Accuracy Assessment of Monte Carlo Particle Transport Using Jaw-Interaction–Based Event Control for Multileaf Collimator Systems In Medical Linear Accelerators
Abstract
Purpose
Monte Carlo (MC) methods are widely regarded as the gold standard for radiotherapy dose calculation; however, high computational cost can limit routine clinical application. This study evaluated two jaw-interaction–based event-control techniques in terms of absolute dose accuracy and efficiency compared with the standard analog simulation technique (SAST).
Methods
Simulations were performed with Geant4 (v11.1.1) using a TrueBeam™ (Varian Medical Systems) linear accelerator head. Two techniques were evaluated: the Jaw-Entry Termination Technique (JETT), in which particle transport is terminated upon jaw entry, and the Jaw-Interaction–filtered Transmission Technique (JITT), in which only particles traversing the jaws without interaction are transported. Across photon beam energies from 4 to 15 MV, absolute dose deviations from SAST were evaluated at the field center and directly beneath the jaws, together with simulation time. Evaluations were performed for various field sizes under fully open (40 × 40 cm²) and jaw-matched multileaf collimator configurations.
Results
For JETT, field-center absolute dose deviations were reduced when transitioning from fully open to jaw-matched configurations. In contrast, JITT maintained high accuracy, with dose deviations within 1.0% regardless of MLC configuration, field size, or photon beam energy. For both techniques, in a 10 × 10 cm² field across all energy levels, absolute dose deviations beneath the jaws outside the field were within 1.4%, and penumbra distance-to-agreement was approximately 0.1 mm at the field edge. JETT achieved a maximum speedup of 80-fold, whereas JITT provided an approximately 5-fold reduction in simulation time.
Conclusion
The techniques demonstrated trade-offs between dosimetric accuracy and computational efficiency. JETT achieved substantial acceleration at the expense of larger dose deviations, whereas JITT preserved SAST-level accuracy with meaningful speedup. Jaw-interaction–based event control offers a framework for accelerating MC dose calculations and may be suitable for clinical applications depending on accuracy requirements.