Background: Amyloid imaging data is often quantified using a simplified method where target region to cerebellum
reference region uptake ratios (SUVR) are computed. In this study, we investigate the use of pons as an alternative
Methods: Data from the previously reported [18F]flutemetamol phase II study was used. 27 patients with earlystage
AD, 20 with MCI and 25 healthy volunteers (HV) underwent an [18F]flutemetamol PET scan (acquisition
window 85-115 min post-injection). Five of the AD subjects had a second (retest) scan performed within
1-2 weeks and 20 of the AD and the 20 MCI subjects also underwent a [11C]PIB scan (acquisition window 40-70
min post-injection). The “30-min PET sum” images were co-registered with the subject’s MRI and all data was
spatially normalized to MNI space where a volume of interest (VOI) atlas was applied. SUVRs were computed
for VOIs corresponding to frontal, lateral temporal and parietal cortices, and anterior cingulated and posterior
cingulate/precuneus. In addition, SUVR for a composite region (COM) was computed by averaging SUVRs for the
individual VOIs. SUVRs were computed using the cerebellar cortex (SUVRCER) and pons (SUVRPON) as reference
regions. We compared the different reference methods by investigating discrimination between the AD and HV
subject groups, test-retest variability and the correlation between [18F]flutemetamol and [11C]PIB.
Results: Both methods showed similar capability in discriminating between AD and HV (p-value SUVRCER=1.54E-11,
p-value SUVRPON=1.11E-9 for the COM region). The average [18F]flutemetamol test-retest variability for the COM
region was 1.7±0.7% for SUVRCER and 1.1±0.9% for SUVRPON. Correlations between [18F]flutemetamol and [11C]
PIB in the 20 AD and 20 MCI subjects were high in all cortical regions using SUVRCER values (Pearson R=0.91 for
COM) and very high using SUVRPON values (Pearson R=0.99 for COM).
Conclusion: The results show that pons is a suitable alternative reference region to the cerebellar cortex for
computation of SUVR values, with similar discrimination capability and test-retest variability and with high
correlation between [18F]flutemetamol and [11C]PIB.