Tabular PET data

This section provides a broad overview of the tabular data associated with ADNI PET scans. Specifically, we will be covering the following data products:

Numerical summary data in tabular csv format - including regional SUVR measures for both tau and amyloid PET.
Tables containing detailed acquisition and quality control information for individual scans.
Tables related to visual PET reads and the amyloid status disclosure process in ADNI4

Note that PET images in DICOM format - including raw, pre-processed, and post-processed scans - are also available on the LONI IDA, but outside the purview of this document.

Introduction to PET in ADNI

PET scan acquisition has been a critical part of the ADNI study since its inception, and the types of PET data collected has evolved over time. Amyloid PET scans have been acquired since the beginning of the study, while Tau PET was introduced to the protocol during the ADNI2 phase.

The figure below shows which scans were collected at different times across the different phases of ADNI, and which PET tracers were used during different time frames:

During each phase of ADNI, different participant cohorts underwent PET imaging at different intervals. The following table gives an overview of the relevant PET imaging schedule information:

Data organization

The tabular PET data can be found under the ‘Imaging’ tab of the study files interface on the IDA, divided into three sections:

PET image acquisition
PET image analysis
PET image quality

In addition to these sections, which contain technical information and metadata regarding PET scans themselves, ADNI4 introduced a new procedure for returning amyloid positivity results to participants on the basis of visual reads, and a number of tables associated with this procedure are stored on the IDA. Tables associated with this process can be found under the ‘Enrollment’ tab.

PET image analysis results

Tables in this section contain numerical summary information derived from PET images by the ADNI PET core based at UC Berkeley, as well as analyses carried out by external investigators.

A number of key pieces of information are captured in these tables, including centiloid (CL) values for amyloid PET, and standardized uptake value ratios (SUVr) for FDG, amyloid, and tau PET.

The most noteworthy tables in this section are the analysis results from the ADNI PET core at UC Berkeley, and these are the tables that will be covered in depth here. The other tables available in this section contain results of analyses carried out by external labs. For more information on that data, investigators should consult the documentation available on the IDA, as well as any publications associated with the projects.

Note that amyloid and Tau PET scans have all been processed to a common uniform resolution of 6mm^3 FWHM, to the extent possible, for the purposes of analysis, since smoothing affects PET tracer quantification. A minority of scans early in ADNI were acquired at a resolution worse than 6mm and could not be re-processed at 6mm, but for these scans, the PET Core has calculated region- and tracer-specific 6mm-to-8mm transformations and applied them in the UC Berkeley datasets in order to report a complete set of PET image SUVrs at 6mm resolution. Details of the smoothing process are described in the smoothing transformation methods document presented below, which can also be found on the IDA:

Amyloid PET numerical summaries (UCBERKELEY_AMY_6MM)

This table contains measurements derived from amyloid PET scans across all phases of ADNI, with the exception of PiB which was a limited sample (~100 scans) during ADNI1. This includes participant amyloid PET positivity status based on tracer-specific SUVr cutoffs, cortical summary SUVrs and CLs based on whole cerebellum intensity normalization, as well as additional reference regions, regional SUVRs, and FreeSurfer-based volumes that correspond to the regions used for sampling PET.

Documentation for this table is available on the IDA, alongside this table, and is included here as well for convenience:

Usage Notes

This table contains SUVRs from a number of different amyloid PET tracers. SUVRs resulting from different tracers are not directly comparable, and centiloids should be used to compare across tracers. This warning is repeated multiple times within the data set documentation, within the table itself, and further down in this very section.

The QC_FLAG field indicates the QC status of the image being assessed, based on visual inspection. Entries of 0 indicate a QC fail, and an entry of -1 indicates that the image has not been visually assessed.

the PROCESSDATE field indicates the date on which the analysis was processed by the PET core lab at UC Berkeley, including the application of the smoothing transform to 6mm, and it has no correspondence with the date on which the image was acquired. The SCANDATE field contains the date of acquisition.

The SCANDATE field should be used as a point of reference when combining PET results with tables from other modalities. Imaging dates and protocols do not necessarily correspond to the dates and protocols of clinical visits, biofluid draws, etc.

