medRxiv (Cold Spring Harbor Laboratory),
Journal Year:
2024,
Volume and Issue:
unknown
Published: Aug. 9, 2024
Abstract
The
development
of
Artificial
Intelligence
(AI)
in
the
healthcare
sector
is
generating
a
great
impact.
However,
one
primary
challenges
for
implementation
this
technology
access
to
high-quality
data
due
issues
collection
and
regulatory
constraints,
which
synthetic
an
emerging
alternative.
This
Scoping
review
analyses
reviews
from
past
10
years
three
different
databases
(i.e.,
PubMed,
Scopus,
Web
Science)
identify
domains
where
are
currently
generated,
motivations
behind
their
creation,
future
uses,
limitations,
types
data.
A
total
13
main
were
identified,
with
Oncology,
Neurology,
Cardiology
being
most
frequently
mentioned.
Five
principal
uses
also
identified.
Furthermore,
it
was
found
that
predominant
type
generated
unstructured,
particularly
images.
Finally,
several
work
directions
suggested,
including
exploring
new
less
commonly
used
(e.g.,
video
text),
developing
evaluation
benchmark
standard
generative
models
specific
domains.
Research Square (Research Square),
Journal Year:
2023,
Volume and Issue:
unknown
Published: Nov. 8, 2023
Reliable
differentiation
of
uveal
melanoma
and
choroidal
nevi
is
crucial
to
guide
appropriate
treatment,
preventing
unnecessary
procedures
for
benign
lesions
ensuring
timely
treatment
potentially
malignant
cases.
The
purpose
this
study
validate
deep
learning
classification
nevi,
evaluate
the
effect
color
fusion
options
on
performance.
O
melanoma
conjuntival
é
uma
neoplasia
maligna,
que
geralmente
se
apresenta
como
lesão
nodular
pigmentada.
Casos
variantes
com
diversas
formas
atípicas
podem
atrasar
a
identificação.
Com
o
intuito
de
auxiliar
médico
no
diagnóstico
precoce,
minimizando
os
riscos
ao
paciente,
este
trabalho
tem
objetivo
realizar
um
estudo
comparativo
algoritmos
para
classificar
tumores
melanocíticos
conjuntivais.
Para
isso,
foram
avaliados
modelos
baseados
em
Redes
Neurais
Convolucionais
classificação
binária
e
multiclasse
dos
tumores,
partir
VGG16,
Xception
MobileNetV2,
utilizando
técnica
Transfer
Learning
melhorar
generalização
modelos.
final
da
imagem,
foi
realizada
abordagem
baseada
assembleia
classificadores,
composta
pelos
PMC,
SVM
KNN.
utilizou
base
dados
406
imagens,
aplicando
técnicas
balanceamento
dados,
SMOTE
ADASYN.
encontrar
modelo
melhor
desempenho,
usada
validação
cruzada
5-folds.
Considerando
todos
testes
realizados,
Ensemble
MobileNetV2
obteve
melhores
resultados.
medRxiv (Cold Spring Harbor Laboratory),
Journal Year:
2024,
Volume and Issue:
unknown
Published: Aug. 9, 2024
Abstract
The
development
of
Artificial
Intelligence
(AI)
in
the
healthcare
sector
is
generating
a
great
impact.
However,
one
primary
challenges
for
implementation
this
technology
access
to
high-quality
data
due
issues
collection
and
regulatory
constraints,
which
synthetic
an
emerging
alternative.
This
Scoping
review
analyses
reviews
from
past
10
years
three
different
databases
(i.e.,
PubMed,
Scopus,
Web
Science)
identify
domains
where
are
currently
generated,
motivations
behind
their
creation,
future
uses,
limitations,
types
data.
A
total
13
main
were
identified,
with
Oncology,
Neurology,
Cardiology
being
most
frequently
mentioned.
Five
principal
uses
also
identified.
Furthermore,
it
was
found
that
predominant
type
generated
unstructured,
particularly
images.
Finally,
several
work
directions
suggested,
including
exploring
new
less
commonly
used
(e.g.,
video
text),
developing
evaluation
benchmark
standard
generative
models
specific
domains.
