Received: 16 January 2018 | Revised: 19 April 2018 | Accepted: 1 June 2018

DOI: 10.1002/acm2.12399

RADIATION ONCOLOGY PHYSICS

Agility MLC transmission optimization in the Monaco

treatment planning system

Michael Roche | Robert Crane | Marcus Powers | Timothy Crabtree

The Department of Medical Physics, The

Townsville Cancer Centre, Douglas, Abstract
Queensland, Australia The Monaco Monte Carlo treatment planning system uses three‐beam model com-
Author to whom correspondence should be ponents to achieve accuracy in dose calculation. These components include a virtual
addressed. Michael Roche source model (VSM), transmission probability filters (TPFs), and an x‐ray voxel Monte
E-mail: [email protected]
Carlo (XVMC) engine to calculate the dose in the patient. The aim of this study was
Funding Information to assess the TPF component of the Monaco TPS and optimize the TPF parameters
Townsville Hospital Health Service
using measurements from an Elekta linear accelerator with an Agility™ multileaf col-
limator (MLC). The optimization began with all TPF parameters set to their default
value. The function of each TPF parameter was characterized and a value was
selected that best replicated measurements with the Agility™ MLC. Both vendor
provided fields and a set of additional test fields were used to create a rigorous sys-
tematic process, which can be used to optimize the TPF parameters. It was found
that adjustment of the TPF parameters based on this process resulted in improved
point dose measurements and improved 3D gamma analysis pass rates with Octa-
vius 4D. All plans calculated with the optimized beam model had a gamma pass rate
of > 95% using criteria of 2% global dose/2 mm distance‐to‐agreement, while some
plans calculated with the default beam model had pass rates as low as 88.4%. For
measured point doses, the improvement was particularly noticeable in the low‐dose
regions of the clinical plans. In these regions, the average difference from the
planned dose reduced from 4.4 ± 4.5% to 0.9 ± 2.7% with a coverage factor (k = 2)
using the optimized beam model. A step‐by‐step optimization guide is provided at
the end of this study to assist in the optimization of the TPF parameters in the
Monaco TPS. Although it is possible to achieve good clinical results by randomly
selecting TPF parameter values, it is recommended that the optimization process
outlined in this study is followed so that the transmission through the TPF is charac-
terized appropriately.

PACS
87.55.D-, 87.55.de, 87.55.Qr, 87.56.bd

KEY WORDS
Agility MLC, MLC transmission, Monaco, TPF, TPS

----------------------------------------------------------------------------------------------------------------------------------------------------------------------
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.
© 2018 Queensland Health. Journal of Applied Clinical Medical Physics published by Wiley Periodicals, Inc. on behalf of American Association of Physicists in Medicine.

J Appl Clin Med Phys 2018; 19:5:473–482 wileyonlinelibrary.com/journal/jacmp | 473

474 | ROCHE ET AL.

