I
A
E
S
I
n
t
e
r
n
at
io
n
al
Jou
r
n
al
of
A
r
t
if
ic
ia
l
I
n
t
e
ll
ig
e
n
c
e
(
I
J
-
AI
)
V
ol
.
14
, N
o.
3
,
J
une
2025
, pp.
2132
~
2145
I
S
S
N
:
2252
-
8938
,
D
O
I
:
10.11591/
ij
a
i.
v
14
.i
3
.pp
2132
-
2145
2132
Jou
r
n
al
h
om
e
page
:
ht
tp
:
//
ij
ai
.
ia
e
s
c
or
e
.c
om
E
ar
l
y d
e
t
e
c
t
i
on
an
d
c
l
ass
i
f
i
c
at
i
on
of
b
on
e
m
ar
r
ow
c
h
an
ge
s i
n
l
u
m
b
ar
ve
r
t
e
b
r
a
e
u
si
n
g m
ac
h
i
n
e
l
e
ar
n
i
n
g t
e
c
h
n
i
q
u
e
s
Y
as
ir
H
u
s
s
e
in
S
h
ak
ir
1
, T
io
n
g S
ie
h
K
io
n
g
1,
2
, C
h
ai
P
h
in
g C
h
e
n
1,
3
1
C
ol
l
e
ge
of
G
r
a
dua
t
e
S
t
udi
e
s
, U
ni
ve
r
s
i
t
i
T
e
na
ga
N
a
s
i
ona
l
(
U
N
I
T
E
N
)
, K
a
j
a
ng,
M
a
l
a
ys
i
a
2
I
ns
t
i
t
ut
e
of
S
us
t
a
i
na
bl
e
E
ne
r
gy, U
ni
ve
r
s
i
t
i
T
e
na
ga
N
a
s
i
ona
l
(
U
N
I
T
E
N
)
, K
a
j
a
n
g, M
a
l
a
ys
i
a
3
D
e
pa
r
t
m
e
nt
E
l
e
c
t
r
i
c
a
l
a
nd E
l
e
c
t
r
oni
c
s
E
ngi
ne
e
r
i
ng, U
ni
ve
r
s
i
t
i
T
e
na
ga
N
a
s
i
ona
l
(
U
N
I
T
E
N
)
, K
a
j
a
ng, M
a
l
a
ys
i
a
A
r
t
ic
le
I
n
f
o
A
B
S
T
R
A
C
T
A
r
ti
c
le
h
is
to
r
y
:
R
e
c
e
iv
e
d
J
un 8, 2024
R
e
vi
s
e
d
F
e
b 14, 2025
A
c
c
e
pt
e
d
M
a
r
15, 2025
Bone
marrow
changes
in
lumbar
vertebrae
(BMCLVB)
have
emerged
as
a
significant
correlation
of
chronic
low
back
pain
(CLBP)
severity,
esp
ecially
in
patients
with
comorbid
conditions
like
HIV,
osteoporosis,
and
cancer.
Identifying
these
correlations
not
only
aids
governments
and
health
insurance
providers
but
also
facilitate
s
early
treatment
for
those
a
t
risk.
However,
challenges
lurk
due
to
the
unavailability
and
quality
of
hea
lthcare
data.
The
collaborat
ion
between
data
science
and
artificial
intell
igence,
particularly
machine
learning
(ML)
,
has
propelled
biomedical
re
search
forward.
So
far,
accessing
and
processing
hospital
and
clinical
data
r
emains
a
hurdle.
In
doing
so
it
aims
to
provide
an
opportunity
for
early
interv
ention
and
treatment.
In
addition,
the
goal
of
the
current
study
was
to
ove
rcome
data
shortcomings
using
advanced
ML
techniques
to
unlock
co
mplex
magnetic
resonanc
e
imaging
(
MRI
)
features.
We
believe
that
extendi
ng
the
dataset
with
that
obtained
from
an
Iraqi
hospital
will
not
only
as
sist
in
diagnosing
BMCLVB
but
also
fill
the
gap
between
data
science
and
healthcare.
Above
all,
the
upgrade
is
intended
to
empower
bio
medical
research and increase t
he chances of succes
sful pati
ent treatment
.
K
e
y
w
o
r
d
s
:
B
one
m
a
r
r
ow
pr
e
di
c
ti
on
D
im
e
ns
io
na
li
ty
r
e
duc
ti
on
F
e
a
tu
r
e
e
xt
r
a
c
ti
on
H
e
a
lt
hc
a
r
e
M
a
c
hi
ne
l
e
a
r
ni
ng
M
a
gne
ti
c
r
e
s
on
a
nc
e
i
m
a
gi
ng
This is an
open
acce
ss artic
le unde
r the
CC BY
-
SA
license.
C
or
r
e
s
pon
di
n
g A
u
th
or
:
Y
a
s
ir
H
us
s
e
in
S
ha
ki
r
C
ol
le
ge
of
G
r
a
dua
te
S
tu
di
e
s
(
C
O
G
S
)
,
U
ni
ve
r
s
it
i
T
e
na
ga
N
a
s
io
na
l
(
U
N
I
T
E
N
)
K
a
ja
ng, M
a
la
y
s
ia
E
m
a
il
:
ya
s
s
e
r
he
s
s
e
in
19855@
gm
a
il
.c
om
1.
I
N
T
R
O
D
U
C
T
I
O
N
T
he
s
tu
dy
in
ve
s
ti
ga
te
s
th
e
e
a
r
ly
de
te
c
to
r
a
nd
c
la
s
s
if
ie
r
bone
m
a
r
r
ow
c
ha
nge
s
in
lu
m
ba
r
v
e
r
te
br
a
e
(
B
M
C
L
V
B
)
us
in
g
m
a
gne
ti
c
r
e
s
ona
nc
e
im
a
gi
ng
(
M
R
I
)
da
ta
a
nd
a
ddr
e
s
s
in
g
a
ke
y
c
li
ni
c
a
l
c
ha
ll
e
ng
e
in
di
a
gnos
in
g
lu
m
ba
r
ve
r
te
br
a
e
.
B
y
im
pr
ovi
ng
c
la
s
s
if
ic
a
ti
on
te
c
hni
que
s
,
th
e
r
e
s
e
a
r
c
h
lo
oks
to
im
pr
ove
th
e
di
a
gnos
ti
c
a
c
c
ur
a
c
y
a
nd
out
c
om
e
of
pa
ti
e
nt
s
.
T
hi
s
s
tu
dy
a
ls
o
a
ddr
e
s
s
e
s
a
n
im
por
ta
nt
ga
p
on
th
e
li
te
r
a
tu
r
e
(
B
M
C
L
V
B
)
pa
r
ti
c
ul
a
r
ly
in
th
e
a
ppl
ic
a
ti
on
of
m
a
c
hi
ne
le
a
r
ni
ng
(
M
L
)
to
c
la
s
s
if
ic
a
ti
on,
a
f
ie
ld
c
ur
r
e
nt
ly
c
ons
tr
a
in
e
d
by
th
e
la
c
k
of
publ
ic
ly
a
va
il
a
bl
e
da
ta
s
e
ts
.
B
M
C
L
V
B
w
a
s
r
e
la
te
d
to
c
hr
oni
c
lo
w
ba
c
k
pa
in
(
C
L
B
P
)
,
a
nd
i
t
w
a
s
a
s
s
oc
ia
te
d
w
it
h
in
c
r
e
a
s
e
d
di
s
e
a
s
e
s
e
v
e
r
it
y
in
th
os
e
w
ho
ha
ve
H
I
V
a
nd
os
te
opor
os
is
th
e
r
e
f
or
e
,
B
M
C
L
V
B
m
a
y
s
e
r
ve
a
s
a
n
im
a
gi
ng
bi
om
a
r
ke
r
f
or
in
di
vi
dua
ls
w
it
h
di
f
f
e
r
e
nt
s
ke
le
ta
l,
m
e
ta
bol
ic
,
a
nd
he
m
a
to
lo
gi
c
a
l
il
ln
e
s
s
e
s
a
s
a
s
ur
r
oga
te
in
th
e
r
a
py
r
e
s
pons
e
[
1]
,
[
2
]
.
I
t
is
pos
s
ib
le
f
or
a
f
r
a
c
tu
r
e
in
vol
vi
ng
th
e
ve
r
ba
l
be
h
a
vi
or
(
VB
)
or
vi
r
tu
a
l
a
s
s
i
s
ta
nt
(
VA
)
to
c
a
us
e
ne
r
ve
pr
e
s
s
ur
e
w
h
e
n
th
e
s
pa
c
e
b
e
twe
e
n
th
e
ve
r
te
br
a
l
bodi
e
s
is
r
e
duc
e
d
a
nd
th
e
in
te
r
ve
r
te
br
a
l
di
s
c
s
(
I
V
D
)
m
a
y
s
li
p
or
he
r
ni
a
te
.
A
s
a
r
e
s
ul
t,
bul
gi
ng
or
s
li
ppe
d
di
s
c
s
put
s
tr
e
s
s
on
lo
c
a
l
ne
r
ve
s
,
pr
oduc
in
g
ti
ngl
in
g
s
e
ns
a
ti
on
in
th
e
le
gs
a
nd,
in
th
e
oc
c
ur
r
e
nc
e
of
ne
r
ve
in
ju
r
y,
lo
w
e
r
body
he
lp
le
s
s
ne
s
s
[
3]
,
[
4]
.
T
he
bone
m
a
r
r
ow
is
onl
y
in
c
ha
r
ge
of
oxyge
n
tr
a
ns
por
t,
c
oa
gul
a
ti
on,
a
nd
im
m
uni
ty
.
I
t
c
om
pr
is
e
s
he
m
a
to
poi
e
ti
c
c
e
ll
s
(
m
e
ga
ka
r
yoc
yt
e
s
,
ly
m
phoi
d,
e
r
yt
hr
oc
yt
e
s
a
nd
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
E
ar
ly
de
te
c
ti
on and c
la
s
s
if
ic
at
io
n of
bone
m
ar
r
o
w
c
hange
s
i
n l
um
bar
v
e
r
t
e
br
ae
…
(
Y
as
ir
H
us
s
e
in
Shak
i
r
)
2133
m
ye
lo
id
c
e
ll
s
,
a
nd
pl
ur
ip
ot
e
nt
m
e
s
e
nc
hym
a
l
s
te
m
c
e
ll
s
)
th
a
t
in
te
r
a
c
t
w
it
h
bone
m
a
r
r
ow
a
di
poc
yt
e
s
in
a
w
e
ll
-
or
ga
ni
z
e
s
ig
na
l
w
or
k s
uppl
ie
d by bony tr
a
be
c
ul
a
e
bor
de
r
e
d by
a
f
ib
r
ous
r
e
ti
c
ul
um
[
3]
.
M
R
I
gi
ve
s
a
non
-
in
va
s
iv
e
vi
e
w
in
to
th
e
bone
m
a
r
r
ow
by
m
a
ppi
ng
lo
c
a
l,
r
e
gi
ona
l,
or
s
ys
te
m
ic
a
lt
e
r
a
ti
ons
in
m
a
r
r
ow
w
a
te
r
pos
it
io
n
due
to
it
s
une
qua
le
d
s
e
ns
it
iv
it
y
to
th
e
p
r
e
s
e
nc
e
of
f
a
t
a
nd
th
e
r
e
la
ti
ve
ly
lo
w
e
f
f
e
c
t
of
m
in
e
r
a
li
z
e
d
ti
s
s
u
e
on
M
R
I
s
ig
na
l
[
5]
,
[
6]
.
T
he
p
a
ti
e
nt
r
e
c
e
iv
e
s
a
n
a
ppr
opr
ia
te
s
ur
gi
c
a
l
th
e
r
a
py
tr
e
a
tm
e
nt
,
s
uc
h
a
s
de
c
om
pr
e
s
s
io
n,
to
r
e
li
e
ve
pr
e
s
s
ur
e
on
th
e
ne
r
ve
s
,
c
ons
e
que
nt
ly
im
pr
ovi
ng
th
e
pa
ti
e
nt
'
s
qua
li
ty
of
li
f
e
(
Q
oL
)
[
7
]
.
T
he
di
a
gnos
ti
c
pr
oc
e
s
s
of
pe
r
s
is
te
nt
lo
w
ba
c
k
pa
in
ta
k
e
s
e
xt
e
ns
iv
e
e
f
f
or
t,
ye
t
im
a
gi
ng
te
s
ts
li
k
e
X
-
r
a
ys
,
M
R
I
s
c
a
ns
,
a
nd
C
T
s
c
a
ns
h
e
lp
t
he
c
li
ni
c
ia
n
to
obt
a
in
s
p
e
c
if
ic
s
pi
na
l
a
il
m
e
nt
in
f
or
m
a
ti
on
a
nd
r
e
la
te
th
is
da
ta
to
pa
in
s
ym
pt
om
s
[
8]
,
[
9]
.
C
u
r
r
e
nt
ly
,
th
e
y
a
r
e
two
m
e
th
ods
e
xi
s
t
to
da
y
f
or
s
pi
na
l
s
ur
gi
c
a
l
a
nd
r
a
di
ol
ogy
di
a
gnos
is
e
it
he
r
th
r
ough
hum
a
n
e
xpe
r
ti
s
e
or
te
c
hni
c
a
l
c
om
put
e
r
-
ba
s
e
d
im
a
ge
a
na
ly
s
is
a
ppr
oa
c
he
s
.
T
he
di
a
gnos
ti
c
r
e
por
t
pr
ovi
de
s
in
f
or
m
a
ti
on
a
bout
bone
m
a
r
r
ow
c
ha
nge
s
in
lu
m
ba
r
ve
r
te
br
a
l
a
nd
a
ls
o
c
onf
ir
m
s
th
e
di
a
gnos
is
w
hi
le
th
e
s
pi
na
l
s
ur
ge
ons
a
na
ly
z
e
pa
ti
e
nt
s
th
r
ough
c
li
ni
c
a
l
m
e
a
ns
ba
s
e
d
on
di
s
pl
a
ye
d
s
ym
pt
om
s
.
T
h
e
a
s
s
e
s
s
m
e
nt
of
s
pi
n
a
l
a
li
gn
m
e
nt
a
nd
s
pi
na
l
de
f
or
m
a
ti
on
a
lo
ng
w
it
h
s
pi
na
l
ba
la
nc
e
s
ta
nd
s
a
s
a
c
r
uc
ia
l
a
s
pe
c
t
be
c
a
us
e
s
c
ie
nt
if
ic
e
vi
de
nc
e
p
r
ove
s
th
a
t
va
r
ia
ti
ons
in
s
pi
na
l
pos
tu
r
e
di
r
e
c
tl
y
im
pa
c
t
lu
m
ba
r
s
tr
uc
tu
r
e
s
ta
bi
li
ty
a
s
w
e
ll
a
s
pr
e
s
s
ur
e
di
s
tr
ib
ut
io
n
[
10]
.
O
ve
r
ti
m
e
,
r
e
s
e
a
r
c
he
r
s
ha
ve
di
s
c
ov
e
r
e
d
a
l
in
k be
twe
e
n c
ha
nge
d
s
pi
na
l
a
li
gnm
e
nt
a
nd l
um
ba
go
[
11]
.
M
e
d
i
c
a
l
im
a
gi
n
g
ha
s
a
d
v
a
n
c
e
d
s
ig
ni
f
i
c
a
n
tl
y
in
r
e
c
e
n
t
ye
a
r
s
,
w
it
h
M
R
I
b
e
in
g
a
k
e
y
c
o
m
p
on
e
nt
i
n
th
e
d
i
a
gn
o
s
i
s
a
nd
e
v
a
lu
a
ti
on
of
m
a
ny
il
ln
e
s
s
e
s
.
T
h
e
c
a
t
e
go
r
i
z
a
t
io
n
o
f
B
M
C
L
V
B
i
s
c
r
u
c
ia
l
b
ut
di
f
f
i
c
ul
t
a
s
s
i
gn
m
e
n
t
a
m
o
ng
a
r
e
a
s
of
i
nt
e
r
e
s
t
in
m
e
di
c
a
l
i
m
a
g
in
g
. T
h
e
r
e
i
s
a
c
l
e
a
r
v
a
c
uu
m
i
n t
h
e
b
o
dy
o
f
l
i
te
r
a
t
ur
e
d
e
s
p
it
e
s
i
gn
if
i
c
a
n
c
e
o
f
th
i
s
f
i
e
ld
.
A
l
t
ho
ug
h
lu
m
ba
r
v
e
r
t
e
br
a
l
i
m
a
g
in
g
a
n
d
c
l
a
s
s
if
ic
a
t
io
n
h
a
v
e
b
e
e
n
s
ub
j
e
c
t
of
s
e
v
e
r
a
l
r
e
s
e
a
r
c
h
i
n
v
e
s
t
ig
a
t
i
on
s
,
th
or
ou
g
h
w
o
r
k
o
n
B
M
C
L
V
B
c
l
a
s
s
i
f
i
c
a
t
io
n
i
s
c
on
s
p
i
c
u
ou
s
ly
l
a
c
ki
ng
.
