This implies that the correlation between the two images obtained from the different X-ray energy is negligible. These observations allowed us to consider flood-field images used in DE NNPS analysis as independent processes. It is worth noting that Equation (15) described the measured DE NNPS. The effects of various wf, t Sn, ξ, and NR filtering on the DE NNPS are shown in Fig. The ACNR filtering showed higher DE MTF performance, with respect to the other NR filtering at the spatial frequencies above 2.3 mm -1 (i.e., high-frequency), as shown in Fig. Hence, these characteristics resulted from the subtraction of two MTFs having different spatial-resolution characteristics, and HE images were filtered by GNR, which reduces noise but degrades spatial resolution. All NR filtering decreased the DE MTF, the GNR filtering showed boost-up characteristics at the spatial frequencies 1.3 mm -1 because of Gaussian filtering of the HE image. 5C, the DE MTF was nearly independent of ξ, except for ξ=0.1. Therefore, it is important to analyze the X-ray interaction through Monte Carlo N-particles (MCNP) simulations (e.g., particle-tracking tally). 5B, DE MTF degraded by increasing Sn-filter thickness could be affected by scatter X-rays and characteristic X-rays. From DE MTF analysis with increasing the t Sn in Fig. As wf increased, the DE MTF was decreased. The DE MTF was largely dependent on wf used for reconstruction as shown in Fig. 4D, the SDNR performance of ACNR was superior when ξ was less than 0.3, whereas the SDNR performance of MNR and GNR was superior when ξ was greater than 0.4.įig. Particularly, reflecting the results of Fig. All the NR filtering suppressed image noise, and improved SDNR performance. Without NR filtering, ξ=0.3 represented the optimal ξ. 4F shows the SDNR performance for ξ used in DE reconstruction. The highest ACNR performance result supports the results of reducing the size of random noise, such as yellow profiles in columns 1 and 4 of Figs. The results are similar to those reported by Warp and Dobbins, who demonstrated a reduction in the noise component at various spatial frequencies for these, and other noise reduction algorithms. The SDNR performance of the ACNR increased the most for t Sn, as shown in Fig. A larger energy separation between the two energies with t Sn enhanced the SDNR, and the performances of GNR and MNR had similar performance with increasing t Sn. However, the noise performance of ACNR was superior when ξ was less than 0.3, while the noise performance of MNR and GNR was superior with the fine difference, when ξ was greater than 0.4, as shown in Fig. 3C, the ACNR showed a better noise performance of artificial-nodule-enhanced DE images than the MNR and GNR as a function of the t Sn. Hence, NR filtering maintained the SD value almost constant. The effects of t Sn and ξ on SD performance were nearly negligible, as shown in Fig. 4 shows signal difference (SD), noise, and SDNR calculated from the artificial-nodule-enhanced DE images for various combinations of t Sn and ξ.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |