Inter
national
J
our
nal
of
Electrical
and
Computer
Engineering
(IJECE)
V
ol.
10,
No.
6,
December
2020,
pp.
5974
5984
ISSN:
2088-8708,
DOI:
10.11591/ijece.v10i6.pp5974-5984
r
5974
Medical
vision:
W
eb
and
mobile
medical
image
r
etrie
v
al
system
based
on
google
cloud
vision
I
K
etut
Gede
Darma
Putra,
Dewa
Made
Sri
Arsa,
I
Gusti
Ngurah
Dwi
v
a
Hardijaya,
I
Gede
Galang
Surya
Prabawa,
I
Made
Aris
Satia
W
idiatmika
Information
T
echnology
,
F
aculty
of
Engineering,
Uni
v
ersitas
Udayana,
Indonesia
Article
Inf
o
Article
history:
Recei
v
ed
Jul
28,
2019
Re
vised
Apr
1,
2020
Accepted
May
2,
2020
K
eyw
ords:
Image
retrie
v
al
Medical
imaging
Mobile
application
W
eb
application
ABSTRA
CT
The
application
of
information
technology
is
rapidly
utilized
in
the
medical
system.
There
is
also
a
massi
v
e
de
v
elopment
in
the
automatic
method
for
recogni
zing
and
detecting
objects
in
the
real
w
orld.
In
this
study
,
we
present
a
syste
m
called
Medical
V
ision
which
is
designed
for
people
who
has
no
e
xpertise
in
medical.
Medical
V
i-
sion
is
a
web
and
mobile-
based
application
to
gi
v
e
an
initial
kno
wledge
in
a
medical
image.
This
system
has
5
features;
object
detection,
web
detection,
object
labeling,
safe
search,
and
image
properties.
These
features
are
run
by
embedding
Google
V
i-
sion
API
in
the
system.
W
e
e
v
aluate
this
system
by
observing
the
result
of
some
medical
images
which
inputted
into
the
system.
The
results
sho
wed
that
our
system
presents
a
promising
performance
and
able
to
gi
v
e
rele
v
ant
information
related
to
the
gi
v
en
image.
Copyright
c
2020
Insitute
of
Advanced
Engineeering
and
Science
.
All
rights
r
eserved.
Corresponding
A
uthor:
I
K
etut
Gede
Darma
Putra,
Information
T
echnology
,
F
aculty
of
Engineering,
Uni
v
ersitas
Udayana,
Mengwitani,
Bali,
Indonesia.
Email:
ikgdarmaputra@unud.ac.id
1.
INTR
ODUCTION
In
this
globalization
era,
technology
has
been
in
touch
in
numerous
li
ving
aspects.
In
education,
massi
v
e
technological
de
v
elopment
brings
a
ne
w
learning
method
which
is
called
online
learning
or
distance
learning.
Barbara
et
al.
[1]
sho
w
that
students
who
tak
e
online
learning
performed
better
than
those
who
tak
e
f
ace-to-f
ace
instruction.
In
the
medical
field,
moreo
v
er
,
intelligent
technology
is
adapted
to
present
and
analyze
the
medical
image
to
help
the
reader
(doctor
,
nurse,
etc.)
in
making
the
right
decision.
A
computerized
analysis
system
w
as
firstly
initiat
edby
Lusted
in
1960s.
He
sho
wed
that
an
automatic
system
could
be
used
to
determine
the
abnormalit
y
in
chest
photofluorograms
[2].
Others
w
ork
later
studied
a
computer
analysis
and
diagnosis
on
bone
cancer
image
[3].
Since
then,
v
arious
computer
-assisted
diagnosis
(CAD)
systems
were
de
v
eloped
in
medical
image.
When
designing
CAD,
the
characteristic
of
image
has
to
address
firstly
to
ensure
the
kind
of
method
needed
to
impro
v
e
the
performance
of
the
system.
Medical
image
is
tak
en
from
a
high-end
medical
de
vice
which
cannot
g
ain
by
human
vision
capabilities.
This
image
has
tw
o
characteristics;
high
resolution
and
high
pix
el
depth
[4].
Ho
we
v
er
,
in
a
certain
condition,
the
produced
image
is
not
clear
enough
because
of
noises.
Therefore,
impro
ving
the
quality
of
the
image
is
necessary
to
deli
v
er
v
aluable
information
to
the
doctor
.
In
adv
ance,
an
intelligent
system
can
be
used
to
pro
vide
an
early
diagnosis
for
them.
T
o
deal
with
medical
image
challenges,
v
arious
research
tried
to
addopt
an
intelligent
method
to
pro-
cess
and
pro
vide
analysis
automatically
.
In
breast
cancer
,
the
research
area
is
detecting
cancer
in
h
yperspectral
imaging
[5],
breast
cancer
classification
[6,
7],
optical
imaging
and
augmented
reality
visualization
[8,
9].
Then,
automatic
methods
were
b
uild
to
Lung
diseases
on
CT
images
[10,
11].
Moreo
v
er
,
similar
methods
were
also
created
to
analyze
skin
diseases
from
skin
image
[12-21].
If
we
tak
e
a
look
in
more
detail
on
the
method,
J
ournal
homepage:
http://ijece
.iaescor
e
.com
Evaluation Warning : The document was created with Spire.PDF for Python.
