ESTRO 2020 Abstract book

S821 ESTRO 2020

while a 3 levels approach ( 70-63-54 Gy) was employed for HN treatments. Both algorithms were applied with no human interference. Plan evaluations accounted for dosimetric parameters with DVH based metrics suggested by QUANTEC and adopted in the clinic. In addition, an evaluation metric for plan quality was integrated in the software; it was derived from the same clinical goals giving a unique Percentage Score Value (PSV%), taking into account for both PTV coverage and OARs sparing. Using an interactive Python visual analog scale, a Clinical Blind Score (CBS) study was performed, i.e. the clinician had to blindly select which plan was deemed most suitable and the strength of the difference was expressed in percentage. Results Minimum requirements for PTVs coverage and maximum allowed overdosages were always satisfied for both tumors and both algorithms, therefore not reported below. For both prostate and HN GPS plans: no statistically and clinically significant improvements were found on average for GPS plans over IOA plans, as visible in figure 1. For prostate IOA plans: rectum D 15% and D mean were 3.1% and 5.7% lower and left and right femoral heads D mean were 4.1% reduced (both p<0.05, as visible in figure 1). Mean PSV was improved by 2.9% and computation time was reduced from 24 to 11 min (both p<0.05). CBS was 4.7% in favor of IOA plans, but without statistical significance, and IOA plan was selected over GPS plan for 13/20 patients. For HN IOA plans: spinal cord D 1cc and Esophagus D 1cc were 9.9% and 7.9% lower, respectively, and oral cavity D 35% and D 65% were both lower by 12% (all p<0.05, as visible in figure 2). PSV was 1% worsen, but not statistically significant, and computation time was reduced from 21 to 14 min (p<0.001). CBS was 6.7% in favor of IOA plans but without statistical significance, and for 13/20 patients IOA plans were selected.

create an optimized set of trajectories for VMAT optimization. These trajectories were used to create an APBI treatment plan in the Eclipse treatment planning system (PO 13.6).

Results This novel optimization algorithm yielded three arcs for VMAT optimization for the test patient. The resulting treatment plan had a PTV coverage of 99.7%, and a conformity index of 1.11. OAR metrics are shown in Table 1. The volume of the ipsilateral lung receiving less than 9 Gy was 1.4%, and the mean dose to the heart was 0.13 Gy. These values compare favourably with the dosimetric constraints used in the clinical trials on 3D-conformal APBI and robotic APBI.

Conclusion Applying a non-coplanar optimization framework to APBI provided a clinically acceptable plan based on constraints current literature. Future work will involve using this framework for a larger subset of patients and comparing results with techniques that are currently used for APBI. PO-1448 Fully-automated VMAT treatment planning: head to head between genetic and iterative algorithms C. Fiandra 1 , S. Zara 2 , A. Alparone 2 , P. Franco 1 , S. Bartoncini 3 , L. Rossi 4 , B. Heijman 4 , U. Ricardi 1 1 University of Turin School of Medicine, Radiation Oncology, Turin, Italy ; 2 Tecnologie Avanzate, Research and Development, Turin, Italy ; 3 AOU Città della Salute e della Scienza, Radiation Oncology, Turin, Italy ; 4 Erasmus MC Cancer Institute, Radiation Oncology, Rotterdam, The Netherlands Purpose or Objective To Develop a new Iterative Optimization Algorithm (IOA) within a fully automated treatment planning software and compare results with those of a well-established genetic optimizer in the setting of prostate and head and neck VMAT treatments. Material and Methods A beta version of a new iterative optimization algorithm was implemented on the RayStation treatment planning system (Version 8B) using Python code within the platform named Genetic Planning Solution (GPS version 2.1). The IOA algorithm starts from an initial value of EUD for each organ at risk to generate a sequence of improving approximate solutions, in which the n-th approximation is derived from the previous ones. We compared 20 prostate and 20 head and neck (HN) VMAT GPS plans with newly developed IOA plans. Prostate patients were irradiated with a SIB technique delivering 70.2-61.1 Gy in 26 fractions

Conclusion