Clinical use of a commercial Monte Carlo treatment planning system for electron beams

Auto-commissioning of a Monte Carlo electron beam model with application to photon MLC shaped electron fields

Objective.Presently electron beam treatments are delivered using dedicated applicators. An alternative is the usage of the already installed photon multileaf collimator (pMLC) enabling efficient electron treatments. Currently, the commissioning of beam models is a manual and time-consuming process. In this work an auto-commissioning procedure for the Monte Carlo (MC) beam model part representing the beam above the pMLC is developed for TrueBeam systems with electron energies from 6 to 22 MeV.Approach.The analytical part of the electron beam model includes a main source representing the primary beam and a jaw source representing the head scatter contribution each consisting of an electron and a photon component, while MC radiation transport is performed for the pMLC. The auto-commissioning of this analytical part relies on information pre-determined from MC simulations, in-air dose profiles and absolute dose measurements in water for different field sizes and source to surface distances (SSDs). For validation calculated and measured dose distributions in water were compared for different field sizes, SSDs and beam energies for eight TrueBeam systems. Furthermore, a sternum case in an anthropomorphic phantom was considered and calculated and measured dose distributions were compared at different SSDs.Main results.Instead of the manual commissioning taking up to several days of calculation time and several hours of user time, the auto-commissioning is carried out in a few minutes. Measured and calculated dose distributions agree generally within 3% of maximum dose or 2 mm. The gamma passing rates for the sternum case ranged from 96% to 99% (3% (global)/2 mm criteria, 10% threshold).Significance.The auto-commissioning procedure was successfully implemented and applied to eight TrueBeam systems. The newly developed user-friendly auto-commissioning procedure allows an efficient commissioning of an MC electron beam model and eases the usage of advanced electron radiotherapy utilizing the pMLC for beam shaping.

Implementation and application of a Monte Carlo model for an in vivo micro computed tomography system

Micro computed tomography (µCT) scanners are used to create high-resolution images and to quantify properties of the scanned objects. While modern µCT scanners benefit from the cone beam geometry, they are compromised by scatter radiation. This work aims to develop a Monte Carlo (MC) model of a µCT scanner in order to characterize the scatter radiation in the detector plane. The EGS++ framework with the MC code EGSnrc was used to simulate the particle transport through the main components of the XtremeCT (SCANCO Medical AG, Switzerland). The developed MC model was based on specific information of the manufacturer and was validated against measurements. The primary and the scatter radiation were analyzed and by implementing a dedicated tracing method, the scatter radiation was subdivided into different scatter components. The comparisons of measured and simulated transmission values for different absorber and filter combinations result in a mean difference of 0.2% ± 1.4%, with a maximal local difference of 3.4%. The reconstructed image of the phantom based on measurements agrees well with the image reconstructed using the MC model. The local contribution of scattered radiation is up to 10% of the total radiation in the detector plane and most of the scattered particles result from interactions in the scanned object. The MC simulations show that scatter radiation contains information about the structure of the object. In conclusion, a MC model for a µCT scanner was successfully validated and applied to analyze the characteristics of the scatter radiation for a µCT scanner.

Dosimetric validation of Monaco treatment planning system on an Elekta VersaHD linear accelerator

The purpose of this study is to perform dosimetric validation of Monaco treatment planning system version 5.1. The Elekta VersaHD linear accelerator with high dose rate flattening filter‐free photon modes and electron energies was used in this study. The dosimetric output of the new Agility head combined with the FFF photon modes warranted this investigation into the dosimetric accuracy prior to clinical usage. A model of the VersaHD linac was created in Monaco TPS by Elekta using commissioned beam data including percent depth dose curves, beam profiles, and output factors. A variety of 3D conformal fields were created in Monaco TPS on a combined Plastic water/Styrofoam phantom and validated against measurements with a calibrated ion chamber. Some of the parameters varied including source to surface distance, field size, wedges, gantry angle, and depth for all photon and electron energies. In addition, a series of step and shoot IMRT, VMAT test plans, and patient plans on various anatomical sites were verified against measurements on a Delta⁴ diode array. The agreement in point dose measurements was within 2% for all photon and electron energies in the homogeneous phantom and within 3% for photon energies in the heterogeneous phantom. The mean ± SD gamma passing rates of IMRT test fields yielded 93.8 ± 4.7% based on 2% dose difference and 2 mm distance‐to‐agreement criteria. Eight previously treated IMRT patient plans were replanned in Monaco TPS and five measurements on each yielded an average gamma passing rate of 95% with 6.7% confidence limit based on 3%, 3 mm gamma criteria. This investigation on dosimetric validation ensures accuracy of modeling VersaHD linac in Monaco TPS thereby improving patient safety.

