Diagnostics,
Journal Year:
2023,
Volume and Issue:
13(4), P. 667 - 667
Published: Feb. 10, 2023
Developers
and
users
of
artificial-intelligence-based
tools
for
automatic
contouring
treatment
planning
in
radiotherapy
are
expected
to
assess
clinical
acceptability
these
tools.
However,
what
is
‘clinical
acceptability’?
Quantitative
qualitative
approaches
have
been
used
this
ill-defined
concept,
all
which
advantages
disadvantages
or
limitations.
The
approach
chosen
may
depend
on
the
goal
study
as
well
available
resources.
In
paper,
we
discuss
various
aspects
acceptability’
how
they
can
move
us
toward
a
standard
defining
new
autocontouring
Advances in Radiation Oncology,
Journal Year:
2024,
Volume and Issue:
9(5), P. 101470 - 101470
Published: Feb. 8, 2024
PurposeManual
contour
work
for
radiation
treatment
planning
takes
significant
time
to
ensure
volumes
are
accurately
delineated.
The
use
of
artificial
intelligence
with
deep
learning
based
autosegmentation
(DLAS)
models
has
made
itself
known
in
recent
years
alleviate
this
workload.
It
is
used
organs
at
risk
(OAR)
contouring
consistency
performance
and
saving.
purpose
study
was
evaluate
the
current
published
data
DLAS
clinical
target
volume
(CTV)
contours,
identify
areas
improvement,
discuss
future
directions.MethodologyA
literature
review
performed
by
utilizing
key
words
"Deep
Learning"
AND
("Segmentation"
OR
"Delineation")
"Clinical
Target
Volume"
an
indexed
search
into
PubMed.
A
total
154
articles
on
criteria
were
reviewed.
considered
model
used,
disease
site,
targets
contoured,
guidelines
utilized,
overall
performance.ResultsOf
53
investigating
CTV,
only
6
before
2020.
Publications
have
increased
years,
46
between
2020-2023.
cervix
(n=19)
prostate
(n=12)
studied
most
frequently.
Most
studies
(n=43)
involved
a
single
institution.
Median
sample
size
130
patients
(range:
5-1,052).
common
metrics
utilized
measure
Dice
similarity
coefficient
(DSC)
followed
Hausdorff
distance.
Dosimetric
seldom
reported
(n=11).
There
also
variability
specific
(RTOG,
ESTRO,
others).
had
good
CTV
multiple
sites,
showing
DSC
values
>0.7.
delineated
faster
compared
manual
contouring.
However,
some
contours
still
required
least
minor
edits,
require
improvement.ConclusionsDLAS
demonstrates
capability
completing
plans
efficiency
accuracy.
developed
validated
institutions
using
developing
institutions.
about
years.
Future
need
include
larger
datasets
different
patient
demographics,
stages,
validation
multi-institutional
settings,
inclusion
dosimetric
performance.
Manual
directions.
Of
improvement.
Medical Physics,
Journal Year:
2023,
Volume and Issue:
50(5), P. 2625 - 2636
Published: Feb. 22, 2023
Abstract
Background
Stereotactic
body
radiation
therapy
(SBRT)
of
central
lung
tumors
with
photon
or
proton
has
a
risk
increased
toxicity.
Treatment
planning
studies
comparing
accumulated
doses
for
state‐of‐the‐art
treatment
techniques,
such
as
MR‐guided
radiotherapy
(MRgRT)
and
intensity
modulated
(IMPT),
are
currently
lacking.
Purpose
We
conducted
comparison
MRgRT,
robustly
optimized
non‐adaptive
IMPT,
online
adaptive
IMPT
tumors.
A
special
focus
was
set
on
analyzing
the
to
bronchial
tree,
parameter
linked
high‐grade
toxicities.
Methods
Data
18
early‐stage
tumor
patients,
treated
at
0.35
T
MR‐linac
in
eight
five
fractions,
were
analyzed.
Three
gated
scenarios
compared:
(S1)
(S2)
(S3)
IMPT.
The
plans
recalculated
reoptimized
daily
imaging
data
acquired
during
over
all
fractions.
Accumulated
dose‐volume
histogram
(DVH)
parameters
gross
volume
(GTV),
lung,
heart,
organs‐at‐risk
(OARs)
within
2
cm
target
(PTV)
extracted
each
scenario
compared
Wilcoxon
signed‐rank
tests
between
S1
&
S2,
S3.
