ESTRO 38 Abstract book

S492 ESTRO 38

in 30 fractions. The median [range] gross tumour volume (GTV) was 29 cm 3 [4-472] and the planning target volume (PTV) was 259 cm 3 [51-1069]. The clinical plans (CP0) consisted of 4-8 IMRT fields. Beam angles were set individually for each patient plan by different dose planners. The plans should comply with constraints for spinal cord (D max <45Gy), esophagus (D 1cm3 <100%), body (D max <110%), lungs (MLD<19Gy, V20Gy<35% and V5Gy<60%) and heart (V40Gy<30%, V25Gy< 50%). Retrospectively, an experienced planner made 74 new treatment plans (EP0). Beam angles and optimization constraints were selected so that dose to the lungs and heart was minimized, still securing full target coverage (V95%>99%) while minimizing conformity index (CI). These EP0 plans were used to generate a RP model (RP1). For each patient, two new plans were made using RP1 (CP1 and EP1) based on the field configurations of CP0 and EP0. A second RP model (RP2) was created from the optimal plan (CP1 or EP1) for each of the 74 patients. This final RP model was used to generate the final plan (F2) for each patient. Dosimetric parameters were compared using a Wilcoxon signed rank test. p< 0.05 was considered significant. Results

Conclusion PSO allows for automatic generation of VMAT plans with comparable plan quality compared to manually optimized plans. However, further research is needed concerning optimization parameters and the optimal number of particles and generations to fully explore the potential of PSO for automatic planning. The plan quality score has to be further refined to find reasonable and clinical acceptable compromises between PTV and OAR doses.

Poster: Physics track: Treatment planning: applications

PO-0920 Knowledge-based planning significantly reduces dose to organs at risk for lung cancer L. Hoffmann 1 , M.M. Knap 2 , D.S. Møller 1 1 Aarhus University Hospital, Department of Oncology- Medical Physics, Aarhus, Denmark ; 2 Aarhus University Hospital, Department of Oncology, Aarhus, Denmark Purpose or Objective The introduction of knowledge-based planning has been shown to result in reduced variability in treatment plans and for some patient groups also reduced dose to organs at risk (OARs). RapidPlan (RP) has been introduced into the Eclipse treatment planning system (Varian Medical Systems). The aim of the present study was to investigate if a general reduction in dose to OARs was achievable by using a RP model. Furthermore, for all patients two different beam angle selections were compared. Material and Methods

Beam angle selection and dose distribution is shown for one patient in Fig 1, where a clear reduction in dose to heart, lungs and esophagus is seen due to optimal beam directions for EP0 vs CP0. Further reduction is obtained by applying the RP model. In Fig. 2, the median value of mean lung/heart/esophagus dose for all 74 patients decreases significantly from CP0 to EP0 due to optimal beam angle selection. The application of RP1 adds a significant reduction in mean doses to lungs and esophagus (D CP1 99% and BodyD max <110%. The CI decreases significantly between CP0 and F2 or EP0 and F2, meaning that the RP model makes the plans more conformal in addition to OARs sparing. Conclusion Introduction of RapidPlan significantly reduces dose to lungs, heart and esophagus. The highest reduction is seen when RP is combined with optimal beam angles. The conformity index is significantly reduced, while PTV dose coverage is maintained for RP plans. Further reduction is seen by creation of a second RP model (RP2).

Seventy-four consecutive lung cancer patients were selected for configuration of the RP model. The patient group consisted of 49 NSCLC patients treated by 60-66 Gy in 30-33 fractions and 25 SCLC patients treated by 45 Gy