Int
J
Elec
&
Comp
Eng
ISSN:
2088-8708
r
5975
deep
learning
has
been
chosen
recently
and
massi
v
ely
as
one
of
the
methods
for
automatically
analyzing
the
medical
image
[22-39].
Deep
learning
has
been
widely
used
to
analyze
medical
images
in
skin
lesion
classifica-
tion
[19-21],
breast
cancer
class
ification
[23-26],
and
melanoma
detection
[27-29].
From
those
research,
their
methods
are
b
uilt
for
the
medical
en
vironment
for
specific
task.
Since
adv
ancing
method
in
machine
learning,
lik
e
deep
learning,
and
a
v
ailability
tons
of
image
processing
libraries
which
can
be
used
freely
,
a
system
which
can
assist
human
to
impro
v
e
their
understanding
on
the
medical
image
is
highly
needed.
In
this
paper
,
dif
ferent
with
pre
vious
research,
we
proposed
a
sys
tem
with
deep
analysis
on
medi
cal
image.
Our
system
is
design
to
produce
e
xtensi
v
e
analysis
on
a
gi
v
en
medical
image.
Our
system
can
detect
medical
object
and
analyze
it
for
better
understanding
on
it.
W
e
design
our
system
to
run
on
web-based
system.
T
o
impro
v
e
mobility
of
our
system,
we
also
de
v
elope
a
mobile-based
application.
These
syst
ems
are
embedded
fi
v
e
features.
The
first
feature
is
medical
object
detection.
Its
purpose
is
detecting
all
objects
occurred
in
the
scene.
The
second
feature
is
medical
website
detection
which
is
designed
to
search
related
articles
or
images
associate
with
the
gi
v
en
image.
The
third
feature
is
object
labeling.
This
feature
performs
entity
cate
gorization
and
rank
the
results
base
d
on
confidence
score.
The
fourth
feature
is
safe
search
which
aims
to
classify
the
type
of
content
related
to
the
gi
v
en
image.
The
last
feature
is
image
properties.
It
pro
vides
the
detail
of
the
image
based
on
pix
el
information.
This
study
is
written
as
follo
ws.
In
section
2.
we
pro
vide
brief
information
about
Google
V
is
ion.
Then,
we
present
our
proposed
system
in
section
3.
After
that,
in
section
4.
we
present
the
detail
of
system
implementation,
the
result
of
the
e
xperiments,
and
some
discussions
of
the
result.
Then,
we
gi
v
e
the
conclusion
of
this
study
and
insight
for
future
research
in
section
5.
2.
GOOGLE
CLOUD
VISION
API
Google
cloud
vision
API
is
a
service
from
Google
cloud
platform
(GCP)
that
can
pro
vide
an
anal
ysis
of
an
image.
This
API
w
as
released
on
May
18,
2017,
with
Machi
n
e
Learning
and
Big
Data
technology
which
became
the
engine
behind
it.
Cloud
V
ision
API
is
used
to
identify
objects
in
an
image
such
as
te
xt,
symbols,
and
types
of
product
objects
digitally
.
Google
has
man
y
scenarios
in
the
Cloud
V
ision
API.
F
or
e
xample
de
v
elopers
can
use
Cloud
V
i
sion
API
to
detect
whether
there
is
a
mobile
in
the
image,
detect
inappropriate
content,
analyze
some
o
ne
’
s
emotions
recorded
in
the
image,
and
e
xtract
the
writing.
Cloud
V
ision
API
supports
the
detection
of
objects
in
images
using
the
same
technology
on
Google
Photos
so
that
de
v
elopers
can
find
out
the
names
of
objects
in
a
photo.
Besides,
this
API
can
be
used
to
a
v
oid
i
nappropriate
image
content
detected
with
Google
SafeSearch.
Cloud
V
ision
API
can
also
be
used
to
analyze
people’
s
emotions
and
detect
v
arious
logos
from
f
amous
products.
Google’
s
API
is
also
able
to
detect
the
letters
contained
in
images
with
automatic
language
identification.
3.
PR
OPOSED
SYSTEM
Generally
,
the
process
of
our
proposed
system
can
be
seen
in
Figure
1a.
The
image
will
be
uploaded
to
our
syst
em.
Our
engine
will
analyze
it
and
pro
vide
se
v
eral
informations.
W
e
realized
these
information
as
features.
W
e
designed
fi
v
e
features
on
our
system.
Those
features
are
object
detection,
label
detection,
web
detection,
image
properties,
and
safe
search.
Each
feature
deli
v
ers
dif
ferent
information
to
the
user
.
Those
features
will
be
e
xplained
as
follo
ws:
(a)
Object
Detection:
This
feature
is
used
to
detect
objects
which
can
be
found
in
the
uploaded
image.
The
detected
objects
will
be
rounded
with
a
square.
(b)
Label
Detection:
F
or
each
image
uploaded
by
the
user
,
Medical
V
ision
will
analyze
the
type
of
content
occurred
in
it.
F
or
e
xample,
an
actinic
k
eratosis
image
is
uploaded
to
the
system.
This
system
will
produce
se
v
eral
labels
correlated
to
the
image
which
ha
v
e
been
sorted
by
their
confidence
score.