F
u
r
t
h
e
r
m
or
e
,
c
on
s
tr
uc
ti
o
n
a
n
d
a
s
s
e
s
s
m
e
nt
o
f
m
od
e
l
s
f
or
th
i
s
p
a
r
t
i
c
u
l
a
r
jo
b
a
r
e
m
a
d
e
m
o
r
e
di
f
f
i
c
ul
t
by
th
e
l
a
c
k
of
a
s
p
e
c
i
a
li
z
e
d,
p
ub
li
c
l
y
a
c
c
e
s
s
i
bl
e
d
a
t
a
ba
s
e
.
T
h
e
e
s
t
a
bl
i
s
h
m
e
n
t
of
th
i
s
n
e
w
d
a
ta
s
e
t
r
e
pr
e
s
e
n
t
s
a
s
u
b
s
t
a
nt
ia
l
a
dv
a
n
c
e
m
e
n
t
in
t
he
f
i
e
ld
o
f
B
M
C
L
V
B
u
s
i
n
g
d
e
e
p
l
e
a
r
ni
ng
te
c
hn
iq
u
e
s
.
T
h
e
m
a
i
n
n
ov
e
lt
y
a
nd
c
o
n
tr
i
bu
ti
o
n
s
of
o
ur
r
e
s
e
a
r
c
h
a
r
e
a
s
f
ol
l
ow
s
.
T
hi
s
s
tu
dy
a
im
s
a
t
th
e
e
a
r
ly
de
te
c
ti
on
a
nd
c
l
a
s
s
if
ic
a
ti
on
of
B
M
C
L
V
B
in
or
de
r
to
e
nha
nc
e
th
e
di
a
gnos
is
a
nd
c
la
s
s
if
iy
of
th
e
di
s
e
a
s
e
.
B
e
s
id
e
s
c
r
e
a
ti
ng
a
ne
w
M
R
I
da
ta
s
e
t
a
nd
d
a
ta
f
or
B
M
C
L
V
B
c
la
s
s
if
ic
a
ti
on,
th
is
w
or
k
a
l
s
o
pr
opos
e
s
a
n
nove
l,
a
ut
om
a
t
e
d
a
ppr
oa
c
h
f
or
B
M
C
L
V
B
c
la
s
s
if
ic
a
ti
on.
T
he
s
tr
a
t
e
g
y
in
c
l
ud
e
s
M
R
I
-
ba
s
e
d
d
a
t
a
a
c
qui
s
i
ti
o
n
w
it
h
hi
s
to
gr
a
m
of
or
ie
nt
e
d
gr
a
di
e
nt
s
(
HOG
)
in
f
e
a
tu
r
e
e
xt
r
a
c
ti
on
a
nd
f
ur
th
e
r
r
e
du
c
t
io
n
w
it
h
pr
i
nc
ip
a
l
c
om
p
on
e
n
t
a
n
a
l
y
s
i
s
(
P
C
A
)
,
li
n
e
a
r
di
s
c
r
im
in
a
nt
a
n
a
l
ys
i
s
(
L
D
A
)
a
n
d
in
d
e
p
e
n
de
nt
c
om
pon
e
n
t
a
na
ly
s
i
s
(
I
C
A
)
a
nd
a
l
s
o u
s
e
d
m
ul
ti
p
le
c
l
a
s
s
if
i
e
r
s
a
r
e
e
m
pl
oy
e
d
to
e
nh
a
n
c
e
c
la
s
s
if
ic
a
ti
on
a
c
c
ur
a
c
y
w
i
th
th
e
go
a
l
of
e
s
t
a
bl
is
hi
n
g
B
M
C
L
V
B
a
s
a
r
obu
s
t
a
c
c
e
s
s
ib
le
bi
om
a
r
k
e
r
in
c
l
in
i
c
a
l
s
e
tt
in
g
s
.
I
n
th
e
pr
e
s
e
nt
pa
pe
r
,
th
e
c
ont
e
nt
is
di
vi
de
d
in
to
s
ix
s
e
c
ti
ons
.
S
e
c
ti
on
2
w
il
l
di
s
c
u
s
s
pr
io
r
s
tu
di
e
s
in
di
s
e
a
s
e
pr
e
di
c
ti
on
or
c
la
s
s
if
ic
a
ti
on
a
nd
pr
io
r
w
or
k,
s
e
c
ti
on
3
gi
ve
s
a
br
ie
f
te
c
hni
c
a
l
ba
c
kgr
ound
of
th
e
m
e
th
ods
us
e
d
in
de
e
p
le
a
r
ni
ng
a
nd
m
e
di
c
a
l
im
a
ge
s
of
s
e
ve
r
a
l
da
ta
s
e
ts
.
T
he
e
xpl
a
n
a
ti
on
a
bout
th
e
pr
opos
e
d
a
nd
th
e
e
xpe
r
im
e
nt
a
l
out
c
om
e
is
s
ho
w
n
in
s
e
c
ti
on
4,
f
ol
lo
w
e
d
by
th
e
s
e
c
ti
on
5,
w
hi
c
h
di
s
c
us
s
e
s
th
e
c
r
it
ic
a
l
a
na
ly
z
e
. S
e
c
ti
on 6 looks
t
o t
he
f
ut
ur
e
a
nd outl
in
e
s
f
ur
th
e
r
r
e
s
e
a
r
c
h di
r
e
c
ti
ons
f
or
t
hi
s
pr
oj
e
c
t.
2.
P
R
E
V
I
O
U
S
WORKS
I
n
th
is
pa
r
t,
g
iv
e
n
th
e
la
c
k
of
pr
io
r
r
e
s
e
a
r
c
h
ta
c
kl
in
g
th
e
is
s
ue
,
w
e
w
il
l
pr
e
s
e
nt
a
n
ove
r
vi
e
w
o
f
th
e
r
e
le
va
nt
li
te
r
a
tu
r
e
on
di
s
e
a
s
e
c
l
a
s
s
if
ic
a
ti
on
a
nd
pr
e
di
c
ti
on
in
br
oa
d
te
r
m
s
.
L
it
e
r
a
tu
r
e
c
a
n
be
s
tr
uc
tu
r
e
d
in
m
ul
ti
pl
e
w
a
ys
.
W
e
c
a
n
c
a
te
gor
iz
e
th
e
m
ba
s
e
d
on
th
e
di
s
e
a
s
e
t
he
y
a
ddr
e
s
s
,
s
uc
h
a
s
C
O
V
I
D
-
19,
tu
m
or
b
r
a
in
,
or
ba
s
e
d on the
t
ype
of
m
e
th
od us
e
d, f
or
i
ns
ta
nc
e
, t
r
a
di
ti
ona
l
ML
-
ba
s
e
d a
nd de
e
p l
e
a
r
ni
ng
-
ba
s
e
d.
2.1.
T
r
ad
it
io
n
al
m
ac
h
in
e
l
e
ar
n
in
g
-
b
as
e
d
ML
a
lg
or
it
hm
s
ha
ve
be
e
n
br
oa
dl
y
c
onne
c
te
d
ove
r
di
s
ti
nc
ti
ve
s
pa
c
e
s
,
ju
s
t
li
ke
th
e
c
la
s
s
if
ic
a
ti
on
a
nd
de
te
c
ti
on
th
e
r
a
pe
ut
ic
s
a
nd
a
na
ly
s
i
s
.
A
l
-
D
hi
e
f
e
t
al
.
[
12]
,
p
r
opos
e
d
c
la
s
s
if
yi
ng
dys
phonia
di
s
e
a
s
e
in
th
e
pa
th
ol
ogy
de
te
c
ti
on
a
ppl
ic
a
ti
on
by
us
in
g
th
e
m
e
l
f
r
e
que
nc
y
c
e
ps
tr
a
l
c
oe
f
f
ic
ie
nt
a
nd
th
e
na
iv
e
B
a
ye
s
m
e
th
od
a
s
a
c
la
s
s
if
ie
r
c
la
s
s
f
r
om
th
e
nor
m
a
l
a
nd
a
bnor
m
a
l
c
la
s
s
by
th
e
S
a
a
r
br
uc
ke
n
voi
c
e
da
ta
ba
s
e
.
B
y
tr
a
in
in
g
m
ode
ls
on
a
da
ta
s
e
t
th
a
t
is
publ
ic
ly
a
c
c
e
s
s
ib
le
,
th
e
r
e
s
e
a
r
c
h
e
r
s
in
ve
s
ti
ga
te
how
e
f
f
e
c
ti
ve
th
e
s
e
a
lg
or
it
hm
s
a
r
e
w
it
h
a
nd
w
it
hout
th
e
us
e
of
da
t
a
a
ugm
e
nt
a
ti
on
te
c
hni
que
s
.
A
dd
it
io
na
ll
y,
U
ys
a
l
a
nd
O
z
tu
r
k
[
13]
s
ugge
s
te
d
th
e
us
e
of
ML
f
or
di
a
gnos
in
g
A
lz
he
im
e
r
'
s
di
s
e
a
s
e
(
A
D
)
,
in
c
lu
di
ng
s
uppor
t
ve
c
to
r
m
a
c
hi
ne
(
S
V
M
)
,
K
-
ne
a
r
e
s
t
ne
ig
hbor
s
(
KNN
)
,
lo
gi
s
ti
c
r
e
gr
e
s
s
io
n
(
LR
)
,
de
c
is
io
n
tr
e
e
(
DT
)
,
gr
a
m
-
ne
ga
ti
ve
ba
c
te
r
ia
(
G
N
B
)
,
a
nd
r
e
a
s
ona
bl
e
a
nd
non
-
di
s
c
r
im
in
a
to
r
y
(
R
AND)
h
a
ve
be
e
n
u
ti
li
z
e
d
in
ta
s
k
s
in
vol
vi
ng
r
e
gr
e
s
s
io
n
a
nd
c
la
s
s
if
ic
a
ti
on.
ha
ve
be
e
n
us
in
g
c
la
s
s
if
ic
a
ti
on
a
nd
r
e
gr
e
s
s
io
n
ta
s
ks
.
M
ul
ti
pl
e
bi
om
a
r
ke
r
s
us
e
d
to
ge
th
e
r
is
how
m
os
t
r
e
s
e
a
r
c
h
e
r
s
a
gr
e
e
is
th
e
be
s
t
u
s
e
a
nd
tr
e
a
tm
e
nt
f
or
th
e
d
ia
gnos
is
of
c
ondi
ti
ons
,
r
a
th
e
r
th
a
n
th
e
s
in
gl
e
e
va
lu
a
ti
on
of
one
pr
oc
e
s
s
a
lo
ne
.
T
he
s
tu
dy a
l
s
o
r
e
in
f
or
c
e
s
t
h
a
t
bot
h
g
e
n
de
r
a
nd a
tr
op
hy l
e
ve
l
h
e
a
vi
ly
i
nf
l
u
e
n
c
e
di
a
gn
os
ti
c
j
u
dgm
e
n
t
s
. U
s
in
g (
K
N
N
)
pr
ov
id
e
s
a
m
a
x
im
u
m
a
c
c
ur
a
c
y o
f
98%
f
or
t
he
di
a
gn
o
s
ti
c
gr
oup
w
h
e
n
u
s
in
g
br
a
i
n
r
e
gi
on
,
a
g
e
,
a
n
d
g
e
n
de
r
vo
lu
m
e
s
.
H
o
w
e
ve
r
,
de
s
p
it
e
t
h
is
no
ve
l
ty
,
th
e
r
e
i
s
s
t
il
l
a
la
c
k
of
a
d
e
e
p
e
r
und
e
r
s
t
a
nd
in
g of
t
he
c
onn
e
c
ti
on
b
e
tw
e
e
n
AD
a
nd
c
og
ni
ti
ve
d
e
c
li
ne
s
in
th
e
l
e
f
t
h
e
m
i
s
ph
e
r
e
of
t
h
e
br
a
in
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
, V
ol
.
14
, N
o.
3
,
J
une
2025
:
2132
-
2145
2134
G
uda
dhe
e
t
al
.
[
14]
us
e
d
A
nov
e
l
ML
-
ba
s
e
d
m
e
th
od
f
or
c
a
t
e
go
r
iz
in
g
c
e
r
e
br
a
l
bl
e
e
di
ng
i
n
c
om
put
e
d
to
m
ogr
a
p
hy
im
a
ge
s
is
e
x
a
m
in
e
d,
a
nd
pr
ogr
e
s
s
in
t
he
c
a
te
gor
i
z
a
ti
on
of
in
tr
a
c
r
a
ni
a
l
he
m
or
r
h
a
ge
in
C
T
s
c
a
n
im
a
ge
s
i
s
s
how
n
in
t
hi
s
w
or
k.
T
he
y
c
om
bi
ne
t
e
xt
ur
e
-
ba
s
e
d
gr
e
y
-
le
ve
l
c
o
-
o
c
c
ur
r
e
nc
e
m
a
tr
ix
(
G
L
C
M
)
f
e
a
tu
r
e
s
w
it
h
tr
a
ns
f
or
m
-
ba
s
e
d
f
e
a
tu
r
e
s
,
li
ke
di
s
c
r
e
te
w
a
ve
l
e
t
tr
a
n
s
f
or
m
a
nd
di
s
c
r
e
t
e
c
os
in
e
tr
a
n
s
f
or
m
,
by
ut
il
i
z
in
g
ML
c
la
s
s
if
i
e
r
s
,
a
nd
th
e
s
e
c
ha
r
a
c
te
r
i
s
ti
c
s
a
r
e
c
om
bi
n
e
d,
th
e
y pr
ovi
d
e
j
oi
nt
f
e
a
tu
r
e
s
e
t
s
,
w
hi
c
h
im
pr
ov
e
c
la
s
s
if
ic
a
ti
on
a
c
c
ur
a
c
y
o
ve
r
s
e
pa
r
a
te
a
p
pr
oa
c
he
s
.
T
hi
s
s
tu
dy
of
f
e
r
i
ng
a
no
te
w
or
th
y
a
c
c
om
pl
i
s
hm
e
nt
a
nd
c
o
m
pa
r
e
s
th
e
e
f
f
ic
a
c
y
of
th
i
s
in
te
gr
a
te
d
f
e
a
tu
r
e
m
e
th
od
to
tr
a
d
it
io
n
a
l
pr
o
c
e
d
ur
e
s
a
n
d
th
e
c
om
bi
ne
d
f
e
a
tu
r
e
s
e
t
of
w
a
v
e
le
t
s
tr
a
ns
f
or
m
w
it
h G
L
C
M
a
c
hi
e
ve
s
t
h
e
b
e
s
t
a
c
c
ur
a
c
y of
87.22%
w
h
e
n a
p
pl
y t
h
e
r
a
ndom f
or
e
s
t
(
R
F
)
c
la
s
s
if
i
e
r
.
C
a
s
tr
o
e
t
al
.
[
15]
in
ve
nt
e
d
a
te
c
hni
que
to
im
p
r
ove
f
e
a
tu
r
e
s
e
le
c
ti
on
in
A
D
s
ta
ge
c
la
s
s
if
ic
a
ti
on
to
in
c
r
e
a
s
e
a
c
c
ur
a
c
y.
T
o
pr
e
pa
r
e
a
nd
s
e
gm
e
nt
s
tr
uc
tu
r
a
l
a
nd
f
unc
ti
ona
l
br
a
in
M
R
I
da
ta
,
a
non
-
hum
a
n
c
onne
c
to
m
e
pr
oj
e
c
t
pr
e
a
m
bl
e
a
ppr
oa
c
h
is
us
e
d
a
nd
m
ul
ti
pl
e
ML
c
la
s
s
if
ie
r
s
,
a
c
ol
le
c
ti
on
of
c
r
it
ic
a
l
ne
twor
k
pa
r
a
m
e
te
r
s
th
a
t
c
ha
nge
s
ig
ni
f
ic
a
nt
ly
th
r
oughout
a
s
pe
c
if
ic
pr
o
gr
e
s
s
iv
e
c
onne
c
ti
on
is
c
ol
le
c
te
d.
T
h
e
a
ve
r
a
g
e
a
c
c
ur
a
c
y
of
th
e
bi
na
r
y
c
la
s
s
if
ic
a
ti
on
f
or
H
a
da
m
a
r
d
c
la
s
s
if
ie
r
(
HC
)
a
ga
in
s
t
e
a
r
ly
m
il
d
c
ogni
ti
ve
im
pa
ir
m
e
nt
(
E
M
C
I
)
,
H
C
vs
la
te
M
C
I
(
L
M
C
I
)
,
H
C
ve
r
s
us
A
D
,
86.0%
ve
r
s
us
E
M
C
I
ve
r
s
us
L
M
C
I
,
87.0%
ve
r
s
us
L
M
C
I
ve
r
s
us
A
D
,
a
nd
88.5%
ve
r
s
u
s
E
M
C
I
v
e
r
s
us
A
D
i
s
s
how
n
be
lo
w
.
