Updated computational performance evaluation of ICRP mesh-type reference computational phantoms using PHITS, Geant4,...

Area of Science:

• Medical Physics
• Computational Dosimetry
• Radiation Transport Simulation

Background:

• The International Commission on Radiological Protection (ICRP) is transitioning to mesh-type reference computational phantoms (MRCPs) for updated dose coefficient calculations.
• Evaluating the computational performance of various Monte Carlo radiation transport codes with these new phantoms is crucial for practical implementation.

Purpose of the Study:

• To systematically assess the computational performance of adult and pediatric MRCPs using updated Monte Carlo codes.
• To compare initialization time, memory usage, computation time, and multi-threading efficiency across different codes.
• To provide guidance for selecting the optimal Monte Carlo code for dose calculations with MRCPs.

Main Methods:

• Evaluated PHITS (v3.31), Geant4 (v11.01.p02), MCNP6.3, and EGSnrc (v2025a) using adult and pediatric MRCPs.
• Assessed performance metrics including initialization time, memory usage, computation time (photons, electrons, neutrons), and multi-threading.
• Compared MRCP performance against existing voxel-type reference computational phantoms (VRCPs).

Main Results:

• EGSnrc had the longest initialization but fastest computation for photons/electrons ≤10 MeV; PHITS was fastest for neutrons.
• PHITS and EGSnrc showed low memory usage (<2 GB), while Geant4 was highest (13-14 GB).
• MCNP6.3 showed longer computation times for MRCPs vs. VRCPs but excellent multi-threading performance.

Conclusions:

• PHITS and Geant4 offer competitive or improved computation times with MRCPs compared to VRCPs for certain scenarios.
• EGSnrc shows efficiency for specific particle types and pediatric MRCPs, while MCNP6.3 excels in multi-threading.
• The findings offer practical insights for researchers selecting Monte Carlo codes for ICRP MRCP-based dose calculations.