Received: 28 July 2020 | Revised: 21 December 2020 | Accepted: 25 January 2021

DOI: 10.1002/acm2.13206

RADIATION ONCOLOGY PHYSICS

Characteristics and limitations of a secondary dose check

software for VMAT plan calculation

Andrew J. Shepard | Sean P. Frigo

Department of Human Oncology, School of

Medicine and Public Health, University of Abstract
Wisconsin-Madison, Madison, WI, USA Purpose: To assess the implementation, accuracy, and validity of the dosimetric leaf
Author to whom correspondence should be gap correction (DLGC) in Mobius3D VMAT plan calculations.
addressed. Andrew J. Shepard Methods: The optimal Mobius3D DLGC was determined for both a TrueBeam with
E-mail: [email protected]
a Millennium multi-leaf collimator and a TrueBeamSTx with a high-definition multi-
Funding information This research did not leaf collimator. By analyzing a broad series of seven VMAT plans and comparing the
receive any specific grant from funding
agencies in the public, commercial, or not- calculated to the measured dose delivered to a cylindrical phantom, optimal DLGC
for-profit sectors. values were determined by minimizing the dose difference for both the collection of
Present address all plans, as well as for each plan individually. The effects of plan removal from the
Andrew J. Shepard, Department of Radiation
optimization of the collective DLGC value, as well as plan-specific DLGC values,
Oncology, University of Iowa, Iowa City, IA,
USA were explored to determine the impact of plan suite design on the final DLGC
determination.
Results: Optimal collective DLGC values across all energies were between −0.71
and 0.89 mm for the TrueBeam, and between 0.35 and 1.85 mm for the True-
BeamSTx. The dose differences ranged between −6.1% and 2.6% across all plans
when the optimal collective DLGC values were used. On a per-plan basis, the plan-
specific optimal DLGC values ranged from −4.36 to 2.35 mm for the TrueBeam,
and between −1.83 and 2.62 mm for the TrueBeamSTx. Comparing the plan-speci-
fic optimal DLGC to the average absolute leaf position from the central axis for
each plan, a negative correlation was observed.
Conclusions: The optimal DLGC determination depends on the plans investigated,
making it essential for users to utilize a suite of test plans that encompasses the full
range of expected clinical plans when determining the optimal DLGC value. Valida-
tion of the secondary dose calculation should always be based on measurements,
and not a comparison with the primary TPS. Varying disagreement with measure-
ments across plans for a single DLGC value indicates potential limitations in the
Mobius3D MLC model.

KEY WORDS
dosimetric leaf gap, leaf offset, MLC model, Mobius3D, secondary dose check

----------------------------------------------------------------------------------------------------------------------------------------------------------------------
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited.
© 2021 The Authors. Journal of Applied Clinical Medical Physics published by Wiley Periodicals, Inc. on behalf of American Association of Physicists in Medicine.

216 | wileyonlinelibrary.com/journal/jacmp J Appl Clin Med Phys 2021; 22:3:216–223