Improved efficiency in Monte Carlo simulation for passive-scattering proton therapy

The aim of this work was to improve the computational efficiency of Monte Carlo simulations when tracking protons through a proton therapy treatment head. Two proton therapy facilities were considered, the Francis H Burr Proton Therapy Center (FHBPTC) at the Massachusetts General Hospital and the Crocker Lab eye treatment facility used by University of California at San Francisco (UCSFETF). The computational efficiency was evaluated for phase space files scored at the exit of the treatment head to determine optimal parameters to improve efficiency while maintaining accuracy in the dose calculation. For FHBPTC, particles were split by a factor of 8 upstream of the second scatterer and upstream of the aperture. The radius of the region for Russian roulette was set to 2.5 or 1.5 times the radius of the aperture and a secondary particle production cut (PC) of 50 mm was applied. For UCSFETF, particles were split a factor of 16 upstream of a water absorber column and upstream of the aperture. Here, the radius of the region for Russian roulette was set to 4 times the radius of the aperture and a PC of 0.05 mm was applied. In both setups, the cylindrical symmetry of the proton beam was exploited to position the split particles randomly spaced around the beam axis. When simulating a phase space for subsequent water phantom simulations, efficiency gains between a factor of 19.9  ±  0.1 and 52.21  ±  0.04 for the FHTPC setups and 57.3  ±  0.5 for the UCSFETF setups were obtained. For a phase space used as input for simulations in a patient geometry, the gain was a factor of 78.6  ±  7.5. Lateral-dose curves in water were within the accepted clinical tolerance of 2%, with statistical uncertainties of 0.5% for the two facilities. For the patient geometry and by considering the 2% and 2mm criteria, 98.4% of the voxels showed a gamma index lower than unity. An analysis of the dose distribution resulted in systematic deviations below of 0.88% for 20% of the voxels with dose of 20% of the maximum or more.

Comprehensive evaluation and clinical implementation of commercially available Monte Carlo dose calculation algorithm

A commercial electron Monte Carlo (eMC) dose calculation algorithm has become available in Eclipse treatment planning system. The purpose of this work was to evaluate the eMC algorithm and investigate the clinical implementation of this system. The beam modeling of the eMC algorithm was performed for beam energies of 6, 9, 12, 16, and 20 MeV for a Varian Trilogy and all available applicator sizes in the Eclipse treatment planning system. The accuracy of the eMC algorithm was evaluated in a homogeneous water phantom, solid water phantoms containing lung and bone materials, and an anthropomorphic phantom. In addition, dose calculation accuracy was compared between pencil beam (PB) and eMC algorithms in the same treatment planning system for heterogeneous phantoms. The overall agreement between eMC calculations and measurements was within 3%/2 mm, while the PB algorithm had large errors (up to 25%) in predicting dose distributions in the presence of inhomogeneities such as bone and lung. The clinical implementation of the eMC algorithm was investigated by performing treatment planning for 15 patients with lesions in the head and neck, breast, chest wall, and sternum. The dose distributions were calculated using PB and eMC algorithms with no smoothing and all three levels of 3D Gaussian smoothing for comparison. Based on a routine electron beam therapy prescription method, the number of eMC calculated monitor units (MUs) was found to increase with increased 3D Gaussian smoothing levels. 3D Gaussian smoothing greatly improved the visual usability of dose distributions and produced better target coverage. Differences of calculated MUs and dose distributions between eMC and PB algorithms could be significant when oblique beam incidence, surface irregularities, and heterogeneous tissues were present in the treatment plans. In our patient cases, monitor unit differences of up to 7% were observed between PB and eMC algorithms. Monitor unit calculations were also preformed based on point‐dose prescription. The eMC algorithm calculation was characterized by deeper penetration in the low‐density regions, such as lung and air cavities. As a result, the mean dose in the low‐density regions was underestimated using PB algorithm. The eMC computation time ranged from 5 min to 66 min on a single 2.66 GHz desktop, which is comparable with PB algorithm calculation time for the same resolution level.

