I
AE
S In
t
er
na
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI
)
Vo
l.
15
,
No
.
2
,
A
p
r
il 2
0
2
6
,
p
p
.
1
5
9
3
~
1
6
0
4
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
15
.i
2
.
p
p
1
5
9
3
-
1
6
0
4
1593
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
Fet
a
l
o
rg
a
n de
tec
tion usin
g
f
ea
ture
enhancem
en
t
w
it
h atten
tion
a
nd residua
l bloc
k
Nus
wil
B
er
no
lia
n
1
,
Siti
N
ur
m
a
ini
2,
3
,
Ade
I
ria
ni Sa
pitr
i
2,
3
,
Anni
s
a
Da
rm
a
wa
hy
un
i
2,
3
,
M
uh
a
m
m
a
d Na
ufa
l R
a
chm
a
t
ull
a
h
2,
3
,
B
a
m
ba
ng
T
utuk
o
2,
3
,
F
irda
us
F
irda
us
2,
3
1
D
e
p
a
r
t
me
n
t
o
f
F
e
t
o
m
a
t
e
r
n
a
l
, D
r
.
M
o
h
a
mm
a
d
H
o
e
si
n
G
e
n
e
r
a
l
H
o
s
p
i
t
a
l
,
P
a
l
e
m
b
a
n
g
,
I
n
d
o
n
e
si
a
2
A
r
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
-
M
e
d
i
c
a
l
C
e
n
t
e
r
o
f
E
x
c
e
l
l
e
n
c
e
,
U
n
i
v
e
r
s
i
t
a
s Sr
i
w
i
j
a
y
a
,
P
a
l
e
m
b
a
n
g
,
I
n
d
o
n
e
si
a
3
I
n
t
e
l
l
i
g
e
n
t
S
y
s
t
e
m
R
e
s
e
a
r
c
h
G
r
o
u
p
,
U
n
i
v
e
r
si
t
a
s
S
r
i
w
i
j
a
y
a
,
P
a
l
e
m
b
a
n
g
,
I
n
d
o
n
e
s
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ma
r
1
4
,
2
0
2
5
R
ev
is
ed
J
an
2
5
,
2
0
2
6
Acc
ep
ted
Feb
6
,
2
0
2
6
Th
e
ra
p
i
d
a
d
v
a
n
c
e
m
e
n
ts
in
f
e
tal
u
lt
ra
so
n
o
g
ra
p
h
y
h
a
v
e
si
g
n
ifi
c
a
n
tl
y
e
n
h
a
n
c
e
d
p
re
n
a
tal
d
iag
n
o
sis
in
re
c
e
n
t
y
e
a
rs.
De
e
p
lea
rn
i
n
g
(DL)
a
rc
h
it
e
c
tu
re
s
h
a
v
e
fu
rt
h
e
r
stre
a
m
li
n
e
d
th
e
p
r
o
c
e
ss
o
f
o
r
g
a
n
d
e
tec
ti
o
n
,
imp
ro
v
e
d
d
ia
g
n
o
stic
a
c
c
u
ra
c
y
,
a
n
d
re
d
u
c
e
d
o
b
se
rv
e
r
d
e
p
e
n
d
e
n
c
y
.
Th
is
stu
d
y
p
ro
p
o
se
s
a
c
o
m
p
u
ter
-
a
id
e
d
DL
a
p
p
r
o
a
c
h
fo
r
fe
tal
o
rg
a
n
se
g
m
e
n
tatio
n
u
sin
g
th
e
y
o
u
o
n
l
y
l
o
o
k
o
n
c
e
(YO
LO)
a
lg
o
rit
h
m
,
a
sta
te
-
of
-
t
h
e
-
a
rt
m
e
th
o
d
fo
r
o
b
jec
t
d
e
tec
ti
o
n
a
n
d
ima
g
e
se
g
m
e
n
tatio
n
.
Th
is stu
d
y
id
e
n
t
ifi
e
d
a
n
d
c
las
sified
1
5
fe
tal
o
rg
a
n
s
,
in
c
l
u
d
i
n
g
th
e
u
m
b
il
ica
l
v
e
i
n
,
st
o
m
a
c
h
,
a
b
d
o
m
e
n
,
b
ra
i
n
(tran
s
-
c
e
re
b
e
ll
u
m
,
tran
s
-
th
a
lam
i
c
,
a
n
d
tran
s
-
v
e
n
tri
c
u
lar
re
g
io
n
s),
fe
m
u
r,
h
e
a
d
,
t
h
o
ra
x
(c
h
e
st
c
a
v
it
y
),
h
e
a
rt
(c
ircu
m
fe
re
n
c
e
,
left
a
tri
u
m
,
left
v
e
n
tri
c
le
,
rig
h
t
a
tri
u
m
,
ri
g
h
t
v
e
n
tri
c
le),
a
n
d
a
o
r
ta.
We
c
o
m
p
a
re
d
t
h
e
p
e
rfo
rm
a
n
c
e
o
f
YO
LOv
7
,
YO
LOv
8
,
YO
LOv
9
,
a
n
d
YO
LOv
1
1
a
rc
h
it
e
c
tu
re
s.
T
h
e
re
su
lt
s
sh
o
we
d
t
h
a
t
YO
LOv
9
o
u
t
p
e
rfo
r
m
e
d
YO
LOv
7
,
YO
LO
v
8
,
a
n
d
YO
LOv
1
1
a
c
h
iev
in
g
m
AP5
0
a
n
d
m
AP9
5
s
c
o
re
s
o
f
9
1
.
9
0
%
a
n
d
9
4
.
5
0
%
,
re
sp
e
c
ti
v
e
ly
.
Th
is p
e
rfo
rm
a
n
c
e
su
r
p
a
ss
e
s
p
re
v
i
o
u
s
st
u
d
ies
th
a
t
fo
c
u
se
d
o
n
c
las
sif
y
in
g
o
n
l
y
a
li
m
it
e
d
n
u
m
b
e
r
o
f
fe
tal
o
r
g
a
n
s.
K
ey
w
o
r
d
s
:
C
o
m
p
u
ter
v
is
io
n
Dec
is
io
n
Dee
p
lear
n
in
g
I
n
s
tan
ce
s
eg
m
en
tatio
n
Me
d
ical
im
ag
in
g
Ultr
aso
u
n
d
YOL
O
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
:
Sit
i N
u
r
m
ain
i
I
n
tellig
en
t Sy
s
tem
R
esear
ch
Gr
o
u
p
,
Facu
lty
o
f
C
o
m
p
u
ter
Sci
en
ce
,
Un
iv
er
s
itas
Sriwijay
a
Palem
b
an
g
,
I
n
d
o
n
esia
E
m
ail: sit
i_
n
u
r
m
ain
i@
u
n
s
r
i.a
c.
id
1.
I
NT
RO
D
UCT
I
O
N
Fetal
g
r
o
wth
is
th
e
m
o
s
t
r
ap
id
p
h
ase
o
f
g
r
o
wth
in
h
u
m
an
s
,
an
d
ex
p
o
n
en
tial
g
r
o
wth
o
cc
u
r
s
b
y
s
ix
f
o
ld
s
b
etwe
en
2
2
an
d
4
0
wee
k
s
[
1
]
.
A
p
r
en
atal
u
ltra
s
o
u
n
d
i
s
th
e
p
r
im
a
r
y
d
iag
n
o
s
tic
to
o
l
f
o
r
e
x
am
in
in
g
f
etal
g
r
o
wth
[
2
]
.
A
p
r
o
p
en
s
ity
s
co
r
e
(
PS
)
c
r
o
s
s
-
s
ec
tio
n
al
an
aly
s
is
o
f
ea
c
h
f
etal
o
r
g
an
is
ex
te
n
s
iv
ely
r
eq
u
ir
ed
to
ev
alu
ate
f
etal
g
r
o
wth
.
I
n
cu
r
r
e
n
t
clin
ical
p
r
ac
tice,
th
is
ass
ess
m
en
t
is
r
o
u
tin
ely
p
er
f
o
r
m
ed
b
y
ca
p
tu
r
in
g
im
ag
es
o
f
th
e
f
etal
o
r
g
an
s
in
u
ter
o
a
n
d
m
ea
s
u
r
in
g
th
e
s
ize
o
f
v
ar
i
o
u
s
an
ato
m
ical
s
tr
u
ctu
r
es
[
3
]
.
Ho
wev
er
,
in
h
e
r
en
t
ch
allen
g
es
with
PS
im
ag
in
g
—
s
u
ch
as
ac
o
u
s
tic
s
h
ad
o
ws,
s
p
ec
k
le
n
o
is
e,
m
o
tio
n
b
lu
r
,
a
n
d
u
n
clea
r
b
o
u
n
d
ar
ies
—
ca
n
also
co
n
tr
ib
u
te
to
th
e
lo
w
d
etec
tio
n
r
at
es
o
f
s
ca
n
n
in
g
p
lan
es
o
f
f
et
al
o
r
g
an
s
[
2
]
,
[
4
]
.
Acc
u
r
ate
ass
ess
m
en
t
an
d
d
ete
ctio
n
o
f
th
e
s
ca
n
n
in
g
p
lan
e
o
f
f
etal
o
r
g
an
s
is
ess
en
tial
f
o
r
e
f
f
ec
tiv
e
p
r
eg
n
a
n
cy
m
an
ag
em
en
t
[
5
]
.
Gu
i
d
in
g
th
e
p
r
en
atal
u
ltra
s
o
u
n
d
p
r
o
b
e
t
h
r
o
u
g
h
th
e
in
tr
icate
f
etal
an
a
to
m
y
to
o
b
tain
th
e
co
r
r
ec
t
s
ca
n
n
in
g
p
lan
e
o
f
th
e
f
etal
o
r
g
an
d
em
an
d
s
ex
ten
s
iv
e
tr
ain
in
g
an
d
d
ee
p
k
n
o
w
led
g
e
[
2
]
.
Ma
n
u
al
n
av
ig
atio
n
ca
n
b
e
p
r
o
n
e
to
e
r
r
o
r
s
d
u
e
to
t
h
e
s
tr
u
ctu
r
al
s
im
ilar
ities
b
etwe
en
p
lan
es
o
f
th
e
s
am
e
f
etal
o
r
g
an
,
m
ak
in
g
th
e
p
r
o
ce
s
s
b
o
th
c
h
al
len
g
in
g
an
d
tim
e
-
co
n
s
u
m
in
g
[
3
]
.
An
a
u
to
m
ated
alg
o
r
ith
m
ca
n
b
e
u
tili
ze
d
to
d
etec
t
ac
cu
r
ate
f
etal
o
r
g
a
n
s
,
s
tr
ea
m
lin
in
g
s
u
b
s
eq
u
en
t
r
et
r
o
s
p
ec
tiv
e
an
aly
s
is
.
