1

Statements
Purpose Statement: The purpose of this study is to determine if scripting can improve treatment
planning efficiency for multiple brain lesion SRS VMAT cases by reducing planning time
commitment.
Problem Statement: The problem is that multiple brain lesion SRS VMAT cases require an
extensive time commitment to create optimization structures and add multiple objectives to the
optimizer, leading to decreased treatment planning efficiency.
Hypotheses: Researchers tested the hypotheses that using Eclipse Scripting Application
Programming Interface (ESAPI) will improve treatment planning efficiency for multiple brain
lesion SRS cases by (H1) reducing the total contouring time, (H2) reducing the optimizer
preparation time, (H3) reducing the total number of optimizations, and (H4) reducing the overall
planning time. Researchers tested the null hypotheses that using Eclipse Scripting Application
Programming Interface (ESAPI) will not improve treatment planning efficiency for multiple
brain lesion SRS cases by (H10) reducing the total contouring time, (H20) reducing the
optimizer preparation, (H30) reducing the total number of optimizations, and (H40) reducing the
overall planning time.

H1: The first research hypothesis (H1) is that using scripting will improve planning efficiency by
reducing the total contouring time for multiple brain lesion SRS cases.
H10: The first null hypothesis (H10) is that using scripting will not improve planning efficiency
by reducing the total contouring time for multiple brain lesion SRS cases.
H2: The second research hypothesis (H2) is that using scripting will improve planning efficiency
by reducing optimizer preparation time for multiple brain lesion SRS cases.
H20: The second null hypothesis (H20) is that using scripting will not improve planning
efficiency by reducing optimizer preparation time for multiple brain lesion SRS cases.
H3: The third research hypothesis (H3) is that using scripting will improve planning efficiency
by reducing the total number of optimizations for multiple brain lesion SRS cases.
H30: The third null hypothesis (H30) is that using scripting will not improve planning efficiency
by reducing the total number of optimizations for multiple brain lesion SRS cases.
H4: The fourth research hypothesis (H4) is that using scripting will improve planning efficiency
by reducing overall planning time for multiple brain lesion SRS cases.
H40: The fourth null hypothesis (H40) is that using scripting will not improve planning
efficiency by reducing overall planning time for multiple brain lesion SRS cases.
Change Matrix
Title of Capstone: Improved planning efficiency in multi-lesion stereotactic radiosurgery
(SRS) and radiotherapy (SRT) volumetric modulated arc therapy (VMAT) cases using
 2

Eclipse scripting

Group: 8
Reviewer’s recommendation How addressed Page numbers
where change
appears
A1. The entire document should be 1.5cm Removed extra space under header Pg. 1
line spacing. No extra spacing under
headers.
A2. Citation should be here after the et al; Added citation and removed comma Pg. 1
remove comma here after I rearranged

A3. I might say users or departments since Changed “clinics” to “departments” Pg. 1
you use clinics in next sentence

A4. I think you'll need to add a sentence or Added two sentences about what scripting Pg. 1
two here about what scripting is. is

A5. Maybe require technical skills? Changed sentence wording from “for tasks Pg. 1
Maybe we should just say ... they are that do not necessarily put their skills to
required to dedicate time and effort for non- work” to “to less advanced tasks”
technical tasks. ?? I'm not sure if non-
technical is the right word. What they are
doing is technical, but repetitive,
monotonous, or perhaps less advanced.
IDK.
A6. Again, but maybe menial? Changed “non-technical” to “ iterative” Pg. 1
No, I think they got it right when the put
monotonous and time-consuming at the end
of this paragraph. I think repetitive or
monotonous is what we are trying to
emphasize. Thoughts?
A7. add citation here after et al Added citation Pg. 2

A8. optimization structures is plural. So Changed “optimization structures are a Pg. 2

you need to put are useful tools... or say useful tool” to “creating optimization
'creating optimization structures is a useful structures is a useful tool”
tool..'

A9. wordy. Perhaps improve dose Changed “deliver improved dose delivery” Pg. 2
delivery to “improve dose delivery”

A10. Nishele, do these need a reference to Added “Varian” before “HyperArc and Pg. 1
Varian in parentheses or anything? RapidPlan”
Yes - I would put Varian before these
B1. Recognized as AMA abbreviation Changed quality assurance to QA Pg. 1
B2. This 2 statements contradict each other. Reworded to remove contradiction Pg. 2
You want a hot center but not in the GTV?
I would reword.
 3

B3. I don’t understand what you’re trying
to say here.
B4. This is an assumption. Have you met
all medical dosimetrists? Modify.
B5. Consider how you will incorporate
figures and tables into this section for your
next outline/drafts.
B6. I would break this section up into more
sections: Patient Selection and Simulation,
Contouring and Target Volumes, Scripting
Development
B7. This entire section needs to be
reworked. You should only focus on the
steps necessary to generate the script. It
seems there is data about the patient
population used in the study, contour
generation etc. but I’m understanding that
this was not related to how the script was
developed. Only focus on that here. You
may need another section talking about
these factors.
B8. I am confused by this statement. Given
that this is listed under scripting
development, were these patients used to
develop the script? Or were they the
patients used to test your dosimetry
participants? Clarify here.
B9. Who completed these?
B10. Spell out at first use
B11. I would move this toward beginning
of paragraph
B12. Reword to scholarly sentence
Reworded for clarity on margin between Pg. 2
GTV and PTV
Modified sentence to indicate a possibility Pg. 2
instead of an assumption
Added figures into outline and draft Pg. 2
Broke it up into sections Pg. 2
Reworked sections. Now the first major Pg. 2
section is about scripting
development/function and the second
section is about testing the script
Reworded sentence to indicate that the Pg. 2
patients were selected to test the
usefulness of the script during the
planning process
Explained that the CMD who wrote the Pg. 3
script also created the contours
NRG is not an acronym, so all Pg. 3
organizations that make up the combined
organization were included
Moved the sentence toward the beginning Pg. 3
of the paragraph
Reworded sentence for clarity Pg. 3

