This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the results posted in the ClinicalTrials.gov record; calculations added by Clinical Biostats are labelled as such.
1. Trial at a Glance
This page covers the registry results for Comparative Study of Conventional MR Images With Synthetically Reconstructed MR Images of the Brain (NCT05425927), a na na study. The primary outcome is Sensitivity of Pathological Findings (Full Analysis Set).
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT05425927 |
| Phase | NA |
| Condition | Clinical Indication for Brain MRI, Healthy Volunteers |
| Allocation | Not applicable (single group) |
| Design model | Single Group |
| Masking | None (open label) |
| Enrollment | 189 |
| Arms in the primary outcome | n/a |
| Lead sponsor | SyntheticMR |
| Results posted | Yes |
2. Clinical Question
Population
People undergoing brain MRI for a clinical indication and healthy volunteers are included in the registry’s condition descriptions. The primary outcome is defined for the full analysis set, but the registry record does not specify its participant-level inclusion criteria or its exact composition.
Intervention
Brain MR images are post-processed using SyMRI 15 (3D) to generate synthetic MR images. The registry identifies the software as a device intervention; it does not describe a separate therapeutic intervention.
Comparator
Conventional MR images are compared with the synthetic images generated from the acquired brain MRI data. This is a comparison of image types, not a randomized comparison between treatment arms.
Primary question
For participants with pathological findings, how sensitively do readers identify those findings using synthetic images compared with conventional images, with the site-determined diagnosis serving as ground truth?
The registry describes a prospective, multicenter, multireader investigation. Both types of images were included in each reading period, while each image was evaluated only once per reader. The stated primary outcome is sensitivity of pathological findings in the full analysis set. Sensitivity concerns detection among cases with the finding; it does not by itself describe how often readers classify people without the finding as positive.
The registry also lists outcomes concerning classification, sensitivity and selectivity for adult and pediatric populations. These indicate additional planned questions about reader performance and population subgroups, but their numerical results are not reported in the registry record. The record does not establish that the comparison was designed to show equivalence, noninferiority or superiority. Its listed outcome and design details therefore support describing the comparison, but not declaring one image type better or worse.
3. Trial Design
This is a completed, single-group study, with enrollment of 189. The registry lists allocation as NA and masking as none, and explicitly notes that this is not a randomized comparison. The comparison is instead between conventional brain MR images and synthetic MR images generated by post-processing with SyMRI 15 (3D). Accordingly, the image types should not be read as separate, randomized treatment arms.
The official title describes a prospective, multicenter, multireader investigation. Brain images and necessary MRI information were acquired on the exam day, with the primary outcome timeframe describing acquisition on 1 day. Synthetic images were generated before the first reading. Readers assessed images in two reading periods: the first at week 0-2 and the second at week 7-8, with a wash-out period of 4 weeks between readings. Both periods contained synthetic and conventional images, and each image was evaluated only once per reader.
The reference for identifying pathological findings was the site-determined diagnosis. The primary outcome was specified for the full analysis set, but its definition is not reported in the registry record. The exact number of readers, reader qualifications, sites, participant flow and image allocation procedures are also not reported in the registry record.
The record gives a start date of 2022-06-27, a primary completion date of 2024-02-10 and a status of completed. It lists SyntheticMR as sponsor and phase as NA. The registry record does not report a randomized control group or a conventional-image-only arm. No inference about treatment effects is appropriate from this design: its focus is a comparison of diagnostic image readings.
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
| Stage | Neurological MRI |
|---|---|
| STARTED | 189 |
| COMPLETED | 189 |
| NOT COMPLETED | 0 |
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.
