T
E
L
K
O
M
NIKA
T
elec
o
mm
un
ica
t
io
n Co
m
pu
t
i
ng
E
lect
ro
nics
a
nd
Co
ntr
o
l
Vo
l.
24
,
No
.
5
,
Octo
b
er
20
26
,
p
p
.
1
7
2
8
~
1
7
4
4
I
SS
N:
1
6
9
3
-
6
9
3
0
,
DOI
: 1
0
.
1
2
9
2
8
/
T
E
L
KOM
NI
K
A
.
v
24
i
5
.
27976
1728
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//telko
mn
ika
.
u
a
d
.
a
c.
i
d
M
RI seg
m
e
ntatio
n and
clas
sifica
tio
n using
1
-
D and
2
-
D
e
m
pi
rica
l
m
o
de
d
eco
m
po
sitio
n
G
ia
ng
H
o
ng
Ng
uy
en
1,
2
,
Yen
T
hi H
o
a
ng
H
ua
1
,
Ng
a
n Vuo
ng
T
hu
y
Ng
uy
en
1
,
L
iet
Va
n
Da
ng
3
1
D
e
p
a
r
t
me
n
t
o
f
P
h
y
si
c
s
a
n
d
C
o
mp
u
t
e
r
S
c
i
e
n
c
e
,
F
a
c
u
l
t
y
o
f
P
h
y
si
c
s a
n
d
En
g
i
n
e
e
r
i
n
g
P
h
y
si
c
s,
U
n
i
v
e
r
si
t
y
o
f
S
c
i
e
n
c
e
,
V
i
e
t
N
a
m N
a
t
i
o
n
a
l
U
n
i
v
e
r
si
t
y
,
H
o
C
h
i
M
i
n
h
C
i
t
y
,
V
i
e
t
n
a
m
2
D
e
p
a
r
t
me
n
t
o
f
G
e
n
e
r
a
l
Ed
u
c
a
t
i
o
n
,
C
a
o
T
h
a
n
g
T
e
c
h
n
i
c
a
l
C
o
l
l
e
g
e
,
H
o
C
h
i
M
i
n
h
C
i
t
y
,
V
i
e
t
n
a
m
3
D
e
p
a
r
t
me
n
t
o
f
N
a
t
u
r
e
,
D
o
n
g
N
a
i
U
n
i
v
e
r
si
t
y
,
T
a
m H
i
e
p
W
a
r
d
,
D
o
n
g
N
a
i
C
i
t
y
,
V
i
e
t
n
a
m
Art
i
cle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
A
p
r
1
,
2
0
2
6
R
ev
i
s
ed
J
u
l
22
,
2
0
2
6
A
cc
ep
ted
A
u
g
28
,
2026
T
h
is
stu
d
y
p
ro
p
o
se
s
a
n
e
w
f
ra
m
e
w
o
rk
f
o
r
b
ra
in
m
a
g
n
e
ti
c
re
so
n
a
n
c
e
im
a
g
in
g
(
M
RI
)
t
u
m
o
r
d
e
tec
ti
o
n
a
n
d
c
las
sif
ica
ti
o
n
u
si
n
g
1
-
D
a
n
d
2
-
D
e
m
p
iri
c
a
l
m
o
d
e
d
e
c
o
m
p
o
siti
o
n
(EM
D
)
/
b
id
ime
n
sio
n
a
l
e
m
p
iri
c
a
l
m
o
d
e
d
e
c
o
m
p
o
siti
o
n
(
BEM
D)
.
T
h
e
se
a
d
a
p
ti
v
e
a
lg
o
rit
h
m
s
o
p
ti
m
ize
i
m
a
g
e
d
e
n
o
isi
n
g
,
se
g
m
e
n
tatio
n
a
n
d
f
e
a
tu
re
e
x
tra
c
ti
o
n
.
T
u
m
o
rs
a
re
d
e
tec
t
e
d
u
sin
g
a
g
lo
b
a
l
th
re
sh
o
l
d
d
e
riv
e
d
f
ro
m
EM
D
-
g
e
n
e
ra
ted
b
im
o
d
a
l
c
u
rv
e
s.
A
v
isu
a
l
g
e
o
m
e
tr
y
g
ro
u
p
1
9
(
V
G
G
1
9
)
m
o
d
e
l
e
x
trac
ts
f
e
a
tu
re
s
f
ro
m
th
e
re
g
io
n
s
o
f
in
tere
st
(ROI)
a
c
ro
ss
b
o
th
d
e
n
o
ise
d
im
a
g
e
s
a
n
d
BEM
D
c
o
m
p
o
n
e
n
ts.
T
h
e
m
e
th
o
d
w
a
s
v
a
li
d
a
ted
o
n
th
e
p
u
b
li
c
F
ig
sh
a
re
b
ra
in
t
u
m
o
r
d
a
tas
e
t
.
F
o
r
se
g
m
e
n
tatio
n
,
th
e
p
ro
p
o
se
d
m
e
th
o
d
p
e
rf
o
rm
e
d
sim
il
a
rl
y
to
m
u
lt
il
e
v
e
l
Otsu
,
Ka
p
u
r
-
Ha
r
m
o
n
ic,
a
n
d
K
-
m
e
a
n
s
a
lg
o
rit
h
m
s.
It
a
c
h
iev
e
d
a
n
a
v
e
r
a
g
e
p
r
e
c
isio
n
o
f
9
6
.
7
0
%
,
slig
h
t
ly
traili
n
g
K
-
m
e
a
n
s
o
f
9
7
.
0
7
%
,
a
n
d
a
se
n
siti
v
it
y
o
f
9
1
.
0
7
%
,
o
u
t
p
e
rf
o
rm
in
g
K
-
m
e
a
n
s
b
u
t
f
a
ll
in
g
b
e
h
i
n
d
Ka
p
u
r
a
n
d
Otsu
.
N
o
tab
ly
,
it
a
c
h
iev
e
d
th
e
h
i
g
h
e
st
F
1
-
sc
o
re
o
f
9
3
.
7
0
%
,
ty
in
g
th
e
Otsu
tec
h
n
i
q
u
e
.
T
h
e
se
re
su
lt
s
h
ig
h
li
g
h
t
e
n
h
a
n
c
e
d
t
u
m
o
r
lo
c
a
li
z
a
ti
o
n
,
re
f
lec
ti
n
g
se
g
m
e
n
tatio
n
th
a
t
c
lo
se
ly
a
li
g
n
s
w
it
h
th
e
g
ro
u
n
d
tru
th
.
F
o
r
tu
m
o
r
c
las
sif
ica
ti
o
n
,
e
n
se
m
b
le
lea
rn
in
g
w
a
s
u
se
d
to
d
e
v
e
lo
p
n
in
e
m
o
d
e
ls,
a
ll
o
f
w
h
ich
su
rp
a
ss
e
d
9
3
%
a
c
c
u
ra
c
y
;
o
n
e
a
c
h
iev
e
d
1
0
0
%
,
a
n
d
t
h
re
e
re
a
c
h
e
d
9
6
.
7
%
.
T
h
e
p
ro
p
o
se
d
m
e
th
o
d
o
u
tp
e
rf
o
rm
e
d
se
v
e
n
sta
te
-
of
-
th
e
-
a
rt
tec
h
n
i
q
u
e
s
w
it
h
a
p
e
a
k
a
c
c
u
ra
c
y
o
f
9
9
.
0
2
%
,
d
e
m
o
n
stra
ti
n
g
it
s c
li
n
i
c
a
l
v
iab
il
it
y
.
O
v
e
ra
ll
,
in
teg
ra
ti
n
g
EM
D an
d
BEM
D
a
lg
o
rit
h
m
s
p
ro
v
i
d
e
s
a
su
p
e
ri
o
r
f
ra
m
e
w
o
rk
f
o
r
b
ra
in
t
u
m
o
r
se
g
m
e
n
tatio
n
a
n
d
c
las
sif
ica
ti
o
n
c
o
m
p
a
re
d
trad
it
io
n
a
l
m
e
th
o
d
s.
K
ey
w
o
r
d
s
:
B
id
im
e
n
s
io
n
al
e
m
p
ir
ical
m
o
d
e
d
ec
o
m
p
o
s
itio
n
B
id
im
e
n
s
io
n
al
i
n
tr
i
n
s
ic
m
o
d
e
f
u
n
ctio
n
E
m
p
ir
ical
m
o
d
e
d
ec
o
m
p
o
s
itio
n
E
m
p
ir
ical
w
av
ele
t tr
an
s
f
o
r
m
V
GG
19
-
la
y
er
m
o
d
el
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
:
Gian
g
Ho
n
g
Ng
u
y
en
Dep
ar
t
m
en
t o
f
Gen
er
al
E
d
u
ca
t
io
n
,
C
ao
T
h
an
g
T
ec
h
n
ica
l Co
l
leg
e,
Ho
C
h
i M
in
h
C
it
y
65
Hu
y
n
h
T
h
u
c
K
h
an
g
,
Sai G
o
n
W
ar
d
,
Ho
C
h
i M
in
h
C
it
y
,
Vietn
a
m
E
m
ail:
n
g
u
y
en
h
o
n
g
g
ia
n
g
@
ca
o
th
an
g
.
ed
u
.
v
n
; 2
2
n
3
8
1
0
1
@
s
t
u
d
en
t.
h
c
m
u
s
.
ed
u
.
v
n
1.
I
NT
RO
D
UCT
I
O
N
T
h
e
b
r
ain
is
a
v
ital
b
u
t
co
m
p
lex
o
r
g
an
co
n
ta
in
i
n
g
ap
p
r
o
x
i
m
atel
y
8
6
b
illi
o
n
n
e
u
r
o
n
s
,
c
o
n
tr
o
l
lin
g
v
o
lu
n
tar
y
a
n
d
in
v
o
l
u
n
tar
y
f
u
n
ctio
n
s
t
h
r
o
u
g
h
t
h
e
s
o
m
atic
a
n
d
au
to
n
o
m
ic
n
er
v
o
u
s
s
y
s
te
m
s
.
B
r
ain
in
j
u
r
ies
ar
e
class
i
f
ied
as
tr
a
u
m
at
ic
o
r
n
o
n
-
tr
au
m
a
tic,
w
it
h
b
r
ain
tu
m
o
r
s
r
ep
r
esen
tin
g
a
m
aj
o
r
m
ed
ical
c
o
n
ce
r
n
.
T
h
e
Glo
b
al
C
an
ce
r
Ob
s
er
v
ato
r
y
(
G
L
OB
O
C
A
N
)
2
0
2
2
r
ep
o
r
t
in
d
icate
d
ab
o
u
t
3
2
1
,
7
3
1
g
lo
b
al
b
r
ain
tu
m
o
r
ca
s
es,
r
es
u
lti
n
g
in
2
4
8
,
5
0
0
d
ea
th
s
,
m
ai
n
l
y
i
n
Asi
a
[
1
]
.
E
ar
ly
d
etec
tio
n
s
i
g
n
i
f
ica
n
tl
y
i
m
p
r
o
v
e
s
p
atien
t
o
u
tc
o
m
e
s
,
b
u
t
t
h
e
s
k
u
ll
s
tr
u
ct
u
r
e
m
a
k
es
d
iag
n
o
s
is
c
h
allen
g
in
g
.
No
n
-
i
n
v
a
s
i
v
e
i
m
a
g
in
g
tec
h
n
iq
u
es,
p
ar
ticu
lar
l
y
m
ag
n
etic
r
eso
n
an
ce
i
m
a
g
in
g
(
MRI
)
,
p
lay
a
n
i
m
p
o
r
tan
t
r
o
le
b
ec
au
s
e
th
e
y
p
r
o
v
id
e
s
af
e
an
d
h
ig
h
-
r
eso
l
u
tio
n
i
m
ag
es
[
2
]
.
A
cc
u
r
ate
MRI
i
n
ter
p
r
etatio
n
d
ep
en
d
s
o
n
r
ad
io
lo
g
is
t
e
x
p
er
tis
e
an
d
i
m
a
g
e
q
u
al
it
y
.
T
o
im
p
r
o
v
e
d
iag
n
o
s
tic
ac
c
u
r
ac
y
,
co
m
p
u
ter
-
a
id
ed
d
iag
n
o
s
i
s
(
C
A
D
x
)
s
y
s
te
m
s
h
a
v
e
b
ee
n
d
ev
elo
p
ed
.
T
h
ese
s
y
s
te
m
s
t
y
p
icall
y
in
cl
u
d
e
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
MRI
s
eg
men
ta
tio
n
a
n
d
cla
s
s
ifica
tio
n
u
s
in
g
1
-
D
a
n
d
2
-
D
emp
ir
ica
l m
o
d
e
…
(
Gia
n
g
Ho
n
g
N
g
u
ye
n
)
1729
p
r
ep
r
o
ce
s
s
in
g
,
s
e
g
m
e
n
tat
io
n
,
an
d
class
i
f
icat
io
n
u
s
i
n
g
tr
ad
i
tio
n
al
m
ac
h
i
n
e
lear
n
i
n
g
(
ML
)
o
r
d
ee
p
lear
n
in
g
(
DL
)
m
e
th
o
d
s
.
P
r
e
p
r
o
ce
s
s
in
g
en
h
a
n
ce
s
i
m
a
g
e
q
u
a
lit
y
an
d
r
e
m
o
v
e
s
ar
tif
ac
ts
to
e
n
s
u
r
e
ac
cu
r
ac
y
in
s
u
b
s
eq
u
en
t
s
tag
e
s
.
Se
g
m
en
ta
tio
n
t
h
e
n
i
s
o
lates
th
e
t
u
m
o
r
t
o
r
e
v
ea
l
i
ts
g
eo
m
etr
y
a
n
d
lo
ca
tio
n
,
wh
ile
cla
s
s
i
f
icat
io
n
ca
teg
o
r
izes
it
as
b
en
i
g
n
,
m
ali
g
n
a
n
t,
o
r
in
d
icativ
e
o
f
a
s
p
ec
if
ic
d
is
ea
s
e.
T
w
o
f
i
n
al
s
ta
g
es
en
h
a
n
ce
au
to
m
ated
w
o
r
k
f
lo
w
s
b
y
r
ed
u
ci
n
g
d
iag
n
o
s
tic
er
r
o
r
s
,
ac
ce
ler
atin
g
cli
n
i
ca
l
ass
e
s
s
m
en
ts
,
a
n
d
i
m
p
r
o
v
i
n
g
p
atie
n
t
o
u
tco
m
e
s
[
3
]
,
[
4
]
.