F
or
the
actinic
k
eratosis
case,
the
detected
labels
will
be
finger
,
skin,
hand,
and
thumb,
joint,
nail,
gesture,
flesh,
and
wrist.
(c)
W
eb
Detection:
This
feature
is
used
to
recommend
related
websites
which
e
xplain
the
image
in
more
detail.
T
w
o
types
cate
gory
of
website
are
produced;
partially
matched
images
and
pages.
Beside
of
this
links,
tw
o
more
information
are
gi
v
en
in
this
feature.
The
first
information
is
the
best
guess
la-
bel
which
mostly
matched
for
the
gi
v
en
image.
Secondly
,
this
feature
presents
entities
information.
These
entities
present
related
diseases
related
to
the
image.
F
or
e
xample,
gi
v
en
an
image
labeled
as
actinic
k
eratoses
,
our
system
will
produce
se
v
eral
entities;
actinic
k
eratosis,
k
eratosis,
actinic
cheilitis,
Medical
vision:
W
eb
and
mobile
medical
ima
g
e
r
etrie
val
system...
(I
K
etut
Gede
Darma
Putr
a)
Evaluation Warning : The document was created with Spire.PDF for Python.
5976
r
ISSN:
2088-8708
skin
cancer
,
a
precancerous
condition,
k
eratoses,
seborrheic
k
eratosis,
therap
y
,
skin,
and
lesion.
The
first
label
has
the
highest
confidence
score.
(d)
Image
Properties:
Image
properties
feature
is
used
to
analyze
the
image
based
on
its
pix
el
v
alue.
This
feature
gi
v
es
the
detail
information
of
entities
produced
in
web
detection
feature.
F
or
each
entity
,
three
v
alues
are
computed;
RGB,
score,
and
pix
el
fraction.
(e)
Safe
Search:
Safe
search
feature
is
used
to
cate
gorize
t
he
image
into
content
classes.
In
this
feature,
the
image
will
cate
gorize
into
adult
content,
spoof
content,
medical
content,
violenc
e
content,
and
rac
y
content.
F
or
each
cate
gory
,
a
score
will
be
gi
v
en
as
v
ery
lik
ely
,
lik
ely
,
possible,
unlik
ely
,
and
v
ery
unlik
ely
.
F
or
e
xample,
the
image
in
Figure
1b
will
be
annotated
lik
ely
as
Medical
content,
possible
as
V
iolence
and
Rac
y
content,
and
v
ery
unlik
ely
as
Adult
and
Spoof
content.
(a)
(b)
Figure
1.
a)
Medical
image
will
be
inputted
to
our
system
for
analysis
process
on
fi
v
e
features
and
b)
Example
of
medical
image:
actinic
k
eratosis
disease
Figure
2.
The
architecture
of
medical
vision
system
Int
J
Elec
&
Comp
Eng,
V
ol.
10,
No.
6,
December
2020
:
5974
–
5984
Evaluation Warning : The document was created with Spire.PDF for Python.
Int
J
Elec
&
Comp
Eng
ISSN:
2088-8708
r
5977
T
o
impro
v
e
ease
of
access,
we
designed
our
system
in
the
form
of
a
web-based
system
and
android
application.
As
seen
in
Figure
2,
the
user
can
upload
an
image
through
both
applications.
This
image
will
be
sent
to
a
web
serv
er
.
This
serv
er
mak
es
a
request
to
Google
V
ision
Serv
er
usi
ng
Google
V
i
sion
API.
After
that,
this
Google
V
ision
Serv
er
produces
the
image
references
and
send
it
back
to
the
web
serv
er
in
JSON
file.
Lastly
,
the
web
serv
er
will
send
this
result
to
the
user
.
4.
RESUL
T
AND
DISCUSSION
As
stated
pre
viously
,
Medical
V
ision
w
as
de
v
eloped
in
a
web-based
and
mobi
le-based
s
ystem.
The
web-based
system
w
as
de
v
eloped
under
HTML
whil
e
the
mobile
application
w
as
designed
for
Android
only
.
As
sho
wn
in
Figure
2,
the
uploaded
image
in
both
bro
wser
and
mobile-apps
will
be
processed
in
the
web
serv
er
where
process
this
image
using
Google
V
ision
API.
W
e
e
v
aluate
this
system
by
inputting
se
v
eral
medical
images
into
the
system;
act
inic
k
eratosis,
b
ullous
pemphigoid,
chick
enpox,
eczema,
herpes
zoster
,
impetigo,
k
eloid,
k
eratoacanthoma,
lichen
planus,
melanoma,
pustular
psoriasis,
seborrheic
k
eratosis,
and
tinea
barbae.
4.1.
W
eb-based
implementation
The
e
xample
of
our
implementation
can
be
seen
in
Figure
3-8.
In
this
e
xample,
we
test
our
syst
em
by
inputting
the
Actinic
K
eratosis
image.
Actinic
k
eratosis
(AK)
is
a
disease
that
can
occur
in
our
skin
caused
by
ultra
violet
radiation.
As
sho
wn
in
Figure
3,
we
upload
a
picture
which
is
classified
as
AK.
The
object
analysis
tells
that
the
image
is
labeled
as
Person
with
confidence
score
by
86
:
17%
.
