Session Item

To study the potential of daily online dose-guided adaptations by varying only the iso-center and couch angles.

We previously introduced a method for dose-guided daily iso-center shifts that resulted in Pareto-optimal shifts when trying to reconciliate the daily dose (as predicted by a daily image prior to treatment) with the planned dose. We showed that it was possible in certain cases to improve on the shifts suggested by image-guidance alone. Furthermore, we demonstrated that trade-offs between minimizing the deviations from planned dose in different planning structures do arise and that they need to be addressed. We have extended this method to include the possibility to change the couch angle by 5 degrees in each direction. We also allow a shift in any of the 3 iso-center coordinates by 3mm in each direction. For a fixed treatment couch angle, the set of plans given by shifting the iso-center position alone determines a Pareto front. By varying the angle of the couch by discrete angle steps (of 1 degree) we obtain twenty possible Pareto fronts from which a user (or automated procedure) would have to choose from. We greatly reduce this number of fronts to any a select few by determining promising correction parameters heuristically. Finally, we demonstrate how to deal with multiple Pareto fronts in a straight-forward decision-making procedure.

We apply this dose-guided positioning correction procedure to 68 retrospective cases of various disease sites (liver, lung, pelvis). We compare V95% of CTV of plans produced using our correction method with the plans that were created by dosimetrists using CBCT- or MV-based image-guided correction alone. The dose-guided correction was done manually by navigating the calculated Pareto fronts and trying to restore the planned CTV coverage. We found that V95% could be improved by about 3% on average. More interestingly, however, in all cases V95% could be restored to at least 88% of prescription using our dose-guided positioning, whereas it could only be restored to 80% in some cases without our method.

We are able to quickly determine good patient setup corrections (iso-center and couch angle) to correct for changes in patient anatomy. We provide Pareto navigation (a multicriteria decision-making tool) on multiple Pareto fronts to balance trade-offs when trying to restore original plan goals. Alternatively, this decision-making can be automized. There are clinical cases which would benefit greatly (as measured by V95% of the CTV) using this online dose-guided patient positioning correction method.