Inclusion of a priori information in frequency space for quantitative SPECT imaging

A Wiener filter has been designed for noise suppression in SPECT. To implement this filter, the authors used a square-root transform and designed a butterfly window and a weighting window to improve the estimation of the power spectra of the noise, the measurement, and the signal. These two windows were derived based on the frequency distribution of noise-free sinograms. Simulations and experiments were performed using the Hoffman brain phantom. The results show the advantage over the traditional implementation of low pass filter.