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.
15
,
No
.
3
,
J
u
n
e
20
25
,
p
p
.
3028
~
3
0
3
8
I
SS
N:
2088
-
8
7
0
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijece.
v
15
i
3
.
pp
3
0
2
8
-
3
0
3
8
3028
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
Leukem
ia
de
tect
i
o
n using
SegNet
and fas
ter
regio
n
-
b
a
sed
co
nv
o
lutiona
l neural network
Della
Rea
s
a
Va
lia
v
ee
t
il,
K
a
nim
o
zhi T
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
o
n
i
c
s a
n
d
C
o
mm
u
n
i
c
a
t
i
o
n
En
g
i
n
e
e
r
i
n
g
,
V
e
l
T
e
c
h
R
a
n
g
a
r
a
j
a
n
D
r
.
S
a
g
u
n
t
h
a
l
a
R
&D
I
n
st
i
t
u
t
e
o
f
S
c
i
e
n
c
e
a
n
d
Te
c
h
n
o
l
o
g
y
,
C
h
e
n
n
a
i
,
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
Ap
r
1
2
,
2
0
2
4
R
ev
is
ed
J
an
7
,
2
0
2
5
Acc
ep
ted
Ma
r
3
,
2
0
2
5
P
re
v
e
n
ti
o
n
o
f
c
a
n
c
e
r
is
m
o
stly
a
tt
a
in
e
d
b
y
su
r
v
e
il
lan
c
e
o
f
t
h
e
tran
s
fo
rm
a
ti
o
n
z
o
n
e
s.
W
h
it
e
b
l
o
o
d
c
e
ll
s
(
WBC
s
)
a
re
e
sta
b
li
sh
e
d
i
n
t
h
e
b
o
n
e
m
a
rro
w
a
n
d
in
tem
p
e
ra
te
g
ro
wt
h
o
f
WBC
lea
d
s
to
leu
k
e
m
ia.
He
m
a
to
lo
g
ists
e
x
a
m
in
e
th
e
m
icro
sc
o
p
ic
ima
g
e
s
in
m
a
n
u
a
l
m
e
th
o
d
fo
r
p
re
d
icti
n
g
leu
k
e
m
ia,
b
u
t
it
is
v
e
ry
c
o
m
p
lex
p
ro
c
e
ss
a
n
d
wi
th
o
u
t
a
n
y
g
u
a
ra
n
tee
d
f
o
r
a
c
c
u
ra
te.
In
th
is
p
ro
p
o
se
d
stu
d
y
,
d
e
e
p
lea
rn
in
g
tec
h
n
iq
u
e
s
i
n
v
o
lv
e
d
to
se
g
m
e
n
t
a
n
d
c
las
sify
th
e
t
h
re
e
ty
p
e
s
o
f
leu
k
e
m
ia
li
k
e
a
c
u
te
l
y
m
p
h
o
c
y
ti
c
leu
k
e
m
ia
(ALL
),
a
c
u
te
m
y
e
l
o
i
d
leu
k
e
m
ia
(AML
)
a
n
d
c
h
ro
n
ic
ly
m
p
h
o
c
y
ti
c
leu
k
e
m
ia
(CLL
)
u
sin
g
t
h
e
Bio
G
p
s
d
a
tas
e
t.
Th
e
p
u
rp
o
se
o
f
d
e
e
p
lea
rn
i
n
g
in
m
e
d
ica
l
sc
ien
c
e
e
n
h
a
n
c
e
s
th
e
a
c
c
u
ra
c
y
a
n
d
p
re
c
isio
n
o
f
d
e
term
in
in
g
leu
k
e
m
ia
i
n
e
a
rly
sta
g
e
s.
In
th
is
stu
d
y
,
i
n
tro
d
u
c
i
n
g
a
si
g
m
o
i
d
s
tretc
h
in
g
(S
S
)
in
p
i
x
e
l
e
n
h
a
n
c
e
m
e
n
t
fo
r
p
re
p
ro
c
e
ss
in
g
;
S
e
g
Ne
t
(S
t)
is
c
o
m
fo
rt
to
e
x
trac
t
th
e
stru
c
t
u
ra
l
fe
a
tu
re
s
o
f
th
e
leu
k
o
c
y
tes
a
n
d
t
o
se
g
m
e
n
t
th
e
n
o
rm
a
l
a
n
d
b
las
t
c
e
ll
s
fo
r
a
c
lea
r
c
las
sifica
ti
o
n
;
fa
ste
r
re
g
i
o
n
-
b
a
se
d
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
two
r
k
(
fa
ste
r
R
-
CNN
)
c
a
rried
u
n
d
e
r
th
e
p
r
o
c
e
ss
o
f
c
las
sifica
ti
o
n
a
n
d
o
p
ti
m
iza
ti
o
n
d
o
n
e
b
y
d
ra
g
o
n
f
ly
a
lg
o
rit
h
m
.
T
h
e
re
su
lt
o
f
t
h
is
w
o
rk
a
c
h
iev
e
s
b
e
st
a
c
c
u
ra
c
y
re
late
d
to
th
e
e
x
isti
n
g
tec
h
n
iq
u
e
s
o
f
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
two
r
k
(
CNN
)
su
c
h
a
s
su
p
p
o
rt
v
e
c
to
r
m
a
c
h
i
n
e
(S
VM),
k
-
n
e
a
re
st
n
e
i
g
h
b
o
rs
(
k
NN
)
a
n
d
Ba
y
e
sia
n
m
o
d
e
l.
Th
is
stu
d
y
a
c
h
iev
e
s
th
e
a
c
c
u
ra
c
y
ra
te
o
f
9
7
%
,
p
re
c
isio
n
ra
te
o
f
9
4
%
a
n
d
se
n
siti
v
it
y
ra
te o
f
9
0
%
re
sp
e
c
ti
v
e
ly
wi
th
l
o
w co
m
p
lex
it
y
.
K
ey
w
o
r
d
s
:
B
io
Gp
s
Dr
ag
o
n
f
l
y
alg
o
r
ith
m
Fas
ter
r
eg
io
n
-
b
ased
co
n
v
o
l
u
tio
n
al
n
e
u
r
al
n
etwo
r
k
L
eu
k
em
ia
Seg
Net
Sig
m
o
id
s
tr
etch
in
g
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Della
R
ea
s
a
Val
iav
ee
til
Dep
ar
tm
en
t o
f
E
lectr
o
n
ics an
d
C
o
m
m
u
n
icatio
n
E
n
g
in
ee
r
i
n
g
,
Vel
T
ec
h
R
an
g
ar
ajan
Dr
.
Sag
u
n
th
ala
R
&
D
I
n
s
titu
te
o
f
Scien
ce
an
d
T
ec
h
n
o
lo
g
y
Av
ad
i,
C
h
en
n
ai,
I
n
d
ia
E
m
ail:
d
ellar
ea
s
a@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
L
eu
k
em
ia
is
a
class
if
icatio
n
o
f
b
lo
o
d
ca
n
ce
r
in
wh
ic
h
ab
n
o
r
m
al
wh
ite
b
lo
o
d
ce
lls
(
W
B
C
)
f
o
r
m
in
th
e
b
o
n
e
m
ar
r
o
w
[
1
]
.
Acc
o
r
d
in
g
t
o
Glo
b
al
C
an
ce
r
Ob
s
er
v
ato
r
y
(
GL
OB
OC
AN)
,
an
esti
m
ated
9
,
0
0
0
p
eo
p
les
ar
e
af
f
ec
ted
b
y
l
e
u
k
em
ia
in
2
0
2
1
.
Fig
u
r
e
1
p
r
esen
ts
m
icr
o
s
co
p
ic
im
ag
es
o
f
d
if
f
er
e
n
t
ty
p
es
o
f
leu
k
em
ia.
Acu
te
ly
m
p
h
o
b
last
ic
leu
k
em
ia
(
AL
L
)
is
ch
ar
ac
ter
ized
b
y
th
e
p
r
esen
ce
o
f
im
m
atu
r
e
ly
m
p
h
o
b
last
s
in
Fig
u
r
e
1
(
a)
.
Acu
te
m
y
elo
id
le
u
k
em
ia
(
AM
L
)
,
wh
ich
p
r
esen
ts
ir
r
eg
u
lar
ly
s
h
ap
ed
n
u
cleo
li
a
n
d
m
y
el
o
b
last
s
is
s
h
o
wn
in
Fig
u
r
e
1
(
b
)
.
C
h
r
o
n
ic
l
y
m
p
h
o
cy
tic
leu
k
em
ia
(
C
L
L
)
is
ch
ar
ac
ter
ized
b
y
m
atu
r
e,
s
m
all
ly
m
p
h
o
c
y
tes
with
a
d
is
tin
ctiv
e,
“
s
m
u
d
g
e
d
”
ap
p
ea
r
an
ce
in
Fig
u
r
e
1
(
c)
.
An
i
n
cr
ea
s
ed
n
u
m
b
er
o
f
m
y
elo
id
p
r
ec
u
r
s
o
r
ce
lls
in
v
a
r
io
u
s
s
tag
es
o
f
d
if
f
e
r
en
tiatio
n
is
s
h
o
wn
in
Fig
u
r
e
1
(
d
)
.
AL
L
ar
is
es
in
ch
ild
r
en
a
g
e
g
r
o
u
p
b
etwe
e
n
2
t
o
8
y
ea
r
s
o
l
d
,
AM
L
ar
is
es
at
an
y
ag
e,
b
u
t
m
o
s
t
co
m
m
o
n
i
n
ag
e
g
r
o
u
p
b
et
wee
n
2
to
2
0
y
ea
r
s
o
ld
,
C
L
L
tak
es
p
lace
th
e
ag
e
g
r
o
u
p
o
f
ab
o
v
e
7
0
y
ea
r
s
o
ld
an
d
r
a
r
ely
a
f
f
ec
ts
th
e
tee
n
ag
e
r
s
,
ch
r
o
n
ic
m
y
elo
id
le
u
k
em
ia
(
C
ML
)
em
an
ates
at
th
e
ag
e
g
r
o
u
p
o
f
6
0
to
6
5
y
e
ar
s
.
Her
e,
AM
L
is
r
a
p
id
ly
s
p
r
ea
d
in
g
i
n
to
th
e
o
th
e
r
b
lo
o
d
ce
lls
.
On
e
o
f
m
o
s
t
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
Leu
ke
mia
d
etec
tio
n
u
s
in
g
S
eg
N
et
a
n
d
fa
s
ter r
eg
io
n
-
b
a
s
ed
c
o
n
vo
lu
tio
n
a
l
…
(
Della
R
e
a
s
a
V
a
lia
ve
etil
)
3029
co
m
m
o
n
test
s
f
o
r
d
etec
tin
g
le
u
k
em
ia
is
th
e
s
m
ea
r
test
[
2
]
,
[
3
]
.
B
lo
o
d
s
am
p
les
ar
e
o
b
tain
e
d
an
d
s
u
b
jecte
d
to
v
ar
io
u
s
test
s
in
o
r
d
er
to
p
r
ed
i
ct
leu
k
em
ia.
T
h
e
test
s
ar
e:
i)
co
m
p
lete
b
lo
o
d
co
u
n
t
with
d
i
f
f
er
en
tial:
u
s
ed
to
d
eter
m
in
e
th
e
co
u
n
ts
o
f
d
if
f
er
en
t
ty
p
es
o
f
leu
k
o
cy
tes
an
d
f
o
r
ca
teg
o
r
izin
g
th
e
ty
p
e
o
f
ca
n
ce
r
;
ii)
f
lo
w
cy
to
m
etr
y
:
t
h
is
test
d
eter
m
in
e
s
th
e
ty
p
e
o
f
ca
n
ce
r
ce
ll
d
u
e
t
o
th
e
p
r
esen
ce
o
r
a
b
s
en
ce
o
f
p
r
o
tein
m
ar
k
er
s
o
n
th
e
ce
ll
s
u
r
f
ac
e;
an
d
iii)
p
er
ip
h
er
al
s
m
ea
r
ex
am
in
atio
n
:
th
is
ex
am
in
atio
n
lo
o
k
s
f
o
r
v
ar
iati
o
n
s
in
th
e
n
u
m
b
e
r
an
d
s
h
ap
e
o
f
leu
k
o
cy
tes
[
4
]
,
[
5
]
.
E
ar
ly
d
etec
tio
n
o
f
leu
k
em
ia
i
s
a
v
er
y
ch
allen
g
in
g
task
f
o
r
h
em
ato
lo
g
is
ts
in
n
o
wa
d
ay
s
.
T
h
er
ef
o
r
e,
h
em
ato
lo
g
is
ts
ex
am
in
e
b
l
o
o
d
tis
s
u
es
u
n
d
er
an
o
p
tical
m
icr
o
s
co
p
e
o
n
a
c
o
n
s
is
ten
t
lev
el
to
ac
cu
r
ately
class
if
y
an
d
d
etec
t
b
last
ce
lls
.
I
t
is
d
if
f
icu
lt
to
s
ep
ar
ate
a
n
d
id
e
n
tify
th
e
cy
to
p
lasm
an
d
n
u
cleu
s
is
en
g
ag
e
d
with
th
e
s
eg
m
en
tatio
n
a
p
p
r
o
a
ch
[
6
]
–
[
8
]
.
R
ec
en
tly
d
ee
p
lear
n
in
g
(
DL
)
an
d
m
ac
h
in
e
lear
n
in
g
(
ML
)
tech
n
i
q
u
es
p
la
y
a
v
ital
r
o
le
s
u
ch
as
an
aly
ze
th
e
m
ed
ical
im
ag
e,
m
ed
ical
im
ag
e
class
if
icatio
n
an
d
ca
n
ce
r
d
et
ec
tio
n
[
9
]
–
[
1
1
]
.
I
n
g
o
in
g
to
d
ee
p
er
lay
er
,
th
er
e
m
u
s
t
b
e
litt
le
d
if
f
icu
lt
to
tr
ain
ed
th
e
n
etwo
r
k
s
[
1
2
]
.
