Reference Beam Data Research Articles

End-to-End Verification and Dosimetric Comparative Study of Brainlab Monte Carlo and Eclipse Acuros XB Photon Dose Calculation Algorithm for VMAT SRS Planning

We commissioned the Brainlab Monte Carlo Dose Calculation Algorithm for VMAT SRS planning and verified it by measurements and comparisons with two other algorithms: Pencil beam and Eclipse Acuros. The end-to-end verification and dosimetric comparative study results are reported in this abstract. Brainlab Elements Cranial SRS (v3.0) provides single lesion planning using VMAT optimization, thus allowing dose modulation with MLC leaf positioning, dose rate, and gantry speed. We commissioned and performed quality assurance for the available dose calculation algorithms, including Pencil Beam (PB) algorithm with only reference beam data (RBM PB) and PB with full beam data (Elements PB), and Monte Carlo (MC) algorithm XVMC. Calculation between the two were compared to each other and to Acuros XB dose calculation algorithm by Varian. Open field MLC, static cone, dynamic MLC, and conical cone plans were calculated with each algorithm. IBA CC01 ionization chamber was used for each measurement, and small field correction factors as determined in TRT-483 were applied. End-to-end tests were performed using an anthropomorphic stereotactic end-to-end verification (STEEV) phantom from CIRS and a phantom/diode array, StereoPHAN/SRSMapCHECK from SunNuclear. Each phantom was CT scanned with 0.5 mm slice thickness. Four clinical plans were optimized in Brainlab Elements Treatment Planning System (TPS). Each plan was calculated with the following algorithms: Elements PB, RBM PB, MC, and Acuros XB with 1 mm dose grid size. The delivered dose was measured with a SunNuclear SRSMapCHECK diode array detector. The calculated dose was compared with the measured dose by Gamma analysis with DTA 1%/1mm, 2%/1mm, and 3%/1mm. The results of point dose measurements for field size larger than 1 × 1 cm2 among measured dose, dose calculated by pencil beam algorithms, and dose calculated by Monte Carlo algorithm were in excellent agreement (1.45% ± 0.7%). For the small rectangular field sizes (0.4 × 1, 0.2 × 1 cm2), Monte Carlo calculation showed superior accuracy and robustness to pencil beam algorithm when compared with measured dose, with percent dose difference 13% and -2.2%, respectively. Monte Carlo calculation have higher gamma results compared with pencil beam algorithm due more accurate calculations with heterogenous materials. Acuros XB showed comparable accuracy with MC in terms of gamma passing rate. The SRS MapCHECK measurement agrees dosimetrically with the plans calculated by Pencil Beam algorithm and the Monte Carlo algorithm XVMC by Brainlab, as well as the Acuros XB dose calculation algorithm by Varian. Point dose calculation for small field dosimetry and gamma analysis using SRS MapCHECK both showed that Monte Carlo algorithm is more robust and is superior in regions of heterogenous materials. Therefore, we recommend that Monte Carlo algorithm for dose calculation be used in clinical SRS and SRT cases.

Neural Representation of Linear Accelerator Beam Data from a Single Reference Dataset for Fast Commissioning and Quality Assurance

<h3>Purpose/Objective(s)</h3> We hypothesize that linear accelerator (Linac) beam properties including lateral profiles and percentage depth dose (PDD) can be modeled through a neural representation network using a single vendor-provided reference beam dataset. Such beam model can be used to predict clinical Linac beam data distributions from sparse beam measurements, which has the potential of supporting fast commissioning and routine quality assurance (QA) in the clinic. <h3>Materials/Methods</h3> A sinusoidal neural representation network that consists of multilayer perceptron and periodic activation function was used to learn a mapping between 2D coordinates (1D spatial locations x field sizes) during beam scanning and the corresponding percentage dose. The network was trained using a vendor-provided 6 MV reference beam dataset only and evaluated using clinical beam datasets that consist of ten 6MV beam profiles at 10 cm depth and seven 6MV percentage depth dose (PDD) curves collected from multiple clinics. Only beam data with 30 × 30 cm² field size were assumed available and used to update the network weights. Beam data with 10 × 10 cm² were then predicted by the network and compared to the ground truth of clinically acquired data by calculating both the relative mean absolute error and the 1D gamma passing rate with a passing criteria of 1%/1 mm. <h3>Results</h3> Compared to vendor-provided reference beam data, the network-predicted beam data showed improved consistency with the clinically acquired data. The averaged mean relative absolute error and averaged gamma passing rate between network-predicted dose profile/reference beam dose profile and the clinical measurements were 1.9%/2.5% and 95%/90% respectively. The averaged mean relative absolute error and averaged gamma passing rate between network-predicted PDD/reference beam PDD and the clinical measurements were 0.6%/0.5% and 100%/97% respectively. <h3>Conclusion</h3> Neural representation network is capable of modeling inherent Linac beam properties and predicting clinical beam data accurately from sparse beam measurements. The proposed beam model has the potential of simplifying the clinical beam scanning process with much reduced beam data collection.

