Home
Search
Collections
Journals
About
Contact us
My IOPscience
Experimentally validated pencil beam scanning source model in TOPAS
This content has been downloaded from IOPscience. Please scroll down to see the full text. 2014 Phys. Med. Biol. 59 6859 (http://iopscience.iop.org/0031-9155/59/22/6859) View the table of contents for this issue, or go to the journal homepage for more Download details: IP Address: 128.59.222.107 This content was downloaded on 24/07/2017 at 13:50 Please note that terms and conditions apply.
You may also be interested in: Experimental characterization of two-dimensional spot profiles for two proton pencil beam scanning nozzles Liyong Lin, Christopher G Ainsley, Timothy D Solberg et al. Experimental characterization of two-dimensional pencil beam scanning proton spot profiles Liyong Lin, Christopher G Ainsley and James E McDonough A novel technique for measuring the low-dose envelope of pencil-beam scanning spot profiles Liyong Lin, Christopher G Ainsley, Thierry Mertens et al. Validation of nuclear models in Geant4 using the dose distribution of a 177 MeV proton pencil beam David C Hall, Anastasia Makarova, Harald Paganetti et al. Monte Carlo study of the potential reduction in out-of-field dose using a patient-specific aperture in pencil beam scanning proton therapy Stephen J Dowdell, Benjamin Clasie, Nicolas Depauw et al. A Monte Carlo PBS model for proton treatment plan simulation L Grevillot, D Bertrand, F Dessy et al. Experimental characterization of the low-dose envelope of spot scanning proton beams Gabriel O Sawakuchi, X Ronald Zhu, Falk Poenisch et al.
Institute of Physics and Engineering in Medicine Phys. Med. Biol. 59 (2014) 6859–6873
Physics in Medicine & Biology doi:10.1088/0031-9155/59/22/6859
Experimentally validated pencil beam scanning source model in TOPAS Liyong Lin, Minglei Kang, Timothy D Solberg, Christopher G Ainsley and James E McDonough Department of Radiation Oncology, University of Pennsylvania, Philadelphia, PA, USA E-mail: [email protected] Received 15 May 2014, revised 8 September 2014 Accepted for publication 2 October 2014 Published 28 October 2014 Abstract
The presence of a low-dose envelope, or ‘halo’, in the fluence profile of a proton spot can increase the output of a pencil beam scanning field by over 10%. This study evaluated whether the Monte Carlo simulation code, TOPAS 1.0-beta 8, based on Geant4.9.6 with its default physics list, can predict the spot halo at depth in phantom by incorporating a halo model within the proton source distribution. Proton sources were modelled using three 2D Gaussian functions, and optimized until simulated spot profiles matched measurements at the phantom surface out to a radius of 100 mm. Simulations were subsequently compared with profiles measured using EBT3 film in Solidwater® phantoms at various depths for 100, 115, 150, 180, 210 and 225 MeV proton beams. Simulations predict measured profiles within a 1 mm distance to agreement for 2D profiles extending to the 0.1% isodose, and within 1 mm/1% Gamma criteria over the integrated curve of spot profile as a function of radius. For isodose lines beyond 0.1% of the central spot dose, the simulated primary spot sigma is smaller than the measurement by up to 15%, and can differ by over 1 mm. The choice of particle interaction algorithm and phantom material were found to cause ~1 mm range uncertainty, a maximal 5% (0.3 mm) difference in spot sigma, and maximal 1 mm and ~2 mm distance to agreement in isodoses above and below the 0.1% level, respectively. Based on these observations, therefore, the selection of physics model and the application of Solidwater® as water replacement material in simulation and measurement should be used with caution. Keywords: Monte Carlo simulation, proton therapy, pencil beam scanning, treatment planning, Solidwater® (Some figures may appear in colour only in the online journal) 0031-9155/14/226859+15$33.00 © 2014 Institute of Physics and Engineering in Medicine Printed in the UK & the USA
6859
L Lin et al
Phys. Med. Biol. 59 (2014) 6859
1. Introduction Compared to analytical-algorithm-based treatment planning systems (TPSs) (Lomax et al 2004), Monte Carlo based TPSs (Tourovsky et al 2005) are superior in modelling the dose distribution around complex density interfaces, especially for pencil beam scanning (PBS) treatments. PBS treatments deliver proton spots in multiple energy layers. The number of spots in a PBS field can vary from less than one thousand to greater than ten thousand. Therefore PBS treatments require accurate characterization and modelling of spot profiles within the TPS. Spot profiles have been described by Pedroni et al (2005) as a primary Gaussian distribution followed by a broad non-Gaussian tail or ‘halo’. Soukup et al (2005) compared spot profiles and depth doses of Bragg peaks in a water phantom calculated by a pencil beam algorithm to those simulated by Geant4 (Agostinelli et al 2003) and concluded that pencil beam algorithms can achieve a 1 mm/3% dose accuracy in homogenous and simple heterogeneous phantoms but not in sophisticated clinical scenarios. Paganetti et al (2004) and Kimstrand et al (2007) described Geant4 simulations of a double-scattering nozzle and an experimental PBS nozzle, respectively. For the dedicated PBS nozzle developed by Ion Beam Applications (Louvain-la-Neuve, Belgium), Grevillot et al (2010) found good agreement in the Bragg peak depth dose between Geant4 simulation and measurement, but observed up to ~20% underestimation of the primary spot sigma in simulation compared with measurement in a PMMA phantom. Grevillot et al also observed differences of up to 15% in the primary spot sigma generated by different simulation platforms (MCNPX, PHITS, Szymanowski) (Grevillot et al 2010). They reported that MCNPX tended to overestimate the spot sigma while PHITS appeared to be the most accurate. Subsequently, Sawakuchi et al (2010a, 2010b) reported a more accurately calculated spot sigma by incorporating more large-angle single scattering events in a newer Multiple Coulomb Scattering algorithm in MCNPX. To study profiles extending beyond the primary Gaussian component of the spot, Sawakuchi et al (2010b) used MCNPX to simulate the contribution of nuclear interactions within the nozzle to spot profiles in a water phantom; simulation results were compared to measurement for the Hitachi ProBeat delivery system (Hitachi, Ltd, Tokyo, Japan). Similarly, Clasie et al (2012) studied the contribution of nuclear interactions to the integral Bragg peak in a water phantom and was able to achieve good agreement with measurement by adjusting the parameters within Geant4. Thus, Monte Carlo simulation is a critical tool in developing approximation methods for TPSs and standalone dose calculation engines (Gillin et al 2009, Grevillot et al 2011, Li et al 2012, Zhu et al 2013). The model of Sawakuchi et al (2010b) simulates the beam beginning at the middle of two scanning magnets using the source’s Gaussian sigmas as free parameters. In contrast, we investigate the feasibility of modelling a proton source located at the phantom surface using 2D Gaussian parameters extracted from measured spot profiles in air. Such an approach does not require modelling of complex particle transport within the IBA dedicated PBS nozzle that, unlike the Hitachi nozzle, has two upstream focusing quadruples (Farr et al 2013). Also, starting particle transport from the phantom surface minimizes the uncertainty from the free parameters used in modelling of the angular spread distribution otherwise needed to derive downstream in-air spot profiles. Furthermore, starting Monte Carlo simulations using a proton source distribution that matches simulated and measured spot profiles at the phantom surface for both the primary and tail regions, allows any differences to be separated into two components: inaccurate source modelling from the treatment nozzle and incorrect particle interactions within the phantom. Rather than focusing solely on the Gaussian sigma of the primary component of the spot profile, this investigation looks closely at the low-dose envelope, or 6860