Mo
r
eo
v
e
r
,
s
u
ch
au
to
m
atio
n
en
s
u
r
es r
ep
r
o
d
u
cib
ilit
y
an
d
eli
m
in
ates b
o
th
in
ter
-
an
d
i
n
tr
a
-
o
b
s
er
v
er
v
ar
ia
b
ilit
y
[
3
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
15
,
No
.
2
,
Ap
r
il 2
0
2
6
:
1
5
9
3
-
1
6
0
4
1594
Ultr
aso
u
n
d
tech
n
o
lo
g
y
p
lay
s
a
v
ital
r
o
le
in
g
y
n
ec
o
lo
g
y
,
s
er
v
in
g
as
a
k
ey
to
o
l
f
o
r
ass
ess
in
g
f
etal
d
ev
elo
p
m
e
n
t
an
d
i
d
en
tify
in
g
p
o
ten
tial
co
m
p
licatio
n
s
d
u
r
in
g
p
r
eg
n
an
cy
.
On
e
o
f
its
ess
en
tial
ap
p
licatio
n
s
in
m
ed
icin
e
is
th
e
d
etec
tio
n
an
d
m
o
n
ito
r
in
g
o
f
m
ater
n
al
-
f
etal
h
ea
lth
p
lan
s
.
C
o
m
p
u
ter
-
aid
e
d
d
ee
p
lear
n
in
g
(
DL
)
is
r
eg
ar
d
ed
as
th
e
f
o
r
em
o
s
t
ar
tific
ial
in
tellig
en
ce
(
AI
)
tech
n
i
q
u
e
in
th
e
d
etec
tio
n
a
n
d
m
o
n
i
to
r
in
g
o
f
m
ater
n
al
-
f
etal
h
ea
lth
p
lan
s
[
6
]
–
[
1
0
]
W
ith
th
e
r
is
e
o
f
DL
alg
o
r
ith
m
s
,
m
an
y
im
ag
e
d
etec
tio
n
an
d
r
e
co
g
n
itio
n
task
s
ca
n
n
o
w
b
e
au
to
m
ated
,
r
e
d
u
cin
g
th
e
n
ee
d
f
o
r
h
ig
h
ly
s
k
illed
o
p
er
ato
r
s
[
2
]
.
Yo
u
o
n
ly
lo
o
k
o
n
ce
(
YOL
O)
,
as
a
r
ep
r
esen
tativ
e
o
f
DL
alg
o
r
ith
m
,
is
th
e
latest m
eth
o
d
o
f
o
b
je
ct
d
etec
tio
n
an
d
m
ed
ical
im
ag
e
s
eg
m
en
tatio
n
[
1
1
]
.
T
h
e
s
tr
u
ctu
r
e
o
f
YOL
O
is
s
im
p
le
an
d
ef
f
icien
t,
allo
win
g
i
t
to
d
ir
ec
tly
o
u
tp
u
t
t
h
e
p
o
s
itio
n
an
d
class
o
f
th
e
b
o
u
n
d
in
g
b
o
x
(
B
B
o
x
)
th
r
o
u
g
h
its
n
eu
r
al
n
etwo
r
k
[
1
2
]
.
Ku
m
ar
an
d
Go
el
[
1
3
]
ad
o
p
ted
Y
OL
Ov
4
ar
ch
itectu
r
e
to
d
etec
t
an
d
class
if
y
th
e
h
ea
lth
y
an
d
u
n
h
ea
lth
y
f
etal
b
r
ain
with
th
e
ac
cu
r
ac
y
o
f
9
7
.
2
7
%.
Sap
itri
et
a
l
.
[
1
4
]
p
r
o
p
o
s
ed
YOL
Ov
7
ar
ch
itect
u
r
e
to
d
etec
t
f
etal
ca
r
d
iac
s
u
b
s
tr
u
ctu
r
e
o
b
jects,
an
d
th
e
r
esu
lts
ac
h
iev
ed
th
e
h
ig
h
est
m
ea
n
av
er
ag
e
p
r
ec
is
io
n
(
m
AP)
o
f
8
2
.
1
0
%.
R
am
la
et
a
l
.
[
1
5
]
ex
p
l
o
r
ed
YOL
O
ar
ch
itectu
r
e
f
o
r
f
etal
h
ea
d
cir
c
u
m
f
er
e
n
ce
at
d
if
f
e
r
e
n
t
g
estatio
n
al
a
g
e.
Da
n
d
il
e
t
al
.
[
1
6
]
i
d
en
tifie
d
f
etal
h
ea
d
,
a
r
m
,
h
ea
r
t
a
n
d
b
o
d
y
u
s
in
g
YOL
Ov
5
ar
c
h
itectu
r
e
w
ith
av
er
ag
e
F1
-
s
co
r
e
o
f
9
0
.
2
5
%.
T
h
e
p
r
ev
i
o
u
s
s
tu
d
ies
s
h
o
wed
th
e
r
esu
lts
r
ep
o
r
te
d
ar
e
n
o
tewo
r
th
y
.
Nev
er
th
eless
,
to
b
est
o
u
r
k
n
o
wled
g
e,
th
e
p
r
e
v
io
u
s
s
tu
d
ies
o
n
ly
lim
ited
t
o
class
if
y
f
etal
o
r
g
an
,
wh
ic
h
ev
alu
ate
f
etal
im
a
g
e
a
n
d
its
an
n
o
tatio
n
it
in
ca
teg
o
r
ies.
H
o
wev
er
,
its
ap
p
r
o
ac
h
d
o
es
n
o
t
p
r
o
v
id
e
in
f
o
r
m
atio
n
r
eg
ar
d
in
g
wh
er
e
th
e
f
etal
o
r
g
an
is
lo
ca
ted
[
2
]
.
Ad
d
itio
n
a
lly
,
th
e
n
u
m
b
er
o
f
f
etal
o
r
g
a
n
o
b
jects st
u
d
ied
was lim
ited
.
T
h
er
ef
o
r
e
,
th
is
s
tu
d
y
p
r
o
p
o
s
ed
two
m
ain
task
s
with
th
e
in
s
tan
ce
s
eg
m
en
tatio
n
:
co
m
b
in
ed
im
ag
e
id
e
n
tific
atio
n
a
n
d
s
eg
m
en
tatio
n
in
o
r
d
er
to
d
eter
m
in
e
wh
ic
h
f
et
al
o
r
g
an
o
b
jects
ar
e
p
r
esen
t
i
n
f
etal
im
ag
es
an
d
to
d
elin
ea
te
ea
ch
f
etal
o
r
g
a
n
o
b
ject
b
y
ass
ig
n
in
g
a
u
n
iq
u
e
l
ab
el
to
e
v
er
y
p
ix
el
co
r
r
esp
o
n
d
in
g
to
a
s
p
ec
if
ic
o
b
ject.
Dr
iv
e
n
b
y
th
e
lim
itatio
n
s
o
f
cu
r
r
en
t
s
tate
-
of
-
t
h
e
-
ar
t
m
e
th
o
d
s
in
th
is
f
ield
,
th
e
co
n
tr
ib
u
tio
n
o
f
th
is
s
tu
d
y
is
au
t
o
m
atic
d
etec
tin
g
o
f
15
-
f
etal
o
r
g
an
o
b
jects
th
at
co
n
s
is
ted
o
f
th
e
(
u
m
b
ilical
v
ein
,
s
to
m
ac
h
,
an
d
a
b
d
o
m
en
)
,
b
r
ai
n
(
tr
an
s
-
ce
r
e
b
ellu
m
,
tr
an
s
-
th
alam
ic,
an
d
tr
an
s
-
v
e
n
tr
icu
lar
)
,
f
em
u
r
,
h
ea
d
,
th
o
r
ax
(
ch
est
ca
v
ity
,
h
ea
r
t
cir
cu
m
f
er
en
ce
)
,
an
d
h
ea
r
t
(
lef
t
atr
ial,
lef
t
v
en
tr
icu
lar
,
r
ig
h
t
atr
ial,
r
ig
h
t
v
e
n
tr
icu
lar
,
a
n
d
ao
r
ta)
.
T
h
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
is
b
u
ilt
u
p
o
n
th
e
YOL
O
ar
ch
itectu
r
e,
a
n
eu
r
al
n
etwo
r
k
wid
ely
u
tili
ze
d
in
m
ed
ical
im
ag
e
p
r
o
ce
s
s
in
g
an
d
co
m
p
u
ter
v
is
io
n
ap
p
licatio
n
s
.
T
h
is
s
tu
d
y
aim
s
to
ad
v
a
n
ce
m
ater
n
al
h
ea
lth
ca
r
e
b
y
en
h
an
cin
g
th
e
ac
cu
r
ac
y
an
d
ac
ce
s
s
ib
ilit
y
o
f
m
ater
n
al
h
ea
lth
s
er
v
ices.
2.
M
AT
E
R
I
AL
S AN
D
M
E
T
H
O
D
T
h
is
s
tu
d
y
aim
s
to
estab
lis
h
an
ef
f
icien
t
class
if
ier
to
d
e
tect
f
etal
u
ltra
s
o
u
n
d
p
la
n
s
.
T
h
is
s
tu
d
y
em
p
lo
y
s
in
s
tan
ce
s
eg
m
en
tati
o
n
(
s
eg
m
en
tatio
n
an
d
class
if
icatio
n
)
f
o
r
f
etal
o
r
g
an
o
b
jects.
T
h
e
f
etal
o
r
g
a
n
o
b
jects
co
n
s
is
ted
o
f
s
ix
p
ar
ts
,
i.e
.
,
ab
d
o
m
en
,
b
r
ai
n
,
f
e
m
u
r
,
h
ea
d
,
th
o
r
ax
,
an
d
h
ea
r
t.
T
h
e
r
esear
ch
m
eth
o
d
o
l
o
g
y
o
f
th
is
s
tu
d
y
ca
n
b
e
v
is
u
al
ized
in
Fig
u
r
e
1
,
wh
ich
lis
ts
:
i
)
d
ata
p
r
ep
ar
atio
n
o
f
f
e
tal
o
b
ject
im
a
g
es,
ii
)
d
ata
an
n
o
tatio
n
,
iii
)
in
s
tan
c
e
s
eg
m
en
tatio
n
with
YOL
O
al
g
o
r
ith
m
s
,
an
d
iv
)
p
er
f
o
r
m
an
ce
ev
alu
atio
n
.
Fig
u
r
e
1
.
T
h
e
s
tag
es o
f
f
etal
o
r
g
an
s
eg
m
e
n
tatio
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
F
eta
l o
r
g
a
n
d
etec
tio
n
u
s
in
g
fea
tu
r
e
en
h
a
n
ce
men
t wi
th
a
tten
t
io
n
a
n
d
r
esid
u
a
l b
lo
ck
(
N
u
s
w
i
l B
ern
o
lia
n
)
1595
2
.
1
.
Da
t
a
prepa
ra
t
i
o
n
T
h
is
s
tu
d
y
i
n
teg
r
ates
two
d
ata
s
ets,
co
m
p
r
is
in
g
b
o
th
p
u
b
lic
a
n
d
p
r
iv
ate
s
o
u
r
ce
s
.
