Positron emission tomography (PET) with the radiotracer Pittsburgh Compound-B ([11C]PiB) is one of the main tools for imaging amyloid plaques in Alzheimer's disease. The gold standard for quantification is the DVR parameter, which requires an hour of dynamic recording and complex kinetic modeling. In practice, the simplified SUVR measure is used more often, computed from a short scan window, but it carries a systematic error: because the tracer washes out of tissues at different rates, SUVR overestimates amyloid burden, and the error depends on the window timing and on the burden itself. Previous correction methods worked only under equilibrium conditions, which for [11C]PiB are reached late, meaning they could not shorten the protocol. This work fills that gap by adapting a recently proposed generalized SUVR correction that does not require equilibrium to amyloid imaging.
The researchers used data from 88 older adults in the Harvard Aging Brain Study (HABS), divided into three groups by amyloid burden level: low, intermediate, and high. For each group they built an empirical model linking the tracer washout rate from the target region to its SUVR: in the low-amyloid group the rate was taken as the rate in the cerebellum (reference region) plus a constant; in the intermediate group a piecewise linear function of the inverse SUVR was used; in the high-amyloid group a power-law model with two parameters. Model parameters were estimated on a training subsample (a third of the data) and tested on an independent validation set. Then, using a correction formula that accounts for washout rates in the target and reference regions, they computed a corrected SUVRc for two time windows: 30–45 and 45–60 minutes after injection. Results were compared with the reference DVR obtained from 60-minute dynamic data using a simplified reference tissue model.
The mean SUVR bias relative to DVR ranged from 4.8% to 14.7% depending on group and time window, and it grew as the window lengthened. After correction, the SUVRc bias dropped to a range of −0.15% to 2.2%, and the variability of the bias decreased by a factor of 1.5 to 2. In the high-amyloid group, the bias reduction was statistically significant for 8 of 11 regions with the 30–45 minute window and for all regions with the 45–60 minute window. Lin's concordance coefficient between SUVRc and DVR rose from 0.954 to 0.987 (30–45 min window) and from 0.918 to 0.989 (45–60 min window). In longitudinal cohorts, SUVRc reflected changes in amyloid burden more accurately: for small changes (5–10%), SUVR overestimated the increase while SUVRc gave values close to DVR; for large changes (30–70%), SUVR underestimated while SUVRc nearly matched DVR.
The work shows that SUVR correction can be adapted to amyloid PET and works across the full range of pathology, from minimal to high burden. Unlike neighboring studies that either use machine learning for image processing or predict events from indirect signs, this work relies on kinetic principles and offers simple parametric models that can be transferred to other cohorts. This is the first case where SUVR correction has been applied to [11C]PiB and validated on independent data with a reference DVR. The method cuts the required scan time to 15 minutes without loss of accuracy, which matters for clinical practice and large trials.
The method can be extended to other amyloid radiotracers, especially [18F]Flutafuranol, whose kinetics are close to those of [11C]PiB. Using the correction to harmonize data across different tracers before conversion to the Centiloid scale is a promising direction. It may also be possible to replace the empirical washout rate models with neural networks trained on dynamic PET data. New high-sensitivity scanners with a long axial field of view could make it possible to estimate washout rates directly from activity curves, simplifying the method.
The method will affect Alzheimer's disease diagnosis by making it possible to obtain accurate quantitative measures of amyloid burden in 15 minutes instead of an hour. It will also boost the statistical power of clinical trials of anti-amyloid drugs and improve tracking of small amyloid changes during therapy.
Validate the method on large cohorts with arterial blood sampling to confirm accuracy against the two-tissue compartment model. Adapt the empirical models to other radiotracers and test them on data from different centers.
The key unsolved problem in amyloid neuroimaging is harmonizing quantitative measures across different tracers and scanning protocols. This method directly addresses the time dependence of SUVR, which is one source of between-center variability. In addition, accurate measurement of small changes in amyloid burden is critical for assessing therapy efficacy in the early stages of Alzheimer's disease, where changes are small.
Pittsburgh Compound B ([11C]PiB) is named after the University of Pittsburgh, where it was developed, and is a derivative of the dye thioflavin-T, which binds to amyloid fibrils. The half-life of carbon-11 is only 20 minutes, so scanning must begin right after injection, and a short 30–45 minute protocol fits within a single patient visit.
Not published on arXiv. Here are its page, the PDF and the formalization: you can check it yourself, not only read our account.