L Lin et al
Phys. Med. Biol. 59 (2014) 6859
‘halo’, simulated in phantom by TOPASwith the initial halo based on in-air dose measurements. At depth in phantom, the halo distribution originates predominately from the treatment nozzle for low energy proton beams. Therefore the determination of the halo source distribution is crucial for accurate characterization of spot profiles for low energy proton beams. In contrast, for high energy proton beams the halo distribution is predominantly generated from particle interaction in the phantom. Recent progress has allowed experimental determination of the halo component (Lin et al 2013a, 2013b , 2014).This study attempts to determine whether the combination of a good proton source model coupled with an appropriate particle interaction algorithm can predict the halo distribution to a clinically acceptable level for selected depths and proton energies. While the dose levels below 1% of the central spot dose may be insignificant for a single spot, the summation of the broad tails from multiple spots can affect a typical field’s overall output by up to 4% and 7% for the IBA dedicated and universal nozzles, respectively, for a 100 MeV proton beam. Similar for all nozzles, the output can increase by over 10% for a 225 MeV proton beam (Gillin et al 2009, Sawakuchi et al 2010a, 2010b, Grevillot et al 2011, Clasie et al 2012, Li et al 2012, Zhu et al 2013, Lin et al 2013a, 2013b, 2014). Accounting for this phenomenon requires an accurate PBS source model and robust particle interaction algorithms to characterize spot tails over the region from 1% to 0.01% of the central spot dose, which may extend spatially outwards to 100 mm radius from the spot centre. Here we study if it is feasible to incorporate the measurement-based halo emanating from the treatment nozzle into a Monte Carlo TPS, neglecting the perturbation caused by in-air angular spread. Although this approach cannot differentiate the origin of the spatial distribution or calculate the energy spectrum of the halo from a treatment nozzle, it can test if a Monte Carlo TPS is accurate enough for clinical dose distribution calculation by reporting the residual disagreement between simulation and measurement from the phantom surface to the end of proton range. 2. Methods and materials Spot profiles were measured in a Solidwater® phantom (Gammex, Inc., Wisconsin, USA) using Gafchromic EBT3 film (Ashland Specialty Ingredients, New Jersey, USA) for six selected energies: 100, 115, 150, 180, 210 and 225 MeV from the IBA dedicated PBS nozzle. The water equivalent thickness (WET) of each Solidwater® phantom was experimentally determined to be 1.03 with an uncertainty of 0.5–1% (see Discussion). Within the Monte Carlo simulation, Solidwater® was modelled according to the vendor-supplied specification as a mixture of six elements, ‘hydrogen’, ‘carbon’, ‘nitrogen’, ‘oxygen’, ‘chlorine’ and ‘calcium’, in weight fractions of 0.0809, 0.6717, 0.0241, 0.1988, 0.0014 and 0.0231, respectively, with a density of 1.044 g cm−3. In contrast to the PMMA used by Grevillot et al (2010), Solidwater® phantoms are widely used in proton therapy quality assurance (Arjomandy et al 2008) in combination with film and ionization chamber arrays. Simulations were performed in both water and Solidwater® phantoms, however, the simulation results are reported for the former because water is the reference material for clinical dosimetry. The equivalence of water and Solidwater® in terms of lateral scattering properties and the validity of the WET ratio along the depth direction were investigated for selected scenarios in this study. Phantom surface locations at isocentre and 270 mm upstream of isocentre were selected to encompass the range of patient surface locations observed in clinical practice. Films were placed in the phantom at 50 mm increments to the end of the range, whereupon two films were placed with a 5 mm spacing. The pair/magnification method (Farr et al 2013, Lin et al 2013a, 6861
L Lin et al
Phys. Med. Biol. 59 (2014) 6859
Figure 1. The process to derive from measurement the 2D Gaussian parameters of
the proton source model in TOPAS. The top left panel shows the isodose contours of the measured 150 MeV proton spot profile of the IBA dedicated PBS nozzle in air at isocentre. The bottom right panel shows the 1D comparison of integrated curve of spot profiles versus radius between TOPAS simulation (S0) and measurement (M0). The bottom left panel shows the 2D profile comparison between simulation (solid lines) and measurement (dashed lines) and lists the average primary Gaussian sigma derived from simulation and measurement. The isodose contours are 50% (cyan), 10% (black), 1% (red), 0.1% (green) and 0.01% (blue) of the central dose.
2013b) was applied to determine the composite 2D spot profiles in air and in phantom for the selected energies. The resulting kernels extend over an area of 200 mm × 200 mm and were constructed with a 0.5 mm resolution. To calculate the fractional energy collected as a function of radius, 2D asymmetric dose kernels were converted into symmetric kernels and used to integrate spot profiles outwards from the centre (Lin et al 2013, 2014). The process of extracting 2D Gaussian parameters from the measured profiles extending to the level of 10–4 of the central dose and to 100 mm in radius is described in figure 1. Three Gaussian functions are chosen to express the broad tail of the proton source model up to 100 mm radius, as there is only ~1% contribution to the integral dose beyond 3σ of the primary Gaussian. The first primary Gaussian is chosen to be almost circular and the second Gaussian typically characterizes the elongation of the tail. If the isodose at 0.01% of the central spot dose does not follow the elongation of the isodose at 0.1% of the central spot dose, the model is further modified by the third Gaussian. For proton energies above 115 MeV, a single Gaussian distribution is used to model the energy spectrum of the spot profiles, while at 115 MeV and below, three slightly different Gaussian energy spreads are used, corresponding to the three spatial spot profile Gaussian functions (see Discussion). The three Gaussian source models were independently simulated in TOPAS 1.0-beta 8 (Perl et al 2012), a simulation tool based on Geant4.9.6, and the weighted sum of the simulation results from the three Gaussian sources was performed using an in-house Matlab program. Twenty million particle histories were run for each of the three sources utilizing 10% of the central spot dose), only the protons from the first Gaussian functions are used to study the average proton energy in the phantom for the 115, 150 and 210 MeV beams. The primary proton energies are averaged in 0.5 mm radial increments. Figure 6 shows that proton energies in the primary spot regions are larger than 40 MeV for most depths; near the end of range, however, the energy falls below 30 MeV for the 115 MeV beams and below 20 MeV for the 150 and 210 MeV beams. The smaller disagreement in spot sigma between simulation and measurement near the end of range compared to other depths is caused by quenching effects of Gafchromic EBT3 film, which results in the spot size measured with film being smaller than reality for proton energies below 40 MeV. Beyond primary spot regions, the average proton energy changes gradually or 6867
L Lin et al
Phys. Med. Biol. 59 (2014) 6859
Figure 5. Comparison of the curves of integrated spot profile versus radius up to 100 mm between simulation and measurement in Solidwater® phantom at depths of 50, 90 and 95 mm for the 115 MeV proton beam; at depths of 50, 100, 150 and 155 mm for the 150 MeV beam; and at depths of 50, 100, 200, 270 and 275 mm for the 210 MeV beam. The phantom surface is at isocentre. The red ‘+’ and blue ‘x’ markers represent simulation and measurement, respectively.
