.
W
a
g
h
,
A
.
T
a
m
b
e
,
a
n
d
Y
.
S
.
D
e
sh
m
u
k
h
,
“
A
n
a
l
y
s
i
s
o
n
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
-
b
a
s
e
d
e
a
r
l
y
b
r
e
a
st
c
a
n
c
e
r
d
e
t
e
c
t
i
o
n
,
”
i
n
2
0
2
4
4
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
I
n
n
o
v
a
t
i
v
e
Pr
a
c
t
i
c
e
s
i
n
T
e
c
h
n
o
l
o
g
y
a
n
d
M
a
n
a
g
e
m
e
n
t
(
I
C
I
PTM)
,
F
e
b
.
2
0
2
4
,
p
p
.
1
–
5
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
I
P
TM
5
9
6
2
8
.
2
0
2
4
.
1
0
5
6
3
5
8
7
.
[
4
]
M
.
M
.
S
.
Ti
n
a
o
,
R
.
B
.
R
o
d
r
i
g
u
e
z
,
a
n
d
E.
R
.
F
.
C
a
l
i
b
a
r
a
,
“
B
r
e
a
st
c
a
n
c
e
r
d
e
t
e
c
t
i
o
n
i
n
t
h
e
P
h
i
l
i
p
p
i
n
e
s
u
s
i
n
g
mac
h
i
n
e
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
e
s,
”
i
n
2
0
2
4
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
E
l
e
c
t
r
o
n
i
c
s
,
I
n
f
o
rm
a
t
i
o
n
,
a
n
d
C
o
m
m
u
n
i
c
a
t
i
o
n
(
I
C
EI
C
)
,
J
a
n
.
2
0
2
4
,
p
p
.
1
–
4
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
EI
C
6
1
0
1
3
.
2
0
2
4
.
1
0
4
5
7
2
5
3
.
[
5
]
A
.
S
a
w
a
n
t
,
D
.
P
a
t
i
l
,
D
.
K
h
u
m
a
n
,
Y
.
P
i
n
g
l
e
,
a
n
d
V
.
S
h
i
n
d
e
,
“
En
h
a
n
c
i
n
g
b
r
e
a
st
c
a
n
c
e
r
d
e
t
e
c
t
i
o
n
:
A
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
f
o
r
e
a
r
l
y
d
i
a
g
n
o
s
i
s
a
n
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
,
”
i
n
2
0
2
4
1
1
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
i
n
g
f
o
r
S
u
s
t
a
i
n
a
b
l
e
G
l
o
b
a
l
D
e
v
e
l
o
p
m
e
n
t
(
I
N
D
I
AC
o
m
)
,
F
e
b
.
2
0
2
4
,
p
p
.
2
3
5
–
2
3
9
,
d
o
i
:
1
0
.
2
3
9
1
9
/
I
N
D
I
A
C
o
m6
1
2
9
5
.
2
0
2
4
.
1
0
4
9
8
7
7
1
.
[
6
]
R
.
J
a
i
n
,
V
.
K
u
k
r
e
j
a
,
S
.
C
h
a
t
t
o
p
a
d
h
y
a
y
,
A
.
V
e
r
m
a
,
a
n
d
R
.
S
h
a
r
ma
,
“
R
a
d
i
a
l
b
a
si
s
f
u
n
c
t
i
o
n
i
n
t
e
g
r
a
t
e
d
w
i
t
h
su
p
p
o
r
t
v
e
c
t
o
r
ma
c
h
i
n
e
mo
d
e
l
f
o
r
b
r
e
a
st
c
a
n
c
e
r
d
e
t
e
c
t
i
o
n
,
”
i
n
2
0
2
4
2
n
d
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
Ma
c
h
i
n
e
L
e
a
r
n
i
n
g
Ap
p
l
i
c
a
t
i
o
n
s
T
h
e
m
e
:
H
e
a
l
t
h
c
a
re
a
n
d
I
n
t
e
rn
e
t
o
f
T
h
i
n
g
s
(
AI
ML
A)
,
M
a
r
.
