Amyloid and Tau Brain PET: Quantification, the Centiloid Scale, and the Physicist's Role

Introduction

Amyloid and tau brain PET have crossed the line from research instruments to clinically consequential tests, and that shift has made harmonized quantification a medical physics priority, not an academic luxury. With anti-amyloid monoclonal antibody therapies now influencing real treatment decisions, the question is no longer simply "is the scan positive?" but "how much amyloid, measured how, and comparable to what?" The answer depends on a chain of physics: scanner calibration, reconstruction, region definition, and the transform that places a result on a common scale. ¹²

That common scale is the Centiloid. By anchoring amyloid burden between young, amyloid-negative controls (0) and typical Alzheimer's disease patients (100), the Centiloid scale lets a florbetaben result from one center be compared to a florbetapir result from another. But the elegance of a single 0–100 axis hides a dependency: the transform only produces trustworthy numbers when the underlying standardized uptake value ratio (SUVR) is itself accurate and reproducible. ¹

This guide explains how amyloid and tau PET are quantified, how the Centiloid scale works, where the tracers differ, and why the medical physicist is central to making quantitative brain PET defensible. DRPS supports nuclear medicine and PET programs across Florida, Maryland, Virginia, Washington DC, California, and Nevada through PET/CT and nuclear medicine physics services.

Topic Explanation

What amyloid and tau PET measure

Amyloid PET images beta-amyloid plaque burden; tau PET images aggregated tau neurofibrillary pathology. Both are hallmark proteinopathies of Alzheimer's disease, but they behave differently in space and time. Amyloid tends to accumulate diffusely across neocortex and is well summarized by a single global value. Tau follows a more stereotyped, stage-dependent spread (often described using Braak-stage-inspired regions) and resists collapse into one number. ²³

Key terms used throughout this guide:

• Radiopharmaceutical — for amyloid, the F-18 tracers florbetapir, florbetaben, and flutemetamol, and the C-11 research tracer Pittsburgh Compound-B (PiB); for tau, the F-18 tracer flortaucipir.
• SUV (standardized uptake value) — tissue activity concentration normalized to injected activity and body weight.
• SUVR (SUV ratio) — the ratio of target-region uptake to a reference region such as the cerebellum, the basis for amyloid quantification.
• Centiloid (CL) — the harmonized 0–100 amyloid burden scale derived from SUVR. ¹

Why harmonization is the whole game

A florbetaben SUVR and a florbetapir SUVR are not directly comparable: they use different tracers, kinetics, and often different reference regions. Without harmonization, a "1.4" from one site means something different from a "1.4" at another. The Centiloid project solved this by defining a standard analysis pipeline and a method to scale any non-standard pipeline onto the same axis. The payoff is cross-tracer, cross-center, and longitudinal comparability — but only if each site's calibration and reconstruction are controlled. ¹⁴ For the upstream uptake-timing decisions that affect any quantitative PET, see our guide to PET uptake time.

Key Technical Principles

From SUV to SUVR

The standardized uptake value normalizes the measured activity concentration $C(t)$ in a region to the injected activity $A_0$ and patient weight $w$:

For amyloid imaging, the clinically robust quantity is the ratio of a cortical target region to a reference region, which cancels many global scaling and calibration factors:

The choice of reference region — whole cerebellum, cerebellar gray matter, pons, or a composite white-matter region for longitudinal work — materially affects the SUVR and therefore the downstream Centiloid value. This is one reason harmonization must specify the pipeline, not just the tracer. ¹

The Centiloid transform

The Centiloid scale applies a linear transform that anchors the endpoints. For an individual SUVR, with ${\text{SUVR}}_{\text{YC-0}}$ the mean of young amyloid-negative controls and ${\text{SUVR}}_{\text{AD-100}}$ the mean of typical Alzheimer's disease patients:

By construction, the young-control mean maps to 0 CL and the typical-AD mean maps to 100 CL. Values can fall below 0 or above 100; the scale is anchored, not bounded. ¹

A worked Centiloid example

Suppose a validated pipeline establishes ${\text{SUVR}}_{\text{YC-0}}=1.00$ and ${\text{SUVR}}_{\text{AD-100}}=2.00$ for a given tracer and reference region. A patient scan yields ${\text{SUVR}}_{\text{IND}}=1.30$. Then:

A result of 30 CL sits above commonly cited positivity thresholds, which generally fall in the range of roughly 20–30 CL depending on tracer, reference region, and whether the cutoff is tuned for sensitivity or specificity. A study comparing Centiloid quantification with expert visual reads reported the highest concordance at a cutoff near 18 CL overall (and near 24 CL for florbetaben), with 93% agreement against visual interpretation — illustrating both the power and the tracer-dependence of these thresholds. ⁴

Comparing the tracers

The short C-11 half-life of PiB ties it to facilities with an on-site cyclotron, which is why the F-18 amyloid tracers dominate clinical practice. For the broader isotope landscape, see common PET and radiopharmaceutical therapy isotopes.

