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
.
6
,
Decem
b
er
20
25
,
p
p
.
5
3
7
1
~
5
3
7
9
I
SS
N:
2088
-
8
7
0
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijece.
v
15
i
6
.
pp
5
3
7
1
-
5
3
7
9
5371
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
Dete
c
tion o
f
b
reas
t
ca
ncer
wit
h
ens
emble lea
rning
us
ing
ma
g
netic
reso
na
n
ce imag
ing
Swa
t
i N
a
dk
a
rni
1
,
K
ev
in No
r
o
nh
a
2
1
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
,
S
h
a
h
a
n
d
A
n
c
h
o
r
K
u
t
c
h
h
i
E
n
g
i
n
e
e
r
i
n
g
C
o
l
l
e
g
e
,
M
u
m
b
a
i
,
I
n
d
i
a
2
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
o
n
i
c
s a
n
d
T
e
l
e
c
o
mm
u
n
i
c
a
t
i
o
n
E
n
g
i
n
e
e
r
i
n
g
,
S
t
.
F
r
a
n
c
i
s I
n
st
i
t
u
t
e
o
f
T
e
c
h
n
o
l
o
g
y
,
M
u
mb
a
i
,
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
10
,
2
0
2
4
R
ev
is
ed
Au
g
18
,
2
0
2
5
Acc
ep
ted
Sep
15
,
2
0
2
5
De
sp
it
e
n
o
tab
le
p
ro
g
re
ss
in
m
e
d
i
c
in
e
a
lo
n
g
wi
th
tec
h
n
o
lo
g
y
,
t
h
e
d
e
a
th
s
d
u
e
to
b
re
a
st
c
a
n
c
e
r
a
re
in
c
re
a
sin
g
ste
a
d
il
y
.
Th
is
p
a
p
e
r
p
r
o
p
o
se
s
a
fra
m
e
wo
rk
to
a
id
t
h
e
e
a
rly
d
e
tec
ti
o
n
o
f
les
io
n
s
in
b
re
a
st
wit
h
m
a
g
n
e
ti
c
re
so
n
a
n
c
e
ima
g
in
g
(
M
RI).
T
h
e
wo
r
k
h
a
s
b
e
e
n
c
a
rried
o
u
t
u
sin
g
d
iff
u
sio
n
we
ig
h
te
d
ima
g
i
n
g
(DWI)
a
n
d
d
y
n
a
m
ic
c
o
n
tras
t
e
n
h
a
n
c
e
d
-
m
a
g
n
e
ti
c
re
so
n
a
n
c
e
ima
g
in
g
(DCE
-
M
RI).
Da
ta
a
u
g
m
e
n
tatio
n
h
a
s
b
e
e
n
in
c
o
rp
o
ra
ted
t
o
e
n
larg
e
th
e
d
a
ta
se
t
c
o
ll
e
c
ted
fr
o
m
a
re
p
u
ted
h
o
sp
it
a
l.
De
e
p
lea
rn
i
n
g
h
a
s
b
e
e
n
im
p
lem
e
n
ted
u
sin
g
th
e
e
n
se
m
b
le
o
f
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
(CNN
).
Am
o
n
g
st
th
e
in
d
i
v
id
u
a
l
CNN
m
o
d
e
ls,
th
e
y
o
u
o
n
l
y
lo
o
k
o
n
c
e
(YO
LO)
CNN
y
ield
e
d
th
e
h
ig
h
e
st
p
e
rf
o
rm
a
n
c
e
with
a
n
a
c
c
u
ra
c
y
o
f
9
3
.
4
%
,
s
e
n
siti
v
it
y
o
f
9
3
.
4
4
%
,
s
p
e
c
ifi
c
it
y
o
f
9
3
.
3
3
%
,
a
n
d
F
1
-
sc
o
re
o
f
9
3
.
4
4
%
.
Us
in
g
Hu
n
g
a
rian
o
p
ti
m
iza
ti
o
n
,
a
p
p
r
o
p
riate
se
lec
ti
o
n
o
f
in
d
iv
id
u
a
l
CNN
a
rc
h
it
e
c
tu
re
s
to
f
o
rm
th
e
e
n
se
m
b
le
o
f
CNN
wa
s
p
o
ss
ib
le.
T
h
e
e
n
se
m
b
le
m
o
d
e
l
e
n
h
a
n
c
e
d
p
e
rfo
rm
a
n
c
e
with
9
5
.
8
7
%
a
c
c
u
ra
c
y
,
9
5
.
0
8
%
s
e
n
siti
v
it
y
,
9
6
.
6
7
%
s
p
e
c
ifi
c
it
y
,
a
n
d
F
1
-
sc
o
re
o
f
9
5
.
8
7
%
.
K
ey
w
o
r
d
s
:
B
r
ea
s
t
ca
n
ce
r
d
etec
tio
n
C
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
Dee
p
lear
n
in
g
E
n
s
em
b
le
l
ea
r
n
in
g
Hu
n
g
ar
ian
o
p
tim
izatio
n
Ma
g
n
etic
r
eso
n
an
ce
i
m
a
g
in
g
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Swati N
ad
k
ar
n
i
Dep
ar
tm
en
t o
f
I
n
f
o
r
m
atio
n
T
e
ch
n
o
lo
g
y
,
Sh
ah
an
d
A
n
ch
o
r
K
u
tch
h
i E
n
g
in
ee
r
in
g
C
o
lleg
e
Mu
m
b
ai,
I
n
d
ia
E
m
ail:
s
wati.
n
ad
k
ar
n
i@
s
ak
ec
.
ac
.
in
1.
I
NT
RO
D
UCT
I
O
N
T
h
e
n
u
m
b
e
r
o
f
in
d
iv
i
d
u
als
d
etec
ted
with
b
r
ea
s
t
ca
n
ce
r
as
well
as
th
e
n
u
m
b
er
o
f
d
ea
t
h
s
r
esu
ltin
g
f
r
o
m
it
ar
e
o
n
th
e
r
is
e.
I
n
2
0
2
1
,
in
th
e
Un
ited
States
o
f
Am
er
ica
(
USA)
,
th
e
n
u
m
b
er
o
f
b
r
ea
s
t
ca
n
ce
r
p
atien
ts
was
ap
p
r
o
x
im
ately
0
.
4
cr
o
r
es
[
1
]
.
I
n
I
n
d
ia
,
d
etec
tin
g
th
e
d
is
ea
s
e
at
an
ad
v
an
ce
d
s
tag
e
is
o
n
e
o
f
th
e
m
o
s
t
f
r
eq
u
e
n
t
f
ac
to
r
s
co
n
tr
i
b
u
tin
g
to
th
e
h
ig
h
d
ea
th
r
ate,
an
d
it
ca
n
b
e
b
ec
au
s
e
o
f
in
a
d
eq
u
ate
k
n
o
wled
g
e,
u
n
f
in
is
h
ed
tr
ea
tm
e
n
t p
lan
s
,
a
n
d
r
estricte
d
av
ailab
ilit
y
o
f
ef
f
i
cien
t c
ar
e
at
r
u
r
al
ca
n
ce
r
ce
n
ter
s
[
2
]
.
I
n
th
e
in
itial
s
tag
es
o
f
b
r
ea
s
t
ca
n
ce
r
,
th
e
r
e
m
a
y
n
o
t
b
e
an
y
s
y
m
p
to
m
s
o
b
s
er
v
ed
.
Var
ia
b
le
s
h
ap
es,
p
o
s
itio
n
s
,
an
d
d
im
e
n
s
io
n
s
m
a
k
e
th
e
d
etec
tio
n
d
if
f
ic
u
lt.
T
h
i
s
r
esear
ch
s
u
g
g
ests
a
d
ee
p
-
lear
n
in
g
ap
p
r
o
ac
h
f
o
r
th
e
d
etec
tio
n
o
f
b
r
ea
s
t
lesi
o
n
s
u
s
in
g
m
ag
n
etic
r
eso
n
an
ce
im
ag
in
g
(
MRI)
.
Ar
tific
ial
in
tellig
en
ce
h
as
n
o
t
ev
o
lv
ed
to
t
h
e
ex
ten
t
th
at
it
c
an
b
e
in
d
ep
en
d
e
n
tly
u
s
ed
f
o
r
m
ed
ical
co
n
s
id
er
atio
n
s
b
u
t
m
ay
b
e
u
s
ed
to
ass
is
t
m
ed
ical
p
r
ac
titi
o
n
er
s
[
3
]
.
Sy
s
tem
s
u
tili
zin
g
d
ee
p
lear
n
in
g
alg
o
r
ith
m
s
f
o
r
h
av
e
b
ee
n
im
p
lem
en
ted
to
h
el
p
d
o
cto
r
s
.
T
h
e
co
m
p
le
x
ity
o
f
th
e
s
y
s
tem
is
lo
w.
Dee
p
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
et
wo
r
k
(
C
NN
)
ass
is
ts
in
d
etec
tin
g
an
d
ca
teg
o
r
izin
g
lesi
o
n
s
u
s
in
g
im
ag
e
an
aly
s
is
[
4
]
.
Kh
an
et
a
l.
[
5
]
p
r
o
p
o
s
ed
a
m
o
d
el,
wh
er
e
p
r
ed
ic
tio
n
was
d
o
n
e
b
y
f
u
s
in
g
th
e
f
e
atu
r
es
ex
tr
ac
ted
b
y
C
NN
f
r
o
m
f
o
u
r
v
iews
o
f
m
a
m
m
o
g
r
am
s
in
to
o
n
e
la
y
er
an
d
ac
h
iev
ed
an
ac
cu
r
ac
y
o
f
9
2
.
2
9
%.
W
an
g
et
a
l.
[
6
]
f
u
s
ed
th
e
d
e
n
s
ity
f
ea
tu
r
es,
te
x
tu
r
e
f
ea
tu
r
es,
an
d
m
o
r
p
h
o
lo
g
ical
f
ea
tu
r
es
to
b
o
o
s
t
th
e
ac
cu
r
ac
y
o
f
th
e
C
NN
m
o
d
el.
T
o
e
x
c
e
r
p
t
f
e
a
t
u
r
e
s
,
Z
h
a
n
g
e
t
a
l
.
[
7
]
e
m
p
l
o
y
e
d
u
n
i
v
a
r
i
a
t
e
a
n
d
m
u
l
t
i
v
a
r
i
a
t
e
l
o
g
i
s
ti
c
r
e
g
r
e
s
s
i
o
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
.
6
,
Decem
b
e
r
20
25
:
5
3
7
1
-
5
3
7
9
5372
e
v
a
l
u
a
t
i
o
n
s
f
o
r
m
u
l
ti
p
a
r
a
m
e
tr
i
c
b
r
e
as
t
M
R
I
.
P
a
r
e
k
h
e
t
a
l
.
[
8
]
d
e
v
e
l
o
p
e
d
a
s
y
s
t
e
m
u
s
i
n
g
S
SA
E
f
o
r
m
u
l
t
i
p
a
r
a
m
e
t
r
i
c
b
r
e
a
s
t
MR
I
.
Ho
w
e
v
e
r
,
t
h
e
i
r
s
y
s
t
e
m
f
ac
e
d
t
h
e
d
r
a
w
b
a
c
k
t
h
a
t
it
n
e
e
d
e
d
e
x
h
a
u
s
t
i
v
e
c
o
m
p
u
t
a
t
i
o
n
s
.
Gu
llo
et
a
l.
[
9
]
em
p
h
asized
th
at
th
e
tr
ain
in
g
o
f
th
e
d
ee
p
lear
n
in
g
m
o
d
els
o
u
g
h
t
to
b
e
d
o
n
e
o
n
m
ass
iv
e
an
d
d
iv
er
s
if
ied
d
ata
s
ets
to
h
av
e
ac
cu
r
ate
d
etec
tio
n
.
Au
r
n
a
et
a
l.
[
1
0
]
m
er
g
e
d
two
d
atasets
an
d
ac
h
iev
ed
th
e
g
en
er
ality
o
f
th
e
ir
wo
r
k
o
n
MRI
b
r
ain
tu
m
o
r
s
u
s
in
g
C
NN.
Acc
o
r
d
in
g
to
S
aleh
et
a
l.
[
1
1
]
th
e
d
ee
p
r
ec
u
r
r
e
n
t
n
e
u
r
al
n
etwo
r
k
(
R
NN)
th
at
was
o
p
tim
ized
u
s
in
g
th
e
u
n
iv
a
r
iate
tech
n
i
q
u
e
an
d
f
ea
tu
r
es
th
at
wer
e
ch
o
s
en
p
er
f
o
r
m
e
d
th
e
b
est.
Mu
d
u
li
et
a
l.
