C. Baker

A national dosimetry audit for stereotactic ablative radiotherapy in lung

BACKGROUND AND PURPOSE A UK national dosimetry audit was carried out to assess the accuracy of Stereotactic Ablative Body Radiotherapy (SABR) lung treatment delivery. METHODS AND MATERIALS This mail-based audit used an anthropomorphic thorax phantom containing nine alanine pellets positioned in the lung region for dosimetry, as well as EBT3 film in the axial plane for isodose comparison. Centres used their local planning protocol/technique, creating 27 SABR plans. A range of delivery techniques including conformal, volumetric modulated arc therapy (VMAT) and Cyberknife (CK) were used with six different calculation algorithms (collapsed cone, superposition, pencil-beam (PB), AAA, Acuros and Monte Carlo). RESULTS The mean difference between measured and calculated dose (excluding PB results) was 0.4±1.4% for alanine and 1.4±3.4% for film. PB differences were -6.1% and -12.9% respectively. The median of the absolute maximum isodose-to-isodose distances was 3mm (-6mm to 7mm) and 5mm (-10mm to +19mm) for the 100% and 50% isodose lines respectively. CONCLUSIONS Alanine and film is an effective combination for verifying dosimetric and geometric accuracy. There were some differences across dose algorithms, and geometric accuracy was better for VMAT and CK compared with conformal techniques. The alanine dosimetry results showed that planned and delivered doses were within ±3.0% for 25/27 SABR plans.

OC-0154: UK SABR Consortium Lung Dosimetry Audit; relative dosimetry results

Purpose/Objective: The UK SABR Consortium QA group conducted a postal dosimetry audit of SABR lung plans at 21 UK centres. The purpose of this was to verify the accuracy of calculated dose distributions, improve confidence of centres in the early stages of implementing lung SABR and to establish a benchmark QA method. Here the results of the GafChromic film relative dosimetry arm of the audit are given. Materials and Methods: Individual centres were asked to plan a treatment to a pre-defined PTV in the CIRS Thorax phantom, using their clinical method and prescription dose. EBT3 GafChromic film was used to measure an axial plane of dose. Pins in the phantom facilitated alignment of the film and calculated dose planes. Gantry linac and Cyberknife centres were audited, using a variety of TPS with pencil beam, AAA, CCC, Acuros and Monte Carlo algorithms. Scanned films were compared to dose distributions calculated by the individual centres, using single red-channel dosimetry and a purpose-built Matlab application. Centres were also asked to irradiate additional calibration films to provide output-normalised optical density to dose calibration. Measured and calculated isodoses corresponding to 120, 100, 70 and 50% of prescription dose were compared (figure 1), and conformity and maximum distance to agreement were measured. For the areas bound by the 100, 50 and 30% calculated isodoses, local gamma analysis, mean gamma and gamma pass rate (at 3%, 2mm) and a mean dose comparison was performed. The latter was compared to the alanine dosimetry results. Results: The dosimetry of the calibration films was reproducible to ±0.9% (1.S.D), for doses ranging from 4.3 to 26.9 Gy. The audit relative dosimetry results are reported in table 1. Mean dose differences within the 100% calculated isodose line agreed well with alanine dosimetry; -0.1 ± 2.0 % (1.S.D). Gamma pass rates (%) and mean gamma results varied with some outlying measurements, mostly caused by small dose deviations within the PTV or at low doses. Isodose line agreement (figure 1) was generally much closer at the 70 and 100% dose levels, indicated by the lower S.D. (table 1, column 5). The exception was the centre using a pencil beam algorithm, where the measured prescription dose covered a significantly smaller area than that calculated, consistent with the algorithm’s known limitations calculating dose in low density lung surrounding tumour. Conclusions: Of the 21 UK centres audited, 74% of measurements were within ±3% agreement compared to calculated doses. Where appropriate, outlying centres have been offered support from the QA Group to bring their results into line. The EBT3 GafChromic film was found to be highly suited to a postal audit, reliably giving detailed information about the geometric and dosimetric accuracy of treatment.

