Translational Development of Photon-counting CT Imaging: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Translational Development of Photon-counting CT Imaging (NCT03878134), a na na study. The primary outcome is Mean Difference in Diagnostic Image Quality Grading of Photon Counting CT vs Conventional CT Images..

2. Clinical Question

3. Trial Design

This was a completed, single-group study of an investigational photon-counting CT scanner. The registry lists phase and allocation as NA and masking as none. Its design description says that participants underwent conventional CT for clinical reasons and additional PCCT imaging during the same examination time. The conventional images from each patient were used as an internal reference for that patient’s PCCT images. This structure makes the intended comparison within-person; it is not a randomized comparison between separate groups.

The record lists enrollment of 67 and gives a study start date of 2019-10-01 and a primary completion date of 2022-01-18. It identifies cancer, inflammatory disease, vascular diseases and metabolic bone disease among the conditions represented. The population description refers to patients with various clinical conditions undergoing routine conventional CT, with extra PCCT images obtained for comparison. The intervention is listed as a device, the PHOTON COUNTING CT Scanner.

The primary outcome was a mean difference in relative image-quality grading between PCCT and conventional CT images. Expert radiologist observers compared images using visual grading analysis. The description states that grading incorporates technical aspects of image acquisition and display, including contrast, noise, artifacts and post-processing. It does not report the number of observers, how observer ratings were combined, or the statistical procedure used to account for ratings from the same patient or observer.

The registry record lists one outcome measure and no analyses. It does not report a sample-size justification for the study’s actual design, a statistical analysis plan, or the outcome results. Any sample-size planning illustration based on an observed effect should therefore be distinguished from the trial design: the stated illustration assumptions use equal allocation between two arms, whereas the described image comparison uses each patient’s conventional images as an internal reference.

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

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Patients with various conditions including cancer who underwent their routine conventional CT scan with additional PCCT images obtained for comparison.

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.

Population: 11 patients could not be scanned on the PCCT investigational scanner after being enrolled due to scanner downtime.

6. Primary Endpoint

The registry lists the primary outcome as Mean Difference in Diagnostic Image Quality Grading of Photon Counting CT vs Conventional CT Images., measured over 2 years and 3 months, 27 months.

The primary endpoint is the mean difference in diagnostic image-quality grading for PCCT images versus conventional CT images. Expert radiologist observers assigned relative grades using visual grading analysis. The scale ranges from 1, meaning images are not usable because of severe limitations for diagnostic use, to 5, meaning excellent image quality with no limitations for clinical diagnosis. Higher scores represent better image quality.

This endpoint is used to turn a qualitative comparison of image appearance into an ordered numerical rating that can be summarized across image assessments. The scale’s description includes technical features such as contrast, noise, artifacts and post-processing. It is therefore intended to capture a broad judgment of image quality rather than a single technical characteristic. The within-patient reference is important to the question: the observers compare PCCT images with conventional CT images from the same patient, acquired during the same examination.

As a general statistical point, reporting a mean difference can describe the average direction and size of a score contrast, but interpretation also depends on variability and uncertainty. The registry record does not report the observed mean difference, its standard deviation, a confidence interval or a test result. It also does not report how the ordinal ratings were aggregated into the stated mean measure. Without those details, the endpoint definition can be described, but the size and precision of the observed comparison cannot be assessed.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The statistical information registered for this study is limited. The primary measure is described as a mean difference in image-quality grading, with standard deviation as the dispersion measure. No outcome analysis is listed, and the registry record does not report a statistical analysis plan or numerical result. The concepts below explain what these methods mean in general; they should not be read as a claim that a particular procedure was used in this study.

What does a mean difference describe?

In general, a mean difference summarizes the average contrast between two measurements or conditions. Here, the endpoint description defines the contrast as the relative image-quality grading of PCCT images versus conventional CT images. The sign and interpretation depend on the subtraction order used by the analysis. The registry names the comparison but does not specify the arithmetic order for the reported difference. The scale definition does clarify that higher grades indicate better image quality.

A mean is one way to summarize a set of ratings, but it does not show how consistently observers rated the images or whether ratings varied widely. A measure of dispersion, such as a standard deviation, describes spread around the mean in general. The registry identifies standard deviation as the dispersion measure, but it does not report its value for the primary outcome. The standard deviation would describe variability, not uncertainty in the estimated mean difference; a confidence interval is commonly used to express that latter uncertainty.

Why does the within-patient comparison matter?

In general, paired data arise when two measurements are linked, such as measurements from the same person. The record describes conventional and PCCT images from each patient, with conventional images used as an internal reference. A paired analysis accounts for that link by analyzing within-person contrasts, rather than treating the two sets of images as unrelated groups. This can be useful because characteristics shared by a patient may affect both image sets.

That general principle does not identify the method used for this study. The record does not report whether the analysis used paired differences, a model for repeated observations, or another approach. It also does not state whether multiple image regions or ratings per participant were included in the endpoint. When data have more than one rating per patient, treating every rating as independent can understate uncertainty if ratings from the same patient are correlated. A suitable analysis generally needs to reflect the structure of the data and the intended unit of inference.

