Stochastic versus deterministic kernel-based superposition approaches for dose calculation of intensity-modulated...

Area of Science:

• Medical Physics
• Radiation Oncology
• Computational Imaging

Background:

• Radiation arc therapy dose calculations traditionally approximate continuous arcs with static beams.
• Intensity-modulated arc therapy (IMAT) necessitates finer angular spacing due to rapid variations in beam shape and dose per degree.
• Traditional deterministic dose calculation methods face computational challenges with increased beam counts in IMAT.

Purpose of the Study:

• To introduce and evaluate a Monte Carlo kernel-superposition technique (MCKS) for accurate dose computation in rotational radiation therapy.
• To assess the computational efficiency of MCKS for IMAT plans requiring fine angular sampling.
• To compare MCKS performance against established deterministic methods like collapsed-cone convolution.

Main Methods:

• Developed a homegrown Monte Carlo kernel-superposition (MCKS) technique for dose calculation in rotational delivery.
• Generated IMAT plans with 36 static beams, then interpolated to finer angular intervals for dose calculation.
• Utilized random sampling of photons in MCKS to enable efficient computation with a large number of beams.

Main Results:

• MCKS demonstrated insignificant increases in computation time even with up to 720 interpolated beams.
• Dose computation for ten IMRT cases using MCKS took approximately 15-30 minutes per case on a single CPU.
• MCKS proved computationally faster than the collapsed-cone algorithm for plans with over 43 beams.

Conclusions:

• MCKS effectively addresses the need for fine angular spacing in IMAT dose calculations.
• The technique offers a practical and efficient solution for clinical implementation of rotational radiation therapy.
• MCKS provides a significant computational advantage over traditional methods for complex arc therapy plans.