Results
GTV
D
98%
above
prescribed
dose
patients
scenarios.
Significant
reductions
(
p
<
0.05)
mean
ipsilateral
(S2:
–8%;
S3:
–23%)
heart
–79%;
–83%)
observed
both
S1.
tree
0.1cc
significantly
lower
S3
(S1:
48.1
Gy;
39.2
=
0.005),
but
not
different
S2
45.0
0.094),
smaller
OARs
1–2
PTV
30.2
S2:
24.6
23.1
Gy),
1
PTV.
Conclusions
significant
sparing
potential
MRgRT
close,
direct
proximity
identified.
near‐maximum
Online
achieved
MRgRT.
Journal of Applied Clinical Medical Physics,
Journal Year:
2023,
Volume and Issue:
24(7)
Published: March 15, 2023
Recently,
target
auto-segmentation
techniques
based
on
deep
learning
(DL)
have
shown
promising
results.
However,
inaccurate
delineation
will
directly
affect
the
treatment
planning
dose
distribution
and
effect
of
subsequent
radiotherapy
work.
Evaluation
geometric
metrics
alone
may
not
be
sufficient
for
accuracy
assessment.
The
purpose
this
paper
is
to
validate
performance
automatic
segmentation
with
dosimetric
try
construct
new
evaluation
comprehensively
understand
dose-response
relationship
from
perspective
clinical
application.A
DL-based
model
was
developed
by
using
186
manual
modified
radical
mastectomy
breast
cancer
cases.
resulting
DL
were
used
generate
alternative
contours
in
a
set
48
patients.
Auto-plan
reoptimized
ensure
same
optimized
parameters
as
reference
Manual-plan.
To
assess
impact
auto-segmentation,
only
common
but
also
spatial
distance
relative
volume
(
RV${R}_V$
)
used.
Correlations
performed
Spearman's
correlation
between
changes.Only
strong
(|R2
|
>
0.6,
p
<
0.01)
or
moderate
0.4,
Pearson
established
traditional
metric
three
indices
(conformity
index,
homogeneity
mean
dose).
For
organs
at
risk
(OARs),
inferior
no
significant
found
differences.
Furthermore,
we
that
OARs
affected
boundary
error
instead
target.Current
could
reflect
certain
degree
variation.
find
contour
variations
do
lead
dosimetry
changes,
clinically
oriented
more
accurately
how
quality
affects
should
constructed.
Physics and Imaging in Radiation Oncology,
Journal Year:
2023,
Volume and Issue:
27, P. 100454 - 100454
Published: June 6, 2023
Normal
tissue
sparing
in
radiotherapy
relies
on
proper
delineation.
While
manual
contouring
is
time
consuming
and
subject
to
inter-observer
variability,
auto-contouring
could
optimize
workflows
harmonize
practice.
We
assessed
the
accuracy
of
a
commercial,
deep-learning,
MRI-based
tool
for
brain
organs-at-risk
delineation.Thirty
adult
tumor
patients
were
retrospectively
manually
recontoured.
Two
additional
structure
sets
obtained:
AI
(artificial
intelligence)
AIedit
(manually
corrected
auto-contours).
For
15
selected
cases,
identical
plans
optimized
each
set.
used
Dice
Similarity
Coefficient
(DSC)
mean
surface-distance
(MSD)
geometric
comparison
gamma
analysis
dose-volume-histogram
dose
metrics
evaluation.
Wilcoxon
signed-ranks
test
was
paired
data,
Spearman
coefficient(ρ)
correlations
Bland-Altman
plots
assess
level
agreement.Auto-contouring
significantly
faster
than
(1.1/20
min,
p
<
0.01).
Median
DSC
MSD
0.7/0.9
mm
0.8/0.5
AIedit.
correlated
with
size
(ρ
=
0.76,
0.01),
higher
large
structures.
pass
rate
74%
(71-81%)
Plan_AI
82%
(75-86%)
Plan_AIedit,
no
correlation
or
MSD.
Differences
between
Dmean_AI
Dmean_Ref
≤
0.2
Gy
(p
0.05).
The
difference
moderately
DSC.
Bland
Altman
plot
showed
minimal
discrepancy
(0.1/0)
reference
Dmean/Dmax.The
AI-model
good
structures,
but
developments
are
required
smaller
ones.
Auto-segmentation
faster,
minor
differences
distribution
caused
by
variations.
