I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
pu
t
er
E
ng
ineering
(
I
J
E
CE
)
Vo
l.
14
,
No
.
6
,
Dec
em
b
er
20
24
,
p
p
.
6
6
2
8
~
6
6
4
1
I
SS
N:
2088
-
8
7
0
8
,
DOI
:
1
0
.
1
1
5
9
1
/ijece.
v
14
i
6
.
pp
6
6
2
8
-
6
6
4
1
6628
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
Im
plementa
tion o
f
inno
v
a
tive a
ppr
o
a
ch f
o
r det
ecting
brain
tumo
rs
in
ma
g
net
ic reson
a
nce ima
g
ing
using
NeuroF
usio
nNet
mo
del
Arpit
ha
K
o
t
t
e
1
,
Sy
ed
Sh
a
bb
ee
r
Ahm
a
d
2
1
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
O
sma
n
i
a
U
n
i
v
e
r
s
i
t
y
,
H
y
d
e
r
a
b
a
d
,
I
n
d
i
a
2
D
e
p
a
r
t
me
n
t
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
M
u
f
f
a
k
h
a
m J
a
h
C
o
l
l
e
g
e
o
f
En
g
i
n
e
e
r
i
n
g
a
n
d
Te
c
h
n
o
l
o
g
y
,
H
y
d
e
r
a
b
a
d
,
I
n
d
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ma
r
1
,
2
0
2
4
R
ev
is
ed
J
u
l 2
8
,
2
0
2
4
Acc
ep
ted
Au
g
1
4
,
2
0
2
4
Th
e
g
o
a
l
o
f
th
is
stu
d
y
is t
o
c
re
a
te
a
stro
n
g
s
y
ste
m
th
a
t
c
a
n
q
u
ic
k
ly
d
e
tec
t
a
n
d
p
re
c
ise
ly
c
las
sify
b
ra
in
tu
m
o
rs,
wh
ich
is
e
ss
e
n
ti
a
l
fo
r
imp
ro
v
in
g
trea
tme
n
t
re
su
lt
s.
Th
e
st
u
d
y
u
se
s
a
d
v
a
n
c
e
d
ima
g
e
p
ro
c
e
ss
in
g
tec
h
n
iq
u
e
s
a
n
d
t
h
e
Ne
u
ro
F
u
si
o
n
Ne
t
d
e
e
p
lea
rn
i
n
g
m
o
d
e
l
to
a
c
c
u
ra
tely
se
g
m
e
n
t
d
a
t
a
fro
m
th
e
b
ra
in
t
u
m
o
r
se
g
m
e
n
tati
o
n
(
B
RATS
)
d
a
tas
e
t,
p
re
se
n
ti
n
g
a
d
e
tailed
m
e
th
o
d
o
lo
g
y
.
T
h
e
o
b
jec
ti
v
e
is
to
c
re
a
te
a
h
ig
h
-
p
re
c
isio
n
sy
ste
m
th
a
t
su
rp
a
ss
e
s
c
u
rre
n
t
m
e
th
o
d
s
i
n
k
e
y
p
e
rfo
rm
a
n
c
e
m
e
tri
c
s.
Ne
u
ro
F
u
sio
n
Ne
t
d
e
m
o
n
stra
tes
o
u
tstan
d
in
g
a
c
c
u
r
a
c
y
o
f
9
9
.
2
1
%
,
a
s
we
ll
a
s
i
m
p
re
ss
iv
e
sp
e
c
ifi
c
it
y
a
n
d
se
n
siti
v
it
y
ra
tes
o
f
9
9
.
1
7
%
a
n
d
9
9
.
3
8
3
%
,
re
sp
e
c
ti
v
e
ly
,
e
x
c
e
e
d
in
g
p
re
v
i
o
u
s
b
e
n
c
h
m
a
rk
s.
Th
e
fin
d
in
g
s
e
m
p
h
a
siz
e
th
e
sy
ste
m
'
s
a
b
il
it
y
to
g
re
a
tl
y
e
n
h
a
n
c
e
th
e
d
iag
n
o
sti
c
p
ro
c
e
ss
,
e
n
a
b
li
n
g
e
a
rly
i
n
terv
e
n
ti
o
n
a
n
d
u
lt
ima
tely
imp
r
o
v
in
g
p
a
ti
e
n
t
c
a
re
in
b
ra
i
n
t
u
m
o
r
d
e
tec
ti
o
n
a
n
d
c
las
s
ifi
c
a
ti
o
n
.
K
ey
w
o
r
d
s
:
B
r
ain
tu
m
o
r
d
etec
tio
n
B
r
ain
tu
m
o
r
s
eg
m
e
n
tatio
n
d
ataset
E
ar
ly
in
ter
v
e
n
tio
n
I
m
ag
e
p
r
o
ce
s
s
in
g
Me
d
ical
im
ag
in
g
Neu
r
iFu
s
io
n
Net
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
:
Ar
p
ith
a
Ko
tte
Dep
ar
tm
en
t o
f
C
o
m
p
u
ter
Scie
n
ce
an
d
E
n
g
in
ee
r
in
g
,
Osma
n
ia
Un
iv
er
s
ity
Hy
d
er
ab
ad
,
I
n
d
ia
E
m
ail:
k
o
ttear
p
ith
a8
7
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
B
r
ain
tu
m
o
r
s
p
r
esen
t
a
co
n
s
id
er
ab
le
h
ea
lth
h
az
ar
d
o
n
a
g
lo
b
al
s
ca
le,
im
p
o
s
in
g
a
co
n
s
id
er
a
b
le
s
tr
ain
o
n
b
o
th
in
d
iv
id
u
als
an
d
h
ea
lt
h
ca
r
e
s
y
s
tem
s
[
1
]
–
[
4
]
.
Du
e
to
th
e
co
m
p
lex
ch
ar
ac
ter
is
tics
o
f
th
e
h
u
m
a
n
b
r
ain
an
d
th
e
p
o
ten
tial
f
o
r
f
atalities
as
s
o
ciate
d
with
tu
m
o
r
s
,
d
ilig
en
t
s
u
r
v
eillan
ce
an
d
tim
ely
id
en
tific
atio
n
ar
e
im
p
er
ativ
e
[
5
]
–
[
9
]
.
T
h
e
s
ig
n
if
ican
t
in
cid
e
n
ce
o
f
b
r
ain
tu
m
o
r
s
,
in
c
o
n
ju
n
ctio
n
with
th
eir
f
r
eq
u
en
tly
asy
m
p
to
m
atic
ad
v
a
n
ce
m
en
t,
em
p
h
asizes
th
e
u
r
g
e
n
t
r
e
q
u
ir
em
en
t
f
o
r
d
ep
en
d
ab
le
d
etec
tio
n
tech
n
iq
u
es
[
1
0
]
–
[
1
2
]
.
T
h
e
p
e
r
n
icio
u
s
c
h
a
r
ac
ter
is
tics
o
f
th
ese
tu
m
o
r
s
ar
e
n
o
t
lim
ited
to
th
eir
ab
ilit
y
t
o
im
p
air
co
g
n
itiv
e
f
u
n
ctio
n
;
th
e
y
also
p
o
s
e
s
ig
n
if
ican
t
h
ea
lth
r
is
k
s
an
d
h
av
e
d
etr
im
en
tal
ef
f
ec
ts
o
n
o
v
er
a
ll
well
-
b
ein
g
.
I
t
is
cr
u
cial
to
co
m
p
r
e
h
en
d
t
h
e
h
ea
lth
h
az
ar
d
s
th
at
ar
e
lin
k
ed
to
b
r
ain
tu
m
o
r
s
in
o
r
d
er
t
o
d
ev
elo
p
ef
f
icac
io
u
s
d
etec
tio
n
m
eth
o
d
o
l
o
g
ies
[
1
3
]
.
T
h
e
s
ev
e
r
ity
o
f
th
e
cir
cu
m
s
t
an
ce
s
tem
s
f
r
o
m
th
e
d
if
f
ic
u
lties
ass
o
ciate
d
with
ea
r
ly
d
etec
tio
n
o
f
th
ese
m
ali
g
n
an
cies,
wh
en
t
h
er
ap
e
u
tic
alter
n
ativ
es
ar
e
at
th
eir
m
o
s
t
p
r
o
m
is
in
g
[
1
4
]
.
T
h
is
ar
ticle
ex
am
in
es
t
h
e
c
r
itical
h
ea
lth
h
az
ar
d
s
p
r
esen
te
d
b
y
b
r
ain
tu
m
o
r
s
,
in
v
esti
g
ates
th
e
r
ea
s
o
n
s
b
eh
in
d
th
ei
r
leth
ality
,
an
d
em
p
h
asizes
th
e
cr
iticality
o
f
ea
r
l
y
d
etec
tio
n
[
1
5
]
.
T
h
e
s
tu
d
y
ad
d
itio
n
al
ly
p
r
esen
ts
an
all
-
en
co
m
p
ass
in
g
a
p
p
r
o
ac
h
t
h
at
i
n
co
r
p
o
r
ates
f
ea
tu
r
e
ex
tr
ac
tio
n
,
im
ag
e
p
r
ep
r
o
ce
s
s
in
g
,
a
n
d
a
h
y
b
r
id
d
ee
p
lear
n
in
g
m
o
d
el
ca
lled
Neu
r
o
Fu
s
io
n
Net
to
ac
h
iev
e
p
r
ec
is
e
s
eg
m
en
tatio
n
an
d
class
if
icatio
n
.
T
h
e
liter
atu
r
e
o
n
b
r
ain
tu
m
o
r
s
eg
m
en
tatio
n
an
d
class
if
icati
o
n
in
m
ed
ical
im
ag
in
g
r
ev
ea
l
s
s
ev
er
al
lim
itatio
n
s
.
Ny
o
et
a
l
.
[
1
6
]
h
i
g
h
lig
h
ted
th
e
s
en
s
itiv
ity
o
f
m
o
r
p
h
o
lo
g
ical
o
p
e
r
atio
n
s
to
im
ag
e
q
u
ality
,
co
m
p
lex
tu
m
o
r
s
h
ap
es,
n
o
is
e
lev
els,
an
d
im
ag
in
g
ar
tifa
cts.
Po
lat
an
d
Gü
n
g
en
[
1
7
]
em
p
h
asized
th
e
d
ep
en
d
en
cy
o
f
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
I
mp
leme
n
ta
tio
n
o
f i
n
n
o
v
a
tive
a
p
p
r
o
a
c
h
fo
r
d
etec
tin
g
b
r
a
in
t
u
mo
r
s
in
…
(
A
r
p
ith
a
K
o
tte
)
6629
tr
an
s
f
er
lear
n
in
g
o
n
th
e
av
ail
ab
ilit
y
an
d
q
u
ality
o
f
lab
ele
d
d
ata
an
d
th
e
s
im
ilar
ity
b
etwe
en
th
e
s
o
u
r
ce
an
d
tar
g
et
d
o
m
ain
s
,
wh
ich
af
f
ec
t
s
p
er
f
o
r
m
an
ce
.
Hasan
ah
et
a
l
.
[
1
8
]
n
o
ted
th
e
s
en
s
itiv
ity
o
f
th
eir
m
ac
h
in
e
lear
n
in
g
-
b
ased
s
tr
ateg
y
to
im
a
g
e
q
u
ality
d
if
f
er
en
ce
s
,
r
elian
c
e
o
n
well
-
an
n
o
tated
d
atasets
,
a
n
d
th
e
ch
allen
g
e
o
f
g
en
er
alizin
g
to
n
ew
clin
ical
s
ettin
g
s
.
Hu
an
g
et
a
l
.
[
1
9
]
p
o
in
ted
o
u
t
t
h
e
s
u
s
ce
p
tib
ilit
y
o
f
th
ei
r
C
NN
an
d
co
m
p
lex
n
etwo
r
k
m
o
d
el
to
d
at
a
q
u
ality
f
lu
ctu
atio
n
s
an
d
in
ter
p
r
etab
ilit
y
is
s
u
es.
Kh
alil
et
a
l
.
[
2
0
]
d
is
cu
s
s
ed
th
e
co
m
p
u
tatio
n
al
co
m
p
le
x
ity
an
d
r
eso
u
r
ce
d
em
an
d
s
o
f
t
h
eir
tw
o
-
s
tep
d
r
a
g
o
n
f
l
y
m
ec
h
an
is
m
,
wh
ich
m
ay
lim
it
its
p
r
ac
tical
ap
p
licatio
n
d
esp
ite
h
ig
h
ac
cu
r
ac
y
.
K
r
is
h
n
a
m
m
al
a
n
d
R
aj
a
[
2
1
]
i
d
e
n
ti
f
i
ed
c
h
al
le
n
g
es
i
n
i
n
t
er
p
r
e
ti
n
g
C
u
r
v
e
let
-
p
r
e
p
r
o
c
ess
e
d
f
ea
tu
r
e
m
a
p
s
a
n
d
u
n
d
e
r
s
t
an
d
i
n
g
c
li
n
ic
all
y
r
el
ev
an
t
i
n
f
o
r
m
a
tio
n
,
a
f
f
ec
t
in
g
t
h
e
ac
c
ep
ta
n
c
e
o
f
t
h
ei
r
m
e
th
o
d
.
