Dynamic multi-bed FDG PET imaging: Feasibility and optimization

Multi-Bed FDG PET/CT as applied to oncologic imaging is currently widely and routinely used for assessment of localized and metastatic disease involvement. In the past, based on conventional (single-bed) dynamic PET imaging, standard tracer kinetic modeling techniques have been developed to estimate the FDG uptake rate K_i. However, routine clinical multi-bed FDG PET imaging commonly involves a single time frame per bed, i.e. static imaging, and the standardized uptake value (SUV), a surrogate of metabolic activity, is employed to estimate the uptake rate K_i. The accuracy depends on two conditions: (i) in the voxel or region of interest, contribution of non-phosphorylated FDG is negligible relative to phosphorylated FDG, and (ii) time integral of plasma FDG concentration is proportional to injected dose divided by lean body mass, which can fail in clinical FDG PET imaging and pose problems in differentiating malignant from benign tumors. The objective of the proposed work is to facilitate, for the fist time, a transition from static to dynamic multi-bed FDG PET/CT imaging in clinically feasible times where, given the challenge of sparse temporal sampling at each bed, novel dynamic acquisition schemes should be employed to yield quantitative whole-body imaging of FDG uptake. Thus, a set of novel dynamic multi-bed PET image acquisition schemes have been modeled, using Monte Carlo simulations, to quantitatively evaluate the clinical feasibility of the method and optimize the number of passes per bed and the total study duration. It has been determined that a data acquisition scheme consisting of 6 whole-body passes and constant time frames of 45sec produces parametric images with the optimal noise vs. bias performance. Finally, clinical whole-body patient data have been acquired dynamically and results demonstrate the potential of the proposed method in enhancing treatment response monitoring capabilities of clinical PET studies.