[
1
2
]
s
tated
th
at
b
y
ch
o
o
s
in
g
t
h
e
s
am
p
les
eq
u
ally
f
r
o
m
ea
c
h
class
,
th
e
p
r
o
b
lem
o
f
im
b
alan
c
e
in
class
es c
an
b
e
av
o
id
ed
.
T
h
e
d
is
s
im
ilar
ity
o
f
th
e
clas
s
if
ier
s
ca
n
b
e
u
tili
ze
d
to
en
h
an
ce
th
e
class
if
icatio
n
p
er
f
o
r
m
an
ce
b
y
em
p
lo
y
in
g
en
s
em
b
le
lear
n
in
g
[
1
3
]
.
T
o
d
escr
ib
e
it
in
s
im
p
l
er
ter
m
s
,
a
n
en
s
em
b
le
m
o
d
el
lead
s
to
p
r
ed
ictio
n
s
th
at
ar
e
m
o
r
e
ac
cu
r
ate
co
m
p
a
r
ed
to
th
o
s
e
o
f
a
s
in
g
le
m
o
d
el
b
y
co
m
b
in
in
g
m
u
ltip
le
o
f
th
em
.
I
n
E
n
s
em
b
le
lear
n
in
g
,
th
e
r
esu
ltan
t p
r
ed
icti
o
n
o
f
th
e
en
s
em
b
le
is
o
b
tain
ed
b
y
u
s
in
g
av
er
ag
in
g
o
f
th
e
p
r
e
d
ictio
n
s
o
f
s
elec
ted
C
NN
m
o
d
els
o
r
b
y
u
s
in
g
v
o
tin
g
.
I
n
v
o
tin
g
,
eith
er
th
e
m
ajo
r
ity
is
co
n
s
id
er
ed
o
r
weig
h
ted
o
u
tp
u
t
is
ev
alu
ate
d
[
1
4
]
.
H
u
n
g
a
r
ian
o
p
tim
izatio
n
en
s
u
r
es
th
e
o
p
tim
al
s
elec
tio
n
of
class
if
ier
s
f
o
r
th
e
en
s
em
b
le
[
1
5
]
.
2.
M
E
T
H
O
D
T
h
is
s
ec
tio
n
d
escr
ib
es
th
e
d
e
tails
o
f
d
ata
ac
q
u
is
itio
n
.
A
b
r
ief
o
v
er
v
iew
o
f
th
e
C
NN
ar
ch
itectu
r
es
u
s
ed
f
o
llo
ws.
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
o
lo
g
y
is
d
is
cu
s
s
ed
later
in
th
is
s
ec
tio
n
.
2
.
1
.
Da
t
a
a
cquis
it
io
n
Sev
er
al
im
ag
in
g
tech
n
iq
u
es
ar
e
av
ailab
le
f
o
r
t
h
e
id
en
tifi
ca
tio
n
o
f
b
r
ea
s
t
ca
n
ce
r
.
MRI
is
a
n
o
n
-
in
v
asiv
e
im
ag
in
g
tech
n
o
lo
g
y
.
Sm
all
lesi
o
n
s
wh
o
s
e
m
ea
s
u
r
e
is
lo
wer
th
an
1
cm
m
ay
n
o
t
n
ec
ess
ar
ily
b
e
d
etec
ted
b
y
b
r
ea
s
t
u
ltra
s
o
u
n
d
[
1
6
]
.
T
h
e
s
en
s
itiv
ity
o
f
c
o
m
b
in
atio
n
s
cr
ee
n
in
g
,
ac
co
r
d
in
g
to
th
e
r
esear
ch
er
s
,
was
9
6
.
2
%
,
co
m
p
ar
ed
to
7
9
.
7
%
with
MRI
an
d
4
8
.
1
%
with
m
am
m
o
g
r
ap
h
y
[
1
7
]
.
M
R
I
m
ay
m
in
im
ize
u
n
n
ec
ess
ar
y
b
io
p
s
ies
[
1
8
]
.
Acc
o
r
d
in
g
to
o
u
r
r
e
v
iew,
th
e
r
is
in
g
tr
en
d
o
f
DC
E
-
MRI
u
tili
za
tio
n
h
as
b
ee
n
o
b
s
er
v
ed
i
n
r
ec
en
t
y
ea
r
s
[
1
9
]
.
A
s
ec
tio
n
o
f
in
f
lated
g
r
a
y
in
ten
s
ity
a
p
p
ea
r
s
i
n
th
e
DC
E
-
MRI
b
ec
au
s
e
th
e
u
n
u
s
u
al
ti
s
s
u
es
ab
s
o
r
b
h
ig
h
er
co
n
tr
ast
ag
en
ts
in
c
o
m
p
ar
is
o
n
with
th
e
u
s
u
al
tis
s
u
es.
Dif
f
u
s
io
n
weig
h
ted
im
ag
in
g
(
DW
I
)
,
wh
en
co
m
b
in
ed
with
d
y
n
am
ic
c
o
n
tr
ast
-
en
h
a
n
c
ed
MRI
(
DC
E
-
MRI)
,
b
o
o
s
ts
th
e
s
y
s
tem
’
s
p
er
f
o
r
m
an
ce
.
MRI
s
ca
n
s
n
am
ely
DC
E
-
MRI
an
d
DW
I
wer
e
ch
o
s
en
as
th
e
in
p
u
t
to
o
u
r
s
y
s
tem
af
ter
an
aly
zin
g
all
th
ese
asp
ec
ts
.
T
h
e
MRI
im
a
g
es
ar
e
av
ailab
le
in
t
h
e
d
i
g
ital
im
ag
in
g
a
n
d
c
o
m
m
u
n
ica
tio
n
s
in
m
e
d
icin
e
(
DI
C
OM
)
f
o
r
m
at.
T
h
e
R
ad
i
An
t
DI
C
OM
Viewe
r
is
d
o
wn
lo
ad
ed
t
o
v
iew
MRI
im
a
g
es.
A
s
p
ec
im
en
o
f
th
e
ca
n
ce
r
o
u
s
an
d
b
e
n
ig
n
im
ag
e
s
is
d
is
p
lay
ed
in
Fig
u
r
e
s
1
(
a)
an
d
1
(
b
)
r
esp
ec
tiv
el
y
.
Data
co
llectio
n
f
o
r
o
u
r
r
esear
ch
h
as b
ee
n
d
o
n
e
f
r
o
m
a
r
ep
u
ted
h
o
s
p
ital,
n
am
el
y
Nan
av
ati
Ma
x
Su
p
er
Sp
ec
ialty
H
o
s
p
ital,
Mu
m
b
ai.
(
a)
(
b
)
Fig
u
r
e
1
.
MRI
im
ag
es
(
a
)
s
am
p
le
o
f
m
alig
n
an
t im
ag
es
an
d
(
b
)
s
am
p
le
o
f
m
alig
n
an
t im
ag
e
s
2
.
2
.
CNN
a
rc
hite
ct
ures
W
ith
th
e
in
cr
ea
s
e
in
r
esear
ch
in
ar
tific
ial
in
tellig
en
ce
a
n
d
b
etter
co
m
p
u
tin
g
f
ac
ilit
ies,
th
e
u
s
e
o
f
C
NN
in
m
ed
ical
im
ag
in
g
h
a
s
in
cr
ea
s
ed
ex
ten
s
iv
ely
.
C
o
n
t
r
ar
y
to
tr
a
d
itio
n
al
m
ac
h
i
n
e
lear
n
in
g
alg
o
r
ith
m
s
r
eq
u
ir
in
g
m
an
u
al
f
ea
tu
r
e
ex
t
r
ac
tio
n
,
C
NN
ca
n
p
er
f
o
r
m
f
ea
tu
r
e
ex
tr
ac
tio
n
.
C
NN
m
o
d
els’
weig
h
t
-
s
h
ar
in
g
ab
ilit
y
an
d
s
p
ar
s
e
co
n
n
ec
tiv
ity
r
ed
u
ce
tr
ai
n
in
g
tim
es a
n
d
co
s
ts
.
T
h
e
p
r
e
-
tr
ain
ed
m
o
d
els
an
d
th
e
k
n
o
wled
g
e
ac
q
u
ir
ed
ca
n
b
e
u
s
ed
in
th
e
p
r
o
ce
s
s
o
f
tr
an
s
f
er
lear
n
in
g
,
th
u
s
elim
in
atin
g
th
e
r
e
q
u
ir
e
m
en
t
f
o
r
en
o
r
m
o
u
s
v
o
l
u
m
e
s
of
d
ata.
T
h
e
C
NN
ar
ch
itectu
r
es
th
at
wer
e
im
p
lem
en
ted
in
o
u
r
r
esear
ch
wer
e
Den
s
eNe
t
-
2
0
1
,
Ma
tC
o
n
v
Net,
VGGN
et,
I
n
ce
p
tio
n
-
V3
,
Alex
Net,
an
d
y
o
u
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
Dete
ctio
n
o
f b
r
ea
s
t c
a
n
ce
r
w
ith
en
s
emb
le
lea
r
n
in
g
u
s
in
g
ma
g
n
etic
r
eso
n
a
n
ce
ima
g
in
g
(
S
w
a
ti N
a
d
ka
r
n
i
)
5373
o
n
ly
lo
o
k
o
n
c
e
(
YOL
O)
C
NN.
T
h
e
ch
o
ice
o
f
C
NN
was
estab
lis
h
ed
o
n
o
u
r
an
aly
s
is
r
esu
lts
in
[
1
5
]
.
T
h
e
s
elec
ted
C
NN
m
o
d
els
wer
e
co
m
p
u
tatio
n
all
y
ef
f
icien
t
a
n
d
s
atis
f
ac
to
r
ily
ex
ec
u
ted
u
s
in
g
o
n
e
g
r
ap
h
ical
p
r
o
ce
s
s
in
g
u
n
it (
GPU
)
.
T
h
is
m
in
im
al
co
m
p
lex
ity
m
ak
es th
e
s
y
s
tem
a
p
o
ten
tial so
lu
tio
n
a
t r
u
r
al
ar
ea
s
.
2
.
2
.
1
.
DenseNet
-
201
E
ac
h
lay
er
’
s
o
u
tp
u
t
in
th
e
De
n
s
eNe
t
m
o
d
el
is
lin
k
e
d
to
th
e
f
o
llo
win
g
la
y
er
to
p
r
o
m
o
te
f
e
atu
r
e
r
e
u
s
e.
T
h
e
Den
s
eNe
t
m
o
d
el
co
m
p
r
is
es
d
if
f
er
en
t
m
o
d
u
les
s
u
ch
as
Den
s
eBl
o
ck
,
co
m
p
o
s
ite
lay
er
,
tr
an
s
itio
n
lay
er
,
an
d
g
r
o
wth
r
ate
.
L
esio
n
s
o
f
d
if
f
e
r
en
t
d
im
en
s
io
n
s
co
u
ld
b
e
p
o
s
s
ib
ly
d
etec
ted
b
y
ex
tr
ac
tio
n
o
f
L
o
w
-
lev
el
a
n
d
in
ter
m
ed
iate
-
lev
el
f
ea
tu
r
es
u
s
in
g
Den
s
eNe
t
C
NN
[
2
0
]
.
Den
s
eNe
t
-
2
0
1
h
as
2
0
1
lay
e
r
s
with
alm
o
s
t
2
0
m
illi
o
n
p
ar
am
eter
s
.
2
.
2
.
2
.
I
ncept
io
n
-
V3
T
h
e
in
ce
p
tio
n
n
et
ar
c
h
itectu
r
e
h
as 4
8
lay
er
s
with
2
4
m
illi
o
n
p
ar
am
eter
s
.
Go
o
g
L
eNe
t is an
o
th
er
n
am
e
f
o
r
I
n
ce
p
tio
n
-
V3
.
An
i
n
ce
p
tio
n
m
o
d
u
le
h
as th
e
p
o
ten
tial to
ex
tr
ac
t m
u
lti
-
lev
el
f
ea
tu
r
es.
T
h
e
I
n
ce
p
tio
n
m
o
d
u
le
led
to
a
d
e
p
letio
n
in
n
etwo
r
k
p
ar
am
eter
s
.
2
.
2
.
3
.
M
a
t
Co
nv
Net
MA
T
L
AB
f
u
n
ctio
n
s
en
ab
le
th
e
b
u
ild
in
g
o
f
C
NN
m
o
d
el
s
wh
ich
h
av
e
g
o
o
d
a
d
ap
tab
i
lity
[
2
1
]
.
Ma
tC
o
n
v
Net
allo
ws
f
ast
p
r
o
t
o
ty
p
in
g
o
f
f
r
esh
C
NN
ar
ch
itectu
r
es.
