Comparative evaluation of lesion detectability for 6 PET imaging platforms using a highly reproducible whole-body phantom with (22)Na lesions and localization ROC analysis.
The lesion detectability performance of 6 PET imaging platforms has been compared using a highly reproducible whole-body phantom and localization receiver operating characteristic (LROC) analysis. A realistic whole-body phantom consisting of brain, thorax with lungs and liver, and pelvis with bladder was assembled and outfitted with 27 semipermanent (22)Na lesions of various sizes and activity concentrations. The background compartments were reproducibly filled with (18)F solutions. The phantom was imaged under the condition of equal emission scan time on 7 PET platforms: Advance, HR+, HR961, C-PET, IRIX, MCD, and AXIS. Imaging data were processed using manufacturer-supplied software and defaults, and LROC evaluation was performed using 11 human observers. Near-nominal counting rates were obtained for the NaI systems, and the bismuth germanate (BGO) systems were operated well below nominal counting rates. The BGO systems provided the highest lesion detection performance, followed by the large-area dedicated NaI system, and hybrid PET gamma cameras. Lesion detectability was highly dependent on lesion size, with all systems exhibiting similar performance for 16-mm lesions but differentiated performance for lesions < or =12 mm. Reconstruction methodology can have a significant effect on lesion detectability. PET lesion detectability performance is correlated with system cost and imaging characteristics. For a particular imaging task, care should be taken to ensure that the scanner being used is appropriate and that the scan time is adjusted accordingly to ensure good lesion detectability.