Filtering of SPECT reconstructions made using Bellini's attenuation correction method: a comparison of three pre-reconstruction filters and a post-reconstruction Wiener filter

The authors evaluate a three-dimensional (3-D) Wiener filter which is used to restore SPECT (single photon emission computed tomography) reconstructions made using a method of attenuation correction described by S. Bellini et al. Its performance is compared to that of several preconstruction filters: (1) the one-dimensional (1-D) Butterworth, (2) the two-dimensional (2-D) Butterworth, and (3) a 2-D Wiener filter. An approximation to a clinical liver spleen study was used as the source distribution, and an algorithm which accounts for the depth- and distance-dependent blurring in SPECT was used to compute noise-free projections. To study the effect of filtering method on tumor detection accuracy, a 2-cm-diameter, cool spherical tumor (40% contrast) was placed at a known, but random, location within the liver. Projection sets for ten tumor locations were computed, and five noise realizations of each set were obtained by introducing Poisson noise.<
>