T
h
e
c
o
m
b
in
atio
n
o
f
th
ese
d
atasets
is
in
ten
d
ed
t
o
e
n
h
an
ce
d
ata
d
iv
er
s
ity
an
d
e
x
p
an
d
th
e
s
ize
o
f
th
e
t
r
ain
in
g
s
et,
th
er
eb
y
im
p
r
o
v
in
g
m
o
d
el
r
o
b
u
s
tn
ess
an
d
ac
cu
r
ac
y
.
B
y
lear
n
in
g
f
r
o
m
m
o
r
e
d
iv
er
s
e
s
am
p
les,
th
e
p
r
o
p
o
s
ed
m
o
d
els
ar
e
b
etter
ab
le
to
id
en
tify
v
ar
i
o
u
s
f
etal
p
la
n
es
an
d
c
o
n
tr
ib
u
te
to
ad
v
a
n
ce
s
in
p
r
e
n
atal
h
ea
lth
ca
r
e
,
p
ar
ticu
lar
ly
in
f
etal
ass
es
s
m
en
t a
n
d
d
etec
tio
n
.
T
h
e
d
atasets
u
s
ed
in
th
is
s
tu
d
y
co
n
s
is
t o
f
two
s
o
u
r
ce
s
.
i)
T
h
e
p
u
b
lic
d
ataset
was
o
b
tain
ed
f
r
o
m
m
u
ltip
le
h
o
s
p
itals
in
B
ar
ce
lo
n
a,
Sp
ain
[
1
7
]
.
I
t
co
m
p
r
is
es
1
2
,
4
0
0
two
-
d
im
en
s
io
n
al
u
ltra
s
o
u
n
d
i
m
ag
es
co
llected
f
r
o
m
8
9
6
p
r
e
g
n
an
t
wo
m
en
.
T
h
ese
im
ag
es
a
r
e
ca
teg
o
r
ize
d
in
to
f
iv
e
f
etal
o
b
jects:
th
o
r
a
x
,
f
em
u
r
,
a
b
d
o
m
en
,
b
r
ain
,
an
d
h
ea
d
.
All
im
ag
es
wer
e
m
a
n
u
all
y
an
n
o
tated
b
y
a
g
y
n
ec
o
lo
g
y
s
p
ec
ialis
t.
ii)
T
h
e
p
r
iv
ate
d
ataset
was
co
llected
at
Dr
.
Mo
h
am
m
a
d
Ho
esin
Gen
er
al
Ho
s
p
ital
in
I
n
d
o
n
esia
an
d
co
n
tain
s
8
0
0
two
-
d
im
en
s
io
n
al
u
ltra
s
o
u
n
d
im
ag
es f
o
cu
s
in
g
o
n
th
e
f
etal
h
ea
r
t.
T
ab
le
1
s
u
m
m
ar
izes
th
e
d
is
tr
ib
u
tio
n
o
f
s
ix
f
etal
o
b
jects
a
cr
o
s
s
th
e
two
d
atasets
an
d
p
r
esen
ts
th
e
n
u
m
b
er
o
f
s
am
p
les
allo
ca
ted
to
th
e
tr
ain
in
g
,
v
alid
atio
n
,
an
d
test
in
g
s
ets.
T
o
r
ed
u
ce
b
ias
an
d
p
r
ev
e
n
t
o
v
er
f
itti
n
g
d
u
r
i
n
g
tr
ai
n
in
g
,
th
e
DL
m
o
d
els
wer
e
tr
ain
ed
u
s
in
g
b
o
th
th
e
tr
ain
in
g
an
d
v
ali
d
atio
n
d
ata.
Mo
d
el
p
er
f
o
r
m
an
ce
was
th
en
ev
alu
ated
o
n
t
h
e
test
in
g
s
et,
wh
i
ch
co
n
s
is
ts
o
f
p
r
e
v
io
u
s
ly
u
n
s
ee
n
s
am
p
les.
T
h
is
ev
alu
atio
n
s
tr
ateg
y
en
s
u
r
es
th
at
th
e
m
o
d
el
d
em
o
n
s
tr
ates
g
o
o
d
g
en
er
aliza
tio
n
ca
p
ab
ilit
y
a
n
d
d
o
es
n
o
t
o
v
er
f
it
th
e
tr
ain
in
g
d
ata.
T
ab
le
1
.
T
h
e
f
etal
o
b
ject
im
a
g
es d
is
tr
ib
u
tio
n
o
f
tr
ai
n
in
g
a
n
d
test
s
et
D
a
t
a
s
e
t
O
b
j
e
c
t
s
C
l
a
s
s
N
u
mb
e
r
o
f
d
a
t
a
Tr
a
i
n
i
n
g
se
t
v
a
l
i
d
a
t
i
o
n
se
t
Te
st
i
n
g
s
e
t
P
u
b
l
i
c
(
B
a
r
c
e
l
o
n
a
S
p
a
i
n
H
o
sp
i
t
a
l
s)
A
b
d
o
me
n
U
mb
i
l
i
c
a
l
v
e
i
n
(
U
V
)
,
s
t
o
mac
h
(
S
TO
)
,
a
b
d
o
me
n
(
A
B
D
)
1
7
6
24
28
H
e
a
d
H
e
a
d
(
H
D
)
7
0
9
1
5
0
3
6
1
B
r
a
i
n
Tr
a
n
s
-
c
e
r
e
b
e
l
l
u
m (T
C
)
,
t
r
a
n
s
-
t
h
a
l
a
mi
c
(
TT),
t
r
a
n
s
-
v
e
n
t
r
i
c
u
l
a
r
(
TV
)
8
1
1
2
0
5
3
9
7
F
e
mu
r
F
e
mu
r
(
F
R
)
2
6
0
40
44
Th
o
r
a
x
C
h
e
st
c
a
v
i
t
y
(
C
C
)
,
h
e
a
r
t
c
i
r
c
u
mf
e
r
e
n
c
e
(
H
C
)
2
2
3
77
87
P
r
i
v
a
t
e
(
D
r
.
M
o
h
a
mm
a
d
H
o
e
si
n
G
e
n
e
r
a
l
H
o
sp
i
t
a
l
,
I
n
d
o
n
e
si
a
)
H
e
a
r
t
Le
f
t
a
t
r
i
a
l
(
LA
)
,
l
e
f
t
v
e
n
t
r
i
c
u
l
a
r
(
LV
)
,
r
i
g
h
t
a
t
r
i
a
l
(
R
A
)
,
r
i
g
h
t
v
e
n
t
r
i
c
u
l
a
r
(
R
V
)
,
a
o
r
t
a
(
A
O
)
2
5
3
30
26
2
.
2
.
Da
t
a
a
nn
o
t
a
t
io
n
T
h
e
an
n
o
t
ati
o
n
p
r
o
ce
s
s
g
e
n
e
r
ates
g
r
o
u
n
d
t
r
u
th
d
at
a
b
y
c
o
n
v
e
r
ti
n
g
u
l
tr
as
o
u
n
d
im
ag
es
in
t
o
s
em
an
tic
s
eg
m
e
n
t
ati
o
n
m
as
k
s
u
s
in
g
t
h
e
Py
th
o
n
L
ab
elM
e
li
b
r
ar
y
(
s
ee
Fig
u
r
e
2
)
.
I
n
th
is
s
tu
d
y
,
R
o
b
o
f
lo
w
w
as
em
p
l
o
y
e
d
to
d
eli
n
e
ate
p
o
ly
g
o
n
al
r
e
g
i
o
n
s
co
r
r
es
p
o
n
d
i
n
g
t
o
f
eta
l
o
r
g
a
n
s
,
as il
lu
s
tr
ate
d
i
n
F
ig
u
r
e
2
,
wh
ic
h
s
h
o
ws
f
etal
o
r
g
an
o
b
je
cts
wi
th
s
t
a
n
d
ar
d
p
l
an
es
o
f
a
b
d
o
m
e
n
as
i
n
Fi
g
u
r
e
2
(
a)
,
h
ea
d
as
i
n
Fi
g
u
r
e
2
(
b
)
,
b
r
ai
n
as
i
n
Fi
g
u
r
e
2
(
c)
,
f
e
m
u
r
as
i
n
Fi
g
u
r
e
2
(
d
)
,
t
h
o
r
ax
as
i
n
Fi
g
u
r
e
2
(
e
)
,
an
d
h
e
a
r
t
as
i
n
Fi
g
u
r
e
2
(
f
)
.
A
ll
g
r
o
u
n
d
t
r
u
t
h
u
lt
r
as
o
u
n
d
im
a
g
es
a
r
e
p
r
o
v
i
d
e
d
i
n
jo
in
t
p
h
o
t
o
g
r
a
p
h
i
c
ex
p
e
r
ts
g
r
o
u
p
(
J
P
E
G/J
PG
)
f
o
r
m
at.
2
.
3
.
M
o
del
ev
a
lua
t
i
o
n
I
n
s
tan
ce
s
eg
m
en
tatio
n
in
v
o
lv
es
p
r
ed
ictin
g
b
o
th
th
e
o
b
ject
i
n
s
tan
ce
s
an
d
th
eir
co
r
r
esp
o
n
d
in
g
b
in
ar
y
s
eg
m
en
tatio
n
m
ask
s
,
co
m
b
i
n
in
g
o
b
ject
d
etec
tio
n
,
an
d
s
em
an
tic
s
eg
m
en
tatio
n
in
o
n
e
p
r
o
c
ess
[
1
8
]
.
T
h
is
m
ea
n
s
th
at
in
s
tan
ce
s
eg
m
en
tatio
n
ca
n
b
e
u
s
ed
s
im
u
ltan
eo
u
s
ly
to
c
o
m
p
lete
co
u
n
tin
g
task
s
f
o
r
d
if
f
er
en
t
f
etal
o
b
jects.
YOL
O
is
a
n
o
v
el
o
n
e
-
s
tag
e
e
n
d
-
to
-
e
n
d
DL
m
o
d
el
f
o
r
in
s
ta
n
ce
s
eg
m
en
tatio
n
[
1
9
]
.
YOL
O
is
d
esig
n
ed
f
o
r
f
ast
an
d
ac
cu
r
ate
o
b
ject
d
etec
tio
n
i
n
b
o
th
i
m
ag
es
an
d
v
id
eo
s
.
I
t
s
p
lits
th
e
im
ag
e
in
to
a
g
r
id
an
d
p
r
ed
icts
B
B
o
x
an
d
class
p
r
o
b
ab
ilit
ies
f
o
r
ea
ch
g
r
id
ce
ll,
e
n
ab
lin
g
th
e
d
etec
tio
n
o
f
m
u
ltip
le
o
b
jects
in
a
s
in
g
le
p
ass
.
