Improving PET Image Quality and Quantification by Using Motion Correction, Parametric Imaging and MAP Reconstruction: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Improving PET Image Quality and Quantification by Using Motion Correction, Parametric Imaging and MAP Reconstruction (NCT04417998), a non-randomized phase 4 study. The primary outcome is Motion Correction.

2. Clinical Question

The study asks whether motion-corrected PET images acquired while a subject is moving have image quality similar to a stationary reference image. The registry defines image quality for this question through the cumulative absolute relative difference in standard uptake values in grey matter voxels. It then measures the percentage of voxels whose difference from the stationary reference is greater than 10%; 0% is defined as ideal for this measure.

The primary outcome applies to the Motion Correction arm only. The record also describes a Parametric Imaging arm, intended to evaluate data acquisition and image-processing workflow in people undergoing brain or whole-body oncologic PET/CT scans. The study title and listed intervention additionally identify MAP reconstruction, but the registry record does not report a corresponding arm description in the available details. These aims address imaging methods and workflow, not a comparison of clinical treatment outcomes.

The study is non-randomized, so its arms should not be interpreted as like-for-like randomized comparison groups. The registry record does not report an observed value for the primary outcome, or a statistical comparison establishing whether motion-corrected images met a prespecified criterion for similarity. It also does not report a numerical definition of acceptable similarity. Consequently, the stated research question can be described, but the posted record does not provide enough outcome data to determine how closely the observed images met that aim.

3. Trial Design

The registry identifies this study as Phase 4, completed, parallel, non-randomized, and unmasked. The recorded enrollment is 80. The study began on 2020-06-03, and its primary completion date is 2023-12-20. Mayo Clinic is listed as sponsor. The conditions are Brain Imaging and Whole Body Imaging.

The brief summary describes an assessment of new software available on a new PET/CT scanner in the Molecular Imaging Research PET/CT Facility, together with an evaluation of data acquisition and image-processing workflow. The listed interventions are 18F-FDG, 18F-AV1451, and the Siemens E7 Reconstruction Tool. The record names Motion Correction and Parametric Imaging arms. For Motion Correction, it describes prospective data collection from subjects already enrolled in Mayo Clinic Rochester research study 08-005553 and scheduled for scanning on the Siemens Biograph Vision 600 PET/CT system. The description of this arm is incomplete in the registry record. The Parametric Imaging description concerns prospective data collection from subjects undergoing brain or whole-body oncologic PET/CT scans.

The primary outcome is assigned to the Motion Correction arm only. The registry record also identifies MAP Reconstruction in its adverse-event group listings, but a detailed description of that arm is not reported in the registry record. The record does not explain how the listed interventions map to each arm, whether all participants received each intervention, or whether any participant contributed data to more than one aim. Those details should not be inferred from the intervention list.

Because allocation was non-randomized and masking was absent, the design does not establish exchangeability of groups or blinded assessment. The stated comparison for the primary endpoint is between motion-corrected images and a stationary-reference reconstruction, not a randomized comparison between study arms. The registry record does not report a protocol-level account of how scan acquisition, reconstruction settings, or other potential sources of variation were controlled. It also does not report a rationale for the enrollment target or a sample-size justification.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Overall Study

Analysis population for the primary outcome, as defined in the registry: This outcome measure was collected and analyzed for the Motion Correction arm only.

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Motion Correction, measured over 2 years.

The primary endpoint is the percentage of grey matter voxels with a cumulative absolute relative difference greater than 10% in standard uptake values, comparing images with and without motion correction against a reference image acquired while the subject was stationary. The outcome is summarized as a mean and standard deviation across subjects, and the timeframe is listed as 2 years. The registry describes 0% as ideal for this measure.

This endpoint turns differences across many voxels into a subject-level summary that can describe how much of the evaluated grey matter image exceeds the stated difference threshold. The threshold focuses attention on voxels with larger relative differences from the stationary reference, while the percentage expresses the extent of such voxels in the image. As a continuous measure, its mean describes the average percentage across subjects, and its standard deviation describes how much those subject-level percentages vary around that average.

