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
.
2
,
A
p
r
il
20
26
,
p
p
.
536
~
548
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
2
.
27576
536
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//jo
u
r
n
a
l.u
a
d
.
a
c.
id
/in
d
ex
.
p
h
p
/TELK
OM
N
I
K
A
A co
m
pa
ra
tive M
RI
-
ba
sed stu
dy
of
ResNe
t
-
1
5
2
and
no
v
el deep
lea
rning
appro
a
ches for early
Al
z
h
ei
m
e
r’s d
isea
se cl
a
ss
ificatio
n
K
elv
in L
eo
na
rdi K
o
hs
a
s
ih,
O
ct
a
ra
P
riba
di,
Andy
,
Da
niel S
m
it
h S
un
a
rio
D
e
p
a
r
t
me
n
t
o
f
I
n
f
o
r
mat
i
c
s
En
g
i
n
e
e
r
i
n
g
,
S
T
M
I
K
TI
M
E,
M
e
d
a
n
,
I
n
d
o
n
e
si
a
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
u
l
10
,
2025
R
ev
i
s
ed
Dec
11
,
2
0
2
5
A
cc
ep
ted
J
an
30
,
2
0
2
6
A
lzh
e
i
m
e
r’s
d
ise
a
se
(
A
D)
is
th
e
lea
d
in
g
c
a
u
se
o
f
d
e
m
e
n
ti
a
,
m
a
k
in
g
e
a
rly
-
sta
g
e
d
e
tec
ti
o
n
e
ss
e
n
ti
a
l
f
o
r
ti
m
e
l
y
in
terv
e
n
ti
o
n
.
M
o
st
p
ri
o
r
stu
d
ies
h
a
v
e
f
o
c
u
se
d
o
n
b
i
n
a
ry
A
D
c
las
si
f
ic
a
ti
o
n
,
w
h
ich
li
m
it
s
se
n
siti
v
it
y
to
d
ise
a
se
p
ro
g
re
ss
io
n
.
T
h
is
stu
d
y
a
d
d
re
ss
e
d
th
is
g
a
p
b
y
e
v
a
lu
a
ti
n
g
w
h
e
th
e
r
tailo
re
d
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
(
CNN
)
a
rc
h
it
e
c
tu
re
s
c
o
u
l
d
im
p
ro
v
e
sta
g
e
-
a
wa
re
c
la
ss
i
f
ica
ti
o
n
u
sin
g
a
p
u
b
l
icly
a
v
a
il
a
b
le
m
a
g
n
e
ti
c
re
so
n
a
n
c
e
i
m
a
g
in
g
(
M
RI
)
d
a
tas
e
t
c
o
n
tain
i
n
g
3
5
,
9
8
4
im
a
g
e
s
a
c
ro
ss
f
o
u
r
d
iag
n
o
stic
c
a
teg
o
ries
.
T
h
e
d
a
tas
e
t
u
n
d
e
rw
e
n
t
g
ra
y
sc
a
le
c
o
n
v
e
rsio
n
,
re
siz
in
g
,
c
o
n
tras
t
e
n
h
a
n
c
e
m
e
n
t,
n
o
rm
a
li
z
a
ti
o
n
,
a
n
d
c
las
s
b
a
lan
c
in
g
p
rio
r
t
o
m
o
d
e
l
d
e
v
e
lo
p
m
e
n
t.
F
o
u
r
m
o
d
e
ls
w
e
r
e
train
e
d
a
n
d
c
o
m
p
a
re
d
:
Re
s
Ne
t
-
1
5
2
,
a
c
u
sto
m
m
u
lt
icla
ss
C
NN
,
a
o
n
e
-
vs
-
one
(Ov
O)
m
o
d
e
l,
a
n
d
a
o
n
e
-
vs
-
re
st
(O
v
R)
m
o
d
e
l.
P
e
rf
o
rm
a
n
c
e
wa
s
m
e
a
su
re
d
u
sin
g
a
c
c
u
ra
c
y
,
p
re
c
isi
o
n
,
re
c
a
ll
,
F
1
sc
o
re
,
a
n
d
c
o
n
f
u
sio
n
-
m
a
tri
x
–
b
a
se
d
m
e
tri
c
s.
T
h
e
c
u
sto
m
m
u
lt
icla
ss
CN
N
a
c
h
iev
e
d
th
e
stro
n
g
e
st
p
e
rf
o
r
m
a
n
c
e
,
y
ield
in
g
th
e
h
ig
h
e
s
t
a
c
c
u
ra
c
y
a
n
d
b
a
lan
c
e
d
re
su
l
ts
a
c
ro
ss
a
ll
e
v
a
lu
a
ti
o
n
m
e
tri
c
s.
T
h
e
se
f
in
d
in
g
s
d
e
m
o
n
stra
te
th
e
v
a
lu
e
o
f
sy
ste
m
a
ti
c
a
ll
y
c
o
m
p
a
rin
g
d
e
c
o
m
p
o
siti
o
n
stra
te
g
ies
f
o
r
m
u
lt
i
-
sta
g
e
A
lzh
e
i
m
e
r’s
d
e
tec
ti
o
n
a
n
d
h
ig
h
li
g
h
t
t
h
e
p
o
ten
ti
a
l
o
f
th
e
p
r
o
p
o
se
d
a
p
p
r
o
a
c
h
t
o
e
n
h
a
n
c
e
e
a
rl
y
d
iag
n
o
stic
su
p
p
o
rt.
F
u
tu
re
w
o
r
k
m
a
y
in
c
o
rp
o
ra
te
m
u
lt
im
o
d
a
l
in
p
u
ts
o
r
h
y
b
rid
a
rc
h
it
e
c
tu
re
s
to
im
p
ro
v
e
se
n
siti
v
it
y
to
su
b
tl
e
stru
c
t
u
ra
l
c
h
a
n
g
e
s
a
n
d
f
u
rth
e
r
stre
n
g
th
e
n
c
li
n
ica
l
a
p
p
l
ica
b
il
it
y
.
K
ey
w
o
r
d
s
:
C
o
n
v
o
lu
tio
n
al
n
e
u
r
al
n
e
t
w
o
r
k
Dee
p
lear
n
in
g
Me
d
ical
i
m
a
g
e
class
if
ica
tio
n
Mu
tli
-
cla
s
s
clas
s
i
f
icatio
n
T
r
an
s
f
er
lear
n
in
g
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Kelv
i
n
L
eo
n
ar
d
i K
o
h
s
a
s
i
h
Dep
ar
t
m
en
t o
f
I
n
f
o
r
m
atic
s
E
n
g
in
ee
r
i
n
g
,
ST
MI
K
T
I
ME
Me
r
b
ab
u
Stre
et,
Me
d
an
C
it
y
,
No
r
th
Su
m
atr
a
2
0
2
1
2
,
I
n
d
o
n
e
s
ia
E
m
ail:
k
el
v
in
leo
n
ar
d
i@
s
t
m
ik
-
ti
m
e.
ac
.
id
1.
I
NT
RO
D
UCT
I
O
N
A
d
v
an
ce
s
in
i
n
f
o
r
m
atio
n
tec
h
n
o
lo
g
y
a
n
d
co
m
p
u
ter
s
cie
n
c
e
h
av
e
ac
ce
ler
ated
p
r
o
g
r
ess
in
m
ed
ical
d
ata
p
r
o
ce
s
s
in
g
an
d
t
h
e
ap
p
licatio
n
o
f
d
ee
p
lear
n
in
g
(
D
L
)
f
o
r
d
iag
n
o
s
in
g
n
e
u
r
o
lo
g
ic
al
d
is
o
r
d
er
s
u
s
in
g
n
eu
r
o
i
m
ag
in
g
tech
n
iq
u
es
[
1
]
,
[
2
]
.
A
s
DL
tec
h
n
iq
u
es
h
av
e
b
ec
o
m
e
i
n
te
g
r
ated
in
to
n
e
u
r
o
im
ag
in
g
r
esear
ch
[
3
]
,
th
e
av
a
ilab
ilit
y
o
f
lar
g
e,
h
i
g
h
-
q
u
alit
y
d
ata
s
ets
[
4
]
h
as
f
u
r
th
e
r
en
ab
led
th
e
ea
r
lier
an
d
m
o
r
e
ac
cu
r
ate
d
etec
tio
n
o
f
n
e
u
r
o
d
eg
en
er
ati
v
e
d
is
ea
s
e
s
.
A
lz
h
ei
m
er
’
s
d
is
ea
s
e
(
A
D)
is
t
h
e
m
o
s
t
co
m
m
o
n
ca
u
s
e
o
f
d
em
en
tia,
r
esp
o
n
s
ib
l
e
f
o
r
6
0
–
8
0
%
o
f
ca
s
es
a
m
o
n
g
o
ld
er
ad
u
lts
[
5
]
.
A
D
is
a
p
r
o
g
r
ess
i
v
e
an
d
ir
r
ev
er
s
ib
le
n
e
u
r
o
d
eg
en
er
ati
v
e
d
is
o
r
d
er
lead
in
g
to
co
g
n
iti
v
e
d
ec
li
n
e
an
d
ev
e
n
t
u
al
m
o
r
talit
y
[
6
]
,
[
7
]
.
A
t
p
r
ese
n
t,
n
o
cu
r
ati
v
e
tr
ea
t
m
e
n
t
e
x
is
t
s
,
a
n
d
av
ailab
le
t
h
er
ap
ies
s
u
ch
as
ad
u
ca
n
u
m
ab
,
leca
n
e
m
ab
,
an
d
d
o
n
an
e
m
ab
h
av
e
o
n
l
y
b
ee
n
s
h
o
w
n
to
s
lo
w
p
r
o
g
r
ess
io
n
w
h
e
n
ad
m
i
n
i
s
ter
e
d
in
t
h
e
ea
r
l
y
p
h
ase
s
,
p
ar
ticu
lar
l
y
m
ild
co
g
n
iti
v
e
i
m
p
air
m
en
t
(
MCI)
o
r
e
ar
l
y
MCI
(
E
MCI)
[
2
]
,
[
8
]
–
[
1
0
]
.
Nev
er
th
e
less
,
ea
r
l
y
s
y
m
p
to
m
s
ar
e
n
o
n
s
p
ec
i
f
ic
an
d
o
f
te
n
m
is
tak
e
n
f
o
r
n
o
r
m
al
ag
in
g
[
1
1
]
,
s
o
d
iag
n
o
s
i
s
i
s
f
r
eq
u
en
tl
y
d
ela
y
ed
u
n
ti
l
t
h
e
m
o
d
er
ate
o
r
s
ev
er
e
s
tag
e
s
,
w
h
en
i
n
ter
v
e
n
tio
n
s
ar
e
les
s
ef
f
ec
tiv
e
[
1
2
]
.
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
A
co
mp
a
r
a
tive
MRI
-
b
a
s
ed
s
tu
d
y
o
f R
esN
et
-
1
5
2
a
n
d
n
o
ve
l d
e
ep
lea
r
n
in
g
…
(
K
elvin
Leo
n
a
r
d
i Ko
h
s
a
s
ih
)
537
Ma
g
n
e
tic
r
eso
n
a
n
ce
i
m
a
g
i
n
g
(
MRI)
is
cr
u
cial
f
o
r
ex
a
m
i
n
in
g
b
r
ai
n
an
a
to
m
y
a
n
d
d
iag
n
o
s
i
n
g
n
eu
r
o
lo
g
ica
l
d
is
o
r
d
er
s
.
I
t
h
elp
s
id
en
tify
ab
n
o
r
m
al
r
eg
i
o
n
s
[
1
3
]
.
I
ts
h
ig
h
r
eso
lu
ti
o
n
en
ab
les
clea
r
d
if
f
er
e
n
tiatio
n
b
et
w
ee
n
g
r
a
y
an
d
w
h
i
te
m
a
tter
.
MRI
al
s
o
r
ev
ea
ls
s
tr
u
ct
u
r
al
alter
atio
n
s
s
u
ch
a
s
h
ip
p
o
ca
m
p
al
atr
o
p
h
y
,
w
id
e
n
ed
s
u
lci,
an
d
v
en
tr
ic
u
lar
en
lar
g
e
m
e
n
t
t
h
at
ac
co
m
p
a
n
y
d
is
ea
s
e
p
r
o
g
r
ess
i
o
n
[
2
]
,
[
1
2
]
,
[
1
4
]
.