stays nearly constant as a function of radius within the ‘shoulder’ region, which is widest at the depths between half and two-thirds of the proton range of the highest energy (210 MeV) proton beam. The statistical error is within ~2% for the isodoses >0.1% and 7% for the 0.01% isodose level. The curve of average proton energy is plotted beyond the 0.01% isodose with dash lines and cut off when the number of protons drops below 20. In the methods section, we presented an approach to derive the weight and sigma of the 2D Gaussian functions for the tail components. In figure 7, we present a method to determine whether the tail components have a lower energy than the primary component by comparing in-air spot profiles on the central axis to those off axis. We hypothesize that Landau fluctuations (Ulmer 2007) in the energy lost in the degrader and the nozzle generate lower energy protons that are subsequently deflected more than the average distance by the ratio of the momentum of the tail components to that of the primary component. When the proton spots are deflected 100 mm away from the x and y axes, no difference is observed between the central axis and the off-axis locations for the 210 MeV proton beam (figure 7(a)). In contrast, for the 115 MeV proton beam, up to 2 mm differences are observed at the far end of the y-axis of the diamond-shaped 0.01% isodose contour (figure 7(c)). As the y-axis is the focusing direction of the downstream quadruple, we speculate that the low energy protons are preferentially located there due to imperfect focusing. Here we construct a hypothetical spectrum (linear scale in figure 7(e) and log scale in figure 7(f)) to account for the approximately 4% lower energy (i.e. ~2% lower momentum) as reflected in 2% more scanning distance in figure 7(c). To approximate the low energy Landau tails the energy spectrum is expressed as the summation of three Gaussian functions centred at 115 MeV, an intermediate energy, and 110 MeV with the corresponding weights (95.5%, 3% and 1.5%) of the three spatial Gaussian functions in table 1. 6868
L Lin et al
Phys. Med. Biol. 59 (2014) 6859
Figure 6. Average simulated proton energy versus radius from the spot centre in a Solidwater® phantom at depths of 50, 90 and 95 mm for the 115 MeV proton beam; at depths of 50, 100, 150 and 155 mm for the 150 MeV beam; and at depths of 50, 100, 200, 270 and 275 mm for the 210 MeV beam. The numbers on the upper right corner of the plots are the number of primary protons at this depth for the beam. The dash part of the curve is beyond isodose of 0.01%.
Wider energy spreads are used for the lower energy components to achieve a smooth energy spectrum. Subsequent simulation demonstrates that adopting such an energy spectrum within the source model can improve the depth to agreement by 5–6 mm for isodoses below 0.1% at a depth of 90 mm near the end of range of the 115 MeV proton beam. Without such a spectrum, simulated isodoses below 0.1% of the central dose would be broader than measurement. The clinical impact of primary and tail components in the source distributions are investigated for the Bragg peaks of the 115 and 210 MeV proton beams. The low energy tail components cause approximately 2% loss in the peak value of the 115 MeV Bragg peak (figure 7(d)). The energy spread is iteratively determined to be 0.6% and 0.35% to match the measured shape of Bragg peaks as closely as possible for 115 MeV and 210 MeV proton beams, respectively. The energy spread data presented here are similar to those reported by Grevillot et al (2010) for similar energies in the IBA dedicated PBS nozzle in Essen, Germany. At 210 MeV, there is no observable difference between the simulated Bragg peaks that include the primary component only and those that have primary and tail components. Note there is maximal 1.5 mm range difference between simulated and measured Bragg peaks and that we have shifted the simulated Bragg peaks to simplify comparison. Figure 8 compares profiles at various depths for 115 and 210 MeV proton beams in Solidwater® and water, and between the two electromagnetic physics models of TOPAS. A WET ratio of 1.03 is used to convert the depths in Solidwater® to those in water; the uncertainty in the WET ratio is presented in figure 9. For all depths of the 115 MeV proton beam (figure 8(a)), Solidwater® yields a smaller spot than water and G4EmstandardPhysics option 4 predicts a larger spot than G4EmstandardPhysics option 3 with ~0.1 mm difference of spot sigma. For the studied depths of the 210 MeV proton beam (figure 8(b)), Solidwater® 6869
L Lin et al
Phys. Med. Biol. 59 (2014) 6859
Figure 7. Comparison of two 2D spot profiles and depth doses of 210 MeV (a), (b) and 115 MeV (c), (d) Bragg peaks (p, t1, t2 stands for the primary, secondary and tertiary Gaussian contributions, respectively). For each energy, one spot profiles were measured in air at the location that is +100 mm and −100 mm away from x and y axes, and the other was measured on the central axis of the isocentre plane. The isodose contours are 50% (cyan), 10% (black), 1% (red), 0.1% (green) and 0.01% (blue) of the central dose. The solid lines are for the off-axis location and the dashed lines are for the central axis location. The inferred initial energy spectrum of the 115 MeV proton beam is displayed in (e) linear and (f) log scales from 105 to 125 MeV.
produces smaller primary spots than water, and G4EmstandardPhysics option 4 generates larger primary spots than the default G4EmstandardPhysics option 3 with up to 5% (0.3 mm) difference. The difference of simulated isodoses between Solidwater® and water phantoms is smaller than 1 mm at all dose levels above 0.1% but can exceed 1 mm below the 0.1% level for the 210 MeV proton beam. Because of the low dose gradient of the integral curves at such large radii, the dose difference reflected in the integral curve would be less than 1% (see fi gure 5). On the other hand, the difference of simulated isodoses between G4EmstandardPhysics option 4 and G4EmstandardPhysics option 3 was smaller than 1 mm DTA above 0.1% and ~2 mm DTA at the 0.01% isodose level for most depths except for 270 mm, where the difference of isodose contours below 0.1% reached ~2 mm DTA potentially due to a change of the lateral scattering mechanism and range uncertainty caused by the different electromagnetic physics models used [8–10, 27]. Another potential reason for the small difference of spot profile between Solidwater® and water is that the geometrical length of Solidwater® is smaller than water and the initial beam divergence, which is on the order of 3 mrad, was neglected. A WET ratio of 1.03 approximates Solidwater® to water along the depth direction. As this ratio may change with proton energy and causes uncertainty in the corresponding WET in simulation, figure 9 compares the Bragg peaks simulated in Solidwater® and water for 115 and 210 MeV proton beams. The scoring volumes use a radius of 10 cm to ensure collection of all the deposited energy. The Bragg peaks in water fall within the scaled Bragg peaks in 6870
L Lin et al
Phys. Med. Biol. 59 (2014) 6859
Figure 8. Comparison of simulated profiles in water and Solidwater® (top) and between the G4EmstandardPhysics option 3 (default physics list of TOPAS) and option 4 (bottom), for (a) 115 MeV and (b) 210 MeV proton beams at the depths (in mm) indicated in the bottom, right of the panels. SW stands for Solidwater®, W for Water, E3 for G4EmstandardPhysics option 3 and E4 for G4EmstandardPhysics option 4. Average sigma values are listed for SW versus W and E4 versus E3.