Then,
in
Figure
4,
the
labels
predicted
by
our
system
sho
ws
that
the
image
is
related
to
finger
,
skin,
hand,
thumb,
joint,
nail,
gesture,
flesh,
and
whist
where
the
highest
score
is
obtained
as
finger
.
Interestingly
,
in
web
detection
feature,
our
system
accurately
predicts
the
image
as
AK
with
confidence
score
by
87
:
34%
follo
wed
by
k
eratos,
artinic
chelitis,
skin
cancer
,
a
precancerous
condition,
k
eratoses,
seborrheic
k
erat
osis,
therap
y
,
skin,
and
lesion
as
sho
wn
in
Figure
5.
When
the
web
detection
result
is
click
ed,
the
page
will
be
directed
to
the
ne
w
page
which
will
sho
w
some
information
related
to
the
image.
In
the
safe
search
feature,
furthermore,
the
system
will
gi
v
e
a
general
type
of
content
for
the
gi
v
en
image
as
sho
wn
in
Figure
8.
F
or
AK
image,
the
result
is
classified
lik
ely
as
medical
content,
possible
for
violence
and
rac
y
content,
and
v
ery
unlik
ely
as
adult
and
spoof
content.
4.2.
Mobile-based
implementation
The
implementation
of
Medical
V
ision
in
mobile
apps
can
be
seen
in
Figure
9.
This
image
is
actually
hands
which
were
af
fected
by
Leprosy
disease
in
https://www
.who.int/lep/disease/en/.
This
disease
is
sho
wed
that
it
infects
the
skin
b
ut
it
also
may
af
fect
peripheral
nerv
es.
As
sho
wn
in
Figure
9a,
the
system
detects
tw
o
objects
in
the
gi
v
en
image.
Those
objects
are
glo
v
e
and
animal
where
glo
v
e
has
the
highest
confidence
score.
In
label
detection,
the
labels
results
sho
w
that
it
highly
related
to
the
image
q
ua
litati
v
ely
.
Interestingly
,
the
web
detection
presents
that
the
im
age
labeled
as
Leprosy
with
81
:
49%
e
vidence
score.
The
apps
also
pro
vide
a
similar
image
from
the
website.
Figure
3.
Object
detection
result
for
actinic
k
eratosis
image
Medical
vision:
W
eb
and
mobile
medical
ima
g
e
r
etrie
val
system...
(I
K
etut
Gede
Darma
Putr
a)
Evaluation Warning : The document was created with Spire.PDF for Python.
5978
r
ISSN:
2088-8708
Figure
4.
Label
detection
result
for
actinic
k
eratosis
image
Figure
5.
W
eb
detection
result
for
actinic
k
eratosis
image
Figure
6.
Example
pre
vie
w
when
the
link
in
web
detection
result
w
as
click
ed
Int
J
Elec
&
Comp
Eng,
V
ol.
10,
No.
6,
December
2020
:
5974
–
5984
Evaluation Warning : The document was created with Spire.PDF for Python.
Int
J
Elec
&
Comp
Eng
ISSN:
2088-8708
r
5979
Figure
7.
Image
properties
result
for
actinic
k
eratosis
image
Figure
8.
Safe
search
result
for
actinic
k
eratosis
image
(a)
(b)
Medical
vision:
W
eb
and
mobile
medical
ima
g
e
r
etrie
val
system...
(I
K
etut
Gede
Darma
Putr
a)
Evaluation Warning : The document was created with Spire.PDF for Python.
5980
r
ISSN:
2088-8708
(c)
(d)
Figure
9.
Implementation
result
of
Medical
vision
in
mobile
application,
a)
Object
detection,
b)
Label
detection,
c)
W
eb
detection,
and
d)
Similar
image
(continue)
4.3.
P
erf
ormance
analysis
The
pre
vious
subsection
sho
ws
the
e
xample
of
the
implementation
of
Medical
V
ision
in
web-based
and
mobile-based
system.
Beside
of
that,
we
e
v
aluate
our
system
performance
for
13
medical
images.
W
e
g
ather
the
results
in
T
able
1
The
first
column
presents
the
actual
label
of
the
image.
In
the
second
col-
umn,
the
detected
object
for
each
image
is
presented
aside
with
the
highes
t
confidence
score.
From
those
result,
most
of
the
objects
are
highly
correlated
in
qualitati
v
ely
.
Ho
we
v
er
,
in
pustular
psoriasis
and
seborrheic
k
eratosis,
the
object
detection
results
are
not
fit
the
actual
image.
In
pustular
psoriasis,
a
w
atermelon
object
is
detected
while
a
bak
ed
good
is
identified
in
seborrheic
k
eratosis.
If
we
observ
e,
ho
we
v
er
,
the
pustular
psoriasis
image,
this
disease
will
mak
e
a
pattern
in
the
skin
and
has
a
similar
pattern
lik
e
w
atermelon.
The
label
detection
res
ults
sho
w
promising
result
where
all
of
them
presents
high
relation
to
the
disease.
In
the
majority
,
the
skin
label
has
the
highest
confidence
score.
F
or
actinic
k
eratosis,
the
sample
image
which
is
sho
wn
in
Figure
3,
is
classified
as
a
finger
.
An
interesting
result
is
sho
wn
in
the
web
search
feature.
In
T
able
1
(see
appendix)
most
guess
labels
w
as
closely
matched
the
actual
image
labels.