YOL
O
ac
h
iev
es
h
ig
h
s
p
ee
d
an
d
e
f
f
ic
ien
cy
b
y
elim
in
atin
g
t
h
e
n
ee
d
f
o
r
r
eg
io
n
p
r
o
p
o
s
al
n
etwo
r
k
s
[
2
0
]
.
T
h
is
s
tu
d
y
co
m
p
ar
ed
YOL
O
ar
ch
itectu
r
e
s
s
u
ch
as YO
L
Ov
7
,
YOL
Ov
8
,
YOL
Ov
9
,
an
d
YOL
Ov
1
1
ar
c
h
itectu
r
es.
2
.
4
.
At
t
ent
io
n m
ec
ha
nis
m
B
y
in
teg
r
atin
g
R
esNet
b
lo
ck
s
,
th
e
m
o
d
el
b
en
ef
its
f
r
o
m
im
p
r
o
v
ed
f
ea
tu
r
e
ex
tr
ac
tio
n
ca
p
ab
i
liti
es
d
u
e
to
d
ee
p
er
n
etwo
r
k
la
y
er
s
[
2
1
]
,
[
2
2
]
.
I
n
ad
d
itio
n
,
c
o
n
v
o
lu
tio
n
al
b
lo
c
k
atten
tio
n
m
o
d
u
le
(
C
B
AM
)
is
an
atten
tio
n
m
ec
h
a
n
is
m
th
at
en
h
an
ce
s
th
e
r
e
p
r
esen
tatio
n
al
p
o
wer
o
f
co
n
v
o
lu
tio
n
al
n
e
u
r
al
n
etwo
r
k
s
(
C
NNs
)
b
y
f
o
cu
s
in
g
o
n
im
p
o
r
tan
t
f
ea
tu
r
e
s
alo
n
g
b
o
th
th
e
c
h
an
n
el
a
n
d
s
p
atial
d
im
en
s
io
n
s
.
I
t
s
eq
u
en
tia
lly
ap
p
lies
ch
an
n
el
an
d
s
p
atial
atten
tio
n
to
r
ef
in
e
f
ea
tu
r
e
m
a
p
s
,
en
a
b
lin
g
th
e
n
etwo
r
k
to
co
n
ce
n
tr
ate
o
n
m
o
r
e
i
n
f
o
r
m
ativ
e
p
ar
ts
o
f
th
e
in
p
u
t
d
ata
[
2
2
]
,
[
2
3
]
.
T
h
e
o
v
er
all
ar
ch
itectu
r
e
o
f
th
e
p
r
o
p
o
s
ed
m
o
d
el,
in
co
r
p
o
r
atin
g
r
esid
u
al
b
lo
ck
s
an
d
C
B
AM
,
i
s
illu
s
tr
ated
in
Fig
u
r
e
3
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
15
,
No
.
2
,
Ap
r
il 2
0
2
6
:
1
5
9
3
-
1
6
0
4
1596
(
a)
(
b
)
(
c)
(
d
)
(
e)
(f)
Fig
u
r
e
2
.
Fetal
o
r
g
an
o
b
jects with
s
tan
d
ar
d
p
lan
es o
f
(
a)
ab
d
o
m
en
,
(
b
)
h
ea
d
,
(
c)
b
r
ain
,
(
d
)
f
em
u
r
,
(
e)
th
o
r
a
x
,
an
d
(
f
)
h
ea
r
t
Fig
u
r
e
3
.
A
n
ew
ar
ch
itectu
r
e
o
f
YOL
O
with
r
esid
u
al
b
lo
ck
-
C
B
AM
2
.
5
.
M
et
ric
ev
a
lua
t
io
n
T
h
e
ev
alu
atio
n
m
etr
ics
f
o
r
a
n
o
b
ject
d
etec
tio
n
m
o
d
el
ass
ess
its
ab
ilit
y
to
ac
cu
r
ately
d
etec
t
an
d
lo
ca
lize
o
b
jects with
in
an
im
a
g
e.
Fo
r
ev
alu
ati
n
g
th
e
p
er
f
o
r
m
an
ce
o
f
f
etal
o
b
ject
d
etec
tio
n
,
th
is
s
tu
d
y
was u
s
ed
in
ter
s
ec
tio
n
o
v
er
u
n
io
n
(
I
o
U
)
an
d
m
AP.
I
o
U,
also
r
e
f
er
r
e
d
t
o
as
th
e
J
ac
ca
r
d
in
d
e
x
,
m
ea
s
u
r
es
th
e
ac
cu
r
ac
y
b
y
ca
lcu
latin
g
th
e
o
v
e
r
lap
b
etwe
en
th
e
p
r
e
d
icted
B
B
o
x
a
n
d
t
h
e
g
r
o
u
n
d
tr
u
t
h
b
o
x
.
I
o
U
f
o
r
c
o
m
p
ar
in
g
s
im
ilar
ity
b
etwe
en
two
ar
b
itra
r
y
s
h
ap
es
,
⊆
∈
ℝ
is
attain
ed
b
y
(
1
)
[
2
4
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
F
eta
l o
r
g
a
n
d
etec
tio
n
u
s
in
g
fea
tu
r
e
en
h
a
n
ce
men
t wi
th
a
tten
t
io
n
a
n
d
r
esid
u
a
l b
lo
ck
(
N
u
s
w
i
l B
ern
o
lia
n
)
1597
=
|
∩
|
|
∪
|
(
1
)
L
et
d
e
n
o
te
t
h
e
g
r
o
u
n
d
tr
u
t
h
an
d
r
ep
r
esen
t
th
e
p
r
ed
ict
ed
o
u
t
p
u
t.
B
y
co
m
p
a
r
in
g
th
e
I
o
U
v
alu
e
with
a
p
r
ed
ef
in
e
d
t
h
r
esh
o
ld
,
a
d
ete
ctio
n
ca
n
b
e
class
if
ied
as
co
r
r
ec
t
o
r
in
co
r
r
ec
t.
A
d
etec
tio
n
i
s
co
n
s
id
er
ed
co
r
r
ec
t
wh
en
I
o
U
≥
; o
th
e
r
wis
e,
it is
d
ee
m
ed
in
c
o
r
r
ec
t w
h
e
n
I
o
U
<
.
m
AP
is
a
s
tan
d
ar
d
ev
alu
atio
n
m
etr
ic
u
s
ed
to
ass
ess
th
e
p
er
f
o
r
m
an
ce
o
f
o
b
ject
d
etec
tio
n
m
o
d
els
ac
r
o
s
s
all
o
b
ject
class
es.
I
t
c
o
n
s
id
er
s
b
o
t
h
p
r
ec
is
io
n
an
d
r
ec
all
at
v
ar
io
u
s
I
o
U
t
h
r
esh
o
l
d
s
f
o
r
ea
ch
class
.
A
h
ig
h
er
m
AP
v
alu
e
in
d
icate
s
s
u
p
er
io
r
o
v
e
r
all
d
etec
tio
n
p
er
f
o
r
m
an
ce
.
T
h
e
c
o
r
r
esp
o
n
d
in
g
f
o
r
m
u
latio
n
is
g
iv
en
as
(
2
)
[
2
5
]
.
=
1
∑
=
1
(
2
)
W
ith
b
ein
g
th
e
av
er
ag
e
p
r
ec
is
io
n
in
th
e
i
-
t
h
class
an
d
N
is
th
e
to
tal
n
u
m
b
e
r
o
f
class
es b
ein
g
ev
alu
ated
.
2
.
6
.
P
l
a
t
f
o
r
m
T
h
e
ex
p
er
im
en
t
wo
r
k
s
o
n
I
n
t
el(
R
)
C
o
r
e(
™
)
i9
-
1
4
9
0
0
K
(
3
2
C
PUs
)
,
3
.
2
GHz
R
AM
,
an
d
NVI
DI
A
GeFo
r
ce
R
T
X
2
0
8
0
T
i.
All
e
x
p
er
im
en
ts
wer
e
r
u
n
o
n
W
in
d
o
ws
1
1
Pro
6
4
B
it.
Py
th
o
n
co
d
e
u
s
in
g
VS
C
o
d
e,
Py
to
r
ch
,
Nu
m
Py
,
Pan
d
as,
Scik
itLe
ar
n
,
Ma
tp
lo
tlib
,
Seab
o
r
n
,
an
d
R
o
b
o
f
l
o
w
was u
s
ed
.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
is
s
ec
tio
n
p
r
esen
ts
a
d
escr
ip
tiv
e
an
aly
s
is
an
d
d
is
cu
s
s
es
th
e
p
r
o
p
o
s
ed
m
o
d
el’
s
r
esu
lts
f
o
r
f
etal
o
b
ject
d
etec
tio
n
.
T
h
is
s
tu
d
y
co
m
p
ar
ed
YOL
Ov
7
,
YOL
Ov
8
,
YOL
Ov
9
,
an
d
YOL
OV1
1
to
o
b
tain
th
e
b
est
m
o
d
el
f
o
r
1
5
-
o
b
ject
f
etal
o
r
g
an
s
eg
m
en
tatio
n
.
T
a
b
le
2
p
r
e
s
en
ts
th
e
r
esu
lts
o
f
YOL
O
m
o
d
els
f
o
r
f
etal
o
r
g
an
s
eg
m
en
tatio
n
with
m
AP5
0
a
n
d
m
AP9
5
.
T
h
e
m
AP5
0
an
d
m
AP9
5
m
ea
n
m
AP
wer
e
ca
lcu
lated
at
an
I
o
U
th
r
esh
o
ld
o
f
0
.
5
0
an
d
0
.
9
5
,
r
esp
ec
tiv
ely
.
As
s
ee
n
in
T
a
b
le
2
,
YOL
Ov
9
o
u
t
p
er
f
o
r
m
ed
YOL
Ov
7
,
YOL
Ov
8
,
an
d
YOL
Ov
1
1
,
with
th
e
m
A
P5
0
B
B
o
x
ac
h
iev
e
d
was
9
1
.
9
0
%
an
d
m
AP9
5
was
9
4
.
5
0
%,
wh
ile
th
e
m
AP5
0
m
ask
was
9
1
.
7
0
%,
an
d
m
AP
9
0
was
9
4
.
4
0
%.
T
ab
le
2
also
co
m
p
ar
es
d
if
f
er
e
n
t
YOL
O
m
o
d
els
b
ased
o
n
I
o
U
m
etr
ics.
T
h
e
I
o
U
o
f
YOL
Ov
9
is
clo
s
e
to
YOL
Ov
8
,
with
o
n
ly
o
0
.
0
3
%
d
if
f
e
r
en
ce
.
T
h
e
r
e
is
n
o
s
ig
n
if
ican
t
ch
an
g
e,
an
d
it
h
as
p
r
o
v
e
n
YOL
Ov
9
s
till
o
u
tp
er
f
o
r
m
s
all
YOL
O
m
o
d
els.