1 | INTRODUCTION allowing for confidence in the all aspect of the TPF transmission
characterization.
In modern radiotherapy, treatment planning systems (TPSs) are used
to generate dose distributions with the aim of maximizing tumor
control and minimizing normal tissue complications. Traditional for- 2 | MATERIALS AND METHODS
ward based treatment planning has been supplemented by inverse
planning, which uses dose optimization techniques including intensity The TPFs in the Monaco TPS are characterized by both geometry
1
modulated radiotherapy (IMRT) and volumetric modulated arc ther- and the probability of particle transmission. For the primary collima-
apy (VMAT),2,3 to satisfy user specified criteria. To achieve the tor, a parameter is set in the vendor modeling process to establish
appropriate target coverage and respect the dose constraint criteria the angle beyond which photons and electrons are attenuated by
for organs at risk, both IMRT and VMAT use many irregularly shaped 99%.9 This parameter is not editable by the user from the Monaco
fields defined by multileaf collimators (MLCs). MLCs have been rou- TPS. However, for the secondary collimators, the TPFs are editable
tinely used in radiotherapy over the past 20 years.4,5 Desirable MLC through several parameters13 used to define the transmission proba-
design characteristics include low intraleaf and interleaf transmission, bilities through various regions of the beam modifiers. Editing a num-
a small tongue and groove effect, a small leaf width, accurate and ber of these transmission parameters appropriately can help
fast leaf positioning, and most importantly, reproducibility. Repro- differentiate the variation in transmission through the distinct MLC
ducibility is paramount in an MLC system as this attribute allows for regions. This includes the transmission through the body of the
accurate characterization of the MLCs in the TPS, which in turn MLCs, between adjacent MLCs and through the MLC tips. Fig. 1
facilitates accurate IMRT and VMAT deliveries. illustrates a 2D representation of the MLC TPF and identifies the
The Monaco 5.11.01 Monte Carlo (MC) treatment planning sys- various MLC regions. In reality, the TPF is three‐dimensional where
tem (IMPAC Medical Systems, Inc., Maryland Heights, MO (a sub- the leaf transmission determines its thickness. This thickness is then
sidiary of Elekta AB, Stockholm, Sweden)) uses three‐beam model divided into 11 equally spaced transmission planes so that the trans-
components to achieve accuracy in dose calculation. First, the linear mission of oblique photons can be calculated more accurately.9 The
accelerator photon beam is approximated using a virtual source TJaw transmission and TJaw Tip Leakage TPF parameters are used
model (VSM) consisting of a primary photon source, a scatter photon to determine the transmission through the jaws that travel trans-
source, and an electron contamination source.6–9 The VSM is used verse to the direction of leaf motion.
instead of MC transport through the components of the linear accel- Table 1 displays the TPF parameters investigated in the optimiza-
erator to speed up the calculation. Second, the primary collimator, tion process. More TPF parameters exist in the Monaco TPS; how-
jaws, and MLC are modeled using transmission probability filters ever, they were either for adjusting the position of the collimator
(TPFs).6–9 Similar to the VSM, the TPFs are used instead of direct planes in the VSM or for increasing the backscatter from the collima-
MC simulation to significantly reduce calculation times. Finally, x‐ray tors. Adjustments to these TPF parameters were not investigated.
voxel Monte Carlo (XVMC)10 is used to calculate the dose in the Table 1 also displays four MLC models with varying parameters. The
patient model defined by the patient CT dataset. “Default” model summarizes the TPF parameters when set to their
This study concentrates on the optimization of the TPF using default value and Model A summarizes the TPF parameters post
measurements from an Elekta linear accelerator with an Agility™ completion of the optimization process. Two additional models,
MLC (Elekta AB, Stockholm, Sweden).11,12 To aid with the opti- Model B and Model C, are included to specifically demonstrate the
mization of the TPF, several predesigned fields known as the effects of adjusting the MLC Leakage and Leaf Groove Width TPF
ExpressQA package13 have been provided by the vendor. Although parameters. Model B is identical to Model A except the MLC Leak-
these fields can aid with the optimization, a set of additional test age TPF parameter has been adjusted and Model C is identical to
fields are recommended in this study which will simplify and Model A except the Leaf Groove Width TPF parameter has been
improve the TPF optimization process. The optimization of the TPF adjusted.
™ 14
for the Agility MLC has been previously described and an alter-
native “potential recipe for MLC modeling” was recommended.
2.A | TPF Optimization
However, in this study, no specific details were provided on the
purpose of a number of the TPF parameters defined in the Mon- The optimization began with all TPF parameters set to their default
aco TPS. It was also suggested that certain TPF parameters can be value. The function of each parameter was investigated and a value
unrealistically adjusted. Point dose measurements and gamma anal- was selected that best replicated measurements with the Agility™
ysis showed that this method provides adequate clinical results; MLC. A number of the TPF parameters are linked; as a result, the
however, setting unrealistic values for TPF parameters is not opti- parameters were iteratively adjusted throughout the optimization
mal. This study endeavors to identify an improved TPF optimization process to improve the agreement between measurements and TPS
process where each TPF parameter can be optimized resulting in a calculations. The number of iterations required in the optimization
simplified post modeling optimization process. Using this method, process is significantly dependent on the users experience with the
the fundamentals of the transmission modeling can be guaranteed, TPF in Monaco. An in‐experienced user will most likely have to
ROCHE ET AL. | 475

F I G . 1 . A 2D representation of the geometry of the MLC TPF in the Monaco TPS, including a selection of TPF parameters. Leaf Transmission
defines the fractional transmission through an MLC; Leaf Groove Width defines the extent of the leaf groove region outside the MLC; Interleaf
Leakage and Leaf Transmission define the increase in transmission between adjacent leaves; Leaf Tip Leakage and Leaf Transmission define the
increase in transmission due to the curvature of the MLC tips; MLC Leakage, Interleaf Leakage and Leaf Transmission define the increase in
transmission at the corner of the MLC tips; and MLC Offset defines the difference between the prescribed leaf position and the actual value used
for dose calculation.