The
web
search
entities
are
also
mostly
related
to
the
actual
image
label.
F
or
impetigo,
k
eratoacanthoma,
pustular
psoriasis,
and
seborrheic
k
eratosis,
web
search
feature
produces
uncorrelated
labels.
Ho
we
v
er
,
their
entities
sho
w
that
the
results
for
impetigo
and
pustular
psoriasis
images
are
not
fit
b
ut
others
match
up
the
labels.
This
labeling
error
may
occur
because
of
model
limitation.
The
safe
search
feature
also
sho
ws
promising.
The
results
present
that
all
test
images
cate
gorize
as
medical
content,
b
ut
not
all
images
predict
VL.
In
actinic
k
eratosis,
the
image
classifies
lik
ely
as
medical
image.
From
those
results,
there
is
a
relation
between
medical
and
violence
content.
When
the
image
labeled
as
medical
content,
it
also
will
be
said
as
violence
cont
ent.
F
or
e
xample,
impetigo
image
is
predicted
VL
as
medical
content
and
L
as
medical
content.
5.
CONCLUSION
In
this
study
,
we
ha
v
e
been
presented
our
system
which
is
called
Medical
V
ision.
Our
system
is
utilized
Google
Cloud
V
ision
API
for
processing
the
image.
This
system
w
as
b
uilt
for
common
people
with
less
kno
wledge
in
the
medical
image.
Based
on
our
e
v
aluation,
our
system
may
w
ork
properly
in
the
gi
v
en
test
case.
This
system
still
needs
some
enhancements
in
the
image
processing
method.
In
Google
Cloud
V
ision,
the
model
w
as
trained
using
v
arious
objects,
so
it
will
not
co
v
er
all
of
the
medical
objects.
Therefore,
a
ne
w
model
is
recommended
to
be
b
uilt
in
the
future.
Int
J
Elec
&
Comp
Eng,
V
ol.
10,
No.
6,
December
2020
:
5974
–
5984
Evaluation Warning : The document was created with Spire.PDF for Python.
Int
J
Elec
&
Comp
Eng
ISSN:
2088-8708
r
5981
APPENDIX
T
able
1.
The
result
of
our
system
analysis,
VU:
v
ery
unlik
ely
,
U:
unlik
ely
,
P:
possible,
L:
lik
ely
,
VL:
v
ery
lik
ely
Actual
Image
label
Object
Detection
Label
Detection
W
eb
Search
(Best
guess
labels)
W
eb
Search
(Entities)
Safe
Search
Actinic
K
eratosis
Person
:
86.17
Finger
:
96.17
K
eratose
Acinique
Actinic
K
eratosis
:
83.34
Adult
:
VU
K
eratosis
Spoof
:
VU
Actinic
chellitis
Medical
:
L
Skin
cancer
V
iolence
:
P
Precancerous
condition
Rac
y
:
P
Bullous
Pemhigoid
Egg
:
53,34
Skin
:
97.21
Bullous
Pemphigoid
Bullous
pemphigoid
:
73.01
Adult
:
VU
Nose
Linear
IgA
b
ullous
dermatosis
Spoof
:
VU
Ja
w
Bullous
dermatoses
Medical
:
VL
Mouth
Skin
condition
V
iolence
:
VL
Flesh
Cicatricial
pemphigoid
Rac
y
:
P
Chick
enpox
Skin
:
96,61
Chick
enpox
Chick
enpox
:
88.36
Adult
:
VU
Pink
Smallpox
Spoof
:
VU
Close-up
V
aricella
zoster
virus
Medical
:
VL
Peach
Attenuated
v
accine
V
iolence
:
L
Flesh
MMR
v
accine
Rac
y
:
VU
Eczhema
Person
:
95.12
Skin
:
97.37
Eczema
on
hands
Dermatitis
:
72.24
Adult
:
VU
Hand
Dyshidrosis
Spoof
:
VU
Fing
Hand
eczema
Medical
:
VL
Close-up
Eczema
V
iolence
:
P
Flesh
Dermatitis
Rac
y
:
P
Herpes
zoster
Skin
:
97.47
Herpes
Zoster
Shingles
:
73.59
Adult
:
U
Close-up
V
aricella
zoster
virus
Spoof
:
VU
Muscle
Chick
enpox
Medical
:
VL
Lip
Herpe
x
simple
x
virus
V
iolence
:
VL
Flesh
Therap
y
Rac
y
:
P
Impetigo
Skin
:
97
Close-up
Stock
photograph
y
Adult
:
U
Finger
Impetigo
Spoof
:
VU
Hand
Science
photo
library
Medical
:
VL
Flesh
-
V
iolence
:
L
Joint
Thumb
Rac
y
:
VL
K
eloid
Lipstick
Skin
Hypertropic
and
k
eloid
scar
Scar
Adult
:
VU
Scar
K
eloid
Spoof
:
VU
Lip
Hypertropic
scar
Medical
:
VL
Cheek
Sur
gery
V
iolence
:
L
Close-up