T
hi
s
s
tu
dy
c
om
pa
r
e
s
D
T
,
K
N
N
,
a
nd
e
ns
e
m
bl
e
t
e
c
hni
que
s
. S
V
M
r
out
in
e
ly
s
ur
pa
s
s
e
s
t
he
s
e
a
ppr
oa
c
he
s
i
n t
e
r
m
s
of
a
c
c
ur
a
c
y.
L
ik
e
t
hi
s
w
or
k,
E
l
-
S
a
pp
a
gh
e
t
al
.
[
16
]
e
v
a
l
u
a
t
e
d
h
ow
f
i
ve
w
e
l
l
-
kn
o
w
n
ML
a
lg
or
it
hm
s
n
a
m
e
l
y
L
R
,
D
T
,
K
N
N
,
R
F
,
a
n
d
S
V
M
e
s
t
im
a
t
e
d
th
e
d
e
v
e
l
op
m
e
n
t
of
A
D
.
T
h
e
s
tu
d
y
u
s
e
s
di
f
f
e
r
e
nt
t
yp
e
s
of
c
o
s
t
-
e
f
f
i
c
i
e
nt
t
im
e
-
s
e
r
ie
s
c
h
a
r
a
c
t
e
r
i
s
ti
c
s
th
a
t
c
o
m
b
in
e
p
a
t
i
e
n
t
de
m
o
gr
a
ph
i
c
s
a
l
ong
w
it
h
m
e
d
i
c
a
ti
o
n
s
t
a
k
e
n
a
nd
a
s
s
e
s
s
m
e
n
t
s
of
c
o
g
ni
ti
v
e
p
e
r
f
or
m
a
n
c
e
a
nd
e
x
i
s
t
i
ng
m
e
di
c
a
l
c
o
n
di
ti
on
s
.
T
h
e
s
e
m
o
d
e
l
s
e
x
e
c
u
t
e
f
o
ur
c
l
a
s
s
if
i
c
a
t
i
on
d
ut
i
e
s
w
hi
c
h
c
o
n
s
i
s
t
of
N
D
a
n
d pr
odr
o
m
a
l
M
C
I
a
n
d s
u
bj
e
c
ti
v
e
M
C
I
a
nd A
D
.
R
e
p
hr
a
s
e
th
e
RF
m
od
e
l
a
c
h
i
e
v
e
d
h
ig
h a
c
c
u
r
a
c
y
o
f
9
0
.5
1%
a
m
o
ng
ML
a
lg
or
it
hm
s
in
th
i
s
s
t
ud
y
.
T
h
e
r
e
s
e
a
r
c
h
d
e
m
o
n
s
tr
a
t
e
s
t
h
a
t
hi
gh
pr
e
di
c
t
iv
e
v
a
l
u
e
e
x
i
s
t
s
f
or
e
v
e
r
y
m
o
d
e
l
w
hi
c
h
i
n
c
or
p
or
a
te
s
c
o
m
or
b
id
it
y
a
n
d
m
e
d
ic
a
ti
on
v
a
r
i
a
bl
e
s
a
n
d
o
th
e
r
e
a
r
l
y
-
s
ta
g
e
f
e
a
t
ur
e
s
t
o
p
e
r
f
o
r
m
b
e
tt
e
r
t
h
a
n
R
F
m
o
d
e
l
s
.
D
u
tt
a
a
nd
B
a
n
dy
op
a
dh
y
a
y
[
17]
d
e
te
c
t
e
d
a
br
a
i
n c
a
n
c
e
r
c
l
a
s
s
i
f
i
e
r
t
o
t
a
c
kl
e
t
h
i
s
a
n
d i
s
s
u
e
d
t
h
e
i
r
p
r
o
p
o
s
a
l
a
dv
o
c
a
ti
n
g f
or
th
e
u
s
e
of
e
n
s
e
m
b
l
e
-
b
a
s
e
d
M
L
a
l
gor
i
t
hm
s
.
T
he
y
in
tr
od
u
c
e
t
h
e
N
G
-
B
o
o
s
t
a
lg
or
it
hm
,
c
o
u
pl
e
d
w
i
th
a
(
5
-
f
o
ld
)
s
t
r
a
ti
f
i
e
d
c
r
o
s
s
v
a
l
id
a
ti
on
a
p
pr
o
a
c
h
,
a
s
th
e
pr
im
a
r
y
c
la
s
s
if
i
e
r
m
o
d
e
l
f
or
a
ut
om
a
t
e
d
t
um
or
d
e
t
e
c
t
io
n,
a
nd
c
on
tr
a
s
t
s
ML
t
e
c
h
ni
qu
e
s
s
u
c
h
a
s
R
F
,
G
li
o
bl
a
s
to
m
a
(
GB
)
,
e
nh
a
n
c
e
d
T
um
or
N
e
t
(
ET
)
,
a
n
d
a
d
a
p
ti
v
e
b
o
o
s
t
in
g
(
A
d
a
B
)
.
I
t
i
s
w
o
r
t
h
no
ti
n
g,
N
G
-
B
o
o
s
t
y
i
e
l
d
s
th
e
m
o
s
t
hop
e
f
u
l
r
e
s
u
lt
,
a
c
hi
e
vi
ng
a
n
a
c
c
ur
a
c
y
of
9
8.
5%
.
K
ha
n
e
t
al
.
[
18]
d
is
c
us
s
e
d
th
e
is
s
ue
s
in
vol
ve
d
in
di
a
gnos
in
g
m
ig
r
a
in
e
he
a
da
c
he
s
,
a
c
om
m
on
a
nd
c
om
pl
e
x
ne
ur
ova
s
c
ul
a
r
di
s
e
a
s
e
.
T
h
e
y
c
onvi
nc
in
gl
y
a
r
gue
th
a
t
th
e
s
ubj
e
c
ti
ve
m
e
a
s
ur
e
m
e
nt
s
of
pa
in
in
te
n
s
it
y
do
not
pr
ovi
de
a
n
a
c
c
ur
a
te
di
a
gnos
is
,
a
nd
he
nc
e
th
e
ne
w
m
e
t
hods
of
ne
w
m
e
a
s
ur
e
m
e
nt
s
w
e
r
e
im
pe
r
a
ti
ve
.
M
L
, i
n t
ur
n, m
or
e
s
pe
c
if
ic
a
ll
y t
he
ML
m
ode
ls
, ha
ve
be
e
n pr
ove
n t
o be
qui
te
s
uc
c
e
s
s
f
ul
i
n s
e
ve
r
a
l
ot
he
r
f
ie
ld
s
.
T
hus
,
th
e
ir
a
ppl
ic
a
ti
on
to
he
a
lt
hc
a
r
e
is
a
l
s
o
de
m
a
ndi
ng,
c
ons
i
de
r
in
g
th
e
s
hor
ta
ge
of
r
e
s
our
c
e
s
.
T
he
y
pr
e
di
c
t
a
nd
c
la
s
s
if
y
th
e
m
a
ny
ki
nds
of
m
ig
r
a
in
e
s
us
in
g
s
t
a
te
-
of
-
th
e
-
a
r
t
ML
te
c
hni
que
s
,
s
u
c
h
a
s
S
V
M
,
K
N
N
,
R
F
,
di
s
tu
r
ba
nc
e
s
to
r
m
ti
m
e
in
d
e
x
(
D
S
T
)
,
a
nd
de
e
p
ne
ur
a
l
ne
twor
ks
(
DNN
)
.
T
he
s
e
a
r
e
tr
a
in
e
d
on
publ
ic
da
ta
s
e
ts
,
bot
h
w
it
h
a
nd
w
it
hout
da
ta
a
ugm
e
nt
a
ti
on,
to
e
nh
a
nc
e
m
ig
r
a
in
e
di
a
gnos
is
.
T
he
r
e
s
ul
ts
s
ho
w
im
pr
ove
m
e
nt
in
th
e
r
a
nge
of
99.66%
a
c
c
ur
a
c
y
of
D
N
N
,
94.60%
of
S
V
M
,
97.1
0%
of
K
N
N
,
88.20%
of
D
S
T
,
a
nd
98.50%
of
R
F
. T
he
pr
ove
t
he
t
r
a
ns
f
or
m
a
ti
ona
l
pow
e
r
of
M
L
i
n m
ig
r
a
in
e
di
a
gnos
ti
c
s
.
K
a
r
a
m
t
i
e
t
al
.
[
1
9]
r
e
v
ie
w
f
oc
u
s
e
s
pr
i
m
a
r
il
y
o
n
t
he
pr
obl
e
m
of
c
e
r
vi
c
a
l
c
a
nc
e
r
,
p
a
r
t
ic
ul
a
r
ly
it
s
bur
d
e
n
a
nd
th
e
i
m
p
a
c
t
of
c
e
r
v
ic
a
l
c
a
n
c
e
r
on
w
om
e
n
i
n
t
he
d
e
v
e
lo
pi
n
g
w
or
ld
.
H
ow
e
v
e
r
,
m
i
s
s
i
ng
va
lu
e
s
in
e
xi
s
ti
ng
da
t
a
s
e
t
s
f
or
a
u
to
m
a
t
e
d
de
te
c
t
io
n
of
t
e
n
c
o
m
pr
om
i
s
e
ML
m
o
de
l
p
e
r
f
or
m
a
n
c
e
. I
n
th
i
s
w
or
k, t
he
a
ut
h
or
s
s
ug
ge
s
t
a
n
a
ut
om
a
t
e
d
s
y
s
t
e
m
f
o
r
f
or
e
c
a
s
ti
ng
c
e
r
vi
c
a
l
c
a
n
c
e
r
a
n
d
d
e
a
l
a
pt
l
y w
it
h
m
i
s
s
in
g
v
a
lu
e
s
by
u
s
in
g
s
ynt
he
ti
c
m
in
or
i
ty
ove
r
s
a
m
pl
i
ng
t
e
c
hni
qu
e
(
S
M
O
T
E
)
f
e
a
tu
r
e
s
to
r
e
a
c
h
a
hi
g
h
r
a
te
of
a
c
c
ur
a
c
y.
T
o
de
a
l
w
it
h
m
i
s
s
in
g
va
lu
e
s
,
th
e
s
y
s
t
e
m
e
m
pl
o
y
s
a
n
e
n
s
e
m
b
le
vo
ti
n
g
c
la
s
s
if
ie
r
w
i
th
th
r
e
e
ML
m
od
e
l
s
u
s
in
g
K
N
N
i
m
p
ut
e
r
a
nd
S
M
O
T
E
u
p
-
s
a
m
pl
e
d
f
e
a
t
ur
e
s
.
W
he
n
tr
a
in
e
d
on
K
N
N
-
im
pu
te
d
S
M
O
T
E
f
e
a
tu
r
e
s
,
th
e
pr
op
o
s
e
d
m
o
de
l
g
a
in
s
a
r
e
m
a
r
k
a
bl
e
99.
99%
f
or
t
he
f
o
ur
m
e
tr
i
c
s
of
b
ot
h
a
c
c
u
r
a
c
y
,
pr
e
c
i
s
i
on
,
r
e
c
a
ll
,
a
nd
F
1
s
c
or
e
.
E
x
pe
r
im
e
nt
a
l
r
e
s
ul
t
s
d
e
m
o
n
s
tr
a
t
e
th
e
pr
opo
s
e
d
m
o
de
l
i
s
m
or
e
e
f
f
e
c
ti
ve
a
s
c
om
p
a
r
e
d
to
a
f
e
w
e
xi
s
ts
ML
a
lg
or
it
h
m
s
u
nd
e
r
di
f
f
e
r
e
nt
s
c
e
n
a
r
io
s
.
T
a
ng
e
t
al
.
[
20]
in
ve
s
ti
ga
te
d
th
e
us
e
of
qua
nt
it
a
ti
ve
c
ha
r
a
c
t
e
r
is
ti
c
s
ta
k
e
n
f
r
om
c
he
s
t
C
T
s
c
a
n
s
to
c
a
te
gor
iz
e
C
O
V
I
D
-
19
pa
ti
e
nt
s
.
T
h
e
y
us
e
d
a
c
ol
le
c
ti
on
of
C
T
s
c
a
n
s
f
r
om
C
O
V
I
D
-
19
pa
ti
e
nt
s
to
e
xt
r
a
c
t
va
r
ia
bl
e
s
s
u
c
h
a
s
te
xt
ur
e
,
s
ha
p
e
,
a
nd
in
te
n
s
it
y.
M
L
m
ode
ls
w
e
r
e
tr
a
in
e
d
a
nd
e
va
lu
a
te
d
f
or
C
O
V
I
D
-
19
c
la
s
s
if
ic
a
ti
on
a
nd
hi
gh
a
c
c
ur
a
c
y
in
di
s
c
r
im
in
a
ti
ng
be
twe
e
n
p
a
ti
e
nt
s
.
U
s
in
g
3
-
f
ol
d
c
r
os
s
-
va
li
da
ti
on,
th
e
R
F
m
ode
l
pe
r
f
or
m
e
d
w
e
ll
,
a
c
hi
e
vi
ng
tr
ue
po
s
it
iv
e
a
nd
ne
ga
ti
ve
r
a
te
s
of
0.933
a
nd
0.745,
a
c
c
ur
a
c
y
of
0.875,
a
nd
a
n
a
r
e
a
unde
r
th
e
c
ur
ve
(
AUC
)
of
0.91.
T
he
s
e
r
e
s
ul
ts
hi
ghl
ig
ht
th
e
e
f
f
ic
ie
nc
y
of
us
in
g
qua
nt
it
a
ti
ve
C
T
im
a
ge
f
e
a
tu
r
e
s
f
or
C
O
V
I
D
-
19 c
la
s
s
if
ic
a
ti
on.
2.2.
D
e
e
p
l
e
an
in
g t
e
c
h
n
ic
al
-
b
as
e
d
D
e
e
p
le
a
r
ni
ng
a
ppr
oa
c
he
s
,
s
u
c
h
a
s
c
onvolut
io
na
l
ne
ur
a
l
ne
two
r
ks
(
C
N
N
)
,
a
r
e
c
om
m
onl
y
ut
il
iz
e
d
to
id
e
nt
if
y
a
nd
c
la
s
s
if
y
m
e
di
c
a
l
im
a
ge
s
.
R
a
m
o
s
e
t
al
.
[
21]
pr
ovi
de
a
s
pe
c
i
a
l
c
om
put
e
r
-
a
id
e
d
de
te
c
ti
on
(
C
A
D
)
s
ys
te
m
f
or
th
e
a
ut
om
a
ti
c
de
te
c
ti
on
of
br
e
a
s
t
c
a
nc
e
r
.
I
m
a
ge
s
e
gm
e
nt
a
ti
on,
m
a
nua
l
f
e
a
tu
r
e
e
xt
r
a
c
ti
on
(
e
.g.,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
E
ar
ly
de
te
c
ti
on and c
la
s
s
if
ic
at
io
n of
bone
m
ar
r
o
w
c
hange
s
i
n l
um
bar
v
e
r
t
e
br
ae
…
(
Y
as
ir
H
us
s
e
in
Shak
i
r
)
2135
H
O
G
,
uni
f
o
r
m
lo
c
a
l
b
in
a
r
y
pa
tt
e
r
ns
(
u
L
B
P
)
)
,
a
nd
de
e
p
f
e
a
tu
r
e
e
xt
r
a
c
ti
on
us
in
g
D
e
ns
e
-
N
e
t
201
a
r
e
a
ll
in
c
lu
de
d
in
th
e
s
y
s
te
m
.
A
hybr
id
te
c
hni
que
c
om
bi
ne
s
de
e
p
a
nd
ha
ndm
a
de
c
ha
r
a
c
t
e
r
is
ti
c
s
w
it
h
s
ta
ti
s
ti
c
a
l
c
r
it
e
r
ia
a
nd
us
e
d
g
e
ne
ti
c
a
lg
or
it
hm
s
(
G
A
)
a
nd
m
ut
ua
ll
y
be
n
e
f
ic
ia
l
s
e
le
c
ti
on
a
r
e
us
e
d
to
c
hoos
e
th
e
m
o
s
t
im
por
ta
nt
tr
a
it
s
.
T
he
p
e
r
f
or
m
e
d
w
e
ll
on
m
a
m
m
ogr
a
phy
a
nd
ul
tr
a
s
ound
da
ta
s
e
ts
(
a
c
c
ur
a
c
y
97.6%
,
pr
e
c
is
io
n
98%
,
r
e
c
a
ll
98%
,
F
1
-
s
c
or
e
98%
,
a
nd
in
de
x
ba
la
nc
e
d
a
c
c
ur
a
c
y
(
I
B
A
)
95%
)
.
T
he
s
ugge
s
te
d
C
A
D
out
pe
r
f
or
m
s
c
ur
r
e
nt
a
ppr
oa
c
he
s
a
nd
s
how
s
pr
om
is
e
f
or
c
li
ni
c
a
l
a
ppl
ic
a
ti
on
s
.
T
he
r
e
s
e
a
r
c
h
w
or
k
d
one
by
J
a
ve
d
e
t
a
l
.