General: a voxel-based image metric can provide a structured way to quantify image differences, but the measure represents only the features captured by its definition. It is not, by itself, a complete measure of diagnostic performance or clinical benefit. Interpretation also depends on how the reference image, motion-corrected reconstruction, and eligible voxels were defined and processed. The registry record does not report the observed mean or standard deviation, or further details of those processing choices, so the endpoint's observed magnitude and precision cannot be assessed from the posted results.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary endpoint is a continuous measure summarized by its mean and standard deviation. Its underlying subject-level quantity is the percentage of grey matter voxels whose cumulative absolute relative difference in standard uptake values is greater than 10% when images with and without motion correction are compared with a stationary reference. The registry states that 0% is ideal. These details define what the endpoint is intended to measure, but the observed mean and standard deviation are not reported in the registry record.

What does the percentage represent?

Each image contains voxels, and the endpoint evaluates the differences in standard uptake values for grey matter voxels. The absolute relative difference expresses the size of a difference relative to a reference value without retaining its direction. The registry then applies a threshold: it counts the percentage of voxels whose cumulative absolute relative difference is greater than 10%. A subject with a smaller percentage has fewer evaluated voxels above that threshold according to this metric. This summary does not show the exact difference for every voxel, and it does not indicate whether the differences are spatially clustered or spread across the image.

General: threshold-based summaries can make a complex distribution easier to describe, but they also discard information. Values just below and just above a threshold are treated differently by the count, even if they are close, while larger differences above the threshold are not distinguished by the percentage alone. The registry record does not report additional distributional summaries or voxel-level results that would clarify those patterns.

Why report a mean and standard deviation?

The mean is the arithmetic average of the subject-level percentages. It describes the average endpoint value in the analyzed subjects, not the result for every subject. The standard deviation describes how spread out those subject-level values are around the mean. It is a measure of variability among observations, not the same as the uncertainty in the estimated mean.

General: a confidence interval for a mean addresses uncertainty in the population mean under the assumptions of the analysis, whereas a standard deviation describes variation among individual observations. The registry record does not report a confidence interval for this endpoint. It also does not give the number of subjects included in its calculation, so the precision of a mean cannot be assessed from the listed enrollment alone.

What is the role of the stationary reference?

The endpoint uses images acquired when the subject was instructed to remain still as the reference. The purpose is to assess whether reconstructed images acquired during movement, with motion correction, have image quality similar to that reference. This is an image comparison, not a comparison of clinical outcomes or a randomized treatment effect. The reference condition is important because the measured difference depends on how it was acquired and reconstructed. Further technical details about the reference and reconstruction process are not reported in the registry record.

General: reference-based measurements can be affected by measurement error in both the image being evaluated and the reference. If either is variable, the calculated difference may reflect more than the imaging method of interest. The registry describes the intended reference condition but does not report repeatability assessments or other estimates of measurement error.

How should similarity be assessed?

The study description uses the word similar, while the primary endpoint defines a threshold for identifying voxels with larger differences. A small endpoint value would indicate that a smaller share of the evaluated voxels exceeded that threshold. However, the registry record does not report an acceptable margin for similarity, a formal decision rule, or an inferential analysis for this outcome. Without such details and the observed data, the stated similarity aim cannot be translated into a statistical conclusion.

General: a formal similarity or equivalence assessment usually requires a prespecified margin that defines how much difference is acceptable, together with an estimate and uncertainty interval evaluated against that margin. A failure to detect a difference is not, by itself, proof of similarity. The registry record does not describe such a framework for this endpoint, so it would be inappropriate to infer one.

What do the design labels imply?

The study is described as non-randomized and unmasked. General: random allocation can help balance measured and unmeasured participant characteristics between comparison groups, while masking can reduce some forms of assessment or performance bias. These protections are not described as features of this study. The primary endpoint nevertheless has a within-imaging reference comparison as described in the registry; it should not be confused with a randomized between-group trial comparison.

The registry lists a sample-size planning illustration based on the observed effect, with a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. Such planning values are illustrations, not observed power. The observed primary effect and numerical planning outputs are not reported in the registry record, so no numerical planning conclusion can be evaluated. The record also does not report the study's own sample-size justification or a protocol-level statistical analysis plan.