therap
y
Rac
y
:
P
K
eratoacanthoma
Skin
Close
up
K
eratoacanthoma
Adult
:
VU
Nose
Spoof
:
VU
Close-up
Medical
:
VL
Flesh
V
iolence
:
VL
No
v
el
Rac
y
:
U
Linchen
Planus
Skin
Lichen
planus
skin
Lichen
planus
Adult
:
VU
Joint
Spoof
:
VU
Hand
Medical
:
VL
Close-up
V
iolence
:
VL
Scar
Rac
y
:
P
Melanoma
Lipstick
Skin
Melanoma
skin
cancer
Skin
cancer
Adult
:
VU
Bro
wn
Melanoma
Spoof
:
VU
EyeBro
w
Cancer
Medical
:
VL
Cheek
Melanoc
yte
V
iolence
:
VL
Lip
Dermatology
Rac
y
:
P
Pustular
psoriasis
W
atermelon
Skin
Red
meat
Apple
iphone
8
Adult
:
VU
Flesh
Apple
Spoof
:
VU
Red
meat
Pustule
Medical
:
VL
W
ater
Ps
oriasis
V
iolence
:
L
Close-up
Fine
art
america
Rac
y
:
U
Seborrheic
k
eratosis
Bak
ed
Good
Skin
Rock
Seborrheic
k
eratosis
Adult
:
VU
Coockie
K
eratosis
Spoof
:
VU
Dessert
-
Medical
:
P
F
ood
Actinic
K
eratosis
V
iolence
:
U
Snack
Melanoma
Rac
y
:
VU
T
inea
Barbae
Lipstick
Skin
T
inea
barbae
T
inea
barbae
Adult
:
VU
Lip
-
Spoof
:
VU
F
ace
Mycosi
s
Medical
:
VL
Chin
T
inea
caphis
V
iolence
:
L
Nose
T
inea
corporis
Rac
y
:
L
Medical
vision:
W
eb
and
mobile
medical
ima
g
e
r
etrie
val
system...
(I
K
etut
Gede
Darma
Putr
a)
Evaluation Warning : The document was created with Spire.PDF for Python.
5982
r
ISSN:
2088-8708
A
CKNO
WLEDGEMENT
This
research
w
as
funded
by
Ministry
of
Research,
T
echnology
,
and
Higher
Education
of
the
Republic
of
Indonesia
with
grant
number
492.67/UN14.4.A/L
T/2019
entitled
by
”Smart
Medic
al
Record
(Sistem
Rekam
Medis
Elektronik
Online
T
erinte
grasi
deng
an
Dukung
an
Fingerprint,
T
ele
Image
Medicine,
Smart
Card,
dan
GIS
Google
Map)”.
REFERENCES
[1]
B.
Means,
et
al.,
”Ev
aluation
of
e
vidence-based
practices
in
online
learning:
A
meta-analysis
and
re
vie
w
of
online
learning
studies,
”
Centre
for
Learning
T
echnology
,
2009.
[2]
L.
B.
Lusted,
“Logical
analysis
in
roentgen
diagnosis:
memorial
fund
lect
ure,
”
Radiology
,
v
ol.
74,
no.
2,
pp.
178–193,
1960.
[3]
G.
S.
Lodwick,
”Radiographic
diagnosis
and
grading
of
bone
tumors,
with
comments
on
computer
e
v
al-
uation,
”
Proceedings
National
Cancer
Conference,
v
ol.
5,
pp.
369–380,
1964.
[4]
A.
Ortiz,
J.
Gorriz,
J.
Ramırez,
D.
Salas-Gonzalez,
and
J.
M.
Llamas-Elvira,
“T
w
o
fully-unsupervised
methods
for
mr
brain
image
se
gmentation
using
som-based
strate
gies,
”
Applied
Soft
Computing,
v
ol.
13,
no.
5,
pp.
2668–2682,
2013.
[5]
R.
Pik
e,
G.
Lu,
D.
W
ang,
Z.
G.
Chen,
and
B.
Fei,
“
A
minimum
spanning
forest-based
method
for
nonin-
v
asi
v
e
cancer
detection
with
h
yperspectral
imaging,
”
IEEE
T
ransactions
on
Biomedical
Engineering,
v
ol.
63,
no.
3,
pp.
653–663,
2015.
[6]
S.
Reis,
et
al.,
”Automated
classification
of
breast
cancer
stroma
maturity
from
histological
images,”
IEEE
T
ransactions
on
Biomedical
Engineering,
v
ol.
64,
no.
10,
pp.
2344–2352,
2017.
[7]
J.
Xu,
L.
Xiang,
Q.
Liu,
H.
Gilmore,
J.
W
u,
J.
T
ang,
and
A.
Madabhushi,
“Stack
ed
sparse
autoencoder
(ssae)
for
nuclei
detection
on
breast
cancer
histopathology
images,
”
IEEE
transactions
on
medical
imag-
ing,
v
ol.
35,
no.
1,
pp.
119–130,
2015.
[8]
X.
Chen,
D.
Y
ang,
F
.
Sun,
X.
Cao,
and
J
.
Liang,
“
Adapti
v
ely
alternati
v
e
light-transport-model-based
three-
dimensional
optic
al
imaging
for
longitudinal
and
quantitati
v
e
monitoring
of
g
astric
cancer
in
li
v
e
animal,
”
IEEE
T
ransactions
on
Biomedical
Engineering,
v
ol.
63,
no.
10,
pp.
2095–2107,
2015.
[9]
Y
.
