Purpose: Metal artifacts are a major concern in computed tomography (CT) imaging with orthopaedic implants. Our intention was to compare already established methods (like raising kV, dual-energy computed tomography [DECT]) with newer methods (iterative metal artifact reduction [IMAR], tin filter) and a combination of both. Our aim was to assess which technique provides the best image quality possible while reducing dose at the same time.
Methods and Materials: Four human cadavers with different orthopaedic implants were tested, using full-dose (FD; computed tomography dose index [CDTI]: 10 mGy) and low dose (LD; CDTI: 3.3 mGy) scan protocols. Besides conventional CT we used the tin-filter technique, DECT, and software-based (iMAR) methods, using a third-generation DECT scanner. Evaluation was done with a six-part Likert scale for objective (e.g., contrast-to-noise ratio) and subjective parameters.
Results: In all four cadavers, FD tin-filter scans showed the best overall results with an average of 1.3 (FD 150 kV protocol: 1.72; LD 150 kV tin-filter protocol: 2.3; FD dual-energy protocol: 2.63). Looking only at metal artifact reduction, the best results were obtained using the DECT technique (FD as well as LD). However, DECT images suffered from high imaging noise that reduced their overall rating. IMAR software improved image quality in most cases, especially in standard protocols with 120 kV. In some case further artifacts were induced by using these algorithms.
Conclusion: Although DECT showed the best metal artifact reduction, it suffered from image noise that reduced the overall accessibility of fine trabecular bone structures. Tin-filter technology proved to be superior providing good image quality of the adjacent bony structures near orthopaedic implants. Using a LD 150-kV tin-filter scan significantly reduced dose by two thirds compared with a normal scan while still providing sufficient image quality and good metal artifact reduction. Image quality was improved in most cases by using IMAR, especially in standard scans with 120 kV. However, it sometimes induced further artifacts in other methods, particularly in tin-filter scans.