Solidwater® using WET ratios of 1.025 and 1.03. Simulation indicates the range overestimation is ~0.5–1 mm difference between the scaled Bragg peaks in Solidwater® using WET ratio of 1.03 and the peaks in water. When the depth is more than 10 mm from the proton range, the relative dose difference is on the order of 2%, with differences up to 4% and 6% approaching 6871
L Lin et al
Phys. Med. Biol. 59 (2014) 6859
Figure 9. Comparison of Bragg peaks in water and Solidwater® phantoms for 210 MeV (left) and 115 MeV (right) proton beams. WET ratios at 1.03 and 1.025 are used to scale the Bragg peaks (SW1.03 and SW1.025) in Solidwater® to those in a water phantom. The relative percentage differences, (W − SW1.03)/W and (SW1.025 − W)/W, are plotted between the scaled Bragg peaks and the Bragg peaks in water.
the Bragg peak, for 115 MeV and 210 MeV proton beams, respectively. Therefore the application of a Solidwater® phantom in TOPAS simulations and measurements as a replacement for water phantom should be used with caution given to the factors indicated above. 5. Conclusion TOPAS Monte Carlo simulations, based on Geant4.9.6, using a proton source model located at the phantom surface and approximated with three Gaussian functions to incorporate the halo, can predict measured 2D profiles within 1 mm DTA down to the 0.1% isodose level and the 1D integral curve of spot profile versus radius within a 1 mm/1% Gamma criteria for the studied depths and energies except near the end of proton range. By matching profiles at the phantom surface and comparing simulated profiles to measurement at various depths, the accuracy of the particle interaction algorithm in phantom was investigated. The difference between simulation and measurement can be up to 15% for the primary spot sigma and exceed 1 mm at isodoses lower than 0.1% of the central spot dose. As the choice of particle interaction physics model and phantom material can both cause ~1 mm range uncertainty, maximal 5% (0.3 mm) difference of spot sigma, maximal 1 mm DTA of isodose above 0.1% level, and ~2 mm DTA of isodose below 0.1% level, the selection of physics model and the application of Solidwater® as a water replacement phantom material in simulation and measurement should be cautioned with the above indicated limitations. Acknowledgments The author Liyong Lin would like to thank Olivier De Wilde from the Research and Development Division of Ion Beam Applications (Louvain-la-Neuve, Belgium) for productive discussions about machine design and beam tuning processes.
6872
L Lin et al
Phys. Med. Biol. 59 (2014) 6859
References Agostinelli S et al 2003 Geant4—a simulation toolkit Nucl. Instrum. Methods A 506 250–303 Arjomandy B et al 2008 Use of a 2D ionization chamber array for proton therapy beam quality assurance Med. Phys. 35 3889–95 Arjomandy B et al 2010 Energy dependence and dose response of Gafchromic EBT2 film over a wide range of photon, electron, and proton beam energies Med. Phys. 37 1942–7 Clasie B et al 2012 Golden beam data for proton pencil-beam scanning Phys. Med. Biol. 57 1147–58 Farr J et al 2013 Fundamental radiological and geometric performance of two types of proton beam modulated discrete scanning systems Med. Phys. 40 2101–8 Gillin M et al 2009 Commissioning of the discrete spot scanning proton beam delivery system at the University of Texas M D Anderson Cancer Center, Proton Therapy Center, Houston Med. Phys. 37 154–63 Grevillot L et al 2010 Optimization of GEANT4 settings for proton pencil beam scanning simulations using GATE Nucl. Instrum. Methods Phys. Res. B 268 3295–305 Grevillot L et al 2011 A Monte Carlo proton pencil beam scanning treatment planning simulation using GATE/Geant4 Phys. Med. Biol. 56 5203–19 Harms W et al 1998 A software tool for the qualitative evaluation of 3D dose calculation algorithms Med. Phys. 25 1830–7 Kimstrand P et al 2007 A beam source model for scanned proton beams Phys. Med. Biol. 52 3151–68 Li Y et al 2012 Beyond Gaussians: a study of single-spot modeling for scanning proton dose calculation Phys. Med. Biol. 57 983–97 Lin L et al 2013a A novel technique for measuring the low-dose envelope of pencil beam scanning spot profiles Phys. Med. Biol. 58 N171–80 Lin L, Ainsley C and McDonough J 2013b Experimental determination of 2D spot pencil beam scanning profiles Phys. Med. Biol. 58 6193–204 Lin L et al 2014 Experimental determination of 2D spot pencil beam scanning profiles for two proton pencil beam scanning nozzles Phys. Med. Biol. 59 493–504 Lomax A et al 2004 Treatment planning and verification of proton therapy using spot scanning: initial experiences Med. Phys. 31 3150–7 Martisikova M and Jäkel O 2010 Dosimetric properties of Gafchromic EBT films in monoenergetic medical ion beams Phys. Med. Biol. 55 3741–51 Paganetti H 2012 Range uncertainties in proton therapy and the role of Monte Carlo simulations Phys. Med. Biol. 57 R99–R117 Paganetti H et al 2004 Accurate Monte Carlo simulations for nozzle design, commissioning and quality assurance for a proton radiation therapy facility Med. Phys. 31 2107–18 Pedroni E et al 2005 Experimental characterization and physical modeling of the dose distribution of scanned proton pencil beams Phys. Med. Biol. 50 541–61 Perl J et al 2012 TOPAS: an innovative proton Monte Carlo platform for research and clinical applications Med. Phys. 39 6818–37 Reinhardt S, Hillbrand M and Assmann F 2012 Comparison of Gafchromic EBT2 and EBT3 films for clinical photon and proton beams Med. Phys. 39 5257–63 Sawakuchi G et al 2010a Experimental characterization of the low-dose envelop of spot scanning proton beams Phys. Med. Biol. 55 3467–78 Sawakuchi G et al 2010b An MCNPX Monte Carlo model of a discrete spot scanning proton beam therapy nozzle Med. Phys. 37 4960–70 Soukup M, Fippel M and Alber M 2005 A pencil beam algorithm for intensity modulated proton therapy derived from Monte Carlo simulations Phys. Med. Biol. 50 5089–104 Tourovsky A, Lomax A and Pedroni E 2005 Monte Carlo calculation for spot scanned proton therapy Phys. Med. Biol. 50 971–9 Ulmer W 2007 Theoretical aspects of energy-range relations, stopping power and energy straggling of protons Radiat. Phys. Chem. 76 1089–107 Zhao L and Das I J 2010 GafChromic EBT film dosimetry in proton beams Phys. Med. Biol. 55 N291–301 Zhu R et al 2013 Commissioning dose computation models for spot scanning proton beams in water for a commercially available treatment planning system Med. Phys. 40 041723
6873
Search
Collections
Journals
About
Contact us
My IOPscience
Experimentally validated pencil beam scanning source model in TOPAS
This content has been downloaded from IOPscience. Please scroll down to see the full text. 2014 Phys. Med. Biol. 59 6859 (http://iopscience.iop.org/0031-9155/59/22/6859) View the table of contents for this issue, or go to the journal homepage for more Download details: IP Address: 128.59.222.107 This content was downloaded on 24/07/2017 at 13:50 Please note that terms and conditions apply.