T
h
e
ch
allen
g
e
o
f
w
o
r
k
in
g
i
s
to
d
ev
elo
p
m
et
h
o
d
s
t
h
at
ar
e
b
o
th
ac
cu
r
ate
a
n
d
p
r
ac
tical
to
m
a
x
i
m
ize
th
eir
clin
ica
l
u
tili
t
y
.
C
u
r
r
en
tl
y
,
r
esear
ch
in
C
A
Dx
s
y
s
te
m
s
h
ea
v
i
l
y
e
m
p
h
as
izes
d
ee
p
lear
n
in
g
.
Ho
w
ev
er
,
th
e
n
ee
d
f
o
r
lar
g
e,
lab
eled
d
atase
ts
an
d
h
i
g
h
co
m
p
u
tatio
n
al
p
o
w
er
o
f
te
n
r
estricts
it
s
ap
p
licatio
n
in
d
ev
elo
p
in
g
r
eg
io
n
s
.
C
o
n
s
eq
u
e
n
t
l
y
,
tr
ad
itio
n
al
an
d
h
y
b
r
id
tech
n
iq
u
e
s
f
o
r
s
eg
m
en
tatio
n
an
d
class
i
f
ic
atio
n
r
e
m
ain
h
ig
h
l
y
r
elev
an
t.
Ho
w
ev
er
,
w
h
e
n
ap
p
lied
t
o
b
r
ain
MRI,
th
ey
f
ac
e
ch
alle
n
g
e
s
s
u
c
h
as
n
o
is
e,
in
te
n
s
it
y
v
ar
iatio
n
s
,
an
d
an
ato
m
ica
l
co
m
p
le
x
it
y
.
T
h
ese
is
s
u
es
lead
to
t
h
r
ee
k
e
y
li
m
ita
tio
n
s
:
(
i)
g
lo
b
al
th
r
es
h
o
ld
in
g
i
s
o
f
te
n
i
n
s
u
f
f
icie
n
t
f
o
r
d
etec
tin
g
tu
m
o
r
s
;
(
ii)
m
u
ltil
ev
el
t
h
r
esh
o
ld
i
n
g
r
eq
u
ir
es
a
tr
ial
-
an
d
-
er
r
o
r
ap
p
r
o
a
ch
to
d
eter
m
in
e
o
p
ti
m
al
th
r
es
h
o
ld
s
;
an
d
(
iii)
f
ea
tu
r
es
e
x
tr
ac
ted
f
r
o
m
d
en
o
i
s
ed
r
eg
io
n
s
o
f
in
ter
est
(
R
OI
s
)
m
a
y
o
v
er
l
o
o
k
s
u
b
tle
cli
n
ical
d
etails.
T
o
ad
d
r
ess
th
ese
li
m
itatio
n
s
,
t
h
is
p
ap
er
p
r
o
p
o
s
es
a
n
o
v
el
f
r
a
m
e
w
o
r
k
u
s
i
n
g
e
m
p
ir
ical
m
o
d
e
d
ec
o
m
p
o
si
tio
n
(
E
MD
,
1
-
D)
an
d
b
id
i
m
en
s
io
n
al
e
m
p
ir
ical
m
o
d
e
d
ec
o
m
p
o
s
it
io
n
(
B
E
MD
,
2
-
D)
f
o
r
s
ig
n
al
a
n
d
i
m
a
g
e
p
r
o
ce
s
s
i
n
g
.
T
h
ese
ad
ap
tiv
e,
m
u
l
tire
s
o
lu
tio
n
,
d
ata
-
d
r
iv
en
tec
h
n
iq
u
es
d
ec
o
m
p
o
s
e
c
o
m
p
le
x
n
o
n
-
li
n
ea
r
an
d
n
o
n
-
s
tatio
n
ar
y
d
ata
in
to
in
tr
i
n
s
ic
m
o
d
e
f
u
n
c
tio
n
s
(
I
M
Fs
)
an
d
b
id
im
e
n
s
io
n
al
in
tr
in
s
ic
m
o
d
e
f
u
n
c
tio
n
s
(
B
I
MFs)
th
r
o
u
g
h
a
“si
f
ti
n
g
p
r
o
ce
s
s
”
w
it
h
o
u
t
r
el
y
i
n
g
o
n
f
i
x
ed
b
asis
f
u
n
ctio
n
s
.
B
ec
au
s
e
e
ac
h
I
MF
a
n
d
B
I
MF
r
ep
r
esen
t
o
s
cillatio
n
s
at
d
is
ti
n
ct
f
r
eq
u
en
c
ies,
t
h
is
p
r
o
ce
s
s
s
i
g
n
i
f
ica
n
tl
y
e
n
h
a
n
ce
s
t
h
e
ca
p
tu
r
e
o
f
s
tr
u
ct
u
r
al
an
d
tex
t
u
r
al
d
etails
ac
r
o
s
s
d
i
f
f
er
en
t
s
ca
le
s
,
i
m
p
r
o
v
i
n
g
o
v
er
all
a
n
al
y
s
i
s
.
T
h
is
s
t
u
d
y
i
n
tr
o
d
u
ce
s
a
tu
m
o
r
d
etec
tio
n
ap
p
r
o
ac
h
th
at
i
n
te
g
r
ates
t
h
e
e
m
p
ir
ical
w
a
v
elet
tr
an
s
f
o
r
m
(
E
W
T
)
an
d
E
MD
,
w
h
i
le
u
s
i
n
g
B
E
MD
alo
n
g
s
id
e
v
is
u
al
g
eo
m
etr
y
g
r
o
u
p
1
9
(
VGG1
9
)
f
o
r
f
ea
tu
r
e
ex
tr
ac
tio
n
.
T
h
e
f
o
llo
w
i
n
g
s
ec
tio
n
r
e
v
ie
w
s
p
r
io
r
w
o
r
k
o
n
tr
ad
itio
n
al
an
d
h
y
b
r
id
s
eg
m
e
n
t
atio
n
an
d
clas
s
i
f
icatio
n
m
et
h
o
d
s
.
Sh
e
m
an
to
et
al
.
[
5
]
d
ev
elo
p
ed
HOFilter
f
o
r
MRI
t
u
m
o
r
d
e
tectio
n
,
u
tili
zi
n
g
a
m
ed
ian
f
i
lt
er
(
ap
p
lied
f
i
v
e
ti
m
es)
a
n
d
a
g
lo
b
al
2
5
5
-
s
ca
le
Ots
u
t
h
r
es
h
o
ld
.
T
h
e
th
r
esh
o
ld
is
d
i
v
id
ed
in
to
f
o
u
r
i
n
cr
e
m
en
ta
l
g
r
o
u
p
s
,
p
r
o
d
u
cin
g
9
4
.
4
6
%
a
cc
u
r
ac
y
o
n
1
9
8
Kag
g
le
d
ataset
M
R
I
s
.
H
o
w
e
v
er
,
th
e
e
x
ac
t
cr
iter
ia
f
o
r
t
h
e
ad
d
ed
th
r
es
h
o
ld
v
alu
e
s
r
e
m
ain
u
n
e
x
p
lain
ed
.
S
h
ar
m
a
e
t
a
l
.
[
6
]
p
r
o
p
o
s
ed
a
h
y
b
r
id
m
u
ltil
e
v
el
t
h
r
es
h
o
ld
in
g
tech
n
iq
u
e
p
o
w
er
ed
b
y
t
h
e
d
y
n
a
m
ic
o
p
p
o
s
ite
b
ald
ea
g
le
s
ea
r
ch
a
lg
o
r
it
h
m
.
T
h
is
m
et
h
o
d
u
s
es K
ap
u
r
’
s
en
tr
o
p
y
t
o
d
eter
m
i
n
e
o
p
ti
m
a
l
th
r
es
h
o
ld
v
al
u
es.
T
ested
o
n
MRIs
f
r
o
m
t
h
e
Fi
g
s
h
ar
e
d
at
aset
f
o
r
t
u
m
o
r
d
etec
tio
n
,
th
e
p
r
o
p
o
s
ed
a
p
p
r
o
ac
h
d
em
o
n
s
tr
ated
s
u
p
er
io
r
p
er
f
o
r
m
an
ce
o
v
er
th
e
B
ald
E
ag
le
S
ea
r
ch
an
d
m
u
lt
ip
le
o
th
er
m
u
l
tilev
el
t
h
r
es
h
o
ld
in
g
tech
n
iq
u
es.
Ak
an
et
a
l
.
[
7
]
p
r
o
p
o
s
ed
a
h
y
b
r
id
th
r
ee
-
d
i
m
en
s
io
n
al
(
3D
)
s
eg
m
e
n
tatio
n
m
et
h
o
d
u
s
in
g
h
i
s
to
g
r
a
m
s
m
o
o
th
in
g
,
m
u
lt
i
m
o
d
al
p
ar
ticl
e
s
w
ar
m
o
p
ti
m
izatio
n
,
a
n
d
n
o
n
-
E
u
clid
ea
n
p
ix
el
cl
u
s
ter
in
g
.
E
v
alu
a
ted
o
n
th
r
ee
d
atasets
,
t
h
e
m
o
d
el
ac
h
iev
ed
9
5
.
5
2
%
ac
cu
r
ac
y
an
d
8
5
.
1
5
%
r
ec
all.
Kr
in
id
i
s
et
a
l
.
[
8
]
p
r
o
p
o
s
ed
d
eter
m
i
n
i
n
g
th
e
g
lo
b
al
i
m
a
g
e
th
r
es
h
o
ld
u
s
in
g
t
h
e
E
MD
o
f
t
h
e
p
r
o
b
ab
ilit
y
m
as
s
f
u
n
ctio
n
(
P
MF)
.
T
h
e
y
co
n
s
tr
u
cted
a
th
r
es
h
o
ld
in
g
cu
r
v
e
b
y
s
u
m
m
i
n
g
t
h
e
I
MFs,
ex
cl
u
d
in
g
th
e
f
i
r
s
t
an
d
last
.
Ho
w
e
v
er
,
th
i
s
m
eth
o
d
is
li
m
ited
to
s
in
g
le
-
p
ea
k
ed
d
is
tr
ib
u
t
io
n
s
a
n
d
ca
n
n
o
t
s
u
p
p
o
r
t
m
u
ltil
e
v
el
th
r
esh
o
ld
i
n
g
.
R
ao
an
d
R
ed
d
y
[
9
]
ad
ap
ted
th
is
cu
r
v
e
-
b
u
ild
i
n
g
ap
p
r
o
ac
h
,
alth
o
u
g
h
it
lac
k
s
clea
r
p
ea
k
a
n
d
v
alle
y
d
ef
i
n
itio
n
s
.
T
o
d
eter
m
in
e
t
h
e
n
u
m
b
er
o
f
clu
s
ter
s
an
d
i
n
itial
ce
n
ter
s
f
o
r
f
u
zz
y
C
-
m
ea
n
s
s
e
g
m
e
n
tatio
n
,
th
e
y
ap
p
lied
a
d
o
u
b
le
-
c
u
t
te
ch
n
iq
u
e;
h
o
w
e
v
er
,
th
is
ap
p
r
o
ac
h
d
e
m
a
n
d
s
h
i
g
h
c
o
m
p
u
tatio
n
al
r
eso
u
r
c
es.
L
a
k
s
h
m
i
et
a
l
.
[
1
0
]
u
s
ed
b
i
later
al
f
ilter
i
n
g
a
n
d
ME
DU
-
Net+
f
o
r
n
o
is
e
r
e
m
o
v
a
l
an
d
s
e
m
a
n
tic
s
eg
m
e
n
tatio
n
i
n
b
r
ain
MRI
a
n
al
y
s
i
s
.
T
h
ey
u
s
ed
R
es
Net5
0
f
o
r
f
ea
tu
r
e
ex
tr
ac
tio
n
an
d
a
B
a
y
esia
n
r
eg
u
lar
ize
d
ML
tech
n
iq
u
e,
b
r
ain
,
f
o
r
t
u
m
o
r
ass
e
s
s
m
e
n
t.
Af
ter
f
in
e
-
t
u
n
i
n
g
b
r
ain
’
s
p
ar
a
m
eter
s
,
th
eir
m
eth
o
d
o
lo
g
y
class
i
f
ied
M
R
I
i
m
ag
e
s
f
r
o
m
t
h
e
Ka
g
g
le
d
ataset
i
n
to
t
h
r
ee
t
u
m
o
r
t
y
p
es
a
n
d
o
n
e
n
o
n
-
t
u
m
o
r
t
y
p
e,
ac
h
ie
v
i
n
g
a
n
accu
r
ac
y
o
f
9
7
.
7
5
%
af
ter
1
0
0
0
ep
o
ch
s
.
A
a
m
ir
et
a
l
.
[
1
1
]
p
r
o
p
o
s
ed
a
h
y
b
r
id
b
r
ain
tu
m
o
r
cla
s
s
i
f
icatio
n
f
r
a
m
e
w
o
r
k
u
s
i
n
g
an
is
o
tr
o
p
ic
Gau
s
s
ia
n
d
en
o
i
s
in
g
,
m
o
r
p
h
o
lo
g
ical
en
h
a
n
ce
m
en
t,
U
-
Net
r
eg
io
n
o
f
i
n
ter
est
(
R
OI
)
ex
tr
ac
tio
n
,
an
d
m
u
lti
s
c
ale
Den
s
eNe
t
f
ea
t
u
r
e
e
x
tr
ac
ti
o
n
.
T
h
e
cla
s
s
i
f
ica
tio
n
u
s
ed
r
a
n
d
o
m
f
o
r
est
(
R
F),
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
i
n
e
(
SVM
)
,
an
d
K
-
n
ea
r
est
n
ei
g
h
b
o
r
s
(
K
NN)
m
o
d
els,
ac
h
iev
in
g
ac
c
u
r
ac
ies
o
f
9
9
.
6
7
%
o
n
th
e
Fi
s
h
ar
e
d
ataset
a
n
d
9
9
.
9
4
%
o
n
Kag
g
le.
Ku
m
ar
et
a
l
.
[
1
2
]
im
p
r
o
v
ed
b
r
ain
M
R
I
class
i
f
icat
io
n
b
y
in
te
g
r
atin
g
i
m
a
g
e
en
h
a
n
ce
m
e
n
t
w
it
h
E
W
T
.