| Characteristic | Unit | Neurological MRI Synthetic |
|---|---|---|
| Participants analysed | 189 | |
| Age, Continuous | years | 35.4 (SD 23.1) |
| Age, Customized: Pediatric (<18 years) | Participants | 63 |
| Age, Customized: Adults (>= 18) | Participants | 126 |
| Sex: Female, Male: Female | Participants | 107 |
| Sex: Female, Male: Male | Participants | 82 |
| Race and Ethnicity Not Collected | Participants | — |
| Region of Enrollment: United States | participants | 189 |
| Pathological finding: Congential malformations | Participants | 13 |
| Pathological finding: Infectious/inflammatory/demyelinating | Participants | 20 |
| Pathological finding: Intracranial neoplasms | Participants | 27 |
| Pathological finding: Toxic and metabolic conditions | Participants | 0 |
| Pathological finding: Traumatic lesions | Participants | 4 |
| Pathological finding: Vascular disorders of the brain | Participants | 26 |
| Pathological finding: Neuro degenerative disorders and hydrocephalus | Participants | 13 |
| Pathological finding: Normal/no findings | Participants | 86 |
| Anaesthesia used: Yes | Participants | 5 |
| Anaesthesia used: No | Participants | 184 |
6. Primary Endpoint
The registry lists the primary outcome as Sensitivity of Pathological Findings (Full Analysis Set), measured over Conventional images and necessary MRI information acquired from participant on exam day (1 day). Synthetic images generated before first reading. Analysis by readers in two readings, first reading: week 0-2, second reading: week 7-8.
| Element | Registry entry |
|---|---|
| Result type | Number |
| Measure of spread | 95% Confidence Interval |
| Unit | Proportion of participants |
| Endpoint classification | Binary Endpoint |
The primary endpoint is sensitivity of pathological findings in the full analysis set. As described in the registry, sensitivity is the number of participants with pathological findings correctly identified, divided by the number of participants with pathological findings assessed by a reader. The site-determined diagnosis is the ground truth against which each reader’s classification is judged. The endpoint is specified for both synthetic and conventional MR images.
Sensitivity is useful when the question is whether an image-reading approach identifies cases that are known to have the finding. It focuses on true positive classifications among people with the condition. It does not, on its own, describe the rate of positive classifications among people without the finding, nor does it summarize every aspect of diagnostic performance. The registry separately lists selectivity and classification outcomes, but numerical results for these are not reported in the registry record.
The stated dispersion measure for the primary endpoint is a 95% confidence interval. In general, a confidence interval communicates the uncertainty around an estimated quantity; it is not a guarantee that the true value falls within any particular interval. The registry record does not report a numerical sensitivity estimate or its confidence interval, so the size and precision of the primary comparison cannot be assessed from the posted results.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The record describes a comparison of reader assessments from conventional and synthetic brain MR images, using the site-determined diagnosis as the reference. Its primary outcome is sensitivity for pathological findings. However, no numerical primary estimate or confidence interval is reported in the registry record. As a result, the posted information does not establish the observed level of sensitivity for either image type or the size and direction of any difference between them.
The study is listed as completed with enrollment of 189, and its record lists 17 outcome measures but 0 analyses. These registry entries describe the registry entry; they do not substitute for outcome estimates. In particular, completion status does not show that a result was favorable, unfavorable or precise. Nor does the enrollment count reveal how many participants had relevant findings, how many were included in each endpoint denominator or how many readings contributed to an estimate.
The comparison has features that matter when interpreting diagnostic performance. Both image types were included in each reading period, and the images were evaluated by readers rather than assigned as separate participant groups. The reading schedule included a wash-out period, but the registry record does not report the statistical analysis used to account for dependence among readings, readers, participants or image types. It also does not report reader-level estimates or a model for combining reader results. Without these details and the numerical findings, precision and consistency across readers cannot be evaluated.
The endpoint is sensitivity, which addresses detection among participants with pathological findings. It is not a complete description of classification performance. The registry lists selectivity and classification outcomes, including adult and pediatric outcomes, but their results are not reported in the registry record. The absence of these results means that the posted information cannot show how sensitivity relates to other dimensions of diagnostic classification.
The record identifies a 95% confidence interval as the primary endpoint’s dispersion measure, but does not provide the interval itself. In general, the interval is needed to judge how uncertain an estimate is; the point estimate alone would not be enough to assess precision. Because neither is reported, no conclusion about the magnitude or precision of the primary result can be drawn. The design information supports describing what was compared and how readings were scheduled, but it cannot replace the missing outcome analyses.
The registry’s planning illustration is distinct from observed trial evidence. It is described as using the observed effect, a two-sided significance level of 0.05, equal allocation between two arms and a normal approximation. Those assumptions do not describe the listed single-group image-comparison design, and planning estimates are not observed power or trial results. Numerical planning estimates are not reported in the registry record.
