I
AE
S
I
nte
rna
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI
)
Vo
l.
6
,
No
.
2
,
J
u
n
e
2
0
1
7
,
p
p
.
56
~
65
I
SS
N:
2252
-
8938
,
DOI
: 1
0
.
1
1
5
9
1
/i
j
ai.
v
6
.
i2
.
p
p
56
-
6
5
56
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
jo
u
r
n
a
l.c
o
m/o
n
lin
e/in
d
ex
.
p
h
p
/I
J
A
I
A Dec
isio
n
Sy
ste
m
for P
redic
ting
Dia
betes using
Ne
ura
l
Netw
o
rk
s
K
.
Cha
nd
a
na
Ra
ni,
Y.
P
ra
s
a
nth
De
p
a
rt
m
e
n
t
o
f
Co
m
p
u
ter S
c
ien
c
e
En
g
in
e
e
rin
g
,
K
o
n
e
ru
L
a
k
sh
m
a
y
a
Un
iv
e
rsity
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Feb
9
,
2
0
1
7
R
ev
i
s
ed
A
p
r
13
,
2
0
1
7
A
cc
ep
ted
Ma
y
2
2
,
2
0
1
7
Dia
b
e
ti
c
re
ti
n
o
p
a
t
h
y
(DR)
i
s
a
n
e
y
e
f
i
x
e
d
il
l
c
o
m
p
lete
b
y
th
e
i
m
p
a
irm
e
n
t
o
f
p
o
ly
g
e
n
ic
d
iso
rd
e
r
a
n
d
th
a
t
w
e
p
u
rc
h
a
se
d
t
o
a
c
k
n
o
w
led
g
e
it
b
e
f
o
re
o
f
c
a
len
d
a
r
f
o
r
se
n
sib
le
trea
t
m
e
n
t.
On
th
e
se
li
n
e
s,
2
so
c
ial
o
c
c
a
sio
n
s
w
e
re
p
e
rc
e
iv
e
d
,
sp
e
c
if
ic
a
ll
y
n
o
n
-
p
r
o
li
f
e
ra
t
iv
e
d
iab
e
ti
c
re
ti
n
o
p
a
t
h
y
(NP
DR),
p
ro
li
f
e
ra
ti
v
e
d
iab
e
ti
c
re
ti
n
o
p
a
th
y
(P
DR).
Du
rin
g
th
is
p
a
p
e
r,
to
d
isse
c
t
d
iab
e
ti
c
re
ti
n
o
p
a
th
y
,
3
m
o
d
e
ls
li
k
e
P
ro
b
a
b
il
isti
c
Ne
u
ra
l
f
ra
m
e
w
o
rk
(P
NN
),
Ba
y
e
sia
n
Clas
si
f
ic
a
ti
o
n
a
n
d
S
u
p
p
o
rt
v
e
c
to
r
m
a
c
h
in
e
(S
VM)
sq
u
a
re
m
e
a
su
re
p
ictu
re
d
a
n
d
th
e
ir
d
is
p
lay
s
sq
u
a
re
m
e
a
su
re
th
o
u
g
h
t
-
a
b
o
u
t.
T
h
e
l
iv
e
o
f
th
e
u
n
w
e
ll
n
e
ss
u
n
f
o
ld
w
it
h
in
th
e
m
e
m
b
ra
n
e
a
re
o
f
ten
re
c
o
g
n
ize
d
b
y
a
n
a
ly
ti
c
th
e
e
le
m
e
n
ts
o
f
th
e
m
e
m
b
ra
n
e
.
T
h
e
e
le
m
e
n
ts
li
k
e
v
e
in
s,
h
e
m
o
rrh
a
g
e
s
o
f
NP
DR
im
a
g
e
a
n
d
e
x
u
d
a
tes
o
f
P
DR
ima
g
e
sq
u
a
re
m
e
a
su
re
o
ff
f
ro
m
th
e
u
n
re
f
in
e
d
p
h
o
to
s
v
ictim
iza
ti
o
n
th
e
ico
n
p
r
e
p
a
re
tec
h
n
iq
u
e
s,
f
e
d
to
t
h
e
c
las
sif
ier
f
o
r
g
a
th
e
rin
g
a
c
o
m
p
lete
o
f
3
5
0
str
u
c
tu
re
p
h
o
t
o
s
w
e
re
u
se
d
,
o
u
t
o
f
th
a
t1
0
0
w
e
re
u
se
d
f
o
r
d
e
sig
n
in
g
a
n
d
2
5
0
p
ict
u
r
e
s
w
e
r
e
u
se
d
f
o
r
tes
ti
n
g
.
Ex
p
lo
ra
to
ry
r
e
su
lt
s
sh
o
w
th
a
t
P
NN
h
a
s
a
n
a
c
c
u
ra
c
y
o
f
8
9
.
6
%
Ba
y
e
s
Clas
si
f
ier
in
c
o
rp
o
ra
tes
a
e
x
a
c
tn
e
ss
o
f
9
4
.
4
%
a
n
d
S
V
M
h
a
s
a
n
e
x
a
c
ti
tu
d
e
o
f
9
7
.
6
%
.
W
h
a
t
i
s
m
o
re
o
u
r
s
y
ste
m
is
e
q
u
a
ll
y
c
o
n
ti
n
u
e
ru
n
n
i
n
g
o
n
1
3
0
p
ict
u
re
s
o
p
e
n
f
ro
m
"
DI
A
RE
T
DB0
:
Ev
a
lu
a
ti
o
n
D
a
tab
a
se
a
n
d
P
r
o
c
e
d
u
re
f
o
r
Dia
b
e
ti
c
Re
ti
n
o
p
a
t
h
y
"
a
n
d
a
lso
th
e
re
su
lt
s
sh
o
w
th
a
t
P
NN
in
c
o
rp
o
ra
tes
a
e
x
a
c
tn
e
ss
o
f
8
7
.
6
9
%
Ba
y
e
s
Clas
si
f
ier
h
a
s
a
n
a
c
c
u
ra
c
y
o
f
9
0
.
7
6
%
a
n
d
S
V
M
h
a
s
a
p
re
c
isio
n
o
f
9
5
.
3
8
%
.
K
ey
w
o
r
d
:
A
cc
u
r
ac
y
P
r
o
b
ab
ilis
tic
n
eu
r
al
n
et
w
o
r
k
B
ay
e
s
ian
cla
s
s
i
f
ica
tio
n
Sen
s
iti
v
it
y
Sp
ec
if
icit
y
Su
p
p
o
r
t v
ec
to
r
m
ac
h
i
n
e
Co
p
y
rig
h
t
©
2
0
1
7
In
stit
u
te o
f
A
d
v
a
n
c
e
d
E
n
g
i
n
e
e
rin
g
a
n
d
S
c
ien
c
e
.
Al
l
rig
h
ts
re
se
rv
e
d
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
K.
C
h
a
n
d
an
a
R
an
i,
Dep
ar
t
m
en
t o
f
C
o
m
p
u
ter
Scie
n
ce
E
n
g
i
n
ee
r
in
g
,
Ko
n
er
u
L
ak
s
h
m
a
y
a
U
n
iv
er
s
it
y
,
Gr
ee
n
Field
s
,
Vad
d
es
w
ar
a
m
,
T
ad
e
p
alli,
Gu
n
t
u
r
-
5
2
2
5
0
2
,
An
d
h
r
a
P
r
ad
esh
.
E
m
ail:c
h
a
n
d
u
.
k
a
n
d
r
u
@
g
m
ai
l.c
o
m
1.
I
NT
RO
D
UCT
I
O
N
Diab
etes is
a
g
e
t
-
co
llect
iv
el
y
o
f
m
e
tab
o
lic
ill
n
ess
e
s
i
n
w
h
ic
h
a
m
a
n
h
a
s
h
ig
h
g
l
u
co
s
e,
b
o
th
i
n
m
ild
o
f
th
e
m
an
n
er
t
h
at
t
h
e
f
r
a
m
e
d
o
es
n
o
w
n
o
t
m
ak
e
s
u
f
f
ic
ien
t
in
s
u
li
n
,
o
r
b
ec
au
s
e
th
e
ce
lls
d
o
n
o
t
r
esp
o
n
d
to
t
h
e
in
s
u
li
n
t
h
at
is
m
ad
e.
Diab
etic
r
etin
o
p
ath
y
is
o
n
e
o
f
t
h
e
r
e
g
u
lar
p
er
p
lex
ities
o
f
d
iab
etes.
I
t
’
s
f
ar
a
n
s
u
p
er
a
n
d
f
o
r
th
e
m
o
s
t
co
m
p
o
n
en
t
s
p
r
ea
d
e
y
e
d
is
ea
s
e.
I
t
h
u
r
t
s
t
h
e
li
ttl
e
v
ei
n
s
w
it
h
i
n
t
h
e
r
eti
n
a
co
m
i
n
g
to
f
r
u
itio
n
i
n
lo
s
s
o
f
i
m
a
g
i
n
ati
v
e
a
n
d
p
r
escien
t.
T
h
e
p
er
il
o
f
th
e
i
n
f
ec
tio
n
w
i
ll
in
cr
ea
s
e
w
i
th
a
g
e
a
n
d
th
u
s
l
y
,
m
o
d
estl
y
m
atu
r
ed
an
d
m
o
r
e
o
r
g
an
ized
d
iab
etics
ar
e
s
lan
ted
to
Diab
etic
R
eti
n
o
p
ath
y
.
No
n
-
p
r
o
lif
er
ati
v
e
d
iab
etic
r
etin
o
p
ath
y
is
a
n
ea
r
l
y
d
u
r
atio
n
o
f
d
iab
etic
r
e
tin
o
p
ath
y
.
In
th
i
s
lev
el,
m
i
n
o
r
v
ein
s
in
s
id
e
t
h
e
r
etin
a
s
p
ill
b
lo
o
d
o
r
f
lu
id
.
T
h
e
d
is
ch
ar
g
i
n
g
f
l
u
id
r
ea
s
o
n
s
th
e
r
etin
a
to
s
w
ell
o
r
to
s
h
ap
e
s
to
r
es
k
n
o
w
n
as
ex
u
d
ates.
P
r
o
lif
er
ativ
e
d
iab
etic
r
etin
o
p
ath
y
,
P
DR
i
s
a
s
tr
i
v
e
b
y
t
h
e
e
y
e
to
cr
ea
te
o
r
r
esu
p
p
l
y
th
e
r
eti
n
a
w
it
h
n
e
w
s
e
lects
v
ess
e
ls
(
n
eo
v
asc
u
lar
izatio
n
)
,
b
ec
au
s
e
o
f
ir
r
esp
ec
tiv
e
o
f
h
o
w
y
o
u
lo
o
k
at
it
f
in
i
s
h
o
f
t
h
e
r
eti
n
al
b
lo
o
d
d
eliv
er
.
Sh
o
ck
in
g
l
y
,
t
h
e
n
e
w
,
o
d
d
v
e
in
s
d
o
n
o
t
r
e
-
s
u
p
p
l
y
th
e
r
eti
n
a
w
i
th
co
m
m
o
n
s
tr
ea
m
m
a
ch
in
e,
b
u
t
ex
p
e
n
d
co
m
p
ete
n
tl
y
an
d
ar
e
r
eg
u
lar
l
y
j
o
in
ed
b
y
w
a
y
o
f
s
ca
r
t
is
s
u
e
th
at
m
a
y
w
r
in
k
le
o
r
is
o
late
th
e
r
eti
n
a.
On
th
is
p
ap
er
,
an
a
u
to
m
a
ted
tech
n
iq
u
e
f
o
r
ac
cu
m
u
la
tin
g
o
f
th
e
ill
n
es
s
d
iab
etic
r
eti
n
o
p
ath
y
t
h
e
u
s
e
o
f
f
u
n
d
u
s
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
A
Dec
is
io
n
S
ystem
fo
r
P
r
ed
ictin
g
Dia
b
etes u
s
in
g
N
eu
r
a
l Netw
o
r
ks
…
(
K
.
C
h
a
n
d
a
n
a
R
a
n
i
)
57
i
m
a
g
es
is
p
r
o
v
e
n
.
W
e
u
s
ed
th
e
r
etin
al
f
u
n
d
u
s
p
i
x
ass
e
m
b
led
o
n
th
e
"
A
r
av
in
d
E
y
e
r
ec
u
p
er
at
io
n
w
o
r
k
p
lace
an
d
P
o
s
tg
r
ad
u
te
I
n
s
tit
u
te
o
f
O
p
th
al
m
o
lo
g
y
"
,
C
u
d
d
al
o
r
e
av
en
u
e
T
h
av
a
lak
u
p
p
a
m
J
u
n
c
tio
n
,
P
o
n
d
ich
er
r
y
.