T
h
ese
f
ea
tu
r
e
s
s
u
p
p
o
r
t
ac
cu
r
ate
d
iag
n
o
s
i
s
an
d
ef
f
ec
t
iv
e
d
is
e
ase
m
o
n
ito
r
in
g
[
1
5
]
.
Desp
ite
its
d
iag
n
o
s
tic
v
alu
e,
MRI
in
ter
p
r
etatio
n
r
e
m
ai
n
s
i
n
h
er
e
n
tl
y
s
u
b
j
ec
tiv
e
a
n
d
d
ep
en
d
en
t
o
n
e
x
p
er
t
j
u
d
g
m
e
n
t,
w
h
ic
h
i
n
tr
o
d
u
ce
s
v
ar
iab
ilit
y
an
d
p
o
ten
tial
d
ia
g
n
o
s
tic
i
n
co
n
s
is
te
n
c
y
[
1
2
]
.
T
h
ese
ch
alle
n
g
es
h
i
g
h
l
ig
h
t
a
g
r
o
w
in
g
n
ee
d
f
o
r
au
to
m
ated
,
ar
tif
icia
l
in
telli
g
e
n
ce
(
AI
)
-
s
u
p
p
o
r
ted
an
al
y
s
i
s
th
at
ca
n
i
m
p
r
o
v
e
ac
cu
r
ac
y
an
d
r
e
d
u
ce
o
b
s
er
v
er
v
ar
iab
ilit
y
.
DL
s
y
s
te
m
s
h
av
e
d
e
m
o
n
s
tr
at
ed
in
cr
ea
s
in
g
ef
f
ec
tiv
e
n
es
s
ac
r
o
s
s
v
ar
io
u
s
r
esear
ch
d
o
m
ai
n
s
an
d
ar
e
p
ar
ticu
lar
l
y
u
s
e
f
u
l
f
o
r
an
al
y
zin
g
s
tr
u
ctu
r
al
b
r
ai
n
ch
a
n
g
e
s
ass
o
ciate
d
w
it
h
AD
[
1
6
]
.
DL
al
s
o
i
m
p
r
o
v
e
s
d
iag
n
o
s
t
ic
ef
f
icien
c
y
b
y
r
ed
u
cin
g
ti
m
e,
co
s
t,
an
d
m
a
n
u
al
ef
f
o
r
t
[
1
7
]
.
Ho
w
e
v
er
,
p
r
i
o
r
r
esear
ch
o
n
A
D
d
etec
tio
n
u
s
in
g
D
L
h
as
n
o
t
y
et
ac
h
ie
v
ed
o
p
ti
m
al
e
f
f
i
c
ien
c
y
[
1
8
]
.
C
o
m
p
ar
ati
v
e
s
t
u
d
ies
e
v
alu
a
tin
g
y
o
u
o
n
l
y
lo
o
k
o
n
ce
(
YOL
O
)
v
ar
ia
n
ts
,
Den
s
eNe
t,
r
esid
u
al
n
et
w
o
r
k
(
R
esNet
)
,
v
i
s
u
al
g
eo
m
et
r
y
g
r
o
u
p
(
VGG
)
,
E
f
f
icien
tNet,
an
d
v
i
s
io
n
tr
a
n
s
f
o
r
m
er
s
i
n
d
icate
th
at
al
th
o
u
g
h
h
i
g
h
-
ca
p
ac
it
y
f
ea
tu
r
e
ex
tr
ac
to
r
s
ac
h
iev
e
s
tr
o
n
g
p
er
f
o
r
m
a
n
ce
i
n
b
in
ar
y
AD
cla
s
s
i
f
icatio
n
,
th
e
y
o
f
te
n
s
tr
u
g
g
le
in
m
u
lt
i
-
s
ta
g
e
tas
k
s
,
esp
ec
iall
y
at
MCI
o
r
E
MCI
lev
els
w
h
er
e
in
ter
-
c
lass
d
if
f
er
en
ce
s
ar
e
s
u
b
tle
[
1
9
]
–
[
2
1
]
.
E
v
en
m
o
d
els
s
u
c
h
as
Den
s
eNe
t
-
1
6
9
an
d
R
esNet
-
50
s
h
o
w
r
ed
u
ce
d
d
is
cr
i
m
in
at
io
n
ab
ilit
y
w
h
e
n
r
eq
u
ir
ed
to
d
is
tin
g
u
i
s
h
clo
s
el
y
r
elate
d
Alzh
ei
m
er
’
s
s
tag
e
s
[
2
2
]
.
Mu
lti
m
o
d
al
MRI
–
p
o
s
itro
n
em
is
s
io
n
to
m
o
g
r
ap
h
y
(
P
E
T
)
f
u
s
io
n
ap
p
r
o
ac
h
es
h
av
e
d
e
m
o
n
s
tr
ated
i
m
p
r
o
v
ed
r
o
b
u
s
tn
es
s
an
d
in
ter
p
r
etab
ilit
y
,
p
ar
ticu
lar
ly
w
h
en
co
m
b
i
n
ed
w
it
h
ex
p
lai
n
ab
ilit
y
tech
n
iq
u
e
s
;
I
n
co
n
tr
ast,
th
ese
m
et
h
o
d
s
s
till
d
o
n
o
t
co
n
s
is
te
n
tl
y
o
u
tp
er
f
o
r
m
o
p
ti
m
ized
s
i
n
g
le
-
m
o
d
alit
y
C
NN
s
f
o
r
m
u
lt
i
-
clas
s
s
ta
g
i
n
g
ta
s
k
s
[
2
3
]
.
Ov
er
all,
r
ep
o
r
te
d
ac
cu
r
a
cies
co
m
m
o
n
l
y
r
a
n
g
e
b
et
w
ee
n
7
0
%
an
d
8
6
%,
w
it
h
m
o
s
t
s
tu
d
ies
f
o
c
u
s
i
n
g
o
n
b
in
ar
y
A
D
d
etec
tio
n
r
at
h
er
th
an
f
i
n
e
-
g
r
ain
ed
s
ta
g
i
n
g
[
1
9
]
,
[
2
2
]
–
[
2
5
]
.
T
h
is
tr
en
d
h
i
g
h
li
g
h
t
s
a
p
er
s
is
ten
t
g
ap
i
n
d
ev
elo
p
in
g
m
o
d
els ca
p
ab
le
o
f
r
eliab
l
y
d
if
f
er
en
tia
tin
g
ea
r
l
y
Alzh
ei
m
er
’
s
s
ta
g
es.
T
o
a
d
d
r
ess
th
ese
li
m
i
tatio
n
s
,
th
i
s
s
t
u
d
y
e
v
al
u
ate
s
w
h
et
h
e
r
tailo
r
ed
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
et
w
o
r
k
(
C
NN)
ar
ch
itect
u
r
es
ca
n
i
m
p
r
o
v
e
f
i
n
e
-
g
r
ai
n
ed
,
s
tag
e
-
a
w
ar
e
A
D
c
lass
if
icatio
n
,
a
n
ar
ea
w
h
er
e
ex
i
s
ti
n
g
DL
ap
p
r
o
ac
h
es
r
em
ai
n
li
m
ited
.
U
s
in
g
a
p
u
b
licl
y
a
v
ailab
le
MR
I
d
ataset
co
m
p
r
is
i
n
g
3
5
,
9
8
4
i
m
ag
e
s
ac
r
o
s
s
f
o
u
r
di
ag
n
o
s
t
ic
ca
te
g
o
r
ies,
t
h
e
d
ataset
w
as
p
r
ep
r
o
ce
s
s
ed
u
s
i
n
g
g
r
a
y
s
ca
le
co
n
v
er
s
io
n
,
r
esizin
g
,
co
n
tr
ast
en
h
a
n
ce
m
en
t,
n
o
r
m
aliza
tio
n
,
an
d
class
b
alan
ci
n
g
.
T
h
r
ee
d
ec
o
m
p
o
s
itio
n
s
tr
ateg
ies
-
m
u
lticlas
s
,
o
n
e
-
vs
-
o
n
e
(
O
v
O
)
,
an
d
o
n
e
-
vs
-
r
e
s
t
(O
v
R
)
-
w
er
e
i
m
p
le
m
e
n
ted
alo
n
g
s
id
e
a
R
esNet
-
1
5
2
tr
an
s
f
er
-
lear
n
i
n
g
b
aseli
n
e
to
en
ab
le
a
s
y
s
te
m
atic
a
n
d
co
n
tr
o
lled
c
o
m
p
ar
is
o
n
.
T
h
e
n
o
v
elt
y
o
f
t
h
is
w
o
r
k
lies
in
its
s
tr
u
ctu
r
ed
ev
alu
a
tio
n
o
f
t
h
ese
s
tr
ateg
ie
s
o
n
a
lar
g
e
an
d
b
alan
ce
d
MRI
co
r
p
u
s
,
ai
m
ed
at
i
m
p
r
o
v
in
g
s
ta
g
e
-
s
p
ec
i
f
ic
s
en
s
i
tiv
it
y
w
h
ile
m
ai
n
tai
n
in
g
co
m
p
u
tatio
n
al
f
e
asib
ilit
y
.
I
m
p
r
o
v
in
g
clas
s
i
f
ica
tio
n
g
r
a
n
u
lar
it
y
i
s
clin
ica
ll
y
i
m
p
o
r
tan
t
b
ec
au
s
e
p
r
o
g
n
o
s
is
,
r
esp
o
n
s
e
to
t
h
er
ap
y
,
an
d
el
ig
ib
il
it
y
f
o
r
d
is
ea
s
e
-
m
o
d
i
f
y
i
n
g
tr
ea
t
m
e
n
t
s
d
ep
en
d
h
ea
v
i
l
y
o
n
ac
c
u
r
ate
ea
r
l
y
-
s
ta
g
e
d
if
f
er
en
tiat
io
n
.
T
h
e
p
r
o
p
o
s
ed
s
tag
e
-
a
w
ar
e
DL
f
r
a
m
e
w
o
r
k
th
er
e
f
o
r
e
o
f
f
er
s
e
n
h
an
ce
d
d
iag
n
o
s
tic
v
alu
e
f
o
r
s
u
p
p
o
r
tin
g
ea
r
l
y
d
ete
ctio
n
o
f
AD
.
2.
M
E
T
H
O
D
T
h
is
s
tu
d
y
co
m
p
ar
e
s
th
e
p
er
f
o
r
m
a
n
ce
o
f
R
es
Net
-
1
5
2
,
ap
p
lied
th
r
o
u
g
h
tr
an
s
f
er
lear
n
i
n
g
,
w
it
h
a
cu
s
to
m
D
L
m
o
d
el
t
h
a
t
e
m
p
lo
y
s
m
u
lticla
s
s
d
ec
o
m
p
o
s
it
io
n
s
tr
ateg
ies,
in
c
lu
d
i
n
g
O
v
O
an
d
O
v
R
,
f
o
r
A
lz
h
ei
m
er
’
s
d
etec
tio
n
f
r
o
m
MRI
s
ca
n
s
.
T
h
e
m
et
h
o
d
o
lo
g
y
i
s
s
tr
u
ct
u
r
ed
in
to
f
i
v
e
p
h
as
es:
d
ata
ac
q
u
is
itio
n
,
p
r
ep
r
o
ce
s
s
in
g
,
clas
s
b
alan
c
in
g
an
d
d
ata
s
et
p
ar
titi
o
n
i
n
g
,
m
o
d
el
ar
ch
itectu
r
e
w
i
th
tr
ain
i
n
g
co
n
f
i
g
u
r
atio
n
,
an
d
m
o
d
el
e
v
al
u
atio
n
s
u
p
p
o
r
ted
b
y
s
tati
s
tical
a
n
al
y
s
i
s
,
as s
u
m
m
ar
ized
in
Fig
u
r
e
1
.
Fig
u
r
e
1
.
Me
th
o
d
r
esear
ch
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
.
2
,
A
p
r
il
20
26
:
5
3
6
-
548
538
2
.
1
.
Da
t
a
a
cquis
it
io
n
T
h
e
d
ataset
u
s
ed
in
th
is
s
tu
d
y
w
a
s
o
b
tain
ed
f
r
o
m
th
e
p
u
b
licl
y
a
v
ailab
le
Kag
g
le
A
lz
h
ei
m
er
’
s
Di
s
ea
s
e
MRI
d
ataset,
w
h
ic
h
p
r
o
v
id
es
a
co
m
p
r
eh
e
n
s
i
v
e
co
llectio
n
o
f
b
r
ain
MRI
s
ca
n
s
lab
eled
ac
c
o
r
d
in
g
to
co
g
n
itiv
e
co
n
d
itio
n
.
