Fred: a GPU-accelerated fast-Monte Carlo code for rapid treatment plan recalculation in ion beam therapy

Area of Science:

• Medical Physics
• Radiation Oncology
• Computational Science

Background:

• Ion beam therapy offers precise dose delivery for tumors, sparing healthy tissue.
• Accurate dose calculation is critical for treatment planning.
• Traditional treatment planning software uses simplified models; Monte Carlo (MC) simulations offer higher accuracy but are computationally intensive.

Purpose of the Study:

• To develop a novel Monte Carlo simulation platform, "fred", for ion beam therapy treatment planning.
• To enhance computational efficiency for MC simulations using GPU hardware.
• To improve the accuracy of dose and spatial distribution calculations in heterogeneous environments.

Main Methods:

• Developed "fred", a new MC simulation platform utilizing a class II MC algorithm for particle transport in a 3D voxel grid.
• Implemented effective models for particle-medium interactions, balancing accuracy and computational cost.
• Leveraged general-purpose GPU (GPGPU) programming for accelerated particle tracing.

Main Results:

• Fred's proton beam transport in water shows dose-depth distributions within 1-2% of standard MC codes at the Bragg peak.
• Lateral dose tails are reproduced within 2% compared to measurements, validated by field size factor tests.
• The GPU-accelerated tracing kernel achieves high performance (10 million primary particles/sec on a single card), enabling rapid treatment plan recalculations.

Conclusions:

• The "fred" platform provides a computationally efficient and accurate solution for ion beam therapy treatment planning.
• GPU acceleration significantly reduces simulation time, making advanced MC methods more accessible for clinical use.
• This advancement supports more precise radiation delivery, potentially improving patient outcomes in tumor radiation therapy.