T
h
e
r
etin
al
p
h
o
to
is
ta
k
en
i
n
s
id
e
th
e
R
GB
s
tr
u
ct
u
r
e
b
y
w
a
y
o
f
f
u
n
d
u
s
d
ig
ital
ca
m
er
a.
A
f
u
n
d
u
s
ca
m
er
a
o
r
r
etin
al
d
i
g
ica
m
i
s
a
s
p
ec
if
ic
l
o
w
elec
tr
icit
y
a
m
p
l
if
y
i
n
g
to
o
l
w
it
h
a
j
o
in
ed
ca
m
er
a
p
r
o
p
o
s
ed
to
i
m
a
g
e
w
it
h
in
s
u
r
f
ac
e
o
f
th
e
e
y
e,
w
h
ich
in
c
lu
d
e
t
h
e
r
eti
n
a,
o
p
tic
cir
cle,
m
ac
u
la,
an
d
r
et
u
r
n
ed
s
u
b
m
it.
T
h
e
s
n
ap
s
h
o
t
s
h
ad
b
ee
n
o
b
s
er
v
ed
u
s
in
g
a
C
a
n
o
n
T
o
p
C
o
n
T
R
C
-
5
0
E
X
w
i
th
Nico
n
r
eti
n
al
d
ig
ica
m
a
t
a
ar
ea
-
of
-
p
o
in
t
o
f
v
ie
w
i
s
1
2
8
0
×1
0
2
4
in
2
4
b
it
J
P
E
G
o
u
tlin
e.
T
h
e
ap
p
r
aisal
o
f
t
h
e
p
r
o
p
o
s
ed
m
ec
h
an
ized
w
illp
o
w
er
co
u
r
s
e
o
f
ac
tio
n
o
f
d
iab
etic
r
etin
o
p
ath
y
h
a
s
b
ee
n
d
o
n
e
b
y
m
ea
n
s
o
f
th
e
u
s
e
o
f
a
r
ec
r
ea
tio
n
p
la
n
o
f
2
5
0
f
u
n
d
u
s
s
n
ap
s
h
o
t
s
t
h
at
'
s
a
m
i
x
o
f
tr
ad
itio
n
al,
NP
D
R
an
d
P
DR
i
m
p
a
cted
p
h
o
to
g
r
ap
h
s
.
T
h
e
p
r
in
cip
le
i
m
a
g
e
is
ch
a
n
g
ed
o
v
er
to
d
i
m
s
ca
le
i
m
a
g
e.
Af
ter
t
h
at
,
b
en
d
y
h
is
to
g
r
a
m
co
n
f
o
r
m
i
t
y
i
s
r
elate
d
to
i
m
p
r
o
v
e
th
e
m
a
n
y
-
s
id
ed
f
i
n
e
o
f
t
h
e
p
ictu
r
e.
Via
t
h
e
n
,
Dis
cr
et
e
W
av
elet
r
e
m
o
d
el
(
DW
T
)
is
ass
o
ciate
d
a
n
d
t
h
e
e
x
ten
t
o
f
t
h
e
p
h
o
to
i
s
r
ed
u
ce
d
i
n
to
h
alf
o
f
as
6
4
0
×
5
1
2
[
3
0
]
w
it
h
th
e
aid
o
f
th
e
n
C
o
o
r
d
in
ated
ch
an
n
el
r
ea
ctio
n
(
MFR
)
is
as
s
o
ciate
d
w
i
th
d
i
m
i
n
is
h
i
n
g
t
h
e
n
o
is
e
in
t
h
e
i
m
a
g
e.
A
t
u
lt
i
m
a
te,
Fu
zz
y
C
-
i
m
p
lie
s
b
u
n
c
h
in
g
i
s
h
o
o
k
ed
u
p
to
co
m
p
o
n
en
t
t
h
e
v
e
in
s
i
n
s
id
e
t
h
e
i
m
ag
e.
in
th
e
w
a
k
e
o
f
p
r
ep
r
o
ce
s
s
in
g
o
f
s
n
a
p
s
h
o
ts
is
f
i
n
is
h
ed
,
f
ac
to
r
s
,
f
o
r
in
s
ta
n
ce
,
R
ad
iu
s
,
Dia
m
eter
,
p
lace
,
A
r
c
d
u
r
atio
n
,
C
o
g
n
izan
ce
a
n
g
le
an
d
1
/2
ar
e
a
ar
e
co
m
p
u
ted
f
o
r
ea
ch
p
h
o
t
o
g
r
ap
h
.
A
t
t
h
at
f
ac
to
r
De
m
o
n
s
tr
atin
g
tech
n
iq
u
es
lik
e
P
NN,
B
a
y
e
s
p
r
in
cip
l
e
an
d
SVM
ar
e
ap
p
lied
an
d
t
h
e
ir
ex
h
ib
itio
n
s
a
r
e
ch
ec
k
ed
o
u
t.
A
t
lo
n
g
f
i
n
al,
th
e
p
h
o
to
g
r
ap
h
s
ar
e
o
r
g
a
n
ized
i
n
to
3
g
at
h
er
i
n
g
s
to
b
e
s
p
ec
i
f
ic,
tr
ad
itio
n
a
l
p
h
o
to
g
r
ap
h
,
No
n
-
P
r
o
lif
er
ativ
e
Diab
etic
R
eti
n
o
p
ath
y
(
NP
DR
)
a
n
d
P
r
o
lif
er
ativ
e
Diab
etic
R
eti
n
o
p
ath
y
.
Fig
u
r
e
1
.
B
lo
ck
d
iag
r
a
m
f
o
r
C
o
m
p
ar
is
o
n
b
et
w
ee
n
t
h
r
ee
C
la
s
s
if
ier
s
f
o
r
D
ia
g
n
o
s
i
s
o
f
D
R
T
h
e
r
elax
atio
n
o
f
th
is
p
ap
er
co
n
s
is
ts
a
s
ta
k
es
a
f
ter
.
Sect
io
n
2
d
ep
icts
th
e
ass
o
ciate
d
p
ain
ti
n
g
s
.
Seg
m
en
t
th
r
ee
clar
i
f
ies
t
h
e
p
r
ep
r
o
ce
s
s
in
g
o
f
p
ictu
r
es.
R
eg
io
n
f
o
u
r
clar
if
ies
t
h
e
d
etail
ex
tr
ac
tio
n
.
Seg
m
e
n
t
f
iv
e
clar
if
ies
th
e
e
le
m
e
n
t
v
al
u
es
g
o
t.
Sectio
n
6
p
o
r
tr
a
y
s
t
h
e
g
r
o
u
p
in
g
o
f
D
R
d
is
o
r
d
er
u
ti
lis
i
n
g
as
s
is
t
Vec
to
r
g
ad
g
et.
Vicin
it
y
7
clar
if
ies
P
r
o
b
ab
ilis
tic
Ne
u
r
al
co
m
m
u
n
it
y
a
n
d
p
h
ase
8
o
f
f
er
s
B
a
y
es
ian
clas
s
.
R
eg
i
on
9
d
ep
icts
t
h
e
co
n
s
eq
u
e
n
ce
s
a
n
d
d
ialo
g
.
A
r
ea
1
0
g
iv
es
t
h
e
b
elie
f
.
Fig
u
r
e
1
g
iv
e
s
t
h
e
r
ec
tan
g
u
lar
o
u
tli
n
e
f
o
r
ex
a
m
in
a
tio
n
o
f
th
r
ee
class
if
ier
s
f
o
r
d
eter
m
i
n
at
io
n
o
f
Diab
etic
R
et
in
o
p
ath
y
.
1
.
1
.
Rela
t
ed
Wo
rk
I
n
s
id
e
th
e
m
id
s
t
o
f
th
e
late
y
e
ar
s
,
th
er
e
w
er
e
d
iv
er
s
e
r
esear
ch
o
n
cu
s
to
m
d
e
s
ig
n
ed
f
i
n
is
h
o
f
d
iab
etic
r
etin
o
p
ath
y
t
h
e
u
s
a
g
e
o
f
a
p
air
h
i
g
h
l
ig
h
t
s
an
d
tec
h
n
iq
u
e
s
.
D.
Vallab
h
a
et
al.
[
1
]
p
r
o
p
o
s
ed
a
p
r
o
ce
d
u
r
e
f
o
r
r
o
b
o
tized
ac
k
n
o
w
led
g
m
e
n
t a
n
d
r
eq
u
est o
f
v
a
s
cu
la
r
ab
n
o
r
m
a
liti
es i
n
Diab
etic
R
e
tin
o
p
at
h
y
t
h
e
u
s
e
o
f
s
ca
le
a
n
d
p
r
esen
tatio
n
s
p
ec
i
f
ic
Gab
o
r
ch
an
n
el
b
an
k
s
.
R
.
Siv
a
k
u
m
ar
e
t
al.
[
2
]
f
am
il
iar
a
m
ac
h
in
e
with
r
eq
u
est
d
iab
etic
r
etin
o
p
ath
y
s
u
b
j
ec
ts
f
r
o
m
ad
j
u
s
t
m
e
n
ts
i
n
v
is
u
al
e
v
o
k
ed
ca
p
ac
it
y
g
h
o
s
t
q
u
a
n
tit
ies.
By
us
in
g
W
alter
et
al.
[
3
]
ex
u
d
ate
s
ar
e
d
eter
m
in
ed
t
h
e
u
s
a
g
e
o
f
t
h
eir
ex
ce
s
s
i
v
e
d
i
m
in
is
h
s
tag
e
a
s
s
o
r
t
m
e
n
t,
a
n
d
th
eir
s
tr
u
ctu
r
e
s
ar
e
d
eter
m
in
ed
w
it
h
th
e
aid
o
f
a
p
p
r
o
ac
h
f
o
r
m
o
r
p
h
o
lo
g
ical
r
e
v
a
m
p
i
n
g
m
et
h
o
d
o
lo
g
ies.
HT
Ng
u
y
e
n
l
et
al.
[
4
]
p
r
o
p
o
s
ed
a
m
u
lti
la
y
er
n
o
u
r
i
s
h
m
en
t
a
h
ea
d
f
r
a
m
e
w
o
r
k
f
o
r
th
e
g
a
m
e
p
la
n
o
f
D
R
.
Ma
r
ía
Gar
cía
et
al.
[
5
]
u
s
ed
a
m
u
ltil
a
y
er
p
er
ce
p
tr
o
n
(
ML
P
)
class
i
f
ier
to
g
et
a
c
lo
s
in
g
d
iv
is
io
n
o
f
to
u
g
h
E
x
u
d
ates
w
it
h
i
n
t
h
e
p
ict
u
r
e.
I
n
[
6
]
,
P
.
V.
Nag
es
w
ar
a
r
ao
et
al.
p
r
o
p
o
s
ed
ev
er
y
o
th
er
ap
p
r
o
ac
h
f
o
r
p
r
o
tei
n
s
p
o
r
t
p
lan
in
li
g
h
t
o
f
a
P
r
o
b
ab
ilis
tic
Neu
r
al
n
et
w
o
r
k
an
d
h
i
g
h
l
ig
h
t
s
elec
tio
n
.
S.
C
h
au
d
h
u
r
y
et
al.
[
7
]
ad
d
r
ess
th
e
is
s
u
e
o
f
p
er
ce
iv
i
n
g
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
IJ
-
AI
Vo
l.
6
,
No
.
1
,
Ma
r
ch
2
0
1
7
:
56
–
65
58
v
ein
s
i
n
r
eti
n
al
p
h
o
to
s
.
T
h
e
y
h
av
e
u
s
ed
th
e
th
o
u
g
h
t o
f
co
m
p
o
s
ed
s
tatio
n
f
o
r
d
is
clo
s
u
r
e
o
f
s
ig
n
s
a
n
d
s
y
m
p
to
m
s
to
r
ec
o
g
n
i
ze
p
iece
w
is
e
d
ir
ec
t
p
ar
ts
o
f
v
ei
n
s
in
r
eti
n
al
i
m
a
g
e
s
a
n
d
cr
ea
ted
1
2
o
n
e
i
n
al
l
a
t
y
p
e
p
r
ep
ar
atio
n
s
to
ch
ase
d
o
w
n
v
es
s
el
r
eg
io
n
s
al
o
n
g
s
id
e
ea
ch
an
d
e
v
er
y
f
ea
s
ib
le
h
ea
d
in
g
.
I
n
[
8
]
A
l
ir
ez
a
Osa
r
eh
et
al.
ask
ed
t
h
e
d
iv
id
ed
ar
ea
s
in
to
t
w
o
d
is
j
o
in
t
in
s
tr
u
ctio
n
s
,
ex
u
d
ate
s
an
d
n
o
n
-
e
x
u
d
ate
s
,
tak
i
n
g
a
g
a
n
d
er
o
n
th
e
ex
ec
u
tio
n
o
f
n
u
m
er
o
u
s
cla
s
s
i
f
ier
s
.