[
22]
s
p
e
c
if
i
c
a
ll
y
in
i
d
e
nt
if
ic
a
ti
on
a
nd
c
l
a
s
s
if
ic
a
ti
on
of
lu
ng
f
oc
a
l
op
a
c
it
y
(
2023)
,
D
L
te
c
hni
q
ue
s
li
ke
D
N
N
a
n
d
s
t
a
c
k
a
ut
o
e
nc
od
e
r
c
l
a
s
s
if
ie
r
s
ha
v
e
be
e
n
de
f
i
ne
d
f
or
th
e
id
e
nt
if
i
c
a
ti
o
n
a
nd
c
la
s
s
if
i
c
a
ti
on
of
lu
ng
f
o
c
a
l
op
a
c
it
y.
T
hi
s
is
in
di
c
a
ti
v
e
of
th
e
f
a
c
t
th
a
t
de
e
p
le
a
r
ni
ng
a
lg
or
it
hm
s
a
r
e
in
de
e
d
u
s
e
f
ul
i
n
th
e
f
i
e
ld
of
m
e
d
ic
a
l
im
a
g
e
a
n
a
l
ys
i
s
.
M
or
e
ov
e
r
,
f
a
c
ia
l
im
a
ge
s
w
e
r
e
e
m
p
lo
ye
d
w
he
n
it
c
om
e
s
to
p
a
in
r
e
c
ogni
ti
on
a
nd
pa
in
l
e
ve
l
d
if
f
e
r
e
nt
i
a
ti
on
ut
il
iz
i
ng
ML
.
M
un
a
di
e
t
al
.
[
23]
pr
opo
s
e
e
nha
n
c
in
g
th
e
D
L
m
ode
ls
f
or
di
a
gno
s
in
g
T
B
by
in
c
r
e
a
s
in
g
im
a
g
e
q
ua
li
ty
.
T
he
f
ol
lo
w
in
g
w
or
k
s
w
e
r
e
di
s
c
us
s
e
d
:
C
on
tr
a
s
t
li
m
it
e
d
a
d
a
pt
iv
e
hi
s
to
gr
a
m
e
q
ua
li
z
a
ti
o
n
(
C
L
A
H
E
)
,
hi
gh
-
f
r
e
que
n
c
y
e
m
pha
s
i
s
f
il
te
r
i
ng
(
H
E
F
)
,
a
nd
un
s
h
a
r
p
m
a
s
k
in
g
(
U
M
)
.
R
e
s
-
N
e
t
a
n
d
E
f
f
ic
i
e
nt
-
N
e
t
t
r
a
ns
f
e
r
le
a
r
ni
ng
m
od
e
l
s
u
s
e
d e
nha
nc
e
d
im
a
ge
s
f
or
m
or
e
e
f
f
i
c
ie
nt
c
la
s
s
if
i
c
a
ti
on
c
om
p
a
r
e
d
to
tr
a
d
it
io
n
a
l
im
a
g
e
s
.
U
s
in
g
th
e
S
h
e
nz
h
e
n
da
t
a
s
e
t
to
te
s
t
th
e
m
od
e
l,
a
c
c
ur
a
c
y
a
nd
t
he
A
U
C
m
e
tr
ic
s
w
e
r
e
a
s
hi
gh
a
s
89
.
92%
a
n
d
9
4.
8%
a
nd
7%
,
r
e
s
pe
c
ti
ve
l
y,
e
ndor
s
in
g
im
a
ge
e
nha
n
c
e
m
e
nt
a
s
a
t
ool
t
o
bui
ld
a
b
e
tt
e
r
, m
or
e
pow
e
r
f
ul
s
ys
te
m
f
or
di
a
gn
os
i
ng
T
B
to
g
e
t
be
tt
e
r
d
e
te
c
ti
on
.
S
h
a
r
m
a
e
t
al
.
[
24
]
pr
e
s
e
n
t
a
n
d
e
x
a
m
i
n
e
d
i
v
e
r
s
e
C
N
N
a
r
c
h
i
t
e
c
tu
r
e
s
d
e
s
i
g
n
e
d
f
o
r
f
e
a
t
ur
e
e
x
t
r
a
c
t
i
o
n
a
n
d
c
l
a
s
s
i
f
i
c
a
t
io
n
i
n
th
e
r
e
a
l
m
of
p
n
e
u
m
on
i
a
d
e
t
e
c
t
i
on
.
A
dd
i
ti
o
n
a
ll
y,
t
h
e
y
e
x
p
l
or
e
d
h
o
w
d
a
t
a
s
e
t
s
i
z
e
a
f
f
e
c
t
s
C
N
N
p
e
r
f
or
m
a
n
c
e
vi
a
t
r
a
in
i
n
g
m
o
d
e
l
s
o
n
b
o
th
o
r
ig
i
n
a
l
a
n
d
a
ug
m
e
n
t
e
d
d
a
t
a
s
e
t
s
.
O
u
r
f
in
d
i
n
g
s
,
w
hi
c
h
r
e
v
e
a
l
e
d
v
a
r
yi
n
g
t
e
s
t
i
n
g
a
c
c
ur
a
c
i
e
s
a
m
o
ng
d
if
f
e
r
e
n
t
m
o
d
e
l
c
o
n
f
i
g
ur
a
t
i
o
n
s
,
a
s
s
u
r
e
t
h
e
p
o
t
e
n
t
i
a
l
o
f
C
N
N
s
t
o
e
n
h
a
n
c
e
p
n
e
u
m
on
i
a
d
e
t
e
c
t
i
o
n.
S
p
e
c
i
f
i
c
a
l
l
y,
t
h
e
y
a
c
h
i
e
v
e
d
m
a
g
n
if
i
c
e
n
t
a
c
c
u
r
a
c
y
of
9
0%
.
A
m
o
n
g
a
u
t
h
or
s
,
S
u
n
g
h
e
e
t
h
a
a
nd
S
h
a
r
m
a
[
25]
h
a
s
t
he
t
a
s
k
of
di
a
gn
o
s
i
ng
th
e
di
a
b
e
t
e
s
r
e
la
t
e
d
s
ym
pt
o
m
s
t
o
d
e
te
c
t
s
ig
n
s
w
h
ic
h
a
r
e
v
i
s
ib
le
in
bl
o
od
v
e
s
s
e
l
s
of
e
y
e
s
.
T
h
e
i
r
t
e
c
h
ni
qu
e
i
nv
ol
v
e
d
u
s
i
ng
a
de
e
p
f
e
a
tu
r
e
C
N
N
,
w
hi
c
h
w
a
s
s
p
e
c
if
i
c
a
ll
y
d
e
s
i
g
ne
d
f
o
r
d
e
t
e
c
t
in
g
t
he
s
e
m
a
r
k
e
r
s
.
T
a
k
i
ng
a
n
e
x
a
m
pl
e
,
t
h
e
m
od
e
l
pr
ov
e
d
t
o
w
or
k
w
i
th
a
c
c
ur
a
t
e
t
o
a
n
i
m
p
r
e
s
s
i
v
e
9
7%
i
n
d
i
a
g
n
o
s
i
ng
d
i
a
be
ti
c
c
o
n
di
t
i
on
s
.
T
h
i
s
s
t
ud
y pr
o
ve
s
t
h
a
t
d
e
e
p
l
e
a
r
ni
n
g i
s
e
f
f
e
c
t
iv
e
i
n
m
e
di
c
a
l
di
a
gn
o
s
t
i
c
s
w
hi
l
e
a
l
s
o
a
t
t
e
m
pt
in
g
t
o
r
e
v
e
a
l
th
e
c
a
p
a
b
il
it
i
e
s
of
d
e
t
h
a
t
m
a
k
e
it
p
o
s
s
ib
l
e
to
i
d
e
n
ti
f
y
s
i
gn
s
of
i
ll
n
e
s
s
e
s
by
a
n
a
ly
z
in
g
t
he
ir
v
i
s
u
a
l
p
a
t
t
e
r
n
s.
D
e
vul
a
pa
ll
i
e
t
al
.
[
26]
e
m
pl
oye
d
G
a
bor
m
ul
ti
s
c
a
le
te
xt
ur
e
f
e
a
tu
r
e
s
in
c
onj
unc
ti
on
w
it
h
a
pr
e
-
t
r
a
in
e
d
G
oogl
e
ne
t
m
ode
l
a
s
a
f
e
a
tu
r
e
e
xt
r
a
c
to
r
.
F
r
om
th
e
e
xt
e
ns
iv
e
im
a
ge
c
ol
le
c
ti
on,
th
e
f
in
a
l
f
e
a
tu
r
e
ve
c
to
r
w
il
l
be
us
e
d
to
e
xt
r
a
c
t
th
e
r
e
qui
r
e
d
p
ic
tu
r
e
da
ta
.
I
t
ha
s
out
pe
r
f
or
m
e
d
s
ta
te
-
of
-
th
e
-
a
r
t
te
c
hni
que
s
w
it
h
a
n
a
c
c
ur
a
c
y
of
91%
.
T
hi
s
w
a
s
done
by
S
ha
nka
r
e
t
al
.
[
27]
,
w
ho
m
a
de
a
s
ig
ni
f
ic
a
nt
a
tt
e
m
pt
a
t
id
e
nt
if
yi
ng
C
O
V
I
D
-
19
by
a
na
ly
z
in
g
c
he
s
t
X
-
r
a
y
pi
c
tu
r
e
s
.
T
he
ir
a
ppr
oa
c
h
in
vol
ve
s
th
e
f
ol
lo
w
in
g
s
te
ps
:
pr
e
-
pr
oc
e
s
s
in
g,
w
hi
c
h
in
vol
ve
s
W
e
in
e
r
f
il
te
r
in
g,
f
e
a
tu
r
e
e
xt
r
a
c
ti
on
w
it
h
G
L
C
M
,
ge
n
e
r
a
li
z
e
d
lo
w
-
r
a
nk
m
ode
ls
(
G
L
R
M
)
,
a
nd
lo
c
a
l
bi
na
r
y
pa
tt
e
r
ns
(
L
B
P
)
,
a
nd
th
e
f
in
a
l
c
la
s
s
if
ic
a
ti
on
w
it
h
a
n
a
r
ti
f
ic
ia
l
ne
ur
a
l
ne
twor
ks
(
ANN
)
.
W
hi
le
f
e
a
tu
r
e
s
e
le
c
ti
on
w
a
s
pe
r
f
or
m
e
d
w
it
h
gr
e
a
t
c
a
r
e
to
s
e
le
c
t
th
e
opt
im
a
l
f
e
a
tu
r
e
s
ubs
e
ts
,
th
is
m
ode
l
s
how
e
d
pot
e
nt
ia
l
to
im
pr
ove
th
e
c
la
s
s
if
ic
a
ti
on
e
f
f
ic
ie
nc
y
b
e
twe
e
n
in
f
e
c
te
d s
a
m
pl
e
s
a
nd
h
e
a
lt
hy
one
s
.
T
hi
s
w
or
k
a
l
s
o
r
e
ve
a
l
s
th
e
ve
r
s
a
ti
li
ty
of
D
L
i
n s
ol
vi
ng me
di
c
in
a
l
di
a
gnos
is
pr
obl
e
m
s
, e
s
p
e
c
ia
ll
y t
he
e
m
e
r
ge
nt
he
a
lt
h c
ha
ll
e
nge
s
i
n t
he
w
or
ld
.
T
he
r
e
s
e
a
r
c
h
c
ons
is
ti
ng
of
th
e
de
te
c
ti
on
of
c
or
ona
r
y
a
r
te
r
y
s
te
nos
i
s
w
a
s
a
l
s
o
c
a
r
r
ie
d
out
by
G
le
s
s
ge
n
e
t
al
[
28]
.
A
r
e
tr
os
pe
c
ti
ve
s
tu
dy
w
it
h
217
e
m
e
r
ge
nc
y
r
oom
pa
ti
e
nt
s
e
va
lu
a
te
d
a
or
ti
c
C
T
A
im
a
ge
s
f
or
vi
s
ua
l
ta
ll
y
by
two
r
e
a
de
r
s
.
T
he
de
e
p
le
a
r
ni
ng
m
ode
l
(
D
L
M
)
pr
ovi
de
d
a
n
in
de
pe
nde
nt
c
la
s
s
if
ic
a
ti
on
of
c
or
ona
r
y
a
r
te
r
ie
s
a
nd
th
e
im
a
ge
s
a
s
e
it
he
r
hi
gh
-
qua
li
ty
(
HQ
)
or
lo
w
qua
li
ty
(
LQ
)
;
a
c
c
ur
a
c
y:
s
e
ns
it
iv
it
y,
92.9%
;
s
pe
c
if
ic
it
y,
95.7%
;
a
nd
ne
t
pr
e
s
e
nt
va
lu
e
(
N
P
V
)
,
99.4
%
f
or
pa
ti
e
nt
s
w
it
h
H
Q
im
a
ge
s
.
A
lt
hough
it
m
a
y
ha
ve
s
li
ght
ly
lo
w
e
r
s
e
ns
it
iv
it
y
in
th
e
p
a
ti
e
nt
gr
oup
w
it
h
th
e
le
s
s
e
r
d
e
f
in
it
io
n
of
th
e
L
Q
im
a
ge
s
,
th
e
di
a
gnos
ti
c
e
f
f
ic
ie
nc
y
di
d
not
c
ha
nge
.
N
ot
a
bl
y,
th
e
D
L
M
in
th
i
s
r
e
ga
r
d
e
xa
c
tl
y
ta
r
ge
te
d
a
ll
A
C
S
pa
ti
e
nt
s
a
nd
I
C
A
-
c
onf
ir
m
e
d
s
te
nos
e
s
w
hi
le
hi
ghl
ig
ht
in
g
th
e
m
oda
li
ty
’
s
pos
s
ib
le
s
tr
e
ngt
h
s
,
pr
im
a
r
il
y,
in
c
lu
di
ng
th
e
de
li
ve
r
y
of
a
n
a
c
c
e
s
s
ib
le
non
-
pa
r
a
m
e
tr
ic
ve
r
y
hi
gh
N
P
V
f
or
t
he
s
ubs
ta
nt
ia
l
c
or
ona
r
y
s
t
e
nos
is
in
a
nt
ic
a
nc
e
r
pe
pt
id
e
(
A
C
P
)
pa
ti
e
nt
s
.
G
ot
ti
pa
ll
a
a
nd
Y
a
ll
a
[
29
]
f
ur
th
e
r
in
ve
s
ti
ga
te
d
a
ne
w
nove
l
hybr
id
ML
a
ppr
oa
c
h
to
a
ddr
e
s
s
in
g
di
m
e
ns
io
n
a
li
ty
pr
obl
e
m
.
T
hi
s
r
e
li
a
bl
e
a
lg
or
it
hm
c
om
bi
ne
s
th
e
B
e
e
s
A
lg
or
it
hm
a
nd
I
C
A
f
or
opt
im
a
l
pe
r
f
or
m
a
nc
e
.
A
s
a
ne
xt
s
te
p,
a
f
te
r
e
s
ta
bl
is
hi
ng
th
e
f
e
a
tu
r
e
r
e
duc
ti
on,
va
r
io
us
c
la
s
s
if
ie
r
s
li
ke
K
N
N
,
S
V
M
,
L
R
,
a
nd
R
F
.
T
he
a
s
pe
c
ts
w
e
r
e
in
s
pe
c
te
d
to
e
va
lu
a
te
th
e
e
f
f
ic
ie
nc
y
of
th
e
s
e
c
la
s
s
if
ie
r
s
a
c
c
ur
a
c
y,
m
e
m
or
y c
ons
um
pt
io
n, pr
e
c
is
io
n, r
e
c
a
ll
va
lu
e
s
,
a
nd F
1 s
c
or
e
.
3.
P
R
O
P
O
S
E
A
P
P
R
O
A
C
H
E
S
T
he
a
im
of
our
s
tu
dy
is
to
c
ol
le
c
t
e
nough
M
R
I
-
ba
s
e
d
bon
e
m
a
r
r
ow
c
ha
nge
da
t
a
to
b
e
u
s
e
d
f
or
ML
m
ode
l
de
ve
lo
pm
e
nt
.
I
t
c
oul
d
a
ls
o
be
ut
il
iz
e
d
a
s
a
c
om
pl
e
m
e
n
ta
r
y
a
s
s
is
t
f
or
r
a
di
ol
ogi
s
ts
or
c
li
ni
c
ia
ns
w
he
n
di
f
f
e
r
e
nt
ia
ti
ng
to
m
a
ke
in
te
ll
ig
e
nt
de
c
is
io
ns
a
nd
w
is
e
r
.
W
e
w
i
ll
w
or
k
w
it
h
our
da
ta
s
e
t
(
B
M
C
L
V
B
)
,
a
nd
th
e
e
xpe
r
im
e
nt
a
l
pr
oc
e
s
s
c
om
pr
is
e
s
s
e
ve
r
a
l
s
te
ps
,
th
e
pr
opos
e
d
m
ode
l,
a
nd
e
xpe
r
im
e
nt
a
l
m
e
th
ods
ut
il
iz
e
d
a
s
s
how
n i
n F
ig
ur
e
1.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
, V
ol
.