C
h
e
n
g
e
t
a
l
.
[
2
2
]
m
e
n
ti
o
n
e
d
t
h
e
s
en
s
iti
v
it
y
o
f
t
h
eir
c
o
n
v
o
l
u
t
io
n
al
c
ap
s
u
l
e
n
et
wo
r
k
t
o
i
m
a
g
e
q
u
ali
ty
,
t
h
e
n
ee
d
f
o
r
s
ig
n
i
f
ic
a
n
t
c
o
m
p
u
tati
o
n
a
l r
es
o
u
r
ce
s
,
a
n
d
p
o
te
n
t
ial
i
n
te
r
p
r
eta
b
ilit
y
c
h
all
en
g
es
d
u
e
to
t
h
e
c
o
m
p
l
e
x
it
y
o
f
th
e
m
o
d
el.
L
astl
y
,
A
b
i
wi
n
a
n
d
a
et
a
l
.
[
2
3
]
h
i
g
h
li
g
h
te
d
th
e
s
en
s
iti
v
i
ty
o
f
C
NNs
to
i
m
a
g
e
q
u
ali
ty
f
l
u
ct
u
ati
o
n
s
an
d
p
o
t
e
n
t
ial
b
iases
i
n
t
h
e
tr
ai
n
i
n
g
d
a
ta
,
i
m
p
ac
t
in
g
t
h
e
m
o
d
e
l'
s
e
f
f
ec
t
iv
e
n
ess
.
T
h
e
liter
atu
r
e
o
n
b
r
ain
tu
m
o
r
s
eg
m
en
tatio
n
an
d
class
if
icati
o
n
in
m
ed
ical
im
ag
in
g
r
ev
ea
l
s
s
ev
er
al
lim
itatio
n
s
,
in
clu
d
in
g
s
en
s
itiv
ity
to
im
a
g
e
q
u
ality
,
r
elian
ce
o
n
lab
eled
d
ata,
c
o
m
p
u
tatio
n
al
co
m
p
lex
ity
,
an
d
in
ter
p
r
etab
ilit
y
is
s
u
es.
T
h
ese
ch
allen
g
es
h
in
d
er
th
e
d
ev
e
lo
p
m
en
t
o
f
r
o
b
u
s
t,
g
en
e
r
aliza
b
le,
an
d
ef
f
icien
t
m
o
d
els
f
o
r
ac
cu
r
ate
an
d
tim
ely
b
r
ain
tu
m
o
r
d
etec
tio
n
.
T
o
ad
d
r
ess
th
ese
lim
itat
io
n
s
,
t
h
is
s
tu
d
y
m
ak
es
a
s
ch
o
lar
ly
co
n
tr
ib
u
tio
n
to
th
e
d
o
m
ain
o
f
b
r
ain
tu
m
o
r
d
etec
tio
n
an
d
class
if
icatio
n
th
r
o
u
g
h
th
e
in
tr
o
d
u
ctio
n
o
f
a
s
o
p
h
is
ticated
an
d
in
teg
r
ated
ap
p
r
o
ac
h
th
at
m
er
g
es
im
ag
e
p
r
ep
r
o
ce
s
s
in
g
,
f
ea
tu
r
e
e
x
tr
ac
tio
n
,
an
d
th
e
Neu
r
o
Fu
s
io
n
Net
m
o
d
el.
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
o
lo
g
y
ef
f
ec
tiv
ely
tack
les
th
e
p
r
ess
in
g
r
eq
u
ir
em
en
ts
f
o
r
p
r
ec
is
e
an
d
tim
ely
id
en
tific
atio
n
o
f
b
r
ain
tu
m
o
r
s
,
co
n
s
id
er
i
n
g
th
e
co
m
p
lex
ch
ar
ac
ter
is
tics
o
f
th
e
h
u
m
an
b
r
ai
n
an
d
th
e
g
r
av
e
h
ea
lth
h
az
ar
d
s
lin
k
e
d
to
s
u
ch
g
r
o
wth
s
.
B
y
em
p
lo
y
in
g
th
e
Neu
r
o
Fu
s
io
n
Net
m
o
d
el,
wh
ich
is
a
h
y
b
r
i
d
d
ee
p
en
co
d
er
U
-
Net
ar
ch
itectu
r
e,
im
ag
e
s
eg
m
en
tatio
n
a
n
d
class
if
icatio
n
ar
e
ac
co
m
p
lis
h
ed
in
a
s
o
p
h
is
ticated
m
an
n
er
.
T
h
is
ap
p
r
o
ac
h
o
v
er
c
o
m
es
th
e
lim
itatio
n
s
o
f
p
r
io
r
s
tu
d
ies
b
y
en
h
an
cin
g
im
ag
e
q
u
ality
s
en
s
itiv
ity
,
r
ed
u
cin
g
r
elian
ce
o
n
la
b
eled
d
ata,
im
p
r
o
v
in
g
g
e
n
er
aliza
tio
n
ac
r
o
s
s
clin
ical
s
ettin
g
s
,
ad
d
r
ess
in
g
co
m
p
u
tatio
n
al
co
m
p
lex
ity
,
a
n
d
o
f
f
er
in
g
b
ette
r
in
ter
p
r
etab
ilit
y
.
T
h
e
ex
h
a
u
s
tiv
e
f
ea
tu
r
e
s
u
m
m
a
r
y
o
b
tain
ed
th
r
o
u
g
h
g
r
a
y
lev
el
co
-
o
cc
u
r
r
en
ce
m
atr
ix
(
GL
C
M
)
an
aly
s
is
p
r
o
v
id
es
a
th
o
r
o
u
g
h
ev
alu
atio
n
o
f
p
ar
ticu
la
r
s
am
p
les,
th
er
eb
y
en
h
an
cin
g
o
u
r
co
m
p
r
eh
en
s
io
n
o
f
th
e
in
tr
icate
attr
ib
u
tes
o
f
b
r
ain
tu
m
o
r
s
.
I
n
its
en
tire
ty
,
th
i
s
s
tu
d
y
en
d
ea
v
o
r
s
to
p
r
o
v
id
e
a
s
u
b
s
tan
tial
co
n
tr
i
b
u
tio
n
to
th
e
en
h
an
ce
m
en
t
o
f
m
eth
o
d
o
lo
g
ies
u
tili
ze
d
in
t
h
e
d
etec
tio
n
o
f
b
r
ain
tu
m
o
r
s
.
I
t u
n
d
er
s
co
r
es th
e
c
r
iticality
o
f
tim
ely
id
en
tific
atio
n
to
f
ac
ilit
ate
ef
f
icac
io
u
s
m
ed
ica
l in
ter
v
en
tio
n
s
an
d
im
p
r
o
v
e
p
atien
t p
r
o
g
n
o
s
es.
T
h
e
r
esear
ch
ar
ticle
is
d
iv
id
e
d
in
to
f
iv
e
p
a
r
ts
.
T
h
e
in
tr
o
d
u
ctio
n
p
r
esen
ts
a
b
r
o
ad
o
v
er
v
iew
o
f
th
e
h
ea
lth
h
az
ar
d
s
ass
o
ciate
d
wit
h
b
r
ain
tu
m
o
r
s
,
em
p
h
asizin
g
th
e
cr
u
cial
im
p
o
r
tan
ce
o
f
ea
r
l
y
d
etec
tio
n
d
u
e
to
th
eir
p
o
ten
tial
m
o
r
tality
an
d
n
eg
ativ
e
ef
f
ec
ts
o
n
co
g
n
itiv
e
f
u
n
ctio
n
.
T
h
e
r
e
v
iew
o
f
e
x
is
tin
g
wo
r
k
s
o
n
b
r
ain
tu
m
o
r
d
etec
tio
n
,
class
if
icatio
n
,
an
d
r
elate
d
ap
p
r
o
ac
h
es
d
iv
es
in
to
ex
is
tin
g
wo
r
k
s
,
h
ig
h
lig
h
tin
g
g
ap
s
in
th
e
cu
r
r
en
t
s
tate
o
f
r
esear
ch
.
Sectio
n
3
d
escr
ib
es
th
e
s
u
g
g
ested
s
y
s
tem
an
d
alg
o
r
ith
m
,
g
iv
in
g
a
co
m
p
r
eh
e
n
s
iv
e
m
eth
o
d
o
l
o
g
y
f
o
r
r
eliab
le
b
r
ain
tu
m
o
r
s
eg
m
e
n
tatio
n
th
at
in
cl
u
d
es
im
ag
e
p
r
ep
r
o
ce
s
s
in
g
,
f
ea
tu
r
e
ex
tr
ac
tio
n
,
a
n
d
th
e
Neu
r
o
Fu
s
io
n
Net
m
o
d
el.
Sectio
n
4
g
o
es
o
v
er
th
e
e
x
p
er
i
m
en
tal
in
v
esti
g
atio
n
s
an
d
p
er
f
o
r
m
an
ce
ev
alu
atio
n
in
d
etail,
in
clu
d
in
g
th
e
d
ataset
u
s
ed
,
ex
p
er
im
en
tal
s
etu
p
,
an
d
o
u
tco
m
es.
T
h
e
co
n
clu
s
io
n
an
d
f
u
t
u
r
e
s
co
p
e
s
ec
tio
n
s
u
m
m
ar
izes
th
e
im
p
o
r
tan
t
f
in
d
in
g
s
,
ex
a
m
in
es
th
e
lim
its
o
f
th
e
p
r
o
p
o
s
ed
s
y
s
tem
,
an
d
p
r
o
p
o
s
es
f
u
tu
r
e
r
esear
ch
p
ath
s
,
u
n
d
e
r
lin
in
g
t
h
e
s
tu
d
y
'
s
im
p
o
r
tan
ce
in
ex
p
an
d
in
g
b
r
ain
tu
m
o
r
d
etec
tio
n
tech
n
o
lo
g
ies
an
d
im
p
r
o
v
in
g
p
atien
t
o
u
tco
m
es.
2.
L
I
T
E
R
AT
U
RE
SU
RVE
Y
Ny
o
et
a
l
.
[
1
6
]
r
ec
e
n
tly
p
r
esen
ted
an
in
n
o
v
ativ
e
a
p
p
r
o
ac
h
in
th
e
f
ield
o
f
m
ed
ical
im
ag
in
g
,
s
p
ec
if
ically
f
o
r
s
eg
m
en
tin
g
b
r
ain
ca
n
ce
r
s
f
r
o
m
m
a
g
n
etic
r
eso
n
an
ce
im
ag
in
g
(
M
R
I
)
im
ag
es.
T
h
eir
m
eth
o
d
o
l
o
g
y
co
m
b
i
n
es
th
r
es
h
o
ld
in
g
an
d
m
o
r
p
h
o
lo
g
y
tech
n
iq
u
es,
r
esu
ltin
g
in
a
n
im
p
r
ess
iv
e
9
0
%
ac
cu
r
ac
y
r
ate.
T
h
is
ac
h
iev
em
e
n
t
r
ep
r
e
s
en
ts
a
s
ig
n
if
ican
t
ad
v
an
ce
m
en
t
in
th
e
f
ield
,
p
r
o
m
is
in
g
i
m
p
r
o
v
e
d
d
ia
g
n
o
s
tic
ca
p
ab
ilit
ies
an
d
p
o
s
s
ib
ly
ass
is
tin
g
in
tr
ea
tm
en
t
p
lan
n
i
n
g
f
o
r
p
atien
ts
with
b
r
ain
tu
m
o
r
s
.
Mo
r
p
h
o
lo
g
ical
o
p
er
atio
n
s
ar
e
s
en
s
itiv
e
to
v
ar
iatio
n
s
in
im
ag
e
q
u
ality
an
d
m
ay
s
tr
u
g
g
le
with
co
m
p
le
x
tu
m
o
r
s
h
a
p
es
o
r
o
v
er
lap
p
i
n
g
s
tr
u
ctu
r
es.
Fu
r
th
e
r
m
o
r
e,
n
o
is
e
lev
els,
im
ag
in
g
ar
tifa
cts,
an
d
v
ar
iatio
n
s
in
tu
m
o
r
ch
a
r
ac
ter
is
tics
ca
n
all
h
av
e
a
n
im
p
ac
t o
n
th
e
ef
f
icac
y
o
f
m
o
r
p
h
o
l
o
g
ical
tec
h
n
iq
u
es.
Po
lat
an
d
Gü
n
g
en
[
1
7
]
p
r
esen
ted
a
u
n
iq
u
e
tech
n
iq
u
e
to
b
r
ai
n
tu
m
o
r
d
etec
tio
n
in
2
0
2
1
b
y
ap
p
ly
in
g
tr
an
s
f
er
lear
n
in
g
m
et
h
o
d
o
lo
g
y
.
T
h
eir
s
tu
d
y
was
a
h
u
g
e
s
u
cc
ess
,
with
an
asto
u
n
d
in
g
o
v
e
r
all
ac
cu
r
ac
y
p
er
ce
n
tag
e
o
f
9
9
.
0
2
%.
T
h
e
r
e
lian
ce
o
n
th
e
av
ailab
ilit
y
an
d
q
u
ality
o
f
lab
ele
d
d
ata
u
n
i
q
u
e
to
b
r
ain
tu
m
o
r
s
is
o
n
e
n
o
tewo
r
th
y
d
r
awb
ac
k
.
