Type-B Uncertainty Evaluation of Monte-Carlo Dose Calculations for a Clinical Linac
Abstract
Purpose
To determine the Type-B uncertainty of dose calculations using a Monte-Carlo model of a clinical linac and to quantify the contributions of different sources of uncertainty.
Methods
A Monte Carlo model of an Elekta Synergy LINAC, producing a 6 MV filtered photon beam, was commissioned, achieving 100 % agreement using a (1 mm, 1 %) gamma passing rate along the percentage depth dose (PDD) curve and both transverse dose profiles for multiple square field sizes. Subsequently, following the procedure of the JCGM Guide to the Expression of Uncertainty in Measurement (GUM), the combined standard uncertainty was determined by carrying out extensive Monte Carlo simulations of dose distributions in a water phantom using EGSnrc, while varying individual parameters representing distinct sources of uncertainty.
Results
The spatially resolved standard uncertainty of the relative dose distributions was determined for three different field sizes. The largest uncertainty contribution near the surface along the PDD was found to be below 1 % and arises from uncertainties in the primary electron properties. Beyond the depth of dose maximum, uncertainties in photon interaction cross sections contributed most to the total uncertainty (< 1 % within the observed depth range of 25 cm. Along the transverse dose profiles, significant uncertainties arose from multiple sources (in-plane < 8 %, cross-plane < 12 %). The product of the spatially resolved combined standard uncertainty and the relative dose value at each respective position resulted in deviations below 1 % across the entire PDD and 4 % across the transverse dose profiles.
Conclusion
Although Monte Carlo simulations are widely regarded as the ground truth for dose calculations, they exhibit non-statistical uncertainties. For the LINAC model discussed in this work, a conservative uncertainty budget was established that yields small overall uncertainties.