Abstract Book

S932

ESTRO 37

had occurred in the unusual '60%” phase. Knowledge of the maximum exhalation is necessary to determine the peak-exhale location of the tumor. It should be noted that this could be a potential source of errors. MIP images of patients reconstructed using both systems were compared among them. Mean values of DSC in patients were 0,952±0,020. . Conclusion The innovative binning of the Deviceless4D requires new testing methods that are suited for exactly those types of sorting systems. Proposed CSP device is capable of this task. The way to minimize the number of cine stitching points should be also discussed, as they affect reconstructed images. Obtained results have shown that, despite the differences in reconstruction techniques, both of the studied systems allow for correct determination of the volumes of interest. EP-1740 3D dose reconstruction in the patient from VARIAN EPID images for IMRT and VMAT treatments F. Younan 1 , J. Mazurier 1 , X. Franceries 2,3 , D. Franck 1 1 Oncorad Garonne, Haute Garonne, Toulouse, France 2 Centre de Recherche en Cancérologie de Toulouse CRCT, Haute Garonne, TOULOUSE, France 3 Laboratoire Plasma et conversion d'Energie LAPLACE, Haute Garonne, Toulouse, France Purpose or Objective The EPID panel (electronic portal imaging device) is increasingly used for dosimetric purposes in external radiotherapy. Two methods are currently available to verify the absorbed dose in the patient: the direct method, which consists on predicting the fluence, acquired with EPID, and the indirect one which redistribute the absorbed dose into the patient. This latter, also referred as a back-projection method, is used in our study to reconstruct the 3D absorbed dose matrix in the patient from EPID images for IMRT and VMAT fields. Material and Methods Images are acquired with the clinac 23iX aS-1000 imager (Varian) and a 6MV beam. To compensate for the lack of uniformity of the detector response and to take into account the influence of the backscattered photons coming from the robotic gantry, a pre-treatment is applied to the portal images. A calibration step is then performed to transform the gray levels of the pixels into absorbed dose in water via a response function. Correction kernels are also used in this step to redistribute the absorbed dose according to the field size and to take into account the penumbra. The next step is the retro-projection of the dose into the patient for all EPID parallel plans with a 1 mm resolution, for each gantry angle. Our algorithm, developed with Matlab version R2015b takes into account the photons scattered into the patient, the attenuation, the beam hardening with depth, the build-up effect, and the beam’s divergence into the patient. Finally, the total dose is obtained by summing the 3D dose associated for each gantry angulation. The estimated dose from the EPID was compared to the one calculated by the TPS (Treatment Planning System) with a local and global gamma index of 3% and 3mm also developed with Matlab. Our algorithm has been tested for 20 IMRT and VMAT prostate and head and neck treatments in a homogeneous phantom (cylinder of 25 cm diameter). Results The global gamma index obtained is greater than 99% and more than 90% for local gamma index (3%,3mm). Conclusion

We have developed and validated a reconstruction algorithm that can be used to verify the 3D dose distribution for the IMRT and VMAT fields from in vivo EPID images. EP-1741 X-ray/Light field Congruence with varying gantry angle using the EPID F. Tato de las Cuevas 1 , J. Yuste López 1 , F. Fernández Belmonte 1 , I. Ribót Hernández 1 1 Hosp. Univ. de Canarias, Medical Physics Dept., Santa Cruz de Tenerife, Spain Purpose or Objective To develop a method to perform X-ray/Light field Congruence ( XLC ) QA using the EPID (Electronic Portal Imaging Device). The method must be fast, accurate and capable of make the analysis for different gantry angles. Material and Methods The LINACs: 1) CLINAC 2100 with Millenium MLC with 6 MV and an EPID with 512x384 pixels, 2) Elekta Synergy with Agility MLC, 6 MV and EPID with 1024x1024 pixels. Field size used is 20x20 cm. A jig is made with a Bearing ball (BB) in its centre and 4 moving radiopaque discs. The jig is aligned with the crosshair light projection over the EPID, the discs are placed inside and tangent the light field. The moving discs allow to work with nominal field size and permit the images to be acquired in clinical mode and exported from the network easily. The images are analysed with an In- House Software ( IHS ): - Rotation correction of the image is performed (when angle > 0.1º). - Radiation field limits are obtained with the 50 % pixel value (PV). - 3 profiles parallel to every field edge are acquired. Discs edge points are obtained with subpixel precision from the 50 % PV. These six disc edge points are used to obtain geometrically every disc centre. Light field edge is obtained from the nominal disc radio and centre. - The BB position is obtained from Gaussian curve fit of cross-profiles over the max PV in central ROI, this gives the cross-hair light field projection over the EPID. Validation of IHS 1. Image J software and IHS are used to perform XLC measurements of 10 EPID Elekta images. XLC analysis is made with RIT113 software and Vidar VXR-16 scanner. Every operator marks each light field edge twice, punching the envelope with a sharp needle. Radiation field edge is determined with the 50% of the central dose. XLC is measured with RIT software tools. LINACs QA EPID images are acquired following the steps described above. 15 tests are made in each LINAC (one test/day) for 3 gantry angles (0, 90, 270º), i.e.: 45 images per LINAC. Results IHS validation 1. The image analysis with the algorithm (IHS) and with the external software (Image J) gives discrepancies in XLC of less than one pixel. 2. Fig. 1 shows the congruence spread is slightly bigger for the film than for the IHS . 2.