C
o
m
p
lex
m
o
d
els
m
a
y
b
e
b
u
ilt
o
n
lar
g
e
d
atasets
with
Ma
tC
o
n
v
Net
'
s
ef
f
ec
tiv
e
ex
ec
u
tio
n
o
n
GPU
an
d
ce
n
tr
al
p
r
o
ce
s
s
in
g
u
n
it
(
C
PU
)
.
2
.
2
.
4
.
YO
L
O
I
n
y
o
u
o
n
ly
lo
o
k
o
n
ce
(
YOL
O
)
C
NN,
b
o
u
n
d
in
g
b
o
x
lo
ca
tio
n
s
an
d
class
p
r
o
b
ab
ilit
ies
ar
e
f
o
u
n
d
s
im
u
ltan
eo
u
s
ly
b
y
b
u
ild
in
g
a
s
in
g
le
n
etwo
r
k
t
o
h
a
n
d
le
o
b
ject
r
ec
o
g
n
itio
n
as
a
r
e
g
r
ess
io
n
[
2
2
]
.
T
h
e
ar
ch
itectu
r
e
o
f
YOL
O
-
V2
C
NN
u
s
ed
in
o
u
r
p
r
o
ject
is
illu
s
tr
ated
in
Fig
u
r
e
2
.
As
a
r
esu
lt,
YOL
O
C
NN
ca
n
d
etec
t
o
b
jects
q
u
ick
ly
.
On
e
m
ajo
r
b
en
ef
it
o
f
YOL
O
is
it
s
ca
p
ac
ity
to
g
en
er
alize
ac
r
o
s
s
m
u
ltip
le
im
ag
es.
R
ec
tan
g
u
lar
ar
ea
s
o
f
in
te
r
est
(
R
OI
)
f
o
r
tu
m
o
r
s
wer
e
lab
eled
.
YOL
O
-
V2
u
s
es
a
s
in
g
le
n
e
u
r
al
n
etwo
r
k
p
ass
to
ac
co
m
p
lis
h
d
etec
tio
n
.
YOL
O
-
V2
r
em
o
v
es
th
e
n
ee
d
f
o
r
c
o
m
p
lex
co
m
p
o
n
e
n
ts
lik
e
r
eg
io
n
p
r
o
p
o
s
al
n
etwo
r
k
s
an
d
m
u
lti
-
s
tag
e
tr
ain
in
g
,
s
tr
ea
m
lin
in
g
th
e
ar
ch
itectu
r
e
f
o
r
ea
s
ier
im
p
lem
en
tatio
n
an
d
f
a
s
ter
d
ep
lo
y
m
en
t
i
n
p
r
o
d
u
ctio
n
en
v
ir
o
n
m
en
ts
.
Fig
u
r
e
2
.
YOL
O
-
V2
C
NN
a
r
c
h
itectu
r
e
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
.
6
,
Decem
b
e
r
20
25
:
5
3
7
1
-
5
3
7
9
5374
2
.
2
.
5
.
Alex
Net
Kr
izh
ev
s
k
y
et
a
l.
[
2
3
]
p
r
esen
ted
a
C
NN
m
o
d
el
n
am
ed
Alex
Net,
wh
ich
h
as
eig
h
t
lay
er
s
an
d
is
th
e
f
ir
s
t
C
NN
in
h
is
to
r
y
to
m
ar
k
th
e
tu
r
n
o
f
r
esear
c
h
er
s
’
at
ten
tio
n
b
ac
k
t
o
d
ee
p
lea
r
n
in
g
.
Alex
Net’
s
d
e
ep
ar
ch
itectu
r
e
c
o
m
p
r
is
es
s
ev
er
al
co
n
v
o
lu
tio
n
al,
p
o
o
lin
g
,
a
n
d
f
u
lly
c
o
n
n
ec
ted
lay
er
s
a
n
d
h
as
6
0
m
illi
o
n
p
ar
am
eter
s
.
T
h
e
a
r
ch
itectu
r
e
e
n
ab
les it to
lear
n
co
m
p
lex
f
ea
tu
r
es.
2
.
2
.
6
.
VG
G
-
19
T
h
e
v
is
u
al
g
r
a
p
h
ics
g
r
o
u
p
(
V
GG)
d
ev
elo
p
e
d
th
e
VGGN
et
m
o
d
el.
An
in
c
r
ea
s
e
in
th
e
d
e
p
th
o
f
th
e
n
etwo
r
k
was
in
tr
o
d
u
ce
d
to
im
p
r
o
v
e
ac
cu
r
ac
y
.
I
n
o
u
r
r
esear
c
h
,
VGG
-
1
9
is
im
p
lem
en
ted
wh
ich
h
as
1
9
lay
er
s
.
T
h
e
in
cr
ea
s
in
g
ly
wid
esp
r
ea
d
u
s
e
o
f
VGG
-
1
9
ca
n
b
e
attr
ib
u
te
d
to
its
s
tr
o
n
g
f
ea
t
u
r
e
ex
tr
ac
tio
n
ca
p
ab
ilit
ies.
2
.
3
.
M
et
ho
do
lo
g
y
Fig
u
r
e
3
r
ep
r
esen
ts
th
e
p
r
o
ce
s
s
f
lo
w
d
iag
r
am
o
f
o
u
r
p
r
o
p
o
s
e
d
ap
p
r
o
a
ch
.
T
h
e
b
lo
ck
s
in
o
u
r
p
r
o
p
o
s
ed
m
o
d
el
in
clu
d
e
d
ata
ac
q
u
is
itio
n
,
d
ata
p
r
o
ce
s
s
in
g
,
en
s
em
b
le
lear
n
in
g
,
an
d
o
u
t
p
u
t
.
Sin
ce
d
ata
ac
q
u
is
itio
n
h
as
b
ee
n
d
is
cu
s
s
ed
p
r
ev
i
o
u
s
ly
,
th
e
r
est
o
f
th
e
b
l
o
ck
s
ar
e
d
is
cu
s
s
ed
in
th
e
n
ex
t
s
u
b
s
ec
tio
n
s
.
A
b
r
ief
d
escr
ip
tio
n
ab
o
u
t th
e
h
ar
d
war
e
an
d
s
o
f
twar
e
r
eq
u
ir
e
m
en
ts
is
p
r
esen
ted
.
Fig
u
r
e
3
.
Pro
ce
s
s
f
lo
w
d
iag
r
a
m
o
f
o
u
r
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
2
.
3
.
1
.
Da
t
a
pro
ce
s
s
ing
3
D
C
NN
m
o
d
els
wer
e
n
o
t
im
p
lem
en
ted
co
n
s
id
er
in
g
th
eir
h
ig
h
co
m
p
u
tatio
n
al
co
m
p
lex
ity
.
2
D
C
NN
ar
ch
itectu
r
es
ca
n
n
o
t b
e
u
tili
ze
d
d
ir
ec
tly
with
MRI
in
p
u
ts
as th
e
d
ata
is
th
r
ee
-
d
im
e
n
s
io
n
al.
Sin
ce
th
e
MRI
d
ata
is
in
th
e
3
D
f
o
r
m
at,
th
e
3
D
MRI
d
ata
was
m
an
u
ally
s
lice
d
to
o
b
tain
2
D
MRI
im
ag
es.
T
h
e
f
ir
s
t
f
ew
an
d
last
f
ew
f
r
am
es
wer
e
n
o
t
in
clu
d
ed
in
th
e
s
am
p
lin
g
,
as
th
er
e
was
less
in
f
o
r
m
atio
n
i
n
th
o
s
e
f
r
a
m
es.
T
h
e
s
am
p
lin
g
was d
o
n
e
r
an
d
o
m
ly
.
T
h
e
s
am
p
lin
g
was m
an
u
ally
d
o
n
e
to
ca
p
tu
r
e
th
e
f
r
am
es h
av
i
n
g
r
elev
a
n
t in
f
o
r
m
atio
n
.
T
h
e
in
p
u
t
s
izes
m
ay
d
i
f
f
er
d
ep
e
n
d
i
n
g
o
n
th
e
C
NN
m
o
d
el.
I
m
ag
es
ca
n
b
e
ea
s
ily
r
esized
t
o
a
s
q
u
ar
e
s
h
ap
e
to
en
s
u
r
e
th
at
th
eir
wid
th
a
n
d
h
eig
h
t
ar
e
th
e
s
am
e,
wh
ich
m
a
k
es
th
e
p
r
o
ce
s
s
o
f
h
an
d
lin
g
d
ata
ea
s
ier
.
T
h
e
MRI
in
p
u
t
s
am
p
les
wer
e
r
escaled
to
m
atch
th
e
d
im
en
s
io
n
s
o
f
th
e
C
NN
in
p
u
t.
Mo
d
els
m
ay
a
d
ap
t
b
ette
r
g
en
er
aliza
tio
n
to
n
ew
d
ata
u
s
in
g
d
ata
au
g
m
en
ta
tio
n
.
Data
au
g
m
en
tatio
n
is
s
ig
n
if
ican
t
wh
en
t
h
er
e
is
a
lack
o
f
g
r
o
u
n
d
-
tr
u
th
d
ata
o
r
wh
en
g
ath
e
r
in
g
n
atu
r
al
d
a
ta
is
tim
e
-
co
n
s
u
m
in
g
o
r
co
s
tly
.
Ov
er
f
itti
n
g
is
lo
wer
ed
b
y
d
ata
au
g
m
en
tatio
n
.
E
x
p
an
s
io
n
o
f
th
e
m
a
g
n
itu
d
e
o
f
th
e
d
ataset,
d
ata
au
g
m
en
t
atio
n
u
s
in
g
v
a
r
io
u
s
o
p
e
r
atio
n
s
lik
e
f
lip
p
in
g
an
d
r
o
tatio
n
is
in
co
r
p
o
r
ated
.
T
h
e
d
ata
au
g
m
en
tatio
n
in
cr
ea
s
ed
th
e
m
ag
n
itu
d
e
o
f
th
e
d
ata
b
y
eig
h
t f
o
ld
s
.
2
.
3
.
2
.
H
a
rdwa
re
a
nd
s
o
f
t
wa
r
e
re
qu
irem
ent
s
T
h
e
im
p
lem
e
n
tatio
n
o
f
o
u
r
p
r
o
ject
is
d
o
n
e
u
s
in
g
a
n
I
n
tel
C
o
r
e
i7
Pro
ce
s
s
o
r
,
R
AM
o
f
8
GB
,
Nv
id
ia
GeFo
r
ce
GT
X
1
6
5
0
T
i
GPU
an
d
1
T
B
h
ar
d
d
is
k
.
An
o
p
t
im
al
en
v
ir
o
n
m
e
n
t
en
ab
lin
g
d
ee
p
lear
n
in
g
u
s
in
g
tr
ain
in
g
m
o
d
els
an
d
d
ep
lo
y
m
en
t
is
o
f
f
er
ed
b
y
MA
T
L
AB
.
Ou
r
p
r
o
je
ct
was
e
x
e
cu
te
d
u
s
i
n
g
t
h
e
d
e
e
p
l
ea
r
n
i
n
g
to
o
l
b
o
x
f
r
o
m
MA
T
L
AB
R
2
0
2
2
a
i
n
s
t
all
ed
o
n
W
i
n
d
o
ws
1
1
.
Dee
p
le
ar
n
i
n
g
t
o
o
lb
o
x
e
n
a
b
l
es
t
h
e
i
m
p
le
m
e
n
t
ati
o
n
o
f
p
r
e
-
tr
ai
n
e
d
C
NN
m
o
d
els
.
2
.
3
.
3
.
E
ns
em
ble
m
o
del o
f
C
NN
T
h
e
en
s
em
b
le
m
o
d
el
o
f
th
r
ee
C
NN
cla
s
s
if
ier
s
wa
s
im
p
lem
en
ted
to
b
o
o
s
t
p
er
f
o
r
m
an
ce
.
T
h
e
s
tep
s
in
v
o
lv
ed
i
n
d
ev
elo
p
in
g
e
n
s
em
b
le
lear
n
in
g
a
r
e
illu
s
tr
ated
in
Fig
u
r
e
4
.
I
n
d
e
p
en
d
e
n
t
C
NN
m
o
d
els
ar
e
tr
ain
ed
f
o
llo
wed
b
y
test
in
g
th
em
.
M
er
g
in
g
th
e
p
r
e
d
ictio
n
s
f
r
o
m
s
ev
er
al
C
NN
in
cr
ea
s
es
th
e
o
v
er
all
ac
cu
r
ac
y
b
y
d
ec
r
ea
s
in
g
th
e
p
o
s
s
ib
ilit
y
o
f
er
r
o
r
s
.