9. Statistical Methods Explained
The primary endpoint is a proportion: sensitivity for pathological findings. Its numerator is the number of participants with findings correctly identified by a reader, and its denominator is the number of participants with findings assessed by that reader. The registry states that conventional and synthetic images are evaluated against a site-determined diagnosis. This definition makes clear that the measure concerns detection among participants considered to have pathological findings; it is not a general accuracy measure that combines all kinds of correct and incorrect classifications.
What does sensitivity measure?
In general, sensitivity describes the fraction of cases with a condition that a test or assessment classifies as positive. A higher sensitivity estimate means that a larger share of the reference-positive cases was identified in the data being analyzed. It does not by itself tell readers how often an assessment is positive when the reference indicates no finding. That complementary question requires a measure such as selectivity, which is separately listed among the registry outcomes. A measure should therefore be interpreted in the context of the clinical question and the chosen reference standard.
Here, the site-determined diagnosis is the stated ground truth. In general, a reference standard anchors the classification being evaluated, but conclusions depend on how that reference is established and applied. Details about the diagnostic process, adjudication, or handling of uncertain diagnoses are not reported in the registry record. The record also does not provide the number of reference-positive participants or the number contributing to each reader’s endpoint denominator.
Why is the confidence interval relevant?
The registry names a 95% confidence interval as the dispersion measure for the primary endpoint. In general, a confidence interval gives a range of values compatible with the data under the method used to calculate it, and conveys uncertainty around an estimate. Its width is informative about precision, while its location indicates the range of estimated values. The registry record does not report the actual sensitivity estimates or intervals, so readers cannot assess either the endpoint’s precision or the range of values consistent with the results.
For a comparison between image types, estimates need to be considered on a common basis. The registry description indicates that both types were assessed in both reading periods, but each image was evaluated only once per reader. It does not state the statistical procedure used to compare the estimates, whether the comparison was paired at the participant or reader level, or how the analysis handled the resulting dependence. Those details are important because repeated observations from the same participant or reader are generally not independent.
Why can reader and participant dependence matter?
In general, observations are clustered when multiple measurements relate to the same person, reader or site. Measurements within a cluster can be more alike than measurements from unrelated clusters. Treating all observations as independent when they are correlated can make uncertainty estimates misleading. A suitable analysis may account for clustering or pairing, depending on the study structure and estimand. The registry describes a multireader investigation and a two-reading schedule, but the analysis method and any adjustment for dependence are not reported in the registry record.
The wash-out period is a design feature intended to separate reading periods. In general, a wash-out interval can reduce the influence of memory from an earlier assessment, but its existence does not establish that carryover or recognition was eliminated. The registry reports the wash-out period as 4 weeks and says that both types of images were included in each reading period. It does not describe whether readers could recognize the same participant or image across periods, or how any such recognition was assessed.
What does over-dispersion mean in general?
For proportion outcomes, a simple model may assume that observations vary according to a specified probability distribution. In general, over-dispersion means that the observed variability is greater than the model expects. It can arise when observations are more heterogeneous or more correlated than assumed. Ignoring it can make standard errors too small and confidence intervals too narrow. Whether over-dispersion was assessed or addressed in this study is not reported in the registry record; the endpoint description alone does not identify the model used.
What can the planning illustration establish?
The registry describes a planning illustration based on the observed effect, a two-sided significance level of 0.05, equal allocation between two arms and a normal approximation. It also states that the effect to detect is a stated fraction of the observed effect. In general, such calculations explore how a hypothetical design might perform under specified assumptions. They do not establish the trial’s observed result, and sample-size planning figures are not observed power. The equal-allocation, two-arm assumption is not the listed single-group design, so the illustration should not be treated as a description of how this study enrolled or assigned participants.
The registry record does not report the analysis model, hypothesis test, handling of multiple readers, confidence-interval method, missing-data method or prespecified statistical analysis plan. Consequently, the endpoint definition can be explained, but the specific inferential method behind an estimate cannot be evaluated from the posted information.
10. Multiplicity, Missing Data and Analysis Populations
The registry lists 17 outcome measures and 0 analyses. The primary outcome is sensitivity of pathological findings in the full analysis set. Secondary outcomes include classification of pathological findings, adult sensitivity, adult selectivity, adult classification, pediatric sensitivity, pediatric selectivity and pediatric classification. Numerical analyses for these outcomes are not reported in the registry record.
