MRI- and DTI-guided PET image reconstruction: Application to imaging with [11C]DPA-713

Objectives To investigate the use of anatomy-assisted PET image reconstruction using modalities of MRI and diffusion tensor imaging (DTI) as applied to parametric imaging for clinical research.

Methods PET images continue to be affected by noise and the partial volume effect (PVE). An approach to reduce simultaneously both effects is to incorporate anatomical information within the PET image reconstruction task. Our extensive simulations of Bayesian PET reconstructions incorporating an anato-functional joint entropy (JE) prior have revealed considerable promise, moving beyond conventional approaches that assume over-simplified correlations between anatomy and radiopharmaceutical uptake. We applied this technique to [11C]DPA-713 PET and MRI imaging studies of healthy control subjects (n=3). DTI imaging was additionally performed on each subject (as the study aims to also investigate changes in white matter tract in controls vs. schizophrenic patients). We hypothesized that utilization of MRI and/or DTI images within PET reconstruction could impact and enhance PET imaging. Eighteen regions-of-interest (ROIs) were considered. Parametric K1 and distribution volume (DV) images were generated using a standard 2-tissue 5-parameter compartmental model and a linear regression with spatial constraint (LRSM) algorithm, and were statistically compared.

Results In comparison to standard 3D-OSEM reconstructed images, MRI- and DTI-guided PET images depicted significant qualitative differences, demonstrating considerably sharper distinction between various regions. Ongoing quantitative analysis has revealed increases in nearly all regions of the parametric images for both DV (by 17%; P<0.02) and K1 (by 27%; P<0.001).

Conclusions Anatomy-guided image reconstruction can substantially impact quality and quantification of PET images, and have the potential to enhance parametric PET imaging significantly