B13. aren't we already in the process at this Reworded sentence for clarity Pg. 3
point? I would clarify this a bit for those of
us who have not used scripting
B14. Assign? Changed to “assign” Pg. 3

B15. To fill? Added “to fill” Pg. 3

B16. I'm not sure this is necessary - or Omitted unnecessary information and Pg. 4
maybe just say all treatment plans were potential identifier
completed by board certified medical
dosimetrists and include in following
paragraph. What do you think
Ashley/Nishele?
Yes, this could be considered an identifier
so remove.
B17. This seems oddly out of place since Omitted Zoom reference here Pg. 4
this is not considered a planning step. I
would bring up zoom later.
 4

B18. This is where we need to learn about Included brief information about cases Pg. 4
the patients selected for the study. selected
B19. Go back to your purpose statement. Deleted information that is not relevant to Pg. 4
You are not considering anything related to the purpose statement (treatment
a patient treatment. Only focus on content machines, algorithm, etc)
related to that purpose statement.
B20. I would omit this since it has more to Omitted two sentences Pg. 4
do with delivery than planning
B 21. I would add this into second sentence Added information into second sentence Pg. 4
with version and grid
B 22. I might say "set and adjust clinical Changed wording to suggested phrase Pg. 4
goal priorities during the optimization
process"
B 23. What was the purpose of this? Added a sentence explaining why there Pg. 4
Explain to the reader was a 7+ day period between plan
attempts
B 24. Not scholarly writing. reword Changed “completed its function” to “was Pg. 4
complete”
B 25. Each patient? Or each participant? Removed “each patient” Pg. 4

AMA Referencing Quick Guide Checklist

Task Submis Submissio Submissio Submissio Submissio Submissio Submissio
sion n Date: n Date: n Date: n Date: n Date: n Date:
Date: 7/12/22 8/2/22
6/21/22
Manuscript
Written in past tense? ☒ ☒ ☒ ☐ ☐ ☐ ☐

Written in size 12,

☒ ☒ ☒ ☐ ☐ ☐ ☐
Times New Roman
font
Paragraphs include at ☒ ☒ ☒ ☐ ☐ ☐ ☐
least 3 sentences
Page numbers?

**The default font

for page numbers is
☒ ☒ ☒ ☐ ☐ ☐ ☐
Calibri, size 11 even
after you have
changed the font in
your paper so make
sure to check
Spell out ☒ ☒ ☒ ☐ ☐ ☐ ☐
abbreviation at first
use if not recognized
by AMA

***Remember that
you may add/subtract
content with each
draft so something
that once spelled out
might be removed
 5

and need to spelled

out again
Spell out numbers
and abbreviations
that begin a sentence?

**If an abbreviation
☐
must be spelled out to ☒ ☒ ☒ ☐ ☐ ☐
begin a sentence, do
not include the
abbreviation in
parentheses after
words unless this is
the first use.
Numeric values when
referring to numbers ☒ ☒ ☒ ☐ ☐ ☐ ☐
in sentence (“3”, not
“three”)
Reference
superscripts after ☒ ☒ ☒ ☐ ☐ ☐ ☐
each sentence I used
a reference?
OAR is properly
defined as organS at
risk.

**This is a common
☒ ☒ ☒ ☐ ☐ ☐ ☐
mistake, even in
journal publications.
By saying OARs, you
are implying organs
at risks which doesn’t
make sense
If I directly cited an
author, did I
immediately include ☒ ☒ ☒ ☐ ☐ ☐ ☐
the reference
superscript following
the author’s name?
Tables and figures
are referenced in-text
☐ ☐ ☒ ☐ ☐ ☐ ☐
directly following the
sentence (….(Figure
1).
All terms must be ☒ ☒ ☒ ☐ ☐ ☐ ☐
spelled out in the
abstract and
manuscript at first
use

**So if you refer to

and spell out VMAT
in the abstract, you
must also define the
 6

term again in the

manuscript
Scholarly writing is
appropriate

**Do not use terms ☒ ☒ ☒ ☐ ☐ ☐ ☐

such as max, cord,
rad onc, simmed etc.
Spell out these terms
and avoid slang
All reference of our
profession should be
written as “medical
dosimetrist” not just
“dosimetrist.” ☒ ☒ ☒ ☐ ☐ ☐ ☐

**Remember that
there are other types
of dosimetrists
Is my paper
formatted according
☐ ☐
the instructions? Case ☒ ☒ ☒ ☐ ☐
study vs. Research
Paper
Reference Page
Page break before
☒ ☒ ☒ ☐ ☐ ☐ ☐
this section?
Capitalize the first
letter of the first
☒ ☒ ☒ ☐ ☐ ☐ ☐
word in the title
only
Abbreviate and
italicize the ☒ ☒ ☒ ☐ ☐ ☐ ☐
journal?
Year, volume,
issue and page
number written
without any
spaces?