Collection and analysis of photon beam data for Varian C-series linear accelerators: a potential reference beam data set.

This study aimed to collect and analyze photon beam data for the Varian C-series linear accelerators (Varian Medical Systems, Palo Alto, CA, USA). We evaluated the potential of the average data to be used as reference beam data for the radiotherapy treatment planning system commissioning verification. We collected 20 data sets for 4 and 6 MV photon beams, and 40 data sets for a 10 MV photon beam generated by the Varian C-series machines, which contained the percent depth dose (PDD), off-center ratio (OCR), and output factor (OPF) from 20 institutions. The average for each of the data types was calculated across the 20 machines. Dose differences from the average for PDD at the dose fall-off region were less than 1.0%. Relative differences from the average for the OPF data were almost within 1.0% for all energies and field sizes. For OCR data in the flat regions, the standard deviation of the dose differences from the average was within 1.0%, excluding that of the 30 × 30 mm2 field size being approximately 1.5%. For all energies and field sizes, the distance to agreement from the average in the OCR penumbra regions was less than 1.0mm. The average data except for the small field size found in this study can be used as reference beam data for verifying users' commissioning results.

Commissioning of the Mobius3D independent dose verification system for TomoTherapy.

In radiation therapy, a secondary independent dose verification is an important component of a quality control system. Mobius3D calculates three‐dimensional (3D) patient dose using reference beam data and a collapsed cone convolution algorithm and analyzes dose‐volume histogram automatically. There are currently no published data on commissioning and determining tolerance levels of Mobius3D for TomoTherapy. To verify the calculation accuracy and adjust the parameters of this system, we compared the measured dose using an ion chamber and film in a phantom with the dose calculated using Mobius3D for nine helical intensity‐modulated radiation therapy plans, each with three nominal field widths. We also compared 126 treatment plans used in our institution to treat prostate, head‐and‐neck, and esophagus tumors based on dose calculations by treatment planning system for given dose indices and 3D gamma passing rates with those produced by Mobius3D. On the basis of these results, we showed that the action and tolerance levels at the average dose for the planning target volume (PTV) at each treatment site are at μ ± 2σ and μ ± 3σ, respectively. After adjusting parameters, the dose difference ratio on average was −0.2 ± 0.6% using ion chamber and gamma passing rate with the criteria of 3% and 3 mm on average was 98.8 ± 1.4% using film. We also established action and tolerance levels for the PTV at the prostate, head‐and‐neck, esophagus, and for the organ at risk at all treatment sites. Mobius3D calculations thus provide an accurate secondary dose verification system that can be commissioned easily and immediately after installation. Before clinical use, the Mobius3D system needs to be commissioned using the treatment plans for patients treated in each institution to determine the calculational accuracy and establish tolerances for each treatment site and dose index.

Validation of secondary dose calculation system with manufacturer-provided reference beam data using heterogeneous phantoms.

The Mobius3D (M3D) system (Mobius Medical Systems) is a second-check dosimetry system. We investigated the dose calculation accuracy of this system using heterogeneous phantoms with reference beam data provided by the manufacturer using simple and patient plans. We compared the dose distributions between M3D and the treatment planning system, as well as the measurements in solid water phantoms, heterogeneous phantoms, and patient plans for Varian and Elekta accelerators. The M3D results agreed well with the measurements in the solid water phantoms for the simple plans. However, the accuracy of M3D appeared to depend on the type of accelerator, as indicated by the slight differences in the dose measurements. Furthermore, the M3D dose measurements differed by 5-10% in the lung and bone regions. Regarding the patient plans, we confirmed that M3D is reasonably accurate as a second-check system, despite the slight accelerator type-dependent dose difference.

