Commissioning of a synchrotron-based proton beam therapy system for use with a Monte Carlo treatment planning system

Abstract

This work tackles the commissioning and validation of a novel combination of a synchrotron-based proton beam therapy system (Hitachi, Ltd.) for use with a Monte Carlo treatment planning system (TPS). Four crucial aspects in this configuration have been investigated: (1) Monte Carlo-based correction performed by the TPS to the measured integrated depth-dose curves (IDD), (2) circular spot modelling with a single Gaussian function to characterize the synchrotron physical spot, which is elliptical, (3) the modelling of the range shifter that enables using only one set of measurements in open beams, and (4) the Monte Carlo dose calculation model in small fields. Integrated depth-dose curves were measured with a PTW Bragg peak chamber and corrected, with a Monte Carlo model, to account for energy absorbed outside the detector. The elliptical spot was measured by IBA Lynx scintillator, EBT3 films and PTW microDiamond. The accuracy of the TPS (RayStation, RaySearch Laboratories) at spot modelling with a circular Gaussian function was assessed. The beam model was validated using spread-out Bragg peak (SOBP) fields. We took single-point doses at several depths through the central axis using a PTW Farmer chamber, for fields between 2 × 2cm and 30 × 30cm. We checked the range-shifter modelling from open-beam data. We tested clinical cases with film and an ioni- zation chamber array (IBA Matrix). Sigma differences for spots fitted using 2D images and 1D profiles to elliptical and circular Gaussian models were below 0.22 mm. Differences between SOBP measurements at single points and TPS calculations for all fields between 5 × 5 and 30 × 30cm were below 2.3%. Smaller fields had larger differences: up to 3.8% in the 2 × 2cm field. Mean differences at several depths along the central axis were generally below 1%. Differences in range- shifter doses were below 2.4%. Gamma test (3%, 3 mm) results for clinical cases were generally above 95% for Matrix and film. Approaches for modelling synchrotron proton beams have been validated. Dose values for open and range- shifter fields demonstrate accurate Monte Carlo correction for IDDs. Elliptical spots can be successfully modelled using a circular Gaussian, which is accurate for patient calculations and can be used for small fields. A double-Gaussian spot can improve small-field calculations. The range-shifter modelling approach, which reduces clinical commissioning time, is adequate