B
y
ex
tr
ac
tin
g
e
n
er
g
y
an
d
e
n
tr
o
p
y
f
ea
tu
r
e
s
f
r
o
m
E
W
T
s
u
b
b
an
d
s
,
th
e
y
b
u
ilt
r
o
b
u
s
t
f
ea
t
u
r
e
v
ec
to
r
s
to
tr
ain
SVM
an
d
L
P
B
o
o
s
t
class
i
f
ier
s
.
E
v
al
u
ated
o
n
2
8
0
i
m
ag
e
s
,
b
o
th
SVM
(
u
s
i
n
g
L
itt
le
w
o
o
d
-
P
ale
y
-
b
a
s
e
d
E
W
T
)
an
d
L
P
B
o
o
s
t
(
u
s
in
g
ten
s
o
r
-
b
ased
E
W
T
)
ac
h
i
ev
ed
an
ac
cu
r
ac
y
o
f
9
6
.
4
3
%.
Gu
d
ig
ar
et
a
l
.
[
1
3
]
c
o
n
d
u
cted
a
r
ev
ie
w
o
n
Dec
o
m
p
o
s
itio
n
T
ec
h
n
iq
u
e
s
in
M
R
I
an
al
y
s
i
s
,
co
m
p
ar
in
g
B
E
MD
an
d
v
ar
iatio
n
al
m
o
d
e
d
ec
o
m
p
o
s
it
io
n
(
VM
D)
f
o
r
class
i
f
icatio
n
.
T
h
e
y
cla
s
s
i
f
i
ed
th
e
d
ata
u
s
i
n
g
b
is
p
ec
tr
al
f
ea
t
u
r
es,
s
u
p
er
v
is
ed
n
ei
g
h
b
o
r
h
o
o
d
p
r
o
j
ec
tio
n
e
m
b
ed
d
in
g
,
an
d
a
n
SVM
cla
s
s
i
f
ier
.
VM
D
ac
h
ie
v
ed
a
m
u
lti
-
clas
s
ac
c
u
r
ac
y
o
f
9
0
.
6
8
%,
o
u
tp
er
f
o
r
m
i
n
g
B
E
MD
’
s
8
4
.
4
7
%.
Ho
w
e
v
er
,
s
p
ec
i
f
ic
d
etails
o
n
f
ea
t
u
r
e
ex
tr
ac
tio
n
co
m
p
o
n
e
n
t
s
ar
e
lack
in
g
.
A
d
d
itio
n
all
y
,
Sah
u
et
a
l
.
[
1
4
]
,
Deb
et
a
l
.
[
1
5
]
,
an
d
Gh
az
i
et
a
l
.
[
1
6
]
u
s
ed
B
E
MD
to
class
i
f
y
M
R
I
,
eso
p
h
ag
ea
l
i
m
ag
e
s
,
a
n
d
m
a
m
m
o
g
r
ap
h
y
i
m
a
g
es,
r
e
s
p
ec
tiv
el
y
.
T
h
is
w
ill
b
e
d
is
c
u
s
s
ed
in
f
u
r
t
h
er
d
etail
in
co
m
p
ar
i
s
o
n
w
it
h
ex
is
ti
n
g
2
D
-
E
MD
ap
p
r
o
ac
h
es
.
B
y
ex
p
lo
r
in
g
B
E
MD
f
o
r
f
ea
tu
r
e
ex
tr
ac
tio
n
,
t
h
e
latest t
h
r
ee
s
t
u
d
ies o
f
f
er
cr
itical
p
er
s
p
ec
tiv
es o
n
m
o
d
er
n
s
ig
n
a
l a
n
d
i
m
a
g
e
p
r
o
ce
s
s
in
g
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6930
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
,
Vo
l.
24
,
No
.
5
,
Octo
b
e
r
20
26
:
1
7
2
8
-
1744
1730
T
h
e
p
ap
e
r
s
in
th
e
liter
atu
r
e
r
ev
ie
w
co
n
f
ir
m
t
h
e
s
tated
li
m
its
o
f
tr
ad
itio
n
al
an
d
h
y
b
r
id
m
et
h
o
d
s
.
T
h
is
s
tu
d
y
ai
m
s
to
ad
d
r
ess
th
e
s
e
li
m
i
tatio
n
s
.
O
u
r
k
e
y
co
n
tr
i
b
u
tio
n
s
in
c
lu
d
e:
(
i)
d
ev
elo
p
i
n
g
an
M
R
I
t
u
m
o
r
s
eg
m
e
n
tatio
n
f
r
a
m
e
w
o
r
k
t
h
at
u
s
e
s
E
W
T
an
d
E
MD
f
o
r
g
lo
b
al
th
r
es
h
o
ld
d
etec
tio
n
;
(
ii)
hi
gh
-
lev
el
f
ea
t
u
r
es
ar
e
ex
tr
ac
ted
b
y
VGG1
9
f
r
o
m
a
c
o
m
b
i
n
atio
n
o
f
t
h
e
R
OI
s
f
r
o
m
d
en
o
is
ed
i
m
a
g
es
a
n
d
th
e
B
I
M
Fs
v
ia
B
E
MD
;
an
d
(
iii)
d
esig
n
i
n
g
h
ig
h
-
p
er
f
o
r
m
i
n
g
,
m
u
lt
i
-
cla
s
s
clas
s
i
f
icatio
n
m
o
d
els
u
s
in
g
e
n
s
e
m
b
le
b
ag
g
i
n
g
an
d
b
o
o
s
tin
g
m
et
h
o
d
s
.
T
h
ese
co
n
tr
ib
u
tio
n
s
ai
m
to
d
eliv
er
a
r
o
b
u
s
t
C
A
D
x
s
y
s
te
m
t
h
at
m
i
n
i
m
ize
s
d
iag
n
o
s
tic
m
is
c
lass
if
ica
tio
n
,
estab
li
s
h
in
g
a
s
tr
o
n
g
f
o
u
n
d
atio
n
f
o
r
clin
ica
l d
ep
lo
y
m
en
t.
T
h
e
r
em
ai
n
d
er
o
f
th
is
p
ap
er
is
s
tr
u
c
tu
r
ed
as
f
o
llo
w
s
.
Secti
o
n
2
d
e
s
cr
ib
es
th
e
d
ata,
eq
u
i
p
m
e
n
t,
a
n
d
m
et
h
o
d
o
lo
g
y
.
Sectio
n
3
p
r
esen
ts
ex
p
er
i
m
e
n
tal
r
esu
l
ts
an
d
d
is
cu
s
s
io
n
.
Sectio
n
4
p
r
o
v
id
es
co
n
clu
s
io
n
s
an
d
f
u
tu
r
e
w
o
r
k
.
2.
M
E
T
H
O
D
2
.
1
.
Da
t
a
a
nd
equip
m
ent
T
h
e
Fig
s
h
ar
e
b
r
ain
t
u
m
o
r
d
ataset
co
n
s
is
t
s
o
f
3
,
0
6
4
T
1
-
w
ei
g
h
ted
co
n
tr
ast
-
e
n
h
a
n
ce
d
MRI
s
f
r
o
m
2
3
3
p
atien
ts
,
co
llected
at
Nan
f
a
n
g
Ho
s
p
ital
a
n
d
T
ian
j
in
Me
d
i
ca
l
Un
i
v
er
s
it
y
i
n
C
h
in
a
b
et
wee
n
2
0
0
5
an
d
2
0
1
0
.
E
ac
h
5
1
2
×
5
1
2
im
a
g
e
f
ea
t
u
r
es
an
i
n
-
p
la
n
e
r
eso
l
u
tio
n
o
f
0
.
4
9
×
0
.
4
9
m
m
2
.
T
h
e
co
llecti
o
n
in
c
lu
d
es
1
,
4
2
6
g
lio
m
a
s
,
9
3
0
p
itu
itar
y
t
u
m
o
r
s
,
an
d
7
0
6
m
en
in
g
io
m
as,
a
n
d
p
r
o
v
id
es t
u
m
o
r
lab
els,
p
atie
n
t I
Ds,
o
r
ig
i
n
al
i
m
a
g
es,
an
d
m
a
s
k
d
ata
[
1
7
]
,
[
1
8
]
.
Fo
r
o
u
r
ex
p
er
i
m
e
n
t
s
,
n
i
n
e
i
m
a
g
e
s
w
er
e
s
elec
ted
f
o
r
s
eg
m
e
n
tatio
n
,
an
d
1
5
0
i
m
ag
e
s
(
5
0
p
er
tu
m
o
r
t
y
p
e)
w
er
e
u
s
ed
f
o
r
clas
s
if
icatio
n
.
P
r
io
r
to
s
eg
m
en
tatio
n
a
n
d
f
ea
t
u
r
e
ex
tr
ac
ti
o
n
,
all
i
m
ag
e
s
w
er
e
en
h
a
n
ce
d
.
A
ll
ex
p
er
i
m
en
ts
w
er
e
co
n
d
u
cted
in
M
A
T
L
AB
o
n
a
lap
to
p
eq
u
ip
p
e
d
w
it
h
an
I
n
tel
C
o
r
e
i5
p
r
o
ce
s
s
o
r
an
d
8
GB
R
A
M,
u
s
i
n
g
T
h
ir
u
m
a
lais
a
m
y
’
s
f
ast
an
d
a
d
ap
tiv
e
B
E
MD
co
d
e
[
1
9
]
an
d
an
is
o
tr
o
p
ic
d
if
f
u
s
io
n
f
ilter
i
n
g
(
A
D
F)
co
d
e
[
2
0
]
.
2
.
2
.
M
et
ho
ds
2
.
2
.
1
.
B
a
ck
g
ro
un
d t
heo
ries
T
h
e
in
tr
o
d
u
ctio
n
p
r
o
v
id
ed
a
b
r
ief
o
v
er
v
ie
w
o
f
E
MD
an
d
B
E
MD
.
T
h
ese
alg
o
r
ith
m
s
co
n
v
er
t
a
s
ig
n
al
o
r
an
i
m
ag
e
in
to
I
MFs
o
r
B
I
MFs,
w
it
h
ea
c
h
I
M
F
a
n
d
B
I
MF
s
atis
f
y
in
g
cr
iter
ia:
(
i)
t
h
e
n
u
m
b
er
o
f
ex
tr
e
m
a
a
n
d
ze
r
o
c
r
o
s
s
in
g
s
m
u
s
t
eit
h
er
b
e
eq
u
al
o
r
d
if
f
er
b
y
at
m
o
s
t
o
n
e
;
an
d
(
ii)
t
h
e
m
ea
n
v
al
u
e
o
f
t
h
e
u
p
p
er
an
d
lo
w
er
en
v
elo
p
es
m
u
s
t
b
e
ze
r
o
.
T
h
e
E
MD
alg
o
r
it
h
m
o
p
er
ates
th
r
o
u
g
h
t
w
o
lo
o
p
s
:
th
e
in
n
e
r
lo
o
p
d
eter
m
in
e
s
in
d
iv
id
u
al
I
MFs
u
s
in
g
a
“
s
i
f
ti
n
g
p
r
o
ce
s
s
”,
w
h
ic
h
in
v
o
l
v
es
i
d
en
tify
i
n
g
lo
ca
l
m
a
x
i
m
a
an
d
m
i
n
i
m
a,
co
n
s
tr
u
cti
n
g
u
p
p
er
an
d
lo
w
er
e
n
v
elo
p
es,
an
d
iter
ativ
e
l
y
s
u
b
tr
ac
tin
g
t
h
e
m
ea
n
u
n
til
th
e
I
M
F
co
n
d
itio
n
s
ar
e
s
ati
s
f
ied
,
w
h
ile
th
e
o
u
ter
lo
o
p
ex
tr
ac
t
s
s
u
cc
es
s
iv
e
I
M
Fs
a
n
d
th
e
f
i
n
al
r
e
s
id
u
e.
T
h
e
B
E
MD
alg
o
r
ith
m
is
a
n
ex
te
n
s
i
o
n
o
f
E
M
D
to
t
w
o
d
im
e
n
s
io
n
s
.
Ho
w
e
v
er
,
w
h
ile
E
MD
u
s
e
s
1
-
D
en
v
elo
p
e
in
ter
p
o
latio
n
,
B
E
MD
u
s
es
s
u
r
f
ac
e
in
ter
p
o
latio
n
[
2
1
]
.
A
lg
o
r
it
h
m
1
s
h
o
w
s
t
h
e
E
MD
alg
o
r
ith
m
.
A
l
g
o
r
i
t
h
m
1
.
E
MD
alg
o
r
it
h
m
(
n
o
te:
cr
iter
ia
o
f
I
MF
(
i)
an
d
(
ii)
in
tex
t
)
Step 1:
Initialization
Set input S
(
)
Set
0
=
(
)
j = 1 (index of IMF)
Step 2: I
nner loop (Sifting process
-
Single IMF
)
0
=
−
1
(
)
m
=
1
(i) Find local extreme: local extreme in
−
1
(
)
(ii) Building envelopes: upper and lower envelopes:
-
Interpolating local extreme by cubic sp
l
ine,
(
)
(
)
(iii) Computing mean envelopes:
−
1
(
)
=
[
(
)
+
(
)
]
2
iv) Oscillation component:
(
)
=
−
1
(
)
−
−
1
(
)
(v) S
top condition of sifting process:
-
if
:
(
)
=
=
:
False
−
1
(
)
=
(
)
=
+
1
repeat Step 2(i)
-
if
:
(
)
=
=
:
True (obtain IMF)
goto Step 3
Step 3
: Outer loop (Multiple IMFs and Residue)
(i) Store IMF:
(
)
=
(
)
(ii) Update residue:
(
)
=
−
1
(
)
−
(
)
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
MRI
s
eg
men
ta
tio
n
a
n
d
cla
s
s
ifica
tio
n
u
s
in
g
1
-
D
a
n
d
2
-
D
emp
ir
ica
l m
o
d
e
…
(
Gia
n
g
Ho
n
g
N
g
u
ye
n
)
1731
(iii) Stop condition:
-
if
(
)
: multiple extreme
Set
=
+
1
go to Step 2
-
if
(
)
: monotonic trend:
=
(
)
Algorithm stop
Fin
all
y
,
t
h
e
in
p
u
t d
ata
is
m
o
d
e
led
b
y
(
1
)
an
d
(
2
)
:
−
Fo
r
E
MD
:
(
)
=
∑
(
)
=
1
+
(
)
(
1
)
w
h
er
e,
(
)
: th
e
i
n
p
u
t s
ig
n
al
,
(
)
: th
e
i
th
I
MF
an
d
(
)
: th
e
f
i
n
al
r
esid
u
e.