I
n
[
9
]
W
o
n
g
L
i
Yu
n
et
al.
as
k
ed
t
h
e
4
e
y
e
s
ick
n
e
s
s
es
t
h
e
u
s
e
o
f
a
3
-
la
y
er
s
u
s
te
n
a
n
ce
ah
ea
d
n
e
u
r
al
f
r
a
m
e
w
o
r
k
.
I
n
[
1
3
]
,
R
.
Priy
a
a
n
d
P
.
A
r
u
n
a
u
s
ed
SVM
f
o
r
t
h
e
ac
k
n
o
w
le
d
g
m
e
n
t
o
f
d
iab
etic
r
etin
o
p
ath
y
tier
s
t
h
e
u
s
e
o
f
s
h
ad
in
g
f
u
n
d
u
s
p
ix
.
I
n
[
1
4
]
p
c
p
r
o
v
id
ed
m
o
to
r
ized
cr
i
m
s
o
n
d
a
m
ag
e
d
is
t
in
g
u
is
h
i
n
g
p
r
o
o
f
s
o
m
e
a
s
s
i
s
tan
ce
w
it
h
b
e
in
g
ac
h
ie
v
ed
o
n
d
i
g
itized
tr
a
n
s
p
ar
en
cies
[
1
5
]
.
T
h
is
p
ap
er
in
s
p
ec
ts
a
n
d
p
r
o
p
o
s
es
a
co
u
r
s
e
o
f
ac
tio
n
o
f
i
n
a
g
r
ea
t
g
lo
b
al
ad
j
u
s
ted
m
o
r
p
h
o
lo
g
ic
al
m
a
n
ag
er
s
f
o
r
u
s
e
f
o
r
e
x
u
d
at
e
ac
k
n
o
w
led
g
m
en
t
o
n
d
iab
etic
r
etin
o
p
ath
y
p
atie
n
t
’
s
n
o
n
-
ex
te
n
d
ed
u
n
d
er
s
tu
d
y
an
d
co
f
f
ee
-
s
ep
ar
atio
n
p
ics
.
[
1
6
]
Dep
icted
a
m
et
h
o
d
f
o
r
ac
tu
al
l
y
p
er
ce
i
v
in
g
n
e
w
v
e
s
s
el
s
o
n
th
e
o
p
tic
p
la
te
th
e
u
s
a
g
e
o
f
r
etin
a
l
p
h
o
to
g
r
ap
h
y
.
T
h
e
'
p
lan
n
ed
ch
an
n
el
r
esp
o
n
s
e
'
ap
p
r
o
ac
h
is
a
f
o
r
t
h
e
m
ax
i
m
u
m
co
m
p
o
n
e
n
t
u
s
ed
f
o
r
m
at
p
r
i
m
ar
il
y
b
ased
tech
n
iq
u
e
t
h
at
m
ak
e
s
u
s
e
o
f
a
p
la
n
o
f
2
d
Ga
u
s
s
ia
n
b
it
s
w
i
th
an
ad
j
u
s
ted
p
er
io
d
an
d
p
r
o
lo
g
u
e
to
r
e
m
o
d
el
t
h
e
v
es
s
els.
I
n
[
1
7
]
,
a
p
o
s
t
h
an
d
li
n
g
g
ad
g
et,
i
n
atti
tu
d
e
o
f
asp
ec
t
d
is
ti
n
g
u
i
s
h
in
g
ev
id
en
ce
,
is
r
elate
d
to
u
n
d
er
s
t
an
d
h
ar
d
ex
u
d
ate
s
f
r
o
m
co
tto
n
w
o
o
l
s
p
o
ts
an
d
u
n
iq
u
e
r
elics.
I
n
[
1
8
]
Keith
A
.
Go
at
g
u
y
et
al.
d
elin
ea
te
d
a
f
r
am
e
w
o
r
k
f
o
r
g
en
er
all
y
d
ed
u
cti
n
g
n
e
w
v
es
s
els
at
th
e
o
p
tic
cir
cle
th
e
u
s
a
g
e
o
f
r
eti
n
al
p
ictu
r
es.
A
l
iaa
A
b
d
el
-
Hale
i
m
et
al.
[
2
3
]
d
is
p
lay
ed
a
f
r
a
m
e
w
o
r
k
to
th
er
ef
o
r
e
p
er
ce
iv
e
th
e
lo
ca
tio
n
o
f
th
e
OD
in
m
o
d
er
n
i
ze
d
r
etin
al
f
u
n
d
u
s
p
h
o
to
g
r
ap
h
s
.
T
h
e
f
r
a
m
e
w
o
r
k
s
tar
ts
v
ia
n
o
r
m
aliz
in
g
r
ad
ian
c
e
an
d
d
if
f
er
en
ce
all
t
h
r
o
u
g
h
t
h
e
p
ictu
r
e
th
e
u
s
a
g
e
o
f
illu
m
i
n
atio
n
lev
el
in
g
an
d
a
d
ap
tab
le
h
is
to
g
r
a
m
e
v
e
n
in
g
o
u
t f
r
a
m
e
w
o
r
k
s
s
ep
ar
ate.
2.
RE
S
E
ARCH
M
E
T
H
O
D
I
n
r
ec
o
g
n
izi
n
g
v
ar
iatio
n
s
f
r
o
m
th
e
n
o
r
m
co
n
n
ec
ted
w
i
th
f
u
n
d
u
s
i
m
ag
e,
th
e
p
ict
u
r
e
m
u
s
t
b
e
p
r
e
-
h
an
d
led
w
it
h
k
ee
p
in
g
i
n
m
i
n
d
th
e
en
d
g
o
al
to
r
ev
is
e
t
h
e
u
n
e
v
en
e
n
li
g
h
ten
m
e
n
t,
n
o
t
ad
eq
u
ate
d
if
f
er
e
n
ce
i
n
th
e
m
id
d
le
o
f
ex
u
d
ates
an
d
p
ictu
r
e
f
o
u
n
d
atio
n
p
i
x
els
a
n
d
th
e
v
ici
n
it
y
o
f
cla
m
o
r
w
it
h
in
t
h
e
in
f
o
r
m
atio
n
f
u
n
d
u
s
p
ictu
r
e.
T
h
e
m
et
h
o
d
s
f
o
r
p
r
ep
r
o
ce
s
s
in
g
in
cl
u
d
e
g
r
a
y
s
ca
le
T
r
an
s
f
o
r
m
atio
n
,
A
d
ap
ti
v
e
His
to
g
r
a
m
E
q
u
al
izatio
n
,
Dis
cr
ete
W
av
elet
c
h
an
g
e,
Gau
s
s
ia
n
Ma
tch
ed
f
ilter
R
e
s
p
o
n
s
e
an
d
Fu
zz
y
C
m
ea
n
s
.
B
u
n
c
h
in
g
f
o
r
d
iv
is
io
n
o
f
v
ei
n
s
.
2
.
1
.
G
ra
y
Sca
le
Co
nv
er
s
io
n
Gr
a
y
s
ca
le
p
h
o
to
g
r
ap
h
s
ar
e
u
n
iq
u
e
f
r
o
m
o
n
e
-
p
iece
b
i
-
to
n
al
h
i
g
h
co
m
p
ar
is
o
n
i
m
a
g
es,
w
h
ich
i
n
t
h
e
p
u
ttin
g
o
f
lap
to
p
i
m
ag
i
n
g
ar
e
i
m
a
g
es
w
it
h
s
i
m
p
l
y
t
h
e
t
w
o
c
o
lo
r
in
g
s
,
d
ar
k
i
s
h
,
a
n
d
w
h
ite
(
l
ik
e
w
i
s
e
r
ef
er
r
ed
to
as
b
i
le
v
el
o
r
p
ar
allel
p
h
o
to
s
)
.
Gr
e
y
s
ca
le
i
m
ag
e
s
h
av
e
s
e
v
er
al
s
u
n
s
h
ad
es
o
f
d
i
m
in
th
e
ce
n
ter
.
Gr
e
y
s
ca
le
p
i
x
ar
e
r
eg
u
lar
l
y
t
h
e
e
f
f
ec
t
o
f
m
ea
s
u
r
in
g
th
e
elec
tr
ici
t
y
o
f
m
ild
a
t
ev
er
y
p
i
x
el
i
n
a
s
o
litar
y
b
an
d
o
f
t
h
e
elec
tr
o
m
ag
n
etic
v
ar
iet
y
(
e.
g
.
in
f
r
ar
ed
,
o
b
v
io
u
s
m
ild
,
s
h
i
n
y
,
an
d
s
o
f
o
r
th
.
)
,
a
n
d
in
s
u
c
h
i
n
s
ta
n
ce
s
th
e
y
'r
e
m
o
n
o
ch
r
o
m
atic
ap
p
r
o
p
r
iate
w
h
ile
j
u
s
t
a
g
i
v
en
r
ec
u
r
r
en
ce
is
s
tu
c
k
.
An
y
w
a
y
s
,
li
k
e
w
i
s
e
th
e
y
ca
n
b
e
m
ix
ed
f
r
o
m
a
f
u
l
l
s
h
ad
i
n
g
p
ictu
r
e;
s
ee
th
e
ar
ea
ap
p
r
o
x
i
m
ate
l
y
ch
a
n
g
i
n
g
o
v
er
to
g
r
a
y
s
ca
le.
S
h
o
w
n
i
n
th
e
Fi
g
u
r
e
2
an
d
Fig
u
r
e
3
.
Fig
u
r
e
2
(a
-
d
)
.
Or
ig
in
al
D
R
af
f
ec
ted
E
y
e
I
m
a
g
e
Fig
u
r
e
3
(a
-
d
)
.
E
y
e
I
m
a
g
e
a
f
te
r
Gr
ey
Sca
le
C
o
n
v
er
s
io
n
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
A
Dec
is
io
n
S
ystem
fo
r
P
r
ed
ictin
g
Dia
b
etes u
s
in
g
N
eu
r
a
l Netw
o
r
ks
…
(
K
.
C
h
a
n
d
a
n
a
R
a
n
i
)
59
A
d
ap
tab
le
h
i
s
to
g
r
a
m
p
ar
it
y
wh
ich
is
u
s
ed
to
ad
v
an
ce
co
m
p
lex
it
y
in
p
ict
u
r
es
i
s
ass
o
ciate
d
w
it
h
th
e
d
i
m
s
ca
le
c
h
an
g
ed
o
v
er
e
y
e
p
ictu
r
e.
C
o
n
s
id
er
a
r
u
n
n
in
g
s
u
b
p
ictu
r
e
W
o
f
N
×
N
p
ix
el
s
co
n
ce
n
tr
ated
o
n
a
p
ix
el
P
(
i,
j
)
,
th
e
p
h
o
to
is
f
ilt
er
ed
to
m
ak
e
a
n
o
th
er
s
u
b
p
ictu
r
e
P
o
f
(
N
×
N)
p
ix
els
as
d
em
o
n
s
tr
ated
b
y
t
h
e
ex
a
m
in
at
io
n
u
n
d
er
n
ea
th
,
A
d
v
an
ce
d
p
ictu
r
e
g
e
t
r
ea
d
y
h
a
s
u
p
h
eld
in
t
h
e
ch
a
n
g
e
an
d
r
ec
o
v
er
y
o
f
p
ictu
r
es
f
o
r
o
v
er
a
q
u
ar
ter
o
f
a
ce
n
tu
r
y
.
T
h
is
w
o
r
k
f
o
c
u
s
e
s
o
n
o
n
e
p
ar
ticu
lar
est
i
m
atio
n
a
m
o
n
g
th
e
v
ar
io
u
s
o
p
e
n.
I
n
p
ar
ticu
lar
,
it
w
ill
b
e
p
r
escr
ib
ed
th
at
t
h
e
co
u
n
t
b
ein
g
all
u
d
ed
to
ca
n
b
e
u
p
g
r
ad
ed
w
it
h
t
h
e
u
til
izatio
n
o
f
d
ata
f
r
o
m
an
al
ter
n
ate
b
r
an
c
h
o
f
p
r
o
g
r
a
m
m
i
n
g
b
u
ild
in
g
.
T
o
en
g
ag
e
th
e
p
er
u
s
er
to
b
etter
f
at
h
o
m
th
e
co
m
p
u
tatio
n
,
cr
itical
estab
lis
h
m
en
t in
f
o
r
m
at
i
o
n
w
ill b
e
s
h
o
w
n
.
Mo
d
er
n
ize
d
p
ictu
r
e
g
et
r
ea
d
y
f
r
a
m
e
w
o
r
k
s
h
ad
th
e
ir
g
i
g
an
t
ic
b
eg
in
n
i
n
g
i
n
th
e
1
9
6
0
s
w
it
h
t
h
e
s
p
ac
e
p
r
o
g
r
am
.