14
, N
o.
3
,
J
une
2025
:
2132
-
2145
2136
F
ig
ur
e
1
.
F
lo
w
c
ha
r
t
of
t
he
pr
opos
e
d m
ode
l
a
nd e
xpe
r
im
e
nt
a
l
m
e
th
ods
us
e
d
3.1.
C
ol
le
c
t
e
d
d
at
as
e
t
M
R
I
b
on
e
m
ar
r
ow
c
h
an
ge
s
i
n
l
u
m
b
a
r
ve
r
t
e
b
r
ae
B
M
C
L
V
da
t
a
s
e
t,
c
om
pr
is
i
ng
1018
M
R
I
s
c
a
n
s
f
r
om
134
B
M
C
L
V
B
pa
ti
e
nt
s
w
ho
t
e
s
t
e
d
f
or
bone
m
a
r
r
ow
c
ha
nge
s
a
nd w
e
r
e
r
e
f
e
r
r
e
d t
o
th
e
tr
a
in
in
g
-
th
e
r
a
pe
ut
i
c
A
l
-
K
a
f
e
e
l
s
upe
r
s
pe
c
ia
lt
y
ho
s
pi
t
a
l
(
K
S
S
H
)
,
I
r
a
q,
f
r
om
A
pr
il
10,
20
22,
to
S
e
pt
e
m
be
r
2,
202
3. T
h
e
i
m
a
gi
n
g
pr
ot
oc
ol
c
ons
i
s
t
e
d of
T
1
-
w
e
ig
ht
e
d,
T
2
-
w
e
i
ght
e
d, a
nd
f
a
t
-
s
up
pr
e
s
s
e
d s
e
qu
e
nc
e
s
w
it
h
a
s
l
ic
e
th
ic
kne
s
s
of
5 m
m
a
nd a
m
a
tr
ix
s
iz
e
of
256
×
256
pi
xe
ls
.
T
h
e
i
m
a
ge
s
w
e
r
e
s
to
r
e
d
in
D
I
C
O
M
f
or
m
a
t
a
nd
a
nonymi
z
e
d
be
f
or
e
a
na
ly
s
i
s
,
a
nd
th
e
a
ge
s
of
th
e
pa
ti
e
nt
s
r
a
n
ge
d
f
r
o
m
18
to
80
ye
a
r
s
ol
d
.
T
he
pr
e
s
e
nt
s
tu
dy
w
a
s
c
o
nduc
te
d
a
c
c
or
di
n
g
to
t
he
g
ui
de
li
n
e
s
pr
ovi
d
e
d
b
y
th
e
in
s
ti
tu
ti
on’
s
r
e
vi
e
w
boa
r
d
a
nd
th
e
d
e
c
l
a
r
a
ti
on
of
H
e
ls
i
nki
pr
in
c
ip
l
e
s
r
e
ga
r
d
in
g
e
th
ic
a
l
c
on
duc
t
of
r
e
s
e
a
r
c
h
in
vol
vi
n
g
hum
a
n
s
ubj
e
c
t
s
. A
c
c
or
d
in
g t
o t
he
f
in
d
in
g
s
of
t
hi
s
s
t
udy,
B
M
C
L
V
B
di
a
g
nos
i
s
i
s
ba
s
e
d
on c
l
in
ic
a
l
s
ig
n
s
a
nd
s
ym
pt
om
s
,
bl
ood
a
na
ly
s
i
s
,
M
R
I
s
c
a
n
s
,
a
n
d
x
-
r
a
y
bo
ne
in
s
pe
c
ti
on.
T
he
u
s
ua
l
in
t
e
r
ve
nt
io
n
e
nt
a
il
s
lo
ng
s
pa
n
do
s
a
ge
of
a
nt
ib
io
ti
c
or
a
nt
if
u
nga
l
c
h
e
m
ic
a
ls
a
nd
, a
t
t
im
e
s
,
s
ur
g
e
r
y t
o
dr
a
in
out
a
b
s
c
e
s
s
or
e
v
e
n r
e
m
ove
a
f
f
e
c
te
d
bon
e
.
A
pa
ti
e
nt
is
id
e
nt
if
ie
d
a
s
pos
it
iv
e
f
or
B
M
C
L
V
B
if
th
e
y
m
e
e
t
one
or
m
or
e
th
e
f
ol
lo
w
in
g
c
r
it
e
r
ia
:
i)
d
is
pl
a
yi
ng
s
u
s
pi
c
io
us
c
li
ni
c
a
l
s
ym
pt
om
s
(
s
u
c
h
a
s
f
a
ti
gue
,
f
e
ve
r
,
ba
c
k
pa
in
,
ne
ur
ol
ogi
c
a
l
is
s
u
e
s
,
a
nd
le
g
pa
in
)
in
c
onj
unc
ti
on
w
it
h
pos
it
iv
e
M
R
I
f
in
di
ngs
a
nd
i
i)
e
xhi
bi
ti
ng
s
us
pi
c
io
us
la
bor
a
to
r
y
f
in
di
ngs
(
s
uc
h
a
s
dys
pl
a
s
ti
c
c
h
a
nge
s
li
ke
m
e
ga
lo
bl
a
s
ti
c
r
ubr
ic
a
te
s
,
ne
ut
r
ophi
l
hype
r
s
e
gm
e
nt
a
ti
on,
or
gi
a
nt
pl
a
te
le
ts
,
w
hi
c
h
m
a
y i
ndi
c
a
te
m
ye
lo
dys
pl
a
s
ti
c
s
yndr
om
e
or
c
e
r
ta
in
m
ye
lo
id
l
e
u
ke
m
ia
s
)
a
lo
ng w
it
h pos
it
iv
e
M
R
I
r
e
s
ul
ts
.
T
hr
e
e
e
xpe
r
t
r
a
di
ol
ogi
s
ts
pe
r
f
or
m
e
d
r
e
a
di
ngs
to
de
te
r
m
in
e
w
h
e
th
e
r
th
e
im
a
ge
s
di
s
pl
a
ye
d
nor
m
a
l
o
r
a
bnor
m
a
l
B
M
C
L
V
B
f
in
di
ngs
.
W
e
ba
s
e
d
our
a
s
s
e
s
s
m
e
nt
on
t
he
m
or
e
pr
oba
bl
e
di
a
gnos
is
id
e
nt
if
ie
d
by
th
e
a
gr
e
e
m
e
nt
be
twe
e
n
two
r
a
di
ol
ogi
s
ts
w
he
n
th
e
ir
in
di
vi
dua
l
di
a
gnos
e
s
va
r
ie
d
f
r
om
th
e
ir
pe
e
r
s
.
I
m
a
ge
s
w
it
h
vi
s
ib
le
le
s
io
ns
f
r
om
ot
he
r
di
s
e
a
s
e
s
or
known
m
a
s
s
e
s
to
ge
th
e
r
w
it
h
th
os
e
of
poor
qua
li
ty
a
nd
s
ubs
ta
nda
r
d
r
a
di
a
ti
on w
e
r
e
e
li
m
in
a
te
d f
r
om
t
he
s
tu
dy.
3.2.
P
r
e
-
p
r
oc
e
s
s
in
g
T
h
e
r
a
d
io
lo
g
y
i
m
a
g
e
s
e
x
po
r
t
e
d
h
a
d
s
i
z
e
s
o
f
a
p
pr
ox
im
a
t
e
ly
5
27
K
B
,
a
nd
th
e
f
or
m
a
t
s
w
e
r
e
i
n
D
I
C
O
M
w
i
th
a
p
ix
e
l
r
e
s
o
lu
ti
on
of
5
12
×
5
12
.
T
h
e
i
m
a
g
e
s
w
e
r
e
th
e
r
e
f
or
e
c
r
op
p
e
d
i
n
J
P
G
f
o
r
m
a
t
to
e
n
h
a
n
c
e
p
a
ti
e
nt
a
n
o
ny
m
i
ty
a
n
d
t
o
m
a
t
c
h
m
o
s
t
pr
og
r
a
m
m
i
ng
s
t
a
nd
a
r
d
l
ib
r
a
r
i
e
s
. E
ve
n
w
h
e
r
e
t
h
e
f
i
l
e
s
i
z
e
w
a
s
r
e
du
c
e
d
to
92
K
B
.
3.2.1.
R
e
s
iz
e
i
m
age
s
A
t
th
is
pha
s
e
,
B
M
in
di
f
f
e
r
e
nt
r
e
s
iz
e
s
a
nd
pr
ope
r
ti
e
s
is
nor
m
a
li
z
e
d.
S
pe
c
if
ic
a
ll
y,
th
e
im
a
ge
s
a
r
e
r
e
s
iz
e
d
to
224×
224.
I
n
th
is
pr
oc
e
s
s
,
in
th
e
ne
xt
s
te
p,
a
ll
im
a
ge
f
e
a
tu
r
e
s
a
r
e
pr
ovi
de
d
th
a
t
ha
ve
th
e
s
a
m
e
s
iz
e
a
nd pr
ope
r
ty
i
n f
e
a
tu
r
e
e
xt
r
a
c
ti
on, f
a
c
il
it
a
ti
ng c
ons
is
te
nt
a
na
ly
s
i
s
.
3.2.2.
C
on
ve
r
t
i
m
age
s
t
o gr
ay
s
c
al
e
G
r
a
ys
c
a
le
c
onve
r
s
io
n
i
s
one
of
th
e
m
o
s
t
ba
s
i
c
or
in
it
ia
l
im
a
ge
p
r
oc
e
s
s
in
g a
nd
is
of
te
n
a
dopt
e
d
a
s
th
e
f
ir
s
t
f
il
te
r
in
im
a
ge
pr
oc
e
s
s
in
g.
T
hi
s
i
s
ve
r
s
a
ti
le
f
or
im
a
ge
s
th
r
o
ugh
r
e
duc
in
g
th
e
a
m
ount
of
c
ol
or
in
f
or
m
a
ti
on
but
m
a
in
ta
in
in
g
s
tr
uc
tu
r
a
l
f
e
a
tu
r
e
s
th
r
ough
th
e
c
onve
r
s
io
n
e
le
m
e
nt
c
a
ll
e
d
gr
a
ys
c
a
le
.
T
hi
s
he
lp
s
to
s
im
pl
if
y
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
E
ar
ly
de
te
c
ti
on and c
la
s
s
if
ic
at
io
n of
bone
m
ar
r
o
w
c
hange
s
i
n l
um
bar
v
e
r
t
e
br
ae
…
(
Y
as
ir
H
us
s
e
in
Shak
i
r
)
2137
be
c
a
us
e
it
a
id
s
in
th
e
e
li
m
in
a
ti
on
of
c
om
pl
e
x
c
om
put
a
ti
ona
l
a
c
ti
vi
ty
th
a
t
c
a
n
m
a
s
k
r
a
di
c
a
l
pa
tt
e
r
ns
a
nd
te
xt
ur
e
[
30]
.
T
he
out
put
im
a
ge
i
s
bl
a
c
k
a
nd
w
hi
te
w
hi
le
th
e
in
te
ns
it
y
va
r
ie
s
f
r
om
0
to
255
(
le
a
s
t
a
nd
gr
e
a
te
s
t
va
lu
e
s
r
e
s
pe
c
ti
ve
ly
)
w
hi
c
h m
e
a
ns
i
t
c
a
n e
a
s
il
y b
e
a
na
ly
z
e
d l
a
te
r
.
3.3.
F
e
at
u
r
e
e
xt
r
ac
t
io
n
u
s
in
g h
is
t
ogr
am
of
or
ie
n
t
e
d
gr
ad
ie
n
t
s
P
r
ope
r
f
or
th
e
obj
e
c
ti
ve
of
our
s
tu
dy,
w
e
a
ppl
ie
d
H
O
G
a
s
a
f
e
a
tu
r
e
e
xt
r
a
c
ti
on
on
th
e
te
xt
ur
a
l
a
nd
s
ha
pe
s
of
M
R
I
im
a
ge
s
.
I
t
c
a
lc
ul
a
te
s
gr
a
di
e
nt
or
ie
nt
a
ti
ons
in
th
e
lo
c
a
li
z
e
d
r
e
gi
ons
of
a
n
im
a
ge
th
a
t
th
e
n
id
e
nt
if
ie
s
th
e
e
dge
s
a
nd
c
or
ne
r
s
of
a
n
im
a
ge
[
31]
.
T
h
e
im
a
ge
is
us
ua
ll
y
c
onve
r
te
d
to
gr
a
y
s
c
a
le
a
nd
f
e
a
tu
r
e
e
xt
r
a
c
ti
on oc
c
ur
s
t
hr
ough HO
G
w
hi
c
h
s
ta
nds
f
or
H
O
G
. H
O
G
p
a
r
a
m
e
te
r
s
c
ons
is
t
of
O
r
ie
nt
a
ti
ons
:9
, P
ix
e
ls
pe
r
c
e
ll
:(
8,
8)
,
c
e
ll
s
pe
r
bl
oc
k:
(
2,
2)
,
T
r
a
ns
f
or
m
s
qua
r
e
r
oot
:
Y
e
s
,
B
lo
c
k
nor
m
a
li
z
a
ti
on
m
e
th
od:
L
2
-
H
ys
.
T
he
e
qua
ti
on f
or
c
a
lc
ul
a
ti
ng H
O
G
f
e
a
tu
r
e
s
i
s
a
s
(
1)
:
(
,
)
=
=
√
(
x
,
y
)
2
+
(
,
)
2
(
1)
3.4.
D
im
e
n
s
io
n
al
it
y r
e
d
u
c
t
io
n
t
e
c
h
n
iq
u
e
s
R
e
duc
ti
on
in
di
m
e
n
s
io
ns
is
a
pr
oc
e
s
s
w
he
r
e
th
e
ove
r
a
ll
vol
um
e
of
a
va
il
a
bl
e
f
e
a
tu
r
e
s
(
or
di
m
e
ns
io
ns
)
is
r
e
duc
e
d
w
hi
le
m
a
ki
ng
s
ur
e
th
a
t
th
e
da
ta
’
s
im
por
ta
nt
c
ha
r
a
c
te
r
is
ti
c
s
a
r
e
r
e
ta
in
e
d.
T
hi
s
pr
oc
e
s
s
is
c
r
uc
i
a
l
in
de
a
li
ng
w
it
h
hi
gh
-
di
m
e
ns
io
na
l
da
ta
,
w
hi
c
h
c
a
n
s
uf
f
e
r
f
r
om
is
s
ue
s
s
uc
h
a
s
th
e
c
ur
s
e
of
di
m
e
ns
io
na
li
ty
,
in
c
r
e
a
s
e
d
c
om
put
a
ti
ona
l
c
om
pl
e
xi
ty
a
nd
ove
r
f
it
ti
ng
.
I
nc
lu
de
th
is
pr
oc
e
s
s
is
c
r
uc
ia
l
f
or
de
a
li
ng
w
it
h
hi
gh
-
di
m
e
ns
io
na
l
da
ta
,
w
hi
c
h
c
a
n
s
uf
f
e
r
f
r
om
is
s
ue
s
s
uc
h
a
s
th
e
c
ur
s
e
of
di
m
e
n
s
io
na
li
ty
,
in
c
r
e
a
s
e
d
c
om
put
a
ti
ona
l
c
om
pl
e
xi
ty
, a
nd ove
r
f
it
ti
ng a
nd i
nc
lu
de
:
3.4.1.
P
r
in
c
ip
al
c
om
p
on
e
n
t
an
al
ys
is
T
o
de
c
r
e
a
s
e
th
e
num
be
r
th
e
da
ta
w
hi
le
m
a
xi
m
iz
in
g
th
e
va
r
ia
nc
e
,
to
r
e
a
c
h
a
li
m
it
,
w
e
us
e
d
P
C
A
.
I
n
e
s
s
e
n
c
e
,
P
C
A
c
a
n
a
s
s
is
t
u
s
in
id
e
nt
if
yi
ng
m
or
e
s
ig
ni
f
ic
a
nt
w
a
ys
of
e
xpl
a
in
in
g
our
da
ta
.
I
t
he
lp
s
r
e
duc
in
g
th
e
c
om
pl
e
xi
ty
e
s
s
e
nt
ia
l
f
or
da
ta
r
e
pr
e
s
e
nt
a
ti
on be
a
r
in
g only m
in
or
i
nf
or
m
a
ti
ona
l
lo
s
s
.
3.4.2.
L
in
e
ar
d
is
c
r
i
m
in
an
t
an
al
ys
is
L
D
A
w
a
s
us
e
d
a
s
th
e
s
e
c
ond
f
or
m
of
f
e
a
tu
r
e
e
xt
r
a
c
ti
on
und
e
r
th
e
c
a
te
gor
y
of
s
up
e
r
vi
s
e
d
li
ne
a
r
m
e
th
ods
th
a
t
is
us
e
d
f
or
di
m
e
ns
io
na
li
ty
r
e
duc
ti
on.
