Evaluation of an electron Monte Carlo dose calculation algorithm for electron beams

Abstract

The electron Monte Carlo (eMC) dose calculation algorithm of the Eclipse treatment planning system is based heavily upon Monte Carlo simulation of the linac head and modeling of the linac beam characteristics with minimal measurement of beam data. Commissioning of the eMC algorithm on multiple identical linacs provided a unique opportunity to systematically evaluate the algorithm with actual measurements of clinically relevant beam and dose parameters. In this study, measured and eMC calculated dose distributions were compared both along and perpendicular to electron beam direction for electron energy/applicator/ depth combination using measurement data from four Varian CLINAC 21EX linear accelerators (Varian Medical Systems, Palo Alto, CA). Cutout factors for sizes down to 3 × 3 cm were also compared. Comparisons between the measurement and the eMC calculated values show that the R90, R80, R50, and R10 values mostly agree within 3 mm. Measure and calculated bremsstrahlung dose Dx correlates well statistically although eMC calculated Dx values are consistently smaller than the measured, with maximum discrepancy of 1% for the 20 MeV electron beams. Surface dose agrees mostly within 2%. Field width and penumbra agree mostly within 3 mm. Calculation grid size is found to have a significant effect on the dose calculation. A grid size of 5 mm can produce erroneous dose distributions. Using a grid size of 2.5 mm and a 3% accuracy specified for the eMC to stop calculation iteration, the absolute output agrees with measurements within 3% for field sizes of 5 × 5 cm or larger. For cutout of 3 × 3 cm, however, the output disagreement can reach 8%. Our results indicate that the eMC algorithm in Eclipse provides acceptable agreement with measurement data for most clinical situations. Calculation grid size of 2.5 mm or smaller is recommended.