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 Feature Detection Study for Validation of Hamamatsu NanoZoomer S360MD Digital Slide Scanner System (NCT03992768), a na na study. The primary outcome is Overall Percentage Agreement of Repeatability and Reproducibility Within Scanner, Between Scanners, and Sites..
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT03992768 |
| Phase | NA |
| Condition | Pathology |
| Allocation | Not applicable (single group) |
| Design model | Single Group |
| Masking | None (open label) |
| Enrollment | 540 |
| Arms in the primary outcome | n/a |
| Lead sponsor | Hamamatsu Photonics K.K. |
| Results posted | Yes |
2. Clinical Question
Population
The registry record describes scans or fields of view containing histological features, assessed by reading pathologists. The intra-scanner substudy used 378 features, read by 3 reading pathologists in 3 sessions each. The inter-scanner substudy also used 378 features and 3 reading pathologists in 3 sessions each. The inter-site population description is incomplete in the registry record.
Intervention
The intervention was whole slide imaging using the Hamamatsu NanoZoomer S360MD Digital Slide Scanner System. The study objective was to evaluate how consistently histological features were detected when scanning and reading procedures were repeated within scanners, among scanners at a site, and across sites.
Comparator
This was a single-group device validation study, not a randomized comparison against a control device or treatment. The comparisons were between reads or scans under different repeatability and reproducibility conditions. The registry record does not report a separate comparator arm.
Primary question
The primary question was how much positive agreement was observed when feature detection was compared across repeated reads or scanning conditions. The endpoint was Average Positive Agreement Rate, assessed for intra-scanner, inter-scanner, and inter-site precision. The registry record does not report the observed agreement estimates, confidence intervals, or statistical test results, so the size and precision of the agreement cannot be determined from the recorded results.
3. Trial Design
This was a completed, single-group device study of the Hamamatsu NanoZoomer S360MD Digital Slide Scanner System. The registry record describes the study objective as evaluating repeatability and reproducibility in the detection of histological features using whole slide imaging. Enrollment was 540. The recorded phase was NA, allocation was NA, and masking was NONE. Because the record identifies a single-group design and no allocation, the study should not be interpreted as a randomized treatment comparison.
The primary outcome was assessed through three precision substudies. Intra-scanner precision involved 3 separate scans on 1 scanner at the site. Inter-scanner precision involved scanning 378 features on 3 scanners at the site. Inter-site precision involved scanning 378 fields of view on 1 scanner at different sites. The record also states that features were read by 3 reading pathologists in 3 sessions each for the intra-scanner and inter-scanner substudies. The description of readers for the inter-site substudy is incomplete in the registry record.
The study start date was 2019-06-12, and the primary completion date was 2021-09-30. The overall status is COMPLETED. The registry record does not report the number of sites, the full inter-site reading procedures, or the results for the primary endpoint. No randomized control group is listed, so the central design question is consistency of detection under repeated imaging and reading conditions rather than a difference between treatment groups.
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 | NanoZoomer Whole Slide Imaging |
|---|---|
| STARTED | 540 |
| COMPLETED | 540 |
| NOT COMPLETED | 0 |
Analysis population for the primary outcome, as defined in the registry: Intra-scanner Precision: 3 separate scans on 1 scanner of 378 features read by 3 reading pathologists in 3 sessions each Inter-scanner Precision: 1 scan each of 378 features on 3 scanners read by 3 reading pathologists in 3 sessions each Inter-site Precision: 378 FOVs scanned on 1 scanner read by 3 reading pathologists at 3 sites in 1 session each
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 | NanoZoomer Whole Slide Imaging |
|---|---|---|
| Participants analysed | 540 | |
| Age, Categorical: <=18 years | Participants | — |
| Age, Categorical: Between 18 and 65 years | Participants | — |
| Age, Categorical: >=65 years | Participants | — |
| Sex: Female, Male: Female | Participants | — |
| Sex: Female, Male: Male | Participants | — |
| Race and Ethnicity Not Collected | Participants | — |
| Region of Enrollment: United States | Participants | 540 |
6. Primary Endpoint
The registry lists the primary outcome as Overall Percentage Agreement of Repeatability and Reproducibility Within Scanner, Between Scanners, and Sites., measured over 1 day.