T A B L E 1 TPF parameters for various TPS models used in this MLC, this should be 1 mm at the leaf bank plane.12 The first step in
study. the MLC optimization process is to check the closed leaf gap on the
linear accelerator. This was done with a feeler gauge, following the
MLC TPF Parameter Default Model A Model B Model C
recommended vendor procedure.15 Next, the transmission through
Static Leaf Gap (mm) 0.10 0.10 0.10 0.10
the closed leaf gap was measured at isocenter, on central axis, and
Leaf Transmission 0.0030 0.0032 0.0032 0.0032
at multiple positions off axis, with Gafchromic EBT3 film (Ashland
Inter Leaf Leakage 3.0 7.0 7.0 7.0
Specialty Products, NJ). A source‐to‐detector distance (SDD)
TJaw Transmission 0.0050 0.0032 0.0032 0.0032
of 100.0 cm was used for all measurements; and 5 cm of Solid
TJaw Tip Leakage 1.03 1.03 1.03 1.03
Water® Model 457 (Gammex, WI) was placed on top. Results were
MLC Offset (mm) 0.00 −0.05 −0.05 −0.05 normalized to the output on central axis of a 10 × 10 cm2 field,
Leaf Tip Leakage 1.10 1.18 1.18 1.18 measured under the same conditions. Measurements were then
MLC Leakage 0.00 0.00 0.10 0.00 compared to transmission values calculated in the Monaco TPS and
Leaf Groove Width (mm) 0.40 0.40 0.40 1.1 the effect of modifying the Static Leaf Gap TPF parameter was

Underlined italics indicate TPF parameters that have been adjusted from investigated.
their default value.

2.A.2 | Secondary Collimator Transmission

modify each TPF parameter individually, resulting in a significant Once the closed leaf gap was set on the linear accelerator, MLC
number of iterations before the optimal TPF parameters are deter- transmission and diaphragm transmission measurements were per-
mined. More experienced user may be able to modify multiple formed. To measure the leaf bank transmission, the MLCs were
parameters at once, reducing the number of iterations required. closed at 15 cm off axis and a point dose was measured on central
Both the vendor provided fields and a set of additional test fields axis. To differentiate the intraleaf and interleaf transmission, a high‐
were used throughout the optimization. TPF parameters were not resolution profile was measured perpendicular to the direction of
modified if their default value was determined to be appropriate. All MLC travel. The Y diaphragm transmission was measured at a point
measurements were taken at a gantry angle of 0° and a collimator on central axis with the diaphragms closed off axis at −12.5 cm and
angle of 0°, while all fields were calculated in Monaco with a 1 mm with the MLCs parked behind the thickest section of the
grid size and a statistical uncertainty of 0.25% per control point. The diaphragms.11
IEC 61217 geometrical convention is used throughout this paper. Point dose measurements were taken in an MP3 water phantom
(PTW, Freiburg, Germany) using both an FC65‐G Farmer type ioniza-
tion chamber (IBA, Schwarzenbruck, Germany) and a PinPoint 31014
2.A.1 | Minimum Leaf Gap
ionization chamber (PTW, Freiburg, Germany). The long axes of the
The minimum leaf gap or closed leaf gap is defined as the minimum chambers were placed perpendicular to the direction of leaf motion.
allowable separation between opposing leaves. For the Agility™ Relative measurements of the interleaf transmission were performed
476 | ROCHE ET AL.

with Gafchromic EBT3 film in Solid Water®. All measurements were

2.A.5 | Tongue and Groove Effect
taken with an SDD of 100 cm, at a depth of 5 cm and were normal-
ized to the output on central axis of a 10 × 10 cm2 field under the The Agility™ MLCs have no tongue or groove, the leaf sides are flat
same conditions. Measurements were compared to transmission val- with a constant gap of 90 μm between adjacent leaves.11 To reduce
ues calculated in the Monaco TPS and the Leaf Transmission, TJaw interleaf transmission, these gaps are defocused from the x‐ray
Transmission, and Interleaf Leakage TPF parameters were adjusted source with the introduction of an angle in the Agility™ MLCs, creat-
to match the measured values. ing an effective tongue and groove. Two in‐house fields were cre-
ated to determine the effective tongue and groove effect of the