F
u
r
th
er
m
o
r
e,
th
e
r
elev
an
ce
an
d
s
im
ilar
ity
b
etwe
en
th
e
s
o
u
r
ce
d
o
m
ai
n
(
t
h
e
p
r
e
-
tr
ain
ed
d
ataset)
an
d
th
e
tar
g
et
d
o
m
ain
(
th
e
b
r
ain
tu
m
o
r
d
ataset)
s
tr
o
n
g
ly
in
f
lu
en
ce
s
t
h
e
p
er
f
o
r
m
a
n
ce
o
f
tr
an
s
f
er
lear
n
in
g
m
eth
o
d
s
.
W
h
en
th
er
e
is
a
s
ig
n
if
ican
t
d
is
s
im
ilar
ity
,
th
e
ef
f
ec
tiv
en
ess
o
f
tr
a
n
s
f
er
lear
n
in
g
m
a
y
d
ec
r
ea
s
e,
an
d
t
h
e
m
o
d
el
m
ay
n
o
t g
en
e
r
alize
well
to
th
e
s
p
ec
if
ic
ch
ar
ac
ter
is
tics
o
f
b
r
ain
t
u
m
o
r
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
e
r
20
24
:
6
6
2
8
-
6
6
4
1
6630
Hasan
ah
et
a
l
.
[
1
8
]
p
u
b
lis
h
ed
a
m
ac
h
in
e
lear
n
in
g
(
ML
)
-
b
ase
d
s
tr
ateg
y
f
o
r
b
r
ain
tu
m
o
r
class
if
icatio
n
in
2
0
2
1
,
g
iv
i
n
g
a
co
m
p
r
e
h
en
s
iv
e
m
eth
o
d
o
lo
g
y
th
at
in
c
o
r
p
o
r
ates
f
ilter
in
g
,
co
n
to
u
r
in
g
,
a
n
d
th
r
esh
o
ld
in
g
in
to
th
e
tu
m
o
r
a
n
aly
s
is
p
r
ep
r
o
ce
s
s
in
g
an
d
s
eg
m
en
tatio
n
s
ta
g
es.
T
h
e
s
tated
av
er
ag
e
ac
cu
r
ac
y
o
f
9
5
.
8
3
%
d
em
o
n
s
tr
ates
a
h
ig
h
lev
el
o
f
s
u
cc
ess
.
T
h
e
lim
itatio
n
h
er
e
is
th
e
s
en
s
itiv
ity
to
d
if
f
er
en
ce
s
in
im
ag
e
q
u
ality
an
d
its
r
elian
ce
o
n
th
e
av
ailab
ilit
y
o
f
well
-
an
n
o
tated
an
d
d
iv
er
s
e
d
atasets
f
o
r
tr
ain
in
g
.
Fu
r
th
e
r
m
o
r
e,
th
e
m
o
d
el'
s
g
en
er
aliza
tio
n
to
n
ew,
p
r
ev
io
u
s
ly
u
n
s
ee
n
d
atasets
is
cr
u
ci
al,
b
ec
au
s
e
p
er
f
o
r
m
an
ce
m
ay
v
ar
y
in
d
if
f
e
r
en
t
clin
ical
s
ettin
g
s
.
Hu
an
g
et
a
l
.
[
1
9
]
p
u
b
lis
h
ed
co
n
v
o
l
u
tio
n
al
n
eu
r
al
n
etwo
r
k
s
an
d
co
m
p
lex
n
etwo
r
k
s
(
C
NNBC
N)
in
2
0
2
0
,
in
d
icatin
g
a
s
ig
n
if
ican
t
p
r
o
g
r
ess
in
th
e
f
ield
o
f
n
e
u
r
al
n
etwo
r
k
-
b
ased
tech
n
iq
u
es.
T
h
e
s
u
g
g
ested
m
o
d
el
co
m
b
in
es
th
e
ch
ar
ac
ter
is
tics
o
f
co
n
v
o
lu
ti
o
n
al
n
eu
r
al
n
etwo
r
k
s
(
C
NN
s
)
,
wh
ich
ar
e
wid
ely
ac
k
n
o
wled
g
ed
f
o
r
th
eir
ef
f
icac
y
in
im
ag
e
-
r
elate
d
task
s
,
with
co
m
p
licated
n
etw
o
r
k
p
r
in
cip
les.
T
h
e
m
eth
o
d
at
tain
ed
an
ac
c
u
r
ac
y
o
f
9
5
.
4
9
%.
T
h
e
m
o
d
el'
s
s
u
s
ce
p
tib
ilit
y
to
f
lu
ctu
atio
n
s
in
d
ata
q
u
ality
an
d
p
o
ten
tial
in
ter
p
r
et
ab
ilit
y
d
if
f
icu
lties
,
p
ar
ticu
lar
ly
in
s
o
p
h
is
ticated
n
etwo
r
k
-
b
ased
s
y
s
tem
s
.
T
h
e
d
ev
elo
p
m
en
t
b
y
Kh
alil
e
t
a
l
.
[
2
0
]
o
f
a
two
-
s
tep
d
r
a
g
o
n
f
ly
m
ec
h
an
is
m
s
y
s
tem
is
a
s
ig
n
if
ican
t
ad
v
an
ce
m
e
n
t
with
a
n
ac
c
u
r
ac
y
o
f
9
8
.
2
0
%.
Ho
wev
e
r
,
it
is
c
r
itical
to
r
ec
o
g
n
ize
a
s
ig
n
i
f
ican
t
d
is
ad
v
a
n
tag
e
o
f
th
is
s
tr
ateg
y
.
T
h
e
co
m
p
u
tatio
n
al
co
m
p
le
x
ity
o
f
th
e
two
-
s
tep
d
r
a
g
o
n
f
ly
m
ec
h
an
is
m
m
a
y
ca
u
s
e
p
r
o
b
lem
s
in
ter
m
s
o
f
p
r
o
ce
s
s
in
g
tim
e
an
d
r
eso
u
r
ce
r
eq
u
ir
em
en
ts
.
W
h
ile
th
e
p
r
o
ce
s
s
's
in
tr
icac
ies
co
n
tr
ib
u
te
to
h
ig
h
ac
cu
r
ac
y
,
th
ey
m
a
y
im
p
o
s
e
a
co
m
p
u
tatio
n
al
co
s
t,
r
estrictin
g
its
ap
p
licab
ilit
y
.
T
h
is
d
is
ad
v
an
tag
e
em
p
h
asizes
th
e
im
p
o
r
tan
ce
o
f
s
tr
ik
in
g
a
p
er
f
ec
t
b
alan
ce
b
etwe
en
s
eg
m
en
tatio
n
ac
c
u
r
ac
y
an
d
p
r
o
c
ess
in
g
p
er
f
o
r
m
an
ce
,
esp
ec
ially
in
m
ed
ical
im
ag
e
an
aly
s
is
,
wh
er
e
s
p
ee
d
y
r
ep
lies
ar
e
cr
itical
f
o
r
g
o
o
d
clin
ica
l
d
ec
is
io
n
-
m
ak
in
g
.
T
h
is
co
m
p
u
tatio
n
al
b
ar
r
ier
m
u
s
t
b
e
o
v
er
co
m
e
in
o
r
d
e
r
to
in
cr
ea
s
e
th
e
o
v
er
all
p
r
ac
ticality
an
d
u
tili
ty
o
f
th
e
p
r
o
p
o
s
ed
d
r
a
g
o
n
f
l
y
m
ec
h
a
n
is
m
in
th
er
ap
e
u
tic
ap
p
licatio
n
s
.
Kr
is
h
n
am
m
al
an
d
R
aja
[
2
1
]
u
s
ed
cu
r
v
elet
-
p
r
ep
r
o
ce
s
s
ed
f
ea
tu
r
e
m
ap
s
to
tr
ain
a
C
NN
f
o
r
b
r
ain
im
ag
in
g
d
ataset
ca
teg
o
r
izatio
n
.
W
h
ile
th
eir
ap
p
r
o
ac
h
co
n
tr
ib
u
tes
g
r
ea
tly
to
t
h
e
f
ield
o
f
m
ed
ical
p
ictu
r
e
ca
teg
o
r
izatio
n
,
s
o
m
e
d
r
awb
ac
k
s
r
elate
d
with
th
e
m
eth
o
d
s
u
s
ed
in
th
is
wo
r
k
m
u
s
t
b
e
a
ck
n
o
wled
g
e
d
.
T
h
e
in
ter
p
r
etab
ilit
y
o
f
f
ea
tu
r
es
ex
t
r
ac
ted
in
th
e
c
u
r
v
elet
d
o
m
ain
is
o
n
e
s
u
ch
lim
ita
tio
n
.
C
u
r
v
elet
tr
an
s
f
o
r
m
s
ar
e
ef
f
ec
tiv
e
in
ca
p
tu
r
in
g
f
in
e
d
et
ails
an
d
ed
g
es
in
im
ag
es,
b
u
t
th
eir
ap
p
licab
ilit
y
in
th
e
co
n
t
ex
t
o
f
b
r
ain
im
ag
e
ca
teg
o
r
izatio
n
ca
n
b
e
ch
allen
g
in
g
.
Un
d
er
s
tan
d
in
g
t
h
e
clin
ically
r
elev
an
t
in
f
o
r
m
atio
n
in
cl
u
d
ed
in
th
e
c
u
r
v
ele
t
d
o
m
ain
f
ea
t
u
r
es
co
u
ld
b
e
cr
it
ical
f
o
r
u
n
d
er
s
tan
d
in
g
th
e
m
o
d
el'
s
d
ec
is
io
n
-
m
ak
in
g
p
r
o
ce
s
s
an
d
g
u
ar
an
teein
g
clin
ical
ac
ce
p
tan
ce
.
As
m
en
tio
n
ed
in
t
h
eir
s
tu
d
y
,
C
h
en
g
et
a
l
.
[
2
2
]
p
r
o
p
o
s
ed
a
u
n
iq
u
e
tech
n
iq
u
e
to
tu
m
o
r
ca
t
eg
o
r
izatio
n
in
2
0
1
9
b
y
in
tr
o
d
u
ci
n
g
a
n
ad
v
an
ce
d
co
n
v
o
l
u
tio
n
al
ca
p
s
u
le
n
etwo
r
k
(
C
o
n
v
C
ap
s
)
ar
ch
itectu
r
e.
T
h
is
ar
ch
itectu
r
e
e
x
ten
d
s
class
ic
C
NNs
b
y
u
s
in
g
ca
p
s
u
le
n
etwo
r
k
s
,
a
f
o
r
m
o
f
n
eu
r
al
n
etwo
r
k
m
o
d
u
le
in
tr
o
d
u
ce
d
to
allev
iate
C
NN
co
n
s
tr
ain
ts
,
s
u
ch
as
is
s
u
es
with
h
ier
a
r
ch
ical
r
elatio
n
s
h
ip
s
b
etwe
en
f
ea
tu
r
es
.
W
h
ile
r
ea
ch
in
g
a
class
if
icatio
n
ac
cu
r
ac
y
o
f
9
3
.
5
%
in
d
icate
s
th
e
m
o
d
el'
s
u
s
ef
u
ln
ess
,
s
ig
n
if
ican
t
lim
its
m
u
s
t
b
e
co
n
s
id
er
ed
.
T
h
ese
m
ay
in
clu
d
e
im
ag
e
q
u
ality
s
en
s
itiv
ity
,
th
e
n
ec
ess
ity
f
o
r
s
ig
n
if
ican
t
co
m
p
u
tatio
n
a
l
r
eso
u
r
ce
s
d
u
r
in
g
tr
ain
in
g
,
an
d
p
o
ten
tial in
ter
p
r
e
tab
ilit
y
ch
allen
g
es d
u
e
t
o
th
e
c
o
m
p
lex
ity
o
f
ca
p
s
u
le
n
etwo
r
k
d
esig
n
s
.
As
s
h
o
wn
in
th
eir
r
esear
ch
A
b
iwin
an
d
a
et
a
l
.
[
2
3
]
,
attem
p
ted
th
e
ch
allen
g
e
o
f
tu
m
o
r
ca
teg
o
r
izatio
n
in
2
0
1
8
b
y
im
p
lem
e
n
tin
g
a
C
NN.
T
h
e
ac
cu
r
ac
y
o
f
8
4
.
1
9
%
in
d
icate
s
a
r
ea
s
o
n
ab
ly
ef
f
ec
tiv
e
p
er
f
o
r
m
an
ce
.
T
h
e
d
is
ad
v
an
tag
es
o
f
C
NNs
ar
e
th
eir
s
en
s
itiv
ity
to
f
lu
ct
u
atio
n
s
in
im
ag
e
q
u
ality
an
d
p
o
t
en
tial
b
iases
in
th
e
tr
ain
in
g
d
ata.
3.
P
RO
P
O
SE
D
SYS
T
E
M
T
h
e
in
itial
s
tep
o
f
th
e
s
y
s
tem
u
n
d
er
c
o
n
s
id
er
atio
n
,
as
illu
s
tr
ated
in
Fig
u
r
e
1
,
is
to
o
b
tain
t
h
e
o
r
ig
i
n
al
b
r
ain
tu
m
o
r
im
a
g
e
(
(
,
)
)
f
r
o
m
th
e
B
R
AT
S
d
atab
ase
[
2
4
]
.
