The goal of this study was to evaluate the effect on the average standardized uptake value (avgSUV) and maximum standardized uptake value (maxSUV) of changing the number of iterations in the reconstruction process on studies acquired with PET/CT.
Methods: Data from 50 human tumors were acquired on a PET/CT scanner, using the CT portion for attenuation correction. Reconstruction was performed using the 2-dimensional reconstruction method of ordered-subsets expectation maximization (OSEM) with 28 subsets and with 1, 2, 3, 4, 5, 10, 20, and 40 iterations. The standardized uptake value (SUV) of the studies was analyzed by positioning a region of interest tightly around the tumor and reproducing the same area on all same-study iterations for SUV measurements.
Results: The differences in mean avgSUV and mean maxSUV were statistically different across different iteration groups. SUV data demonstrated that the avgSUV measurements have the most significant differences between 1 versus 2 iterations and 2 versus 3 iterations. The P values for these comparisons were less then 0.001. For maxSUV, all differences had P values less than 0.001. There also was a systematic increase in the SUVs as the number of iterations increased. The avgSUV increased at early iterations (less than 5), with just 50%-60% increasing after 5 iterations. However, maxSUV increased systematically at early iterations, and this trend continued as the number of iterations increased.
Conclusion: The OSEM algorithm converges sooner for avgSUV than for maxSUV. The likely reason is that avgSUV depends on low-frequency features that are recovered with fewer iterations. The differences in maxSUV were likely due to noise, which increased with the number of iterative updates, and to increased resolution and recovery of high-frequency features (i.e., tumor heterogeneity) with a larger number of iterations. Factors that determine the quantitative accuracy of iterative reconstruction may have played an additional role. Given the continued change in maxSUV with iterations, great care must be taken in selecting the number of iterative updates when using it to assess tumors and their response to chemotherapy and radiation therapy. Because 2-5 iterations with 8-28 subsets are being used in clinical settings, these data are pertinent when comparing the SUVs of a tumor before and after therapy.