Weighted schemes applied to 3D-OSEM reconstruction in PET

In order to better preserve the Poisson characteristics of PET data when using 3D OSEM iterative reconstruction, the authors have compared images reconstructed by various data weighting schemes: unweighted, attenuation weighted, attenuation and normalization weighted OSEM 3D, FORE+unweighted and attenuation weighted OSEM 2D using 3DRP and FORE+FBP images as reference. Due to better noise modeling, 3D weighted OSEM schemes are expected to produce better images (i.e. noise-bias trade-off) especially at low statistics, in short frames of dynamic studies and when bed overlap is suboptimal, as in 3D wholebody studies. These reconstruction methods were implemented on the ECAT Exact HR and quantification was validated on clinical data. Their influence on physiological parameter measurements was examined on a dynamic brain receptor study with ¹¹C-Flumazenil and a whole-body study with ¹⁸FDG. The 3D weighted OSEM schemes provide unbiased results in volume of interests. The resulting noise reduction may lead to more accurate registration, acquisition of shorter frames and generation of parametric images