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.
12
,
No
.
1
,
Feb
r
u
ar
y
20
22
,
p
p
.
303
~
310
I
SS
N:
2088
-
8
7
0
8
,
DOI
: 1
0
.
1
1
5
9
1
/
ijece
.
v
12
i
1
.
pp
3
0
3
-
3
1
0
303
J
o
ur
na
l
ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
Deep segm
en
tatio
n of t
h
e liver an
d
the
he
pa
tic
tumo
r
s from
a
bdo
men
tom
o
g
ra
phy
ima
g
es
Ner
m
ee
n E
lm
ena
ba
wy
1
,
M
er
v
a
t
E
l
-
Sed
dek
2
,
H
o
s
s
a
m
E
l
-
Din M
o
u
s
t
a
f
a
1
,
Ahm
ed
E
lna
k
ib
1
1
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
o
n
i
c
s a
n
d
C
o
m
mu
n
i
c
a
t
i
o
n
s
En
g
i
n
e
e
r
i
n
g
,
F
a
c
u
l
t
y
o
f
En
g
i
n
e
e
r
i
n
g
,
M
a
n
so
u
r
a
U
n
i
v
e
r
si
t
y
,
M
a
n
s
o
u
r
a
,
Eg
y
p
t
2
D
e
p
a
r
t
me
n
t
o
f
C
o
mm
u
n
i
c
a
t
i
o
n
s
a
n
d
El
e
c
t
r
o
n
i
c
s
En
g
i
n
e
e
r
i
n
g
,
M
i
sr
H
i
g
h
e
r
I
n
st
i
t
u
t
e
o
f
E
n
g
i
n
e
e
r
i
n
g
a
n
d
T
e
c
h
n
o
l
o
g
y
,
M
a
n
s
o
u
r
a
,
E
g
y
p
t
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Dec
1
,
2
0
2
0
R
ev
is
ed
Au
g
3
,
2
0
2
1
Acc
ep
ted
Au
g
1
4
,
2
0
2
1
A
p
ip
e
li
n
e
d
fra
m
e
wo
rk
is
p
r
o
p
o
se
d
fo
r
a
c
c
u
ra
te,
a
u
t
o
m
a
ted
,
sim
u
lt
a
n
e
o
u
s
se
g
m
e
n
tatio
n
o
f
th
e
li
v
e
r
a
s
w
e
ll
a
s
th
e
h
e
p
a
ti
c
t
u
m
o
rs
fr
o
m
c
o
m
p
u
te
d
to
m
o
g
ra
p
h
y
(CT)
ima
g
e
s.
T
h
e
in
tro
d
u
c
e
d
fra
m
e
wo
rk
c
o
m
p
o
se
d
o
f
th
re
e
p
ip
e
li
n
e
d
le
v
e
ls.
F
irst
,
tw
o
d
iff
e
re
n
t
tran
sfe
r
s
d
e
e
p
c
o
n
v
o
l
u
ti
o
n
a
l
n
e
u
ra
l
n
e
two
rk
s
(CNN
)
a
re
a
p
p
li
e
d
to
g
e
t
h
ig
h
-
lev
e
l
c
o
m
p
a
c
t
fe
a
tu
r
e
s
o
f
CT
ima
g
e
s.
S
e
c
o
n
d
,
a
p
ix
e
l
-
wise
c
las
sifier
is
u
se
d
to
o
b
tain
tw
o
o
u
tp
u
t
-
c
las
sified
m
a
p
s
fo
r
e
a
c
h
CNN
m
o
d
e
l.
F
i
n
a
ll
y
,
a
f
u
sio
n
n
e
u
r
a
l
n
e
two
r
k
(F
NN
)
is
u
se
d
to
in
te
g
ra
te
th
e
tw
o
m
a
p
s.
Ex
p
e
rime
n
tatio
n
s
p
e
rfo
r
m
e
d
o
n
t
h
e
M
ICCAI’2
0
1
7
d
a
tab
a
se
o
f
t
h
e
li
v
e
r
tu
m
o
r
se
g
m
e
n
tati
o
n
(LI
T
S
)
c
h
a
ll
e
n
g
e
,
re
su
lt
i
n
a
d
ice
sim
il
a
rit
y
c
o
e
ffici
e
n
t
(DSC)
o
f
9
3
.
5
%
f
o
r
th
e
se
g
m
e
n
tatio
n
o
f
th
e
li
v
e
r
a
n
d
o
f
7
4
.
4
0
%
fo
r
t
h
e
se
g
m
e
n
tatio
n
o
f
t
h
e
les
io
n
,
u
si
n
g
a
5
-
f
o
ld
c
ro
ss
-
v
a
li
d
a
ti
o
n
sc
h
e
m
e
.
Co
m
p
a
ra
ti
v
e
re
su
lt
s
wit
h
t
h
e
sta
te
-
of
-
th
e
-
a
rt
tec
h
n
iq
u
e
s
o
n
t
h
e
sa
m
e
d
a
ta
s
h
o
w
th
e
c
o
m
p
e
t
in
g
p
e
rfo
rm
a
n
c
e
o
f
th
e
p
ro
p
o
se
d
fra
m
e
wo
rk
fo
r
sim
u
lt
a
n
e
o
u
s li
v
e
r
a
n
d
tu
m
o
r
se
g
m
e
n
tatio
n
.
K
ey
w
o
r
d
s
:
C
o
m
p
u
ted
to
m
o
g
r
a
p
h
y
Dee
p
lear
n
in
g
L
iv
er
Seg
m
en
tatio
n
T
u
m
o
r
s
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Ah
m
ed
E
ln
ak
i
b
Dep
ar
tm
en
t o
f
E
lectr
o
n
ics an
d
C
o
m
m
u
n
icatio
n
s
E
n
g
in
ee
r
in
g
,
Facu
lty
o
f
E
n
g
in
ee
r
i
n
g
Ma
n
s
o
u
r
a
Un
iv
er
s
ity
E
lg
o
m
h
o
u
r
ia
St.,
Ma
n
s
o
u
r
a
3
5
5
1
6
,
E
g
y
p
t
E
m
ail:
n
ak
ib
@
m
an
s
.
ed
u
.
e
g
1.
I
NT
RO
D
UCT
I
O
N
Acc
o
r
d
in
g
to
th
e
W
o
r
ld
Hea
lt
h
Or
g
an
izatio
n
(
W
HO)
,
liv
er
ca
n
ce
r
is
th
e
m
ain
ca
u
s
e
o
f
ca
n
ce
r
d
ea
t
h
s
am
o
n
g
all
t
y
p
es
o
f
ca
n
ce
r
s
.
W
o
r
ld
wid
e,
ar
o
u
n
d
8
0
0
,
0
0
0
ca
s
es
o
f
liv
er
ca
n
ce
r
ar
e
d
iag
n
o
s
ed
ea
c
h
y
ea
r
,
ac
co
u
n
tin
g
f
o
r
ar
o
u
n
d
7
0
0
,
0
0
0
d
ea
th
s
[
1
]
.
I
n
2
0
1
9
,
th
e
A
m
er
ican
C
an
ce
r
So
ciety
(
AC
S)
esti
m
ated
ar
o
u
n
d
4
2
,
0
3
0
n
ew
ca
s
es
f
o
r
p
r
im
ar
y
liv
er
ca
n
ce
r
an
d
in
tr
a
h
ep
atic
b
ile
d
u
ct
ca
n
ce
r
in
th
e
Un
ited
States
,
with
ar
o
u
n
d
3
1
,
7
8
0
d
ea
t
h
s
[
1
]
.
T
h
ese
m
etr
i
cs r
ef
lect
th
e
ep
id
em
ic
in
f
latio
n
o
f
liv
e
r
ca
n
ce
r
.
C
o
m
p
u
ted
to
m
o
g
r
ap
h
y
(
C
T
)
im
ag
in
g
is
u
s
u
ally
u
s
ed
f
o
r
liv
er
s
eg
m
en
tatio
n
an
d
/o
r
li
v
er
ca
n
ce
r
d
etec
tio
n
.
Ho
wev
er
,
m
a
n
u
al
s
eg
m
en
tatio
n
o
f
th
e
liv
e
r
an
d
/o
r
th
e
liv
er
tu
m
o
r
s
f
r
o
m
C
T
im
ag
es c
o
n
s
u
m
es a
lo
t
o
f
tim
e
an
d
s
u
f
f
er
s
f
r
o
m
o
b
s
er
v
er
v
a
r
iab
ilit
y
.
T
h
e
r
ef
o
r
e
,
t
h
e
d
esig
n
o
f
ef
f
i
cien
t
c
o
m
p
u
ter
aid
ed
d
ia
g
n
o
s
tic
(
C
AD)
s
y
s
tem
s
,
to
ass
is
t
th
e
r
ad
io
lo
g
is
ts
f
o
r
liv
er
s
eg
m
en
tat
io
n
an
d
/o
r
liv
e
r
ca
n
ce
r
s
eg
m
e
n
tatio
n
,
is
a
wid
ely
in
v
esti
g
ated
o
p
en
r
esear
ch
p
r
o
b
lem
.
T
h
r
o
u
g
h
o
u
t
liter
atu
r
e,
d
if
f
er
en
t
m
eth
o
d
o
lo
g
ies
h
av
e
b
ee
n
u
tili
ze
d
f
o
r
liv
er
s
eg
m
en
tatio
n
an
d
/o
r
f
o
r
liv
er
ca
n
ce
r
s
eg
m
en
tatio
n
.
T
h
ese
m
eth
o
d
s
ca
n
b
e
ca
teg
o
r
ized
as
t
r
ad
itio
n
al
m
eth
o
d
s
o
r
d
ee
p
lear
n
in
g
m
eth
o
d
s
.
T
r
ad
itio
n
al
ap
p
r
o
ac
h
es
u
s
u
all
y
ex
tr
ac
t
f
ea
tu
r
es,
e.
g
.
,
in
ten
s
i
ty
,
tex
tu
r
e,
s
h
ap
e,
f
r
o
m
liv
e
r
C
T
im
ag
e
s
an
d
u
s
e
a
class
if
ier
b
ased
o
n
th
ese
f
ea
tu
r
es
t
o
p
e
r
f
o
r
m
th
e
s
eg
m
en
tatio
n
p
r
o
ce
s
s
.
On
th
e
o
th
er
h
an
d
,
d
ee
p
lear
n
in
g
m
eth
o
d
s
u
s
u
ally
u
s
e
co
n
v
o
l
u
tio
n
al
n
e
u
r
al
n
etwo
r
k
s
(
C
NN)
th
at
co
n
s
is
t
o
f
a
n
u
m
b
er
o
f
co
n
v
o
l
u
tio
n
al
lay
er
s
f
o
r
ex
tr
ac
ti
n
g
l
o
w
-
lev
el
an
d
h
ig
h
-
lev
el
f
ea
tu
r
es
f
o
r
th
e
liv
er
C
T
im
ag
es
an
d
f
u
lly
c
o
n
n
ec
ted
lay
er
s
t
o
en
co
d
e
a
c
o
m
p
ac
t f
e
atu
r
e
s
et
f
o
r
th
e
s
eg
m
e
n
tatio
n
p
r
o
ce
s
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.