| Element | Registry entry |
|---|---|
| Result type | Number |
| Measure of spread | 95% Confidence Interval |
| Unit | Average Positive Agreement Rate |
| Endpoint classification | not reduced to a single comparison |
The primary endpoint is titled Overall Percentage Agreement of Repeatability and Reproducibility Within Scanner, Between Scanners, and Sites. Its unit is Average Positive Agreement Rate (APA), and the timeframe is 1 day. The registry record defines APA by comparing two reads and classifying them into four possible outcomes: a means both reads detect the feature; b means read 1 fails to detect it; c means read 2 fails to detect it; and d means both reads fail to detect it. The stated formula is APA = 2a/(2a+b+c).
APA focuses on positive agreement: it reflects the degree to which repeated assessments both identify a feature, while its denominator includes cases where only one read identifies the feature. The both-negative outcome, d, is not part of the stated formula. This makes the endpoint distinct from a measure that counts agreement on both positive and negative classifications equally.
In general, repeatability describes consistency under repeated measurements within the same measurement setting, while reproducibility addresses consistency across changed measurement conditions. The study applied APA to scans repeated within a scanner, scans among scanners at a site, and scans across sites. The registry record identifies a 95% confidence interval as the dispersion measure, but it does not report the APA estimates or their confidence intervals. Therefore, the recorded endpoint definition explains what was intended to be measured, not the observed level of agreement.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The registry record describes a completed validation study and specifies how the primary endpoint was to be calculated. It does not report a numerical APA result, a 95% confidence interval, or another primary-outcome estimate. The registry lists 1 outcome measure and 0 analyses. Accordingly, the posted results do not establish the magnitude or precision of positive agreement for any of the three scanning conditions.
The endpoint is framed around positive agreement rather than overall agreement across positive and negative classifications. APA gives weight to cases where both reads detect a feature and accounts for cases where only one read detects it. Because both reads failing to detect a feature is not included in the stated formula, the measure does not by itself describe agreement on negative findings. An APA estimate, if reported, would need to be interpreted as a measure of positive detection consistency under the relevant comparison, not as a complete description of all classification agreement.
The study separates intra-scanner, inter-scanner, and inter-site conditions. Those conditions address different sources of variation, so a result for one would not automatically establish consistency for another. However, no separate estimates are reported in the registry record. The incomplete inter-site population description also limits what can be established about the exact reading setup for that substudy.
The record reports enrollment of 540, but does not report how many observations or participants contributed to each endpoint analysis, or whether any planned observations were excluded. Enrollment should not be treated as the endpoint analysis denominator. The adverse-event record lists 0 at risk and 0 affected for serious and other events in the NanoZoomer Whole Slide Imaging group, with 0 deaths. Those recorded zeros do not provide a basis for a broad safety conclusion, particularly because the registry does not describe an at-risk population for that reporting.
The sample-size assumptions describe planning illustrations based on an observed effect, using a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. Those are planning assumptions, not the design of this single-group validation study. The registry record does not report the resulting planning estimates or an observed effect for this endpoint. Any such illustration should not be interpreted as observed power or as evidence of the study's precision.
9. Statistical Methods Explained
What does the agreement measure represent?
The primary measure is Average Positive Agreement Rate. The registry record defines it using paired reads and four possible outcomes. The symbol a denotes cases where both reads detect a feature; b and c denote cases where one read fails to detect the feature; and d denotes cases where both reads fail to detect it. The stated calculation, APA = 2a/(2a+b+c), emphasizes positive findings and disagreement about positive detection. The both-negative category, d, is not part of this calculation.
In general, agreement measures describe consistency between assessments; they do not by themselves establish that either assessment is correct. A comparison against a reference standard or an independent truth determination would address a different question. The registry record does not report a reference-standard comparison for the primary endpoint. Consequently, APA should be interpreted as the specified agreement measure, not as accuracy against an established truth.
Why distinguish repeatability from reproducibility?