**If you didn’t

find one listed, ☒ ☒ ☒ ☐ ☐ ☐ ☐
consider
completing another
literature search
review. If you
cannot find one,
reach out to
instructor for help
Doi? ☒ ☒ ☒ ☐ ☐ ☐ ☐

**Remember that
most publications
have doi numbers
 7

now so if you do
not locate one on
the original article,
complete another
literature search to
find it.
Format dois like
this:
http://doi.org...
☐
☒ ☒ ☒ ☐ ☐ ☐
**Remember this
has changed from
last semester
Listed in
chronological
☒ ☒ ☒ ☐ ☐ ☐ ☐
order as they are
referenced in text
Figures and Tables
Page break before
☐ ☐ ☒ ☐ ☐ ☐ ☐
each section?
Each heading is
bolded and
☐ ☐ ☒ ☐ ☐ ☐ ☐
centered for each
section
If 2 figures are
related, they are to
☐ ☐ ☒ ☐ ☐ ☐ ☐
be labeled as A
and B.
Captions are
written in
complete sentences
☐ ☐ ☒ ☐ ☐ ☐ ☐
and single spaced
starting with
“Figure 1”
Figure captions
appear after the ☐ ☐ ☒ ☐ ☐ ☐ ☐
figure
Table captions
appear before the ☐ ☐ ☒ ☐ ☐ ☐ ☐
figure
All patient
identifying
information is
☐ ☐ ☒ ☐ ☐ ☐ ☐
blocked and fused
with the original
image
All table axis,
labels and legends
are in Times New ☐ ☐ ☒ ☐ ☐ ☐ ☐
Roman, size 12
font
Any DVHs include ☐ ☐ ☒ ☐ ☐ ☐ ☐
structure labels
directly on the
 8

DVH
Vertical lines are
removed from ☐ ☐ ☒ ☐ ☐ ☐ ☐
tables
Single line spacing
used for figure and ☐ ☐ ☒ ☐ ☐ ☐ ☐
table descriptions
 9

Improved planning efficiency in multiple brain lesion SRS VMAT cases using Eclipse
scripting
Rebecca Barrett, BA, CMD; Rob Hale, BS, CPhT; Nishele Lenards, PhD, CMD, RT(R)(T),
FAAMD; Ashley Hunzeker, MS, CMD; Sabrina Zeiler, MS, CMD, RT(T)
Medical Dosimetry Program at the University of Wisconsin – La Crosse

ABSTRACT
Though dosimetry has a multitude of treatment modalities, software, and workflows to aid in
the treatment planning process, treatment planners are still responsible for several tedious and
monotonous tasks that could decrease their planning efficiency. The purpose of this study was to
determine if scripting could improve treatment planning efficiency for multiple brain lesion
stereotactic radiosurgery (SRS) volumetric arc therapy (VMAT) cases by reducing planning time
commitment. A script was developed for multiple brain lesion SRS cases using Eclipse Scripting
Application Programming Interface (ESAPI) with the intention of improving treatment planning
efficiency by creating optimization structures (OSs) and importing prescription and suggested
OS metrics to the optimizer. Nine treatment planners were each provided with 3 different
multiple brain lesion, single-isocenter SRS cases and created 2 plans for each case: 1 using the
SRS script, and 1 without using the SRS script. There were 54 treatment plans developed,
totaling 27 plan comparisons. Statistical analyses of planning time commitment with and without
the SRS script were performed using RStudio. The mean and median planning times with and
without the SRS script were compared using a paired T-test and Wilcoxon Signed Rank (WSR)
test, respectively, and effect size was evaluated using Cohen’s classification. Using the SRS
script resulted in statistically significant reduction in total contouring time (11.3 vs 2.8 minutes,
P < 0.001), optimizer preparation time (7.7 vs. 2.1 minutes, P < 0.001), and overall planning
time (105.1 vs. 77.9 minutes, P < 0.001). This study concluded that scripts developed using
ESAPI offer an opportunity to improve treatment planning efficiency by reducing the planning
time commitment for treatment planners.

Keywords: Stereotactic Radiosurgery, Scripting, Treatment Planning Efficiency, Planning Time

Commitment, Contouring, Optimizer Preparation
 10

Introduction
Radiation has been used as a form of cancer therapy since the beginning of the twentieth
century and has since consistently progressed to improve patient treatment. With increased
complexities, medical dosimetrists have been required to adapt and improve treatment planning
efficiency on a multitude of levels. For example, stereotactic treatments for brain metastases,
such as radiosurgery (SRS), were originally limited to single lesion treatments but are now
commonly used to treat multi-lesion cases with a single isocenter. As these plans allow for
precise delivery of high doses, complexity of the treatment planning process has become ever
more complicated.1 Many new technologies, such as Varian HyperArc and RapidPlan, are
available for clinics to purchase to improve the efficiency of the medical dosimetrist’s daily
tasks, resulting in a more automated and proficient treatment planning procedure.2,3 Fung et al4
found a decrease in overall planning time when using organs at risk (OAR) auto-contouring
versus manual contouring for head and neck cases, for example. Though these technologies are
useful tools that can simplify medical dosimetry tasks, many are still being studied, are not
available to all clinics, and require multiple rounds of QA testing to implement in the radiation
oncology workflow.
Eclipse scripting offers planning automation with the flexibility of direct editing at the
hands of individual departments.5 Scripting is the use of coding for advanced computer
programming in the treatment planning process and can be shared through online portals,
allowing clinics to utilize them without a designated on-site coder. Additionally, these Eclipse
scripts can be as simple or complex as planners find useful and executed directly through
Eclipse, offering an opportunity to effectively improve planning efficiency.6 Given that medical
dosimetrists have many tedious responsibilities, they are required to dedicate time and effort to
less advanced tasks which, in turn, could reduce productivity and limit the number of treatment
plans completed per week.7 With Eclipse scripting, Teruel et al8 saved approximately 2-3 hours
in plan setup tasks using a total body irradiation planning script that generated multi-isocenter
volumetric modulated arc therapy (VMAT) treatment plans. Additionally, Zhang et al9 identified
that auto-planning for nasopharyngeal carcinoma was an effective way to save time on treatment
planning while achieving clinically acceptable plans with lower overall monitor units.
Researchers have demonstrated the ability of treatment planning automation through scripting as
an effective time-saving option, especially for complex cases with several monotonous tasks.4,8.9
 11

