TY - JOUR T1 - A more clinically relevant assessment of PET spatial resolution. JF - Journal of Nuclear Medicine JO - J Nucl Med SP - 175 LP - 175 VL - 62 IS - supplement 1 AU - Madelyn Zimmer AU - Scott Leonard AU - Michelle Gruchot AU - Gary Dillehay Y1 - 2021/05/01 UR - http://jnm.snmjournals.org/content/62/supplement_1/175.abstract N2 - 175Objectives: Spatial resolution testing of PET scanners, per the National Electric Manufacturers Association (NEMA) protocol, is performed using a very specific method. A 1 mm or smaller point source is imaged in air (with no attenuation or scatter) and reconstructed by filtered back projection (FBP) in a sub-clinical pixel size. The point source is located and acquired in specified locations in the PET field of view (FOV) utilizing a vendor provided application. Resolution assessment is performed by the same application, providing full width half max (FWHM) measurements in millimeters (mm) for each location. The aim of this project was to assess PET spatial resolution in a more clinically relevant method, utilizing clinical reconstruction methods and corrections (3D iterative reconstruction, attenuation, and scatter). Methods: A phantom study was performed to assess system resolution in a more clinically relevant manner utilizing typical clinical reconstruction methods on a Siemens Biograph PET-MR. A 22Na, 0.25 mm point source was first imaged in air, per the NEMA protocol, and reconstructed with FBP as well as with the system’s enhanced reconstruction algorithms. The point source was placed approximately 6 cm horizontally from the center of the field of view (FOV) and imaged for two million counts. The result of the measurement was then compared with the vendor provided analysis application. To assess the system spatial resolution in more clinical-like conditions, the 22Na point source was glued to the largest (37 mm) bulb of a NEMA Image Quality PET phantom. When imaged, this placed the source in approximately the same location as the measurement in air. The phantom was filled with water and imaged for two million counts. This data set was reconstructed with a 3D iterative algorithm using typical clinical parameters - 172x172 matrix (4.2 mm pixel, 3 iterations, 21 subsets, 3 mm Gaussian smoothing filter). Additional reconstructions were performed for comparison and potential application in clinical imaging. Results: Data acquired with and without the reconstructions were analyzed and FWHM values were calculated which are shown in Table 1. This experiment’s NEMA protocol resolution of 3.9 mm, 3.7 mm (x, y respectively) FWHM matches the results of the vendor provided application, demonstrating the validity of the analysis technique. Utilizing the enhanced techniques of 3D iterative reconstruction and full corrections dramatically improves resolution to 1.87 mm and 1.80 mm FWHM. As expected, FWHM increases when the point source is imaged in the phantom (3.08 mm, 3.04 mm). Decreasing the matrix size to the clinically utilized 172x172 (4.17 mm pixel) increases the FWHM to 4.66 mm and 4.64 mm. Increasing the matrix size to 256x256 (2.8 mm pixel) only marginally improves the FWHM to 4.13 mm and 4.58 mm. Conclusions: Although important, there are other aspects of PET imaging beyond spatial resolution (image contrast, image quality) to be assessed. In the absence of a more complete assessment of diagnostic image quality, further experiment of image quality should be performed before attempting to utilize increased matrix sizes to improve spatial resolution. ER -