Path-Length-Based Monte Carlo Simulation for Primary Proton Fluence In Dose Calculation
Poster
PrimaryProtonAAPM26QX.pdf Hosted by AAPM · 0.8 MBAbstract
Purpose
Proton therapy offers superior therapeutic ratio due to its favorable dose distribution, but accurate dose calculation remains challenging. Current methods rely on either pencil-beam algorithms (fast but inaccurate in heterogeneous media) or Monte Carlo simulations (accurate but computationally expensive). This work introduces a primary proton model and a fluence calculation method to enable faster, accurate dose calculation in clinical applications.
Methods
Particles in proton therapy exhibit distinct propagation characteristics. We classify them as primary protons (continuous small-scale energy loss and direction changes) and secondary particles (wider directional distribution). Following the collapsed-cone convolution/superposition framework, secondary particle fluence is derived from primary proton fluence. Using Geant4 as ground truth, we define primary protons by energy loss per step in a homogeneous medium using a threshold of 1.2×(2×P₉₉−P₉₈), where P₉₉ and P₉₈ are the 99th and 98th percentiles of stepwise energy loss. We traced 250 MeV protons and collected comprehensive energy and directional statistics. Scattering angular distributions were approximated as a function of path length despite a moderate energy spread at equal path lengths. We validated our approach against Geant4 using a custom Monte Carlo program that traces primary protons by path length and samples directional changes from collected statistics.
Results
We calculated the primary proton fluence of a 250 MeV infinitesimal incident beam using both Geant4 and our custom Monte Carlo method in a water phantom. Our simulation achieved an 18.36× speedup over Geant4 (14 vs. 257 seconds). Due to rapid beam spreading, fluence values beyond the incident region are low; we benchmarked using logarithmic scale. Our method achieved gamma passing rates of 95.11%, 98.18%, and 99.78% at criteria 1.5%/1.5 mm, 2.0%/2.0 mm, and 3.0%/3.0 mm, respectively.
Conclusion
We successfully developed a path-length-based Monte Carlo simulation for primary proton fluence, achieving a significant computational speedup while maintaining high accuracy.