12
,
No
.
1
,
Feb
r
u
ar
y
20
22
:
3
0
3
-
310
304
Fo
r
th
e
task
o
f
liv
er
s
eg
m
en
t
atio
n
,
a
p
r
elim
in
ar
y
s
tep
in
m
an
y
C
AD
s
y
s
tem
s
f
o
r
liv
er
ca
n
c
e
r
[
2
]
a
n
d
l
i
v
e
r
f
i
b
r
o
s
i
s
[
3
]
,
d
i
f
f
e
r
e
n
t
t
r
a
d
i
t
i
o
n
a
l
a
n
d
d
e
e
p
l
e
a
r
n
i
n
g
m
e
t
h
o
d
s
h
a
v
e
b
e
e
n
a
p
p
l
i
e
d
.
F
o
r
e
x
a
m
p
l
e
,
B
a
r
s
t
u
g
a
n
et
a
l.
[
4
]
u
s
ed
a
s
u
p
e
r
-
p
ix
el
lin
ea
r
iter
ativ
e
clu
s
ter
in
g
ap
p
r
o
ac
h
an
d
Ad
aBo
o
s
t
alg
o
r
ith
m
to
s
eg
m
en
t
th
e
liv
er
,
ac
h
iev
in
g
a
DSC
o
f
9
2
.
1
3
%
o
n
1
6
a
b
d
o
m
e
n
C
T
test
im
ag
es.
Mu
th
u
s
wam
y
a
n
d
Ka
n
m
a
n
i
[
5
]
ex
tr
ac
te
d
th
e
liv
er
f
r
o
m
C
T
im
ag
es
b
ased
o
n
in
ten
s
ity
th
r
esh
o
ld
in
g
,
f
u
z
zy
c
-
m
ea
n
s
clu
s
ter
in
g
,
an
d
co
n
n
ec
ted
co
m
p
o
n
e
n
t
an
aly
s
is
.
Fo
r
ex
am
p
le,
C
h
an
g
et
a
l.
[
6
]
s
eg
m
en
ted
t
h
e
tu
m
o
r
s
u
s
in
g
a
r
e
g
io
n
g
r
o
win
g
al
g
o
r
ith
m
.
A
b
in
ar
y
lo
g
is
tic
r
eg
r
ess
io
n
an
aly
s
is
b
ased
o
n
ex
tr
ac
ted
tex
tu
r
e
,
s
h
ap
e,
an
d
k
in
etic
cu
r
v
e
f
ea
tu
r
es
wer
e
f
u
r
th
er
p
er
f
o
r
m
ed
to
class
if
y
th
e
s
eg
m
en
ted
tu
m
o
r
s
(
B
en
ig
n
o
r
Ma
lig
n
an
t)
.
I
n
C
h
leb
u
s
et
a
l.
[
7
]
,
a
m
o
d
if
ied
U
-
ne
[
8
]
ar
ch
itectu
r
e
was
u
s
ed
,
c
o
n
s
is
ts
o
f
f
o
u
r
r
eso
lu
tio
n
lev
els,
f
o
r
liv
er
tu
m
o
r
s
s
eg
m
e
n
tatio
n
.
Yu
an
[
9
]
u
s
ed
a
h
ier
ar
ch
ical
d
ee
p
f
u
lly
co
n
v
o
l
u
tio
n
al
-
d
ec
o
n
v
o
l
u
tio
n
al
n
eu
r
a
l
n
etwo
r
k
s
(
C
DNN)
f
o
r
tu
m
o
r
s
eg
m
en
tatio
n
.
An
in
itial
liv
er
s
eg
m
en
tatio
n
was p
r
o
v
id
e
d
u
s
in
g
a
s
im
p
le
C
DNN
m
o
d
el.
T
h
e
s
eg
m
en
ted
liv
er
r
eg
io
n
was
r
e
f
in
ed
u
s
in
g
an
o
th
e
r
C
DN
N
to
f
in
d
th
e
f
in
al
liv
er
s
eg
m
en
tatio
n
e
n
h
an
ce
d
b
y
h
is
to
g
r
am
eq
u
aliz
atio
n
.
T
h
en
a
t
h
ir
d
C
DNN
i
s
ap
p
lied
f
o
r
tu
m
o
r
s
e
g
m
en
tatio
n
.
B
i
et
a
l.
[
1
0
]
u
s
ed
a
d
ee
p
r
esid
u
al
n
etwo
r
k
s
(
R
esNet)
f
o
r
liv
er
an
d
lesi
o
n
s
s
eg
m
en
tatio
n
.
Gr
u
b
er
et
a
l.
[
1
1
]
a
p
p
lied
,
s
eq
u
e
n
tial
ly
,
two
U
-
n
et
[
8
]
n
etwo
r
k
s
f
o
r
liv
er
an
d
lesi
o
n
s
s
eg
m
en
tatio
n
.
W
an
g
et
a
l.
[
1
2
]
a
3
D
atlas
-
b
ased
m
o
d
el
f
o
r
liv
er
s
eg
m
en
tatio
n
.
Sh
i
et
a
l.
[
1
3
]
u
tili
ze
d
a
d
ef
o
r
m
a
b
le
s
h
ap
e
liv
er
s
eg
m
e
n
tatio
n
m
eth
o
d
.
So
n
g
et
a
l.
[
1
4
]
im
p
lem
en
te
d
a
m
o
d
i
f
ied
U
-
Net
m
o
d
el
f
o
r
li
v
e
r
s
eg
m
en
tatio
n
.
Alth
o
u
g
h
th
e
m
eth
o
d
s
p
r
esen
ted
in
th
e
liter
atu
r
e
ac
h
iev
ed
g
o
o
d
r
esu
lts
,
th
e
ac
cu
r
ac
y
is
s
till
a
n
ee
d
to
b
e
im
p
r
o
v
ed
.
T
h
e
p
r
esen
t
s
tu
d
y
p
r
esen
ts
a
d
ee
p
l
ea
r
n
in
g
s
y
s
tem
f
o
r
s
im
u
ltan
e
o
u
s
liv
er
a
n
d
t
u
m
o
r
s
eg
m
en
tatio
n
u
s
in
g
C
NN
m
o
d
elin
g
.
T
h
e
m
ai
n
co
n
t
r
ib
u
tio
n
s
o
f
th
is
wo
r
k
a
r
e
as f
o
llo
ws:
−
I
n
v
esti
g
atin
g
d
if
f
er
en
t
d
ee
p
lear
n
in
g
ar
c
h
itectu
r
es
f
o
r
liv
er
an
d
tu
m
o
r
s
eg
m
en
tatio
n
(
i.e
.
,
Den
s
en
et
an
d
FC
N
-
Alex
Net)
−
Ap
p
ly
in
g
a
3
D
n
ar
r
o
w
-
b
a
n
d
o
f
th
e
in
p
u
t im
ag
es to
en
h
an
ce
t
h
e
d
ee
p
tr
ai
n
in
g
−
Usi
n
g
a
s
m
ar
t f
u
s
io
n
o
f
two
C
NN
ar
ch
itectu
r
es to
im
p
r
o
v
e
t
h
e
s
eg
m
en
tatio
n
q
u
ality
−
Per
f
o
r
m
an
ce
e
v
alu
atio
n
o
n
th
e
MI
C
C
AI
’
2
0
1
7
ch
allen
g
e
liv
e
r
tu
m
o
r
s
eg
m
en
tatio
n
(
L
I
T
S)
d
atab
ase.
T
h
e
s
tr
u
ctu
r
e
o
f
th
is
p
ap
e
r
is
as
f
o
llo
win
g
.
Sectio
n
2
p
r
esen
ts
th
e
s
u
g
g
ested
s
y
s
tem
f
o
r
s
im
u
ltan
eo
u
s
liv
er
a
n
d
tu
m
o
r
s
eg
m
en
tatio
n
.
Sectio
n
3
s
u
m
m
ar
izes
th
e
p
r
o
p
o
s
ed
s
y
s
tem
r
esu
lts
as
well
as
th
e
co
m
p
ar
ativ
e
r
esu
lts
to
th
e
cu
r
r
en
t state
-
of
-
th
e
-
ar
t te
c
h
n
iq
u
es.
Fin
ally
,
s
ec
tio
n
4
co
n
clu
d
es t
h
e
p
ap
er
.
2.
M
E
T
H
O
DS
T
h
e
p
r
o
p
o
s
ed
f
r
am
ewo
r
k
p
r
o
ce
s
s
es
a
r
aw
im
ag
e
th
r
o
u
g
h
t
h
r
ee
s
tag
es
as
s
h
o
w
n
in
Fig
u
r
e
1
.
First,
f
ea
tu
r
es
ar
e
ex
tr
ac
ted
f
r
o
m
r
aw
im
ag
es,
with
o
u
t
p
r
e
p
r
o
ce
s
s
in
g
s
tep
s
,
b
y
in
v
esti
g
atin
g
two
d
if
f
er
en
t
C
NN
m
o
d
els.
Seco
n
d
,
a
p
ix
el
-
wis
e
class
if
icatio
n
lay
er
is
ap
p
lied
.
Fin
ally
,
a
s
m
ar
t
f
u
s
io
n
o
f
th
e
o
u
tp
u
ts
o
f
th
e
two
C
NN
m
o
d
els
i
s
p
er
f
o
r
m
ed
u
s
in
g
a
n
eu
r
al
n
etwo
r
k
(
NN)
to
p
r
o
v
id
e
th
e
f
i
n
al
s
im
u
ltan
eo
u
s
liv
er
an
d
tu
m
o
r
s
eg
m
en
tatio
n
m
ap
,
co
n
tain
i
n
g
th
r
ee
o
u
tp
u
t l
ab
els:
b
ac
k
g
r
o
u
n
d
(
B
G)
,
liv
er
,
an
d
lesi
o
n
.
Fig
u
r
e
1
.
Pro
p
o
s
ed
f
r
am
ewo
r
k
f
o
r
th
e
liv
er
an
d
lesi
o
n
s
s
eg
m
en
tatio
n
with
th
r
ee
s
tag
es: f
ea
t
u
r
e
ex
tr
ac
tio
n
u
s
in
g
d
ee
p
lea
r
n
in
g
,
class
if
icatio
n
b
ased
o
n
p
ix
el
-
wi
s
e
tech
n
iq
u
e,
an
d
s
m
ar
t f
u
s
io
n
2
.
1
.
F
ea
t
ure
e
x
t
ra
c
t
io
n
Her
ein
,
two
p
r
e
-
tr
ain
e
d
C
NNs
ar
e
u
s
ed
to
g
et
th
e
f
ea
t
u
r
es
o
f
th
e
liv
er
an
d
its
lesi
o
n
s
;
Den
s
en
et
[
1
5
]
an
d
th
e
f
u
lly
co
n
n
ec
ted
n
etwo
r
k
(
FC
N)
u
s
in
g
Alex
n
et
(
FC
N
-
Alex
n
et
[
1
6
]
)
.
T
h
e
Den
s
en
et
m
o
d
el
c
o
n
s
is
ts
o
f
a
d
o
wn
-
s
am
p
lin
g
p
ath
an
d
u
p
-
s
am
p
lin
g
p
ath
.