−
Fo
r
B
E
MD
:
(
,
)
=
∑
(
,
)
+
(
,
)
=
1
(
2
)
w
h
er
e,
(
,
)
: th
e
i
n
p
u
t i
m
a
g
e,
(
,
)
: th
e
i
th
B
I
MFs an
d
(
,
)
: th
e
f
i
n
al
r
esi
d
u
e.
I
MFs
an
d
B
I
MFs
r
ep
r
ese
n
t
o
s
cillato
r
y
co
m
p
o
n
e
n
t
s
r
an
g
i
n
g
f
r
o
m
h
ig
h
to
lo
w
f
r
eq
u
en
cie
s
,
co
r
r
esp
o
n
d
in
g
to
s
h
o
r
t
-
ter
m
(
f
in
e
te
x
tu
r
e)
a
n
d
lo
n
g
-
ter
m
(
lar
g
er
s
tr
u
c
tu
r
al
c
h
an
g
es)
v
ar
iati
o
n
s
,
w
h
ile
t
h
e
f
i
n
a
l
r
esid
u
e
(
)
r
ef
lects t
h
e
o
v
er
all
tr
en
d
o
f
th
e
s
ig
n
al
o
r
th
e
b
ac
k
g
r
o
u
n
d
o
f
t
h
e
i
m
a
g
e.
E
W
T
co
m
b
i
n
es
p
r
in
cip
le
s
o
f
w
a
v
elet
tr
a
n
s
f
o
r
m
a
n
d
E
M
D
to
id
en
ti
f
y
f
r
eq
u
e
n
c
y
s
u
p
p
o
r
ts
in
t
h
e
Fo
u
r
ier
s
p
ec
tr
u
m
a
n
d
co
n
s
tr
u
ct
ad
ap
tiv
e
w
av
e
let
f
ilter
s
.
U
n
li
k
e
E
MD
,
w
h
ich
r
elies
o
n
te
m
p
o
r
al
ex
tr
e
m
a,
E
W
T
an
al
y
ze
s
th
e
s
i
g
n
al
in
t
h
e
f
r
eq
u
e
n
c
y
d
o
m
a
in
.
I
t i
n
v
o
l
v
es f
o
u
r
s
tep
s
: (
i)
p
er
f
o
r
m
i
n
g
f
a
s
t f
o
u
r
ier
tr
an
s
f
o
r
m
(
FF
T
)
f
o
r
m
ag
n
it
u
d
e
s
p
ec
tr
u
m
a
n
al
y
s
i
s
,
(
ii)
s
e
g
m
en
t
in
g
t
h
e
s
p
ec
tr
u
m
b
y
d
etec
tin
g
s
i
g
n
i
f
ican
t
s
p
ec
tr
al
p
ea
k
s
,
(
iii)
co
n
s
tr
u
cti
n
g
b
an
d
-
p
as
s
f
i
lter
s
u
s
i
n
g
L
ittl
e
w
o
o
d
–
P
aley
f
u
n
ctio
n
s
,
an
d
(
iv
)
ap
p
l
y
i
n
g
t
h
e
f
il
ter
s
to
ex
tr
ac
t
e
m
p
ir
ical
m
o
d
es
[
2
2
]
.
(
)
=
(
0
,
)
∅
1
(
)
+
∑
(
,
)
(
)
=
1
(
3
)
w
h
er
e,
(
)
:
th
e
in
p
u
t,
:
in
n
er
p
r
o
d
u
ct,
∅
1
(
)
:
th
e
s
ca
li
n
g
f
u
n
c
tio
n
,
(
)
:
th
e
e
m
p
ir
ical
w
a
v
elet
f
u
n
cti
o
n
s
,
an
d
: th
e
co
r
r
esp
o
n
d
in
g
tr
an
s
f
o
r
m
co
ef
f
ic
ien
t
s
.
E
W
T
d
ec
o
m
p
o
s
es
a
s
i
g
n
al
in
to
t
w
o
co
m
p
o
n
en
t
s
:
a
n
ap
p
r
o
x
i
m
ate
co
m
p
o
n
e
n
t
(
c
A
)
a
n
d
a
d
etailed
co
m
p
o
n
e
n
t
(
cD)
.
P
r
o
ce
s
s
in
g
ca
n
b
e
p
er
f
o
r
m
ed
o
n
o
n
e
o
r
b
o
th
co
m
p
o
n
en
ts
d
ep
en
d
in
g
o
n
th
e
in
te
n
d
ed
r
esu
lt.
Th
e
d
iag
r
am
o
f
th
e
E
W
T
alg
o
r
ith
m
is
s
h
o
w
n
i
n
Fi
g
u
r
e
1
.
Fig
u
r
e
1
.
Diag
r
a
m
o
f
th
e
E
W
T
2
.
2
.
2
.
P
ro
po
s
ed
m
et
ho
d
T
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
u
s
es
E
MD
an
d
B
E
MD
,
b
ec
au
s
e
th
e
y
ar
e
d
ata
-
d
r
iv
en
m
e
th
o
d
s
,
t
h
eir
r
esu
lti
n
g
co
m
p
o
n
e
n
t
s
I
MFs
an
d
B
I
MF
s
ar
e
d
ata
-
ad
ap
tiv
e,
cr
ea
ted
d
ir
ec
tl
y
f
r
o
m
lo
ca
l
d
ata
f
ea
t
u
r
e
s
b
ased
o
n
v
ar
y
in
g
f
r
eq
u
en
c
ies;
th
e
s
e
co
m
p
o
n
e
n
t
s
r
ev
ea
l
h
id
d
en
tex
t
u
r
es
in
th
e
d
ata
at
d
if
f
er
en
t
s
ca
le
s
[
2
3
]
.
T
h
er
ef
o
r
e,
s
p
ec
if
ic
f
ea
t
u
r
es
ca
n
b
e
s
elec
ted
,
m
a
k
in
g
t
h
e
m
h
ig
h
l
y
u
s
e
f
u
l
f
o
r
an
al
y
z
in
g
co
m
p
le
x
i
m
a
g
es,
p
ar
ticu
lar
l
y
m
ed
ical
i
m
a
g
es.
Str
u
ctu
r
all
y
,
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
o
lo
g
y
i
s
ex
ec
u
ted
th
r
o
u
g
h
t
h
r
ee
m
ai
n
s
ta
g
es:
i
m
ag
e
e
n
h
an
ce
m
e
n
t,
s
eg
m
e
n
tatio
n
,
a
n
d
class
i
f
icat
i
o
n
.
I
m
ag
e
e
n
h
an
ce
m
e
n
t
is
p
er
f
o
r
m
ed
in
th
r
ee
s
tep
s
;
(
i)
s
k
u
l
l
s
t
r
ip
p
in
g
:
c
r
o
p
3
–
4
p
ix
els
f
r
o
m
th
e
i
m
a
g
e
b
o
r
d
er
s
to
r
em
o
v
e
t
h
e
li
g
h
t
b
o
x
,
th
en
ap
p
l
y
t
h
r
es
h
o
ld
in
g
a
n
d
m
o
r
p
h
o
lo
g
ical
o
p
er
ato
r
s
to
s
tr
ip
th
e
s
k
u
ll
[
2
4
]
;
(
ii)
n
o
is
e
r
ed
u
ctio
n
:
t
h
e
i
m
a
g
e
i
s
p
r
o
ce
s
s
ed
u
s
i
n
g
B
E
MD
an
d
an
an
is
o
tr
o
p
ic
d
if
f
u
s
io
n
f
ilte
r
;
an
d
(
iii)
c
o
n
tr
as
t
en
h
a
n
ce
m
en
t:
t
op
-
h
at
a
n
d
b
o
tto
m
-
h
at
m
o
r
p
h
o
lo
g
ica
l
tr
an
s
f
o
r
m
s
ar
e
u
s
ed
.
Fi
n
all
y
,
w
a
v
el
et
f
u
s
io
n
i
n
te
g
r
ates
th
e
r
esu
lts
o
f
s
tep
s
(
ii)
an
d
(
iii
)
[
2
5
]
.
A
n
i
m
a
g
e
h
i
s
t
o
g
r
a
m
r
e
p
r
e
s
e
n
t
s
t
h
e
f
r
e
q
u
e
n
c
y
d
i
s
t
r
i
b
u
t
i
o
n
o
f
p
i
x
e
l
i
n
t
e
n
s
i
t
i
e
s
.
T
h
e
P
M
F
i
s
d
e
f
i
n
e
d
a
s
[
8
]
:
(
)
=
ℎ
(
)
×
(
4
)
w
h
er
e,
ℎ
(
)
:
th
e
h
is
to
g
r
a
m
,
(
)
:
th
e
to
tal
n
u
m
b
er
o
f
p
ix
els
i
n
th
e
i
m
ag
e
;
f
o
r
an
8
-
b
it
g
r
a
y
s
ca
le
i
m
ag
e,
∈
[
0
,
255
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6930
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
,
Vo
l.
24
,
No
.
5
,
Octo
b
e
r
20
26
:
1
7
2
8
-
1744
1732
T
h
e
P
MF
r
ef
lects
b
ac
k
g
r
o
u
n
d
,
o
b
j
ec
ts
,
an
d
b
o
u
n
d
ar
ies
th
r
o
u
g
h
its
p
ea
k
s
a
n
d
v
alle
y
s
.
Ho
w
e
v
er
,
d
eter
m
in
i
n
g
th
e
t
h
r
es
h
o
ld
d
ir
ec
tl
y
f
r
o
m
(
)
is
d
if
f
ic
u
lt
d
u
e
t
o
n
o
is
e
an
d
h
i
g
h
-
f
r
eq
u
e
n
c
y
p
ix
els.
T
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
in
clu
d
es:
(
i)
c
o
m
p
u
t
in
g
th
e
P
MF
f
r
o
m
th
e
h
i
s
to
g
r
a
m
;
(
ii)
s
m
o
o
th
in
g
th
e
P
MF:
E
W
T
d
ec
o
m
p
o
s
es
th
e
P
MF
i
n
to
a
p
p
r
o
x
im
a
tio
n
(
c
A
)
a
n
d
d
e
tail
(
cD)
co
m
p
o
n
en
ts
.
D
is
ca
r
d
in
g
t
h
e
d
etail
-
h
ea
v
y
,
n
o
is
y
cD
co
m
p
o
n
en
t
w
h
ile
r
e
tain
i
n
g
th
e
lo
w
-
f
r
eq
u
e
n
c
y
c
A
co
m
p
o
n
en
t
y
ield
s
a
s
m
o
o
t
h
e
d
P
MF
f
o
r
th
e
n
e
x
t
s
tep
;
(
iii)
c
o
n
s
tr
u
cti
n
g
a
th
r
es
h
o
ld
cu
r
v
e:
d
ec
o
m
p
o
s
in
g
th
e
s
m
o
o
th
ap
p
r
o
x
i
m
at
io
n
c
A
v
ia
E
MD
y
ield
s
s
ev
er
al
I
MFs
an
d
a
r
esid
u
e.
Sin
ce
c
A
is
s
m
o
o
th
,
th
e
h
ig
h
es
t
-
f
r
eq
u
e
n
c
y
co
m
p
o
n
e
n
t
1
is
v
ir
tu
all
y
n
o
is
e
-
f
r
ee
an
d
clea
r
l
y
is
o
late
s
t
w
o
d
is
ti
n
ct
m
o
d
e
s
:
o
n
e
co
r
r
esp
o
n
d
in
g
to
b
r
ain
s
tr
u
ct
u
r
e
an
d
an
o
t
h
er
to
tu
m
o
r
p
r
esen
ce
.
C
o
n
s
eq
u
en
tl
y
,
1
is
s
elec
ted
as
th
e
t
h
r
es
h
o
ld
cu
r
v
e
ℎ
f
o
r
d
eter
m
in
in
g
t
h
e
g
lo
b
al
th
r
es
h
o
l
d
;
an
d
(
iv
)
t
u
m
o
r
d
etec
tio
n
:
t
h
e
t
h
r
es
h
o
ld
f
r
o
m
t
h
e
ℎ
cu
r
v
e,
d
e
f
i
n
ed
as
t
h
e
p
ix
el
v
a
lu
e
at
t
h
e
m
i
n
i
m
u
m
v
al
le
y
b
et
w
ee
n
t
h
e
t
w
o
p
ea
k
s
,
is
u
s
ed
to
b
in
ar
ize
t
h
e
d
en
o
is
ed
MRI
f
o
r
r
a
w
t
u
m
o
r
d
etec
ti
o
n
.
Su
b
s
eq
u
en
tl
y
,
m
o
r
p
h
o
lo
g
ical
o
p
er
ato
r
s
ar
e
a
p
p
lied
to
r
ef
in
e
th
e
s
e
g
m
en
tat
i
o
n
.
Fo
llo
w
i
n
g
s
e
g
m
en
tat
io
n
,
t
h
e
tu
m
o
r
ce
n
ter
is
id
en
ti
f
ied
,
an
d
a
R
OI
m
ea
s
u
r
in
g
2
2
4
×
2
2
4
p
ix
els
i
s
ex
tr
ac
ted
ar
o
u
n
d
t
h
e
t
u
m
o
r
i
n
t
h
e
en
h
a
n
ce
d
MRI,
t
h
en
B
E
MD
is
ap
p
lied
to
th
e
R
OI
t
o
g
en
er
ate
B
I
MFs
:
(
i)
f
ea
tu
r
e
ex
tr
ac
tio
n
:
t
h
e
R
OI
s
o
f
t
w
o
s
p
ec
if
ic
B
I
MFs
ar
e
s
elec
ted
an
d
in
teg
r
ated
w
it
h
th
e
en
h
an
ce
d
MRI
’
s
R
OI
to
cr
ea
te
a
r
ed
-
g
r
ee
n
-
b
lu
e
(
R
GB
)
i
m
ag
e
w
it
h
e
n
h
a
n
ce
d
f
ea
tu
r
e
in
p
u
ts
.
