Purpose Christchurch Hospital has implemented EPIgray, an EPID based in-vivo dosimetry (IVD) sys- tem for dosimetric verification of treatment. The current EPIgray clinical tolerance of ±5% has been set for all sites treated using 3D conformal radiation therapy (3DCRT). This project aims to optimise the use of EPIgray for treatment verification of 3DCRT breast treatments, by estab- lishing a site specific clinical tolerance, reducing false positive results and investigating causes of error.

Methods A retrospective study was performed in which 30 previously treated 3DCRT breast patients were re-analysed within EPIgray with the current tolerance. EPID based IVD systems are susceptible to setup errors on treatment. Positional corrections in EPIgray were performed to identify groups of patients at increased risk of having setup errors on treatment, and the impact of these on the EPIgray result. For controlled tests to validate the findings of the ret- rospective study, a phantom representing breast and lung tissue was designed. The phantom was positioned for treatment with setup errors of known magnitude in all orthogonal planes to compare EPIgray reconstructed doses against other routinely used dosimeters. An uncertain- ties investigation was also conducted in which the uncertainties associated with the EPIgray reconstructed doses were determined specifically for the treatment parameters used for 3DCRT breast treatments.

Results The retrospective study has revealed that right breast patients and large breast patients were more prone to setup errors than left breast patients and small breast patients. The phantom study revealed that EPIgray reported doses are in agreement with the delivered dose. Whilst the accuracy of EPIgray was determined to be systematically offset from other dosimeters, the results have indicated that EPIgray can be reliably used with appropriate tolerances. The re- sults of the uncertainties investigation have shown that the intervention clinical tolerance for 3DCRT breast treatments can be reduced to ±3.5% with the aim of detecting potential treat- ment errors whilst still accounting for inherent uncertainties associated with EPIgray and the uncertainties associated with the variations in treatment.

Conclusions This project has clinically validated the use of EPIgray for treatment verification of 3DCRT breast treatments. A new intervention tolerance of ±3.5% was recommended for all 3DCRT breast patients. This tolerance is smaller than the current intervention tolerance of ±5.0% and would aid in the reduction of false results.