Efficient Monte Carlo simulation of multileaf collimators using geometry-related variance-reduction techniques.

Area of Science:

• Medical Physics
• Computational Physics

Background:

• Monte Carlo simulations are crucial for accurate radiotherapy dose calculations.
• Simulating complex geometries like multileaf collimators (MLCs) can be computationally intensive, limiting efficiency.

Purpose of the Study:

• To present a novel technique, the movable-skin method, for accelerating Monte Carlo simulations of MLCs.
• To introduce AUTOLINAC, a code that automates MLC simulation setup for the PENELOPE Monte Carlo code.

Main Methods:

• The movable-skin method modifies geometry processing without altering physical leaf shapes, simulating radiation transport accurately in 'skin' zones and approximately in others.
• AUTOLINAC code automates geometry file generation and parameter selection for variance reduction techniques in PENELOPE simulations.
• Adaptive variance-reduction techniques are employed to enhance simulation speed.

Main Results:

• The movable-skin method combined with AUTOLINAC and PENELOPE enables simulation of an entire linac with a closed MLC in approximately two hours.
• Achieved 2% statistical uncertainty (1 sigma) for absorbed dose in water using a 2.8 GHz processor and 2x2x2 mm³ voxel size.
• Simulated MLC configurations showed excellent agreement with experimental measurements.

Conclusions:

• The movable-skin method significantly accelerates Monte Carlo simulations for radiotherapy, making complex linac modeling more feasible.
• AUTOLINAC streamlines the simulation workflow, enhancing the applicability of Monte Carlo methods in clinical settings.
• The developed techniques provide a robust and efficient approach for accurate dose prediction in radiation therapy.