F
in
al
l
y
,
h
ig
h
-
le
v
el
f
ea
t
u
r
es
ar
e
ex
tr
ac
ted
f
r
o
m
th
i
s
in
te
g
r
ated
R
GB
i
m
ag
e
an
d
MRI
’
s
R
OI
u
s
in
g
th
e
f
c7
la
y
er
o
f
a
p
re
-
tr
ain
ed
VGG1
9
m
o
d
el,
g
en
er
ati
n
g
4
0
9
6
f
ea
tu
r
es
p
er
MRI
s
ca
n
;
a
n
d
(
ii)
c
lass
i
f
ic
atio
n
:
t
h
e
d
ataset
i
s
s
tan
d
ar
d
ized
an
d
s
p
lit
i
n
to
tr
ain
i
n
g
a
n
d
test
in
g
s
u
b
s
et
s
at
an
8
0
:2
0
r
atio
.
Fin
all
y
,
en
s
e
m
b
le
lear
n
i
n
g
tec
h
n
iq
u
es
ar
e
ap
p
lied
,
in
teg
r
atin
g
p
ar
allel
b
ase
m
o
d
el
tr
ain
in
g
v
i
a
b
ag
g
i
n
g
w
i
th
s
eq
u
e
n
tial e
r
r
o
r
co
r
r
ec
tio
n
th
r
o
u
g
h
b
o
o
s
ti
n
g
[
2
6
]
–
[
2
8
]
.
T
u
m
o
r
d
e
t
e
ct
i
o
n
an
d
c
la
s
s
if
i
ca
t
i
o
n
r
e
s
u
lt
s
a
r
e
ev
a
lu
at
e
d
ag
ai
n
s
t
t
h
e
d
a
t
as
e
t
’
s
g
r
o
u
n
d
t
r
u
t
h
,
u
s
in
g
t
h
e
f
o
l
l
o
w
in
g
s
ta
n
d
a
r
d
p
e
r
f
o
r
m
an
ce
m
et
r
i
cs
s
u
ch
as
a
c
cu
r
a
cy
,
p
r
ec
i
s
i
o
n
,
s
en
s
i
t
iv
ity
,
s
p
e
cif
i
c
ity
,
an
d
F
1
-
s
c
o
r
e
[
2
9
]
:
=
+
+
+
+
(
5
)
=
+
(
6
)
=
+
(
7
)
=
+
(
8
)
1
−
=
2
.
∗
+
(
9
)
w
h
er
e,
: tr
u
e
p
o
s
iti
v
e,
: tr
u
e
n
e
g
ati
v
e,
: f
al
s
e
p
o
s
itiv
e
a
n
d
: f
a
ls
e
n
e
g
ati
v
e.
T
h
e
f
lo
w
d
iag
r
a
m
o
f
t
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
is
illu
s
tr
ated
in
Fig
u
r
e
2
.
Fig
u
r
e
2
.
T
h
e
f
lo
w
d
iag
r
a
m
o
f
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
MRI
s
eg
men
ta
tio
n
a
n
d
cla
s
s
ifica
tio
n
u
s
in
g
1
-
D
a
n
d
2
-
D
emp
ir
ica
l m
o
d
e
…
(
Gia
n
g
Ho
n
g
N
g
u
ye
n
)
1733
3.
RE
SU
L
T
3
.
1
.
T
u
m
o
r
det
ec
t
io
n
T
h
is
s
ec
tio
n
p
r
esen
ts
tu
m
o
r
d
etec
tio
n
r
esu
lts
f
o
r
n
in
e
MRI
i
m
a
g
es
ac
r
o
s
s
th
r
ee
tu
m
o
r
t
y
p
e
s
,
b
eg
in
n
i
n
g
w
i
th
t
h
e
p
r
o
ce
s
s
i
n
g
s
tep
s
f
o
r
MRI
-
2315.
Step
s
f
o
r
tu
m
o
r
d
etec
tio
n
o
f
MRI
-
2315
:
−
I
m
ag
e
e
n
h
a
n
ce
m
e
n
t a
n
d
(
)
cu
r
v
e
Fig
u
r
e
3
(
a)
p
r
esen
ts
MRI
-
2
3
1
5
(
g
lio
m
a)
,
w
h
ic
h
w
as
in
itiall
y
s
k
u
ll
-
s
tr
ip
p
ed
[
2
4
]
.
I
ts
q
u
ality
w
a
s
th
e
n
en
h
a
n
ce
d
u
s
i
n
g
B
E
MD
an
d
A
D
F
f
o
r
n
o
is
e
r
ed
u
ctio
n
an
d
m
o
r
p
h
o
lo
g
ical
co
n
tr
ast
en
h
an
ce
m
e
n
t
[
2
5
]
,
as
illu
s
tr
ated
in
Fig
u
r
e
3
(
b
)
.
T
h
e
(
)
d
ep
icted
in
Fig
u
r
e
3
(
c)
d
i
s
p
la
y
s
i
n
d
is
ti
n
ct
v
alle
y
s
t
h
at
co
m
p
lica
te
th
r
es
h
o
ld
s
elec
tio
n
.
C
o
n
s
eq
u
e
n
tl
y
,
t
h
i
s
cu
r
v
e
is
u
s
ed
to
co
n
s
tr
u
ct
th
e
t
h
r
es
h
o
ld
in
g
cu
r
v
e.
−
S
m
o
o
th
i
n
g
(
)
u
s
in
g
E
W
T
T
h
e
E
W
T
d
ec
o
m
p
o
s
es
t
h
e
(
)
in
to
t
w
o
p
ar
ts
:
a
h
ig
h
-
f
r
eq
u
en
c
y
d
etail
co
m
p
o
n
e
n
t
(
cD
)
(
Fig
u
r
e
3
(
d
)
,
s
o
lid
lin
e)
co
n
tai
n
in
g
n
o
is
e
an
d
f
i
n
e
d
et
ails
,
w
h
ic
h
is
d
is
ca
r
d
ed
an
d
a
lo
w
-
f
r
eq
u
e
n
c
y
ap
p
r
o
x
im
a
tio
n
co
m
p
o
n
e
n
t
(
cA
)
.
T
h
is
p
r
eser
v
es
th
e
m
a
in
s
t
r
u
ctu
r
e
o
f
(
)
w
h
ile
r
ed
u
ci
n
g
n
o
is
e,
r
esu
lti
n
g
in
a
s
m
o
o
t
h
ed
(
)
illu
s
tr
ated
b
y
t
h
e
s
o
lid
lin
e
i
n
Fi
g
u
r
e
3
(
e)
,
w
h
er
ea
s
th
e
o
r
ig
in
al
(
)
ap
p
ea
r
s
as
a
d
ash
ed
lin
e.
−
B
u
ild
in
g
a
th
r
es
h
o
ld
cu
r
v
e
u
s
i
n
g
E
MD
Fig
u
r
e
3
(
f
)
s
h
o
w
s
t
h
e
d
ec
o
m
p
o
s
itio
n
o
f
th
e
s
m
o
o
t
h
ed
(
)
(
cA
)
in
to
I
MFs
a
n
d
a
r
esid
u
e
(
)
u
s
i
n
g
E
MD
.
I
MF
s
r
ep
r
esen
t
o
s
cillatio
n
s
at
d
i
f
f
er
en
t
f
r
eq
u
e
n
cies
f
r
o
m
h
ig
h
to
lo
w
,
a
n
d
t
h
e
in
d
icate
s
t
h
e
tr
en
d
.
Du
e
to
th
e
d
en
o
is
i
n
g
an
d
s
m
o
o
th
i
n
g
i
n
p
u
t,
t
h
e
n
1
,
th
e
h
ig
h
e
s
t
-
f
r
eq
u
e
n
c
y
co
m
p
o
n
en
t,
b
ec
o
m
e
s
n
o
is
e
-
f
r
ee
a
n
d
f
o
r
m
s
a
b
i
m
o
d
al
cu
r
v
e,
m
ak
i
n
g
it
s
u
itab
le
f
o
r
th
r
es
h
o
ld
d
etec
tio
n
.
ℎ
(
)
=
1
(
)
(
1
0
)
F
ig
u
r
e
3
(
g
)
il
lu
s
t
r
a
t
es
th
e
ℎ
(
)
c
u
r
v
e
,
w
h
ich
c
o
n
t
a
in
s
tw
o
p
e
ak
s
c
o
r
r
e
s
p
o
n
d
in
g
t
o
b
r
a
in
t
i
s
s
u
e
a
n
d
t
u
m
o
r
r
eg
i
o
n
s
.
T
h
e
th
r
esh
o
l
d
=
81
is
d
e
te
r
m
in
e
d
at
th
e
m
i
n
im
u
m
v
a
l
le
y
b
e
tw
ee
n
th
e
tw
o
p
e
ak
s
.
S
im
i
la
r
ly
,
F
ig
u
r
e
3
(
h
)
s
h
o
w
s
th
e
ℎ
(
)
c
u
r
v
e
f
r
o
m
th
e
K
r
in
i
d
is
te
ch
n
i
q
u
e
[
8
]
.
T
h
i
s
a
p
p
r
o
a
ch
s
u
m
s
m
u
l
ti
p
l
e
I
M
F
s
(
f
r
o
m
I
MF
2
t
o
I
MF
m
ax
-
1
)
v
i
a
E
M
D
t
o
b
u
il
d
a
th
r
es
h
o
l
d
i
n
g
cu
r
v
e
w
ith
1
=
80
.
A
l
th
o
u
g
h
t
h
e
t
h
r
esh
o
l
d
s
o
f
t
h
e
t
w
o
m
e
th
o
d
s
a
r
e
c
o
n
s
i
d
er
e
d
e
q
u
al
,
K
r
in
i
d
is
’
s
cu
r
v
e
ex
h
i
b
i
t
s
a
s
u
b
-
p
e
ak
,
an
d
t
h
e
I
MF
s
u
m
m
a
ti
o
n
l
im
its
it
s
a
d
a
p
t
a
b
i
l
ity
t
o
m
u
l
t
il
ev
e
l
th
r
es
h
o
l
d
in
g
[
9
]
.
C
o
n
v
e
r
s
ely
,
th
e
p
r
o
p
o
s
e
d
m
e
th
o
d
i
n
t
eg
r
at
e
s
E
W
T
a
n
d
I
M
F
u
s
in
g
o
n
ly
1
,
g
e
n
e
r
a
t
in
g
a
c
l
ea
r
,
b
im
o
d
a
l
c
u
r
v
e
w
ith
o
u
t
s
u
b
-
p
e
ak
s
.
On
c
e
d
et
r
e
n
d
e
d
,
i
t
p
r
o
d
u
c
es
a
m
u
l
t
im
o
d
a
l
c
u
r
v
e
th
at
f
a
c
i
li
t
at
e
s
f
i
n
d
i
n
g
m
u
l
ti
l
ev
el
th
r
e
s
h
o
l
d
s
f
o
r
b
r
a
in
t
is
s
u
e
an
d
tu
m
o
r
l
ay
e
r
an
a
ly
s
is
.
−
T
u
m
o
r
d
etec
tio
n
a
n
d
co
m
p
ar
is
o
n
T
h
i
s
t
h
r
esh
o
l
d
i
d
e
n
t
if
i
es
th
e
r
aw
tu
m
o
r
d
e
te
c
t
i
o
n
,
o
v
e
r
l
a
i
d
o
n
th
e
o
r
ig
in
a
l
MR
I
in
F
ig
u
r
e
3
(
i
)
(
1
)
.
F
ig
u
r
e
3
(
i
)
(2
-
5
)
p
r
es
en
ts
a
co
m
p
a
r
is
o
n
o
f
th
e
p
r
o
p
o
s
e
d
m
e
t
h
o
d
’
s
r
es
u
l
ts
ag
ai
n
s
t
th
r
e
e
r
e
f
e
r
en
c
e
t
ec
h
n
i
q
u
es
:
O
t
s
u
’
s
m
u
l
ti
l
ev
el
th
r
esh
o
l
d
[
3
0
]
,
K
a
p
u
r
’
s
m
et
h
o
d
w
i
th
o
p
t
im
a
l
h
a
r
m
o
n
i
c
s
e
a
r
ch
[
3
1
]
,
a
n
d
K
-
m
e
an
s
c
lu
s
t
e
r
in
g
[
3
2
]
.
C
o
l
o
r
c
o
d
i
n
g
i
n
d
i
c
a
t
e
s
:
d
e
t
e
c
t
e
d
t
u
m
o
r
s
(
w
h
it
e
)
,
d
if
f
e
r
en
c
es
(
g
r
ay
)
,
a
n
d
g
r
o
u
n
d
t
r
u
th
(
g
r
e
en
cu
r
v
e
)
.
Po
s
t
-
p
r
o
c
e
s
s
in
g
u
s
e
s
c
o
n
s
is
t
en
t
m
o
r
p
h
o
l
o
g
i
c
a
l
o
p
e
r
a
t
i
o
n
s
.
V
is
u
a
l
a
s
s
e
s
s
m
en
t
in
d
ic
a
t
es
th
at
th
e
O
t
s
u
m
eth
o
d
c
l
o
s
el
y
m
a
tc
h
e
s
t
h
e
g
r
o
u
n
d
t
r
u
th
,
f
o
ll
o
w
e
d
b
y
t
h
e
p
r
o
p
o
s
e
d
a
p
p
r
o
a
c
h
a
n
d
Ka
p
u
r
’
s
m
e
th
o
d
,
an
d
f
in
a
lly
K
-
m
e
an
s
.
Pe
r
f
o
r
m
an
ce
m
e
t
r
i
cs
a
r
e
s
u
m
m
a
r
i
z
e
d
in
T
a
b
l
e
1
.
3
.
1
.
1
.
T
u
m
o
r
det
ec
t
io
n f
o
r
t
he
re
m
a
i
nin
g
eig
ht
M
RIs
Fig
u
r
e
4
ill
u
s
tr
ates
t
h
e
t
u
m
o
r
d
etec
tio
n
r
esu
l
ts
f
o
r
ei
g
h
t
s
elec
ted
M
R
I
i
m
a
g
es,
co
m
p
ar
in
g
t
h
e
p
r
o
p
o
s
ed
tech
n
iq
u
e
a
g
a
i
n
s
t
t
h
r
ee
r
ef
er
en
ce
m
et
h
o
d
s
.