PACS number: 87.55.K‐

VMC++ validation for photon beams in the energy range of 20-1000 keV

PURPOSE

In high energy teletherapy, VMC++ is known to be a very accurate and efficient Monte Carlo (MC) code. In principle, the MC method is also a powerful dose calculation tool in other areas in radiation oncology, e.g., brachytherapy or orthovoltage radiotherapy. However, VMC++ is not validated for the low-energy range of such applications. This work aims in the validation of the VMC++ MC code for photon beams in the energy range between 20 and 1000 keV.

METHODS

Dose calculations were performed in different 40 x 40 x 40 cm3 phantoms of different materials. Dose distributions of monoenergetic (ranging from 20 to 1000 keV) 10 x 10 and 2 x 2 cm2 parallel beams were calculated. Voxel sizes of 4 x 4 x 4 and 1 x 1 x 1 mm3 were used for the dose calculations. The resulting dose distributions were compared to those calculated using EGSnrc, which is used as a golden standard in this work.

RESULTS

At energies between 100 and 1000 keV, EGSnrc and VMC++ calculated dose distributions agree within the statistical uncertainty of about 1% (1sigma). At energies < or = 50 keV, dose differences of up to 1.6% (in % of D(max)) occur when VMC++ and EGSnrc are compared. Turning off Rayleigh scattering, binding effects for Compton scattering, and the atomic relaxation after photoelectric absorption in EGSnrc (all not implemented in VMC++) leads to an agreement between both MC codes within statistical uncertainty. Further, using the KERMA approximation feature implemented in VMC++ leads to very efficient simulations in the energy range between 20 and 1000 keV.

CONCLUSIONS

Further improvements for very low energies in accuracy of VMC++ could be achieved by implementing Rayleigh scattering, binding effects for Compton scattering, and the atomic relaxation after photoelectric absorption. Implementation into VMC++ of KERMA approximation has been validated.

Assessment of computerized treatment planning system accuracy in calculating wedge factors of physical wedged fields for 6 MV photon beams

Wedge filters are commonly used in external beam radiotherapy to achieve a uniform dose distribution within the target volume. The main objective of this study was to investigate the accuracy of the beam modifier algorithm of Theraplan plus (TPP version 3.8) treatment planning system and to confirm that either the beam hardening, beam softening and attenuation coefficients along with wedge geometry and measured wedge factor at single depth and multiple fields sizes can be the replacement of wedged profile and wedged cross-sectional data or not. In this regard the effect of beam hardening and beam softening was studied with physical wedges for 6 MV photons. The Normalized Wedge Factors (NWFs) were measured experimentally as well as calculated with the Theraplan plus, as a function of depth and field size in a water phantom for 15°, 30°, 45°, and 60° wedge filters. The beam hardening and softening was determined experimentally by deriving the required coefficients for all wedge angles. The TPP version 3.8 requires wedge transmission factor at single depth and multiple field sizes. Without incorporating the hardening and softening coefficients the percent difference between measured and calculated NFWs was as high as 7%. After the introduction of these parameters into the algorithm, the agreement between measured and TPP (V 3.8) calculated NWFs were improved to within 2 percent for various depths. Similar improvement was observed in TPP version 3.8 while calculating NWFs for various field sizes when the required coefficients were adjusted. In conclusion, the dose calculation algorithm of TPP version 3.8 showed good accuracy for a 6 MV photon beam provided beam hardening and softening parameters are taken into account. From the results, it is also concluded that, the beam hardening, beam softening and attenuation coefficients along with wedge geometry and measured wedge factor at single depth and multiple fields sizes can be the replacement of wedged profile and wedged cross-sectional data in the TPS. The study also indicated that by ignoring the beam softening and beam hardening will result in an inaccurate dose to the target volume of the patient.

Quantitative and analytical comparison of isodose distributions for shaped electron fields from ADAC Pinnacle treatment planning system and Monte Carlo simulations

We assess the accuracy of ADAC Pinnacle(3) commercial treatment planning system (TPS) in computation of isodose distributions for shaped electron fields. The assessment is based on comparison of dose profiles generated by TPS and a Monte Carlo model for different beam energies, applicator sizes, and percentages of field blocking. Dose differences of up to 14% are observed at the depth of maximum dose. These discrepancies, often ignored in clinical evaluations, are attributable to inadequate modeling of scatter from applicators and blocks by TPS.