Why scanner physics sets the ceiling on accuracy

Centiloid math is only as good as the SUVR feeding it, and SUVR is only as good as the scanner. Quantitative accuracy depends on NEMA NU 2 performance (spatial resolution, sensitivity, count-rate behavior, image quality), accurate dose-calibrator-to-scanner cross-calibration, and a controlled reconstruction recipe (iterations, subsets, post-filter, point-spread-function modeling, and time-of-flight). Differences in reconstruction alone can shift SUVR enough to move a borderline case across a threshold. This is why harmonization standards specify acquisition and reconstruction, and why the NEMA NU 2 performance testing that underpins quantitative PET is non-negotiable. ⁷

Clinical Impact

The arrival of disease-modifying anti-amyloid therapy has raised the stakes on every part of the quantitative chain. Amyloid PET now helps confirm amyloid pathology before initiating therapy and can support monitoring of amyloid reduction. When a number can gate access to an expensive, not-risk-free treatment, the difference between a well-calibrated 22 CL and a poorly calibrated 18 CL is not academic. ²

Quantification also strengthens the reliability of borderline reads. Visual interpretation remains the validated, primary clinical method for approved tracers, but the concordance literature shows that disagreements cluster exactly where they matter most — cases with focal or early uptake near threshold. Reporting a harmonized Centiloid value alongside the visual read gives the interpreting physician a defensible, reproducible second axis of evidence, and the highest reliability comes from combining both. ⁴

Tau PET adds complementary, stage-relevant information about where disease has progressed, which correlates more closely with symptoms than amyloid burden does. Because tau is regionally distributed, its clinical value depends on consistent region definition and reconstruction even more than amyloid does — another place where physics discipline directly shapes clinical meaning. ³ The same harmonization mindset that governs Ga-68 PSMA PET imaging in oncology applies here in neurology.

Longitudinal monitoring raises the bar further. When amyloid PET is used to track change over time — for example, to document amyloid reduction during therapy — the relevant quantity is the difference between two Centiloid measurements, and that difference inherits the noise of both scans. A reconstruction recipe that drifted between the baseline and follow-up study, a different uptake time, or a switched reference region can manufacture an apparent change that has nothing to do with biology. For longitudinal work, harmonized white-matter reference regions are often preferred precisely because they are more stable across sessions, and the test–retest variability of the chosen pipeline should be characterized so that a reported change can be judged against the measurement's own uncertainty. This is the difference between a number that looks precise and one that is actually reliable. ¹⁴

Practical Optimization Tips

A defensible quantitative brain PET program rests on a few disciplined practices.

1. Lock down acquisition and reconstruction

• Follow the SNMMI/EANM amyloid PET procedure standard for uptake time, scan duration, and acquisition. ⁸
• Define and freeze a reconstruction recipe for quantitative work; do not let it drift with software upgrades without re-validation.
• Document time-of-flight and point-spread-function settings, because they affect SUVR.

2. Cross-calibrate and verify

• Maintain dose-calibrator-to-scanner cross-calibration and verify it on a schedule.
• Run NEMA-based and routine QC (uniformity, SUV recovery) so that quantitative output stays anchored. ⁷

3. Validate the analysis pipeline

• Use a Centiloid-validated pipeline and confirm your site reproduces published anchor values before reporting CL clinically. ¹
• Fix the reference region per protocol and report it; a switch from whole cerebellum to a composite region changes the number.