T
o
r
e
f
r
e
s
h
our
m
e
m
or
y
r
e
c
a
ll
,
L
D
A
w
a
s
he
lp
f
ul
in
a
w
a
y
th
a
t
it
e
nha
nc
e
d
be
twe
e
n
c
la
s
s
va
r
ia
nc
e
to
w
it
hi
n
c
la
s
s
va
r
ia
nc
e
r
a
ti
o
.
T
hi
s
im
pr
ove
m
e
nt
c
ont
r
ib
ut
e
d
to
b
e
tt
e
r
c
la
s
s
s
e
pa
r
a
bi
li
ty
i
n our
da
ta
s
e
t.
3.4.3.
I
n
d
e
p
e
n
d
e
n
t
c
om
p
on
e
n
t
a
n
al
ys
is
W
e
a
ls
o
us
e
d
a
not
h
e
r
te
c
hni
que
c
a
ll
e
d
I
C
A
w
hos
e
obj
e
c
ti
ve
w
a
s
to
f
ur
th
e
r
c
ut
dow
n
on
th
e
di
m
e
ns
io
na
li
ty
of
our
M
R
I
im
a
ge
da
ta
s
e
t
to
di
s
c
ove
r
th
e
la
t
e
nt
s
ta
ti
s
ti
c
a
ll
y
in
de
p
e
nde
nt
s
our
c
e
s
c
ont
a
in
e
d
he
r
e
in
.
I
C
A
w
a
s
p
a
r
ti
c
ul
a
r
ly
he
lp
f
ul
in
e
s
ti
m
a
ti
ng
a
nd
s
e
pa
r
a
ti
n
g
a
non
-
G
a
us
s
ia
n
s
ig
na
l
.
T
hi
s
s
ig
n
a
l
m
ig
ht
be
in
c
lu
de
d i
n t
he
s
a
m
pl
e
.
3.5. Dat
as
e
t
s
p
li
t
T
he
s
pl
it
th
e
da
ta
s
e
t
in
to
th
r
e
e
s
ub
s
e
ts
:
tr
a
in
in
g,
va
li
da
ti
on,
a
n
d
te
s
t
s
e
ts
,
w
he
r
e
da
t
a
a
r
e
di
vi
de
d
f
or
le
a
r
ni
ng
th
e
m
ode
l,
a
s
s
e
s
s
in
g
th
e
m
ode
l'
s
a
c
c
ur
a
c
y
ba
s
e
d
on
t
he
uns
e
e
n
da
ta
of
th
e
va
li
da
ti
on,
a
nd
th
e
f
in
a
l
a
s
s
e
s
s
m
e
nt
of
th
e
m
ode
l
on
th
e
un
s
e
e
n
te
s
t
da
ta
.
T
he
da
ta
s
e
t
in
th
is
s
tu
dy
in
vol
ve
d
a
s
a
m
pl
e
s
iz
e
of
one
th
ous
a
nd
a
nd
e
ig
ht
e
e
n.
H
e
nc
e
,
w
e
pa
r
ti
ti
one
d
70%
of
th
e
da
ta
,
e
qua
l
to
(
712
c
a
s
e
s
)
,
in
to
th
e
tr
a
in
in
g
s
e
t
f
o
r
our
m
ode
ls
’
tr
a
in
in
g.
T
he
f
in
a
l
s
e
t
of
da
ta
,
w
hi
c
h
w
a
s
a
ls
o
th
e
va
li
da
ti
on
s
e
t,
c
ont
a
in
e
d
20%
o
f
th
e
s
ubj
e
c
ts
(
203 c
a
s
e
s
)
, a
nd t
he
l
a
s
t,
t
he
hol
d
-
out
s
e
t,
c
ons
is
te
d of
10%
o
f
t
he
da
ta
(
103 c
a
s
e
s
)
, w
hi
c
h w
a
s
not
e
xpos
e
d t
o
th
e
tr
a
in
in
g
a
nd
th
e
c
r
os
s
-
va
li
da
ti
on
a
nd
w
a
s
us
e
d
to
r
a
te
th
e
ge
ne
r
a
l
pe
r
f
or
m
a
nc
e
of
th
e
m
ode
ls
a
s
s
how
n
in
T
a
bl
e
1.
T
a
bl
e
1
.
S
pl
it
da
ta
s
e
t
in
f
or
m
a
ti
on
D
a
t
a
s
e
t
N
um
be
r
of
c
a
s
e
s
P
e
r
c
e
nt
a
ge
of
t
ot
a
l
da
t
a
s
e
t
(%)
T
r
a
i
ni
ng
712
70
V
a
l
i
da
t
i
on
203
20
T
e
s
t
103
10
T
ot
a
l
1
,
018
100
3.6.
C
la
s
s
if
i
c
at
io
n
I
n
th
e
c
la
s
s
if
ic
a
ti
on
pr
oc
e
s
s
,
th
e
K
N
N
,
S
V
M
,
DT
,
m
ul
ti
-
la
ye
r
pe
r
c
e
pt
r
on
(
M
L
P
)
,
RF
,
hi
s
to
gr
a
m
(
H
G
)
,
gr
a
di
e
nt
boos
ti
ng
(
GB
)
,
A
da
B
,
pa
s
s
iv
e
a
ggr
e
s
s
iv
e
(
P
A
)
,
a
nd
L
R
m
e
th
ods
w
e
r
e
us
e
d
to
tr
a
in
th
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
, V
ol
.
14
, N
o.
3
,
J
une
2025
:
2132
-
2145
2138
id
e
nt
if
ie
d
f
e
a
tu
r
e
s
.
T
he
s
e
te
c
hni
que
s
f
a
ll
unde
r
tr
a
di
ti
ona
l
m
e
th
odol
ogi
e
s
of
le
a
r
ni
ng,
a
nd
th
e
r
e
s
ul
ts
th
a
t
c
a
n
be
a
c
hi
e
ve
d
a
r
e
hi
ghl
y
be
n
e
f
ic
ia
l
in
bi
na
r
y
a
nd
m
ul
ti
c
la
s
s
f
or
m
a
ti
ons
.
E
ve
n
th
ough
a
r
ti
f
ic
ia
l
le
a
r
ni
ng
m
e
th
ods
ha
ve
be
e
n
gr
a
dua
ll
y
a
dva
n
c
in
g
in
r
e
c
e
nt
ye
a
r
s
,
le
a
di
n
g
to
r
e
duc
e
d
ut
il
iz
a
ti
on
of
tr
a
di
ti
ona
l
le
a
r
ni
ng
m
e
th
ods
,
it
is
pi
vot
a
l
to
r
e
m
e
m
be
r
th
a
t
c
la
s
s
i
c
a
l
le
a
r
ni
ng
m
e
t
hods
a
r
e
ju
s
t
a
s
e
f
f
e
c
ti
ve
a
s
w
e
ll
.
O
ur
r
e
s
ul
ts
s
ig
ni
f
y
th
a
t
ut
il
iz
in
g
f
a
c
to
r
s
li
ke
f
1
-
s
c
or
e
,
a
nd
a
c
c
ur
a
c
y
a
ki
n
to
our
w
or
k
m
a
y
de
m
on
s
tr
a
te
th
a
t
tr
a
di
ti
ona
l
le
a
r
ni
ng
te
c
hni
que
s
,
s
uc
h
a
s
c
la
s
s
ic
a
l
ML
,
m
a
y
a
l
s
o
pr
ovi
de
f
r
ui
tf
ul
r
e
s
ul
ts
.
A
ddi
ti
ona
ll
y,
w
e
’
ve
di
s
c
ove
r
e
d
th
a
t
w
he
n
a
ppl
yi
ng
ML
,
it
is
pos
s
ib
le
to
ta
ke
a
dva
nt
a
ge
of
c
l
a
s
s
ic
a
l
le
a
r
ni
ng
m
e
th
ods
f
or
th
os
e
ki
nds
of
ta
s
ks
th
a
t
r
e
qui
r
e
l
ow
c
om
pl
e
xi
ty
a
nd r
a
pi
d r
e
s
ul
ts
.
3.6.1.
K
-
n
e
ar
e
s
t
n
e
ig
h
b
or
s
I
n
te
r
m
s
of
pur
pos
e
,
K
N
N
is
one
of
th
e
e
a
s
ie
s
t
to
unde
r
s
ta
nd
a
nd
m
os
t
pow
e
r
f
ul
a
lg
or
it
hm
s
us
e
d
in
c
la
s
s
if
ic
a
ti
on
a
nd
id
e
nt
if
ic
a
ti
on
c
onc
e
r
ns
.
I
t
s
or
ts
da
ta
ba
s
e
d
ont
o
KNN
in
a
f
e
a
tu
r
e
s
pa
c
e
.
T
he
out
put
is
a
c
la
s
s
m
e
m
be
r
s
hi
p
w
he
r
e
th
e
s
a
m
pl
e
is
c
la
s
s
if
ie
d
by
a
vot
in
g
s
ys
te
m
a
nd
th
e
vot
e
s
a
r
e
c
a
s
t
by
ne
ig
hbor
s
.
T
he
a
lg
or
it
hm
us
e
d
to
im
pl
e
m
e
nt
c
lu
s
t
e
r
in
c
r
e
m
e
nt
a
l
i
s
nonpa
r
a
m
e
tr
ic
a
nd
of
th
e
la
z
y
le
a
r
ni
ng
ty
pe
,
w
hi
c
h
im
pl
ie
s
th
a
t
th
e
da
ta
di
s
tr
ib
ut
io
n
doe
s
not
m
a
ke
a
ny
a
s
s
um
pt
io
ns
a
nd
doe
s
not
m
a
ni
pul
a
te
th
e
tr
a
in
in
g
da
ta
poi
nt
s
t
o m
a
ke
c
e
r
ta
in
i
nf
e
r
e
nc
e
s
[
32]
.
3.6.2.
S
u
p
p
or
t
v
e
c
t
or
m
ac
h
in
e
O
ne
of
th
e
m
os
t
e
f
f
e
c
ti
ve
a
nd
w
id
e
ly
u
s
e
d
s
upe
r
vi
s
e
d
l
e
a
r
ni
ng
a
lg
or
it
hm
s
,
S
V
M
pe
r
f
or
m
s
w
e
ll
in
bot
h
c
la
s
s
if
ic
a
ti
on
a
nd
f
unc
ti
on
a
ppr
oxi
m
a
ti
on.
A
lt
hough
it
is
ne
a
r
ly
e
qu
a
ll
y
a
ppl
ic
a
bl
e
to
a
ll
f
or
m
s
of
pr
obl
e
m
s
,
it
is
m
os
t
f
r
e
que
nt
ly
a
ppl
ie
d
to
c
la
s
s
if
ic
a
ti
on
pr
obl
e
m
s
.
T
he
im
pl
e
m
e
nt
a
ti
on
of
th
e
S
V
M
a
lg
or
it
hm
in
vol
ve
s
pl
a
c
in
g
e
a
c
h
it
e
m
of
th
e
da
ta
s
e
t
on
th
e
n
-
di
m
e
ns
io
na
l
pl
a
ne
,
w
he
r
e
n
i
s
th
e
num
b
e
r
of
a
tt
r
ib
ut
e
s
in
th
e
da
ta
s
e
t,
w
it
h
th
e
qua
li
ti
e
s
of
a
tt
r
ib
ut
e
s
b
e
in
g
th
e
m
e
a
s
ur
e
of
th
e
c
oor
di
na
te
a
xi
s
.
N
e
xt
,
th
e
c
la
s
s
if
ic
a
ti
on
s
te
p i
s
c
a
r
r
ie
d out, which involve
s
i
d
e
nt
if
yi
ng t
he
pl
a
ne
t
ha
t
m
a
xi
m
a
ll
y s
e
pa
r
a
te
s
t
he
t
w
o
c
la
s
s
e
s
[
33]
.
3.6.3.
D
e
c
is
io
n
t
r
e
e
DT
unde
r
go
le
a
r
ni
ng
a
lg
or
it
hm
s
th
a
t
c
a
n
onl
y
be
e
m
pl
oye
d
in
th
os
e
pr
obl
e
m
s
w
he
r
e
th
e
ta
r
ge
t
va
r
ia
bl
e
is
a
lr
e
a
dy
de
f
in
e
d
or
gi
ve
n,
th
e
s
e
a
r
e
m
o
r
e
us
e
f
ul
in
e
xpl
or
a
to
r
y
ty
pe
o
f
p
r
obl
e
m
s
li
ke
c
la
s
s
if
ic
a
ti
on
pr
obl
e
m
s
.
S
in
c
e
it
c
a
n
e
s
ti
m
a
te
a
ny
f
unc
ti
ona
l
f
or
m
,
it
is
s
ui
ta
bl
e
f
or
bot
h
c
ont
in
uous
a
nd
c
a
te
gor
ic
a
l
in
put
a
nd
out
put
va
r
ia
bl
e
s
.
T
hi
s
a
lg
or
it
hm
m
e
a
ns
s
pl
it
ti
ng
th
e
popula
ti
on
in
to
two
or
m
or
e
hom
oge
nous
gr
oups
in
a
c
c
or
da
nc
e
w
it
h t
he
m
os
t
di
s
ti
ngui
s
hi
ng f
e
a
tu
r
e
/i
nput
va
r
ia
bl
e
[
34]
.
3.6.4.
M
u
lt
i
-
l
aye
r
p
e
r
c
e
p
t
r
on
M
L
P
c
on
ta
in
s
m
u
lt
i
pl
e
p
e
r
c
e
p
tr
o
ns
be
c
a
u
s
e
it
i
s
c
om
po
s
e
d
of
m
or
e
th
a
n
o
ne
pe
r
c
e
pt
r
on
.
T
h
e
tr
u
e
c
om
put
a
t
io
n
a
l
po
w
e
r
of
M
L
P
e
xi
s
t
s
b
e
t
w
e
e
n
i
t
s
in
put
a
n
d
ou
tp
ut
l
a
y
e
r
s
in
v
a
r
i
ou
s
h
id
d
e
n
la
ye
r
s
w
hi
c
h
a
l
ong
w
it
h
o
ne
hi
dd
e
n
l
a
y
e
r
.
T
h
e
s
e
,
a
lo
ng
w
it
h
on
e
h
id
de
n l
a
y
e
r
,
e
n
a
bl
e
a
ny
c
on
ti
n
uo
us
f
u
nc
ti
o
n
a
p
pr
ox
im
a
ti
on
[
35]
.
3.6.5.
R
an
d
om
f
or
e
s
t
RF
is
a
n
e
ns
e
m
bl
e
le
a
r
ni
ng
m
e
th
od
f
or
c
la
s
s
if
ic
a
ti
on,
r
e
gr
e
s
s
io
n,
a
nd
ot
he
r
ta
s
ks
th
a
t
w
or
k
by
c
ons
tr
uc
ti
ng
a
m
ul
ti
tu
de
of
DT
a
t
tr
a
in
in
g
ti
m
e
.
I
t
out
put
s
th
e
c
la
s
s
th
a
t
is
th
e
m
ode
of
th
e
c
la
s
s
e
s
(
c
la
s
s
if
ic
a
ti
on)
or
m
e
a
n
pr
e
di
c
ti
on
(
r
e
gr
e
s
s
io
n)
of
th
e
in
di
vi
dua
l
tr
e
e
s
.
R
a
ndom
de
c
is
io
n
f
or
e
s
ts
c
or
r
e
c
t
th
e
DT
ha
bi
t
of
ove
r
f
it
ti
ng t
o t
he
ir
t
r
a
in
in
g s
e
t
[
36]
.
3.6.6.
H
is
t
ogr
am
-
b
as
e
d
gr
ad
ie
n
t
b
oos
t
in
g
H
is
to
gr
a
m
-
ba
s
e
d
gr
a
di
e
nt
boos
ti
ng
(
H
G
B
)
is
a
ML
a
ppr
oa
c
h
f
or
c
la
s
s
if
ic
a
ti
on
a
nd
r
e
gr
e
s
s
io
n
th
a
t
e
xt
e
nds
th
e
gr
a
di
e
nt
boos
ti
ng
m
a
c
hi
n
e
a
r
c
hi
te
c
tu
r
e
.
I
t
c
on
s
tr
uc
ts
a
n
e
xt
r
a
m
ode
l
in
a
f
or
w
a
r
d
s
t
a
ge
-
w
is
e
a
ppr
oa
c
h,
a
ll
ow
s
f
or
th
e
opt
im
iz
a
ti
on
of
a
ny
di
f
f
e
r
e
nt
ia
bl
e
lo
s
s
f
unc
ti
ons
.
A
ddi
ti
ona
ll
y,
it
us
e
s
hi
s
to
gr
a
m
s
to
a
c
c
e
le
r
a
te
t
he
l
e
a
r
ni
ng pr
oc
e
s
s
[
37]
.
3.6.7.