SHEPARD AND FRIGO | 217

1 | INTRODUCTION rounded leaf tip of the MLC. There is no consideration of leaf height
and no leaf offset table. Alternatively, Mobius3D utilizes a full
Secondary dose calculations are utilized to serve as an independent rounded leaf-tip calculation using ray-tracing and considering the leaf
verification of the primary treatment planning system (TPS). This inde- height to define the transmission function, which results in a smooth
pendence is enhanced when the secondary check arrives in a pre-con- falloff of the fluence through the rounded leaf end.3
figured state with a standard beam model for a given machine/energy There have been investigations reported in the literature regard-
class. In this case, it is recommended that the user makes minimal, if ing the commissioning, validity and implementation of Mobius3D for
any, modifications to the software’s model configuration. When use as a secondary dose calculation tool.4–8 On average, prior works
changes are made, it is essential to have an understanding of the modi- saw typical agreement relative to ion chamber measurements in a
fiable factors and their impact on dose calculations, as well as how phantom of <2%,4,6–8 though differences as large as 5.5% were
parameter value variation relates to a measurement scenario. reported.8 These authors focused on the overall implementation of
Mobius3D (Varian Medical Systems Inc., Palo Alto, CA) is a sec- the software and validated the use of Mobius3D as a suitable sec-
ondary check software that provides quality assurance calculations ondary check software for treatment planning systems used within
for a full range of clinical photon and electron plans. For photon their clinic. Hillman et al. also investigated the Mobius3D MLC
beams, Mobius3D utilizes a CT dataset along with plan and structure model as applied specifically to SRS/SBRT treatments and looked at
information (exported from the primary TPS) to perform an indepen- a revised MLC model.9 Further, they investigated an alternative
dent convolution-superposition dose calculation for comparison with DLGC determination method that relied on sweeping gap plans,
the dose calculated by the primary treatment planning system.1 Ven- though noted that ultimately a DLGC optimization based on patient
dor-specific beam models which contain a set of default values are plans is likely necessary.9 Work by Kim et al. investigated the MLC
provided. Outside of the necessary linac calibration conditions, it is modeling accuracy in Mobius3D, observing higher uncertainties in
recommended that the beam model is maintained as provided unless the dose calculation for small fields, and noting that careful optimiza-
large variations are observed between the reference and measured tion of the DLGC is necessary for optimal performance.10
percent depth dose (PDD) and off-axis ratios (OAR). This work presents in-depth the determination and implications
Customization of the beam-model may be facilitated with an of the DLGC for a full range of clinically relevant plans, including
auto-modeling feature at the expense of reducing the independence both large-field VMAT and small-field SBRT cases. The optimization
of the beam model.2 In addition to built-in parameters, there is a ser- of the Mobius3D DLGC for both a TrueBeam with a Millennium
ies of machine configuration parameters whose values the user can MLC (MMLC) and a TrueBeamSTx with a high-definition MLC
adjust, including those related to the determination of treatment (HDMLC) emphasizes the necessity for a representative set of test
times (dose rate), deliverability (max MU/field), clearance (distance plans in DLGC optimization. The results show how a test plan suite
between isocenter and collimator surface; collimator radius), and in general plays in to TPS validation.
dose (dosimetric leaf gap correction). Of these parameters, the dosi-
metric leaf gap correction (DLGC) is the only one that will directly
2 | MATERIALS AND METHODS
affect the dose calculation for MLC-based plans, and subsequently
the dose calculation accuracy of the system.
2.A | Collective DLGC optimization
The Mobius DLG parameter contains two parts, an internal value
(DLGI), which is not exposed to the user and is determined by auto- The optimal DLGC value was determined for a TrueBeam with an
modeling routines executed by the vendor, and a correction (DLGC) MMLC and energies of flattened 6, 10, and 15 MV, as well as unflat-
which the user can vary. Prior to calculation, each MLC leaf position tened 6 and 10 MV. It was also determined for a TrueBeamSTx with
value in Mobius is moved by an amount of DLG/2 = an HDMLC and the same energies except 15 MV flattened. The DLGC
(DLGI + DLGC)/2. There are specific DLGI and DLGC values for was determined for each energy on each machine independently fol-
each machine and energy combination. According to Varian docu- lowing an optimization process that was a variation of that recom-
mentation, the DLGC in Mobius3D is modifiable by the user on a mended by Mobius3D,2 where our approach utilized multiple ion
per-machine, per-energy basis to help account for differences chamber measurements (in target, both on- and off-axis). The optimal
between real-world dosimetry and the inherent Mobius MLC trans- collective DLGC value for each energy was determined by comparing
mission model.2 Making the DLGC more positive increases the calcu- the Mobius3D calculated dose at several DLGC values to the mea-
lated dose, while making it more negative decreases the calculated sured ion chamber dose for a series of test plans.
dose. Essentially, the Mobius3D DLG can be interpreted as a single Seven VMAT test plans were utilized throughout the optimiza-
entry to a leaf offset table within the software, where the DLGC is tion: four geometrically based, and three anatomically based. The
used to tune the internal DLGI to the institution’s preference. four geometrically based plans were based on recommendations
An important distinction when considering the Mobius3D DLG is made in TG-119 and contained either a central cylinder, lateral cylin-
that it is not defined in the same manner as the DLG in the Eclipse der, large cylinder, or c-shape target structure.11 The anatomically
treatment planning system (DLGE). In Eclipse, the DLGE is an inher- based plans consisted of a representative head and neck, lung SBRT,
ent part of the step-wise transmission function utilized to model the and chest wall targets. The plans were chosen to cover a range of
218 | SHEPARD AND FRIGO

