2022 44th Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC),
Год журнала:
2023,
Номер
unknown, С. 1 - 5
Опубликована: Июль 24, 2023
Non-linear
least
square
minimization
algorithms
are
often
employed
to
solve
Diffuse
Optical
Tomography
(DOT)
inverse
problem.
However,
it
is
time-consuming
calculate
the
Jacobian
matrix.
This
work
has
proposed
a
data-driven
neural
network
method
improve
computational
efficiency.
The
singular
value
decomposition
compute
updated
and
mapping
from
boundary
measurements
values
based
on
convolutional
(CNN)
learned
obtain
values.
validated
with
3D
numerical
simulation
data.
We
have
demonstrated
that
approach
can
save
computation
time
compared
Adjoint
method,
reconstructed
absorption
coefficient
close
method.Clinical
Relevance—
These
results
not
focused
clinical
relevance
currently,
but
in
future
may
be
helpful
accelerant
DOT
reconstruction
clinic.
Applied Physics B,
Год журнала:
2024,
Номер
130(9)
Опубликована: Авг. 29, 2024
Computational
methods
have
been
established
as
cornerstones
in
optical
imaging
and
holography
recent
years.
Every
year,
the
dependence
of
on
computational
is
increasing
significantly
to
extent
that
components
are
being
completely
efficiently
replaced
with
at
low
cost.
This
roadmap
reviews
current
scenario
four
major
areas
namely
incoherent
digital
holography,
quantitative
phase
imaging,
through
scattering
layers,
super-resolution
imaging.
In
addition
registering
perspectives
modern-day
architects
above
research
areas,
also
reports
some
latest
studies
topic.
codes
pseudocodes
presented
for
a
plug-and-play
fashion
readers
not
only
read
understand
but
practice
algorithms
their
data.
We
believe
this
will
be
valuable
tool
analyzing
trends
predict
prepare
future
holography.
Journal of Lightwave Technology,
Год журнала:
2024,
Номер
42(10), С. 3631 - 3641
Опубликована: Фев. 5, 2024
Interest
in
optical
wireless
communications
(OWC)
as
a
possible
complement
to
RF
technology
has
increased
recently.
However,
the
propagation
of
beams
through
atmosphere
distorts
beam's
amplitude
and
phase,
resulting
information
loss
significant
noise.
Arising
due
combination
multiple
absorption
scattering
events,
distortion
beam
makes
communication
difficult.
In
some
cases,
orbital
angular
momentum
light
(OAM),
along
with
various
deep
learning
algorithms
(DL),
could
be
helpful
mitigate
problem
provide
high-capacity
links.
this
work,
we
propagate
Laguerre-Gaussian
(LG)
different
topological
charges
(l)
under
diffuse
turbulent
conditions
develop
deep-learning
classification
network
characterize
'l'
LG
beam.
The
proposed
method
is
later
implemented
using
laboratory
setup
demonstrating
on
table.
results
show
that
algorithm
can
identify
modes
high
accuracy
even
when
propagates
highly
media.
To
demonstrate
robustness
OWC
system,
small
grayscale
images
are
transmitted
over
channel.
A
bit
error
rate
(BER)
only
2.3
×
10
-4
9.7
for
tabletop
experiment
simulations,
respectively.
demonstrated
low
BER
system
suggests
promising
applications
secure
reliable
data
transmission
adverse
atmospheric
conditions,
highlighting
potential
advancing
technologies.
Bioengineering,
Год журнала:
2023,
Номер
10(3), С. 382 - 382
Опубликована: Март 21, 2023
Diffuse
optical
tomography
(DOT)
is
a
non-invasive
method
for
detecting
breast
cancer;
however,
it
struggles
to
produce
high-quality
images
due
the
complexity
of
scattered
light
and
limitations
traditional
image
reconstruction
algorithms.
These
algorithms
can
be
affected
by
boundary
conditions
have
low
imaging
accuracy,
shallow
depth,
long
computation
time,
high
signal-to-noise
ratio.
However,
machine
learning
potentially
improve
performance
DOT
being
better
equipped
solve
inverse
problems,
perform
regression,
classify
medical
images,
reconstruct
biomedical
images.
In
this
study,
we
utilized
model
called
"XGBoost"
detect
tumors
in
inhomogeneous
breasts
applied
post-processing
technique
based
on
genetic
programming
accuracy.
