RESUMO
PURPOSE: To investigate various issues for clinical implementation of aSi EPID panels for IMRT/VMAT QA. METHODS: Six linacs are used in our clinic for EPID-based plan QA; two Varian Truebeams, two Varian 2100 series, two Elekta Infiniti series. Multiple corrections must be accounted for in the calibration of each panel for dosimetric use. Varian aSi panels are calibrated with standard dark field, flood field, and 40×40 diagonal profile for beam profile correction. Additional corrections to account for off-axis and support arm backscatter are needed for larger field sizes. Since Elekta iViewGT system does not export gantry angle with images, a third-party inclinometer must be physically mounted to back of linac gantry and synchronized with data acquisition via iViewGT PC clock. A T/2 offset correctly correlates image and gantry angle for arc plans due to iView image time stamp at the end of data acquisition for each image. For both Varian and Elekta panels, a 5 MU 10×10 calibration field is used to account for the nonlinear MU to dose response at higher energies. Acquired EPID images are deconvolved via a high pass filter in Fourier space and resultant fluence maps are used to reconstruct a 3D dose 'delivered' to patient using DosimetryCheck. Results are compared to patient 3D dose computed by TPS using a 3D-gamma analysis. RESULTS: 120 IMRT and 100 VMAT cases are reported. Two 3D gamma quantities (Gamma(V10) and Gamma(PTV)) are proposed for evaluating QA results. The Gamma(PTV) is sensitive to MLC offsets while Gamma(V10) is sensitive to gantry rotations. When a 3mm/3% criteria and 90% or higher 3D gamma pass rate is used, all IMRT and 90% of VMAT QA pass QA. CONCLUSIONS: After appropriate calibration of aSi panels and setup of image acquisition systems, EPID based 3D dose reconstruction method is found clinically feasible.