• DocumentCode
    1097130
  • Title

    Development of a 3D gradient-based method for volume quantitation in SPECT

  • Author

    Long, David T. ; King, Michael A. ; Gennert, Michael A.

  • Author_Institution
    Dept. of Nucl. Med., Massachusetts Univ. Med. Center, Worcester, MA, USA
  • Volume
    38
  • Issue
    2
  • fYear
    1991
  • fDate
    4/1/1991 12:00:00 AM
  • Firstpage
    748
  • Lastpage
    754
  • Abstract
    A three-dimensional (3-D) method for quantifying organ and lesion volume in SPECT (single photon emission computed tomography) is described. A 3-D edge-detection technique based on a 3-D gradient operator is used to define the boundary of a radionuclide distribution in typical low-contrast, high-noise SPECT images. The method combines pre- and post-reconstruction filtering of the SPECT images with a 3-D adaptation of interpolative background subtraction to provide adequate noise reduction for reliable boundary definition. Simulation studies have shown that the method provides a direct measurement of relatively small volumes in SPECT images with good reproducibility and minimal operator bias. The method is applicable to quantifying the volume of hot and cold lesions
  • Keywords
    computerised tomography; picture processing; radioisotope scanning and imaging; volume measurement; 3D edge-detection technique; 3D gradient-based method; SPECT; cold lesions; hot lesions; image filtering; interpolative background subtraction; lesion volume; medical diagnostic imaging; nuclear medicine; operator bias; organ volume; radionuclide distribution; reliable boundary definition; single photon emission computed tomography; volume quantitation; Adaptive filters; Biomedical imaging; Computer science; Filtering; Image edge detection; Image resolution; Image segmentation; Lesions; Noise reduction; Nuclear medicine;
  • fLanguage
    English
  • Journal_Title
    Nuclear Science, IEEE Transactions on
  • Publisher
    ieee
  • ISSN
    0018-9499
  • Type

    jour

  • DOI
    10.1109/23.289384
  • Filename
    289384