Room: AAPM ePoster Library
Purpose: To reduce the uncertainty in dose-to-water electronic portal imaging detector (EPID) pixel calibration by developing a more direct formalism based on a film measurement made inside the EPID panel and detailed Monte Carlo (MC) simulations.
Methods: To relate EPID images to absolute dose in the patient, a typical empirical approach involving dozens of discontinuous measurements and 2D-interpolation was used to calibrate pixel response in terms of primary dose-to-water. To improve the accuracy of this calibration methodology, a PerkinElmer XRD-1640 amorphous silicon (a-Si) EPID was disassembled and dimensionally-measured to model the flat-panel detector in detail using the EGSnrc DOSXYZnrc usercode. As a first stage validation of the MC model, two triple-channel calibrated sheets of Gafchromic EBT3 film were centrally positioned between constituent detector layers, 1 mm upstream and 4 mm downstream from the a-Si array prior to reassembly of the EPID. The panel was then irradiated a 6 MV beam and an EPID image set was concurrently acquired.
Results: The difference in the measured and simulated ratio of film doses was 0.7 %, consistent with measurement uncertainties. From the upstream film and EPID measurement, a 2D pixel sensitivity map was calculated as the ratio of film dose to pixel value corrected for the EPID dark field. In comparing this map to lateral and longitudinal profiles of pixel sensitivity derived from >20 individual EPID measurements and assembled piecewise according to the established empirical method, an average agreement of better than 0.5 % is observed between the pixels in this range.
Conclusion: A direct means of calibrating an a-Si EPID in terms of dose-to-water involving a single measurement and accurate MC simulations is proposed. Reducing the number of required measurements and directly measuring 2D distributions rather than relying on interpolating piecewise data will improve the overall accuracy of EPID dose calibration.