ESTRO 36 Abstract Book

S771 ESTRO 36 2017 _______________________________________________________________________________________________

i) collimator angle modified 1º, 2º and 3º, ii) X-jaws modified +2 mm, +5 mm, -2 mm and +5 mm, iii) Y-jaws modified 5 and -5 mm, iv) gantry modified 2º in 4 control points (CP), 1º in 8 CP, 1º in all the control points and finally 2º in all the control points, v) in other plans total UM, that is, total dose, was modified by 1, 2, 3, 5 and 10%, vi) UM for 4 individual CP were modified by 10%, 10% for 8 CP and 20% for 8 CP, vii) the position of all the leafs were modified +0.5, -0.5, +1 and -1 mm, viii) in other plans leafs were modified in each CP in a random way with a maximum displacement of 1, 2 and 5 mm and finally ix) leafs were modified in a random way with a maximum displacement of 2, 5 and 10 mm but using the same displacement for a particular leaf in all the CP. We have compared the dose obtained with the EPID with that calculated by the Pinnacle TPS collapsing the VMAT plans and using the Epiqa software. The gamma 3%/3mm, 2%/2mm and 2%/1 mm have been obtained. We have looked also for visual differences between the dose obtained with the EPID and that obtained with the Pinnacle TPS. Results The analysis of the data shows that some errors can be detected, such as the collimator errors, some X and Y jaws errors, leafs with a systematic displacement, plans with a difference in total dose of 3, 5 and 10% error. Some errors in plans were very hard to detect or undetectable such as that plans with different UM in some control points, different gantry positions, total dose difference of 1 or 2%, and random errors in the leaves in each control point. Conclusion The aS1200 EPID of TrueBeam 2.0 Linac plus the Epiqa software is capable to detect errors in the irradiation of treatment plans although some other errors are undetectable by this system. This makes EPID and interesting dosimetric equipment for the QA of VMAT plans. EP-1461 Scintillator dosimetry reveals lung tumor size dependency of 6 MV AAA dose calculations W. Ottosson 1 , P. Sibolt 1,2 , C.F. Behrens 1 , C.E. Andersen 2 1 Herlev Hospital, Radiotherapy Research Unit- Department of Oncology, Herlev, Denmark 2 Technical University of Denmark, Radiation Physics- Center for Nuclear Technologies, Roskilde, Denmark Purpose or Objective Radiotherapy for lung cancer generally has a poor prognosis. Motion during imaging and treatment is a major challenge, but also other factors may contribute to the poor prognosis. One such factor is the ability of current treatment planning systems to accurately compute absorbed dose to tumors in the thorax region where large heterogeneities are present. The current study was designed to experimentally address the question: What is the agreement between actual delivered dose and computed dose using the Anisotropic-Analytical-Algorithm (AAA) in Eclipse treatment planning system for a thoracic- like geometry with tumors of different sizes? This is an important question given the widespread use of AAA and the changes in tumor sizes both over the course of treatment, and from patient-to-patient. Material and Methods An in-house developed thoracic-like phantom, enabling measurements of radiotherapy under well-defined conditions, was used. The phantom has a body of PMMA and can be filled with inserts of various materials, including simulated spherical lung tumors made of PMMA (ranging from 1-8 cm in diameter) which are embedded in low-density balsa wood that simulates lung-tissue. 14 different phantom setups underwent CT scanning, structure delineation, and treatment planning. 56 isocentric treatments of different complexity and phantom configurations were calculated using AAA. Treatment techniques investigated included single

conventional field technique, four-field conventional box technique, five-field intensity-modulated radiotherapy and dual-arc volumetric-modulated arc technique. To perform accurate dosimetry under these non-reference conditions, point measurements were carried out using water-equivalent, organic plastic scintillator detectors (PSDs), positioned in the center of the PMMA tumors. Dose differences between measurements and AAA calculations were calculated. Results Considerable tumor-size dependence was observed. For tumor sizes ≤ 2 cm, the dose deviations between AAA calculations and PSD measurements were 7.4±1.8% (median ± 1SD). For larger tumor sizes (3-8 cm in diameter) corresponding dose deviations were 4.2±1.4%. For the most homogeneous setup, the dose deviations were insignificant (0.3±0.6%). The results were essentially independent of treatment technique. Conclusion This study suggests a systematic tumor-size dependent dose calculation error for treatment planning on small tumor sizes in heterogeneous setups. This may originate from imperfections in the AAA algorithm. The largest dose deviations were observed for the smallest tumor sizes. Although, it is well known that AAA has issues in heterogeneous regions, we are not aware of any previous experimental study demonstrating a similar systematic tumor-size effect. The effect is large enough to potentially have implications for lung cancer treatment planning. Monte Carlo simulations are currently being conducted in order to verify these findings. EP-1462 The impact on VMAT optimization using Type C vs B algorithms for patients with temporary gas pockets B. Smulders 1 , J. Thomsen 1 , P. Munck Af Rosenschöld 1 1 Rigshospitalet, Department of Oncology, Copenhagen, Denmark Purpose or Objective In our clinic, we have introduced a type C (i.e. Monte Carlo-like) dose calculation algorithm and dose to medium as standard practice. Previous work has shown difference between type B and type C calculation algorithms for dose calculation in high and very low density areas. However, little attention has been given to study the robustness of the treatment plans optimized using type C algorithms during the course of radiotherapy. In particular, the initial CT scan used for radiotherapy treatment planning can contain temporary gas pockets inside the target volume for patients with tumours in the pelvis that later disappear during the course of radiotherapy. In this study we are interested to explore the dosimetric impact of applying the type C dose calculation algorithm for patients treated in the pelvic area using VMAT, where gas pockets appear and disappear in the rectum.

Material and Methods Ten clinical cervix cancer patients were selected for this study. The patients had different sizes of gas pockets on the planning CT. The treatment plans were optimized and calculated using one type B and one type C dose calculation algorithm (Eclipse, AAA and Acuros XB(Dose to Medium), respectively). Gas pockets of these patients