the clinical scenarios expected at our institution. An overview of

each plan and the corresponding target structure is provided in
Table 1.
All plans were delivered to a cylindrical TomoTherapy “cheese”
phantom (diameter = 30 cm, length = 18 cm; Accuray Inc., Sunny-
vale, CA), which allowed for the placement of six A1SL ion chambers
(rcav = 2.00 mm; collecting volume = 0.053 cm3; Standard Imaging
Inc., Middleton, WI) laterally along the central slice of the phantom.
The ion chambers were positioned such that measurements were
acquired at representative target (high-dose) locations for each of
the plans investigated. An example of the setup used for measure-
ment acquisition is shown in Fig. 1. Our configuration differed
slightly from the recommendation in Mobius3D documentation for
the determination of the DLGC, which recommended the use of the
Mobius Verification Phantom and a single cylindrical ion chamber.2
The decision to use the cheese phantom with multiple A1SL ion F I G . 1 . Tomotherapy “cheese” phantom measurement setup. The
phantom was positioned with the ion chamber inserts located
chambers was made based on the availability of equipment in our
laterally across the central slice. Ion chambers were adjusted on a
clinic, the desire to acquire multiple measurements for each plan at a per-plan basis such that they were in the target location for each
variety of positions, and the utilization of smaller ion chamber mea- plan.
surement volumes.
For each plan, the ion chambers were placed in high-level, low-
The explored DLGC values investigated were chosen to ensure that
gradient dose regions corresponding to the treatment target volume.
the optimal collective DLGC value would be contained within the
The charge was measured and converted to absolute dose using
range of these values, and interpolation to an optimal result would
appropriate chamber correction and calibration factors. Absolute
be possible.
dose uncertainty for the A1SL was estimated to be 1%.12 The per-
To facilitate dose comparisons, ROI structures representative of
cent difference between the calculated and measured dose was cal-
the collection volumes of the ion chambers were defined in the
culated for each chamber location, defined as
RayStation (v7.0.0.19) treatment planning system (RaySearch, Stock-
ðCalculated  MeasuredÞ holm, Sweden) and exported to Mobius3D. The average dose
Percent Difference ¼ 100 
Measured

This formulation considers the measurement as the reference

and produces a result that gives a direct indication of whether the
calculation is higher or lower than the measurement. Chambers were
only considered if they had a measured dose of >80% of the maxi-
mum calculated dose.
For each machine/energy/MLC/plan combination, Mobius3D
(v2.1) calculations were performed for DLGC values between −2.00
and 2.15 mm, in variable steps as shown in Fig. 2, and were then
compared to measurements by calculating the percent difference.

T A B L E 1 Target structure details. Relevant volume and positional

statistics for the target structure in each plan used for DLGC
optimization.
Absolute distance
Plan Target volume [cm3] from isocenter [cm]
Central cylinder 221.5 0.01
C-Shape 274.2 0.26
F I G . 2 . Average Dose Difference vs. Collective DLGC. The
Large cylinder 1616.2 0.02
average dose difference relative to measurements across all seven
Lateral cylinder 221.5 10.00 test plans for a given collective DLGC, machine, and energy. The
Neck 329.3 3.63 optimal collective DLGC for each machine/energy was determined
by fitting the data linearly and identifying the value for 0.0%
Lung SBRT 15.5 0.21
absolute average dose difference. Each data point represents the
Chest wall 1634.2 4.13
percent difference averaged across the seven test plans.
SHEPARD AND FRIGO | 219

difference across eligible chambers was first calculated for each plan, and average value for a given plan/energy combination. We then
followed by an average across all plans to determine the total aver- observed the level of agreement between the primary TPS and the
age percentage dose difference at a given DLGC value. A linear fit secondary check software relative to measurements for the test
was applied to the total average dose difference (all plans and eligi- plans, to help aid in the determination of relevant action criteria
ble chambers for a given machine/energy) for either four or five when comparing the secondary check software to the primary TPS.
DLGC exploration values, and based on the fit, the collective DLGC
corresponding to the minimum absolute dose difference was taken
3 | RESULTS
to be the optimal value for the specific energy and machine.

