Abstract Book

S1016

ESTRO 37

suggested that 4Drobust is more robust for the chest wall ( p=0.045 ) and LungV5 ( p=0.001 ). Among the three pairs, the results showed statistically insignificance for GTVmin, GTVmean and GTVmax ( p>0.05 ).

configured using 5 training patients, based on their VMAT manual plans. Subsequently, VMAT auto and VMAT manual plans were compared for an independent set of 30 evaluation patients through dose-volume parameter analysis and blinded physician scoring. Results All VMAT auto and VMAT manual plans met all OAR constraints and were considererd clinically acceptable. OAR sparing and plan conformity were substantially improved in VMAT auto plans compared to VMAT manual . Mean doses were statistically significantly lower for rectum, bladder and bowel with differences (VMAT manual - VMAT auto ) of 4.5 ± 2.2 Gy, 10.7 ± 3.2 Gy and 2.1 ± 1.2 Gy, respectively. Right femoral head D 2% was on average 2.2 Gy higher in VMAT manual , whereas left femoral head D 2% was equal. Target volume doses were similar in both optimization approaches with no statistical difference in D mean of the prostate-PTV and a slightly higher coverage V 95% in VMAT manual (98.3% vs. 97.7%). Average dose-volume histograms are depicted in Figure 1. In the blinded review the radiation oncologist preferred the VMAT auto plan for 27 patients and the VMAT manual plan for the remaining 3 patients. Manual planning time was reduced by 76 minutes on average through VMAT auto .

Conclusion Uncertainty under respiration is an important issue in lung SBRT. Our results suggested that 4Drobust optimization provides more robustness in terms of doses to OARs against respiratory motion and should be implemented in lung SBRT. EP-1878 Fully automated multi-criterial VMAT plan optimization for prostate cancer whole-pelvic radiotherapy M. Buschmann 1,2 , A.W. Sharfo 3 , J. Penninkhof 3 , Y. Seppenwoolde 1,2 , G. Goldner 1 , D. Georg 1,2 , S. Breedveld 3 , B. Heijmen 3 1 Medical University of Vienna, Department of Radiotherapy, Vienna, Austria 2 Medical University of Vienna, Christian Doppler Laboratory for Medical Radiation Research for Radiation Oncology, Vienna, Austria 3 Erasmus MC Cancer Institute, Department of Radiation Oncology, Rotterdam, The Netherlands Purpose or Objective Inclusion of pelvic lymph nodes in prostate radiotherapy results in large and complex target volumes with a concave shape. This study evaluates the benefit of automated VMAT optimization (VMAT auto ) for the whole- pelvic radiotherapy technique compared to manual planning (VMAT manual ) for patients treated with a simultaneously integrated boost technique. Material and Methods An in-house multi-criterial optimizer was used as a preoptimizer for the Monaco treatment planning system (Elekta AB, Stockholm, Sweden) to fully automatically generate the VMAT auto plans. Patients were treated with a simultaneous integrated boost VMAT plan, delivering 60 Gy and 50 Gy in 25 fractions to the prostate and the pelvic lymph nodes, respectively, followed by a sequential-boost VMAT plan, delivering 13 Gy in 5 fractions to the prostate. The autoplanning system was

Conclusion Fully automated VMAT plan optimization for whole-pelvic prostate radiotherapy with large, concave target volumes was feasible, and resulted in substantially reduced OAR doses compared to manual planning by an expert, especially for the bladder. Manual planning time was reduced by more than 70 minutes. EP-1879 Implementation of GPU technology into the Varian Eclipse TPS for Fast Dose Calculation T. Aland 1 , M. Jones 2 , A. Walsh 1 , A. Piccini 2 , A. Devlin 2 , S. Thompson 3 , M. Sabel 3 1 Icon Group, Radiation Oncology Centres, Brisbane, Australia 2 Icon Group, ICON Plan, Brisbane, Australia 3 Varian Medical Systems, Treatment Planning, Palo Alto, USA Purpose or Objective The purpose of this study was to validate a pre-clinical release of the first implementation of GPU (Graphics Processing Unit) technology on the Varian Eclipse treatment planning system platform.