T
h
is
im
ag
e
is
th
en
r
esized
u
s
i
n
g
b
ilin
ea
r
in
ter
p
o
latio
n
to
attain
a
s
tan
d
ar
d
ized
d
im
en
s
io
n
o
f
5
1
2
×
5
1
2
.
Af
ter
r
esizin
g
th
e
im
ag
e,
lu
m
in
an
ce
-
p
r
eser
v
in
g
R
GB
to
g
r
ay
s
ca
le
m
ap
p
in
g
is
u
tili
ze
d
to
co
n
v
er
t
it
to
g
r
ay
s
ca
le.
Fo
llo
win
g
th
is
,
in
o
r
d
er
to
r
ed
u
ce
n
o
is
e,
a
non
-
lo
ca
l
m
ea
n
s
(
NL
M)
f
ilter
is
im
p
lem
en
ted
o
n
th
e
g
r
a
y
s
ca
le
im
ag
e.
Su
b
s
eq
u
e
n
tly
,
p
ix
el
n
o
r
m
aliza
tio
n
elim
in
atio
n
is
e
x
ec
u
ted
with
th
e
p
u
r
p
o
s
e
o
f
im
p
r
o
v
in
g
co
n
tr
ast
an
d
en
h
an
cin
g
p
h
o
s
p
h
o
r
u
s
v
alu
es,
th
er
eb
y
p
r
o
m
o
tin
g
p
ix
el
v
alu
e
u
n
if
o
r
m
ity
.
Fo
llo
win
g
n
o
r
m
aliza
tio
n
,
f
r
eq
u
en
cy
c
o
m
p
o
n
en
ts
ar
e
ca
p
tu
r
e
d
u
s
in
g
a
2
D
d
is
cr
ete
wav
elet
tr
an
s
f
o
r
m
(
DW
T
)
o
n
th
e
im
ag
e.
Featu
r
e
ex
tr
ac
tio
n
is
p
er
f
o
r
m
e
d
o
n
th
e
DW
T
-
tr
an
s
f
o
r
m
ed
im
ag
e
v
i
a
GL
C
M
an
aly
s
i
s
,
wh
ich
y
ield
s
in
f
o
r
m
atio
n
r
eg
a
r
d
in
g
th
e
f
ea
tu
r
es
in
v
o
lv
ed
.
Af
ter
f
ee
d
in
g
th
e
GL
C
M
-
d
er
iv
e
d
f
ea
tu
r
es
in
to
th
e
Neu
r
o
Fu
s
io
n
Net
m
o
d
el,
w
h
ich
is
a
h
y
b
r
id
d
ee
p
en
co
d
e
r
-
U
-
Net,
p
ix
el
s
eg
m
en
tatio
n
,
d
ee
p
au
to
e
n
co
d
e
r
tr
ain
in
g
,
an
d
im
ag
e
class
if
icatio
n
ar
e
p
e
r
f
o
r
m
ed
.
An
e
x
h
au
s
tiv
e
f
ea
tu
r
e
s
u
m
m
ar
y
is
em
p
lo
y
ed
to
ev
al
u
ate
th
e
m
o
d
el,
wh
ich
f
ac
ilit
ates
a
th
o
r
o
u
g
h
an
aly
s
is
o
f
its
p
er
f
o
r
m
an
ce
o
n
p
a
r
ticu
lar
s
am
p
les.
B
y
s
h
o
wca
s
in
g
th
e
m
o
d
el'
s
p
r
o
f
icien
cy
in
p
r
e
p
r
o
ce
s
s
in
g
,
n
o
is
e
r
ed
u
ctio
n
,
an
d
tu
m
o
r
class
if
icatio
n
,
th
is
co
m
p
r
eh
en
s
iv
e
s
tr
ateg
y
p
r
o
v
id
es a
r
eliab
le
f
r
am
ewo
r
k
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
I
mp
leme
n
ta
tio
n
o
f i
n
n
o
v
a
tive
a
p
p
r
o
a
c
h
fo
r
d
etec
tin
g
b
r
a
in
t
u
mo
r
s
in
…
(
A
r
p
ith
a
K
o
tte
)
6631
Fig
u
r
e
1
.
Pro
p
o
s
ed
s
y
s
tem
ar
c
h
itectu
r
e
T
h
e
ex
p
er
im
en
tal
s
etu
p
f
o
r
th
is
s
tu
d
y
in
cl
u
d
es
a
p
er
s
o
n
al
co
m
p
u
ter
(
PC
)
with
an
I
n
tel
(
R
)
C
o
r
e
(
T
M)
i5
-
8
2
6
5
U
C
PU
r
u
n
n
in
g
at
a
b
ase
f
r
eq
u
en
cy
o
f
1
.
6
0
GHz
(
with
a
m
ax
im
u
m
tu
r
b
o
f
r
e
q
u
en
c
y
o
f
1
.
8
0
GHz
)
,
8
.
0
0
GB
o
f
R
AM
(
with
7
.
8
5
GB
av
ailab
le
f
o
r
u
s
e)
,
a
n
d
a
6
4
-
b
it
o
p
er
at
in
g
s
y
s
tem
with
an
x64
-
b
ased
p
r
o
ce
s
s
o
r
.
T
h
e
u
tili
ze
d
to
o
ls
co
m
p
r
is
e
MA
T
L
AB
R
2
0
2
2
b
eq
u
i
p
p
ed
with
im
ag
e
p
r
o
ce
s
s
in
g
,
co
m
p
u
ter
v
is
io
n
,
an
d
d
ee
p
lea
r
n
in
g
to
o
lb
o
x
e
s.
T
h
e
d
etailed
d
is
cu
s
s
io
n
o
f
ab
o
v
e
s
y
s
tem
ar
ch
itectu
r
e
as
(
1
)
.
(
,
)
r
ep
r
esen
ts
th
e
o
r
ig
in
al
im
ag
e
s
ca
n
o
b
tain
ed
f
r
o
m
th
e
B
R
AT
S
d
atab
ase
[
2
5
]
.
B
y
r
esizin
g
t
h
e
im
ag
e,
a
n
ew
im
ag
e
d
en
o
t
ed
(
′
,
′
)
is
g
en
er
ated
with
th
e
ex
ac
t
d
im
en
s
io
n
s
o
f
5
1
2
×
5
1
2
.
B
ilin
ea
r
in
ter
p
o
latio
n
is
a
p
r
ev
alen
t
tech
n
iq
u
e
u
tili
ze
d
in
th
e
p
r
o
ce
s
s
o
f
r
esizin
g
im
a
g
e
s
v
ia
in
ter
p
o
latio
n
.
B
y
ap
p
ly
i
n
g
a
s
et
o
f
eq
u
atio
n
s
to
a
s
a
m
p
le
o
f
th
e
o
r
ig
in
al
im
ag
e,
th
e
r
esized
im
a
g
e
is
g
e
n
er
ated
.
T
h
e
ex
p
r
ess
io
n
f
o
r
b
ilin
ea
r
in
ter
p
o
latio
n
is
as:
(
′
,
′
)
=
∑
∑
(
,
)
⋅
(
′
,
)
⋅
(
′
,
)
1
=
0
1
=
0
(
1
)
w
h
er
e:
-
′
an
d
′
ar
e
th
e
co
o
r
d
in
ates in
th
e
r
esized
im
ag
e.
=
⌊
′
⌋
+
(
2
)
=
⌊
′
⌋
+
(
3
)
-
(
′
,
)
and
(
′
,
)
ar
e
t
he
l
i
n
ea
r
i
n
t
e
r
pol
at
i
on w
ei
ght
s.
In
o
r
d
e
r
to
co
n
v
er
t
th
e
r
esize
d
im
ag
e
to
g
r
a
y
s
ca
le,
co
lo
r
m
ap
p
in
g
is
p
er
f
o
r
m
ed
.
C
o
m
m
o
n
ly
u
s
ed
is
lu
m
in
an
ce
-
p
r
eser
v
in
g
co
n
v
er
s
io
n
f
r
o
m
R
GB
to
g
r
ay
s
ca
le,
wh
ich
is
m
ath
em
atica
lly
ex
p
r
ess
ed
as
(
4
)
.
(
′
,
′
)
=
0
.
299
⋅
(
′
,
′
)
+
0
.
587
⋅
(
′
,
′
)
+
0
.
114
⋅
(
′
,
′
)
(
4
)
w
h
er
e:
-
,
,
an
d
ar
e
th
e
r
e
d
,
g
r
ee
n
,
an
d
b
lu
e
c
h
an
n
els o
f
th
e
r
esized
im
ag
e.
-
T
h
e
weig
h
ts
0
.
2
9
9
,
0
.
5
8
7
,
an
d
0
.
1
1
4
ar
e
u
tili
ze
d
in
th
is
f
o
r
m
u
la;
th
ey
ar
e
ty
p
ical
v
alu
es
f
o
r
co
n
v
e
r
tin
g
R
GB
to
g
r
ay
s
ca
le
wh
ile
m
ain
tain
in
g
lu
m
in
a
n
ce
.
C
o
n
tin
u
in
g
f
r
o
m
th
e
p
r
ev
io
u
s
s
tag
e,
p
r
o
v
id
e
th
e
im
ag
e
in
g
r
ay
s
ca
le
f
o
r
th
e
f
o
llo
win
g
o
n
e.
Su
b
s
eq
u
en
tly
,
a
NL
M
f
ilter
i
s
im
p
lem
en
ted
o
n
th
e
m
o
n
o
c
h
r
o
m
atic
r
e
n
d
itio
n
o
f
th
e
im
ag
e.
Fo
r
d
en
o
is
in
g
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
e
r
20
24
:
6
6
2
8
-
6
6
4
1
6632
im
ag
es,
th
e
non
-
lo
ca
l
m
ea
n
s
f
ilter
is
f
r
eq
u
en
tly
ap
p
lied
.
T
h
e
f
ilter
ed
g
r
a
y
s
ca
le
im
ag
e
(
′
,
′
)
ca
n
b
e
r
ep
r
esen
ted
as th
e
f
ilter
e
d
im
ag
e
(
′
,
′
)
v
ia
NL
M.
(
′
,
′
)
=
1
(
′
,
′
)
∑
(
′
)
⋅
(
′
,
′
)
⋅
(
(
′
)
,
(
′
)
)
′
∈
(
5
)
w
h
er
e
is
th
e
s
ea
r
ch
win
d
o
w
ar
o
u
n
d
th
e
p
ix
el
(
′
,
′
)
;
′
′
r
ep
r
esen
ts
th
e
co
o
r
d
in
ates in
th
e
s
ea
r
ch
wi
n
d
o
w
;
(
′
,
′
)
is
a
s
p
atial
Gau
s
s
ian
weig
h
t
;
(
(
′
)
,
(
′
)
)
is
an
in
ten
s
ity
s
im
ilar
it
y
weig
h
t
;
an
d
(
′
,
′
)
is
a
n
o
r
m
aliza
tio
n
f
a
cto
r
.
T
h
e
s
p
atial
g
au
s
s
ian
weig
h
t
(
′
,
′
)
is
g
iv
en
b
y
(
6
)
.
(
′
,
′
)
=
−
∣
∣
′
−
(
′
,
′
)
∣
∣
2
ℎ
2
(
6
)
w
h
e
r
e
ℎ
c
o
n
t
r
o
l
s
t
h
e
s
p
a
ti
a
l
e
x
te
n
t
o
f
t
h
e
n
e
i
g
h
b
o
r
h
o
o
d
a
n
d
t
h
e
i
n
t
e
n
s
i
t
y
s
i
m
i
la
r
i
t
y
w
e
i
g
h
t
(
(
′
)
,
(
′
)
)
i
s
g
i
v
e
n
b
y
(
7
)
.
(
(
′
)
,
(
′
)
)
=
−
∣
∣
(
′
)
−
(
′
)
∣
∣
2
ℎ
2
(
7
)
wh
er
e
h
in
t c
o
n
tr
o
ls
th
e
in
ten
s
i
ty
s
im
ilar
ity
.
Fin
ally
,
th
e
n
o
r
m
aliza
tio
n
f
ac
to
r
(
′
,
′
)
is
g
iv
en
b
y
(
8
)
.
(
′
,
′
)
=
∑
(
′
,
′
)
⋅
(
(
′
)
,
(
′
)
)
′
∈
(
8
)
T
h
e
NL
M
f
ilter
en
h
an
ce
s
th
e
o
v
er
all
q
u
ality
o
f
an
im
ag
e
an
d
aid
s
in
n
o
is
e
p
r
ed
ictio
n
th
r
o
u
g
h
th
e
in
co
r
p
o
r
atio
n
o
f
n
o
n
-
lo
ca
l
s
im
ilar
ities
p
r
esen
t
in
th
e
im
ag
e.
T
h
e
s
u
b
s
eq
u
en
t
o
p
er
atio
n
f
o
llo
win
g
th
e
ap
p
licatio
n
o
f
th
e
NL
M
f
ilte
r
is
p
ix
el
n
o
r
m
aliza
tio
n
elim
in
atio
n
.
