A Study of IMRT Pre-Treatment Dose Verification Using a-Si Electronic Portal Imaging Devices

Abstract

<p>Intensity-Modulated Radiation Treatment (IMRT) requires patient-specific quality assurance measurements, which can benefit from the convenience of using an Electronic Portal Imaging Device (EPID) for dose verification. However, EPIDs have limitations stemming from the non-uniform backscatter due to the support-arm as well as from scatter, glare, and an increased sensitivity to low-energy photons. None of these effects is typically accounted for in a treatment planning system (TPS) model, resulting in errors in calculated EPID response of up to 6%. This work addresses the non-uniform backscatter by directly incorporating a support-arm backscatter region into the TPS geometry. The shape of the backscatter region is adjusted iteratively until the TPS-calculated flood-field planar dose matches the flood-field EPID image The scatter, glare and increased low-energy response are addressed by using a radially-dependent Point-Spread Function (Kernel). The kernel is fitted using a least-squares method so that it best reproduces the EPID-acquired image for a checkerboard field. The backscatter-correction method is implemented for a Varian Clinac equipped with a 40 cm x 30 cm (512 x 384 pixel) EPID and a Pinnacle³ TPS and tested for several rectangular and IMRT fields. The scatter, glare and energy-response correction kernel is implemented and tested for a simulated checkerboard field and a simulated IMRT field. Agreement between the EPID-measured image and TPS-calculated planar dose map is seen to improve from 6% to 2% when the backscatter region is added to the Pinnacle³ model. Agreement between the simulated EPID images and simulated TPS images is improved from 14% to approx. 1% when the radially-dependent kernel is used. Simultaneous application of both the backscatter region and Point-Spread Function is a promising direction for future investigations.</p>