Room: Exhibit Hall | Forum 9
Purpose: Electronic noise plays an important role in low dose electronic portal imaging device (EPID) images. When modeling with Monte Carlo (MC), electronic noise should be included in the overall noise estimation. In this study, an EPID MC model of a clinical imager is created with electronic noise included and validated against measurement.
Methods: A validated EPID model was built with the Geant4 application for tomographic emission (GATE). Both radiative and optical photon transport were included. Electronic noise was obtained empirically by subtraction between a single frame dark field image and an averaged dark field image (over 200 frames). By fitting the histogram of pixel noise response with a Gaussian distribution, electronic noise was expressed statistically with mean value of 0 and standard deviation of 3.15. Thus, an electronic noise image can be reconstructed by randomly generating pixel responses following the same mean and standard deviation. This is then added to a MC simulated EPID image, which has been normalized to a measured image. We validated the electronic noise estimation against measurement using a standard Las Vegas phantom.
Results: We created a MC model of the Varian as1200 EPID in GATE and validated it using experimental data. We found that adding electronic noise to MC-generated images doesnâ€™t substantially affect contrast, with differences between 0.5-1.5% for 0.25 MU. However, CNR accuracy is greatly improved with changes in absolute difference reducing from 8.5%-16% to 2.5%-6.5% for 0.1-0.5 MU acquisitions.
Conclusion: Including empirically derived electronic noise in the MC EPID model demonstrates better agreement with low dose measurements. This model can be used to generate realistic low dose EPID images for evaluation of imaging techniques like MV-CBCT.
Funding Support, Disclosures, and Conflict of Interest: This work was supported, in part, by award number R01CA188446 from the National Institutes of Health and a research grant from Varian Medical Systems, Inc.