The
proposed
algorithm
was
tested
using
simulated
measurements
from
complex
evaluated
cosine
similarity
metrics
root
mean
square
error
loss.
results
showed
that
use
XGBoost
could
lead
more
accurate
detection
compared
methods,
with
reconstructed
having
an
average
than
0.97
±
0.07
around
0.1270
0.0031
ground
truth.
Journal of Biomedical Optics,
Год журнала:
2023,
Номер
28(02)
Опубликована: Фев. 6, 2023
SignificanceThe
machine
learning
(ML)
approach
plays
a
critical
role
in
assessing
biomedical
imaging
processes
especially
optical
(OI)
including
segmentation,
classification,
and
reconstruction,
intending
to
achieve
higher
accuracy
efficiently.AimThis
research
aims
develop
an
end-to-end
deep
framework
for
diffuse
(DOI)
with
multiple
datasets
detect
breast
cancer
reconstruct
its
properties
the
early
stages.ApproachThe
proposed
Periodic-net
is
nondestructive
(DL)
algorithm
reconstruction
evaluation
of
inhomogeneities
inverse
model
high
accuracy,
while
boundary
measurements
are
calculated
by
solving
forward
problem
sources/detectors
arranged
uniformly
around
circular
domain
various
combinations,
16
×
15,
20
19,
36
35
measurement
setups.ResultsThe
results
image
on
numerical
phantom
demonstrate
that
network
provides
higher-quality
images
greater
amount
small
details,
superior
immunity
noise,
sharper
edges
reduction
artifacts
than
other
state-of-the-art
competitors.ConclusionsThe
highly
effective
at
simultaneous
properties,
i.e.,
absorption
reduced
scattering
coefficients,
optimizing
time
without
degrading
inclusions
localization
quality.
Applied Intelligence,
Год журнала:
2022,
Номер
53(13), С. 16519 - 16539
Опубликована: Дек. 10, 2022
Abstract
Noninvasive
assessment
of
skin
structure
using
hyperspectral
images
has
been
intensively
studied
in
recent
years.
Due
to
the
high
computational
cost
classical
methods,
such
as
inverse
Monte
Carlo
(IMC),
much
research
done
with
aim
machine
learning
(ML)
methods
reduce
time
required
for
estimating
parameters.
This
study
aims
evaluate
accuracy
and
estimation
speed
ML
this
purpose
compare
them
traditionally
used
adding-doubling
(IAD)
algorithm.
We
trained
three
models
–
an
artificial
neural
network
(ANN),
a
1D
convolutional
(CNN),
random
forests
(RF)
model
predict
seven
The
were
on
simulated
data
computed
To
improve
predictive
performance,
we
introduced
stacked
dynamic
weighting
(SDW)
combining
predictions
all
individually
models.
SDW
was
by
only
handful
real-world
spectra
top
ANN,
CNN
RF
that
data.
Models
evaluated
based
estimated
parameters’
mean
absolute
error
(MAE),
considering
surface
inclination
angle
comparing
fitted
IAD
On
data,
lowest
MAE
achieved
(0.0030),
while
vivo
measured
(0.0113).
shortest
estimate
parameters
single
spectrum
93.05
μ
s.
Results
suggest
algorithms
can
produce
accurate
estimates
human
optical
near
real-time.
Biomedical Optics Express,
Год журнала:
2023,
Номер
14(4), С. 1818 - 1818
Опубликована: Март 23, 2023
Double
integrating
sphere
measurements
obtained
from
thin
ex
vivo
tissues
provides
more
spectral
information
and
hence
allows
full
estimation
of
all
basic
optical
properties
(OPs)
theoretically.
However,
the
ill-conditioned
nature
OP
determination
increases
excessively
with
reduction
in
tissue
thickness.
Therefore,
it
is
crucial
to
develop
a
model
for
that
robust
noise.
Herein,
we
present
deep
learning
solution
precisely
extract
four
OPs
real-time
tissues,
leveraging
dedicated
cascade
forward
neural
network
(CFNN)
each
an
additional
introduced
input
refractive
index
cuvette
holder.
The
results
show
CFNN-based
enables
accurate
fast
evaluation
OPs,
as
well
robustness
Our
proposed
method
overcomes
highly
restriction
can
distinguish
effects
slight
changes
measurable
quantities
without
any
priori
knowledge.