You may also be interested in: Experimental characterization of two-dimensional spot profiles for two proton pencil beam scanning nozzles Liyong Lin, Christopher G Ainsley, Timothy D Solberg et al. Experimental characterization of two-dimensional pencil beam scanning proton spot profiles Liyong Lin, Christopher G Ainsley and James E McDonough A novel technique for measuring the low-dose envelope of pencil-beam scanning spot profiles Liyong Lin, Christopher G Ainsley, Thierry Mertens et al. Validation of nuclear models in Geant4 using the dose distribution of a 177 MeV proton pencil beam David C Hall, Anastasia Makarova, Harald Paganetti et al. Monte Carlo study of the potential reduction in out-of-field dose using a patient-specific aperture in pencil beam scanning proton therapy Stephen J Dowdell, Benjamin Clasie, Nicolas Depauw et al. A Monte Carlo PBS model for proton treatment plan simulation L Grevillot, D Bertrand, F Dessy et al. Experimental characterization of the low-dose envelope of spot scanning proton beams Gabriel O Sawakuchi, X Ronald Zhu, Falk Poenisch et al.
Institute of Physics and Engineering in Medicine Phys. Med. Biol. 59 (2014) 6859–6873
Physics in Medicine & Biology doi:10.1088/0031-9155/59/22/6859
Experimentally validated pencil beam scanning source model in TOPAS Liyong Lin, Minglei Kang, Timothy D Solberg, Christopher G Ainsley and James E McDonough Department of Radiation Oncology, University of Pennsylvania, Philadelphia, PA, USA E-mail: [email protected] Received 15 May 2014, revised 8 September 2014 Accepted for publication 2 October 2014 Published 28 October 2014 Abstract
The presence of a low-dose envelope, or ‘halo’, in the fluence profile of a proton spot can increase the output of a pencil beam scanning field by over 10%. This study evaluated whether the Monte Carlo simulation code, TOPAS 1.0-beta 8, based on Geant4.9.6 with its default physics list, can predict the spot halo at depth in phantom by incorporating a halo model within the proton source distribution. Proton sources were modelled using three 2D Gaussian functions, and optimized until simulated spot profiles matched measurements at the phantom surface out to a radius of 100 mm. Simulations were subsequently compared with profiles measured using EBT3 film in Solidwater® phantoms at various depths for 100, 115, 150, 180, 210 and 225 MeV proton beams. Simulations predict measured profiles within a 1 mm distance to agreement for 2D profiles extending to the 0.1% isodose, and within 1 mm/1% Gamma criteria over the integrated curve of spot profile as a function of radius. For isodose lines beyond 0.1% of the central spot dose, the simulated primary spot sigma is smaller than the measurement by up to 15%, and can differ by over 1 mm. The choice of particle interaction algorithm and phantom material were found to cause ~1 mm range uncertainty, a maximal 5% (0.3 mm) difference in spot sigma, and maximal 1 mm and ~2 mm distance to agreement in isodoses above and below the 0.1% level, respectively. Based on these observations, therefore, the selection of physics model and the application of Solidwater® as water replacement material in simulation and measurement should be used with caution. Keywords: Monte Carlo simulation, proton therapy, pencil beam scanning, treatment planning, Solidwater® (Some figures may appear in colour only in the online journal) 0031-9155/14/226859+15$33.00 © 2014 Institute of Physics and Engineering in Medicine Printed in the UK & the USA
6859
L Lin et al
Phys. Med. Biol. 59 (2014) 6859
1. Introduction Compared to analytical-algorithm-based treatment planning systems (TPSs) (Lomax et al 2004), Monte Carlo based TPSs (Tourovsky et al 2005) are superior in modelling the dose distribution around complex density interfaces, especially for pencil beam scanning (PBS) treatments. PBS treatments deliver proton spots in multiple energy layers. The number of spots in a PBS field can vary from less than one thousand to greater than ten thousand. Therefore PBS treatments require accurate characterization and modelling of spot profiles within the TPS. Spot profiles have been described by Pedroni et al (2005) as a primary Gaussian distribution followed by a broad non-Gaussian tail or ‘halo’. Soukup et al (2005) compared spot profiles and depth doses of Bragg peaks in a water phantom calculated by a pencil beam algorithm to those simulated by Geant4 (Agostinelli et al 2003) and concluded that pencil beam algorithms can achieve a 1 mm/3% dose accuracy in homogenous and simple heterogeneous phantoms but not in sophisticated clinical scenarios. Paganetti et al (2004) and Kimstrand et al (2007) described Geant4 simulations of a double-scattering nozzle and an experimental PBS nozzle, respectively. For the dedicated PBS nozzle developed by Ion Beam Applications (Louvain-la-Neuve, Belgium), Grevillot et al (2010) found good agreement in the Bragg peak depth dose between Geant4 simulation and measurement, but observed up to ~20% underestimation of the primary spot sigma in simulation compared with measurement in a PMMA phantom. Grevillot et al also observed differences of up to 15% in the primary spot sigma generated by different simulation platforms (MCNPX, PHITS, Szymanowski) (Grevillot et al 2010). They reported that MCNPX tended to overestimate the spot sigma while PHITS appeared to be the most accurate. Subsequently, Sawakuchi et al (2010a, 2010b) reported a more accurately calculated spot sigma by incorporating more large-angle single scattering events in a newer Multiple Coulomb Scattering algorithm in MCNPX. To study profiles extending beyond the primary Gaussian component of the spot, Sawakuchi et al (2010b) used MCNPX to simulate the contribution of nuclear interactions within the nozzle to spot profiles in a water phantom; simulation results were compared to measurement for the Hitachi ProBeat delivery system (Hitachi, Ltd, Tokyo, Japan). Similarly, Clasie et al (2012) studied the contribution of nuclear interactions to the integral Bragg peak in a water phantom and was able to achieve good agreement with measurement by adjusting the parameters within Geant4. Thus, Monte Carlo simulation is a critical tool in developing approximation methods for TPSs and standalone dose calculation engines (Gillin et al 2009, Grevillot et al 2011, Li et al 2012, Zhu et al 2013). The model of Sawakuchi et al (2010b) simulates the beam beginning at the middle of two scanning magnets using the source’s Gaussian sigmas as free parameters. In contrast, we investigate the feasibility of modelling a proton source located at the phantom surface using 2D Gaussian parameters extracted from measured spot profiles in air. Such an approach does not require modelling of complex particle transport within the IBA dedicated PBS nozzle that, unlike the Hitachi nozzle, has two upstream focusing quadruples (Farr et al 2013). Also, starting particle transport from the phantom surface minimizes the uncertainty from the free parameters used in modelling of the angular spread distribution otherwise needed to derive downstream in-air spot profiles. Furthermore, starting Monte Carlo simulations using a proton source distribution that matches simulated and measured spot profiles at the phantom surface for both the primary and tail regions, allows any differences to be separated into two components: inaccurate source modelling from the treatment nozzle and incorrect particle interactions within the phantom. Rather than focusing solely on the Gaussian sigma of the primary component of the spot profile, this investigation looks closely at the low-dose envelope, or 6860
L Lin et al
Phys. Med. Biol. 59 (2014) 6859