E
n
s
em
b
les
p
er
f
o
r
m
b
ett
er
o
n
u
n
s
ee
n
o
r
n
o
is
y
d
ata
b
e
ca
u
s
e
th
ey
ar
e
less
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
Dete
ctio
n
o
f b
r
ea
s
t c
a
n
ce
r
w
ith
en
s
emb
le
lea
r
n
in
g
u
s
in
g
ma
g
n
etic
r
eso
n
a
n
ce
ima
g
in
g
(
S
w
a
ti N
a
d
ka
r
n
i
)
5375
p
r
o
n
e
to
o
v
er
f
it
th
e
tr
ain
in
g
s
et.
E
n
s
em
b
les
av
er
ag
e
o
u
t
th
e
er
r
o
r
s
m
ad
e
b
y
in
d
iv
i
d
u
al
C
NN
m
o
d
els,
p
r
o
d
u
cin
g
m
o
r
e
r
eliab
le
p
r
ed
i
ctio
n
s
.
T
h
e
m
ajo
r
it
y
v
o
tin
g
al
g
o
r
ith
m
h
as
b
ee
n
in
c
o
r
p
o
r
ated
i
n
to
en
s
em
b
le
lear
n
in
g
.
T
h
is
m
et
h
o
d
is
also
ca
lled
h
ar
d
v
o
tin
g
an
d
is
r
o
b
u
s
t
d
esp
ite
b
ein
g
s
tr
aig
h
tf
o
r
w
ar
d
.
W
ith
ea
ch
b
ase
class
if
ier
m
o
d
el,
a
n
in
p
u
t
is
g
iv
en
,
an
d
an
o
u
tp
u
t
class
lab
el
is
p
r
ed
icte
d
.
T
h
e
m
o
s
t
f
r
e
q
u
en
t
ca
teg
o
r
y
lab
el
am
o
n
g
i
n
d
iv
id
u
al
p
r
ed
ictio
n
s
d
ec
id
es
th
e
f
in
al
e
n
s
em
b
le
p
r
e
d
ictio
n
.
Ass
ig
n
m
en
t
p
r
o
b
lem
s
will
alwa
y
s
h
av
e
an
id
ea
l
s
o
l
u
tio
n
in
r
esp
o
n
s
e
to
th
e
Hu
n
g
ar
ian
alg
o
r
ith
m
.
Acc
o
r
d
in
g
to
th
e
n
atu
r
e
o
f
th
e
p
r
o
b
lem
,
th
e
o
b
jectiv
e
f
u
n
ctio
n
v
alu
e
o
r
to
tal
co
s
t
is
eith
er
m
ax
im
ized
o
r
m
in
im
iz
ed
.
Hu
n
g
a
r
ian
o
p
tim
izatio
n
i
s
u
s
ed
to
ch
o
o
s
e
th
e
C
NN
m
o
d
els
h
av
in
g
th
e
h
ig
h
est
p
e
r
f
o
r
m
an
ce
to
f
o
r
m
th
e
en
s
em
b
le
.
T
h
e
f
o
u
n
d
ati
o
n
o
f
th
e
Hu
n
g
ar
ia
n
m
eth
o
d
is
th
e
id
ea
th
at
th
e
o
p
tim
u
m
s
o
lu
tio
n
to
th
e
r
esu
ltan
t
ass
ig
n
m
en
t
p
r
o
b
lem
is
th
e
s
am
e
as
th
e
p
r
o
b
lem
its
elf
an
d
v
ice
v
er
s
a
if
a
co
n
s
tan
t
is
ad
d
ed
to
ea
ch
ele
m
en
t
o
f
a
r
o
w
a
n
d
co
l
u
m
n
o
f
th
e
co
s
t
m
atr
ix
.
Ma
jo
r
ity
v
o
ti
n
g
is
in
co
r
p
o
r
ated
to
y
ield
th
e
f
in
al
o
u
tp
u
t.
Fig
u
r
e
4
.
Sch
em
atic
d
iag
r
am
f
o
r
en
s
em
b
le
lear
n
in
g
2
.
3
.
4
.
O
utput
b
lo
ck
Du
al
class
if
icatio
n
an
d
m
u
lti
p
le
-
class
class
if
icatio
n
ar
e
t
wo
f
u
n
d
am
e
n
tal
ty
p
es
o
f
cla
s
s
if
icatio
n
.
B
ased
o
n
a
g
iv
e
n
co
llectio
n
o
f
lab
eled
ex
am
p
les,
a
b
in
ar
y
c
lass
if
ier
is
a
f
o
r
m
o
f
class
if
icatio
n
tech
n
i
q
u
e
th
at
esti
m
ates
b
in
ar
y
lab
els
(
e.
g
.
,
-
1
o
r
1
)
f
o
r
an
y
n
ewly
u
n
s
ee
n
ex
am
p
les.
I
t
b
u
ild
s
a
class
if
ie
r
th
at
g
iv
es
a
n
ew
d
ata
p
o
in
t
o
n
e
o
f
two
p
o
s
s
ib
le
lab
els.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
e
s
elec
tio
n
o
f
h
y
p
er
p
ar
a
m
e
ter
s
is
a
cr
u
cial
s
tep
in
d
esig
n
in
g
d
ee
p
lea
r
n
in
g
m
o
d
els.
T
o
im
p
r
o
v
e
tr
ain
in
g
,
b
atc
h
n
o
r
m
aliza
tio
n
ca
n
r
ed
u
ce
a
m
o
d
el’
s
s
en
s
itiv
ity
to
its
s
tar
tin
g
weig
h
ts
.
A
b
atch
s
ize
o
f
th
ir
ty
-
two
was
u
s
ed
to
im
p
lem
e
n
t
b
atch
n
o
r
m
aliza
tio
n
.
T
h
e
tr
ai
n
in
g
-
test
in
g
r
atio
was
s
et
at
7
0
:3
0
.
At
0
.
0
0
1
,
th
e
lear
n
in
g
r
ate
was k
e
p
t c
o
n
s
tan
t.
T
o
f
o
r
ce
th
e
r
em
ain
in
g
n
eu
r
o
n
s
in
a
la
y
er
to
lear
n
m
o
r
e
r
o
b
u
s
t
f
ea
tu
r
es,
d
r
o
p
o
u
t
o
p
er
ates
b
y
r
an
d
o
m
l
y
elim
in
atin
g
a
p
o
r
ti
o
n
o
f
th
e
n
eu
r
o
n
s
in
th
e
lay
er
d
u
r
in
g
tr
ain
in
g
.
As
a
r
esu
lt,
th
er
e
is
les
s
co
-
d
ep
e
n
d
en
c
y
b
etwe
en
n
eu
r
o
n
s
an
d
n
o
ch
an
ce
o
f
o
n
e
o
f
th
em
u
p
d
atin
g
th
e
er
r
o
r
s
o
f
an
o
th
e
r
.
Dr
o
p
o
u
t
lo
wer
s
o
v
er
f
itti
n
g
an
d
en
h
an
c
es
g
en
er
aliza
tio
n
a
b
ilit
ies.
T
h
is
h
elp
s
u
s
g
et
b
etter
r
esu
lts
o
n
d
atasets
th
at
h
av
e
n
o
t b
ee
n
s
ee
n
b
ef
o
r
e.
T
h
e
d
r
o
p
o
u
t f
ac
to
r
was m
ain
tain
ed
at
0
.
5
.
T
h
e
ex
p
er
im
en
tatio
n
was
d
o
n
e
u
s
in
g
1
0
ep
o
c
h
s
an
d
ap
p
ly
in
g
th
e
B
ac
k
p
r
o
p
ag
atio
n
alg
o
r
ith
m
.
Ad
ap
tiv
e
m
o
m
e
n
t
esti
m
atio
n
(
ADAM
)
,
a
v
ar
ian
t
o
f
Sto
c
h
asti
c
g
r
ad
ien
t
d
escen
t,
was
u
s
ed
f
o
r
p
a
r
am
eter
u
p
d
atin
g
in
tr
ain
in
g
.
ADAM
u
s
es les
s
m
em
o
r
y
an
d
is
a
co
m
p
u
tatio
n
ally
ef
f
icien
t o
p
tim
ize
r
.
T
h
e
in
d
iv
id
u
al
C
NN
ar
ch
itectu
r
es
wer
e
m
o
d
el
f
itted
with
th
e
MRI
d
ataset,
h
av
in
g
1
2
0
b
en
ig
n
s
am
p
les
an
d
1
2
2
m
alig
n
a
n
t
s
a
m
p
les.
T
h
e
p
er
f
o
r
m
a
n
ce
ev
alu
atio
n
was
d
o
n
e
b
ased
o
n
th
e
tr
u
e
n
e
g
ativ
es
,
tr
u
e
p
o
s
itiv
es,
f
alse
p
o
s
itiv
es,
an
d
f
alse
n
eg
ativ
es
v
alu
es
th
at
w
er
e
co
m
p
u
ted
f
r
o
m
th
e
co
n
f
u
s
io
n
m
atr
ices
.
T
h
e
p
er
f
o
r
m
an
ce
ev
alu
atio
n
was
d
o
n
e
u
s
in
g
t
h
e
m
etr
ics
n
am
el
y
s
en
s
itiv
ity
,
ac
cu
r
ac
y
,
s
p
ec
if
icity
,
an
d
F1
-
s
co
r
e.
T
h
e
clin
ical
s
ig
n
if
ican
ce
o
f
th
e
m
etr
ics is
d
is
cu
s
s
ed
b
elo
w.
Sen
s
itiv
ity
is
a
s
y
s
tem
’
s
ab
ilit
y
to
c
o
r
r
ec
tly
i
d
en
tify
p
atien
ts
with
a
d
is
ea
s
e.
T
h
e
ter
m
Sp
ec
if
icity
is
a
s
y
s
tem
’
s
ab
ilit
y
to
co
r
r
ec
tly
s
o
r
t
p
er
s
o
n
s
with
o
u
t
illn
ess
.
Fals
e
n
eg
ativ
es
an
d
f
alse
p
o
s
itiv
es
ar
e
v
iewe
d
s
im
ilar
ly
in
ac
cu
r
ac
y
.
Misclass
if
y
in
g
s
o
m
e
ca
s
es
m
ay
h
av
e
v
ar
iab
le
im
p
licatio
n
s
b
ased
o
n
th
e
n
atu
r
e
o
f
th
e
s
itu
atio
n
.
Par
ticu
lar
ly
in
th
e
d
iag
n
o
s
is
o
f
a
d
is
ea
s
e,
a
f
alse
n
eg
ativ
e
m
ig
h
t
b
e
m
o
r
e
d
an
g
er
o
u
s
in
co
m
p
ar
is
o
n
with
a
f
alse
p
o
s
itiv
e.
Acc
u
r
ac
y
ca
n
b
e
d
ec
ep
tiv
e
in
s
itu
atio
n
s
wh
en
o
n
e
class
ex
ce
ed
s
th
e
o
th
er
s
.
A
m
o
d
el
co
u
ld
,
f
o
r
in
s
tan
ce
,
p
r
ed
ict
ju
s
t
th
e
m
ajo
r
ity
class
an
d
s
till
h
av
e
a
h
ig
h
ac
cu
r
ac
y
r
ati
n
g
.
Ho
wev
er
,
th
e
F1
-
s
co
r
e
will
ac
cu
r
ately
r
ep
r
esen
t
th
e
m
o
d
el’
s
p
er
f
o
r
m
a
n
ce
in
ev
er
y
class
.
T
h
e
m
o
d
el’
s
ca
p
ac
ity
to
ac
cu
r
ately
id
en
tify
p
o
s
itiv
e
p
atien
ts
wh
ile
m
in
im
izin
g
f
als
e
p
o
s
itiv
es
an
d
f
alse
n
eg
ativ
es
is
ev
alu
ated
f
air
ly
b
y
th
e
F1
-
s
co
r
e.
Hu
n
g
ar
ian
o
p
tim
izatio
n
to
ch
o
o
s
e
b
est
C
NN
m
o
d
els
Selecte
d
class
if
ier
s
to
f
o
r
m
e
n
s
em
b
le
Ma
jo
r
ity
v
o
tin
g
T
r
ain
in
g
/Te
s
tin
g
I
n
d
ep
e
n
d
en
t
C
NN
Mo
d
els
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
.
6
,
Decem
b
e
r
20
25
:
5
3
7
1
-
5
3
7
9
5376
T
h
is
r
esear
ch
is
d
o
n
e
to
h
el
p
r
ad
io
lo
g
is
ts
in
m
ak
in
g
f
aster
d
ec
is
io
n
s
.
T
h
is
s
y
s
tem
is
p
r
o
p
o
s
ed
to
aid
th
e
p
atien
ts
esp
ec
ially
in
r
u
r
al
ar
ea
s
wh
er
e
th
er
e
is
a
s
ca
r
c
ity
o
f
ex
p
er
t
r
ad
io
lo
g
is
ts
.
