ESTRO 2024 - Abstract Book

S3457

Physics - Dose calculation algorithms

ESTRO 2024

3221

Digital Poster

Expansion of OpenMC Monte Carlo Functionality for Boron Neutron Capture Therapy Dose Computation

Perry Young 1 , Dave Dewitt 2 , Warren Kilby 3 , Jed Styron 1 , Ales Necas 4 , Chad Lee 3

1 TAE Technologies, Nuclear Engineering, Foothill Ranch, CA, USA. 2 TAE Life Sciences, Software Development, Irvine, CA, USA. 3 TAE Life Sciences, Clinical Support, Irvine, CA, USA. 4 TAE Technologies, Plasma Sciences, Foothill Ranch, CA, USA

Purpose/Objective:

Dosimetry for Boron Neutron Capture Therapy (BNCT) involves the production and transport of neutrons, photons, and multiple high-LET particle species. OpenMC 1 is a modern Monte Carlo (MC) code that provides speed advantages over other codes currently used for BNCT dose computation, but lacks some necessary features for patient dosimetry and requires careful baseline comparisons with established MC codes. The OpenMC dose engine is integrated with RayStation 2 ’s BNCT module, which collects all dose into four component groups (hydrogen dose, nitrogen dose, boron dose and gamma dose). Careful testing of all possible neutron-induced reactions in elements expected in human tissues is necessary to avoid underestimating total physical and biologically weighted doses. The present work describes baseline testing of OpenMC against existing MC codes, expansion of code functionality to improve photon dose computational speed and accuracy, and calculational speed comparisons with other codes used for BNCT.

Material/Methods:

Software was written to convert three-dimensional DICOM voxel and material data into input files for OpenMC and MCNP 3 , the MC code used for baseline testing. Doses for neutron-induced reactions, including elastic scattering and (n,p) conversions, were calculated in both codes using voxel-specific KERMAs integrated over the energy-dependent neutron flux using a track length estimator (TLE). The TLE for photons was available for MCNP, but was unavailable in OpenMC when testing began. All energetically-allowed neutron and photon reactions were analyzed in all elements expected in human tissues to determine what reactions may contribute significantly to the total dose but not correspond to an established RayStation group. Calculations were also performed with the code Phits 4 , and speed comparisons were performed with both OpenMC and MCNP.

Results:

The enabled method for calculating dose in OpenMC at the start of testing was a collision estimator instead of TLE. Our testing showed re-introduction of TLE to OpenMC improved uncertainty relative to the collision estimator by 70%, with no impact on calculation time. Timing studies with MCNP and Phits showed up to +140% and +1500% timing increases compared to OpenMC. Testing was performed with using a single lateral beam for a target in the cranium, voxel size of 1.3mm and 10 8 source neutrons.