Assessing a universal neocortical mask for Centiloid quantification

Abstract

AbstractBackgroundThe Centiloid (CL) project was developed to harmonise the quantification of Aβ‐PET scans to a unified scale. The CL neocortical mask was defined using 11C‐PiB, overlooking potential differences in regional distribution among Aβ tracers. We created a universal mask using an independent dataset of 5 Aβ tracers, and investigated its impact on inter‐tracer agreement, tracer variability and group separation.MethodUsing data from the ADOPIC study (AIBL, ADNI and OASIS3), age‐matched pairs of mild Alzheimer’s disease (AD) (MMSE = 20‐24;CL>25) and healthy controls (MMSE> = 28;CDR = 0;CL<15) were selected: 18F‐Florbetapir (FBP:147 pairs), 18F ‐Florbetaben (FBB:22 pairs), 18F ‐Flutemetamol (FLT:10 pairs), 18F ‐NAV (NAV:42 pairs), 11C‐PiB (PIB:63 pairs). The PET images were smoothed to a uniform 8mm resolution to reduce the influence of different scanner sharpness on the derived masks. The images were then spatially normalised using the SPM CL pipeline and transformed into SUVR (whole cerebellum). For each tracer, the mean AD‐HC difference image was computed and mirrored. The threshold for each difference image was optimised to maximise the overlap with the standard CL mask. The Universal mask was defined as the intersection of all 5 masks. It was then used to recalibrate each tracer into Centiloids. The Universal mask and associated transforms were evaluated on the GAAIN head‐to‐head calibration datasets in terms of inter‐tracer agreement and variance in the young controls (YC). They were also evaluated on the baseline ADOPIC dataset (N = 3565) with the effect‐size between HC/AD and HC/MCI.ResultThe overlap between each tracer specific mask was high (mean Dice = 0.82). The Universal mask was 26% smaller than the Standard one, but the overlap was high (Dice = 0.74). The Universal mask led to a small reduction in the variance of the YC in most tracers (‐3.4%) and a small increase in the R2 between each of the 11C‐PiB/18F‐tracer pairs (+0.24%). In ADOPIC, it led to a higher effect‐size between HC/AD (1.43 vs 1.42) and HC/MCI (0.71 vs 0.70).ConclusionThe universal CL mask led to an increase in inter‐tracer agreement and group separation. Those increases were however relatively small indicating that the existing standard CL mask is suitable for the quantification of all Aβ tracers.