B
y
en
g
ag
in
g
ad
v
a
n
ce
d
DL
m
o
d
els
s
u
ch
as
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
(
C
NN
)
[
1
3
]
,
R
eg
Net
[
1
4
]
,
L
STM
[
1
5
]
,
YOL
O
[
1
6
]
an
d
s
o
o
n
u
tili
ze
d
to
d
etec
tin
g
leu
k
em
i
a.
I
n
th
is
p
a
p
er
,
p
ix
el
b
r
ig
h
t
n
ess
(
s
ig
m
o
id
s
tr
etch
in
g
)
tech
n
iq
u
e
is
u
s
ed
f
o
r
p
r
ep
r
o
ce
s
s
in
g
,
Seg
-
n
et
in
DL
u
s
ed
f
o
r
s
eg
m
en
tatio
n
an
d
f
ea
tu
r
e
ex
tr
ac
tio
n
.
Fo
r
class
if
icatio
n
,
f
aster
r
eg
io
n
-
b
ased
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
e
two
r
k
(
f
aster
R
-
C
NN)
i
s
ap
p
li
ed
an
d
to
r
ed
u
ce
th
e
r
esp
o
n
d
i
n
g
tim
e
an
d
r
ea
ch
es
h
ig
h
er
ac
c
u
r
ac
y
.
T
h
i
s
p
a
p
e
r
w
a
s
i
n
c
o
r
p
o
r
a
t
e
d
as f
o
l
l
o
w
s
f
i
v
e
s
e
ct
i
o
n
s
: p
a
r
t
2
d
e
p
i
c
t
s
b
r
i
e
f
a
b
o
u
t
t
h
e
e
x
is
ti
n
g
te
c
h
n
i
q
u
e
s
.
P
r
o
p
o
s
e
d
f
e
a
t
u
r
e
e
x
t
r
a
c
ti
o
n
,
s
e
g
m
e
n
t
a
t
i
o
n
,
c
la
s
s
i
f
ic
a
t
i
o
n
a
n
d
o
p
t
i
m
i
z
a
ti
o
n
t
e
c
h
n
i
q
u
es
e
x
p
l
a
i
n
e
d
i
n
t
h
e
p
a
r
t
3
.
P
e
r
f
o
r
m
a
n
c
e
a
n
a
l
y
s
i
s
a
n
d
c
o
m
p
a
r
i
s
o
n
p
a
r
t
d
o
n
e
i
n
p
a
r
t
4
a
n
d
p
a
r
t
5
c
o
n
c
l
u
d
e
s
w
it
h
a
c
o
n
c
l
u
s
i
o
n
.
(
a)
(
b
)
(
c)
(
d
)
Fig
u
r
e
1
.
Mic
r
o
s
co
p
ic
im
ag
es
o
f
d
if
f
e
r
en
t ty
p
es o
f
leu
k
em
ia
(
a)
AL
L
,
(
b
)
AM
L
,
(
c
)
C
L
L
,
(
d
)
C
ML
2.
RE
L
AT
E
D
WO
RK
S
I
n
r
ec
en
t
d
a
y
s
,
s
ev
er
al
DL
tech
n
iq
u
es
wer
e
p
u
t
f
o
r
war
d
b
y
r
esear
ch
er
s
m
ain
ly
t
o
im
p
r
o
v
e
th
e
ac
cu
r
ac
y
o
f
b
lo
o
d
ce
ll
im
a
g
es.
T
h
ese
ad
v
an
ce
m
e
n
ts
p
r
im
ar
ily
f
o
cu
s
o
n
im
p
r
o
v
in
g
th
e
class
if
icatio
n
,
s
eg
m
en
tatio
n
,
an
d
d
etec
tio
n
o
f
d
if
f
er
e
n
t
ty
p
es
o
f
b
lo
o
d
ce
lls
p
ar
ticu
lar
ly
f
o
r
d
iag
n
o
s
in
g
h
em
ato
lo
g
ical
d
is
o
r
d
er
s
s
u
ch
as
leu
k
e
m
ia.
Fu
r
th
er
m
o
r
e
,
ad
v
a
n
ce
d
p
r
ep
r
o
ce
s
s
in
g
tech
n
iq
u
es
ar
e
e
n
h
a
n
cin
g
th
e
r
o
b
u
s
tn
ess
an
d
g
e
n
er
aliza
b
ilit
y
o
f
d
ee
p
le
ar
n
in
g
m
o
d
els in
th
e
a
n
aly
s
is
o
f
b
lo
o
d
ce
ll im
ag
es.
T
h
an
h
et
a
l.
[
1
7
]
h
ad
p
u
t
f
o
r
war
d
to
DL
tech
n
i
q
u
es
u
tili
z
ed
to
d
is
tin
g
u
is
h
n
o
r
m
al
a
n
d
an
o
m
alo
u
s
ca
s
es.
I
n
p
r
ep
r
o
ce
s
s
in
g
,
h
is
to
g
r
am
e
q
u
aliza
tio
n
in
v
o
l
v
es
to
u
p
g
r
ad
e
th
e
c
o
n
tr
ast
o
f
a
p
o
o
r
b
r
ig
h
tn
ess
im
ag
e
an
d
p
r
ed
ictin
g
t
h
e
d
is
tr
ib
u
tio
n
o
f
p
ix
el
d
en
s
ities
,
tr
an
s
latio
n
al
o
p
er
atio
n
s
ar
e
u
s
ed
to
s
h
if
t
an
im
a
g
e
alo
n
g
b
o
th
X
an
d
Y
ax
is
with
co
r
r
esp
o
n
d
in
g
d
is
p
lace
m
en
t
v
alu
es a
n
d
m
id
d
le
o
f
ea
c
h
ax
is
.
I
m
ag
e
r
ef
lectio
n
p
r
o
ce
s
s
also
in
clu
d
ed
.
B
u
t
ac
tiv
atio
n
f
u
n
ctio
n
is
n
o
t
im
p
r
o
v
ed
,
s
o
it
r
ea
ch
es
lo
w
ac
cu
r
ac
y
.
R
ajesh
an
d
Sath
iam
o
o
r
th
y
[
1
8
]
h
ad
d
ev
elo
p
ed
g
en
etic
b
ased
k
-
n
ea
r
est
n
eig
h
b
o
r
(
G
-
k
NN
)
alg
o
r
ith
m
in
v
o
l
v
ed
to
cl
ass
if
y
th
e
leu
k
em
ia.
Her
ein
in
teg
r
ates
th
e
g
en
eti
c
alg
o
r
ith
m
(
GA)
an
d
k
-
n
e
ar
est
n
eig
h
b
o
r
(
k
NN)
alg
o
r
i
th
m
.
Pre
p
r
o
ce
s
s
in
g
f
r
am
ewo
r
k
co
m
p
o
s
ed
o
f
two
p
r
o
ce
s
s
es,
i
m
ag
e
n
o
is
e
ca
n
ce
lin
g
b
y
m
ed
ia
n
f
ilter
a
p
p
r
o
ac
h
an
d
e
n
h
an
ce
s
th
e
im
ag
e
b
y
G
-
k
NN
alg
o
r
ith
m
.
I
t
ch
o
o
s
es
th
e
b
est
k
v
al
u
e
with
m
in
im
u
m
m
is
class
if
icati
o
n
r
ate.
B
u
t,
Sm
all
n
u
m
b
er
o
f
d
atasets
o
n
ly
f
ee
d
.
Sh
af
iq
u
e
an
d
T
e
h
s
in
[
1
9
]
h
a
d
p
r
o
p
o
s
ed
a
Alex
Net
is
ac
tiv
ated
to
i
d
en
tify
AL
L
in
a
n
au
to
m
atic
m
an
n
er
a
n
d
class
if
y
its
s
u
b
ty
p
es.
T
h
e
im
ag
es
wer
e
tak
e
n
b
y
p
u
b
lic
av
ailab
le
d
atasets
.
Fo
u
r
s
ets
o
f
d
atasets
h
ad
b
ee
n
n
o
te
d
as
d
if
f
er
en
t
co
lo
r
s
(
r
ed
-
g
r
ee
n
-
b
l
u
e
(
R
GB
)
,
h
u
e,
s
atu
r
atio
n
,
v
alu
e
(
HSV)
,
lu
m
in
an
ce
,
ch
r
o
m
in
a
n
ce
b
lu
e,
ch
r
o
m
in
an
ce
r
ed
(
Ycb
C
r
)
an
d
h
ig
h
b
it
r
ate
(
HB
R
)
)
.
Fo
r
th
e
to
tal
d
ata
s
ets,
AL
L
d
etec
tio
n
was
g
o
o
d
b
u
t,
th
e
class
if
icatio
n
was
lo
we
r
th
a
n
th
e
R
GB
im
ag
e
d
atasets
.
Ah
m
e
d
et
a
l.
[
2
0
]
h
ad
p
r
o
p
o
s
ed
C
NN
i
s
u
s
ed
to
id
en
tify
th
e
v
a
r
ieties o
f
leu
k
em
ia.
T
h
e
d
atase
ts
ar
e
p
ick
in
g
u
p
b
y
th
e
two
c
o
m
m
o
n
l
y
av
ailab
le
leu
k
em
ia
d
ata
s
o
u
r
ce
s
ar
e
AL
L
im
ag
e
d
atab
ase
an
d
Ash
I
m
ag
e
B
an
k
.
Featu
r
e
e
x
tr
ac
tio
n
s
ca
r
r
y
o
f
f
co
n
v
o
l
u
tio
n
al
an
d
p
o
o
lin
g
lay
er
.
S
to
ch
asti
c
g
r
ad
ien
t
d
escen
t (
SGD)
an
d
ADAM
o
p
tim
izer
s
ar
e
ap
p
lied
.
Ku
m
ar
et
a
l.
[
2
1
]
h
ad
p
r
o
p
o
s
e
d
a
Den
s
e
C
NN
f
r
a
m
ewo
r
k
to
class
if
y
th
e
two
t
y
p
es
o
f
leu
k
em
ia
s
u
ch
as
AL
L
an
d
m
u
ltip
le
m
y
elo
m
a
(
MM
)
.
T
h
e
d
atasets
h
ad
b
ee
n
co
llected
f
r
o
m
SMS
Sp
am
r
esear
ch
.
Her
e,
d
ata
au
g
m
en
tatio
n
i
n
tr
o
d
u
ce
d
two
p
r
o
ce
s
s
es,
f
ir
s
t
o
n
e
is
r
o
tatin
g
th
e
im
ag
es
co
r
r
esp
o
n
d
in
g
to
ce
r
tain
d
eg
r
ee
s
an
d
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.
15
,
No
.
3
,
J
u
n
e
20
25
:
3
0
2
8
-
3
0
3
8
3030
s
ec
o
n
d
o
n
e
is
u
p
g
r
a
d
in
g
o
n
ly
th
e
ed
g
es
o
r
b
o
u
n
d
ar
ies
o
f
th
e
o
r
ig
in
al
im
ag
e.
Un
iv
ar
iate
f
e
atu
r
e
s
elec
tio
n
h
ad
b
ee
n
in
t
r
o
d
u
ce
d
an
d
s
elec
ts
th
e
f
ea
tu
r
es
b
ased
o
n
u
n
iv
ar
ia
te
s
tatis
tical
test
s
.
B
u
t
its
co
m
p
u
tatio
n
al
tim
e
is
m
o
r
e.
L
o
ey
et
a
l.
[
2
2
]
h
ad
d
ev
elo
p
ed
two
p
r
o
p
o
s
ed
m
o
d
e
ls
u
s
in
g
tr
an
s
f
er
lear
n
in
g
f
o
r
d
etec
tin
g
leu
k
em
ia.
T
h
e
d
atasets
ar
e
tak
en
f
r
o
m
k
ag
g
le
an
d
ASH
im
ag
e
b
an
k
.
T
h
e
f
ir
s
t
tech
n
iq
u
e
en
tails
ex
tr
a
ctin
g
f
ea
tu
r
es
f
r
o
m
in
p
u
t
p
h
o
to
s
an
d
attain
in
g
th
e
co
r
r
esp
o
n
d
in
g
p
ar
am
eter
s
o
f
th
e
f
in
al
FC
lay
er
b
e
f
o
r
e
ac
ce
s
s
in
g
th
e
class
if
icatio
n
p
ar
t,
wh
er
ea
s
t
h
e
n
ex
t
s
tep
te
n
d
s
to
n
etwo
r
k
-
f
in
e
tu
n
in
g
p
r
o
ce
d
u
r
e
.
T
h
e
f
ir
s
t
class
if
icatio
n
m
eth
o
d
h
as
th
r
ee
s
ev
er
al
s
tep
s
n
am
ely
:
I
m
ag
e
p
r
ep
r
o
ce
s
s
in
g
b
y
u
s
in
g
R
GB
,
Featu
r
e
ex
tr
ac
tio
n
b
y
u
s
in
g
Alex
Net,
SVM
an
d
lin
ea
r
d
is
cr
im
in
an
ts
(
L
Ds)
m
o
d
e
ls
ar
e
u
tili
ze
d
f
o
r
class
if
ic
atio
n
.
T
h
e
s
ec
o
n
d
class
if
icatio
n
m
o
d
el
h
as
o
n
ly
d
u
al
s
tep
s
:
f
ir
s
t
s
tep
is
s
am
e
as
th
at
o
f
f
i
r
s
t
m
o
d
el
an
d
Ale
x
Net
is
in
d
u
ce
d
f
o
r
ed
g
e
d
etec
tio
n
as we
ll a
s
id
en
tific
atio
n
.
Her
ein
,
s
ec
o
n
d
m
o
d
el
r
ea
ch
es th
e
lo
w
p
er
f
o
r
m
a
n
c
e
m
etr
ics.
Dasar
ir
aju
et
a
l.
[
2
3
]
h
ad
d
ev
is
ed
th
e
r
an
d
o
m
f
o
r
est
(
R
F)
a
lg
o
r
ith
m
u
s
ed
to
d
etec
t
an
d
cl
ass
if
y
th
e
u
n
d
ev
el
o
p
ed
leu
k
o
cy
tes.
