Though
artificial
intelligence
(AI)
has
been
used
in
nuclear
medicine
for
more
than
50
years,
progress
made
deep
learning
(DL)
and
machine
(ML),
which
have
driven
the
development
of
new
AI
abilities
field.
ANNs
are
both
medicine.
Alternatively,
if
3D
convolutional
neural
network
(CNN)
is
used,
inputs
may
be
actual
images
that
being
analyzed,
rather
a
set
inputs.
In
medicine,
reimagines
reengineers
field's
therapeutic
scientific
capabilities.
Understanding
concepts
CNN
U-Net
context
provides
deeper
engagement
with
clinical
research
applications,
as
well
ability
to
troubleshoot
problems
when
they
emerge.
Business
analytics,
risk
assessment,
quality
assurance,
basic
classifications
all
examples
simple
ML
applications.
General
SPECT,
PET,
MRI,
CT
benefit
from
advanced
DL
applications
classification,
detection,
localization,
segmentation,
quantification,
radiomic
feature
extraction
utilizing
CNNs.
An
ANN
analyze
small
dataset
at
same
time
traditional
statistical
methods,
bigger
datasets.
Nuclear
medicine's
practices
largely
unaffected
by
introduction
(AI).
Clinical
landscapes
fundamentally
altered
advent
professionals
must
now
least
an
elementary
understanding
principles
such
networks
(ANNs)
(CNNs).
Physica Medica,
Год журнала:
2021,
Номер
83, С. 122 - 137
Опубликована: Март 1, 2021
This
review
sets
out
to
discuss
the
foremost
applications
of
artificial
intelligence
(AI),
particularly
deep
learning
(DL)
algorithms,
in
single-photon
emission
computed
tomography
(SPECT)
and
positron
(PET)
imaging.
To
this
end,
underlying
limitations/challenges
these
imaging
modalities
are
briefly
discussed
followed
by
a
description
AI-based
solutions
proposed
address
challenges.
will
focus
on
mainstream
generic
fields,
including
instrumentation,
image
acquisition/formation,
reconstruction
low-dose/fast
scanning,
quantitative
imaging,
interpretation
(computer-aided
detection/diagnosis/prognosis),
as
well
internal
radiation
dosimetry.
A
brief
algorithms
fundamental
architectures
used
for
is
also
provided.
Finally,
challenges,
opportunities,
barriers
full-scale
validation
adoption
improvement
quality
accuracy
PET
SPECT
images
clinic
discussed.
European Journal of Nuclear Medicine and Molecular Imaging,
Год журнала:
2021,
Номер
48(8), С. 2405 - 2415
Опубликована: Янв. 25, 2021
Tendency
is
to
moderate
the
injected
activity
and/or
reduce
acquisition
time
in
PET
examinations
minimize
potential
radiation
hazards
and
increase
patient
comfort.
This
work
aims
assess
performance
of
regular
full-dose
(FD)
synthesis
from
fast/low-dose
(LD)
whole-body
(WB)
images
using
deep
learning
techniques.
Diagnostics,
Год журнала:
2025,
Номер
15(3), С. 397 - 397
Опубликована: Фев. 6, 2025
Radioimmunotherapy
(RIT)
is
a
novel
cancer
treatment
that
combines
radiotherapy
and
immunotherapy
to
precisely
target
tumor
antigens
using
monoclonal
antibodies
conjugated
with
radioactive
isotopes.
This
approach
offers
personalized,
systemic,
durable
treatment,
making
it
effective
in
cancers
resistant
conventional
therapies.
Advances
artificial
intelligence
(AI)
present
opportunities
enhance
RIT
by
improving
precision,
efficiency,
personalization.
AI
plays
critical
role
patient
selection,
planning,
dosimetry,
response
assessment,
while
also
contributing
drug
design
classification.
review
explores
the
integration
of
into
RIT,
emphasizing
its
potential
optimize
entire
process
advance
personalized
care.