T
h
e
d
o
wn
-
s
am
p
lin
g
p
at
h
ex
tr
ac
te
d
th
e
s
em
a
n
tic
f
ea
tu
r
es
th
en
th
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
Dee
p
s
eg
men
ta
tio
n
o
f t
h
e
liver
a
n
d
th
e
h
ep
a
tic
tu
mo
r
s
fr
o
m
…
(
N
ermeen
E
lmen
a
b
a
w
y
)
305
up
-
s
am
p
lin
g
p
ath
is
tr
ain
ed
to
r
ec
o
v
e
r
th
e
im
ag
e
r
eso
lu
tio
n
o
f
th
e
in
p
u
t
at
t
h
e
o
u
tp
u
t
o
f
t
h
e
m
o
d
el.
Fig
u
r
e
2
s
h
o
ws th
e
ar
ch
itectu
r
e
o
f
th
e
Den
s
en
et
m
o
d
el.
Fig
u
r
e
2
.
Sch
em
atic
d
iag
r
am
s
f
o
r
th
e
s
tr
u
ct
u
r
es o
f
Den
s
n
et
m
o
d
el
[
1
5
]
FC
N
-
Alex
n
et
co
n
s
is
t
s
o
f
an
en
co
d
er
,
a
d
ec
o
d
er
,
a
n
d
a
p
ix
e
l
-
wis
e
clas
s
if
ier
as
s
h
o
wn
in
Fig
u
r
e
3
.
T
h
e
jo
b
o
f
th
e
en
co
d
er
is
to
ex
tr
ac
t
h
ig
h
-
lev
el
c
o
m
p
ac
t
d
e
ep
lear
n
in
g
f
ea
tu
r
es
f
r
o
m
th
e
ab
d
o
m
en
liv
er
C
T
im
ag
es.
Fo
r
th
e
FC
N
-
Alex
n
et
m
o
d
el,
th
e
s
tag
e
o
f
t
h
e
en
co
d
er
co
n
s
is
ts
o
f
f
iv
e
Alex
n
et’
s
lay
er
s
as
s
h
o
wn
in
T
ab
le
1
,
with
n
o
f
u
lly
c
o
n
n
ec
ted
lay
er
s
as
s
h
o
wn
i
n
Fig
u
r
e
3
.
T
h
e
jo
b
o
f
t
h
e
d
ec
o
d
er
is
to
p
er
f
o
r
m
d
ec
o
n
v
o
lu
tio
n
al
s
tep
s
to
g
et
e
x
tr
ac
ted
f
ea
t
u
r
es
with
th
e
s
am
e
d
im
en
s
io
n
s
as
t
h
e
in
p
u
t
im
a
g
e.
T
o
p
e
r
f
o
r
m
th
e
s
eg
m
en
tatio
n
p
r
o
ce
s
s
,
a
class
i
f
icatio
n
lay
er
b
ased
o
n
p
ix
el
-
wis
e
tech
n
iq
u
e
is
f
u
r
th
er
u
s
ed
,
to
lab
el
ea
ch
p
ix
el
in
th
e
C
T
in
p
u
t im
a
g
e
in
to
o
n
e
o
f
th
r
ee
lab
els:
lesi
o
n
,
liv
er
,
o
r
b
ac
k
g
r
o
u
n
d
.
Fig
u
r
e
3
.
Sch
em
atic
d
iag
r
am
s
f
o
r
th
e
s
tr
u
ct
u
r
es o
f
FC
N
-
Alex
n
et
[
1
6
]
T
h
e
p
r
e
-
tr
ain
ed
Den
s
en
et
an
d
Alex
n
et
ar
e
tr
ain
ed
o
n
th
e
I
m
ag
eNe
t
lar
g
e
-
s
ca
le
v
is
u
al
r
e
co
g
n
itio
n
ch
allen
g
e
2
0
1
2
(
I
L
SVR
C
2
0
1
2
)
d
ataset
th
at
is
co
m
p
o
s
ed
o
f
1
0
m
illi
o
n
tr
ain
in
g
im
ag
es
o
f
s
ize
2
2
4
x
2
2
4
f
r
o
m
m
o
r
e
th
an
1
0
0
0
s
u
b
jects.
T
h
e
Den
s
en
et
n
etwo
r
k
co
m
p
o
s
ed
o
f
co
n
tig
u
o
u
s
d
e
n
s
e
b
lo
ck
s
.
T
h
er
e
ar
e
a
tr
an
s
itio
n
lay
er
s
(
co
n
v
o
lu
tio
n
al
lay
er
s
an
d
av
er
ag
e
p
o
o
lin
g
)
b
etwe
e
n
th
e
c
o
n
tig
u
o
u
s
d
en
s
e
b
lo
c
k
s
with
m
o
r
e
th
an
2
0
m
illi
o
n
p
ar
a
m
eter
s
.
T
h
e
s
ize
o
f
th
e
f
ea
tu
r
e
m
ap
an
d
th
e
d
en
s
e
b
lo
ck
is
s
im
ilar
to
b
e
co
n
ca
ten
ated
ea
s
ily
.
T
h
er
e
ar
e
a
g
lo
b
al
a
v
er
ag
e
p
o
o
lin
g
an
d
a
So
f
tMa
x
class
if
ier
at
th
e
en
d
o
f
th
e
last
d
en
s
e
b
lo
ck
.
Alex
n
et
m
o
d
el
is
co
m
p
o
s
ed
o
f
f
iv
e
co
n
v
o
lu
ti
o
n
al
lay
er
s
an
d
th
r
ee
FC
lay
er
s
with
m
o
r
e
th
an
6
2
.
4
m
illi
o
n
p
ar
a
m
eter
s
.
Mo
r
e
d
etailed
o
f
ea
ch
m
o
d
el
ca
n
b
e
f
o
u
n
d
in
[
1
5
]
,
[
1
6
]
,
r
esp
ec
ti
v
ely
.
W
e
ap
p
lied
th
e
two
m
o
d
els
(
Den
s
en
et
an
d
FC
N
-
Alex
n
et)
in
th
e
p
r
o
p
o
s
ed
s
y
s
tem
,
s
in
ce
th
eir
d
ec
o
d
er
s
p
r
o
d
u
ce
o
u
t
p
u
ts
th
at
ar
e
o
f
t
h
e
s
am
e
d
im
en
s
io
n
s
as
th
e
in
p
u
t
im
ag
e
,
wh
ich
s
u
its
th
e
task
o
f
s
eg
m
en
tati
o
n
.
I
n
a
d
d
itio
n
,
th
e
y
h
a
v
e
s
h
o
wn
o
u
ts
tan
d
in
g
p
er
f
o
r
m
an
ce
f
o
r
s
ev
er
al
r
ela
ted
m
ed
ical
a
p
p
licatio
n
s
,
s
u
ch
as
lu
n
g
s
eg
m
e
n
tatio
n
[
1
7
]
,
[
1
8
]
,
p
u
lm
o
n
ar
y
ca
n
ce
r
o
u
s
d
etec
tio
n
[
1
9
]
,
f
ac
e
r
ec
o
g
n
itio
n
[
2
0
]
,
b
r
ain
ca
n
ce
r
[
2
1
]
a
n
d
d
ia
b
etic
r
etin
o
p
at
h
y
[
2
2
]
.
2
.
2
.
Cla
s
s
if
ica
t
io
n
A
p
ix
el
-
wis
e
class
if
ier
is
ap
p
lied
af
ter
ea
ch
m
o
d
el’
s
d
ec
o
d
er
to
lab
el
th
e
s
eg
m
en
ted
o
u
tp
u
t
im
ag
e.
T
h
e
p
ix
el
-
wis
e
class
if
ier
is
co
m
p
o
s
ed
o
f
two
lay
er
s
:
a
So
f
t
Ma
x
lay
er
an
d
a
weig
h
te
d
lay
er
to
p
er
f
o
r
m
p
ix
el
-
wis
e
c
lass
if
icatio
n
.
T
h
e
So
f
t
Ma
x
lay
er
is
co
m
p
o
s
ed
o
f
th
r
ee
So
f
tMa
x
n
o
d
es
p
er
ea
ch
i
m
ag
e
p
ix
el,
p
r
o
v
id
in
g
th
e
p
r
o
b
ab
ilit
ies o
f
th
e
th
r
ee
lab
els:
lesi
o
n
,
liv
er
,
o
r
b
ac
k
g
r
o
u
n
d
,
as
in
(
1
)
.
(
)
=
∑
(
1
)
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.
12
,
No
.
1
,
Feb
r
u
ar
y
20
22
:
3
0
3
-
310
306
wh
er
e
d
en
o
tes
th
e
i
n
p
u
t
at
th
e
s
o
f
tm
ax
n
o
d
e
an
d
(
.
)
d
e
n
o
tes
th
e
o
u
tp
u
t p
r
o
b
a
b
ilit
y
o
f
th
e
So
f
tMa
x
n
o
d
e.
T
h
e
weig
h
ts
o
f
th
e
p
ix
el
cla
s
s
if
icatio
n
lay
er
ar
e
tr
ain
e
d
u
s
in
g
th
e
L
I
T
S
d
atab
ase.
B
a
s
ed
o
n
th
e
lar
g
est
So
f
tMa
x
p
r
o
b
a
b
ilit
y
,
th
e
p
ix
e
l
-
wis
e
clas
s
if
icati
o
n
lay
er
p
r
o
v
id
es
th
e
f
in
al
o
u
tp
u
t
lab
el
f
o
r
ea
ch
p
ix
el
to
b
e
eith
er
lesi
o
n
,
liv
er
,
o
r
b
ac
k
g
r
o
u
n
d
.
2
.
3
.
F
us
io
n neura
l net
wo
rk
(
F
NN)
T
o
in
v
esti
g
ate
th
e
p
o
ten
tial
o
f
f
u
s
in
g
t
h
e
ex
tr
ac
te
d
d
ee
p
lear
n
in
g
f
ea
t
u
r
es
f
r
o
m
th
e
two
u
ti
lized
d
ee
p
lear
n
in
g
m
o
d
els
(
D
e
n
s
en
et
an
d
FC
N
-
Alex
n
et)
,
an
FNN
is
d
esig
n
ed
to
in
teg
r
ate
th
e
s
tr
en
g
th
o
f
ea
c
h
m
o
d
el.
T
h
e
p
r
o
p
o
s
ed
FNN
co
n
s
is
ts
o
f
an
in
p
u
t
la
y
er
,
o
n
e
f
u
lly
co
n
n
ec
ted
h
id
d
en
lay
er
,
a
n
d
an
o
u
t
p
u
t
lay
er
as
s
h
o
wn
in
Fig
u
r
e
4
.
T
h
e
i
n
p
u
t
la
y
er
o
f
th
e
FNN
co
n
s
is
ts
o
f
th
e
t
wo
in
p
u
t
lab
ele
d
im
ag
es
(
f
r
o
m
th
e
o
u
t
p
u
ts
o
f
t
h
e
Den
s
en
et
an
d
th
e
FC
N
-
Alex
n
et
m
o
d
els).