‘halo’, simulated in phantom by TOPASwith the initial halo based on in-air dose measurements. At depth in phantom, the halo distribution originates predominately from the treatment nozzle for low energy proton beams. Therefore the determination of the halo source distribution is crucial for accurate characterization of spot profiles for low energy proton beams. In contrast, for high energy proton beams the halo distribution is predominantly generated from particle interaction in the phantom. Recent progress has allowed experimental determination of the halo component (Lin et al 2013a, 2013b , 2014).This study attempts to determine whether the combination of a good proton source model coupled with an appropriate particle interaction algorithm can predict the halo distribution to a clinically acceptable level for selected depths and proton energies. While the dose levels below 1% of the central spot dose may be insignificant for a single spot, the summation of the broad tails from multiple spots can affect a typical field’s overall output by up to 4% and 7% for the IBA dedicated and universal nozzles, respectively, for a 100 MeV proton beam. Similar for all nozzles, the output can increase by over 10% for a 225 MeV proton beam (Gillin et al 2009, Sawakuchi et al 2010a, 2010b, Grevillot et al 2011, Clasie et al 2012, Li et al 2012, Zhu et al 2013, Lin et al 2013a, 2013b, 2014). Accounting for this phenomenon requires an accurate PBS source model and robust particle interaction algorithms to characterize spot tails over the region from 1% to 0.01% of the central spot dose, which may extend spatially outwards to 100 mm radius from the spot centre. Here we study if it is feasible to incorporate the measurement-based halo emanating from the treatment nozzle into a Monte Carlo TPS, neglecting the perturbation caused by in-air angular spread. Although this approach cannot differentiate the origin of the spatial distribution or calculate the energy spectrum of the halo from a treatment nozzle, it can test if a Monte Carlo TPS is accurate enough for clinical dose distribution calculation by reporting the residual disagreement between simulation and measurement from the phantom surface to the end of proton range. 2. Methods and materials Spot profiles were measured in a Solidwater® phantom (Gammex, Inc., Wisconsin, USA) using Gafchromic EBT3 film (Ashland Specialty Ingredients, New Jersey, USA) for six selected energies: 100, 115, 150, 180, 210 and 225 MeV from the IBA dedicated PBS nozzle. The water equivalent thickness (WET) of each Solidwater® phantom was experimentally determined to be 1.03 with an uncertainty of 0.5–1% (see Discussion). Within the Monte Carlo simulation, Solidwater® was modelled according to the vendor-supplied specification as a mixture of six elements, ‘hydrogen’, ‘carbon’, ‘nitrogen’, ‘oxygen’, ‘chlorine’ and ‘calcium’, in weight fractions of 0.0809, 0.6717, 0.0241, 0.1988, 0.0014 and 0.0231, respectively, with a density of 1.044 g cm−3. In contrast to the PMMA used by Grevillot et al (2010), Solidwater® phantoms are widely used in proton therapy quality assurance (Arjomandy et al 2008) in combination with film and ionization chamber arrays. Simulations were performed in both water and Solidwater® phantoms, however, the simulation results are reported for the former because water is the reference material for clinical dosimetry. The equivalence of water and Solidwater® in terms of lateral scattering properties and the validity of the WET ratio along the depth direction were investigated for selected scenarios in this study. Phantom surface locations at isocentre and 270 mm upstream of isocentre were selected to encompass the range of patient surface locations observed in clinical practice. Films were placed in the phantom at 50 mm increments to the end of the range, whereupon two films were placed with a 5 mm spacing. The pair/magnification method (Farr et al 2013, Lin et al 2013a, 6861
L Lin et al
Phys. Med. Biol. 59 (2014) 6859
Figure 1. The process to derive from measurement the 2D Gaussian parameters of
the proton source model in TOPAS. The top left panel shows the isodose contours of the measured 150 MeV proton spot profile of the IBA dedicated PBS nozzle in air at isocentre. The bottom right panel shows the 1D comparison of integrated curve of spot profiles versus radius between TOPAS simulation (S0) and measurement (M0). The bottom left panel shows the 2D profile comparison between simulation (solid lines) and measurement (dashed lines) and lists the average primary Gaussian sigma derived from simulation and measurement. The isodose contours are 50% (cyan), 10% (black), 1% (red), 0.1% (green) and 0.01% (blue) of the central dose.
2013b) was applied to determine the composite 2D spot profiles in air and in phantom for the selected energies. The resulting kernels extend over an area of 200 mm × 200 mm and were constructed with a 0.5 mm resolution. To calculate the fractional energy collected as a function of radius, 2D asymmetric dose kernels were converted into symmetric kernels and used to integrate spot profiles outwards from the centre (Lin et al 2013, 2014). The process of extracting 2D Gaussian parameters from the measured profiles extending to the level of 10–4 of the central dose and to 100 mm in radius is described in figure 1. Three Gaussian functions are chosen to express the broad tail of the proton source model up to 100 mm radius, as there is only ~1% contribution to the integral dose beyond 3σ of the primary Gaussian. The first primary Gaussian is chosen to be almost circular and the second Gaussian typically characterizes the elongation of the tail. If the isodose at 0.01% of the central spot dose does not follow the elongation of the isodose at 0.1% of the central spot dose, the model is further modified by the third Gaussian. For proton energies above 115 MeV, a single Gaussian distribution is used to model the energy spectrum of the spot profiles, while at 115 MeV and below, three slightly different Gaussian energy spreads are used, corresponding to the three spatial spot profile Gaussian functions (see Discussion). The three Gaussian source models were independently simulated in TOPAS 1.0-beta 8 (Perl et al 2012), a simulation tool based on Geant4.9.6, and the weighted sum of the simulation results from the three Gaussian sources was performed using an in-house Matlab program. Twenty million particle histories were run for each of the three sources utilizing 10% of the central spot dose), only the protons from the first Gaussian functions are used to study the average proton energy in the phantom for the 115, 150 and 210 MeV beams. The primary proton energies are averaged in 0.5 mm radial increments. Figure 6 shows that proton energies in the primary spot regions are larger than 40 MeV for most depths; near the end of range, however, the energy falls below 30 MeV for the 115 MeV beams and below 20 MeV for the 150 and 210 MeV beams. The smaller disagreement in spot sigma between simulation and measurement near the end of range compared to other depths is caused by quenching effects of Gafchromic EBT3 film, which results in the spot size measured with film being smaller than reality for proton energies below 40 MeV. Beyond primary spot regions, the average proton energy changes gradually or 6867
L Lin et al
Phys. Med. Biol. 59 (2014) 6859
Figure 5. Comparison of the curves of integrated spot profile versus radius up to 100 mm between simulation and measurement in Solidwater® phantom at depths of 50, 90 and 95 mm for the 115 MeV proton beam; at depths of 50, 100, 150 and 155 mm for the 150 MeV beam; and at depths of 50, 100, 200, 270 and 275 mm for the 210 MeV beam. The phantom surface is at isocentre. The red ‘+’ and blue ‘x’ markers represent simulation and measurement, respectively.