I
t
f
ac
ilit
ates
d
ec
i
s
io
n
-
m
ak
in
g
an
d
o
f
f
er
s
p
r
ec
is
e
in
s
ig
h
ts
,
wh
ich
lo
wer
s
e
x
p
en
s
e
s
an
d
d
if
f
icu
lties
.
T
h
e
p
e
r
f
o
r
m
an
ce
o
f
th
e
C
NN
m
o
d
els is
s
u
m
m
ar
ized
in
T
a
b
le
1
.
T
h
e
p
er
f
o
r
m
an
ce
m
etr
ics
co
n
s
id
er
ed
f
o
r
th
e
H
u
n
g
a
r
ian
o
p
tim
izatio
n
wer
e
s
en
s
itiv
ity
,
s
p
ec
if
icity
,
an
d
F1
-
Sco
r
e
.
All
th
e
elem
e
n
ts
wer
e
n
o
r
m
alize
d
b
etwe
en
0
an
d
1
a
n
d
later
n
eg
ated
to
m
ax
im
ize
th
e
to
tal
co
s
t.
T
ab
le
2
r
e
p
r
esen
ts
th
e
o
p
tim
al
co
s
t
m
atr
ix
r
esu
ltin
g
f
r
o
m
Hu
n
g
ar
ian
o
p
tim
izatio
n
.
T
h
e
o
p
tim
al
v
alu
e
o
b
tain
ed
is
2
.
9
3
4
4
.
Acc
o
r
d
in
g
to
th
e
o
p
tim
izatio
n
r
esu
lts
,
th
e
in
d
iv
id
u
al
C
NN
m
o
d
els
th
at
o
u
tp
er
f
o
r
m
th
e
o
th
er
C
NN
m
o
d
els
ar
e
YOL
O,
Ma
tC
o
n
v
Net,
an
d
Alex
Net,
wh
ich
ar
e
ch
o
s
en
to
f
o
r
m
th
e
en
s
em
b
le.
E
n
s
em
b
le
lear
n
in
g
em
p
lo
y
s
m
ajo
r
ity
v
o
ti
n
g
.
T
h
e
s
im
u
latio
n
r
esu
lts
co
m
p
r
is
i
n
g
o
f
co
n
f
u
s
io
n
m
atr
ix
a
n
d
r
ec
eiv
er
o
p
er
atin
g
ch
ar
ac
ter
is
tic
(
R
OC
)
cu
r
v
e
ar
e
s
h
o
wn
in
Fig
u
r
es
5
(
a)
an
d
5
(
b
)
r
esp
ec
tiv
ely
.
T
h
e
p
er
f
o
r
m
an
ce
m
etr
ics
o
f
th
e
p
r
o
p
o
s
ed
en
s
em
b
le
s
y
s
tem
ar
e
s
h
o
wn
in
T
a
b
le
3
.
An
im
p
a
ct
an
aly
s
is
is
d
o
n
e
to
ch
e
ck
t
h
e
im
p
r
o
v
em
en
t
i
n
p
er
f
o
r
m
an
ce
d
u
e
to
th
e
ap
p
licatio
n
o
f
th
e
Hu
n
g
ar
ia
n
o
p
tim
izatio
n
to
f
o
r
m
th
e
en
s
em
b
le.
T
h
e
m
ea
n
p
er
f
o
r
m
an
ce
m
etr
ics
with
th
e
v
ar
io
u
s
C
NN
m
o
d
els
a
r
e
ev
al
u
ated
an
d
co
m
p
ar
e
d
with
th
e
p
er
f
o
r
m
an
ce
m
etr
ic
ac
h
iev
ed
b
y
th
e
en
s
em
b
le
m
o
d
el.
T
ab
le
3
illu
s
tr
ates
t
h
e
p
er
f
o
r
m
an
ce
im
p
r
o
v
em
e
n
t.
T
h
e
p
er
f
o
r
m
an
ce
ac
h
iev
ed
b
y
in
tr
o
d
u
cin
g
th
e
en
s
em
b
le
is
f
ar
s
u
p
er
i
o
r
t
o
t
h
at
o
f
in
d
i
v
id
u
al
C
NN
m
o
d
e
ls
.
C
o
n
s
id
er
in
g
th
e
p
er
ce
n
tag
e
im
p
r
o
v
em
en
t
in
t
h
e
ac
cu
r
ac
y
m
etr
ic,
s
in
ce
th
e
p
-
v
alu
e
is
m
u
c
h
less
th
an
0
.
0
5
,
t
h
e
im
p
r
o
v
em
e
n
t
is
s
tatis
t
ically
s
ig
n
if
ican
t
at
th
e
9
5
%
co
n
f
id
e
n
ce
lev
el.
T
h
e
co
n
f
id
en
ce
in
te
r
v
als
d
o
n
o
t
co
in
cid
e,
wh
ich
f
u
r
th
e
r
s
u
p
p
o
r
ts
th
at
th
e
d
if
f
er
e
n
ce
is
s
tatis
t
ically
m
ea
n
in
g
f
u
l.
T
ab
le
1
.
T
h
e
p
er
f
o
r
m
a
n
ce
o
f
C
NN
m
o
d
els
C
N
N
mo
d
e
l
S
e
n
s
i
t
i
v
i
t
y
S
p
e
c
i
f
i
c
i
t
y
F1
-
S
c
o
r
e
A
c
c
u
r
a
c
y
D
e
n
seN
e
t
-
2
0
1
0
.
9
5
0
8
0
.
7
6
6
6
0
.
8
7
2
0
0
.
8
6
0
0
I
n
c
e
p
t
i
o
n
-
V3
0
.
9
3
4
4
0
.
8
6
6
6
0
.
9
0
4
8
0
.
9
0
1
0
M
a
t
C
o
n
v
N
e
t
1
.
0
0
0
0
0
.
8
3
3
3
0
.
9
2
3
1
0
.
9
1
7
0
VGG
-
19
0
.
7
2
1
3
0
.
9
8
3
3
0
.
8
3
0
2
0
.
8
5
1
0
A
l
e
x
N
e
t
0
.
7
2
1
3
1
.
0
0
0
0
0
.
8
3
8
0
0
.
8
6
0
0
Y
O
LO
0
.
9
3
4
4
0
.
9
3
3
3
0
.
9
3
4
4
0
.
9
3
4
0
T
ab
le
2
.
T
h
e
o
p
tim
al
c
o
s
t m
atr
ix
C
N
N
M
o
d
e
l
S
e
n
s
i
t
i
v
i
t
y
S
p
e
c
i
f
i
c
i
t
y
F1
-
S
c
o
r
e
D
e
n
seN
e
t
-
2
0
1
0
.
9
5
0
8
0
.
7
6
6
6
0
.
8
7
2
0
I
n
c
e
p
t
i
o
n
-
V3
0
.
9
3
4
4
0
.
8
6
6
6
0
.
9
0
4
8
M
a
t
C
o
n
v
N
e
t
1
.
0
0
0
0
0
.
8
3
3
3
0
.
9
2
3
1
VGG
-
19
0
.
7
2
1
3
0
.
9
8
3
3
0
.
8
3
0
2
A
l
e
x
N
e
t
0
.
7
2
1
3
1
.
0
0
0
0
0
.
8
3
8
0
Y
O
LO
0
.
9
3
4
4
0
.
9
3
3
3
0
.
9
3
4
4
(
a)
(
b
)
Fig
u
r
e
5
.
T
h
e
s
im
u
latio
n
r
esu
l
ts
(
a)
co
n
f
u
s
io
n
m
atr
ix
a
n
d
(
b
)
th
e
R
OC
c
u
r
v
e
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
Dete
ctio
n
o
f b
r
ea
s
t c
a
n
ce
r
w
ith
en
s
emb
le
lea
r
n
in
g
u
s
in
g
ma
g
n
etic
r
eso
n
a
n
ce
ima
g
in
g
(
S
w
a
ti N
a
d
ka
r
n
i
)
5377
T
ab
le
3
.
T
h
e
p
er
f
o
r
m
a
n
ce
im
p
r
o
v
em
en
t a
n
aly
s
is
o
f
th
e
p
r
o
p
o
s
ed
en
s
em
b
le
s
y
s
tem
P
e
r
f
o
r
ma
n
c
e
m
e
t
r
i
c
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
(
%
)
F1
-
s
c
o
r
e
(
%
)
I
n
d
i
v
i
d
u
a
l
C
N
N
a
v
e
r
a
g
e
8
8
.
7
2
8
7
.
7
0
8
9
.
7
2
8
8
.
3
8
P
r
o
p
o
se
d
e
n
s
e
m
b
l
e
m
o
d
e
l
9
5
.
8
7
9
5
.
0
8
9
6
.
6
7
9
5
.
8
7
I
mp
r
o
v
e
m
e
n
t
7
.
1
5
7
.
3
8
6
.
9
5
7
.
4
9
T
h
e
p
r
o
p
o
s
ed
e
n
s
em
b
le
s
y
s
tem
is
co
n
tr
asted
with
o
th
er
ex
i
s
tin
g
s
y
s
tem
s
with
r
eg
a
r
d
s
o
f
ac
cu
r
ac
y
,
as
s
h
o
wn
in
T
ab
le
4
.
T
h
e
r
ese
ar
ch
wo
r
k
d
o
n
e
o
n
ly
o
n
MRI
d
ata
was
co
n
s
id
er
ed
.
T
h
e
m
et
h
o
d
s
im
p
lem
en
ted
an
d
th
e
d
ataset
u
s
ed
ar
e
s
p
ec
i
f
ied
.
T
h
e
d
ata
p
r
esen
ted
in
th
e
tab
les
s
h
o
ws
th
at
th
e
p
r
o
p
o
s
ed
en
s
em
b
le
s
y
s
tem
s
u
r
p
ass
ed
th
e
ex
is
tin
g
m
o
d
els,
p
r
o
p
o
s
ed
b
y
o
th
er
r
esear
ch
er
s
.
T
ab
le
4
.
C
o
m
p
a
r
is
o
n
o
f
th
e
e
n
s
em
b
le
s
y
s
tem
with
ex
is
tin
g
s
y
s
tem
s
R
e
f
e
r
e
n
c
e
A
c
c
u
r
a
c
y
(
%
)
M
e
t
h
o
d
D
a
t
a
s
e
t
Y
e
a
r
S
u
n
e
t
a
l
.
[
2
4
]
9
2
.
3
7
C
r
o
ss
-
a
t
t
e
n
t
i
o
n
m
u
l
t
i
-
b
r
a
n
c
h
n
e
t
D
C
E
-
M
R
I
2
0
2
3
A
b
d
u
l
l
a
h
e
t
a
l
.
[
2
5
]
9
0
.
0
0
C
N
N
M
u
l
t
i
p
a
r
a
m
e
t
r
i
c
M
R
I
2
0
2
5
Zh
e
n
g
e
t
a
l
.
[
2
6
]
8
2
.
4
0
P
C
M
M
–
N
e
t
M
u
l
t
i
p
a
r
a
m
e
t
r
i
c
M
R
I
2
0
2
4
A
sh
f
a
q
e
t
a
l
.
[
2
7
]
9
0
.
5
7
C
N
N
M
R
I
2
0
2
2
P
r
o
p
o
se
d
e
n
s
e
m
b
l
e
m
o
d
e
l
9
5
.
8
7
En
se
mb
l
e
o
f
C
N
N
D
C
E
-
M
R
I
,
D
W
I
2
0
2
5
4.
CO
NCLU
SI
O
N
T
h
e
p
r
o
p
o
s
ed
co
m
p
u
ter
-
ai
d
ed
d
esig
n
tec
h
n
iq
u
e
u
tili
ze
s
th
e
m
u
ltip
ar
am
etr
ic
MRI
d
ata
b
y
ex
p
lo
itin
g
th
e
ad
d
itio
n
al
in
f
o
r
m
atio
n
p
r
e
s
en
ted
b
y
DC
E
-
MRI
an
d
D
W
I
-
MRI
f
o
r
th
e
b
r
ea
s
t
tu
m
o
r
s
d
etec
tio
n
.
T
h
e
d
ata
was
g
ath
er
ed
f
r
o
m
Nan
a
v
ati
Ma
x
Su
p
er
Sp
ec
iality
Ho
s
p
ita
l.
Dr
o
p
o
u
t
a
n
d
d
ata
a
u
g
m
en
ta
tio
n
wer
e
em
p
lo
y
ed
to
av
o
id
o
v
er
f
itti
n
g
.
E
x
p
e
r
i
m
en
tatio
n
was
d
o
n
e
i
n
d
ep
e
n
d
en
tly
u
s
in
g
C
NN.