3.A | Collective DLGC optimization

2.B | Individual plan DLGC optimization
The average dose difference between the calculated and measured
In addition to determining the optimal collective DLGC, the optimal dose across all plans investigated is presented in Fig. 2 for a range
DLGC for each individual plan was also calculated in an identical of explored DLGC values for each energy and machine. The linear
manner, as an investigation into the variability of the DLGC based fits had an average R2 of 0.998. The optimal collective DLGC, the
on plan characteristics. As opposed to the collective DLGC, the plan- value at which the fit equaled an average dose difference of 0.0%,
specific values will be referred to in this work as the plan-specific for all machines and energies is presented in Table 2.
DLGC. Although the optimal collective DLGC corresponded to the value
The plan-specific values were calculated for each machine and at which the average dose difference of all plans was equal to zero,
energy independently based on the line of best-fit for each individual there were variations in the dose difference within a given machine
plan. It was observed that some of the optimal plan-specific DLGC and energy when the optimal collective value was used. This is
values fell outside of the initial values set for the collective DLGC shown in Figs. 3(a) and 3(b), which present the average dose differ-
optimization, so spot-check calculations were performed for those ence for each plan using the optimal collective DLGC for each
plans which required extrapolation. This was in order to verify the energy on the TrueBeam and TrueBeamSTx, respectively. Addition-
validity of the best-fit curve. In the scenario that the extrapolated ally, these figures display the average dose difference across all
value still resulted in a dose difference of >1% (estimate of absolute beam energies for each plan investigated.
dose determination uncertainty for A1SL12), the new data point was It was observed that at the optimal collective DLGC, the average
added to the analysis, and a new best-fit for that specific plan was dose difference relative to measurements (1 standard deviation)
determined. when considering plan and energy independently was −0.1%  2.0%
The importance of each plan to the determination of the collec- (range: −6.1% to 2.6%) for the TrueBeam, and −0.0%  1.5% (range:
tive DLGC was assessed by performing a collective DLGC optimiza- −4.0% to 2.2%) for the TrueBeamSTx. Considering only the magni-
tion while excluding single plans from the analysis, in order to assess tude of the dose difference relative to measurements, the average
the potential impact of test plan selection. across all energies and plans investigated was 1.4%  1.4% (range:
0.0%–6.1%) for the TrueBeam, and 1.1%  0.9% (range: 0.1% to
4.0%) for the TrueBeamSTx. At the plan level, the most significant
2.C | Plan-specific DLGC correlation to plan
dose differences occurred for the lateral cylinder and lung SBRT
complexity metrics
plans. Mobius3D consistently calculated doses greater than measure-
The optimal values for the plan-specific DLGC were assessed relative ments for the lateral cylinder plan, and less than measurements for
to several plan metrics to assess whether any relevant correlation the lung SBRT plan. The average magnitude of the dose difference
between DLGC values and plan metrics was present. These metrics across both machines was 2.0%  0.6% for the lateral cylinder plan,
were calculated based on work by Desai et al. and included the aver- and −3.4%  1.3% for the lung SBRT plan.
age absolute leaf position off the central axis (ALPCA), mean aper-
ture displacement (MAD), and modulation complexity score (MCS),
3.B | Individual plan DLGC optimization
among others.13
Due to the dose differences observed for individual plans at the
optimal collective DLGC, an identical DLGC analysis was performed
2.D | Comparison with RayStation calculations
To investigate how the dose differences observed in Mobius3D T A B L E 2 Optimal collective DLGC. The optimal collective DLGC
compared with the primary TPS at our institution, all seven plans value for each energy and machine.

were initially calculated in RayStation using standard clinical planning Optimal collective DLGC [mm]
procedures, and then exported to Mobius3D. The calculated average
6X 10X 15X 6FFF 10FFF
dose to relevant A1SL ion chamber structures in RayStation was
TrueBeam −0.02 0.89 −0.46 −0.71 0.07
compared to the corresponding average values to the same struc-
TrueBeamSTx 1.61 0.35 – 1.85 0.53
tures in Mobius3D. The differences were analyzed for their range
220 | SHEPARD AND FRIGO

F I G . 4 . Plan-Specific DLGC Analysis. The dose difference between

the calculated and measured dose at varying DLGC values for each plan
as calculated for the 6 MV flattened beam model on the TrueBeam. The
points indicate explicit calculations that were performed, while the solid
line indicates the line of best fit for each plan.