Sp
ec
if
icall
y
,
Fi
g
u
r
e
4
(
a)
s
h
o
w
s
t
h
e
s
k
u
ll
-
s
tr
ip
p
ed
an
d
en
h
a
n
ce
d
M
R
I
,
w
h
ile
Fi
g
u
r
e
4
(
b
)
p
r
esen
ts
th
e
(
)
cu
r
v
e
(
d
ash
-
li
n
e)
alo
n
g
s
id
e
its
s
m
o
o
t
h
ed
v
er
s
io
n
(
s
o
lid
lin
e)
.
Fig
u
r
e
4
(
c)
d
eta
ils
th
e
b
i
m
o
d
al
c
u
r
v
e
ℎ
(
)
d
er
iv
ed
f
r
o
m
1
(
)
u
s
i
n
g
E
MD
w
h
ic
h
estab
lis
h
es
t
h
e
th
r
es
h
o
ld
.
Fig
u
r
e
4
(
d
)
(
1
)
p
r
esen
ts
th
e
r
a
w
d
etec
ted
tu
m
o
r
h
i
g
h
li
g
h
ted
in
r
ed
o
n
th
e
o
r
ig
in
a
l
MRI.
Fig
u
r
es
4
(
d
)
(2
–
5
)
d
e
p
ict
th
e
r
ef
i
n
ed
tu
m
o
r
s
g
en
e
r
ated
b
y
t
h
e
p
r
o
p
o
s
ed
alg
o
r
i
th
m
an
d
t
h
e
th
r
e
e
co
m
p
ar
is
o
n
m
eth
o
d
s
.
Vi
s
u
a
l
ev
alu
a
tio
n
i
n
d
icate
s
t
h
at
t
h
e
Ots
u
m
et
h
o
d
alig
n
s
m
o
s
t
c
lo
s
el
y
w
it
h
t
h
e
g
r
o
u
n
d
tr
u
th
,
f
o
llo
w
ed
b
y
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
,
Kap
u
r
’
s
m
et
h
o
d
,
an
d
K
-
m
ea
n
s
,
r
esp
ec
tiv
el
y
.
W
ith
in
t
h
e
g
lio
m
a
g
r
o
u
p
,
i
m
a
g
es
2
4
0
7
an
d
2
7
3
0
p
ar
ticu
lar
ly
i
m
a
g
e
2
4
0
7
d
is
p
la
y
ed
th
e
lar
g
e
s
t
d
is
cr
ep
an
cies
b
et
w
ee
n
th
e
d
etec
ted
tu
m
o
r
s
a
n
d
th
e
g
r
o
u
n
d
tr
u
t
h
co
m
p
ar
ed
to
o
th
er
g
r
o
u
p
s
.
Deta
i
led
p
er
f
o
r
m
an
ce
m
etr
ic
s
ar
e
s
u
m
m
ar
ized
in
T
ab
le
1
.
3
.
1
.
2
.
P
er
f
o
rm
a
nce
m
et
ric
s
I
n
MRI
s
ca
n
s
,
b
r
ain
t
u
m
o
r
s
a
r
e
ty
p
ical
l
y
m
u
ch
s
m
al
ler
th
a
n
th
e
s
u
r
r
o
u
n
d
in
g
ti
s
s
u
es,
ca
u
s
i
n
g
tr
u
e
n
eg
at
iv
e
(
T
N)
v
alu
es
to
d
o
m
i
n
ate
a
n
d
i
n
f
la
te
b
o
th
ac
c
u
r
ac
y
a
n
d
s
p
ec
i
f
icit
y
to
n
ea
r
1
.
T
o
av
o
id
th
es
e
m
is
lead
in
g
r
es
u
lts
,
d
etec
tio
n
p
er
f
o
r
m
a
n
ce
i
s
i
n
s
tead
ev
a
lu
a
ted
u
s
i
n
g
p
r
ec
is
io
n
,
s
e
n
s
i
tiv
it
y
,
an
d
t
h
e
F1
-
s
co
r
e
(
T
a
b
le
1
)
.
T
h
e
av
er
ag
e
p
r
ec
is
i
o
n
o
f
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
r
ea
ch
ed
9
6
.
7
0
%,
w
h
ich
i
s
s
li
g
h
tl
y
lo
w
er
t
h
an
t
h
e
K
-
m
ea
n
s
m
et
h
o
d
(
9
7
.
0
7
%).
I
n
ter
m
s
o
f
s
e
n
s
iti
v
it
y
,
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
ac
h
iev
ed
9
1
.
0
7
%
s
u
r
p
ass
i
n
g
K
-
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6930
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
,
Vo
l.
24
,
No
.
5
,
Octo
b
e
r
20
26
:
1
7
2
8
-
1744
1734
m
ea
n
s
(
8
9
.
2
7
%)
b
u
t tr
ailin
g
th
e
Kap
u
r
an
d
Ots
u
m
et
h
o
d
s
.
B
ec
au
s
e
t
h
e
d
i
f
f
er
e
n
ce
s
i
n
p
r
ec
i
s
io
n
a
n
d
s
en
s
iti
v
it
y
b
et
w
ee
n
th
e
s
e
m
et
h
o
d
s
ar
e
m
in
o
r
,
b
o
th
Ots
u
’
s
m
et
h
o
d
an
d
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
b
o
th
estab
lis
h
an
o
p
ti
m
a
l
tr
ad
e
-
o
f
f
b
et
w
ee
n
th
e
t
w
o
m
e
tr
ics
(
r
an
k
in
g
s
ec
o
n
d
an
d
t
h
ir
d
)
.
C
o
n
v
er
s
el
y
,
t
h
e
Kap
u
r
a
n
d
K
-
m
ea
n
s
m
o
d
el
s
lack
th
i
s
b
alan
ce
(
r
an
k
in
g
f
ir
s
t
an
d
f
o
u
r
t
h
,
tr
ad
e
-
o
f
f
)
.
As
a
r
esu
lt,
th
e
p
r
o
p
o
s
ed
ap
p
r
o
a
ch
s
h
ar
e
s
th
e
to
p
-
r
an
k
ed
F1
-
s
co
r
e
w
it
h
Ot
s
u
(
9
3
.
7
0
% v
s
.
9
3
.
7
1
%).
T
h
is
d
e
m
o
n
s
tr
ate
s
t
h
at
b
o
th
m
et
h
o
d
s
e
f
f
ec
tiv
el
y
b
alan
ce
th
e
s
eg
m
e
n
tatio
n
o
f
d
etec
ted
tu
m
o
r
s
an
d
g
r
o
u
n
d
tr
u
th
s
,
an
d
m
i
n
i
m
ize
f
alse d
etec
tio
n
s
[
3
3
]
.
T
h
e
th
r
esh
o
ld
v
alu
es
i
n
co
lu
m
n
4
o
f
T
ab
le
1
in
d
icate
th
at
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
is
m
o
r
e
co
n
s
is
ten
t
w
it
h
Ots
u
’
s
m
u
lt
ilev
el
m
eth
o
d
th
an
w
it
h
th
e
Kap
u
r
-
Har
m
o
n
ic
m
et
h
o
d
.
A
lth
o
u
g
h
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
is
g
lo
b
a
l,
it
p
r
o
v
id
es
th
r
esh
o
ld
s
co
m
p
ar
ab
le
to
th
e
s
tan
d
ar
d
Ots
u
m
u
ltil
e
v
el
tec
h
n
iq
u
e.
T
h
is
s
u
cc
ess
f
u
ll
y
b
y
p
as
s
es
t
h
e
tr
ial
-
an
d
-
er
r
o
r
p
r
o
ce
s
s
ty
p
ical
l
y
r
eq
u
ir
ed
to
d
eter
m
i
n
e
th
e
o
p
ti
m
al
th
r
e
s
h
o
ld
o
f
m
u
ltil
e
v
el
th
r
es
h
o
ld
in
g
,
h
i
g
h
lig
h
ti
n
g
th
e
p
r
ac
tical
ef
f
ec
ti
v
e
n
es
s
o
f
t
h
e
p
r
o
p
o
s
ed
a
p
p
r
o
ac
h
.
C
o
m
p
ar
is
o
n
w
it
h
e
x
is
ti
n
g
E
MD
ap
p
r
o
ac
h
es:
t
h
e
r
esu
lts
wer
e
co
m
p
ar
ed
w
it
h
t
h
o
s
e
o
f
t
h
e
m
o
d
i
f
ied
Kr
in
d
in
i
s
th
r
es
h
o
ld
in
g
m
e
th
o
d
d
etailed
in
[
3
4
]
.
I
n
th
at
s
tu
d
y
,
th
e
au
t
h
o
r
s
s
u
m
m
ed
f
r
o
m
2
to
−
2
in
s
tead
o
f
2
to
5
as
Kr
in
d
in
i
s
et
a
l
.
d
id
[
8
]
,
ev
alu
atin
g
th
eir
ap
p
r
o
ac
h
o
n
n
in
e
MRI
i
m
a
g
es
f
r
o
m
t
h
e
Fig
s
h
ar
e
d
atase
t.
W
h
ile
th
e
y
r
ep
o
r
ted
an
av
er
ag
e
p
r
ec
is
io
n
o
f
8
0
.
1
2
%,
s
en
s
it
iv
i
t
y
o
f
9
8
.
4
%,
an
d
F1
-
s
co
r
e
o
f
8
8
.
8
5
%,
o
u
r
ap
p
r
o
ac
h
ac
h
iev
ed
9
6
.
7
0
%,
9
1
.
0
7
%,
an
d
9
3
.
7
0
%,
r
esp
ec
tiv
el
y
,
d
e
m
o
n
s
tr
atin
g
o
u
r
m
eth
o
d
’
s
o
v
er
all
s
u
p
er
io
r
it
y
.
C
o
m
p
u
tatio
n
al
ti
m
e:
t
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
ac
h
ie
v
ed
a
r
u
n
t
i
m
e
o
f
ap
p
r
o
x
i
m
atel
y
1
.
7
1
s
ec
o
n
d
s
,
r
an
k
i
n
g
th
ir
d
a
m
o
n
g
th
e
f
o
u
r
m
et
h
o
d
s
.
I
t
w
as
f
a
s
ter
th
a
n
th
e
Kap
u
r
-
Har
m
o
n
ic
s
ea
r
ch
(
~3
.
9
1
s
)
,
b
u
t
s
lo
w
er
th
an
t
h
e
Ot
s
u
(
~0
.
2
5
s
)
an
d
K
-
m
ea
n
s
(
~0
.
2
4
s
)
m
et
h
o
d
s
,
as d
etailed
in
th
e
f
i
n
al
co
lu
m
n
o
f
T
ab
le
1
.
(
a)
(
b
)
(
c)
(
d
)
(
e)
(
f
)
(
g
)
(
h
)
(
i)
Fig
u
r
e
3
.
T
u
m
o
r
d
etec
tio
n
s
te
p
s
f
o
r
MRI
-
2
3
1
5
:
(
a)
o
r
ig
in
al
i
m
a
g
e,
(
b
)
d
en
o
is
ed
im
a
g
e,
(
c)
(
)
cu
r
v
e,
(
d
)
E
W
T
d
etail
co
m
p
o
n
en
t c
D
,
(
e)
E
W
T
ap
p
r
o
x
i
m
atio
n
co
e
f
f
icie
n
t c
A
u
s
ed
as s
m
o
o
t
h
ed
(
)
,
(
f
)
I
MFs
an
d
r
esid
u
e
u
s
i
n
g
E
MD
,
(
g
)
ℎ
(
)
cu
r
v
e
o
f
t
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
,
(
h
)
ℎ
(
)
cu
r
v
e
f
r
o
m
Kr
i
n
id
i
s
(
f
o
r
r
ef
er
en
ce
)
[
8
]
,
an
d
(
i)
tu
m
o
r
d
etec
tio
n
r
esu
lts
o
f
t
h
e
p
r
o
p
o
s
ed
m
e
th
od
.
P
an
els s
h
o
w
:
(
1
)
r
a
w
tu
m
o
r
d
etec
tio
n
o
v
er
laid
o
n
th
e
o
r
ig
i
n
al
i
m
ag
e,
a
n
d
(2
-
5)
r
ef
in
ed
tu
m
o
r
d
etec
tio
n
u
s
in
g
:
(
2
)
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
,
(
3
)
Ots
u
’
s
m
u
lti
lev
el
t
h
r
es
h
o
l
d
in
g
,
(
4
)
Kap
u
r
’
s
m
u
ltil
e
v
el
t
h
r
esh
o
ld
in
g
,
an
d
(
5
)
K
-
m
ea
n
s
clu
s
ter
i
n
g
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
MRI
s
eg
men
ta
tio
n
a
n
d
cla
s
s
ifica
tio
n
u
s
in
g
1
-
D
a
n
d
2
-
D
emp
ir
ica
l m
o
d
e
…
(
Gia
n
g
Ho
n
g
N
g
u
ye
n
)
1735
T
u
mo
r
M
R
I
e
n
h
a
n
c
e
me
n
t
(
)
c
u
r
v
e
s
ℎ
(
)
c
u
r
v
e
F
o
u
r
t
u
mo
r
s i
d
e
n
t
i
f
i
e
d
3
10
2
2
5
2
4
0
7
2
7
3
0
1
0
2
0
1
0
2
3
1
5
1
7
(
a)
(
b
)
(
c)
(
d
)
Fig
u
r
e
4
.
R
esu
lts
o
f
t
h
e
t
u
m
o
r
d
etec
tio
n
p
r
o
ce
s
s
w
er
e
o
b
tain
ed
f
o
r
th
e
r
e
m
ai
n
in
g
ei
g
h
t M
R
I
i
m
ag
e
s
:
(
a)
MRI
en
h
a
n
ce
m
en
t
w
it
h
s
k
u
l
l strip
p
in
g
,
(
b
)
(
)
cu
r
v
e
s
an
d
s
m
o
o
t
h
ed
v
er
s
io
n
s
u
s
i
n
g
E
W
T
,
(
c)
ℎ
(
)
cu
r
v
e
d
er
iv
ed
f
r
o
m
1
(
)
u
s
in
g
E
M
D,
an
d
(
d
)
t
u
m
o
r
d
ete
ctio
n
r
esu
lts
.