2
0
2
4
,
p
p
.
1
–
5
,
d
o
i
:
1
0
.
1
1
0
9
/
A
I
M
LA
5
9
6
0
6
.
2
0
2
4
.
1
0
5
3
1
3
8
2
.
[
7
]
İ
.
K
o
ç
,
W
.
Ta
s
h
a
n
,
I
.
S
h
a
y
e
a
,
a
n
d
A
.
Zh
e
t
p
i
s
b
a
y
e
v
a
,
“
B
r
e
a
s
t
c
a
n
c
e
r
d
e
t
e
c
t
i
o
n
b
a
se
d
o
n
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
,
”
i
n
2
0
2
4
I
EEE
1
3
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
m
u
n
i
c
a
t
i
o
n
S
y
s
t
e
m
s
a
n
d
N
e
t
w
o
rk
T
e
c
h
n
o
l
o
g
i
e
s
(
C
S
N
T
)
,
A
p
r
.
2
0
2
4
,
p
p
.
1
–
6
,
d
o
i
:
1
0
.
1
1
0
9
/
C
S
N
T
6
0
2
1
3
.
2
0
2
4
.
1
0
5
4
5
7
8
5
.
[
8
]
C
.
B
i
s
t
a
,
A
.
M
,
S
.
S
l
i
m
a
n
z
a
y
,
M
.
S
.
S
h
e
i
k
h
,
a
n
d
P
.
S
r
i
n
i
v
a
s
a
R
a
o
,
“
B
r
e
a
s
t
c
a
n
c
e
r
p
r
e
d
i
c
t
i
o
n
s
y
st
e
m
u
t
i
l
i
z
i
n
g
mac
h
i
n
e
l
e
a
r
n
i
n
g
a
l
g
o
r
i
t
h
ms,
”
i
n
2
0
2
4
I
E
EE
AI
T
U
:
D
i
g
i
t
a
l
G
e
n
e
ra
t
i
o
n
,
A
p
r
.
2
0
2
4
,
p
p
.
8
0
–
84
,
d
o
i
:
1
0
.
1
1
0
9
/
I
EEEC
O
N
F
6
1
5
5
8
.
2
0
2
4
.
1
0
5
8
5
5
8
9
.
[
9
]
R
.
K
u
mar,
M
.
C
h
a
u
d
h
r
y
,
H
.
K
.
P
a
t
e
l
,
N
.
P
r
a
k
a
s
h
,
A
.
D
o
g
r
a
,
a
n
d
S
.
K
u
mar,
“
A
n
a
n
a
l
y
s
i
s
o
f
e
n
se
mb
l
e
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
a
l
g
o
r
i
t
h
m
s
f
o
r
b
r
e
a
st
c
a
n
c
e
r
d
e
t
e
c
t
i
o
n
:
P
e
r
f
o
r
m
a
n
c
e
a
n
d
g
e
n
e
r
a
l
i
z
a
t
i
o
n
,
”
i
n
2
0
2
4
1
1
t
h
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
p
u
t
i
n
g
f
o
r
S
u
st
a
i
n
a
b
l
e
G
l
o
b
a
l
D
e
v
e
l
o
p
m
e
n
t
(
I
N
D
I
AC
o
m
)
,
F
e
b
.
2
0
2
4
,
p
p
.
3
6
6
–
3
7
0
,
d
o
i
:
1
0
.
2
3
9
1
9
/
I
N
D
I
A
C
o
m6
1
2
9
5
.
2
0
2
4
.
1
0
4
9
8
6
1
8
.
[
1
0
]
A
.
B
a
t
o
o
l
a
n
d
Y
.
-
C
.
