Total hip prosthesis metal-artifact suppression using iterative deblurring reconstruction.

Abstract

PURPOSE: CT of total joint prostheses is limited by metal artifact produced mainly by missing projection data. Iterative deblurring reconstruction is less sensitive to missing projection data than filtered backprojection. A software CT simulator was used to compare total hip prosthesis images reconstructed using standard filtered backprojection, filtered backprojection after linear interpolation of missing data, and iterative deblurring. METHOD: Unilateral and bilateral total hip replacements with metal-backed or all-polyethylene acetabular prostheses were simulated using bone, metal, and polyethylene annuli and circles of metal and water (soft tissue). Material attenuation properties were taken from the literature. The simulation assumed that no X-rays penetrated the metal. Simulated projection data were reconstructed using filtered backprojection, filtered backprojection after linear interpolation of missing data, and iterative deblurring. Visual observations and objective region-of-interest analyses were made. RESULTS: Even with no X-rays penetrating the metal, iterative deblurring produced almost no visible artifact within the bone or soft tissues. Bone edge detection and sizing were more easily and accurately done from the iterative deblurring images. All reconstruction techniques underestimated bone and water CT attenuation values. Metal artifact was worse for the bilateral or metal-backed prostheses. CONCLUSION: Iterative deblurring generated nearly metal-artifact-free images in this simulation. Filtered backprojection, even after linear interpolation, produced typical clinical metal-artifact images.