Statistical analysis of incomplete circular orbits for breast pinhole SPECT imaging
Breast SPECT imaging with pinhole collimation provides high resolution and sensitivity when the pinhole aperture is in close proximity to the breast. Sampling artifacts can be limited by more nearly satisfying the Tuy complete-sampling requirement. In this study we used clinical count density scans to analyze quantitatively the effects of three acquisition geometries. A Trionix clinical scanner was used with rotation range limited to 180 degrees to simulate the presence of a torso. An isolated breast phantom containing two spherical lesions was mounted on a tiltable platform attached to the patient bed. The orbits were: A) single half-circular orbit (Single HCO), B) three parallel half circles separated axially (Triple parallel HCO), and C) three half circles with the plane of rotation for two half circles tilted +30 and -30 degrees relative to the middle (Triple Tilted HCO). Multiple scans were taken for all orbits to perform statistical analysis. ML-EM was used for reconstruction. Our analysis shows that for the central lesion, orbits B and C yield 31% and 28% higher peak contrast than orbit A. For the side lesion, orbit C gives a 55% higher peak contrast than A and 70% higher than B. Signal-to-noise ratios are dependent on the lesion location. The contrast analysis suggests that triple half-circular orbits are superior to the single half-circular orbit, especially, when two of the circles are tilted so that a larger part of the breast volume stays in the field of view during acquisition. © 2004 IEEE.