SU‐E‐T‐211: Comparison of Seven New TrueBeam Linacs with Enhanced Beam Data Conformance Using a Beam Comparison Software Tool

Purpose:To develop an independent software tool to assist in commissioning linacs with enhanced beam conformance, as well as perform ongoing QA for dosimetrically equivalent linacs.Methods:Linac manufacturers offer enhanced beam conformance as an option to allow for clinics to complete commissioning efficiently, as well as implement dosimetrically equivalent linacs. The specification for enhanced conformance includes PDD as well as profiles within 80% FWHM. Recently, we commissioned seven Varian TrueBeam linacs with enhanced beam conformance. We developed a software tool in Visual Basic to allow us to load the reference beam data and compare our beam data during commissioning to evaluate enhanced beam conformance. This tool also allowed us to upload our beam data used for commissioning our dosimetrically equivalent beam models to compare and tweak each of our linac beams to match our modelled data in Varian's Eclipse TPS. This tool will also be used during annual QA of the linacs to compare our beam data to our baseline data, as required by TG‐142.Results:Our software tool was used to check beam conformance for seven TrueBeam linacs that we commissioned in the past six months. Using our tool we found that the factory conformed linacs showed up to 3.82% difference in their beam profile data upon installation. Using our beam comparison tool, we were able to adjust the energy and profiles of our beams to accomplish a better than 1.00% point by point data conformance.Conclusion:The availability of quantitative comparison tools is essential to accept and commission linacs with enhanced beam conformance, as well as to beam match multiple linacs. We further intend to use the same tool to ensure our beam data conforms to the commissioning beam data during our annual QA in keeping with the requirements of TG‐142.

SU-E-T-532: Validation and Implementation of Model-Based Patient Specific Quality Assurance Using Mobius3D and MobiusFX in a Clinical Setting

Purpose: This work carries out the commissioning and validation of the Mobius3D and MobiusFX software tools, which can replace the time-consuming measurement-based patient specific quality assurance (PSQA). Methods: The beam model supplied by Mobius3D was validated against a 21EX linac's beam measured data. Complex patient (VMAT) plans using Eclipse treatment planning system (TPS) was used to test the consistency between Mobius3D (calculates dose using patient image and field data) and MobiusFx (calculates dose using treatment dynalog files). Dose difference and gamma analysis (3%/3mm) between Mobius3D and MobiusFx were used to assess treatment plan and treatment delivery consistency. An end-to-end test was performed to validate Mobius3D and MobiusFx against ion chamber measurements. Effect of the dosimetric leaf gap (DLG) on Mobius3D dose calculation was additionally investigated. Results: Mobius3D beam model parameters matched within 1%-3% with our beam measured data. A comparison of Mobius3D and MobiusFx dose matrices for VMAT planned prostate cases showed (0.33±0.07)% mean dose difference with gamma values above 95%. The end-to-end test showed dose differences of 1% between Mobius3D and MobiusFx. Dependence of Mobius3D dose calculation upon DLG was explored by introducing a ±0.5 mm change in the default value for DLG. This change resulted in agreement differences above 2% Conclusion: Use of reference beam data would appear to speed up commissioning process for the clinical implementation of Mobius3D. However, careful consideration is required when comparing the information provided by the software, since large dose variations can be seen when the proper parameters are not optimized. The plan and delivered dose were in good agreement; hence MobiusFx has the potential to significantly speed up the PSQA process and at the same time helps to verify treatment parameters that are not possible with measurement-based PSQA.

SU‐E‐T‐03: 3D GPU‐Accelerated Secondary Checks of Radiation Therapy Treatment Plans

Purpose:Redundant treatment verifications in conformal and intensity‐modulated radiation therapy techniques are traditionally performed with single point calculations. New solutions can replace these checks with 3D treatment plan verifications. This work describes a software tool (Mobius3D, Mobius Medical Systems) that uses a GPU‐accelerated collapsed cone algorithm to perform 3D independent verifications of TPS calculations.Methods:Mobius3D comes with reference beam models for common linear accelerators. The system uses an independently developed collapsed cone algorithm updated with recent enhancements. 144 isotropically‐spaced cones are used for each voxel for calculations. These complex calculations can be sped up by using GPUs. Mobius3D calculate dose using DICOM information coming from TPS (CT, RT Struct, RT Plan RT Dose). DVH‐metrics and 3D gamma tests can be used to compare both TPS and secondary calculations. 170 patients treated with all common techniques as 3DCFRT (including wedged), static and dynamic IMRT and VMAT have been successfully verified with this solution.Results:Calculation times are between 3–5 minutes for 3DCFRT treatments and 15–20 for most complex dMLC and VMAT plans. For all PTVs mean dose and 90% coverage differences are (1.12±0.97)% and (0.68±1.19)%, respectively. Mean dose discrepancies for all OARs is (0.64±1.00)%. 3D gamma (global, 3%/3 mm) analysis shows a mean passing rate of (97.8 ± 3.0)% for PTVs and (99.0±3.0)% for OARs. 3D gamma pasing rate for all voxels in CT has a mean value of (98.5±1.6)%.Conclusion:Mobius3D is a powerful tool to verify all modalities of radiation therapy treatments. Dose discrepancies calculated by this system are in good agreement with TPS. The use of reference beam data results in time savings and can be used to avoid the propagation of errors in original beam data into our QA system. GPU calculations permit enhanced collapsed cone calculations with reasonable calculation times.

