Parameterization of Brachytherapy Source Phase Space Files for GPU-Accelerated Monte Carlo Particle Transport Simulation
Abstract
Purpose
The study aims to develop a parameterized brachytherapy source geometry modeling framework for GPU-based Monte Carlo particle transport simulation with an integrated DICOM workflow.
Methods
The framework was developed for two Ir192 HDR brachytherapy source models GammaMedPlus and VS2000. Phase-space data were generated using egs_brachy Monte Carlo toolkit and parameterized using the information (e.g., particle location, direction and energy) stored in the phase space file through several probability distributions. These distributions were loaded to GPU-based MC package and particles were randomly sampled from them. DICOM RT plan was used to define dwell positions within the workflow. The sampling methodology was validated by comparing the generated particle distributions with the input phase-space data. Dose distributions were calculated on 3D voxelized water phantom and radial dose analysis was performed to validate geometric dose falloff.
Results
The results showed the developed tool accurately converts 459 MB phase space file into 78 KB file. GPU-accelerated Monte Carlo phase-space sampling efficiently reproduced the original particle distributions, while radial dose analysis confirmed proper inverse-square dose falloff behavior.
Conclusion
By using the parameterized source representation, computational memory requirements are significantly reduced, enabling practical GPU implementation for clinical brachytherapy Monte Carlo dose calculations.