Using ANN for thermal neutron shield designing for BNCT treatment room

Area of Science:

• Medical Physics
• Radiation Oncology
• Materials Science

Background:

• Occupational radiation protection is crucial in therapy rooms, especially for techniques like Boron Neutron Capture Therapy (BNCT).
• Escaping neutrons from the beam shaping assembly (BSA) pose a risk, necessitating effective shielding for treatment room doors.

Purpose of the Study:

• To design and evaluate an alternative shielding material to lead for BNCT treatment room doors.
• To effectively absorb thermal neutrons originating from the BSA after traversing room walls and mazes.

Main Methods:

• Simulation of a polymer composite and polyethylene thermal neutron shield using Geant4 Monte Carlo code.
• Prediction of neutron flux and dose values using an artificial neural network (ANN) for rapid parameter optimization.
• Testing ANN stability with Gaussian distributed uncertainties.

Main Results:

• The optimized shielding material, determined by ANN, effectively reduces the radiation dose behind the door below occupational exposure limits.
• ANN calculations enabled simultaneous optimization of shielding parameters (thickness, composition).
• ANN proved to be a reliable and efficient tool for predicting dosimetric results, surpassing time-consuming Monte Carlo simulations.

Conclusions:

• Developed shielding material offers a viable alternative to conventional lead shielding for BNCT facilities.
• Artificial neural networks provide a powerful and efficient method for optimizing radiation shielding designs and predicting dosimetric outcomes.