Ph
o
s
p
h
o
r
u
s
v
alu
es
a
r
e
in
ten
d
e
d
to
b
e
en
h
an
ce
d
o
r
n
o
r
m
alize
d
in
ac
co
r
d
an
ce
with
s
p
ec
if
ic
cr
i
ter
ia.
T
h
is
p
r
o
c
ess
s
h
o
u
ld
b
e
r
ep
r
esen
te
d
b
y
m
ath
em
atica
l
eq
u
atio
n
s
.
Giv
e
n
th
e
f
ilter
e
d
im
ag
e
(
′
,
′
)
,
th
e
p
ix
el
n
o
r
m
aliza
tio
n
elim
in
atio
n
ca
n
b
e
ex
p
r
ess
ed
as
(
9
)
.
(
′
,
′
)
=
(
′
,
′
)
−
(
9
)
wh
er
e
n
o
r
m
alize
d
(
′
,
′
)
is
th
e
n
o
r
m
alize
d
p
ix
el
v
alu
e
at
co
o
r
d
in
ates
(
′
,
′
)
;
is
th
e
m
ea
n
p
ix
el
v
alu
e
o
f
th
e
en
tire
im
ag
e
o
r
a
l
o
ca
l
n
ei
g
h
b
o
r
h
o
o
d
;
a
n
d
is
th
e
s
tan
d
a
r
d
d
ev
iatio
n
o
f
p
ix
el
v
alu
es
in
th
e
e
n
tire
im
a
g
e
o
r
a
lo
ca
l n
ei
g
h
b
o
r
h
o
o
d
.
T
h
is
p
r
o
ce
d
u
r
e
f
ac
ilit
ates
th
e
n
o
r
m
aliza
tio
n
o
f
p
ix
el
v
al
u
es,
th
er
eb
y
in
cr
e
asin
g
th
eir
u
n
i
f
o
r
m
ity
an
d
p
o
ten
tially
en
h
an
ci
n
g
th
e
im
a
g
e'
s
o
v
er
all
co
n
tr
ast.
Fu
r
th
er
m
o
r
e,
th
e
f
o
r
m
u
las
f
o
r
d
eter
m
in
in
g
th
e
m
ea
n
μ
an
d
s
tan
d
ar
d
d
e
v
iatio
n
ar
e
as
(
1
0
)
an
d
(
1
1
)
.
=
1
.
∑
∑
(
′
,
′
)
′
=
1
′
=
1
(
1
0
)
=
√
1
.
∑
∑
(
(
′
,
′
)
−
2
)
′
=
1
′
=
1
(
1
1
)
wh
er
e
an
d
ar
e
th
e
d
im
en
s
io
n
s
o
f
th
e
im
ag
e.
T
h
e
eq
u
atio
n
s
p
r
o
v
id
ed
illu
s
tr
ate
th
e
p
ix
el
n
o
r
m
aliza
tio
n
eli
m
in
atio
n
p
r
o
ce
d
u
r
e,
in
wh
ich
th
e
v
alu
e
o
f
ea
ch
p
ix
el
is
n
o
r
m
alize
d
u
s
in
g
th
e
im
ag
e'
s
m
ea
n
an
d
s
tan
d
ar
d
d
e
v
iatio
n
,
o
r
a
n
eig
h
b
o
r
h
o
o
d
'
s
av
er
ag
e.
T
h
e
n
o
r
m
alize
d
im
a
g
e
is
s
u
b
jecte
d
to
th
e
2
D
DW
T
f
o
llo
win
g
th
e
elim
in
atio
n
o
f
p
i
x
e
l
n
o
r
m
a
l
i
z
a
t
i
o
n
.
D
W
T
i
s
a
m
a
t
h
e
m
a
t
i
c
a
l
o
p
e
r
a
t
i
o
n
t
h
a
t
c
o
n
v
e
r
t
s
a
n
i
m
a
g
e
i
n
t
o
i
t
s
c
o
m
p
o
n
e
n
t
f
r
e
q
u
e
n
c
i
e
s
.
T
h
e
2
D
D
W
T
o
f
t
h
e
n
o
r
m
a
l
i
z
e
d
i
m
a
g
e
s
h
a
l
l
b
e
d
e
n
o
t
e
d
a
s
(
′
,
′
)
.
U
n
d
e
r
t
h
e
a
s
s
u
m
p
t
i
o
n
o
f
a
s
o
l
i
t
a
r
y
l
e
v
e
l
o
f
d
e
c
o
m
p
o
s
i
t
i
o
n
,
t
h
e
D
W
T
c
o
e
f
f
i
c
i
e
n
t
s
m
a
y
b
e
c
a
l
c
u
l
a
t
e
d
u
t
i
l
i
z
i
n
g
t
h
e
(
1
2
)
-
(
1
5
)
:
(
′
,
′
)
=
∑
∑
ℎ
(
)
⋅
ℎ
(
)
⋅
(
2
′
−
,
2
′
−
)
(
1
2
)
(
′
,
′
)
=
∑
∑
(
)
⋅
ℎ
(
)
⋅
(
2
′
−
,
2
′
−
)
(
1
3
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
I
mp
leme
n
ta
tio
n
o
f i
n
n
o
v
a
tive
a
p
p
r
o
a
c
h
fo
r
d
etec
tin
g
b
r
a
in
t
u
mo
r
s
in
…
(
A
r
p
ith
a
K
o
tte
)
6633
(
′
,
′
)
=
∑
∑
ℎ
(
)
⋅
(
)
⋅
(
2
′
−
,
2
′
−
)
(
1
4
)
(
′
,
′
)
=
∑
∑
(
)
⋅
(
)
⋅
(
2
′
−
,
2
′
−
)
(
1
5
)
T
h
e
ap
p
r
o
x
im
atio
n
,
h
o
r
izo
n
ta
l
d
etail,
v
er
tical
d
etail,
a
n
d
d
i
ag
o
n
al
d
etail
co
ef
f
icien
ts
ar
e
d
en
o
ted
b
y
,
,
,
an
d
,
r
esp
ec
tiv
ely
.
B
y
d
ec
o
m
p
o
s
in
g
th
e
im
a
g
e
in
to
its
f
r
eq
u
en
c
y
co
m
p
o
n
e
n
ts
,
th
is
p
r
o
ce
d
u
r
e
e
f
f
ec
tiv
ely
ca
p
tu
r
es
in
f
o
r
m
atio
n
at
b
o
th
lo
w
an
d
h
ig
h
f
r
e
q
u
en
cies.
T
o
s
u
m
m
ar
ize,
th
e
m
ath
em
atica
l
f
o
r
m
u
latio
n
s
f
o
r
th
e
two
-
d
im
en
s
io
n
al
DW
T
en
tail
co
n
v
o
lu
tio
n
o
p
er
ati
o
n
s
u
tili
zin
g
f
ilter
co
ef
f
icien
ts
o
f
lo
w
-
p
ass
an
d
h
ig
h
-
p
ass
ty
p
es.
T
h
is
p
r
o
ce
s
s
y
ield
s
d
etail
an
d
ap
p
r
o
x
im
atio
n
c
o
ef
f
icien
ts
th
at
co
r
r
esp
o
n
d
to
d
is
tin
ct
o
r
ien
tatio
n
s
.
T
h
e
s
u
b
s
eq
u
en
t
p
r
o
ce
d
u
r
e,
f
o
llo
win
g
th
e
2
D
DW
T
,
is
th
e
ap
p
licatio
n
o
f
th
e
GL
C
M
tech
n
iq
u
e
f
o
r
f
ea
tu
r
e
ex
tr
ac
tio
n
.
GL
C
M
q
u
an
tifie
s
th
e
s
p
atial
r
elatio
n
s
h
ip
s
b
etwe
en
p
ix
el
v
alu
es
in
an
im
ag
e
as
a
tex
tu
r
e
an
aly
s
is
tech
n
iq
u
e.
Giv
en
th
e
wav
elet
-
tr
an
s
f
o
r
m
ed
im
a
g
e
(
′
,
′
)
,
let's
d
en
o
te
th
e
GL
C
M
as
(
,
)
,
wh
er
e
is
th
e
d
is
tan
ce
an
d
is
th
e
an
g
le
b
etwe
en
p
ix
el
p
ai
r
s
.
T
h
e
GL
C
M
at
d
is
tan
ce
an
d
an
g
le
ca
n
b
e
co
m
p
u
ted
u
s
in
g
as (
1
6
)
:
(
,
)
=
∑
∑
(
′
,
′
)
⋅
(
′
+
,
′
+
)
′
′
′
(
1
6
)
w
h
er
e
=
⋅
(
)
an
d
=
⋅
(
)
.
T
h
e
GL
C
M
is
o
f
ten
n
o
r
m
alize
d
,
an
d
t
h
e
n
o
r
m
alize
d
GL
C
M
is
g
iv
en
b
y
(
1
7
)
.
(
,
,
,
)
=
(
,
)
∑
GL
CM
(
,
)
,
(
1
7
)
At
th
is
ju
n
ctu
r
e,
th
e
GL
C
M
p
r
o
v
id
es
th
e
f
ea
t
u
r
es
ex
p
lain
e
d
b
elo
w,
am
o
n
g
o
t
h
er
tex
tu
r
e
ch
ar
ac
ter
is
tics
th
at
ca
n
b
e
e
x
tr
ac
ted
.
Ass
ig
n
ed
t
o
p
ix
el
v
alu
es
with
in
an
im
ag
e,
th
e
m
ea
n
s
ig
n
if
ies
th
e
av
e
r
ag
e
in
ten
s
ity
.
T
h
e
m
ea
n
o
f
th
e
lu
m
in
a
n
ce
lev
els
in
an
MRI
s
ca
n
o
f
th
e
b
r
ain
ca
n
p
o
ten
tially
y
ield
in
s
ig
h
ts
in
to
th
e
r
elativ
e
ab
u
n
d
a
n
ce
o
f
v
ar
io
u
s
tis
s
u
e
ty
p
es.
On
e
illu
s
tr
atio
n
o
f
th
is
is
h
o
w
d
is
cr
e
p
an
cies
in
av
er
a
g
e
in
ten
s
ity
co
u
ld
p
o
ten
tially
s
ig
n
if
y
d
is
tin
ct
ce
r
eb
r
al
ar
ch
itectu
r
es o
r
p
ath
o
lo
g
ies.
=
∑
⋅
(
,
,
,
)
,
(
1
8
)
T
h
e
s
tan
d
ar
d
d
ev
iatio
n
q
u
an
ti
f
ies
th
e
d
is
p
er
s
io
n
o
r
s
p
r
ea
d
o
f
th
e
in
ten
s
ities
o
f
in
d
i
v
id
u
a
l
p
ix
els.
A
h
ig
h
s
tan
d
ar
d
d
ev
iatio
n
in
a
b
r
ain
MRI
s
ca
n
m
ay
s
er
v
e
as
an
in
d
icato
r
o
f
r
e
g
io
n
s
ex
h
ib
itin
g
d
iv
er
s
e
in
ten
s
ities
,
p
o
ten
tially
s
ig
n
if
y
in
g
ab
n
o
r
m
alities
o
r
f
lu
ctu
ati
o
n
s
in
tis
s
u
e
ch
ar
ac
ter
is
tics
.
R
eg
io
n
s
with
a
lo
w
s
tan
d
ar
d
d
e
v
iatio
n
m
ay
b
e
in
d
icativ
e
o
f
h
o
m
o
g
e
n
eity
.
=
√
∑
(
−
)
2
⋅
(
,
,
,
)
,
(
1
9
)
E
n
tr
o
p
y
q
u
a
n
tifie
s
th
e
u
n
ce
r
t
ain
ty
o
r
r
an
d
o
m
n
ess
o
f
th
e
p
i
x
el
v
alu
es.
Hig
h
en
tr
o
p
y
o
n
b
r
ain
MRI
s
ca
n
s
m
ay
in
d
icate
r
eg
io
n
s
w
ith
m
ix
ed
tis
s
u
e
ty
p
es
o
r
c
o
m
p
lex
s
tr
u
ctu
r
es,
w
h
er
ea
s
lo
w
en
tr
o
p
y
m
ay
im
p
ly
m
o
r
e
u
n
if
o
r
m
r
eg
io
n
s
.
T
h
e
e
n
tr
o
p
y
o
f
ce
r
e
b
r
al
tis
s
u
e
ca
n
b
e
u
tili
ze
d
to
q
u
a
n
tify
its
co
m
p
le
x
ity
an
d
te
x
tu
r
e.
=
−
∑
(
,
,
,
)
⋅
(
(
,
,
,
)
+
)
,
(
2
0
)
T
h
e
v
ar
ian
ce
o
f
a
p
ix
el'
s
in
te
n
s
ity
r
ep
r
esen
ts
its
d
eg
r
ee
o
f
v
ar
iab
ilit
y
.
Hig
h
v
ar
ian
ce
in
b
r
ain
MRI
s
ca
n
s
m
ay
in
d
icate
r
eg
io
n
s
c
o
n
tain
in
g
a
v
a
r
iety
o
f
tis
s
u
e
ty
p
es
o
r
p
ath
o
lo
g
ical
c
o
n
d
it
io
n
s
,
wh
er
ea
s
lo
w
v
ar
ian
ce
m
a
y
s
u
g
g
est
r
eg
io
n
s
th
at
ar
e
m
o
r
e
h
o
m
o
g
en
e
o
u
s
.
