Abstract Book

S962

ESTRO 37

beams 1, 2 . The focus of this work was to characterise WE- EPID response under flattening filter free (FFF) beams and implement a new model for transit dosimetry using a conventional treatment planning system (TPS). Material and Methods A standard amorphous silicon (a-Si) EPID was modified to a WE-EPID 2 configuration by replacing the metal- plate/phosphor screen situated above the photodiode detector with a 3 x 40 x 40 cm 3 solid water slab. The dose response linearity of the WE-EPID was evaluated using the integrated pixel value per monitor unit (MU) over 5–1000 MU at three different dose rates (nominal, 1000 MU/min and 550 MU/min). A clinical TPS was used to calculate dose to the WE-EPID in its conventional EPID position behind the phantom/patient, with the ‘’extended phantom’’ concept 3 enabling dose calculation at the EPID position. The accuracy of TPS dose calculations at the EPID plane in transit geometry was first evaluated for different field sizes and distance from the beam axis by comparison with dose measured using a 2D ion-chamber array (ICA) and then the WE-EPID. Following basic dose response tests, clinical volumetric modulated arc fields (VMAT) with gantry collapsed to zero were measured. The EPID images were corrected for dark signal and pixel sensitivity 4 and converted to dose using a single dose calibration factor. The 2D dose evaluation was conducted using 3%/3 mm gamma index criteria. All the experiments were conducted with 10MV FFF beam at 150 cm source to detector distance. Results The pixel dose response down to 5 MU was within 2 % of the response at 1000 MU. The WE- EPID agreed with ICA and TPS to within 2 % for field size and off-axis response with open fields in the transit configuration (figure 1). The WE-EPID does not exhibit dose rate response. The average percentage Gamma pass rates for the absolute dose images of the WE-EPID and ICA for all VMAT fields were > 96%(3%/3mm criteria) respectively. The TPS calculated absolute dose for the same VMAT fields also show similar gamma pass rates. Conclusion The accuracy of transit dose measurements with the WE- EPID design was confirmed by close agreement with reference ICA measurements. The advantage of having a WE-EPID for FFF beams is that it does not saturate, unlike conventional EPIDs for high dose rate FFF beams. This study also demonstrates the feasibility of incorporation of the WE-EPID into a commercial TPS for in vivo dosimetry of FFF photon beams. Reference 1.Vial et al Med Phys 35 , 4362-74 (2008) 2.Deshpande et al Med Phys 42, 1753-64 (2015) 3. McNutt et al Med Phys 23 , 1381-92 (1996) 4.Greer Med Phys 32 , 3558-68 (2005) EP-1792 A non-measuring way to compare pre- treatment QA devices and set gamma analysis parameters. M. Gizynska 1 , D. Blatkiewicz 1 , M. Bukat 1 , M. Gil- Ulkowska 1 , S. Maluszczak 1 , A. Paciorkiewicz 1 , D. Szałkowski 1 , A. Walewska 1 1 The Maria Sklodowska-Curie Memorial Cancer Center, Medical Physics Department, Warsaw, Poland Purpose or Objective The IMRT technique is widely used in radiotherapy of many cancer sites. Pre-treatment verification of such plans is very important and in some countries even obligatory. The pre-treatment verification is done to test the machine performance in order to detect errors that would be clinically relevant. Rangel et al. (2010) and Steers et al. (2016) showed the methodology of introducing known errors into the RT plans in order to select gamma criteria on the basis of clinical relevance. The weak point of their studies was unknown

measurement uncertainty and its influence on gamma results. As an alternative we propose to use non- measuring way for testing known errors during delivery of the IMRT plan. Material and Methods Three QA devices were investigated: ArcCHECK (SunNuclear), Octavius (PTW) and EPID (Varian). The Python script changing the dose distribution calculated in TPS into the artificial measurement file has been prepared for each device. This artificial measurement had the same detectors resolution as in the real measurement. We introduced MLC errors into 40 IMRT plans for 4 cancer sites (10 plans each). Head & Neck and Brain with 6MV beams while Prostate and Gynecology plans with 15MV beams. There were introduced two types of systematic MLC errors: gap width and gap position (±0.5, ±1, ±2 and ±3mm for both banks of MLC). DVHs were calculated in order to detect MLC errors with the clinically relevant influence on dose distribution. Artificial measurements were prepared for each plan with MLC error. We performed gamma analysis with criteria: 1%/1mm, 2%/2mm, 3%/3mm both for local and max (global) gamma. Dose threshold was set to 5% for gamma analysis despite of the device. Results Artificial measurements preparation lasts between 1 to 5min. depending on device type and number of fields. We assumed 2% change in PTV mean dose, PTV D98% lower than 95% and PTV D2% greater than 107% as clinically relevant. In Fig. 1 Octavius 2D percent of passing points for gamma analysis is shown. The shaded area shows MLC errors with not clinically relevant change in PTV dose distribution. For different cancer sites different percent of passing points level is required in order to detect errors with clinical impact on PTV dose. In Fig. 2 the comparison of percent of passing points for 2mm/2% max gamma analysis is shown for all tested devices.

Fig. 1 Percent of passing points for Octavius 2D, different gamma criteria and cancer sites.

Fig. 2 Percent of passing points for different devices and the same gamma criteria.