ESTRO 36 Abstract Book

S432 ESTRO 36 _______________________________________________________________________________________________

lines. Negative MLC errors were not performed at Institution 2, due to differing equipment. The automated VMAT plans from institution 3 were similar in pass rate to the manually planned VMAT for collimator errors, despite the difference (higher magnitude for manual VMAT plans) in error magnitude. This could be caused by the higher MLC modulation in the automated plans. Conclusion Not all deliberately introduced errors were discovered for VMAT plans using a typical 3%/3mm global gamma pass rate (for 10% threshold with correction off). Consistency between institutions was low for plans assessed utilising differing devices and software. A 2%/2mm global analysis was most sensitive to errors. PO-0809 A 3D polymer gel dosimeter coupled to a patient-specific anthropomorphic phantom for proton therapy M. Hillbrand 1 , G. Landry 2 , G. Dedes 2 , E.P. Pappas 3 , G. Kalaitzakis 4 , C. Kurz 2 , F. Dörringer 2 , K. Kaiser 2 , M. Würl 2 , F. Englbrecht 2 , O. Dietrich 5 , D. Makris 3 , E. Pappas 6 , K. Parodi 2 1 Rinecker Proton Therapy Center, Medical Physics, Munich, Germany 2 Ludwig-Maximilians-Universität München, Department of Medical Physics, Munich, Germany 3 National and Kapodistrian University of Athens, Medical Physics Laboratory- Medical School, Athens, Greece 4 University of Crete, Department of Medical Physics, Heraklion, Greece 5 Ludwig-Maximilians-Universität München, Department of Radiology, Munich, Germany 6 Technological Educational Institute, Radiology & Radiotherapy Department, Athens, Greece Purpose or Objective The high conformity of proton therapy (PT) dose distributions, attributed to protons stopping in the target, is also the main source of uncertainty of the modality. PT is sensitive to errors in relative stopping power to water (RSP) uncertainties and to density changes caused by organ motion. The ability to verify PT dose distributions in 3D with a high resolution is therefore a key component of a safe and effective PT program. Existing 2D dosimetric methods suffer from shortcomings attributed to LET dependence, positioning uncertainties, limited spatial resolution and their intrinsic 2D nature. Recent advances in polymer gel dosimetry coupled to 3D printing technology have enabled the production of high resolution, patient specific dosimetry phantoms. So far this approach has not been tested for PT. Material and Methods A 3D-printed hollow head phantom derived from real CT data was filled with VIPAR6 polymer gel and CT scanned for pencil beam scanning (PBS) treatment planning, following RSP characterization of the gel and the 3D printer bone mimicking material (see Figure 1). All irradiations of phantoms were carried out at the Rinecker Proton Therapy Center in Munich, which is dedicated for PBS. An anterior oblique SFUD plan was used to cover a centrally located cerebral PTV, following the standard operating procedures of the PT facility. The field was crossing the paranasal sinuses (see Figure 2A) to test the TPS modelling of heterogeneities. 3D maps of the T2 relaxation time were obtained from subsequent MR scanning of the phantom and were converted to relative dose. The dose response linearity and proton range were verified using separate mono-energetic irradiations of cubic phantoms filled with gel from the same batch. Relative dose distributions were compared to the TPS predictions using gamma analysis.

different equipment to simulated machine errors and explores the role of different planning approaches to this Material and Methods VMAT plans were generated for a selected patient in Pinnacle 3 at three institutions, as per their local protocol. An automated VMAT plan was also generated by institution 3 using Pinnacle 3 Autoplanning. Simulated machine errors were deliberately introduced to the plans utilising Python. These included collimator (°), MLC field size (mm) and MLC shift (mm) errors of 5, 2, 1, -1, -2 and -5 units. Error- introduced plans were then recalculated and reviewed. The DVH metrics listed in Table 1 were deemed unacceptable if their differences relative to the relevant baseline plan were above the tolerances listed. Plans were considered unacceptable if any one or more of the limits were exceeded. Table 1. DVH metrics and limits. For each error type (i.e. in Collimator, C; MLC shift, S; MLC Field Size, FS), the smallest error plans that were deemed unacceptable were delivered within the given institution; on an Elekta Linac, measured using an Arccheck for institutions 1 and 3, and on a Varian Linac , measured using a Delta4 for institution 2. Gamma analysis was performed in SNC Patient version 6.6 or Delta4 software respectively, utilising a 3%/3mm and 2%/2mm global gamma pass rate (10% isodose threshold with correction off). Before each set of measurements, MLC checks and a complex benchmark patient test were used to ensure the Linacs' performances were within normal range. Results The global 3%/3mm gamma pass is able to detect the majority of unacceptable plans; however some plans with significant errors still pass. Interestingly the error type/s that passed differed at differing institutions (Figure 1).

Figure 1. The smallest error plans (including Collimator (C), MLC shift (S), and MLC Field Size (FS) error) which exceeded global gamma pass rates. Errors detected if the gamma pass rate was < 95% (for 3%/3mm) or <88% (2%/2mm). Plans that passed are illustrated above the red