T
h
e
ac
c
u
r
ac
ies
ac
h
iev
ed
b
y
Den
s
eNe
t
-
201,
I
n
ce
p
tio
n
-
V3
,
Ma
tC
o
n
v
Net,
VGG
-
1
9
,
Alex
Net
an
d
YOL
O
C
NN
ar
e
0
.
8
6
,
0
.
9
0
1
,
0
.
9
1
7
,
0
.
8
5
1
,
0
.
8
6
,
an
d
0
.
9
3
4
r
esp
ec
t
iv
ely
.
T
h
e
YOL
O
C
NN
co
u
ld
attain
th
e
t
o
p
m
o
s
t
ac
c
u
r
ac
y
o
f
9
3
.
4
%
co
m
p
ar
ed
with
th
e
o
th
er
C
NN
m
o
d
els.
Fo
llo
win
g
th
e
e
x
p
er
im
e
n
tatio
n
with
i
n
d
iv
id
u
al
C
NN
m
o
d
els
,
th
e
r
esu
lts
wer
e
o
p
tim
i
ze
d
u
s
in
g
Hu
n
g
ar
ian
o
p
tim
izatio
n
f
o
r
c
h
o
o
s
in
g
t
h
e
b
est
C
NN
m
o
d
e
ls
to
f
o
r
m
th
e
e
n
s
em
b
le.
T
h
e
e
n
s
em
b
le
s
y
s
tem
s
ig
n
if
ican
tly
o
u
tp
er
f
o
r
m
e
d
th
e
in
d
iv
id
u
al
C
NN
m
o
d
els
a
n
d
o
th
er
ex
is
tin
g
s
y
s
tem
s
w
ith
an
ac
cu
r
ac
y
o
f
9
5
.
8
7
%.
T
h
e
p
er
ce
n
ta
g
e
im
p
r
o
v
em
en
t
i
n
th
e
Acc
u
r
ac
y
m
et
r
ic
with
th
e
e
n
s
em
b
le
m
o
d
el
o
v
er
th
e
av
er
a
g
e
o
f
in
d
iv
id
u
al
C
NN
m
o
d
els is
7
.
1
5
%,
wh
ich
is
s
tatis
tical
ly
s
ig
n
if
ican
t.
5.
L
I
M
I
T
AT
I
O
NS A
N
D
F
UT
U
RE
WO
RK
A
lim
itatio
n
is
th
at
th
e
s
am
p
lin
g
was
m
an
u
ally
d
o
n
e
to
ca
p
tu
r
e
th
e
f
r
am
es.
Sam
p
lin
g
ca
n
b
e
au
to
m
ated
.
I
n
ca
s
e
co
m
p
u
tati
o
n
al
r
eso
u
r
ce
s
ar
e
av
aila
b
le,
3
D
C
NN
m
o
d
els
ca
n
b
e
im
p
lem
en
ted
in
f
u
t
u
r
e.
Fu
tu
r
e
wo
r
k
co
u
ld
in
v
o
lv
e
d
etec
tio
n
an
d
class
if
icatio
n
b
y
f
etch
in
g
d
ata
f
r
o
m
m
u
ltip
le
h
o
s
p
itals
.
T
h
e
d
ata
f
r
o
m
v
ar
io
u
s
m
o
d
alities
lik
e
m
am
m
o
g
r
a
m
s
alo
n
g
with
MRI
ca
n
b
e
co
n
s
id
er
ed
.
Als
o
,
o
th
er
s
eq
u
en
ce
s
lik
e
T1
-
weig
h
ted
,
an
d
T
2
-
weig
h
te
d
MRI
s
eq
u
en
ce
s
m
ay
b
e
ap
p
l
ied
.
Sin
ce
,
C
NN
m
o
d
els
r
ely
o
n
v
ast
am
o
u
n
ts
o
f
d
ata,
th
er
e
is
a
wid
e
s
co
p
e
f
o
r
r
esear
ch
e
r
s
to
co
llect
th
e
d
ata,
clea
n
an
d
lab
el
it.
Fu
r
th
er
m
ac
h
in
e
lear
n
in
g
alg
o
r
ith
m
s
co
u
ld
b
e
co
m
b
in
e
d
with
C
NN
m
o
d
els.
T
h
e
o
th
er
co
m
p
u
tatio
n
ally
h
u
n
g
r
y
C
NN
m
o
d
els ca
n
b
e
u
s
ed
f
o
r
im
p
r
o
v
is
atio
n
in
p
er
f
o
r
m
a
n
ce
.
T
h
e
m
in
o
r
ity
class
m
ay
b
e
as
s
ig
n
ed
b
iased
weig
h
ts
as
th
er
e
ar
e
im
b
ala
n
ce
d
class
es.
W
o
r
k
ca
n
b
e
d
o
n
e
f
o
r
n
o
is
e
r
em
o
v
al
with
l
o
w
-
q
u
ality
,
n
o
is
y
,
o
r
p
o
o
r
ly
p
r
ep
r
o
ce
s
s
ed
im
ag
es.
T
h
e
in
te
g
r
atio
n
o
f
C
NN
in
to
cu
r
r
en
t
clin
ical
wo
r
k
f
lo
ws
r
e
q
u
ir
es
v
alid
atio
n
,
a
p
p
r
o
v
als
a
n
d
p
h
y
s
ician
tr
ain
in
g
.
AI
d
r
i
v
en
d
iag
n
o
s
is
r
aises
ad
d
itio
n
al
eth
ical
an
d
leg
al
is
s
u
es,
wh
ich
n
ee
d
to
b
e
ad
d
r
ess
ed
to
en
s
u
r
e
th
e
p
r
o
ject
to
b
e
ac
tu
ally
im
p
lem
en
ted
in
t
r
u
e
s
en
s
e.
ACK
NO
WL
E
DG
M
E
N
T
T
h
e
au
th
o
r
s
ar
e
g
r
atef
u
l
to
Dr
.
Dee
p
ak
Patk
ar
,
an
d
D
r
.
Mitu
s
h
a
Ver
m
a,
f
r
o
m
Nan
a
v
ati
Su
p
er
Sp
ec
iality
Ho
s
p
ital,
f
o
r
th
e
in
v
alu
ab
le
h
elp
p
r
o
v
id
ed
f
o
r
d
ata
co
llectio
n
an
d
an
n
o
tatio
n
.
T
h
e
au
th
o
r
s
ar
e
in
d
eb
ted
to
th
e
eth
ical
an
d
s
cien
tific
r
ev
iew
co
m
m
ittee
to
p
r
o
v
id
e
an
o
p
p
o
r
tu
n
ity
to
wo
r
k
in
th
is
ar
ea
an
d
ap
p
r
o
v
e
th
e
r
esear
c
h
wo
r
k
.
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
.
6
,
Decem
b
e
r
20
25
:
5
3
7
1
-
5
3
7
9
5378
RE
F
E
R
E
NC
E
S
[
1
]
B
.
S
a
t
h
i
y
a
b
h
a
m
a
e
t
a
l
.
,
“
A
n
o
v
e
l
f
e
a
t
u
r
e
se
l
e
c
t
i
o
n
f
r
a
m
e
w
o
r
k
b
a
s
e
d
o
n
g
r
e
y
w
o
l
f
o
p
t
i
mi
z
e
r
f
o
r
m
a
mm
o
g
r
a
m
i
ma
g
e
a
n
a
l
y
s
i
s,
”
N
e
u
ra
l
C
o
m
p
u
t
i
n
g
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
3
3
,
n
o
.
2
1
,
p
p
.
1
4
5
8
3
–
1
4
6
0
2
,
N
o
v
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
0
7
/
s
0
0
5
2
1
-
0
2
1
-
0
6
0
9
9
-
z.
[
2
]
S
.
P
.
S
o
mas
h
e
k
h
a
r
e
t
a
l
.
,
“
A
ss
o
c
i
a
t
i
o
n
o
f
b
r
e
a
st
s
u
r
g
e
o
n
s
o
f
I
n
d
i
a
(
A
B
S
I
)
p
r
a
c
t
i
c
a
l
c
o
n
se
n
s
u
s
st
a
t
e
me
n
t
,
r
e
c
o
m
men
d
a
t
i
o
n
s,
a
n
d
g
u
i
d
e
l
i
n
e
s
f
o
r
t
h
e
t
r
e
a
t
me
n
t
o
f
b
r
e
a
st
c
a
n
c
e
r
i
n
I
n
d
i
a
2
0
2
1
—
I
n
d
i
a
n
s
o
l
u
t
i
o
n
s
f
o
r
I
n
d
i
a
n
p
r
o
b
l
e
ms
,
”
I
n
d
i
a
n
J
o
u
rn
a
l
o
f
S
u
r
g
e
r
y
,
v
o
l
.
8
4
,
n
o
.
S
3
,
p
p
.
5
7
3
–
5
8
4
,
O
c
t
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
2
2
6
2
-
0
2
1
-
0
3
1
6
0
-
y.
[
3
]
S
.
M
.
R
y
u
e
t
a
l
.
,
“
A
u
t
o
ma
t
e
d
l
a
n
d
m
a
r
k
i
d
e
n
t
i
f
i
c
a
t
i
o
n
f
o
r
d
i
a
g
n
o
si
s
o
f
t
h
e
d
e
f
o
r
mi
t
y
u
s
i
n
g
a
c
a
s
c
a
d
e
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
l
a
t
N
e
t
)
o
n
w
e
i
g
h
t
-
b
e
a
r
i
n
g
l
a
t
e
r
a
l
r
a
d
i
o
g
r
a
p
h
s
o
f
t
h
e
f
o
o
t
,
”
C
o
m
p
u
t
e
rs
i
n
Bi
o
l
o
g
y
a
n
d
M
e
d
i
c
i
n
e
,
v
o
l
.
1
4
8
,
p
.
1
0
5
9
1
4
,
S
e
p
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
m
p
b
i
o
me
d
.
2
0
2
2
.
1
0
5
9
1
4
.
[
4
]
J.
I
q
b
a
l
e
t
a
l
.
,
“
R
e
i
m
a
g
i
n
i
n
g
h
e
a
l
t
h
c
a
r
e
:
u
n
l
e
a
s
h
i
n
g
t
h
e
p
o
w
e
r
o
f
a
r
t
i
f
i
c
i
a
l
i
n
t
e
l
l
i
g
e
n
c
e
i
n
m
e
d
i
c
i
n
e
,
”
C
u
r
e
u
s
,
v
o
l
.
1
5
,
n
o
.
9
,
S
e
p
.
2
0
2
3
,
d
o
i
:
1
0
.
7
7
5
9
/
c
u
r
e
u
s.
4
4
6
5
8
.
[
5
]
H
.
N
a
si
r
K
h
a
n
,
A
.
R
.
S
h
a
h
i
d
,
B
.
R
a
z
a
,
A
.
H
.
D
a
r
,
a
n
d
H
.
A
l
q
u
h
a
y
z
,
“
M
u
l
t
i
-
v
i
e
w
f
e
a
t
u
r
e
f
u
s
i
o
n
b
a
s
e
d
f
o
u
r
v
i
e
w
s
m
o
d
e
l
f
o
r
mammo
g
r
a
m
c
l
a
ss
i
f
i
c
a
t
i
o
n
u
s
i
n
g
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
EE
E
Ac
c
e
ss
,
v
o
l
.
7
,
p
p
.
1
6
5
7
2
4
–
1
6
5
7
3
3
,
2
0
1
9
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
1
9
.
2
9
5
3
3
1
8
.
[
6
]
Z.
W
a
n
g
e
t
a
l
.
,
“
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
e
x
t
r
e
me
l
e
a
r
n
i
n
g
ma
c
h
i
n
e
b
a
sed
o
n
f
e
a
t
u
r
e
f
u
s
i
o
n
W
i
t
h
C
N
N
d
e
e
p
f
e
a
t
u
r
e
s,
”
I
EEE
Ac
c
e
ss
,
v
o
l
.
7
,
p
p
.
1
0
5
1
4
6
–
1
0
5
1
5
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
2
7
9
5
.
[
7
]
J.
Z
h
a
n
g
e
t
a
l
.
,
“
M
u
l
t
i
p
a
r
a
met
r
i
c
M
R
I
-
b
a
sed
r
a
d
i
o
m
i
c
s
n
o
m
o
g
r
a
m
f
o
r
p
r
e
o
p
e
r
a
t
i
v
e
p
r
e
d
i
c
t
i
o
n
o
f
l
y
m
p
h
o
v
a
s
c
u
l
a
r
i
n
v
a
s
i
o
n
a
n
d
c
l
i
n
i
c
a
l
o
u
t
c
o
mes
i
n
p
a
t
i
e
n
t
s
w
i
t
h
b
r
e
a
st
i
n
v
a
si
v
e
d
u
c
t
a
l
c
a
r
c
i
n
o
ma,
”
E
u
r
o
p
e
a
n
Ra
d
i
o
l
o
g
y
,
v
o
l
.
