Uncertainty quantification analysis and optimization for proton therapy beam lines.

Abstract Since many years proton therapy is an effective treatment solution against deep-seated tumors. A precise quantification of sources of uncertainty in each proton therapy aspect (e.g. accelerator, beam lines, patient positioning, treatment planning) is of profound importance to increase the accuracy of the dose delivered to the patient. Together with Monte Carlo techniques, a new research field called Uncertainty Quantification (UQ) has been recently introduced to verify the robustness of the treatment planning. In this work we present the first application of UQ as a method to identify typical errors in the transport lines of a cyclotron-based proton therapy facility and analyze their impact on the properties of the therapeutic beams. We also demonstrate the potential of UQ methods in developing optimized beam optics solutions for high-dimensional problems. Sensitivity analysis and surrogate models offer a fast way to exclude unimportant parameters frcomplex optimization problems such as the design of a superconducting gantry performed at Paul Scherrer Institute in Switzerland.