2.A.3 | Leaf Offset Agility™ MLCs. The field shapes in Fig. 6(a) and 6(b) were created so
that the transmission through the effective tongue and groove
The next step in the optimization process is to define the Leaf region could be determined and compared for (a) beamlets of vary-
Offset TPF parameter. The Leaf Offset is described as the differ- ing size and (b) under a range of leaves extended into the field at
ence between the prescribed leaf position, and the actual value multiple locations. These two fields along with the vendor provided
used for dose calculation and should be adjusted to match the FOURL field (Fig. 6(c)) were used to aid in the optimization of the
machine‐specific MLC calibration.13 To identify an appropriate Leaf Groove Width TPF parameter. All measurements were taken at
value for the Leaf Offset, measurements were performed with a an SDD of 100 cm, at a depth of 5 cm in Solid Water® with
Gafchromic EBT3 film in Solid Water® using the 3ABUT vendor Gafchromic EBT3 film. Measurements were compared to values cal-
provided predesigned field. All measurements were taken at an culated in the Monaco TPS under the same conditions and the effect
SDD of 100 cm, at 5 cm depth. The dose at the junctions formed of adjusting the Leaf Groove Width parameters was investigated.
by the leaf tips of each segment was matched to that calculated
in the Monaco TPS by adjusting the Leaf Tip Leakage and Leaf
Offset TPF parameters. The effect of adjusting the MLC Leakage 2.B | Validation of TPF Optimization
parameter was also investigated. To validate the TPF optimization, measured point doses and 3D
dose matrices for a number of clinical IMRT and VMAT plans

2.A.4 | Dosimetric Leaf Gap were compared to those calculated in the TPS. Point dose mea-
surements were made in the IMRT Matrix Phantom T40026
The dosimetric leaf gap (DLG) has been described as the differ- (PTW, Freiburg, Germany) using a 0.125 cc Semiflex 31010 ioniza-
ence between the nominal field width defined by the MLC leaves tion chamber (PTW, Freiburg, Germany). In total, 30 point dose
and the full width half maximum (FWHM) of the dose profile, measurements were made using 10 IMRT and VMAT plans cre-
measured parallel to the direction of leaf motion.16–18 It is possible ated for various anatomical sites. In the TPS, the dose grid resolu-
to measure the DLG using an integral dose method which relates tion was set to 2 mm and the statistical uncertainty was set to
the width of the nominal MLC field to the integral dose of its 3.0% per control point. The ESTRO recommended confidence limit
profile.16 To measure the DLG of the Agility™ MLC, five sliding of ±3% for ion chamber measurements19 was used to identify
window fields with fixed widths from 15 mm to 4 mm were deliv- passing points and measurements were compared to calculations
ered. Point dose measurements were taken in an MP3 water with the default and optimized beam models. Point doses were
phantom using both an FC65‐G Farmer type ionization chamber divided into two categories; low dose and high dose. A low‐dose
and a PinPoint 31014 ionization chamber. All measurements were region was considered to be any region with a dose lower than
taken at an SDD of 100 cm, at a depth of 5 cm and were nor- 50% of the maximum planned dose and a high‐dose region was
malized to the output on central axis of a 10 × 10 cm2 field under considered to be any region with a dose greater than 50%.
the same conditions. Transmission through the MLCs contributing 3D dose matrices were reconstructed with Octavius® 4D (PTW,
to the measured dose was subtracted for each sliding window Freiburg, Germany) from measurements with the Octavius® 1500
field using Eq. (1), detector T10044 (PTW, Freiburg, Germany). The array was placed in
corr
Scp;sw ¼ Scp;sw MLCT ð1 ðw=LÞÞ (1) the Octavius® 4D rotational phantom and all fields were delivered
with the planned gantry angles. The Octavius® 4D rotational phan-
corr
where Scp;sw is the corrected sliding window total scatter factor, tom was modeled in the TPS as a cylindrical phantom with a uniform
Scp;sw is the uncorrected sliding window total scatter factor, MLCT density using the CT dataset supplied by PTW. The relative electron
is the MLC leaf bank transmission measured in section 2.A.2, w is density (RED) of the Octavius® phantom was set to 1.016. The sta-
the sliding window width, and L is the length of the dynamic field, tistical uncertainty and dose grid size were set to 3.0% per control
which in this case was 100 mm. The dose under a 10 mm and a point and 2 mm, respectively, for all dose calculations. Dose distribu-
20 mm sliding window field, measured under the same conditions, tions were analyzed using VeriSoft v.7.0.1.30 (PTW, Freiburg, Ger-
was also matched to that calculated in the Monaco TPS. This was many) using a gamma20 criteria of both 3% global dose, 3 mm
possible by adjusting the Leaf Tip Leakage and MLC Offset TPF distance‐to‐agreement, and 2% global dose, 2 mm distance‐to‐agree-
parameters. ment where global dose is defined as the maximum dose in the
ROCHE ET AL. | 477

