Automatic, machine-agnostic, convolution-based beam, and fluence modeling for Monte Carlo independent dose...

Area of Science:

• Medical Physics
• Radiotherapy
• Computational Dosimetry

Background:

• Monte Carlo (MC) simulations offer accurate dose calculations for radiotherapy quality assurance (QA).
• Challenges in beam modeling and commissioning have limited widespread clinical adoption of MC methods.
• MC is crucial for online adaptive radiotherapy where traditional QA is impractical.

Purpose of the Study:

• To develop a generic, automated, convolution-based method for MC dose calculation beam and fluence modeling.
• To derive all necessary linear accelerator (LINAC) head information from water phantom measurements, minimizing prior knowledge.
• To replace conventional particle transport with a direct convolution method for modeling beam modulation effects.

Main Methods:

• Beam modeling uses percent depth dose (PDD), diagonal profiles, and output factors from water phantom measurements.
• Energy spectrum and electron contamination are optimized to match reference PDD.
• Fluence is calculated by convolving beam modulation data with a kernel; dose is computed via a point source model.

Main Results:

• Achieved >97% gamma passing rates (2%/2mm) against reference beam data.
• Calculated doses for clinical plans showed >97% gamma passing rates (3%/2mm) compared to treatment planning systems.
• Average calculation time for clinical plans was under 1 minute.

Conclusions:

• The machine-agnostic, convolution-based method allows efficient, automated commissioning for various photon beam machines.
• Fluence map-based dose calculation achieves sub-minute efficiency for diverse treatment plans.
• This approach can accelerate the adoption of MC-based QA for online adaptive radiotherapy.