Improving photon energy fluence measurement with an amorphous silicon EPID

The present work comprises a study of the physics that underlies the image formation process of a commercial a-Si EPID, the Varian aS500, with particular emphasis on its adaptation for two-dimensional megavoltage dosimetry. Specifically, the work focuses on blur kernel elucidation and dosimetric calibration.; A comprehensive model of the aS500 was developed for Monte Carlo simulations, allowing detector-specific response data to be calculated for direct comparison with corresponding experimental results. Additional, unrelated Monte Carlo work that led to the development of single event spectra for five therapeutic radionuclides is also described.; In this work, we present a comprehensive blur kernel for the aS500 EPID. Monte Carlo techniques were used to derive a dose kernel and an optical kernel, which were combined into an overall blur kernel for 6 and 15 MV photon beams. Experimental measurements of the line spread function verified the kernel shape. Kernel performance was gauged by comparing EPID image profiles (pre- and post-deconvolution) with in-air profiles measured with a diamond detector. Quantitative comparisons demonstrate the relative importance of the optical kernel. Empirical and semi-empirical kernels are shown to closely approach the performance of the comprehensive kernel. A follow-up investigation developed a diamond detector blur kernel and showed that deconvolution of the volume averaging effects for this detector had little influence on results, except for very small fields.; This work also investigated dose calibration curve variability with beam quality for the aS500 EPID. An over-response of the aS500 to low energy (<1 MeV) photons was found to be the main cause of this variability. When a 6 MV beam is hardened by steel shot attenuators, the EPID response can be as much as 8% lower than that for an open field. A Monte Carlo study suggested that this difference is due to spectral effects, and that it could be reduced by the addition of an external copper plate. Experimental results confirm that a copper plate ∼0.7 cm thick placed 15 cm above the EPID can reduce calibration curve differences to <4%, but at the expense of reduced contrast-to-noise ratio and modulation transfer.