Diagnostics,
Год журнала:
2024,
Номер
14(2), С. 181 - 181
Опубликована: Янв. 14, 2024
Radiotheranostics
refers
to
the
pairing
of
radioactive
imaging
biomarkers
with
therapeutic
compounds
that
deliver
ionizing
radiation.
Given
introduction
very
promising
radiopharmaceuticals,
radiotheranostics
approach
is
creating
a
novel
paradigm
in
personalized,
targeted
radionuclide
therapies
(TRTs),
also
known
as
radiopharmaceuticals
(RPTs).
Radiotherapeutic
pairs
targeting
somatostatin
receptors
(SSTR)
and
prostate-specific
membrane
antigens
(PSMA)
are
increasingly
being
used
diagnose
treat
patients
metastatic
neuroendocrine
tumors
(NETs)
prostate
cancer.
In
parallel,
radiomics
artificial
intelligence
(AI),
important
areas
quantitative
image
analysis,
paving
way
for
significantly
enhanced
workflows
diagnostic
theranostic
fields,
from
data
processing
clinical
decision
support,
improving
patient
selection,
personalized
treatment
strategies,
response
prediction,
prognostication.
Furthermore,
AI
has
potential
tremendous
effectiveness
dosimetry
which
copes
complex
time-consuming
tasks
RPT
workflow.
The
present
work
provides
comprehensive
overview
application
radiotheranostics,
focusing
on
SSTR-
or
PSMA-targeting
radioligands,
describing
fundamental
concepts
specific
imaging/treatment
features.
Our
review
includes
ligands
radiolabeled
by
68Ga,
18F,
177Lu,
64Cu,
90Y,
225Ac.
Specifically,
contributions
via
towards
improved
acquisition,
reconstruction,
response,
segmentation,
restaging,
lesion
classification,
dose
estimation
well
ongoing
developments
future
directions
discussed.
European Journal of Nuclear Medicine and Molecular Imaging,
Год журнала:
2024,
Номер
51(6), С. 1516 - 1529
Опубликована: Янв. 25, 2024
Abstract
Purpose
Accurate
dosimetry
is
critical
for
ensuring
the
safety
and
efficacy
of
radiopharmaceutical
therapies.
In
current
clinical
practice,
MIRD
formalisms
are
widely
employed.
However,
with
rapid
advancement
deep
learning
(DL)
algorithms,
there
has
been
an
increasing
interest
in
leveraging
calculation
speed
automation
capabilities
different
tasks.
We
aimed
to
develop
a
hybrid
transformer-based
model
that
incorporates
multiple
voxel
S
-value
(MSV)
approach
voxel-level
[
177
Lu]Lu-DOTATATE
therapy.
The
goal
was
enhance
performance
achieve
accuracy
levels
closely
aligned
Monte
Carlo
(MC)
simulations,
considered
as
standard
reference.
extended
our
analysis
include
(SSV
MSV),
thereby
conducting
comprehensive
study.
Methods
used
dataset
consisting
22
patients
undergoing
up
4
cycles
MC
simulations
were
generate
reference
absorbed
dose
maps.
addition,
formalism
approaches,
namely,
single
(SSV)
MSV
techniques,
performed.
A
UNEt
TRansformer
(UNETR)
DL
architecture
trained
using
five-fold
cross-validation
MC-based
Co-registered
CT
images
fed
into
network
input,
whereas
difference
between
(MC-MSV)
set
output.
results
then
integrated
revive
Finally,
maps
generated
by
MSV,
SSV,
quantitatively
compared
at
both
level
organ
(organs
risk
lesions).
Results
showed
slightly
better
(voxel
relative
absolute
error
(RAE)
=
5.28
±
1.32)
RAE
5.54
1.4)
outperformed
SSV
7.8
3.02).
Gamma
pass
rates
99.0
1.2%,
98.8
1.3%,
98.7
1.52%
DL,
respectively.
computational
time
highest
(~2
days
single-bed
SPECT
study)
DL-based
other
approaches
terms
efficiency
(3
s
SPECT).