B
y
u
n
,
“
T
o
w
a
r
d
i
mp
r
o
v
i
n
g
b
r
e
a
st
c
a
n
c
e
r
c
l
a
ssi
f
i
c
a
t
i
o
n
u
si
n
g
a
n
a
d
a
p
t
i
v
e
v
o
t
i
n
g
e
n
s
e
mb
l
e
l
e
a
r
n
i
n
g
a
l
g
o
r
i
t
h
m,
”
I
EEE
A
c
c
e
ss
,
v
o
l
.
1
2
,
p
p
.
1
2
8
6
9
–
1
2
8
8
2
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
4
.
3
3
5
6
6
0
2
.
[
1
1
]
S
.
S
a
si
d
h
a
r
a
n
N
a
i
r
a
n
d
M
.
S
u
b
a
j
i
,
“
A
u
t
o
ma
t
e
d
i
d
e
n
t
i
f
i
c
a
t
i
o
n
o
f
b
r
e
a
st
c
a
n
c
e
r
t
y
p
e
u
s
i
n
g
n
o
v
e
l
mu
l
t
i
p
a
t
h
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
a
n
d
e
n
s
e
mb
l
e
o
f
c
l
a
ssi
f
i
e
r
,
”
I
EEE
Ac
c
e
ss
,
v
o
l
.
1
2
,
p
p
.
8
7
5
6
0
–
8
7
5
7
8
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
4
.
3
4
1
5
4
8
2
.
[
1
2
]
K
.
S
.
R
a
o
e
t
a
l
.
,
“
I
n
t
e
l
l
i
g
e
n
t
u
l
t
r
a
so
u
n
d
i
ma
g
i
n
g
f
o
r
e
n
h
a
n
c
e
d
b
r
e
a
st
c
a
n
c
e
r
d
i
a
g
n
o
si
s
:
En
s
e
mb
l
e
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
st
r
a
t
e
g
i
e
s
,
”
I
EEE
Ac
c
e
ss
,
v
o
l
.
1
2
,
p
p
.
2
2
2
4
3
–
2
2
2
6
3
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
4
.
3
3
5
8
4
4
8
.
[
1
3
]
D
.
N
.
El
s
h
e
a
k
h
,
O
.
M
.
F
a
h
my
,
M
.
F
a
r
o
u
k
,
K
.
E
z
z
a
t
,
a
n
d
A
.
R
.
El
d
a
ma
k
,
“
A
n
e
a
r
l
y
b
r
e
a
st
c
a
n
c
e
r
d
e
t
e
c
t
i
o
n
b
y
u
si
n
g
w
e
a
r
a
b
l
e
f
l
e
x
i
b
l
e
se
n
s
o
r
s a
n
d
a
r
t
i
f
i
c
i
a
l
i
n
t
e
l
l
i
g
e
n
t
,
”
I
E
EE
A
c
c
e
ss
,
v
o
l
.
1
2
,
p
p
.
4
8
5
1
1
–
4
8
5
2
9
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
4
.
3
3
8
0
4
5
3
.
[
1
4
]
S
.
S
.
K
o
s
h
y
a
n
d
L.
J.
A
n
b
a
r
a
s
i
,
“
LM
H
i
st
N
e
t
:
Le
v
e
n
b
e
r
g
–
M
a
r
q
u
a
r
d
t
b
a
se
d
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
f
o
r
c
l
a
ssi
f
i
c
a
t
i
o
n
o
f
b
r
e
a
st
c
a
n
c
e
r
h
i
s
t
o
p
a
t
h
o
l
o
g
i
c
a
l
i
ma
g
e
s
,
”
I
E
EE
Ac
c
e
ss
,
v
o
l
.
1
2
,
p
p
.
5
2
0
5
1
–
5
2
0
6
6
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
4
.
3
3
8
5
0
1
1
.
[
1
5
]
J.
W
a
n
g
e
t
a
l
.
