Variance Propagation For Spect With Energy-Weighted Acquisition

Compton scatter reduces image contrast and increases noise in single photon emission computed tomography (SPECT). A compensation approach, called energy-weighted acquisition (EWA) can reduce the effects of scatter in SPECT imaging. With EWA the detected events are not only weighted according to their energies, but they are also redistributed spatially. The design of specific weighting functions can be assisted by understanding how random noise is propagated through the EWA and filtered backprojection (FBP) algorithms. The purpose of this research has been to compare the noise characteristics of SPECT images which are reconstructed from data acquired by both EWA and normal windowed acquisition (NWA). The comparisons used are: 1) propagation of variance, and 2) measurement of the noise power spectrum (NPS). For a selected T1-201 weighting function, the analytically predicted variance (neglecting covariance terms) has been compared with the variance from experimental SPECT scans of a phantom. Reasonable agreement was observed between predicted and measured %rms noise. The NPS for EWA was skewed slightly with respect to the NWA spectrum. Further research is justified to determine the clinical significance of these noise characteristics. © 1991 IEEE

Duke Scholars

Author Ronald J. Jaszczak Radiology