Single-Energy Retrospective Metal Artifact Reduction Using Adaptive Thresholding for Metal Implants in the Abdomen
Sujithraj Dommaraju1, Masoud Nakhaei, Da Zhang
1From the Department of Radiology, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA.
Objective:
To assess impact of single-energy metal artifact reduction (SEMAR) algorithm utilizing retrospective adaptive thresholding in reducing metal artifacts in the abdomen and pelvis.
Methods:
In this prospective institutional review board-approved, Health Insurance Portability and Accountability Act-compliant study, 90 patients with various metals (n = 97) on computed tomography of abdomen and pelvis (Canon Medical, Aquilion ONE and PRIME) scanned 07/2017-09/2018 with SEMAR retrospectively applied were included. Density was measured in the near and far field to the metals. Density standard deviation (SD), representing artifact severity, was compared with and without SEMAR applied. Two trained human observers independently evaluated severity of artifacts on a five-point scale (0, no artifact; 5, severe artifact).
Results:
The SEMAR significantly decreased artifact severity in the near field of high-density metal implants (SD of 204 ± 101HU without vs. 66 ± 40HU with SEMAR, P < 0.001). In the far field, the artifact severity was similar (40 ± 31HU without vs. 36 ± 27HU with SEMAR, P = 0.41). Artifact severity was decreased adjacent to low-density metal in the near field (SD of 86 ± 56HU without vs 49 ± 30HU with SEMAR, P < 0.001). In the far field to the low-density metals artifact severity was similar (33 ± 29HU without vs. 31 ± 27HU with SEMAR, P = 0.79). Subjectively, artifacts severity decreased for high-density metals in near field by 1.3 ± 1.0, and in far field by 0.7 ± 0.7 and for low-density metals in the near field by 0.7 ± 1.0, far field 0.4 ± 0.5, all P < 0.05.
Conclusions:
The SEMAR retrospective algorithm with adaptive thresholding subjectively and objectively reduced near-field artifacts generated by high- and low-density metals.
