Abstract Book

S130

ESTRO 37

properly selection of patients through MRI and PET to avoid overtreatment in metastatic patients. PV-0256 Fast and insightful bi-objective HDR prostate brachytherapy planning N.H. Luong 1 , T. Alderliesten 2 , B.R. Pieters 2 , A. Bel 2 , Y. Niatsetski 3 , P.A.N. Bosman 1 1 Centrum Wiskunde & Informatica, Life Sciences & Health, Amsterdam, The Netherlands 2 Academic Medical Center - University of Amsterdam, Radiation Oncology, Amsterdam, The Netherlands 3 Elekta, Elekta Brachytherapy, Veenendaal, The Netherlands Purpose or Objective The goal in prostate brachytherapy (BT) planning is to select the best trade-off between target coverage and organ sparing. Searching for the set of best possible trade-off plans has great potential because it facilitates planners to insightfully compare plans and select the most appropriate plan for each patient. To find such plans automatically, we introduce a bi-objective optimization model that describes separate objectives for target coverage and organ sparing. Dose-volume indices (DVIs) are key in planning. DVIs are typically computed using randomly sampled dose calculation (DC) points. For final evaluation, many points are used for the sake of accuracy, e.g., Oncentra Brachy (Elekta) uses by default 500,000 points. For optimization, typical automated BT planning methods (e.g., IPSA, HIPO), use about 100 times less points for the sake of speed. We studied the relation between the number of DC points and the quality of plans obtained with our bi- objective approach and studied whether a novel multi- resolution scheme can reduce the total planning time. Material and Methods We derived separate coverage and sparing objectives based on DVIs of a clinical protocol (Table 1). To solve this bi-objective optimization model, we used the Gene- pool Optimal Mixing Evolutionary Algorithm (GOMEA). For each of 18 HDR prostate BT cases, we conducted 5 experiments: 4 with 2,500, 5,000, 10,000, and 20,000 DC points and 1 with a multi-resolution scheme in which the number of points is gradually increased during optimization from 5,000 to 20,000. Final results were re- evaluated using 500,000 DC points. The quality of the obtained set of plans was scored by the hypervolume indicator, which quantifies the area covered by the set of plans. We compared the DVIs of the automatically generated plans with the clinical plans that were obtained in 30-60 minutes by experienced planners using IPSA/HIPO, followed by graphical optimization.

obtained as when using 20,000 points from the start. After 5 minutes, for all patients, the DVIs of the plans found with the multi-resolution scheme are already better than the DVIs of the clinical plan and close to plans found using 20,000 points after 1 hour (Fig. 1).

Conclusion By combining our novel bi-objective optimization approach for BT planning with a multi-resolution scheme for the number of DC points, many high-quality plans that insightfully identify trade-offs between target coverage and organ sparing can be obtained within a few minutes, ensuring clinical usability. PV-0257 Functional sub-structures of lower urinary tract in cervix cancer: contouring and dose distribution S. Spampinato 1 , L. Fokdal 1 , E. Marinovskij 2 , S. Axelsen 3 , E.M. Pedersen 4 , R. Pötter 5 , J. Lindegaard 1 , K. Tanderup 1 1 Aarhus University Hospital, Department of Oncology, Aarhus, Denmark 2 Aarhus University Hospital, The MR Research Center, Aarhus, Denmark 3 Aarhus University Hospital, Department of Obstetrics and Gynaecology, Aarhus, Denmark 4 Aarhus University Hospital, Røntgen og Skanning, Aarhus, Denmark 5 Vienna General Hospital- Medical Univerisy of Vienna, Department of Radiotherapy, Vienna, Austria Purpose or Objective Radiotherapy related urinary morbidity is a complex phenomenon consisting of various clinical endpoints (i.e. frequency, cystitis, incontinence, bleeding, fistula) that may be related to various anatomical sub-structures. However, dose to the bladder and correlation with morbidity is currently mainly evaluated through contouring of the outer wall. The aim of this study is to investigate contouring and dose evaluation in sub- structures potentially responsible for urinary morbidity in Locally Advanced Cervical Cancer (LACC) after radiochemotherapy with Image Guided Adaptive Assuming that structures related to the bladder base are essential for frequency and incontinence, a methodology for contouring bladder subvolumes (trigone, bladder neck, urethra) was established. Structures were contoured for each BT fraction and DVH parameters were extracted: outer bladder wall (D 2cm3 , D0 .1cm3 ), ICRU Bladder point, trigone (D 2cm3 , D 0.1cm3 ), bladder neck (D 0.1cm3 ), and urethra (D 0.1cm3 , D50). Furthermore, dose to Brachytherapy (IGABT). Material and Methods

Results Using fewer DC points, better hypervolume scores are obtained in the first 5-10 minutes. More points lead to better results, but it takes longer to achieve good plans. Differences between using 5,000 and 20,000 DC points are substantial, signifying that for successful bi-objective optimization, more DC points are needed than for single- objective optimization (e.g., IPSA and HIPO). Using the multi-resolution scheme, better results are obtained after 5 minutes. After one hour, the same results are