3.C | Plan-specific DLGC correlation to plan

complexity metrics
To assess the potential dependence of the DLGC on plan characteris-
tics and MLC position distribution, the average absolute leaf positions
from the central axis (ALPCA), mean aperture displacement (MAD),
and modulation complexity score (MCS) were calculated for each plan.
The optimal plan-specific DLGC value averaged over all energies was
compared to each plan metric. The Pearson correlation coefficient was
−0.69, −0.00, and −0.02 for the ALPCA, MAD, and MCS, for the True-
Beam, and −0.97, −0.33, and −0.39 for the TrueBeamSTx. Low corre-
lation values were observed between the plan-specific DLGC values
F I G . 3 . Plan Specific Dose Deviation at Optimal Collective DLGC.
The dose deviation between Mobius3D calculated dose and and both the MAD and MCS metrics; however, there was a relatively
measurement at the optimal collective DLGC for each plan and high correlation observed with the ALPCA. Figure 6 shows the
energy on the (a) TrueBeam with MMLC and (b) TrueBeamSTx with energy-averaged plan-specific DLGC values compared to the ALPCA,
HDMLC is presented. Measurements for the lateral cylinder plan along with a linear best-fit. Plans with a smaller leaf position from the
was noted to be consistently lower than calculations, while
central axis typically required a more positive DLGC, demonstrating a
measurements for the lung SBRT plan was consistently higher than
negative linear trend. To demonstrate the sensitivity associated with
calculations; others show mixed agreement.
each plan to varying DLGC values, the error bars presented in Fig. 6
are representative of a 1% uncertainty in the dose measurement.
on a per-plan basis. This calculation was performed for each Note that plans with a lower DLGC sensitivity, as demonstrated by a
machine and energy combination, and an example of the analysis lower slope in Fig. 4, correspondingly had a larger optimal DLGC
for the 6 MV beam on the TrueBeam model is provided in Fig. 4. uncertainty for a constant 1% dose uncertainty.
All extrapolated values were spot-checked, and the percent differ-
ence was within 0.6% at the extrapolated DLGC value. Consider-
3.D | Effects of plan exclusion from optimization
ing each plan and energy individually resulted in DLGC values
between −5.38 and 4.59 mm. The optimal plan-specific DLGC aver- To assess the potential impact that plan selection can have on the
aged over all energies is shown in Fig. 5, with values ranging from optimal collective DLGC determination, the same analysis was
−4.36 to 2.62 mm. By presenting these data as an average across repeated after removing a single plan and the change in the collec-
all energies, it helps isolate the plan-specific variability in the values tive DLGC was noted. This was performed independently for an
observed. exclusion of both the lung SBRT plan and the lateral cylinder plan
SHEPARD AND FRIGO | 221

F I G . 5 . Optimal DLGC vs. Plan. The optimal plan-specific DLGC

across all energies for a given machine is presented for each plan.
The error bars represent the standard deviation of the dose
difference for all energies for a given plan.

F I G . 7 . Effects of Plan Exclusion. The change in the optimal DLGC

while excluding the lung SBRT or lateral cylinder plan as compared
to the analysis including all plans is provided in (a). The dosimetric
effect of applying the DLGC determined while excluding the given
plan from the analysis is shown in (b).

F I G . 6 . Optimal DLGC vs. Average Leaf Position off Central Axis. increase in the collective DLGC. Further, applying the plan-removal
The optimal DLGC averaged across all energies is compared to the collective DLGC value to the two respective plans resulted in
MLC positioning characteristics of the plan. As the average leaf increases in the absolute dose differences of 0.6  0.4% and
position off the central became larger, the optimal DLGC became 0.2  0.1% for the lung SBRT plan and the lateral cylinder plan,
more negative. DLGC error bars are representative of a 1%
respectively. These changes in the absolute dose difference observed
dosimetric uncertainty and ALPCA error bars are given by the
with the plans excluded from the collective DLGC determination is
standard deviation across all energies.
demonstrated in Fig. 7(b).