T
h
e
ess
e
n
ti
al
to
u
p
d
ate
th
e
w
a
y
o
f
th
e
p
h
o
to
g
r
ap
h
s
r
etu
r
n
ed
b
y
t
h
e
ea
r
l
y
s
p
ac
e
test
s
en
er
g
i
ze
d
r
esear
ch
ar
o
u
n
d
th
er
e.
A
t
f
ir
s
t
t
h
i
s
d
ev
elo
p
m
e
n
t
w
a
s
ass
o
ciate
d
w
it
h
s
p
ac
e
i
m
a
g
er
y
,
th
e
n
ag
a
in
it
w
as
s
o
o
n
co
m
p
r
eh
e
n
d
ed
th
at
d
if
f
e
r
en
t
ex
te
n
ts
co
u
ld
m
o
r
eo
v
er
f
av
o
r
ab
le
p
o
s
itio
n
.
I
n
th
is
w
a
y
,
th
e
n
ec
e
s
s
itie
s
o
f
th
e
h
elp
f
u
l
f
ie
ld
p
u
lled
in
th
e
th
o
u
g
h
t
o
f
au
th
o
r
ities
.
T
h
e
s
y
s
te
m
s
u
ti
lized
b
y
th
e
s
p
ac
e
b
u
s
in
e
s
s
w
er
e
as
s
o
ciate
d
.
W
ith
th
e
p
r
ese
n
ce
o
f
t
h
e
X
-
b
ar
an
d
later
Nu
clea
r
Ma
g
n
e
tic
R
e
s
o
n
a
n
ce
I
m
ag
i
n
g
(
NM
R
I
o
r
MRI)
,
Po
s
itro
n
E
m
is
s
io
n
T
o
m
o
g
r
ap
h
y
(
P
E
T
Scan
s
)
,
C
o
m
p
u
ter
ized
H
elp
ed
T
o
m
o
g
r
ap
h
y
(
C
A
T
Scan
s
)
,
a
n
d
Ultr
a
-
s
o
u
n
d
I
m
ag
i
n
g
,
t
h
e
m
ea
s
u
r
e
o
f
r
e
m
ed
i
al
d
ata
e
x
te
n
d
ed
ess
e
n
tia
ll
y
.
I
n
t
h
e
m
ea
n
ti
m
e
au
th
o
r
itie
s
s
ea
r
ch
ed
f
o
r
b
etter
s
y
s
te
m
s
to
u
p
d
ate
th
ese
p
h
o
to
s
.
T
h
e
test
d
ata
u
tili
ze
d
f
o
r
th
i
s
w
o
r
k
b
eg
a
n
f
r
o
m
th
r
ee
o
n
e
o
f
a
k
i
n
d
s
o
r
ts
o
f
d
ata
co
llectio
n
f
r
a
m
e
w
o
r
k
s
.
T
h
e
m
id
-
ar
ea
p
ictu
r
e
is
o
n
e
c
u
t
o
f
a
C
A
T
ch
an
n
el
.
T
h
e
lef
t
s
tan
d
in
g
u
p
to
h
ea
d
is
cu
t
n
u
m
b
er
5
0
o
f
a
1
0
9
-
cu
t
MRI
ch
ec
k
.
T
h
e
p
r
iv
ileg
e
s
tan
d
in
g
u
p
to
h
ea
d
n
a
m
ed
"
b
r
ain
"
is
an
en
tr
y
w
a
y
p
ict
u
r
e
tak
e
n
i
n
t
h
e
m
id
s
t
o
f
R
ad
io
t
h
er
ap
y
T
r
ea
t
m
en
t.
E
ac
h
ar
ch
i
v
e
g
et
s
i
n
co
n
tact
f
o
r
d
ev
elo
p
m
e
n
t
in
a
n
u
n
r
e
f
in
ed
p
ictu
r
e
p
o
s
itio
n
.
I
n
u
n
r
e
f
in
ed
p
ictu
r
e
o
u
t
lin
e
t
h
e
f
o
r
ce
q
u
alities
ar
e
co
n
tain
ed
in
a
r
ec
o
r
d
in
s
ec
ti
o
n
g
en
u
i
n
e
s
o
licitat
io
n
,
w
it
h
d
if
f
er
e
n
t
p
lan
es
b
ei
n
g
s
ec
u
r
ed
co
n
s
ec
u
ti
v
el
y
f
r
o
m
f
r
o
n
t to
b
ac
k
.
3.
DIS
CR
E
T
E
WA
VE
L
E
T
T
R
ANSF
O
RM
T
h
e
ex
ch
an
g
e
o
f
a
s
i
g
n
i
s
ac
tu
all
y
o
n
e
g
r
ea
ter
s
o
r
t o
f
s
p
ea
k
i
n
g
to
t
h
e
s
i
g
n
al.
I
t d
o
es n
o
t e
x
ch
an
g
e
t
h
e
f
ac
ts
co
n
te
n
t
m
ater
ial
p
r
ese
n
t
in
s
id
e
t
h
e
s
i
g
n
al.
T
h
e
D
is
cr
et
e
W
av
elet
tr
a
n
s
f
o
r
m
(
DW
T
)
,
w
h
ic
h
is
i
n
li
g
h
t
o
f
s
u
b
-
b
an
d
co
d
in
g
,
is
f
o
u
n
d
to
y
ield
a
b
r
ief
ca
lcu
la
tio
n
o
f
W
av
elet
r
e
m
o
d
el
.
It
is
w
h
a
tev
er
h
o
w
e
v
er
to
u
g
h
to
ac
tu
alize
a
n
d
d
ec
r
ea
s
es
th
e
c
alcu
latio
n
ti
m
e
an
d
ass
et
s
r
eq
u
ir
ed
.
W
av
elet
alter
n
ate
d
ec
a
y
s
a
s
i
g
n
in
to
a
n
ar
r
an
g
e
m
en
t o
f
p
r
e
m
is
e
ca
p
ac
ities
.
3
.
1
.
Neura
l Sy
s
t
e
m
A
Ne
u
tr
al
f
r
a
m
e
w
o
r
k
is
a
co
u
r
s
e
o
f
ac
tio
n
o
f
ch
ar
ac
ter
i
s
ti
cs
all
r
elate
d
b
y
p
o
in
t
ch
an
g
e
s
th
at
h
a
v
e
in
d
is
ti
n
g
u
is
h
ab
le
li
m
i
t
o
r
f
itn
ess
[
1
]
.
E
ac
h
ce
n
ter
ad
d
r
ess
es
a
q
u
alit
y
g
at
h
er
i
n
g
a
n
d
ea
ch
lin
e
id
en
t
if
ie
s
w
it
h
th
e
ch
a
n
g
e
j
o
in
in
g
t
w
o
p
r
o
g
r
ess
io
n
s
.
Fair
-
mi
n
d
ed
f
r
a
m
e
wo
r
k
s
ca
n
b
e
co
n
s
id
er
ed
as
h
ig
h
,
le
v
el
lev
el
s
in
a
w
ellb
ei
n
g
s
ce
n
e.
I
n
t
h
e
m
id
s
t
o
f
u
n
p
r
ej
u
d
iced
ad
v
an
ce
m
en
t,
q
u
alitie
s
ca
n
s
el
f
-
as
s
e
r
tiv
el
y
g
o
th
r
o
u
g
h
o
b
j
ec
tiv
e
f
r
a
m
e
w
o
r
k
s
a
n
d
cr
o
s
s
ter
r
ito
r
ies
o
f
co
u
r
s
e
o
f
ac
ti
o
n
s
p
ac
e
w
h
ic
h
m
a
y
h
a
v
e
r
es
u
lts
f
o
r
f
o
r
ce
a
n
d
ev
o
lv
ab
ilit
y
.
Un
p
r
ej
u
d
iced
f
r
a
m
e
w
o
r
k
s
ex
is
t
i
n
h
ea
lt
h
s
ce
n
es
s
i
n
ce
p
r
o
tein
s
ar
e
s
o
lid
to
ch
an
g
es.
T
h
is
p
r
o
m
p
t
s
cr
ea
ted
f
r
a
m
e
wo
r
k
s
o
f
c
h
ar
ac
ter
is
t
ics
o
f
p
r
o
p
o
r
tio
n
ate
li
m
i
t,
as
s
o
ciate
d
b
y
f
air
m
u
tatio
n
s
[
2
-
3]
.
P
r
o
tein
s
ar
e
i
m
p
en
etr
ab
le
to
ch
an
g
e
s
f
o
llo
w
i
n
g
v
ar
io
u
s
g
r
o
u
p
in
g
s
ca
n
w
r
in
k
le
i
n
to
s
i
g
n
if
ica
n
tl
y
p
r
ac
ticall
y
id
en
tical
es
s
en
tia
l
f
o
ld
s
[
4
]
.
A
p
r
o
tein
g
r
asp
s
a
co
n
f
i
n
ed
ass
e
m
b
l
in
g
o
f
n
ea
r
b
y
co
m
p
lian
ce
s
s
i
n
ce
th
o
s
e
co
n
f
o
r
m
er
s
h
av
e
lo
w
er
i
m
p
e
r
ativ
en
e
s
s
t
h
an
cr
ea
ted
an
d
m
is
-
gi
v
en
w
a
y
s
tates
(
Δ
Δ
G
o
f
f
o
ld
in
g
)
[
5
-
6]
.
T
h
is
is
ex
p
er
t
b
y
a
s
ca
tter
e
d
,
in
ter
n
al
ar
r
an
g
e
m
en
t
o
f
s
u
p
p
o
r
tiv
e
af
f
ilia
tio
n
s
(
h
y
d
r
o
p
h
o
b
ic,
p
o
lar
an
d
co
v
alen
t)
[
7
]
.
P
r
o
tein
h
elp
er
h
ea
lt
h
in
e
s
s
r
esu
lts
f
r
o
m
f
e
w
s
in
g
le
c
h
a
n
g
e
s
b
ei
n
g
s
ati
s
f
ac
t
o
r
ily
d
an
g
er
o
u
s
to
ex
ch
a
n
g
e
o
f
f
li
m
it.
P
r
o
te
in
s
h
av
e
in
lik
e
m
an
n
er
p
r
o
g
r
ess
ed
to
av
o
id
ag
g
r
eg
at
io
n
[
8
]
as n
o
t c
o
m
p
letel
y
b
r
o
k
e
n
d
o
w
n
p
r
o
tein
s
ca
n
j
o
in
to
o
u
tli
n
e
g
i
g
an
tic,
r
eiter
atin
g
,
i
n
s
o
l
u
b
le
p
r
o
tein
f
ib
r
ils
a
n
d
m
as
s
es
[
9
]
.
T
h
er
e
is
af
f
ir
m
at
io
n
th
a
t
p
r
o
tein
s
s
h
o
w
n
eg
at
iv
e
f
r
a
m
e
w
o
r
k
p
ar
ts
to
d
im
i
n
i
s
h
t
h
e
p
r
esen
tatio
n
o
f
ag
g
r
eg
ate
s
lan
ted
b
eta
-
s
h
e
et
s
u
b
j
ec
ts
in
th
e
ir
s
tr
u
ctu
r
e
s
[
1
0
]
.
A
d
d
itio
n
all
y
,
t
h
er
e
is
s
o
m
e
co
n
f
ir
m
a
tio
n
t
h
at
th
e
in
h
er
ited
co
d
e
its
el
f
m
a
y
b
e
s
tr
ea
m
li
n
ed
s
u
ch
t
h
at
m
o
s
t
g
u
id
e
c
h
a
n
g
e
s
lead
to
w
ar
d
s
i
m
ilar
a
m
i
n
o
ac
id
s
(
co
n
s
er
v
at
iv
e)
[1
1
-
12]
.
T
o
g
eth
er
t
h
ese
co
m
p
o
n
en
ts
m
a
k
e
a
m
o
v
e
m
e
n
t
o
f
h
ea
lth
ef
f
ec
t
s
o
f
c
h
a
n
g
e
s
t
h
at
co
n
ta
in
s
a
h
ig
h
d
eg
r
ee
o
f
u
n
p
r
ej
u
d
iced
an
d
a
b
o
u
t
f
a
ir
ch
a
n
g
es.
Un
p
r
ej
u
d
iced
f
r
a
m
e
w
o
r
k
s
ar
e
a
s
u
b
s
e
t
o
f
t
h
e
g
r
o
u
p
i
n
g
s
i
n
p
r
o
g
r
ess
io
n
s
p
ac
e
t
h
at
h
a
v
e
r
elativ
e
li
m
i
t,
alo
n
g
th
e
s
e
li
n
e
s
s
h
ap
e
a
w
id
e,
le
v
el
i
n
a
h
ea
lth
s
ce
n
e.