Pu
b
licly
ac
ce
s
s
ib
le
d
atasets
ar
e
e
m
p
lo
y
ed
,
an
d
t
h
e
s
eg
m
e
n
tatio
n
m
et
h
o
d
in
clu
d
es
p
ictu
r
e
f
o
r
m
at
co
n
v
er
s
io
n
a
n
d
s
tr
u
ctu
r
al
p
r
o
ce
s
s
es
to
f
r
ag
m
en
t
th
e
ch
ar
ac
te
r
is
tics
o
f
th
e
b
lo
o
d
ce
ll'
s
n
u
cleu
s
an
d
cy
t
o
p
lasm.
I
n
e
v
er
y
im
ag
e,
1
4
f
ea
tu
r
es
an
d
2
w
h
ich
a
r
e
o
f
n
ew
n
u
cleu
s
-
co
lo
r
ed
f
ea
tu
r
es
wer
e
ex
tr
ac
ted
.
Her
e
in
o
n
ly
lim
ited
m
o
r
p
h
o
lo
g
ical
f
ea
tu
r
es we
r
e
ad
d
ed
.
Mo
s
t o
f
th
e
d
ata
wer
e
im
b
alan
ce
d
an
d
d
id
n
o
t d
etec
t
th
e
v
ar
ian
ce
o
f
leu
k
em
ia.
Das
et
a
l.
[
2
4
]
h
ad
d
ev
elo
p
ed
a
G
L
C
M
(
g
r
ay
lev
el
co
-
o
cc
u
r
r
en
c
e
m
atr
ix
)
an
d
g
lr
l
m
(
g
r
ay
lev
el
r
u
n
len
g
t
h
m
atr
ix
)
alg
o
r
ith
m
s
ar
e
em
p
lo
y
ed
to
ex
tr
ac
t
th
e
n
u
cleu
s
ch
ar
ac
ter
is
tics
an
d
d
etec
t
AL
L
.
T
h
e
d
atasets
wer
e
tak
en
b
y
AL
L
I
m
ag
e
Data
b
ase.
T
h
e
SVM
is
to
cla
s
s
if
y
th
e
W
B
C
s
.
H
er
e,
C
L
AHE
ap
p
lied
to
u
p
g
r
ad
e
th
e
s
am
p
le
q
u
ality
.
B
u
t
SVM
tak
es
m
o
r
e
tim
e
t
o
tr
ain
th
e
lar
g
e
d
atasets
.
Sh
ah
ee
n
et
a
l.
[
2
5
]
h
a
d
d
ev
is
ed
th
e
id
en
tific
atio
n
o
f
AM
L
u
s
in
g
Alex
Net
an
d
L
e
n
et
-
5
m
o
d
els
an
d
c
o
m
p
ar
e
d
t
h
e
p
er
f
o
r
m
a
n
ce
o
f
th
ese
b
o
th
m
o
d
els.
T
h
e
d
atasets
wer
e
d
er
iv
ed
f
r
o
m
Ace
v
ed
o
et
a
l.
W
h
en
c
o
m
p
ar
i
n
g
t
h
e
b
o
th
two
m
o
d
els
o
n
th
e
p
er
f
o
r
m
an
ce
an
al
y
s
is
,
th
e
s
ec
o
n
d
m
o
d
el
r
ea
c
h
es
lo
w
ac
cu
r
ac
y
.
B
u
t,
Alex
Net
r
ea
ch
es
h
ig
h
ac
cu
r
ac
y
an
d
d
etec
t
o
n
ly
o
n
e
t
y
p
e
o
f
leu
k
e
m
ia
lik
e
AM
L
.
Fro
m
th
is
s
tu
d
y
v
ar
io
u
s
d
ee
p
lear
n
in
g
m
eth
o
d
s
ar
e
co
m
f
o
r
tab
le
to
g
et
h
ig
h
ac
c
u
r
ac
y
.
I
n
th
is
p
r
o
p
o
s
ed
s
y
s
tem
,
Seg
Net
is
a
ty
p
e
o
f
C
NN
ar
ch
itectu
r
e
u
s
ed
t
o
g
et
h
ig
h
ac
c
u
r
ac
y
an
d
lo
w
m
em
o
r
y
s
p
ac
e
f
o
r
b
o
u
n
d
in
g
b
o
x
es.
3.
P
RO
P
O
SE
D
SYS
T
E
M
I
n
th
is
r
esear
ch
,
d
ee
p
lear
n
i
n
g
tech
n
iq
u
es
in
v
o
lv
ed
to
s
eg
m
en
t
an
d
class
if
y
th
e
th
r
ee
ty
p
es
o
f
leu
k
em
ia
lik
e
AL
L
,
AM
L
an
d
C
L
L
u
s
in
g
th
e
B
io
Gp
s
d
ataset.
T
h
e
p
u
r
p
o
s
e
o
f
d
ee
p
le
ar
n
in
g
in
m
ed
ical
s
cien
ce
en
h
an
ce
s
th
e
ac
cu
r
a
cy
an
d
p
r
ec
is
io
n
o
f
d
eter
m
i
n
in
g
le
u
k
em
ia
in
ea
r
ly
s
tag
es.
I
n
th
is
s
tu
d
y
,
in
tr
o
d
u
cin
g
a
SS
in
p
ix
el
en
h
an
ce
m
en
t f
o
r
p
r
ep
r
o
ce
s
s
in
g
; St is co
m
f
o
r
t to
ex
tr
ac
t th
e
s
tr
u
ctu
r
al
f
ea
tu
r
es o
f
th
e
leu
k
o
cy
tes
an
d
to
s
eg
m
en
t
th
e
n
o
r
m
al
an
d
b
last
ce
lls
f
o
r
a
clea
r
class
if
icatio
n
;
f
aster
R
-
C
NN
ca
r
r
ied
u
n
d
er
th
e
p
r
o
ce
s
s
o
f
class
if
icatio
n
an
d
o
p
tim
izatio
n
d
o
n
e
b
y
d
r
ag
o
n
f
ly
alg
o
r
ith
m
.
Fig
u
r
e
2
s
h
o
ws
th
e
o
v
er
all
p
r
o
ce
s
s
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
.
Fig
u
r
e
2
.
Sch
em
atic
d
iag
r
am
o
f
p
r
o
p
o
s
ed
m
eth
o
d
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
Leu
ke
mia
d
etec
tio
n
u
s
in
g
S
eg
N
et
a
n
d
fa
s
ter r
eg
io
n
-
b
a
s
ed
c
o
n
vo
lu
tio
n
a
l
…
(
Della
R
e
a
s
a
V
a
lia
ve
etil
)
3031
3
.
1
.
Da
t
a
a
cquis
it
io
n
B
lo
o
d
s
am
p
le
im
ag
es
in
v
o
lv
e
d
in
th
is
s
tu
d
y
ar
e
tak
e
n
f
r
o
m
th
e
co
m
m
o
n
ly
av
ailab
le
B
io
G
PS
d
ataset
lib
r
ar
y
[
2
6
]
.
Fo
u
r
ty
p
es
o
f
le
u
k
em
ia
im
a
g
es
with
s
ize
o
f
2
5
6
0
×
1
9
2
0
in
B
MP
f
o
r
m
at
c
o
n
s
titu
te
th
e
d
ataset
[
2
7
]
.
T
h
e
m
e
r
g
ed
f
o
r
m
s
o
f
im
ag
es a
r
e
in
ten
d
e
d
to
tr
ain
t
h
e
p
r
o
p
o
s
ed
m
o
d
el
to
d
ec
id
e
th
e
ty
p
es o
f
leu
k
em
ia.
3
.
2
.
Da
t
a
prepro
ce
s
s
ing
T
h
e
g
o
al
o
f
p
r
e
p
r
o
ce
s
s
in
g
is
t
o
im
p
r
o
v
e
t
h
e
im
ag
e
b
y
in
h
ib
itin
g
u
n
wan
te
d
d
is
to
r
tio
n
s
o
r
en
h
an
cin
g
s
o
m
e
s
p
ec
if
ic
f
ea
tu
r
es
th
at
ar
e
ess
en
tial
f
o
r
f
u
r
th
er
p
r
o
ce
s
s
in
g
an
d
an
aly
s
is
task
s
.
Her
e
in
,
two
tech
n
iq
u
es
s
u
ch
as
s
ig
m
o
id
s
tr
etch
in
g
in
p
ix
el
b
r
ig
h
tn
ess
tr
an
s
f
o
r
m
atio
n
an
d
im
ag
e
clea
n
in
g
a
r
e
in
v
o
lv
ed
.
(
i)
Sig
m
ar
o
id
s
tr
etch
in
g
in
th
e
tr
an
s
f
o
r
m
at
io
n
o
f
p
ix
el
b
r
ig
h
t
n
ess
:
T
h
e
ch
ar
ac
ter
is
tics
o
f
th
e
p
ix
el
its
elf
d
ictate
th
e
tr
an
s
f
o
r
m
atio
n
a
n
d
p
ix
el
b
r
i
g
h
tn
ess
.
T
h
e
co
n
g
r
u
en
t
v
alu
e
o
f
th
e
in
p
u
t
p
ix
el
is
th
e
s
in
g
le
f
ac
to
r
th
at
d
eter
m
in
es
th
e
o
u
t
p
u
t
p
ix
el'
s
v
alu
e
in
p
ix
el
b
r
ig
h
t
n
ess
tr
an
s
f
o
r
m
atio
n
.
T
h
e
s
ig
m
o
id
f
u
n
c
tio
n
is
a
n
o
n
-
lin
ea
r
ac
tiv
atio
n
f
u
n
cti
o
n
th
at
is
co
n
t
in
u
o
u
s
.
(
,
)
=
1
1
+
(
∗
(
−
(
,
)
)
)
(
1
)
T
h
e
ab
o
v
e
(
1
)
im
p
lies
(
,
)
-
o
r
i
g
in
al
im
ag
e;
(
,
)
-
im
p
r
o
v
e
d
p
ix
el
v
alu
e;
c
-
co
n
t
r
ast
f
ac
to
r
;
t
-
th
r
esh
o
ld
v
alu
e
.
B
y
alter
in
g
t
h
e
co
n
tr
ast
f
ac
to
r
‘
n
’
a
n
d
th
r
e
s
h
o
ld
v
alu
e
‘
t’
it
is
p
o
s
s
ib
le
to
tailo
r
r
eg
u
late
th
e
o
v
er
all
im
ag
e
q
u
ality
.
(
ii)
I
m
a
g
e
clea
n
in
g
:
i
t
is
also
r
eq
u
ir
ed
to
b
e
p
e
r
f
o
r
m
ed
.
T
h
e
ter
m
s
o
lid
ity
(
S)
r
eq
u
i
r
ed
to
b
e
m
ea
s
u
r
ed
f
o
r
im
ag
e
cle
an
in
g
.
E
ac
h
c
o
m
p
o
n
en
t
with
a
s
o
lid
ity
v
alu
e
less
er
th
an
th
e
th
r
esh
o
ld
v
alu
e
is
elim
in
ated
.
T
h
e
f
o
r
m
u
la
f
o
r
s
o
lid
ity
(
S)
is
(
2
)
=
(
2
)
3
.
3
.
E
x
t
r
a
ct
ing
f
ea
t
ures
T
h
e
aim
o
f
th
is
p
h
ase
was
to
cr
ea
te
a
s
et
o
f
d
escr
ip
tio
n
s
th
at
co
u
ld
b
e
u
s
ed
t
o
cl
ass
if
y
th
e
leu
k
o
cy
tes.
I
t
h
elp
s
to
m
in
im
ize
th
e
q
u
an
tity
o
f
r
ed
u
n
d
an
t
d
ata
f
r
o
m
th
e
d
atasets
.
Her
ein
,
S
eg
N
et
is
in
v
o
lv
ed
ex
tr
ac
tin
g
th
e
s
tr
u
ctu
r
al
f
ea
t
u
r
es
o
f
th
e
cy
to
p
lasm
an
d
n
u
clei.
T
h
e
s
tr
u
ctu
r
al
f
ea
tu
r
es
o
f
n
u
cleu
s
lik
e
co
n
v
ex
ity
,
cir
cu
lar
ity
an
d
co
n
v
ex
ity
ar
e
id
en
tifie
d
in
wh
ich
h
elp
s
to
d
etec
t
th
e
leu
k
em
ia.
T
h
e
f
o
llo
win
g
(
3
)
-
(
5
)
ar
e
r
elate
d
to
s
tr
u
ctu
r
al
f
ea
tu
r
es:
(
)
=
ℎ
(
3
)
(
)
=
ℎ
(
4
)
(
)
=
(
)
2
4
(
)
(
5
)
3
.
4
.
Seg
m
ent
a
t
i
o
n
Dete
ctin
g
an
d
class
if
y
in
g
th
e
o
b
jects
is
th
e
m
o
s
t
v
ital
r
o
le
in
a
co
m
p
u
ter
v
is
io
n
.
T
h
e
g
o
al
o
f
th
is
p
r
o
ce
s
s
is
to
s
eg
r
eg
ate
th
e
o
b
j
ec
t
in
th
e
im
ag
e.
I
n
th
is
p
r
o
ce
s
s
,
Seg
N
et
i
s
in
tr
o
d
u
ce
d
to
f
r
a
g
m
en
t
th
e
b
last
an
d
n
o
r
m
al
ce
ll
o
f
th
e
W
B
C
.
T
h
e
m
ain
co
n
t
r
ib
u
tio
n
in
Seg
N
et
was
to
av
o
id
tr
an
s
p
o
s
ed
co
n
v
o
lu
tio
n
,
b
ec
au
s
e
it
lead
s
to
u
n
ev
en
o
v
er
la
p
.
Fig
u
r
e
3
s
h
o
ws
th
e
ar
ch
itectu
r
e
d
ia
g
r
am
o
f
Seg
N
et
.