A design of a DICOM-RT-based tool box for nonrigid 4D dose calculation

The study was aimed to introduce a design of a DICOM‐RT‐based tool box to facilitate 4D dose calculation based on deformable voxel‐dose registration. The computational structure and the calculation algorithm of the tool box were explicitly discussed in the study. The tool box was written in MATLAB in conjunction with CERR. It consists of five main functions which allow a) importation of DICOM‐RT‐based 3D dose plan, b) deformable image registration, c) tracking voxel doses along breathing cycle, d) presentation of temporal dose distribution at different time phase, and e) derivation of 4D dose. The efficacy of using the tool box for clinical application had been verified with nine clinical cases on retrospective‐study basis. The logistic and the robustness of the tool box were tested with 27 applications and the results were shown successful with no computational errors encountered. In the study, the accumulated dose coverage as a function of planning CT taken at end‐inhale, end‐exhale, and mean tumor position were assessed. The results indicated that the majority of the cases (67%) achieved maximum target coverage, while the planning CT was taken at the temporal mean tumor position and 56% at the end‐exhale position. The comparable results to the literature imply that the studied tool box can be reliable for 4D dose calculation. The authors suggest that, with proper application, 4D dose calculation using deformable registration can provide better dose evaluation for treatment with moving target. PACS number(s): 87.55.khThe study was aimed to introduce a design of a DICOM-RT-based tool box to facilitate 4D dose calculation based on deformable voxel-dose registration. The computational structure and the calculation algorithm of the tool box were explicitly discussed in the study. The tool box was written in MATLAB in conjunction with CERR. It consists of five main functions which allow a) importation of DICOM-RT-based 3D dose plan, b) deformable image registration, c) tracking voxel doses along breathing cycle, d) presentation of temporal dose distribution at different time phase, and e) derivation of 4D dose. The efficacy of using the tool box for clinical application had been verified with nine clinical cases on retrospective-study basis. The logistic and the robustness of the tool box were tested with 27 applications and the results were shown successful with no computational errors encountered. In the study, the accumulated dose coverage as a function of planning CT taken at end-inhale, end-exhale, and mean tumor position were assessed. The results indicated that the majority of the cases (67%) achieved maximum target coverage, while the planning CT was taken at the temporal mean tumor position and 56% at the end-exhale position. The comparable results to the literature imply that the studied tool box can be reliable for 4D dose calculation. The authors suggest that, with proper application, 4D dose calculation using deformable registration can provide better dose evaluation for treatment with moving target. PACS number(s): 87.55.kh.

PD-0450: A more exact approach to uncertainties in radiobiological parameter estimation

early effects. ICS distributions can be determined with Monte Carlo simulations at nanometric scale and this is usually done by considering liquid water cross-sections within the target volume. In this work, we propose to quantify the differences between ICS distributions obtained in a water medium with a realistic geometry of the DNA molecule compared to a DNAlike medium. The needed cross-sections of DNA-like materials have been measured and parameterized in the framework of a European Joint Research project. Materials and Methods: In this work, Geant4-DNA processes, included in the general Geant4 Monte Carlo (MC) toolkit, have been used to perform the simulations. Up to this work, Geant4-DNA only offers the possibility of using liquid water cross-sections for the target material. To take into account the new DNA-like materials, we implemented new specific models in the Geant4-DNA processes. These models use the cross-sections tables derived from the experimental data measured at PTB: tetrahydrofuran (THF), trimethylphosphate (TMP) pyrimidine (PY) and Purine (PU). Two different kinds of simulations were performed in this work: First, a homogeneous cylinder equivalent to 10 base pairs (bp) (d=2.3 nm, h =3.4nm) was filled with a target material defined as a combination of 20 nucleotides and their corresponding water shell (Γ=12 per nucleotide). The objective was to evaluate the difference in the number of ionizations obtained in the backbone region (sugar-phosphate materials) when using this medium compared to only liquid water cross-sections in a setup where the regions corresponding to those materials are evaluated from a realistic DNA geometrical target. Therefore, in a second set of simulations, a geometry representing 10 bp DNA segment was designed in this work with realistic volumes and positions calculated from PDB files. The ICS distribution taking place in the sugar and the phosphate volumes of this geometry was then compared to the ICS distribution obtained in the homogeneous cylinder were the THF and TMP cross-sections were used. Results: This work shows that the shape of the ICS distributions in both calculations present a very similar trend, evolving in the same manner with the projectile energy. Nevertheless, important differences were found in the mean value of these distributions that can rise to more than a factor of 2 higher in the case of using the DNA cross-sections instead of liquid water within realistic volumes. Conclusions: The use of DNA materials cross-sections in dedicated MC simulations seems to be necessary to a good estimation of the ICS complexity on DNA molecule that cannot be achieved only with liquid water materials and realistic geometries of the target.