Metal Artifact Reduction Computed Tomography of Arthroplasty Implants: Effects of Combined Modeled Iterative Reconstruction and Dual-Energy Virtual Monoenergetic Extrapolation at Higher Photon Energies
Iman Khodarahmi,Reham R Haroun,Moses Lee,George S K Fung,Matthew K Fuld,Lew C Schon,Elliot K Fishman,Jan Fritz,Reham R. Haroun,George S.K. Fung,Matthew K. Fuld,Elliot K. Fishman
DOI: https://doi.org/10.1097/RLI.0000000000000497
IF: 6.7
2019-10-04
Investigative Radiology
Abstract:Objective
The aim of this study was to compare the effects of combined virtual monoenergetic extrapolation (VME) of dual-energy computed tomography data and iterative metal artifact reduction (iMAR) at higher photon energies on low- and high-density metal artifacts and overall image quality of the ankle arthroplasty implants with iMAR, weighted filtered back projection (WFBP), and WFBP-based VME.
Materials and Methods
Total ankle arthroplasty implants in 6 human cadaver ankles served as surrogates for arthroplasty implants. All specimens underwent computed tomography with a 2 × 192-slice dual-source computed tomography scanner at tube voltages of 80 and tin-filtered 150 kVp to produce mixed 120 kVp equivalent polychromatic and virtual monoenergetic extrapolated images at 150 and 190 keV (VME 150 and VME 190, respectively). By implementing the WFBP and iMAR reconstruction algorithms on polychromatic, VME 150 and VME 190 data, 6 image datasets were created: WFBP-Polychromatic, iMAR-Polychromatic, WFBP-VME 150, WFBP-VME 190, iMAR-VME 150, and iMAR-VME 190. High-density and low-density artifacts were separately quantified with a threshold-based computer algorithm. After anonymization and randomization, 2 observers independently ranked the datasets for overall image quality. Repeated measures analysis of variance, Friedman, and Cohen weighted κ tests were applied for statistical analysis. A conservative P value of less than 0.001 was considered statistically significant.
Results
iMAR-VME 190 keV and iMAR-VME 150 keV created the least amount of high-density artifacts (all P < 0.001), whereas iMAR-Polychromatic was the most effective method to mitigate low-density streaks (P < 0.001). For low- and high-density artifacts, polychromatic iMAR acquisition was superior to WFBP-VME 150 keV and WFBP-VME 190 keV (all P < 0.001). On sharp kernel reconstructions, readers ranked the overall image quality of iMAR-Polychromatic images highest (all P < 0.001). Similarly, on soft tissue kernel reconstructions, readers ranked iMAR-Polychromatic images highest with a statistically significant difference over other techniques (all P < 0.001), except for iMAR-VME 150 keV (P = 0.356).
Conclusions
In computed tomography imaging of ankle arthroplasty implants, iMAR reconstruction results in fewer metal artifacts and better image quality than WFBP reconstruction for both polychromatic and virtual monoenergetic data. The combination of iMAR and VME at higher photon energies results in mixed effects on implant-induced metal artifacts, including decreased high-density and increased low-density artifacts, which in combination does not improve image quality over iMAR reconstruction of the polychromatic data. Our results suggest that, for ankle arthroplasty implants, the highest image quality is obtained by iMAR reconstruction of the polychromatic data without the need to implement VME at high-energy levels.
radiology, nuclear medicine & medical imaging