In general, repeatability concerns consistency when measurement conditions are kept similar, whereas reproducibility concerns consistency when some conditions change. The study applies these ideas to imaging and reading. Its intra-scanner substudy repeats scans within a scanner at a site. Its inter-scanner substudy compares scans among scanners at a site. Its inter-site substudy concerns scans at different sites. These are separate settings, and agreement in one setting cannot be assumed to represent agreement in the others.
The intra-scanner description specifies 3 separate scans on 1 scanner and 378 features read by 3 reading pathologists in 3 sessions each. The inter-scanner description specifies 1 scan each of 378 features on 3 scanners, with the same stated reader and session arrangement. The inter-site description specifies 378 fields of view scanned on 1 scanner at different sites, but the remaining population details are incomplete in the registry record.
What does the confidence interval add?
The primary endpoint names a 95% confidence interval as its dispersion measure. In general, a confidence interval communicates uncertainty around an estimated quantity: narrower intervals indicate greater precision under the method used, while wider intervals indicate less precision. An interval does not guarantee that the true value lies within its bounds, and its interpretation depends on the statistical procedure and assumptions. The registry record does not report the endpoint estimate, the interval limits, or the method used to calculate the interval. The degree of precision therefore cannot be evaluated from the posted results.
What can be inferred from the analysis record?
The registry record lists 1 outcome measure and 0 analyses. This indicates that no analysis results are listed in the record; it does not establish whether analyses were or were not conducted outside the registry reporting. The record does not report an analysis plan, the handling of repeated observations, or whether estimates account for dependence among reads, features, scanners, or sites. In general, observations that share readers, equipment, or sites may be correlated rather than independent. Accounting for such dependence can affect standard errors and confidence intervals, but the registry record does not state how it was handled.
The record describes a sample-size planning illustration using 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. These assumptions describe a planning calculation, not the study's reported design. In general, planning estimates depend on the assumed effect and statistical model. They are not observed power, and they do not replace an analysis of the data collected. The registry record does not report the calculation results, an observed endpoint effect, or the trial's own sample-size justification.
10. Multiplicity, Missing Data and Analysis Populations
The primary outcome description covers three related precision settings: scans within a scanner, scanners within a site, and scans between sites. The registry record lists 1 outcome measure and 0 analyses, but it does not report separate results for these settings, a hierarchy among them, or a plan for handling multiple comparisons. It is therefore not possible to determine whether any multiplicity procedure was planned or used. In general, when several related hypotheses are assessed, multiplicity methods can help control the chance of false-positive findings across those assessments; whether that concern applies in a particular way depends on the prespecified analysis plan, which is not reported in the registry record.
The registry record does not report missing-data counts, reasons for missingness, exclusions, or an imputation method. It also does not report how many observations contributed to each endpoint estimate. Enrollment of 540 is a study-level figure and cannot be taken as the denominator for any specific APA analysis. In general, missing observations can affect both the composition of the analyzed data and the precision of an estimate, especially if missingness is related to the feature or measurement condition.
The record does not identify an analysis population, such as a defined set of scans, reads, features, or participants included in each analysis. It also does not explain how repeated reads or scans were combined, or how dependence among assessments was handled. These details are important for interpreting an agreement estimate and its confidence interval. Since neither estimates nor analysis methods are reported, the posted results do not permit an assessment of the consequences of missing data, the analysis population, or multiplicity.
11. What the Result Does, and Does Not, Mean
The study addresses a measurement question: whether histological features are detected consistently under repeated whole slide imaging conditions using the Hamamatsu NanoZoomer S360MD Digital Slide Scanner System. Its stated endpoint is positive agreement, with separate intended assessments for repeat scans within a scanner, scans among scanners at a site, and scans across sites. This is a device-validation objective, not a randomized comparison of clinical treatments.
The registry record establishes the planned measurement framework and reports enrollment of 540, study dates, and completion status. It does not report the primary APA estimates or their 95% confidence intervals. As a result, the posted results do not quantify how much agreement was observed or how precisely it was estimated. The endpoint definition alone cannot show whether agreement was high or low.
