Scientific Reports,
Год журнала:
2024,
Номер
14(1)
Опубликована: Июнь 24, 2024
Abstract
Sarcopenic
obesity
(SO)
is
characterized
by
concomitant
sarcopenia
and
presents
a
high
risk
of
disability,
morbidity,
mortality
among
older
adults.
However,
predictions
based
on
sequential
neural
network
SO
studies
the
relationship
between
physical
fitness
factors
are
lacking.
This
study
aimed
to
develop
predictive
model
for
in
adults
focusing
factors.
A
comprehensive
dataset
Korean
participating
national
programs
was
analyzed
using
networks.
Appendicular
skeletal
muscle/body
weight
defined
as
an
anthropometric
equation.
Independent
variables
included
body
fat
(BF,
%),
waist
circumference,
systolic
diastolic
blood
pressure,
various
The
dependent
variable
binary
outcome
(possible
vs
normal).
We
hyperparameter
tuning
stratified
K-fold
validation
optimize
model.
prevalence
significantly
higher
women
(13.81%)
than
men,
highlighting
sex-specific
differences.
optimized
Shapley
Additive
Explanations
analysis
demonstrated
accuracy
93.1%,
with
BF%
absolute
grip
strength
emerging
most
influential
predictors
SO.
highly
accurate
adults,
emphasizing
critical
roles
strength.
identified
BF,
strength,
sit-and-reach
key
predictors.
Our
findings
underscore
nature
importance
its
prediction.
IEEE Access,
Год журнала:
2023,
Номер
11, С. 41003 - 41018
Опубликована: Янв. 1, 2023
Skin
cancer
is
a
prevalent
form
of
malignancy
globally,
and
its
early
accurate
diagnosis
critical
for
patient
survival.
Clinical
evaluation
skin
lesions
essential,
but
it
faces
challenges
such
as
long
waiting
times
subjective
interpretations.
Deep
learning
techniques
have
been
developed
to
tackle
these
assist
dermatologists
in
making
more
diagnoses.
Prompt
treatment
vital
prevent
progression
potentially
life-threatening
consequences.
The
use
deep
algorithms
can
improve
the
speed
accuracy
diagnosis,
leading
earlier
detection
treatment.
Additionally,
reduce
workload
healthcare
professionals,
allowing
them
concentrate
on
complex
cases.
goal
this
study
was
develop
reliable
(DL)
prediction
models
classification;
(i)
deal
with
typical
severe
class
imbalance
problem,
which
arises
because
skin-affected
patients'
significantly
smaller
than
healthy
class;
(ii)
interpret
model
output
better
understand
decision-making
mechanism
(iii)
Propose
an
End-to-End
smart
system
through
android
application.
In
comparison
examination
six
well-known
classifiers,
effectiveness
proposed
DL
technique
explored
terms
metrics
relating
both
generalization
capability
classification
accuracy.
A
used
HAM10000
dataset
optimized
CNN
identify
seven
forms
cancer.
trained
using
two
optimization
functions
(Adam
RMSprop)
three
activation
(Relu,
Swish,
Tanh).
Furthermore,
XAI-based
lesion
developed,
incorporating
Grad-CAM
Grad-CAM++
explain
model's
decisions.
This
help
doctors
make
informed
diagnoses
their
stages,
82%
0.47%
loss
Mathematics,
Год журнала:
2024,
Номер
12(7), С. 1030 - 1030
Опубликована: Март 29, 2024
The
medical
sciences
are
facing
a
major
problem
with
the
auto-detection
of
disease
due
to
fast
growth
in
population
density.
Intelligent
systems
assist
professionals
early
detection
and
also
help
provide
consistent
treatment
that
reduces
mortality
rate.
Skin
cancer
is
considered
be
deadliest
most
severe
kind
cancer.
Medical
utilize
dermoscopy
images
make
manual
diagnosis
skin
This
method
labor-intensive
time-consuming
demands
considerable
level
expertise.