As
s
h
o
w
n
i
n
Fi
g
u
r
e
2
,
th
e
d
atase
t
i
n
cl
u
d
e
s
f
o
u
r
d
iag
n
o
s
t
ic
ca
te
g
o
r
ies:
n
on
-
d
e
m
e
n
ted
(
ND)
,
v
er
y
m
ild
d
e
m
e
n
ted
(
VM
D)
,
m
ild
d
em
e
n
ted
(
MD
)
,
an
d
m
o
d
er
ate
d
e
m
en
ted
(
Mo
d
D)
.
T
h
ese
ca
teg
o
r
ies
r
ep
r
esen
t
p
r
o
g
r
ess
iv
e
le
v
els
o
f
co
g
n
iti
v
e
d
ec
lin
e,
r
an
g
i
n
g
f
r
o
m
n
o
r
m
al
co
g
n
iti
v
e
f
u
n
ctio
n
to
m
o
r
e
ad
v
an
ce
d
d
e
m
en
ti
a
s
y
m
p
to
m
s
,
an
d
b
r
o
a
d
ly
co
r
r
esp
o
n
d
to
th
e
ea
r
ly
to
m
o
d
er
ate
s
tag
es
o
f
AD
p
r
o
g
r
ess
io
n
.
I
n
to
tal,
th
e
d
ataset
co
m
p
r
is
e
s
3
5
,
9
8
4
MRI
i
m
a
g
e
s
,
d
is
tr
ib
u
ted
a
s
f
o
llo
w
s
:
9
,
6
0
0
i
m
a
g
es
(
2
8
.
2
%)
f
o
r
ND,
8
,
9
6
0
i
m
ag
e
s
(
2
6
.
4
%)
f
o
r
VM
D,
8
,
9
6
0
im
a
g
es
(
2
6
.
4
%)
f
o
r
MD
,
an
d
6
,
4
6
4
im
a
g
es
(
1
9
.
0
%)
f
o
r
Mo
d
D.
A
ll
i
m
ag
es
ar
e
p
r
o
v
id
ed
in
a
s
tan
d
ar
d
ized
f
o
r
m
at
an
d
d
i
m
en
s
io
n
,
m
i
n
i
m
izi
n
g
v
ar
iab
i
lit
y
a
n
d
en
s
u
r
i
n
g
s
ea
m
le
s
s
i
n
teg
r
at
io
n
i
n
to
t
h
e
p
r
ep
r
o
ce
s
s
in
g
p
ip
elin
e.
T
h
e
d
ataset
w
a
s
ch
o
s
e
n
b
ec
au
s
e
it
is
o
p
en
ly
ac
ce
s
s
ib
le,
w
h
ic
h
g
u
ar
an
tee
s
r
ep
r
o
d
u
cib
ilit
y
a
n
d
tr
an
s
p
ar
e
n
c
y
.
I
t
also
h
as
a
r
elati
v
el
y
b
alan
ce
d
d
is
tr
ib
u
tio
n
ac
r
o
s
s
i
m
p
air
m
e
n
t
s
tag
e
s
,
w
h
ic
h
s
u
p
p
o
r
ts
r
o
b
u
s
t
m
o
d
el
tr
ain
i
n
g
a
n
d
ev
al
u
atio
n
.
A
d
d
i
tio
n
all
y
,
t
h
e
d
ataset
o
f
f
er
s
h
i
g
h
-
q
u
a
lit
y
lab
eli
n
g
,
en
ab
lin
g
r
eliab
le
s
u
p
er
v
i
s
ed
lear
n
in
g
.
I
t
s
d
is
tr
ib
u
tio
n
ac
r
o
s
s
f
o
u
r
d
iag
n
o
s
tic
ca
teg
o
r
ies
f
u
r
th
er
allo
w
s
t
h
is
s
tu
d
y
to
ad
d
r
ess
co
m
m
o
n
c
h
a
llen
g
es
i
n
m
ed
ical
i
m
a
g
i
n
g
r
esear
ch
,
i
n
cl
u
d
in
g
cla
s
s
i
m
b
ala
n
ce
a
n
d
in
ter
-
clas
s
s
i
m
ilar
it
y
,
t
h
r
o
u
g
h
tar
g
eted
p
r
ep
r
o
ce
s
s
in
g
a
n
d
d
ata
au
g
m
e
n
t
atio
n
s
tr
ate
g
ies
.
Fig
u
r
e
2
.
R
ep
r
esen
tati
v
e
MRI
b
r
ain
s
ca
n
s
a
m
p
les
f
r
o
m
t
h
e
d
ataset
2
.
2
.
Cla
s
s
ba
la
ncing
a
nd
da
t
a
s
et
pa
rt
it
io
nin
g
T
h
e
i
n
it
i
a
l
d
a
t
as
et
e
x
h
i
b
i
t
e
d
c
la
s
s
im
b
al
an
c
e
,
w
it
h
ND
c
o
n
t
a
in
in
g
9
,
6
0
0
s
am
p
le
s
,
M
D
8
,
9
6
0
s
am
p
le
s
,
V
M
D
8
,
9
6
0
s
am
p
l
e
s
,
an
d
M
o
d
D
6
,
4
6
4
s
am
p
l
e
s
,
w
h
i
ch
c
o
u
ld
b
i
a
s
t
h
e
m
o
d
e
l
t
o
w
a
r
d
m
a
jo
r
i
ty
cl
a
s
s
es
.
I
n
th
is
a
p
p
r
o
a
c
h
,
th
e
m
a
jo
r
i
ty
c
la
s
s
s
am
p
l
es
w
e
r
e
r
an
d
o
m
ly
d
i
s
c
a
r
d
ed
u
n
t
il
c
l
as
s
s
iz
e
s
w
e
r
e
e
q
u
a
li
ze
d
,
t
h
e
r
e
b
y
r
e
d
u
c
in
g
b
i
a
s
t
o
w
a
r
d
d
o
m
i
n
a
n
t
c
a
t
eg
o
r
ie
s
,
w
h
i
ch
i
s
c
o
n
s
i
s
t
en
t
w
it
h
th
e
f
u
n
d
am
en
t
a
l
p
r
in
c
i
p
l
es
o
f
r
e
-
b
a
l
an
cin
g
m
e
th
o
d
s
[
2
6
]
,
[
2
7
]
.
A
f
ix
e
d
r
an
d
o
m
s
e
ed
(
4
2
)
w
as
u
s
e
d
t
o
e
n
s
u
r
e
r
e
p
r
o
d
u
c
i
b
i
li
ty
.
T
h
e
b
a
l
a
n
c
e
d
d
a
t
a
s
e
t
w
a
s
s
u
b
s
e
q
u
e
n
t
ly
d
i
v
i
d
e
d
in
t
o
8
0
%
f
o
r
tr
ain
in
g
a
n
d
2
0
%
f
o
r
test
in
g
to
s
u
p
p
o
r
t r
o
b
u
s
t
m
o
d
el
lear
n
in
g
an
d
o
b
jectiv
e
p
er
f
o
r
m
a
n
ce
ass
es
s
m
en
t
o
n
u
n
s
ee
n
d
ata.
T
h
e
class
d
is
tr
ib
u
tio
n
o
f
t
h
e
d
ataset
b
ef
o
r
e
an
d
af
ter
th
e
r
e
-
b
al
an
cin
g
p
r
o
ce
d
u
r
e
is
p
r
esen
ted
in
Fig
u
r
e
3
.
Fi
g
u
r
e
3
(
a)
s
h
o
w
s
t
h
e
clas
s
d
is
tr
i
b
u
tio
n
b
ef
o
r
e
t
h
e
r
e
-
b
ala
n
ci
n
g
p
r
o
ce
d
u
r
e,
w
h
ile
Fig
u
r
e
3
(
b
)
illu
s
tr
ates t
h
e
d
is
tr
ib
u
tio
n
a
f
ter
th
e
r
e
-
b
alan
ci
n
g
p
r
o
ce
d
u
r
e
.
2
.
3
.
Da
t
a
prepro
ce
s
s
ing
B
e
f
o
r
e
m
o
d
e
l
t
r
ai
n
in
g
,
a
l
l
MR
I
im
ag
es
u
n
d
e
r
w
en
t
a
s
ta
n
d
a
r
d
iz
e
d
p
r
e
p
r
o
c
e
s
s
in
g
p
i
p
el
in
e
c
o
n
s
is
t
in
g
o
f
:
−
Gr
a
y
s
ca
le
co
n
v
er
s
io
n
to
r
ed
u
ce
co
m
p
u
tatio
n
a
l c
o
m
p
lex
it
y
wh
ile
p
r
eser
v
i
n
g
s
tr
u
c
tu
r
al
i
n
f
o
r
m
atio
n
.
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
A
co
mp
a
r
a
tive
MRI
-
b
a
s
ed
s
tu
d
y
o
f R
esN
et
-
1
5
2
a
n
d
n
o
ve
l d
e
ep
lea
r
n
in
g
…
(
K
elvin
Leo
n
a
r
d
i Ko
h
s
a
s
ih
)
539
−
R
esizi
n
g
to
2
2
4
×2
2
4
p
ix
els to
en
s
u
r
e
u
n
i
f
o
r
m
in
p
u
t d
i
m
e
n
s
i
o
n
s
f
o
r
th
e
C
NN
m
o
d
el
[
2
8
]
.
−
C
o
n
tr
ast
e
n
h
an
ce
m
e
n
t
u
s
i
n
g
c
o
n
tr
ast
li
m
ited
ad
ap
tiv
e
h
i
s
to
g
r
a
m
eq
u
aliza
tio
n
(
C
L
A
HE
)
t
o
i
m
p
r
o
v
e
lo
ca
l
co
n
tr
ast an
d
h
ig
h
li
g
h
t s
u
b
tle
a
n
ato
m
ica
l f
ea
t
u
r
es r
ele
v
an
t to
A
D
cla
s
s
i
f
icat
io
n
[
1
8
]
,
[
2
9]
.
−
B
ilater
al
f
ilter
i
n
g
to
s
u
p
p
r
ess
n
o
is
e
w
h
ile
p
r
eser
v
i
n
g
ed
g
e
d
etails.
−
Min
–
m
a
x
n
o
r
m
aliza
tio
n
to
s
ca
le
p
ix
el
in
te
n
s
ities
i
n
to
t
h
e
[
0
,
1
]
r
an
g
e
f
o
r
s
tab
le
n
et
w
o
r
k
o
p
ti
m
izat
io
n
.
−
B
atch
w
is
e
p
r
o
ce
s
s
i
n
g
o
f
3
0
0
im
ag
e
s
p
er
c
y
cle
to
m
ai
n
tai
n
m
e
m
o
r
y
ef
f
ic
ien
c
y
d
u
r
i
n
g
p
r
ep
r
o
ce
s
s
in
g
.
(
a)
(
b
)
Fig
u
r
e
3
.
I
n
itial a
n
d
b
alan
ce
d
class
d
is
tr
ib
u
tio
n
s
o
f
th
e
AD
i
m
ag
e
d
ata
s
et;
(
a)
i
n
itial i
m
b
al
an
ce
d
d
is
tr
ib
u
tio
n
an
d
(
b
)
b
alan
ce
d
d
is
tr
ib
u
tio
n
af
ter
u
n
d
er
s
a
m
p
li
n
g
2
.
4
.
M
o
del a
rc
hite
ct
ure
a
nd
t
ra
ini
ng
co
nfig
ura
t
io
n
2
.
4
.
1
.
Resnet
1
5
2
R
esNet
-
1
5
2
is
a
d
ee
p
v
ar
ian
t
o
f
th
e
R
esNe
t
f
a
m
il
y
t
h
at
ad
d
r
ess
es
g
r
ad
ien
t
d
eg
r
ad
atio
n
in
v
er
y
d
ee
p
m
o
d
el
s
th
r
o
u
g
h
r
esid
u
a
l
lear
n
in
g
w
it
h
id
en
tit
y
s
h
o
r
tcu
t
co
n
n
ec
tio
n
s
[
3
0
]
.
