Accurate Event-Driven Motion Compensation Incorporating All Detected Events

This work develops and investigates a formalism for accurate motion-compensated reconstruction, including elaborate consideration of scattered and random coincidences, which at the same time is particularly feasible in the context of high-resolution PET. The method takes into consideration normally-detected projection data which are not detected due to motion. Furthermore, it incorporates information from all detected events, particularly those which, following correction for motion, fall outside the FoV (e.g. axially or through detector gaps), thus satisfying a mathematical requirement, elaborated in the text, that would allow accurate motion averaging of sensitivity factors in image-space (as opposed to projection-space). The proposed method has been extensively validated using phantom experiments as well as realistic simulations of a new mathematical brain phantom developed in this work.