A 2D couch attenuation model for in vivo EPID transit dosimetry

Transit-based in vivo EPID dose reconstruction algorithms need to correct for the couch attenuation at the exit side of the patient. In this study, two couch attenuation models are presented. The simple 0D model applies a single pixel-independent correction which is gantry angle and energy dependent. The more complex 2D model incorporates the dimensions and the treatment position of the couch to apply a pixel-specific correction by calculating the couch radiological thickness during delivery. The models were first described and then validated with measurements made on a phantom against measured couch attenuation maps and TPS data. There was a critical range of gantry angle values in which 2D models were required to accurately reconstruct the delivered dose. This was corroborated with the in vivo 2D dose verification results of 200 IMRT anterior fields. Without a couch attenuation model, the in vivo reconstructed isocentre dose value was on average 2.6% +/- 1.0% (1SD) lower. In the in vivo 3D dose verification of 80 VMAT arcs and 75 IMRT treatments, both models showed comparable results. Without a couch attenuation model, the in vivo reconstructed PTVD50 value was on average 1.2% +/- 0.5% (1SD) lower. The 0D and 2D couch attenuation models are generic and directly suitable for implementation in an in vivo EPID dosimetry solution.