THE EVOLUTIONARY APPROACH FOR TUMOR DOSE WITH FOTELP-VOX TRANSPORT SIMULATIONS

Abstract

This study introduces an evolutionary approach for minimizing risks to organs-at-risk during tumor therapy, utilizing the FOTELP-VOX program in voxel-based transport simulations (author R.Ilić). The use of voxels in the FOTELP program requires spatial constraints within a parallelepiped for targeted irradiation with a particle source. Particle interactions with the initial voxel are identified based on voxel density. The technique continues until the particle's fate is complete, verifying interactions on that path and modeling these processes. To simulate particle delivery, it uses Monte Carlo techniques. When particles are carried from an external source through the human body, the absorbed dose has a 3-D distribution. The CT data is used to characterize the anatomy of the patient. The current methodology is based on a manual trial-and-error approach. In order to expedite the discovery of an optimized solution, we explored various optimization strategies, such as random search, Bayesian optimization (BO), and genetic algorithm (GA), within the framework of FOTELP-VOX. By evaluating these approaches, our research seeks to identify the most effective strategy for minimizing risks to organs-at-risk during radiation exposure. Two novel methodologies, namely FOTELP-VOX-BO and FOTELP-VOX-GA, are proposed. The introduction of FOTELP-VOX-BO and FOTELP-VOX-GA methodologies further expands the research's potential applications and relevance within the domain of Monte Carlo transport simulations