Performance comparison of smoothing and gamma priors for transmission tomography

The authors introduced (T. Hsiao et al., “Joint-MAP Reconstruction/Segmentation for Transmission Tomography Using Mixture-Models as Priors”, Proc. IEEE Nuclear Science Symposium and Medical Imaging Conference, II, p. 1689-93, 1998) a Bayesian method for transmission tomography that used a novel pointwise prior in the form of a mixture of gamma distributions. Here, the authors compare the performance their pointwise prior with that of a smoothing prior in the context of a weak tumor detection task for emission PET. The reprojected attenuation maps from each method are used for attenuation correction of 2D PET emission data. The authors´ focus is on the problem of low counts in the transmission scan