ESTRO 38 Abstract book

S485 ESTRO 38

Szałkowski 1 , A. Walewska 1 1 Maria Skłodowska-Curie Memorial Cancer Center and Institute of Oncology, Medical Physics Department, Warsaw, Poland ; 2 University of Warsaw Faculty of Physics, Department of Biomedical Physics, Warsaw, Poland Purpose or Objective An important question which should be answered prior to pre-treatment measurement is whether we want to check linac performance and detect all MLC errors or do we want to detect only those errors with clinically relevant impact on dose distribution. Some groups reported capability of different gamma parameters to detect clinically relevant errors. In our work we propose to use Receiver Operating Characteristic (ROC) in order to compare three different QA devices and choose gamma analysis parameters – especially tolerance level for percent of passing points (PPP). Material and Methods Forty clinical IMRT plans (10 plans for each of tumor localizations: gynecology, prostate, brain, head and neck) were used in our study. All plans were clinically accepted and used for patients treatment. Sixteen known MLC errors of different magnitude were introduced for each plan. Dose distribution was recalculated in TPS (Eclipse AAA 13.6.23) on the patient CT. We created verification plan for all plans (original and with error) for three QA devices: ArcCHECK (SunNuclear), Octavius 729 (PTW), EPID (Varian). In order to do the bias-free evaluation, artificial measurements were created based on the dose calculations in TPS. Artificial measurement is a file created with a Python script on the basis of TPS dicom dose distribution which mimics the real measurement file. All artificial measurements were compared with original verification plan in adequate software (SNCPatient, Verisoft, Portal Dosimetry). Gamma analysis was performed with 5% threshold. Tested parameters were: 3mm/3%, 2mm/2%, 1mm/1% for both max and local evaluation. From dose comparison between the clinical plan and error plans we flagged each of 680 plans with Positive or Negative, where Positive meant clinically relevant change in PTV dose (i.e. D98%<95%, D2%>107% or ∆Dmean>2%). While changing the gamma evaluation percent of passing points tolerance level we calculated sensitivity (True Positive Ratio) and specificity (True Negative Ratio) for all tested devices. ROC curves were evaluated for each site and gamma parameters. Results In Figure 1 ROC curves for 3mm/3% max gamma evaluation are shown. Octavius 2D (with 729 array) performs slightly better than EPID. Area under ROC curve for ArcCHECK is smaller thus showing lower performance in detecting clinically relevant errors. It can be seen that 98% tolerance for Octavius and EPID does not balance between Sensitivity and Specificity. ROC curves obtained for 3mm/3% local gamma evaluation (Figure 2) show that ArcCHECK tolerance level should be set to 90% while EPID and Octavius tolerance levels should be 95%-98%. For these parameter all detectors performed similarly.

Conclusion ROC analysis for pre-treatment QA device can help in setting the tolerance level of PPP for gamma analysis aimed to detect clinically relevant dose distribution change. Our study showed that for 3mm/3% gamma evaluation ArcCHECK tolerance level should be set to lower value than that for Octavius and EPID.

Poster: Physics track: Radiation protection, secondary tumour induction and low dose

PO-0912 Effect of Heart Anatomy on Radiation Related Cardiac Risk in the Childhood Cancer Survivor Study R. Howell 1 , J. Bates 2 , L. Qi 3 , B. Hoppe 2 , C. Lee 4 , S.A. Smith 1 , K.C. Oeffinger 5 , L.S. Constine 6 , W.M. Leisenring 7 , D.A. Mulrooney 8 , T.M. Gibson 8 , G.T. Armstrong 8 , Y. Yasui 8 1 University of Texas at MD Anderson Cancer Center, Radiation Physics, Houston- TX, USA ; 2 University of Florida, Radiation Oncology, Gainsville- FL, USA ; 3 University of Alberta-, Public Health Sciences, Edmonton- AB, Canada ; 4 Radiation Epidemiology Branch, National Cancer Institute, Bethesda- MD, USA ; 5 Duke Universit, Department of Medicine, Durham- NC, USA ; 6 University of Rochester Medical Center, Department of Radiation Oncology, Rochester- NY, USA ; 7 Fred Hutchinson Cancer Research Center, Clinical Research Division, Seattle- WA, USA ; 8 St. Jude Children’s Research Hospital, Department of Epidemiology and Cancer Control, Memphis- TN, USA