Click here to


Are you sure ?

Yes, do it No, cancel

Universal Orbits for Metal Artifact Elimination

G Gang1, T Russ2, Y Ma3, C Toennes4, L Schad5, C Weiss6, T Ehtiati7, J Siewerdsen8, J Stayman9*, (1) Johns Hopkins University, Baltimore, MD, (2) University of Heidelberg, ,,DE, (3) Johns Hopkins University, ,,(4) University Of Heidelberg, ,,DE, (5) University of Heidelberg, ,,DE, (6) Johns Hopkins University, Baltimore, ,(7) Siemens Medical Solutions USA, Baltimore, ,(8) Johns Hopkins University, Baltimore, MD, (9) Johns Hopkins University, Baltimore, MD


(Sunday, 7/12/2020)   [Eastern Time (GMT-4)]

Room: AAPM ePoster Library

Purpose: Metal artifacts are a persistent cone-beam CT problem with artifact reduction schemes predominantly relying algorithms to fix the data. However, metal-occluded anatomical details cannot generally be recovered using such approaches. X-rays passing through metal can exhibit severe photon starvation and can effectively yield missing data. Using a data completeness quality metric, we seek to find novel, non-circular source-detector trajectories that permit collection of complete data, regardless of the location and shape of the metal. Such a universal orbit would have distinct advantage over traditional trajectories by effectively eliminating missing data and metal artifacts.

Methods: We implemented a numerical of Tuy’s condition for local data completeness as the total percentage of Radon planes intersecting the source trajectory at a point. We accounted for missing data attributed to metal by discounting any intersections that pass through metal. Various non-circular trajectories were optimized according to this metric. Orbits were tested in a physical phantom with background clutter and two sets of metal objects: 1) three metal balls each surrounded by 3D printed radial line pairs, and 2) spine fusion hardware including pedicle screws. The phantoms were imaged on a custom cone-beam CT test bench and a commercial robotic C-arm.

Results: We found that many non-circular orbits that can achieve similar levels of completeness. While simple sinusoidal orbits perform well, larger source-detector tilts yield better performance, and specific sinusoidal frequencies are more optimal – particularly when this leads to increased data redundancy. We observe highly reduced metal streaks in both scenarios (even without explicit metal artifact correction).

Conclusion: We have demonstrated the potential a universal orbit to be robust to metal throughout the CT field-of-view. The approach solves the metal artifact problem, not through algorithmic processing of bad data, but instead changes the data acquisition to avoid bad data in the first place.

Funding Support, Disclosures, and Conflict of Interest: Funding from NIH, Siemens Healthineers, Fischer Imaging, Canon Medical Research Partnerships with GE Healthcare, Philips, United Imaging


Not Applicable / None Entered.


Not Applicable / None Entered.

Contact Email