Overall, the registry defines a continuous image-quality endpoint and its intended summary statistics. It does not provide the numerical endpoint results or analysis details needed to assess estimation, precision, similarity, or robustness.

10. Multiplicity, Missing Data and Analysis Populations

The registry record lists three outcome measures and zero analyses. It provides the definition of the primary Motion Correction endpoint, but details of the other outcome measures are not reported in the registry record. A statistical plan specifying how outcomes were prioritized or analyzed is also not reported.

Multiplicity arises when several outcomes or statistical tests are assessed: the chance of at least one apparently notable result can increase as more tests are considered. General: methods such as prespecifying a primary outcome or adjusting inference across tests can address different multiplicity concerns, but the appropriate approach depends on the study aims and analysis plan. The registry record does not report a multiplicity strategy, adjustment, or set of analysis-specific hypotheses. With no posted analyses, the record does not support assessment of how multiplicity was handled.

The record does not report missing-data counts for the primary endpoint, the number of subjects with usable images, or reasons for unavailable measurements. Methods for handling missing or incomplete scans, exclusions, and technically unusable images are not reported in the registry record. The analysis population and any deviations from the intended population are also not reported. Enrollment alone cannot establish how many subjects contributed to the endpoint calculation.

General: missingness can affect interpretation when the availability of an outcome is related to participant characteristics, scan conditions, or the measurement itself. The direction and size of any such effect cannot be assessed without information on missingness and the analysis approach. Those details are not reported in the registry record.

The registry identifies the primary outcome as collected and analyzed for the Motion Correction arm only. It does not report a population definition beyond that arm-level description, such as whether all enrolled subjects with a scan were included or whether specific data-quality criteria were applied. These limitations prevent a full appraisal of analysis populations, missing-data handling, and the scope of any inference.

11. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

The adverse-event listings report zero serious adverse events and zero other adverse events for the Motion Correction, Parametric Imaging, and MAP Reconstruction groups. The numbers listed at risk are 30 for Motion Correction, 20 for Parametric Imaging, and 30 for MAP Reconstruction. The listings also report zero deaths in each group. These are the posted counts; they should not be interpreted as proof that no events occurred outside the reported collection or reporting period.

The primary aim concerns imaging methods and workflow. The adverse-event figures are safety summaries and do not provide information about the primary image-quality endpoint. The registry record does not report event definitions, the period during which events were collected, ascertainment procedures, or group-specific follow-up. It also does not provide a narrative of events or further detail about how the groups were formed.

General: zero reported events means that no events were recorded in the corresponding registry count; it does not establish zero underlying risk. Interpretation depends on who was monitored, how long they were followed, and how events were defined and collected. The listed at-risk counts should not be treated as the number of subjects with evaluable primary imaging outcomes, since the registry record does not report that connection. The safety counts also do not make the non-randomized groups suitable for a like-for-like comparison.

12. What the Result Does, and Does Not, Mean

The registry record documents a completed study designed to assess PET image quality and data-processing workflow, including a motion-correction endpoint based on differences from a stationary reference. It specifies the endpoint's unit, threshold, direction, summary statistics, and the arm to which it applies. Those details make the intended measurement understandable.

What the record does not provide is the observed mean or standard deviation for the primary outcome. It also does not report a confidence interval, an inferential result, or a numerical criterion for deciding whether the images were sufficiently similar. Therefore, the posted information cannot establish the magnitude or precision of the measured image differences, or whether the study met its stated similarity aim.

The result should not be interpreted as a clinical treatment effect. The primary endpoint is an image-based measure, and the study is non-randomized and unmasked. The registry record does not report a randomized comparison or clinical outcome showing that an imaging method changes diagnosis, care, or health outcomes. The listed interventions and study arms describe imaging-related procedures and substances, but the record does not provide enough arm-specific detail to infer their use or effects.

General: technical performance measures can inform understanding of an imaging workflow, but they do not automatically establish clinical utility. A measured image difference is not equivalent to diagnostic accuracy, and diagnostic accuracy is not itself equivalent to improved patient outcomes. The registry record does not report those further outcomes.