SUVRs resulting from different tracers are not directly comparable. centiloids should be used to compare results across tracers. Further information on the methods used to derive centiloid scale across multiple tracers can be found in Landau et al. 2025, and in the amyloid PET methods doc available above.

Tau PET numerical summaries

There are two tables containing numerical summaries of tau PET scans across all phases of ADNI - one table (UCBERKELEY_TAUPVC_6MM) includes measurements derived after applying partial volume correction (PVC), while the other table (UCBERKELEY_TAU_6MM) includes results obtained without any PVC.

More detailed information about these tables can be found in the associated documentation, which is available on the IDA and is reproduced here:

Usage notes

This table contains SUVRs from a number of different amyloid PET tracers. SUVRs and thresholds resulting from different tracers are not directly comparable, and CLs and/or positivity status (+/-) should be used when merging data across tracers. This warning is repeated multiple times within the data set documentation, within the table itself, and further down in this very section. Harmonization of tau PET SUVrs across tracers is an ongoing area of research in the PET Core (see Landau et al. Alz & Dem 2025).

The PROCESSDATE field indicates the date on which the analysis was processed by the PET core lab at UC Berkeley, including the application of the smoothing transform to 6mm, and it has no correspondence with the date on which the image was acquired. The SCANDATE field contains the date of acquisition.

A fuzzy merge should be used to combine this table with other data sources. The nearest observations in time can be matched by comparing the SCANDATE field with the field that captures the date of the other observation of interest (e.g. VISDATE/EXAMDATE for clinical date).

SUVRs resulting from different tracers are not directly comparable.

FDG PET numerical summaries (UCBERKELEYFDG_8mm)

FDG PET scans were conducted in all phases up to and including ADNI3, and were discontinued in ADNI4. This table contains SUVr measurements associated with a number of derived metaROIs. Note that FDG PET images were smoothed to 8mm^3 FHWM (the original common resolution in earlier phases of ADNI) and unlike amyoid & tau PET scans were not reprocessed at 6mm^63.

More information on these metaROIs can be found in the associated documentation, which is included below and available on the IDA

Usage notes

The UCBERKELEYFDG_8mm table is not formatted in ‘wide’ format. Instead of each participant having one row per visit, each participant/visit combination is associated with multiple rows - one for each metaROI.

PET acquisition

These tables (as the name of the section suggests) contain information about the acquisition of PET scans, including details of tracer administration.

Amyloid

Amyloid PET metadata is split between three tables due to variations in tracer usage across different phases of ADNI. Specifically, for each phase, we have:

ADNI1: PIB Scan Information (PIBMETA)
ADNIGO/2: AV-45 Scan Information (AV45META)
ADNI3/4: Amyloid PET Scan Information (AMYMETA)

All of these tables contain similar information, sometimes with different field names - a summary of how some critical information is recorded is represented in the table below

	PIBMETA (ADNI1)	AV45META (ADNIGO/2)	AMYMETA (ADNI3/4)
Date of scan	EXAMDATE	EXAMDATE	SCANDATE
Was the scan conducted? (y/n)	PBCONDCT	PMCONDCT	DONE
Was the radiotracer administered? (y/n)	May be recorded as protocol variation in the PBVARIAT field	RADTRACER	RADTRACER
Time of tracer injection	PBPIBTIME	PMINJTIME	INJTIME
Time of scan	PBSCTIME	PMSCTIME	SCANTIME

Information on scanner make/model is recorded explicitly in the PIBMETA and AV45META tables, while it is absent in the AMYMETA table. Getting this information for ADNI3/4 requires cross-referencing image metadata on the IDA.

Tau (TAUMETA)

Data on tau PET acquisition across all phases is contained in a single table, Tau PET Scan Information (TAUMETA).

FDG

This is split across 3 tables, similarly to amyloid PET information. FDG information can be found in the PETMETA_ADNI1, PETMETA_ADNI2, and PETMETA3 tables for the ADNI 1, ADNIGO/2, and ADNI3 phases, respectively.

PET quality

These tables contain information regarding the quality of acquired PET images and series, based on a multi-step quality control (QC) process carried out by Bob Koeppe (Univ Michigan).

