S456 3rd ESTRO Forum 2015 Conclusions: An experimental set-up for dose determination in heterogeneous media in proton beam therapy was developed. Various dose engines using different techniques and computer technologies were compared to data. Overall, MC simulations show better agreement with experimental data than pencil beam algorithms, including the superfast MCSQUARE and Raysearch-MC, which is promising for future clinical use. The data obtained in the phantom experiment will be used as a basis to interpret disagreements between dose distributions calculated by MC and analytical algorithms in clinical cases. This study is financially supported by the Walloon Region and IBA under the project name IVDGPT, convention number 1217662. PO-0889 A multi-institutional study for the evaluation of DIR algorithms for structure delineation in virtual phantoms C. Fiandra 1 , G. Loi 2 , M. Fusella 1 , E. Lanzi 3 , G. Orlandi 4 , L. Strigari 5 , L.C. Orlandini 6 , M. Stasi 7 , M. Paiusco 8 , E. Menghi 9 , F. Lucio 10 , A. Carosi 11 1 Università di Torino, Department of Oncology, Torino, Italy 2 University Hospital Maggiore della Carità, Department of Medical Physics, Novara, Italy 3 Tecnologie Avanzate, R & D Department, Torino, Italy 4 Ospedale Civile Giuseppe Mazzini, Department of Medical Physics, Teramo, Italy 5 Regina Elena National Cancer Institute, Department of Medical Physics, Roma, Italy 6 Centro Oncologico Fiorentino - CFO, Department of Medical Physics, Sesto Fiorentino, Italy 7 Mauriziano Hospital, Department of Medical Physics, Torino, Italy 8 Veneto Institute of Oncology IOV - IRCCS, Department of Medical Physics, Padova, Italy 9 Istituto Scientifico Romagnolo per lo Studio e la Cura dei Tumori (IRST) IRCCS, Department of Medical Physics, Meldola, Italy 10 'Santa Croce e Carle' Hospital, Department of Medical Physics, Cuneo, Italy 11 Tor Vergata University General Hospital, Department of Medical Physics, Roma, Italy Purpose/Objective: This study investigates the accuracy of various algorithms for deformable image registration (DIR), to propagate regions of interest (ROIs) in virtual phantoms using different commercial systems. This work is a preliminary step to provide a consistent quality assurance check among different institutions on the use of DIR before its clinical implementation for ROIs and dose propagation in adaptive radiotherapy. Materials and Methods: Ten Italian institutions with 4 available commercial solutions provided data to assess the agreement of DIR-propagated ROIs with automatically drown ROIs considered as ground-truth for the comparison. The DIR algorithms were tested on a specific virtual phantom made of three cylinders of different gray density inside a uniform cube phantom (CT1, Fig 1a) with two data sets obtained by specific Deformation Vector Field (DVF) applied to the reference data set (CT2 and CT3). The different software used in this study are based on various algorithms: a multi- resolution modified basis Spline (B-Spline), a radial basis function, an intensity-based free-form, an Hybrid intensity and structure based and a Biomechanical model based. The DIR-mapped ROIs were then compared with the reference ROIs using the Dice Similarity Coefficent (DSC) and the Mean of the Hausdorff Distance (MHD). Results: Figure 1a shows examples of the DIR-propagated ROIs while figure 1b shows the DSC for the three considered ROIs. Mean values for DSC were 0.93±0.08, 0.94±0.04 and 0.92±0.12 respectively for the three considered ROIs (Cylinder 1, Cylinder 2 and Cylinder 3) on CT2; 0.93±0.08, 0.94±0.04 and 0.94±0.11 were the average values on CT3. Regarding MHD values we obtain 0.14±0.23, 0.19±0.31 and 0.22±0.47 for CT2 while 0.09±0.15, 0.08±0.06 and 0.17±0.39 are the values obtained for CT3. Minimum and maximum values for DSC and MHD resulted respectively 0.65 and 1.23mm Conclusions: Although the different algorithms used in this study are significantly dissimilar in their approach to brought to you by CORE View metadata, citation and similar papers at core.ac.uk provided by Elsevier - Publisher Connector