Evaluation of different MRI-based anatomical priors for PET brain imaging

Abstract

Image reconstruction in emission tomography may benefit from the use of anatomical side information obtained with other imaging modalities in the same subject. One way to implement this, is to use the anatomical image for defining the a-priori distribution in a maximum-a-posteriori reconstruction algorithm. In this contribution, we use the PET-SORTEO Monte Carlo simulator to evaluate three different anatomical priors for PET brain imaging, using MRI for the anatomical image. The priors are: 1) a prior based on a segmentation of the MRI image; 2) the joint entropy prior; 3) a prior (proposed by Bowsher et al.) that encourages smoothness within a position dependent neighborhood, computed from the MRI image. The two latter priors do not rely on an explicit segmentation, which makes them more generally applicable than a segmentation-based prior. The three priors produced a compromise between noise and bias that was significantly better than that obtained with post-smoothed MLEM. The performance of the joint entropy prior was slightly worse than that of the other two priors. In contrast to the joint entropy prior, the Bowsher prior is easily tuned and does not pose convergence problems due to local maxima.