Proton therapy is a form of radiotherapy where protons are used instead of the commonly used photon. Protons have the property that the dose delivered falls of rapidly beyond a certain range. Combining a set of proton beams with different intensities and energy can be used to adequately cover the tumor region and at the same time sparing the surrounding healthy tissue or organs at risk such as the eye nerve, specific glands or the spinal chord.
Deterministic Dose Calculation
Dose calculation is at the heart of radiotherapy. Currently, dose distributions are obtained in the clinic by fast approximate models such as pencil-beam models. Such models are generally accurate in fairly homogeneous tissue. They however exhibit loss of accuracy inn heterogeneous materials. The alternative Monte Carlo method is accurate but way too expensive for clinical application. We study deterministic methods for dose calculation in proton therapy. This deterministic approach is based on direct discretization of the Boltzmann equation in space, angle and energy and subsequent solution of the complete radiation field on this computational grid by efficient iterative methods. Being based on the Boltzmann model, we will be able to achieve near Monte Carlo method accuracy at much reduced computational cost and without the associated statistical noise. Our research in this area is strongly linked to our work in the area of adaptive transport. We are currently combining our capabilities to focus the available computational efforts in specific spatial regions and in certain directions to adequately resolve the proton radiation field with recently developed Fokker-Planck discretization method for the continuous scatter and continuous slowing down models.
Uncertainty Quantification and Robust Planning
Proton therapy has the potential to be a highly accurate treatment. With the ability to deliver a highly localized dose comes an increased sensitivity when compared to conventional photon therapy. Patient setup errors, breathing and other motion, stopping power uncertainties and gradual changes in anatomy therefore may have large consequences. Part of our research focuses on quantitative analysis of the effects of setup and range errors on the dose distribution. Innovative polynomial chaos methods are applied where a model is constructed that replaces the dose engine. For specific (systematic and random) setup and range errors the dose can be reconstructed orders of magnitude faster than with the actual dose engine. Using the PCE method and the resulting surrogate dose-engine, a comprehensive uncertainty quantification (robustness analysis) of proton therapy plans is possible. Furthermore, research that would otherwise be too cpu-intensive is now within reach. An example of the latter is the determination of robustness-recipes where consistent input parameters are determined for robust planning software for head and neck cancer and prostate patients.
Some references
Z. Perko, S.R. van der Voort, S. van de Water, C.M.H. Hartman, M. Hoogeman, D. Lathouwers, Fast and accurate sensitivity analysis of IMPT treatment using polynomial chaos expansion, Physics in Medicine and Biology 61(12), pp. 1373-1381, 2016.
S.R. van der Voort, S. van de Water, Z.Perko, B. Heijmen, D. Lathouwers, M. Hoogeman, Robustness recipes for minimax robust optimization in intensity modulated proton therapy for orphanyngeal cancer patients, International Journal of Radiation Oncology, Biology and Physics, 95(1), pp. 163-170, 2016.