Multiple brain lesion SRS cases tend to be classified as complex given the varying sizes and
locations of the target volumes as well as the multitude of optimization structures (OSs)
commonly utilized during the planning process.1
For multiple brain lesion SRS cases, OSs are useful tools to ensure sufficient dose
coverage in the center of the target, a reduction of hotspots outside of the gross tumor volume
(GTV), and a uniform, sharp dose fall-off outside of the planning target volume (PTV). Using
only targets, OAR, and normal tissue objectives in the optimizer can yield acceptable plans;
however, some medical dosimetrists find OSs useful in many cases, especially multiple brain
lesion SRS. Creating these additional structures and planning objectives requires the medical
dosimetrist to dedicate more time to the planning process to create a clinically acceptable plan.10
Highly complex plans, like multiple brain lesion SRS VMAT cases, can improve dose delivery
and clinical outcomes; the problem, however, is that these cases require an extensive time
commitment to create OSs and add multiple objectives to the optimizer, leading to decreased
treatment planning efficiency. Implementing a script that creates optimization structures and
adds these structures, targets, and OAR to the optimizer could offer a time-reducing solution to
tedious tasks that medical dosimetrists are required to complete. The purpose of this study was to
determine if scripting could improve treatment planning efficiency for multiple brain lesion SRS
VMAT cases by reducing planning time commitment. Researchers tested the hypotheses that
using Eclipse Scripting Application Programming Interface (ESAPI) will improve treatment
planning efficiency for multiple brain lesion SRS cases by (H1A) reducing the total contouring
time, (H2A) reducing the optimizer preparation time, (H3A) reducing the total number of
optimizations and (H4A) reducing the overall planning time.
Materials and Methods
Scripting Development
The SRS script used in the current study was developed primarily for multiple brain
lesion SRS cases. The SRS script was written by a practicing board certified medical dosimetrist
(CMD). The scripting process utilized the C# programming language and ESAPI, a scripting
application integrated in the Varian Eclipse treatment planning system.5,6,7 Using ESAPI, C#
scripts can access several aspects of patient data from Eclipse, including plan, image, and
structure set data.5
 12

The SRS script required the user to select the appropriate ARIA prescription for the plan.
With access to the structure set assigned to the plan and the ARIA prescription selected by the
user, the SRS script performed 4 primary functions: created 4 OSs, imported suggested
objectives for these OSs into the plan’s optimizer based on the prescription dose, imported all
prescription metrics to the plan’s optimizer, and assigned normal tissue objective (NTO)
parameters. The 4 OSs created by the SRS script include a structure that represents all PTVs
minus all GTVs, a structure that represents a sum of 1.0 mm squares at the center of each GTV,
wall structures with varying margins on the sum of the PTVs, and bridging structures for PTVs
that are proximal to one another (Figure 1, Table 1). The SRS script recognized the PTVs and
GTVs in the structure set based on structures that start with “GTV” and “PTV”. Descriptions of
each OS and how each was created are summarized in Table 1 and examples of each OS within
the structure set are shown in Figure 1.
The SRS script imported 3 sets of objectives to the optimizer following OS creation. The
first set of objectives was prescription metrics. These included all target and OAR metrics listed
in the user-selected ARIA prescription. These prescription metric objectives were used for
monitoring purposes only and were imported to the optimizer with a priority of zero. The second
set of objectives included OS and target objectives suggested by the SRS script to maintain target
coverage while optimally utilizing the OSs (Table 2). These objectives were scaled based on the
prescription dose. The final objective added to the optimizer was the NTO, predefined in the
SRS script with a 0.4 cm distance from target border, 40% start dose, 4% end dose, 0.5 fall-off,
and 150 priority.
Patient Selection and Simulation
To test the SRS script’s ability to decrease total contouring time, optimizer preparation
time, total number of optimizations, and overall planning time, cases with multiple brain lesions
were planned both with and without the SRS script. This study was a retrospective analysis of 3
anonymized cases: case 1 had 5 brain lesions, case 2 had 9 brain lesions, and case 3 had 13 brain
lesions. All cases were simulated in a supine position with a neutral head position, immobilized
with the Qfix Encompass SRS Fibreplast System, and set up using a stereotactic tolerance. All
cases were planned in Varian Eclipse treatment planning system.
Treatment Planning
 13

Several controls were implemented to ensure that all time metrics were measured specific
to the planning process. Controls included contours, beam definition, avoidance sectors or
structures, calculation parameters, and dose level. Contoured structures predefined for all cases,
regardless of script utilization, included the body, brain, brain minus all GTVs, brainstem, optic
chiasm, optic nerves, eyes, normal tissue (NT) structures, all GTVs, all PTVs, and a sum PTV
structure. The NT structures were defined as an 8.0 mm wall around each GTV, and some NT
structures were combined given a 6.0-8.0 mm proximity between GTVs. Target volumes, OAR,
and planning objectives were created in accordance with guidelines from Radiation Therapy
Oncology Group (RTOG) and the established organization combining the National Surgical
Adjuvant Breast and Bowel Project (NSABP), RTOG, and Gynecologic Oncology Group (GOG)
called NRG.11
The body contour included the Qfix Encompass masking system per the participating
clinic’s standards. The treatment technique was single-isocenter SRS VMAT. All plans were
completed using 6 MV flattening filter free beam energy. Isocenter, gantry rotations, collimator
rotations, and couch rotations were predefined for each case based on what was clinically used,
both with and without the SRS script (Table 3). The plans for case 1 utilized an avoidance sector
from 40-320° for the coplanar beam and the plans for case 2 and case 3 utilized a triangular entry
avoidance structure that included the eyes, nasal cavities, and mouth (Table 3). The decision to
use an avoid entry structure versus sector was based on what was clinically used for each case.
Calculation parameters were preset, including a calculation grid size of 0.1 cm, convergence
mode set to “On”, and a calculation volume reset to ensure proper calculation volume fit to the
body contour. The prescription for all plans was 2000 cGy in 1 fraction. Planning objectives
were applied according to the clinic’s guidelines.
All cases were completed by a total of 9 treatment planners, either a CMD or medical
physicist. All 9 participating planners had experience in multiple brain lesion SRS treatment
planning with each participant completing 2 plans per case, 6 plans in total. Twenty-seven plans
were completed with the SRS script and 27 plans were completed without the SRS script for a
total of 54 plans and 27 plan comparisons. There was a mandatory period of 7 or more days
between planning the same case with and without the SRS script to reduce the potential influence
of a participant’s first planning process on their second planning process.
Time Comparison
 14

