Fully 3D PET image reconstruction for a spatially-varying system response and very low counting statistics.

A dedicated PET-system has been integrated into the heavy ion tumour facility at the Gesellschaft für Schwerionenforschung (GSI) in Darmstadt in order to monitor and control the applied dose distribution during the irradiation. The positron camera has strongly spatially-varying system behaviour due to its limited angle design and the presence of detector gaps. Furthermore, the expected counting statistics are limited by the doses applied per therapy fraction.
A three-dimensional Maximum-Likelihood Estimator algorithm has been adapted to this imaging situation. Corrections for activity outside of the field of view, parallax errors, randoms as well as normalization factors have been implemented. The algorithm has been validated using ²²Na reference sources. Resolution, edge detectability, geometrical fidelity of size and position have been chosen as figures of merit. Results of the performance studies as well as an example of the algorithm's application to patient data are presented.