Improved Clinical Parametric Imaging Using List-Mode Reconstruction via Resolution Modeling

Image reconstruction with comprehensive modeling of the acquisition system response needs to be devised for clinical imaging to fully benefit from the new generation of high resolution tomographs. An iterative list-mode OP-OSEM algorithm including an experimental stationary spatial resolution model (RM-OP-OSEM) of a high resolution tomograph has previously been developed [Reader et al., 2005]. In this work it is compared to a classical OP-OSEM algorithm (OP-OSEM). We investigated the performance of these two algorithms on image quality and on the quantification of regions of interests. We also assessed their impact on the quantification of biological parameters in a clinical study of the dopamine transporter. Realizations from an experimental phantom study were used to investigate contrast and noise characteristics of the images. The binding potential of a selective tracer of the dopamine transporter was assessed in anatomical regions of interest in a clinical study. In the phantom experiment, we observe a slower convergence of the RM-OP-OSEM algorithm characterized by a higher contrast recovery for the same level of statistical noise. Altogether, RM-OP-OSEM allows the recovery of contrast levels that cannot be reached without resolution modeling (RM). Visual inspection of the images indicates better recovery of the smallest spheres and better delineation of the structures. Statistical noise with RM is characterized by much lower variance at the voxel level. In a homogeneous region, higher positive correlation and lower negative correlations with neighbours voxels are also observed, which leads to lower spatial variance. The clinical images reconstructed with RM shows better delineation of the cortical and sub-cortical structures in both the time averaged and parametric images. The binding potential in the striatum is also increased, similar to what observed in the phantom study, suggesting improved quantification due to lowered partial volume effects (PVEs).