T
h
e
in
f
o
r
m
atio
n
p
r
o
v
id
e
d
b
y
th
e
m
ea
n
an
d
s
tan
d
ar
d
d
e
v
iatio
n
is
s
u
p
p
lem
en
ted
b
y
v
ar
ian
ce
.
=
∑
,
(
−
)
2
⋅
(
,
,
,
)
(
2
1
)
A
m
ea
s
u
r
e
o
f
th
e
s
p
atial
u
n
i
f
o
r
m
ity
o
f
p
ix
el
in
ten
s
ities
,
s
m
o
o
th
n
ess
q
u
an
tifie
s
th
is
.
W
h
en
u
s
in
g
b
r
ain
MRI,
a
h
i
g
h
s
m
o
o
th
n
es
s
v
alu
e
in
d
icate
s
h
o
m
o
g
en
e
o
u
s
r
eg
io
n
s
,
wh
ich
ar
e
d
e
f
in
ed
as
ad
jace
n
t
p
ix
els
with
co
m
p
ar
ab
le
in
ten
s
ities
.
T
h
is
ca
n
f
ac
ilit
ate
th
e
id
en
tific
atio
n
o
f
r
eg
io
n
s
co
n
tain
in
g
tis
s
u
e
with
co
n
s
is
ten
t
p
r
o
p
er
ties
.
ℎ
=
1
−
1
1
+
∑
(
−
)
2
⋅
(
,
,
,
)
,
(
2
2
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
e
r
20
24
:
6
6
2
8
-
6
6
4
1
6634
C
o
n
tr
ast
q
u
an
tifie
s
th
e
d
is
p
ar
ity
in
th
e
in
ten
s
ities
o
f
ad
jac
en
t
p
ix
els.
E
lev
ated
co
n
tr
ast
v
alu
es
in
b
r
ain
MRI
s
ca
n
s
ca
n
d
is
tin
g
u
is
h
b
o
u
n
d
a
r
ies
b
etwe
en
d
is
tin
ct
tis
s
u
e
ty
p
es
o
r
s
tr
u
ctu
r
es
.
T
h
er
eb
y
f
ac
ilit
atin
g
th
e
d
etec
tio
n
o
f
p
ath
o
lo
g
ical
r
eg
io
n
s
o
r
a
n
ato
m
ical
f
ea
tu
r
es.
=
∑
(
−
)
2
⋅
(
,
,
,
)
,
(
2
3
)
T
h
e
lin
ea
r
r
elatio
n
s
h
ip
b
etwe
en
p
ix
el
in
ten
s
ities
in
o
p
p
o
s
ite
d
ir
ec
tio
n
s
is
q
u
an
tifie
d
b
y
c
o
r
r
elatio
n
.
C
o
r
r
elatio
n
in
b
r
ai
n
MRI
ca
n
o
f
f
er
v
alu
ab
le
in
s
ig
h
ts
r
eg
a
r
d
i
n
g
th
e
ar
r
an
g
em
e
n
t a
n
d
c
o
n
g
r
u
en
ce
o
f
a
n
ato
m
ical
co
m
p
o
n
en
ts
.
Fo
r
in
s
tan
ce
,
a
lo
w
co
r
r
elatio
n
m
ay
in
d
icat
e
th
at
f
ea
t
u
r
es
ar
e
n
o
t
alig
n
ed
,
wh
er
ea
s
a
h
i
g
h
co
r
r
elatio
n
m
a
y
in
d
icate
th
at
f
ea
tu
r
es a
r
e
alig
n
ed
.
=
∑
,
(
−
Mea
n
)
⋅
(
−
Mea
n
)
⋅
G
L
C
M
(
,
,
,
)
S
t
d
D
ev
⋅
S
t
d
D
ev
(
2
4
)
T
h
e
en
er
g
y
o
f
th
e
GL
C
M
is
eq
u
al
to
th
e
ag
g
r
eg
ate
o
f
th
e
s
q
u
ar
ed
elem
en
ts
.
W
ith
in
th
e
d
o
m
ain
o
f
b
r
ain
MRI
s
ca
n
s
,
elev
ated
e
n
er
g
y
v
al
u
es
s
ig
n
if
y
a
g
r
ea
ter
lik
elih
o
o
d
o
f
s
p
ec
if
ic
p
ix
el
p
air
s
o
cc
u
r
r
in
g
.
T
h
is
m
ay
b
e
c
o
r
r
elate
d
with
r
e
g
io
n
s
o
f
u
n
if
o
r
m
tex
t
u
r
e
o
r
d
is
tin
ct
p
atter
n
s
p
r
esen
t in
t
h
e
im
ag
e.
=
∑
,
G
L
C
M
(
,
,
,
)
2
(
2
5
)
T
h
e
a
f
o
r
e
m
en
tio
n
ed
f
ea
tu
r
e
s
ar
e
p
r
e
s
en
tly
s
u
p
p
l
ied
to
th
e
d
ee
p
le
ar
n
i
n
g
m
o
d
el
Ne
u
r
o
Fu
s
i
o
n
N
et,
a
h
y
b
r
id
d
ee
p
en
co
d
er
-
U
-
Ne
t
m
o
d
el
th
at
ai
m
s
t
o
a
ch
i
ev
e
ac
cu
r
at
e
b
r
ain
tu
m
o
r
s
eg
m
en
ta
tio
n
.
T
h
e
in
i
ti
al
p
h
a
s
e
o
f
th
e
d
ee
p
lea
r
n
in
g
m
o
d
el
,
N
eu
r
o
Fu
s
io
n
Ne
t
i
s
d
ed
ica
ted
to
im
ag
e
c
la
s
s
if
i
ca
tio
n
u
ti
li
zin
g
a
h
y
b
r
id
en
co
d
er
-
d
ec
o
d
e
r
s
tr
u
ctu
r
e.
Fo
r
im
ag
e
cla
s
s
if
i
ca
tio
n
,
a
t
ar
g
e
t
c
la
s
s
i
f
i
ca
t
io
n
an
d
a
to
ler
an
c
e
lev
el
ar
e
es
ta
b
l
is
h
ed
.
Af
t
er
d
ete
r
m
i
n
in
g
th
e
d
im
en
s
io
n
s
o
f
t
h
e
im
p
o
r
te
d
i
m
ag
es
,
t
h
ey
ar
e
r
e
s
i
ze
d
to
a
s
q
u
ar
e
f
o
r
m
a
t
to
en
s
u
r
e
u
n
if
o
r
m
i
ty
.
Fo
llo
win
g
th
is
,
th
e
m
o
d
el
is
p
u
t
th
r
o
u
g
h
th
e
p
r
o
ce
s
s
o
f
t
r
a
in
in
g
a
d
ee
p
a
u
to
en
co
d
er
.
E
n
co
d
er
a
n
d
d
ec
o
d
er
co
m
p
o
n
e
n
ts
o
f
th
e
a
r
ch
itectu
r
e
h
a
v
e
b
ee
n
tr
ai
n
ed
t
o
r
ec
o
g
n
ize
c
o
m
p
r
ess
ed
r
ep
r
e
s
en
tatio
n
s
o
f
in
p
u
t
im
ag
es.
R
ec
o
n
s
tr
u
ctio
n
lo
s
s
is
r
ed
u
ce
d
wh
en
e
v
er
p
o
s
s
ib
le
d
u
r
in
g
tr
ai
n
in
g
.
Af
ter
t
h
e
au
t
o
en
co
d
e
r
h
as
b
ee
n
tr
ain
ed
,
p
i
x
el
s
eg
m
en
tatio
n
is
ex
ec
u
ted
u
tili
zin
g
a
U
-
Net
t
h
at
in
co
r
p
o
r
ates
a
d
ee
p
en
c
o
d
er
in
to
its
p
at
h
o
f
co
n
v
er
g
en
ce
.
Fu
r
th
er
e
n
co
d
i
n
g
lay
er
s
ar
e
in
co
r
p
o
r
ated
in
to
th
e
U
-
Net
ar
c
h
itectu
r
e
in
o
r
d
er
to
a
u
g
m
e
n
t
its
ca
p
ab
ilit
y
o
f
ca
p
tu
r
in
g
co
m
p
l
ex
f
ea
tu
r
es.
T
h
e
ca
p
ab
ilit
ies
o
f
th
e
m
o
d
el
ar
e
s
u
b
s
eq
u
en
tl
y
in
v
esti
g
ated
with
r
esp
ec
t
to
co
lo
r
q
u
an
tizatio
n
a
n
d
v
is
u
aliza
tio
n
.
An
alg
o
r
ith
m
is
em
p
lo
y
ed
to
q
u
an
tify
th
e
c
o
lo
r
o
f
an
im
ag
e
to
a
s
et
o
f
s
ix
teen
d
is
tin
ct
h
u
es;
th
e
r
esu
ltan
t
im
ag
e
is
v
is
u
ally
p
r
esen
ted
in
o
r
d
er
to
o
f
f
er
a
n
u
n
d
e
r
s
tan
d
in
g
o
f
th
e
d
im
in
is
h
ed
co
l
o
r
p
alette.
Fo
llo
win
g
th
is
,
th
e
im
p
lem
e
n
tatio
n
o
f
co
lo
r
s
p
ac
e
co
n
v
er
s
io
n
an
d
s
o
f
t
m
ax
e
n
tr
o
p
y
clas
s
if
icatio
n
o
cc
u
r
s
.
An
a
d
d
itio
n
al
im
ag
e
is
im
p
o
r
ted
an
d
s
u
b
s
eq
u
e
n
tly
co
n
v
er
ted
to
a
d
is
tin
ct
co
lo
r
r
ep
r
esen
tatio
n
.
T
h
e
s
o
f
tm
ax
en
tr
o
p
y
m
eth
o
d
is
s
u
b
s
eq
u
en
tly
im
p
lem
en
ted
to
cl
ass
if
y
p
ix
els,
with
p
r
o
b
a
b
ilit
ies
b
ein
g
ass
ig
n
ed
to
ea
ch
class
ac
co
r
d
in
g
to
th
e
co
l
o
r
in
f
o
r
m
atio
n
o
f
th
e
p
ix
els.
Ou
r
p
r
o
p
o
s
ed
m
et
h
o
d
o
lo
g
y
ad
d
r
ess
es
k
ey
lim
itatio
n
s
in
b
r
ai
n
tu
m
o
r
s
eg
m
en
tatio
n
a
n
d
cla
s
s
if
icatio
n
b
y
in
cr
ea
s
in
g
im
ag
e
q
u
ality
r
o
b
u
s
tn
ess
,
d
ec
r
ea
s
in
g
r
elian
ce
o
n
lab
eled
d
ata,
o
p
tim
i
zin
g
co
m
p
u
tatio
n
al
ef
f
icien
cy
,
an
d
im
p
r
o
v
in
g
in
te
r
p
r
etab
ilit
y
.
W
e
u
s
e
a
NL
M
f
il
ter
an
d
p
ix
el
n
o
r
m
aliza
tio
n
elim
in
atio
n
to
r
ed
u
ce
n
o
is
e
an
d
im
p
r
o
v
e
co
n
t
r
ast,
m
ak
in
g
th
e
m
o
d
el
less
s
en
s
itiv
e
to
im
ag
e
v
a
r
iatio
n
s
.
T
h
e
d
ee
p
au
to
en
co
d
er
in
t
h
e
Neu
r
o
Fu
s
io
n
Net
m
o
d
el
r
e
d
u
ce
s
r
elian
ce
o
n
lar
g
e
am
o
u
n
ts
o
f
lab
eled
d
ata
b
y
lea
r
n
in
g
co
m
p
r
ess
ed
r
ep
r
esen
tatio
n
s
,
r
esu
ltin
g
in
b
etter
g
e
n
er
aliza
tio
n
.
T
h
e
co
m
b
in
atio
n
o
f
a
d
ee
p
en
c
o
d
er
a
n
d
a
U
-
Net
ar
ch
itectu
r
e
o
p
tim
izes
co
m
p
u
tatio
n
al
ef
f
icien
c
y
a
n
d
ac
cu
r
ac
y
,
allo
win
g
f
o
r
p
r
ec
is
e
s
eg
m
en
tatio
n
with
o
u
t
ex
ce
ed
in
g
c
o
m
p
u
tatio
n
al
d
em
an
d
s
.
Deta
iled
f
ea
tu
r
e
ex
tr
ac
tio
n
u
s
in
g
GL
C
M
a
n
aly
s
is
im
p
r
o
v
es
in
ter
p
r
etab
ilit
y
b
y
p
r
o
v
id
i
n
g
in
f
o
r
m
atio
n
ab
o
u
t
tex
tu
r
e
an
d
s
p
atial
r
elatio
n
s
h
ip
s
.