All treatment planning processes were screen recorded using Zoom to track total
contouring times, optimizer preparation times, total number of optimizations, and overall
planning times. Screen recordings were diligently reviewed to mark start and stop times for each
task, except for total number of optimizations which were incrementally noted throughout the
planning process. If pauses of >30 seconds were detected while reviewing the screen recordings
during any point in the planning process, these pause times were subtracted out accordingly from
the appropriate time measurement metric.
Total contouring time was measured in both the contouring and external beam planning
workspaces depending on where the planner completed the task. In the contouring workspace,
contouring time began when the planner opened the workspace. In the external beam planning
workspace, contouring time began when the planner selected the freehand drawing tool.
Contouring time ended when the planner stopped contouring and either saved their work, exited
the contouring workspace, or started another task such as reoptimizing or plan copying. For plans
using the SRS script, the script running time was included in the total contouring time. Script
running time started when the planner selected Tools > Scripts… and ended when the script’s
Completed notice appeared on the screen. For plans without the SRS script, total contouring time
referred only to the sum of time spent contouring for each contouring instance.
Optimizer preparation time started when the planner opened the optimizer for the first
time and ended when the planner clicked Start VMAT Optimization. For plans with the SRS
script, the planner could begin optimization without changing the objectives that the SRS script
entered. Planners also had the option to spend time editing these parameters or adding additional
objectives if desired. Aside from predefined avoidances, plans without the SRS script had an
empty optimizer, requiring planners to add as many or as little objectives as desired to start
optimization. Planners could add objectives either manually or using an objective template. If
planners exited the optimizer to save the plan or to contour a new optimization structure, for
example, multiple optimization preparation times were recorded and added together. Optimizer
preparation time was recorded for the first optimization only.
The number of optimizations were tracked during the entire planning process for each
plan. The total number of optimizations was determined by the amount of completed
optimizations needed to complete the plan. If an optimization was started and stopped before
 15

completion, thus not contributing to the final plan, that optimization was not included in the final
count of optimizations.
The overall planning time measurement began when the planner started working on the
plan, either by copying the original plan, going into the contouring workspace, or opening the
SRS script. The measurement ended when the planner copied the plan and labeled it FINAL.
Overall planning times included the contouring time, script running time, optimizer preparation
time, and all optimizations.
Planners were aware of the functions that the SRS script performed. They were informed
that the SRS script’s functions were meant to be useful suggestions as opposed to creating a fully
automated planning process. Planners were permitted to set and adjust clinical goal priorities
during the optimization process both with and without the SRS script. They could also change
pre-planning parameters, such as aperture size controller or MR level after intermediate dose
calculation, given that they adjusted these parameters for plans both with and without the SRS
script.
Statistical Analysis
Each plan was individually evaluated to collect appropriate time stamp data for this
study. The data for each hypothesis was evaluated for normality using the Shapiro-Wilk test.
Total contouring time was the only data metric that showed a normal distribution. For this data
measurement, researchers used a paired T-test to determine whether the contouring time needed
with the SRS script was significantly different from that needed without the SRS script. All other
data measurements showed a non-normal distribution and so were evaluated for statistical
significance using the Wilcoxon Signed Rank test (WSR). These data measurements included
optimizer preparation time, total number of optimizations, and overall planning time. The effect
size (d), evaluated for the practical significance of each data measurement comparison, was
|Z|
calculated as d= where Z is the standardized z-score and N g is the number of pairs, corrected
Ng
with Hedges’ g for small samples.12 Cohen’s classification was used to evaluate the overall effect
size with < 0.2 as small, 0.2-0.8 as medium, and > 0.8 as large.13 Statistical analyses were
completed using RStudio, version 4.2.1. R packages openxlsx, psych, readxl, tidyverse, and zoo
were used.
Results
 16

Total Contouring Time

The total contouring time comparison for plans completed with and without the SRS
script was evaluated using a paired T-test. With the SRS script, total contouring time was
reduced in 25 of the 27 plan comparisons and was increased in 2 of the 27 plan comparisons,
with 2 outliers observed for plans with the SRS script (Figure 2A). The mean total contouring
time percent reduction with the SRS script was 69.2% for case 1, 75.1% for case 2, and 77.4%
for case 3, with similar results when comparing medians. The total contouring time standard
deviation decreased by 2.8 minutes for case 1, 3.5 minutes for case 2, and 4.5 minutes for case 3
(Table 4). The shortest script running time for plans with the SRS script was 0.6 minutes and the
longest script running time was 2.4 minutes. The mean total contouring time without the SRS
script was 11.3 minutes while the mean total contouring time with the SRS script was reduced to
2.8 minutes. The mean percent reduction using the SRS script was therefore 75.1%, with a
similar result when comparing medians (Table 5). The paired T-test showed statistically
significant differences for total contouring time (P < 0.001) with a large effect size (Table 5).
Given the statistically significant total contouring time reduction with the SRS script, the null
hypothesis (H10) was rejected.
Optimizer Preparation Time
The optimizer preparation time comparison for plans completed with and without the
SRS script was evaluated using a WSR test. With the SRS script, optimizer preparation time was
reduced in all 27 plan comparisons (Figure 2B). The mean optimizer preparation time percent
reduction with the SRS script was 67.4% for case 1, 73.6% for case 2, and 75.0% for case 3, with
similar results when comparing medians. The optimizer preparation time standard deviation
decreased by 2.5 minutes for case 1, 0.4 minutes for case 2, and 1.5 minutes for case 3 (Table 4).
The mean optimizer preparation time without the SRS script was 7.7 minutes while the mean
optimizer preparation time with the SRS script was reduced to 2.1 minutes The mean percent
reduction using the SRS script was therefore 72.3%, with a similar result when comparing
medians (Table 5). The WSR test showed statistically significant differences for optimizer
preparation time (P < 0.001) with a large effect size (Table 5). Given the statistically significant
optimizer preparation time reduction with the SRS script, the null hypothesis (H20) was rejected.
Total Number of Optimizations
 17