T
h
e
en
c
o
d
er
n
etwo
r
k
r
esem
b
les
to
th
e
th
ir
teen
co
n
v
o
lu
tio
n
al
lay
er
s
in
th
e
v
er
y
d
ee
p
co
n
v
o
lu
ti
o
n
al
(
VGG
1
6
)
n
etwo
r
k
with
o
u
t
f
u
lly
co
n
n
ec
ted
lay
er
s
in
ten
d
e
d
f
o
r
im
ag
e
class
if
icat
io
n
.
Fig
u
r
e
4
d
em
o
n
s
tr
ates
th
e
u
p
-
s
am
p
lin
g
o
f
Seg
N
et
.
T
h
e
d
ec
o
d
er
u
s
es
p
o
o
lin
g
in
d
ices
o
f
co
r
r
esp
o
n
d
i
n
g
m
ax
im
u
m
p
o
o
lin
g
s
tep
s
to
p
e
r
f
o
r
m
u
p
-
s
am
p
lin
g
.
Fig
u
r
e
3
s
h
o
ws th
e
ar
ch
itectu
r
e
o
f
Seg
N
et.
Fo
r
s
ettin
g
th
e
win
d
o
w
2
×
2
,
s
t
r
id
e
2
wh
ich
is
d
o
n
e,
t
h
en
th
e
o
u
tco
m
e
o
f
th
e
p
o
o
lin
g
lay
er
i
s
s
am
p
led
b
y
th
e
f
ac
to
r
2
.
B
ef
o
r
e
s
u
b
-
s
am
p
lin
g
,
t
h
e
b
o
u
n
d
ar
y
ch
ar
ac
ter
is
tics
m
u
s
t
b
e
r
etain
ed
in
t
h
e
en
co
d
er
f
ea
tu
r
e
m
ap
s
.
Usi
n
g
m
ax
im
al
p
o
o
lin
g
,
ea
ch
d
ec
o
d
er
in
th
e
n
etwo
r
k
u
p
s
am
p
les
th
e
m
ap
o
f
f
ea
tu
r
es.
T
h
e
m
ap
s
o
f
s
p
ar
s
e
f
ea
tu
r
es
ar
e
p
r
o
d
u
ce
d
a
t
th
is
s
tep
.
T
h
en
,
th
e
n
o
r
m
al
c
ells
an
d
b
last
ce
ll
s
ar
e
d
iv
id
ed
in
to
two
d
if
f
er
en
t
im
ag
es a
n
d
with
co
n
g
r
u
en
t c
o
l
o
r
s
.
3
.
5
.
Cla
s
s
if
ica
t
io
n
I
n
th
is
s
ec
tio
n
,
th
e
im
ag
es
ar
e
r
ec
o
g
n
ized
u
tili
zin
g
f
aster
R
-
C
NN.
Fas
ter
R
-
C
NN
is
ch
o
o
s
in
g
to
ac
h
iev
e
h
i
g
h
ac
c
u
r
ac
y
an
d
r
el
iab
ilit
y
f
o
r
d
eter
m
in
i
n
g
le
u
k
e
m
ia.
Firstl
y
,
th
e
e
x
tr
ac
ted
f
ea
tu
r
es
ar
e
tr
ain
e
d
to
th
e
f
aster
R
-
C
NN.
T
h
e
in
p
u
t
im
ag
e
is
f
ed
in
to
p
r
e
tr
ain
ed
o
r
in
itialized
C
N
N
to
g
en
er
ate
a
f
ea
tu
r
e.
Her
e,
VGG1
6
is
in
v
o
lv
e
d
to
C
NN
b
lo
ck
.
T
h
e
r
e
g
io
n
p
r
o
p
o
s
al
n
etwo
r
k
g
iv
es
r
is
e
to
p
r
o
p
o
s
al
f
o
r
th
is
r
eg
io
n
,
in
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.
15
,
No
.
3
,
J
u
n
e
20
25
:
3
0
2
8
-
3
0
3
8
3032
wh
ich
th
e
o
b
ject
lies
;
a
m
i
n
i
n
etwo
r
k
is
s
lid
o
v
er
t
h
r
o
u
g
h
a
co
n
v
o
lu
tio
n
al
f
ea
tu
r
e
t
h
at
is
th
e
o
u
tco
m
e
o
f
t
h
e
f
in
al
lay
er
an
d
it tak
es less
co
m
p
u
tatio
n
al
tim
e.
Fin
ally
,
th
e
FC
lay
er
s
wer
e
ac
cu
s
to
m
ed
to
class
if
ies
th
e
tar
g
et
an
d
ad
ju
s
t th
e
b
o
u
n
d
in
g
b
o
x
o
f
th
e
tar
g
et
th
at
m
ea
n
s
to
class
if
y
th
e
ty
p
es
o
f
leu
k
e
m
ia.
=
(
6
)
T
h
e
(
6
)
d
e
p
icts
th
e
in
ter
s
ec
tio
n
o
f
u
n
i
o
n
.
Her
e
b
y
,
-
ar
ea
o
f
in
ter
s
ec
tio
n
b
etwe
en
an
ch
o
r
an
d
g
r
o
u
n
d
tr
u
th
b
o
x
;
-
ar
ea
o
f
u
n
io
n
o
f
th
e
a
n
ch
o
r
.
I
t
d
eliv
er
s
t
h
e
class
if
icatio
n
o
n
th
e
b
last
ce
ll
s
ev
er
ity
an
d
ch
r
o
n
ic
le
v
el
f
o
r
class
if
y
in
g
th
e
le
u
k
em
ia
ca
s
es.
T
h
e
f
aster
R
-
C
NN
ar
ch
ite
ctu
r
e
is
s
h
o
wn
in
Fig
u
r
e
5
.
Fig
u
r
e
3
.
Ar
c
h
itectu
r
e
o
f
Seg
Net
Fig
u
r
e
4
.
Up
-
s
am
p
lin
g
o
f
Seg
Net
Fig
u
r
e
5
.
Fas
ter
R
C
NN
ar
ch
itectu
r
e
3
.
6
.
O
pti
m
iza
t
io
n
E
s
s
en
tially
,
o
p
tim
izatio
n
is
a
u
n
iq
u
e
m
eth
o
d
o
f
p
r
o
b
le
m
s
o
lv
in
g
wh
er
e
ce
r
tain
o
b
j
ec
tiv
es
ar
e
f
u
lf
illed
b
y
ad
ju
s
tin
g
th
e
n
e
u
r
al
n
etwo
r
k
'
s
in
ter
n
al
p
ar
am
eter
weig
h
ts
.
wh
er
e
th
e
o
p
tim
izatio
n
is
d
o
n
e
u
s
in
g
th
e
d
r
a
g
o
n
f
ly
tech
n
iq
u
e.
T
h
e
weig
h
t
ch
an
g
in
g
in
t
h
e
o
p
tim
izatio
n
p
r
o
ce
s
s
is
th
e
p
at
h
to
p
ar
ity
o
f
ex
p
lo
r
atio
n
an
d
ex
p
lo
itatio
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
Leu
ke
mia
d
etec
tio
n
u
s
in
g
S
eg
N
et
a
n
d
fa
s
ter r
eg
io
n
-
b
a
s
ed
c
o
n
vo
lu
tio
n
a
l
…
(
Della
R
e
a
s
a
V
a
lia
ve
etil
)
3033
=
−
∑
=
1
−
(
7
)
T
h
e
(
7
)
im
p
lies
th
e
ca
lcu
latio
n
f
o
r
s
ep
ar
ati
o
n
.
Hith
er
,
-
d
e
p
icts
th
e
cu
r
r
en
t
lo
ca
tio
n
o
f
t
h
e
in
d
iv
id
u
al;
-
d
ep
icts
th
e
lo
ca
tio
n
f
o
r
th
e
ℎ
n
eig
h
b
o
r
i
n
g
elem
en
t;
M
-
d
e
p
icts
th
e
to
tal
co
u
n
t
o
f
in
d
iv
id
u
a
l
in
th
e
elem
en
t
th
r
o
n
g
an
d
-
d
ep
icts
th
e
s
ep
ar
a
tio
n
m
o
tio
n
f
o
r
th
e
ℎ
in
d
iv
id
u
al
.
=
∑
=
1
(
8
)
T
h
e
ab
o
v
e
(
8
)
d
en
o
tes
th
e
alig
n
m
en
t
ca
lcu
latio
n
.
Hith
er
,
is
th
e
alig
n
m
en
t m
o
tio
n
f
o
r
t
h
e
ℎ
in
d
iv
id
u
al
a
n
d
is
th
e
v
elo
city
f
o
r
th
e
ℎ
n
eig
h
b
o
r
in
g
elem
e
n
t.
=
∑
=
1
−
(
9
)
T
h
e
(
9
)
im
p
lies
th
e
ad
h
esio
n
c
alcu
latio
n
an
d
is
th
e
ad
h
esio
n
f
o
r
ℎ
in
d
iv
id
u
al,
is
th
e
n
eig
h
b
o
r
h
o
o
d
s
ize,
is
th
e
lo
ca
tio
n
o
f
ℎ
n
eig
h
b
o
r
in
g
elem
en
t
an
d
is
th
e
cu
r
r
en
t
elem
en
t
in
d
iv
id
u
al.
T
h
e
two
m
o
r
e
f
ea
tu
r
es
ar
e
ad
d
e
d
to
im
p
r
o
v
e
th
e
p
er
f
o
r
m
an
ce
o
f
d
r
ag
o
n
f
ly
alg
o
r
ith
m
is
m
em
o
r
y
-
b
ased
h
y
b
r
id
d
r
ag
o
n
f
ly
alg
o
r
ith
m
.
Fin
ally
,
it h
elp
s
to
ac
h
iev
e
t
h
e
co
r
r
ec
t im
ag
e
f
o
r
m
at,
h
ig
h
o
p
tim
al
v
alu
e
an
d
c
o
n
v
e
r
g
en
ce
s
p
ee
d
.
3
.
7
.
Det
ec
t
io
n
T
h
e
c
l
a
s
s
i
f
i
e
d
i
m
a
g
e
s
a
r
e
i
d
e
n
t
i
f
i
e
d
f
o
r
t
h
e
t
h
r
e
e
t
y
p
e
s
o
f
l
e
u
k
e
m
i
a
.
T
h
e
d
i
f
f
e
r
e
n
c
e
w
a
s
p
r
e
d
i
c
t
e
d
a
s
a
s
h
a
p
e
,
s
p
o
n
g
i
n
e
s
s
o
f
t
h
e
t
i
s
s
u
e
s
a
n
d
m
u
l
t
i
p
l
e
o
f
t
h
e
b
l
a
s
t
c
e
l
l
s
.
I
t
i
s
u
s
e
d
t
o
d
i
s
p
l
a
y
t
h
e
f
i
n
a
l
r
e
s
u
l
t
o
f
t
h
e
t
e
s
t
d
a
t
a
.
4.
RE
SU
L
T
AND
DI
SCUS
SI
O
NS
T
h
e
ex
p
er
im
en
tal
s
etu
p
o
f
th
i
s
p
ap
er
was
im
p
lem
en
ted
b
y
MA
T
L
AB
2
0
1
9
a.
I
n
th
is
r
esu
lt
an
aly
s
is
,
th
e
b
lo
o
d
s
m
ea
r
im
ag
es
ar
e
t
ak
en
f
r
o
m
B
io
Gp
s
d
ataset
to
d
etec
t
at
th
e
leu
k
em
ia
at
t
h
e
ea
r
ly
s
tag
es.
T
h
e
p
r
o
p
o
s
ed
d
ee
p
lea
r
n
in
g
-
b
ase
d
m
eth
o
d
was
ev
alu
ated
u
s
i
n
g
v
ar
io
u
s
p
er
f
o
r
m
a
n
ce
m
et
r
ics
to
en
s
u
r
e
its
ef
f
ec
tiv
en
ess
in
id
e
n
tify
in
g
ab
n
o
r
m
al
b
lo
o
d
ce
lls
.
B
ased
o
n
co
llected
d
ata,
th
e
p
r
o
p
o
s
ed
m
eth
o
d
was
ass
ess
ed
b
ased
o
n
s
p
ec
if
icity
,
p
r
ec
is
io
n
,
F1
s
co
r
e,
r
ec
all,
a
n
d
ac
cu
r
ac
y
.
4
.
1
.
P
re
pro
ce
s
s
ing
I
t
is
ap
p
lied
to
en
h
an
ce
th
e
i
m
ag
e
q
u
ality
.
I
n
th
is
s
tu
d
y
,
s
ig
m
o
id
s
tr
etch
in
g
is
u
s
ed
to
e
n
h
an
ce
th
e
co
n
tr
ast.
Her
e,
is
th
e
av
er
ag
e
in
ten
s
ity
v
alu
e
an
d
is
th
e
av
er
ag
e
in
ten
s
ity
o
f
im
a
g
e.
=
−
(
1
0
)
So
m
e
b
lo
o
d
s
m
ea
r
s
am
p
les
u
s
u
ally
ac
q
u
ir
ed
with
th
e
in
t
en
t
o
f
en
h
an
cin
g
co
n
tr
ast
an
d
class
if
icatio
n
ar
e
ac
h
iev
ed
u
s
in
g
th
e
m
eth
o
d
s
ar
e
m
en
tio
n
ed
ab
o
v
e.
T
h
e
co
n
tr
ast
m
ea
s
u
r
es
o
f
f
iv
e
im
ag
es
ar
e
d
ep
icted
as
a
T
ab
le
1
.
T
h
e
ab
o
v
e
tab
le
an
d
g
r
ap
h
s
ar
e
clea
r
ly
s
h
o
ws
th
at
th
e
m
ea
s
u
r
e
o
f
co
n
tr
ast
i
s
h
ig
h
f
o
r
s
ig
m
o
id
s
tr
etch
in
g
th
an
h
is
to
g
r
am
eq
u
aliza
tio
n
m
eth
o
d
a
n
d
f
u
zz
y
l
o
g
ic
-
b
ased
m
eth
o
d
.