W
ith
1
5
2
la
y
er
s
o
r
g
an
ized
in
to
b
o
ttlen
ec
k
b
l
o
ck
s
,
it
p
r
o
v
i
d
e
s
h
i
e
r
a
r
ch
i
c
a
l
f
e
a
tu
r
e
ex
t
r
ac
t
i
o
n
t
h
at
is
ef
f
e
ct
iv
e
f
o
r
d
e
t
ec
t
in
g
s
u
b
t
l
e
m
o
r
p
h
o
l
o
g
i
ca
l
c
h
an
g
es
in
b
r
a
in
MR
I
s
ca
n
s
ass
o
c
i
a
te
d
w
it
h
e
a
r
ly
AD
[
3
1
]
,
[
3
2
]
.
B
ey
o
n
d
i
ts
w
i
d
es
p
r
e
a
d
u
s
e
,
R
es
N
et
-
1
5
2
w
a
s
s
e
l
e
ct
e
d
i
n
th
is
s
tu
d
y
b
e
ca
u
s
e
r
e
c
e
n
t
w
o
r
k
o
n
d
e
e
p
r
es
i
d
u
a
l
an
d
c
o
n
v
o
lu
ti
o
n
al
a
r
ch
i
t
e
ct
u
r
e
s
ap
p
l
i
e
d
t
o
s
t
r
u
ct
u
r
al
M
R
I
h
as
s
h
o
w
n
th
at
v
e
r
y
d
ee
p
n
e
t
w
o
r
k
s
c
an
c
a
p
t
u
r
e
d
i
s
t
r
i
b
u
te
d
a
n
d
f
in
e
-
g
r
a
in
e
d
p
a
tt
e
r
n
s
o
f
b
r
a
i
n
at
r
o
p
h
y
t
h
at
ar
e
cr
itical
f
o
r
ac
cu
r
ate
AD
s
t
ag
in
g
[
3
3
]
.
I
n
th
is
s
tu
d
y
,
R
e
s
N
e
t
-
1
5
2
w
as
im
p
lem
en
t
e
d
u
s
in
g
t
r
an
s
f
e
r
l
e
ar
n
in
g
w
ith
p
r
e
t
r
a
in
e
d
I
m
ag
e
N
et
w
e
ig
h
ts
,
w
h
e
r
e
c
o
n
v
o
lu
t
i
o
n
a
l
l
ay
e
r
s
w
e
r
e
f
r
o
z
en
t
o
p
r
e
s
e
r
v
e
g
en
e
r
a
l
f
e
atu
r
e
r
e
p
r
e
s
en
t
at
i
o
n
s
a
n
d
a
t
a
s
k
-
s
p
e
c
if
i
c
c
l
a
s
s
if
ic
a
t
i
o
n
h
e
a
d
w
as
t
r
ain
e
d
o
n
th
e
A
l
zh
e
im
e
r
’
s
d
a
t
as
e
t
[
3
1
]
,
[
3
2
]
,
[
3
4
]
.
T
h
e
c
l
a
s
s
if
ic
a
t
i
o
n
h
e
a
d
c
o
n
s
i
s
t
e
d
o
f
a
g
l
o
b
a
l
a
v
e
r
ag
e
p
o
o
l
in
g
l
ay
e
r
,
a
d
e
n
s
e
l
ay
e
r
w
i
th
1
0
2
4
n
eu
r
o
n
s
an
d
r
e
c
tif
i
e
d
li
n
e
a
r
u
n
i
t
(
R
eL
U
)
a
c
t
iv
at
i
o
n
,
an
d
a
f
in
al
d
en
s
e
l
ay
e
r
w
it
h
f
o
u
r
n
eu
r
o
n
s
an
d
s
o
f
tm
ax
a
c
tiv
a
t
i
o
n
t
o
c
l
a
s
s
if
y
t
h
e
d
i
ag
n
o
s
t
i
c
c
a
t
eg
o
r
ie
s
.
T
h
e
m
o
d
el
w
as
t
r
a
in
e
d
u
s
i
n
g
th
e
a
d
a
p
ti
v
e
m
o
m
en
t
e
s
t
im
a
ti
o
n
(
A
d
am
)
o
p
t
im
i
z
e
r
w
ith
c
at
eg
o
r
i
c
a
l
c
r
o
s
s
-
e
n
t
r
o
p
y
l
o
s
s
a
n
d
a
b
a
t
ch
s
i
z
e
o
f
3
2
.
E
a
r
ly
s
t
o
p
p
in
g
w
as
a
p
p
l
i
e
d
t
o
p
r
e
v
en
t
o
v
e
r
f
i
tt
in
g
an
d
t
o
r
e
s
t
o
r
e
th
e
b
es
t
w
eig
h
t
s
.
T
h
i
s
c
o
n
f
ig
u
r
at
i
o
n
a
l
ig
n
s
w
ith
es
t
a
b
li
s
h
e
d
t
r
an
s
f
e
r
l
e
a
r
n
in
g
p
r
a
ct
i
c
es
in
m
e
d
ic
a
l
im
ag
e
an
aly
s
is
[
3
2
]
,
[
3
4
]
.
2
.
4
.
2
.
Cus
t
o
m
cla
s
s
if
ier
C
NNs
h
a
v
e
d
e
m
o
n
s
tr
ated
s
tr
o
n
g
e
f
f
ec
tiv
e
n
es
s
i
n
n
e
u
r
o
im
ag
in
g
b
ec
au
s
e
th
e
y
lear
n
h
ier
ar
ch
ica
l
s
p
atial
f
ea
tu
r
es
d
ir
ec
tl
y
f
r
o
m
MRI
d
ata
[
2
]
,
[
3
5
]
,
[
3
6
]
.
P
r
io
r
s
tu
d
ies
a
ls
o
i
n
d
i
ca
te
t
h
at
task
-
s
p
ec
i
f
ic
C
NN
ar
ch
itect
u
r
es
ar
e
b
etter
s
u
ited
to
ca
p
tu
r
e
d
is
ea
s
e
-
r
elate
d
m
o
r
p
h
o
lo
g
ical
p
atter
n
s
,
o
f
f
er
i
n
g
i
m
p
r
o
v
ed
s
tab
ilit
y
an
d
in
ter
p
r
etab
ilit
y
i
n
m
ed
ical
i
m
a
g
in
g
ap
p
licatio
n
s
[
3
7
]
–
[
4
0
]
.
I
n
th
is
s
tu
d
y
,
a
cu
s
to
m
m
u
lticla
s
s
C
NN
w
as
d
ev
el
o
p
ed
u
s
in
g
s
tac
k
ed
co
n
v
o
lu
tio
n
al
b
lo
ck
s
f
o
r
f
ea
t
u
r
e
ex
tr
ac
tio
n
an
d
f
u
ll
y
co
n
n
ec
ted
l
a
y
er
s
w
ith
d
r
o
p
o
u
t
to
en
h
a
n
ce
g
e
n
er
aliza
tio
n
p
er
f
o
r
m
a
n
ce
[
4
1
]
,
[
4
2
]
.
A
f
i
n
al
s
o
f
t
m
ax
la
y
er
p
r
o
d
u
ce
s
p
r
ed
ictio
n
s
ac
r
o
s
s
f
o
u
r
d
iag
n
o
s
t
ic
ca
teg
o
r
ies b
ased
o
n
s
tr
u
ct
u
r
al
MRI
f
ea
tu
r
es
[
4
3
]
.
A
co
n
tr
o
lled
co
m
p
ar
is
o
n
f
o
r
m
u
lti
-
s
ta
g
e
A
lz
h
ei
m
er
’
s
clas
s
if
icatio
n
w
as
ac
h
ie
v
ed
b
y
i
m
p
le
m
en
t
in
g
t
w
o
ad
d
itio
n
al
f
o
r
m
u
latio
n
s
.
T
h
e
Ov
O
ap
p
r
o
ac
h
tr
ain
s
in
d
iv
id
u
al
p
air
w
i
s
e
class
i
f
ier
s
t
h
at
r
ef
in
e
d
ec
is
io
n
b
o
u
n
d
ar
ies
b
et
w
ee
n
s
p
ec
if
ic
c
lass
p
air
s
,
w
h
ile
th
e
Ov
R
f
o
r
m
u
latio
n
tr
ain
s
s
ep
ar
ate
m
o
d
e
ls
to
d
is
tin
g
u
is
h
o
n
e
d
iag
n
o
s
t
ic
ca
te
g
o
r
y
f
r
o
m
t
h
e
r
e
m
ai
n
in
g
c
lass
e
s
.
B
o
th
a
p
p
r
o
ac
h
es
em
p
l
o
y
c
o
m
p
a
c
t CN
N
b
ac
k
b
o
n
e
s
an
d
f
o
l
l
o
w
e
s
t
a
b
li
s
h
e
d
m
u
l
ti
c
l
ass
l
e
a
r
n
in
g
p
r
a
c
t
ic
e
s
in
m
e
d
ic
a
l
im
ag
in
g
[
4
4
]
–
[
4
8
]
.
D
e
t
a
i
l
e
d
a
r
ch
i
te
ct
u
r
al
s
p
e
c
if
i
c
a
ti
o
n
s
,
i
n
cl
u
d
i
n
g
l
ay
e
r
c
o
m
p
o
s
it
i
o
n
,
o
u
t
p
u
t
d
im
en
s
i
o
n
s
,
a
n
d
p
a
r
am
ete
r
c
o
u
n
ts
,
a
r
e
p
r
o
v
i
d
e
d
in
T
a
b
l
e
s
1
t
o
4
.
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
.
2
,
A
p
r
il
20
26
:
5
3
6
-
548
540
T
ab
le
1
.
L
ay
er
s
o
f
cu
s
to
m
m
u
l
ticlass
C
NN
Ty
p
e
D
e
scri
p
t
i
o
n
P
a
r
a
ma
t
e
r
s
O
u
t
p
u
t
s
h
a
p
e
I
n
p
u
t
l
a
y
e
r
I
n
p
u
t
_
1
0
(
N
o
n
e
,
2
2
4
,
2
2
4
,
1
)
C
o
n
v
b
l
o
c
k
1
2
×
C
o
n
v
(
3
2
)
+
B
a
t
c
h
N
o
r
m
+
M
a
x
P
o
o
l
9
,
8
2
4
(
N
o
n
e
,
1
1
0
,
1
1
0
,
3
2
)
C
o
n
v
b
l
o
c
k
2
2
×
C
o
n
v
(
6
4
)
+
B
a
t
c
h
N
o
r
m
+
M
a
x
P
o
o
l
5
5
,
9
3
6
(
N
o
n
e
,
5
3
,
5
3
,
6
4
)
C
o
n
v
b
l
o
c
k
3
2
×
C
o
n
v
(
1
2
8
)
+
B
a
t
c
h
N
o
r
m +
M
a
x
P
o
o
l
2
2
2
,
4
6
4
(
N
o
n
e
,
2
4
,
2
4
,
1
2
8
)
C
o
n
v
b
l
o
c
k
4
1
×
C
o
n
v
(
2
5
6
)
+
B
a
t
c
h
N
o
r
m +
M
a
x
P
o
o
l
2
9
6
,
1
9
2
(
N
o
n
e
,
1
1
,
1
1
,
2
5
6
)
F
l
a
t
t
e
n
F
l
a
t
t
e
n
0
(
N
o
n
e
,
3
0
9
7
6
)
D
e
n
se
(
f
c
1
)
D
e
n
se
(
5
1
2
)
1
5
,
8
6
0
,
2
2
4
(
N
o
n
e
,
5
1
2
)
D
r
o
p
o
u
t
D
r
o
p
o
u
t
(
0
.
5
)
0
(
N
o
n
e
,
5
1
2
)
D
e
n
se
(
f
c
2
)
D
e
n
se
(
2
5
6
)
1
3
1
,
3
2
8
(
N
o
n
e
,
2
5
6
)
D
r
o
p
o
u
t
D
r
o
p
o
u
t
(
0
.
5
)
0
(
N
o
n
e
,
2
5
6
)
O
u
t
p
u
t
l
a
y
e
r
D
e
n
se
(
4
,
so
f
t
max
)
1
,
0
2
8
(
N
o
n
e
,
4
)
T
ab
le
2
.
