The Effect of Beam Energy Separation on Dual Energy Radiography

Dual-energy radiography generates separate soft-tissue and bone radiographic images by combining non-linearly high and low energy images, utilizing the different energy-dependent attenuation characteristics of bone and soft tissue. These images have reduced structured noise, making the identification of anatomy and the diagnosis of pathology less difficult. OBJECTIVE: This study compared dual energy radiographic reconstructions for five different values of the beam energy separation. METHODS: A medical General Electric x-ray generator was used to make sets of exposures of two overlying aluminum and Lucite stepwedges at 50 kVp paired with 120, 100, 90, 80, and 70 kVp images. The detector system was Fujifilm FugiIP cassettes type C, processed with a Fujifilm FCR XG5000 reader. The images were exported as RAW files and reconstructed using NIH ImageJ software and Java code written by one of the authors. 7 pairs of images comprised each set of energy combinations. One pair was used to determine the coefficients of the reconstruction algorithm and the other 6 generated reconstruction images for measurement. The pixel values along the steps in the reconstructed images were fit by linear regression to the height of the steps and the RMS was calculated. RESULTS: In the order of decreasing kVp separation, the 6 pairs of reconstructed images fit the phantom steps heights with mean R-squared values for aluminum of 0.9996, 0.9993, 0.9995, 0.9988, and 0.9974 and for Lucite of 0.9958, 0.9897, 0.9845, 0.9818, and 0.9680. All were highly significant, p<0.01. The RMS values were for aluminum were 0.1841, 0.2590, 0.2855, 0.3209, and 0.3995 and for Lucite were 0.5902, 0.8921, 1.3155, 1.1817, and 1.5214. CONCLUSIONS: The decrease in separation of beam energies from 50-120 kVp to 50-70 kVp resulted in a less than a 3% decrease in R-squared values but the RMS increased 117% for aluminum and 158% for Lucite.