The total number of optimizations comparison for plans completed with and without the
SRS script was evaluated using a WSR test. With the SRS script, the total number of
optimizations was reduced in 10 of the 27 plan comparisons, unchanged in 14 of the 27 plan
comparisons, and increased in 3 of the 27 plan comparisons (Figure 2C). The mean total number
of optimizations percent reduction with the SRS script was 4.8% for case 1, 21.2% for case 2,
and 10.3% for case 3 and the median total number of optimizations percent reduction was 0%,
33.3%, and 33.3% respectively (Table 3). The total number of optimizations standard deviation
decreased with the SRS script by 0.3 for case 1 and 0.1 for case 2 (Table 4). In contrast, the total
number of optimizations standard deviation increased by 0.4 for case 3 with the SRS script
(Table 4). The mean total number of optimizations without the SRS script was 3.1 while the
mean total number of optimizations with the SRS script was reduced to 2.7. The mean percent
reduction using the SRS script was therefore 13.3%, with a similar result when comparing
medians (Table 5). The WSR test did not show statistically significant differences for total
number of optimizations (P = 1.333) and showed a medium effect size (Table 5). Given the
statistically insignificant reduction in total number of optimizations with the SRS script, the null
hypothesis (H30) failed to be rejected.
Overall Planning Time
The overall planning time comparison for plans completed with and without the SRS
script was evaluated using a WSR test. With the SRS script, overall planning time was reduced
in 22 of the 27 plan comparisons and was increased in 5 of the 27 plan comparisons, with 2
outliers observed for plans with the SRS script (Figure 2D). The mean overall planning time
percent reduction with the SRS script was 12.3% for case 1, 34.0% for case 2, and 26.5% for
case 3, with similar results when comparing medians for cases 2 and 3 (Table 4). For case 1, the
median overall planning time showed a percent increase of 8.2% with the SRS script (Table 4).
The overall planning time standard deviation decreased by 10.6 minutes for case 1 and 16.7
minutes for case 2 while it increased by 9.0 minutes for case 3 with the SRS script (Table 4). The
mean overall planning time without the SRS script was 105.1 minutes while the mean overall
planning time with the SRS script was reduced to 77.9 minutes. The mean percent reduction
using the SRS script was therefore 25.9%, with a similar result when comparing medians (Table
5). The WSR test showed statistically significant differences for overall planning time (P <
 18

0.001) with a medium effect size (Table 5). Given the statistically significant overall planning
time reduction with the SRS script, the null hypothesis (H40) was rejected.
Discussion
The results of this study illustrated a substantial improvement in planning efficiency
using Eclipse scripting for OS contouring and optimizer preparation, as evidenced by the
statistically significant differences in total contouring time, optimizer preparation time, and
overall planning time. Moreover, the standard deviations across all time metrics for every case
when using the SRS script (aside from overall planning time for case 3) decreased. Though an
increase in the number of lesions correlated with increased time commitment for each time
metric regardless of the SRS script use, the decrease in standard deviations suggests that not only
is planning time reduced when using the SRS script, but the reduction in planning time is
consistent across cases, regardless of plan complexity (Table 4). The overall planning time
standard deviation for case 3 did not follow this trend. This was likely due to the 2 outliers noted
in the results (Figure 2D, Table 4). These 2 outliers in overall planning time with the SRS script
could be due to planner discretion regarding what is considered a clinically acceptable plan.
Additionally, researchers in the current study demonstrated that one of the major benefits
to scripting is offering efficiency in OS contouring. Given that the total contouring time and
overall planning time were both significantly reduced, the results imply that planners reduce their
planning time commitment when using the SRS script for contouring optimization structures
(Figures 2A, 2D). Similar results were discussed by Rusu et al,6 who noted that RapidPlan
eliminated the need for planners to generate optimization structures, thereby requiring less active
working time in the treatment planning process. Fung et al4 also noted that manual contouring
time could account for up to 35% of overall planning time, though the researchers tested auto-
contouring for OAR, as opposed to OSs, for nasopharyngeal carcinoma. Teruel et al8 used an
automatic scripting process to create target and optimization contours for total body irradiation
cases and they estimated that the automatically generated contours saved 30-50 minutes in
overall planning time. One potential limitation of that study, which was also discussed in the
Fung et al4 study, was with automatically generated structures and possible distortion
emphasizing that contours should be reviewed before proceeding with automatically generated
structures.
 19