Erratum to "Technical note: Accuracy checks of physical beam modifier factors algorithm used in computerized treatment planning system for a 15 MV photon beam" [Rep. Pract. Oncol. Radiother. 14 (2009) 214-220]

[This corrects the article DOI: 10.1016/j.rpor.2009.12.002.].

Dosimetry of a small air cavity for clinical electron beams: A Monte Carlo study

This study investigated dosimetric changes in a water phantom when a small air cavity was presented at the central axis of a clinical electron beam. We used 6-, 9-, and 16-MeV electron beams with a 10 x 10 cm(2) applicator and cutout produced by a Varian 21 EX linear accelerator. Percentage depth doses (PDDs) for different depths (0.5-7 cm), thicknesses (2-10 mm), and widths (1-5 cm) of air cavities were calculated using Monte Carlo simulations (EGSnrc code) validated by film measurements. By comparing PDDs of phantoms with and without the air cavity, it was found that when the depth or thickness of cavity was changed, the PDD curve below the cavity was shifted with a distance equal to the thickness of the cavity. However, when the width of the air cavity was changed, both the PDD curve and its slope within and below the cavity were changed. A larger width of the air cavity resulted in a shallower PDD curve within the cavity. The slope of the PDD curve below the cavity tended towards a value as the width of the air cavity was increased to 3-5 cm for the 6-, 9-, and 16-MeV electron beams. The dependence of the depth dose on the width of the air cavity is a result of the contribution of the electron side scattering in the water surrounding the cavity. The change in depth dose resulting from the presence of an air cavity can cause discrepancies between the calculated and actual dose during radiotherapy, unless the effects of the air cavity are properly characterized during treatment planning. From the dosimetry data in this study, neglecting an air cavity of 1-cm thickness in the build-up region of a 6-MeV electron beam resulted in a delivered dose 10-12% larger than the original prescription. Delivered doses 3% and 6% higher than the prescribed dose were observed when doses were prescribed at R(80) for a 16-MeV electron beam. These results were obtained by neglecting air cavities with thicknesses equal to 2 and 4 mm, respectively, at a depth of 5 cm.

Modulated electron radiotherapy treatment planning using a photon multileaf collimator for post-mastectomized chest walls

BACKGROUND AND PURPOSE

To evaluate the feasibility of using a photon MLC (xMLC) for modulated electron radiotherapy treatment (MERT) as an alternative to conventional post-mastectomy chest wall (CW) irradiation. A Monte Carlo (MC) based planning system was developed to overcome the inaccuracy of the 'pencil beam' algorithm. MC techniques are known to accurately calculate the dose distributions of electron beams, allowing the explicit simulation of electron interactions within the MLC.

MATERIALS AND METHODS

Four real clinical CW cases were planned using MERT which were compared with the conventional electron treatments based on blocks and by a straightforward approach using the MLC, and not the blocks (as an intermediate step to MERT) to shape the same segments with SSD between 60 and 70 cm depending on PTV size. MC calculations were verified with an array of ionization chambers and radiochromic films in a solid water phantom.

RESULTS

Tests based on gamma analysis between MC dose distributions and radiochromic film measurements showed an excellent agreement. Differences in the absolute dose measured with a plane-parallel chamber at a reference point were below 3% for all cases. MERT solution showed a better PTV coverage and a significant reduction of the doses to the organs at risk (OARs).

CONCLUSION

MERT can effectively improve the current electron treatments by obtaining a better PTV coverage and sparing healthy tissues. More directly, block-shaped treatments could be replaced by MLC-shaped non-modulated segments providing similar results.

The use of a diode matrix in commissioning activities for electron energies > or = 9 MeV: a feasibility study