T
h
is
co
m
p
r
eh
en
s
iv
e
ap
p
r
o
ac
h
y
ield
s
a
r
o
b
u
s
t,
ef
f
i
cien
t,
an
d
in
ter
p
r
eta
b
le
s
o
lu
tio
n
th
at
im
p
r
o
v
es
d
iag
n
o
s
tic
p
r
ec
is
io
n
an
d
aid
s
in
h
ea
lth
ca
r
e
d
ec
is
io
n
-
m
ak
in
g
.
3
.
1
.
P
r
o
po
s
ed
Neuro
F
us
io
n
Net
m
o
del
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
as
s
h
o
w
n
in
Fig
u
r
e
2
in
teg
r
ates
a
d
e
ep
en
c
o
d
er
an
d
a
U
-
Net
ar
ch
itectu
r
e
to
ad
d
r
ess
im
ag
e
p
r
o
ce
s
s
in
g
ch
al
len
g
es,
with
a
p
ar
ticu
lar
f
o
cu
s
o
n
b
r
ain
tu
m
o
r
class
if
icatio
n
.
B
y
ca
p
italizin
g
o
n
th
e
r
esp
ec
tiv
e
s
tr
en
g
th
s
o
f
b
o
th
co
m
p
o
n
en
ts
,
th
is
in
teg
r
atio
n
s
ee
k
s
to
im
p
r
o
v
e
th
e
ac
cu
r
ac
y
o
f
f
ea
t
u
r
e
ex
tr
ac
tio
n
an
d
s
eg
m
en
tatio
n
.
By
em
p
lo
y
in
g
a
d
ee
p
en
c
o
d
e
r
,
co
m
p
lex
h
ier
a
r
ch
ical
f
ea
tu
r
es
ca
n
b
e
ex
tr
ac
ted
f
r
o
m
th
e
in
p
u
t
im
ag
es,
en
ab
li
n
g
th
e
ca
p
tu
r
e
o
f
m
in
u
te
d
etails
th
at
m
ay
b
e
cr
itical
f
o
r
p
r
ec
is
e
clas
s
if
icatio
n
.
I
n
ad
d
itio
n
,
th
e
U
-
Net'
s
in
tr
in
s
ic
ca
p
ab
ilit
y
f
o
r
p
ix
el
-
wis
e
s
eg
m
en
tatio
n
f
ac
ilit
ates
th
e
r
ec
o
n
s
tr
u
ctio
n
o
f
h
ig
h
-
r
eso
lu
tio
n
s
eg
m
en
tatio
n
m
as
k
s
v
ia
a
s
tr
u
ctu
r
ed
d
ec
o
d
in
g
p
ath
.
T
h
e
in
co
r
p
o
r
atio
n
o
f
a
d
ee
p
au
to
en
co
d
er
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
I
mp
leme
n
ta
tio
n
o
f i
n
n
o
v
a
tive
a
p
p
r
o
a
c
h
fo
r
d
etec
tin
g
b
r
a
in
t
u
mo
r
s
in
…
(
A
r
p
ith
a
K
o
tte
)
6635
en
h
an
ce
s
th
e
r
o
b
u
s
tn
ess
o
f
f
ea
tu
r
e
lear
n
in
g
b
y
e
n
ab
lin
g
a
m
o
r
e
ac
cu
r
ate
d
ep
ictio
n
o
f
in
tr
i
ca
te
p
atter
n
s
with
in
th
e
im
ag
e
d
ata.
B
y
ca
p
italizin
g
o
n
th
e
co
m
b
in
e
d
s
tr
en
g
th
s
o
f
th
ese
co
m
p
o
n
en
ts
,
th
e
ar
ch
it
ec
tu
r
e
o
f
th
e
m
o
d
el
p
r
o
v
id
es a
c
o
m
p
r
e
h
en
s
iv
e
s
o
l
u
tio
n
f
o
r
class
if
icatio
n
an
d
im
ag
e
p
r
o
ce
s
s
in
g
task
s
.
T
h
e
h
y
b
r
id
m
o
d
el
d
ev
elo
p
e
d
b
y
co
m
b
in
in
g
a
d
ee
p
en
c
o
d
er
an
d
a
U
-
Net
ar
ch
itectu
r
e
is
wh
at
d
is
tin
g
u
is
h
es
th
e
p
r
o
p
o
s
ed
m
o
d
el
an
d
en
ab
les
it
to
ef
f
ec
tiv
ely
tack
le
th
e
c
o
m
p
lex
ities
o
f
im
ag
e
p
r
o
ce
s
s
in
g
task
s
.
B
y
in
teg
r
atin
g
a
d
ee
p
en
co
d
er
,
th
e
m
o
d
el'
s
ab
ilit
y
to
d
etec
t
s
u
b
tle
ch
ar
ac
ter
i
s
tics
i
s
im
p
r
o
v
ed
,
f
ac
ilit
atin
g
th
e
ac
q
u
is
itio
n
o
f
n
u
an
ce
d
d
ata
th
at
is
v
ital
f
o
r
p
r
ec
is
e
class
if
icatio
n
.
T
h
e
in
teg
r
atio
n
o
f
th
is
co
m
b
in
atio
n
e
n
h
an
ce
s
th
e
U
-
Net
f
r
am
ewo
r
k
with
s
o
p
h
is
ticated
f
ea
tu
r
e
ex
t
r
ac
tio
n
f
u
n
ct
io
n
alities
,
th
er
eb
y
in
tr
o
d
u
cin
g
a
f
r
esh
v
iewp
o
in
t.
Fu
r
th
er
m
o
r
e,
th
e
in
c
o
r
p
o
r
atio
n
o
f
a
d
ee
p
a
u
to
en
c
o
d
er
p
r
ese
n
ts
a
n
o
v
el
asp
ec
t
th
at
en
h
an
ce
s
th
e
ad
ap
tab
ilit
y
an
d
ef
f
icac
y
o
f
th
e
m
o
d
el
wh
e
n
ac
q
u
ir
in
g
k
n
o
wled
g
e
o
f
co
m
p
lex
p
atter
n
s
.
T
h
is
h
as
th
e
p
o
te
n
tial
to
en
h
an
ce
d
iag
n
o
s
tic
p
r
ec
is
io
n
a
n
d
ass
is
t
h
ea
lth
ca
r
e
p
r
ac
titi
o
n
er
s
in
m
a
k
in
g
well
-
in
f
o
r
m
ed
ju
d
g
m
en
ts
.
I
n
g
en
er
al,
th
e
p
r
o
p
o
s
ed
m
o
d
el
em
b
o
d
ies
a
n
in
n
o
v
ativ
e
an
d
m
u
tu
ally
b
en
ef
ici
al
am
alg
am
atio
n
o
f
d
ee
p
lear
n
in
g
elem
en
ts
c
u
s
to
m
ized
to
ad
d
r
ess
th
e
p
ar
ticu
l
ar
co
m
p
lex
ities
ass
o
ciate
d
w
ith
m
ed
ical
im
ag
e
class
if
icatio
n
an
d
p
r
o
ce
s
s
in
g
.
Fig
u
r
e
2
.
Pro
p
o
s
ed
Neu
r
o
Fu
s
io
n
Net
m
o
d
el
Alg
o
r
ith
m
.
Neu
r
o
Fu
s
io
n
Net
m
o
d
el
#
Step
1
: I
m
a
g
e
class
if
icatio
n
with
U
-
Net
an
d
lo
ad
in
g
_
=
.
.
.
=
.
.
.
=
_
(
.
.
.
)
_
=
_
_
(
)
#
Step
2
: T
r
ain
i
n
g
o
f
d
ee
p
au
t
o
en
co
d
e
r
_
=
.
.
.
_
=
_
_
(
_
,
_
)
#
Step
3
: Pix
el
s
eg
m
en
tatio
n
u
s
in
g
U
-
Net
_
=
_
(
.
.
.
)
_
=
_
(
_
)
#
Step
4
: V
is
u
aliza
tio
n
af
ter
c
o
lo
r
q
u
an
tizatio
n
_
=
_
(
_
,
_
=
16
)
_
(
_
)
#
Step
5
: Co
n
v
er
s
io
n
o
f
c
o
lo
r
s
p
ac
e
an
d
s
o
f
tm
ax
en
tr
o
p
y
cla
s
s
if
icatio
n
_
=
_
(
.
.
.
)
_
=
_
_
(
_
)
_
=
_
_
(
_
)
#
Step
6
: Bra
in
tu
m
o
r
class
if
icatio
n
with
co
m
b
i
n
ed
d
ee
p
en
c
o
d
er
an
d
U
-
Net
_
,
=
_
_
(
.
.
.
)
_
=
_
_
_
(
)
_
=
_
_
(
_
,
_
,
)
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
I
n
Fig
u
r
e
3
,
th
e
in
p
u
t
im
ag
e
(
(
,
)
)
f
r
o
m
th
e
B
R
AT
S
d
atab
ase
is
o
b
tain
ed
f
ir
s
t,
r
e
p
r
esen
tin
g
a
n
o
r
ig
in
al
b
r
ain
s
ca
n
.
T
h
is
im
a
g
e'
s
d
im
en
s
io
n
s
ar
e
th
en
s
ca
led
to
5
1
2
×
5
1
2
,
r
esu
ltin
g
in
F
ig
u
r
e
4
,
th
e
r
esized
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
e
r
20
24
:
6
6
2
8
-
6
6
4
1
6636
im
ag
e
(
(
′
,
′
)
)
.
Du
r
in
g
th
e
r
esizin
g
p
r
o
ce
s
s
,
b
ilin
ea
r
i
n
ter
p
o
latio
n
i
s
u
s
ed
to
o
b
tain
th
e
n
ew
s
p
at
ial
co
o
r
d
in
ates
(
′
,
′
)
u
s
in
g
a
s
et
o
f
in
(
1
)
.
T
h
is
en
lar
g
ed
im
ag
e
will
b
e
th
e
s
tar
tin
g
p
o
in
t
f
o
r
ad
d
itio
n
al
p
r
o
ce
s
s
in
g
p
r
o
ce
d
u
r
es.
Mo
v
in
g
o
n
t
o
Fig
u
r
es
5
an
d
6
,
th
e
s
ca
led
im
ag
e
is
co
lo
r
-
m
ap
p
ed
to
c
o
n
v
e
r
t
it
to
g
r
a
y
s
ca
le.
As
s
h
o
wn
in
(
4
)
,
th
e
co
lo
r
m
ap
p
in
g
ap
p
r
o
ac
h
u
s
ed
is
a
lu
m
i
n
an
ce
-
p
r
eser
v
i
n
g
tr
a
n
s
f
er
f
r
o
m
R
GB
to
g
r
ay
s
ca
le.
T
h
e
weig
h
ts
0
.
2
9
9
,
0
.
5
8
7
,
an
d
0
.
1
1
4
ar
e
ap
p
lied
to
th
e
s
h
r
u
n
k
im
ag
e'
s
r
ed
,
g
r
ee
n
,
a
n
d
b
l
u
e
ch
an
n
els,
r
esu
ltin
g
in
co
lo
r
-
m
ap
p
e
d
p
ictu
r
es
(
(
′
,
′
)
)
.
T
h
ese
g
r
ay
s
ca
le
r
ep
r
esen
tatio
n
s
ar
e
cr
itical
f
o
r
t
h
e
d
e
n
o
is
in
g
an
d
f
ea
tu
r
e
ex
tr
ac
tio
n
p
r
o
ce
s
s
es th
at
f
o
llo
w.
Fig
u
r
e
3
.
I
n
p
u
t
i
m
ag
e
Fig
u
r
e
4
.
R
esized
i
m
ag
e
Fig
u
r
e
5
.
C
o
lo
r
m
ap
p
e
d
i
m
ag
e
Fig
u
r
e
6
.
C
o
lo
r
m
ap
p
e
d
i
m
ag
e
T
h
e
ef
f
ec
t
o
f
a
p
p
ly
in
g
th
e
NL
M
f
ilter
to
th
e
g
r
a
y
s
ca
le
im
ag
e
is
s
h
o
wn
in
Fig
u
r
e
7
,
wh
ich
r
esu
lts
in
th
e
NL
M
f
ilter
ed
o
u
tp
u
t
(
(
′
,
′
)
)
.
T
h
e
NL
M
f
ilter
,
as
d
ef
in
ed
in
(
5
)
,
im
p
r
o
v
es
im
ag
e
q
u
ality
an
d
d
ec
r
ea
s
es
n
o
is
e
b
y
tak
i
n
g
n
o
n
-
lo
ca
l
s
im
ilar
ities
in
th
e
im
ag
e
in
to
ac
co
u
n
t.
T
h
is
d
en
o
is
in
g
p
h
ase
is
cr
itical
f
o
r
in
cr
ea
s
in
g
th
e
p
r
ec
is
io
n
o
f
s
u
b
s
eq
u
en
t a
n
aly
s
is
.
I
n
Fig
u
r
e
8
,
th
e
NL
M
-
f
ilter
ed
im
ag
e
is
s
u
b
jecte
d
to
th
e
p
ix
el
n
o
r
m
aliza
tio
n
elim
in
atio
n
m
eth
o
d
.
In
(
9
)
s
h
o
ws
h
o
w
to
n
o
r
m
alize
ea
ch
p
ix
el
in
th
e
NL
M
-
f
ilter
ed
im
ag
e
u
s
in
g
th
e
m
ea
n
an
d
s
tan
d
ar
d
d
ev
iatio
n
o
f
th
e
en
tire
im
ag
e
o
r
a
lo
ca
l
n
eig
h
b
o
r
h
o
o
d
.