I
m
p
ar
tial
p
r
o
g
r
ess
io
n
ca
n
in
t
h
is
w
a
y
b
e
im
a
g
i
n
ed
as
a
p
eo
p
le
d
if
f
u
s
i
n
g
f
r
o
m
o
n
e
co
u
r
s
e
o
f
ac
tio
n
o
f
co
llectio
n
ce
n
ter
s
,
th
r
o
u
g
h
th
e
u
n
p
r
ej
u
d
iced
f
r
am
e
w
o
r
k
,
to
a
n
o
th
er
p
ac
k
o
f
p
r
o
g
r
ess
io
n
ce
n
ter
s
.
E
n
s
u
i
n
g
to
th
e
b
i
g
g
er
p
iece
o
f
ad
v
an
ce
m
en
t
i
s
t
h
o
u
g
h
t
to
b
e
n
e
u
tr
al,
[
1
4
-
1
5
]
an
i
m
m
ea
s
u
r
ab
le
d
eg
r
ee
o
f
v
al
u
e
c
h
a
n
g
e
i
s
t
h
e
i
m
p
r
o
v
e
m
en
t
h
o
w
ev
er
c
lear
in
g
i
m
p
ar
tial
f
r
a
m
e
w
o
r
k
s
.
T
h
e
m
o
r
e
f
a
ir
n
e
ig
h
b
o
r
s
a
g
at
h
er
i
n
g
h
a
s
,
t
h
e
a
ll t
h
e
m
o
r
e
e
f
f
ec
ti
v
e
to
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
IJ
-
AI
Vo
l.
6
,
No
.
1
,
Ma
r
ch
2
0
1
7
:
56
–
65
60
ch
an
g
es
it
is
s
i
n
ce
c
h
an
g
es
will
m
o
s
t
li
k
el
y
b
asicall
y
f
air
-
m
i
n
d
ed
l
y
c
h
an
g
e
o
v
er
it
in
to
a
lik
e
w
i
s
e
u
tili
tar
ia
n
s
eq
u
en
ce
[
1
]
.
I
n
d
ee
d
,
if
t
h
er
e
ar
e
g
en
er
o
u
s
d
if
f
er
en
ce
s
b
et
wee
n
t
h
e
a
m
o
u
n
t o
f
i
m
p
ar
tial
n
e
ig
h
b
o
r
s
o
f
d
if
f
er
en
t
co
u
r
s
es
o
f
ac
tio
n
w
it
h
i
n
a
n
eu
tr
al
f
r
a
m
e
w
o
r
k
,
th
e
m
as
s
e
s
is
ex
p
ec
ted
to
cr
ea
te
to
w
ar
d
s
th
ese
in
cr
ed
ib
le
p
r
o
g
r
ess
io
n
s
.
T
h
is
i
s
r
ar
el
y
c
alled
cir
cu
m
-
ab
s
e
n
ce
o
f
p
r
ed
is
p
o
s
itio
n
a
n
d
id
en
ti
f
ies
w
it
h
t
h
e
i
m
p
r
o
v
e
m
e
n
t
o
f
m
as
s
es
a
lo
n
g
w
a
y
f
r
o
m
s
lo
p
es
in
t
h
e
h
ea
lth
s
ce
n
e.
E
ag
er
n
ess
f
o
r
th
e
tr
ad
e
b
et
w
ee
n
g
en
e
tic
b
u
o
y
a
n
d
d
eter
m
in
at
io
n
h
as
b
ee
n
ar
o
u
n
d
s
in
ce
t
h
e
1
9
3
0
s
w
h
en
t
h
e
m
o
v
in
g
e
v
e
n
in
g
o
u
t
s
p
ec
u
latio
n
p
r
o
p
o
s
ed
th
at
i
n
a
co
u
p
le
o
f
cir
c
u
m
s
ta
n
ce
s
,
i
n
h
e
r
ited
b
u
o
y
co
u
ld
s
u
p
p
o
r
t
later
f
le
x
ib
le
ev
o
l
u
tio
n
[
2
0
]
.
A
lt
h
o
u
g
h
t
h
e
s
p
ec
if
ic
s
o
f
th
e
th
eo
r
y
w
er
e
f
o
r
th
e
m
o
s
t
p
ar
t
d
is
cr
ed
ited
,
[
2
1
]
it
p
u
lled
in
in
s
ig
h
t
f
u
ln
e
s
s
w
it
h
r
esp
ec
t
to
th
e
p
r
o
b
ab
ilit
y
th
at
b
u
o
y
co
u
ld
m
a
k
e
d
ar
k
as
s
o
r
t
m
en
t
th
a
t,
h
o
w
ev
er
u
n
p
r
ej
u
d
iced
to
cu
r
r
en
t
li
m
it,
m
a
y
i
m
p
ac
t
d
ec
is
io
n
f
o
r
n
e
w
li
m
it
s
(
ev
o
lv
ab
ili
t
y
)
.
3.2.
R
adic
a
l
bas
i
s
f
un
ct
i
on
I
n
p
er
ce
p
tr
o
n
-
s
o
r
t
o
r
ch
est
r
ates
,
th
e
ac
tiv
atio
n
o
f
co
v
er
ed
u
n
i
ts
r
elies
o
n
u
p
o
n
t
h
e
s
p
o
t
th
i
n
g
b
et
w
ee
n
th
e
in
f
o
r
m
at
io
n
v
ec
to
r
an
d
a
w
ei
g
h
t
v
ec
to
r
.
I
n
th
i
s
lo
ca
tio
n
w
e
w
i
ll
lo
o
k
at
R
B
Fs
,
f
r
a
m
e
w
o
r
k
s
w
h
er
e
th
e
ac
tiv
atio
n
o
f
co
v
er
ed
u
n
its
r
e
lies
o
n
u
p
o
n
t
h
e
d
i
v
is
io
n
b
et
w
ee
n
t
h
e
i
n
f
o
r
m
a
tio
n
v
ec
to
r
an
d
a
m
o
d
el
v
ec
to
r
R
ad
ial
r
ea
s
o
n
li
m
it
s
h
a
v
e
d
i
f
f
er
en
t
c
h
ar
m
i
n
g
p
r
o
p
er
ties
.
T
h
er
e
ex
is
ts
s
tr
o
n
g
r
elatio
n
s
h
ip
w
it
h
d
if
f
er
en
t
in
v
e
s
ti
g
ati
v
e
co
n
tr
o
ls
.
T
h
ese
f
u
s
e
li
m
i
t
g
a
u
g
e,
r
eg
u
lar
izatio
n
th
eo
r
y
,
t
h
ick
n
e
s
s
es
ti
m
a
tio
n
a
n
d
in
s
er
tio
n
i
n
t
h
e
r
eg
io
n
o
f
u
p
r
o
ar
[
B
is
h
o
p
,
1
9
9
5
]
R
B
Fs
co
n
s
id
er
an
u
n
m
is
ta
k
ab
le
i
n
ter
p
r
etatio
n
o
f
th
e
in
t
er
n
al
r
ep
r
esen
tatio
n
m
ad
e
b
y
th
e
h
id
la
y
er
g
et
r
ea
d
y
co
m
p
u
tatio
n
s
f
o
r
R
B
Fs
ar
e
b
asicall
y
s
n
ap
p
ier
th
a
n
th
o
s
e
f
o
r
ML
P
s
A
n
d
,
a
s
w
e
w
i
ll
s
ee
to
d
a
y
,
m
o
s
t
o
f
t
h
e
s
e
co
u
n
ts
h
a
v
e
s
tar
t
i
n
g
n
o
w
b
ee
n
p
r
esen
ted
i
n
p
ast
ad
d
r
ess
e
s
.
Ou
t
s
p
r
ea
d
r
ea
s
o
n
li
m
it
s
ar
e
s
u
s
ten
a
n
ce
f
o
r
w
ar
d
f
r
a
m
e
w
o
r
k
s
co
n
tai
n
i
n
g
a
co
v
e
r
ed
lay
er
o
f
w
in
d
i
n
g
s
eg
m
e
n
t
s
an
d
a
y
ield
la
y
er
o
f
s
tr
aig
h
t
n
e
u
r
o
n
s
.
T
h
e
t
w
o
R
B
F
la
y
er
s
p
as
s
o
n
ab
s
o
l
u
tel
y
p
ar
tic
u
lar
p
ar
ts
[
Ha
y
k
i
n
,
1
9
9
9
]
.
T
h
e
co
v
er
ed
la
y
er
p
er
f
o
r
m
s
a
n
o
n
-
s
tr
ai
g
h
t
ch
an
g
e
o
f
d
ata
s
p
ac
e
th
e
co
n
s
eq
u
e
n
t
co
v
er
ed
s
p
ac
e
is
o
r
d
in
ar
il
y
o
f
h
ig
h
er
d
i
m
en
s
io
n
al
it
y
t
h
a
n
t
h
e
d
ata
s
p
ac
e.
T
h
e
y
ield
la
y
er
p
er
f
o
r
m
s
s
tr
aig
h
t
b
ac
k
s
lid
e
to
en
v
i
s
io
n
t
h
e
p
in
ed
f
o
r
tar
g
ets
.
W
h
y
u
s
e
a
n
o
n
-
d
ir
ec
t
ch
an
g
e
to
o
k
af
ter
b
y
a
s
tr
ai
g
h
t
o
n
e?
Sp
r
ea
d
'
s
th
eo
r
y
o
n
t
h
e
d
is
ti
n
ctn
e
s
s
o
f
ca
s
es
"
A
b
r
ain
b
o
g
g
li
n
g
o
u
t
lin
e
r
e
q
u
est
is
s
u
e
ca
s
t
in
a
h
i
g
h
-
d
im
en
s
io
n
a
l
s
p
ac
e
n
o
n
-
s
tr
ai
g
h
t
f
o
r
w
ar
d
l
y
w
ill
m
o
s
t
lik
el
y
b
e
s
tr
aig
h
tl
y
u
n
m
is
ta
k
ab
le
th
a
n
i
n
a
lo
w
-
d
i
m
e
n
s
io
n
al
s
p
ac
e"
A
s
w
e
w
i
ll
d
i
s
co
v
er
i
n
t
w
o
o
r
t
h
r
ee
ad
d
r
ess
es,
th
is
to
a
g
r
ea
t
d
eg
r
ee
s
a
m
e
d
is
p
u
te
is
at
th
e
f
o
ca
l
p
o
in
t
o
f
B
o
ls
ter
Vec
to
r
Ma
c
h
in
e
s
.
R
B
Fs
ar
e
to
m
ak
e
s
u
r
e
o
n
e
o
f
th
e
b
it
li
m
it
s
m
o
s
t
t
y
p
ical
l
y
u
s
ed
as
a
p
iece
o
f
SVMs!
R
B
Fs
ar
e
g
en
er
a
ll
y
ar
r
an
g
ed
af
ter
a
m
u
tt
f
r
a
m
e
w
o
r
k
th
at
w
o
r
k
s
in
t
w
o
s
tag
e
s
o
r
tim
e
s
ca
les
[
Ha
y
k
i
n
,
1
9
9
9
]
Un
s
u
p
er
v
is
ed
d
eter
m
in
a
tio
n
o
f
R
B
F
co
r
e
in
ter
ests
a.
R
B
F
ce
n
ter
s
ar
e
p
ick
ed
w
i
th
a
s
p
ec
if
ic
e
n
d
g
o
al
to
ar
r
an
g
e
th
e
s
ca
tter
i
n
g
o
f
p
la
n
n
i
n
g
ca
s
es
i
n
th
e
d
ata
h
ig
h
li
g
h
t sp
ac
e
b.
T
h
is
is
th
e
f
u
n
d
a
m
en
ta
l
s
tep
i
n
g
et
r
ea
d
y
,
r
eg
u
lar
l
y
p
er
f
o
r
m
ed
in
a
m
o
d
er
ate
iter
ativ
e
m
an
n
er
c.
Fo
r
tu
n
ate
l
y
,
d
if
f
er
e
n
t s
y
s
te
m
s
s
h
o
w
ed
in
p
ast lo
ca
tio
n
s
ca
n
b
e
u
tili
ze
d
to
d
ea
l
w
it
h
t
h
is
i
s
s
u
e
d.
R
eg
u
lated
f
ig
u
r
i
n
g
o
f
y
ield
v
e
cto
r
s
e.
Hid
d
en
-
to
-
y
ie
ld
w
ei
g
h
t
v
ec
to
r
s
ar
e
s
et
o
u
t
to
m
i
n
i
m
ize
t
h
e
en
tire
s
q
u
ar
ed
s
cr
e
w
u
p
b
et
w
ee
n
t
h
e
R
B
F
y
ield
s
a
n
d
th
e
lo
o
k
ed
f
o
r
tar
g
e
ts
f.
Sin
ce
t
h
e
y
ie
ld
s
ar
e
s
tr
aig
h
t,
t
h
e
p
er
f
ec
t
w
ei
g
h
t
s
ca
n
b
e
p
r
ep
ar
ed
.