Ov
er
all,
YOL
Ov
9
ten
d
s
to
s
h
o
w
a
s
tab
le
p
er
f
o
r
m
a
n
ce
with
m
A
P5
0
,
m
AP9
5
,
a
n
d
I
o
U.
YOL
Ov
7
u
n
d
er
p
er
f
o
r
m
s
co
m
p
ar
e
d
to
YOL
Ov
8
an
d
YOL
Ov
9
,
m
ak
in
g
it
less
o
p
tim
al
f
o
r
f
etal
o
r
g
an
s
eg
m
en
t
atio
n
.
Hen
ce
,
th
is
s
tu
d
y
p
r
o
p
o
s
ed
th
e
YOL
Ov
9
m
o
d
el
f
o
r
f
etal
o
r
g
an
s
eg
m
e
n
t
atio
n
.
T
ab
le
2
.
T
h
e
B
B
o
x
an
d
m
ask
r
esu
lts
with
YOL
Ov
7
,
YOL
Ov
8
,
YOL
Ov
9
,
an
d
YOL
Ov
1
1
M
o
d
e
l
B
B
o
x
(
%)
M
a
s
k
(
%)
I
o
U
mA
P
5
0
mA
P
9
5
mA
P
5
0
mA
P
9
5
Y
O
LO
v
7
8
0
.
5
1
5
2
.
2
6
8
0
.
2
7
5
0
.
0
0
7
0
.
0
3
Y
O
LO
v
8
9
4
.
4
0
7
2
.
6
0
9
4
.
3
0
6
7
.
0
0
8
0
.
4
8
Y
O
LO
v
9
9
1
.
9
0
9
4
.
5
0
9
1
.
7
0
9
4
.
4
0
8
0
.
4
5
Y
O
LO
v
1
1
9
3
.
9
0
7
2
.
7
0
9
3
.
4
0
6
7
.
2
0
7
9
.
7
5
Fig
u
r
e
4
p
r
esen
ts
th
e
co
m
p
ar
i
s
o
n
o
f
th
e
p
er
f
o
r
m
a
n
ce
o
f
th
r
ee
YOL
O
m
o
d
els
(
YOL
Ov
7
,
YOL
Ov
8
,
YOL
Ov
9
,
an
d
YOL
OV1
1
)
a
cr
o
s
s
1
5
-
o
b
ject
f
etal
o
r
g
an
s
.
T
h
e
y
-
a
x
is
r
ep
r
esen
ts
m
AP5
0
,
wh
ile
th
e
x
-
ax
is
r
ep
r
esen
ts
d
if
f
er
en
t
o
b
ject
c
lass
es,
lab
eled
with
ab
b
r
ev
iatio
n
s
.
I
n
Fig
u
r
e
4
(
a)
,
YOL
Ov
9
(
o
r
a
n
g
e)
a
n
d
YOL
Ov
1
1
(
g
r
ee
n
)
ten
d
to
h
a
v
e
h
ig
h
er
o
r
s
im
ilar
m
AP5
0
v
alu
es
co
m
p
ar
e
d
to
YOL
Ov
7
(
lig
h
t
b
lu
e)
.
I
n
s
o
m
e
ca
s
es,
YOL
Ov
9
(
o
r
an
g
e)
ap
p
ea
r
s
s
lig
h
tly
ah
ea
d
o
f
YOL
Ov
8
(
d
ar
k
b
lu
e)
in
p
er
f
o
r
m
a
n
ce
.
YOL
Ov
8
an
d
YOL
Ov
9
b
o
th
r
ea
ch
n
ea
r
1
0
0
% m
AP5
0
in
m
u
ltip
le
class
es (
e.
g
.
,
"Ab
d
",
"FR
",
"HD
",
"RV",
"Ao
")
.
YOL
Ov
7
h
as
n
o
ticea
b
ly
lo
wer
p
er
f
o
r
m
an
ce
in
ce
r
tain
class
es
(
e.
g
.
,
"
Sto
",
"T
_
C
",
an
d
"T
_
T
")
.
YO
L
Ov
8
an
d
YOL
Ov
9
p
er
f
o
r
m
n
ea
r
ly
eq
u
ally
,
with
YOL
Ov
9
h
a
v
in
g
a
s
lig
h
t
ed
g
e
in
s
o
m
e
ca
s
es.
I
n
Fig
u
r
e
4
(
b
)
with
th
e
m
AP5
0
m
ask
,
YOL
Ov
9
(
o
r
an
g
e
)
an
d
YOL
Ov
1
1
(
g
r
ee
n
)
p
er
f
o
r
m
b
etter
th
an
YOL
Ov
7
(
lig
h
t
b
lu
e)
in
m
o
s
t
class
es.
YOL
Ov
7
s
h
o
ws
n
o
ticea
b
ly
lo
wer
p
er
f
o
r
m
an
ce
i
n
ce
r
tain
c
ateg
o
r
ies,
esp
ec
ially
"Sto
"
an
d
"T
_
C
"
.
YOL
Ov
9
(
o
r
an
g
e)
an
d
YOL
Ov
8
(
d
ar
k
b
lu
e)
h
a
v
e
s
im
ilar
p
er
f
o
r
m
an
c
e,
b
u
t
in
s
o
m
e
ca
s
es,
YOL
Ov
9
is
s
lig
h
tly
ah
ea
d
.
Ma
n
y
class
es
(
e.
g
.
,
"Ab
d
",
"
FR
",
"HD
",
"RV",
"Ao
")
r
ea
ch
n
ea
r
1
0
0
%
m
AP5
0
Ma
s
k
.
B
ased
o
n
o
v
er
all
Fig
u
r
e
4
,
YOL
Ov
9
s
lig
h
tly
le
ad
s
in
s
o
m
e
ca
s
es.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
15
,
No
.
2
,
Ap
r
il 2
0
2
6
:
1
5
9
3
-
1
6
0
4
1598
(
a)
(
b
)
Fig
u
r
e
4
.
Per
f
o
r
m
an
c
e
co
m
p
ar
is
o
n
o
f
YOL
O
m
o
d
els
(
a)
m
A
P5
0
co
m
p
a
r
is
o
n
ac
r
o
s
s
1
5
f
eta
l o
r
g
an
class
es
an
d
(
b
)
m
AP5
0
m
ask
co
m
p
a
r
is
o
n
ac
r
o
s
s
th
e
s
am
e
class
es
B
ased
o
n
th
e
co
m
p
ar
is
o
n
r
e
s
u
lts
ab
o
v
e,
th
is
s
tu
d
y
h
as
ex
p
lo
r
ed
t
h
e
YOL
OV8
an
d
YOL
Ov
9
cu
s
to
m
m
o
d
el
with
r
esid
u
al
b
lo
ck
an
d
C
B
AM
.
T
ab
le
3
p
r
esen
ts
th
e
h
ig
h
est
I
o
U
(
8
1
.
0
1
)
is
ac
h
iev
ed
b
y
YOL
Ov
8
-
R
esNetB
lo
ck
C
B
A
M,
in
d
icatin
g
its
s
u
p
er
io
r
ab
ilit
y
in
d
etec
tin
g
an
d
lo
ca
lizin
g
o
b
jects
ac
cu
r
ately
.
YOL
Ov
9
(
9
1
.
9
0
m
AP5
0
)
la
g
s
b
eh
in
d
th
e
YOL
Ov
8
v
ar
i
atio
n
s
b
u
t
ex
ce
ls
in
m
AP9
5
(
9
4
.
5
0
)
,
s
h
o
win
g
s
tr
o
n
g
er
g
en
er
aliza
tio
n
at
h
ig
h
er
p
r
ec
is
io
n
.
YOL
Ov
9
(
9
4
.
4
0
m
AP9
5
)
is
th
e
b
est
in
h
i
g
h
-
p
r
ec
is
io
n
s
eg
m
en
tatio
n
.
YOL
Ov
9
p
e
r
f
o
r
m
s
well
at
h
ig
h
p
r
ec
is
io
n
(
m
AP9
5
)
,
m
ea
n
i
n
g
it'
s
b
ett
er
s
u
ited
f
o
r
task
s
r
eq
u
ir
in
g
h
ig
h
ly
ac
cu
r
ate
d
ete
ctio
n
s
.
F
i
g
u
r
e
5
p
r
es
e
n
ts
t
h
e
l
e
a
r
n
i
n
g
c
u
r
v
e
s
f
o
r
e
a
c
h
m
o
d
el
(
YO
L
Ov
7
,
Y
L
O
v
8
,
Y
OL
O
v
8
-
C
B
AM
,
Y
O
L
O
v
8
R
es
N
e
tB
l
o
c
k
C
B
A
M
,
Y
OL
O
v
8
R
es
N
et
XB
l
o
c
k
C
B
A
M
,
YO
L
O
v
9
,
YO
L
O
v
9
-
C
B
AM
,
an
d
Y
O
L
O
v
1
1
)
.
A
l
e
a
r
n
i
n
g
c
u
r
v
e
r
e
f
e
r
s
t
o
a
p
l
o
t
th
a
t
i
l
l
u
s
t
r
at
e
s
h
o
w
a
m
o
d
e
l'
s
m
A
P
c
h
a
n
g
es
o
v
e
r
a
r
a
n
g
e
o
f
c
o
n
d
i
t
i
o
n
s
,
s
u
c
h
as
t
h
e
n
u
m
b
e
r
o
f
t
r
a
i
n
i
n
g
e
p
o
c
h
s
.
T
h
i
s
c
u
r
v
e
s
h
o
ws
h
o
w
t
h
e
m
AP
p
e
r
f
o
r
m
a
n
c
e
o
f
a
m
o
d
e
l
i
m
p
r
o
v
e
s
o
r
p
l
a
te
a
u
s
a
s
i
t
l
e
a
r
n
s
f
r
o
m
t
h
e
d
a
t
a
o
v
e
r
t
i
m
e
.
A
s
p
r
e
s
e
n
te
d
i
n
F
i
g
u
r
e
5
,
w
i
t
h
1
0
0
e
p
o
c
h
s
,
t
h
e
l
e
a
r
n
i
n
g
c
u
r
v
e
r
e
f
e
r
s
t
o
t
h
e
m
o
d
e
l
p
e
r
f
o
r
m
i
n
g
w
e
l
l
o
n
b
o
t
h
t
r
a
i
n
in
g
a
n
d
v
a
l
i
d
a
t
i
o
n
s
e
ts
w
i
t
h
o
u
t
u
n
d
e
r
f
i
t
t
i
n
g
o
r
o
v
e
r
f
i
t
ti
n
g
.
S
p
ec
i
f
i
c
a
ll
y
,
Fi
g
u
r
e
5
(
a
)
s
h
o
w
s
t
h
e
B
B
o
x
r
e
s
u
l
ts
,
w
h
i
l
e
F
i
g
u
r
e
5
(
b
)
p
r
e
s
e
n
t
s
t
h
e
m
a
s
k
r
e
s
u
l
t
s
.
