Optimization of a real time dual-energy subtraction technique based on a flat panel detector

Purpose: To determine the optimal x‐ray spectra for dual‐energy subtraction with a flat panel detector (FPD), and to quantify the effects of dynamic filtration and FPD gain settings on image quality, tube loading, patient exposure, and overall system performance. Method and Materials: A simulation study was performed using available empirically determined data of x‐ray spectra. The lungs and mediastinum of the chest were modeled with 12.5 cm and 20 cm thick regions of Lucite, respectively. Coronary calcification was modeled with a 1 mm thickness of bone‐equivalent plastic. The FPD was modeled as an ideal detector with a 600 micron thick layer of CsI. Scatter was not considered in this study. Low and high energy images were normalized to a desired energy deposit in the detector dependent on FPD gain. A figure‐of‐merit (FOM) was used t6o quantify the overall system performance. The effects of various silver filter thicknesses (0–1000 microns), high to low energy image signal ratios (1–8), and two dual‐energy noise reduction algorithms were evaluated for their effect on image quality, patient entrance exposure, tube loading, and FOM improvement. Results: As the thickness of the high‐energy filter increased, image contrast, contrast‐to‐noise ratio, FOM and tube loading increased. Patient exposure was reduced by approximately 10% for the range of filter thicknesses studied. The FOM was maximized with a FPD signal ratio of approximately 3 without application of dual‐energy noise reduction. However, dual gain operation did not show any improvement after noise reduction. Conclusion: An optimal dynamic filter combined with dual‐energy noise reduction can improve the system FOM by a factor of 5. There is no further improvement in the system FOM when dual detector gain is used in conjunction with dual‐energy noise reduction. © 2005, American Association of Physicists in Medicine. All rights reserved.