Monte Carlo evaluation of uncertainties in photon and electron TG-51 absorbed dose calibration

The accuracy of dose delivery to all patients treated with medical linacs depends on the accuracy of beam calibration. Dose delivery cannot be any more accurate than this. Given the importance of this, it seems worthwhile taking another look at the expected uncertainty in TG-51 photon dose calibration and a first look at electron calibration. This work builds on the 2014 addendum to TG-51 for photons and adds to it by also considering electrons. In that publication, estimates were made of the uncertainty in the dose calibration. In this paper, we take a deeper look at this important issue.

The methodology used here is more rigorous than previous determinations as it is based on Monte Carlo simulation of uncertainties. It is assumed that mechanical QA has been performed following TG-142 prior to beam calibration and that there are no uncertainties that exceed the tolerances specified by TG-142.

Despite the different methodology and assumptions, the estimated uncertainty in photon beam calibration is close to that in the addendum. The careful user should be able to easily reach a 95% confidence interval (CI) of ± 2.3% for photon beam calibration with standard instrumentation. For electron beams calibrated with a Farmer chamber, the estimated uncertainties are slightly larger, and the 95% CI is ±2.6% for 6 MeV and slightly smaller than this for 18 MeV. There is no clear energy dependence in these results. It is unlikely that the user will be able to improve on these uncertainties as the dominant factor in the uncertainty resides in the ion chamber dose calibration factor $$ . For both photons and electrons, reduction in the ion chamber depth uncertainty below about 0.5 mm and SSD uncertainty below 1 mm have almost no effect on the total dose uncertainty, as uncertainties beyond the user's control totally dominate under these circumstances.