In the current study, several comparisons that did not follow the observed trends included
the decrease of overall planning time median using the SRS script versus not using the SRS
script for case 1 and the standard deviation decrease for case 3 overall planning time (Tables 4,
5). Furthermore, results were inconsistent for total number of optimizations, which likely
correlates with the statistically insignificant results (Figure 2C, Table 4). These discrepancies
could be a result of variation in the participating planners’ overall multiple brain lesion SRS
experience level or their confidence level in their ability to produce a clinically acceptable plan.
Planner-dependent factors were also noted in the Fung et al4 study when discussing the
contouring processes completed by planners. The ending point measurement for both overall
planning time and total number of optimizations were defined once the planner deemed their
plan clinically acceptable, which was a subjective decision. The variability noted in the case-by-
case data for overall planning time and total number of optimizations was less surprising
compared to total contouring time and optimizer preparation time, which had more concrete,
objective ending point measurements. Finally, the total number of optimizations was measured
incrementally with >50% of the measurements at <3; therefore, showing data trends and possibly
statistical significance would likely require a larger number of plan comparisons.
Conclusion
Highly complex plans, like multiple brain lesion SRS VMAT cases, can improve dose
delivery and clinical outcomes; the problem, however, is that these cases require an extensive
time commitment to create OSs and add multiple objectives to the optimizer, leading to
decreased treatment planning efficiency. The purpose of this study was to determine if scripting
could improve treatment planning efficiency for multiple brain lesion SRS VMAT cases by
reducing planning time commitment. Researchers demonstrated that treatment planning
efficiency was significantly improved by reducing planners’ contouring time, optimizer
preparation time, and overall planning time. There was a correlation in script effectiveness and
plan complexity, as the 2 most complex cases showed an improvement of 30-40 minutes in
average total planning time using the SRS script. The greatest effects were for total contouring
time and optimizer preparation time, demonstrating that the SRS script’s functions especially
aided in the beginning stages of the planning process. Total planning time was significantly
reduced as well, though the total number of optimizations was not.
 20

There are limitations worth noting for future research. The study was performed at a
single institution with a limited sample size of patients and medical dosimetrists. In the future, a
scripting effectiveness study would benefit from a larger sample of patients and more
involvement from multiple institutions with a greater number of participating medical
dosimetrists and planning styles. Additionally, researchers in the current study did not evaluate
plan quality through target coverage, OAR sparing, or total monitor units, but rather allowed the
participant medical dosimetrists to decide when their plan was clinically acceptable. An
evaluation of metrics with and without a script like the SRS script could offer insight to the plan
quality of script-aided plans.
Scripting offers a wide variety of opportunities for automation ranging from full plan
automation to contouring to patient setup.8 Both simple and complex automation provide
treatment planners with a time-saving solutions for contouring and optimizer preparation, leading
to decreased overall planning time commitment. Based on the significance of current and prior
research, it would be ideal if scripts for other treatment sites and techniques, including SRT,
SBRT, and other intensity-modulated treatment planning, were available.
 21

References
1. Desai DD, Cordrey IL, Johnson EL. Efficient optimization of R50% when planning multiple
cranial metastases simultaneously in single isocenter SRS/SRT. J Appl Clin Med Phys.
2021;22(6):71-82. https://doi.org/10.1002/acm2.13254
2. Saw CB, Li S, Battin F, McKeague J, Peters CA. External beam planning module of Eclipse
for external beam radiation therapy. Med Dosim. 2018;43(2):195-204.
https://doi.org/10.1016/j.meddos.2018.03.003
3. Yoder T, Hsai AT, Xu Z, Stessin A, Ryu S. Usefulness of EZFluence Software for
radiotherapy planning of breast cancer treatment. Med Dosim. 2019;44(4):339-343.
https://doi.org/10.1016/j.meddos.2018.12.001
4. Fung NTC, Hung WM, Sze CK, Lee MCH, Ng WT. Automatic segmentation for adaptive
planning in nasopharyngeal carcinoma IMRT: Time, geometrical, and dosimetric
analysis. Med Dosim. 2020;45(1):60-65. https://doi.org/10.1016/j.meddos.2019.06.002
5. Pyyry EJ, Keranen W. Varian APIs: A handbook for programming in the Varian oncology
software ecosystem. Palo Alto, CA: Varian; 2018.
https://varianapis.github.io/VarianApiBook.pdf
6. Rusu I, Roeske J, Solanki A, Kang H. Fully automated planning and delivery of
hippocampal-sparing whole brain irradiation. Med Dosim. 2022;47(1):8-13.
https://doi.org/10.1016/j.meddos.2021.06.004
7. Wang H, Rea A, Rudek B, Chen T, McCarthy A, Barbee D. Automatic couch position
calculation using eclipse scripting for external beam radiotherapy. J Appl Clin Med Phys.
2021;22(2):77-84. https://doi.org/10.1002/acm2.13159
8. Teruel JR, Taneja Sameer, et al. Automatic treatment planning for VMAT-based total body
irradiation using Eclipse scripting. J Appl Clin Med Phys. 2021;22(3);119-130.
https://doi.org/10.1002/acm2.13189
9. Zhang Q, Peng Y, Song X, Yu H, Wang L, Zhang S. Dosimetric evaluation of automatic and
manual plans for early nasopharyngeal carcinoma for radiotherapy. Med Dosim.
2019;45(1):13-20. https://doi.org/10.1016/j.meddos.2019.05.005
10. Bell JP, Patel P, Higgins, K, McDonald MW, Roper J. Fine-tuning the normal tissue
objective in eclipse for lung stereotactic body radiation therapy. Med Dosim. 2017;43(4):344-
350. https://doi.org/10.1016/j.meddos.2017.11.004
 22

11. Kruser TJ, Bosch WR, Badiyan SN, et al. NRG brain tumor specialists consensus guidelines
for glioblastoma contouring. J Neurooncol. 2019;143(1):157-166. doi:10.1007/s11060-019-
03152-9
12. Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. Hillsdale, NJ:
Lawrence Erlbaum Associates; 1988.
13. Hedges LV, Olkin I. Statistical Methods for Meta-Analysis. New York, NY: Academic
Press; 1985.
 23