Amyloid

Similar to the acquisition tables above, QC information for amyloid PET scans is split across three tables, corresponding to historical tracer usage. Specifically, for each phase we have:

ADNI1: PIB QC Tracking (PIBQC)
ADNIGO/2: AV-45 PET QC Tracking (AV45QC)
ADNI3/4: Amyloid PET QC (AMYQC)

There is a great deal of information captured in these tables, and like the acquisition metadata tables listed above there is some variance in how that information is recorded, with tables in earlier phases sometimes recording more granular information.

One field of particular importance indicates whether or not a scan passed QC. This field is named PASS in the PIBQC and AV45QC tables, and SCANQLTY in the AMYQC table.

Tau (TAUQC)

There is one table containing QC information for Tau PET that applies to scans across all phases. Please consult the data dictionary for more detailed information on its contents.

FDG (PETQC)

Unlike the acquisition metadata, QC information for FDG PET is contained to one singular table across all the phases in which it was administered.

Visual read results & amyloid status disclosure

In ADNI4, participants are given the option of learning the results of their amyloid PET imaging.

The tables related to this procedure - including the results of visual reads of amyloid PET scans - can be found under the ‘ENROLLMENT’ section of the study files interface, as indicated in the figure below:

This is a new process for ADNI4, and more information can be found in the ADNI4 procedures manual and clinical protocol. A relevant excerpted from the procedures manual that gives a general outline of the disclosure process is presented below:

Further details on the methods and procedures for conducting visual reads are available in the relevant methods doc, which is available on the IDA and also included here for reference:

Amyloid PET visual read (AMYREAD)

This table contains the results of visual read assessments of participant amyloid PET scans. Images are categorized as either ‘elevated’ or ‘non-elevated’, on the basis of observed cortical tracer binding. The results of this assessment are recorded in the OUTCOME field.

This visual assessment - NOT the centiloid-based cutoffs present in the analysis results tables - is used to determine a participant’s amyloid positivity status for the purposes of disclosure.

Usage notes

The final result of the visual read process is given in the OUTCOME field. Investigators who are strictly interested in the visual read result should use the OUTCOME field. The remaining fields in the table can be seen as a sort of ‘flowchart’ through the visual read process.

Because the assessments in this table are based on visual reads and not on centiloid quantification, it is possible for the two methods to produce different assessments of amyloid positivity. If the two methods are in agreement, the result of the visual assessment (captured in the OUTCOME field) will also appear in the CONGRU field. If the two are not in agreement, the field will contain an NA value.

To clarify, since this represents a potential point of confusion: 1 and 0 in the CONGRU indicate whether the scan was elevated or not elevated if both the visual read and the quantitative read are in agreement. A 0 in the CONGRU fields does not indicate that the visual read and the quantitative assessment are in disagreement. The CONGRU is NA if the two methods are not in agreement. If CONGRU is populated with a non-NA value, then the visual read is in agreement with the quantitative assessment.

In cases where it is particularly difficult to visually assess an image, a consensus meeting is convened and the scan is reviewed by a team. If a scan requires a consensus meeting, it is indicated in the CONSENS field.

If a consensus meeting is called, then the resulting decision (elevated vs. not elevated) is reflected in the CONSENSRES field.

Disclosure visit questionnaires

This return of results process is new to ADNI4, and a great deal of interesting information is available regarding the disclosure process itself - including feedback from both participants and clinicians.

Participants are asked to complete brief questionnaires before and after learning of their results (PREAMYDISC and POSTAMYDISC, repsectively), while the clinician conducting the visit is asked to complete a questionnaire as well (AMYDISC).

Investigators who are interested in the return of results process are strongly encouraged to consult the data dictionaries for these tables to see the exact contents, as it is quite varied and interesting. Some topics covered include:

participants’ self-assessed risk of developing dementia, asked both before and after learning of their results
clinician confidence in participant understanding of their results
details of the visit, including the setting, presence of loved ones, etc.
did the participant share their results after obtaining them? If so, with who?
topics discussed during the disclosure visit
participant and clinician satisfaction with the disclosure process

Other associated instruments

There are also a number of other instruments that are specifically collected in association with the disclosure process.

These assessments include:

Impact of events scale (IES)
Mini-State Trait Anxiety Inventory (STAI) for Adults
Perceived Stress Scale
RYFF Psychological Well-Being: Purpose in Life

The results of these questionnaires can be found in the Assessments -> Neuropsychological section of the study files interface.