3.3.
Mul
t
i
-
l
ay
er
pe
rc
ep
t
ron
I
n
th
e
ac
co
m
p
a
n
y
in
g
t
w
o
lo
ca
tio
n
s
w
e
w
i
ll
lo
o
k
at
M
u
lti
-
L
a
y
er
P
er
ce
p
tr
o
n
s
(
ML
P
s
)
w
h
ic
h
ar
e
m
o
r
e
v
iab
le
th
a
n
t
h
e
Si
n
g
le
-
L
a
y
er
m
o
d
el
s
w
h
ich
f
ab
r
icate
s
tr
ai
g
h
t
d
ec
is
io
n
li
m
its
.
I
t
w
i
ll
b
e
s
e
en
th
a
t
M
L
P
s
ca
n
b
e
ar
r
an
g
ed
as
a
d
is
cr
i
m
i
n
ati
v
e
m
o
d
el
to
y
ield
clas
s
b
ac
k
s
id
es.
M
L
P
s
ar
e
n
a
m
ed
a
s
o
r
t
o
f
A
r
ti
f
icial
Neu
r
al
Net
w
o
r
k
:
t
h
e
co
m
p
u
tatio
n
is
p
er
f
o
r
m
ed
u
s
in
g
a
co
u
r
s
e
o
f
ac
tio
n
o
f
(
v
ar
io
u
s
)
d
ir
ec
t
u
n
its
w
it
h
w
ei
g
h
ted
r
elatio
n
s
h
ip
b
et
w
ee
n
th
e
m
.
I
n
ad
d
itio
n
th
er
e
ar
e
lear
n
in
g
f
ig
u
r
in
g
s
to
s
et
th
e
esti
m
ati
o
n
s
o
f
th
e
w
ei
g
h
ts
ad
d
itio
n
all
y
,
t
h
e
s
a
m
e
b
asic
s
tr
u
ct
u
r
e
(
w
it
h
d
if
f
er
en
t
weig
h
t
q
u
alitie
s
)
is
p
r
ep
ar
ed
to
p
er
f
o
r
m
v
ar
io
u
s
ass
i
g
n
m
e
n
ts
.
I
n
th
i
s
a
n
d
t
h
e
g
o
in
g
w
it
h
lo
ca
tio
n
w
e
w
ill
co
n
s
id
er
O
v
er
all
s
tr
u
c
tu
r
e
o
f
m
u
lti
-
la
y
er
p
er
ce
p
tr
o
n
s
Dec
is
io
n
co
n
f
in
e
s
t
h
at
th
e
y
c
an
o
u
t
lin
e,
T
r
ain
in
g
C
r
iter
ia
,
Net
w
o
r
k
s
a
s
b
ac
k
p
r
o
b
ab
i
lity
esti
m
ato
r
s
,
B
asi
c
E
r
r
o
r
b
ac
k
-
s
p
r
ea
d
g
et
r
ea
d
y
f
i
g
u
r
i
n
g
,
I
m
p
r
o
v
ed
g
et
r
ea
d
y
m
eth
o
d
o
lo
g
ies
,
Dee
p
Ne
u
r
al
f
r
am
e
w
o
r
k
s
Fro
m
th
e
p
ast
s
lid
e
w
e
ca
n
s
ee
th
at
th
e
a
m
o
u
n
t
o
f
co
n
ce
aled
la
y
er
s
ch
o
o
s
es
t
h
e
d
ec
is
io
n
f
u
r
t
h
est
r
ea
ch
es
t
h
at
ca
n
b
e
m
ad
e.
I
n
p
ick
i
n
g
t
h
e
a
m
o
u
n
t o
f
la
y
er
s
th
e
g
o
in
g
w
i
t
h
t
h
o
u
g
h
ts
ar
e
m
ad
e.
a.
Mu
lti
-
la
y
er
f
r
a
m
e
w
o
r
k
s
ar
e
h
a
r
d
er
to
g
et
r
ea
d
y
th
a
n
s
i
n
g
le
la
y
er
f
r
a
m
e
w
o
r
k
s
.
b.
A
t
w
o
la
y
er
f
r
a
m
e
w
o
r
k
(
o
n
e
c
o
n
ce
aled
)
w
it
h
s
i
g
m
o
id
al
s
tar
t
li
m
it
s
ca
n
d
e
m
o
n
s
tr
ate
a
n
y
d
e
cisi
o
n
li
m
it
.
T
w
o
la
y
er
f
r
a
m
e
w
o
r
k
s
ar
e
m
o
s
t
t
y
p
icall
y
u
s
ed
as
a
p
iece
o
f
illu
s
tr
atio
n
af
f
ir
m
atio
n
(
t
h
e
co
v
er
ed
la
y
er
h
a
v
i
n
g
s
ig
m
o
id
al
in
c
itati
o
n
li
m
i
ts
)
.
W
h
at
n
u
m
b
er
o
f
u
n
its
to
h
a
v
e
in
ea
c
h
la
y
er
?
a.
T
h
e
am
o
u
n
t o
f
y
ield
u
n
i
ts
is
c
o
n
s
is
ten
tl
y
co
n
tr
o
lled
b
y
t
h
e
q
u
an
t
it
y
o
f
y
ie
ld
class
es.
b.
T
h
e
am
o
u
n
t o
f
in
p
u
ts
i
s
co
n
tr
o
lled
b
y
t
h
e
a
m
o
u
n
t o
f
in
f
o
r
m
atio
n
esti
m
atio
n
s
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
A
Dec
is
io
n
S
ystem
fo
r
P
r
ed
ictin
g
Dia
b
etes u
s
in
g
N
eu
r
a
l Netw
o
r
ks
…
(
K
.
C
h
a
n
d
a
n
a
R
a
n
i
)
61
c.
T
h
e
a
m
o
u
n
t
o
f
h
id
u
n
its
is
a
f
r
a
m
e
w
o
r
k
i
s
s
u
e.
T
h
e
is
s
u
es
ar
e:
ex
o
r
b
itan
tl
y
f
e
w
,
th
e
f
r
am
e
w
o
r
k
w
o
n
't
s
h
o
w
co
m
pl
ex de
c
i
s
i
on
l
i
m
i
t
s;
exc
es
s
i
v
el
y
, t
he
f
r
am
ework
.
4.
M
O
RP
H
O
L
O
G
I
CA
L
P
RO
CE
SS
I
N
G
E
r
o
s
io
n
in
v
o
lv
e
s
t
h
e
eli
m
i
n
ati
o
n
(
alter
atio
n
)
o
f
p
ix
el
s
at
t
h
e
r
i
m
s
o
f
ar
ea
s
,
a
s
a
n
ex
a
m
p
le
ch
an
g
i
n
g
b
in
ar
y
1
v
al
u
e
to
0
,
at
th
e
s
a
m
e
ti
m
e
as
d
ilat
io
n
i
s
t
h
e
o
p
p
o
s
ite
m
a
n
n
er
w
it
h
ar
ea
s
g
r
o
w
i
n
g
o
u
t
f
r
o
m
t
h
eir
b
ar
r
ier
s
.
T
h
o
s
e
m
et
h
o
d
s
ar
e
o
f
ten
f
in
is
h
ed
u
s
i
n
g
a
f
o
r
m
o
f
k
er
n
el
ca
lled
a
s
tr
u
ct
u
r
al
d
etail.
A
s
tr
u
c
tu
r
al
ele
m
e
n
t
is
an
N
×
N
k
er
n
el
w
i
th
en
tr
ies
ca
te
g
o
r
is
ed
in
lin
e
w
it
h
a
b
in
ar
y
s
c
h
e
m
e,
n
o
r
m
all
y
a
s
ze
r
o
o
r
1
.
I
f
all
en
tr
ies
ar
e
co
d
ed
1
th
en
th
e
s
tr
u
ctu
r
al
d
etail
is
a
s
tr
o
n
g
s
q
u
ar
e
b
lo
ck
,
th
e
m
id
d
le
o
f
th
at
is
l
aid
o
v
er
ea
ch
p
ix
el
in
s
id
e
t
h
e
s
u
p
p
l
y
p
ictu
r
e
i
n
f
l
i
p
.
S
h
o
w
n
i
n
Fi
g
u
r
e
4
.
Fig
u
r
e
4
.
T
h
e
Stru
ctu
r
al
Deta
i
l is a
Stro
n
g
Sq
u
ar
e
B
lo
ck
T
h
e
k
in
g
d
o
m
o
f
t
h
e
b
asic
f
a
cto
r
w
o
u
ld
p
o
s
s
ib
l
y
d
if
f
er
,
f
o
r
in
s
tan
ce
as
a
v
er
tical
b
ar
,
d
eg
r
ee
b
ar
,
m
o
v
e
s
h
ap
e
o
r
a
p
u
r
ch
aser
c
h
ar
ac
ter
ized
la
y
o
u
t.
I
n
s
id
e
th
e
ev
en
t
t
h
at
e
n
lar
g
e
m
e
n
t
i
s
t
r
ailed
b
y
m
ea
n
s
o
f
d
is
in
te
g
r
atio
n
th
e
m
an
n
er
i
s
p
o
r
tr
ay
ed
as
a
f
i
n
al
o
p
er
atio
n
,
w
h
il
s
t E
r
o
s
io
n
to
o
k
a
f
ter
b
y
m
ea
n
s
o
f
e
n
lar
g
e
m
e
n
t
is
k
n
o
w
n
a
s
o
p
en
i
n
g
.
T
h
ese
s
t
r
ateg
ies
ar
e
ac
tu
a
ll
y
n
o
t
s
y
m
m
etr
ic,
a
n
d
o
n
th
is
w
a
y
ar
e
f
o
r
th
e
m
o
s
t
p
ar
t
n
o
w
n
o
t
r
ev
er
s
ib
le.
s
tar
ti
n
g
w
ip
es
o
u
t
litt
le
a
n
d
ex
tr
a
s
li
m
co
m
p
o
n
en
t
s
,
b
r
in
g
i
n
g
ap
p
r
o
x
i
m
ate
l
y
s
m
o
o
t
h
er
ed
g
ed
ar
ea
s
,
ev
en
a
s
s
h
u
tti
n
g
lik
e
w
is
e
s
m
o
o
th
e
s
t
s
h
ap
es
h
o
w
ev
er
m
ak
e
s
n
ar
r
o
w
co
n
s
tr
ain
ed
co
m
p
o
n
en
ts
lar
g
er
an
d
w
ip
es o
u
t little
g
ap
s
an
d
r
estra
in
ed
cr
ev
ices.
Gu
id
e
v
ec
to
r
d
ev
ice
(
SVM)
i
s
a
d
ir
ec
ted
lear
n
i
n
g
v
er
s
io
n
w
it
h
a
ass
o
ciate
d
lear
n
i
n
g
f
i
g
u
r
in
g
t
h
at
co
u
ld
s
ep
ar
ate
in
f
o
r
m
at
io
n
a
n
d
n
o
tice
o
u
tli
n
es
t
h
at
ar
e
t
h
en
u
s
ed
f
o
r
b
ac
k
s
lid
e
e
x
a
m
i
n
atio
n
ad
d
itio
n
all
y
,
co
llectin
g
.
I
f
w
e
ta
k
e
a
d
ir
ec
tio
n
o
f
m
o
v
e
m
en
t
o
f
p
lan
n
i
n
g
in
s
ta
n
ce
s
,
w
h
er
ei
n
ea
ch
i
s
s
e
p
a
r
ate
as
f
itti
n
g
i
n
w
it
h
o
n
e
o
f
p
o
r
tr
ay
ed
g
r
o
u
p
i
n
g
s
,
a
SVM
g
et
p
r
ep
ar
ed
c
o
u
n
t
w
ill
b
u
ild
i
n
g
u
p
a
v
er
s
io
n
to
b
e
ab
le
t
o
d
is
p
en
s
e
n
e
w
ca
s
es
to
o
n
e
o
r
th
e
o
th
er
ass
o
ciatio
n
,
m
ak
in
g
it
n
o
n
-
p
r
o
b
ab
ilis
tic
t
w
o
f
o
ld
d
ir
ec
t
class
i
f
ier
.
