Computer-assisted beam modeling for particle therapy.

Area of Science:

• Medical Physics
• Radiation Oncology
• Computational Physics

Background:

• Accurate Monte Carlo (MC) beam models are crucial for precise proton and carbon ion therapy planning.
• Current methods for generating these models can be user-dependent and time-consuming.
• A need exists for a more automated and straightforward approach to MC beam model generation.

Purpose of the Study:

• To develop a computer-driven, less user-dependent method for generating MC beam models of scanned proton and carbon ion beam delivery systems.
• To enable simple and straightforward generation of these complex models.

Main Methods:

• Experimental measurements of proton and carbon ion beam properties (depth dose, spot size) were performed.
• An automated regularization-based optimization process (AUTO-BEAM) in GATE/Geant4 was used to tune beam parameters (energy, energy spread, beam sigma, divergence, emittance).
• The method was validated against independent measurements and applied to multiple beam lines across three particle therapy centers.

Main Results:

• MC beam models achieved excellent agreement with measurements: particle beam ranges within 0.2 mm and spot sizes within 0.4%.
• Dose calculations showed agreement within 1.7% for a representative clinical proton case.
• Beam model generation for protons was completed within three working days, demonstrating efficiency.

Conclusions:

• The developed method is independent of beam optics, particle type, and geometry, proving its versatility.
• The approach is suitable for non-expert users with minimal interaction required.
• Validated beam models for different systems showed good agreement, confirming the method's reliability.