Iterative algorithms for metal artifact reduction in children with orthopedic prostheses: preliminary results.

BACKGROUND: Increased computational power allows computed tomography (CT) software to process very advanced mathematical algorithms to generate better quality images at lower doses. One such algorithm, iterative metal artifact reduction (iMAR) has proven to decrease metal artifacts seen in CT images of adults with orthopedic implants.

OBJECTIVES: To evaluate artifact reduction capability of the algorithm in lower-dose pediatric CT compared to our routine third-generation advanced modeled iterative reconstruction (ADMIRE) algorithm.

MATERIALS AND METHODS: Thirteen children (11-17 years old) with metal implants underwent routine clinically indicated CT. Data sets were reconstructed with an iMAR algorithm. Hounsfield units and image noise were measured in bone, muscle and fat in the streak artifact (near the implant) and at the greatest distance from the artifact (far from the implant). A regression model compared the effects of the algorithm (standard ADMIRE vs. iMAR) near and far from the implant.

RESULTS: Near the implant, Hounsfield units with iMAR were significantly different in our standard ADMIRE vs. iMAR for bone, muscle and fat (P<0.001). Noise was significantly different in standard ADMIRE vs. iMAR in bone (P<0.003). Far from the implant, Hounsfield units and noise were not significantly different for ADMIRE vs. iMAR, for the three tissue types.

CONCLUSION: These preliminary results demonstrate that iMAR algorithms improves Hounsfield units near the implant and decreases image noise in bone in low-dose pediatric CT. It does this without changing baseline tissue density or noise far from the implant.