SVM
m
o
d
el
i
s
th
e
i
llu
s
tr
atio
n
o
f
ca
s
es
d
ef
i
n
ed
as
ce
n
ter
s
i
n
s
p
ac
e
w
h
i
ch
ca
n
b
e
m
ap
p
ed
s
o
t
h
at
t
h
e
i
n
s
ta
n
ce
s
o
f
t
h
e
n
u
m
er
o
u
s
ch
ar
ac
ter
izatio
n
s
m
a
y
b
e
p
ar
ce
led
b
y
m
ea
n
s
o
f
an
u
n
m
is
ta
k
ab
le
clef
t
t
h
i
s
is
a
s
i
m
m
ea
s
u
r
ab
le
as
m
a
y
w
an
t
to
s
en
s
ib
l
y
b
e
r
eg
u
l
ar
.
s
p
ar
k
lin
g
s
p
ec
i
m
en
s
ar
e
th
en
m
ap
p
ed
in
to
t
h
at
s
a
m
e
ar
e
a
an
d
f
o
r
esee
n
th
a
t
m
i
g
h
t
h
a
v
e
a
n
ic
h
e
w
i
th
an
o
r
d
er
co
n
s
id
er
in
g
w
h
ic
h
f
ac
e
t
o
f
t
h
e
cle
f
t
t
h
e
y
f
all
at
th
e
h
elp
Vec
to
r
d
ev
ice
(
SVM)
is
a
e
x
ce
p
tio
n
al
i
n
e
leg
an
ce
a
m
as
s
in
g
tech
n
iq
u
e
d
is
p
la
y
ed
i
n
1
9
9
2
t
h
r
o
u
g
h
B
o
s
er
,
Gu
y
o
n
,
an
d
Vap
n
ik
.
T
h
e
S
VM
clas
s
i
f
ier
is
f
o
r
th
e
m
o
s
t
ele
m
en
t
s
u
s
ed
as
a
b
it o
f
b
io
in
f
o
r
m
atic
s
(
ex
ce
p
t,
co
n
tr
o
ls
)
in
v
ie
w
o
f
its
e
x
ce
s
s
i
v
e
p
r
ec
is
io
n
.
I
t
m
a
y
f
in
d
o
u
t a
n
d
m
an
a
g
e
t
h
e
e
x
ce
s
s
i
v
e
-
d
i
m
e
n
s
io
n
al
f
ac
ts
,
as a
n
ex
a
m
p
le,
q
u
ali
t
y
ex
p
r
ess
io
n
a
n
d
ex
ib
ilit
y
in
s
h
o
w
i
n
g
d
i
f
f
er
e
n
t
w
ells
p
r
i
n
g
s
o
f
s
tati
s
tics
.
A
tr
en
d
y
o
r
d
er
n
am
ed
as
co
m
p
o
n
en
t
s
tr
u
ct
u
r
es,
SVM
i
n
s
h
ap
e
in
with
t
h
at
g
r
o
u
p
i
n
g
.
T
h
is
p
ar
t
f
r
a
m
e
w
o
r
k
is
a
n
esti
m
at
io
n
w
h
ic
h
d
ep
en
d
s
o
n
u
p
o
n
d
ata
m
ai
n
l
y
t
h
r
o
u
g
h
s
p
o
t
t
h
i
n
g
.
Fo
r
th
is
s
it
u
atio
n
,
a
s
ec
ti
o
n
r
estrict
ca
n
s
u
p
p
la
n
t
t
h
i
s
s
p
o
t
is
s
u
e
a
n
d
it
'
l
l
d
is
ce
r
n
t
h
e
b
it
co
m
p
o
n
e
n
t
i
n
s
o
m
e
h
i
g
h
d
i
m
e
n
s
io
n
al
s
e
g
m
en
t
s
p
ac
e
p
er
h
ap
s
.
I
t
h
as
t
w
o
p
u
r
p
o
s
es
o
f
h
o
b
b
y
:
f
ir
s
tl
y
,
it
m
a
y
d
eli
v
er
n
o
n
-
d
ir
ec
t
s
elec
tio
n
li
m
i
ts
t
h
e
u
s
e
o
f
th
e
o
n
es
f
r
a
m
e
w
o
r
k
s
w
h
ic
h
ca
n
b
e
p
r
o
p
o
s
ed
f
o
r
d
ir
ec
t
class
i
f
ier
s
.
Mo
r
eo
v
er
,
u
s
e
o
f
b
it
li
m
it
s
w
ill
al
lo
w
th
e
cu
s
to
m
er
to
ap
p
ly
c
lass
if
ier
t
o
th
e
s
tati
s
tic
s
th
a
t
h
av
e
n
'
t
a
n
y
d
escr
ib
ed
alter
ed
d
i
m
en
s
io
n
al
v
ec
to
r
s
p
ac
e
ill
u
s
tr
atio
n
.
T
h
e
m
ax
i
m
u
m
f
itt
in
g
s
p
ec
i
m
e
n
o
f
s
u
ch
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
IJ
-
AI
Vo
l.
6
,
No
.
1
,
Ma
r
ch
2
0
1
7
:
56
–
65
62
r
ec
o
r
d
s
in
b
io
in
f
o
r
m
atic
s
is
p
r
o
g
r
ess
io
n
,
eith
er
,
p
r
o
tein
o
r
DNA
t
h
en
a
g
ai
n
p
r
o
tein
s
t
r
u
ctu
r
e.
T
h
e
u
s
e
o
f
SVMs
f
ea
s
ib
l
y
r
eq
u
ir
es
an
ap
p
r
ec
iatio
n
o
f
w
a
y
s
t
h
e
y
p
ai
n
ti
n
g
s
.
ev
e
n
as
p
u
tti
n
g
in
p
lace
a
SVM
th
e
p
r
o
n
ee
d
s
to
s
ettle
o
n
d
is
ti
n
ct
s
elec
tio
n
s
:
th
at
is
t
h
e
s
ati
s
f
ac
to
r
y
tech
n
iq
u
e
to
p
r
ep
r
o
ce
s
s
s
tatis
tics
,
w
h
at
p
iece
o
u
g
h
t
to
b
e
u
s
ed
,
an
d
in
s
id
e
t
h
e
r
e
m
ai
n
in
g
,
p
u
tt
in
g
i
n
p
lace
th
e
p
ar
a
m
eter
s
o
f
SVM
an
d
p
iece
.
Du
m
b
f
o
u
n
d
ed
p
ick
s
ca
n
also
r
ea
p
to
a
o
u
ts
tan
d
i
n
g
d
ip
lo
m
a
r
ed
u
ce
d
e
x
ec
u
t
io
n
.
O
u
r
p
o
in
t
is
to
o
u
t
f
it
t
h
e
clie
n
t
w
it
h
a
e
n
tire
n
o
tio
n
o
f
t
h
e
s
elec
tio
n
s
f
u
r
th
er
m
o
r
e
t
o
o
f
f
er
s
ta
n
d
ar
d
u
s
e
r
eg
u
latio
n
s
.
E
ac
h
o
n
e
o
f
t
h
e
s
p
ec
i
m
en
s
ex
h
ib
ited
h
ad
b
ee
n
in
tr
o
d
u
ce
d
u
s
i
n
g
t
h
e
P
y
M
L
d
ev
ice
m
aster
i
n
g
e
n
v
ir
o
n
m
e
n
t,
w
h
ic
h
f
o
c
u
s
e
s
o
n
co
m
p
o
n
en
t
tec
h
n
iq
u
e
s
a
n
d
SVMs.
S
h
o
w
n
i
n
Fi
g
u
r
e
5
.
P
R
OB
A
B
I
L
I
ST
I
C
NE
UR
AL
NE
T
W
OR
K
P
NN
is
co
n
s
tan
t
l
y
u
s
ed
as
a
b
it
o
f
g
a
m
e
p
lan
i
s
s
u
es
[
4
]
w
h
ile
an
d
ata
is
o
u
t
t
h
er
e,
th
e
m
ai
n
la
y
er
r
eg
is
ter
s
t
h
e
d
etac
h
m
e
n
t
f
r
o
m
th
e
f
ac
ts
v
ec
to
r
to
th
e
ar
r
an
g
e
m
e
n
t
f
ac
t
s
v
ec
to
r
s
.
T
h
is
co
n
v
e
y
s
a
v
ec
to
r
w
h
er
ein
its
p
ar
ts
i
n
d
icate
h
o
w
clo
s
ed
th
e
r
ec
o
r
d
s
is
to
th
e
p
la
n
n
i
n
g
s
ta
tis
tic
s
.
T
h
e
s
ec
o
n
d
o
n
e
la
y
er
w
h
o
les t
h
e
d
eter
m
in
at
io
n
f
o
r
ea
ch
m
a
g
n
i
f
ice
n
ce
o
f
i
n
p
u
ts
a
n
d
co
n
v
e
y
s
its
in
ter
n
et
y
ield
as
a
v
ec
t
o
r
o
f
p
r
o
b
ab
ilit
ies.
Su
b
s
eq
u
e
n
tl
y
,
a
f
i
g
h
t
ch
a
n
g
e
r
estrict
at
th
e
y
ield
o
f
t
h
e
s
ec
o
n
d
la
y
er
p
ick
s
th
e
f
i
n
e
o
f
th
e
s
e
p
o
s
s
ib
ilit
ies,
a
n
d
p
r
o
d
u
ce
s
a
1
(
ef
f
ec
tiv
e
id
e
n
ti
f
i
ca
tio
n
)
f
o
r
th
at
ele
g
an
ce
a
n
d
a
0
(
n
eg
ativ
e
id
)
f
o
r
n
o
n
-
f
o
cu
s
e
d
o
n
class
es.
I
n
f
o
r
m
a
tio
n
la
y
er
:
E
ac
h
n
e
u
r
o
n
i
n
th
e
i
n
f
o
r
m
ati
o
n
la
y
er
id
en
ti
f
ie
s
w
it
h
a
m
a
r
k
er
v
ar
iab
le.
I
n
o
b
v
io
u
s
v
ar
i
ab
les,
N
-
1
n
eu
r
o
n
s
ar
e
u
s
ed
w
h
en
t
h
er
e
ar
e
N
n
u
m
b
er
o
f
g
r
o
u
p
in
g
s
.
I
t
r
eg
u
la
tes
t
h
e
ex
te
n
t
o
f
t
h
e
q
u
al
ities
b
y
s
u
b
tr
ac
tin
g
th
e
ce
n
ter
a
n
d
d
is
co
n
n
ec
ti
n
g
b
y
t
h
e
i
n
ter
q
u
ar
tile
r
a
n
g
e.
B
y
t
h
en
th
e
d
ata
n
e
u
r
o
n
s
s
u
p
p
o
r
t
th
e
q
u
al
ities
to
ea
c
h
o
f
th
e
n
eu
r
o
n
s
in
t
h
e
co
v
er
ed
l
a
y
er
.
P
lan
la
y
er
:
T
h
is
la
y
er
co
n
ta
in
s
o
n
e
n
eu
r
o
n
f
o
r
ea
ch
ca
s
e
i
n
t
h
e
r
ea
d
in
e
s
s
d
ata
s
et.
I
t
s
to
r
es
th
e
e
s
ti
m
at
io
n
s
o
f
th
e
m
ar
k
er
v
ar
iab
les
f
o
r
t
h
e
ca
s
e
n
ea
r
b
y
t
h
e
g
o
al
w
o
r
th
.
A
d
i
s
g
u
is
ed
n
eu
r
o
n
e
n
r
o
lls
t
h
e
E
u
c
li
d
ea
n
p
ar
titi
o
n
o
f
t
h
e
ex
a
m
in
at
io
n
f
r
o
m
t
h
e
n
eu
r
o
n
'
s
ce
n
ter
p
o
in
t
a
n
d
af
ter
t
h
at
ap
p
lies
t
h
e
R
B
F
s
e
g
m
e
n
t
li
m
it
u
s
in
g
t
h
e
s
ig
m
a
v
alu
e
s
.
Su
m
m
atio
n
la
y
er
:
Fo
r
P
NN
f
r
a
m
e
w
o
r
k
s
t
h
er
e
is
o
n
e
ill
u
s
tr
atio
n
n
e
u
r
o
n
f
o
r
ea
ch
g
r
o
u
p
in
g
o
f
t
h
e
g
o
al
v
ar
iab
le.
T
h
e
g
en
u
i
n
e
tar
g
et
ch
ar
ac
ter
iz
atio
n
o
f
ea
ch
r
ea
d
in
e
s
s
ca
s
e
i
s
s
ec
u
r
ed
w
i
th
ea
c
h
h
id
d
en
n
e
u
r
o
n
;
t
h
e
w
e
ig
h
ted
q
u
alit
y
leav
in
g
a
co
v
er
ed
n
e
u
r
o
n
is
m
a
n
a
g
ed
j
u
s
t
to
t
h
e
s
a
m
p
le
n
eu
r
o
n
th
a
t
id
en
ti
f
ie
s
w
i
th
t
h
e
co
v
er
ed
n
eu
r
o
n
's g
r
o
u
p
i
n
g
.
T
h
e
s
a
m
p
l
e
n
eu
r
o
n
s
in
co
r
p
o
r
ate
th
e
q
u
al
ities
f
o
r
t
h
e
class
t
h
e
y
id
en
ti
f
y
w
it
h
.
Yield
la
y
er
:
T
h
e
y
ield
la
y
er
co
n
s
id
er
s
t
h
e
w
e
ig
h
ted
v
o
tes
f
o
r
ea
ch
tar
g
et
class
ac
cu
m
u
lated
i
n
t
h
e
il
lu
s
tr
atio
n
la
y
er
an
d
u
s
es t
h
e
g
r
ea
te
s
t v
o
t
e
t
o
p
r
ed
ict
th
e
g
o
al
ar
r
an
g
e
m
en
t.
A
s
s
h
o
w
n
in
F
ig
u
r
e
6
to
Fig
u
r
e
9
.
5.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
e
ex
ec
u
tio
n
o
f
o
u
r
u
s
e
o
f
th
e
Q
A
i
s
ev
al
u
ated
w
it
h
a
w
e
ll
k
n
o
w
n
b
e
n
ch
m
ar
k
in
g
tech
n
iq
u
e.
T
h
e
en
tire
QA
s
h
ap
e
i
s
h
o
ld
s
tr
o
llin
g
o
n
1
3
0
p
ics
o
p
en
f
r
o
m
"
DI
AR
E
T
DB
0
:
ass
ess
m
en
t
Data
b
ase
an
d
tech
n
iq
u
e
f
o
r
Diab
etic
R
eti
n
o
p
ath
y
"
.
[
3
1
]
A
cc
o
r
d
in
g
l
y
b
y
m
ea
n
s
o
f
m
ak
i
n
g
u
s
e
o
f
th
e
p
r
o
p
o
s
ed
d
atab
ase
,
it
is
w
it
h
o
u
t
f
u
r
th
er
ad
o
p
o
s
s
ib
le
to
d
o
n
'
t
f
o
r
g
et
o
u
r
f
i
g
u
r
i
n
g
s
,
an
d
co
r
r
esp
o
n
d
in
g
l
y
,
d
is
co
v
er
th
eir
ad
v
an
ce
m
en
t
f
o
r
d
ev
elo
p
m
e
n
t
alter
n
a
te
f
r
o
m
th
e
ex
a
m
i
n
a
tio
n
r
esear
c
h
f
o
cu
s
es
to
t
h
e
r
e
m
ed
ial
p
r
ac
tice.
T
h
e
r
an
g
e
o
f
th
e
p
h
o
to
s
in
th
is
d
atab
ase
is
1
5
0
0
×
1
1
5
2
.
F
o
r
th
e
DI
A
R
E
T
DB
0
d
ata
b
ase
th
e
ef
f
ec
ts
o
f
t
h
e
g
ath
er
i
n
g
m
et
h
o
d
ar
e
s
h
o
w
ed
u
p
in
T
ab
le
1
an
d
T
a
b
le
2
e
x
h
ib
it
s
t
h
e
e
n
d
r
esu
l
t
o
f
Af
f
e
ctab
ilit
y
,
Sp
ec
i
f
icit
y
an
d
p
er
ce
n
t o
f
p
r
ec
is
io
n
t
h
e
u
s
e
o
f
T
ab
le
1
.
R
esu
lt o
f
Se
n
s
iti
v
it
y
,
Sp
ec
if
icit
y
,
% o
f
A
cc
u
r
a
y
T
ab
le
2
.
E
x
h
ib
its
th
e
e
n
d
R
es
u
lt o
f
A
f
f
ec
tab
ili
t
y
,
Sp
ec
i
f
icit
y
an
d
P
er
ce
n
t o
f
P
r
ec
is
io
n
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
A
Dec
is
io
n
S
ystem
fo
r
P
r
ed
ictin
g
Dia
b
etes u
s
in
g
N
eu
r
a
l Netw
o
r
ks
…
(
K
.
C
h
a
n
d
a
n
a
R
a
n
i
)
63
Fig
u
r
e
5
.
R
OC
C
u
r
v
e
f
o
r
C
o
m
p
ar
is
o
n
o
f
P
NN
b
ay
e
s
an
d
SV
M
Scre
en
s
ho
t
s
:
Scre
en
s
ho
t
s
:
(
a)
(
b
)
Fig
u
r
e
6
.
Sn
ap
s
h
o
t o
f
an
E
y
e
I
m
ag
e
a
f
ter
Hi
s
to
g
r
a
m
E
q
u
alis
atio
n
(
a)
P
r
o
ce
s
s
o
f
P
NN
R
esu
lts
(
b
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
IJ
-
AI
Vo
l.
6
,
No
.
1
,
Ma
r
ch
2
0
1
7
:
56
–
65
64
Fig
u
r
e
7
.
E
y
e
I
m
ag
e
a
f
ter
Ma
t
ch
ed
Fil
ter
s
SCRE
E
NSH
O
T
S F
O
R
SVM
CL
ASS
I
F
I
C
AT
I
O
N
Fig
u
r
e
8
.
L
o
ad
ed
C
o
lo
u
r
E
y
e
I
m
ag
e
Fig
u
r
e
9
.
E
y
e
I
m
ag
e
a
f
ter
His
t
o
g
r
a
m
E
q
u
ali
s
atio
n
s
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
A
Dec
is
io
n
S
ystem
fo
r
P
r
ed
ictin
g
Dia
b
etes u
s
in
g
N
eu
r
a
l Netw
o
r
ks
…
(
K
.
C
h
a
n
d
a
n
a
R
a
n
i
)
65
RE
F
E
R
E
NC
E
S
[1
]
D.
V
a
ll
a
b
h
a
,
R.
Do
ra
iraj,
K.
R.
Na
m
u
d
u
ri
a
n
d
H.T
h
o
m
p
so
n
,
“
A
u
to
m
a
ted
De
te
c
ti
o
n
a
n
d
Clas
sif
ica
ti
o
n
o
f
V
a
sc
u
lar
A
b
n
o
rm
a
li
ti
e
s
in
Dia
b
e
ti
c
R
e
ti
n
o
p
a
th
y
”
,
38
th
Asil
o
ma
r
Co
n
fer
e
n
c
e
o
n
S
i
g
n
a
ls,
S
y
ste
ms
a
n
d
Co
m
p
u
t
e
rs
,
v
o
l.
2
,
p
p
.
1
6
2
5
–
1
6
2
9
,
2
0
0
4
.
[2
]
R.
S
iv
a
k
u
m
a
r,
G
.
Ra
v
in
d
ra
n
,
M
.
M
u
th
a
y
y
a
,
S
.
Lak
sh
m
in
a
ra
y
a
n
a
n
a
n
d
C.
U.
V
e
lm
u
ru
g
h
e
n
d
ra
n
,
“
Dia
b
e
ti
c
Re
ti
n
o
p
a
t
h
y
Clas
si
f
ica
ti
o
n
”
,
IEE
E
In
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
Co
n
v
e
rg
e
n
t
T
e
c
h
n
o
l
o
g
ies
fo
r
th
e
Asia
-
Pa
c
if
i
c
Reg
io
n
,
V
o
l.
1
,
p
p
.
2
0
5
-
2
0
8
,
2
0
0
3
.
[3
]
T
.
W
a
lt
e
r,
J.
C.
Kle
in
,
P
.
M
a
ss
in
a
n
d
A
.
Erg
in
a
y
,
“
A
c
o
n
tri
b
u
ti
o
n
o
f
i
m
a
g
e
p
ro
c
e
ss
in
g
to
th
e
d
iag
n
o
sis
o
f
d
iab
e
ti
c
re
ti
n
o
p
a
th
y
-
d
e
tec
ti
o
n
o
f
e
x
u
d
a
te
s
in
c
o
lo
r
f
u
n
d
u
s
im
a
g
e
s
o
f
th
e
h
u
m
a
n
re
ti
n
a
”
,
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
M
e
d
ica
l
.
Ima
g
i
n
g
”
,
V
o
l
.
2
1
,
N
o
.
1
0
,
p
p
.
1
2
3
6
–
1
2
4
3
,
2
0
0
2
.
[4
]
H.
T
.
Ng
u
y
e
n
,
M
.
B
u
tl
e
r,
A
.
Ro
y
c
h
o
u
d
h
ry
,
A
.
G
.
S
h
a
n
n
o
n
,
J.
F
l
a
c
k
a
n
d
P
.
M
it
c
h
e
ll
,
“
Clas
sif
ica
t
io
n
o
f
d
iab
e
ti
c
re
ti
n
o
p
a
th
y
u
sin
g
n
e
u
ra
l
n
e
tw
o
rk
s”
,
1
8
th
A
n
n
u
a
l
I
n
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
f
th
e
IEE
E
En
g
in
e
e
rin
g
in
M
e
d
icin
e
a
n
d
Bi
o
lo
g
y
S
o
c
iety
,
Vo
l.
4
,
p
p
.
1
5
4
8
–
1
5
4
9
,
1
9
9
6
.
[5
]
M
a
ría
Ga
rc
ía,
Ro
b
e
rto
Ho
rn
e
ro
,
Clara
I.
S
a
n
c
h
e
z
,
M
a
ría
I.
L
o
p
e
z
a
n
d
A
n
a
Díe
z
,
“
F
e
a
tu
re
Ex
trac
ti
o
n
a
n
d
S
e
lec
ti
o
n
f
o
r
th
e
A
u
to
m
a
ti
c
De
te
c
ti
o
n
o
f
Ha
rd
Ex
u
d
a
tes
in
Re
ti
n
a
l
I
m
a
g
e
s
”
,
Pro
c
e
e
d
in
g
s
o
f
th
e
2
9
th
A
n
n
u
a
l
In
te
rn
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
f
th
e
IEE
E
E
n
g
i
n
e
e
rin
g
in
M
e
d
ici
n
e
a
n
d
B
io
l
o
g
y
S
o
c
i
e
ty Ci
te In
ter
n
a
t
io
n
a
le
,
p
p
.
4
9
6
9
–
4
9
7
2
,
2
0
0
7
.
[6
]
P
.
V.
Na
g
e
s
w
a
r
a
Ra
o
,
T
.
U
m
a
De
v
i,
D.
S
.
V
.
G
.
K.
Ka
lad
h
a
r,
G
.
R.
S
rid
h
a
r,
A
ll
a
m
A
p
p
a
Ra
o
,
“
A
P
ro
b
a
b
il
isti
c
Ne
u
ra
l
Ne
t
w
o
rk
A
p
p
ro
a
c
h
f
o
r
p
r
o
tein
S
u
p
e
rf
a
m
il
y
Clas
si
f
ica
ti
o
n
”
,
J
o
u
rn
a
l
o
f
T
h
e
o
re
ti
c
a
l
a
n
d
A
p
p
l
ied
In
fo
rm
a
t
io
n
T
e
c
h
n
o
l
o
g
y
,
Vo
l.
6
,
No
.
1
,
p
p
.
1
0
1
,
2
0
0
9
.
[7
]
S
.
Ch
a
u
d
h
u
ri,
S
.
Ch
a
tt
e
rjee
,
N.
Ka
tz,
M
.
Ne
lso
n
a
n
d
M
.
G
o
ld
b
a
i
m
,
“
De
tec
ti
o
n
o
f
b
lo
o
d
v
e
ss
e
ls
in
re
ti
n
a
l
im
a
g
e
s
u
sin
g
tw
o
-
d
im
e
n
sio
n
a
l
m
a
tch
e
d
f
i
lt
e
rs”
,
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
M
e
d
ica
l
Ima
g
in
g
,
Vo
l.
8
,
N
o
.
3
,
p
p
.
2
6
3
–
2
6
9
,
1
9
8
9
.
[8
]
A
li
re
z
a
Os
a
r
e
h
,
M
a
ji
d
M
irm
e
h
d
i,
Ba
rry
T
h
o
m
a
s
a
n
d
Rich
a
rd
M
a
rk
h
a
m
,
“
Clas
si
f
ica
ti
o
n
a
n
d
L
o
c
a
li
sa
ti
o
n
o
f
Dia
b
e
ti
c
-
Re
late
d
Ey
e
Dise
a
se
”
,
Pro
c
e
e
d
in
g
s
o
f
th
e
7
th
Eu
r
o
p
e
a
n
Co
n
fer
e
n
c
e
o
n
Co
mp
u
ter
Vi
si
o
n
,
p
p
.
5
0
2
–
5
1
6
,
2
0
0
2
.
[9
]
W
o
n
g
L
i
Yu
n
,
U.
Ra
jen
d
ra
Ac
h
a
r
y
a
,
Y.
V
.
V
e
n
k
a
tes
h
,
Ca
ro
l
in
e
Ch
e
e
c
,
L
i
m
Ch
o
o
M
i
n
a
n
d
E.
Y.
K.
Ng
“
Id
e
n
ti
f
ica
ti
o
n
o
f
d
if
f
e
re
n
t
sta
g
e
s o
f
d
iab
e
ti
c
re
ti
n
o
p
a
t
h
y
.
Evaluation Warning : The document was created with Spire.PDF for Python.