Metal Artifact Reduction in Lumbar Spine Instrumentation

Objective: Photon-counting detector (PCD) computed tomography (CT) systems allow the reconstruction of virtual monoenergetic images (VMIs) from inherently available spectral data. However, qualitative and quantitative evaluations of VMI in combination with tin-prefiltration (Sn) and iterative metal artifact reduction (iMAR), as well as their comparison to energy-integrating detector (EID) CT, are limited; therefore, the objective was to compare their performance qualitatively and quantitatively in the setting of lumbar spine instrumentation. Materials and Methods: Retrospectively, 100 clinical Sn CT scans after lumbar spine fusion surgery (50 EID and 50 PCD) were analyzed. Metal artifacts were quantitatively and qualitatively assessed in EID standard (EID std ), PCD standard (PCD std ), and PCD VMI at 120 keV (PCD VMI120 ), PCD iMAR (PCD iMAR ), and combined PCD VMI at 120 keV + iMAR (PCD VMI120+iMAR ). Regions of interest were placed in the most hyperdense and hypodense areas near metal artifacts, and the attenuation and SD/noise (SD) of Hounsfield Units (HU) were assessed as a quantitative artifact surrogate. For qualitative evaluation, 2 independent readers rated 6 image features on a 4-point Likert Scale. Results: Seventy-four patients were included, of whom 26 underwent both an EID and a PCD CT. Radiation doses were significantly lower ( P < 0.01) for PCD CT median CTDIvol of 6.4 mGy (4.6 to 8.4) compared with EID CT median CTDI vol of 12.5 mGy (11.1 to 13.7). Hypodense artifact attenuation was significantly reduced ( P < 0.01) with PCD iMAR (HU -260), PCD VMI120 (-176), and PCD VMI120+IMAR (-60) compared with EID std (-626). Hyperdense artifact attenuation was significantly reduced ( P < 0.01) in PCD VMI120 (HU 103) and PCD VMI120+IMAR (-1) compared with EID std (577). Qualitative artifact extent was rated significantly less severe in PCD iMAR and PCD VMI120+iMAR images compared with EID std ( P ≤ 0.01), while trabecular and cortical bone visibility was rated lower in PCD iMAR (2.73) and PCD VMI120 (2.63), P < 0.01, and lowest in PCD VMI120+iMAR (1.82) compared with EID std (3.19), P < 0.01. Significantly ( P < 0.01) more silhouette artifacts were present in the 100 kVp subgroup compared with the 140 kVp subgroup in PCD VMI120 and PCD VMI120+iMAR , and significantly ( P < 0.01) more dark zone artifacts in the PCD VMI120 series at 100 kVp compared with 140 kVp. Conclusion: PCD CT with VMI 120 or iMAR achieved artifact reduction comparable to EID CT for lumbar spine metal artifact reduction, at nearly half the radiation dose. Combining VMI120 + iMAR provided the strongest qualitative and quantitative artifact reduction, but at the cost of the lowest cortical and trabecular bone visibility. 140 kVp acquisitions should be favored over 100 kVp acquisitions in PCD to reduce the occurrence of “dark zone” and “silhouette artifacts”, suggesting a fixed tube voltage instead of an automatic tube voltage setting in instrumented lumbar spine imaging. VMI reconstruction of tin-filtered lumbar spine CT effectively reduces metal artifacts, with PCD favored over EID for its inherent spectral capabilities and the potential for substantial dose reduction.