P
an
el
s
s
h
o
w
:
(
1
)
r
aw
t
u
m
o
r
d
etec
tio
n
o
v
er
laid
o
n
th
e
o
r
ig
i
n
al
i
m
ag
e,
a
n
d
(2
-
5)
r
ef
in
ed
tu
m
o
r
d
etec
tio
n
u
s
in
g
:
(
2
)
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
,
(
3
)
Ots
u
’
s
m
u
lti
lev
el
t
h
r
es
h
o
l
d
in
g
,
(
4
)
Kap
u
r
’
s
m
u
ltil
e
v
el
t
h
r
esh
o
ld
in
g
,
a
n
d
(
5
)
K
-
m
ea
n
s
clu
s
ter
i
n
g
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6930
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
,
Vo
l.
24
,
No
.
5
,
Octo
b
e
r
20
26
:
1
7
2
8
-
1744
1736
T
ab
l
e
1
.
T
h
e
p
er
f
o
r
m
an
ce
m
et
r
ics f
o
r
th
e
f
o
u
r
s
e
g
m
e
n
tat
io
n
alg
o
r
ith
m
s
o
n
n
i
n
e
MRI
s
No
M
R
I
M
e
t
h
o
d
T
h
r
e
sh
o
l
d
P
r
e
.
(
%)
S
e
n
.
(
%)
F1
-
S
c
o
re
(
%)
T
i
me
(
s)
G
l
i
o
mas
2
3
1
5
P
r
o
p
o
se
d
A
.
81
9
9
.
9
6
8
7
.
5
1
9
3
.
3
2
1
.
7
3
M
u
l
t
i
-
O
t
s
u
2
3
5
7
89
1
2
9
9
9
.
9
2
8
8
.
2
8
9
3
.
7
4
0
.
2
7
K
a
p
u
r
-
Ha
r.
81
194
9
9
.
9
6
8
7
.
5
1
9
3
.
3
2
2
.
8
0
K
-
m
e
a
n
s
3
C
l
u
st
e
r
1
0
0
8
4
.
4
0
9
1
.
5
4
0
.
2
9
2
4
0
7
P
r
o
p
o
se
d
A
.
45
9
7
.
7
2
7
9
.
9
5
8
7
.
9
5
1
.
4
1
M
u
l
t
i
-
O
t
s
u
1
2
3
0
45
68
9
7
.
7
2
7
9
.
9
5
8
7
.
9
5
0
.
2
4
K
a
p
u
r
-
H
a
r
.
46
1
1
3
1
8
4
9
7
.
8
1
7
9
.
3
3
8
7
.
6
0
3
.
9
7
K
-
m
e
a
n
s
3
c
l
u
st
e
r
s
9
7
.
9
9
7
8
.
7
5
8
7
.
3
2
0
.
2
6
2
7
3
0
P
r
o
p
o
se
d
A
.
41
9
2
.
7
7
8
7
.
6
9
9
0
.
1
6
1
.
4
6
M
u
l
t
i
-
O
t
s
u
1
0
2
0
2
9
3
4
42
9
3
.
4
0
8
6
.
9
7
9
0
.
0
7
0
.
1
4
K
a
p
u
r
-
H
a
r
.
40
5
8
9
1
1
3
2
1
7
3
2
1
4
9
2
.
4
1
8
8
.
8
2
9
0
.
5
8
3
.
6
2
K
-
m
e
a
n
s
4
c
l
u
st
e
r
s
9
2
.
7
7
8
7
.
6
9
9
0
.
1
6
0
.
4
9
M
e
n
i
n
g
i
o
mas
3
P
r
o
p
o
se
d
A
.
88
9
5
.
6
8
9
2
.
8
2
9
4
.
2
3
1.
61
M
u
l
t
i
-
O
t
s
u
2
6
6
2
90
9
6
.
3
1
8
9
.
8
6
9
2
.
9
8
0
.
2
5
K
a
p
u
r
-
H
a
r
.
85
153
9
5
.
1
6
9
4
.
0
9
9
4
.
6
2
4
.
2
6
K
-
m
e
a
n
s
3
c
l
u
st
e
r
s
9
5
.
3
2
9
3
.
5
0
9
4
.
4
0
0
.
2
5
10
P
r
o
p
o
se
d
A
.
38
9
5
.
3
7
9
6
.
1
7
9
5
.
7
7
1
.
6
6
M
u
l
t
i
-
O
t
s
u
14
3
7
60
9
4
.
0
4
9
6
.
9
8
9
5
.
4
8
0
.
2
6
K
a
p
u
r
-
H
a
r
.
35
9
0
1
4
6
2
0
1
8
9
.
9
4
9
8
.
1
7
9
3
.
8
7
3
.
4
8
K
-
m
e
a
n
s
3
c
l
u
st
e
r
s
9
9
.
9
3
8
1
.
7
9
8
9
.
9
6
0
.
2
0
2
2
5
P
r
o
p
o
se
d
A
.
41
9
4
.
5
1
9
5
.
8
6
9
5
.
1
8
1
.
7
0
M
u
l
t
i
-
O
t
s
u
1
6
43
67
9
5
.
1
7
9
5
.
1
2
9
5
.
1
4
0
.
3
3
K
a
p
u
r
-
H
a
r
.
40
8
8
1
7
2
9
4
.
2
4
9
6
.
2
4
9
5
.
2
3
3
.
2
8
K
-
m
e
a
n
s
3
c
l
u
st
e
r
s
9
6
.
8
4
9
1
.
6
9
9
4
.
2
0
0
.
2
1
P
i
t
u
i
t
a
r
y
1
0
2
0
P
r
o
p
o
se
d
A
.
41
9
6
.
5
1
9
5
.
3
8
9
5
.
9
4
2
.
2
0
M
u
l
t
i
-
O
t
s
u
1
6
41
59
9
6
.
5
1
9
5
.
3
8
9
5
.
9
4
0
.
1
7
K
a
p
u
r
-
H
a
r
.
4
0
8
2
1
6
9
9
5
.
1
9
9
6
.
9
4
9
6
.
0
6
3
.
3
6
K
-
m
e
a
n
s
3
c
l
u
st
e
r
s
9
5
.
7
4
9
5
.
3
9
9
5
.
5
6
0
.
6
1
1
0
2
3
P
r
o
p
o
se
d
A
.
34
9
8
.
5
0
9
1
.
8
3
9
5
.
0
5
1
.
5
2
M
u
l
t
i
-
O
t
s
u
1
2
2
2
3
2
4
7
60
9
7
.
4
5
9
3
.
5
4
9
5
.
4
5
0
.
2
3
K
a
p
u
r
-
H
a
r
.
3
0
7
0
1
1
6
1
6
3
2
0
9
9
4
.
7
0
9
5
.
2
7
9
4
.
9
8
3
.
1
5
K
-
m
e
a
n
s
3
c
l
u
st
e
r
9
7
.
7
8
9
3
.
0
9
9
5
.
3
8
0
.
3
6
1
5
1
7
P
r
o
p
o
se
d
A
.
46
9
9
.
2
5
9
2
.
3
8
9
5
.
6
9
2
.
1
2
M
u
l
t
i
-
O
t
s
u
1
3
3
0
44
89
9
8
.
7
2
9
4
.
7
2
9
6
.
6
8
0
.
3
4
K
a
p
u
r
-
H
a
r
.
44
9
3
1
3
4
1
7
4
2
1
5
9
8
.
7
2
9
4
.
7
2
9
6
.
6
8
7
.
3
0
K
-
m
e
a
n
s
4
c
l
u
st
e
r
s
9
7
.
2
8
9
6
.
8
0
9
7
.
0
4
0
.
3
6
A
v
e
r
a
g
e
KPIs
P
r
o
p
o
se
d
A
.
9
6
.
7
0
(
i
i
)
9
1
.
0
7
(
i
i
i
)
9
3
.
7
0
(
i
)
1
.
7
1
2
2
(
i
i
i
)
M
u
l
t
i
-
O
t
s
u
9
6
.
5
8
(
i
i
i
)
9
1
.
2
0
(
i
i
)
9
3
.
7
1
(
i
)
0
.
2
4
7
8
(
i
)
K
a
p
u
r
-
H
a
r
.
9
5
.
3
5
(
i
v
)
9
2
.
3
4
(
i
)
9
3
.
6
6
(
i
i
i
)
3
.
9
1
3
3
(
i
v
)
K
-
m
e
a
n
s
9
7
.
0
7
(
i
)
8
9
.
2
3
(
i
v
)
9
2
.
8
4
(
i
v
)
0
.
3
3
6
7
(
i
i
)
N
o
t
e
s:
R
o
man
n
u
me
r
a
l
s
d
e
n
o
t
e
t
h
e
o
r
d
e
r
o
f
t
h
e
a
p
p
r
o
a
c
h
e
s a
c
c
o
r
d
i
n
g
t
o
e
a
c
h
p
e
r
f
o
r
man
c
e
me
t
r
i
c
:
(
i
)
f
i
r
st
,
(
i
i
)
se
c
o
n
d
,
(
i
i
i
)
t
h
i
r
d
,
a
n
d
(
i
v
)
f
o
u
r
t
h
B
ased
o
n
th
e
p
r
esen
tatio
n
s
,
c
o
n
s
id
er
in
g
b
o
th
ac
cu
r
ac
y
a
n
d
co
m
p
u
tatio
n
t
i
m
e,
t
h
e
Ots
u
’
s
m
eth
o
d
is
th
e
m
o
s
t
e
f
f
icie
n
t.
Ho
w
ev
er
,
w
h
e
n
s
i
m
p
licit
y
i
s
also
f
ac
to
r
ed
in
,
o
u
r
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
is
th
e
b
es
t.
I
t
ac
h
iev
e
s
co
m
p
ar
ab
le
ac
cu
r
ac
y
w
i
th
o
u
t
r
el
y
i
n
g
o
n
th
e
p
r
ed
ef
in
ed
cr
iter
ia
th
at
th
e
t
h
r
ee
r
ef
er
en
ce
m
et
h
o
d
s
m
u
s
t
u
s
e.
3
.
2
.
Cla
s
s
if
ica
t
io
n
Data
p
r
ep
ar
atio
n
:
b
ec
au
s
e
s
u
b
tle
f
ea
t
u
r
es
ar
e
o
f
ten
h
id
d
en
d
ee
p
w
ith
in
en
h
a
n
ce
d
MRI
s
ca
n
s
,
ex
tr
ac
ti
n
g
th
e
m
d
ir
ec
tl
y
f
r
o
m
th
e
R
OI
o
f
ten
f
al
ls
s
h
o
r
t.
T
h
e
r
ef
o
r
e,
B
E
MD
is
ap
p
lied
to
th
e
R
OI
to
g
en
er
ate
B
I
MFs;
ea
ch
B
I
MF
co
n
tai
n
s
in
s
tan
ta
n
eo
u
s
a
m
p
lit
u
d
e
an
d
f
r
eq
u
e
n
c
y
in
f
o
r
m
a
tio
n
,
al
lo
w
i
n
g
t
h
e
m
o
d
el
to
ad
ap
t
to
lo
ca
l
im
a
g
e
v
ar
iatio
n
s
ac
r
o
s
s
d
if
f
er
en
t
s
ca
les.
I
n
o
u
r
class
i
f
icatio
n
,
2
an
d
3
w
er
e
s
el
ec
ted
b
ec
au
s
e
th
e
y
r
etai
n
th
e
te
x
t
u
r
e
d
etails,
ed
g
es,
an
d
s
tr
u
ctu
r
es
d
o
m
i
n
ati
n
g
t
h
e
m
id
-
f
r
eq
u
en
c
y
r
an
g
e,
w
h
i
le
ex
clu
d
i
n
g
b
o
th
th
e
n
o
i
s
e
(
1
)
an
d
th
e
tr
en
d
(
th
e
last
B
I
MFs).
Fig
u
r
e
s
5
(
a)
to
(
f
)
illu
s
tr
ates
t
h
e
en
h
an
ce
d
MRI,
R
OI
i
m
a
g
es,
a
n
d
th
eir
co
r
r
esp
o
n
d
in
g
B
I
MF
co
m
p
o
n
en
ts
a
n
d
i
n
teg
r
ated
R
OI
.
Sp
ec
if
icall
y
,
t
h
e
i
m
a
g
es
p
r
esen
ted
in
Fi
g
u
r
es
5
(
b
)
an
d
5
(
f
)
w
er
e
u
s
ed
f
o
r
f
ea
t
u
r
e
ex
tr
ac
tio
n
.
C
las
s
i
f
icatio
n
w
a
s
p
er
f
o
r
m
ed
u
s
i
n
g
1
5
0
R
OI
i
m
ag
e
s
f
r
o
m
th
e
b
r
ain
MRI
Fig
s
h
ar
e
d
ataset.
T
h
ese
i
m
a
g
es
w
er
e
p
r
o
ce
s
s
ed
u
s
in
g
B
E
MD
to
ex
tr
ac
t
h
ig
h
-
to
-
m
id
f
r
eq
u
en
c
y
co
m
p
o
n
en
t
s
(
2
an
d
3
)
.
Featu
r
e
ex
tr
ac
tio
n
in
p
u
ts
i
n
cl
u
d
ed
th
e
R
GB
-
R
OI
i
m
ag
e
s
an
d
co
m
b
in
atio
n
s
o
f
t
h
e
R
G
B
-
R
OI
w
it
h
R
GB
-
B
I
MFs.
T
h
ese
in
p
u
t
s
w
er
e
an
al
y
ze
d
u
s
i
n
g
a
p
r
e
-
tr
ain
ed
VGG1
9
m
o
d
el
to
ex
tr
ac
t
h
ig
h
-
l
ev
el
f
ea
t
u
r
es
f
r
o
m
a
f
u
ll
y
co
n
n
ec
ted
f
c
la
y
er
,
r
es
u
lt
in
g
i
n
a
f
ea
tu
r
e
v
ec
to
r
o
f
s
ize
150
×
4
0
9
6
.
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
MRI
s
eg
men
ta
tio
n
a
n
d
cla
s
s
ifica
tio
n
u
s
in
g
1
-
D
a
n
d
2
-
D
emp
ir
ica
l m
o
d
e
…
(
Gia
n
g
Ho
n
g
N
g
u
ye
n
)
1737
(
a)
(
b
)
(
c)
(
d
)
(
e)
(
f
)
Fig
u
r
e
5
.