entire analyzed volume. Again, measurements were compared to cal- leaf gap does. This is due to the leaf design which results in the reduc-
culations with both the default and optimized beam models. Gamma tion of the transmission through the closed leaf gap as illustrated in
analysis results for the 3D dose volume greater than 50% of the Fig. 2(c). The Monaco TPS successfully modeled the reduction in the
maximum delivered dose are displayed. radiation transmission when the closed leaf gap is moved off axis.
However, if the SS does not physically close the MLCs, it is likely that
the TPS will overestimate the transmission off axis under very small
3 | RESULTS AND DISCUSSION leaf gaps. No clinically significant difference in modeling the closed
leaf gap was identified between the default and optimized beam
Due to the many small beamlets created by the Monaco TPS in com- model.
plex IMRT plans, it is important that the closed leaf gap on the linear Fig. 3 illustrates the measured MLC transmission compared to
accelerator is appropriately set and that the TPS correctly models this that calculated in the Monaco TPS. The transmission under the X1
behavior. Initially, the closed leaf gap measured with the feeler gauge and X2 leaf bank was measured to be 0.60 ± 0.02% (k = 2) with the
was 1.45 ± 0.04 mm (k = 2) at the leaf bank plane. The physical leaf FC65‐G ionization chamber and 0.58 ± 0.02% (k = 2) with the Pin-
gap was reduced to 0.8 ± 0.04 mm (k = 2) so that the measured Point ionization chamber. The transmission measured with the
transmission matched the Monaco TPS. The closed leaf gap was not FC65‐G ionization chamber was higher as the 23.1 mm length of the
reduced below 0.8 mm due to the increased likelihood of MLC colli- sensitive volume sampled both the intraleaf and interleaf transmis-
sions. As shown in Fig. 2(b), which illustrates the relative transmission sion. The Leaf Transmission and Interleaf Leakage TPF parameters
through the closed leaf gap at isocenter, even with the new closed were iteratively adjusted to match the film and ionization chamber
leaf gap the Monaco TPS underestimated the transmission under the measurements. The transmission under the Y1 and Y2 jaw were
closed leaves by nearly 20%. The Static Leaf Gap TPF parameter was measured to be 0.38% ± 0.02% (k = 2) and the TJaw Transmission
increased from its default value of 0.1 mm; however, no change in TPF parameter was adjusted to match the measurement. It should
the transmission under the closed leaves was observed. This is be noted that the TPF does not modify particle energies for both
because the Static Leaf Gap parameter is used by the Monaco Static the MLC and jaw transmission. As a result, when the individual TPF
MLC Sequencer (SS) and does not affect the transmission through the parameters are specified using the above methodology, the com-
TPF.21
Although the closed leaf gap does not physically change as the bined transmission is not truly the product of the two filters. Adjust-
MLCs move off axis, the radiation transmission through the closed ments of the TJaw Transmission, Leaf Transmission and Interleaf

F I G . 2 . Transmission through the closed leaf gap compared to the Monaco TPS: (a) MLC positioning for the closed leaf gap field, (b) profiles
of the transmission through the closed leaf gap at isocenter, and (c) variation in the transmission through the closed leaf gap with off axis
distance.
478 | ROCHE ET AL.

FIG. 3. Leaf transmission profile measured perpendicular to the direction of leaf travel, compared to profiles calculated in the Monaco TPS.

F I G . 4 . Dose distributions calculated and measured with the 3ABUT vendor test beam: (a) measured dose distribution, (b)–(d) profiles from
the relevant dashed lines in (a). The data labels in (d) apply to (b) and (c).

