ESTRO 2023 - Abstract Book

S1766

Digital Posters

ESTRO 2023

(i.e. Dconst(Vconst)) for both plans for each OAR. These relative ratios (RR) are then summed (RR1+RR2) and should be <= 100%. The relativity implies that constraints do not need to have the same Vconst enabling the combination of hypo- and regular-constraints. We compared the PM with two different strategies: a 3D and a single-value approach. Firstly, the 3D method (3DM) uses a home-written Varian Eclipse script and the LQ model to convert 3D doses from plan1 to the fractionation of plan2 (f2), after which rigid registration and dose summation yields the total dose. Total OAR doses are then evaluated using the constraints for f2. Secondly, the single-value method (SVM), which is more widely used, takes the sum of single critical dose values from each fractionation scheme, after an LQ conversion of the single dose value of plan1 to f2. For each OAR and constraint type, we plotted the RR per method and we calculated the true positive/negative rates and accuracy of the PM and SVM. For now, we ignored OAR repair in between treatments, only considered single re-irradiations, and assumed alpha/beta=3 for all OARs.

Results In Fig 1 the relation between critical OAR doses for the different methods is depicted, showing that largest differences occur in regular-hypo re-irradiations. The PM is at least as good at predicting the 3D results as the SVM. This is confirmed by the calculation of the accuracy which is 76% and 74% for DV and Dmax constraints of the SVM, respectively, and 80% and 83% for the PM. Further advantages of the PM are that plans are evaluated by their validated constraints and no image/dose registration or LQ conversion schemes are required. However, there is no radiobiological justification as yet.