T
h
e
h
id
d
en
la
y
er
is
co
m
p
o
s
ed
o
f
a
n
u
m
b
e
r
o
f
1
n
o
d
es,
1
=1
0
0
,
s
elec
ted
d
u
r
in
g
ex
p
er
im
e
n
tatio
n
s
,
all
with
tan
h
ac
tiv
atio
n
f
u
n
ctio
n
s
.
T
h
e
o
u
tp
u
t
lay
e
r
i
s
co
m
p
o
s
ed
o
f
th
e
f
in
a
lly
f
u
s
ed
o
u
tp
u
t
lab
ele
d
im
ag
e
with
th
e
s
am
e
d
im
en
s
io
n
s
as
th
e
in
p
u
t
im
ag
es.
Fig
u
r
e
4
s
h
o
ws
a
ty
p
ical
ex
am
p
le
o
f
f
u
s
io
n
,
wh
er
e
th
e
p
r
o
p
o
s
ed
FNN
was a
b
le
to
en
h
an
ce
th
e
p
er
f
o
r
m
an
ce
o
f
th
e
g
iv
en
ex
a
m
p
le.
Fig
u
r
e
4
.
Ar
c
h
itectu
r
e
o
f
th
e
p
r
o
p
o
s
ed
FNN
2
.
4
.
P
er
f
o
rma
nce
m
et
rics
I
n
o
r
d
er
to
ac
cu
r
ately
ev
alu
a
te
th
e
p
er
f
o
r
m
an
ce
o
f
th
e
p
r
o
p
o
s
ed
s
y
s
tem
f
o
r
th
e
liv
er
an
d
tu
m
o
r
s
eg
m
en
tatio
n
,
two
p
ar
a
m
eter
s
ar
e
u
s
ed
to
ass
ess
th
e
q
u
ality
o
f
s
eg
m
en
tatio
n
:
o
n
e
ar
ea
-
b
a
s
ed
m
etr
ic;
th
e
d
ice
s
im
ilar
ity
co
ef
f
icien
t
(
DS
C
)
,
an
d
a
d
is
tan
ce
-
b
ased
m
etr
ic;
t
h
e
av
er
ag
e
s
y
m
m
etr
ic
s
u
r
f
ac
e
d
is
tan
ce
(
ASSD).
T
h
e
DSC
[
2
3
]
r
ep
r
esen
ts
th
e
ar
ea
o
v
er
lap
b
etwe
en
th
e
s
eg
m
en
ted
im
ag
e
(
S)
an
d
th
e
g
r
o
u
n
d
tr
u
th
(
GT
)
im
ag
e:
(
,
)
=
|
∩
|
/ 0
.
5
(
|
|
+
|
|
)
×
1
0
0
%
(
2
)
wh
er
e
th
e
|
.
|
o
p
er
ato
r
d
en
o
tes th
e
o
b
ject
ar
ea
.
On
th
e
o
th
er
h
a
n
d
,
th
e
ASSD
[
2
4
]
m
ea
s
u
r
es
t
h
e
d
is
tan
ce
b
et
wee
n
th
e
s
eg
m
en
ted
o
b
ject
s
u
r
f
ac
e
a
n
d
its
co
r
r
esp
o
n
d
in
g
GT
s
eg
m
e
n
tatio
n
s
u
r
f
ac
e,
k
n
o
wn
as
th
e
a
v
er
ag
e
o
f
th
e
E
u
clid
ian
d
is
tan
ce
s
,
,
f
r
o
m
(
i)
all
p
o
in
ts
,
,
o
n
th
e
s
u
r
f
ac
e
o
f
t
h
e
s
eg
m
en
ted
o
b
ject
(
)
t
o
th
e
s
u
r
f
ac
e
o
f
th
e
GT
(
)
an
d
(
ii)
all
p
o
in
ts
o
n
th
e
to
:
(
,
)
=1
/ (
|
|
+
|
|
)
×
(
∑
(
x
,
G
Ts
)
∈
+
∑
(
x
,
Ss
)
x
∈
G
T
s
)
×
1
0
0
%
(
3
)
3.
E
XP
E
R
I
M
E
N
T
A
L
RE
SUL
T
S AN
D
D
I
SC
USS
I
O
N
I
n
th
is
s
ec
tio
n
,
th
e
L
I
T
S
ch
all
en
g
in
g
d
atab
ase,
th
e
ex
p
er
im
en
tal
s
etu
p
,
an
d
th
e
co
m
p
ar
at
iv
e
r
esu
lts
to
o
th
er
m
et
h
o
d
s
ar
e
d
etailed
.
3
.
1
.
L
I
T
S
da
t
a
ba
s
e
T
h
e
L
I
T
S
ch
allen
g
in
g
d
atab
as
e
[
2
5
]
,
[
2
6
]
c
o
n
s
is
ts
o
f
1
3
0
co
n
tr
ast
-
en
h
an
ce
d
a
b
d
o
m
in
al
C
T
tr
ain
in
g
s
ca
n
s
co
llected
f
r
o
m
s
ev
en
d
if
f
er
en
t
clin
ical
in
s
titu
tio
n
s
.
T
h
e
tr
ain
in
g
C
T
s
ca
n
s
wer
e
g
iv
en
with
m
an
u
al
s
eg
m
en
tatio
n
s
o
f
th
e
liv
e
r
an
d
liv
er
lesi
o
n
s
d
o
n
e
b
y
t
r
ai
n
e
d
r
ad
io
lo
g
is
ts
.
All
v
o
lu
m
es
co
n
tain
ed
a
d
if
f
er
e
n
t
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
Dee
p
s
eg
men
ta
tio
n
o
f t
h
e
liver
a
n
d
th
e
h
ep
a
tic
tu
mo
r
s
fr
o
m
…
(
N
ermeen
E
lmen
a
b
a
w
y
)
307
n
u
m
b
er
o
f
ax
ial
s
lices
(
4
2
to
1
0
2
6
cr
o
s
s
-
s
ec
tio
n
p
er
v
o
lu
m
e)
,
with
an
o
v
er
all
n
u
m
b
er
o
f
1
6
,
9
1
7
im
ag
es.
T
h
e
s
ize
o
f
ea
ch
C
T
im
ag
e
is
5
1
2
×
5
1
2
p
i
x
els.
Data
d
escr
ip
tio
n
is
d
etailed
in
[
2
5
]
a
n
d
[
2
6
]
.
3
.
2
.
E
x
perim
ent
a
l set
t
ing
Mo
d
el
1
(
Den
s
en
et)
an
d
Mo
d
el
2
(
FC
N
-
Alex
n
et)
ar
e
tr
ain
ed
u
s
in
g
th
e
d
atab
ase
o
f
L
I
T
S
co
m
p
etitio
n
as
f
o
llo
ws:
in
itially
,
all
th
e
e
n
co
d
e
r
’
s
weig
h
ts
ar
e
in
itialized
b
y
tr
an
s
f
e
r
r
in
g
t
h
e
Den
s
en
et
n
etwo
r
k
in
[
1
5
]
an
d
Alex
n
et
i
n
[
1
6
]
p
er
tai
n
ed
weig
h
ts
,
r
esp
ec
tiv
ely
.
I
n
t
h
e
tr
ain
in
g
p
h
ase
,
all
en
co
d
e
r
lay
er
s
an
d
d
ec
o
d
er
lay
e
r
s
ar
e
f
in
e
-
t
u
n
ed
u
s
in
g
th
e
L
I
T
S
d
ata.
T
h
e
tr
ain
in
g
e
p
o
ch
s
ar
e
r
e
p
ea
t
ed
u
n
til
th
e
cr
o
s
s
-
en
tr
o
p
y
lo
s
s
is
v
er
y
s
m
all
o
r
th
e
n
u
m
b
er
o
f
e
p
o
ch
s
e
x
ce
ed
s
3
0
.
I
n
p
u
ts
ar
e
s
h
u
f
f
led
in
e
ac
h
ep
o
c
h
u
s
in
g
a
m
in
i
-
p
atch
s
ize
o
f
5
0
0
.
L
ea
r
n
in
g
r
ates
ar
e
s
et
to
1
0
-
3
f
o
r
m
o
d
el
1
an
d
f
o
r
m
o
d
el
2
to
af
f
o
r
d
h
ig
h
er
p
ar
am
eter
tu
n
in
g
.
FNN
tr
ain
in
g
a
p
p
lie
d
th
e
s
am
e
tr
ain
in
g
s
ettin
g
.
All
tr
ain
in
g
p
h
ases
ar
e
im
p
lem
en
ted
u
s
in
g
MA
T
L
AB
©
2
0
1
8
a.
Ov
er
-
f
itti
n
g
is
av
o
id
ed
b
y
r
ed
u
cin
g
th
e
n
etwo
r
k
'
s
ca
p
ac
ity
b
y
r
em
o
v
in
g
lay
e
r
s
(
f
u
lly
co
n
n
ec
ted
lay
er
in
th
e
p
r
e
-
tr
ai
n
ed
n
etwo
r
k
Alex
n
et
in
m
o
d
e
l
2
)
an
d
r
ed
u
ci
n
g
th
e
n
u
m
b
er
o
f
elem
en
ts
in
th
e
h
id
d
en
la
y
er
s
in
th
e
f
u
s
io
n
n
et
wo
r
k
.
A
Fiv
e
-
f
o
ld
cr
o
s
s
-
v
alid
atio
n
is
u
s
ed
to
ev
alu
ate
th
e
p
r
o
p
o
s
ed
s
y
s
tem
.
T
wo
m
o
d
es
ar
e
u
s
ed
f
o
r
th
e
in
p
u
t
d
ata:
“d
u
p
licate”
an
d
“3
D
n
ar
r
o
w
-
b
a
n
d
”
as
s
h
o
wn
in
Fig
u
r
e
5
.
I
n
t
h
e
“Du
p
licate”
m
o
d
e,
th
e
in
p
u
t
d
ata
is
co
m
p
o
s
ed
o
f
th
r
ee
d
u
p
licated
g
r
ey
lev
el
im
ag
es
at
ea
ch
o
f
th
e
th
r
ee
s
tan
d
ar
d
ch
a
n
n
els
o
f
th
e
u
tili
ze
d
d
ee
p
lear
n
in
g
m
o
d
el.
I
n
th
e
“3
D
n
a
r
r
o
w
-
b
a
n
d
”
m
o
d
e,
in
p
u
t
d
ata
i
s
co
m
p
o
s
ed
o
f
th
r
ee
co
n
s
eq
u
e
n
t
an
ato
m
ical
g
r
ey
lev
el
im
ag
es
to
th
e
p
r
o
p
o
s
ed
s
y
s
tem
(
i.e
.
,
th
e
tar
g
et
im
ag
e
t
o
th
e
ce
n
tr
alize
d
C
NN
m
o
d
el
’
s
in
p
u
t
ch
a
n
n
el
a
n
d
th
e
p
r
ev
io
u
s
an
d
n
ex
t c
r
o
s
s
-
s
ec
tio
n
s
to
ea
ch
s
id
e
ch
a
n
n
el)
.
(
a)
(
b
)
Fig
u
r
e
5
.
