Experimental validation of an EPID dosimetry model for pre-treatment QA.

Modern Radiation Therapy treatment plans complexity requires accurate pre-treatment Quality Assurance (QA). Electronic portal imaging devices (EPIDs), embedded on recent LINACs, are employed for dosimetry due to their excellent spatial resolution, dose linearity and large detector area. Their adoption in clinical practice provides reliable, accurate and fast QA. We describe the validation of an EPID dosimetry model on the commercial software SunNuclear EpidDose. Firstly, we tested the model by irradiating the EPID detector with 9 appropriately MLC shaped static fields, mainly to check leaves transmission. Then, 12 selected dynamic VMAT plans were compared in terms of $\gamma$ index ({3%, 3 ${mm}$) with TPS calculations. Both static and VMAT plans were verified through MapCheck and ArcCheck diode arrays and ionization chamber (IC) at significant plan positions. The difference between the $\gamma$ analysis of the EPID and diode arrays against TPS dose maps allows for a cross-check of the model, while IC measurements give absolute dose comparison. The average $\gamma$ passing rate was $97.8%$ for EPID, $99.7%$ for ArcCheck and $96.6%$ for MapCheck, while the mean dose difference was $-0.6%$.