ESTRO 37 Abstract book

S1185

ESTRO 37

Radiotherapy, Utrecht, The Netherlands 6 Radboud University Medical Center, Radiation Oncology, Nijmegen, The Netherlands 7 Oscar Lambret Comprehensive Cancer Center, Academic Radiation Therapy Department, Lille, France 8 University Hospital Antwerp, Iridium Kankernetwerk Radiotherapy, Antwerp, Belgium 9 GZA Ziekenhuizen, Iridium Kankernetwerk Radiotherapy, Wilrijk, Belgium 10 University Hospital Zurich, Department of Radiation Oncology, Zurich, Switzerland 11 University Hospitals Leuven, Department of Radiation Oncology, Leuven, Belgium 12 Erasmus MC, Department of Radiology and Nuclear Medicine, Rotterdam, The Netherlands Purpose or Objective The Dice similarity coefficient (DSC), defined as the overlap of two volumes divided by their average, is often used to quantify variations in delineation of target volumes and OARs. We studied correlations between DSCs of delineations of the same patient by different clinicians and the resulting dosimetric differences in treatment planning. Fully automated planning was used to reduce variations in the quality and consistency of treatment plans. Material and Methods Ten centers participating in a randomized trial on SBRT for hepatocellular carcinoma (TRENDY trial) submitted delineations of the clinical target volume (CTV = GTV in this trial) and the liver on the same CT scan of the same patient. For delineation, also a diagnostic MR-scan was available. For each of these ten contour sets, a VMAT plan was made, using a fixed 5mm CTV-PTV margin and equal contours for the other OARs (stomach, kidneys, oesophagus, bowel, gallbladder, heart and spinal cord). SBRT plans, delivering 54Gy (D95% of the PTV) in six fractions in line with the trial protocol, were automatically generated. An NTCP for the healthy liver (liver minus CTV), considered crucial for patient safety, was calculated and evaluated. For each dose distribution, the contours used for planning and the other nine contour sets were mutually compared with regards to target dose (D95%-PTV) and liver NTCP (for 10 contour sets, 90 comparisons in total), and correlations between changes in these dosimetric parameters and the corresponding DSCs for PTV and liver were investigated. Results The observed changes in D95%-PTV and NTCP are shown respectively in the figures below. The D95%-PTV, which in all cases was equal to 54Gy when evaluated on the contours used for planning, was on average 10.1 Gy lower when evaluated on the other contour sets (p < 0.001). The NTCP, which ranged from 0% - 2.9% on the contours used for planning, ranged from 0% - 12.5% when evaluated on the other contour sets, and was on average 0.6 percentage point larger (p < 0.001). Poor correlation (R 2 = 0.17) was found between PTV DSCs and changes in D95%-PTV, and hardly any (R 2 = 0.03) between liver DSCs and changes in NTCP.

specifically to each registration or automatic for a large image dataset, such as 4DCT. Our registration QA software opens the possibility of integrating DIR into a clinical enviroment. EP-2143 Influence of number of projections on carbon computed tomography reconstruction D. Shrestha 1 , J. Wang 1 1 UT Southwestern Medical Center, Radiation Oncology, Dallas, USA Purpose or Objective In heavy ion radiation therapy, an accurate localization of the Bragg peak reduces range uncertainties and provides greater dose conformity while sparing the healthy tissues. Computed tomography using carbon ions (carbon CT) can reduce range uncertainty of heavy ions by reconstructing 3D relative electron density map of a patient body directly. The goal of this work is to systematically investigate the influence of the number of projections on the carbon CT image quality reconstructed by an iterative algorithm. Material and Methods The Geant4 toolkit was used to simulate carbon CT projections of two different phantoms with inserts of various materials and line pair (lp) densities ranging from 1.67 lp/cm through 5 lp/cm. For each phantom, a different number of projections, including 15, 30, 45, 60 and 90, were obtained with 10^6 ions/projection using 430 MeV/u carbon ions. An event-by-event reconstruction algorithm based on algebraic reconstruction technique coupled with total variation minimization (ART-TV) was used to reconstruction carbon CT after the phantom boundary was extracted from a first-phase approximate reconstruction without considering the accurate boundary of the object. Quantitative measurements, including the relative mean squared error (RMSE) and range estimation error, were used to characterize the performance of ART- TV with a different number of projections. Results When the number of projections increases from 15 to 90, the average of RMSE of the two phantoms decreases from 6.8% to 2.4% and the range estimation error for 200 MeV/u carbon ions decreases from 2.3 mm to 0.2 mm, respectively. When the number of projections increases from 60 to 90, the average of RMSE of the two phantoms decreases less than 0.1% and the range estimation error for 200 MeV/u carbon ions decreases less than 0.1 mm. Conclusion The quality of carbon CT is greatly affected by the number of projections when the projection number is less than 30. The image quality, measured by RMSE and range estimation error, becomes stable when the number of projections is more than 60. EP-2144 Tumor and OAR delineation variation – Dice coefficient versus dose assessed with automated planning S.J.M. Habraken 1 , A.W.M. Sharfo 1 , J. Buijsen 2 , W.F.A.R. Verbakel 3 , C.J.A. Haasbeek 3 , M.C. Ollers 2 , G.H. Westerveld 4 , N. Van Wieringen 4 , O. Reerink 5 , E. Seravalli 5 , P.M. Braam 6 , M. Wendling 6 , T. Lacornerie 7 , X. Mirabel 7 , R. Weytjens 8 , L. Depuydt 9 , S. Tanadini-Lang 10 , O. Riesterer 10 , K. Haustermans 11 , T. Depuydt 11 , R.S. Dwarkasing 12 , F.E.J.A. Willemssen 12 , B.J.M. Heijmen 1 , A. Méndez Romero 1 1 Erasmus MC Cancer Institute, Department of Radiation Oncology, Rotterdam, The Netherlands 2 MAASTRO Clinic, Department of Radiation Oncology, Maastricht, The Netherlands 3 VU University Medical Center, Radiation Oncology, Amsterdam, The Netherlands 4 Academic Medical Center, Radiotherapy, Amsterdam, The Netherlands 5 University Medical Center Utrecht, Department of