Simultaneous reconstruction of activity distribution and attenuation/scatter distribution in myocardial SPECT

To get good reconstruction from the SPECT data heavily corrupted by noise, attenuation, scattering and blurring, it is essential to find the medium attenuation/scatter distribution of the study. In myocardial SPECT, this problem is more severe because of the complex structure of the chest. The authors present a Bayesian approach which estimate both the activity distribution and the medium distribution simultaneously. Their method consists of three models for the different aspects of SPECT. It is concluded that the deformable template used to model the medium distribution, together with the Markov random field, used to model the activity distribution, provides a way to impose global and local regularizations on the highly ill-posed problem of reconstructing activity distribution from the projection data alone. The Bayesian approach yields good results for reconstructions in myocardial SPECT.