T
h
e
d
ata
ar
e
in
p
u
t t
o
th
e
p
r
o
p
o
s
ed
s
y
s
tem
u
s
in
g
tw
o
m
o
d
es: (
a)
“Du
p
licate”
an
d
(
b
)
“3
D
Nar
r
o
w
-
b
an
d
”.
Or
i
g
in
al
im
ag
e
(
to
p
r
o
w)
an
d
GT
im
ag
e
(
b
o
tto
m
r
o
w)
A
f
iv
e
-
f
o
ld
cr
o
s
s
-
v
alid
atio
n
is
ap
p
lied
to
e
v
alu
ate
th
e
p
r
o
p
o
s
ed
s
y
s
tem
with
two
d
i
f
f
er
en
t
s
ettin
g
s
:
“g
lo
b
al”
an
d
“p
er
ca
s
e”
.
T
h
e
“g
lo
b
al”
s
ettin
g
ap
p
lies
th
e
5
-
f
o
ld
cr
o
s
s
v
alid
atio
n
o
n
th
e
w
h
o
le
1
6
,
9
1
7
im
ag
es
o
f
all
th
e
1
3
0
s
ca
n
s
(
i.e
.
,
3
8
3
test
im
ag
es
(
2
0
%
o
f
im
ag
es)
a
n
d
1
3
,
5
3
4
tr
ain
in
g
im
ag
es
(
8
0
%
o
f
th
e
im
ag
es))
.
On
th
e
o
th
er
s
id
e,
t
h
e
“
p
er
ca
s
e”
s
ettin
g
d
i
v
id
e
th
e
d
ata
b
ase
d
o
n
ca
s
e
(
s
u
b
ject
o
r
s
ca
n
)
a
n
d
ap
p
lies
th
e
5
-
f
o
l
d
cr
o
s
s
-
v
alid
atio
n
o
n
th
e
to
tal
n
u
m
b
er
o
f
1
3
0
s
ep
ar
ate
s
ca
n
s
(
i.e
.
,
2
6
test
s
u
b
jec
ts
’
im
ag
es
(
2
0
%
o
f
th
e
s
ca
n
s
)
an
d
1
0
4
tr
ain
in
g
s
u
b
jects
(
8
0
%
o
f
th
e
s
ca
n
s
)
)
.
C
r
o
s
s
-
en
tr
o
p
y
is
u
s
ed
as
th
e
o
b
jectiv
e
f
u
n
ctio
n
to
tr
ain
t
h
e
n
etwo
r
k
u
s
in
g
ADAM
o
p
tim
i
ze
r
[
2
7
]
.
T
h
e
m
e
d
ian
f
r
e
q
u
en
c
y
b
alan
cin
g
is
u
s
ed
,
w
h
er
e
th
e
weig
h
t
ass
ig
n
ed
to
a
class
in
th
e
lo
s
s
f
u
n
ctio
n
.
3
.
3
.
E
x
perim
ent
a
l r
esu
lt
s
I
n
o
r
d
er
to
ass
ess
q
u
an
titativ
ely
th
e
s
y
s
tem
p
er
f
o
r
m
a
n
ce
,
T
ab
le
1
p
r
o
v
id
es
d
etailed
liv
er
an
d
tu
m
o
r
s
eg
m
en
tatio
n
r
esu
lts
f
o
r
ea
ch
u
tili
ze
d
C
NN
m
o
d
el
(
Den
s
e
n
et
an
d
FC
N
-
Alex
n
et)
as
w
ell
as
th
e
p
r
o
p
o
s
ed
f
u
s
ed
s
y
s
tem
.
C
o
n
s
is
ten
t
with
th
e
v
is
u
al
r
esu
lts
in
Fig
u
r
e
6
,
th
e
p
er
f
o
r
m
an
ce
o
f
FC
N
-
Alex
n
et
m
o
d
el
is
b
ette
r
th
an
th
e
Den
s
en
et
n
etwo
r
k
.
T
h
is
is
d
u
e
to
th
e
ef
f
icien
t
s
i
m
p
ler
s
tr
u
ctu
r
e
o
f
t
h
e
FC
N
-
Alex
n
et
(
its
en
co
d
e
r
co
n
tain
s
o
n
ly
f
iv
e
c
o
n
v
o
lu
tio
n
al
lay
er
s
p
lu
s
2
f
u
lly
co
n
n
ec
t
ed
lay
er
s
,
wh
ich
is
ea
s
y
to
b
e
tr
ain
ed
ef
f
icien
tly
)
co
m
p
ar
ed
to
th
e
Den
s
en
et
(
co
n
tain
s
2
0
1
lay
er
s
[
1
5
]
,
m
ak
in
g
its
tr
ain
in
g
r
ath
er
co
m
p
lex
an
d
ca
u
s
es
o
v
er
f
itti
n
g
)
.
I
n
a
d
d
itio
n
,
T
ab
l
e
s
1,
2
,
a
n
d
Fig
u
r
e
6
s
h
o
w
th
at
th
e
p
r
o
p
o
s
ed
FNN
f
u
s
io
n
f
u
r
th
e
r
im
p
r
o
v
es
th
e
p
er
f
o
r
m
an
ce
.
As
ex
p
ec
ted
,
t
h
e
“3
D
Nar
r
o
w
-
b
an
d
”
m
o
d
e
a
ch
iev
es
b
etter
r
esu
lts
th
an
th
e
“Du
p
licate”
m
o
d
e,
s
in
ce
it
tak
es
in
to
ac
co
u
n
t
a
n
ex
ten
d
ed
3
D
n
ar
r
o
w
-
b
an
d
an
ato
m
ical
in
f
o
r
m
atio
n
o
f
th
e
o
b
ject.
Ho
wev
er
,
T
ab
le
1
s
h
o
ws
th
at
wh
ile
t
h
e
“3
D
Nar
r
o
w
-
b
an
d
”
m
o
d
e
ac
h
i
ev
es
b
etter
r
esu
lts
f
o
r
tu
m
o
r
s
eg
m
en
tatio
n
f
o
r
all
th
e
th
r
ee
co
m
p
ar
e
d
s
y
s
tem
s
(
Den
s
en
et,
FC
N
-
Alex
n
et,
an
d
th
e
p
r
o
p
o
s
ed
s
y
s
tem
)
,
it
f
ails
to
en
h
an
ce
th
e
liv
er
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.
12
,
No
.
1
,
Feb
r
u
ar
y
20
22
:
3
0
3
-
310
308
s
eg
m
en
tatio
n
r
esu
lts
.
T
h
is
is
d
u
e
to
th
e
alm
o
s
t
n
o
s
ig
n
if
ic
an
t
ch
an
g
e
b
etwe
en
t
h
e
liv
e
r
an
ato
m
ies
f
o
r
th
e
co
n
s
eq
u
en
t
im
ag
es,
w
h
ile
tu
m
o
r
a
n
ato
m
y
s
h
o
ws
s
ig
n
if
ican
t
ch
an
g
es
d
u
e
to
its
r
elativ
el
y
s
m
all
s
ize
co
m
p
a
r
ed
with
th
e
liv
er
.
T
ab
le
1
.
L
iv
e
r
an
d
tu
m
o
r
s
eg
m
en
tatio
n
r
esu
lts
f
o
r
ea
c
h
u
tili
ze
d
d
ee
p
lear
n
in
g
m
o
d
el
(
Den
s
en
et,
FC
N
-
Alex
n
et,
an
d
FNN)
.
Fo
r
ea
ch
m
o
d
el,
r
esu
lts
ar
e
co
m
p
ar
ed
f
o
r
two
m
o
d
es
(
“Du
p
licate”
an
d
“
3
D
Nar
r
o
w
-
b
an
d
”
)
M
o
d
e
l
O
b
j
e
c
t
Li
v
e
r
Tu
m
o
r
M
o
d
e
“
G
l
o
b
a
l
”
“
P
e
r
c
a
se
”
“
G
l
o
b
a
l
”
“
P
e
r
c
a
se
”
M
e
t
r
i
c
D
S
C
A
S
S
D
D
S
C
A
S
S
D
D
S
C
A
S
S
D
D
S
C
A
S
S
D
M
o
d
e
l
1
D
e
n
se
n
e
t
D
u
p
l
i
c
a
t
e
8
2
.
8
%
3
.
8
9
7
6
.
5
%
4
.
9
5
6
9
.
7
%
3
.
8
7
6
2
.
6
%
4
.
1
2
N
a
r
r
o
w
-
b
a
n
d
3
D
8
2
.
8
%
3
.
8
9
7
6
.
5
%
4
.
9
5
7
3
.
0
%
2
.
7
6
6
4
.
8
%
3
.
0
7
M
o
d
e
l
2
F
C
N
A
l
e
x
n
e
t
D
u
p
l
i
c
a
t
e
9
6
.
9
%
0
.
8
9
9
1
.
4
%
1
.
3
2
7
6
.
3
%
3
.
2
1
6
6
.
1
%
3
.
8
7
N
a
r
r
o
w
-
b
a
n
d
3
D
9
6
.
9
%
0
.
8
5
9
1
.
4
%
1
.
3
5
7
8
.
2
%
2
.
4
3
6
8
.
9
%
3
.
1
8
Pr
o
p
o
sed:
F
N
N
f
u
si
o
n
D
u
p
l
i
c
a
t
e
9
7
.
2
%
0
.
7
4
9
3
.
5
%
0
.
9
9
7
8
.
8
%
2
.
3
6
7
0
.
0
%
3
.
1
1
N
a
r
r
o
w
-
b
a
n
d
3
D
9
7
.
2
%
0
.
7
2
9
3
.
5
%
0
.
7
7
7
9
.
9
%
0
.
9
2
7
4
.
4
%
0
.
9
9
Fig
u
r
e
6
.
A
“
3
D
Nar
r
o
w
-
b
an
d
”
s
am
p
le
s
eg
m
en
tatio
n
r
esu
lts
.
First co
lu
m
n
co
n
tain
s
th
e
i
n
p
u
t a
n
d
GT
s
eg
m
en
tatio
n
.
Seco
n
d
,
th
ir
d
,
F
o
r
th
an
d
last
co
lu
m
n
s
p
r
o
v
id
es
th
e
r
esu
lts
o
f
Mo
d
el
1
,
Mo
d
el
2
,
p
r
o
p
o
s
ed
FF
N,
an
d
GT
s
eg
m
en
tatio
n
,
r
esp
ec
ti
v
ely
; liv
er
(
f
i
r
s
t r
ow
)
an
d
th
e
t
u
m
o
r
(
s
ec
o
n
d
r
o
w)
3
.
4
.
Co
m
pa
ra
t
iv
e
re
s
ults
R
esu
lts
ar
e
co
m
p
ar
ed
to
th
e
r
elate
d
s
tate
-
of
-
th
e
-
ar
t
m
eth
o
d
s
o
n
th
e
L
I
T
S
co
m
p
etitio
n
d
atab
ase
to
q
u
an
tify
th
e
p
r
o
p
o
s
ed
s
y
s
tem
s
tr
en
g
th
as
s
h
o
wn
in
T
ab
le
2
.
