Metal Artifact Redcution in CT for Surgical Screws: Comparison Between Two Commercially Available Software Packages
R Al-Senan*, M Yester, UAB Medical Center, Birmingham, ALSU-E-I-21 Sunday 3:00PM - 6:00PM Room: Exhibit Hall
Purpose: To compare the performance between two available metal artifact reduction software (MAR) packages from different CT vendors to determine applicability for surgical screws, even though such use is not supported.
Methods: Two vertebral screws (Al and stainless steel) were placed in the center of a water phantom (20 cm x 15 cm), with long axis perpendicular to the scan direction. Scans were performed using a GE Discovery HD750 with Dual Energy (DE) and MARS software, and Philips iCT with iDose4 and O-MAR software. Reconstruction was performed with and without MAR. Assessment was performed qualitatively and quantitatively by measuring the SD in areas surrounding the screws. For the GE scanner, mono-energy reconstructions were performed for 70, 127, and 140 keV in addition to the MARS routine. Thickness of the screws was also measured and compared to the expected size.
Results: DE reduced streak artifacts and MARS further reduced artifacts for both screws. However, a portion of the middle was distorted. SD of region next to the screw tips was reduced up to 97%; whereas SD in the region under the long axis increased in some MARS images by more than 50%. O-MAR reduced streak artifacts but less effectively than MARS but with no loss of the middle section. With O-MAR, in general most images showed reduced SD in both regions by 2% to 54%. Additionally, width was significantly reduced (20% to 50%) by MARS. This effect was not noted with O-MAR.
Conclusion: Performance of MARS and O-MAR differed considerably. MARS effectively reduced streaks along the long axis but introduced distortions in regions near the center of the screw. With O-MAR, streaks were reduced by a lesser degree than MARS along the screws axis, but by higher degree above and below the screws with no distortion at the center.