Abstract
241744
Introduction: The NEMA/IEC PET image quality phantom has served as a valuable tool not only for providing system specifications but for acceptance testing, quality control, algorithm evaluation, site qualification and other tasks. The phantom has internal spheres ranging from 10 mm to 37 mm internal diameter. The continuous improvement in PET hardware and reconstruction algorithms has led to better spatial resolution, sensitivity, and time-of-flight specifications, all driving the need for a modified design with smaller internal spheres. A new prototype concept phantom adapting from the original NEMA/IEC design, has been designed, built, and tested on nine different PET/CT systems.
Methods: The new test phantom includes 6 mm and 8 mm diameter spheres in the central ring (radius = 57 mm) in addition to the larger established sizes 10 mm, 13 mm, 17 mm, and 22 mm. The 6 and 8 mm spheres were constructed with 0.5 mm wall thickness (vs. 1.0 mm for the larger spheres). Sampling a 6 mm sphere in a 3 mm slice results in a range of 27%, depending on how the sphere center is located relative to a slice (compared to a 12% range for 10 mm sphere). For this reason, and for the addition of statistical power in the challenging case of detecting and quantitating very small spheres in images with typical noise, three additional 6 mm and three additional 8 mm spheres were added at a 114 mm radius. Positioning these spheres at a larger radius also challenges PET systems due to worsening intrinsic spatial resolution at larger radius. The axial location of these spheres increments by 1 mm, with the intention of sampling over different locations within the slice plane.The phantom with F-18 solution with a sphere-to-background ratio of 4 and was scanned for 30 min, allowing a single high-count image set as well as an ensemble of ten 3 min images to assess image quality and reproducibility at more typical noise levels. Images were reconstructed with typical clinical parameters.
Results: In an initial qualitative evaluation, the 8mm and 6 mm spheres were discernable in 30 min images on all systems (Figure 1). The 3 min images from all systems revealed the 8 mm spheres clearly and most revealed the 6 mm (Figure 2). The relative intensity of the four 8 mm spheres was variable, as were the four 6 mm (when seen) justifying a multi-sphere approach.
Conclusions: A new PET image quality phantom has been designed and tested to challenge current and future PET systems. Based on the images obtained, it is warranted to have smaller (8 mm and perhaps even 6 mm) spheres for a phantom to be relevant. It may be necessary to have multiple spheres of the smaller sizes and/or multiple scan realizations to make image analysis meaningful.