T
h
e
p
r
o
p
o
s
ed
FNN
f
u
s
i
o
n
s
y
s
tem
ac
h
iev
es
s
u
p
er
io
r
p
er
f
o
r
m
a
n
ce
f
o
r
tu
m
o
r
s
eg
m
en
tatio
n
,
ev
id
e
n
ce
d
b
y
th
e
h
ig
h
est “p
er
ca
s
e”
DSC
a
n
d
th
e
s
m
allest “p
er
ca
s
e”
ASSD
am
o
n
g
all
th
e
co
m
p
ar
ed
m
eth
o
d
s
.
Ho
wev
er
,
th
e
liv
er
s
eg
m
en
tatio
n
r
esu
lts
ar
e
less
th
an
th
e
r
elate
d
m
o
d
els.
T
h
e
clin
ical
i
m
p
o
r
tan
ce
o
f
th
e
ac
cu
r
ate
liv
e
r
s
eg
m
en
tatio
n
is
less
im
p
o
r
ta
n
t
th
an
th
e
ac
cu
r
ate
tu
m
o
r
s
eg
m
en
tatio
n
,
e.
g
.
,
wh
e
n
co
n
s
id
er
in
g
th
e
ca
s
e
o
f
ass
is
tin
g
th
e
r
a
d
io
lo
g
is
ts
in
liv
er
ca
n
ce
r
ca
s
es.
L
ater
,
an
in
v
esti
g
atio
n
o
f
h
o
w
to
in
cr
ea
s
e
th
e
p
er
f
o
r
m
an
ce
will b
e
i
n
tr
o
d
u
ce
d
,
esp
ec
ially
f
o
r
th
e
li
v
er
s
eg
m
en
tatio
n
.
T
ab
le
2
.
C
o
m
p
a
r
ativ
e
r
esu
lts
b
etwe
en
th
e
p
r
o
p
o
s
ed
s
y
s
tem
an
d
th
e
r
elate
d
s
tate
-
of
-
th
e
-
a
r
t
m
eth
o
d
s
u
s
in
g
t
h
e
s
am
e
d
atab
ase,
co
n
s
is
tin
g
o
f
1
3
0
s
ca
n
s
P
a
p
e
r
Ex
p
e
r
i
m
e
n
t
a
l
s
e
t
u
p
M
e
t
h
o
d
“
Pe
r c
a
s
e
”
D
S
C
Li
v
e
r
Tu
m
o
r
Bi
e
t
a
l
.
[
1
0
]
Tr
a
i
n
si
z
e
=
1
1
8
Te
st
si
z
e
=
1
3
C
a
sc
a
d
e
d
R
e
sN
e
t
(
M
u
l
t
i
-
sca
l
e
F
u
s
i
o
n
)
9
5
.
1
%
5
0
.
1
%
El
me
n
a
b
a
w
y
e
t
a
l
.
[
2
8
]
4
-
f
o
l
d
v
a
l
i
d
a
t
i
o
n
Tr
a
i
n
si
z
e
=
9
7
Te
s
t
si
z
e
=
3
3
F
C
N
-
A
l
e
x
n
e
t
w
i
t
h
p
r
e
p
r
o
c
e
ssi
n
g
9
0
.
4
%
6
2
.
4
%
P
r
o
p
o
se
d
f
r
a
m
e
w
o
r
k
5
-
f
o
l
d
v
a
l
i
d
a
t
i
o
n
Tr
a
i
n
si
z
e
=
2
6
Te
st
si
z
e
=
1
0
4
F
u
si
n
g
D
e
n
se
n
e
t
a
n
d
F
C
N
-
A
l
e
x
n
e
t
9
3
.
5
%
7
4
.
4
0
%
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
Dee
p
s
eg
men
ta
tio
n
o
f t
h
e
liver
a
n
d
th
e
h
ep
a
tic
tu
mo
r
s
fr
o
m
…
(
N
ermeen
E
lmen
a
b
a
w
y
)
309
4.
CO
NCLU
SI
O
N
I
n
th
is
p
a
p
er
,
a
C
AD
s
y
s
tem
f
o
r
s
im
u
ltan
eo
u
s
liv
er
a
n
d
t
u
m
o
r
s
eg
m
e
n
tatio
n
is
p
r
esen
te
d
,
b
ased
o
n
th
e
ef
f
icien
t
f
u
s
io
n
o
f
two
d
ee
p
lear
n
in
g
C
NN
m
o
d
els,
t
r
ain
ed
u
s
in
g
3
D
n
a
r
r
o
w
-
b
an
d
d
ata.
T
h
e
s
y
s
tem
p
er
f
o
r
m
an
ce
is
ev
alu
ated
o
n
t
h
e
ch
allen
g
in
g
L
I
T
S
d
atab
ase,
ac
h
iev
in
g
s
u
p
er
io
r
p
er
f
o
r
m
a
n
ce
o
v
er
co
m
p
etin
g
m
eth
o
d
s
f
o
r
liv
er
tu
m
o
r
s
eg
m
en
tatio
n
.
I
n
th
e
f
u
t
u
r
e,
d
if
f
e
r
e
n
t
C
NN
ar
ch
itectu
r
es
a
s
w
ell
as
d
if
f
er
en
t
f
u
s
io
n
m
o
d
els
will b
e
in
v
esti
g
ated
to
im
p
r
o
v
e
t
h
e
s
eg
m
en
tatio
n
ac
c
u
r
ac
y
.
RE
F
E
R
E
NC
E
S
[
1
]
R
.
L.
S
i
e
g
e
l
,
K
.
D
.
M
i
l
l
e
r
,
a
n
d
A
.
Je
mal
,
"
C
a
n
c
e
r
st
a
t
i
s
t
i
c
s,”
C
A:
A
C
a
n
c
e
r
J
o
u
r
n
a
l
f
o
r
C
l
i
n
i
c
i
a
n
s
,
v
o
l
.
6
9
,
p
p
.
7
-
3
4
,
2
0
1
9
,
d
o
i
:
1
0
.
3
3
2
2
/
c
a
a
c
.
2
1
5
5
1
.
[
2
]
L.
H
u
a
n
g
,
M
.
W
e
n
g
,
H
.
S
h
u
a
i
,
Y
.
H
a
n
g
,
J
.
S
u
n
,
a
n
d
F
.
G
a
o
,
“
A
u
t
o
m
a
t
i
c
l
i
v
e
r
s
e
g
m
e
n
t
a
t
i
o
n
f
r
o
m
C
T
i
m
a
g
e
s
u
si
n
g
s
i
n
g
l
e
-
b
l
o
c
k
l
i
n
e
a
r
d
e
t
e
c
t
i
o
n
,
”
Bi
o
m
e
d
R
e
se
a
rc
h
I
n
t
e
r
n
a
t
i
o
n
a
l
,
v
o
l
.
2
0
1
6
,
p
p
.
1
-
1
1
,
2
0
1
6
,
d
o
i
:
1
0
.
1
1
5
5
/
2
0
1
6
/
9
4
2
0
1
4
8
.
[
3
]
E.
B
a
l
,
E.
K
l
a
n
g
,
M
.
A
m
i
t
a
i
,
a
n
d
H
.
G
r
e
e
n
s
p
a
n
,
“
A
u
t
o
ma
t
i
c
l
i
v
e
r
v
o
l
u
me
s
e
g
m
e
n
t
a
t
i
o
n
a
n
d
f
i
b
r
o
si
s
c
l
a
ss
i
f
i
c
a
t
i
o
n
,
”
M
e
d
i
c
a
l
I
m
a
g
i
n
g
2
0
1
8
:
C
o
m
p
u
t
e
r
-
Ai
d
e
d
D
i
a
g
n
o
s
i
s,
I
n
t
e
r
n
a
t
i
o
n
a
l
S
o
c
i
e
t
y
f
o
r
O
p
t
i
c
s
a
n
d
P
h
o
t
o
n
i
c
s
,
2
0
1
8
,
v
o
l
.
1
0
5
7
5
,
A
r
t
.
n
o
.
1
0
5
7
5
0
6
,
d
o
i
:
1
0
.
1
1
1
7
/
1
2
.
2
2
9
4
5
5
5
.
[
4
]
M
.
B
a
r
st
u
g
a
n
,
R
.
C
e
y
l
a
n
,
M
.
S
i
v
r
i
,
a
n
d
H
.
Er
d
o
ğ
a
n
,
“
A
u
t
o
ma
t
i
c
l
i
v
e
r
s
e
g
m
e
n
t
a
t
i
o
n
i
n
a
b
d
o
m
e
n
C
T
i
ma
g
e
s
u
si
n
g
S
LI
C
a
n
d
A
d
a
B
o
o
st
a
l
g
o
r
i
t
h
ms
,
”
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
B
i
o
s
c
i
e
n
c
e
.
Bi
o
c
h
e
m
i
st
r
y
a
n
d
B
i
o
i
n
f
o
rm
a
t
i
c
s,
2
0
1
8
,
p
p
.
2
9
-
1
3
3
,
d
o
i
:
1
0
.
1
1
4
5
/
3
1
8
0
3
8
2
.
3
1
8
0
3
8
3
.
[
5
]
J.
M
u
t
h
u
sw
a
m
y
a
n
d
B
.
K
a
n
m
a
n
i
,
“
O
p
t
i
mi
z
a
t
i
o
n
b
a
se
d
l
i
v
e
r
c
o
n
t
o
u
r
e
x
t
r
a
c
t
i
o
n
o
f
a
b
d
o
mi
n
a
l
c
t
i
ma
g
e
s
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
El
e
c
t
r
i
c
a
l
a
n
d
C
o
m
p
u
t
e
r
En
g
i
n
e
e
r
i
n
g
(
I
J
EC
E)
,
v
o
l
.
8
,
n
o
.
6
,
p
p
.
5
0
6
1
-
5
0
7
0
,
2
0
1
8
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
e
c
e
.
v
8
i
6
.
p
p
.
5
0
6
1
-
5
0
7
0
.
[
6
]
C
.
C
.
C
h
a
n
g
e
t
a
l
.
,
“
C
o
mp
u
t
e
r
-
a
i
d
e
d
d
i
a
g
n
o
s
i
s
o
f
l
i
v
e
r
t
u
m
o
r
s
o
n
c
o
m
p
u
t
e
d
t
o
m
o
g
r
a
p
h
y
i
ma
g
e
s,
”
C
o
m
p
u
t
e
r
Me
t
h
o
d
s
a
n
d
Pro
g
ra
m
s
i
n
B
i
o
m
e
d
i
c
i
n
e
,
v
o
l
.
1
4
5
,
p
p
.
4
5
-
5
1
,
2
0
1
7
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
mp
b
.
2
0
1
7
.
0
4
.
0
0
8
.
[
7
]
G
.
C
h
l
e
b
u
s
,
A
.
S
c
h
e
n
k
,
J
.
H
.
M
o
l
t
z
,
B
.
V
.
G
i
n
n
e
k
e
n
,
H
.
K
.
H
a
h
n
,
a
n
d
H
.
