• DocumentCode
    2637453
  • Title

    Lesion quantification in dual-modality mammography using expectation maximization with attenuation correction

  • Author

    Li, Heng ; Zheng, Yibin ; More, Mitali J. ; Goodale, Patricia J. ; Williams, Mark B.

  • Author_Institution
    Dept. of Electr. & Comput. Eng., Virginia Univ., Charlottesville, VA, USA
  • fYear
    2004
  • fDate
    15-18 April 2004
  • Firstpage
    1392
  • Abstract
    Conventional X-ray/scintillation mammography does not provide the depth information necessary to quantify the location and concentration of lesion activities in breast tissue. Our group is developing a novel tomographic dual modality system that will provide 3D structural and functional information of the breast. Because of constraints on the gantry geometry and acquisition time, only a limited number of views on one side of the breast can be acquired. In addition, the scatter and attenuation of gamma rays by breast tissue is a major problem for the reconstruction. We present an expectation-maximization (EM) method which naturally takes into account all the factors of noise, limited views, scatter, and attenuation. In a preliminary phantom study our quantification technique obtained results agreeing well with the ground truth.
  • Keywords
    cancer; gamma-ray scattering; image reconstruction; mammography; medical image processing; phantoms; statistical analysis; tumours; 3D structural information; EM; X-ray mammography; attenuation correction; breast tissue; expectation maximization; gamma ray; gantry geometry; lesion quantification; phantom study; scintillation mammography; three dimensional functional information; tomographic dual modality system; Attenuation; Breast tissue; Gamma ray detectors; Image reconstruction; Image resolution; Lesions; Mammography; Statistics; X-ray imaging; X-ray scattering;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Biomedical Imaging: Nano to Macro, 2004. IEEE International Symposium on
  • Print_ISBN
    0-7803-8388-5
  • Type

    conf

  • DOI
    10.1109/ISBI.2004.1398807
  • Filename
    1398807