Self-normalization of emission data in 3D PET

We demonstrate that dead-time mismatch between the normalization and emissi on acquisitions in three-dimensional positron emission tomography (3D PET) can introduce high-frequency circular image artifacts. We describe a simple correction scheme for this problem where appropriate correction factors ar e calculated from the emission data itself, thus avoiding the necessity of acquiring normalization data at a range of different count-rates. In this s cheme "pseudo detector efficiencies" are calculated from the emission data by application of a conventional normalization algorithm. The mean systemat ic variation in these pseudo-efficiencies with position within the block-de tector is then calculated, and appropriate correction factors applied to th e original emission data. We show that this "self-normalization'' scheme can remove normalization art ifacts in both uniform and nonuniform phantom studies, and that it is also effective in vivo.