Abstract Book

S1011

ESTRO 37

class solution, using Pareto Fronts to investigate the influence of such parameters on target coverage and rectal dose. Material and Methods A representative prostate case was planned with a SIB technique delivering 52.6Gy to PTV1 (prostate and seminal vesicles plus margin) and 60Gy to PTV2 (prostate alone) in 20 fractions, using Elekta Monaco TPS (v 5.10.02) and 6MV FFF. A baseline VMAT plan was designed and the local Pareto front was sampled by varying the iso-constraint for the rectum cost function in a step-wise fashion such that PTV1 coverage changed from 99.5% to 90%. The Pareto front was constructed for V50Gy for PTV1 against V32.4Gy for the rectum by fitting a polynomial trendline to those data points lying on the convex border for the entire dataset. One VMAT planning parameter at a time was changed with respect to reference plan conditions and Pareto fronts constructed for the different parameter values. Planning parameters include number of arcs, number of sectors in arc, maximum number of control points per arc, arc length, collimator angle, minimum segment width, target margin, statistical uncertainty, and various cost functions for the rectum structure both overlapping with as well as lying outside of PTV1. Results Using a single arc shows the greatest degradation in Pareto front compared to a dual arc and applying two physical beams with one arc each will not produce a Pareto front equivalent to a single beam dual arc (see Fig 1).

Using a tight target margin improves the Pareto front compared to a normal or wide target margin. A statistical uncertainty per calculation of 3% degrades the Pareto front compared to a value of 1%. The serial, parallel and conformality cost functions applied to the mid-dose region of the rectum structure outside of PTV1 show near identical Pareto fronts, whereas the use of sub-optimal parameters for the serial and max dose cost functions for the high-dose region of the rectum overlapping PTV1 has the potential to severely degrade the Pareto front. Conclusion Pareto front comparison demonstrates the impact of various planning parameters that improve or deteriorate plan quality and aids the development of a class solution for prostate VMAT planning. EP-1872 Stepping validation of a KB-model for plan optimization of prostatic/pelvic patients irradiation R. Castriconi 1 , C. Fiorino 1 , S. Broggi 1 , C. Cozzarini 2 , N. Di Muzio 2 , R. Calandrino 1 , G.M. Cattaneo 1 1 IRCCS San Raffaele Scientific Institute, Medical Physics, Milano, Italy 2 IRCCS San Raffaele Scientific Institute, Radiotherapy, Milano, Italy Purpose or Objective To develop and apply a stepping approach for the validation of Knowledge-based (KB) models for planning optimization, starting from the complex and clinically relevant case of the treatment of pelvic nodes and prostatic+seminal vesicles (SV) bed in post-prostatectomy patients (pts). Material and Methods In our protocol three PTVs are treated in two phases, delivering 70/74Gy to prostatic-bed. We focused on the first phase delivering 52.5/60Gy in 30 fractions to nodes/prostate+SV. Fifty-two VMAT (Varian RapidArc) clinical plans optimized by two planners (reference of the protocol) were selected to generate KB model using the RapidPlan (RP) tool implemented in the Eclipse system (v13.5). The resulting DVH prediction model was set on the lower DVH population quartile for each OARs (rectum, bladder, bowel, femoral heads, penile bulb); an optimized template based on the planner experience was carried out for optimization. A stepping-validation was performed: 1) Close-loop: 20 randomly chosen pts used to generate the model; 2) Open-loop: 10 new pts, planned by the same 2 operators; 3) Wide-loop: 20 new pts, planned by 4 additional planners. KB plans were generated following two modalities: RP+planner intervention ( RP ) and fully-automatic RP ( only-RP ). All plans were compared, in terms of OARs/PTVs dose-volume parameters and OARs generalized equivalent uniform dose (gEUD), against the clinical plans ( RA ): values of a for gEUD calculation were set according to the AAPM TG166 report. Two tails paired

Increasing the number of sectors in the arc from 8 to 12 improves the Pareto front. A maximum 90 control points per arc has a Pareto front somewhat improved over lower values (80 and 75), whilst increasing the maximum number of control points per arc beyond a value of 100 increases plan complexity and reduces plan delivery efficiency. Partial arc lengths of 220° and 300° show equivalent Pareto fronts, whereas a full 360° arc shows a degraded Pareto front. Collimator angles of 23° and 45° produce improvements in the Pareto front compared to a collimator angle of 0°. A 1 cm minimum segment width severely degrades the Pareto front with respect the recommended 0.5 cm value (see Fig 2).