Automated
methods
necessary
for
occurrence
hair
air
bubbles
dermoscopic
affects
research
aims
classify
eight
different
types
cancer,
namely
actinic
keratosis
(AKs),
dermatofibroma
(DFa),
melanoma
(MELa),
basal
cell
carcinoma
(BCCa),
squamous
(SCCa),
melanocytic
nevus
(MNi),
vascular
lesion
(VASn),
benign
(BKs).
In
this
study,
we
propose
SNC_Net,
which
integrates
features
derived
from
through
deep
learning
(DL)
models
handcrafted
(HC)
feature
extraction
aim
improving
performance
classifier.
A
convolutional
neural
network
(CNN)
employed
classification.
Dermoscopy
publicly
accessible
ISIC
2019
dataset
utilized
train
validate
model.
proposed
model
compared
four
baseline
models,
EfficientNetB0
(B1),
MobileNetV2
(B2),
DenseNet-121
(B3),
ResNet-101
(B4),
six
state-of-the-art
(SOTA)
classifiers.
With
an
accuracy
97.81%,
precision
98.31%,
recall
97.89%,
F1
score
98.10%,
outperformed
SOTA
classifiers
as
well
models.
Moreover,
Ablation
study
performed
on
its
performance.
therefore
assists
dermatologists
other
detection.
Journal of Experimental & Theoretical Artificial Intelligence,
Год журнала:
2024,
Номер
unknown, С. 1 - 26
Опубликована: Янв. 21, 2024
In
recent
years,
skin
cancer
has
been
the
most
dangerous
disease
noticed
among
people
worldwide.
Skin
should
be
identified
earlier
to
reduce
rate
of
mortality.
Employing
dermoscopic
images
can
identify
and
categorise
effectively.
But,
visual
evaluation
is
a
complex
procedure
done
in
image.
However,
Deep
learning
(DL)
an
efficient
method
for
detection;
however,
segmenting
lesion
automatic
localisation
stage
complicated.
this
paper,
novel
Ladybug
Beetle
Optimization-Double
Attention
Based
Multilevel
1-D
CNN
(LBO-DAM
CNN)
technique
proposed
detect
classify
cancer.
To
improve
type
discriminability,
two
types
attention
modules
are
introduced.
The
Ultra-Lightweight
Subspace
Module
(ULSAM)
utilised
classifying
feature
maps
into
different
stages
validate
frequency
from
image
samples.
multilayer
perceptron
module
(MLPAM)
determined
provide
information
regarding
classification
diminish
noise
unwanted
data.
minimise
data
loss,
it
then
combined
with
hierarchical
complementarity
during
classification.
Second,
modified
MLPAM
used
extract
significant
spaces
network
learning,
select
important
information,
space
redundancy.
Optimization
(LBO)
algorithm
provides
optimal
solution
by
minimising
loss
DAM
architecture.
experimentation
conducted
on
three
datasets
such
as
ISIC2020,
HAM10000,
melanoma
detection
dataset.
experimental
results
revealed
that
compared
existing
methods
IMFO-KELM,
Mask
RCNN,
M-SVM,
DCNN-9,
TL-CNN
datasets.
These
attained
94.56,
92.65,
90.56,
88.65,
95.5
ISIC2020
dataset
but
enhanced
performance
attaining
97.02.
Also,
validation
based
metrics,
namely,
accuracy,
precision,
sensitivity,
F1-score
97.03%,
97.05%,
97.58%,
97.27%
total
500
epochs.
Diagnostics,
Год журнала:
2023,
Номер
13(19), С. 3147 - 3147
Опубликована: Окт. 7, 2023
Skin
lesions
are
essential
for
the
early
detection
and
management
of
a
number
dermatological
disorders.
Learning-based
methods
skin
lesion
analysis
have
drawn
much
attention
lately
because
improvements
in
computer
vision
machine
learning
techniques.
A
review
most-recent
classification,
segmentation,
is
presented
this
survey
paper.