,
“
W
e
a
k
l
y
s
u
p
e
r
v
i
se
d
l
e
si
o
n
d
e
t
e
c
t
i
o
n
a
n
d
d
i
a
g
n
o
si
s
f
o
r
b
r
e
a
s
t
c
a
n
c
e
r
s
w
i
t
h
p
a
r
t
i
a
l
l
y
a
n
n
o
t
a
t
e
d
u
l
t
r
a
so
u
n
d
i
ma
g
e
s
,
”
I
EEE
T
r
a
n
s
a
c
t
i
o
n
s
o
n
M
e
d
i
c
a
l
I
m
a
g
i
n
g
,
v
o
l
.
4
3
,
n
o
.
7
,
p
p
.
2
5
0
9
–
2
5
2
1
,
J
u
l
.
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
T
M
I
.
2
0
2
4
.
3
3
6
6
9
4
0
.
[
1
6
]
J.
A
h
ma
d
,
S
.
A
k
r
a
m
,
A
.
Jaff
a
r
,
M
.
R
a
sh
i
d
,
a
n
d
S
.
M
.
B
h
a
t
t
i
,
“
B
r
e
a
st
c
a
n
c
e
r
d
e
t
e
c
t
i
o
n
u
s
i
n
g
d
e
e
p
l
e
a
r
n
i
n
g
:
A
n
i
n
v
e
s
t
i
g
a
t
i
o
n
u
s
i
n
g
t
h
e
D
D
S
M
d
a
t
a
s
e
t
a
n
d
a
c
u
s
t
o
m
i
z
e
d
A
l
e
x
N
e
t
a
n
d
s
u
p
p
o
r
t
v
e
c
t
o
r
m
a
c
h
i
n
e
,
”
I
EEE
Ac
c
e
ss
,
v
o
l
.
1
1
,
p
p
.
1
0
8
3
8
6
–
1
0
8
3
9
7
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
3
.
3
3
1
1
8
9
2
.
[
1
7
]
S
.
A
z
i
z
,
K
.
M
u
n
i
r
,
A
.
R
a
z
a
,
M
.
S
.
A
l
m
u
t
a
i
r
i
,
a
n
d
S
.
N
a
w
a
z
,
“
I
V
N
e
t
:
T
r
a
n
s
f
e
r
l
e
a
r
n
i
n
g
b
a
s
e
d
d
i
a
g
n
o
s
i
s
o
f
b
r
e
a
s
t
c
a
n
c
e
r
g
r
a
d
i
n
g
u
s
i
n
g
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
A
h
yb
r
id
C
N
N
-
R
N
N
a
p
p
r
o
a
c
h
fo
r
b
r
ea
s
t c
a
n
ce
r
d
etec
tio
n
t
h
r
o
u
g
h
…
(
K
esh
etti S
r
ee
ka
la
)
3093
h
i
s
t
o
p
a
t
h
o
l
o
g
i
c
a
l
i
m
a
g
e
s
o
f
i
n
f
e
c
t
e
d
c
e
l
l
s
,
”
I
E
E
E
A
c
c
e
s
s
,
v
o
l
.
1
1
,
p
p
.
1
2
7
8
8
0
–
1
2
7
8
9
4
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
E
S
S
.
2
0
2
3
.
3
3
3
2
5
4
1
.
[
1
8
]
A
.
G
a
d
e
,
D
.
K
.
D
a
s
h
,
T.
M
.
K
u
m
a
r
i
,
S
.
K
.
G
h
o
s
h
,
R
.
K
.
Tr
i
p
a
t
h
y
,
a
n
d
R
.
B
.
P
a
c
h
o
r
i
,
“
M
u
l
t
i
sca
l
e
a
n
a
l
y
si
s
d
o
mai
n
i
n
t
e
r
p
r
e
t
a
b
l
e
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
f
o
r
d
e
t
e
c
t
i
o
n
o
f
b
r
e
a
s
t
c
a
n
c
e
r
u
si
n
g
t
h
e
r
mo
g
r
a
m
i
ma
g
e
s,
”
I
E
EE
T
ra
n
s
a
c
t
i
o
n
s
o
n
I
n
s
t
ru
m
e
n
t
a
t
i
o
n
a
n
d
Me
a
su
r
e
m
e
n
t
,
v
o
l
.