4. Report transparently

• State the tracer, reference region, pipeline, and the adopted threshold alongside any Centiloid value.
• Pair quantification with the validated visual read rather than replacing it.
• For longitudinal studies, report the test–retest uncertainty of the pipeline so a measured change can be judged against measurement noise. ⁴

5. Train and audit the workflow

• Confirm that technologists follow the frozen uptake-time window and patient-positioning protocol, since both feed directly into SUVR.
• Periodically audit reported Centiloid values against visual reads to catch pipeline drift or region-definition errors early. ⁴

Common pitfalls to avoid

• Comparing raw SUVRs across tracers. Convert to Centiloids first. ¹
• Treating Centiloid thresholds as universal. They are tracer- and pipeline-specific. ⁴
• Ignoring reconstruction drift after upgrades. Re-validate quantitation.
• Reporting a number without its reference region and pipeline. It is not interpretable in isolation.
• Skipping NEMA and cross-calibration QC while trusting the quantitative output anyway.

Regulatory Considerations

Amyloid and tau PET radiopharmaceuticals are byproduct material regulated under the NRC medical-use framework or its Agreement State equivalent, while quantitative interpretation is governed by professional standards and appropriate use criteria. Medical use of F-18 and C-11 radiopharmaceuticals falls under 10 CFR Part 35 (or the equivalent Agreement State program), with dose limits set by 10 CFR Part 20, and authorized-user and dosing requirements that the radiation safety officer and medical physicist help administer. ⁹¹⁰

Key frameworks to reference:

• SNMMI/EANM amyloid PET procedure standard — acquisition, processing, and interpretation guidance for amyloid PET. ⁸
• SNMMI Appropriate Use Criteria for amyloid PET — defines clinically appropriate indications. ²
• NEMA NU 2 — the PET performance standard underpinning quantitative accuracy and cross-scanner comparability. ⁷
• FDA prescribing information — tracer-specific dosing and validated visual reading criteria for approved amyloid and tau radiopharmaceuticals. ⁵⁶
• 10 CFR Part 35 / Part 20 — medical use of byproduct material and radiation protection standards. ⁹¹⁰

In the states DRPS serves, Florida, Maryland, Virginia, California, and Nevada are NRC Agreement States administering their own medical-use programs, while Washington, DC is regulated directly by the NRC. A facility must confirm which authority issues its license and which authorized-user, dosing, and QC expectations apply. Coordinating the quantitative PET program with PET/CT and nuclear medicine physics support, radioactive material license support, and the RSO keeps the clinical and regulatory sides aligned.

Frequently Asked Questions (FAQs)

What is the Centiloid scale?

The Centiloid scale is a standardized 0–100 axis for amyloid PET burden. It linearly transforms a tracer- and pipeline-specific SUVR so that 0 represents the mean of young amyloid-negative controls and 100 the mean of typical Alzheimer's disease patients, allowing results from different tracers and centers to be compared. ¹

How is SUVR different from SUV in brain PET?

SUV normalizes tissue activity concentration to injected activity and body weight. SUVR is a ratio of mean uptake in a target region to a reference region such as the cerebellum. SUVR cancels many global scaling factors and is the basis for amyloid quantification and the Centiloid transform.

Do amyloid and tau PET require quantification, or is visual reading enough?

Approved tracers are interpreted using validated visual reading criteria, which remain the primary clinical method. Quantification with SUVR and the Centiloid scale is a powerful adjunct that improves consistency, supports borderline cases, and enables longitudinal and multi-site comparison, but it must be harmonized. ⁴⁵

Why does the medical physicist matter for amyloid and tau PET?

Quantitative brain PET is only as trustworthy as the scanner calibration, reconstruction, and analysis pipeline behind it. A medical physicist ensures NEMA-based performance, dose-calibrator and scanner cross-calibration, harmonized reconstruction, and pipeline validation so that SUVR and Centiloid values are accurate and reproducible. ⁷

Can you compare amyloid PET results across different tracers?

Yes, but only after conversion to a common scale. Raw SUVR values differ by tracer and reference region. The Centiloid scale was specifically designed to harmonize amyloid burden across tracers and analysis methods. ¹

What Centiloid threshold indicates amyloid positivity?

Published thresholds commonly fall in the range of roughly 20–30 Centiloids, with specific cutoffs depending on tracer, reference region, and whether the goal is sensitivity or specificity. A facility should adopt validated, tracer-appropriate thresholds. ⁴

Is tau PET quantified the same way as amyloid PET?