L
ay
er
s
o
f
cu
s
to
m
O
v
O
b
aselin
e
-
C
NN
Ty
p
e
D
e
scri
p
t
i
o
n
P
a
r
a
ma
t
e
r
s
O
u
t
p
u
t
s
h
a
p
e
I
n
p
u
t
l
a
y
e
r
I
n
p
u
t
_
1
0
(
N
o
n
e
,
2
2
4
,
2
2
4
,
1
)
C
o
n
v
2
D
C
o
n
v
2
D
(
3
2
,
3
×
3
,
R
e
L
U
)
3
2
0
(
N
o
n
e
,
2
2
2
,
2
2
2
,
3
2
)
M
a
x
P
o
o
l
i
n
g
2
D
M
a
x
P
o
o
l
(
2
×
2
)
0
(
N
o
n
e
,
1
1
1
,
1
1
1
,
3
2
)
C
o
n
v
2
D
C
o
n
v
2
D
(
6
4
,
3
×
3
,
R
e
L
U
)
1
8
,
4
9
6
(
N
o
n
e
,
1
0
9
,
1
0
9
,
6
4
)
M
a
x
P
o
o
l
i
n
g
2
D
M
a
x
P
o
o
l
(
2
×
2
)
0
(
N
o
n
e
,
5
4
,
5
4
,
6
4
)
C
o
n
v
2
D
C
o
n
v
2
D
(
1
2
8
,
3
×
3
,
R
e
L
U
)
7
3
,
8
5
6
(
N
o
n
e
,
5
2
,
5
2
,
1
2
8
)
M
a
x
P
o
o
l
i
n
g
2
D
M
a
x
P
o
o
l
(
2
×
2
)
0
(
N
o
n
e
,
2
6
,
2
6
,
1
2
8
)
F
l
a
t
t
e
n
F
l
a
t
t
e
n
0
(
N
o
n
e
,
8
6
,
5
2
8
)
D
e
n
se
(
f
c
)
D
e
n
se
(
6
4
,
R
e
L
U
)
5
,
5
3
7
,
8
5
6
(
N
o
n
e
,
6
4
)
D
r
o
p
o
u
t
D
r
o
p
o
u
t
(
0
.
5
)
0
(
N
o
n
e
,
6
4
)
O
u
t
p
u
t
l
a
y
e
r
D
e
n
se
(
1
,
si
g
mo
i
d
)
65
(
N
o
n
e
,
1
)
T
ab
le
3
.
L
ay
er
s
o
f
cu
s
to
m
O
v
O
d
ee
p
-
b
atch
n
o
r
m
al
izatio
n
(
BN
)
-
C
NN
Ty
p
e
D
e
scri
p
t
i
o
n
P
a
r
a
me
t
e
r
s
O
u
t
p
u
t
s
h
a
p
e
I
n
p
u
t
l
a
y
e
r
I
n
p
u
t
_
1
0
(
N
o
n
e
,
2
2
4
,
2
2
4
,
1
)
C
o
n
v
2
D
C
o
n
v
2
D
(
3
2
,
3
×
3
,
R
e
L
U
)
3
2
0
(
N
o
n
e
,
2
2
2
,
2
2
2
,
3
2
)
B
a
t
c
h
N
o
r
m
B
a
t
c
h
N
o
r
mal
i
z
a
t
i
o
n
(
3
2
)
1
2
8
(
N
o
n
e
,
2
2
2
,
2
2
2
,
3
2
)
M
a
x
P
o
o
l
i
n
g
2
D
M
a
x
P
o
o
l
(
2
×
2
)
0
(
N
o
n
e
,
1
1
1
,
1
1
1
,
3
2
)
C
o
n
v
2
D
C
o
n
v
2
D
(
6
4
,
3
×
3
,
R
e
L
U
)
1
8
,
4
9
6
(
N
o
n
e
,
1
0
9
,
1
0
9
,
6
4
)
B
a
t
c
h
N
o
r
m
B
a
t
c
h
N
o
r
mal
i
z
a
t
i
o
n
(
6
4
)
2
5
6
(
N
o
n
e
,
1
0
9
,
1
0
9
,
6
4
)
M
a
x
P
o
o
l
i
n
g
2
D
M
a
x
P
o
o
l
(
2
×
2
)
0
(
N
o
n
e
,
5
4
,
5
4
,
6
4
)
C
o
n
v
2
D
C
o
n
v
2
D
(
1
2
8
,
3
×
3
,
R
e
L
U
)
7
3
,
8
5
6
(
N
o
n
e
,
5
2
,
5
2
,
1
2
8
)
B
a
t
c
h
N
o
r
m
B
a
t
c
h
N
o
r
mal
i
z
a
t
i
o
n
(
1
2
8
)
5
1
2
(
N
o
n
e
,
5
2
,
5
2
,
1
2
8
)
M
a
x
P
o
o
l
i
n
g
2
D
M
a
x
P
o
o
l
(
2
×
2
)
0
(
N
o
n
e
,
2
6
,
2
6
,
1
2
8
)
C
o
n
v
2
D
C
o
n
v
2
D
(
2
5
6
,
3
×
3
,
R
e
L
U
)
(
e
x
t
r
a
c
o
n
v
o
l
u
t
i
o
n
a
l
b
l
o
c
k
)
2
9
5
,
1
6
8
(
N
o
n
e
,
2
4
,
2
4
,
2
5
6
)
M
a
x
P
o
o
l
i
n
g
2
D
M
a
x
P
o
o
l
(
2
×
2
)
0
(
N
o
n
e
,
1
2
,
1
2
,
2
5
6
)
F
l
a
t
t
e
n
F
l
a
t
t
e
n
0
(
N
o
n
e
,
3
6
,
8
6
4
)
D
e
n
se
(
f
c
1
)
D
e
n
se
(
1
2
8
,
R
e
L
U
)
4
,
7
1
8
,
7
2
0
(
N
o
n
e
,
1
2
8
)
B
a
t
c
h
N
o
r
m
B
a
t
c
h
N
o
r
mal
i
z
a
t
i
o
n
(
1
2
8
)
5
1
2
(
N
o
n
e
,
1
2
8
)
D
r
o
p
o
u
t
D
r
o
p
o
u
t
(
0
.
3
)
0
(
N
o
n
e
,
1
2
8
)
D
e
n
se
(
f
c
2
)
D
e
n
se
(
6
4
,
R
e
L
U
)
8
,
2
5
6
(
N
o
n
e
,
6
4
)
B
a
t
c
h
N
o
r
m
B
a
t
c
h
N
o
r
mal
i
z
a
t
i
o
n
(
6
4
)
2
5
6
(
N
o
n
e
,
6
4
)
D
r
o
p
o
u
t
D
r
o
p
o
u
t
(
0
.
3
)
0
(
N
o
n
e
,
6
4
)
D
e
n
se
(
f
c
3
)
D
e
n
se
(
3
2
,
R
e
L
U
)
2
,
0
8
0
(
N
o
n
e
,
3
2
)
O
u
t
p
u
t
l
a
y
e
r
D
e
n
se
(
1
,
si
g
mo
i
d
)
33
(
N
o
n
e
,
1
)
T
ab
le
4
.
L
ay
er
s
o
f
cu
s
to
m
O
v
R
C
NN
Ty
p
e
D
e
scri
p
t
i
o
n
P
a
r
a
ma
t
e
r
s
O
u
t
p
u
t
s
h
a
p
e
I
n
p
u
t
l
a
y
e
r
I
n
p
u
t
_
1
0
(
N
o
n
e
,
2
2
4
,
2
2
4
,
1
)
C
o
n
v
2
D
C
o
n
v
2
D
(
3
2
,
3
×
3)
3
2
0
(
N
o
n
e
,
2
2
2
,
2
2
2
,
3
2
)
M
a
x
P
o
o
l
i
n
g
2
D
M
a
x
P
o
o
l
(2
×
2)
0
(
N
o
n
e
,
1
1
1
,
1
1
1
,
3
2
)
C
o
n
v
2
D
C
o
n
v
2
D
(
6
4
,
3
×
3)
1
8
,
4
9
6
(
N
o
n
e
,
1
0
9
,
1
0
9
,
6
4
)
M
a
x
P
o
o
l
i
n
g
2
D
M
a
x
P
o
o
l
(2
×
2)
0
(
N
o
n
e
,
5
4
,
5
4
,
6
4
)
C
o
n
v
2
D
C
o
n
v
2
D
(
1
2
8
,
3
×
3)
7
3
,
8
5
6
(
N
o
n
e
,
5
2
,
5
2
,
1
2
8
)
M
a
x
P
o
o
l
i
n
g
2
D
M
a
x
P
o
o
l
(2
×
2)
0
(
N
o
n
e
,
2
6
,
2
6
,
1
2
8
)
F
l
a
t
t
e
n
F
l
a
t
t
e
n
0
(
N
o
n
e
,
8
6
,
5
2
8
)
D
e
n
se
(
f
c
1
)
D
e
n
se
(
5
1
2
)
4
4
,
3
0
2
,
8
4
8
(
N
o
n
e
,
5
1
2
)
D
r
o
p
o
u
t
D
r
o
p
o
u
t
(
0
.
5
)
0
(
N
o
n
e
,
5
1
2
)
O
u
t
p
u
t
l
a
y
e
r
D
e
n
se
(
1
,
si
g
mo
i
d
)
5
1
3
(
N
o
n
e
,
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
A
co
mp
a
r
a
tive
MRI
-
b
a
s
ed
s
tu
d
y
o
f R
esN
et
-
1
5
2
a
n
d
n
o
ve
l d
e
ep
lea
r
n
in
g
…
(
K
elvin
Leo
n
a
r
d
i Ko
h
s
a
s
ih
)
541
R
ep
r
o
d
u
cib
ilit
y
w
a
s
e
n
s
u
r
ed
b
y
s
ta
n
d
ar
d
izin
g
t
h
e
p
r
i
m
ar
y
tr
ain
i
n
g
co
n
f
ig
u
r
atio
n
s
.
A
d
r
o
p
o
u
t
r
ate
o
f
0
.
5
w
as
ap
p
lied
to
r
e
d
u
ce
o
v
er
f
itti
n
g
,
an
d
ea
r
l
y
s
to
p
p
in
g
w
i
th
a
p
atien
ce
o
f
ten
ep
o
ch
s
w
as
u
s
ed
to
p
r
o
m
o
te
s
tab
le
co
n
v
er
g
e
n
ce
.
T
r
ain
in
g
w
a
s
co
n
d
u
cted
u
s
i
n
g
t
h
e
A
d
a
m
o
p
ti
m
izer
w
it
h
a
lear
n
in
g
r
ate
o
f
0
.
0
0
0
1
an
d
a
b
atch
s
iz
e
o
f
3
2
,
s
etti
n
g
s
co
m
m
o
n
l
y
r
ep
o
r
ted
as
ef
f
ec
ti
v
e
f
o
r
MRI
-
b
ased
DL
tas
k
s
.
H
y
p
er
p
ar
am
eter
s
w
er
e
r
ef
in
ed
t
h
r
o
u
g
h
ex
p
lo
r
ato
r
y
g
r
id
s
ea
r
ch
es
t
h
at
v
ar
ied
lear
n
in
g
r
ate,
d
r
o
p
o
u
t
m
a
g
n
it
u
d
e,
an
d
b
atch
s
ize
to
en
s
u
r
e
s
tab
le
o
p
ti
m
izatio
n
ac
r
o
s
s
all
clas
s
if
icatio
n
s
tr
ateg
ie
s
.
2
.
5
.
M
o
del e
v
a
lua
t
i
o
n
T
h
is
s
tu
d
y
is
ev
a
lu
ated
u
s
in
g
a
s
et
o
f
c
o
n
f
u
s
io
n
-
m
atr
i
x
-
b
ased
m
e
tr
ics
th
at
ca
p
tu
r
e
co
m
p
le
m
en
tar
y
asp
ec
ts
o
f
c
lass
if
icatio
n
p
er
f
o
r
m
a
n
ce
.
T
h
ese
i
n
cl
u
d
e
ac
cu
r
ac
y
,
p
r
ec
is
io
n
(
p
o
s
iti
v
e
p
r
ed
ictiv
e
v
al
u
e,
P
P
V)
,
r
ec
all
(
s
en
s
it
iv
i
t
y
)
,
F1
-
s
co
r
e,
s
p
ec
if
icit
y
,
f
alse
p
o
s
iti
v
e
r
ate
(
FP
R
)
,
f
alse
n
e
g
ati
v
e
r
ate
(
F
NR
)
,
an
d
n
e
g
ati
v
e
p
r
ed
ictiv
e
v
alu
e
(
NP
V)
[
8
]
.