T
ab
le
1
.
Me
asu
r
e
o
f
co
n
tr
ast
Per
f
o
r
m
a
n
ce
m
ea
s
u
r
e
a
n
aly
s
is
f
o
r
s
eg
n
et:
En
h
a
n
c
i
n
g
t
e
c
h
n
i
q
u
e
I
mg
1
I
mg
2
I
mg
3
I
mg
4
I
mg
5
H
i
st
o
g
r
a
m e
q
u
a
l
i
z
a
t
i
o
n
0
.
5
8
7
0
.
6
3
6
0
.
4
5
4
0
.
3
8
1
0
.
7
6
5
F
u
z
z
y
l
o
g
i
c
0
.
7
4
5
0
.
8
3
7
0
.
7
6
8
0
.
5
9
8
0
.
8
2
1
S
i
g
m
o
i
d
s
t
r
e
t
c
h
i
n
g
,
c
=
2
1
.
0
3
8
0
.
9
4
5
1
.
1
3
4
0
.
8
5
6
1
.
2
6
4
S
i
g
m
o
i
d
s
t
r
e
t
c
h
i
n
g
,
c
=
5
1
.
1
4
3
1
.
8
9
6
1
.
6
5
4
0
.
9
7
8
1
.
5
2
9
T
h
is
tech
n
iq
u
e
is
a
cr
u
cial
co
m
p
o
n
e
n
t
o
f
th
e
t
o
tal
f
ac
e
r
ec
o
g
n
itio
n
ac
cu
r
ac
y
r
ate
f
o
r
im
p
r
o
v
in
g
f
ac
e
d
etec
tio
n
ac
cu
r
ac
y
wh
ile
r
ed
u
cin
g
th
e
n
u
m
b
e
r
o
f
f
alse
n
eg
ativ
es
an
d
p
o
s
itiv
es.
T
h
e
p
ar
am
eter
s
u
s
ed
f
o
r
d
etec
tio
n
ar
e
AC
C
,
PR
E
,
R
e
F
1
s
co
r
e,
cw,
an
d
wh
ich
a
r
e
d
ef
i
n
ed
as:
(
tr
u
e
p
o
s
itiv
e)
;
(
tr
u
e
n
eg
ativ
e)
;
(
f
alse
p
o
s
itiv
e)
an
d
(
f
alse
n
e
g
ativ
e)
.
T
h
e
(
1
2
)
i
m
p
lies
th
e
f
o
r
m
u
la
f
o
r
(
p
r
ec
i
s
io
n
)
.
C
las
s
weig
h
ts
ar
e
u
s
ed
to
b
ala
n
ce
th
e
class
es.
T
h
e
class
we
ig
h
t
(
cw)
f
o
r
ea
c
h
class
is
f
o
r
m
u
lated
as
(
1
5
)
.
W
h
er
ea
s
,
-
f
r
eq
u
en
cy
o
f
ea
c
h
class
; N
-
th
e
co
u
n
t
o
f
p
i
x
el
in
th
e
class
; T
N
is
th
e
co
u
n
t o
f
p
ix
el
in
th
e
im
ag
e.
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.
15
,
No
.
3
,
J
u
n
e
20
25
:
3
0
2
8
-
3
0
3
8
3034
=
+
+
+
+
(
1
1
)
=
+
(
1
2
)
=
+
(
1
3
)
1
=
2
(
.
)
(
+
)
(
1
4
)
=
1
(
1
5
)
=
(
1
6
)
As
s
h
o
wn
in
T
ab
le
2
,
th
e
b
la
s
t
ce
ll
h
as
th
e
h
ig
h
est
weig
h
t
(
8
.
2
0
3
4
)
b
ec
a
u
s
e
o
f
its
cr
itic
al
r
o
le
in
d
etec
tin
g
leu
k
em
ia,
f
o
llo
wed
b
y
th
e
n
o
r
m
al
ce
ll
(
1
)
an
d
th
e
b
ac
k
g
r
o
u
n
d
(
0
.
5
6
4
3
)
.
Seg
Net
in
co
r
p
o
r
ates
class
weig
h
ts
in
to
th
e
f
in
al
lay
er
o
f
th
e
a
r
ch
itectu
r
e
to
b
alan
ce
th
e
lear
n
in
g
p
r
o
ce
s
s
,
en
s
u
r
i
n
g
t
h
e
m
o
d
el
d
o
es
n
o
t
f
av
o
r
m
o
r
e
f
r
e
q
u
en
t
class
es
an
d
im
p
r
o
v
i
n
g
s
en
s
itiv
ity
to
war
d
s
less
r
ep
r
esen
ted
b
last
ce
lls
.
Acc
o
r
d
in
g
to
T
ab
le
3
,
Seg
Net
h
as
an
o
v
er
al
l
ac
cu
r
ac
y
o
f
0
.
9
6
,
a
n
I
n
ter
s
ec
tio
n
o
v
er
Un
io
n
(
I
o
U
)
o
f
0
.
7
9
,
a
weig
h
ted
I
o
U
o
f
0
.
8
9
,
an
d
a
Me
an
F1
-
s
co
r
e
o
f
0
.
7
8
.
All
o
f
th
ese
m
etr
ics
d
em
o
n
s
tr
ate
Seg
Net's
ef
f
icien
cy
in
s
eg
m
en
tin
g
b
lo
o
d
s
m
ea
r
im
ag
es.
T
ab
le
4
p
r
o
v
id
es
a
b
r
ea
k
d
o
w
n
o
f
class
-
wis
e
p
er
f
o
r
m
a
n
ce
m
etr
ics,
d
em
o
n
s
tr
atin
g
h
ig
h
a
cc
u
r
ac
y
f
o
r
b
last
ce
lls
(
0
9
7
)
,
n
o
r
m
al
ce
lls
(
0
9
5
)
,
an
d
b
ac
k
g
r
o
u
n
d
(
0
9
2
)
.
Ad
d
itio
n
all
y
,
th
e
Me
an
F1
-
s
co
r
es
f
o
r
all
th
r
ee
class
es
s
u
p
p
o
r
t
th
e
m
o
d
el'
s
a
b
ilit
y
to
s
u
cc
ess
f
u
lly
d
is
tin
g
u
is
h
b
etwe
en
d
if
f
er
en
t
ce
ll
ty
p
es.
Her
e
b
last
ce
l
l
r
ea
ch
es th
e
h
ig
h
ac
cu
r
ac
y
th
an
th
e
o
th
er
class
es a
n
d
m
a
x
im
u
m
F1
s
co
r
e
ac
h
iev
es th
e
b
ac
k
g
r
o
u
n
d
.
T
ab
le
2
.
C
lass
weig
h
t o
f
th
r
ee
class
es
N
a
me
C
l
a
s
s we
i
g
h
t
B
l
a
s
t
C
e
l
l
8
.
2
0
3
4
N
o
r
mal
C
e
l
l
1
B
a
c
k
g
r
o
u
n
d
0
.
5
6
4
3
T
ab
le
3
.
Valu
atio
n
m
etr
ics o
f
Seg
Net
A
c
c
u
r
a
c
y
M
e
a
n
I
o
U
W
e
i
g
h
t
e
d
I
o
U
M
e
a
n
F
1
sc
o
r
e
0
.
9
6
0
.
7
9
0
.
8
9
0
.
7
8
T
ab
le
4
.
C
lass
m
etr
ics o
f
th
r
ee
class
es
N
a
me
A
c
c
u
r
a
c
y
I
o
U
M
e
a
n
F
1
s
c
o
r
e
B
l
a
s
t
C
e
l
l
0
.
9
7
0
.
8
8
0
.
7
9
N
o
r
mal
C
e
l
l
0
.
9
5
0
.
8
7
0
.
8
2
B
a
c
k
g
r
o
u
n
d
0
.
9
2
0
.
9
0
0
.
8
4
4
.
2
.
Co
m
pa
ra
t
iv
e
a
na
ly
s
is
I
n
th
is
p
r
o
p
o
s
ed
s
tu
d
y
,
th
e
te
s
tin
g
m
eth
o
d
o
lo
g
y
is
u
s
ed
to
d
etec
t
th
e
leu
k
em
ia
in
ea
r
ly
s
tag
es.
T
h
e
Seg
Net
is
u
s
ed
to
av
o
id
tr
a
n
s
p
o
s
ed
co
n
v
o
lu
tio
n
.
I
t
is
n
o
t
ed
,
th
at
th
e
Seg
Net
r
ea
ch
es
h
ig
h
r
ea
c
h
es
h
ig
h
ac
cu
r
ac
y
th
an
th
e
o
th
e
r
p
r
ev
i
o
u
s
tech
n
iq
u
es
s
u
ch
as
d
u
al
th
r
esh
o
ld
m
eth
o
d
,
en
h
a
n
ce
d
SVM
an
d
B
ay
esian
m
o
d
el.
T
ab
le
5
r
ep
r
esen
ts
th
e
ac
cu
r
ac
y
an
d
p
r
ec
is
io
n
r
esu
lt
s
o
f
th
ese
d
if
f
er
en
t
m
eth
o
d
s
b
y
h
ig
h
lig
h
tin
g
th
e
ad
v
an
tag
es
o
f
Seg
Net
in
ac
h
i
ev
in
g
m
o
r
e
p
r
ec
is
e
an
d
ac
cu
r
ate
b
lo
o
d
ce
ll
class
if
icatio
n
.
T
h
e
ac
cu
r
ac
y
o
f
d
u
al
th
r
esh
o
ld
m
eth
o
d
,
e
n
h
an
ce
d
S
VM
,
an
d
B
ay
esian
m
o
d
el
a
r
e
5
6
%,
6
7
% a
n
d
8
8
%.
T
ab
le
5
.
Acc
u
r
ac
y
an
d
p
r
ec
is
i
o
n
r
esu
lt f
o
r
d
if
f
er
en
t m
eth
o
d
s
M
e
t
h
o
d
s
A
c
c
u
r
a
c
y
P
r
e
c
i
s
i
o
n
D
u
a
l
t
h
r
e
s
h
o
l
d
me
t
h
o
d
56
45
En
h
a
n
c
e
d
S
V
M
67
78
B
a
y
e
s
i
a
n
m
o
d
e
l
88
89
S
e
g
N
e
t
97
94
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
Leu
ke
mia
d
etec
tio
n
u
s
in
g
S
eg
N
et
a
n
d
fa
s
ter r
eg
io
n
-
b
a
s
ed
c
o
n
vo
lu
tio
n
a
l
…
(
Della
R
e
a
s
a
V
a
lia
ve
etil
)
3035
Fro
m
th
e
ab
o
v
e
c
o
m
p
ar
is
o
n
Seg
Net
r
ea
ch
es
th
e
m
ax
im
u
m
ac
cu
r
ac
y
9
7
%
an
d
p
r
e
cisi
o
n
9
4
%
co
m
p
ar
ed
to
o
t
h
er
m
eth
o
d
s
lik
e
d
u
al
th
r
esh
o
ld
m
eth
o
d
,
e
n
h
an
ce
d
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
an
d
B
ay
esian
m
o
d
el.
T
h
e
s
eg
m
en
te
d
im
ag
es
ar
e
s
u
b
jecte
d
to
f
aster
r
eg
io
n
-
b
ased
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
e
two
r
k
(
R
-
C
NN)
to
class
if
y
th
e
leu
k
em
ia
b
ased
o
n
s
ev
er
ity
lev
el
a
n
d
th
e
T
a
b
le
6
r
ep
r
esen
ts
c
o
m
p
ar
is
o
n
o
f
f
aster
R
-
C
NN
class
if
ier
with
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e,
k
n
ea
r
est
n
eig
h
b
o
r
.
T
h
e
f
aster
R
-
C
NN
clas
s
if
ier
f
ir
s
tly
well
tr
ain
ed
th
e
tr
ain
ed
d
ata
an
d
t
h
en
m
o
v
e
to
esti
m
ate
th
e
test
d
ata.
T
ab
le
6
.
co
m
p
ar
is
o
n
o
f
f
aster
R
-
C
NN
wi
th
o
th
er
class
if
ier
s
C
l
a
s
si
f
i
e
r
s
A
c
c
u
r
a
c
y
S
p
e
c
i
f
i
c
i
t
y
S
e
n
s
i
t
i
v
i
t
y
S
V
M
86
89
85
k
NN
92
90
91
F
a
st
e
r
R
-
C
N
N
95
92
90
Acc
o
r
d
in
g
to
T
a
b
le
7
,
th
e
p
r
o
p
o
s
ed
m
o
d
el
e
n
h
an
ce
s
t
h
e
a
cc
u
r
ac
y
o
f
2
%,
0
.
2
3
%
an
d
0
.
5
%
b
etter
th
an
SVM,
R
F,
an
d
Alex
Net
.
As
a
r
esu
lt,
th
e
p
r
ed
icte
d
r
esu
lts
o
f
th
e
p
r
o
p
o
s
ed
m
o
d
el
f
o
r
leu
k
em
ia
r
ec
o
g
n
izin
g
in
i
m
ag
es
ar
e
q
u
i
te
r
eliab
le.
C
o
m
p
ar
e
d
to
tr
ad
i
tio
n
al
m
ac
h
in
e
lear
n
in
g
tech
n
iq
u
es
lik
e
R
F
an
d
SVM,
th
e
p
r
o
p
o
s
ed
d
ee
p
lear
n
in
g
ap
p
r
o
ac
h
p
r
o
v
id
es
m
o
r
e
p
r
ec
is
e
an
d
ac
cu
r
ate
class
if
icatio
n
s
,
r
ed
u
cin
g
t
h
e
lik
elih
o
o
d
o
f
m
is
d
iag
n
o
s
is
.
