ESTRO 36 Abstract Book

S777 ESTRO 36 2017 _______________________________________________________________________________________________

characteristics (80 or 120 leaves) are indicated.

profiles showed qualitatively good agreement between the gel dosimeter, EBT film and RTP data for all PTVs. Conclusion The results indicate that those processes could effectively evaluate geometric and dosimetric accuracy of brain SRT. This study using 3D dosimetry system was useful to validate the 3D dose distributions for patient-specific QA. EP-1473 Improving the accuracy of dosimetry verification by non-uniform backscatter correction in the EPID Y. Md Radzi 1,2 , R.S. Windle 2 , D.G. Lewis 2 , E. Spezi 1,2 1 Cardiff University, School of Engineering, Cardiff, United Kingdom 2 Velindre Cancer Centre, Department of Medical Physics, Cardiff, United Kingdom Purpose or Objective Challenges in improving the accuracy of EPID-based patient dose verification have been widely discussed and remain a key topic of interest for patient safety, as exemplified in the UK by the ‘Towards Safer Radiotherapy’ 2008 report[1]. In particular, one of which is for every radiotherapy centre to have protocols for in vivo dosimetry (IVD) to be used for most patients as recommended in the Annual Report of the Chief Medical Officer for 2006 and it is already a legal requirement in many European Countries [2]. In this presentation, we report on commissioning and implementation of the commercially available Dosimetry Check (DC) [3, 4] system. Particular emphasis has been given to addressing the significant non-uniform backscatter effect from the VARIAN aSi-1000 EPID arm [5, 6]. Material and Methods A backscatter correction matrix was developed by combination of dosimetric information from a set of segmented fields sampling on different positions around the active area of the imager. The matrix was then used to correct EPID images using MATLAB programming scripts. The corrected image was created in DICOM format and exported to Dosimetry Check to read and analyse. Example treatment fields were generated in our Oncentra MasterPlan (OMP) Treatment Planning System (TPS), with several equidistant dose reference points relative to central axis included. A dose comparison given by DC with reference to the TPS was recorded in an auto-generated report. Assessment and comparison undertaken included the (i) asymmetry evaluation of equidistant points before and after correction being applied with respect to TPS, (ii) improvement in segmented IMRT dose profiles after correction, and (iii) OMP-DC pass rate with gamma criterion 3%/3mm[7], as well as 2-D Gamma Volume Histogram (GVH) evaluation on outlined PTVs. Results (i) Correction for non-uniform backscatter improved with overall agreement between fields generated in OMP and those recorded in DC from within 3% to better than 1%. (ii) Agreement between OMP and DC for IMRT dose profiles with a sample Head & Neck case was improved by approximately 3% using the correction methodology ( Table 1 ). (iii) For gamma comparison of fields in OMP and DC with 3%/3mm, pass rates were improved from around 80% to around 90% by the correction method. Similarly in GVH evaluation for the outlined PTVs, pass rate has increased from around 80% to 90% after correction being applied.

Fig 1. Dose volume histogram for the whole body averaged over the 10 patients of this study, comparing every treatment technique. Conclusion The source for higher values of ID and NTID for HT is the larger volume receiving dose below 20 Gy. No differences were found in the election of IMRT delivery. For RapidArc plans, ID and NTID values are similar to IMRT. EP-1472 Dosimetric E2E verification using 3D printing and 3D dosimeter for brain stereotactic radiotherapy M.S. Kim 1 , K.H. Chang 1 , J. Kwak 1 , G.M. Back 1 , T.Y. Kang 1 , S.W. Kim 1 , Y. Ji 1 1 Asan Medical Center- Univ of Ulsan, Radiation Oncology, Seoul, Korea Republic of Purpose or Objective To evaluate the dosimetric accuracy of brain stereotactic radiotherapy (SRT) with a 3D dosimetry system and MRI, we investigated dosimetric end-to-end verification using 3D printing technology and 3D dosimeter. Material and Methods We implemented an anthropomorphic head and neck phantom with a 3D printed insert made using a 3D printer designed by the Autodesk software and two gel-filled spherical glass flasks as a patient having multiple target brain cancer. For the feasibility study of the gel dosimeter, the dose linearity, dose rate dependence, and reproducibility for the gel dosimeter were verified. Gel- filled vials were irradiated with 6 MV beams to acquire a calibration curve of dose relation to R2 (1/T2) values in 9.4T MR images. Graded doses from 0 to 8 Gy with an interval of 2 Gy were delivered. Two PTVs (PTV1,2) were contoured on the MR images of phantom have dosimetric gel tumor. To evaluate geometric and dosimetric accuracy, a treatment plan was created such that D95s for PTV1 and intentional PTV2 were more than the prescribed dose. The intentional PTV2 was produced by intentionally shifting by 5mm from the true target position. 2 arc VMAT plan was created to deliver 35 Gy in 5 fractions. After irradiation, calibration vials and phantom were scanned by 9.4T MRI and then acquired images were analyzed using an ImageJ and DCMTK software libraries. Scanned MRI images of phantom were imported to a treatment planning system and registered to CT images to compare dose distributions. We also compared the agreement result between the planned and the measured data in 1D (ion chamber), 2D (gafchromic film), and 3D (Gel dosimeter). Results The best dose linearity was 0.99 (R 2 ) at 180 TE (ms). Reproducibility and dose rate dependency were less than 2.2% and 3.5%, respectively for 180 TE. Point dose differences in plan vs. ion chamber were 1.08%, 0.47%, and -2.82%, respectively, for PTV 1, 2, and intentional PTV. And its differences between plan and gel were 0.98%, 1.66% and 3.76%, respectively, for PTV 1, 2, and shifted PTV. Gamma passing rates with 3%/3mm criteria were greater than 99% for all plans. Isodose distributions and