How should the rating scale be understood?

The visual grading analysis scale has five ordered categories, from 1 for the lowest quality to 5 for the highest, with 5 described as better. In general, an ordered rating scale provides rank-like information: a higher category means a better grade, but the distance between adjacent categories is not necessarily equal. Calculating a mean treats category steps as if they can be averaged numerically. That may be a practical summary, but its interpretation depends on the scale and analysis rationale. The registry record does not state why the mean was selected, whether another summary was considered, or how observer ratings were combined.

The endpoint description says that observers consider technical factors in image acquisition and display, including contrast, noise, artifacts and post-processing. This broad assessment is useful for describing an overall observer judgment, but it can combine distinct image characteristics. The record does not report separate estimates for those characteristics, so they should not be inferred from a mean image-quality grade.

What does a confidence interval add?

In general, a confidence interval gives a range of values compatible with the data under the statistical model and procedure used. Its width reflects precision: wider intervals indicate less precise estimation, while narrower intervals indicate greater precision. An interval around a mean difference would help readers judge whether the observed contrast is estimated precisely enough to be informative. A confidence interval is not the same as a range containing most individual ratings, and it does not by itself establish clinical importance. No confidence interval for the primary outcome is reported in the registry record.

How should planning assumptions be read?

The record describes sample-size planning illustrations using a two-sided significance level of 0.05, equal allocation between two arms and a normal approximation. It specifies that the effect to detect is a stated fraction of the observed effect. These are planning assumptions, not evidence of an analysis result. A sample-size illustration asks how many participants would be needed under specified assumptions; it is not observed power and does not establish that the study was adequately sized. The record says that non-time-to-event illustrations are expressed as participants per arm. The resulting sample-size values are not reported in the registry record.

Those assumptions describe a two-arm planning calculation, while the study description is a single-group, within-patient comparison. General planning for paired data would need to reflect the variability of within-person differences and the intended paired analysis. The registry does not report such a design-specific justification, so the generic assumptions should not be treated as the study’s own protocol or as a substitute for its analysis plan.

10. Confidence Intervals for Each Arm

No two-arm comparison is shown for this trial, but each arm's mean can be given with a t-based 95% confidence interval computed from the posted mean, spread and number of participants.

R code and output

arm <- c("Patients Enrolled in Protocol")
n   <- c(56)
m   <- c(0.947368421)
sp  <- c(0.478877133)
sd  <- sp      # posted standard deviations
for (i in seq_along(n)) {
  se <- sd[i] / sqrt(n[i]); h <- qt(0.975, n[i] - 1) * se
  cat(sprintf("CARM|%s|%d|%.3f|%.3f|%.3f|%.3f\n", arm[i], n[i], m[i], sd[i], m[i] - h, m[i] + h))
}

CARM  Patients Enrolled in Protocol  56  0.947  0.479  0.819  1.076

11. Multiplicity, Missing Data and Analysis Populations

The registry lists one primary outcome measure and no analyses. It does not report whether there were additional planned endpoints, subgroups, image regions or observer-level comparisons beyond the registered measure. The record also does not describe any procedure to control the risk of false-positive findings when multiple statistical tests are performed. Multiplicity handling is therefore not reported in the registry record.

Missing-data methods are also not described. The record does not state how many participants had image-quality ratings available, whether any images or observer ratings were excluded, or how incomplete observations were handled. Enrollment was 67, while the adverse-event summary gives 56 at risk; these figures refer to different record elements, and the registry does not explain the relationship between them. Neither figure identifies how many observations contributed to the primary endpoint.

The analysis population is not specified. The registry does not say whether the primary analysis was intended to include all enrolled participants, only those with both image types available, or another defined group. It also does not describe exclusions, protocol deviations or handling of unusable images. These omissions limit assessment of potential selection bias and make it impossible to determine from the record whether the endpoint analysis reflects all eligible participants. The status is completed, but completion status alone does not provide information about missing ratings or the analysis population.

12. 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 record lists 56 people at risk in the group titled “Patients Receiving the Intervention PCCT CT Scan.” It records 0 serious events, 0 other adverse events and 0 deaths for that group. These are the posted counts; they should be read alongside the denominator and the limited safety information recorded.

A zero count means that no events of the listed type were recorded among those at risk in this registry summary. It does not establish that no event occurred outside the recorded information, nor does it describe the duration or completeness of safety monitoring. The registry does not report event definitions, collection procedures, follow-up details or a narrative explaining how the safety counts were obtained. It also does not explain why the adverse-event group’s 56 at-risk figure differs from the enrollment of 67.

These safety counts concern imaging exposure in a study of a device; they are separate from the primary image-quality endpoint. The absence of reported events in this summary cannot be used to estimate comparative safety against conventional CT, because the registry does not present a like-for-like randomized safety comparison. It should be interpreted as a description of the counts recorded, not as proof of absence of risk.

