Computational and Mathematical Methods in Medicine,
Год журнала:
2021,
Номер
2021, С. 1 - 18
Опубликована: Март 8, 2021
Aim.
COVID-19
has
caused
large
death
tolls
all
over
the
world.
Accurate
diagnosis
is
of
significant
importance
for
early
treatment.
Methods.
In
this
study,
we
proposed
a
novel
PSSPNN
model
classification
between
COVID-19,
secondary
pulmonary
tuberculosis,
community-captured
pneumonia,
and
healthy
subjects.
entails
five
improvements:
first
n-conv
stochastic
pooling
module.
Second,
neural
network
was
proposed.
Third,
PatchShuffle
introduced
as
regularization
term.
Fourth,
an
improved
multiple-way
data
augmentation
used.
Fifth,
Grad-CAM
utilized
to
interpret
our
AI
model.
Results.
The
10
runs
with
random
seed
on
test
set
showed
algorithm
achieved
microaveraged
F1
score
95.79%.
Moreover,
method
better
than
nine
state-of-the-art
approaches.
Conclusion.
This
will
help
assist
radiologists
make
more
quickly
accurately
cases.
Sensors,
Год журнала:
2023,
Номер
23(2), С. 634 - 634
Опубликована: Янв. 5, 2023
Artificial
intelligence
(AI)
with
deep
learning
models
has
been
widely
applied
in
numerous
domains,
including
medical
imaging
and
healthcare
tasks.
In
the
field,
any
judgment
or
decision
is
fraught
risk.
A
doctor
will
carefully
judge
whether
a
patient
sick
before
forming
reasonable
explanation
based
on
patient's
symptoms
and/or
an
examination.
Therefore,
to
be
viable
accepted
tool,
AI
needs
mimic
human
interpretation
skills.
Specifically,
explainable
(XAI)
aims
explain
information
behind
black-box
model
of
that
reveals
how
decisions
are
made.
This
paper
provides
survey
most
recent
XAI
techniques
used
related
applications.
We
summarize
categorize
types,
highlight
algorithms
increase
interpretability
topics.
addition,
we
focus
challenging
problems
applications
provide
guidelines
develop
better
interpretations
using
concepts
image
text
analysis.
Furthermore,
this
future
directions
guide
developers
researchers
for
prospective
investigations
clinical
topics,
particularly
imaging.
International Journal of Cognitive Computing in Engineering,
Год журнала:
2021,
Номер
2, С. 57 - 64
Опубликована: Фев. 23, 2021
As
one
of
the
most
important
directions
in
field
computer
vision,
facial
emotion
recognition
plays
an
role
people's
daily
work
and
life.
Human
based
on
expressions
is
great
significance
application
intelligent
human-computer
interaction.
However,
current
research
recognition,
there
are
some
problems
such
as
poor
generalization
ability
network
model
low
robustness
system.
In
this
content,
we
propose
a
method
feature
extraction
using
deep
residual
ResNet-50,
which
combines
convolutional
neural
for
recognition.
Through
experimental
simulation
specified
data
set,
it
can
be
proved
that
superior
to
mainstream
models
performance
detection.
Sensors,
Год журнала:
2021,
Номер
21(14), С. 4758 - 4758
Опубликована: Июль 12, 2021
With
the
advances
of
data-driven
machine
learning
research,
a
wide
variety
prediction
problems
have
been
tackled.
It
has
become
critical
to
explore
how
and
specifically
deep
methods
can
be
exploited
analyse
healthcare
data.
A
major
limitation
existing
focus
on
grid-like
data;
however,
structure
physiological
recordings
are
often
irregular
unordered,
which
makes
it
difficult
conceptualise
them
as
matrix.
As
such,
graph
neural
networks
attracted
significant
attention
by
exploiting
implicit
information
that
resides
in
biological
system,
with
interacting
nodes
connected
edges
whose
weights
determined
either
temporal
associations
or
anatomical
junctions.
In
this
survey,
we
thoroughly
review
different
types
architectures
their
applications
healthcare.
We
provide
an
overview
these
systematic
manner,
organized
domain
application
including
functional
connectivity,
structure,
electrical-based
analysis.
also
outline
limitations
techniques
discuss
potential
directions
for
future
research.
International Journal of Intelligent Systems,
Год журнала:
2021,
Номер
37(2), С. 1572 - 1598
Опубликована: Сен. 21, 2021
COVID-19
pneumonia
started
in
December
2019
and
caused
large
casualties
huge
economic
losses.
In
this
study,
we
intended
to
develop
a
computer-aided
diagnosis
system
based
on
artificial
intelligence
automatically
identify
the
chest
computed
tomography
images.
We
utilized
transfer
learning
obtain
image-level
representation
(ILR)
backbone
deep
convolutional
neural
network.
Then,
novel
neighboring
aware
(NAR)
was
proposed
exploit
relationships
between
ILR
vectors.
To
information
feature
space
of
ILRs,
an
graph
generated
k-nearest
neighbors
algorithm,
which
ILRs
were
linked
with
their
ILRs.
Afterward,
NARs
by
fusion
graph.
On
basis
representation,
end-to-end
classification
architecture
called
network
(NAGNN)
proposed.
The
private
public
data
sets
used
for
evaluation
experiments.
Results
revealed
that
our
NAGNN
outperformed
all
10
state-of-the-art
methods
terms
generalization
ability.
Therefore,
is
effective
detecting
COVID-19,
can
be
clinical
diagnosis.
Alexandria Engineering Journal,
Год журнала:
2021,
Номер
61(2), С. 1319 - 1334
Опубликована: Июнь 19, 2021
The
problem
of
respiratory
sound
classification
has
received
good
attention
from
the
clinical
scientists
and
medical
researcher's
community
in
last
year
to
diagnosis
COVID-19
disease.
Artificial
Intelligence
(AI)
based
models
deployed
into
real-world
identify
disease
human-generated
sounds
such
as
voice/speech,
dry
cough,
breath.
CNN
(Convolutional
Neural
Network)
is
used
solve
many
problems
with
machines.
We
have
proposed
implemented
a
multi-channeled
Deep
Convolutional
Network
(DCNN)
for
automatic
human
like
voice,
breath,
it
will
give
better
accuracy
performance
than
previous
models.
applied
multi-feature
channels
data
De-noising
Auto
Encoder
(DAE)
technique,
GFCC
(Gamma-tone
Frequency
Cepstral
Coefficients),
IMFCC
(Improved
Multi-frequency
Coefficients)
methods
on
augmented
extract
deep
features
input
CNN.
approach
improves
system
provides
results
dataset.