M
e
i
n
e
,
“
A
u
t
o
ma
t
i
c
l
i
v
e
r
t
u
m
o
r
se
g
me
n
t
a
t
i
o
n
i
n
C
T
w
i
t
h
f
u
l
l
y
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
a
n
d
o
b
j
e
c
t
-
b
a
se
d
p
o
s
t
p
r
o
c
e
ssi
n
g
,
”
S
c
i
e
n
t
i
f
i
c
r
e
p
o
rt
s,
v
o
l
.
8
,
n
o
.
1
,
2
0
1
8
,
A
r
t
.
n
o
.
1
5
4
9
7
,
d
o
i
:
1
0
.
1
0
3
8
/
s
4
1
5
9
8
-
0
1
8
-
3
3
8
6
0
-
7.
[
8
]
O
.
R
o
n
n
e
b
e
r
g
e
r
,
P
.
F
i
sc
h
e
r
,
a
n
d
T.
B
r
o
x
,
“
U
-
n
e
t
:
C
o
n
v
o
l
u
t
i
o
n
a
l
n
e
t
w
o
r
k
s
f
o
r
b
i
o
me
d
i
c
a
l
i
ma
g
e
se
g
m
e
n
t
a
t
i
o
n
,
”
Pro
c
e
e
d
i
n
g
o
f
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
Me
d
i
c
a
l
I
m
a
g
e
C
o
m
p
u
t
i
n
g
a
n
d
C
o
m
p
u
t
e
r,
v
o
l
.
9
3
5
1
,
2
0
1
5
,
p
p
.
2
3
4
-
2
4
1
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
3
1
9
-
2
4
5
7
4
-
4
_
28
.
[
9
]
Y
.
Y
u
a
n
,
“
H
i
e
r
a
r
c
h
i
c
a
l
c
o
n
v
o
l
u
t
i
o
n
a
l
-
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
s
f
o
r
a
u
t
o
m
a
t
i
c
l
i
v
e
r
a
n
d
t
u
m
o
r
s
e
g
men
t
a
t
i
o
n
,
”
a
r
Xi
v
p
re
p
ri
n
t
a
rXi
v
:
1
7
1
0
.
0
4
5
4
0
,
2
0
1
7
.
[
1
0
]
L.
B
i
,
J
.
K
i
m,
A
.
K
u
mar
,
a
n
d
D
.
F
e
n
g
,
“
A
u
t
o
m
a
t
i
c
l
i
v
e
r
l
e
s
i
o
n
d
e
t
e
c
t
i
o
n
u
s
i
n
g
c
a
sca
d
e
d
d
e
e
p
r
e
si
d
u
a
l
n
e
t
w
o
r
k
s,”
2
0
1
7
,
a
rX
i
v
p
re
p
ri
n
t
a
rXi
v
:
1
7
0
4
.
0
2
7
0
3
.
[
1
1
]
N
.
G
r
u
b
e
r
e
t
a
l
.
,
“
A
j
o
i
n
t
d
e
e
p
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
f
o
r
a
u
t
o
m
a
t
e
d
l
i
v
e
r
a
n
d
t
u
mo
r
s
e
g
me
n
t
a
t
i
o
n
,
”
2
0
1
9
,
a
rX
i
v
p
r
e
p
r
i
n
t
a
rXi
v
:
1
9
0
2
.
0
7
9
7
1
.
[
1
2
]
J.
W
a
n
g
,
Y
.
C
h
e
n
g
,
C
.
G
u
o
,
Y
.
W
a
n
g
,
a
n
d
S
.
Ta
mu
r
a
,
“
S
h
a
p
e
-
i
n
t
e
n
si
t
y
p
r
i
o
r
l
e
v
e
l
se
t
c
o
m
b
i
n
i
n
g
p
r
o
b
a
b
i
l
i
s
t
i
c
a
t
l
a
s
a
n
d
p
r
o
b
a
b
i
l
i
t
y
ma
p
c
o
n
st
r
a
i
n
s
f
o
r
a
u
t
o
mat
i
c
l
i
v
e
r
se
g
m
e
n
t
a
t
i
o
n
f
r
o
m
a
b
d
o
mi
n
a
l
C
T
i
ma
g
e
s
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
j
o
u
rn
a
l
o
f
c
o
m
p
u
t
e
r
a
ssi
s
t
e
d
r
a
d
i
o
l
o
g
y
a
n
d
su
r
g
e
r
y
,
v
o
l
.
1
1
,
n
o
.
5
,
p
p
.
8
1
7
-
8
2
6
,
2
0
1
9
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
1
5
4
8
-
015
-
1
3
3
2
-
9.
[
1
3
]
C
.
S
h
i
.
,
Y
.
C
h
e
n
g
,
F
.
Li
u
,
Y
.
W
a
n
g
,
J
.
B
a
i
,
a
n
d
S
.
Ta
m
u
r
a
,
“
A
h
i
e
r
a
r
c
h
i
c
a
l
l
o
c
a
l
r
e
g
i
o
n
-
b
a
se
d
sp
a
r
se
s
h
a
p
e
c
o
m
p
o
s
i
t
i
o
n
f
o
r
l
i
v
e
r
seg
m
e
n
t
a
t
i
o
n
i
n
C
T
s
c
a
n
s,
”
P
a
t
t
e
r
n
R
e
c
o
g
n
i
t
i
o
n
,
v
o
l
.
5
0
,
p
p
.
8
8
-
1
0
6
,
2
0
1
6
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
a
t
c
o
g
.
2
0
1
5
.
0
9
.
0
0
1
.
[
1
4
]
L.
I
.
S
o
n
g
,
K
.
F
.
G
e
o
f
f
r
e
y
,
a
n
d
H
.
E
.
K
a
i
j
i
a
n
,
“
B
o
t
t
l
e
n
e
c
k
f
e
a
t
u
r
e
s
u
p
e
r
v
i
s
e
d
U
-
N
e
t
f
o
r
p
i
x
e
l
-
w
i
se
l
i
v
e
r
a
n
d
t
u
m
o
r
seg
m
e
n
t
a
t
i
o
n
,
”
Ex
p
e
rt
S
y
st
e
m
s w
i
t
h
Ap
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
1
4
5
,
p
p
.
1
1
3
-
1
3
1
,
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
e
sw
a
.
2
0
1
9
.
1
1
3
1
3
1
.
[
1
5
]
Y
.
Z
h
u
a
n
d
S
.
N
e
w
sam
,
“
D
e
n
se
n
e
t
f
o
r
d
e
n
se
f
l
o
w
,
”
2
0
1
7
I
EEE
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
I
m
a
g
e
Pro
c
e
ss
i
n
g
(
I
C
I
P
)
,
2
0
1
7
,
p
p
.
7
9
0
-
7
9
4
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
I
P
.
2
0
1
7
.
8
2
9
6
3
8
9
.
[
1
6
]
J.
L
o
n
g
,
E.
S
h
e
l
h
a
mer,
a
n
d
T.
D
a
r
r
e
l
l
,
“
F
u
l
l
y
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
t
w
o
r
k
s
f
o
r
s
e
ma
n
t
i
c
s
e
g
m
e
n
t
a
t
i
o
n
,
”
Pr
o
c
.
o
f
t
h
e
I
EEE
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
Vi
s
i
o
n
a
n
d
P
a
t
t
e
r
n
R
e
c
o
g
n
i
t
i
o
n
,
2
0
1
5
,
p
p
.
3
4
3
1
-
3
4
4
0
.
[
1
7
]
A
.
M
i
t
t
a
l
,
R
.
H
o
o
d
a
,
a
n
d
S
.
S
o
f
a
t
,
“
L
F
-
S
e
g
N
e
t
:
A
f
u
l
l
y
c
o
n
v
o
l
u
t
i
o
n
a
l
e
n
c
o
d
e
r
-
d
e
c
o
d
e
r
n
e
t
w
o
r
k
f
o
r
s
e
g
m
e
n
t
i
n
g
l
u
n
g
f
i
e
l
d
s
f
r
o
m
c
h
e
st
r
a
d
i
o
g
r
a
p
h
s,”
W
i
re
l
e
ss
Pe
rs
o
n
a
l
C
o
m
m
u
n
i
c
a
t
i
o
n
s
,
v
o
l
.
1
0
,
p
p
.
5
1
1
-
5
2
9
,
2
0
1
8
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
1
2
7
7
-
0
1
8
-
5
7
0
2
-
9.
[
1
8
]
A
.
K
a
l
i
n
o
v
sk
y
a
n
d
V
.
K
o
v
a
l
e
v
,
“
L
u
n
g
i
ma
g
e
S
e
g
me
n
t
a
t
i
o
n
u
si
n
g
d
e
e
p
l
e
a
r
n
i
n
g
m
e
t
h
o
d
s
a
n
d
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
,
”
XI
I
I
I
n
t
.
C
o
n
f
.
i
n
P
a
t
t
e
r
n
Re
c
o
g
n
i
t
i
o
n
a
n
d
I
n
f
o
rm
a
t
i
o
n
Pr
o
c
e
ss
i
n
g
,
2
0
1
6
,
p
p
.
2
1
-
2
4
.
[
1
9
]
A
.
El
n
a
k
i
b
,
H
.
M
.
A
mer,
a
n
d
F
.
E.
A
b
o
u
-
C
h
a
d
i
,
“
C
o
m
p
u
t
e
r
a
i
d
e
d
d
e
t
e
c
t
i
o
n
s
y
s
t
e
m
f
o
r
e
a
r
l
y
c
a
n
c
e
r
o
u
s
p
u
l
m
o
n
a
r
y
n
o
d
u
l
e
s
b
y
o
p
t
i
m
i
z
i
n
g
d
e
e
p
l
e
a
r
n
i
n
g
f
e
a
t
u
r
e
s,
”
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
2
0
1
9
8
t
h
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
S
o
f
t
w
a
r
e
a
n
d
I
n
f
o
rm
a
t
i
o
n
En
g
i
n
e
e
ri
n
g
,
2
0
1
9
,
p
p
.
7
5
-
7
9
,
d
o
i
:
1
0
.
1
1
4
5
/
3
3
2
8
8
3
3
.
3
3
2
8
8
5
6
.
[
2
0
]
A
.
A
.
M
o
u
st
a
f
a
.
,
A
.
A
.
El
n
a
k
i
b
,
a
n
d
N
.
F
.
A
r
e
e
d
,
“
O
p
t
i
m
i
z
a
t
i
o
n
o
f
d
e
e
p
l
e
a
r
n
i
n
g
f
e
a
t
u
r
e
s
f
o
r
a
g
e
-
i
n
v
a
r
i
a
n
t
f
a
c
e
r
e
c
o
g
n
i
t
i
o
n
,
”
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
En
g
i
n
e
e
r
i
n
g
(
I
J
E
C
E)
,
v
o
l
.
1
0
,
n
o
.
2
,
p
p
.
1
8
3
3
-
1
8
4
1
,
2
0
2
0
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
e
c
e
.
v
1
0
i
2
.
p
p
1
8
3
3
-
1
8
4
1
.
[
2
1
]
S
.
E.
N
a
ss
a
r
,
M
.
A
.
M
o
h
a
m
e
d
,
a
n
d
A
.
