Variational Bayesian inference for quantification of brain PET data at the voxel level

PET is a functional nuclear medicine imaging technique widely used to study in vivo physiological processes in the body, which are targeted by an appropriate tracer, as those used in this study: [11C](R)-rolipram, [11C]WAY100635, [11C]PBR28. The purpose of this work is to evaluate, at the voxel level, the performance of a Bayesian method never used before in PET domain: the Variationa Bayes method. In this study analysis on both simulated and real data are presented. VB is compared with WNLLS