• DocumentCode
    686847
  • Title

    A novel image restoration method assisted by reference image in dual-energy CT

  • Author

    Yuanji Li ; Li Zhang ; Kejun Kang

  • Author_Institution
    Dept. of Eng. Phys., Tsinghua Univ., Beijing, China
  • fYear
    2013
  • fDate
    Oct. 27 2013-Nov. 2 2013
  • Firstpage
    1
  • Lastpage
    4
  • Abstract
    Because of its remarkable material discrimination ability, dual-energy CT (DECT) becomes more and more important in security inspection, medical diagnosis and other industrial areas. Four images can be reconstructed in one dual-energy scan, and the effective atomic number image suffers from decomposition imprecision, system noise and the look-up table errors, so its quality is always poorer than the attenuation images. But the four images should have the same structures. In this paper, we propose a novel image restoration method that uses the attenuation image as reference image to improve the quality of the atomic number image. The structure information/weighting is obtained using the bilateral filtering method in the attenuation image and then we restore the atomic number image. Using the method, the image quality of the effective atomic number is improved prominently and the artifacts can be suppressed obviously. Two experiments were conducted to prove that the method is effective in DECT image processing.
  • Keywords
    computerised tomography; filtering theory; image reconstruction; image restoration; medical image processing; DECT image processing; attenuation image; bilateral filtering method; decomposition imprecision; dual-energy CT; effective atomic number image; image restoration method; look-up table errors; material discrimination ability; medical diagnosis; reference image quality; security inspection; system noise; Attenuation; Computed tomography; Filtering; Image reconstruction; Image restoration; Materials; X-ray imaging;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Nuclear Science Symposium and Medical Imaging Conference (NSS/MIC), 2013 IEEE
  • Conference_Location
    Seoul
  • Print_ISBN
    978-1-4799-0533-1
  • Type

    conf

  • DOI
    10.1109/NSSMIC.2013.6829279
  • Filename
    6829279