Leakage TPF parameters were not required for the remainder of the The MLC Offset and Leaf Tip Leakage TPF parameters were
optimization process as the calculated transmission did not change then iteratively adjusted to match the measured and calculated DLG
significantly while adjusting other TPF parameters. and vendor provided 3ABUT field. Profile measurements through
ROCHE ET AL. | 479

F I G . 5 . Relative dose distributions resulting from DLG measurements and calculations: (a) integral dose for sliding window fields of
decreasing window width, (b) profile taken parallel to the direction of MLC motion under the 1 cm sliding window field, static at isocenter.

various regions of the 3ABUT field are illustrated in Fig. 4, while the 4(d) show measured and calculated profiles through the 3ABUT field.
results from the DLG measurements are illustrated in Fig. 5. Fig. 5(a) As illustrated by the results of Model A, the measured leaf end cor-
illustrates the integral dose for each sliding window decreasing lin- ner transmission was not calculated appropriately by the TPS for the
early with a reduction in the gap width. Extrapolating a linear regres- fully optimized model. This is because no adjustments were made to
sion fit of the data reveals the gap width corresponding to zero the MLC Leakage TPF parameter. Figures 4(b), 4(c) and 4(d) also
charge measurement. Using this methodology, the DLG was mea- show the results of adjusting the MLC Leakage TPF parameter to
sured to be 0.01 ± 0.18 mm (k = 2) and 0.03 ± 0.16 mm (k = 2) 0.10 (Model B). A sharp increase in the dose in the MLC tip region
with the FC65‐G and Pinpoint ionization chambers respectively. is illustrated, with a negligible improvement in the calculation of leaf
However, when the sliding window fields were calculated in the TPS end corner transmission. These results were attributed to the rela-
with the default model, a DLG of 0.30 ± 0.08 mm (k = 2) was calcu- tively large size of the MLC Leakage region in the TPF ((0.5 mm ×
lated. To investigate this discrepancy, the 1 cm sliding window field 2.0 mm) × 2) and as a result, adjustments to the MLC Leakage
was measured under static conditions at isocenter. A profile from parameter are not recommended. To replicate the leaf end corner
this measurement, parallel to the direction of MLC motion, is illus- leakage of the Agility™ MLC with the current TPF, the MLC Leakage
trated in Fig. 5(b). The default model underestimated the dose for parameter would have to be increased further and the MLC Offset
the static field off axis between 7.5 cm to 15.0 cm. To correct this parameter decreased significantly. It is expected that replicating the
in the TPS, the Leaf Tip Leakage TPF parameter was increased. At leaf end corner leakage in this manner will affect small field output
the same time, the MLC Offset TPF parameter was reduced to factors and degrade clinical plan results. However, in future versions
ensure a continued match with the 3ABUT field. Figs. 4(b), 4(c), and of Monaco, an adjustment to the size of the MLC Leakage TPF
480 | ROCHE ET AL.

F I G . 6 . In‐house and vendor provided

fields created to determine the effective
tongue and groove of the Agility™ MLCs:
(a)–(b) measured dose distribution of in‐
house created fields, (c) measured dose
distribution of vendor provided FOURL
field, (d) profiles from the relevant dashed
lines in (a), (e) profiles from the relevant
dashed lines in (b), and (f) profiles from the
relevant dashed lines in (c). The data labels
in (f) apply to (d) and (e).

region may allow for increased accuracy in the calculation of the leaf with measured transmission on the Agility™ MLC. This modification
end corner transmission. It is estimated that a MLC Leakage TPF will likely require the introduction of additional TPF parameters to
region of 0.5 mm × 1.0 mm to replace the current 0.5 mm × 2.0 mm model the transmission in this region.
region would be sufficient to aid with modeling the leaf end corner
transmission on the Agility™ MLC.
Fig. 6 illustrates the two in‐house fields along with the vendor
provided FOURL field used to aid in the optimization of the Leaf
Groove Width TPF parameter. A selection of dose distributions
resulting from the three test fields are also displayed in Fig. 6(d)–(f).
The Leaf Groove Width TPF parameter was adjusted from its default
value of 0.40 to 1.1 to match the dose at the effective tongue and
groove junctions of the vendor provided FOURL field. Model C in
Figs. 6(d), 6(e) and 6(f) illustrates the results of this adjustment.
Fig. 6(d) shows a reduction in dose of approximately 7% under the
1 cm × 3 cm beamlet, while the reduction in dose under the
0.5 cm × 3 cm beamlet was approximately 30%. This reduction in
dose is not desirable and is due to a decrease in the width of the
beamlets as the Leaf Groove Width parameter is increased. Similar
reductions in dose were measured under the 1 cm × 2 cm and
0.5 cm × 2 cm beamlets. Fig. 6(e) illustrates the agreement of the
TPS with measurements across the effective tongue and groove of
the Agility™ MLC. None of the TPS models had the ability to model
the effective tongue and groove accurately; however, Model C pro-
duced the poorest results. As a result, during the optimization of the
TPF no adjustment to the Leaf Groove Width TPF parameter is rec-
ommended. It is also recommended that the vendor provided
FOURL field is not used to match the dose at the effective tongue F I G . 7 . Whisker and box plots of the difference between calculated
and measured point doses (a) high‐dose region and (b) low‐dose
and groove junctions. The limited accuracy of the TPF in replicating
region. “+” illustrates the mean, the inner quartile range (IQR) is
the effective tongue and groove of the Agility™ MLC was noted. A illustrated by the boxes, and the whiskers illustrate the 1.5*IQR. Open
modification to the modeling of the transmission in the TPF through circles are used to display any points outside 1.5*IQR. The TPF
the tongue and groove region is required to create better agreement parameters for each model are displayed in Table 1.
ROCHE ET AL. | 481