(representative of the plans exhibiting the largest dose differences

3.E | Comparison with RayStation calculations
when the optimal collective DLGC value was used).
With plan removal, the change in the optimal collective DLGC Dose calculations were also performed with the RayStation treat-
value ranged from −0.63 to 0.86 mm. The results for all energies are ment planning system for comparison with Mobius3D. On average,
presented in Fig. 7(a). The removal of the lung SBRT plan from the the difference between RayStation calculations and measurements,
analysis always resulted in a decrease in the collective DLGC, while when considering each plan and energy independently, was
the removal of the lateral cylinder plan always resulted in an −0.3  0.9% (range: −2.4% to 1.3%) for the TrueBeam system, and
222 | SHEPARD AND FRIGO

1.1  0.8% (range: −0.4% to 2.6%) for the TrueBeamSTx system. clinics look to potentially implement tighter action criteria, it is
The average magnitude of the dose difference of the RayStation cal- essential that they do so keeping current limitations in mind. If the
culations compared to measurements was 0.7  0.6% (range: goal is to focus the scope on a small number of plans, and the DLGC
0.0%–2.4%) on the TrueBeam system, and 1.1  0.8% (range: has been optimized specifically for the given plan characteristics, it
0.1%–2.6%) on the TrueBeamSTx. may be reasonable to decrease the tolerance values, as in general,
Comparing to Mobius3D using optimized collective DLGC values, relatively good agreement between Mobius3D and measurements
the magnitude of the calculated dose difference relative to measure- was observed. However, when considering a wide range of potential
ments was less for RayStation in 24/35 and 12/28 plans for the plans as was investigated in this work, a tightening of the action cri-
TrueBeam and TrueBeamSTx machines, respectively. For RayStation teria could unduly raise flags due to plan disagreement from calcula-
and Mobius3D calculations relative to measurements for each plan, tion inaccuracy. In this scenario, it may be difficult to truly
it was observed that the RayStation calculations agreed with mea- determine whether disagreement between Mobius3D and the pri-
surements much better for the lung SBRT plan. The average dose mary calculation is due to a problem with the plan, the primary TPS,
difference magnitude across all energies on both machines was or inaccuracies in the Mobius3D dose calculation. This was effec-
1.0  0.7% (range: 0.1%–2.4%), whereas the Mobius3D average tively demonstrated when comparing Mobius3D to RayStation calcu-
dose difference magnitude was 3.4  1.3% (range: 2.1%–6.1%). The lations for the lung SBRT plan. Dose differences in the calculation of
plan that Mobius3D outperformed the RayStation model to the lar- the lung SBRT plan by RayStation and Mobius3D on the order of
gest extent was the chest wall plan, with an average absolute differ- 2%–5% indicate a tightening of the action criteria could result in the
ence of 0.7  0.4% (range: 0.0%–1.2%), compared to 1.2  1.0% onset of false positives. In a similar comparison of RayStation and
(range: 0.0%–2.6%) for the RayStation model. Mobius3D, Kim et al. observed large discrepancies for small-field
Evaluating directly the Mobius3D calculations at the optimal col- cases, which would be characteristic of the lung SBRT case investi-
lective DLGC relative to the RayStation calculations, the average gated in this work.10
dose difference was 0.2%  1.9% (range: −4.8% to 4.3%) for the The large maximum differences observed in the current study