Tau PET also uses SUVR-based quantification, but it is regionally distributed and often summarized using Braak-stage-inspired composite regions rather than a single global value. Standardization of tau quantification is less mature than amyloid, though harmonization efforts are advancing. ³

Key Takeaways

• Quantification is now clinically consequential. Anti-amyloid therapy decisions make accurate amyloid burden measurement matter. ²
• The Centiloid scale harmonizes amyloid PET across tracers and centers by anchoring SUVR between young controls (0) and typical AD (100). ¹
• SUVR — not raw SUV — drives amyloid quantification, and reference-region choice changes the result.
• Thresholds are tracer- and pipeline-specific, typically near 20–30 CL; concordance with visual reading is highest near these cutoffs. ⁴
• Tau PET is regional, summarized with stage-based composites rather than a single global number. ³
• Scanner physics sets the ceiling. NEMA performance, cross-calibration, and frozen reconstruction recipes make quantitation trustworthy. ⁷

Conclusion

Amyloid and tau PET have become decision-grade tests, and the Centiloid scale is what lets a number from one tracer, scanner, and site mean the same thing as a number from another. But the scale is a transform, not a guarantee — it inherits every error in the SUVR it is built on. That is why quantitative brain PET is fundamentally a medical physics undertaking: harmonized acquisition, validated reconstruction, rigorous cross-calibration, and a pipeline that reproduces published anchors.

Facilities that pair validated visual reading with disciplined, harmonized quantification — and that treat the scanner and analysis chain as instruments to be controlled, not black boxes to be trusted — will deliver brain PET that is both clinically useful and defensible.

How DRPS Can Help

Diagnostic Radiation Physics Services supports PET and nuclear medicine programs implementing quantitative brain PET. This includes PET/CT and nuclear medicine physics, NEMA NU 2 performance testing, dose-calibrator and scanner cross-calibration, reconstruction harmonization, Centiloid pipeline validation support, QC program design, and radioactive material license support — all delivered by board-certified medical physicists.

DRPS serves facilities across Florida, Maryland, Virginia, Washington DC, California, and Nevada. To discuss a quantitative brain PET program, contact us or review our service locations.

Related Resources

• PET/CT NEMA NU 2 performance testing
• PET uptake time and its effect on quantification
• Ga-68 PSMA PET imaging
• Time-of-flight PET imaging
• Common PET and radiopharmaceutical therapy isotopes
• PET/CT and nuclear medicine physics services
• Radioactive material license support

References

1. Klunk WE, Koeppe RA, Price JC, et al. The Centiloid Project: standardizing quantitative amyloid plaque estimation by PET. Alzheimer's & Dementia. 2015;11(1):1-15.e1-4. doi:10.1016/j.jalz.2014.07.003. PubMed
2. Johnson KA, Minoshima S, Bohnen NI, et al. Appropriate use criteria for amyloid PET: a report of the Amyloid Imaging Task Force, the SNMMI, and the Alzheimer's Association. Journal of Nuclear Medicine. 2013;54(3):476-490. doi:10.2967/jnumed.113.120618. PubMed
3. Schöll M, Lockhart SN, Schonhaut DR, et al. PET imaging of tau deposition in the aging human brain. Neuron. 2016;89(5):971-982. doi:10.1016/j.neuron.2016.01.028. PubMed
4. Khalafi M, Hojjati SH, Wang XH, et al. Concordance between Centiloid quantification and visual interpretation of amyloid PET scans across the Alzheimer disease continuum. AJNR American Journal of Neuroradiology. 2025;46(9):1884-1892. doi:10.3174/ajnr.A8743. PubMed
5. U.S. Food and Drug Administration. Amyvid (florbetapir F 18), Neuraceq (florbetaben F 18), and Vizamyl (flutemetamol F 18) prescribing information. accessdata.fda.gov
6. U.S. Food and Drug Administration. Tauvid (flortaucipir F 18) prescribing information. 2020. accessdata.fda.gov
7. National Electrical Manufacturers Association. NEMA NU 2: Performance Measurements of Positron Emission Tomographs (PET). Rosslyn, VA: NEMA. nema.org
8. Minoshima S, Drzezga AE, Barthel H, et al. SNMMI procedure standard/EANM practice guideline for amyloid PET imaging of the brain, version 1.0. Journal of Nuclear Medicine. 2016;57(8):1316-1322. doi:10.2967/jnumed.116.174615. PubMed
9. U.S. Nuclear Regulatory Commission. 10 CFR Part 35: Medical Use of Byproduct Material. ecfr.gov
10. U.S. Nuclear Regulatory Commission. 10 CFR Part 20: Standards for Protection Against Radiation. ecfr.gov
11. International Commission on Radiological Protection. Radiation Dose to Patients from Radiopharmaceuticals. ICRP Publication 128. Ann ICRP. 2015;44(2 Suppl). icrp.org