7
2
,
p
p
.
1
–
1
3
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
TI
M
.
2
0
2
3
.
3
3
1
7
9
1
3
.
[
1
9
]
D
.
A
l
s
a
e
d
i
,
M
.
E
l
B
a
d
a
w
e
,
a
n
d
O
.
M
.
R
a
ma
h
i
,
“
A
b
r
e
a
st
c
a
n
c
e
r
d
e
t
e
c
t
i
o
n
sy
s
t
e
m
u
si
n
g
m
e
t
a
su
r
f
a
c
e
s
w
i
t
h
a
c
o
n
v
o
l
u
t
i
o
n
n
e
u
r
a
l
n
e
t
w
o
r
k
:
A
f
e
a
s
i
b
i
l
i
t
y
st
u
d
y
,
”
I
EEE
T
ra
n
s
a
c
t
i
o
n
s
o
n
M
i
c
r
o
w
a
v
e
T
h
e
o
r
y
a
n
d
T
e
c
h
n
i
q
u
e
s
,
v
o
l
.
7
0
,
n
o
.
7
,
p
p
.
3
5
6
6
–
3
5
7
6
,
J
u
l
.
2
0
2
2
,
d
o
i
:
1
0
.
1
1
0
9
/
TM
TT.
2
0
2
2
.
3
1
6
8
3
1
2
.
[
2
0
]
K
.
R
.
S
e
k
h
a
r
,
G
.
R
.
L.
M
.
T
a
y
a
r
u
,
A
.
K
.
C
h
a
k
r
a
v
a
r
t
h
y
,
B
.
G
o
p
i
r
a
j
u
,
A
.
La
k
sh
ma
n
a
r
a
o
,
a
n
d
T.
V
.
S
.
K
r
i
sh
n
a
,
“
A
n
e
f
f
i
c
i
e
n
t
l
u
n
g
c
a
n
c
e
r
d
e
t
e
c
t
i
o
n
m
o
d
e
l
u
s
i
n
g
c
o
n
v
n
e
t
s
a
n
d
r
e
si
d
u
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
s,
”
i
n
2
0
2
4
F
o
u
r
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
A
d
v
a
n
c
e
s
i
n
El
e
c
t
r
i
c
a
l
,
C
o
m
p
u
t
i
n
g
,
C
o
m
m
u
n
i
c
a
t
i
o
n
a
n
d
S
u
s
t
a
i
n
a
b
l
e
T
e
c
h
n
o
l
o
g
i
e
s
(
I
C
AE
C
T
)
,
Ja
n
.
2
0
2
4
,
p
p
.
1
–
5
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
A
EC
T
6
0
2
0
2
.
2
0
2
4
.
1
0
4
6
9
1
8
7
.
[
2
1
]
V
.
P
a
t
e
l
,
V
.
C
h
a
u
r
a
si
a
,
R
.
M
a
h
a
d
e
v
a
,
a
n
d
S
.
P
.
P
a
t
o
l
e
,
“
G
A
R
L
-
N
e
t
:
g
r
a
p
h
b
a
se
d
a
d
a
p
t
i
v
e
r
e
g
u
l
a
r
i
z
e
d
l
e
a
r
n
i
n
g
d
e
e
p
n
e
t
w
o
r
k
f
o
r
b
r
e
a
s
t
c
a
n
c
e
r
c
l
a
ssi
f
i
c
a
t
i
o
n,
”
I
EE
E
A
c
c
e
ss
,
v
o
l
.
1
1
,
p
p
.
9
0
9
5
–
9
1
1
2
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
3
.
3
2
3
9
6
7
1
.
[
2
2
]
F
.