stays nearly constant as a function of radius within the ‘shoulder’ region, which is widest at the depths between half and two-thirds of the proton range of the highest energy (210 MeV) proton beam. The statistical error is within ~2% for the isodoses >0.1% and 7% for the 0.01% isodose level. The curve of average proton energy is plotted beyond the 0.01% isodose with dash lines and cut off when the number of protons drops below 20. In the methods section, we presented an approach to derive the weight and sigma of the 2D Gaussian functions for the tail components. In figure 7, we present a method to determine whether the tail components have a lower energy than the primary component by comparing in-air spot profiles on the central axis to those off axis. We hypothesize that Landau fluctuations (Ulmer 2007) in the energy lost in the degrader and the nozzle generate lower energy protons that are subsequently deflected more than the average distance by the ratio of the momentum of the tail components to that of the primary component. When the proton spots are deflected 100 mm away from the x and y axes, no difference is observed between the central axis and the off-axis locations for the 210 MeV proton beam (figure 7(a)). In contrast, for the 115 MeV proton beam, up to 2 mm differences are observed at the far end of the y-axis of the diamond-shaped 0.01% isodose contour (figure 7(c)). As the y-axis is the focusing direction of the downstream quadruple, we speculate that the low energy protons are preferentially located there due to imperfect focusing. Here we construct a hypothetical spectrum (linear scale in figure 7(e) and log scale in figure 7(f)) to account for the approximately 4% lower energy (i.e. ~2% lower momentum) as reflected in 2% more scanning distance in figure 7(c). To approximate the low energy Landau tails the energy spectrum is expressed as the summation of three Gaussian functions centred at 115 MeV, an intermediate energy, and 110 MeV with the corresponding weights (95.5%, 3% and 1.5%) of the three spatial Gaussian functions in table 1. 6868
L Lin et al
Phys. Med. Biol. 59 (2014) 6859
Figure 6. Average simulated proton energy versus radius from the spot centre in a Solidwater® phantom at depths of 50, 90 and 95 mm for the 115 MeV proton beam; at depths of 50, 100, 150 and 155 mm for the 150 MeV beam; and at depths of 50, 100, 200, 270 and 275 mm for the 210 MeV beam. The numbers on the upper right corner of the plots are the number of primary protons at this depth for the beam. The dash part of the curve is beyond isodose of 0.01%.
Wider energy spreads are used for the lower energy components to achieve a smooth energy spectrum. Subsequent simulation demonstrates that adopting such an energy spectrum within the source model can improve the depth to agreement by 5–6 mm for isodoses below 0.1% at a depth of 90 mm near the end of range of the 115 MeV proton beam. Without such a spectrum, simulated isodoses below 0.1% of the central dose would be broader than measurement. The clinical impact of primary and tail components in the source distributions are investigated for the Bragg peaks of the 115 and 210 MeV proton beams. The low energy tail components cause approximately 2% loss in the peak value of the 115 MeV Bragg peak (figure 7(d)). The energy spread is iteratively determined to be 0.6% and 0.35% to match the measured shape of Bragg peaks as closely as possible for 115 MeV and 210 MeV proton beams, respectively. The energy spread data presented here are similar to those reported by Grevillot et al (2010) for similar energies in the IBA dedicated PBS nozzle in Essen, Germany. At 210 MeV, there is no observable difference between the simulated Bragg peaks that include the primary component only and those that have primary and tail components. Note there is maximal 1.5 mm range difference between simulated and measured Bragg peaks and that we have shifted the simulated Bragg peaks to simplify comparison. Figure 8 compares profiles at various depths for 115 and 210 MeV proton beams in Solidwater® and water, and between the two electromagnetic physics models of TOPAS. A WET ratio of 1.03 is used to convert the depths in Solidwater® to those in water; the uncertainty in the WET ratio is presented in figure 9. For all depths of the 115 MeV proton beam (figure 8(a)), Solidwater® yields a smaller spot than water and G4EmstandardPhysics option 4 predicts a larger spot than G4EmstandardPhysics option 3 with ~0.1 mm difference of spot sigma. For the studied depths of the 210 MeV proton beam (figure 8(b)), Solidwater® 6869
L Lin et al
Phys. Med. Biol. 59 (2014) 6859
Figure 7. Comparison of two 2D spot profiles and depth doses of 210 MeV (a), (b) and 115 MeV (c), (d) Bragg peaks (p, t1, t2 stands for the primary, secondary and tertiary Gaussian contributions, respectively). For each energy, one spot profiles were measured in air at the location that is +100 mm and −100 mm away from x and y axes, and the other was measured on the central axis of the isocentre plane. The isodose contours are 50% (cyan), 10% (black), 1% (red), 0.1% (green) and 0.01% (blue) of the central dose. The solid lines are for the off-axis location and the dashed lines are for the central axis location. The inferred initial energy spectrum of the 115 MeV proton beam is displayed in (e) linear and (f) log scales from 105 to 125 MeV.
produces smaller primary spots than water, and G4EmstandardPhysics option 4 generates larger primary spots than the default G4EmstandardPhysics option 3 with up to 5% (0.3 mm) difference. The difference of simulated isodoses between Solidwater® and water phantoms is smaller than 1 mm at all dose levels above 0.1% but can exceed 1 mm below the 0.1% level for the 210 MeV proton beam. Because of the low dose gradient of the integral curves at such large radii, the dose difference reflected in the integral curve would be less than 1% (see fi gure 5). On the other hand, the difference of simulated isodoses between G4EmstandardPhysics option 4 and G4EmstandardPhysics option 3 was smaller than 1 mm DTA above 0.1% and ~2 mm DTA at the 0.01% isodose level for most depths except for 270 mm, where the difference of isodose contours below 0.1% reached ~2 mm DTA potentially due to a change of the lateral scattering mechanism and range uncertainty caused by the different electromagnetic physics models used [8–10, 27]. Another potential reason for the small difference of spot profile between Solidwater® and water is that the geometrical length of Solidwater® is smaller than water and the initial beam divergence, which is on the order of 3 mrad, was neglected. A WET ratio of 1.03 approximates Solidwater® to water along the depth direction. As this ratio may change with proton energy and causes uncertainty in the corresponding WET in simulation, figure 9 compares the Bragg peaks simulated in Solidwater® and water for 115 and 210 MeV proton beams. The scoring volumes use a radius of 10 cm to ensure collection of all the deposited energy. The Bragg peaks in water fall within the scaled Bragg peaks in 6870
L Lin et al
Phys. Med. Biol. 59 (2014) 6859
Figure 8. Comparison of simulated profiles in water and Solidwater® (top) and between the G4EmstandardPhysics option 3 (default physics list of TOPAS) and option 4 (bottom), for (a) 115 MeV and (b) 210 MeV proton beams at the depths (in mm) indicated in the bottom, right of the panels. SW stands for Solidwater®, W for Water, E3 for G4EmstandardPhysics option 3 and E4 for G4EmstandardPhysics option 4. Average sigma values are listed for SW versus W and E4 versus E3.
