An investigation on analytical methods for correction of distance-dependent resolution variation in 3D SPECT imaging

Spatial resolution variation as a function of distance from collimator surface in single photon emission computed tomography (SPECT) is a major obstacle for quantitative imaging. This work investigated two analytical inversion methods for correcting the distance-dependent resolution variation. The first one derives accurately an inversion formula, but approximates the resolution variation kernel. The second one considers accurately the resolution variation kernel, but approximates the inversion formula. Reconstructed images using the first method showed a better resolution recovery at the central area of field-of-view (FOV), consistent with the approach that the resolution kernel is approximated for near-field regions. The second method restored resolution better at the FOV periphery, consistent with the theory that the formula is approximated for far-field regions. The first method is very sensitive to the approximation. The second one is robust and, therefore, is a better choice for quantitative SPECT imaging.