A
z
o
u
r
a
n
d
A
.
B
o
u
k
e
r
c
h
e
,
“
D
e
s
i
g
n
G
u
i
d
e
l
i
n
e
s f
o
r
m
a
mm
o
g
r
a
m
-
b
a
se
d
c
o
mp
u
t
e
r
-
a
i
d
e
d
s
y
st
e
ms
u
si
n
g
d
e
e
p
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s
,
”
I
EEE
A
c
c
e
ss
,
v
o
l
.
1
0
,
p
p
.
2
1
7
0
1
–
2
1
7
2
6
,
2
0
2
2
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
2
.
3
1
5
1
8
3
0
.
[
2
3
]
Y
.
W
a
n
g
,
L.
Z
h
a
n
g
,
X
.
S
h
u
,
Y
.
F
e
n
g
,
Z
.
Y
i
,
a
n
d
Q
.
L
v
,
“
F
e
a
t
u
r
e
-
se
n
si
t
i
v
e
d
e
e
p
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
f
o
r
m
u
l
t
i
-
i
n
s
t
a
n
c
e
b
r
e
a
s
t
c
a
n
c
e
r
d
e
t
e
c
t
i
o
n
,
”
I
EEE
/
A
C
M
T
ra
n
s
a
c
t
i
o
n
s
o
n
C
o
m
p
u
t
a
t
i
o
n
a
l
Bi
o
l
o
g
y
a
n
d
Bi
o
i
n
f
o
rm
a
t
i
c
s
,
v
o
l
.
1
9
,
n
o
.
4
,
p
p
.
2
2
4
1
–
2
2
5
1
,
Ju
l
.
2
0
2
2
,
d
o
i
:
1
0
.
1
1
0
9
/
TC
B
B
.
2
0
2
1
.
3
0
6
0
1
8
3
.
[
2
4
]
Y
.
L
i
,
J.
W
u
,
a
n
d
Q
.
W
u
,
“
C
l
a
ssi
f
i
c
a
t
i
o
n
o
f
b
r
e
a
s
t
c
a
n
c
e
r
h
i
s
t
o
l
o
g
y
i
m
a
g
e
s
u
s
i
n
g
m
u
l
t
i
-
s
i
z
e
a
n
d
d
i
scri
mi
n
a
t
i
v
e
p
a
t
c
h
e
s
b
a
s
e
d
o
n
d
e
e
p
l
e
a
r
n
i
n
g
,
”
I
EE
E
A
c
c
e
ss
,
v
o
l
.
7
,
p
p
.
2
1
4
0
0
–
2
1
4
0
8
,
2
0
1
9
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
1
9
.
2
8
9
8
0
4
4
.
[
2
5
]
A
.
J.
R
a
o
,
M
.
R
.
B
a
b
u
,
G
.
D
u
r
g
a
R
a
o
,
G
.
P
.
K
u
m
a
r
,
a
n
d
A
.
L
a
k
s
h
m
a
n
a
r
a
o
,
“
I
n
n
o
v
a
t
i
v
e
w
a
y
o
f
i
d
e
n
t
i
f
y
i
n
g
s
k
i
n
c
a
n
c
e
r
m
o
d
e
l
d
e
s
i
g
n
w
i
t
h
F
C
N
N
a
n
d
LSTM
,
”
i
n
2
0
2
4
I
EEE
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
p
u
t
i
n
g
,
P
o
w
e
r
a
n
d
C
o
m
m
u
n
i
c
a
t
i
o
n
T
e
c
h
n
o
l
o
g
i
e
s
(
I
C
2
P
C
T
)
,
F
e
b
.
2
0
2
4
,
p
p
.
2
8
7
–
291
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
2
P
C
T
6
0
0
9
0
.
2
0
2
4
.
1
0
4
8
6
4
1
0
.
[
2
6
]
C.
-
M
.
K
i
m,
R
.
C
.