Quantitative evaluation of a beam‐matching procedure using one‐dimensional gamma analysis

This article provides a quantitative evaluation of Varian Medical Systems' beam matching procedure. A one-dimensional y analysis is employed to investigate the level of agreement of matched beams. A customized concept of one-dimensional gamma evaluation was designed. Our algorithm first performs a "local" fit of the reference and the evaluated datasets. For a particular point on the fitted evaluated curve, the y is derived as the shortest distance between the point and the fitted reference curve. This approach removes variations of the obtained y value related to the discrete character and noise in the original datasets. Criteria of 1 mm distance-to-agreement and 1% dose difference were used to evaluate the level of agreement of according profiles. Relative point and profile measurements were performed for all photon and electron beams of two Varian Clinacs 2100C/D. Matched beams show a good level of agreement. 70% of profiles completely pass the chosen criteria. The analysis of remaining 30% of the profiles demonstrates that measurement error becomes a limiting factor in achieving a better score. The highest obtained y value was 1.70. The quality of beam matching allowed us to treat according beams of both treatment units as "identical" and to use the reference beam data for the new unit. Nevertheless, the vendor's acceptance criteria of beam matching are much more benevolent. It might happen that the acceptance criteria are met, however, resulting quality of beam matching does not allow full interchangeability of beams.

Fast electron beam simulation and dose calculation in radiotherapy

A flexible multiple source model capable of fast reconstruction of clinical electron beams is presented in this paper. A source model considers multiple virtual sources emulating the effect of the accelerator head components. A reference configuration ( 10 MeV and 10×10 cm 2 field size) for a Siemens KD2 linear accelerator was simulated in detail using the GEANT3 Monte Carlo code. Our model allows the reconstruction of other electron energies and field sizes, as well as other beam configurations of similar accelerators, using only the reference beam data. Electron dose calculations with the reconstructed beams were performed in a water phantom and compared with experimental data. An agreement of 1–2%/1– 2 mm was obtained, comparable to the accuracy of full Monte Carlo accelerator simulation. The source model reduces accelerator simulation CPU time by a factor of 7500 relative to full Monte Carlo approaches. The developed model was then interfaced to DPM, a fast radiation transport Monte Carlo code for dose calculation. Dose distributions and dosimetric parameters computed in the presence of air gaps using this procedure agree within 1%/ 1 mm with experimental data. The obtained accuracy corresponds to a significant improvement when compared to the treatment planning system PLATO, currently used in clinical practice. Globally, a two order of magnitude efficiency gain was achieved with respect to the full Monte Carlo approach.

Dosimetry and treatment planning for boron neutron capture therapy

Dosimetric techniques were developed for the determination of the various dose components in phantoms irradiated with an epithermal neutron beam for boron neutron capture therapy (BNCT). Using paired ionization chambers and activation foils, the accuracy (1 standard deviation, SD) in the determination of the absolute thermal neutron fluence, gamma‐ray dose and fast neutron dose amounted to 5%, 5%, and 18%, respectively. Methods were developed which assure an agreement between prescribed and actual obtained values of 1% (1 SD) for the absolute beam output and 5%–10% (1 SD) for the blood–boron concentration during irradiation. Clinical dosimetry studies were performed resulting in a set of reference beam data which are used for quality control, dosimetry intercomparisons and as input for a conventional semi‐empirical treatment planning system (TPS). Using these reference beam data and the electron and photon beam model of the TPS, the various dose components could be calculated fast and with a high accuracy in a homogeneous anthropomorphic head phantom. At the central axis, the agreement between measurements and calculations was within 2%–3% (1 SD). It is concluded that the accuracy in dosimetry, beam output monitoring, blood–boron monitoring and treatment planning is sufficient to start clinical studies of BNCT in Petten, The Netherlands.