Doctor of Philosophy (PhD)


Physics and Astronomy

Document Type



Pencil beam algorithms (PBAs) are often utilized for dose calculation in proton therapy treatment planning because they are fast and accurate under most conditions. However, as discussed in Chapman et al (2017), the accuracy of a PBA can be limited under certain conditions because of two major assumptions: (1) the central-axis semi-infinite slab approximation; and, (2) the lack of material dependence in the nuclear halo model. To address these limitations, we transported individual protons using a class II condensed history Monte Carlo and added a novel energy loss method that scaled the nuclear halo equation in water to arbitrary geometry. Our results indicated significant reductions in primary dose difference distal to laterally finite slab heterogeneities (~15%) compared to our previous model. Furthermore, our improved nuclear halo model decreased the distance-to-agreement (DTA) of the 1% isodose lines near heterogeneities by ~2-7 mm, and resulted in significant in-field improvement for deep air slabs (~2% improvement in total dose at the peak). Evaluation of both of these improvements in more clinically relevant geometries revealed an improved DTA of the 1% isodose line (~0.3-3 mm) and a reduction of maximum dose near the peak (18-27% reduced to 6-15%). Overall, the two modeling improvements made in this work have resulted in a dose model with significantly higher dose calculation accuracy across a wide range of particularly challenging geometries.



Committee Chair

Fontenot, Jonas