Figures

Figure 1. The SRS Research script creates a total of 4 OSs. Figure (A) shows zPTV-GTV_Total
(dark red), Figure (B) shows zGTV_opt_Total (dark purple) with a GTV (red), Figure (C) shows
zWall1 (dark red), zWall2 (purple), and zWall3 (light brown) with 2 PTVs (dark blue), and
Figure (D) shows zBridge_90% (green), zBridge_60% (light purple), and zBridge_30% (orange)
with 3 PTVs (dark blue).
 24

A. B.

C. D.

Figure 2. Side-by-side boxplots of difference in (A) total contouring time, (B) optimizer
preparation time, (C) total number of optimizations, and (D) overall planning time with and
without the SRS script.
 25

Tables

Table 1. Optimization structures created using the SRS script.

Necessary
OS Purpose Creation Cropped Margin
components
Limit hotspot All GTVs
All PTVs and all
zPTV-GTV_Total in GTV to subtracted from all NA
GTVs
PTV margin PTVs
Attain hotspot 1 mm square
All GTVs and their
zGTV_opt in GTVs’ placed at each NA
center coordinates
centers GTV’s center
zWall3 40 mm outer wall 12 mm from PTVs
Limit dose
zWall2 All PTVs 12 mm outer wall 5 mm from PTVs
around PTVs
zWall1 5 mm outer wall 1 mm from PTVs
Distance from PTV 15 mm hexagon
zBridge_30% centers > 35 mm placed at PTV pair 12 mm from PTVs
and < 55 mm center ± 5 slices
Limit dose Distance from PTV 8 mm hexagon 5 mm from PTVs
zBridge_60% between centers > 16 mm placed at PTV pair with zBridge_30%
proximal and < 40 mm center ± 3 slices subtracted
PTVs 1 mm from PTVs
5 mm hexagon
Distance from PTV with zBridge_30%
zBridge_90% placed at PTV pair
centers < 25 mm and zBridge_90%
center ± 2 slices
subtracted
Abbreviations: OS, optimization structure; GTV, gross tumor volume; PTV, planning target volume; NA, not applicable.

Table 2. Metrics and NTO values added to the optimizer through the SRS script. Note that all
prescription metrics were also added to the optimizer with the SRS script but for monitoring
purposes only.
“a”
Structure Objective Dose (%) Volume (%) Priority
parameter
100 100 NA 120
Lower
PTVs 105 99 NA 120
Upper 140 0 NA 50
Lower 100 99 NA 100
zPTV-GTV_Total
Upper 117.5 0 NA 90
zGTV_opt Lower 135 100 NA 100
zWall3 25 NA 20 50
zWall2 Upper gEUD 55 NA 40 50
zWall1 85 NA 30 50
zBridge_30% 25 NA 25 50
zBridge_60% Upper gEUD 55 NA 40 50
zBridge_90% 85 NA 35 50
Abbreviations: PTV, planning target volume; NA, not applicable; gEUD, generalized equivalent uniform dose.
Table 3. Beam definitions that were predefined for cases 1-3.
 26

Number
Case GTVs/PTVs Gantry, ° Collimator, ° Couch, ° Avoidance
of Arcs
179-181 345 0
330-230 40 270 Sector
1 5 4
220-320 50 315 40-320°
30-130 340 45
179-181 355 0
200-300 25 315
2 9 4 Structure
45-145 15 45
140-40 5 90
179-181 5 0
320-220 345 290
3 13 5 230-330 40 325 Structure
20-120 315 35
140-40 285 70
Abbreviations: GTV, gross tumor volume; PTV, planning target volume.

Table 4. Comparison of mean, median and standard deviation data between cases with the SRS
script and without the SRS script. All values for total contouring time, optimizer preparation
time, and overall planning time are given in minutes.
Total contouring Optimizer Total number of Overall planning
Script
Case time preparation time optimizations time
Use
Mean Med SD Mean Med SD Mean Med SD Mean Med SD
With 2.0 1.0 1.8 2.2 2.4 1.4 2.2 2.0 1.3 66.1 67.8 24.5
1
Without 6.4 7.3 4.5 6.9 5.5 3.9 2.3 2.0 1.6 75.4 62.6 35.2
With 2.9 1.5 2.4 1.8 1.0 1.6 2.9 2.0 2.1 76.4 75.4 39.0
2
Without 11.5 12.4 5.8 6.8 6.9 2.0 3.7 3.0 2.2 115.8 104.8 55.7
With 3.6 1.8 3.5 2.3 1.7 2.0 2.9 2.0 2.3 91.3 64.0 63.0
3
Without 16.0 19.3 8.0 9.2 10.1 3.4 3.2 3.0 1.9 124.2 98.4 54.0
Abbreviations: Med, median; SD, standard deviation.

Table 5. Comparison of means and medians for measured metrics with and without the SRS
script and ranks using the Wilcoxon Signed-Rank test. Mean and median data for total
contouring time, optimizer preparation time, and overall planning time are given in minutes.
 27

With the SRS Without the

Negative Positive Tied Effect P-Value
Metric script SRS script
Ranks Ranks Ranks Size Adjusted
Mean Med Mean Med
Total contouring
2.8 1.5 11.3 10.9 2 25 0 1.67 < 0.001
time
Optimizer
2.1 1.7 7.7 6.9 0 27 0 2.29 < 0.001
preparation time
Total number of
2.7 2 3.1 3 1 1 25 0.23 0.133
optimizations
Overall planning
78 67.8 105.1 96.4 3 24 0 0.60 < 0.001
time
Abbreviations: Med, median; SRS, stereotactic radiosurgery.