This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright. In most cases, these works may not be reposted without the explicit permission of the copyright holder.

Search for Publication

Year(s) from:  to 
Keywords (separated by spaces):

Better Anatomical Priors for PET Reconstruction

J.Oakley, J.Missimer and G.Székely
International Meeting on Medical Image Understanding and Applications


Bayesian methods are commonly used to regularise Maximum Likelihood (ML) based Emission Tomography reconstruction algorithms. The prior models employed are either dependent upon some assumed form of activity distribution, on an assumed magnitude of the activity across well defined regions of interest, or both. Assuming approximate levels of the activity levels requires firstly a delineation of the tissued regions, and secondly sensible starting estimates for the level of activity present in each of these compartments. Errors in either of these steps yields an incorrect prior model and a reconstruction solution that reflects this. In order to compensate for this possible weakness in the Bayesian reconstruction process, the following work proposes developing priors based on the emission data itself, constrained according to a more realistic forward model of the activity distribution. The influence of inaccuracies in the prior estimation of compartment activity levels is subsequently removed, the so-called resolution mis-match is avoided, and uncertainties in the segmentation are compensated for. As comparative tests described in this short paper show, bias toward the prior - an oft cited criticism of Bayesian methods - may yet be warranted.

Download in postscript format
  author = {J.Oakley and J.Missimer and G.Székely},
  title = {Better Anatomical Priors for PET Reconstruction},
  booktitle = {International Meeting on Medical Image Understanding and Applications},
  year = {1999},
  keywords = {PET Reconstruction, Priors, Forward Model, Bayesian, Maximum Likelihood Estimation}