G
r
ad
ie
n
t
b
oos
t
in
g
GB
s
e
r
v
e
s
a
s
a
ML
t
e
c
h
n
iq
u
e
t
h
a
t
e
n
a
b
l
e
s
r
e
gr
e
s
s
i
o
n
a
n
d
c
l
a
s
s
i
f
i
c
a
t
i
o
n
b
y
c
o
n
s
tr
u
c
t
in
g
p
r
e
di
c
t
i
on
m
o
d
e
l
s
t
h
r
o
ug
h
m
ul
t
i
pl
e
w
e
a
k
p
r
e
di
c
t
io
n
m
o
d
e
l
s
m
a
d
e
o
f
DT
.
T
h
e
m
o
de
l
i
n
g
pr
o
c
e
s
s
h
a
p
p
e
n
s
s
t
a
g
e
-
by
-
s
t
a
g
e
w
hi
l
e
t
h
e
m
e
t
h
o
d
e
n
a
b
l
e
s
o
pt
i
m
i
z
a
t
i
on
o
f
a
n
y
di
f
f
e
r
e
nt
i
a
b
l
e
l
o
s
s
f
u
n
c
t
io
n
.
T
h
i
s
a
p
p
r
o
a
c
h
c
r
e
a
t
e
s
g
e
n
e
r
a
l
i
z
e
d
r
e
s
u
l
t
s
[
3
8]
.
3.6.8.
A
d
ap
t
iv
e
b
oos
t
in
g
T
he
ML
m
e
ta
-
a
lg
or
it
hm
A
da
B
ope
r
a
te
s
unde
r
th
e
a
c
r
onym
A
da
B
oos
t.
Y
oa
v
F
r
e
und
a
lo
ng
w
it
h
R
obe
r
t
S
c
ha
pi
r
e
de
ve
lo
pe
d
th
i
s
a
lg
or
it
hm
.
T
he
m
e
th
od
w
or
ks
a
lo
ng
w
it
h
m
ul
ti
pl
e
ki
nds
of
le
a
r
ni
ng
a
lg
or
it
hm
s
to
e
nha
nc
e
th
e
ir
pe
r
f
or
m
a
nc
e
r
e
s
ul
ts
.
S
e
ve
r
a
l
w
e
a
k
le
a
r
ni
ng
a
lg
or
it
hm
s
pr
oduc
e
th
e
ir
out
put
w
hi
c
h ge
ts
c
om
bi
ne
d i
nt
o w
e
ig
ht
e
d va
lu
e
s
f
or
m
in
g t
he
boos
te
d
c
la
s
s
if
ie
r
out
put
[
39]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
E
ar
ly
de
te
c
ti
on and c
la
s
s
if
ic
at
io
n of
bone
m
ar
r
o
w
c
hange
s
i
n l
um
bar
v
e
r
t
e
br
ae
…
(
Y
as
ir
H
us
s
e
in
Shak
i
r
)
2139
3.6.9.
P
as
s
iv
e
aggr
e
s
s
iv
e
PA
a
lg
or
it
hm
s
a
r
e
onl
in
e
le
a
r
ni
ng
a
lg
or
it
hm
s
th
a
t
r
e
m
a
in
pa
s
s
iv
e
f
or
pr
ope
r
c
la
s
s
if
ic
a
ti
on
r
e
s
ul
t
s
.
H
ow
e
ve
r
,
th
e
y
be
c
om
e
a
ggr
e
s
s
iv
e
i
n t
he
c
a
s
e
of
a
m
is
ta
ke
, upd
a
ti
ng a
nd modi
f
yi
ng. I
t'
s
a
ppr
opr
ia
te
f
o
r
l
a
r
ge
-
s
c
a
le
l
e
a
r
ni
ng a
nd ma
y b
e
ut
il
iz
e
d f
or
bi
na
r
y or
m
ul
ti
-
c
la
s
s
c
la
s
s
if
ic
a
ti
on
[
40]
.
3.6.10.
L
ogi
s
t
ic
r
e
gr
e
s
s
io
n
LR
is
a
n
a
na
ly
s
is
of
s
t
a
ti
s
ti
c
s
e
m
pl
oye
d
to
a
na
ly
z
e
th
e
va
li
di
t
y
of
a
gi
ve
n
da
ta
w
he
n
one
or
m
or
e
e
xpl
a
na
to
r
y
f
a
c
to
r
s
in
f
lu
e
nc
e
th
e
va
r
ia
bl
e
of
in
te
r
e
s
t.
A
m
e
tr
ic
,
w
hi
c
h
is
a
n
or
de
r
e
d
qu
a
nt
it
a
ti
ve
va
r
ia
bl
e
,
r
e
pr
e
s
e
nt
s
th
e
r
e
s
ul
t
to
m
e
e
t
a
di
c
hot
om
ous
va
r
ia
bl
e
,
in
w
hi
c
h
th
e
r
e
a
r
e
onl
y
two
opt
io
ns
.
I
t
[
41]
ha
s
b
e
e
n
a
ppl
ie
d i
n nume
r
ous
f
ie
ld
s
s
uc
h a
s
s
o
c
ia
l
s
c
ie
n
c
e
s
, m
a
r
ke
ti
ng, a
s
w
e
ll
a
s
i
n t
he
m
e
di
c
a
l
f
ie
ld
a
m
ongs
t
ot
he
r
s
.
4.
R
E
S
U
L
T
S
D
u
r
i
ng
t
h
e
e
x
p
e
r
i
m
e
nt
s
,
t
h
e
s
y
s
t
e
m
w
a
s
c
o
nf
i
g
ur
e
d
w
i
t
h
a
n
I
n
t
e
l
C
o
r
e
(
i
7
-
9
7
0
0
C
P
U
)
,
t
w
o
G
P
U
s
(
T
4
×
2)
,
a
n
d
1
6
G
B
of
R
A
M
.
A
l
l
t
h
e
pr
o
c
e
s
s
e
s
d
e
s
c
r
i
b
e
d
b
y
th
e
f
l
ow
w
e
r
e
d
o
n
e
u
s
i
n
g
K
a
g
g
l
e
.
T
h
e
c
l
a
s
s
i
f
i
c
a
t
io
n
o
u
t
c
om
e
w
a
s
e
v
a
l
u
a
t
e
d
t
o
c
om
p
a
r
e
t
h
e
r
e
s
ul
t
s
o
f
t
h
e
c
l
a
s
s
i
f
i
e
r
s
w
it
h
t
h
e
a
c
tu
a
l
r
e
s
u
lt
s
u
s
i
n
g
a
c
c
ur
a
c
y
,
pr
e
c
i
s
i
o
n
,
r
e
c
a
l
l,
a
n
d
F
1
s
c
o
r
e
.
M
e
tr
i
c
s
u
s
e
d
i
n
t
h
i
s
v
i
e
w
i
n
c
l
u
d
e
t
h
e
f
ol
l
o
w
i
n
g
t
h
a
t
i
s
d
e
s
c
r
i
b
e
d
b
y
(
2
)
t
o
(
5
)
.
T
h
e
e
q
u
a
t
i
o
n
s
d
e
m
o
n
s
t
r
a
t
e
t
h
e
c
a
l
c
u
l
a
t
i
o
n
o
f
t
h
e
s
e
v
a
l
u
e
s
:
w
h
e
r
e
hi
g
h
e
r
v
a
lu
e
f
o
r
e
a
c
h
o
f
th
e
s
e
m
e
tr
i
c
s
i
s
i
n
d
i
c
a
t
i
v
e
o
f
b
e
t
t
e
r
m
od
e
l
p
e
r
f
or
m
a
n
c
e
. H
e
r
e
,
T
P
r
e
p
r
e
s
e
nt
s
tr
u
e
p
o
s
i
ti
v
e
, w
h
i
c
h i
s
t
h
e
T
N
s
ig
n
a
l
i
n
g n
or
m
a
l
s
a
m
pl
e
s
u
c
c
e
s
s
f
ul
l
y c
a
t
e
g
or
i
z
e
d
b
y
t
h
e
m
o
d
e
l
,
a
n
d
T
N
s
i
g
ni
f
i
e
s
t
r
u
e
n
e
g
a
t
iv
e
,
w
h
i
c
h
i
s
th
e
B
M
C
L
V
B
s
a
m
p
l
e
a
c
c
ur
a
t
e
l
y
c
l
a
s
s
i
f
i
e
d
b
y
t
h
e
m
o
d
e
l
;
F
P
i
s
f
or
f
a
l
s
e
po
s
i
t
iv
e
,
w
hi
c
h
i
s
t
h
e
B
M
C
L
V
B
s
a
m
pl
e
m
i
s
c
l
a
s
s
i
f
i
e
d
a
s
n
or
m
a
l
b
y
t
h
e
m
o
d
e
l;
F
N
s
t
a
n
d
s
f
or
f
a
l
s
e
n
e
g
a
t
i
v
e
,
w
hi
c
h
i
s
t
h
e
n
or
m
a
l
s
a
m
p
l
e
i
n
c
or
r
e
c
t
l
y
c
l
a
s
s
if
i
e
d
a
s
B
M
C
L
V
B
b
y
t
h
e
m
od
e
l
.
I
t
i
s
i
m
po
r
t
a
n
t
t
o
no
t
e
t
h
a
t
i
n
o
ur
w
or
k
,
w
e
t
h
in
k
t
h
a
t
B
M
C
L
V
B
s
a
m
pl
e
s
a
r
e
n
e
g
a
t
i
v
e
s
a
m
p
l
e
s
,
w
h
i
l
e
no
r
m
a
l
s
a
m
p
l
e
s
a
r
e
p
o
s
i
t
i
v
e
s
a
m
p
l
e
s
.
=
+
+
+
+
(
2)
=
+
(
3)
=
+
(
4)
1
−
=
2
×
×
+
(
5)
T
he
H
O
G
f
e
a
tu
r
e
e
xt
r
a
c
ti
on
m
e
th
od
a
c
hi
e
ve
d
it
s
hi
ghe
s
t
s
uc
c
e
s
s
pe
r
c
e
nt
a
ge
r
a
te
s
f
r
om
T
a
bl
e
2.
T
he
m
a
r
ks
th
e
be
s
t
p
e
r
f
or
m
a
nc
e
r
a
te
s
w
it
h
bol
df
a
c
e
.
H
O
G
f
e
a
tu
r
e
e
xt
r
a
c
ti
on
w
it
h
LR
c
la
s
s
if
ie
r
ge
ne
r
a
te
d
th
e
hi
ghe
s
t
pe
r
f
or
m
in
g
r
e
s
ul
ts
a
c
c
or
di
ng
to
ta
bl
e
da
ta
.
T
he
pa
ti
e
nt
c
la
s
s
if
ic
a
ti
on
pr
e
c
is
io
n
r
e
a
c
he
d
89.36%
w
hi
le
r
e
c
a
ll
pe
r
f
or
m
e
d
a
t
89.36%
a
nd
th
e
F
1
s
c
or
e
a
c
hi
e
ve
d
89.36%
a
c
c
or
di
ng
to
th
e
r
e
c
or
de
d
ta
bl
e
da
ta
.
T
he
de
te
c
ti
on r
a
te
f
or
nor
m
a
l
a
s
w
e
ll
a
s
a
bnor
m
a
l
c
a
s
e
s
t
he
a
c
c
ur
a
c
y r
e
a
c
he
s
90.29%
.
T
a
bl
e
2
.
C
la
s
s
if
ic
a
ti
on r
e
s
ul
t
s
w
it
h only HOG
A
l
gor
i
t
hm
H
O
G
A
c
c
ur
a
c
y
P
r
e
c
i
s
i
on
R
e
c
a
l
l
F
1
s
c
or
e
K
N
N
H
O
G
0.8058
0.7288
0.9149
0.8113
S
V
M
H
O
G
0.8932
0.9091
0.8511
0.8791
DT
HOG
0.6408
0.6087
0.5957
0.6022
M
L
P
H
O
G
0.8058
0.9655
0.5957
0.7368
RF
HOG
0.8340
0.8947
0.7234
0.7990
H
i
s
t
g
r
a
di
e
nt
H
O
G
0.8350
0.8409
0.7872
0.8132
G
r
a
di
e
nt
b
oos
t
H
O
G
0.7660
0.7556
0.7234
0.7391
A
da
B
oos
t
H
O
G
0.6505
0.6122
0.6383
0.6250
PA
HOG
0.8835
0.8431
0.9149
0.8772
LR
HOG
0.9029
0.8936
0.8936
0.8936
T
a
b
l
e
3
o
f
H
O
G
w
i
th
P
C
A
f
e
a
t
ur
e
s
i
n
di
c
a
t
e
s
t
he
h
ig
h
e
s
t
hi
t
pe
r
c
e
n
t
a
g
e
s
a
nd
w
h
e
r
e
t
h
e
b
e
s
t
m
e
tr
i
c
s
a
r
e
n
o
t
e
d
w
it
h
b
ol
d
m
a
r
k
s
.
T
h
e
r
e
s
u
lt
s
i
nd
i
c
a
t
e
t
h
a
t
f
e
a
t
ur
e
e
xt
r
a
c
ti
o
n
u
s
i
n
g
H
O
G
in
c
o
m
b
in
a
ti
on
w
i
th
t
h
e
P
C
A
r
e
du
c
ti
on
a
lo
ng
w
i
t
h
p
a
s
s
i
v
e
-
a
ggr
e
s
s
iv
e
c
la
s
s
if
i
c
a
ti
o
n
pr
o
du
c
e
s
m
a
xi
m
um
s
uc
c
e
s
s
r
a
te
s
.
T
h
e
pr
e
c
i
s
i
on
r
a
t
e
i
n
p
a
ti
e
nt
i
d
e
n
ti
f
i
c
a
t
io
n
r
e
a
c
h
e
d
9
1
.0
7%
w
hi
l
e
r
e
c
a
ll
a
c
hi
e
v
e
d
9
2
.7
3%
a
n
d F
1
s
c
o
r
e
a
m
ou
n
te
d
t
o 89
.
36%
a
c
c
or
di
n
g
t
o
t
h
e
t
a
b
le
a
na
ly
s
i
s
.
T
h
e
a
c
c
ur
a
c
y
9
1
.8
9%
of
a
l
l
c
or
r
e
c
t
d
e
t
e
c
t
io
ns
b
e
tw
e
e
n
no
r
m
a
l
a
nd
a
b
no
r
m
a
l
s
p
e
c
im
e
n
s
.
A
s
uc
c
e
s
s
pe
r
c
e
nt
a
ge
a
n
a
ly
s
is
of
our
da
ta
th
r
ough
H
O
G
-
L
D
A
c
om
bi
na
ti
on
a
ppe
a
r
s
in
T
a
bl
e
4.
A
m
ong
th
e
r
e
s
ul
ts
th
e
m
os
t
s
ig
ni
f
ic
a
nt
va
lu
e
s
s
ta
nd
out
th
r
ough
bol
d
f
or
m
a
tt
in
g.
A
c
c
or
di
ng
to
th
e
r
e
s
e
a
r
c
h
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
, V
ol
.
14
, N
o.
3
,
J
une
2025
:
2132
-
2145
2140
f
in
di
ngs
H
O
G
f
e
a
tu
r
e
s
to
ge
th
e
r
w
it
h
L
D
A
di
m
e
ns
io
na
li
ty
r
e
du
c
ti
on
a
c
hi
e
ve
m
a
xi
m
um
s
uc
c
e
s
s
r
a
te
s
th
r
ough
K
N
N
c
la
s
s
if
ic
a
ti
on.
T
he
a
bi
li
ty
to
di
f
f
e
r
e
nt
ia
te
pa
ti
e
nt
s
pr
e
c
is
io
n
a
c
hi
e
ve
d
76.59%
s
uc
c
e
s
s
r
a
te
f
ol
lo
w
e
d
by
76.59%
r
e
c
a
ll
,
a
nd
76.59%
F
1
s
c
or
e
a
c
c
or
di
ng
to
th
e
ta
bl
e
e
xa
m
in
a
ti
on.
T
he
a
c
c
ur
a
c
y
w
hi
c
h
m
e
a
s
ur
e
s
c
or
r
e
c
t
de
te
c
ti
on r
a
te
s
ta
nds
a
t
78.64%
f
or
a
ll
nor
m
a
l
a
nd a
bnor
m
a
l
c
a
s
e
s
.
T
a
b
l
e
5
w
h
e
n
u
s
in
g
H
O
G
w
it
h
I
C
A
.
E
a
c
h
e
nt
r
y
in
t
h
e
d
i
s
p
l
a
y
e
d
t
a
b
l
e
i
nd
ic
a
t
e
s
r
e
s
u
lt
s
a
s
t
h
e
h
ig
h
e
s
t
p
o
in
t
b
e
f
or
e
d
if
f
e
r
e
nt
m
e
tr
i
c
i
n
bo
ld
f
a
c
e
.