Si
n
ce
th
e
cla
s
s
d
is
tr
ib
u
tio
n
is
b
a
lan
ce
d
,
o
v
er
all
ac
cu
r
ac
y
i
s
co
n
s
id
er
ed
a
r
eliab
le
in
d
icato
r
o
f
p
er
f
o
r
m
an
ce
.
I
n
ad
d
itio
n
,
w
e
r
ep
o
r
t
p
r
ec
is
io
n
,
r
ec
all,
F1
-
s
co
r
e,
an
d
s
p
ec
if
icit
y
to
p
r
o
v
id
e
co
m
p
le
m
e
n
tar
y
p
er
s
p
ec
ti
v
es.
I
n
th
e
m
u
lticla
s
s
s
etti
n
g
,
p
er
-
c
lass
p
r
ec
is
io
n
,
r
ec
all,
a
n
d
F1
-
s
co
r
e
ar
e
co
m
p
u
ted
b
y
tr
ea
ti
n
g
ea
ch
cla
s
s
a
g
ain
s
t
th
e
r
est
a
n
d
th
e
n
s
u
m
m
ar
i
ze
d
u
s
in
g
m
ac
r
o
,
w
ei
g
h
ted
,
o
r
m
icr
o
av
er
a
g
es,
co
n
s
is
ten
t
w
it
h
s
ta
n
d
ar
d
class
i
f
icatio
n
r
ep
o
r
ts
an
d
co
n
f
u
s
io
n
-
m
atr
ix
a
n
al
y
s
e
s
[
4
9
]
,
[
5
0
]
.
T
h
e
f
o
r
m
u
las
u
s
ed
t
o
co
m
p
u
te
t
h
es
e
m
etr
ics ar
e
as
f
o
llo
w
s
:
=
+
+
+
+
(
1
)
(
)
=
+
(
2
)
/
=
+
(
3
)
1
−
=
2
∗
∗
+
(
4
)
=
+
(
5
)
=
+
(
6
)
=
+
(
7
)
=
+
(
8
)
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
e
d
ataset
w
as
f
ir
s
t
b
ala
n
ce
d
th
r
o
u
g
h
r
an
d
o
m
u
n
d
er
s
a
m
p
li
n
g
,
r
es
u
lti
n
g
in
6
,
4
6
4
i
m
ag
e
s
p
er
class
an
d
a
to
tal
o
f
2
5
,
8
5
6
s
am
p
les.
I
t
w
a
s
th
e
n
s
p
li
t
in
to
8
0
%
f
o
r
tr
ain
i
n
g
a
n
d
2
0
%
f
o
r
test
i
n
g
.
A
ll
i
m
a
g
es
w
er
e
p
r
o
ce
s
s
ed
u
s
in
g
a
s
ta
n
d
ar
d
ized
p
ip
elin
e
th
at
co
n
s
is
ted
o
f
g
r
ay
s
ca
le
co
n
v
er
s
io
n
,
r
esiz
in
g
to
2
2
4
×
2
2
4
p
ix
els,
C
L
AHE
-
b
ased
co
n
tr
ast
en
h
a
n
ce
m
en
t,
b
ilater
al
f
ilter
i
n
g
,
a
n
d
m
i
n
–
m
a
x
n
o
r
m
aliza
tio
n
to
t
h
e
[
0
,
1
]
r
an
g
e.
T
h
is
p
r
o
ce
d
u
r
e
p
r
o
d
u
ce
d
clea
r
er
an
d
m
o
r
e
co
n
s
is
ten
t
M
R
I
i
m
a
g
e
s
,
w
it
h
en
h
a
n
ce
d
v
is
ib
ilit
y
o
f
s
u
lci
an
d
v
e
n
tr
icles,
r
ed
u
ce
d
n
o
is
e,
an
d
s
tan
d
ar
d
ized
d
im
e
n
s
io
n
s
.
R
ep
r
ese
n
tati
v
e
ex
a
m
p
les
o
f
t
h
e
p
r
ep
r
o
ce
s
s
ed
d
ata
ar
e
s
h
o
w
n
in
Fig
u
r
e
4
.
T
h
e
tr
ain
in
g
h
is
to
r
y
o
f
R
es
Ne
t
-
1
5
2
is
p
r
esen
ted
i
n
Fig
u
r
e
5
.
A
l
th
o
u
g
h
t
h
e
m
o
d
el
w
as
al
lo
ca
ted
2
0
0
0
ep
o
ch
s
,
tr
ain
i
n
g
co
n
v
er
g
ed
ea
r
l
y
at
ep
o
ch
8
1
,
as
th
e
ea
r
l
y
s
to
p
p
in
g
cr
iter
io
n
w
a
s
m
et
.
T
h
e
m
o
d
el
r
ea
ch
ed
7
2
%
tr
ain
in
g
ac
cu
r
ac
y
at
t
h
e
f
i
n
al
ep
o
ch
,
w
h
ile
v
ali
d
atio
n
ac
cu
r
ac
y
f
o
llo
w
ed
a
s
i
m
ilar
tr
en
d
w
it
h
m
o
d
er
ate
f
l
u
ctu
a
tio
n
s
,
r
e
f
lecti
n
g
s
en
s
iti
v
it
y
to
d
ata
v
ar
iab
ilit
y
b
u
t
n
o
s
ev
er
e
o
v
er
f
itti
n
g
.
T
h
e
co
r
r
es
p
o
n
d
in
g
lo
s
s
cu
r
v
e
s
s
h
o
w
a
co
n
s
is
te
n
t
d
ec
r
ea
s
e
i
n
tr
ain
i
n
g
lo
s
s
,
w
i
th
v
alid
ati
o
n
lo
s
s
ex
h
ib
iti
n
g
h
i
g
h
er
v
a
r
ian
ce
b
u
t
o
v
er
all
co
n
v
er
g
e
n
ce
w
it
h
tr
ai
n
in
g
lo
s
s
.
T
h
e
s
ec
o
n
d
m
o
d
el
tr
ain
ed
w
a
s
th
e
c
u
s
to
m
C
NN,
w
h
ic
h
e
m
p
lo
y
ed
a
m
u
lt
iclas
s
class
i
f
icatio
n
s
tr
ateg
y
.
T
h
is
m
o
d
el
ac
h
iev
ed
th
e
b
est
p
er
f
o
r
m
a
n
ce
w
it
h
a
t
est
ac
cu
r
ac
y
o
f
9
0
%,
co
n
f
ir
m
i
n
g
i
ts
e
f
f
ec
tiv
e
n
es
s
in
h
an
d
li
n
g
t
h
e
f
o
u
r
-
clas
s
A
l
zh
ei
m
er
’
s
clas
s
i
f
icatio
n
tas
k
.
T
h
e
tr
ain
in
g
h
i
s
to
r
y
s
h
o
w
n
i
n
Fig
u
r
e
6
i
n
d
icate
s
th
at
b
o
th
tr
ai
n
in
g
a
n
d
v
alid
atio
n
ac
cu
r
ac
y
s
tab
ilized
at
ar
o
u
n
d
9
0
%,
w
h
ile
t
h
e
lo
s
s
cu
r
v
es
co
n
v
er
g
ed
s
m
o
o
th
l
y
w
i
th
o
u
t si
g
n
s
o
f
o
v
e
r
f
itti
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
.
2
,
A
p
r
il
20
26
:
5
3
6
-
548
542
Fig
u
r
e
4
.
Sa
m
p
le
i
m
ag
e
s
af
ter
p
r
ep
r
o
ce
s
s
in
g
s
tep
s
Fig
u
r
e
5
.
T
r
ain
in
g
ac
cu
r
ac
y
a
n
d
lo
s
s
cu
r
v
es
o
f
R
esNe
t
-
152
Fig
u
r
e
6
.
T
r
ain
in
g
ac
cu
r
ac
y
a
n
d
lo
s
s
cu
r
v
es o
f
c
u
s
to
m
C
NN
m
u
l
ticlas
s
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
A
co
mp
a
r
a
tive
MRI
-
b
a
s
ed
s
tu
d
y
o
f R
esN
et
-
1
5
2
a
n
d
n
o
ve
l d
e
ep
lea
r
n
in
g
…
(
K
elvin
Leo
n
a
r
d
i Ko
h
s
a
s
ih
)
543
T
h
e
th
ir
d
m
o
d
el
tr
ain
ed
w
a
s
t
h
e
c
u
s
to
m
C
NN
w
it
h
an
Ov
O
class
i
f
icatio
n
s
tr
ate
g
y
.
I
n
t
h
i
s
ap
p
r
o
ac
h
,
s
ix
i
n
d
ep
en
d
en
t
s
b
in
ar
y
C
NN
s
w
er
e
tr
ai
n
ed
,
ea
ch
f
o
cu
s
i
n
g
o
n
a
s
p
ec
if
ic
p
air
o
f
class
e
s
.
T
h
e
tr
ain
i
n
g
h
i
s
to
r
ies
o
f
th
e
s
i
x
class
i
f
ier
s
ar
e
s
h
o
w
n
in
Fi
g
u
r
e
s
7
(
a
)
–
(
f
)
:
tr
ain
in
g
ac
cu
r
ac
y
an
d
lo
s
s
cu
r
v
e
s
o
f
cu
s
to
m
C
NN
O
v
O.
Mo
s
t
class
i
f
ier
s
ac
h
iev
ed
s
ta
b
le
co
n
v
er
g
e
n
ce
,
w
i
th
s
o
m
e
p
air
s
,
s
u
ch
as
ND
v
s
.
Mo
d
D
an
d
MD
v
s
.
Mo
d
D,
r
ea
ch
in
g
v
alid
atio
n
ac
c
u
r
ac
ies
ab
o
v
e
9
0
%.
Ho
w
ev
er
,
cla
s
s
i
f
ier
s
in
v
o
lv
in
g
ea
r
l
y
-
s
tag
e
ca
te
g
o
r
ies,
s
u
ch
a
s
ND
v
s
VM
D
o
r
VM
D
v
s
MD
,
e
x
h
ib
ited
lo
w
er
a
n
d
m
o
r
e
f
l
u
ct
u
atin
g
v
al
id
atio
n
ac
cu
r
ac
y
,
r
ef
l
ec
tin
g
th
e
i
n
h
er
en
t
d
if
f
ic
u
lt
y
i
n
d
is
t
in
g
u
is
h
i
n
g
s
u
b
tle
ch
a
n
g
es
at
t
h
e
s
e
s
ta
g
es
.
T
h
e
class
if
icatio
n
r
ep
o
r
t
f
u
r
th
er
s
u
p
p
o
r
ts
th
es
e
f
i
n
d
in
g
s
,
w
i
th
an
o
v
er
all
ac
c
u
r
ac
y
o
f
8
5
%.
ND
r
ea
c
h
ed
a
n
F1
-
s
co
r
e
o
f
9
3
%,
VM
D
8
9
%,
MD
8
2
%,
an
d
Mo
d
D
9
9
%.
T
h
e
m
ac
r
o
-
av
er
ag
e
F1
w
as
7
3
%,
in
d
icatin
g
v
ar
iab
ilit
y
ac
r
o
s
s
class
es,
wh
ile
th
e
w
eig
h
ted
-
av
er
ag
e
F1
r
ea
ch
ed
9
1
%,
co
n
f
ir
m
i
n
g
t
h
e
s
tr
o
n
g
o
v
er
all
p
er
f
o
r
m
an
ce
o
f
t
h
e
O
v
O
ap
p
r
o
ac
h
.
(
a)
(
b
)
(
c)
(
d
)
(
e)
(
f
)
Fig
u
r
e
7
.
T
r
ain
in
g
ac
cu
r
ac
y
a
n
d
lo
s
s
cu
r
v
es o
f
t
h
e
s
ix
c
u
s
to
m
C
N
N
Ov
O
b
i
n
ar
y
cla
s
s
i
f
ier
s
; (
a)
ND
v
s
VM
D,
(
b
)
ND
v
s
Mo
d
D,
(
c)
VM
D
v
s
Mo
d
D,
(
d
)
ND
v
s
MD
,
(
e)
VM
D
v
s
MD
,
a
n
d
(
f
)
MD
v
s
Mo
d
D
T
h
e
f
in
al
m
o
d
el
d
ev
elo
p
ed
w
a
s
th
e
c
u
s
to
m
C
NN
u
s
in
g
an
Ov
R
s
tr
ate
g
y
,
w
h
er
e
f
o
u
r
b
in
ar
y
class
i
f
ier
s
w
er
e
tr
ai
n
ed
in
d
ep
en
d
en
tl
y
,
ea
ch
i
s
o
lati
n
g
o
n
e
tar
g
et
clas
s
f
r
o
m
th
e
r
e
m
ai
n
i
n
g
ca
te
g
o
r
ies.