Ad
d
itio
n
ally
,
wh
en
co
m
p
ar
e
d
to
Alex
Net,
wh
ich
is
a
wid
ely
u
s
ed
C
NN
ar
ch
itectu
r
e,
f
aster
R
-
C
NN
o
f
f
er
s
en
h
an
ce
d
o
b
ject
lo
ca
lizati
o
n
an
d
s
eg
m
e
n
tatio
n
,
m
ak
in
g
it
m
o
r
e
s
u
itab
le
f
o
r
m
ed
ical
im
ag
in
g
task
s
.
T
ab
le
7
.
C
o
m
p
a
r
is
o
n
o
f
p
r
o
p
o
s
ed
m
o
d
el
with
th
e
e
x
is
tin
g
m
o
d
els
A
u
t
h
o
r
s
Te
c
h
n
i
q
u
e
s
A
c
c
u
r
a
c
y
D
a
sari
r
a
j
u
S
.
e
t
a
l
.
[
2
3
]
RF
9
6
.
7
7
%
D
a
s,
e
t
a
l
.
[
2
4
]
S
V
M
9
5
%
S
h
a
h
e
e
n
M
.
e
t
a
l
.
[
2
5
]
A
l
e
x
N
e
t
9
6
.
5
%
P
r
o
p
o
se
d
m
e
t
h
o
d
F
a
st
e
r
R
-
C
N
N
9
7
%
Acc
o
r
d
in
g
to
T
a
b
le
8
,
to
attain
h
ig
h
lev
els
o
f
p
r
ec
is
io
n
,
cla
s
s
ic
n
etwo
r
k
s
lik
e
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
(
C
NN
)
,
d
ee
p
n
eu
r
al
n
etwo
r
k
(
DNN
)
,
an
d
d
ee
p
b
elief
n
etwo
r
k
(
DB
N
)
u
s
e
a
h
u
g
e
n
u
m
b
er
o
f
p
ar
am
eter
s
,
wh
ich
in
c
r
ea
s
es th
e
d
if
f
icu
lty
.
W
ith
f
ewe
r
p
a
r
am
eter
s
u
s
ed
in
T
ab
le.
8
,
t
h
e
s
u
g
g
ested
m
o
d
el
k
ee
p
s
its
g
o
o
d
p
er
f
o
r
m
an
ce
wh
ile
r
ed
u
cin
g
co
m
p
lex
ity
.
T
o
d
e
m
o
n
s
tr
ate
th
e
s
y
s
tem
'
s
s
p
ee
d
,
th
e
s
u
g
g
ested
m
o
d
el
em
p
lo
y
s
a
r
estricte
d
n
u
m
b
e
r
o
f
GFLO
Ps
.
Ad
d
itio
n
ally
,
th
e
s
u
g
g
ested
m
o
d
el'
s
leu
k
em
ia
d
iag
n
o
s
is
p
r
o
ce
d
u
r
e
in
v
o
lv
in
g
a
d
er
m
o
s
co
p
ic
with
7
.
1
GFLO
PS
tak
es
1
1
1
m
illi
s
ec
o
n
d
s
(
m
s
)
.
C
o
m
p
ar
ed
to
o
th
er
ea
r
lier
D
L
ar
ch
itectu
r
es
s
u
ch
as
C
NN,
DNN,
an
d
DB
N,
th
e
co
m
p
lex
ity
is
r
ad
ically
r
ed
u
ce
d
.
T
h
e
p
r
o
p
o
s
ed
a
p
p
r
o
ac
h
o
u
tp
er
f
o
r
m
s
th
e
cu
r
r
e
n
t
DL
m
eth
o
d
s
b
ased
o
n
th
is
co
m
p
ar
is
o
n
.
T
h
e
co
m
p
u
tatio
n
a
l
co
m
p
lex
ity
o
f
t
h
e
p
r
o
p
o
s
ed
o
p
tim
izatio
n
alg
o
r
ith
m
u
s
in
g
th
e
Dr
ag
o
n
f
ly
Alg
o
r
ith
m
an
d
f
aster
R
-
C
NN
i
s
d
i
s
cu
s
s
ed
.
T
h
e
m
o
d
el
m
ain
tain
s
h
ig
h
p
e
r
f
o
r
m
an
ce
wh
ile
r
ed
u
cin
g
co
m
p
u
tatio
n
al
co
m
p
lex
ity
c
o
m
p
a
r
ed
to
co
n
v
en
tio
n
al
d
ee
p
lear
n
in
g
a
r
ch
itectu
r
es
lik
e
C
NN,
DNN,
an
d
DB
N.
Sp
ec
i
f
ically
:
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
o
p
er
ates
with
7
.
1
GFLO
PS
,
in
d
icatin
g
f
ewe
r
f
l
o
atin
g
-
p
o
in
t
o
p
er
atio
n
s
co
m
p
ar
ed
to
C
NN
(
1
5
.
1
)
,
DNN
(
9
.
3
)
,
an
d
DB
N
(
9
.
5
)
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
ac
h
ie
v
es
a
p
r
o
ce
s
s
in
g
tim
e
o
f
5
4
6
m
s
o
n
th
e
C
PU.
T
h
e
m
o
d
el
p
r
o
ce
s
s
es
im
ag
es
in
1
1
1
m
s
o
n
th
e
GPU.
T
h
is
r
ed
u
ce
d
co
m
p
u
tatio
n
al
lo
ad
d
em
o
n
s
tr
ates
th
e
ef
f
icien
cy
o
f
t
h
e
p
r
o
p
o
s
ed
o
p
tim
izatio
n
s
tr
ateg
y
in
ac
h
iev
in
g
ac
cu
r
ate
leu
k
em
ia
d
etec
tio
n
w
h
ile
m
in
i
m
izin
g
r
eso
u
r
ce
co
n
s
u
m
p
tio
n
.
T
ab
le
8
.
C
o
m
p
u
tatio
n
al
co
m
p
l
ex
ity
co
m
p
a
r
is
o
n
o
f
ex
is
tin
g
m
eth
o
d
s
with
p
r
o
p
o
s
ed
m
o
d
el
M
o
d
e
l
G
F
LO
P
S
C
P
U
(
ms)
G
P
U
(
ms)
C
N
N
1
5
.
1
8
8
8
2
1
5
DNN
9
.
3
6
5
2
1
3
9
D
B
N
9
.
5
7
4
4
1
4
3
R
-
C
N
N
(
o
u
r
s)
7
.
1
5
4
6
1
1
1
4
.
3
.
Clini
ca
l set
t
ing
I
n
th
is
s
ec
tio
n
,
a
clin
ical
im
p
licatio
n
o
f
th
e
p
r
o
p
o
s
ed
f
aster
R
-
C
NN
was
illu
s
tr
ated
f
o
r
ef
f
icien
tly
d
etec
tin
g
th
e
leu
k
e
m
ia
d
is
ea
s
e
u
s
in
g
b
lo
o
d
s
m
ea
r
im
ag
es
lik
ely
in
a
r
ea
l
-
wo
r
ld
clin
ical
s
ettin
g
.
I
t
b
eg
i
n
s
with
th
e
p
atien
t
v
is
itin
g
a
h
o
s
p
ital,
wh
er
e
b
lo
o
d
s
m
ea
r
im
ag
es
ar
e
co
llected
as
in
p
u
t
f
o
r
d
iag
n
o
s
is
.
T
h
ese
b
lo
o
d
s
m
ea
r
im
ag
es
ar
e
th
en
p
r
o
ce
s
s
ed
b
y
th
e
f
aster
R
-
C
NN
m
o
d
el,
wh
ich
an
aly
s
es
th
em
to
p
r
ed
ict
th
e
lo
ca
tio
n
o
f
leu
k
em
ia
.
T
h
e
p
r
e
d
ictio
n
r
esu
lts
ar
e
th
en
s
en
t
b
ac
k
to
th
e
d
o
cto
r
in
th
e
h
o
s
p
ital
f
o
r
ass
is
tin
g
in
r
ef
in
in
g
th
e
p
atien
t'
s
d
iag
n
o
s
is
an
d
ad
v
is
es
o
n
an
ap
p
r
o
p
r
iate
tr
ea
tm
e
n
t
p
lan
f
o
r
th
e
p
atien
t.
T
h
is
s
tr
ea
m
lin
ed
p
r
o
ce
s
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.
15
,
No
.
3
,
J
u
n
e
20
25
:
3
0
2
8
-
3
0
3
8
3036
s
u
p
p
o
r
ts
tim
ely
an
d
ac
cu
r
ate
d
iag
n
o
s
is
to
ass
is
t
in
ef
f
ec
tiv
e
p
atien
t
ca
r
e
in
r
ea
l
-
tim
e.
T
h
is
ap
p
r
o
ac
h
en
h
an
ce
s
ac
cu
r
ac
y
b
y
tar
g
etin
g
s
p
ec
if
ic
leu
k
em
ia
ty
p
es (
b
last
ce
ll,
n
o
r
m
al
ce
ll)
an
d
r
e
d
u
cin
g
d
iag
n
o
s
tic
co
m
p
lex
ity
.
5.
CO
NCLU
SI
O
N
T
h
e
p
r
im
ar
y
in
ten
t
o
f
t
h
is
s
tu
d
y
is
to
s
eg
m
en
t
th
e
n
o
r
m
al
a
n
d
b
last
ce
lls
o
r
u
n
d
ev
el
o
p
ed
ce
lls
in
th
e
p
er
ip
h
er
al
b
lo
o
d
s
m
ea
r
i
m
ag
es
in
to
s
ep
ar
ate
im
ag
es.
Af
ter
,
ex
tr
ac
tin
g
th
e
f
ea
tu
r
es
an
d
f
ee
d
i
n
to
S
eg
N
et.
Seg
n
et
(
St)
r
e
q
u
ir
es
less
m
em
o
r
y
s
p
ac
e
co
m
p
ar
in
g
to
th
e
o
th
er
b
o
u
n
d
in
g
b
o
x
s
eg
m
en
t
atio
n
.
T
h
e
p
r
o
v
id
ed
s
ig
m
o
id
s
tr
etch
in
g
(
SS
)
b
o
o
s
ts
th
e
im
ag
e
co
n
tr
ast
lev
el
wh
ile
also
u
p
g
r
ad
i
n
g
t
h
e
im
a
g
e
q
u
ality
.
Fas
ter
R
-
C
NN
ca
n
b
e
tr
ain
ed
in
all
p
o
s
s
ib
le
way
s
to
d
etec
t
a
n
d
class
i
f
y
th
e
leu
k
em
ia
with
lo
w
e
r
r
o
r
r
ate.
Acc
o
r
d
i
n
g
to
th
e
f
in
al
r
esu
lts
,
it
h
as
b
ee
n
co
n
clu
d
in
g
th
is
p
r
o
p
o
s
ed
f
r
a
m
ewo
r
k
im
p
r
o
v
es
th
e
im
a
g
e
q
u
ality
co
m
p
a
r
e
to
o
th
er
m
eth
o
d
s
.
T
h
e
ac
cu
r
ate
d
etec
tio
n
o
f
th
e
leu
k
em
ia
is
d
o
n
e
u
s
in
g
th
e
co
m
p
ar
ativ
e
a
n
aly
s
is
o
f
ac
cu
r
ac
y
,
s
p
ec
if
icity
an
d
s
en
s
itiv
ity
.
T
h
is
p
r
o
p
o
s
ed
wo
r
k
ac
h
iev
es
th
e
ac
cu
r
ac
y
o
f
9
5
%,
s
p
ec
i
f
icity
o
f
9
2
%
an
d
s
en
s
itiv
ity
o
f
9
0
%
r
esp
ec
tiv
el
y
.
Dete
ctin
g
leu
k
e
m
ia
u
s
in
g
d
ee
p
lear
n
in
g
h
as
p
r
ac
tical
r
ea
l
-
wo
r
ld
ap
p
licatio
n
s
in
im
p
r
o
v
in
g
d
iag
n
o
s
tic
ac
cu
r
ac
y
,
an
d
ac
ce
s
s
ib
ilit
y
.
Dee
p
lear
n
in
g
alg
o
r
ith
m
s
ar
e
an
aly
zin
g
th
e
b
lo
o
d
s
m
ea
r
s
im
ag
es
to
id
en
tify
leu
k
e
m
ic
c
ells
with
h
ig
h
p
r
ec
is
io
n
,
o
f
ten
s
u
r
p
ass
in
g
tr
ad
itio
n
al
m
eth
o
d
s
.
Ad
d
itio
n
ally
,
it
ca
n
b
e
d
ep
lo
y
e
d
in
r
eso
u
r
ce
-
li
m
ited
s
ettin
g
s
th
r
o
u
g
h
au
t
o
m
a
ted
d
iag
n
o
s
tic
to
o
ls
,
ex
p
a
n
d
in
g
ac
ce
s
s
to
q
u
ality
h
ea
lth
ca
r
e
an
d
p
o
ten
tially
im
p
r
o
v
in
g
p
atien
t
o
u
tco
m
es.
I
n
f
u
tu
r
e,
o
u
r
m
o
d
el
im
p
r
o
v
es
th
e
Mu
lti
-
Mo
d
al
Data
Fu
s
io
n
,
an
d
will b
e
e
x
ten
d
ed
with
th
e
ad
v
an
ce
d
r
ec
o
g
n
izin
g
n
etwo
r
k
to
im
p
r
o
v
e
th
e
ac
c
u
r
ac
y
.
ACK
NO
WL
E
DG
M
E
N
T
S
T
h
e
au
t
h
o
r
wo
u
ld
lik
e
to
ex
p
r
ess
h
is
h
ea
r
tf
elt
g
r
atitu
d
e
t
o
th
e
s
u
p
er
v
is
o
r
f
o
r
h
is
g
u
i
d
an
ce
a
n
d
u
n
wav
er
in
g
s
u
p
p
o
r
t d
u
r
in
g
t
h
is
r
esear
ch
f
o
r
h
is
g
u
id
an
ce
an
d
s
u
p
p
o
r
t.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
Au
th
o
r
s
s
tate
n
o
f
u
n
d
in
g
in
v
o
lv
ed
.