The contribution of a commercially available diode matrix (MapCHECK, provided by Sun Nuclear, Melbourne, FL) for the commissioning procedures of the voxel based Monte Carlo (VMC++) algorithm for electron beams of MasterPlan treatment planning system was investigated. The attention is mainly focused on the calculation in homogeneous and heterogeneous phantoms. With this aim, following a data set similar to that proposed by Electron Collaborative Working Group (ECWG), the dose profiles and two-dimensional (2D) dose distributions measured by the diode matrix were compared with the calculated ones using the gamma analysis method with acceptance criteria for the dose difference and the distance to agreement equal to 4% and 4 mm, respectively. The average and standard deviation of the percentage of points satisfying the constraint gamma < or = 1 are 98.3 +/- 4.1% and 99.3 +/- 1.7% for the 9 and 12 MeV electron beam, respectively, showing that the accuracy of MasterPlan electron beam algorithm is good for simple two-dimensional geometries as well as for more complicated three-dimensional ones. The results are in agreement with those reported in literature by Cygler et al. ["Evaluation of the first commercial Monte Carlo dose calculation engine for electron beam treatment planning," Med. Phys. 31, 142-153 (2004)]. In addition, the authors have also analyzed the response of the 2D array in terms of dose profiles at different depths, comparing the results with those obtained in water phantom using an electron diode. The results show that in the low gradient regions there were no deviations larger than the criteria of acceptability set by Van Dyk et al. ["Commissioning and quality assurance of treatment planning computers," Int. J. Radiat. Oncol. Biol. Phys. 26, 261-273 (1993)]; in the high gradient region, the maximum deviations are less than 2 mm with most of the values less than 1 mm. The present article shows that MapCHECK can play a useful role in the commissioning of electron algorithms of treatment planning systems in the evaluation of the 2D dose distributions in homogeneous and heterogeneous phantoms. In fact, it provides accurate results with the merit of expediting the commissioning process by using measuring device that requires minimal setup time and data processing time.

Comparison of conventional inserts and an add-on electron MLC for chest wall irradiation of left-sided breast cancer

BACKGROUND

Collimation of irregularly shaped clinical electron beams is currently based on electron inserts made of low melting point alloys. The present investigation compares a conventional electron applicator with insert and add-on eMLC-based dose distributions in the postoperative chest wall irradiation of left-sided breast cancer.

MATERIAL AND METHODS

Voxel Monte Carlo++ (VMC++) calculated dose distributions related to electron fields were compared with 10 left-sided breast cancer patients after radical mastectomy. The prescription dose was 50 Gy at a build-up maximum. The same dose was prescribed for the ipsilateral axillary, parasternal and supraclavicular lymph nodes that were treated with photons and calculated with a pencil beam algorithm. The insert beams were shaped with 1.5 cm thick Wood's metal electron inserts in an electron applicator of a Varian 2100 C/D linac. Doses for the eMLC-shaped beams were calculated for an eMLC prototype with 2 cm thick and 5 mm wide steel leaves. The same collimator-to-surface distance (CSD) of 5.8 cm was used for both collimators.

RESULTS

The mean PTV dose was slightly higher for the eMLC plans (50.7 vs 49.5 Gy, p<0.001, respectively). The maximum doses assessed by D5% for the eMLC and insert were 60.9 and 59.1 Gy (p<0.001). The difference was due to the slightly higher doses near the field edges for the eMLC. The left lung V20 volumes were 34.5% and 34.0% (p<0.001). There was only a marginal difference in heart doses.

DISCUSSION

Despite a slight increase of maximum dose in PTV the add-on electron MLC for chest wall irradiation results in practically no differences in dose distributions compared with the present insert-based collimation.

Enhancement of electron-beam surface dose with an electron multi-leaf collimator (eMLC): a feasibility study

Use of a water-equivalent bolus in electron-beam radiotherapy is sometimes impractical and non-hygienic. Therefore, the feasibility of applying adjacent narrow beams for producing high surface dose electron beams without a bolus was investigated. Depth dose curves and profiles in water were calculated and measured for 6 and 9 MeV electron-beam segments (width 0.3-1.5 cm, length 10 cm) for source-to-surface distances (SSD) 102 and 105 cm. Segment shaping was performed with an add-on electron multi-leaf collimator prototype attached to the Varian 2100 C/D linac. Dose calculations were performed with the Voxel Monte Carlo++ algorithm. Resulting dose distributions in typical clinical cases were compared with the bolus technique. With a composite segmental field with 1.0 cm wide segments the surface dose was over 90% of the depth dose maximum for both energies. The build-up area practically disappeared with a 0.5 cm wide single beam. This led to decrease in the therapeutic range for composite fields with segment widths smaller than 1.0 cm. The new technique yielded similar surface doses as the bolus technique. The photon contamination was 4% with a 9 x 10 cm(2) field (1.0 cm wide segments) compared to 1% for the respective open field with 9 MeV with a bolus. The calculated dose agreed within 2 mm and 3% of the measured dose in 93.7% and 85.2% of the voxels. Adjacent narrow eMLC beams with a 1.0 cm width are suitable to produce electron fields with high surface dose. Despite a slight nonuniformity in the surface profiles in the lateral part of the field at SSD 102 cm, surface dose and target coverage are comparable with the bolus technique.

