5-year progress on imaging performance of commercial PET scanners: a multi-center study towards harmonization of FDG-PET

Objectives: Whole-body FDG-PET plays key roles in detection, staging, and characterization of malignant tumors. In oncologic clinical studies and trials, a standardized uptake value (SUV) is often used as a quantitative biomarker although it is considerably affected by many factors such as imaging protocols, PET scanner specifications, and reconstruction parameters. On the other hand, clinical PET scanner performance has rapidly progressed by novel technologies such as point-spread function, time-of-flight, and penalized likelihood reconstruction. To establish a feasible and practical methodology for harmonization of FDG-PET, we are investigating quantitative physical performance of recent clinical PET scanners with the image quality phantom. Here, we compared the recent results with the phantom data that had been acquired about 5 years ago with the PET scanners of those days*.

Methods: A total of 19 types of recent commercial PET/CT scanners used in a total of 16 centers were tested in this study. Of these, 15 scanners had TOF measurement capability. A NEMA image quality phantom having six spheres (10-37 mm in diameter) was used for image quality evaluation. The sphere-to-background radioactivity ratio was 4. The background radioactivity concentration was 2.53-2.65 kBq/mL. PET images with various scanning duration (30-300 sec, 1800 sec) were reconstructed using a clinical reconstruction setting in each site. For visual assessment, the PET images were evaluated according to the detectability of the 10-mm hot sphere on a 3-step scale (0, not visualized; 1, visualized, but similar hot spots are observed elsewhere; 2, identifiable) by five observers. We measured the percent contrast of the 10-mm hot sphere (Q_H,10mm), the background variability (N_10mm), the coefficient of variation in the background area (CV_BG), and the relative recovery coefficient (RC). The data were compared with the previous data published in 2014* that had been acquired with a total of 13 PET scanners released before 2011.

Results: On the PET images with 120 sec acquisition, the median value for all scanners was 2.0 for visual detectability score, 28.7% for Q_H,10mm, 7.6% for N_10mm, and 9.6% for CV_BG. Those of the previous data was 0.8 for visual detectability score, 13.7% for Q_H,10mm, 8.3% for N_10mm, and 12.3% for CV_BG*. The relative RCs for the spheres with diameters of 22 mm or smaller were higher than the previous data. Overall, the visual detectability of the 10-mm sphere, contrast, image noise, and RC were clearly better than the previous data.

Conclusions: We have revealed that imaging performance of commercial PET scanners has made remarkable progress in the past 5 years. For practical harmonization of FDG-PET, regular updates of standard phantom test methodology and harmonizing criteria are essential to adapt to new PET scanners with better imaging performance. (*Fukukita H, et al. Ann Nucl Med. 2014;28:693-705)