When a study examines multiple outcomes or subgroup results, multiplicity is a general statistical concern: testing many questions can increase the chance of findings that appear notable by chance. An analysis plan may specify a primary outcome, a testing hierarchy or an adjustment strategy. The registry identifies a primary endpoint, but multiplicity procedures, planned contrasts and any hierarchy among the other outcomes are not reported in the registry record. It is therefore not possible to determine how the listed measures were handled statistically.
The term full analysis set appears in the primary endpoint description, but the population definition and participant flow are not reported in the registry record. The record does not state how missing images, unreadable scans, absent diagnoses or incomplete reader assessments were handled. It also does not report exclusions from the endpoint denominator or whether all enrolled participants contributed to each outcome. Without those details, the analysis population and the potential effect of missing data cannot be assessed.
In general, missing-data methods matter when the availability of observations is related to participant characteristics or outcomes. Excluding incomplete cases, imputing values or using a model that accommodates incomplete data can lead to different estimates under different assumptions. No such method is reported in the registry record. The posted information therefore does not allow readers to judge whether missingness could have affected the sensitivity estimates or their uncertainty.
11. Safety Results
The registry reports serious adverse events as affected participants divided by participants at risk.
| Arm | Participants with serious adverse events | Participants at risk | Percentage |
|---|---|---|---|
| Neurologic Synthetic and Conventional MRI | 0 | 189 | 0.0% |
Deaths
| Arm | Deaths (participants affected of at risk) |
|---|---|
| Neurologic Synthetic and Conventional MRI | 0 of 189 |
Other (non-serious) adverse events
Participants with at least one non-serious event at or above the registry's 0% reporting threshold.
| Arm | Participants affected of at risk |
|---|---|
| Neurologic Synthetic and Conventional MRI | 1 of 189 |
Most frequent non-serious adverse event terms
| Event term | Organ system | Neurologic Synthetic and Conventional MRI |
|---|---|---|
| Device Deficiencies | Product Issues | 1 of 189 |
The registry reports serious adverse events for the Neurologic Synthetic and Conventional MRI group. The serious-event count is 0 among 189 participants at risk, and deaths are recorded as 0. The record also lists 1 other adverse event among 189 participants at risk. The term for that other event is not reported in the registry record.
These counts describe the adverse-event entries in the registry, not a comparison between randomized groups. The study is listed as a single-group investigation, and the record does not provide a separate control group for interpreting these counts. The numbers alone also do not state whether an event was considered related to the image-processing software or study procedures, how events were collected, or how long participants were monitored. Those details are not reported in the registry record.
A count of 0 serious events in this record should be read as the reported count, not as proof that no serious event could occur. Likewise, the count of 1 other event does not disclose its nature or clinical significance. The posted safety figures are limited to the categories and counts recorded; they do not support conclusions beyond those entries.
12. What the Result Does, and Does Not, Mean
This study addresses a diagnostic-imaging question: how readers classify pathological findings when viewing conventional brain MR images compared with synthetic images generated by SyMRI 15 (3D). Its prospective, multicenter, multireader description and scheduled reading periods provide context for the comparison. The primary endpoint, sensitivity, focuses on detection among participants whose site-determined diagnosis indicates a pathological finding.
The registry record does not report a numerical primary estimate, confidence interval or analysis result. It therefore does not show the observed sensitivity for either image type, the size or direction of a comparison, or how precisely any estimate was made. The study’s completed status and enrollment of 189 do not answer these questions. The record also lists additional adult and pediatric outcomes, but their numerical results are not reported in the registry record.
Even if an estimate were available, sensitivity alone would not characterize all aspects of image classification. It does not describe performance among participants without pathological findings, and it does not establish how well a reading approach performs across every type of finding or setting. The registry record does not provide results for the listed selectivity and classification outcomes, or information sufficient to evaluate variation across readers and sites.
The comparison should not be interpreted as a randomized treatment trial. It compares image types within a single-group study and does not list a randomized control arm. The registry record does not report an equivalence or noninferiority conclusion, and the design description alone cannot establish that the two image types perform the same. In the absence of posted numerical results and analytic details, the defensible interpretation is limited to the study question, endpoint definition and reading design—not to a claim that either image type performs better, worse or equivalently.