Sato,
et
al.,
“Image
guidance
of
breast
cancer
sur
gery
using
3-d
ultrasound
images
and
augmented
reality
visualization,
”
IEEE
T
ransac-
tions
on
Medical
Imaging,
v
ol.
17,
no.
5,
pp.
681–693,
1998.
[10]
Y
.
Chunran,
W
.
Y
uan
vuan,
and
G.
Y
i,
“
Automati
c
detection
and
se
gmentation
of
lung
nodule
on
ct
im-
ages,
”
11th
International
Congress
on
Ima
ge
and
Signal
Processing,
BioMedical
Engineering
and
Infor
-
matics,
pp.
1–6,
2018.
[11]
X.
Liu,
L.
Ma,
L.
Song,
Y
.
Zhao,
X.
Zhao,
and
C.
Zhou,
“Recognizing
common
ct
imaging
signs
of
lung
diseases
through
a
ne
w
feature
selection
method
based
on
fisher
criterion
and
genetic
optimization,
”
IEEE
journal
of
biomedical
and
health
informatics,
v
ol.
19,
no.
2,
pp.
635–647,
2014.
[12]
M.
Sade
ghi,
T
.
K.
Lee,
D.
McLean,
H.
Lui,
and
M.
S.
Atkins,
“Detection
and
analysis
of
irre
gu-
lar
streaks
in
dermoscopic
images
of
skin
lesions,
”
IEEE
T
ransactions
on
Medical
Imaging,
v
ol.
32,
no.
5,
pp.
849–861,
2013.
[13]
D.
H.
Chung
and
G.
Sapiro,
“Se
gmenting
skin
lesions
wit
h
partial-dif
ferential-equations-based
image
processing
algorithms,
”
IEEE
transactions
on
Medical
Imaging,
v
ol.
19,
no.
7,
pp.
763–767,
2000.
[14]
R.
Sug
an
ya,
“
An
automated
computer
aided
diagnosis
of
skin
lesions
detection
and
classification
for
dermoscop
y
images,
”
International
Conference
on
Recent
T
rends
in
Information
T
echnology
,
pp.
1–5,
2016.
[15]
A.
H.
Shahin,
A.
Kamal,
and
M.
A.
Elattar
,
“Deep
ensemble
learning
for
skin
lesion
classification
from
dermo-
scopic
images,
”
9th
Cairo
International
Biomedical
Engineering
Conference,
pp.
150–153,
2018.
[16]
T
.
Pole
v
aya
,
R.
Ra
v
odin,
and
A.
Filchenk
o
v
,
“Skin
lesion
primary
morphology
classification
with
end-
to-end
deep
learning
netw
ork,
”
International
Conference
on
Artificial
Intelligence
in
Information
and
Communication,
pp.
247–250,
2019.
[17]
M.
A.
Albahar
,
“Skin
lesion
classification
using
con
v
olutional
neural
netw
ork
with
no
v
el
re
gularizer
,
”
IEEE
Access,
v
ol.
7,
pp.
306–313,
2019.
[18]
K.
Shimizu,
H.
Iyatomi,
M.
E.
Celebi,
K.-A.
Norton,
and
M.
T
anaka,
“F
our
-class
classification
of
skin
lesions
with
task
decomposition
strate
gy
,
”
IEEE
transactions
on
biomedical
engineering,
v
ol.
62,
no.
1,
pp.
274–283,
2014.
Int
J
Elec
&
Comp
Eng,
V
ol.
10,
No.
6,
December
2020
:
5974
–
5984
Evaluation Warning : The document was created with Spire.PDF for Python.
Int
J
Elec
&
Comp
Eng
ISSN:
2088-8708
r
5983
[19]
O.
Ab
uza
g
hl
eh,
B.
D.
Barkana,
and
M.
F
aezipour
,
“Nonin
v
asi
v
e
real-time
automated
skin
lesion
analysis
system
for
melanoma
early
detection
and
pre
v
ention,
”
IEEE
journal
of
translational
engineering
in
health
and
medicine,
v
ol.
3,
pp.
1–12,
2015.
[20]
J.
Ka
w
ahara,
S.
Daneshv
ar
,
G.
Ar
genziano,
and
G.
Hamarneh,
“Se
v
en-point
checklist
and
skin
lesion
classifica-
tion
using
multitask
multimodal
neural
nets,
”
IEEE
journal
of
biomedical
and
health
informat-
ics,
v
ol.
23,
no.
2,
pp.
538–546,
2018.
[21]
F
.
Rundo,
S.
Conoci,
G.
L.
Banna,
A.
Ortis,
F
.
Stanco,
and
S.
Battiato,
“Ev
aluati
on
of
le
v
en-
ber
g–marquardt
neural
netw
orks
and
stack
ed
autoencoders
clustering
for
ski
n
lesion
analysis,
screening
and
follo
w-up,
”
IET
Computer
V
ision,
v
ol.
12,
no.
7,
pp.
957–962,
2018.
[22]
D.
Mulf
ari,
A.
Celesti,
M.
F
azio,
M.
V
illari,
and
A.
Puliafito,
“Using
google
cloud
vision
in
assisti
v
e
technology
scenarios,
”
IEEE
Symposium
on
Computers
and
Communica
tion
(ISCC),
pp.
214–219,
2016.
[23]
P
.
Me
yer
,
V
.
Noblet,
C.
Mazzara,
and
A.