through the tongue and groove region is required before better

agreement with the effective tongue and groove on the Agility™
MLC can be obtained.
6. Verify the accuracy of the optimized beam model by creating
several different IMRT and VMAT plans for sites that will be
treated clinically. Follow a rigorous plan specific quality control
procedure.

4 | CONCLUSION

F I G . 8 . Whisker and box plots of the gamma pass rates (2% global Throughout this study, the ability of the Monaco TPS to model the
dose/2 mm DTA) of the plans for the default and optimized beam transmission through the MLC and jaws of the Agility™ MLC was
model. “+” illustrates the mean, the inner quartile range (IQR) is investigated. Although direct simulation of particles through the
illustrated by the boxes, and the whiskers illustrate the 1.5*IQR.
beam modifiers would be the most accurate method to achieve this,
Open circles are used to display any points outside 1.5*IQR. The
calculation time limitations currently make this impractical. The use
TPF parameters for each model are displayed in Table 1.
of an optimized TPF to model the transmission has been shown to
Once the final optimized model was created, measured point achieve good clinical results for both IMRT and VMAT treatment
doses and 3D dose matrices for a number of clinical IMRT and techniques. The effect of relevant TPF parameters has been pro-
VMAT plans were compared to those calculated in the TPS. Figs. 7(a) vided along with a set of additional test fields which will simplify
and 7(b) illustrate the point dose measurement results for the clinical and improve the TPF optimization process. Although it is possible to
plans. A significant improvement in the number of points passing the achieve good clinical results by randomly selecting TPF parameter
ESTRO recommended confidence limit of ±3% for ion chamber mea- values, it is recommended that the optimization process outlined in
surements was recorded with the optimized beam model. This this study is followed so that the transmission through the TPF is
improvement was particularly noticeable in the low‐dose regions of characterized appropriately. To improve calculation accuracy, poten-
the clinical plans. Fig. 8 illustrates the gamma pass rates for the plans tial future revision of the TPF may look at reducing the size of the
® leaf end corners in the TPF and providing additional TPF parameters
measured with Octavius 4D using the gamma criteria of 2% global
dose, 2 mm DTA with a 10% threshold. All plans calculated with the so that the effective tongue and groove on Agility™ MLC can be
optimized beam model had a gamma pass rate of >95%, while all plans accurately characterized.
calculated with the default beam model had lower gamma pass rates.
For a gamma criteria of 3%/3 mm, there was an insignificant difference
ACKNOWLEDGMENTS
in the pass rates for the default and optimized beam models.
The authors acknowledge the support and funding of the Townsville
Hospital Health Service as well as Dr. Louis Fourie and the staff at
3.A | TPF optimization process the Townsville Cancer Centre, Queensland, Australia.

