CenTauR Project


 

Tau positron emission tomography (PET) imaging is the most recent addition to the arsenal of tools for the in vivo assessment of neurodegenerative proteinopathies.

Differences between tracers, quantitative pipelines and regions of interest lead to disparities in PET-derived quantitative measurements which decrease reproducibility and pose a challenge when trying to compare tau PET outcomes across cohorts or in therapeutic trials that use different tau tracers.

The aim of the CenTauR project is to standardize tau PET quantification by expressing measurements on a universal scale, with units termed CenTauRs. The CenTauR method builds upon and improves the previously proposed CenTauRZ approach (Villemagne et al., Alzheimers Dement 2023), which established a standardized quantification framework based on predefined CenTauR regions of interest (ROIs) and a z-score transformation to rescale tau PET measurements. However, this approach has inherent limitations, as z-score normalization is sensitive to differences in reference population characteristics across cohorts. CenTauR overcomes these limitations by replacing the z-score transformation with conversion equations derived from the Joint Propagation Model (JPM), a nonlinear mixed-effects model that integrates head-to-head, anchor-point, and test–retest tau PET data (Leuzy et al., Alzheimers Dement 2024; Moscoso Rial, Leuzy, et al. Alzheimers Dement 2026). The JPM hence provides a more accurate and robust characterization of inter-tracer relationships, enabling the harmonization of tau PET measurements across tracers and cohorts.

Given the inherent limitations of the CenTauRZ approach, CenTauRZ is being deprecated in favor of the new CenTauR scale. Therefore, tau PET measurements transformed to CenTauRZ should not be aggregated across different tracers, as the underlying z-score normalization does not ensure quantitative equivalence between them. Similarly, cut-points defined on the CenTauRZ scale should not be applied interchangeably across tracers under the assumption that equivalent CenTauRZ values reflect the same underlying biological substrate. Still, cut-points previously established using CenTauRZ using a single tracer and quantification method can be readily converted to the CenTauR scale (Moscoso Rial, Leuzy, et al. Alzheimers Dement 2026).

How to derive CenTauR units?

The CenTauR approach harmonizes tau PET measurements across tracers and quantification methods using tracer- and region-specific conversion equations derived from the Joint Propagation Model (JPM).

There are two possible approaches for obtaining CenTauR units:


Level-A: Direct computation using the Standard Method

Level-A CenTauR computation is the recommended approach for deriving CenTauR units directly from tau PET images that were acquired using a standard acquisition time window (Section 2.1.1 in Moscoso Rial, Leuzy, et al. Alzheimers Dement 2026).

The CenTauR Standard Method specifies tracer-specific tau PET acquisition time windows and a standardized SPM12-based quantification pipeline. The pipeline calculates Standard SUVR (SUVRStd) using an inferior cerebellar gray matter reference region and five predefined CenTauR regions of interest: Universal, Meta Temporal, Mesial Temporal, Temporoparietal, and Frontal. The resulting SUVRStd values are then converted into CenTauR units by applying the corresponding tracer- and ROI-specific JPM conversion equations (Table 1 in Moscoso Rial, Leuzy, et al. Alzheimers Dement 2026).

To facilitate implementation of Level-A CenTauR computation, we provide a freely available CenTauR Standard Method quantification pipeline, distributed as MATLAB code.

The pipeline takes a tau PET image and its corresponding T1-weighted MRI as inputs, performs coregistration and processing of the T1 scan, and automatically generates SUVRStd and CenTauR values for the five standard ROIs. As with any PET quantification pipeline, it is important to inspect the output for accuracy in coregistration and segmentation.


Level-B: Conversion of non-standard tau PET measurements

Level-B CenTauR computation enables researchers to convert previously obtained tau PET measurements into CenTauR units, even when these were generated using different acquisition windows, quantification pipelines, regions of interest, reference regions, or quantitative metrics.

Because the JPM conversion equations are defined specifically for SUVRStd, non-standard measurements must first be calibrated to SUVRStd. This requires deriving a linear conversion equation between the original measurements and SUVRStd, ideally using at least 50 participants spanning a broad range of tau pathology (see Section 3.2 in Moscoso Rial, Leuzy, et al. Alzheimers Dement 2026 for additional details).

For a valid calibration, the relationship must be approximately linear, demonstrate sufficient agreement (R² > 0.7), and involve anatomically comparable regions of interest (see Section 3.2 in Moscoso Rial, Leuzy, et al. Alzheimers Dement 2026 for additional details). When SUVRStd cannot be obtained directly because of a non-standard acquisition window, additional calibration data covering both acquisition windows are required.

Once the non-standard measurements have been transformed into SUVRStd, the corresponding JPM equations (Table 1 in Moscoso Rial, Leuzy, et al. Alzheimers Dement 2026) can be applied to obtain CenTauR units.

Calibration datasets

To facilitate the calibration of non-standard tau PET metrics into SUVRStd, we currently provide reference datasets for [18F]flortaucipir and [18F]MK-6240. These datasets include tau PET and T1-weighted MRI images, together with precomputed SUVRStd values provided in tabular format. Researchers can apply their own quantification methods to the images and compare the resulting measurements with the provided SUVRStd values to derive the calibration equations required for Level-B CenTauR computation.

[18F]Flortaucipir calibration dataset

This dataset contains T1-weighted MRI and [18F]flortaucipir tau PET images from 100 participants in the Mayo Clinic Study of Aging and Mayo Clinic Alzheimer's Disease Research Center, comprising two groups:

  • 50 individuals with probable Alzheimer's disease (25 men and 25 women): Adults aged 65 years or older with an amnestic clinical phenotype, a positive PiB amyloid PET scan, and no additional neurological diagnoses.
  • 50 cognitively unimpaired individuals (25 men and 25 women): Older adults with a negative amyloid PET scan, matched to the Alzheimer's disease group by age and sex.

[18F]MK-6240 calibration dataset

This dataset comprises 50 participants (25 men and 25 women), aged 54–93 years, spanning different stages of the Alzheimer's disease continuum:

  • 15 amyloid-negative cognitively unimpaired individuals: 7 men and 8 women.
  • 24 amyloid-positive individuals with mild cognitive impairment (MCI): 12 men and 12 women.
  • 11 amyloid-positive individuals with Alzheimer's disease dementia: 6 men and 5 women.

To improve reproducibility and consistency across studies, we encourage CenTauR users to use these reference datasets when calibrating non-standard tau PET metrics to SUVRStd, rather than relying on independently assembled calibration cohorts.


Additional calibration datasets for other tau PET tracers will be incorporated as they become available.

CenTauRZ (deprecated)

CenTauRZ was the original approach developed by the CenTauR project to standardize tau PET quantification across radiotracers (Villemagne et al., 2023). It introduced five standardized CenTauR regions of interest and a SPM8-based common quantification framework, expressing tau PET measurements as z-scores relative to tracer-specific reference populations of amyloid-negative cognitively unimpaired individuals.

However, differences in reference population characteristics and sampling variability limit the quantitative comparability of CenTauRZ values across tracers. These limitations are addressed by the new JPM-based CenTauR scale, which provides more accurate inter-tracer harmonization.

CenTauRZ is therefore being deprecated in favor of the new CenTauR scale. However, we continue to provide the original CenTauRZ quantification pipeline to support reproducibility, facilitate comparisons with previously published studies, and enable researchers to reproduce analyses conducted using the earlier framework. For new studies, we recommend adopting the updated CenTauR scale.