Abstract Book

S1027

ESTRO 37

Purpose or Objective Breast tangents remain as one of our field's dominant non-VMAT/IMRT treatment techniques. Modern treatment and optimization techniques are, in many ways, suboptimal for the treatment of tangential superficial disease and may lead to higher low-dose to the lungs and larger volumes of patient receiving radiation that would otherwise be avoided through standard breast tangents. A new approach (Trajectory Optimization in Radiotherapy Using Sectioning (TORUS)) to the treatment of superficial disease has been developed and has been shown to produce plans of higher quality than VMAT and IMRT, with shorter delivery times for sites including chest wall and scalp. In this study, we extend TORUS to generate automated radiation trajectories for left and right sided, intact, breast cancers and compare to conventional tangents, VMAT and IMRT. Material and Methods TORUS avoids degradation of 3D dose optimization quality by mapping the connectedness of target regions from the BEV perspective throughout the space of deliverable coordinates. This connectedness information is then incorporated into a graph optimization problem to define trajectories. The unique usage of two distance functions in this graph optimization permits the TORUS algorithm to generate efficient dynamic trajectories for delivery. 3D dose optimization is performed for trajectories using Aria's Photon Optimizer (version 15.5.1). In this work a new technique for improving coverage of trajectories has been incorporated into the TORUS approach. Results In this study we perform trajectory optimizat ions for 3 intact breast cases, two left and one right sided. Optimizations were performed for which dynamic collimator and gantry motion were permitted, with couch motion constrained to static arrangements. In all cases, TORUS has outperformed VMAT and IMRT with improved organ at risk sparing, conformality, and homogeneity. TORUS has inferred a characteristic class solution with small deviations in starting and stopping positions and couch rotation between cases. Conclusion The TORUS algorithm is able to automatically generate trajectories for breast treatment which have improved plan quality and delivery time than standard IMRT and VMAT treatments. TORUS offers an exciting and promising avenue forward toward increasing the dynamic capabilities of radiation delivery, and may offer improvement over standard breast tangents in difficult to plan cases involving excessive lung and/or heart proximity. EP-1896 Is robust optimization better than virtual bolus method to achieve skin flash in breast VMAT plans? D. Nguyen 1 , C. Corbet 1 , G. Largeron 1 , F. Josserand- Pietri 1 , S. Yossi 2 , M. Khodri 1 1 orlam Macon, Saone Et Loire, Macon, France 2 orlam Charcot, Rhone, Lyon, France Purpose or Objective In this study, we evaluate the virtual bolus method and the Robust Optimization (RO) functionality in Raystation 6.0.24 (RS) treatment planning system to take into account the setup uncertainties and anatomic variations, for Breast VMAT (Volumetric Modulated Arc Therapy) irradiation. Material and Methods Five patients with advanced stage breast cancer were randomly selected for this study. The prescription dose was 50 Gy in 25 fractions. Dosimetric planning goals were determined from normal tissue tolerances and from the clinical experience of radiation oncologists in our clinic. To assess the efficiency of each method, we created VMAT (Nominal), virtual bolus-VMAT and Robust-VMAT

treatment plans for each patient. A setup uncertainty of 15 mm for PTV and OARs was used as robustness setting in the anterior and left or right direction depending on the tumor side. To estimate the robustness of the plan, the isocenter was shifted by 5 mm and 10 mm in the posterior and left or right direction. To simulate anatomical change, we used a virtual bolus of 5mm and 10mm setting in the anterior and left/right direction. The resulting perturbed dose was compared to the nominal plan, optimized without RO or bolus. All plans were compared on the basis of dosimetric end-points, target coverage (D95%) and OAR doses. Results

All the three methods showed good robustness to 5mm shift of the isocenter or with a 5mm anatomical change. In the case of larger changes or offsets, the virtual bolus method seems better than RO and nominal optimization regarding target coverage (p<0,001, t-test).Nominal optimization generates more hot spots than the virtual bolus and RO methods.Regarding OAR, all methods generate the same perturbed dose. Conclusion A virtual bolus and a “robust” optimization methods for breast VMAT plans, introducing a “skin flash” and maximizing the probability of satisfying the planning criteria in the presence of respiratory motion and setup uncertainty, has been evaluated. Only the virtual bolus method maintains target coverage and prevents the increase of hot spots. We observe similar results by calculating on CTs for which we found real anatomical variations. EP-1897 A multi-centre comparison of automated treatment planning for prostate cancer D. Roach 1,2 , C. Rønn Hansen 3,4,5 , G. Wortel 6 , H.R. Jensen 3 , C. Ochoa 7 , E. Damen 6 , P. Vial 2,4,7 , T. Janssen 6 1 University of New South Wales, Faculty of Medicine, Sydney, Australia 2 Ingham Institute for Applied Medical Research, Medical Physics, Liverpool, Australia 3 Odense University Hospital, Laboratory of Radiation Physics, Odense, Denmark 4 University of Sydney, Institute of Medical Physics- School of Physics, Sydney, Australia 5 University of Southern Denmark, Institute of Clinical Research, Odense, Denmark 6 The Netherlands Cancer Institute, Department of Radiation Oncology, Amsterdam, The Netherlands 7 Liverpool and Macarthur Cancer Therapy Centres, Department of Radiation Oncology, Liverpool, Australia Purpose or Objective To investigate if automated treatment planning techniques can be adapted across departments to different local protocols for prostate radiotherapy, providing an assessment of whether such techniques have potential to improve treatment plan quality during future clinical trials. Material and Methods Three treatment centres, two from Europe and one from Australia, and each using Pinnacle’s Autoplanning (AP) software, distributed three prostate patient datasets and local protocols as learning datasets for AP technique development. Centres modified local prostate AP techniques to create protocol compliant AP techniques for each centre. Dose objectives were graded as critical, hard, medium, or soft by the host-centre. The learning