Lallement,
“Surv
e
y
on
deep
learning
for
radiotherap
y
,
”
Com-
puters
in
Biology
and
Medicine,
v
ol.
98,
pp.
126–146,
2018.
[24]
J.
de
la
T
orre,
A.V
alls,
and
D.
Puig,
“
A
deep
learning
interpret
able
classifier
for
diabetic
retinopath
y
disease
grading,
”
Neurocomputing,
2019.
[25]
A.
B.
Le
vine,
C.
Schlosser
,
J.
Gre
w
al,
R.
Coope,
S.
J.
Jones,
and
S.
Y
ip,
“Rise
of
the
machines:
Adv
ances
in
deep
learning
for
cancer
diagnosis,
”
T
rends
in
Cancer
,
v
ol.
5,
no.
3,
pp.
157–169,
2019.
[26]
M.
P
.
McBee,
et
al.,
“Deep
learning
in
radiol
ogy
,
”
Academic
Radiology
,
v
ol.
25,
no.
11,
pp.
1472–1480,
2018.
[27]
J.
Zhang,
Y
.
Xie,
Q.
W
u,
and
Y
.
Xia,
“Medical
image
classification
using
syner
gic
deep
learning,
”
Medical
Image
Analysis,
v
ol.
54,
pp.
10
–
19,
2019.
[28]
R.
W
ason,
“Deep
learning:
Ev
olution
and
e
xpansion,
”
Cogniti
v
e
Systems
Research,
v
ol.
52,
pp.
701–708,
2018.
[29]
A.
Mahbod,
G.
Schaefer
,
I.
Ellinger
,
R.
Eck
er
,
A.
Pitiot,
and
C.
W
ang,
“Fusing
fine-tuned
deep
features
for
skin
lesion
classification,
”
Computerized
Medical
Imaging
and
Graphics,
v
ol.
71,
pp.
19–29,
2019.
[30]
P
.
M.
Burlina,
N.
J.
Joshi,
E.
Ng,
S.
D.
Billings,
A.
W
.
Rebman,
and
J.
N.
Aucott,
“
Automated
detection
of
erythema
migrans
and
other
confounding
skin
lesions
via
deep
learning,
”
Computers
in
Biology
and
Medicine,
v
ol.
105,
pp.
151–156,
2019.
[31]
A.
Dascalu
and
E.
Da
vid,
“Skin
cancer
detection
by
deep
learning
and
sound
analysis
algorithms:
A
prospecti
v
e
clinical
study
of
an
elementary
dermoscope,
”
EBioMedicine,
v
ol.
43,
pp.
107–113,
2019.
[32]
M.
Chen,
P
.
Zhou,
D.
W
u,
L.
Hu,
M.
M.
Hassan,
and
A.
Alamri,
“
Ai-skin
:
Skin
disease
recognition
based
on
self-learning
and
wide
data
collection
through
a
closed
loop
frame
w
ork,
”
Information
Fusion,
2019.
[33]
S.
Khan,
N.
Islam,
Z.
Jan,
I.
U.
Din,
and
J.
J.
P
.
C.
Rodrigues,
“
A
no
v
el
deep
learning
based
frame
w
ork
for
the
detection
and
classification
of
breast
cancer
using
transfer
learning,
”
P
attern
Recognition
Letters,
v
ol.
125,
pp.
1–6,
2019.
[34]
X.
Li,
M.
Radulo
vic,
K.
Kanjer
,
and
K.
N.
Plataniotis,
“Discrimi
nati
v
e
pattern
mining
for
breast
can-
cer
histopathology
image
classification
via
fully
con
v
olutional
autoencoder
,
”
IEEE
Access,
v
ol.
7,
pp.
433–445,2019.
[35]
M.
K.
Elbashir
,
M.
Ezz,
M.
Mohammed,
and
S.
S.
Saloum,
“Lightweight
con
v
olutional
neural
netw
ork
for
breast
cancer
classification
using
rna-seq
gene
e
xpression
data,
”
IEEE
Access,
v
ol.
7,
pp.
338–185,
2019.
[36]
D.
Bardou,
K.
Zhang,
and
S.
M.
Ahmad,
“Classification
of
breast
cancer
based
on
histology
images
using
con
v
olutional
neural
netw
orks,
”
IEEE
Access,
v
ol.
6,
pp.
680–693,
2018.
[37]
S.
H.
Kassani
and
P
.
H.
Kas
sani,
“
A
comparati
v
e
study
of
deep
learning
architectures
on
melanoma
detection,
”
T
issue
and
Cell,
v
ol.
58,
pp.
76–83,
2019.
[38]
A.
A.
Ade
gun
and
S.
V
iriri,
“Deep
learning-based
system
for
automatic
melanoma
detection,
”
IEEE
Access,
2019.
[39]
L.
Y
u,
et
al.,
“
Automated
melanoma
recognition
in
dermoscop
y
images
via
v
ery
deep
residual
netw
orks,
”
IEEE
transactions
on
medical
imaging,
v
ol.
36,
no.
4,
pp.
994–1004,
2016.
Medical
vision:
W
eb
and
mobile
medical
ima
g
e
r
etrie
val
system...
(I
K
etut
Gede
Darma
Putr
a)
Evaluation Warning : The document was created with Spire.PDF for Python.