T
h
e
m
A
P
p
e
r
f
o
r
m
a
n
c
e
s
h
o
u
l
d
c
o
n
v
e
r
g
e
a
n
d
s
t
a
b
i
li
z
e
as
t
h
e
m
o
d
e
l
l
e
a
r
n
s
t
h
e
d
a
t
a
w
e
ll
.
T
ab
le
3
.
T
h
e
r
esu
lts
o
f
YOL
O
ar
ch
itectu
r
es f
o
r
f
etal
o
r
g
an
s
eg
m
en
tatio
n
M
o
d
e
l
B
B
o
x
(
%)
M
a
s
k
(
%)
I
o
U
mA
P
5
0
mA
P
9
5
mA
P
5
0
mA
P
9
5
Y
O
LO
v
8
9
4
.
4
0
7
2
.
6
0
9
4
.
3
0
6
7
.
0
0
8
0
.
4
8
Y
O
LO
v
8
-
C
B
A
M
9
4
.
1
0
7
1
.
8
0
9
4
.
6
0
6
6
.
9
0
8
0
.
6
5
Y
O
LO
v
8
-
R
e
sN
e
t
B
l
o
c
k
C
B
A
M
9
3
.
7
0
8
8
.
8
8
9
3
.
8
0
8
9
.
0
0
8
1
.
0
1
Y
O
LO
v
8
-
R
e
sN
e
t
X
B
l
o
c
k
C
B
A
M
9
4
.
2
0
7
1
.
9
0
9
4
.
2
0
6
7
.
1
0
8
0
.
8
5
Y
O
LO
v
9
9
1
.
9
0
9
4
.
5
0
9
1
.
7
0
9
4
.
4
0
8
0
.
4
5
Y
O
LO
v
9
-
C
B
A
M
9
4
.
0
0
7
2
.
6
0
9
4
.
4
0
6
7
.
8
0
8
0
.
5
8
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
F
eta
l o
r
g
a
n
d
etec
tio
n
u
s
in
g
fea
tu
r
e
en
h
a
n
ce
men
t wi
th
a
tten
t
io
n
a
n
d
r
esid
u
a
l b
lo
ck
(
N
u
s
w
i
l B
ern
o
lia
n
)
1599
(
a)
(
b
)
Fig
u
r
e
5
.
T
h
e
lear
n
in
g
cu
r
v
es
o
f
YOL
O
in
s
tan
ce
s
eg
m
en
tati
o
n
with
m
AP5
0
a
n
d
m
AP9
5
o
f
(
a)
B
B
o
x
an
d
(
b
)
m
ask
T
h
e
p
er
f
o
r
m
an
ce
o
f
m
AP5
0
a
n
d
m
AP9
5
f
o
r
1
5
f
etal
o
r
g
an
o
b
jects
is
il
lu
s
tr
ated
in
Fig
u
r
es
6
an
d
7
,
r
esp
ec
tiv
ely
,
u
tili
zin
g
th
e
p
r
o
p
o
s
ed
YOL
Ov
9
ar
ch
itectu
r
e.
T
h
e
m
o
d
el
d
em
o
n
s
tr
ates
ex
ce
p
tio
n
al
p
er
f
o
r
m
an
c
e
in
d
etec
tin
g
th
e
ab
d
o
m
en
,
h
ea
d
,
an
d
r
ig
h
t
v
en
tr
icu
la
r
(
h
ea
r
t)
s
tr
u
ctu
r
es,
ac
h
iev
in
g
a
p
er
f
ec
t
1
0
0
%
m
AP5
0
f
o
r
th
ese
o
r
g
an
s
.
T
h
is
in
d
icate
s
th
at
th
e
YOL
Ov
9
m
o
d
el
ac
cu
r
a
tely
id
en
tifie
d
a
n
d
lo
ca
lized
t
h
ese
o
r
g
a
n
s
with
at
least
5
0
%
o
v
er
lap
,
r
esu
ltin
g
in
f
lawless
p
r
ec
is
io
n
a
n
d
r
ec
all
u
n
d
er
th
is
I
o
U
th
r
esh
o
ld
(
s
ee
Fig
u
r
e
6
)
.
Ho
wev
er
,
th
e
p
e
r
f
o
r
m
an
ce
in
d
etec
tin
g
th
e
u
m
b
ilical
v
ein
was
s
ig
n
if
ican
tly
lo
wer
,
with
an
m
AP5
0
o
f
o
n
ly
7
4
.
8
%.
T
h
is
d
is
cr
ep
an
c
y
h
ig
h
lig
h
ts
th
e
ch
allen
g
es
ass
o
ciate
d
with
ac
cu
r
ately
d
etec
tin
g
th
e
u
m
b
ilical
v
ein
,
wh
ich
m
ay
r
e
q
u
ir
e
f
u
r
th
e
r
r
ef
i
n
em
en
t o
f
th
e
m
o
d
el.
YOL
Ov
9
in
co
r
p
o
r
ates
ad
v
a
n
c
ed
m
ec
h
an
is
m
s
,
in
clu
d
in
g
p
r
o
g
r
am
m
a
b
le
g
r
a
d
ien
t
in
f
o
r
m
a
tio
n
(
PGI
)
an
d
th
e
g
en
er
alize
d
e
f
f
icien
t
lay
er
ag
g
r
e
g
atio
n
n
etwo
r
k
(
GE
L
AN)
.
T
h
ese
in
n
o
v
atio
n
s
m
itig
ate
in
f
o
r
m
atio
n
lo
s
s
in
d
ee
p
n
eu
r
al
n
etwo
r
k
s
,
th
er
eb
y
en
h
an
cin
g
ef
f
ic
ien
cy
,
ac
cu
r
ac
y
,
an
d
ad
ap
t
ab
ilit
y
[
2
6
]
,
[
2
7
]
.
Sp
ec
if
ically
,
PGI
m
ain
tain
s
cr
itical
in
f
o
r
m
atio
n
f
l
o
w
ac
r
o
s
s
n
etwo
r
k
lay
er
s
,
wh
er
ea
s
GE
L
AN
im
p
r
o
v
es
p
ar
am
eter
ef
f
icien
cy
an
d
r
e
d
u
ce
s
co
m
p
u
tatio
n
al
co
m
p
lex
ity
[
2
7
]
.
Fig
u
r
e
7
illu
s
tr
ates
th
e
p
er
f
o
r
m
an
ce
o
f
m
AP9
5
f
o
r
1
5
f
etal
o
r
g
an
o
b
jects.
Ov
er
all,
ea
ch
class
ac
h
iev
ed
a
b
o
v
e
9
0
%
m
AP9
5
;
h
o
wev
e
r
,
th
e
p
er
f
o
r
m
an
ce
was
n
o
tab
ly
af
f
ec
ted
b
y
th
e
d
etec
tio
n
o
f
t
h
e
u
m
b
ilical
v
ein
.
T
h
is
is
p
r
im
a
r
ily
b
ec
au
s
e
th
e
u
m
b
ilical
v
e
in
is
ty
p
ically
an
is
o
lated
f
in
d
in
g
,
y
et
it
ca
n
b
e
ass
o
ciate
d
with
o
th
er
f
etal
m
alf
o
r
m
atio
n
s
.
An
o
m
alies
r
elate
d
to
t
h
e
u
m
b
ilical
v
ei
n
m
ay
in
clu
d
e
th
e
p
er
s
is
ten
ce
o
f
em
b
r
y
o
lo
g
ical
s
tr
u
ctu
r
es,
ab
n
o
r
m
al
in
s
er
tio
n
p
o
in
ts
,
aty
p
ical
co
u
r
s
es,
an
d
th
e
p
r
esen
ce
o
f
s
u
p
er
n
u
m
e
r
ar
y
v
ess
els.
T
h
ese
co
m
p
lex
ities
ca
n
co
m
p
licat
e
d
etec
tio
n
an
d
s
eg
m
en
tatio
n
,
u
n
d
er
s
co
r
i
n
g
th
e
n
ee
d
f
o
r
im
p
r
o
v
ed
alg
o
r
ith
m
s
th
at
ca
n
b
etter
h
an
d
le
s
u
ch
ca
s
es.
T
h
e
p
r
e
d
icted
r
esu
lts
o
f
YOL
Ov
9
with
m
AP9
5
in
1
5
-
f
etal
o
r
g
an
s
eg
m
en
tatio
n
with
u
n
s
ee
n
s
et
ca
n
b
e
p
r
esen
ted
in
Fig
u
r
e
8
.
T
h
er
e
ar
e
ab
d
o
m
en
,
h
ea
d
,
b
r
ain
,
f
e
m
u
r
,
th
o
r
ax
a
n
d
h
ea
r
t
th
at
p
r
e
s
en
ted
as
d
if
f
er
en
t
co
lo
r
s
.
So
m
e
m
ask
s
in
an
at
o
m
ical
p
ar
ts
in
s
eg
m
en
tatio
n
(
e
.
g
.
,
a
b
d
o
m
en
(
Fig
u
r
e
8
(
a)
,
h
e
ad
(
Fig
u
r
e
8
(
b
)
,
a
n
d
b
r
ain
(
Fig
u
r
e
8
(
c
)
)
co
v
er
al
m
o
s
t
th
e
en
tire
s
tr
u
ctu
r
e,
wh
ile
o
th
er
s
(
e.
g
.
,
f
em
u
r
(
Fig
u
r
e
8
(
d
)
an
d
th
o
r
ax
(
Fig
u
r
e
8
(
e
)
)
ap
p
ea
r
m
o
r
e
r
ef
i
n
ed
.
T
h
is
s
u
g
g
ests
d
if
f
er
e
n
ce
s
in
s
eg
m
en
tatio
n
q
u
ality
ac
r
o
s
s
class
e
s
.
T
h
e
h
ea
r
t
im
ag
e
in
Fig
u
r
e
8
(
f
)
in
clu
d
es
n
u
m
er
ical
v
alu
es,
lik
ely
co
n
f
id
en
ce
s
co
r
es
f
o
r
d
if
f
er
e
n
t
p
ar
ts
(
R
A,
R
V,
an
d
AO)
.
T
h
e
s
eg
m
en
tatio
n
m
ask
s
ap
p
ea
r
well
-
alig
n
ed
with
an
ato
m
ical
s
tr
u
ctu
r
es,
in
d
icatin
g
s
tr
o
n
g
d
etec
tio
n
ac
cu
r
ac
y
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
15
,
No
.
2
,
Ap
r
il 2
0
2
6
:
1
5
9
3
-
1
6
0
4
1600
I
n
r
ec
en
t
y
ea
r
s
,
s
ev
er
al
s
tu
d
ies
h
av
e
im
p
lem
en
ted
th
e
DL
f
o
r
f
etal
o
r
g
an
s
eg
m
en
tatio
n
(
r
ef
er
to
T
ab
le
4
)
.
Do
n
g
et
a
l
.