Even if an APA estimate were available, it would answer a bounded question about positive agreement under the described reading and scanning conditions. It would not, by itself, establish that detected features were correct, that negative findings were consistently classified, or that findings generalize to settings not described in the record. The incomplete inter-site population description and absence of reported analysis details further constrain interpretation.
The adverse-event entries record zeros for at-risk and affected counts and deaths in the listed device group. They should be understood as registry entries, not as a comprehensive demonstration of safety. Likewise, a sample-size planning illustration is not a result from the study and is not observed power. The study's statistical meaning is therefore limited by the difference between a defined endpoint and a reported endpoint estimate: the former is available, while the latter is not reported in the registry record.
12. Why This Trial Matters Statistically
Reliable feature detection is a measurement property, and measurement consistency matters when image-based assessments are used in research or clinical workflows. This study's statistical structure recognizes that consistency may depend on where scanning occurs and which scanner is used. Separating within-scanner, between-scanner, and between-site assessments is useful because each condition represents a distinct reproducibility question.
The use of Average Positive Agreement Rate also makes the target of measurement explicit. It focuses on agreement around positive feature detection rather than treating positive and negative agreement as interchangeable. That specificity can help readers understand what the endpoint is intended to capture. At the same time, it means the endpoint should not be treated as a complete measure of overall classification agreement or as a measure of correctness against a reference standard.
Statistically, repeated reads and scans can create related observations. In general, this dependence matters when estimating uncertainty: an analysis that treats correlated assessments as independent may not represent uncertainty appropriately. The registry record does not report how repeated measurements or clustering by readers, scanners, or sites were handled, so readers cannot assess that aspect of the analysis.
The study is marked completed, but the registry record does not report the primary endpoint results. This distinction is important: a clearly specified validation question is not equivalent to evidence about the observed level of reproducibility. Without estimates and confidence intervals, neither the magnitude of agreement nor its precision can be assessed. The study is statistically relevant for its measurement framework, while the absence of posted results limits the conclusions that can be drawn from its registry record.
13. 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 interpreting this trial is that the registry record defines the primary endpoint but does not report its estimate or 95% confidence interval. No numerical results are listed for intra-scanner, inter-scanner, or inter-site precision, so the observed agreement and its precision cannot be evaluated.
The study description is incomplete for the inter-site substudy: it identifies 378 fields of view scanned on 1 scanner at different sites, but does not fully describe the reading arrangement. The record does not report the number of sites or clarify the complete analysis population. It also does not report how many scans, reads, or features contributed to each analysis.
Statistical methods beyond the endpoint formula and named confidence interval are not reported in the registry record. The record does not describe the confidence-interval calculation, handling of repeated or clustered assessments, missing data, exclusions, or multiplicity. It lists 0 analyses, which means no analysis results appear in the record, but does not establish whether analysis occurred elsewhere.
The enrollment figure of 540 does not clarify the endpoint-specific denominators. The adverse-event entries list zeros for at-risk and affected counts and deaths, but do not describe an at-risk population or provide details for interpreting those entries. The record's planning assumptions are illustrative and do not establish the study's own sample-size justification. Together, these reporting gaps prevent a detailed appraisal of the results or their statistical robustness.
14. Related Tutorials
15. Related Statistical Calculators
16. Sources
- ClinicalTrials.gov: Feature Detection Study for Validation of Hamamatsu NanoZoomer S360MD Digital Slide Scanner System — NCT03992768.
- Registry record last updated 2022-12-22; status last verified 2022-11.
- Results first posted 2022-12-22.
- 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.
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.
17. Record Summary
NCT03992768 is a completed, single-group device study sponsored by Hamamatsu Photonics K.K. It evaluated repeatability and reproducibility of histological feature detection using whole slide imaging with the Hamamatsu NanoZoomer S360MD Digital Slide Scanner System. Enrollment was 540; the study start date was 2019-06-12, and the primary completion date was 2021-09-30.
The primary measure was Average Positive Agreement Rate, with a 95% confidence interval, for within-scanner, between-scanner, and between-site assessments. The record reports the endpoint definition and some substudy procedures, but no primary outcome estimates or analyses. It also does not fully describe the inter-site population or report endpoint-specific analysis details. The posted registry results therefore do not establish the observed level or precision of agreement.