A retrospective evaluation of PET scanner properties utilizing brain Hoffman phantom imaging for F-18-AV-45 study

Objectives This study of 18F-AV-45 (florbetapir F 18) utilized Brain Hoffman phantom (HF) imaging to establish the consistency of scanner performance for brain amyloid multicenter imaging trials.

Methods Fifty-three brain HF scans were obtained with 1mCi F-18 imaged for 10 minutes in PET or PET/CT scanners at different imaging centers. All images are reconstructed by iterative technique (4i/16s) with 3mm post-reconstruction Gaussian filter. Inter-site uniformity is calculated by estimating statistical spread (SP=%SD/Mean) of normalized regional uptake of 7 regions (corresponding to gray and white matter regions of the phantom PET template). Attenuation is calculated by comparing counts in the middle to periphery and scatter estimated by comparing background counts to gray matter.

Results Non-uniformity is highest in the centrum (SP=8%) and lowest in the precuneus (SP=2%) (see table 1). Variance in attenuation (4%) and scatter (1.7%) is low with none of the images showing variation > 15% from the mean. Only 3 phantoms (5.7%) shows variation of uniformity in multiple regions (>1) more than 2 SD.

Conclusions Overall the regional uptake, attenuation and scatter did not vary significantly (<10%) across different phantom images. Highest statistical spread (8%) in Centrum may reflect variability in scanner resolution and phantom properties per se. Scanners with old transmission sources (e.g. HR+) and above average resolution (e.g. HRRT) created increased variability in phantom regional uptake (5.7%) and accounted for the few phantoms that had individual regions that quantitated >2 SD from the mean

• Download figure
• Open in new tab
• Download powerpoint

Table 1: Statistics of regional uniformity of phantom