Iterative Kinetic Parameter Estimation within Fully 4D PET Image Reconstruction

4D PET imaging seeks to estimate kinetic parameters of physiological significance through the generation of a time series of 3D images. Conventionally the time series is reconstructed one frame at a time, and then the kinetic modeling is applied as a post-reconstruction step to estimate the desired parameters. Such a separated approach does not account for the task of kinetic parameter estimation within the reconstruction itself. This work indicates that conventional frame-by-frame maximum likelihood reconstruction in high noise situations is sub-optimal if post-reconstruction kinetic parameter estimation is to be performed. As an alternative, a simple to implement, EM-based iterative reconstruction method is proposed which uses all of the acquired data in every iteration and includes the image-space kinetic parameter estimation process within the reconstruction. The method can accommodate kinetic models of any chosen complexity with relative ease, and can deliver more accurate kinetic parameter estimates than the conventional approach for low-statistics data.