ESTRO 2024 - Abstract Book

S3646

Physics - Dose prediction, optimisation and applications of photon and electron planning

ESTRO 2024

The true dose delivered to the GTV in each planning approach remains to be identified, and will be evaluated in a future study. This will also help to determine the optimal treatment planning strategy.

Keywords: Treatment Planning, Robust Optimization, Margins

2270

Digital Poster

An efficient multicriteria algorithm to decide among various IMRT and VMAT treatment configurations

Helene Krieg, Ina Lammel, Felix Riexinger, Philipp Süss

Fraunhofer Institute for Industrial Mathematics ITWM, Optimization, Kaiserslautern, Germany

Purpose/Objective:

We want to find the best treatment plan among several given treatment types (i.e., which given beam arrangement for IMRT or which given partial or full arcs to use) for any given patient. Several IMRT or VMAT configuration choices may lead to good treatment plans; however given state-of-the-art decision support, it is difficult to answer more detailed questions like “How much tumour coverage would we lose when reducing the treatment time by changing from full to partial arcs?” or “Do we get better goal fulfilment for 9 field IMRT or one full VMAT arc?”. Our new method allows to study the trade-offs between treatment types without having to plan each separately in full.

Material/Methods:

We previously introduced the Patch Approximation algorithm [1] for the simultaneous approximation of several convex Pareto fronts. Given a set of planning objectives for the patient, each IMRT beam arrangement and each VMAT arc set implicitly defines a single Pareto front and the union over all these Pareto fronts defines the overall candidate set of possible treatment plans. Our algorithm determines regions in which one treatment type is inherently worse than another, i.e., it is dominated by a different treatment configuration. In these regions, the algorithm avoids computing more approximation points of the dominated treatment type. In general, the higher the overlap of the individual Pareto fronts of several treatment types, the shorter the plan computation time compared to approximating the Pareto fronts for each treatment configuration individually. Moreover, using Pareto Navigation [2], the trade-offs of the treatment objectives can then be studied interactively across all treatment types simultaneously. The best IMRT setup or VMAT arcs can be chosen automatically based on the selected plan that attains the preferred trade-offs.

Results:

Applying our Patch Approximation algorithm to radiotherapy planning problems indeed shows that there is often a large overlap between the Pareto fronts of different treatment configurations, leading to substantial savings in plan computations.