Room: Exhibit Hall | Forum 4
Purpose: To study the clinical performance of a semi-analytical dose engine developed for spot scanning proton delivery systems that feature small spot sizes.
Methods: The dose engine used a ray-casting pencil beam algorithm with three lateral Gaussian components. It was commissioned using Monte Carlo (MC) data generated by an experimentally well-benchmarked Geant4 code and then fine-tuned using point dose measurements with various combinations of energy layers, field sizes, depths, and lengths of spread-out-Bragg-Peaks. The dose engine used water-equivalent distance to account for inhomogeneity. Ten patients representing different disease sites were randomly selected for validation. Comparisons were done in water with data calculated from a fast MC code and with ionization chamber array measurements for patient-specific QA (including point doses, dose profiles, 2D-3D, and 3D-3D Gamma analysis) and in patient geometries with MC data (including dose volume histogram indices and 3D-3D Gamma analysis).
Results: In-water Gamma passing rates using 3%/3mm criteria for these ten patients were greater than 97% when compared to MC and greater than 95% when compared to dose plane measurements. Excellent agreement was observed for dose profiles and point dose measurements in water. Good agreement between the semi-analytical dose engine and the MC code was also observed in patient geometries. This dose engine has been successfully used at our institution as a fast, independent dose calculation second check for over two years.
Conclusion: A modified ray-casting dose engine with three lateral Gaussian components can accurately calculate dose distributions for our proton delivery system.