Scatter-to-primary ratio and grid performance in contrast-enhanced dual-energy digital mammography using Monte Carlo simulations

Scattered radiation and grid performance were investigated for contrast-enhanced dual-energy mammography (CEDM) using Monte Carlo simulations through analysis of the scatter-to-primary ratio (SPR), grid selectivity (I), Bucky factor (BF), and signal-to-noise ratio improvement factor (KSNR). These parameters were investigated for different grid designs: a 5:1 linear grid (G52D) and a 4:1 cellular grid (G42D), both employed for 2D mammography, and 5:1 and 11:1 linear grids (G5BT and G11BT, respectively), both employed for breast tomosynthesis. Different breast thicknesses, compositions, along with direct (aSe) and indirect (CsI) image receptors were investigated. Additionally, different x-ray spectra were also evaluated, for low-energy (LE) imaging, with the W/Rh and W/Ag combinations for tube potentials between 24-40 kV, and for high-energy imaging (HE), with W/Cu, Mo/Cu and W/Ti combinations between 40-49 kV. The results showed that the HE spectra yielded higher SPR values than the LE spectra, regardless of grid usage. The dependence of SPR on the tube potential varied between the LE and HE spectra and the image receptor. For the LE spectra, the G11BT and G42D grids exhibited lower SPR values, whereas the 5:1 linear grids exhibited higher values. For the HE spectra, higher SPR values were observed with the G5BT and G42D grids and lower values were observed with the G11BT grid. The grid selectivity decreased with tube potential and breast thickness. For the LE beam, the selectivity was highest for the G11BT and G42D grids, and lowest for the G5BT grid. For the HE beam, the selectivity was highest for G11BT and G5BT grids and lowest for the G5BT and G42D grids. The Bucky factor decreased with tube potential, but increased with increasing breast thickness. Higher BF values were observed for the breast tomosynthesis grids than for the 2D imaging grids for both the LE and HE beams. KSNR values increased with tube potential for the LE anode/filter combinations, whereas for the HE combinations, the image receptor influenced this relationship. KSNR values also increased with breast thickness, with the G42D and G11BT grids showing higher values for the LE and HE spectra, respectively. Lower KSNR values were observed for the G5BT grid for both LE and HE beams. The concurrent analysis of KSNR and BF indicated that, for the LE spectrum, the G42D cellular grid exhibited better performance, followed by the G52D, whereas the G5BT and G11BT grids performed similarly. For the HE spectrum, the G42D and G52D grids demonstrated comparable and better overall performance, followed by the G11BT and G5BT grids. The results of this study demonstrate that, in CEDM, the properties of scattered radiation and grid performance differ significantly between the low- and high-energy spectra used.