International Journal of Health Sciences,
Journal Year:
2022,
Volume and Issue:
6(S10), P. 2075 - 2086
Published: Jan. 15, 2022
The
present
review
articles
are
focused
much
on
the
changes
which
have
taken
place
in
imaging
methodologies,
especially
with
regard
to
computed
tomographic
(CT)
relation
early
diagnosis
of
cancer.
background
information
modern
medical
is
provided
article,
starting
naked
eye
inspection
and
its
progressive
into
X
rays,
fluoroscopy,
CT
scans
beyond.
article
gives
basic
working
principles
uses
scan
great
detail
finding
following
up
different
types
advanced
techniques
such
as
high-resolution
tomography
(HRCT),
micro
(μCT)
also
been
covered
paper
where
their
use
studying
bone
structures
other
preclinical
studies
that
involve
high
resolution
has
highlighted.
role
these
management
various
conditions
including
cancer,
cardiovascular
disease,
disorders
nervous
system
examined.
Nonetheless,
risks
scanning
noted
this
review;
particularly,
frequency
exposure
patients
effect
may
after
a
long
period
time.
Cancer Imaging,
Journal Year:
2024,
Volume and Issue:
24(1)
Published: Feb. 29, 2024
Abstract
Background
Prostate-specific
membrane
antigen
(PSMA)
PET/CT
imaging
is
widely
used
for
quantitative
image
analysis,
especially
in
radioligand
therapy
(RLT)
metastatic
castration-resistant
prostate
cancer
(mCRPC).
Unknown
features
influencing
PSMA
biodistribution
can
be
explored
by
analyzing
segmented
organs
at
risk
(OAR)
and
lesions.
Manual
segmentation
time-consuming
labor-intensive,
so
automated
methods
are
desirable.
Training
deep-learning
models
challenging
due
to
the
scarcity
of
high-quality
annotated
images.
Addressing
this,
we
developed
shifted
windows
UNEt
TRansformers
(Swin
UNETR)
fully
segmentation.
Within
a
self-supervised
framework,
model’s
encoder
was
pre-trained
on
unlabeled
data.
The
entire
model
fine-tuned,
including
its
decoder,
using
labeled
Methods
In
this
work,
752
whole-body
[
68
Ga]Ga-PSMA-11
images
were
collected
from
two
centers.
For
pre-training,
652
employed.
remaining
100
manually
supervised
training.
training
phase,
5-fold
cross-validation
with
64
16
validation,
one
center.
testing,
20
hold-out
images,
evenly
distributed
between
centers,
used.
Image
quantification
metrics
evaluated
test
set
compared
ground-truth
conducted
nuclear
medicine
physician.
Results
generates
OARs
lesion
lesion-positive
cases,
mCRPC.
results
show
that
pre-training
significantly
improved
average
dice
similarity
coefficient
(DSC)
all
classes
about
3%.
Compared
nnU-Net,
well-established
medical
segmentation,
our
approach
outperformed
5%
higher
DSC.
This
improvement
attributed
combined
use
fine-tuning,
specifically
when
applied
input.
Our
best
had
lowest
DSC
lesions
0.68
highest
liver
0.95.
Conclusions
We
state-of-the-art
neural
network
followed
fine-tuning
limited
analysis.
generalizable
holds
potential
various
clinical
applications,
enhanced
RLT
patient-specific
internal
dosimetry.
Seminars in Nuclear Medicine,
Journal Year:
2025,
Volume and Issue:
unknown
Published: Feb. 1, 2025
Nuclear
medicine
has
continuously
evolved
since
its
beginnings,
constantly
improving
the
diagnosis
and
treatment
of
various
diseases.
The
integration
artificial
intelligence
(AI)
is
one
latest
revolutionizing
chapters,
promising
significant
advancements
in
diagnosis,
prognosis,
segmentation,
image
quality
enhancement,
theranostics.
Early
AI
applications
nuclear
focused
on
diagnostic
accuracy,
leveraging
machine
learning
algorithms
for
disease
classification
outcome
prediction.
Advances
deep
learning,
including
convolutional
more
recently
transformer-based
neural
networks,
have
further
enabled
precise
segmentation
as
well
low-dose
imaging,
patient-specific
dosimetry
personalized
treatment.
Generative
AI,
driven
by
large
language
models
diffusion
techniques,
now
allowing
process,
interpretation,
generation
complex
medical
images.
Despite
these
achievements,
challenges
such
data
scarcity,
heterogeneity,
ethical
concerns
remain
barriers
to
clinical
translation.
Addressing
issues
through
interdisciplinary
collaboration
will
pave
way
a
broader
adoption
medicine,
potentially
enhancing
patient
care
optimizing
therapeutic
outcomes.
Advances in computational intelligence and robotics book series,
Journal Year:
2025,
Volume and Issue:
unknown, P. 231 - 246
Published: Feb. 28, 2025
The
integration
of
Artificial
Intelligence
(AI)
in
nuclear
medicine
has
revolutionized
personalized
radiopharmaceutical
therapy,
enabling
precise,
patient-centric
approaches
to
treatment.
This
chapter
explores
the
role
AI
optimizing
development
and
application,
focusing
on
its
transformative
impact
therapy
personalization.
It
delves
into
AI-driven
methodologies
for
predicting
biodistribution,
dosimetry,
patient
response,
which
significantly
enhance
effectiveness
safety
therapies.
By
leveraging
machine
learning
algorithms,
this
technology
facilitates
identification
biomarkers,
streamlines
selection
targeted
radiopharmaceuticals,
refines
treatment
planning.
AI's
potential
advancing
theranostics,
combining
diagnostic
imaging
with
therapeutic
applications
improve
disease
targeting
efficacy
is
being
explored.
Ethical
considerations
regulatory
challenges
associated
adoption
field
are
also
discussed.
Research Square (Research Square),
Journal Year:
2025,
Volume and Issue:
unknown
Published: May 16, 2025
Abstract
Following
uveal
melanoma
(UM)
affected
treatment
using
ruthenium-106
brachytherapy,
tumor
thickness
patterns
fall
into
one
of
four
categories:
decrease
(regression),
increase
(recurrence),
stop
(stable),
or
other,
which
are
assessed
in
follow-up
A-mode
and
B-mode
images.
These
critical
indicators
the
tumor’s
response
to
therapy.
This
study
aims
apply
deep
learning
(DL)
models
for
predicting
post-brachytherapy
regression
patterns.
A
cohort
192
patients
participated
this
study.
B-Mode
images
taken
at
time
diagnosis
were
collected,
ophthalmologists
labeled
based
on
results
treatment.
DenseNet121
ResNet34
trained
evaluated
performance
metrics.
achieved
a
macro-average
AUC
0.933,
compared
0.916
ResNet34.
The
per-class
evaluation
showed
that
excelled
all
categories,
providing
superior
predictive
accuracy.
ablation
revealed
best
was
without
pretrained
weights,
dropout
layers
batch
size
32.
Both
demonstrated
strong
classification
capabilities,
with
highest
overall
highlights
potential
DL
UM
undergoing
brachytherapy.
Further
validation
exploration
their
integration
clinical
practice
warranted.