Solidwater® using WET ratios of 1.025 and 1.03. Simulation indicates the range overestimation is ~0.5–1 mm difference between the scaled Bragg peaks in Solidwater® using WET ratio of 1.03 and the peaks in water. When the depth is more than 10 mm from the proton range, the relative dose difference is on the order of 2%, with differences up to 4% and 6% approaching 6871
L Lin et al
Phys. Med. Biol. 59 (2014) 6859
Figure 9. Comparison of Bragg peaks in water and Solidwater® phantoms for 210 MeV (left) and 115 MeV (right) proton beams. WET ratios at 1.03 and 1.025 are used to scale the Bragg peaks (SW1.03 and SW1.025) in Solidwater® to those in a water phantom. The relative percentage differences, (W − SW1.03)/W and (SW1.025 − W)/W, are plotted between the scaled Bragg peaks and the Bragg peaks in water.
the Bragg peak, for 115 MeV and 210 MeV proton beams, respectively. Therefore the application of a Solidwater® phantom in TOPAS simulations and measurements as a replacement for water phantom should be used with caution given to the factors indicated above. 5. Conclusion TOPAS Monte Carlo simulations, based on Geant4.9.6, using a proton source model located at the phantom surface and approximated with three Gaussian functions to incorporate the halo, can predict measured 2D profiles within 1 mm DTA down to the 0.1% isodose level and the 1D integral curve of spot profile versus radius within a 1 mm/1% Gamma criteria for the studied depths and energies except near the end of proton range. By matching profiles at the phantom surface and comparing simulated profiles to measurement at various depths, the accuracy of the particle interaction algorithm in phantom was investigated. The difference between simulation and measurement can be up to 15% for the primary spot sigma and exceed 1 mm at isodoses lower than 0.1% of the central spot dose. As the choice of particle interaction physics model and phantom material can both cause ~1 mm range uncertainty, maximal 5% (0.3 mm) difference of spot sigma, maximal 1 mm DTA of isodose above 0.1% level, and ~2 mm DTA of isodose below 0.1% level, the selection of physics model and the application of Solidwater® as a water replacement phantom material in simulation and measurement should be cautioned with the above indicated limitations. Acknowledgments The author Liyong Lin would like to thank Olivier De Wilde from the Research and Development Division of Ion Beam Applications (Louvain-la-Neuve, Belgium) for productive discussions about machine design and beam tuning processes.
6872
L Lin et al
Phys. Med. Biol. 59 (2014) 6859
References Agostinelli S et al 2003 Geant4—a simulation toolkit Nucl. Instrum. Methods A 506 250–303 Arjomandy B et al 2008 Use of a 2D ionization chamber array for proton therapy beam quality assurance Med. Phys. 35 3889–95 Arjomandy B et al 2010 Energy dependence and dose response of Gafchromic EBT2 film over a wide range of photon, electron, and proton beam energies Med. Phys. 37 1942–7 Clasie B et al 2012 Golden beam data for proton pencil-beam scanning Phys. Med. Biol. 57 1147–58 Farr J et al 2013 Fundamental radiological and geometric performance of two types of proton beam modulated discrete scanning systems Med. Phys. 40 2101–8 Gillin M et al 2009 Commissioning of the discrete spot scanning proton beam delivery system at the University of Texas M D Anderson Cancer Center, Proton Therapy Center, Houston Med. Phys. 37 154–63 Grevillot L et al 2010 Optimization of GEANT4 settings for proton pencil beam scanning simulations using GATE Nucl. Instrum. Methods Phys. Res. B 268 3295–305 Grevillot L et al 2011 A Monte Carlo proton pencil beam scanning treatment planning simulation using GATE/Geant4 Phys. Med. Biol. 56 5203–19 Harms W et al 1998 A software tool for the qualitative evaluation of 3D dose calculation algorithms Med. Phys. 25 1830–7 Kimstrand P et al 2007 A beam source model for scanned proton beams Phys. Med. Biol. 52 3151–68 Li Y et al 2012 Beyond Gaussians: a study of single-spot modeling for scanning proton dose calculation Phys. Med. Biol. 57 983–97 Lin L et al 2013a A novel technique for measuring the low-dose envelope of pencil beam scanning spot profiles Phys. Med. Biol. 58 N171–80 Lin L, Ainsley C and McDonough J 2013b Experimental determination of 2D spot pencil beam scanning profiles Phys. Med. Biol. 58 6193–204 Lin L et al 2014 Experimental determination of 2D spot pencil beam scanning profiles for two proton pencil beam scanning nozzles Phys. Med. Biol. 59 493–504 Lomax A et al 2004 Treatment planning and verification of proton therapy using spot scanning: initial experiences Med. Phys. 31 3150–7 Martisikova M and Jäkel O 2010 Dosimetric properties of Gafchromic EBT films in monoenergetic medical ion beams Phys. Med. Biol. 55 3741–51 Paganetti H 2012 Range uncertainties in proton therapy and the role of Monte Carlo simulations Phys. Med. Biol. 57 R99–R117 Paganetti H et al 2004 Accurate Monte Carlo simulations for nozzle design, commissioning and quality assurance for a proton radiation therapy facility Med. Phys. 31 2107–18 Pedroni E et al 2005 Experimental characterization and physical modeling of the dose distribution of scanned proton pencil beams Phys. Med. Biol. 50 541–61 Perl J et al 2012 TOPAS: an innovative proton Monte Carlo platform for research and clinical applications Med. Phys. 39 6818–37 Reinhardt S, Hillbrand M and Assmann F 2012 Comparison of Gafchromic EBT2 and EBT3 films for clinical photon and proton beams Med. Phys. 39 5257–63 Sawakuchi G et al 2010a Experimental characterization of the low-dose envelop of spot scanning proton beams Phys. Med. Biol. 55 3467–78 Sawakuchi G et al 2010b An MCNPX Monte Carlo model of a discrete spot scanning proton beam therapy nozzle Med. Phys. 37 4960–70 Soukup M, Fippel M and Alber M 2005 A pencil beam algorithm for intensity modulated proton therapy derived from Monte Carlo simulations Phys. Med. Biol. 50 5089–104 Tourovsky A, Lomax A and Pedroni E 2005 Monte Carlo calculation for spot scanned proton therapy Phys. Med. Biol. 50 971–9 Ulmer W 2007 Theoretical aspects of energy-range relations, stopping power and energy straggling of protons Radiat. Phys. Chem. 76 1089–107 Zhao L and Das I J 2010 GafChromic EBT film dosimetry in proton beams Phys. Med. Biol. 55 N291–301 Zhu R et al 2013 Commissioning dose computation models for spot scanning proton beams in water for a commercially available treatment planning system Med. Phys. 40 041723
6873
