Speaker
Description
The optimization of Boron Neutron Capture Therapy (BNCT) requires an accurate computational framework capable of integrating complex neutron source characterization with precise energy deposition modeling. In this context and within the activities of the Geant4INFN project, this work presents recent advances of Geant4 simulations, using the Multi-Threading (MT) architecture and exploiting both condensed history and track-structure approaches.
In particular, we executed simulations on local 48-core computers available at INFN unit of Pavia, to investigate the p+$^9$Be reaction at 5 MeV. To minimize statistical noise and ensure robust validation against experimental data, we generated high-statistics distributions (considering 10$^{10}$−10$^{11}$ primaries). This intensive simulation revealed systematic underestimation of neutron yields when employing the QGSP_BIC_AllHP Physics list, specifically due to the omission of the (p,p′n) reaction channel. To resolve this, we implemented a hybrid approach, integrating simulation outputs with a novel data-driven theoretical model, recently published [1]. This methodology was successfully extended to composite $^9$Be+$^7$Li targets, with results currently accepted for publication [2]. Furthermore, we are extending the native Geant4 General Particle Source framework to enable the sampling of 2D energy-angle correlated distributions with updated fast algorithms.
The intensive utilization of local computing resources was beneficial in optimizing both the model development and the simulation time. We are currently evaluating strategies to scale this framework across multiple nodes, aiming to further increase our computational efficiency for a broader range of reaction channels.
Parallel to macroscopic source modeling, another highly computationally expensive research topic vital for BNCT is track-structure simulation. In this context, we use Geant4-DNA, the track-structure extension of the Geant4 Monte Carlo toolkit, to model and quantify DNA-level radiobiological damage induced by BNCT-relevant radiation. Validated against experimental data from the TRIGA reactor in Pavia, these intensive simulations reconstruct a realistic cellular scenario, linking heterogeneous boron distribution to the number of double-strand breaks (DSBs) produced. Ultimately, both of these high-performance computing efforts are carried out also within the framework of the PNC-PNRR ANTHEM project, supporting the future Accelerator-Based BNCT facility at the ‘L. Vanvitelli’ University of Caserta.
References:
[1] A. Colombi, I. Postuma, S. Bortolussi, V. Vercesi and A. Fontana, A new hybrid model for accurate double differential neutron yield calculations on 9Be thick targets for proton-BNCT applications at low energies. Eur. Phys. J. Plus 140, 1142 (2025) https://doi.org/10.1140/epjp/s13360-025-07017-1 (full text available to read at https://rdcu.be/eSapR)
[2] A. Colombi, I. Postuma, S. Bortolussi, V. Vercesi and A. Fontana, Theoretical study of proton-induced reactions on a composite beryllium–lithium target as a BNCT neutron source. Eur. Phys. J. Plus 141, 324 (2026). https://doi.org/10.1140/epjp/s13360-026-07429-7