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
T
h
is
jo
u
r
n
al
u
s
es
th
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT)
to
r
ec
o
g
n
ize
in
d
iv
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
th
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
llab
o
r
atio
n
.
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
Della
R
ea
s
a
Val
iav
ee
til
✓
✓
✓
✓
✓
✓
K
a
n
i
m
o
z
h
i
T
✓
✓
✓
✓
✓
✓
✓
C
:
C
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
M
:
M
e
t
h
o
d
o
l
o
g
y
So
:
So
f
t
w
a
r
e
Va
:
Va
l
i
d
a
t
i
o
n
Fo
:
Fo
r
mal
a
n
a
l
y
s
i
s
I
:
I
n
v
e
s
t
i
g
a
t
i
o
n
R
:
R
e
so
u
r
c
e
s
D
:
D
a
t
a
C
u
r
a
t
i
o
n
O
:
W
r
i
t
i
n
g
-
O
r
i
g
i
n
a
l
D
r
a
f
t
E
:
W
r
i
t
i
n
g
-
R
e
v
i
e
w
&
E
d
i
t
i
n
g
Vi
:
Vi
su
a
l
i
z
a
t
i
o
n
Su
:
Su
p
e
r
v
i
s
i
o
n
P
:
P
r
o
j
e
c
t
a
d
mi
n
i
st
r
a
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
si
t
i
o
n
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
T
h
e
au
t
h
o
r
s
d
ec
lar
e
th
at
th
e
y
h
av
e
n
o
k
n
o
wn
c
o
m
p
etin
g
f
in
an
cial
in
ter
ests
o
r
p
er
s
o
n
al
r
el
atio
n
s
h
ip
s
th
at
co
u
ld
h
av
e
ap
p
ea
r
ed
t
o
in
f
lu
en
ce
th
e
wo
r
k
r
e
p
o
r
te
d
in
t
h
is
p
ap
er
.
I
NF
O
RM
E
D
CO
NS
E
N
T
T
h
e
au
th
o
r
s
ce
r
tif
y
th
at
th
ey
h
av
e
ex
p
lain
ed
t
h
e
n
at
u
r
e
a
n
d
p
u
r
p
o
s
e
o
f
t
h
is
s
tu
d
y
t
o
th
e
ab
o
v
e
-
n
am
ed
in
d
iv
id
u
al,
an
d
th
ey
h
av
e
d
i
s
cu
s
s
ed
th
e
p
o
ten
tial
b
en
ef
its
o
f
th
is
s
tu
d
y
p
ar
ticip
atio
n
.
T
h
e
q
u
esti
o
n
s
th
e
in
d
iv
id
u
al
h
ad
a
b
o
u
t
th
is
s
tu
d
y
h
av
e
b
ee
n
an
s
wer
ed
,
an
d
we
will
a
lway
s
b
e
av
ailab
l
e
to
ad
d
r
ess
f
u
tu
r
e
q
u
esti
o
n
s
.
E
T
H
I
CAL AP
P
RO
V
AL
Ou
r
r
esear
ch
g
u
id
e
r
e
v
iewe
d
a
n
d
eth
ically
ap
p
r
o
v
ed
th
is
m
a
n
u
s
cr
ip
t f
o
r
p
u
b
lis
h
in
g
in
th
is
j
o
u
r
n
al.
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
Leu
ke
mia
d
etec
tio
n
u
s
in
g
S
eg
N
et
a
n
d
fa
s
ter r
eg
io
n
-
b
a
s
ed
c
o
n
vo
lu
tio
n
a
l
…
(
Della
R
e
a
s
a
V
a
lia
ve
etil
)
3037
DATA AV
AI
L
AB
I
L
I
T
Y
Data
s
h
ar
in
g
n
o
t
ap
p
licab
le
to
th
is
ar
ticle
a
s
n
o
d
atasets
wer
e
g
en
er
ated
o
r
an
aly
ze
d
d
u
r
in
g
th
e
cu
r
r
en
t stu
d
y
.
RE
F
E
R
E
NC
E
S
[
1
]
I
.
K
a
r
u
n
a
r
a
t
h
n
a
,
S
.
B
a
n
d
a
r
a
,
A
.
Ja
y
a
w
a
r
d
a
n
a
,
K
.
D
e
A
l
v
i
s,
P
.
G
u
n
a
se
n
a
,
a
n
d
S
.
G
u
n
a
t
h
i
l
a
k
e
,
“
Le
u
k
e
m
i
a
:
C
l
a
s
si
f
i
c
a
t
i
o
n
,
r
i
sk
f
a
c
t
o
r
s,
a
n
d
d
i
a
g
n
o
st
i
c
c
h
a
l
l
e
n
g
e
s
,
”
R
e
se
a
c
h
G
a
t
e
.
p
p
.
1
–
1
1
,
2
0
2
4
.
d
o
i
:
1
0
.
1
3
1
4
0
/
R
G
.
2
.
2
.
3
6
6
5
7
.
2
9
2
8
8
.
[
2
]
P
.
K
.
M
a
l
l
i
c
k
,
S
.
K
.
M
o
h
a
p
a
t
r
a
,
G
.
-
S
.
C
h
a
e
,
a
n
d
M
.
N
.
M
o
h
a
n
t
y
,
“
C
o
n
v
e
r
g
e
n
t
l
e
a
r
n
i
n
g
–
b
a
s
e
d
m
o
d
e
l
f
o
r
l
e
u
k
e
mi
a
c
l
a
ss
i
f
i
c
a
t
i
o
n
f
r
o
m
g
e
n
e
e
x
p
r
e
ssi
o
n
,
”
P
e
rs
o
n
a
l
a
n
d
U
b
i
q
u
i
t
o
u
s
C
o
m
p
u
t
i
n
g
,
v
o
l
.
2
7
,
n
o
.
3
,
p
p
.
1
1
0
3
–
1
1
1
0
,
J
u
n
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s
0
0
7
7
9
-
0
2
0
-
0
1
4
6
7
-
3.
[
3
]
R
.
N
.
A
g
r
a
w
a
l
a
n
d
V
.
K
a
p
a
d
i
a
,
“
S
y
s
t
e
ma
t
i
c
r
e
v
i
e
w
o
f
d
i
a
g
n
o
s
i
s
a
n
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
t
e
c
h
n
i
q
u
e
s
o
f
a
c
u
t
e
l
y
mp
h
o
b
l
a
st
i
c
l
e
u
k
a
e
m
i
a
a
n
d
a
c
u
t
e
m
y
e
l
o
i
d
l
e
u
k
a
e
mi
a
c
e
l
l
s,
”
I
n
d
i
a
n
J
o
u
r
n
a
l
o
f
N
a
t
u
r
a
l
S
c
i
e
n
c
e
s
,
v
o
l
.
1
4
,
n
o
.
8
2
,
p
p
.
6
8
9
0
5
–
6
8
9
1
7
,
2
0
2
4
.
[
4
]
M
.
A
.
R
e
j
u
l
a
,
B
.
M
.
J
e
b
I
n
,
R
.
S
e
l
v
a
k
u
mar,
S
.
A
mu
t
h
a
,
a
n
d
E.
G
e
o
r
g
e
,
“
D
e
t
e
c
t
i
o
n
o
f
a
c
u
t
e
l
y
m
p
h
o
b
l
a
s
t
i
c
l
e
u
k
e
mi
a
u
si
n
g
a
n
o
v
e
l
b
o
n
e
m
a
r
r
o
w
i
ma
g
e
s
e
g
m
e
n
t
a
t
i
o
n
,
”
T
s
i
n
g
h
u
a
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
3
0
,
n
o
.
2
,
p
p
.
6
1
0
–
6
2
3
,
A
p
r
.
2
0
2
5
,
d
o
i
:
1
0
.
2
6
5
9
9
/
TST.
2
0
2
3
.
9
0
1
0
0
9
9
.
[
5
]
A
.
S
.
N
e
g
m,
O
.
A
.
H
a
ss
a
n
,
a
n
d
A
.
H
.
K
a
n
d
i
l
,
“
A
d
e
c
i
si
o
n
s
u
p
p
o
r
t
s
y
st
e
m
f
o
r
A
c
u
t
e
Le
u
k
a
e
m
i
a
c
l
a
s
si
f
i
c
a
t
i
o
n
b
a
se
d
o
n
d
i
g
i
t
a
l
mi
c
r
o
sc
o
p
i
c
i
m
a
g
e
s,”
Al
e
x
a
n
d
r
i
a
E
n
g
i
n
e
e
ri
n
g
J
o
u
r
n
a
l
,
v
o
l
.
5
7
,
n
o
.
4
,
p
p
.
2
3
1
9
–
2
3
3
2
,
D
e
c
.
2
0
1
8
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
a
e
j
.
2
0
1
7
.
0
8
.
0
2
5
.
[
6
]
L.
P
u
t
z
u
,
G
.
C
a
o
c
c
i
,
a
n
d
C
.
D
i
R
u
b
e
r
t
o
,
“
Le
u
c
o
c
y
t
e
c
l
a
ssi
f
i
c
a
t
i
o
n
f
o
r
l
e
u
k
a
e
mi
a
d
e
t
e
c
t
i
o
n
u
s
i
n
g
i
m
a
g
e
p
r
o
c
e
s
si
n
g
t
e
c
h
n
i
q
u
e
s
,
”
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
i
n
M
e
d
i
c
i
n
e
,
v
o
l
.
6
2
,
n
o
.
3
,
p
p
.
1
7
9
–
1
9
1
,
N
o
v
.
2
0
1
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
a
r
t
me
d
.
2
0
1
4
.
0
9
.
0
0
2
.
[
7
]
S
.
K
h
a
n
,
M
.
S
a
j
j
a
d
,
N
.
A
b
b
a
s
,
J.
E
sco
r
c
i
a
-
G
u
t
i
e
r
r
e
z
,
M
.
G
a
marr
a
,
a
n
d
K
.
M
u
h
a
mm
a
d
,
“
Ef
f
i
c
i
e
n
t
l
e
u
k
o
c
y
t
e
s
d
e
t
e
c
t
i
o
n
a
n
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
i
n
mi
c
r
o
s
c
o
p
i
c
b
l
o
o
d
i
m
a
g
e
s u
s
i
n
g
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
c
o
u
p
l
e
d
w
i
t
h
a
d
u
a
l
a
t
t
e
n
t
i
o
n
n
e
t
w
o
r
k
,
”
C
o
m
p
u
t
e
rs
i
n
Bi
o
l
o
g
y
a
n
d
M
e
d
i
c
i
n
e
,
v
o
l
.
1
7
4
,
M
a
y
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
m
p
b
i
o
me
d
.
2
0
2
4
.
1
0
8
1
4
6
.
[
8
]
M
.
C
l
a
r
o
e
t
a
l
.
,
“
C
o
n
v
o
l
u
t
i
o
n
n
e
u
r
a
l
n
e
t
w
o
r
k
m
o
d
e
l
s
f
o
r
a
c
u
t
e
l
e
u
k
e
m
i
a
d
i
a
g
n
o
s
i
s,”
i
n
2
0
2
0
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
S
y
st
e
m
s
,
S
i
g
n
a
l
s
a
n
d
I
m
a
g
e
Pr
o
c
e
ssi
n
g
(
I
W
S
S
I
P)
,
Ju
l
.
2
0
2
0
,
p
p
.
6
3
–
6
8
.
d
o
i
:
1
0
.
1
1
0
9
/
I
W
S
S
I
P
4
8
2
8
9
.
2
0
2
0
.
9
1
4
5
4
0
6
.
[
9
]
B
.
E
l
sa
y
e
d
e
t
a
l
.
,
“
D
e
e
p
l
e
a
r
n
i
n
g
e
n
h
a
n
c
e
s a
c
u
t
e
l
y
mp
h
o
b
l
a
s
t
i
c
l
e
u
k
e
mi
a
d
i
a
g
n
o
s
i
s
a
n
d
c
l
a
ssi
f
i
c
a
t
i
o
n
u
si
n
g
b
o
n
e
marr
o
w
i
ma
g
e
s,
”
Fro
n
t
i
e
rs
i
n
O
n
c
o
l
o
g
y
,
v
o
l
.
1
3
,
D
e
c
.
2
0
2
3
,
d
o
i
:
1
0
.
3
3
8
9
/
f
o
n
c
.
2
0
2
3
.
1
3
3
0
9
7
7
.
[
1
0
]
R
.
S
u
n
d
a
r
a
s
e
k
a
r
a
n
d
A
.
A
p
p
a
t
h
u
r
a
i
,
“
A
u
t
o
m
a
t
i
c
b
r
a
i
n
t
u
m
o
r
d
e
t
e
c
t
i
o
n
a
n
d
c
l
a
ssi
f
i
c
a
t
i
o
n
b
a
se
d
o
n
I
o
T
a
n
d
mac
h
i
n
e
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s,
”
F
l
u
c
t
u
a
t
i
o
n
a
n
d
N
o
i
s
e
L
e
t
t
e
rs
,
v
o
l
.
2
1
,
n
o
.
0
3
,
J
u
n
.
2
0
2
2
,
d
o
i
:
1
0
.
1
1
4
2
/
S
0
2
1
9
4
7
7
5
2
2
5
0
0
3
0
4
.
[
1
1
]
R
.
S
u
n
d
a
r
a
se
k
a
r
a
n
d
A
.
A
p
p
a
t
h
u
r
a
i
,
“
Ef
f
i
c
i
e
n
t
b
r
a
i
n
t
u
mo
r
d
e
t
e
c
t
i
o
n
a
n
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
u
s
i
n
g
ma
g
n
e
t
i
c
r
e
so
n
a
n
c
e
i
ma
g
i
n
g
,
”
Bi
o
m
e
d
i
c
a
l
P
h
y
s
i
c
s
& E
n
g
i
n
e
e
r
i
n
g
E
x
p
r
e
ss
,
v
o
l
.
7
,
n
o
.
5
,
S
e
p
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
8
8
/
2
0
5
7
-
1
9
7
6
/
a
c
0
c
c
c
.