TrueBeam, and −1.1%  1.6% (range: −5.2% to 1.7%) for the True- are believed to potentially be due to the variability in plan character-
BeamSTx. The largest dose differences between the two calculations istics investigated and the possibility that a single collective DLGC
were present for the lung SBRT plan where Mobius3D consistently value is not adequate. To further investigate the potential variability
calculated a dose lower than RayStation, with an average difference in the DLGC with varying plan characteristics, an analysis was per-
of −3.3%  1.1% (range: −5.2% to −2.2%). formed on a per-plan basis. By optimizing the DLGC for each plan,
energy, and machine, the optimal plan-specific DLGC values ranged
between −5.38 and 4.59 mm. Some of the plans had a shallower
4 | DISCUSSION slope in the dose difference versus DLGC data, indicating that a
wide range of DLGC values may result in a minimal dose difference
Performing the DLGC optimization based on minimizing the calcu- for these types of plans. However, the overall observed variations in
lated dose difference with measurements for seven VMAT plans the optimal plan-specific DLGC indicates how the choice of plans
allowed for determination of a single collective value for each energy used for the commissioning process can have a large impact on the
and machine. As expected, the average dose difference followed a collective DLGC that is ultimately chosen by the user, especially for
linear trend with DLGC. This produced an optimal collective DLGC plans whose slope (sensitivity) is higher.
value that results in a minimum dose difference of −0.1%  2.0%
for the TrueBeam, and −0.0%  1.5% for the TrueBeamSTx. Based Using plans that are representative of the full range of expected clin-
on the way in which the optimization was performed, values very ical plans is essential to the proper selection of the collective DLGC
near 0.0% were expected. They generally are in agreement with parameter value to be used. For instance, Fig. 7(a) shows that when
many of the prior publications addressing Mobius3D implementa- the lung SBRT plan was removed from the optimization, the collec-
tion. 4–8
However, in the determination of the collective DLGC, a rel- tive DLGC was reduced by 0.34  0.17 mm. On the contrary, if the
atively large range of dose difference values about the mean was lateral cylinder were removed, the collective DLGC became more
observed, as well as patterns that were similar across all energies positive, increasing by 0.52  0.21 mm. The ultimate effect of the
and machine/MLC models, but varying by plan. This study observed variability that may result if a non-representative suite of test plans
disagreement between Mobius3D and measurements of up to 6.1% was chosen for collective DLGC determination is demonstrated in
for the TrueBeam and 4.0% for the TrueBeamSTx when the collec- Fig. 7(b). The decision to potentially exclude a given plan from the
tive DLGC parameter value was used. Fontenot saw disagreement as collective DLGC determination therefore can lead to significant
large as 5.5%,8 but most prior publications reported values closer to changes in the resultantly determined optimal DLGC value. This is
2–3%.4,5,7 particularly true when considering cases that may define the limits
Mobius3D is often used with relatively large gamma parameter of treatment, such as very small or large fields and is demonstrated
tolerances of 5%/3 mm, providing an effective means for detecting by the lung SBRT case. Removing it from the collective DLGC deter-
gross errors and differences with the treatment planning system. As mination for the TrueBeam resulted in the average dose difference
SHEPARD AND FRIGO | 223