A systematic study of posterior cervical lymph node irradiation with electrons: Conventional versus customized planning

BACKGROUND

High dose irradiation of the posterior cervical lymph nodes usually employs applied electron fields to treat the target volume and maintain the spinal cord dose within tolerance. In the light of recent advances in elective lymph node localisation we investigated optimization of field shape and electron energy to treat this target volume.

METHODS

In this study, three sequential hypotheses were tested. Firstly, that customization of the electron fields based on the nodal PTV outlined gives better PTV coverage than conventional field delineation. Using the consensus guidelines, customization of the electron field shape was compared to conventional fields based on bony landmarks. Secondly, that selection of electron energy using DVHs for spinal cord and PTV improves the minimum dose to PTV. Electron dose-volume histograms (DVHs) for the PTV, spinal cord and para-vertebral muscles, were generated using the Monte Carlo electron algorithm. These DVHs were used to compare standard vs optimized electron energy calculations. Finally, that combination of field customization and electron energy optimization improves both the minimum and mean doses to PTV compared with current standard practice.

RESULTS

Customized electron beam shaping based on the consensus guidelines led to fewer geographical misses than standard field shaping. Customized electron energy calculation led to higher minimum doses to the PTV. Overall, the customization of field shape and energy resulted in an improved mean dose to the PTV (92% vs 83% p=0.02) and a 27% improvement in the minimum dose delivered to the PTV (45% vs 18% p=0.0009).

CONCLUSIONS

Optimization of electron field shape and beam energy based on current consensus guidelines led to significant improvement in PTV coverage and may reduce recurrence rates.

A virtual-accelerator-based verification of a Monte Carlo dose calculation algorithm for electron beam treatment planning in clinical situations

BACKGROUND AND PURPOSE

The introduction of Monte Carlo (MC) techniques for treatment planning and also for verification purposes will have considerable impact on the radiation therapy planning process. The aim of this work was to use a virtual accelerator to study the performance of a MC-based electron dose calculation algorithm, implemented in a commercial treatment planning system.

METHODS

The performance in phantoms containing air and bone as well as in patient-specific geometries (thorax wall, nose, parotid gland and spinal cord) has been studied.

RESULTS

The agreement between the virtual accelerator and the MC dose calculation algorithm is generally very good. A gamma-evaluation with criteria of 0.03 Gy/3 mm (per Gy at the depth of maximum dose) shows that, even for the worst cases, only a small volume of about 1.5% has gamma>1.0. In the worst case, with the 0.02 Gy/2 mm criteria, about 92% of the volume receiving more than 0.85 Gy per 100 monitor units (MU) has gamma-values <1.0. The corresponding value for the volume receiving more than 0.10 Gy/100 MU is about 98%. For the 18 MeV spinal-cord case, where a 6 x 20 cm2 insert is used, the TPS underestimates the dose outside the primary field due to inadequate modelling of the insert.

CONCLUSION

The possibility of dose calculations in typical patient cases makes the virtual accelerator a powerful tool for validation and evaluation of dose calculation algorithms present in treatment planning systems.

A virtual-accelerator-based verification of a Monte Carlo dose calculation algorithm for electron beam treatment planning in homogeneous phantoms

By introducing Monte Carlo (MC) techniques to the verification procedure of dose calculation algorithms in treatment planning systems (TPSs), problems associated with conventional measurements can be avoided and properties that are considered unmeasurable can be studied. The aim of the study is to implement a virtual accelerator, based on MC simulations, to evaluate the performance of a dose calculation algorithm for electron beams in a commercial TPS. The TPS algorithm is MC based and the virtual accelerator is used to study the accuracy of the algorithm in water phantoms. The basic test of the implementation of the virtual accelerator is successful for 6 and 12 MeV (gamma < 1.0, 0.02 Gy/2 mm). For 18 MeV, there are problems in the profile data for some of the applicators, where the TPS underestimates the dose. For fields equipped with patient-specific inserts, the agreement is generally good. The exception is 6 MeV where there are slightly larger deviations. The concept of the virtual accelerator is shown to be feasible and has the potential to be a powerful tool for vendors and users.