T
he
t
a
b
l
e
d
e
m
o
n
s
tr
a
te
s
t
h
a
t
m
a
x
im
u
m
s
u
c
c
e
s
s
oc
c
ur
s
w
he
n
H
O
G
f
e
a
t
ur
e
s
e
x
tr
a
c
t
d
a
t
a
w
h
il
e
a
p
pl
yi
n
g
I
C
A
di
m
e
n
s
io
n
a
l
it
y
r
e
d
u
c
t
io
n
c
o
m
bi
ne
d
w
i
th
A
d
a
B
oo
s
t
c
l
a
s
s
i
f
i
e
r
m
o
d
e
l
.
E
x
a
m
in
a
ti
on
of
t
h
e
t
a
bl
e
s
h
o
w
s
pr
e
c
i
s
i
on
46
.
15%
s
t
a
n
d
s
a
s
t
he
h
i
gh
e
s
t
r
e
s
u
lt
w
h
il
e
r
e
c
a
l
l
3
8
.2
0%
a
n
d
F
1
s
c
or
e
4
1
.8
0%
a
r
e
a
l
s
o
i
n
c
l
u
de
d
.
A
l
l
N
or
m
a
l
a
n
d
a
bn
or
m
a
l
c
a
s
e
s
h
a
d
a
c
c
u
r
a
c
y
r
a
t
e
o
f
51
.
46%
w
h
e
n
c
or
r
e
c
tl
y
d
e
t
e
c
t
e
d.
T
a
bl
e
3
.
C
la
s
s
if
ic
a
ti
on r
e
s
ul
t
s
w
it
h H
O
G
a
nd P
C
A
A
l
gor
i
t
hm
H
O
G
a
nd P
C
A
A
c
c
ur
a
c
y
P
r
e
c
i
s
i
on
R
e
c
a
l
l
F
1
s
c
or
e
K
N
N
H
O
G
a
nd P
C
A
0.8058
0.7288
0.9149
0.8113
S
V
M
H
O
G
a
nd P
C
A
0.8252
0.7636
0.8936
0.8235
PA
H
O
G
a
nd P
C
A
0.9126
0.9107
0.9273
0.9189
DT
H
O
G
a
nd P
C
A
0.6990
0.6071
0.7907
0.6876
M
L
P
H
O
G
a
nd P
C
A
0.9126
0.8958
0.9149
0.9053
RF
H
O
G
a
nd P
C
A
0.6893
0.9412
0.3404
0.5000
H
i
s
t
gr
a
di
e
nt
boos
t
i
ng
H
O
G
a
nd P
C
A
0.8738
0.7931
0.9787
0.8762
GB
H
O
G
a
nd P
C
A
0.8252
0.7544
0.9149
0.8269
A
da
B
oos
t
H
O
G
a
nd P
C
A
0.7087
0.6269
0.8936
0.7368
LR
H
O
G
a
nd P
C
A
0.9029
0.8936
0.8936
0.8936
T
a
bl
e
4.
C
la
s
s
if
ic
a
ti
on r
e
s
ul
t
s
w
it
h H
O
G
a
nd L
D
A
A
l
gor
i
t
hm
A
c
c
ur
a
c
y
P
r
e
c
i
s
i
on
R
e
c
a
l
l
F
1
s
c
or
e
KNN
H
O
G
a
nd L
D
A
0.7864
0.7659
0.7659
0.7659
S
V
M
H
O
G
a
nd L
D
A
0.7767
0.7609
0.7447
0.7527
PA
H
O
G
a
nd L
D
A
0.7573
0.7200
0.7659
0.7427
DT
H
O
G
a
nd L
D
A
0.6408
0.5926
0.6809
0.6333
M
L
P
H
O
G
a
nd
L
D
A
0.7660
0.7447
0.7447
0.7447
RF
H
O
G
a
nd L
D
A
0.6408
0.5926
0.6809
0.6333
H
i
s
t
gr
a
di
e
nt
boos
t
H
O
G
a
nd
L
D
A
0.7767
0.7727
0.7234
0.7477
GB
H
O
G
a
nd
L
D
A
0.7573
0.7292
0.7447
0.7368
A
da
B
oos
t
H
O
G
a
nd L
D
A
0.7660
0.7347
0.7659
0.7500
LR
H
O
G
a
nd
L
D
A
0.7669
0.7346
0.7659
0.7499
T
a
bl
e
5
.
C
la
s
s
if
ic
a
ti
on r
e
s
ul
t
s
w
it
h H
O
G
a
nd I
C
A
A
l
gor
i
t
hm
A
c
c
ur
a
c
y
P
r
e
c
i
s
i
on
R
e
c
a
l
l
F
1
s
c
or
e
K
N
N
H
O
G
a
nd L
D
A
0.4563
0.3953
0.3617
0.3775
S
V
M
H
O
G
a
nd L
D
A
0.4757
0.4653
1.0
0.6347
M
L
P
H
O
G
a
nd
L
D
A
0.4466
0.4242
0.5957
0.4966
PA
H
O
G
a
nd L
D
A
0.4369
0.3778
0.3617
0.3696
DT
H
O
G
a
nd L
D
A
0.5049
0.45
0.382
0.413
RF
H
O
G
a
nd L
D
A
0.5144
0.4612
0.382
0.4181
H
i
s
t
gr
a
di
e
nt
boos
t
H
O
G
a
nd
L
D
A
0.4757
0.4478
0.6383
0.5279
GB
H
O
G
a
nd
L
D
A
0.5049
0.4773
0.8936
0.6211
A
da
B
oos
t
H
O
G
a
nd L
D
A
0.5146
0.4615
0.382
0.418
LR
H
O
G
a
nd
L
D
A
0.4175
0.3673
0.382
0.3746
5.
D
I
S
C
U
S
S
I
O
N
S
T
he
m
a
in
obj
e
c
ti
ve
of
th
i
s
r
e
s
e
a
r
c
h
pr
oj
e
c
t
is
to
de
ve
lo
p
a
di
a
g
nos
ti
c
s
y
s
te
m
w
hi
c
h
de
te
c
t
s
c
h
a
nge
s
in
B
M
C
L
V
B
dur
in
g
th
e
c
li
ni
c
a
l
pr
oc
e
s
s
.
T
he
s
tu
dy
to
ge
th
e
r
w
it
h
it
s
r
e
s
ul
ts
de
m
on
s
tr
a
te
e
nough
e
vi
de
nc
e
to
a
c
c
om
pl
is
h
th
e
d
e
te
r
m
in
e
d
goa
l.
T
he
s
y
s
te
m
c
om
bi
ne
s
ML
w
it
h
f
e
a
tu
r
e
e
xt
r
a
c
ti
on
th
r
ough
H
O
G
a
nd
di
m
e
ns
io
na
li
ty
r
e
duc
ti
on via
P
C
A
, L
D
A
,
a
nd I
C
A
t
e
c
hni
que
s
a
s
s
how
n by F
ig
ur
e
2.
T
a
bl
e
6
s
h
ow
s
th
e
r
e
s
ul
t
s
obt
a
i
ne
d
f
r
om
our
da
t
a
s
e
t
a
nd
th
e
c
om
pa
r
i
s
on
of
M
L
pe
r
f
or
m
a
nc
e
a
c
r
o
s
s
f
e
a
tu
r
e
e
xt
r
a
c
ti
on
m
e
th
o
ds
a
nd
di
m
e
n
s
io
n
a
li
ty
r
e
duc
ti
on.
T
he
ut
il
iz
a
ti
on
of
p
a
s
s
iv
e
-
a
ggr
e
s
s
iv
e
le
a
r
ni
ng
w
it
h
H
O
G
f
or
f
e
a
tu
r
e
e
x
tr
a
c
ti
on a
nd
P
C
A
f
or
di
m
e
n
s
io
na
li
ty
r
e
duc
ti
on
yi
e
ld
e
d
no
ta
bl
e
r
e
s
ul
ts
.
A
s
our
in
ve
s
ti
ga
ti
on
s
how
s
,
t
hi
s
a
ppr
o
a
c
h
pr
ovi
d
e
s
a
n
e
xt
r
a
or
di
na
r
y
a
c
c
ur
a
c
y
of
9
1.
26%
,
w
hi
c
h
s
ig
ni
f
i
e
s
t
ha
t
it
is
e
f
f
e
c
ti
ve
w
he
n
us
e
d
f
or
t
he
e
va
lu
a
ti
on
of
c
ha
nge
s
in
th
e
bon
e
m
a
r
r
ow
.
B
e
s
i
de
s
,
if
w
e
lo
o
k
a
t
th
e
pr
e
c
is
i
on
pa
r
t
,
it
r
e
a
ds
91.07%
,
in
di
c
a
ti
n
g
t
ha
t
th
i
s
m
ode
l
w
il
l
gr
e
a
tl
y
r
e
d
uc
e
f
a
ls
e
p
os
it
iv
e
s
,
w
hi
c
h
e
vi
d
e
nc
e
s
th
e
pr
o
bl
e
m
,
th
a
t
i
s
,
to
im
pr
ove
th
e
di
a
gno
s
i
s
a
nd
tr
e
a
tm
e
nt
of
pa
ti
e
nt
s
.
E
qu
a
ll
y
im
p
or
ta
nt
ly
,
th
e
hi
gh
r
e
c
a
ll
r
a
te
of
9
2.73%
.
T
he
F
1
s
c
or
e
r
a
te
is
9
1.89%
,
a
nd
it
in
c
r
e
a
s
e
s
c
onf
id
e
nc
e
a
bo
ut
th
e
m
ode
l
in
te
r
m
s
of
a
good
b
a
la
n
c
e
b
e
twe
e
n
th
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
E
ar
ly
de
te
c
ti
on and c
la
s
s
if
ic
at
io
n of
bone
m
ar
r
o
w
c
hange
s
i
n l
um
bar
v
e
r
t
e
br
ae
…
(
Y
as
ir
H
us
s
e
in
Shak
i
r
)
2141
f
a
ls
e
-
pos
i
ti
ve
r
a
t
e
a
nd
t
he
f
a
l
s
e
-
n
e
ga
ti
ve
r
a
te
.
I
n
c
li
n
ic
a
l
pr
a
c
t
ic
e
,
th
e
r
e
i
s
of
te
n
a
ne
e
d
f
or
a
pe
r
f
e
c
t
s
to
r
m
s
c
e
na
r
io
w
he
r
e
s
e
n
s
it
i
vi
ty
a
n
d
s
p
e
c
if
i
c
it
y
a
r
e
b
ot
h
opt
im
a
l,
w
hi
c
h
i
s
a
c
hi
e
v
e
d
he
r
e
.
T
he
r
e
f
or
e
,
th
e
s
tu
d
y
pr
ovi
de
s
e
vi
d
e
nc
e
on
how
th
e
p
a
s
s
iv
e
-
a
ggr
e
s
s
iv
e
l
e
a
r
ni
ng
a
ppr
oa
c
h
a
l
ong
s
id
e
H
O
G
a
nd
P
C
A
c
a
n
be
u
s
e
d
to
e
s
ti
m
a
te
th
e
c
ha
n
ge
i
n
b
one
m
a
r
r
o
w
in
th
e
lu
m
ba
r
ve
r
t
e
br
a
e
f
r
o
m
M
R
I
d
a
ta
.
B
y a
r
ti
f
ic
i
a
l
in
te
ll
ig
e
n
c
e
a
nd
M
L
,
th
is
p
a
pe
r
pr
e
s
e
nt
s
a
r
obu
s
t
a
ppr
o
a
c
h
to
e
nha
nc
e
di
a
gno
s
ti
c
a
c
c
ur
a
c
y
a
n
d
s
uppor
t
c
li
ni
c
ia
ns
dur
in
g
tr
e
a
tm
e
nt
pl
a
nni
n
g.
H
o
w
e
v
e
r
,
f
ut
ur
e
s
tu
di
e
s
a
r
e
n
e
e
de
d
t
o
e
xt
e
nd
a
nd
c
or
r
obor
a
te
th
is
m
e
th
odol
o
gy
in
d
if
f
e
r
e
nt
c
li
ni
c
a
l
c
ont
e
xt
s
t
o c
he
c
k t
h
e
e
xt
e
nt
of
i
t
s
c
r
o
s
s
-
s
e
c
ti
o
na
l
a
nd i
nt
e
r
v
e
nt
io
na
l
tr
a
ns
l
a
ti
on
pot
e
nt
ia
l
.
F
ig
ur
e
2
.
C
om
pa
r
is
on of
ML
a
lg
or
it
hm
pe
r
f
o
r
m
a
nc
e
a
c
r
os
s
f
e
a
t
ur
e
e
xt
r
a
c
ti
on me
th
ods
a
nd dim
e
ns
io
na
li
ty
r
e
duc
ti
on
T
a
b
l
e
6
.
C
o
m
p
a
r
i
s
o
n
o
f
M
L
a
l
g
o
r
i
t
h
m
p
e
r
f
o
r
m
a
n
c
e
a
c
r
o
s
s
f
e
a
t
u
r
e
e
x
t
r
a
c
t
i
o
n
m
e
t
h
o
d
s
a
n
d
d
i
m
e
n
s
i
o
n
a
l
i
t
y
r
e
d
u
c
t
i
o
n
A
l
gor
i
t
hm
a
nd
F
E
+D
R
A
c
c
ur
a
c
y
P
r
e
c
i
s
i
on
R
e
c
a
l
l
F
1 S
c
or
e
K
N
N
H
O
G
0.8058
0.7288
0.9149
0.8113
S
V
M
H
O
G
0.8932
0.9091
0.8511
0.8791
DT
HOG
0.6408
0.6087
0.5957
0.6022
M
L
P
H
O
G
0.8058
0.9655
0.5957
0.7368
RF
HOG
0.8340
0.8947
0.7234
0.7990
H
i
s
t
g
r
a
di
e
nt
H
O
G
0.8350
0.8409
0.7872
0.8132
GB
HOG
0.7660
0.7556
0.7234
0.7391
A
da
B
oos
t
H
O
G
0.6505
0.6122
0.6383
0.6250
PA
HOG
0.8835
0.8431
0.9149
0.8772
LR
HOG
0.9029
0.8936
0.8936
0.8936
K
N
N
H
O
G
a
nd P
C
A
0.8058
0.7288
0.9149
0.8113
S
V
M
H
O
G
a
nd P
C
A
0.8252
0.7636
0.8936
0.8235
PA
H
O
G
a
nd P
C
A
0.9126
0.9107
0.9273
0.9189
DT
H
O
G
a
nd P
C
A
0.6990
0.6071
0.7907
0.6876
M
L
P
H
O
G
a
nd P
C
A
0.9126
0.8958
0.9149
0.9053
RF
H
O
G
a
nd P
C
A
0.6893
0.9412
0.3404
0.5000
H
i
s
t
g
r
a
di
e
nt
H
O
G
a
nd P
C
A
0.8738
0.7931
0.9787
0.8762
GB
H
O
G
a
nd P
C
A
0.8252
0.7544
0.9149
0.8269
A
da
B
oos
t
H
O
G
a
nd P
C
A
0.7087
0.6269
0.8936
0.7368
LR
H
O
G
a
nd P
C
A
0.9029
0.8936
0.8936
0.8936
K
N
N
H
O
G
a
nd L
D
A
0.7864
0.7659
0.7659
0.7659
S
V
M
H
O
G
a
nd L
D
A
0.7767
0.7609
0.7447
0.7527
PA
H
O
G
a
nd L
D
A
0.7573
0.7200
0.7659
0.7427
DT
H
O
G
a
nd L
D
A
0.6408
0.5926
0.6809
0.6333
M
L
P
H
O
G
a
nd
L
D
A
0.7660
0.7447
0.7447
0.7447
RF
H
O
G
a
nd L
D
A
0.6408
0.5926
0.6809
0.6333
H
i
s
t
g
r
a
di
e
nt
H
O
G
a
nd L
D
A
0.7767
0.7727
0.7234
0.7477
GB
H
O
G
a
nd
L
D
A
0.7573
0.7292
0.7447
0.7368
A
da
B
oos
t
H
O
G
a
nd L
D
A
0.7660
0.7347
0.7659
0.7500
LR
H
O
G
a
nd
L
D
A
0.7669
0.7346
0.7659
0.7499
K
N
N
H
O
G
a
nd L
D
A
0.4563
0.3953
0.3617
0.3775
S
V
M
H
O
G
a
nd L
D
A
0.4757
0.4653
1.0
0.6347
M
L
P
H
O
G
a
nd
L
D
A
0.4466
0.4242
0.5957
0.4966
PA
H
O
G
a
nd L
D
A
0.4369
0.3778
0.3617
0.3696
DT
H
O
G
a
nd L
D
A
0.5049
0.4500
0.3820
0.4130
RF
H
O
G
a
nd L
D
A
0.5144
0.4612
0.3820
0.4181
H
i
s
t
g
r
a
di
e
nt
H
O
G
a
nd L
D
A
0.4757
0.4478
0.6383
0.5279
GB
H
O
G
a
nd
L
D
A
0.5049
0.4773
0.8936
0.6211
A
da
B
oos
t
H
O
G
a
nd L
D
A
0.5146
0.4615
0.3820
0.4180
Evaluation Warning : The document was created with Spire.PDF for Python.