EPID-based linear accelerator benchmarking using pixel sensitivity map

Conventional verification of linac performance is conducted by measuring percent depth dose (PDD) as well as beam profiles in a water tank, which is labor intensive and requires expensive beam scanning equipment. We present a novel method to benchmark the linac beam characteristics using pixel sensitivity map (PSM) corrected EPID images and to evaluate the feasibility of EPID-based beam benchmarking. A novel approach was developed to generate a two-dimensional (2D) PSM for EPIDs utilizing an alternating beam and dark-field image acquisition technique. The calculated PSM based on a recursive calculation algorithm was used to correct EPID pixel response. The acceptability of a benchmarked beam is based on its profiles being within predetermined tolerances. The PSM-corrected open field EPID images were compared for three photon energies (6 MV, 10 MV and 6FFF) between three TrueBeam and one Edge linacs with aS1200 imaging panels. The output factors were measured with EPID from 2-2 cm² to 40-40 cm². The differences in beam profiles and output factors from EPID were evaluated. The four dosimetrically similar linacs with less than 1% variation in PDD, profiles and output factors measured in water scans showed excellent agreement in the EPID profile measurements. The 1D Gamma of the PSM-corrected profiles between any two linacs showed 100% passing rate for 6 MV and 6 FFF and >97% for 10 MV with 1mm/1% criteria. The maximum difference of output factors was 0.18% among all the measurements except for 2-2 cm² with 0.6% difference. By preserving the beam dosimetry information, the PSM-corrected EPID images enable the possibility of validating the beam data of linac dosimetry and benchmarking linac performance with the common EPID detector.