Bayesian image reconstruction using image-modeling Gibbs priors

Abstract

We demonstrate that (a) classical methods of image reconstruction from projections can be improved upon by considering the output of such a method as a distorted version of the original image and applying a Bayesian approach to estimate from it the original image, and (b) by selecting an “image-modeling” prior distribution (one from which random samples are likely to share important characteristics of the images of the application area), one can do better than using some other Gibbs priors. Our demonstration is from the area of positron emission tomography of the brain. We present some encouraging results obtained using both simulated and real data. © 1998 John Wiley & Sons, Inc. Int J Imaging Syst Technol, 9, 85–98, 1998