ESTRO 38 Abstract book

S1143 ESTRO 38

justify the selection of electron beam parameters in MC simulations aimed to model small fields and detector specific correction factors. Material and Methods Monte Carlo simulations were performed using the EGSnrc code system. The MC model for the Agility treatment head of the Elekta VersaHD linear accelerator was created and examined in our previous study. The IBA Stereotactic Field Diode was approximated as a silicon chip of relevant dimensions. In the current study, the electron source was presented as a sum of square sub-sources. We divided the electron target area of 4.8 x 4.8 mm 2 into a number of squares of 0.2 x 0.2 mm 2 in the centre region and 0.4 x 0.4 mm 2 in the peripheral region. Then, each part was treated as an independent electron sub-source in (x,y) position on the target surface. The reconstruction of the source intensity distribution was based on a search of the best agreement between simulated and measured data. Measured data was obtained in the form of a 2D array where each value corresponds to an in-air measurement in a certain position of the SFD detector at SSD 100 cm. The weights for each sub-source was optimized to minimize a least-square cost-function:

where A{meas,V} is the measured dose at position V in the 2D array of measured doses; A{MC,V} is the MC dose at position V in the 2D array of simulated doses. At position V in the plane of measurements, A{MC,V} consists of a weighted sum of simulated dose A{MC,V}(x,y) from each sub-source in the (x,y) position on the target surface:

The additional effort for the analyzation is quite low and just takes a few minutes for copying data and starting the tool. All results can be summarized in a printable report file or Excel sheets. The analyzation of the 12 cases with almost 1000 segmented fields resulted in a mean gamma value 0,549 while only 19 segment fields had a maximum gamma of more than 1, based on a 2mm/3% local criteria. Fig 1 shows positions of gamma values >0.8 in the dose. While here results are equally distributed the histogram of leaf pairs (Fig 2) indicates that the area of leaf pair 45 has highest number of >0.8 gamma values of the matrixes. Here it might be worth to look at the leaf calibration even though center leafs are more in use than others. Beside that, patient cases did show good results and thereby supported the positive 3d dose results. A manipulated test case (changed leaf positions and MU settings) was easily detected by gamma index and local dose error results. Conclusion The tool breaks up the complex 3D dose verification measurement to an understandable field based approach and automatically checks the dose delivery in a different way. This might have the potential to easier understand why some patient verification measurements might show higher or lower pass rates. EP-2073 Reconstruction of the electron source distribution using in-air measurements and genetic algorithm E. Borzov 1 , A. Nevelsky 1 , R. Bar-Deroma 1 , I. Orion 2 1 Rambam Health Care Campus - Faculty of Medicine, Oncology, Haifa, Israel ; 2 Ben-Gurion University of the Negev, Nuclear Engineering, Beer Sheva, Israel Purpose or Objective The intensity distribution of a pre-target electron source in Monte Carlo accelerator model is needed for accurate reproducing of measured photon beam data. The purpose of this study is to develop an approach for the electron source reconstruction problem. The approach should

The search for weight factors was performed using a genetic algorithm optimization with rank scaling function and tournament selection function. For benchmarking the proposed technique, the whole procedure was applied in the case where the MC simulation data with the known electron source distribution was used instead of the measured data. Results The benchmark procedure demonstrated that the algorithm reconstructs the source dimensions with accuracy better than 0.01 cm. Using measured data, the matrix of electron intensity distribution was calculated. After a curve fitting procedure, the matrix was found to be elliptical in shape with Gaussian intensity distribution: FWHM x = 0.08 cm and FWHM y = 0.17 cm, where x and y correspond to inplane and crossplane directions. The found source dimensions with the current approach are similar to those found in our previous work using empirical method: FWHM x = 0.10 cm and FWHM y = 0.20 cm. Conclusion Our results demonstrate that the electron source of the Elekta Versa HD linear accelerator has Gaussian elliptical shape. The approach can be used for accurate parameter selection of the electron source when MC simulations are used for calculation of small field correction factors. EP-2074 Can we use Effective Depth for deformable image registration QA alongside the AAPM recommendations? M. Wilson 1 , J. Lui 1 , D. Noble 2,3 , G. Royle 1 , S. Holloway 1,4