increasing from 3.9% to 4.5%. Prior studies by Kim et al. as well as spectrum of plans, it would likely be necessary for the MLC model to be
Hillman et al. noted inaccuracies in Mobius3D calculations for small- improved. Based on these results, MLC model limitations in the second
field deliveries, further strengthening the assertion that it is neces- check software could have the possibility of raising a false positive. Lastly,
sary to ensure the DLGC is optimized for the relevant clinical a comparison between the primary TPS, secondary TPS, and measure-
plans.9,10 ment, points out that one must validate their secondary TPS against mea-
In addition to target volume, the MLC position characteristics in surement, and not use agreement with the primary TPS for this purpose.
the DICOM RT-Plan data were used to further characterize the test
plans, focusing on the average leaf position relative to the central axis.
ACKNOWLEDGMENTS
Essentially, plans with leaf positions further from the central axis
required a more negative plan-specific DLGC. With only the ability to The authors would like to thank Dr. Adam Bayliss, Dr. Jessie Huang-
define a single DLGC, the model may struggle to accurately calculate Vredevoogd, and Stephanie Olson for assistance with plan creation,
dose for plans with characteristics different than what was used for as well as Dr. Jon Hansen and Maxwell Belanger for assistance with
the determination of the collective DLGC. This point is illustrated in data acquisition.
Fig. 6 where the optimal plan-specific DLGC for a given plan (averaged
across all energies) is plotted relative to the ALPCA through the course
CONFLICT OF INTEREST
of the plan. Considering the results observed for plans with varying
characteristics, multiple DLGC values may be necessary to ensure The authors have no conflict of interest to report.
accurate calculation for all clinical scenarios. In the current framework
of the software, this would require creating a separate machine for REFERENCES
plans with different characteristics. Without the ability for the DLGC
1. Childress N, Stevens E, Eklund D, Zhang M. Mobius3D White Paper:
to easily vary with plan characteristics, the accuracy of the software Dose Calculation Algorithm. In.: Mobius Medical Systems; 2012.
may be limited at the boundary extents of the plan characteristic con- 2. Inc.’ VMS. Mobius 3D: Instructions for Use; 2020.
sidered, as was the case for the lung SBRT plan. 3. LoSasso T, Chui CS, Ling CC. Physical and dosimetric aspects of a multi-
This work focused on the implementation of the DLGC for VMAT leaf collimation system used in the dynamic mode for implementing
intensity modulated radiotherapy. Med Phys. 1998;25:1919–1927.
deliveries; however, it is important to consider the potential impact on
4. McDonald DG, Jacqmin DJ, Mart CJ, et al. Validation of a modern
conformal arc deliveries as well. This is particularly relevant at target second-check dosimetry system using a novel verification phantom. J
edges and for small-field conformal arc plans where the DLG will consti- Appl Clin Med Phys. 2017;18:170–177.
tute a larger fraction of the full irradiation field width. Users should be 5. Kodama T, Saito Y, Hatanaka S, Hariu M, Shimbo M, Takahashi T.
Commissioning of the Mobius3D independent dose verification sys-
cautious in the implementation of large positive or negative DLGC values
tem for TomoTherapy. J Appl Clin Med Phys. 2019;20:12–20.
optimized based on larger fields and which could have a substantial 6. Nelson CL, Mason BE, Robinson RC, Kisling KD, Kirsner SM. Com-
impact on small-field conformal arcs. It would be prudent of the user to missioning results of an automated treatment planning verification
verify the validity of DLGC values on a series of conformal arc test plans system. J Appl Clin Med Phys. 2014;15:4838.
7. Clemente-Gutiérrez F, Pérez-Vara C. Dosimetric validation and clini-
as well, if the desire is to implement the same model as was developed
cal implementation of two 3D dose verification systems for quality
for the VMAT plans. This point further emphasizes the necessity to use, assurance in volumetric-modulated arc therapy techniques. J Appl
or at the very least validate, a full representative suite of test plans for Clin Med Phys. 2015;16:5190.
the determination of the proper DLGC. 8. Fontenot JD. Evaluation of a novel secondary check tool for inten-
sity-modulated radiotherapy treatment planning. J Appl Clin Med
Phys. 2014;15:4990.
9. Hillman Y, Kim J, Chetty I, Wen N. Refinement of MLC modeling
5 | CONCLUSION improves commercial QA dosimetry system for SRS and SBRT
patient-specific QA. Med Phys. 2018;45:1351–1359.
10. Kim J, Han MC, Lee E, et al. Detailed evaluation of Mobius3D dose
A Mobius3D DLGC optimization procedure was implemented on both a calculation accuracy for volumetric-modulated arc therapy. Phys
TrueBeam with an MMLC and a TrueBeamSTx with an HDMLC. Optimal Med. 2020;74:125–132.
DLGC parameter values based on all test plans investigated were able to 11. Ezzell GA, Burmeister JW, Dogan N, et al. IMRT commissioning: mul-
be determined through the comparisons of calculated and measured tiple institution planning and dosimetry comparisons, a report from
AAPM Task Group 119. Med Phys. 2009;36:5359–5373.
dose. While the DLGC was able to be optimized for the collective dose
12. Lopes MDC, Santos T, Ventura T, Capela M. Application of the TRS
difference of the plans, the optimal collective DLGC was not the same as 483 code of practice for reference and relative dosimetry in
the optimal DLGC for each plan individually. Ultimately, the DLGC value tomotherapy. Med Phys. 2019;46:5799–5806.
determined from the optimization procedure will be dependent on the 13. Desai VK, Labby ZE, DeWerd LA, Culberson WS. On the implemen-
tation of the plan-class specific reference field using multidimen-
test plan spectrum, making it essential for users to consider how the soft-
sional clustering of plan features and alternative strategies for
ware will be used and the shortcomings that may result from a given test improved dosimetry in modulated clinical linear accelerator treat-
plan suite. In order to achieve an increased accuracy across a full ments. Med Phys. 2020;47:3621–3635.