Use of Automatic Metaheuristic Optimizations for Boron Neutron Capture Therapy

Boron Neutron Capture therapy (BNCT) is a cancer treatment method investigated to reach malignant tumors that have added complications due to shape, size, or location. The unique property of BNCT is that it can deposit an immense dose gradient between the tumor cells and normal cells. This is done by selectively concentrating boron compounds in tumor cells and then irradiating the cells with an epithermal neutron beam. Due to complex nature of neutron interactions, a high fidelity neutronic model is needed to design and predict the outcomes of a BNCT experiment. This work aims to introduce a high level autonomous optimization software, a metaheuristic code (GNOWEE) coupled with a high fidelity neutron transport code (MCNP6.2), to generically optimize a Beam Shaping Assembly (BSA) required for any BNCT scenario. This research work leverages from previous work in the nuclear engineering field on metaheuristic optimization used to expand the optimization parameter space for a flexible BSA design. This thesis describes further efforts made in developing a generalized, problem independent Gnowee/MCNP6.2 software. The new software package includes the ability to design a BSA within a high fidelity model for possible future BNCT application.

The expanded parameter space includes the exploration of different neutron sources (e.g. generator based or proton accelerator), of complex geometries, of optimization based on doses to the target, and cost constraints for possible BNCT treatment. Case studies were investigated and modeled for different tumor depth inside a patient's head setup, which resulted in a gradient of 10x the increase in tumor dose respectively to the healthy tissues. In addition, other models studied include the optimization of a BSA for a boron filled cell culture medium, which demonstrated the possibility to tailor any source to a primarily thermal neutron flux. This situation would demonstrate the first step toward a possible BNCT treatment. The results provide a powerful demonstration of the tailoring capabilities of Gnowee/MCNP6.2 for BNCT application and provide improved capabilities to refine the spectrum of neutrons that are most effective for the treatment of both surface and deep-seated tumors.