I
n
p
u
t M
R
I
f
o
r
VGG1
9
f
ea
tu
r
e
ex
tr
ac
tio
n
: (
a)
en
h
a
n
ce
d
MRI
(
5
1
2
×
5
1
2
)
,
(
b
)
tu
m
o
r
R
OI
(
2
2
4
×
2
2
4
)
,
(
c)
1
o
f
R
OI
w
it
h
n
o
is
e,
(
d
)
2
o
f
R
OI
(
1
)
,
(
e)
3
o
f
R
OI
(
2
)
,
an
d
(
f
)
R
GB
in
te
g
r
ated
R
OI
,
1
,
an
d
2
.
(
n
o
te:
Fig
u
r
e
5
(
b
)
an
d
Fig
u
r
e
5
(
f
)
s
er
v
e
as i
n
p
u
ts
f
o
r
f
ea
tu
r
e
e
x
tr
ac
tio
n
)
Featu
r
e
ex
tr
ac
tio
n
:
a
to
tal
o
f
1
5
0
en
h
an
ce
d
i
m
a
g
es
f
r
o
m
th
e
Fig
s
h
ar
e
B
r
ain
MRI
d
ataset
w
er
e
u
s
e
d
f
o
r
f
ea
tu
r
e
ex
tr
ac
tio
n
.
T
h
e
R
OI
s
w
er
e
ex
tr
ac
ted
an
d
p
r
o
ce
s
s
ed
u
s
i
n
g
B
E
MD
to
o
b
tain
th
e
co
r
r
esp
o
n
d
in
g
2
an
d
3
co
m
p
o
n
e
n
ts
.
T
h
e
in
p
u
t
f
o
r
f
ea
tu
r
e
ex
tr
ac
tio
n
w
as
d
iv
id
ed
in
to
t
w
o
g
r
o
u
p
s
:
(
i)
R
GB
-
MRI
’
s
R
OI
s
a
n
d
(
ii)
a
co
m
b
i
n
ed
i
m
ag
e
o
f
R
GB
-
M
R
I
’
s
R
OI
s
an
d
R
GB
-
’
s
R
OI
s
(
=
2
,
3
)
.
A
p
r
e
-
tr
ain
ed
VGG1
9
m
o
d
el
w
as
u
s
ed
to
ex
tr
ac
t
h
i
g
h
-
le
v
el
f
ea
t
u
r
es
f
r
o
m
t
h
e
f
u
ll
y
co
n
n
ec
ted
f
c
7
la
y
er
,
g
e
n
er
ati
n
g
a
f
ea
t
u
r
e
v
ec
to
r
o
f
s
ize
1
5
0
×
4
0
9
6
f
o
r
ea
ch
d
ata
g
r
o
u
p
.
C
las
s
i
f
icatio
n
:
t
he
d
ata
s
et
w
as
r
an
d
o
m
l
y
d
iv
id
ed
in
to
an
8
0
:2
0
r
atio
f
o
r
th
e
tr
ain
i
n
g
an
d
te
s
t set
s
,
an
d
s
u
b
s
eq
u
en
t
l
y
f
ed
in
to
M
A
T
L
A
B
’
s
f
it
en
s
e
m
b
le
f
u
n
ctio
n
to
cr
ea
te
a
class
if
icatio
n
m
o
d
el
[
3
5
]
.
T
h
e
en
s
e
m
b
l
e
lear
n
in
g
p
r
o
ce
s
s
o
p
ti
m
izes
p
r
ed
ictiv
e
p
er
f
o
r
m
an
ce
b
y
b
alan
ci
n
g
b
ias
an
d
v
ar
ia
n
ce
.
I
t
ac
h
iev
e
s
th
i
s
b
y
ag
g
r
e
g
ati
n
g
b
ase
lear
n
er
s
th
r
o
u
g
h
b
ag
g
i
n
g
tech
n
iq
u
e
s
(
s
u
ch
as
s
ta
n
d
ar
d
b
ag
g
in
g
a
n
d
r
an
d
o
m
Su
b
s
p
ac
e/S
u
b
B
o
o
s
t
)
an
d
s
eq
u
en
t
ial
b
o
o
s
tin
g
alg
o
r
it
h
m
s
(
in
cl
u
d
in
g
T
o
talB
o
o
s
t,
A
d
aB
o
o
s
t,
L
P
B
o
o
s
t,
an
d
R
USB
o
o
s
t)
.
T
h
e
r
esu
lti
n
g
m
o
d
el
clas
s
i
f
ies
th
e
d
ata
in
to
th
r
ee
d
is
tin
ct
d
is
ea
s
e
c
ateg
o
r
ies:
g
lio
m
as,
m
en
in
g
io
m
as,
an
d
p
it
u
itar
y
t
u
m
o
r
s
; a
ch
ie
v
i
n
g
an
ac
c
u
r
ac
y
o
f
at
least 9
3
%
o
n
b
o
th
t
h
e
tr
ai
n
in
g
a
n
d
test
s
et
s
.
3
.
2
.
1
.
Resul
t
s
T
h
is
s
ec
tio
n
d
etails
th
e
b
es
t
-
p
er
f
o
r
m
i
n
g
c
lass
if
ica
tio
n
m
o
d
els
ac
r
o
s
s
t
h
e
t
w
o
i
m
a
g
e
m
o
d
aliti
es
u
s
e
d
f
o
r
f
ea
t
u
r
e
ex
tr
ac
tio
n
.
a.
C
ase
s
t
u
d
y
I
-
f
ea
t
u
r
e
ex
tr
ac
ti
o
n
u
s
i
n
g
VGG1
9
w
it
h
R
GB
-
R
OI
I
n
th
is
ca
s
e,
f
ea
tu
r
e
s
w
er
e
e
x
t
r
ac
ted
d
ir
ec
tly
f
r
o
m
R
GB
-
R
O
I
i
m
a
g
es
a
n
d
s
ta
n
d
ar
d
ized
f
o
r
en
s
e
m
b
le
lear
n
in
g
,
p
r
o
d
u
cin
g
f
o
u
r
h
ig
h
-
ac
cu
r
ac
y
m
o
d
els
w
ith
d
is
ti
n
c
t
b
ase
lea
r
n
er
s
.
B
o
o
s
tin
g
m
o
d
els
w
er
e
ex
ec
u
ted
ev
er
y
1
,
0
0
0
cy
c
les.
Fi
g
u
r
e
6
p
r
esen
ts
t
h
e
co
n
f
u
s
io
n
m
atr
ices
f
o
r
t
h
ese
m
o
d
els
,
w
it
h
Fig
u
r
e
6(
a)
d
em
o
n
s
tr
ati
n
g
a
p
er
f
ec
t 1
0
0
% tr
ain
i
n
g
ac
c
u
r
ac
y
ac
r
o
s
s
all
f
o
u
r
.
Su
b
B
o
o
s
t
m
o
d
el
u
s
i
n
g
KN
N
(
=
5
)
,
allo
w
ed
f
o
r
j
o
in
t
m
u
l
tiv
ar
iab
le
u
p
d
ates.
T
h
is
s
tati
s
tical
tech
n
iq
u
e
w
a
s
h
ig
h
l
y
e
f
f
ic
ien
t
f
o
r
h
i
g
h
-
d
i
m
e
n
s
io
n
a
l
d
atasets
w
h
er
e
t
h
e
f
ea
tu
r
e
s
ize
ex
ce
ed
ed
th
e
s
a
m
p
le
s
ize
,
as
in
th
i
s
s
tu
d
y
;
it
r
esa
m
p
led
d
ata
at
th
e
s
tar
t
o
f
ea
ch
iter
atio
n
[
3
6
]
.
T
h
is
m
o
d
el
ac
h
iev
ed
1
0
0
%
ac
cu
r
ac
y
o
n
th
e
tr
ai
n
i
n
g
s
et.
E
v
alu
a
tio
n
o
n
th
e
te
s
t
s
et
p
r
o
d
u
ce
d
9
6
.
7
%
ac
cu
r
ac
y
(
w
it
h
a
s
i
n
g
le
m
en
i
n
g
io
m
a
m
is
cla
s
s
i
f
ied
as
a
p
itu
itar
y
tu
m
o
r
)
,
9
6
.
3
%
p
r
ec
is
io
n
,
9
7
.
7
9
%
s
en
s
iti
v
it
y
,
a
n
d
a
9
6
.
5
%
F1
-
s
co
r
e
(
Fig
u
r
e
6
(
b
)
)
.
T
h
ese
r
o
b
u
s
t
m
etr
ics
h
i
g
h
l
ig
h
ted
th
e
m
o
d
el
’
s
v
iab
ilit
y
f
o
r
cli
n
ical
ap
p
licatio
n
.
A
d
a
B
o
o
s
t
,
L
PB
o
o
s
t
,
an
d
b
ag
g
in
g
u
s
e
d
d
ec
i
s
i
o
n
t
r
e
es
(
D
T
)
a
s
b
a
s
e
l
ea
r
n
e
r
s
d
u
e
t
o
th
e
i
r
h
ig
h
ly
s
u
i
t
a
b
le
s
t
r
u
c
tu
r
es
f
o
r
ass
em
b
l
in
g
.
W
h
ile
a
l
l
th
r
e
e
m
o
d
e
l
s
a
ch
ie
v
e
d
1
0
0
%
a
c
cu
r
a
cy
o
n
th
e
t
r
a
in
in
g
s
et
(
Fi
g
u
r
e
6
(
a
)
)
,
th
ei
r
t
e
s
t
s
et
a
cc
u
r
a
cy
w
as
9
3
.
3
%
(
w
ith
e
ac
h
y
i
e
l
d
i
n
g
2
m
is
c
l
ass
if
i
ca
t
i
o
n
s
o
u
t
o
f
3
0
)
(
F
ig
u
r
e
s
6
(c
)
-
(
e
)
)
,
w
h
i
ch
w
as
l
o
w
e
r
t
h
a
n
th
at
o
f
th
e
Su
b
B
o
o
s
t
m
o
d
el
.
T
h
is
p
er
f
o
r
m
a
n
ce
g
a
p
o
cc
u
r
r
ed
b
ec
au
s
e
A
d
aB
o
o
s
t
an
d
L
P
B
o
o
s
t
u
s
ed
all
f
ea
t
u
r
es
d
u
r
i
n
g
ea
ch
iter
atio
n
,
m
ea
n
i
n
g
n
o
is
y
f
ea
t
u
r
es
co
u
ld
n
e
g
ati
v
el
y
i
m
p
ac
t
m
o
d
el
o
u
tco
m
es
[
3
7
]
.
Me
an
w
h
ile,
u
n
lik
e
S
u
b
B
o
o
s
t,
w
h
ic
h
ap
p
li
ed
r
esa
m
p
lin
g
iter
ativ
el
y
,
b
ag
g
i
n
g
r
esa
m
p
led
s
u
b
s
ets
o
f
th
e
d
ataset
s
i
m
u
lta
n
eo
u
s
l
y
,
tr
ain
ed
a
s
ep
ar
ate
m
o
d
el
o
n
ea
c
h
s
u
b
s
e
t
in
p
ar
allel,
a
n
d
f
o
c
u
s
ed
o
n
r
ed
u
cin
g
v
ar
ian
ce
.
T
h
er
ef
o
r
e,
it
ty
p
icall
y
p
r
o
d
u
ce
d
lo
w
er
ac
c
u
r
ac
y
t
h
an
Su
b
B
o
o
s
t
[
3
6
]
,
[
3
7
]
.
Nev
er
th
el
es
s
,
w
it
h
p
r
ec
is
io
n
,
s
en
s
iti
v
it
y
,
a
n
d
F1
-
s
co
r
es
r
an
g
in
g
f
r
o
m
9
3
.
3
%
to
9
4
.
2
%,
th
ese
m
o
d
els
d
e
m
o
n
s
tr
ated
a
s
tr
o
n
g
ab
ilit
y
to
av
o
id
m
is
c
lass
if
ica
tio
n
er
r
o
r
s
.
T
h
e
tr
ain
i
n
g
t
i
m
e
s
f
o
r
m
o
d
el
co
n
s
tr
u
ctio
n
o
r
d
er
ed
f
r
o
m
f
aste
s
t
to
s
lo
w
e
s
t
w
er
e
b
ag
g
i
n
g
(
1
.
8
s
)
,
s
u
b
Sp
ac
et
(
2
s
)
,
L
P
B
o
o
s
t
(
1
7
s
)
,
an
d
A
d
aB
o
o
s
t
(
5
5
3
.
6
s
)
.
C
o
n
s
eq
u
e
n
tl
y
,
co
n
s
id
er
in
g
b
o
th
co
m
p
u
tatio
n
al
ef
f
icie
n
c
y
an
d
class
i
f
icatio
n
ac
cu
r
ac
y
,
s
u
b
Sp
ac
e
e
m
er
g
ed
as t
h
e
o
p
ti
m
al
m
o
d
el.
b.
C
ase
s
t
u
d
y
I
I
-
f
ea
t
u
r
e
ex
tr
ac
ti
o
n
u
s
i
n
g
u
s
i
n
g
VG
G1
9
w
it
h
a
co
m
b
i
n
ed
R
GB
-
R
OI
,
R
GB
-
2
an
d
R
GB
-
3
I
n
th
i
s
ca
s
e,
f
ea
t
u
r
es
w
er
e
ex
tr
ac
ted
f
r
o
m
t
h
e
R
GB
-
R
OI
co
m
b
i
n
ed
w
it
h
R
GB
-
2
an
d
R
GB
-
3
.
T
h
is
co
m
b
in
at
io
n
ca
p
tu
r
ed
s
u
b
tle
in
f
o
r
m
atio
n
th
at
m
i
g
h
t
h
av
e
b
ee
n
m
is
s
ed
b
y
u
s
i
n
g
th
e
R
GB
-
R
O
I
alo
n
e.
P
r
io
r
to
m
o
d
el
tr
ain
in
g
,
th
e
ex
tr
ac
ted
f
ea
t
u
r
es
w
er
e
s
tan
d
ar
d
ized
to
en
s
u
r
e
co
n
s
i
s
t
en
t
s
ca
li
n
g
,
an
d
an
en
s
e
m
b
le
lear
n
i
n
g
ap
p
r
o
ac
h
w
it
h
1
,
0
0
0
b
o
o
s
tin
g
c
y
c
les
was
ap
p
lied
.
T
h
e
r
esu
lts
in
d
icat
ed
th
at
th
e
b
ag
g
in
g
,
Evaluation Warning : The document was created with Spire.PDF for Python.