Clinical implications of adopting Monte Carlo treatment planning for CyberKnife.

It is documented that well-modeled Monte Carlo dose calculation algorithms are more accurate than traditional correction-based algorithms or convolution algorithms at predicting dose distributions delivered to heterogeneous volumes. This increased accuracy has clinical implications for CyberKnife, particularly when comparing dose distributions between the ray-tracing and Monte Carlo algorithms. Differences between ray-tracing and Monte Carlo calculations are exacerbated for highly heterogeneous volumes and small field sizes. In this study, the anthropomorphic thorax phantom from the Radiological Physics Center was used to validate the accuracy of the CyberKnife Monte Carlo dose calculation algorithm. Retrospective comparisons of dose distributions calculated by ray-tracing and Monte Carlo were made for a selection of CyberKnife treatment plans; comparisons were based on target coverage and conformality. For highly heterogeneous cases, such as those involving the lungs, the ray-tracing algorithm consistently overestimated the target dose and coverage. In our sample of lung treatment plans, the average target coverage for ray-tracing calculations was 97.7%, while for Monte Carlo, the average coverage dropped to 69.2%. In each plan comparison, the same beam orientations and monitor units were used for both calculations. Significant changes in conformality were also observed. Isodose prescription lines and subsequent target coverage selected for treatment plans calculated with the ray-tracing algorithm may be different from comparable treatment plans calculated with Monte Carlo, and as such, may have clinical implications for dose prescriptions.