Article
Optimization of the RBE-Weighted Dose in Heavy-Ion Therapy
Search Medline for
Authors
Published: | September 24, 2009 |
---|
Outline
Text
Background: From 1997 until 2008 more than 430 cancer patients were successfully treated in the ion therapy pilot project at GSI Helmholtzzentrum für Schwerionenforschung GmbH. Most of the tumors were located in the head or neck region. The treatment planning was performed with the software TRiP which was developed at GSI. A crucial part of the treatment planning is the particle number optimization for the raster-scan system. The goal of the optimization step is to achieve a homogeneous target dose distribution as close as possible to the prescribed dose distribution by an appropriate sparing of critical structures (organs at risk, OAR) like the brainstem or the optical nerve.
Material and methods: The optimization task can be mathematically expressed as the minimization of a multidimensional objective function by means of the least squares method. If biological effects are taken into account, i.e. the absorbed dose is weighted with the relative biological effectiveness (RBE), the objective function is nonlinear in its free parameters and an analytical solution is not available. Hence, numerical methods are required to obtain the desired optimum. We implemented and investigated several standard iterative methods for nonlinear optimization problems.
Results: We examined and compared the performance of the various algorithms in terms of memory requirements, iteration steps and computation times. Sample patient plans from the pilot project yielded acceptable dose distributions for the target and OAR. The computational performance, however, shows major differences. Advantages as well as disadvantages of the algorithms will be discussed.
Conclusion: By selecting an appropriate algorithm it is possible to optimize complete patient plans within about an hour even on less powerful hardware. An outlook on further reduction of resource consumption will be given.