The safety listings show zero recorded serious adverse events, other adverse events, and deaths across the named groups, alongside the stated numbers at risk. These counts provide a limited description of posted safety reporting, not a complete estimate of risk. Taken together, the registry record supports a description of the study aims and measurement plan, but not a numerical conclusion about the primary imaging result, its uncertainty, or its clinical implications.

13. Why This Trial Matters Statistically

This study is statistically notable because it translates an imaging-method question into a defined quantitative endpoint. Rather than relying only on a general description of image quality, the primary measure identifies a voxel population, a reference image, a difference metric, and a threshold. It then summarizes the proportion of voxels exceeding that threshold at the subject level and proposes a mean and standard deviation across subjects. This makes the intended measurement more explicit and potentially easier to compare across imaging workflows.

The endpoint also illustrates why definition and results must be read together. The metric's ideal value is 0%, but the registry record does not report the observed mean or standard deviation, and it does not set out a formal margin for similarity. The endpoint can therefore be described, but its achieved value and relevance to the stated aim cannot be evaluated from the posted record. A clearly defined measure is useful, but it does not substitute for reporting the estimate and its uncertainty.

The design is another important statistical feature. The study is non-randomized and unmasked, and the primary outcome is limited to the Motion Correction arm. The comparator is a stationary image reference rather than a randomized control group. That distinction affects what conclusions the design can support and discourages treating the listed arms as direct, balanced comparisons.

General: imaging studies often involve complex data that are reduced to a manageable number of subject-level summaries. Statistical interpretation depends on transparent reporting of how those summaries were formed, how many subjects contributed, and how uncertainty was quantified. Here, the registry describes the endpoint construction but does not report these numerical analysis details. The case therefore emphasizes that a technically specific endpoint and a completed study status do not by themselves allow readers to assess effect magnitude, precision, or reproducibility.

14. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The most consequential limitation of the registry record for statistical interpretation is the absence of numerical primary outcome results. The observed mean and standard deviation are not reported, nor are a confidence interval, a test result, or subject-level summaries. This prevents assessment of the endpoint's magnitude, variability, and precision.

The registry describes the target as image quality similar to a stationary reference, but an acceptable similarity margin or decision criterion is not reported. The endpoint's 10% voxel threshold defines which differences contribute to the summary; it does not state how small the overall percentage must be for the study aim to count as met. Technical details about how the images and reference were processed are also not reported in the registry record.

The study is non-randomized and unmasked. The record does not report blinding of image assessment, procedures for controlling scan conditions, or adjustments for potential sources of variation. The Motion Correction arm description is incomplete, and a detailed description of the MAP Reconstruction arm is not reported in the registry record. The relationship between the listed interventions and each arm is also not specified in the available details.

Although the overall enrollment is listed as 80, the number of subjects contributing to the primary endpoint is not reported. Missing scans, exclusions, image-quality failures, analysis populations, and missing-data methods are not reported in the registry record. The record also does not provide the study's own sample-size justification or a full statistical analysis plan. These gaps limit evaluation of generalizability, potential bias, precision, and reproducibility.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Improving PET Image Quality and Quantification by Using Motion Correction, Parametric Imaging and MAP Reconstruction — NCT04417998.
• Registry record last updated 2025-01-22; status last verified 2025-01.
• Results first posted 2025-01-22.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

18. Record Summary

The registry record describes a completed, Phase 4, parallel, non-randomized, unmasked study sponsored by Mayo Clinic, with enrollment of 80. Its broad aim is to assess PET software and data acquisition and image-processing workflow. Listed interventions include 18F-FDG, 18F-AV1451, and the Siemens E7 Reconstruction Tool.

The primary outcome is specific to the Motion Correction arm. It measures the percentage of grey matter voxels with cumulative absolute relative differences greater than 10% from a stationary reference, summarized by mean and standard deviation; 0% is ideal for this measure. The registry record does not report the observed primary outcome values, uncertainty estimates, or a formal similarity criterion. It lists zero serious adverse events, zero other adverse events, and zero deaths in each named adverse-event group. Further details about analysis populations, missing data, and statistical methods are not reported in the registry record.