Comparison of Metal Artifact Reduction in Dual- and Single-Source CT: A Vertebral Phantom Study

The objective of this study was to compare the capability of two algorithms for metal artifact reduction and virtual monoenergetic imaging (VME), a metal artifact reduction application for dual-source CT.

A bovine vertebra phantom with 16 artificial osteolyses and two 20 × 4.5 mm stainless steel screws was scanned on two single-source CT scanner and one dual-source CT scanner at a dose identical to the single-source acquisitions. Datasets were reconstructed with a metal artifact reduction algorithm for orthopedic implants (O-MAR, Philips Healthcare), an iterative metal artifact reduction algorithm (iMAR, Siemens Healthineers), and VME. Blinded to the method used for artifact reduction, three independent observers evaluated datasets regarding the extent of metal artifacts using a 4-point scale. Depicted osteolyses were counted and screw diameters measured for each reconstruction. Interobserver variability was evaluated using the Kendall coefficient of concordance for ordinal variables and the intraclass correlation coefficient for continuous data.

VME showed the best metal artifact reduction capability among evaluated methods; overall artifacts were rated 1.08 ± 0.29 for VME, 3.33 ± 0.65 for iMAR, and 3.91 ± 0.29 for O-MAR (p < 0.01). VME resulted in better representation of the cortical bone, trabecular structure, and soft tissue compared with the other two algorithms. VME provided the most realistic reconstruction of screw diameter. However, VME missed osteolyses. Good to almost perfect agreement was achieved for nearly all evaluated attributes.

In our vertebral phantom, VME led to the most detailed representation of the osteosynthesis screw, caused the lowest amount of artifact, and represented the adjacent tissue best. Thus, VME should be considered as an alternative method to evaluate implants when other algorithms fail.