P
a
r
k
,
a
n
d
E.
J.
H
o
n
g
,
“
B
r
e
a
s
t
mass
c
l
a
ssi
f
i
c
a
t
i
o
n
u
s
i
n
g
e
LFA
a
l
g
o
r
i
t
h
m
b
a
s
e
d
o
n
C
R
N
N
d
e
e
p
l
e
a
r
n
i
n
g
m
o
d
e
l
,
”
I
EEE
A
c
c
e
ss
,
v
o
l
.
8
,
p
p
.
1
9
7
3
1
2
–
1
9
7
3
2
3
,
2
0
2
0
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
0
.
3
0
3
4
9
1
4
.
[
2
7
]
G
.
K
u
mari
,
U
.
V
.
R
a
m
e
s
h
,
P
.
P
a
t
t
a
b
h
i
R
a
mam
o
h
a
n
,
K
.
S
.
P
a
v
a
n
i
,
a
n
d
A
.
L
a
k
sh
ma
n
a
r
a
o
,
“
C
a
n
c
e
r
d
e
t
e
c
t
i
o
n
w
i
t
h
e
n
sem
b
l
e
l
e
a
r
n
i
n
g
mo
d
e
l
f
r
o
m
n
o
v
e
l
p
r
e
c
e
d
e
n
c
e
b
a
se
d
a
l
g
o
r
i
t
h
m
s,
”
i
n
2
0
2
3
6
th
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
C
o
n
t
e
m
p
o
r
a
r
y
C
o
m
p
u
t
i
n
g
a
n
d
I
n
f
o
rm
a
t
i
c
s
(
I
C
3
I
)
,
S
e
p
.
2
0
2
3
,
p
p
.
2
1
8
7
–
2
1
9
0
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
3
I
5
9
1
1
7
.
2
0
2
3
.
1
0
3
9
7
6
2
9
.
[
2
8
]
A
.
S
h
a
h
,
“
B
r
e
a
s
t
u
l
t
r
a
so
u
n
d
i
ma
g
e
s
d
a
t
a
s
e
t
,
”
K
a
g
g
l
e
.
h
t
t
p
s
:
/
/
w
w
w
.
k
a
g
g
l
e
.
c
o
m
/
d
a
t
a
se
t
s/
a
r
y
a
sh
a
h
2
k
/
b
r
e
a
st
-
u
l
t
r
a
so
u
n
d
-
i
m
a
g
e
s
-
d
a
t
a
se
t
(
a
c
c
e
sse
d
J
u
l
.
2
3
,
2
0
2
4
)
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
K
e
shetti
S
r
e
e
k
a
la
c
u
rre
n
tl
y
wo
rk
i
n
g
a
s
a
n
a
ss
istan
t
p
ro
fe
ss
o
r
in
t
h
e
De
p
a
rtme
n
t
o
f
CS
E
a
t
M
a
h
a
tma
G
a
n
d
h
i
In
st
it
u
te
o
f
Tec
h
n
o
l
o
g
y
,
Hy
d
e
ra
b
a
d
,
In
d
ia.
S
h
e
re
c
e
iv
e
d
P
h
.
D
.
d
e
g
re
e
in
CS
E
fro
m
JN
TU,
Hy
d
e
ra
b
a
d
,
Tela
n
g
a
n
a
.
S
h
e
h
a
s
p
u
b
li
sh
e
d
5
0
re
se
a
rc
h
p
a
p
e
rs
i
n
re
fe
re
e
d
in
tern
a
ti
o
n
a
l
jo
u
rn
a
ls,
1
6
re
se
a
rc
h
p
a
p
e
rs
i
n
c
o
n
fe
re
n
c
e
s
a
n
d
4
P
a
ten
ts.