3
2
,
n
o
.
6
,
p
p
.
4
0
7
9
–
4
0
8
9
,
Ju
n
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
0
7
/
s0
0
3
3
0
-
0
2
1
-
0
8
5
0
4
-
6.
[
8
]
V
.
S
.
P
a
r
e
k
h
e
t
a
l
.
,
“
M
u
l
t
i
p
a
r
a
me
t
r
i
c
d
e
e
p
l
e
a
r
n
i
n
g
t
i
ss
u
e
si
g
n
a
t
u
r
e
s
f
o
r
a
r
a
d
i
o
l
o
g
i
c
a
l
b
i
o
m
a
r
k
e
r
o
f
b
r
e
a
s
t
c
a
n
c
e
r
:
p
r
e
l
i
mi
n
a
r
y
r
e
su
l
t
s,
”
M
e
d
i
c
a
l
P
h
y
s
i
c
s
,
v
o
l
.
4
7
,
n
o
.
1
,
p
p
.
7
5
–
8
8
,
Ja
n
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
0
2
/
mp
.
1
3
8
4
9
.
[
9
]
R
.
L
o
G
u
l
l
o
,
S
.
Esk
r
e
i
s
-
W
i
n
k
l
e
r
,
E.
A
.
M
o
r
r
i
s,
a
n
d
K
.
P
i
n
k
e
r
,
“
M
a
c
h
i
n
e
l
e
a
r
n
i
n
g
w
i
t
h
m
u
l
t
i
p
a
r
a
me
t
r
i
c
ma
g
n
e
t
i
c
r
e
so
n
a
n
c
e
i
ma
g
i
n
g
o
f
t
h
e
b
r
e
a
st
f
o
r
e
a
r
l
y
p
r
e
d
i
c
t
i
o
n
o
f
r
e
sp
o
n
se
t
o
n
e
o
a
d
j
u
v
a
n
t
c
h
e
m
o
t
h
e
r
a
p
y
,
”
T
h
e
Br
e
a
s
t
,
v
o
l
.
4
9
,
p
p
.
1
1
5
–
1
2
2
,
F
e
b
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
b
r
e
a
st
.
2
0
1
9
.
1
1
.
0
0
9
.
[
1
0
]
N
.
F
.
A
u
r
n
a
,
M
.
A
.
Y
o
u
su
f
,
K
.
A
.
T
a
h
e
r
,
A
.
K
.
M
.
A
z
a
d
,
a
n
d
M
.
A
.
M
o
n
i
,
“
A
c
l
a
ssi
f
i
c
a
t
i
o
n
o
f
M
R
I
b
r
a
i
n
t
u
mo
r
b
a
s
e
d
o
n
t
w
o
st
a
g
e
f
e
a
t
u
r
e
l
e
v
e
l
e
n
s
e
m
b
l
e
o
f
d
e
e
p
C
N
N
m
o
d
e
l
s,
”
C
o
m
p
u
t
e
r
s
i
n
B
i
o
l
o
g
y
a
n
d
Me
d
i
c
i
n
e
,
v
o
l
.
1
4
6
,
p
.
1
0
5
5
3
9
,
J
u
l
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
mp
b
i
o
m
e
d
.
2
0
2
2
.
1
0
5
5
3
9
.
[
1
1
]
H
.
S
a
l
e
h
,
S
.
F
.
A
b
d
-
e
l
g
h
a
n
y
,
H
.
A
l
y
a
mi
,
a
n
d
W
.
A
l
o
sa
i
mi
,
“
P
r
e
d
i
c
t
i
n
g
b
r
e
a
s
t
c
a
n
c
e
r
b
a
se
d
o
n
o
p
t
i
m
i
z
e
d
d
e
e
p
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
,
”
C
o
m
p
u
t
a
t
i
o
n
a
l
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
N
e
u
ro
s
c
i
e
n
c
e
,
v
o
l
.
2
0
2
2
,
p
p
.
1
–
1
1
,
M
a
r
.
2
0
2
2
,
d
o
i
:
1
0
.
1
1
5
5
/
2
0
2
2
/
1
8
2
0
7
7
7
.
[
1
2
]
D
.
M
u
d
u
l
i
,
R
.
D
a
s
h
,
a
n
d
B
.
M
a
j
h
i
,
“
A
u
t
o
ma
t
e
d
d
i
a
g
n
o
s
i
s
o
f
b
r
e
a
st
c
a
n
c
e
r
u
s
i
n
g
m
u
l
t
i
-
m
o
d
a
l
d
a
t
a
se
t
s:
A
d
e
e
p
c
o
n
v
o
l
u
t
i
o
n
n
e
u
r
a
l
n
e
t
w
o
r
k
b
a
se
d
a
p
p
r
o
a
c
h
,
”
Bi
o
m
e
d
i
c
a
l
S
i
g
n
a
l
Pr
o
c
e
ss
i
n
g
a
n
d
C
o
n
t
r
o
l
,
v
o
l
.
7
1
,
p
.
1
0
2
8
2
5
,
Jan
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
b
s
p
c
.
2
0
2
1
.
1
0
2
8
2
5
.
[
1
3
]
N
.
C
h
o
u
h
a
n
,
A
.
K
h
a
n
,
J.
Z.
S
h
a
h
,
M
.
H
u
ss
n
a
i
n
,
a
n
d
M
.
W
.
K
h
a
n
,
“
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
a
n
d
e
m
o
t
i
o
n
a
l
l
e
a
r
n
i
n
g
b
a
s
e
d
b
r
e
a
st
c
a
n
c
e
r
d
e
t
e
c
t
i
o
n
u
si
n
g
d
i
g
i
t
a
l
ma
mm
o
g
r
a
p
h
y
,
”
C
o
m
p
u
t
e
rs
i
n
Bi
o
l
o
g
y
a
n
d
Me
d
i
c
i
n
e
,
v
o
l
.
1
3
2
,
p
.
1
0
4
3
1
8
,
M
a
y
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
m
p
b
i
o
me
d
.
2
0
2
1
.
1
0
4
3
1
8
.
[
1
4
]
A
.
M
.
A
l
q
u
d
a
h
,
S
.
Q
a
z
a
n
,
a
n
d
I
.
S
.
M
a
s
a
d
,
“
A
r
t
i
f
i
c
i
a
l
i
n
t
e
l
l
i
g
e
n
c
e
f
r
a
mew
o
r
k
f
o
r
e
f
f
i
c
i
e
n
t
d
e
t
e
c
t
i
o
n
a
n
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
o
f
p
n
e
u
mo
n
i
a
u
s
i
n
g
c
h
e
st
r
a
d
i
o
g
r
a
p
h
y
i
mag
e
s,”
J
o
u
r
n
a
l
o
f
M
e
d
i
c
a
l
a
n
d
B
i
o
l
o
g
i
c
a
l
En
g
i
n
e
e
ri
n
g
,
v
o
l
.
4
1
,
p
p
.
5
9
9
–
6
0
9
,
J
u
n
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
0
7
/
s
4
0
8
4
6
-
021
-
0
0
6
3
1
-
1.
[
1
5
]
S
.
N
a
d
k
a
r
n
i
a
n
d
K
.
N
o
r
o
n
h
a
,
“
B
r
e
a
st
c
a
n
c
e
r
d
e
t
e
c
t
i
o
n
u
si
n
g
e
n
s
e
m
b
l
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
s,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
E
l
e
c
t
r
i
c
a
l
a
n
d
C
o
m
p
u
t
e
r
E
n
g
i
n
e
e
r
i
n
g
(
I
J
E
C
E)
,
v
o
l
.
1
4
,
n
o
.
1
,
p
p
.
1
0
4
1
–
1
0
4
7
,
F
e
b
.
2
0
2
4
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
e
c
e
.
v
1
4
i
1
.
p
p
1
0
4
1
-
1
0
4
7
.
[
1
6
]
T.
S
h
a
h
a
n
d
S
.
G
u
r
a
y
a
,
“
B
r
e
a
st
c
a
n
c
e
r
scree
n
i
n
g
p
r
o
g
r
a
ms
:
R
e
v
i
e
w
o
f
m
e
r
i
t
s
,
d
e
mer
i
t
s,
a
n
d
r
e
c
e
n
t
r
e
c
o
mm
e
n
d
a
t
i
o
n
s
p
r
a
c
t
i
c
e
d
a
c
r
o
ss
t
h
e
w
o
r
l
d
,
”
J
o
u
r
n
a
l
o
f
Mi
c
r
o
sc
o
p
y
a
n
d
U
l
t
r
a
st
r
u
c
t
u
re
,
v
o
l
.
5
,
n
o
.
2
,
p
p
.
5
9
–
6
9
,
2
0
1
7
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
j
m
a
u
.
2
0
1
6
.
1
0
.
0
0
2
.
[
1
7
]
J.
H
a
l
l
,
“
M
a
mm
o
g
r
a
p
h
y
a
n
d
b
r
e
a
s
t
M
R
I
:
i
s
i
t
t
i
m
e
t
o
e
v
a
l
u
a
t
e
s
t
r
a
t
e
g
i
e
s
a
s
o
p
p
o
se
d
t
o
m
o
d
a
l
i
t
i
e
s?,
”
J
o
u
r
n
a
l
o
f
t
h
e
Am
e
ri
c
a
n
C
o
l
l
e
g
e
o
f
Ra
d
i
o
l
o
g
y
,
2
0
2
4
.
[
1
8
]
D
.
L
e
i
t
h
n
e
r
e
t
a
l
.
,
“
C
l
i
n
i
c
a
l
r
o
l
e
o
f
b
r
e
a
st
M
R
I
n
o
w
a
n
d
g
o
i
n
g
f
o
r
w
a
r
d
,
”
C
l
i
n
i
c
a
l
R
a
d
i
o
l
o
g
y
,
v
o
l
.
7
3
,
n
o
.
8
,
p
p
.
7
0
0
–
7
1
4
,
A
u
g
.
2
0
1
8
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
r
a
d
.
2
0
1
7
.
1
0
.
0
2
1
.
[
1
9
]
S
.
N
a
d
k
a
r
n
i
a
n
d
K
.
N
o
r
o
n
h
a
,
“
A
r
e
v
i
e
w
o
n
mu
l
t
i
p
a
r
a
m
e
t
r
i
c
ma
g
n
e
t
i
c
r
e
s
o
n
a
n
c
e
i
ma
g
i
n
g
f
o
r
t
h
e
d
e
t
e
c
t
i
o
n
o
f
b
r
e
a
s
t
c
a
n
c
e
r
,
”
i
n
2
0
2
0
I
EEE
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
El
e
c
t
r
o
n
i
c
s,
C
o
m
p
u
t
i
n
g
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
(
C
O
N
E
C
C
T
)
,
I
EEE,
Ju
l
.
2
0
2
0
,
p
p
.
1
–
5
,
d
o
i
:
1
0
.
1
1
0
9
/
C
O
N
EC
C
T5
0
0
6
3
.
2
0
2
0
.
9
1
9
8
3
6
2
.
[
2
0
]
Q
.
H
u
,
H
.
M
.
W
h
i
t
n
e
y
,
a
n
d
M
.
L.
G
i
g
e
r
,
“
A
d
e
e
p
l
e
a
r
n
i
n
g
me
t
h
o
d
o
l
o
g
y
f
o
r
i
m
p
r
o
v
e
d
b
r
e
a
s
t
c
a
n
c
e
r
d
i
a
g
n
o
si
s
u
s
i
n
g
mu
l
t
i
p
a
r
a
me
t
r
i
c
M
R
I
,
”
S
c
i
e
n
t
i
f
i
c
R
e
p
o
rt
s
,
v
o
l
.
1
0
,
n
o
.
1
,
p
.
1
0
5
3
6
,
J
u
n
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
3
8
/
s4
1
5
9
8
-
0
2
0
-
6
7
4
4
1
-
4.
[
2
1
]
M
.
A
.
H
o
ssa
i
n
a
n
d
M
.
S
.
A
l
a
m
S
a
j
i
b
,
“
C
l
a
ssi
f
i
c
a
t
i
o
n
o
f
i
ma
g
e
u
si
n
g
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
(
C
N
N
)
,
”
G
l
o
b
a
l
J
o
u
rn
a
l
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
p
p
.
1
3
–
1
8
,
2
0
1
9
,
d
o
i
:
1
0
.
3
4
2
5
7
/
g
j
c
s
t
d
v
o
l
1
9
i
s
2
p
g
1
3
.
