Quality control for electronic portal imaging devices

Full text: The need for objective quality control of electronic portal imaging devices (EPIDs) arises at a number of different stages. The manufacturer should be able to conduct a test before shipping the EPID, to be sure that the device meets predetermined specifications. On installation, the engineer should have a simple and definitive test to ensure that installation is complete. The customer also requires proof that specifications have been met. And a quick and easy test is required for daily or weekly verification that the EPID is operating as well as at installation. All the suppliers of EPIDs provide copies of the 'Las Vegas' phantom, but this is a highly subjective test that was never intended to be used for routine quality control. A quality control test is described which involves acquiring a pair of portal images of a specially designed phantom, followed by totally automatic analysis to provide quantities related to the spatial resolution and the contrast-to-noise ratio. The phantom includes five bar patterns, with spatial frequencies from 0.1 to 0.75 1p/mm. The measured modulation in each pattern is normalized to the value at the lowest frequency, and the spatial frequency f50 corresponding to a 50% decrease in RMIF is used as the measure of spatial resolution. Contrast is measured in regions of the phantom with different attenuations, and noise is obtained from the difference image derived from the original pair. Tests were conducted on EPIDs from all the commercial suppliers. By determining f50 with the phantom placed on the surface of the EPID, and again at isocenter (ISO), it is possible to determine the relative contributions to image blurring from the EPID (E) and from the effect of the finite focal spot size in the linac (S). In all cases the effect of the focal spot is dominant. The results are presented in a table, which shows mean values of the resolution, blurring parameters E and S, the optimal magnification, and the best attainable resolution with the phantom at isocenter and the detector at the optimal magnification. The highest spatial resolution was found with the PortalVision, and results for all EPIDs of this make were very similar. Results from the video-based EPIDs are much more variable. In conclusion, the test has been shown to be simple, fast, objective and sensitive to very small changes