13. Why This Trial Matters Statistically
This investigation is statistically interesting because it evaluates diagnostic image readings rather than comparing separate randomized treatment groups. The same study includes conventional images and synthetic images generated from brain MRI data, and the registry describes both types in each reading period. That structure raises practical analytic questions about how observations from the same participant and reader should be compared and how reader-to-reader variation should be represented.
The primary outcome is sensitivity, a proportion defined against the site-determined diagnosis. Its interpretation depends on the reference classification and on which participants and readings enter the denominator. The registry record identifies the endpoint and a 95% confidence interval as its dispersion measure, but does not report the numerical estimate, interval, analysis method or exact full analysis set definition. The gap between a well-defined question and missing result details is itself important: readers need estimates and their uncertainty to judge the evidence, not just a study label or enrollment count.
The listed adult and pediatric outcomes also illustrate the distinction between an outcome being named and its result being available. The registry lists 17 outcome measures but 0 analyses. Without reported values, readers cannot assess subgroup patterns, compare measures or evaluate multiplicity. A planning illustration cannot fill this gap because it concerns hypothetical assumptions rather than observed performance.
More broadly, this record shows why diagnostic studies require attention to design and analysis together. Reading order, wash-out, reader dependence, the reference standard and missing assessments can all affect interpretation. The registry gives some design details, including the reading windows and wash-out period, but not the methods needed to quantify or resolve these issues. Statistical conclusions should remain bounded by what the posted results actually report.
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 important limitation for interpretation is that numerical outcome analyses are not reported in the registry record. Although the primary endpoint and its 95% confidence interval are specified, no sensitivity estimate or interval is posted. The record lists 17 outcome measures and 0 analyses, leaving the primary comparison and the listed secondary questions without numerical results in the registry entry.
The primary endpoint is described for the full analysis set, but that population is not defined. Participant flow, numbers with pathological findings, endpoint denominators and the number of usable readings are not reported in the registry record. The handling of missing images, missing diagnoses and incomplete reader assessments is also not reported. These omissions prevent assessment of who contributed to the endpoint and whether missing information could have affected the estimates.
Some design features are described, but important analytic details are not. The record does not state the number of readers or sites, the statistical model, how comparisons accounted for repeated observations, or how reader and participant dependence were handled. It also does not explain whether the wash-out period prevented recognition across readings. These are relevant considerations for a multireader comparison, but the available record does not show how they were addressed.
Finally, the single-group design is not a randomized, like-for-like control comparison. The image types are compared within the study, and the record does not report an equivalence or noninferiority framework. The planning illustration’s equal-allocation, two-arm assumption is not the listed study design and should not be treated as evidence about observed performance. Together, these limits mean that the registry supports a description of the question and methods, but not a quantitative conclusion about comparative diagnostic performance.
15. Related Tutorials
16. Related Statistical Calculators
17. Sources
- ClinicalTrials.gov: Comparative Study of Conventional MR Images With Synthetically Reconstructed MR Images of the Brain — NCT05425927.
- Registry record last updated 2025-04-23; status last verified 2025-04.
- Results first posted 2025-04-23.
- 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.
Continue through the Clinical Biostats statistical pathway
Use this trial's endpoint as a starting point for deeper study of effect measures, confidence intervals, and sample-size planning.
18. Record Summary
The record describes a completed prospective, multicenter, multireader study of conventional brain MR images and synthetic images generated with SyMRI 15 (3D). Enrollment was 189. The study is listed as single group, with allocation NA and no masking; the registry explicitly notes that it is not a randomized comparison.
The primary outcome is sensitivity for pathological findings in the full analysis set, using the site-determined diagnosis as ground truth. Images were acquired on the exam day, synthetic images were generated before the first reading, and readers assessed images during two reading periods separated by a wash-out period of 4 weeks. Numerical primary and secondary analyses are not reported in the registry record.
The record lists 17 outcome measures and 0 analyses. It reports 0 serious adverse events among 189 participants at risk, 1 other adverse event among 189 at risk and 0 deaths. The type of the other event is not reported in the registry record.