[
2
2
]
M
.
A
.
A
l
-
m
a
s
n
i
e
t
a
l
.
,
“
S
i
mu
l
t
a
n
e
o
u
s
d
e
t
e
c
t
i
o
n
a
n
d
c
l
a
ssi
f
i
c
a
t
i
o
n
o
f
b
r
e
a
st
mass
e
s
i
n
d
i
g
i
t
a
l
m
a
mm
o
g
r
a
ms
v
i
a
a
d
e
e
p
l
e
a
r
n
i
n
g
Y
O
LO
-
b
a
se
d
C
A
D
s
y
st
e
m,
”
C
o
m
p
u
t
e
r
M
e
t
h
o
d
s
a
n
d
P
ro
g
r
a
m
s
i
n
B
i
o
m
e
d
i
c
i
n
e
,
v
o
l
.
1
5
7
,
p
p
.
8
5
–
9
4
,
A
p
r
.
2
0
1
8
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
m
p
b
.
2
0
1
8
.
0
1
.
0
1
7
.
[
2
3
]
A
.
K
r
i
z
h
e
v
s
k
y
,
I
.
S
u
t
s
k
e
v
e
r
,
a
n
d
G
.
E.
H
i
n
t
o
n
,
“
I
mag
e
N
e
t
c
l
a
ssi
f
i
c
a
t
i
o
n
w
i
t
h
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
s
,
”
C
o
m
m
u
n
i
c
a
t
i
o
n
s
o
f
t
h
e
A
C
M
,
v
o
l
.
6
0
,
n
o
.
6
,
p
p
.
8
4
–
9
0
,
M
a
y
2
0
1
7
,
d
o
i
:
1
0
.
1
1
4
5
/
3
0
6
5
3
8
6
.
[
2
4
]
L.
S
u
n
e
t
a
l
.
,
“
C
r
o
ss
-
a
t
t
e
n
t
i
o
n
mu
l
t
i
-
b
r
a
n
c
h
C
N
N
u
s
i
n
g
D
C
E
-
M
R
I
t
o
c
l
a
ssi
f
y
b
r
e
a
s
t
c
a
n
c
e
r
m
o
l
e
c
u
l
a
r
su
b
t
y
p
e
s,”
Fr
o
n
t
i
e
rs
i
n
O
n
c
o
l
o
g
y
,
v
o
l
.
1
3
,
M
a
r
.
2
0
2
3
,
d
o
i
:
1
0
.
3
3
8
9
/
f
o
n
c
.
2
0
2
3
.
1
1
0
7
8
5
0
.
[
2
5
]
K
.
A
.
A
b
d
u
l
l
a
h
,
S
.
M
a
r
z
i
a
l
i
,
M
.
N
a
n
a
a
,
L.
Esc
u
d
e
r
o
S
á
n
c
h
e
z
,
N
.
R
.
P
a
y
n
e
,
a
n
d
F
.
J.
G
i
l
b
e
r
t
,
“
D
e
e
p
l
e
a
r
n
i
n
g
-
b
a
s
e
d
b
r
e
a
s
t
c
a
n
c
e
r
d
i
a
g
n
o
si
s
i
n
b
r
e
a
st
M
R
I
:
sy
s
t
e
m
a
t
i
c
r
e
v
i
e
w
a
n
d
met
a
-
a
n
a
l
y
s
i
s,
”
E
u
r
o
p
e
a
n
Ra
d
i
o
l
o
g
y
,
v
o
l
.
3
5
,
n
o
.
8
,
p
p
.
4
4
7
4
–
4
4
8
9
,
F
e
b
.
2
0
2
5
,
d
o
i
:
1
0
.
1
0
0
7
/
s
0
0
3
3
0
-
025
-
1
1
4
0
6
-
6.
[
2
6
]
H
.
Z
h
e
n
g
,
L.
J
i
a
n
,
L.
L
i
,
W
.
Li
u
,
a
n
d
W
.
C
h
e
n
,
“
P
r
i
o
r
c
l
i
n
i
c
o
-
r
a
d
i
o
l
o
g
i
c
a
l
f
e
a
t
u
r
e
s
i
n
f
o
r
m
e
d
m
u
l
t
i
-
m
o
d
a
l
M
R
i
m
a
g
e
s
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
n
o
v
e
l
d
e
e
p
l
e
a
r
n
i
n
g
f
r
a
mew
o
r
k
f
o
r
p
r
e
d
i
c
t
i
o
n
o
f
l
y
m
p
h
o
v
a
sc
u
l
a
r
i
n
v
a
si
o
n
i
n
b
r
e
a
s
t
c
a
n
c
e
r
,
”
C
a
n
c
e
r
M
e
d
i
c
i
n
e
,
v
o
l
.
1
3
,
n
o
.
3
,
F
e
b
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
0
2
/
c
a
m
4
.
6
9
3
2
.
[
2
7
]
A
.
A
sh
f
a
q
,
Y
.
W
e
n
h
u
i
,
S
.
J
i
n
h
a
i
,
a
n
d
M
.
U
.
N
a
si
r
,
“
B
r
e
a
s
t
c
a
n
c
e
r
d
i
a
g
n
o
s
i
n
g
e
m
p
o
w
e
r
e
d
w
i
t
h
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
,
”
i
n
2
0
2
2
I
EE
E
8
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
p
u
t
e
r
a
n
d
C
o
m
m
u
n
i
c
a
t
i
o
n
s
(
I
C
C
C
)
,
I
EEE,
D
e
c
.
2
0
2
2
,
p
p
.
2
2
9
2
–
2
2
9
7
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
C
5
6
3
2
4
.
2
0
2
2
.
1
0
0
6
5
7
8
7
.
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
Dete
ctio
n
o
f b
r
ea
s
t c
a
n
ce
r
w
ith
en
s
emb
le
lea
r
n
in
g
u
s
in
g
ma
g
n
etic
r
eso
n
a
n
ce
ima
g
in
g
(
S
w
a
ti N
a
d
ka
r
n
i
)
5379
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
S
wa
ti
N
a
d
k
a
r
n
i
h
o
l
d
s
a
P
h
.
D.
d
e
g
re
e
fr
o
m
t
h
e
De
p
a
rtme
n
t
o
f
El
e
c
tro
n
ics
a
n
d
Tele
c
o
m
m
u
n
ica
ti
o
n
En
g
in
e
e
ri
n
g
fro
m
S
t.
F
ra
n
c
is
I
n
stit
u
te
o
f
Tec
h
n
o
lo
g
y
,
M
u
m
b
a
i
Un
iv
e
rsity
.
Cu
rre
n
tl
y
,
s
h
e
is
a
n
a
ss
o
c
iate
p
ro
fe
ss
o
r
a
n
d
h
e
a
d
i
n
De
p
a
rtme
n
t
o
f
I
n
fo
rm
a
ti
o
n
Tec
h
n
o
l
o
g
y
,
S
h
a
h
a
n
d
An
c
h
o
r
Ku
tch
h
i
En
g
in
e
e
rin
g
C
o
ll
e
g
e
,
M
u
m
b
a
i
Un
i
v
e
rsity
a
n
d
h
a
s
b
e
e
n
tea
c
h
in
g
f
o
r
m
o
re
t
h
a
n
2
5
y
e
a
rs.
S
h
e
h
a
s
p
r
o
c
u
re
d
a
re
se
a
rc
h
g
ra
n
t
fro
m
M
u
m
b
a
i
Un
iv
e
rsity
.
S
h
e
h
a
s
se
v
e
ra
l
p
a
p
e
rs
to
h
e
r
c
re
d
i
t.
S
h
e
a
lso
h
a
s
p
u
b
li
sh
e
d
a
p
a
ten
t.
S
h
e
is
in
tere
ste
d
i
n
s
o
ft
c
o
m
p
u
ti
n
g
,
d
ig
it
a
l
sig
n
a
l
p
ro
c
e
ss
in
g
,
a
n
d
d
e
e
p
lea
rn
in
g
.
S
h
e
h
a
s
m
e
m
b
e
rsh
ip
s
in
p
ro
fe
ss
io
n
a
l
so
c
i
e
ti
e
s
li
k
e
th
e
As
so
c
iati
o
n
f
o
r
C
o
m
p
u
ti
n
g
M
a
c
h
in
e
ry
(ACM),
th
e
In
stit
u
te
o
f
El
e
c
tri
c
a
l
a
n
d
El
e
c
tro
n
ics
En
g
in
e
e
rs
(IE
EE
),
a
n
d
th
e
I
n
d
ian
S
o
c
iety
f
o
r
Tec
h
n
ica
l
E
d
u
c
a
ti
o
n
(IS
TE
)
.
S
h
e
is
v
ice
-
c
h
a
ir
o
f
M
u
m
b
a
i
AC
M
p
ro
fe
ss
io
n
a
l
c
h
a
p
ter
a
n
d
fa
c
u
lt
y
s
p
o
n
so
r
o
f
S
AK
EC
-
ACM
stu
d
e
n
t
c
h
a
p
ter
.
S
h
e
wa
s
a
wa
rd
e
d
a
s
“
M
o
st
In
f
lu
e
n
t
ial
P
ro
fe
ss
o
r”
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
sw
a
ti
.
n
a
d
k
a
rn
i@sa
k
e
c
.
a
c
.
in
.
K
e
v
in
No
r
o
n
h
a
h
o
l
d
s
a
P
h
D
i
n
e
lec
tro
n
ics
a
n
d
tele
c
o
m
m
u
n
ica
ti
o
n
fro
m
M
a
n
i
p
a
l
In
stit
u
te
o
f
Tec
h
n
o
lo
g
y
.
He
is
a
p
r
o
fe
ss
o
r
a
n
d
h
e
a
d
i
n
t
h
e
De
p
a
rtme
n
t
o
f
El
e
c
tro
n
ics
a
n
d
Tele
c
o
m
m
u
n
ica
ti
o
n
En
g
in
e
e
rin
g
,
a
t
S
t.
F
ra
n
c
is
I
n
stit
u
te
o
f
Tec
h
n
o
lo
g
y
,
M
u
m
b
a
i
Un
i
v
e
rsity
,
He
h
a
s b
e
e
n
tea
c
h
in
g
fo
r
m
o
re
th
a
n
2
5
y
e
a
rs.
He
wa
s a
lso
h
a
v
i
n
g
t
h
e
re
sp
o
n
si
b
il
i
ty
o
f
De
a
n
o
f
Ac
a
d
e
m
ics
a
t
S
t.
F
ra
n
c
is
In
stit
u
t
e
o
f
Tec
h
n
o
lo
g
y
.
He
h
a
s
p
u
b
li
sh
e
d
se
v
e
ra
l
p
a
p
e
rs
th
a
t
h
a
v
e
m
a
n
y
c
it
a
ti
o
n
s.
He
g
o
t
a
m
in
o
r
r
e
se
a
rc
h
g
ra
n
t
fro
m
M
u
m
b
a
i
Un
i
v
e
rsity
.
He
is
i
n
tere
ste
d
in
m
e
d
ica
l
ima
g
e
p
ro
c
e
ss
in
g
,
c
y
b
e
r
se
c
u
rit
y
,
c
o
m
p
u
ter
n
e
two
r
k
s,
m
icro
p
ro
c
e
ss
o
rs,
a
n
d
n
e
x
t
g
e
n
e
ra
ti
o
n
n
e
two
r
k
s
.
He
h
o
l
d
s
m
e
m
b
e
rsh
ip
s
o
f
se
v
e
ra
l
p
ro
fe
ss
io
n
a
l
so
c
ieties
li
k
e
t
h
e
In
s
ti
t
u
te
o
f
El
e
c
tri
c
a
l
a
n
d
E
lec
tro
n
ics
En
g
i
n
e
e
rs
(IE
EE
),
a
n
d
t
h
e
In
d
ian
S
o
c
iety
f
o
r
Tec
h
n
ica
l
Ed
u
c
a
ti
o
n
(IS
TE
).
He
is a rec
o
g
n
i
z
e
d
P
h
D g
u
i
d
e
.
He
h
a
s g
u
id
e
d
se
v
e
ra
l
P
G
a
n
d
P
h
D st
u
d
e
n
ts.
He
is
a
lso
th
e
sin
g
le
p
o
in
t
o
f
c
o
n
t
a
c
t
(S
P
OC)
o
f
n
a
ti
o
n
a
l
p
ro
g
ra
m
m
e
o
n
tec
h
n
o
l
o
g
y
e
n
h
a
n
c
e
d
lea
rn
in
g
(NPT
EL
).
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
k
e
v
in
n
o
r
o
n
h
a
@s
fit
.
a
c
.
in
.
Evaluation Warning : The document was created with Spire.PDF for Python.