T
h
e
g
o
al
o
f
th
is
n
o
r
m
aliza
tio
n
is
to
im
p
r
o
v
e
p
i
x
el
u
n
if
o
r
m
ity
an
d
o
v
er
all
im
ag
e
c
o
n
tr
ast.
I
n
Fig
u
r
e
9
,
th
e
f
ea
tu
r
e
ex
tr
ac
tio
n
p
h
ase
em
p
lo
y
s
th
e
tw
o
-
d
im
en
s
io
n
al
DW
T
o
n
t
h
e
n
o
r
m
alize
d
p
ictu
r
e,
y
ield
i
n
g
(
′
,
′
)
.
I
n
(
1
2
)
-
(
1
5
)
d
escr
ib
e
th
e
im
a
g
e'
s
d
ec
o
m
p
o
s
itio
n
in
to
ap
p
r
o
x
im
atio
n
an
d
d
etail
co
ef
f
icien
ts
.
T
h
e
GL
C
M
ap
p
r
o
ac
h
is
th
en
u
s
ed
to
ex
tr
ac
t
t
ex
tu
r
e
in
f
o
r
m
atio
n
f
r
o
m
th
e
wav
elet
-
tr
an
s
f
o
r
m
ed
im
ag
e,
as
d
escr
ib
e
d
i
n
(
1
6
)
.
T
h
e
n
o
r
m
alize
d
GL
C
M
(
(
,
,
,
)
)
is
th
en
co
m
p
u
ted
,
wh
ich
p
r
o
v
i
d
e
s
m
e
a
n
,
s
t
a
n
d
a
r
d
d
e
v
i
a
t
i
o
n
,
e
n
t
r
o
p
y
,
v
a
r
i
a
n
c
e
,
s
m
o
o
t
h
n
e
s
s
,
c
o
n
t
r
a
s
t
,
c
o
r
r
e
l
a
t
i
o
n
,
a
n
d
e
n
e
r
g
y
i
n
f
o
r
m
a
t
i
o
n
.
Fin
ally
,
Fig
u
r
e
1
0
s
h
o
ws
th
e
Neu
r
o
Fu
s
io
n
Net
m
o
d
el'
s
o
u
tp
u
t,
wh
ich
s
h
o
ws
th
e
d
is
co
v
er
e
d
ar
ea
o
f
a
b
r
ain
tu
m
o
r
.
Fo
r
p
ix
el
s
eg
m
en
tatio
n
,
th
e
d
ee
p
lear
n
in
g
m
o
d
el
in
co
r
p
o
r
ates
a
h
y
b
r
id
en
co
d
er
-
d
ec
o
d
er
s
tr
u
ctu
r
e,
a
d
ee
p
au
to
e
n
co
d
er
,
an
d
a
U
-
Net.
T
h
e
m
o
d
el
is
tau
g
h
t
to
r
ec
o
g
n
ize
co
m
p
r
ess
ed
r
ep
r
esen
tatio
n
s
o
f
in
p
u
t im
ag
es a
n
d
th
en
s
eg
m
en
ted
b
r
ain
tu
m
o
r
s
.
T
h
e
m
o
d
el
th
at
was
p
r
o
p
o
s
e
d
was
ev
alu
ated
u
s
in
g
a
d
ata
s
et
co
n
s
is
tin
g
o
f
o
n
e
th
o
u
s
an
d
B
R
AT
S
d
atab
ase
[
2
5
]
MRI
s
ca
n
s
.
I
n
o
r
d
er
to
e
n
h
an
ce
th
e
ef
f
icac
y
o
f
th
e
tr
ai
n
in
g
p
r
o
ce
s
s
,
th
e
m
o
d
el
was
tr
ain
e
d
u
s
in
g
8
0
%
o
f
th
e
im
ag
es,
w
h
ich
allo
wed
it
to
co
m
p
r
eh
e
n
d
th
e
in
tr
icate
d
etails
an
d
f
u
n
d
am
e
n
tal
p
atter
n
s
p
r
esen
t
in
th
e
d
ata.
A
s
u
b
s
et
o
f
2
0
%
o
f
th
e
d
ataset,
co
m
p
r
is
in
g
2
0
0
im
ag
es,
was
s
et
asid
e
f
o
r
ex
h
au
s
tiv
e
test
in
g
.
B
y
s
eg
r
eg
atin
g
th
e
d
ata,
a
c
o
m
p
r
e
h
en
s
iv
e
ass
ess
m
en
t
o
f
t
h
e
m
o
d
el'
s
ef
f
icac
y
o
n
n
o
v
el
an
d
u
n
o
b
s
er
v
e
d
in
f
o
r
m
atio
n
was f
ea
s
ib
le,
th
er
eb
y
g
u
ar
an
teei
n
g
i
ts
ca
p
ac
ity
f
o
r
s
o
u
n
d
g
en
er
aliz
atio
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
I
mp
leme
n
ta
tio
n
o
f i
n
n
o
v
a
tive
a
p
p
r
o
a
c
h
fo
r
d
etec
tin
g
b
r
a
in
t
u
mo
r
s
in
…
(
A
r
p
ith
a
K
o
tte
)
6637
Fig
u
r
e.
7
NL
M
f
ilter
e
d
o
u
t
p
u
t
Fig
u
r
e
8
.
R
e
s
u
l
t
a
n
t
o
f
p
i
x
e
l
n
o
r
m
a
l
i
z
a
t
i
o
n
e
l
i
m
i
n
a
t
i
o
n
Fig
u
r
e
9
.
Featu
r
e
ex
tr
ac
tio
n
u
s
in
g
GL
C
M
Fig
u
r
e
10
.
B
r
ain
tu
m
o
r
d
etec
t
ed
ar
ea
A
co
m
p
r
eh
e
n
s
iv
e
s
u
m
m
ar
y
o
f
th
e
f
ea
tu
r
es
ex
tr
ac
ted
u
til
izin
g
th
e
c
o
m
p
lex
ity
f
ea
tu
r
e
ex
tr
ac
tio
n
tech
n
iq
u
e
is
p
r
o
v
i
d
ed
i
n
T
ab
le
1
o
f
th
e
s
tu
d
y
.
T
h
e
c
o
m
p
u
ted
f
ea
tu
r
es
wer
e
ap
p
lied
to
a
s
u
b
s
et
o
f
f
iv
e
s
am
p
les
th
at
wer
e
d
elib
er
ately
ch
o
s
e
n
f
r
o
m
th
e
lar
g
er
d
ataset.
B
y
u
tili
zin
g
th
is
tar
g
eted
s
u
b
s
et,
an
ex
h
au
s
tiv
e
ev
alu
atio
n
o
f
t
h
e
m
o
d
el'
s
p
er
f
o
r
m
an
ce
ca
n
b
e
co
n
d
u
cte
d
o
n
th
ese
p
ar
ticu
lar
s
am
p
les,
y
ield
in
g
s
ig
n
if
ican
t
in
s
ig
h
ts
in
to
its
ap
titu
d
e
f
o
r
d
e
tectin
g
m
in
u
te
in
tr
icac
ies in
M
R
I
s
ca
n
im
ag
es.
T
ab
le
1
.
E
x
tr
ac
ted
f
ea
tu
r
es o
f
b
r
ain
MR sam
p
le
im
ag
es
B
R
A
TS
d
a
t
a
b
a
se
s
a
m
p
l
e
s
F
e
a
t
u
r
e
s
e
x
t
r
a
c
t
e
d
S
a
mp
l
e
1
S
a
mp
l
e
2
S
a
mp
l
e
3
S
a
mp
l
e
4
S
a
mp
l
e
5
M
e
a
n
0
.
0
0
3
9
0
.
0
0
4
2
3
0
.
0
0
4
1
0
.
0
0
4
2
1
0
.
0
0
3
9
3
S
t
a
n
d
a
r
d
d
e
v
i
a
t
i
o
n
0
.
0
8
2
4
1
0
.
0
8
8
9
7
0
.
0
8
7
8
0
.
0
9
2
3
0
.
0
9
3
3
5
En
t
r
o
p
y
0
.
0
0
7
7
3
0
.
0
0
7
6
8
0
.
0
0
8
3
8
0
.
0
0
7
8
1
2
0
.
0
0
8
1
V
a
r
i
a
n
c
e
2
.
4
7
9
2
.
4
2
4
5
2
.
4
3
4
5
2
.
4
3
1
2
.
4
2
4
S
mo
o
t
h
n
e
ss
0
.
9
3
1
0
.
9
3
2
4
0
.
9
3
2
1
0
.
9
7
1
0
.
9
3
1
C
o
n
t
r
a
st
0
.
3
5
2
0
.
3
7
1
2
0
.
3
6
5
0
.
3
4
5
0
.
3
4
5
C
o
r
r
e
l
a
t
i
o
n
0
.
1
2
0
1
0
.
1
0
9
1
0
.
1
8
6
5
0
.
1
0
4
5
0
.
1
1
2
3
En
e
r
g
y
0
.
8
2
3
4
0
.
8
2
6
3
0
.
8
1
7
5
0
.
8
1
2
9
0
.
8
1
5
7
4
.
1
.
Acc
ura
cy
Acc
u
r
ac
y
is
ca
lcu
lated
b
y
co
m
p
ar
in
g
t
h
e
n
u
m
b
er
o
f
co
r
r
e
ctly
class
if
ied
tu
m
o
r
r
eg
io
n
s
to
th
e
to
tal
n
u
m
b
er
o
f
r
e
g
io
n
s
e
x
am
in
ed
.
Her
e
tr
u
e
p
o
s
itiv
e
d
ep
icts
t
h
e
(
T
P)
(
id
en
tifie
d
T
u
m
o
r
s
)
,
tr
u
e
n
e
g
ativ
e
(
TN
)
d
ep
icts
,
f
alse p
o
s
itiv
e
d
ep
icts
(
FP
)
wh
er
ea
s
f
alse n
eg
ativ
e
(
FN
)
d
ep
icts
(
n
o
t id
e
n
tifie
d
)
.
=
(
+
)
(
+
+
+
)
(
2
6
)
T
h
e
s
u
g
g
ested
Neu
r
o
Fu
s
io
n
N
et
ac
h
iev
es
an
ex
ce
llen
t
ac
c
u
r
ac
y
o
f
9
9
.
2
1
%
in
a
c
o
m
p
lete
ev
alu
atio
n
o
f
ac
cu
r
ac
y
p
er
f
o
r
m
a
n
ce
in
b
r
ain
tu
m
o
r
i
d
en
tific
atio
n
,
as
s
h
o
wn
in
T
a
b
le
2
.
E
x
is
tin
g
m
e
th
o
d
s
s
u
ch
as
Ots
u
th
r
esh
o
ld
in
g
[
1
6
]
,
tr
an
s
f
er
l
ea
r
n
in
g
[
1
7
]
,
m
ac
h
in
e
lear
n
in
g
[
2
6
]
,
C
NNBC
N
[
1
9
]
,
a
n
d
th
e
d
r
a
g
o
n
f
ly
m
ec
h
an
is
m
[
2
0
]
o
u
tp
er
f
o
r
m
t
h
is
.
T
h
e
s
u
p
er
io
r
ity
o
f
Neu
r
o
Fu
s
io
n
Net
d
em
o
n
s
tr
ates
its
ac
cu
r
ac
y
in
d
etec
tin
g
b
r
ain
ca
n
ce
r
s
,
d
em
o
n
s
tr
atin
g
its
p
o
ten
tial
as
a
n
ad
v
an
ce
d
an
d
d
e
p
en
d
a
b
le
m
o
d
el
in
co
m
p
ar
is
o
n
to
co
n
v
en
tio
n
al
an
d
s
tate
-
of
-
th
e
-
ar
t m
eth
o
d
s
.
I
n
ac
co
r
d
a
n
ce
with
T
ab
le
2
,
Fig
u
r
e
1
1
d
ep
icts
th
e
co
m
p
ar
ativ
e
ac
cu
r
ac
y
p
lo
t
f
o
r
b
r
ain
tu
m
o
r
d
etec
tio
n
.
T
h
e
g
r
ap
h
ical
r
ep
r
esen
tatio
n
v
alid
ates
th
e
tab
u
l
ar
f
in
d
in
g
s
b
y
d
em
o
n
s
tr
atin
g
Neu
r
o
Fu
s
io
n
Net'
s
h
ig
h
er
p
er
f
o
r
m
an
ce
w
h
en
c
o
m
p
ar
ed
t
o
o
th
er
ap
p
r
o
ac
h
es.
T
h
e
f
ig
u
r
e
p
r
o
v
id
es
as
a
v
is
u
al
v
alid
atio
n
o
f
th
e
n
u
m
er
ical
ac
c
u
r
ac
y
v
alu
es,
e
m
p
h
asizin
g
t
h
e
s
u
g
g
ested
m
o
d
el'
s
co
n
s
is
ten
t
an
d
am
az
i
n
g
co
r
r
ec
tn
ess
.
T
h
is
Evaluation Warning : The document was created with Spire.PDF for Python.