• DocumentCode
    2198483
  • Title

    Comparison of planar vs. SPECT images for lesion detectability

  • Author

    Sorensen, Eric S. ; Zeng, Gengsheng L. ; Holt, B. ; Scott, Ken ; Karen, Scott ; Ovard, Todd

  • Volume
    3
  • fYear
    2002
  • fDate
    10-16 Nov. 2002
  • Firstpage
    1731
  • Abstract
    This research used computer simulation along with human observer studies to evaluate SPECT and planar imaging for lesion detection. The research initially focused on a simple computer simulated phantom consisting of a large sphere with a lesion placed inside at various depths. A constant lesion to background ratio of 8 to 1 was maintained throughout the trials. Two noise levels were simulated by varying the mean value of the sphere and lesion. Both planar and SPECT images were generated simulating a clinical scan time of 30 minutes. Results showed SPECT images consistently had higher contrast but also higher noise values than planar images. When the lesion was located near the center of the sphere the contrast-to-noise ratio was consistently higher in SPECT images. Research using a realistic computer model of the human torso along with realistic image acquisition and reconstruction techniques was then carried out. Planar images were generated with modeling attenuation, scatter and geometric point response. SPECT images were projected and then reconstructed with an OSEM algorithm accounting for attenuation, the scatter and the geometric point response of the collimator.
  • Keywords
    medical image processing; phantoms; single photon emission computed tomography; 30 min; OSEM algorithm; SPECT images; computer simulation; human observer studies; image acquisition; lesion detectability; planar images; simple computer simulated phantom; Attenuation; Computational modeling; Computer simulation; Humans; Image generation; Image reconstruction; Imaging phantoms; Lesions; Noise level; Scattering;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Nuclear Science Symposium Conference Record, 2002 IEEE
  • Print_ISBN
    0-7803-7636-6
  • Type

    conf

  • DOI
    10.1109/NSSMIC.2002.1239657
  • Filename
    1239657