Image reconstruction parameters and the standardized uptake value ratios in brain amyloid PET.

Abstract

The present study investigated various image reconstruction protocols for amyloid PET using phantom test criteria published by the Japanese Society of Nuclear Medicine (JSNM) and compared them with the composite standardized uptake value ratio (cSUVR) in clinical imaging. Hoffman 3D phantoms and cylindrical phantoms were collected for 30 min according to the JSNM guidelines. Images were created under various reconstruction protocols by three physical evaluation items in the guidelines and were assessed: gray matter/white matter contrast (%contrast), uniformity (SDuROImean), and image noise [coefficient of variation (CV)]. We compared the cSUVR of images reconstructed under 15 protocols using 18F-flutemetamol and 18F-florbetapir in 15 cases each and the guidelines for physical evaluation of reconstruction parameters. No significant differences were observed in cSUVR between reconstruction protocols that satisfied the guidelines' criteria for %contrast and CV and those that did not; however, the visual impression of images differed. SDuROImean, which evaluated uniformity, met the criteria in all data. Reconstruction protocols should be selected appropriately using guidelines and other information, as cSUVR remains largely the same even if the visual impression of the images differs between different reconstruction protocols. When the relationship between %contrast and CV is expressed in terms of several reconstruction protocols, the graph shows a curved shape, and the optimal protocols for both %contrast and CV are near its center. Since cSUVR is similar to optimal parameters, even under parameters outside this range, multiple parameters need to be considered when selecting image reconstruction protocols for amyloid PET.