Abstract Book

S1233

ESTRO 37

Material and Methods 10 curative prostate patients were selected to the study and the variability of delineation within 5 observers was analyzed. 3 RTT-s and 2 radiation oncologist delineated each patient bladder, prostate, rectum and bulb of penis. From the 2 radiation oncologist one is experienced with treating prostate patients and one with treating breast and gastrointestinal cancer patients. Delineation was done in Elekta´s Focal Contouring System. After receiving all contours, analyze was done by an independent person. Volume differences between observers were evaluated. Results Mean bladder size was 154.9 cm 3 (87.7-267.1) and mean standard deviation (SD) was 9.4 (5-19.5), mean bulbus penis size was 2.0 (1-0-3.8) and mean SD was 0.6 (0.4-1), mean prostate size was 55.3 (29.8-83.4) and mean SD was 4.1 (1.9-6.8), mean rectum size was 64.3 (41.9-85.0) and mean SD was 3.8 (1.9-7.2). There was no correlation between size of the OAR and SD. The RTT-s tend to delineate bigger volumes than radiation oncologists. Conclusion Although standard deviations in OAR volume differences were in acceptable ranges, delineation protocol was made. While delineating OAR-s in pelvic area window- level (W/L) protocol should be “Soft tissue” and check for the prostate and bladder line should be in “Brain” W/L protocol. While delineating in axial plane, check in sagittal plane should follow. Rectum contour should start from rectosigmoid junction and should stop in anal sphincter. To conclude, single center inter-observer study is a good tool to determine the differences in delineation and compendious delineation protocol helps to minimize variability among observers. EP-2357 A new perspective to evaluation criteria of IMRT plan verification A. Schwahofer 1 , F. Roden 2 , R. Kussaether 3 , M. Grimm 3 , A. Wallin 2 1 German Cancer Research Center DKFZ, Department of Medical Physics in radiation therapy, Heidelberg, Germany 2 MVZ Vivantes Neukoelln- Berlin, Radiation Therapy, Berin, Germany 3 MedCom, GmbH, Darmstadt, Germany Purpose or Objective The Monte Carlo (MC) based verification software ProSomaCore (MedCom, Darmstadt, Germany) is validated as secondary dose calculation system. Therefore, specific DVH and CT value based evaluation criteria were defined and verified in order to provide an alternative to the gamma criteria. Material and Methods The therapy planning system (TPS) Eclipse (Varian Medical Systems, Palo Alto, US) was used as reference system with the superposition algorithm AAA. 40 patients from four different entities were recalculated with ProSomaCore: Mamma, HNO, Lung and Prostate. Since the algorithms inherently show different behaviour in tissues, the gamma criteria is not reasonable anymore in the typical thresholds to compare plans. Thus, a new evaluation criterion was introduced: the relative dose and density parameter – rDD. DVH and dose distribution were separated into the three density regions air (- 1000 Electronic Poster: RTT track: Treatment planning and dose calculation / QC and QA

/ - 171 HU), soft tissue (- 170 / 149 HU) and bones (> 150 HU). This was considered over the whole calculation volume in ten dose categories (see figure 1) by a self- developed script. Subsequently, a set of assessment possibilities were elaborated.

Results The biggest schematic deviations were observed in bony structures over the whole dose range. The AAA overestimates on average over all dose and density regions compared to the MC algorithm. The evaluation shows exponential dose dependence towards higher deviations (~ 20 %) at lower doses and less deviation at higher dose levels (see figure 2) for all three density regions. DVH analysis shows directly, whether a deviation is located in the PTV or in an OAR. Furthermore, the DVH lines show, if there are higher deviations in air, soft or bony tissue for each structure.

Conclusion In general, the concept of gamma evaluation should be reconsidered. Deviation levels can be defined depending on the dose level (low or high dose region).Different new evaluation scenarios are conceivable: 1) DVH analysis with automatic determination of predefined thresholds and classification of OARs as well as PTVs into rDDs. 2) Gamma analysis could be complemented by the three