Room: Exhibit Hall | Forum 7
Purpose: Prior studies on multicriteria optimization (MCO) suggested constraints be minimally used, only on targets and serial organs. Since practical MCO algorithms only sample the Pareto surface with finite plans, such conventional strategy creates a large surface with coarse resolution, lack of clinically-relevant boundaries for parallel organs. With increasing use of simultaneous integrated boost (SIB), the large number of clinical goals makes navigation on the unbounded surface inefficient and the coarse resolution makes the optimal solution hard to find. This study investigates if applying physician-directed constraints on parallel organs can improve the quality of the starting balanced plan and the final deliverable plan.
Methods: A patient is found for each site treated with SIB, categorized into three levels of difficulty: simple (prostate), intermediate (prostate with node, pancreas) and complex (breast, cranial, anal, gynecological, H&N). The strategies using minimal vs. extensive constraints are explored for IMRT and VMAT using Elekta Agility model in RayStation v6.1. The four plans for each site are navigated to provide similar target coverage, with organ dose limits fulfilled in the order of importance. The quality of the balanced and deliverable plans is assessed by the number and magnitude of the unachieved clinical goals.
Results: In the balanced plan, the extensively-constrained strategy achieves more clinical goals in all difficulty levels, and the unachieved goals are generally less severe. In the deliverable plan, the extensively-constrained strategy is more beneficial for more complex sites. For the most notorious H&N case (nasopharynx with scalp invasion), the extensively-constrained VMAT meets all clinical goals, whereas the minimally-constrained missed 12 goals to make parotid dose clinically acceptable (not achieved). Clinical QA shows no difference between the two strategies.
Conclusion: The extensively-constrained strategy improves planning efficiency by a more accurate balanced plan and dosimetric quality by navigating on a bounded surface with finer resolution.
Optimization, Treatment Planning, Inverse Planning
TH- External beam- photons: VMAT dose optimization algorithms