S
h
e
a
ls
o
p
u
b
li
sh
e
d
t
h
re
e
b
o
o
k
s
ti
tl
e
d
d
a
t
a
stru
c
tu
re
s
a
n
d
a
lg
o
rit
h
m
s
u
si
n
g
C,
th
e
e
ra
o
f
a
rti
ficia
l
in
telli
g
e
n
c
e
a
n
d
m
a
c
h
in
e
lea
rn
i
n
g
a
n
d
AI
a
p
p
li
c
a
ti
o
n
s
.
He
r
a
re
a
s
o
f
re
se
a
rc
h
in
c
l
u
d
e
m
a
c
h
i
n
e
lea
rn
in
g
,
a
rti
ficia
l
i
n
telli
g
e
n
c
e
a
n
d
g
rid
c
o
m
p
u
ti
n
g
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
k
sr
ik
a
la_
c
se
@m
g
it
.
a
c
.
in
.
S
r
il
a
th
a
Ya
l
a
m
a
ti
a
ss
istan
t
p
ro
fe
ss
o
r,
De
p
a
rtme
n
t
o
f
C
o
m
p
u
ter
S
c
ien
c
e
a
n
d
En
g
i
n
e
e
rin
g
,
Ko
n
e
ru
Lak
sh
m
a
iah
Ed
u
c
a
ti
o
n
F
o
u
n
d
a
ti
o
n
,
Va
d
d
e
sw
a
ra
m
,
An
d
h
ra
P
ra
d
e
sh
.
S
h
e
is
p
u
rsu
i
n
g
P
h
.
D
.
i
n
G
a
n
d
i
In
stit
u
te
o
f
Tec
h
n
o
l
o
g
y
a
n
d
M
a
n
a
g
e
m
e
n
t,
Visa
k
h
a
p
a
t
n
a
m
.
S
h
e
h
a
s
1
4
Ye
a
rs
o
f
tea
c
h
in
g
e
x
p
e
rien
c
e
a
n
d
3
y
e
a
rs
o
f
in
d
u
str
y
e
x
p
e
rien
c
e
.
S
h
e
p
u
b
l
ish
e
d
p
a
p
e
rs
i
n
re
p
u
ted
n
a
ti
o
n
a
l
a
n
d
i
n
tern
a
ti
o
n
a
l
jo
u
rn
a
ls
.
He
r
a
re
a
s
o
f
i
n
tere
st
i
n
c
lu
d
e
d
a
ta
m
in
in
g
,
m
a
c
h
i
n
e
lea
rn
in
g
,
c
o
m
p
u
ter
n
e
two
rk
s
,
a
n
d
a
rti
ficia
l
i
n
telli
g
e
n
c
e
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
srilath
a
.
y
a
lam
a
ti
@g
m
a
il
.
c
o
m
.
Ann
e
m
n
e
e
d
i
La
k
sh
m
a
n
a
r
a
o
c
u
rre
n
tl
y
wo
r
k
in
g
a
s
a
ss
istan
t
p
ro
fe
ss
o
r
in
Ad
it
y
a
Un
iv
e
rsity
,
S
u
ra
m
p
a
lem
.
He
c
o
m
p
lete
d
h
is
B.
Tec
h
.
in
C
S
IT
a
n
d
M
.
Tec
h
.
i
n
so
f
twa
re
e
n
g
in
e
e
rin
g
.
He
c
o
m
p
lete
d
P
h
.
D.
in
An
d
h
ra
Un
i
v
e
rsity
,
Vish
a
k
h
a
p
a
tn
a
m
.
His
a
re
a
s
o
f
in
tere
st
a
re
m
a
c
h
in
e
lea
rn
in
g
,
c
y
b
e
r
se
c
u
rit
y
,
d
e
e
p
lea
rn
in
g
.
He
is
a
li
fe
m
e
m
b
e
r
o
f
Co
m
p
u
ter
S
o
c
iety
o
f
In
d
ia (CS
I)
a
n
d
I
S
TE
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
la
x
m
a
n
1
2
1
6
@g
m
a
il
.
c
o
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.