[
1
2
]
D
.
C
a
mp
a
n
a
a
n
d
C
.
H
.
P
u
i
,
“
D
e
t
e
c
t
i
o
n
o
f
mi
n
i
m
a
l
r
e
si
d
u
a
l
d
i
s
e
a
se
i
n
a
c
u
t
e
l
e
u
k
e
m
i
a
:
M
e
t
h
o
d
o
l
o
g
i
c
a
d
v
a
n
c
e
s
a
n
d
c
l
i
n
i
c
a
l
si
g
n
i
f
i
c
a
n
c
e
[
se
e
c
o
m
me
n
t
s]
,
”
Bl
o
o
d
,
v
o
l
.
8
5
,
n
o
.
6
,
p
p
.
1
4
1
6
–
1
4
3
4
,
M
a
r
.
1
9
9
5
,
d
o
i
:
1
0
.
1
1
8
2
/
b
l
o
o
d
.
V
8
5
.
6
.
1
4
1
6
.
b
l
o
o
d
j
o
u
r
n
a
l
8
5
6
1
4
1
6
.
[
1
3
]
B
.
S
i
v
a
sa
n
k
a
r
i
,
M
.
S
h
u
n
m
u
g
a
t
h
a
mm
a
l
,
A
.
A
p
p
a
t
h
u
r
a
i
,
a
n
d
M
.
K
a
v
i
t
h
a
,
“
H
i
g
h
-
t
h
r
o
u
g
h
p
u
t
a
n
d
p
o
w
e
r
-
e
f
f
i
c
i
e
n
t
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
u
s
i
n
g
o
n
e
-
p
a
ss p
r
o
c
e
s
si
n
g
e
l
e
me
n
t
s,
”
J
o
u
r
n
a
l
o
f
C
i
r
c
u
i
t
s
,
S
y
st
e
m
s a
n
d
C
o
m
p
u
t
e
rs
,
v
o
l
.
3
1
,
n
o
.
1
3
,
S
e
p
.
2
0
2
2
,
d
o
i
:
1
0
.
1
1
4
2
/
S
0
2
1
8
1
2
6
6
2
2
5
0
2
2
6
7
.
[
1
4
]
M
.
A
n
l
i
n
S
a
h
a
y
a
I
n
f
a
n
t
T
i
n
u
,
A
.
A
p
p
a
t
h
u
r
a
i
,
a
n
d
N
.
M
u
t
h
u
k
u
mara
n
,
“
D
e
t
e
c
t
i
o
n
o
f
b
r
a
i
n
t
u
mo
u
r
v
i
a
r
e
v
e
r
si
n
g
h
e
x
a
g
o
n
a
l
f
e
a
t
u
r
e
p
a
t
t
e
r
n
f
o
r
c
l
a
ssi
f
y
i
n
g
d
o
u
b
l
e
-
mo
d
a
l
b
r
a
i
n
i
m
a
g
e
s,”
I
ETE
J
o
u
rn
a
l
o
f
Re
s
e
a
rc
h
,
v
o
l
.
7
0
,
n
o
.
8
,
p
p
.
7
0
3
3
–
7
0
4
3
,
A
u
g
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
8
0
/
0
3
7
7
2
0
6
3
.
2
0
2
3
.
2
3
0
1
6
6
3
.
[
1
5
]
D
.
J.
S
a
mu
e
l
R
.
e
t
a
l
.
,
“
R
e
a
l
t
i
me
v
i
o
l
e
n
c
e
d
e
t
e
c
t
i
o
n
f
r
a
m
e
w
o
r
k
f
o
r
f
o
o
t
b
a
l
l
s
t
a
d
i
u
m
c
o
m
p
r
i
si
n
g
o
f
b
i
g
d
a
t
a
a
n
a
l
y
s
i
s
a
n
d
d
e
e
p
l
e
a
r
n
i
n
g
t
h
r
o
u
g
h
b
i
d
i
r
e
c
t
i
o
n
a
l
LSTM
,
”
C
o
m
p
u
t
e
r
N
e
t
w
o
r
k
s
,
v
o
l
.
1
5
1
,
p
p
.
1
9
1
–
2
0
0
,
M
a
r
.
2
0
1
9
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
mn
e
t
.
2
0
1
9
.
0
1
.
0
2
8
.
[
1
6
]
K
.
G
a
y
a
t
h
r
i
,
K
.
P
.
A
.
G
l
a
d
i
s
,
a
n
d
A
.
A
.
M
a
r
y
,
“
R
e
a
l
t
i
me
ma
sk
e
d
f
a
c
e
r
e
c
o
g
n
i
t
i
o
n
u
si
n
g
d
e
e
p
l
e
a
r
n
i
n
g
b
a
s
e
d
y
o
l
o
v
4
n
e
t
w
o
r
k
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
D
a
t
a
S
c
i
e
n
c
e
a
n
d
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
(
I
J
D
S
AI
)
,
v
o
l
.
1
,
n
o
.
1
,
p
p
.
2
6
–
3
2
,
A
u
g
.
2
0
2
3
.
[
1
7
]
T.
T
.
P
.
T
h
a
n
h
,
C
.
V
u
n
u
n
u
,
S
.
A
t
o
e
v
,
S
.
-
H
.
Le
e
,
a
n
d
K
.
-
R
.
K
w
o
n
,
“
L
e
u
k
e
mi
a
b
l
o
o
d
c
e
l
l
i
ma
g
e
c
l
a
ss
i
f
i
c
a
t
i
o
n
u
s
i
n
g
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
C
o
m
p
u
t
e
r
T
h
e
o
ry
a
n
d
En
g
i
n
e
e
ri
n
g
,
v
o
l
.
1
0
,
n
o
.
2
,
p
p
.
5
4
–
5
8
,
2
0
1
8
,
d
o
i
:
1
0
.
7
7
6
3
/
I
JC
TE.
2
0
1
8
.
V
1
0
.
1
1
9
8
.
[
1
8
]
M
.
B
e
n
n
e
t
R
a
j
e
s
h
a
n
d
S
.
S
a
t
h
i
a
m
o
o
r
t
h
y
,
“
C
l
a
ss
i
f
i
c
a
t
i
o
n
o
f
l
e
u
k
e
m
i
a
i
ma
g
e
u
s
i
n
g
g
e
n
e
t
i
c
b
a
se
d
K
-
n
e
a
r
e
s
t
n
e
i
g
h
b
o
r
(
G
-
K
N
N
)
,
”
Asi
a
n
J
o
u
r
n
a
l
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
7
,
n
o
.
2
,
p
p
.
1
1
3
–
1
1
7
,
A
u
g
.
2
0
1
8
,
d
o
i
:
1
0
.
5
1
9
8
3
/
a
j
c
s
t
-
2
0
1
8
.
7
.
2
.
1
8
6
9
.
[
1
9
]
S
.
S
h
a
f
i
q
u
e
a
n
d
S
.
Te
h
si
n
,
“
A
c
u
t
e
l
y
mp
h
o
b
l
a
s
t
i
c
l
e
u
k
e
m
i
a
d
e
t
e
c
t
i
o
n
a
n
d
c
l
a
ssi
f
i
c
a
t
i
o
n
o
f
i
t
s
su
b
t
y
p
e
s
u
s
i
n
g
p
r
e
t
r
a
i
n
e
d
d
e
e
p
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
s,”
T
e
c
h
n
o
l
o
g
y
i
n
C
a
n
c
e
r
Re
s
e
a
rch
&
T
re
a
t
m
e
n
t
,
v
o
l
.
1
7
,
J
a
n
.
2
0
1
8
,
d
o
i
:
1
0
.
1
1
7
7
/
1
5
3
3
0
3
3
8
1
8
8
0
2
7
8
9
.
[
2
0
]
N
.
A
h
m
e
d
,
A
.
Y
i
g
i
t
,
Z
.
I
si
k
,
a
n
d
A
.
A
l
p
k
o
c
a
k
,
“
I
d
e
n
t
i
f
i
c
a
t
i
o
n
o
f
l
e
u
k
e
mi
a
s
u
b
t
y
p
e
s
f
r
o
m
mi
c
r
o
s
c
o
p
i
c
i
ma
g
e
s
u
s
i
n
g
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
,
”
D
i
a
g
n
o
s
t
i
c
s
,
v
o
l
.
9
,
n
o
.
3
,
A
u
g
.
2
0
1
9
,
d
o
i
:
1
0
.
3
3
9
0
/
d
i
a
g
n
o
st
i
c
s
9
0
3
0
1
0
4
.
[
2
1
]
D
.
K
u
mar
e
t
a
l
.
,
“
A
u
t
o
ma
t
i
c
d
e
t
e
c
t
i
o
n
o
f
w
h
i
t
e
b
l
o
o
d
c
a
n
c
e
r
f
r
o
m
b
o
n
e
m
a
r
r
o
w
mi
c
r
o
sc
o
p
i
c
i
m
a
g
e
s
u
s
i
n
g
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
s,”
I
EEE
Ac
c
e
ss
,
v
o
l
.
8
,
p
p
.
1
4
2
5
2
1
–
1
4
2
5
3
1
,
2
0
2
0
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
0
.
3
0
1
2
2
9
2
.
[
2
2
]
M
.
L
o
e
y
,
M
.
N
a
ma
n
,
a
n
d
H
.
Z
a
y
e
d
,
“
D
e
e
p
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
i
n
d
i
a
g
n
o
si
n
g
l
e
u
k
e
m
i
a
i
n
b
l
o
o
d
c
e
l
l
s
,
”
C
o
m
p
u
t
e
rs
,
v
o
l
.
9
,
n
o
.
2
,
A
p
r
.
2
0
2
0
,
d
o
i
:
1
0
.
3
3
9
0
/
c
o
m
p
u
t
e
r
s
9
0
2
0
0
2
9
.
[
2
3
]
S
.
D
a
sari
r
a
j
u
,
M
.
H
u
o
,
a
n
d
S
.
M
c
C
a
l
l
a
,
“
D
e
t
e
c
t
i
o
n
a
n
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
o
f
i
mm
a
t
u
r
e
l
e
u
k
o
c
y
t
e
s
f
o
r
d
i
a
g
n
o
s
i
s
o
f
a
c
u
t
e
m
y
e
l
o
i
d
l
e
u
k
e
mi
a
u
si
n
g
r
a
n
d
o
m
f
o
r
e
s
t
a
l
g
o
r
i
t
h
m,”
B
i
o
e
n
g
i
n
e
e
r
i
n
g
,
v
o
l
.
7
,
n
o
.
4
,
O
c
t
.
2
0
2
0
,
d
o
i
:
1
0
.
3
3
9
0
/
b
i
o
e
n
g
i
n
e
e
r
i
n
g
7
0
4
0
1
2
0
.
[
2
4
]
P
.
K
.
D
a
s,
P
.
J
a
d
o
u
n
,
a
n
d
S
.
M
e
h
e
r
,
“
D
e
t
e
c
t
i
o
n
a
n
d
c
l
a
ssi
f
i
c
a
t
i
o
n
o
f
a
c
u
t
e
l
y
mp
h
o
c
y
t
i
c
l
e
u
k
e
m
i
a
,
”
i
n
2
0
2
0
I
E
EE
-
H
Y
D
C
O
N
,
S
e
p
.
2
0
2
0
,
p
p
.
1
–
5
.
d
o
i
:
1
0
.
1
1
0
9
/
H
Y
D
C
O
N
4
8
9
0
3
.
2
0
2
0
.
9
2
4
2
7
4
5
.
[
2
5
]
M
.
S
h
a
h
e
e
n
e
t
a
l
.
,
“
A
c
u
t
e
my
e
l
o
i
d
l
e
u
k
e
m
i
a
(
A
M
L)
d
e
t
e
c
t
i
o
n
u
si
n
g
A
l
e
x
N
e
t
m
o
d
e
l
,
”
C
o
m
p
l
e
x
i
t
y
,
v
o
l
.
2
0
2
1
,
n
o
.
1
,
J
a
n
.
2
0
2
1
,
d
o
i
:
1
0
.
1
1
5
5
/
2
0
2
1
/
6
6
5
8
1
9
2
.
[
2
6
]
H
.
Li
,
X
.
Z
h
a
o
,
A
.
S
u
,
H
.
Zh
a
n
g
,
J
.
L
i
u
,
a
n
d
G
.
G
u
,
“
C
o
l
o
r
sp
a
c
e
t
r
a
n
sf
o
r
m
a
t
i
o
n
a
n
d
m
u
l
t
i
-
c
l
a
ss
w
e
i
g
h
t
e
d
l
o
ss
f
o
r
a
d
h
e
s
i
v
e
w
h
i
t
e
b
l
o
o
d
c
e
l
l
se
g
me
n
t
a
t
i
o
n
,
”
I
EE
E
A
c
c
e
s
s
,
v
o
l
.
8
,
p
p
.
2
4
8
0
8
–
2
4
8
1
8
,
2
0
2
0
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
0
.
2
9
7
0
4
8
5
.
[
2
7
]
S
.
R
a
i
n
a
,
A
.
K
h
a
n
d
e
l
w
a
l
,
S
.
G
u
p
t
a
,
a
n
d
A
.
Le
e
k
h
a
,
“
B
l
o
o
d
c
e
l
l
s
d
e
t
e
c
t
i
o
n
u
s
i
n
g
f
a
s
t
e
r
-
R
C
N
N
,
”
i
n
2
0
2
0
I
E
EE
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
i
n
g
,
P
o
w
e
r
a
n
d
C
o
m
m
u
n
i
c
a
t
i
o
n
T
e
c
h
n
o
l
o
g
i
e
s
(
G
U
C
O
N
)
,
O
c
t
.
2
0
2
0
,
p
p
.
2
1
7
–
2
2
2
.
d
o
i
:
1
0
.
1
1
0
9
/
G
U
C
O
N
4
8
8
7
5
.
2
0
2
0
.
9
2
3
1
1
3
4
.
Evaluation Warning : The document was created with Spire.PDF for Python.