Organ-wise
percent
errors
1.44
3.05%,
1.18
2.65%,
1.15
2.5%
respectively,
lesion-absorbed
doses.
Conclusion
developed
fast
accurate
map
generation,
outperforming
specifically
heterogenous
regions.
achieved
close
gold
potential
implementation
use
on
large-scale
datasets.
Clinical Nuclear Medicine,
Год журнала:
2021,
Номер
46(11), С. 872 - 883
Опубликована: Июль 8, 2021
Purpose
The
availability
of
automated,
accurate,
and
robust
gross
tumor
volume
(GTV)
segmentation
algorithms
is
critical
for
the
management
head
neck
cancer
(HNC)
patients.
In
this
work,
we
evaluated
3
state-of-the-art
deep
learning
combined
with
8
different
loss
functions
PET
image
using
a
comprehensive
training
set
its
performance
on
an
external
validation
HNC
Patients
Methods
18
F-FDG
PET/CT
images
470
patients
presenting
which
manually
defined
GTVs
serving
as
standard
reference
were
used
(340
patients),
evaluation
(30
testing
(100
from
centers)
these
algorithms.
intensity
was
converted
to
SUVs
normalized
in
range
(0–1)
SUV
max
whole
data
set.
cropped
12
×
cm
subvolumes
isotropic
voxel
spacing
mm
containing
neighboring
background
including
lymph
nodes.
We
approaches
augmentation,
rotation
(−15
degrees,
+15
degrees),
scaling
(−20%,
20%),
random
flipping
(3
axes),
elastic
deformation
(sigma
=
1
proportion
deform
0.7)
increase
number
sets.
Three
networks,
Dense-VNet,
NN-UNet,
Res-Net,
functions,
Dice,
generalized
Wasserstein
Dice
loss,
plus
XEnt
cross-entropy,
sensitivity-specificity,
Tversky,
used.
Overall,
28
networks
built.
Standard
metrics,
similarity,
image-derived
first-order,
shape
radiomic
features,
assessment
Results
best
results
terms
coefficient
(mean
±
SD)
achieved
by
cross-entropy
Res-Net
(0.86
0.05;
95%
confidence
interval
[CI],
0.85–0.87),
Dense-VNet
(0.85
0.058;
CI,
0.84–0.86),
NN-UNet
(0.87
0.86–0.88).
difference
between
not
statistically
significant
(
P
>
0.05).
percent
relative
error
(RE%)
quantification
less
than
5%
more
0.84,
whereas
lower
RE%
(0.41%)
loss.
For
maximum
3-dimensional
diameter
sphericity
all
RE
≤
≤10%,
respectively,
reflecting
small
variability.
Conclusions
Deep
exhibited
promising
automated
GTV
delineation
images.
Different
performed
competitively
when
emerged
reliable
delineation.
Caution
should
be
exercised
clinical
deployment
owing
occurrence
outliers
learning–based
Quantitative Imaging in Medicine and Surgery,
Год журнала:
2021,
Номер
11(6), С. 2792 - 2822
Опубликована: Апрель 12, 2021
Abstract:
Recently,
the
application
of
artificial
intelligence
(AI)
in
medical
imaging
(including
nuclear
medicine
imaging)
has
rapidly
developed.
Most
AI
applications
have
focused
on
diagnosis,
treatment
monitoring,
and
correlation
analyses
with
pathology
or
specific
gene
mutation.
It
can
also
be
used
for
image
generation
to
shorten
time
acquisition,
reduce
dose
injected
tracer,
enhance
quality.
This
work
provides
an
overview
single-photon
emission
computed
tomography
(SPECT)
positron
(PET)
either
without
anatomical
information
[CT
magnetic
resonance
(MRI)].
review
four
aspects,
including
physics,
reconstruction,
postprocessing,
internal
dosimetry.
generating
attenuation
map,
estimating
scatter
events,
boosting
quality,
predicting
map
is
summarized
discussed.