[
2
8
]
p
r
o
p
o
s
ed
ag
g
r
eg
ated
r
esid
u
al
v
is
u
al
b
lo
ck
n
etwo
r
k
(
AR
VB
Net
)
to
d
etec
t f
etal
h
ea
r
t
with
9
3
.
5
2
%
m
AP
an
d
a
test
s
p
ee
d
o
f
1
0
1
f
r
am
e
-
p
er
-
s
ec
o
n
d
(
FP
s
)
.
Nu
r
m
ain
i
et
a
l
.
[
2
9
]
d
etec
t
th
e
f
etal
h
ea
r
t
o
b
ject
with
9
8
%
m
AP
u
s
in
g
R
esNet5
0
.
Als
o
,
th
ey
d
ev
elo
p
ed
th
e
in
s
tan
ce
s
eg
m
en
tatio
n
b
ased
o
n
a
m
ask
r
eg
io
n
-
b
ased
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
(
Ma
s
k
-
R
C
N
N)
ar
ch
itectu
r
e.
Ko
m
atsu
et
a
l
.
[
3
0
]
p
r
o
p
o
s
ed
s
u
p
er
v
is
ed
o
b
ject
d
etec
tio
n
wi
th
n
o
r
m
al
d
ata
o
n
ly
(
SONO)
,
b
ased
o
n
a
C
NN,
to
d
etec
t
ca
r
d
iac
s
u
b
s
tr
u
ctu
r
es
an
d
s
tr
u
ctu
r
al
a
b
n
o
r
m
alities
in
f
etal.
T
h
e
y
o
b
tain
ed
th
e
a
v
er
ag
e
p
r
ec
is
io
n
in
u
m
b
ilical
v
ein
an
d
s
to
m
ac
h
,
9
4
.
4
% a
n
d
9
6
.
9
%,
r
esp
ec
tiv
el
y
.
T
h
ey
also
h
av
e
d
etec
ted
th
e
f
etal
h
ea
r
t,
with
8
5
.
2
2
% m
AP.
Fig
u
r
e
6
.
T
h
e
p
er
f
o
r
m
a
n
ce
o
f
m
AP5
0
in
1
5
-
f
etal
o
r
g
an
o
b
je
cts
Fig
u
r
e
7
.
T
h
e
p
er
f
o
r
m
a
n
ce
o
f
m
AP9
5
in
1
5
-
f
etal
o
r
g
an
o
b
je
cts
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
F
eta
l o
r
g
a
n
d
etec
tio
n
u
s
in
g
fea
tu
r
e
en
h
a
n
ce
men
t wi
th
a
tten
t
io
n
a
n
d
r
esid
u
a
l b
lo
ck
(
N
u
s
w
i
l B
ern
o
lia
n
)
1601
(
a)
(
b
)
(
c)
(
d
)
(
e)
(f)
Fig
u
r
e
8
.
T
h
e
p
r
e
d
icted
r
esu
lts
o
f
YOL
Ov
9
with
m
AP9
5
in
1
5
-
f
etal
o
r
g
an
s
eg
m
e
n
tatio
n
with
u
n
s
ee
n
s
et
o
f
(
a)
ab
d
o
m
en
,
(
b
)
h
ea
d
,
(
c)
b
r
ai
n
,
(
d
)
f
em
u
r
,
(
e)
th
o
r
a
x
,
an
d
(
f
)
h
ea
r
t
T
ab
le
4
p
r
esen
ted
th
e
p
r
o
p
o
s
ed
s
tu
d
y
o
u
tp
e
r
f
o
r
m
ed
t
h
e
p
r
e
v
io
u
s
r
esu
lts
[
2
8
]
–
[
3
0
]
,
with
t
h
e
h
ig
h
est
9
4
.
5
0
%
m
AP
to
d
etec
t
m
u
lti
f
etal
s
u
ch
as
ab
d
o
m
en
,
b
r
ain
,
f
em
u
r
,
h
ea
d
,
th
o
r
ax
,
an
d
h
ea
r
t.
W
ith
th
is
s
tu
d
y
ap
p
r
o
ac
h
,
th
e
i
n
s
tan
ce
s
eg
m
e
n
tatio
n
o
f
YOL
Ov
9
d
eliv
e
r
s
p
ix
el
-
lev
el
m
ask
s
f
o
r
ea
ch
d
etec
ted
f
etal
o
r
g
a
n
o
b
ject.
I
t
wo
u
ld
allo
w
f
o
r
m
u
ch
m
o
r
e
ac
cu
r
ate
lo
ca
lizatio
n
o
f
f
etal
o
r
g
an
s
,
wh
ich
is
p
ar
ti
cu
lar
ly
im
p
o
r
tan
t
in
m
ed
ical
im
ag
in
g
wh
e
r
e
ex
ac
t
b
o
u
n
d
ar
ies
ar
e
cr
itical
f
o
r
d
iag
n
o
s
is
.
Fetal
o
r
g
an
s
ca
n
h
av
e
ir
r
eg
u
lar
an
d
co
m
p
lex
s
h
ap
es
th
at
a
r
e
d
i
f
f
ic
u
lt
to
ca
p
tu
r
e
with
ju
s
t
B
B
o
x
,
b
u
t
th
e
p
r
o
p
o
s
ed
YOL
Ov
9
ca
n
b
etter
ca
p
tu
r
e
th
e
f
in
e
d
etails
o
f
ea
c
h
o
r
g
an
'
s
s
t
r
u
ctu
r
e,
lea
d
in
g
t
o
m
o
r
e
d
etai
led
an
d
i
n
f
o
r
m
ativ
e
an
aly
s
es.
W
ith
o
u
ts
tan
d
in
g
p
er
f
o
r
m
an
ce
,
th
er
e
ar
e
lim
itat
io
n
s
in
th
is
s
tu
d
y
:
i)
lim
ited
t
o
YOL
Ov
7
,
YOL
Ov
8
,
YOL
Ov
9
,
a
n
d
YOL
Ov
1
1
ar
ch
itectu
r
es,
with
C
B
AM
,
R
esNet
an
d
R
e
s
NetX
b
ac
k
b
o
n
es;
an
d
ii)
th
is
s
tu
d
y
d
id
n
o
t
h
an
d
le
th
e
v
ar
iatio
n
s
o
f
f
etal
p
o
s
itio
n
an
d
o
r
ien
tatio
n
;
th
e
f
etal
f
r
eq
u
e
n
tly
ch
an
g
e
p
o
s
itio
n
o
r
o
r
ien
tatio
n
d
u
r
in
g
im
ag
e
ac
q
u
is
itio
n
,
wh
ich
p
o
s
es
a
ch
allen
g
e
f
o
r
o
b
ject
d
etec
tio
n
m
o
d
els,
p
o
ten
tially
lead
in
g
to
r
ed
u
ce
d
d
etec
tio
n
an
d
s
eg
m
en
tatio
n
ac
cu
r
ac
y
.
T
ab
le
4
.
T
h
e
b
en
c
h
m
ar
k
i
n
g
r
e
s
u
lts
o
f
f
etal
o
r
g
an
o
b
ject
d
ete
ctio
n
A
u
t
h
o
r
s
M
e
t
h
o
d
F
e
t
a
l
o
r
g
a
n
o
b
j
e
c
t
s
mA
P
(
%)
D
o
n
g
e
t
a
l
.
[
2
8
]
A
R
V
B
N
e
t
H
e
a
r
t
9
3
.
5
2
K
o
ma
t
su
e
t
a
l
.
[
3
0
]
S
O
N
O
-
C
N
N
U
mb
i
l
i
c
a
l
v
e
i
n
a
n
d
s
t
o
mac
h
,
h
e
a
r
t
8
8
.
7
Th
e
p
r
o
p
o
s
e
d
st
u
d
y
Y
O
LO
v
9
A
b
d
o
me
n
,
b
r
a
i
n
,
f
e
mu
r
,
h
e
a
d
,
t
h
o
r
a
x
,
a
n
d
h
e
a
r
t
9
4
.
5
0
4.
CO
NCLU
SI
O
N
T
h
e
an
aly
s
is
o
f
an
ato
m
ical
s
t
r
u
ctu
r
es
is
a
cr
u
cial
asp
ec
t
o
f
f
etal
o
r
g
an
s
cr
ee
n
in
g
,
as
it
o
f
f
er
s
clea
r
ev
id
en
ce
o
f
p
o
ten
tial
f
etal
m
alf
o
r
m
atio
n
s
.
Fetal
o
r
g
a
n
s
ar
e
f
u
lly
d
ev
el
o
p
ed
b
etwe
en
1
8
an
d
2
2
wee
k
s
o
f
p
r
eg
n
a
n
cy
,
s
u
ch
as
b
r
ain
,
f
ac
e,
n
ec
k
,
c
h
est/
h
ea
r
t,
ab
d
o
m
en
,
s
k
eleta
l
s
y
s
tem
,
p
lace
n
ta,
u
m
b
ilical
co
r
d
,
g
en
italia,
an
d
ce
r
v
ix
.
T
h
is
s
tu
d
y
p
r
o
p
o
s
ed
an
in
s
tan
ce
s
eg
m
en
tatio
n
b
ased
o
n
co
m
p
u
ter
-
ai
d
ed
DL
f
o
r
1
5
-
f
etal
o
r
g
an
o
b
jects,
in
clu
d
e
ab
d
o
m
en
,
b
r
ain
,
f
em
u
r
,
h
ea
d
,
t
h
o
r
a
x
,
an
d
h
ea
r
t.
T
h
is
s
tu
d
y
co
m
p
a
r
ed
v
ar
io
u
s
YOL
O
ar
ch
itectu
r
es,
in
clu
d
in
g
YOL
Ov
7
,
YOL
Ov
8
,
YOL
Ov
9
,
an
d
YOL
Ov
1
1
.
T
h
e
r
esu
lts
s
h
o
wed
th
at
YOL
Ov
9
o
u
tp
er
f
o
r
m
ed
YOL
Ov
7
,
YO
L
Ov
8
,
an
d
YOL
Ov
1
1
,
ac
h
ie
v
in
g
m
AP5
0
a
n
d
m
AP9
5
s
co
r
es
o
f
9
1
.
9
0
%
an
d
9
4
.
5
0
%,
r
esp
ec
tiv
ely
.
T
h
ese
f
in
d
in
g
s
in
d
icate
th
at
YOL
O
v
9
s
u
r
p
ass
es
p
r
ev
io
u
s
r
elate
d
wo
r
k
,
wh
ich
was
lim
ited
to
class
if
y
in
g
f
etal
o
r
g
an
s
.
B
ey
o
n
d
p
er
f
o
r
m
an
ce
im
p
r
o
v
em
e
n
ts
,
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
h
as
th
e
p
o
ten
tial
to
s
u
p
p
o
r
t
clin
ical
f
etal
o
r
g
an
s
cr
ee
n
in
g
b
y
ass
is
tin
g
clin
ician
s
in
ac
c
u
r
ately
l
o
ca
lizin
g
m
u
ltip
l
e
an
ato
m
ical
s
tr
u
ctu
r
es,
th
er
eb
y
r
ed
u
cin
g
o
p
er
ato
r
d
ep
e
n
d
e
n
cy
a
n
d
im
p
r
o
v
in
g
s
cr
ee
n
in
g
ef
f
icien
cy
.