El
n
a
k
i
b
,
“
M
R
I
b
r
a
i
n
t
u
m
o
r
se
g
m
e
n
t
a
t
i
o
n
u
s
i
n
g
d
e
e
p
l
e
a
r
n
i
n
g
,
”
Ma
n
s
o
u
r
a
En
g
i
n
e
e
r
i
n
g
J
o
u
rn
a
l
,
v
o
l
.
4
5
,
n
o
.
4
,
p
p
.
4
5
-
5
4
,
d
o
i
:
1
0
.
2
1
6
0
8
/
B
F
EM
U
.
2
0
2
1
.
1
3
9
4
7
0
.
[
2
2
]
D
.
El
s
a
w
a
h
,
A
.
El
n
a
k
i
b
,
a
n
d
H
.
E.
D
.
S
.
M
o
u
s
t
a
f
a
,
“
D
e
e
p
l
e
a
r
n
i
n
g
g
r
a
d
i
n
g
sy
s
t
e
m
f
o
r
d
i
a
b
e
t
i
c
r
e
t
i
n
o
p
a
t
h
y
u
si
n
g
f
u
n
d
u
s
i
m
a
g
e
s,
”
Ma
n
s
o
u
r
a
E
n
g
i
n
e
e
ri
n
g
J
o
u
rn
a
l
(
ME
J
)
,
v
o
l
.
4
5
,
n
o
.
4
,
p
p
.
1
-
8
,
d
o
i
:
1
0
.
2
1
6
0
8
/
b
f
e
m
u
.
2
0
2
0
.
1
1
8
6
4
6
.
[
2
3
]
K
.
H
.
Zo
u
e
t
a
l
.
,
“
S
t
a
t
i
st
i
c
a
l
v
a
l
i
d
a
t
i
o
n
o
f
i
ma
g
e
s
e
g
m
e
n
t
a
t
i
o
n
q
u
a
l
i
t
y
b
a
se
d
o
n
a
s
p
a
t
i
a
l
o
v
e
r
l
a
p
i
n
d
e
x
1
:
S
c
i
e
n
t
i
f
i
c
r
e
p
o
r
t
s,
”
Ac
a
d
e
m
i
c
Ra
d
i
o
l
o
g
y
,
v
o
l
.
1
1
,
n
o
.
2
,
p
p
.
1
7
8
-
1
8
9
,
2
0
0
4
,
d
o
i
:
1
0
.
1
0
1
6
/
S
1
0
7
6
-
6
3
3
2
(
0
3
)
0
0
6
7
1
-
8.
[
2
4
]
S
.
R
a
sc
h
k
a
,
“
A
n
o
v
e
r
v
i
e
w
o
f
g
e
n
e
r
a
l
p
e
r
f
o
r
m
a
n
c
e
me
t
r
i
c
s
o
f
b
i
n
a
r
y
c
l
a
ssi
f
i
e
r
sy
s
t
e
ms
,
”
a
rX
i
v
p
r
e
p
ri
n
t
a
rX
i
v
:
1
4
1
0
.
5
3
3
0
,
2
0
1
4
.
[
2
5
]
P
.
B
i
l
i
c
e
t
a
l
.
,
“
T
h
e
l
i
v
e
r
t
u
m
o
r
se
g
m
e
n
t
a
t
i
o
n
b
e
n
c
h
m
a
r
k
(
LI
Ts)
,
”
a
rX
i
v
p
r
e
p
ri
n
t
a
r
Xi
v
:
1
9
0
1
.
0
4
0
5
6
,
2
0
1
9
.
[
2
6
]
P
.
C
h
r
i
s
t
.
“
Li
TS
-
l
i
v
e
r
t
u
m
o
r
seg
me
n
t
a
t
i
o
n
c
h
a
l
l
e
n
g
e
.
”
T
h
e
L
I
T
S
w
e
b
si
t
e
.
c
o
m
p
e
t
i
t
i
o
n
s
.
c
o
d
a
l
a
b
.
o
r
g
.
2
0
1
7
.
h
t
t
p
s
:
/
/
c
o
m
p
e
t
i
t
i
o
n
s
.
c
o
d
a
l
a
b
.
o
r
g
/
c
o
m
p
e
t
i
t
i
o
n
s
/
1
7
0
9
4
(
a
c
c
e
sse
d
A
u
g.
2
,
2
0
2
1
)
.
[
2
7
]
J.
D
u
c
h
i
,
E.
H
a
z
a
n
,
a
n
d
Y
.
S
i
n
g
e
r
,
“
A
d
a
p
t
i
v
e
su
b
g
r
a
d
i
e
n
t
me
t
h
o
d
s
f
o
r
o
n
l
i
n
e
l
e
a
r
n
i
n
g
a
n
d
s
t
o
c
h
a
s
t
i
c
o
p
t
i
mi
z
a
t
i
o
n
,
”
J
o
u
r
n
a
l
o
f
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
res
e
a
rc
h
,
v
o
l
.
1
2
,
n
o
.
7
,
p
p
.
2
1
2
1
-
2
1
5
9
.
2
0
1
1
.
[
2
8
]
N
.
A
.
E
l
me
n
a
b
a
w
y
,
A
.
E
l
n
a
k
i
b
,
a
n
d
H
.
M
o
u
st
a
f
a
,
“
D
e
e
p
j
o
i
n
t
se
g
me
n
t
a
t
i
o
n
o
f
l
i
v
e
r
a
n
d
c
a
n
c
e
r
o
u
s
n
o
d
u
l
e
s
f
r
o
m
C
T
i
ma
g
e
s
,
”
2
0
2
0
3
7
t
h
N
a
t
i
o
n
a
l
R
a
d
i
o
S
c
i
e
n
c
e
C
o
n
f
e
re
n
c
e
(
N
R
S
C
)
,
2
0
2
0
,
p
p
.
2
9
6
-
3
0
1
,
d
o
i
:
1
0
.
1
1
0
9
/
N
R
S
C
4
9
5
0
0
.
2
0
2
0
.
9
2
3
5
0
9
7
.
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.
12
,
No
.
1
,
Feb
r
u
ar
y
20
22
:
3
0
3
-
310
310
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Ne
r
m
e
e
n
Elm
e
n
a
b
a
w
y
is
a
g
ra
d
u
a
te
stu
d
e
n
t
in
El
e
c
tro
n
ics
a
n
d
Co
m
m
u
n
ica
ti
o
n
s
E
n
g
in
e
e
rin
g
(ECE
)
De
p
a
rtme
n
t,
F
a
u
lt
y
o
f
En
g
i
n
e
e
rin
g
,
M
a
n
so
u
ra
Un
iv
e
rsity
si
n
c
e
2
0
1
7
.
S
h
e
re
c
e
iv
e
d
h
e
r
BS
c
fr
o
m
th
e
ECE
De
p
a
rtme
n
t
in
2
0
1
3
.
S
h
e
h
a
s
g
a
in
e
d
a
two
-
y
e
a
r
h
a
n
d
-
o
n
e
x
p
e
r
ien
c
e
o
n
m
e
d
ica
l
d
a
ta
a
n
a
ly
sis
d
u
rin
g
h
e
r
re
se
a
rc
h
stu
d
ies
.
Sh
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
e
n
g
_
n
e
rm
e
e
n
a
@y
a
h
o
o
.
c
o
m
.
Mer
v
a
t
El
-
S
e
d
d
e
k
is
a
n
a
ss
istan
t
P
ro
fe
ss
o
r
a
t
th
e
De
p
a
rtme
n
t
o
f
Co
m
m
u
n
ica
ti
o
n
s
a
n
d
El
e
c
tro
n
ic
s
En
g
i
n
e
e
rin
g
,
a
t
t
h
e
h
i
g
h
e
r
in
s
ti
tu
te
o
f
e
n
g
in
e
e
rin
g
a
n
d
tec
h
n
o
l
o
g
y
,
M
a
n
so
u
ra
.
T
h
e
m
a
in
re
se
a
rc
h
p
o
i
n
ts
i
n
c
lu
d
e
ima
g
e
p
ro
c
e
ss
in
g
a
n
d
m
a
c
h
in
e
lea
rn
in
g
.
Sh
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
m
e
rv
a
t.
e
lse
d
d
e
k
@ie
e
e
.
o
rg
.
H
o
ss
a
m
E
l
-
Din
Mo
u
st
a
fa
is
a
n
a
ss
o
c
iate
P
ro
fe
ss
o
r
a
t
t
h
e
De
p
a
rtme
n
t
o
f
El
e
c
tro
n
ics
a
n
d
Co
m
m
u
n
ica
ti
o
n
s
En
g
i
n
e
e
rin
g
,
a
n
d
t
h
e
fo
u
n
d
e
r
a
n
d
e
x
e
c
u
ti
v
e
m
a
n
a
g
e
r
o
f
Bio
m
e
d
ica
l
En
g
in
e
e
rin
g
P
r
o
g
r
a
m
(BM
E)
a
t
th
e
F
a
c
u
lt
y
o
f
En
g
i
n
e
e
rin
g
,
M
a
n
s
o
u
ra
Un
iv
e
rsity
.
T
h
e
m
a
in
re
se
a
rc
h
p
o
in
ts
in
c
l
u
d
e
b
i
o
m
e
d
ica
l
ima
g
i
n
g
,
si
g
n
a
l
p
r
o
c
e
ss
in
g
,
a
n
d
d
e
e
p
lea
rn
in
g
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
h
o
ss
a
m
_
m
o
u
sta
fa
@h
o
tma
il
.
c
o
m
.
Ahm
e
d
Eln
a
k
i
b
is
a
n
a
ss
o
c
iate
p
ro
fe
ss
o
r
in
ECE
d
e
p
a
rt
m
e
n
t,
M
a
n
s
o
u
ra
Un
iv
e
rsity
.
D.
El
n
a
k
ib
h
a
v
e
a
u
t
h
o
re
d
o
r
c
o
-
a
u
th
o
re
d
m
o
re
th
a
n
2
5
j
o
u
r
n
a
l
a
rti
c
les
,
7
b
o
o
k
c
h
a
p
ters
,
a
n
d
3
0
p
e
e
r
-
re
v
iew
e
d
c
o
n
fe
re
n
c
e
p
a
p
e
rs.
D.
El
n
a
k
i
b
is
a
re
g
u
lar
re
v
iew
e
r
f
o
r
t
o
p
in
tern
a
ti
o
n
a
l
m
e
d
ica
l
sig
n
a
l
a
n
a
l
y
sis
jo
u
rn
a
ls
th
a
t
in
c
l
u
d
e
:
M
e
d
ic
a
l
Im
a
g
e
An
a
ly
sis,
IEE
E
Tran
sa
c
ti
o
n
s
o
n
M
e
d
ica
l
Im
a
g
i
n
g
,
a
n
d
Ne
u
r
o
c
o
m
p
u
ti
n
g
.
In
2
0
1
3
,
h
e
h
a
s
a
wa
rd
e
d
t
h
e
Jo
h
n
Ho
u
c
h
e
n
s
P
rize
f
o
r
t
h
e
b
e
st
o
u
tstan
d
i
n
g
d
isse
rtatio
n
.
H
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
n
a
k
ib
@m
a
n
s.e
d
u
.
e
g
.
Evaluation Warning : The document was created with Spire.PDF for Python.