Improved Bayesian image reconstruction in positron emission tomography

Abstract

We extend the Bayesian model (Johnson, Chen, et al., 1990, 1991), previously developed for image reconstruction and restoration, by the introduction of diagonal line sites and the use of symmetrical neighborhood configurations. A computer simulation study was performed to examine the effect of the hyperparameters, the diagonal line sites, and the size of the neighbourhood configuration on the performance of the Bayesian method. We show that for optimal performance, distinct hyperparameters should be used for the intensity sites and line sites. The results also suggested the use of a larger configuration. By comparing the near-optimal restored images, we also demonstrate that the use of diagonal line sites, along with the symmetrical configurations thus made possible, can effectively remove the blocky edge artifacts and produce images of better quality. For image reconstruction applications, in addition to the use of diagonal line sites and symmetrical neighborhood configurations, we suggested using the area-weighted projection/backprojection technique to improve the accuracy of the system response function. Quality of reconstructed images were shown to be improved for both computer simulated and real patient PET data.