Su
ch
a
s
y
s
tem
m
ay
s
er
v
e
as
a
co
m
p
lem
en
tar
y
to
o
l
in
r
o
u
tin
e
u
ltra
s
o
u
n
d
ex
a
m
in
atio
n
s
,
p
ar
ticu
lar
l
y
in
h
ig
h
-
wo
r
k
lo
ad
clin
ical
en
v
ir
o
n
m
e
n
ts
wh
er
e
r
ap
id
an
d
c
o
n
s
is
ten
t
ass
es
s
m
en
ts
ar
e
r
eq
u
ir
ed
.
Ne
v
er
th
ele
s
s
,
th
is
s
tu
d
y
h
as
s
ev
er
al
lim
itatio
n
s
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
was
tr
ain
ed
an
d
e
v
alu
ated
o
n
a
lim
ited
d
ataset,
wh
ich
m
ay
a
f
f
ec
t
its
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
15
,
No
.
2
,
Ap
r
il 2
0
2
6
:
1
5
9
3
-
1
6
0
4
1602
g
en
er
aliza
b
ilit
y
ac
r
o
s
s
d
if
f
er
e
n
t
u
ltra
s
o
u
n
d
d
e
v
ices,
im
ag
in
g
p
r
o
to
c
o
ls
,
an
d
p
atien
t
p
o
p
u
l
atio
n
s
.
Fu
tu
r
e
wo
r
k
will
f
o
cu
s
o
n
ex
p
an
d
i
n
g
th
e
d
ataset,
im
p
r
o
v
in
g
r
o
b
u
s
tn
ess
ac
r
o
s
s
d
iv
er
s
e
clin
ical
s
ce
n
ar
io
s
,
an
d
in
teg
r
atin
g
th
e
p
r
o
p
o
s
ed
f
r
a
m
ewo
r
k
in
to
r
ea
l
-
tim
e
u
ltra
s
o
u
n
d
s
y
s
tem
s
f
o
r
co
m
p
r
eh
en
s
iv
e
clin
ical
v
alid
atio
n
.
ACK
NO
WL
E
DG
M
E
N
T
S
T
h
is
wo
r
k
was
s
u
p
p
o
r
ted
b
y
th
e
I
n
tellig
en
t
Sy
s
tem
R
esear
ch
Gr
o
u
p
(
I
Sy
s
R
G)
,
Un
iv
er
s
itas
Sriwijay
a.
W
e
g
r
atef
u
lly
ac
k
n
o
wled
g
e
Dr
.
Mo
h
a
m
m
ad
Ho
e
s
in
C
en
tr
al
Gen
er
al
Ho
s
p
ital
f
o
r
th
eir
in
v
alu
a
b
le
co
n
tr
ib
u
tio
n
s
an
d
s
u
p
p
o
r
t t
h
r
o
u
g
h
o
u
t th
is
p
r
o
ject.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
T
h
is
r
esear
ch
was
f
u
n
d
e
d
b
y
Un
iv
e
r
s
itas
Sriwijay
a
u
n
d
er
Pro
f
ess
io
n
al
R
esear
ch
Gr
an
t
No
.
0
0
1
6
/UN9
/SK.L
P2
M.
PT/2
0
2
4
,
wh
ich
p
r
o
v
id
ed
t
h
e
n
ec
ess
ar
y
r
eso
u
r
ce
s
a
n
d
to
o
ls
to
d
ev
el
o
p
an
d
v
alid
ate
o
u
r
DL
m
o
d
el
f
o
r
th
e
ass
ess
m
en
t o
f
f
etal
h
ea
r
t e
n
lar
g
em
e
n
t.
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
T
h
is
jo
u
r
n
al
u
s
es
th
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT)
to
r
ec
o
g
n
ize
in
d
iv
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
th
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
llab
o
r
atio
n
.
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
Nu
s
wil Ber
n
o
lian
✓
✓
✓
✓
✓
Sit
i N
u
r
m
ain
i
✓
✓
✓
Ad
e
I
r
ian
i Sap
itri
✓
✓
✓
✓
An
n
is
a
Dar
m
awa
h
y
u
n
i
✓
✓
Mu
h
am
m
ad
Na
u
f
al
R
ac
h
m
atu
llah
✓
✓
✓
B
am
b
an
g
T
u
tu
k
o
✓
✓
Fird
au
s
Fird
au
s
✓
✓
✓
C
:
C
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
M
:
M
e
t
h
o
d
o
l
o
g
y
So
:
So
f
t
w
a
r
e
Va
:
Va
l
i
d
a
t
i
o
n
Fo
:
Fo
r
mal
a
n
a
l
y
s
i
s
I
:
I
n
v
e
s
t
i
g
a
t
i
o
n
R
:
R
e
so
u
r
c
e
s
D
:
D
a
t
a
C
u
r
a
t
i
o
n
O
:
W
r
i
t
i
n
g
-
O
r
i
g
i
n
a
l
D
r
a
f
t
E
:
W
r
i
t
i
n
g
-
R
e
v
i
e
w
&
E
d
i
t
i
n
g
Vi
:
Vi
su
a
l
i
z
a
t
i
o
n
Su
:
Su
p
e
r
v
i
s
i
o
n
P
:
P
r
o
j
e
c
t
a
d
mi
n
i
st
r
a
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
si
t
i
o
n
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
T
h
e
au
th
o
r
s
d
ec
lar
e
th
at
th
e
y
h
av
e
n
o
co
n
f
lict o
f
in
ter
est.
DATA AV
AI
L
AB
I
L
I
T
Y
T
h
e
d
ata
th
at
s
u
p
p
o
r
t
th
e
f
in
d
in
g
s
o
f
th
is
s
tu
d
y
ar
e
av
ailab
l
e
f
r
o
m
th
e
co
r
r
esp
o
n
d
in
g
au
t
h
o
r
,
[
SN
]
,
u
p
o
n
r
ea
s
o
n
ab
le
r
eq
u
est.
RE
F
E
R
E
NC
E
S
[
1
]
S
.
S
h
r
e
st
h
a
,
A
.
Th
a
k
u
r
,
S
.
G
o
y
a
l
,
P
.
G
a
r
g
,
a
n
d
N
.
K
l
e
r
,
“
G
r
o
w
t
h
c
h
a
r
t
s
i
n
n
e
o
n
a
t
e
s
,
”
C
u
rr
e
n
t
Me
d
i
c
i
n
e
Re
se
a
r
c
h
a
n
d
Pr
a
c
t
i
c
e
,
v
o
l
.
6
,
n
o
.
2
,
p
p
.
7
9
–
8
4
,
M
a
r
.
2
0
1
6
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
mr
p
.
2
0
1
6
.
0
3
.
0
0
9
.
[
2
]
R
.
R
.
Ze
g
a
r
r
a
a
n
d
T.
G
h
i
,
“
U
se
o
f
a
r
t
i
f
i
c
i
a
l
i
n
t
e
l
l
i
g
e
n
c
e
a
n
d
d
e
e
p
l
e
a
r
n
i
n
g
i
n
f
e
t
a
l
u
l
t
r
a
s
o
u
n
d
i
ma
g
i
n
g
,
”
U
l
t
r
a
so
u
n
d
i
n
O
b
s
t
e
t
ri
c
s
& G
y
n
e
c
o
l
o
g
y
,
v
o
l
.
6
2
,
n
o
.
2
,
p
p
.
1
8
5
–
1
9
4
,
A
u
g
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
2
/
u
o
g
.
2
6
1
3
0
.
[
3
]
P
.
S
r
i
d
a
r
e
t
a
l
.
,
“
A
u
t
o
m
a
t
i
c
i
d
e
n
t
i
f
i
c
a
t
i
o
n
o
f
m
u
l
t
i
p
l
e
p
l
a
n
e
s
o
f
a
f
e
t
a
l
o
r
g
a
n
f
r
o
m
2
D
u
l
t
r
a
so
u
n
d
i
ma
g
e
s,”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
MICC
AI
W
o
rks
h
o
p
Pe
r
i
n
a
t
a
l
,
Pre
t
e
r
m
Pa
e
d
i
a
t
ri
c
I
m
a
g
e
A
n
a
l
(
PI
P
PI
)
,
2
0
1
6
,
p
p
.
1
–
1
0
.
[
4
]
B
.
R
a
h
m
a
t
u
l
l
a
h
a
n
d
J.
A
.
N
o
b
l
e
,
“
A
n
a
t
o
m
i
c
a
l
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
i
n
f
e
t
a
l
u
l
t
r
a
so
u
n
d
:
c
o
m
p
u
t
e
r
-
e
x
p
e
r
t
a
g
r
e
e
me
n
t
s
,
”
i
n
B
i
o
m
e
d
i
c
a
l
I
n
f
o
rm
a
t
i
c
s
a
n
d
T
e
c
h
n
o
l
o
g
y
(
AC
BI
T
2
0
1
3
)
,
2
0
1
4
,
p
p
.
2
0
7
–
2
1
8
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
6
4
2
-
5
4
1
2
1
-
6
_
1
8
.
[
5
]
M
.
C
.
-
C
a
s
t
r
e
j
o
n
,
T.
Ja
n
ss
o
n
,
a
n
d
T.
L.
P
o
w
e
l
l
,
“
F
e
t
a
l
m
a
t
e
r
n
a
l
u
n
i
t
,
”
i
n
En
c
y
c
l
o
p
e
d
i
a
o
f
Re
p
r
o
d
u
c
t
i
o
n
,
2
0
1
8
,
p
p
.
4
7
7
–
4
8
3
,
d
o
i
:
1
0
.
1
0
1
6
/
B
9
7
8
-
0
-
12
-
8
0
1
2
3
8
-
3
.
6
4
6
7
4
-
3.
[
6
]
E.
M
e
d
j
e
d
o
v
i
c
e
t
a
l
.
,
“
A
r
t
i
f
i
c
i
a
l
i
n
t
e
l
l
i
g
e
n
c
e
a
s
a
n
e
w
a
n
sw
e
r
t
o
o
l
d
c
h
a
l
l
e
n
g
e
s
i
n
m
a
t
e
r
n
a
l
-
f
e
t
a
l
me
d
i
c
i
n
e
a
n
d
o
b
s
t
e
t
r
i
c
s,
”
T
e
c
h
n
o
l
o
g
y
a
n
d
H
e
a
l
t
h
C
a
re
,
v
o
l
.
3
2
,
n
o
.
3
,
p
p
.
1
2
7
3
–
1
2
8
7
,
M
a
y
2
0
2
4
,
d
o
i
:
1
0
.
3
2
3
3
/
TH
C
-
2
3
1
4
8
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.