Macroscopic geometric heterogeneity effects in radiation dose distribution analysis for boron neutron capture therapy.

Calculations of radiation flux and dose distributions for boron neutron capture therapy (BNCT) of brain tumors are typically performed using sophisticated three-dimensional analytical models based on either a homogeneous approximation or a simplified few-region approximation to the actual highly heterogeneous geometry of the irradiation volume. Such models should be validated by comparison with calculations using detailed models in which all significant macroscopic tissue heterogeneities and geometric structures are explicitly represented as faithfully as possible. This paper describes such a validation exercise for BNCT of canine brain tumors. Geometric measurements of the canine anatomical structures of interest for this work were performed by dissecting and examining two essentially identical Labrador retriever heads. Chemical analyses of various tissue samples taken during the dissections were conducted to obtain measurements of elemental compositions for the tissues of interest. The resulting geometry and tissue composition data were then used to construct a detailed heterogeneous calculational model of the Labrador head. Calculations of three-dimensional radiation flux distributions pertinent to BNCT were performed for this model using the TORT discrete-ordinates radiation transport code. The calculations were repeated for a corresponding volume-weighted homogeneous-tissue model. Comparison of the results showed that peak neutron and photon flux magnitudes were quite similar for the two models (within 5%), but that the spatial flux profiles were shifted in the heterogeneous model such that the fluxes in some locations away from the peak differed from the corresponding fluxes in the homogeneous model by as much as 10%-20%. Differences of this magnitude can be therapeutically significant, emphasizing the need for proper validation of simplified treatment planning models.