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

The record identifies an investigational imaging approach and a direct comparison: expert observers were to grade PCCT images against conventional CT images from the same patients. The endpoint is a relative image-quality rating on a scale where 5 is better. The study’s single-group, within-patient structure is relevant to the intended comparison, because each participant’s conventional images serve as an internal reference.

What the record does not establish is the observed image-quality difference. No mean difference, standard deviation, confidence interval or analysis result is reported, and the registry lists no analyses. It is therefore not possible from this record to tell whether the ratings favored PCCT or conventional images, how large any difference was, or how precisely it was estimated. The aim of comparing image quality is not itself a finding.

The rating concerns an overall observer assessment incorporating technical image features. It does not, by itself, establish effects on diagnostic accuracy, patient management, clinical outcomes or treatment decisions. Those are distinct questions, and the registry record does not report them as outcomes. Similarly, the adverse-event summary’s zero counts do not establish comparative safety or prove that no adverse events occurred beyond those captured in that summary.

The planning assumptions in the record are not outcome findings and do not show observed power. They describe an illustration based on an observed effect and a two-arm allocation framework, whereas the study description uses an internal reference within each patient. Without design-specific analysis details and reported endpoint estimates, those assumptions cannot resolve the uncertainty about the study’s result.

The most defensible reading is limited: the registry describes a completed study intended to compare relative image quality across two imaging methods, but it does not make the primary result available. Any conclusion about the direction or importance of the image-quality comparison would require results not reported in the registry record.

14. Why This Trial Matters Statistically

This study’s statistical question is useful because imaging comparisons often involve measurements that are naturally linked: two image sets can be obtained from the same patient. The registry explicitly describes conventional CT images as an internal reference for that patient’s PCCT images. Recognizing this structure helps readers distinguish a within-patient comparison from a randomized comparison between separate groups. The analysis should reflect that pairing, although the registry record does not report which method was used.

The endpoint also illustrates a common measurement challenge. Observer judgments can capture a broad assessment of image quality, incorporating contrast, noise, artifacts and post-processing, but the underlying scale is ordered and subjective. A mean difference can provide a concise summary, while the uncertainty and variability around that summary are essential for judging how stable it is. The record defines the scale and dispersion measure but does not report the numerical estimates needed to assess them.

The study matters statistically, too, because a completed status and a stated primary outcome do not guarantee that the result is available in the registry. Here, the record identifies the intended question but lists no analyses and gives no primary endpoint estimate. This separates protocol-level information from evidence about what the recorded observations showed.

Finally, the record’s planning illustration highlights the importance of matching assumptions to design. Its stated assumptions involve equal allocation between two arms, whereas the clinical comparison uses images from the same participant. In general, planning for paired data depends on the variability of within-person differences. The registry does not report a design-specific justification, so readers should not treat a generic illustration as proof of adequate precision or power for the actual comparison.

15. 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 main limitation for interpretation is that the registry record does not report the primary image-quality result or any statistical analysis. There is no mean difference, standard deviation, confidence interval or test result with which to assess direction, magnitude or precision. The record lists one outcome measure and no analyses, so the study’s endpoint findings cannot be reconstructed from the posted information.

The description does not report the number of expert observers, how their grades were combined, or whether ratings were made independently. It does not specify the number of images or regions assessed per participant, the analysis unit, or how repeated observations were handled. These details are especially relevant when ratings may be clustered within patients or observers. The registry also does not report reliability or agreement information for the observer grades.

The record does not identify the primary analysis population, the number of participants with complete paired images, exclusions, missing ratings or methods for handling missing data. Enrollment of 67 is not enough to establish the number contributing to the primary endpoint. The adverse-event summary’s 56 at-risk figure is not explained in relation to enrollment.

The conditions listed are varied, but the record does not report subgroup estimates or results by condition. It also does not describe image-quality findings by technical feature, so the overall grading endpoint cannot be separated into effects on contrast, noise, artifacts or post-processing. Finally, the planning assumptions described are for equal allocation between two arms and do not specify a design-specific justification for the within-patient comparison. The record does not report a protocol, analysis plan or outcome estimates that would address these uncertainties.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Translational Development of Photon-counting CT Imaging — NCT03878134.
• Registry record last updated 2023-11-01; status last verified 2022-01.
• Results first posted 2023-09-21.
• 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.

19. Record Summary

NCT03878134, “Translational Development of Photon-counting CT Imaging,” is listed as completed. The study evaluated an investigational photon-counting CT scanner in patients receiving clinically indicated conventional CT. Participants underwent conventional and PCCT imaging during the same examination, with conventional images serving as an internal reference.

The registry lists enrollment of 67, a single-group model, no masking, and phase and allocation as NA. Conditions include cancer, inflammatory disease, vascular diseases and metabolic bone disease. The primary measure is the mean difference in relative visual grading of image quality, on a scale from 1 to 5 where 5 is better. The record lists one outcome measure and no analyses.

The primary endpoint result, its variability and its uncertainty are not reported in the registry record. The adverse-event summary records 0 serious events, 0 other adverse events and 0 deaths among 56 at risk. The record does not explain the difference between that at-risk figure and total enrollment or describe the analysis population and missing data.