iMAR dramatically reduces CT artefacts for hip prostheses and increases accuracy of SUVs when used for PET attenuation correction with low-dose CT

Objectives: The benefits of iterative Metal Artifact Reduction (iMAR) are well established for diagnostic CT. The aim of this study was to investigate the effectiveness of iMAR in reducing metal artifacts on low-dose CT of hip prosthesis and improving accuracy of SUV when used for CT-based attenuation-correction (AC) for PET. Methods: A hip prosthesis was placed in a cylindrical uniformity phantom (9.293kg) filled with a homogenous distribution of water and F-18-FDG solution. A one bed-position PET-CT study was acquired on a Siemens Biograph mCT Flow 64, using low-dose localization CT scans with 120kV and a CARE Dose 4D reference of 50mAs (effective). B30 CT reconstructions were made with and without iMAR, and used for AC of PET TOF reconstructions. 4 spherical VOIs (diameter 30mm) were placed in the water-FDG solution around the prosthesis, on CT and PET-AC datasets, and CT HU statistics and PET SUV statistics compared between iMAR and non-iMAR reconstructions. Control scans of the phantom were also performed without prostheses, to determine reference CT HUs and SUV. Results: CT: The mean CT HU for the 4 VOIs in the reference scan (without prosthesis) ranged from -1 to 0HU, and the VOI SD ranged from 22 to 26HU. In the presence of a hip prosthesis, the mean CT HU for the VOIs in the non-iMAR reconstructions ranged from -209 to +106HU, and the SD ranged from 44 to 90HU, with clearly visible metal artifacts that were most severe at the head-neck junction. For the iMAR reconstructions, the mean CT HU for the 4 VOIs ranged from -8 (96% improvement) to +21 (80% improvement), the SD ranged from 27 to 31HU, and the visual appearance showed that the artifacts were dramatically reduced. PET: The mean SUV for the 4 VOIs in the reference scan (without prosthesis) ranged from -2% to +2%. In the presence of a hip prosthesis, the mean SUV for AC PET reconstructions made without iMAR ranged from -17% to +4% with one area of apparently reduced uptake corresponding to the head-neck junction. The mean SUV for AC PET reconstructions made with iMAR ranged from -6% to +1%, and the artifact was no longer visible. Conclusion: For low-dose CT reconstructions, iMAR dramatically reduced CT artifacts for hip prostheses compared with non-iMAR CT reconstructions, and when used for AC PET reconstruction, iMAR reconstructions improved the accuracy of PET SUVs and the artifacts were no longer visually apparent.