The
significance
healthcare
difficulties
physical
inspection
discussed
state-of-the-art
papers
targeting
classification
then
covered
depth
with
goal
correctly
identifying
type
from
dermoscopic,
macroscopic,
other
image
formats.
contribution
limitations
various
techniques
used
selected
study
papers,
including
deep
architectures
conventional
methods,
examined.
looks
into
focused
on
segmentation
that
aimed
to
identify
precise
borders
classify
them
accordingly.
These
make
it
easier
conduct
subsequent
analyses
allow
measurements
quantitative
evaluations.
paper
discusses
well-known
algorithms,
deep-learning-based,
graph-based,
region-based
ones.
difficulties,
datasets,
evaluation
metrics
particular
also
discussed.
Throughout
survey,
notable
benchmark
challenges,
relevant
highlighted,
providing
comprehensive
overview
field.
concludes
summary
major
trends,
potential
future
directions
detection,
aiming
inspire
further
advancements
critical
domain
research.
IEEE Access,
Год журнала:
2023,
Номер
11, С. 72120 - 72133
Опубликована: Янв. 1, 2023
Nowadays,
computer
vision
plays
an
essential
role
in
disease
detection,
computer-aided
diagnosis,
and
patient
risk
identification.
This
is
especially
true
for
skin
cancer,
which
can
be
fatal
if
not
diagnosed
its
early
stages.
For
this
purpose,
several
diagnostic
detection
systems
have
been
created
the
past.
They
were
limited
their
performance
because
of
complicated
visual
characteristics
lesion
images,
included
inhomogeneous
features
hazy
borders.
In
paper,
we
proposed
two
methods
detecting
classifying
dermoscopic
images
into
benign
malignant
tumors.
The
first
method
using
k-nearest
neighbor
(KNN)
as
classifier
when
pretrained
deep
neural
networks
are
used
feature
extractors.
second
one
AlexNet
with
grey
wolf
optimizer,
that
optimizes
AlexNet's
hyperparameters
to
get
best
results.
We
also
tested
approaches
cancer
machine
learning
(ML)
(DL).
ML
approach
artificial
network,
KNN,
support
vector
machine,
Naive
Bayes,
decision
tree.
DL
contains
convolutional
network
networks:
AlexNet,
VGG-16,
VGG-19,
EfficientNet-b0,
ResNet-18,
ResNet-50,
ResNet-101,
DenseNet-201,
Inception-v3,
MobileNet-v2.
Our
experiments
trained
on
4000
from
ISIC
archive
dataset.
outcomes
showed
outperformed
other
approaches.
Accuracy
exceeded
99%
some
models
achieved
99%.
CAAI Transactions on Intelligence Technology,
Год журнала:
2023,
Номер
unknown
Опубликована: Авг. 30, 2023
Abstract
In
computer
vision
applications
like
surveillance
and
remote
sensing,
to
mention
a
few,
deep
learning
has
had
considerable
success.
Medical
imaging
still
faces
number
of
difficulties,
including
intra‐class
similarity,
scarcity
training
data,
poor
contrast
skin
lesions,
notably
in
the
case
cancer.
An
optimisation‐aided
learning‐based
system
is
proposed
for
accurate
multi‐class
lesion
identification.
The
sequential
procedures
start
with
preprocessing
end
categorisation.
step
where
hybrid
enhancement
technique
initially
identification
healthy
regions.
Instead
flipping
rotating
outputs
from
middle
phases
enhanced
are
employed
data
augmentation
next
step.
Next,
two
pre‐trained
models,
MobileNetV2
NasNet
Mobile,
trained
using
transfer
on
upgraded
enriched
dataset.
Later,
dual‐threshold
serial
approach
obtain
combine
features
both
models.
was
variance‐controlled
Marine
Predator
methodology,
which
authors
as
superior
optimisation
method.
top
fused
feature
vector
classified
machine
classifiers.
experimental
strategy
provided
accuracy
94.4%
publicly
available
dataset
HAM10000.
Additionally,
framework
evaluated
compared
current
approaches,
remarkable
results.