ESTRO 37 Abstract book

ESTRO 37

S482

Results For vestibular schwannoma and pituitary adenoma cases, use of the four-pi algorithm significantly lowered the maximum dose to orbits, lenses, optic nerves and chiasm (p<0.05) by adjusting the template to produce gantry arcs that avoided intersection of treatment beams with these OARs. For vestibular schwannoma cases, the maximum dose values of optic structures were approximately 50% of those received without using the four-pi approach, however the effect on brainstem sparing was non-significant. Four-pi also significantly lowered monitor units by selecting treatment geometries that necessitated simpler and generally more open MLC apertures during VMAT delivery. Results show a trade-off with regard to proximal OARs included in the optimization; for example, eliminating the brainstem from the criteria of the four-pi algorithm allowed an improvement of the dosimetry to optic structures. Compared to standard VMAT planning, conformity of the prescription dose (D Rx ) and the compactness of the (Dcompactness of the (D Rx /2) dose level were improved significantly.

Planning in the context of attaining the best feasible treatment planning solution. Material and Methods Knowledge based planning was performed with RP v15.5. Multi-criteria optimisation was performed with MCO v15.5. All plans were normalised to a PTV mean of 100% dose. RP v15.5 generates estimated DVHs and planning objectives based on the library of plans it is built from. MCO v15.5 is a planning trade off tool that calculates multiple plans for each planning objective; this allows the planner to explore the Pareto surface of the ideal dose distribution to determine the optimal trade off between PTV coverage and OAR doses. PlanIQ is an automated plan comparison tool, capable of performing quantitative analysis of treatment plans. PlanIQ predicts the Feasibility of the ideal dose distribution for an individual plan by depositing the prescribed dose in the PTV and calculating dose fall off from the PTV. Feasibility DVH curves are then created which are scored 0 – 1 progressively based on degrading plan quality, Figure 1. It is hypothesised that DVH curves produced by MCO planning should be closer to ideal Pareto surface dose distributions and should thus corr elate closer with Feasibility DVH curves than non-MCO planning methods.

DVHs for ipsi-lateral kidney, contra-lateral kidney, duodenum and liver were correlated with the closest matching Feasibility DVH curve by area under the curves and DICE coefficient. Results The average feasibility score over all 10 patients was 0.2 for RP15.5 and 0.15 for MCO (p <0.0001). The average DICE coefficient comparing the nearest feasibility curve to the planned DVH for all organs was 0.87 with RP15.5 and 0.88 with MCO (p = 0.003), Table 1.

Conclusion The four-pi approach allowed automated selection of couch and gantry parameters that consistently lowered the dose to OARs in stereotactic radiotherapy plans compared to standard planning templates. PO-0904 Comparison of knowledge based planning and multi-criteria optimisation for VMAT pancreas planning P. Houston 1 , N. Laverick 1 1 Beatson West of Scotland Cancer Centre, Radiotherapy, GLASGOW, United Kingdom Purpose or Objective To perform a comparison of a knowledge based planning solution, RapidPlan 15.5 (RP v15.5), and a multi-criteria optimisation solution (MCO v15.5) for VMAT Pancreas

p

value

Average RP15.5 Feasibility Score

Average MCO 15.5 Feasibility Score

(two tailed t- test)

OAR

Ipsilateral kidney

0.41

0.25

0.0002

Contra- lateral kidney

0.08

0.06

0.05

Liver

0.11

0.10

0.03

Duodenum 0.21

0.18

0.12

Average over

0.20

0.15

0.003

all

OARs