Optimizing reconstruction parameters for quantitative I-PET in the presence of therapeutic doses of I.

Background: The goal of this work was to determine the quantitative accuracy and optimal reconstruction parameters for I-PET imaging in the presence of therapeutic levels of I. In this effort, images were acquired on a GE D710 PET/CT scanner using a NEMA IEC phantom with spheres containing I and increasing amounts of I activity in the background. At each activity level, two scans were acquired, one with the phantom centered in the field of view (FOV) and one 11.2 cm off-center. Reconstructions used an ordered subset expectation maximization algorithm with up to 100 iterations of 16 subsets, with and without time-of-flight (TOF) information. Results were evaluated visually and by comparing the I activity relative to the scan performed in the absence of I.

Results: I within the FOV added to the randoms rate, to dead time, and to pile-up within the detectors. Using our standard clinical reconstruction parameters, the image quality and quantitative accuracy suffered at I activities above 1.4 GBq. Convergence rates slowed progressively in the presence of increasing amounts of I for both TOF and nonTOF reconstructions. TOF reconstructions converged more quickly than nonTOF but often towards erroneous concentrations. Iterating nonTOF reconstructions to convergence produced quantitatively accurate images except for the off-center phantom at the very highest level of background I tested.

Conclusions: This study shows that quantitative PET is feasible in the presence of large amounts of I. The high randoms fractions resulted in slow reconstruction convergence and negatively impacted TOF corrections and/or the accuracy of TOF information. Therefore, increased iterations and nonTOF reconstructions are recommended.