T
h
e
tr
ain
i
n
g
b
eh
a
v
io
r
s
o
f
t
h
ese
m
o
d
els
ar
e
illu
s
tr
ated
in
F
ig
u
r
e
s
8
(
a
)
–
(
d
)
.
Fig
u
r
es
8
(
a)
an
d
(
b
)
,
r
e
p
r
esen
tin
g
t
h
e
ND
v
s
r
est
a
n
d
Mo
d
D
v
s
r
est
class
i
f
ier
s
,
s
h
o
w
th
e
m
o
s
t
s
tab
le
co
n
v
er
g
e
n
ce
,
w
ith
tr
ai
n
in
g
a
n
d
v
alid
atio
n
ac
cu
r
ac
ies
co
n
s
is
ten
tl
y
e
x
ce
e
d
in
g
8
5
%.
I
n
co
n
tr
ast,
F
ig
u
r
e
8
(
c)
f
o
r
th
e
MD
v
s
r
est
cla
s
s
i
f
ier
ex
h
ib
it
s
g
r
ea
ter
v
ar
iab
ilit
y
,
w
it
h
n
o
ticea
b
le
f
l
u
ctu
a
tio
n
s
in
v
al
id
atio
n
ac
cu
r
ac
y
,
i
n
d
icati
n
g
th
e
c
h
all
en
g
e
s
i
n
co
r
r
ec
tl
y
s
ep
ar
atin
g
M
Ds
a
m
p
les
f
r
o
m
n
eig
h
b
o
r
in
g
clas
s
es.
Me
an
wh
ile,
Fi
g
u
r
e
8
(
d
)
,
c
o
r
r
esp
o
n
d
i
n
g
to
t
h
e
VM
D
v
s
R
est
clas
s
i
f
ier
,
d
em
o
n
s
tr
ate
s
m
o
d
er
ate
p
er
f
o
r
m
a
n
ce
an
d
r
ev
ea
ls
s
en
s
i
tiv
it
y
to
class
i
m
b
ala
n
ce
an
d
o
v
er
lap
p
in
g
ea
r
l
y
-
s
ta
g
e
f
ea
tu
r
es.
T
h
ese
o
b
s
er
v
atio
n
s
h
i
g
h
lig
h
t
t
h
e
v
ar
y
i
n
g
co
m
p
lex
i
t
y
ac
r
o
s
s
d
iag
n
o
s
tic
b
o
u
n
d
ar
ies
w
it
h
in
t
h
e
O
v
R
f
r
a
m
e
w
o
r
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
.
2
,
A
p
r
il
20
26
:
5
3
6
-
548
544
(
a)
(
b
)
(
c)
(
d
)
Fig
u
r
e
8
.
T
r
ain
in
g
ac
cu
r
ac
y
a
n
d
lo
s
s
cu
r
v
es o
f
t
h
e
f
o
u
r
cu
s
t
o
m
C
NN
O
v
R
class
if
ier
s
; (
a)
ND
v
s
r
est,
(
b
)
Mo
d
D
v
s
r
est,
(
c)
MD
v
s
r
est,
an
d
(
d
)
VM
D
v
s
r
est
A
lt
h
o
u
g
h
t
h
e
c
u
s
to
m
m
u
lticl
ass
C
NN
s
h
o
w
ed
t
h
e
s
tr
o
n
g
e
s
t
an
d
m
o
s
t
co
n
s
i
s
te
n
t
p
er
f
o
r
m
an
ce
,
th
e
co
n
f
u
s
io
n
m
atr
ices
r
e
v
ea
l
w
h
y
t
h
e
O
v
O
a
n
d
O
v
R
m
o
d
els
p
er
f
o
r
m
ed
p
o
o
r
ly
o
n
ea
r
l
y
-
s
tag
e
cla
s
s
es.
B
o
th
ap
p
r
o
ac
h
es
f
r
eq
u
en
tl
y
m
i
s
cla
s
s
i
f
ied
VM
D
an
d
MD
,
w
h
ic
h
r
ef
lects
t
h
e
s
u
b
tle
m
o
r
p
h
o
lo
g
ical
d
if
f
er
e
n
ce
s
b
et
w
ee
n
th
e
s
e
ad
j
ac
en
t
s
tag
e
s
an
d
th
e
ir
o
v
er
lap
w
it
h
n
o
r
m
al
a
g
in
g
.
T
h
e
Ov
R
m
o
d
els
ten
d
ed
to
f
av
o
r
th
e
d
o
m
i
n
a
n
t
“
R
e
s
t”
class
,
r
ed
u
ci
n
g
s
en
s
iti
v
it
y
to
ea
r
l
y
-
s
ta
g
e
ca
s
es,
w
h
i
le
th
e
O
v
O
m
o
d
el
s
p
r
o
d
u
ce
d
in
co
n
s
i
s
ten
t
p
air
w
is
e
o
u
tp
u
ts
th
at
ac
cu
m
u
lated
d
u
r
in
g
v
o
ti
n
g
an
d
in
cr
e
ased
th
e
lik
eli
h
o
o
d
o
f
e
r
r
o
r
p
r
o
p
ag
atio
n
.
T
h
ese
o
b
s
er
v
atio
n
s
i
n
d
icate
th
at
d
ec
o
m
p
o
s
i
tio
n
-
b
ased
s
tr
ateg
ie
s
ar
e
m
o
r
e
v
u
ln
er
ab
le
to
b
o
u
n
d
ar
y
a
m
b
i
g
u
it
y
t
h
an
a
u
n
i
f
ied
m
u
lticla
s
s
ap
p
r
o
ac
h
.
A
p
p
l
y
in
g
s
tati
s
tical
s
i
g
n
if
ic
an
ce
te
s
ts
s
u
ch
a
s
Mc
Ne
m
a
r
’
s
te
s
t
o
r
p
air
ed
b
o
o
ts
tr
ap
c
o
m
p
ar
is
o
n
s
w
o
u
l
d
h
elp
co
n
f
ir
m
w
h
et
h
er
th
e
p
er
f
o
r
m
a
n
ce
d
if
f
er
e
n
ce
s
b
et
w
ee
n
m
o
d
els
ar
e
s
tatis
t
icall
y
m
ea
n
i
n
g
f
u
l.
A
co
n
s
o
lid
ated
co
m
p
ar
is
o
n
ac
r
o
s
s
all
m
o
d
els
is
p
r
esen
ted
to
ev
al
u
ate
o
v
er
all
d
iag
n
o
s
tic
p
er
f
o
r
m
a
n
ce
an
d
er
r
o
r
ch
ar
ac
ter
is
tics
.
T
ab
le
5
s
u
m
m
ar
ize
s
th
e
r
esu
l
ts
an
d
h
ig
h
li
g
h
ts
t
h
e
d
if
f
er
e
n
ce
s
a
m
o
n
g
th
e
m
u
lt
iclas
s
,
Ov
O,
O
v
R
,
a
n
d
R
esNet
-
1
5
2
ap
p
r
o
ac
h
es
.
T
ab
le
5
.
R
esu
lts
p
er
f
o
r
m
an
ce
ev
alu
a
tio
n
M
o
d
e
l
A
p
p
r
o
a
c
h
A
c
c
u
r
a
c
y
(
%)
P
r
e
c
i
si
o
n
(
%)
R
e
c
a
l
l
(
%)
F1
-
S
c
o
r
e
(
%)
S
p
e
c
i
f
i
c
i
t
y
(
%)
F
P
R
(
%)
F
N
R
(
%)
N
P
V
(
%)
R
e
sN
e
t
-
1
5
2
M
u
l
t
i
c
l
a
ss
7
2
72
72
71
91
9
28
91
C
u
s
t
o
m C
N
N
M
u
l
t
i
c
l
a
ss
90
90
90
90
97
3
10
97
C
u
s
t
o
m C
N
N
O
v
O
85
78
85
90
97
3
15
96
C
u
s
t
o
m C
N
N
O
v
R
81
71
81
84
95
5
19
95
T
h
is
s
tu
d
y
p
r
ese
n
ted
a
s
ta
g
e
-
a
w
ar
e
DL
p
ip
elin
e
f
o
r
AD
cla
s
s
if
ica
tio
n
o
n
s
tr
u
ct
u
r
al
MRI,
c
o
m
p
ar
i
n
g
a
tr
an
s
f
er
-
lear
n
in
g
b
aseli
n
e
R
esNet
-
1
5
2
w
it
h
t
h
r
ee
p
r
o
p
o
s
ed
class
i
f
ier
s
:
c
u
s
to
m
C
NN
m
u
lticlas
s
,
c
u
s
to
m
C
NN
Ov
O,
an
d
c
u
s
to
m
C
NN
Ov
R
.
On
th
e
test
s
et,
th
e
cu
s
to
m
m
u
lticla
s
s
C
NN
d
eliv
er
e
d
th
e
s
tr
o
n
g
est
an
d
m
o
s
t
b
ala
n
ce
d
p
er
f
o
r
m
an
ce
,
a
ch
iev
in
g
9
0
%
ac
cu
r
ac
y
w
it
h
co
n
s
is
ten
t
p
r
ec
is
io
n
,
r
ec
all,
an
d
F1
s
co
r
es
ac
r
o
s
s
th
e
n
o
n
-
d
e
m
e
n
ted
,
v
er
y
m
ild
,
m
i
ld
,
an
d
m
o
d
er
ate
s
tag
es.
T
h
is
r
esu
lt
s
u
r
p
ass
ed
th
e
p
er
f
o
r
m
a
n
ce
o
f
Ov
O
at
8
5
%,
Ov
R
at
8
1
%,
an
d
th
e
b
aselin
e
R
e
s
Net
-
1
5
2
at
7
2
%.
Mo
d
el
r
eliab
ilit
y
w
as
f
u
r
t
h
er
s
u
p
p
o
r
ted
b
y
s
tab
le
v
alid
atio
n
tr
en
d
s
,
w
i
th
f
l
u
ct
u
a
tio
n
s
o
f
ap
p
r
o
x
i
m
ate
l
y
±
2
%
f
o
r
th
e
c
u
s
to
m
m
u
ltic
lass
C
N
N
,
±
3
%
f
o
r
th
e
Ov
O
m
o
d
el,
±
4
%
f
o
r
th
e
Ov
R
m
o
d
el,
an
d
±
5
%
f
o
r
R
esNet
-
1
5
2
.
C
o
llectiv
el
y
,
th
e
s
e
f
i
n
d
in
g
s
i
n
d
icate
th
at
th
e
m
u
lticla
s
s
f
o
r
m
u
lat
io
n
is
p
ar
ticu
lar
l
y
e
f
f
ec
tiv
e
f
o
r
ea
r
ly
-
s
ta
g
e
Alzh
ei
m
er
’
s
d
etec
tio
n
,
w
h
er
e
s
u
b
tle
s
tr
u
ct
u
r
al
d
if
f
er
e
n
ce
s
o
f
ten
r
ese
m
b
le
n
o
r
m
al
ag
in
g
an
d
i
n
tr
o
d
u
ce
s
u
b
j
e
ctiv
it
y
i
n
clin
ical
ass
e
s
s
m
e
n
t.
W
h
en
i
n
ter
p
r
eted
in
th
e
co
n
t
ex
t
o
f
ex
is
ti
n
g
MRI
-
b
ased
Alzh
ei
m
er
’
s
class
if
icatio
n
r
ese
ar
ch
,
th
es
e
r
esu
lt
s
s
h
o
w
a
clea
r
ad
v
a
n
c
e
m
en
t.
