Abstract Book

S268

ESTRO 37

Of note, the SpNTCP modeling of real lung RIT was more accurate than LKB, seemingly in agreement with the suggested inhomogeneous RSM of the lungs [Rubin 2013]. In conclusion, we believe that the proposed germinal idea could pave the way toward a new NTCP modeling philosophy.

Proffered Papers: PH 10: Treatment planning 2

OC-0513 In-silico comparison of five automated treatment planning solutions for primary head and neck cancer J. Krayenbuehl 1 , M. Zamburlini 1 , S. Ghandour 2 , M. Pachoud 2 , S. Lang-Tanadini 1 , S. Tol 3 , M. Guckenberger 1 , W. Verbakel 3 1 University Hospital Zürich, Department of Radiation Oncology, Zurich, Switzerland 2 Hôpital Riviera-Chablais, Department of Radiation Oncology, Vevey, Switzerland 3 VU University Medical Center, Department of Radiotherapy, Amsterdam, The Netherlands Purpose or Objective Automated treatment planning and/or optimization systems (ATPS) are in the process of broad clinical implementation in order to reduce the inter-planner variability, improve workflow by reducing the planning time allocated for the optimization process and to improve the plan quality. Different ATPS use different approaches and no comparison has yet been performed In the present in-silico planning study, three radiation oncology departments compared 5 different ATPS: 1) Automatic Interactive Optimizer (AIO) in combination with RapidArc (in-house developed and Varian Medical Systems); 2) Auto-Planning (AP) (Philips Radiation Oncology Systems); 3) RapidPlan (RP) version 13.6 with H&N model from University Hospital A (Varian Medical Systems, Palo Alto, USA); 4) RP version 13.7 combined with scripting for automated setup of fields with H&N model from University Hospital B; 5) Raystation multicriteria optimization algorithm version 5 (RS) (Labora tories AB, Stockholm, Sweden). 7 Randomly select ed locally advanced head and neck cancer cases from Hospital A and 7 from hospital B were used. Two cases for each group were used to get familiar with the target volume, concepts and dose constraints and the other five were included in the planning comparison. A single optimization for each ATPS was performed on the comparison group. P lans were compared based on PTV coverage, mean dose to different organs at risk and effective planning time. An overall ranking was carried out by summing the ranking of different group of DVH parameters (PTV, serial organs,swallowing muscles, swallowing organs and dose conformity). Results All planning systems achieved the PTV dose constraints for the two sets of patients. Averaged over each set, the differences in mean OAR doses between the ATPS were small, always within a standard deviation. Table 1 shows as example the comparison of the averaged mean doses to OAR for the 2 sets between the different ATPS. For the five cases from set A, RS had the best ranking followed by AIO, AP and both RP. For set B, RS ranked best followed by RP_B, AIO and AP ranked equally, followed by RP_A. The effective working time required after volume definition by the clinicians to the end of the optimization process was <1 minute for plans optimized with AIO and RP_B. The effective working time was substantially longer for AP, RP_A and RS with 3-6, 12-17 and 336- between the systems. Material and Methods

On real patients, the area under the ROC curve of the SpNTCP was 0.72 (CI: [0.62-0.81]), significantly higher than 0.61 (CI: [0.51-0.70]) obtained by the LKB ( p <0.002). Conclusion The proposed SpNTCP model, differently from LKB, proved able to distinguish the toxicity effects of given dose distributions in presence of inhomogeneous RSM, even for coincident structure DVHs (Fig. 2). Interestingly, our approach in principle also allows for mapping a posteriori the organ RSM, as suggested by Figs. 1a-b and 2a-b. As expected, the performance of the SpNTCP model, at least in this preliminary format, decreases as the volume effect decreases.