Integration of a model-independent interface for RBE predictions in a treatment planning system for active particle...

Especially for heavier ions such as carbon ions, treatment planning systems (TPSs) for ion radiotherapy depend on models predicting the relative biological effectiveness (RBE) of the particles involved. Such models are subject to intensive research and the choice of the optimal RBE model is a matter of debate. On the other hand TPSs are often strongly coupled to particular RBE models and transition even to extended models of the same family can be difficult. We present here a model-independent interface which allows the unbiased use of any RBE model capable of providing dose-effect curves (even sampled curves) for a TPS. The full decoupling between the RBE model and TPS is based on the beam-mixing model proposed by Lam which is, in contrast to the often-used Zaider-Rossi model, independent of the explicit form of the underlying dose-effect curves. This approach not only supports the refinement of RBE models without adaptations of the TPS--which we demonstrate by means of the local effect model (LEM)--but also allows the comparison of very different model approaches on a common basis. We exemplify this by a comparison between the LEM and a model from the literature for proton RBE prediction.