GPU-based fast Monte Carlo dose calculation for proton therapy.

Area of Science:

• Medical Physics
• Computational Physics

Background:

• Accurate radiation dose calculation is critical for effective proton radiotherapy.
• Monte Carlo (MC) simulation offers high accuracy but is computationally intensive, limiting clinical use.
• Graphics processing units (GPUs) can accelerate complex calculations.

Purpose of the Study:

• To develop and validate a fast MC dose calculation package (gPMC) for proton therapy utilizing GPU acceleration.
• To assess the accuracy and efficiency of gPMC compared to established MC codes.

Main Methods:

• Developed gPMC using a class II condensed history scheme with continuous slowing down approximation on a GPU (CUDA platform).
• Modeled proton transport, including ionization, elastic/inelastic interactions, energy straggling, and multiple scattering.
• Validated gPMC against TOPAS/Geant4 MC code using homogeneous, inhomogeneous phantoms, and a patient case.

Main Results:

• gPMC demonstrated good agreement with TOPAS/Geant4, achieving >98.7% gamma passing rate (2%/2mm criterion) in relevant regions.
• Simulation of 10 million protons to ~1% uncertainty took only 6-22 seconds, a substantial improvement over CPU-based methods.
• The code handles secondary proton transport and local energy deposition for charged fragments and electrons.

Conclusions:

• gPMC provides a highly efficient and accurate solution for proton dose calculation in radiotherapy.
• The GPU-accelerated MC code facilitates the routine clinical application of MC simulations for proton therapy.
• This advancement can enhance the precision and safety of proton radiotherapy treatments.