The below procedure is recommended for the optimization of the

TPF in the Monaco TPS, CONFLICT OF INTEREST

1. Start the optimization process with the default model. No conflicts of interest.
2. Measure and adjust the closed leaf gap on the Agility™ MLC to
match calculations from the Monaco TPS. Be aware that reducing REFERENCES
the closed leaf gap below 1 mm (at the leaf bank plane) can
1. Webb S. The physical basis of IMRT and inverse planning [Electronic
increase the likelihood of MLC collisions. Version]. Br J Radiol. 2003;76(910):678–89.
3. Iteratively adjust the Leaf Transmission, TJaw Transmission and 2. Yu CX, Tang G. Intensity‐modulated arc therapy: principles, technolo-
Interleaf Leakage TPF parameters to match the measured trans- gies and clinical implementation [Electronic Version]. Phys Med Biol.
mission values. 2011;56(5):R31–54.
3. Earl MA, Shepard DM, Naqvi S, Li XA, Yu CX. Inverse planning for
4. Iteratively adjust the Leaf Tip Leakage and MLC Offset TPF
intensity‐modulated arc therapy using direct aperture optimization
parameters to match the DLG and 3ABUT vendor provided field. [Electronic Version]. Phys Med Biol. 2003;48(8):1075–89.
The MLC Leakage TPF parameter may be adjusted, if in future ver- 4. Helyer SJ, Heisig S. Multileaf collimation versus conventional shield-
sions of Monaco the size of the affected TPF regions are reduced. ing blocks: A time and motion study of beam shaping in radiotherapy
[Electronic Version]. Radiother Oncol. 1995;37(1):61–64.
5. In the current version of Monaco (5.11.01), no adjustment to
5. Fernandez EM, Shentall GS, Mayles WPM, Dearnaley DP. The
the Leaf Groove Width TPF parameter is recommended. An acceptability of a multileaf collimator as a replacement for conven-
improvement in the ability of the TPF to model the transmission tional blocks [Electronic Version]. Radiother Oncol. 1995;36(1):65–74.
482 | ROCHE ET AL.

6. Sikora M, Dohm O, Alber M. A virtual photon source model of an 14. Snyder M, Halford R, Knill C, et al. Modeling the Agility MLC in the
Elekta linear accelerator with integrated mini MLC for Monte Carlo Monaco treatment planning system [Electronic Version]. JACMP.
based IMRT dose calculation [Electronic Version]. Phys Med Biol. 2016;17(3):190–202.
2007;52(15):4449–63. 15. Elekta AB. Integrity™ R3.0 for Agility™ Software Upgrade, Installa-
7. Sikora M, Alber M. A virtual source model of electron contamination tion & Configuration Information. Document ID: 1016013 03. 2015.
of a therapeutic photon beam [Electronic Version]. Phys Med Biol. 16. Mullins J, DeBlois F, Syme A. Experimental characterization of the
2009;54(24):7329–44. dosimetric leaf gap [Electronic Version]. Biomed Phys Eng Express.
8. Sikora MP. Virtual source modelling of photon beams for Monte 2016;2(6):1–8.
Carlo based radiation therapy treatment planning [PhD thesis]. Ber- 17. Yao W, Farr JB. Determining the optimal dosimetric leaf gap
gen, Norway: Universitas Bergensis; 2011. setting for rounded leaf‐end multileaf collimator systems by
9. IMPAC Medical Systems Inc. Monaco dose calculation technical ref- simple test fields [Electronic Version]. JACMP. 2014;16
erence. Document ID: LRMMON0001. 2017. (4):65–77.
10. Fippel M. Fast Monte Carlo dose calculation for photon beams 18. Vial P, Oliver L, Greer PB, Baldock C. An experimental investigation
based on the VMC electron algorithm. Med Phys. 1999;26(8):1466– into the radiation field offset of a dynamic multileaf collimator [Elec-
75. tronic Version]. Phys Med Biol. 2006;51(21):5517–5538.
11. Thompson CM, Weston SJ, Cosgrove VC, Thwaites DI. A dosimetric 19. Alber M, Brogi S, De Wagter C, et al. Guidelines for the Verification of
characterization of a novel linear accelerator collimator. Med Phys. IMRT [Electronic Version]. Belgium: ESTRO; 2008.
2014;41(3):031713. 20. Low DA, Harms WB, Mutic S, Purdy JA. A technique for the quanti-
12. Elekta AB. Agility™ and Integrity™ R3.x information for treatment tative evaluation of dose distributions. Med Phys. 1998;25(5):656–
planning systems. Document ID: 1504231 01. 2017. 61.
13. IMPAC Medical Systems Inc. Monaco post modeling adjustment of 21. IMPAC Medical Systems Inc. Monaco Static MLC Sequencer Techni-
MLC parameters. Document ID: LRMMON0003. 2013. cal Reference. Document ID: LRMMON0004. 2014.