P
r
io
r
s
tu
d
ies
u
s
i
n
g
YOL
O
v
5
,
VGG1
6
,
De
n
s
eNe
t
-
1
6
9
,
o
r
R
esNet
-
50
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
A
co
mp
a
r
a
tive
MRI
-
b
a
s
ed
s
tu
d
y
o
f R
esN
et
-
1
5
2
a
n
d
n
o
ve
l d
e
ep
lea
r
n
in
g
…
(
K
elvin
Leo
n
a
r
d
i Ko
h
s
a
s
ih
)
545
t
y
p
icall
y
r
ep
o
r
t
ac
cu
r
ac
ies
r
a
n
g
in
g
f
r
o
m
6
1
to
8
2
%
[
1
9
]
,
[
2
2
]
,
[
2
4
]
.
an
d
m
u
lti
-
s
ta
g
e
m
o
d
els
o
f
te
n
s
tr
u
g
g
le
to
s
ep
ar
ate
ad
j
ac
en
t
s
tag
es
alo
n
g
t
h
e
cli
n
ical
d
e
m
e
n
tia
r
atin
g
s
ca
le
[
5
1
]
.
I
n
co
n
tr
ast,
th
e
p
r
o
p
o
s
ed
cu
s
to
m
m
u
lticla
s
s
C
NN
d
e
m
o
n
s
tr
ate
s
s
tr
o
n
g
er
s
tag
e
-
s
p
ec
if
ic
d
is
c
r
i
m
i
n
atio
n
,
p
r
o
v
id
in
g
e
m
p
ir
ic
al
ev
id
en
ce
t
h
at
a
tailo
r
ed
ar
ch
itectu
r
e
ca
n
le
v
er
ag
e
d
is
ea
s
e
-
r
elate
d
s
tr
u
ct
u
r
al
p
atter
n
s
m
o
r
e
ef
f
ec
ti
v
el
y
th
a
n
co
n
v
en
tio
n
al
tr
an
s
f
er
-
lear
n
i
n
g
b
ac
k
b
o
n
e
s
.
T
h
is
i
m
p
r
o
v
e
m
e
n
t
in
s
ta
g
e
-
a
w
ar
e
p
er
f
o
r
m
an
ce
r
ep
r
esen
t
s
a
k
e
y
n
o
v
el
t
y
o
f
th
e
p
r
esen
t
w
o
r
k
,
as
it
s
h
o
w
s
th
at
a
d
ed
icate
d
m
u
l
ticlas
s
ar
c
h
itect
u
r
e
ca
n
s
i
m
u
ltan
eo
u
s
l
y
en
h
a
n
ce
ac
c
u
r
ac
y
,
r
ed
u
ce
in
ter
-
c
lass
co
n
f
u
s
io
n
,
a
n
d
i
m
p
r
o
v
e
ea
r
l
y
-
s
tag
e
d
etec
ti
o
n
s
en
s
iti
v
it
y
.
T
h
e
clin
ical
r
el
ev
a
n
ce
o
f
th
e
s
e
g
ai
n
s
is
co
n
s
id
er
ab
le.
E
ar
l
y
id
en
t
if
icatio
n
o
f
v
er
y
m
ild
an
d
MD
ca
teg
o
r
ies
is
d
i
f
f
icu
lt
ev
e
n
f
o
r
ex
p
er
ien
ce
d
r
ad
io
lo
g
is
ts
,
y
e
t
th
e
p
r
o
p
o
s
ed
m
o
d
el
ca
p
tu
r
e
d
th
ese
b
o
r
d
er
lin
e
d
is
tin
ct
io
n
s
w
i
th
g
r
ea
ter
r
eliab
ilit
y
.
T
h
is
ca
p
ab
ilit
y
ca
n
s
u
p
p
o
r
t r
ad
io
lo
g
is
ts
b
y
r
ed
u
cin
g
i
n
ter
-
r
ater
v
ar
iab
ilit
y
,
f
la
g
g
i
n
g
s
u
b
tle
m
o
r
p
h
o
lo
g
ica
l
ch
an
g
es
f
o
r
r
e
-
ev
al
u
atio
n
,
an
d
p
r
io
r
itizin
g
p
atien
ts
f
o
r
ad
d
itio
n
al
b
io
m
ar
k
er
test
i
n
g
o
r
ti
m
el
y
i
n
itiat
io
n
o
f
d
is
ea
s
e
-
m
o
d
if
y
i
n
g
t
h
er
ap
y
[
4
1
]
.
No
n
eth
eless
,
s
e
v
er
al
ch
a
llen
g
es
r
e
m
a
in
f
o
r
r
ea
l
-
w
o
r
ld
d
ep
lo
y
m
en
t.
Mo
d
els
tr
ain
ed
o
n
a
s
i
n
g
le
d
ataset
m
a
y
f
ail
to
g
e
n
er
alize
a
cr
o
s
s
s
ca
n
n
er
s
o
r
p
o
p
u
latio
n
s
d
u
e
to
d
o
m
ain
s
h
i
f
t,
an
d
r
elian
ce
s
o
lel
y
o
n
s
tr
u
ctu
r
al
MRI
m
a
y
o
v
er
lo
o
k
ea
r
l
y
f
u
n
ctio
n
al
o
r
m
etab
o
lic
ab
n
o
r
m
ali
ties
d
ete
ctab
le
b
y
P
E
T
o
r
f
u
n
ctio
n
a
l
m
ag
n
etic
r
e
s
o
n
a
n
ce
i
m
a
g
i
n
g
(
f
MRI
)
[
5
2
]
,
[
5
3
]
.
Fu
t
u
r
e
w
o
r
k
s
h
o
u
ld
th
er
e
f
o
r
e
ex
a
m
in
e
cr
o
s
s
-
d
ataset
tr
ai
n
i
n
g
,
d
o
m
ai
n
ad
ap
tatio
n
,
an
d
m
u
lt
i
m
o
d
al
f
u
s
io
n
,
an
d
s
h
o
u
ld
ex
p
lo
r
e
h
y
b
r
id
o
r
en
s
em
b
le
ar
ch
itect
u
r
es
co
m
b
in
ed
w
it
h
ex
p
lai
n
ab
le
A
I
to
f
u
r
t
h
er
im
p
r
o
v
e
r
o
b
u
s
t
n
es
s
an
d
clin
ical
s
af
e
t
y
.
4.
CO
NCLU
SI
O
N
T
h
is
s
tu
d
y
i
n
tr
o
d
u
ce
d
a
s
tag
e
-
a
w
ar
e
DL
f
r
a
m
e
w
o
r
k
f
o
r
AD
cla
s
s
i
f
icatio
n
u
s
i
n
g
s
tr
u
c
tu
r
al
MRI,
d
em
o
n
s
tr
ati
n
g
th
at
th
e
p
r
o
p
o
s
ed
cu
s
to
m
m
u
lticla
s
s
C
NN
ac
h
iev
ed
9
0
%
ac
cu
r
ac
y
an
d
o
u
tp
er
f
o
r
m
ed
b
o
th
th
e
O
v
O
,
O
v
R
m
o
d
el
s
,
an
d
t
h
e
R
esNet
-
1
5
2
tr
an
s
f
er
-
lear
n
i
n
g
b
aselin
e.
T
h
ese
f
in
d
i
n
g
s
h
i
g
h
li
g
h
t
t
h
e
m
o
d
el
’
s
ab
ilit
y
to
ca
p
tu
r
e
s
u
b
tle
m
o
r
p
h
o
lo
g
ical
d
if
f
er
e
n
ce
s
ac
r
o
s
s
A
lz
h
ei
m
er
’
s
s
ta
g
es
a
n
d
its
cli
n
ical
r
elev
a
n
ce
f
o
r
i
m
p
r
o
v
i
n
g
ea
r
l
y
-
s
ta
g
e
d
etec
tio
n
,
w
h
er
e
s
y
m
p
to
m
s
o
f
te
n
m
i
m
ic
n
o
r
m
al
a
g
in
g
.
No
n
e
th
ele
s
s
,
s
ev
er
al
li
m
itat
io
n
s
r
e
m
ain
,
in
cl
u
d
i
n
g
r
elia
n
ce
o
n
a
s
in
g
le
MRI
d
ataset,
p
o
te
n
tial
d
o
m
ai
n
s
h
if
t a
cr
o
s
s
s
ca
n
n
er
s
o
r
p
o
p
u
latio
n
s
,
an
d
th
e
e
x
clu
s
i
v
e
u
s
e
o
f
s
tr
u
ct
u
r
al
MRI,
w
h
ich
m
a
y
li
m
it
s
e
n
s
it
iv
i
t
y
to
ea
r
l
y
p
ath
o
lo
g
ical
ch
a
n
g
e
s
.
Fu
tu
r
e
r
esear
ch
s
h
o
u
ld
in
co
r
p
o
r
ate
m
u
lti
m
o
d
al
i
m
a
g
i
n
g
s
u
c
h
as
P
E
T
o
r
f
MRI,
in
teg
r
ate
clin
ical
an
d
co
g
n
it
iv
e
b
io
m
ar
k
er
s
,
p
er
f
o
r
m
cr
o
s
s
-
d
at
aset
o
r
m
u
ltice
n
ter
v
alid
atio
n
,
an
d
ex
p
lo
r
e
h
y
b
r
id
o
r
en
s
e
m
b
le
ar
ch
itectu
r
e
s
to
en
h
a
n
ce
r
o
b
u
s
tn
e
s
s
.
W
ith
f
u
r
th
er
r
ef
in
e
m
e
n
t
an
d
ex
ter
n
a
l
v
alid
atio
n
,
th
e
p
r
o
p
o
s
ed
f
r
a
m
e
w
o
r
k
h
a
s
th
e
p
o
ten
tial
to
s
er
v
e
as
a
r
eliab
le
d
ec
is
io
n
-
s
u
p
p
o
r
t
to
o
l
th
at
s
tr
en
g
t
h
e
n
s
d
iag
n
o
s
tic
co
n
s
i
s
te
n
c
y
an
d
f
ac
il
itates
ea
r
lier
in
ter
v
e
n
tio
n
i
n
cli
n
ical
p
r
ac
tice.
ACK
NO
WL
E
D
G
M
E
NT
S
T
h
e
au
th
o
r
s
w
o
u
ld
lik
e
to
ex
p
r
ess
th
eir
g
r
atit
u
d
e
to
ST
MI
K
T
I
ME
f
o
r
in
s
ti
tu
t
io
n
al
s
u
p
p
o
r
t
an
d
th
e
r
esear
ch
f
ac
ilit
ie
s
p
r
o
v
id
ed
d
u
r
in
g
th
e
co
m
p
letio
n
o
f
t
h
i
s
s
t
u
d
y
.
T
h
is
w
o
r
k
w
as
s
u
p
p
o
r
ted
b
y
th
e
Dir
ec
to
r
ate
o
f
R
esear
ch
,
T
ec
h
n
o
lo
g
y
,
an
d
C
o
m
m
u
n
it
y
Ser
v
ice
(
DR
T
P
M)
,
Min
is
tr
y
o
f
E
d
u
ca
tio
n
,
C
u
lt
u
r
e,
R
esear
ch
,
an
d
T
ec
h
n
o
lo
g
y
o
f
I
n
d
o
n
e
s
ia
u
n
d
er
R
esear
ch
Gr
an
t
No
.
1
2
2
/C
3
/DT
.
0
5
.
0
0
/P
L
/2
0
2
5
an
d
Su
b
co
n
tr
ac
t
No
.
6
9
/SP
K/L
L
1
/
AL
.
0
4
.
0
3
/P
L
/2
0
2
5
.
F
UNDIN
G
I
NF
O
RM
AT
I
O
N
T
h
is
r
esear
ch
r
ec
eiv
ed
f
i
n
an
c
ial
s
u
p
p
o
r
t
f
r
o
m
th
e
M
in
i
s
tr
y
o
f
E
d
u
ca
tio
n
,
C
u
l
tu
r
e,
R
e
s
ea
r
ch
,
an
d
T
ec
h
n
o
lo
g
y
o
f
I
n
d
o
n
esia.
AUTHO
R
CO
NT
RIB
UT
I
O
NS ST
A
T
E
M
E
NT
T
h
is
j
o
u
r
n
al
u
s
e
s
th
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT
)
to
r
ec
o
g
n
ize
in
d
i
v
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
t
h
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
lla
b
o
r
atio
n
.
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
Kelv
i
n
L
eo
n
ar
d
i
Ko
h
s
as
ih
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
Octa
r
a
P
r
ib
a
d
i
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
An
d
y
✓
✓
✓
✓
✓
✓
✓
✓
✓
Dan
iel
S
m
it
h
S
u
n
ar
io
✓
✓
✓
✓
✓
✓
✓
✓
✓
Evaluation Warning : The document was created with Spire.PDF for Python.