(315 days)
The White IRIS is a leukocyte differential analyzer intended for in vitro diagnostic use in determining the proportional leukocyte count (WBC differential) on peripheral blood specimens that have been flagged by an automated hematology analyzer performing differential counts as well as for peripheral blood specimens for which no automated differential has been performed.
Combining unique cytoprobe, rapid hemacyte fractionation and novel color image analysis, The White IRIS extends automated intelligent microscopy to leukocyte differentiation. It provides flow cytometry precision and microscopical resolution to review specimens flagged by hematology analyzers with differential capabilities or to complement other analyzers without these capabilities. The system includes compartments for closed sampling, rapid leukocyte-rich plasma preparations, cytoprobeinduced metachromasia, and collection and color analysis of leukocyte images, and presents the results as a single-view 500-cell differential on a 20-in touchscreen monitor for examination by a skilled competent observer.
Acceptance Criteria and Device Performance Study for The White IRIS
1. Table of Acceptance Criteria and Reported Device Performance
The provided documentation does not explicitly state formal acceptance criteria with numerical targets for each performance metric. However, the study aims to demonstrate substantial equivalence to the predicate method (Wright-stain/light microscope) in terms of accuracy, clinical sensitivity, and precision. The "acceptance criteria" are thus inferred from the comparisons made and the overall conclusion of "results equal to or better than the reference method."
| Performance Metric | Implied Acceptance Criterion (vs. Wright-Stain Reference) | The White IRIS Reported Performance |
|---|---|---|
| Accuracy | Correlation similar to or better than the reference; Slopes near 1, intercepts near 0 for leukocyte types. | - Correlations (r): Neutrophils (0.98), Lymphocytes (0.98), Monocytes (0.91), Eosinophils (0.98), Basophils (0.87), Segmented Neutrophils (0.93), Bands (0.54), Metamyelocytes (0.84), Myelocytes (0.74), Promyelocytes (0.97), Blasts (0.96), Normal Lymphocytes (0.98), Variant Lymphocytes (0.41). |
| - Overall Agreement (Abnormal Classification): 91.89% (93.51% after removing specific disagreement cases within 95% CI). (Table 9) |
- Agreement (Pre-removal of CI cases): 82.95% (Table 11)
- Cohen's Kappa (4x4 Classification): 0.49 (rejects random agreement). (Table 12) |
| Precision | Standard deviations similar to or better than the reference method. | - SD compared to Wright-Stain (H20-A method): Generally lower or comparable SDs for most cell types (e.g., Neutrophils: 2.72 vs 3.55; Lymphocytes: 3.17 vs 3.68; Monocytes: 1.62 vs 2.29). - Within-run Precision (22 replicates): Neutrophils (± 5.0%), Lymphocytes (± 5.0%), Monocytes (± 2.5%), Eosinophils (± 1.5%), Basophils (± 1.0%) at 95% CI. (Table 6) |
| Clinical Sensitivity | Ability to correctly identify abnormal specimens (defined by H20-A criteria and reference ranges) with comparable or better performance than the reference method. | - The study evaluated 1202 specimens, including those required by H20-A for various abnormal conditions (e.g., Granulocytosis, Monocytosis, Eosinophilia, Lymphocytosis, Immature Cells). - False Negative Rate: 4.79% (2.98% after specific removals). (Table 9)
- False Positive Rate: 15.36% (14.12% after specific removals). (Table 9) |
2. Sample Size Used for the Test Set and Data Provenance
- Sample Size:
- Accuracy: 1,202 normal and abnormal patient specimens (for correlation and overall agreement analysis).
- Precision (H20-A calculation): 1,014 specimens for The White IRIS; 1,277 specimens for Wright-stain.
- Precision (within-run): 22 replicates of the same sample.
- Clinical Sensitivity: 1,202 Leukocyte Differential Summaries (the same set as for accuracy).
- Data Provenance: Not explicitly stated regarding country of origin. The study is described as taking place in a clinical laboratory setting, comparing results to a standard reference method. It indicates a retrospective/cross-sectional design as it involves collecting specimens and analyzing them using both the new device and the predicate method for comparison.
3. Number of Experts Used to Establish the Ground Truth for the Test Set and Their Qualifications
- Number of Experts: Two skilled technologists.
- Qualifications: "Skilled technologists." The document also mentions "competent human observer... well skilled in both the use of the instrument and in the recognition of leukocyte classes." This implies trained medical laboratory professionals with expertise in manual leukocyte differential counting.
4. Adjudication Method for the Test Set
The primary reference method involves two skilled technologists reading manually prepared and observed Wright-stained smears. The analysis directly compares The White IRIS's results (reviewed by these same two technologists) against this "reference method."
While not explicitly stated as a formal "adjudication" in the sense of a third reader resolving discrepancies, the predicate method itself inherently relies on consensus or comparison between the two technologists' readings (implied when the "Wright-Stain Mean (%)" is presented). The study then compares the device's output, post human review, to this established reference.
5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study
- Was it done? Yes, in a limited sense. The study compares the performance of human readers (the two skilled technologists) when using The White IRIS (assisted interpretation) versus their performance using the traditional Wright-stain/light microscope method (unassisted interpretation).
- Effect Size (Improvement with AI vs. without AI assistance): The study reports on the "efficacy" of The White IRIS, highlighting that its use "provides results equal to or better than the reference method with less time expenditure and less biohazard exposure." However, it does not provide a quantifiable effect size specifically on how much human readers improve their accuracy or efficiency when using The White IRIS compared to their standalone performance. It mainly demonstrates equivalence or superiority of the combined system (IRIS + human review) to the manual method. For example, precision (SD) for most cell types is better with The White IRIS.
6. Standalone (Algorithm Only) Performance
- Was it done? No. The White IRIS explicitly requires a "skilled competent observer" to confirm or modify the classification of each cell. The device provides "presumptive" classifications, but human review is an integral part of its intended use. Therefore, a standalone algorithm-only performance assessment was not conducted, as it would not represent the intended clinical workflow.
7. Type of Ground Truth Used
- Expert Consensus / Reference Method: The ground truth for the test set was established by the Wright-stain/light microscope reference method, performed by two skilled technologists. This is a widely accepted, "Approved Standard" (NCCLS document H20-A) in hematology.
8. Sample Size for the Training Set
- The document does not specify a separate training set size. The reported studies (accuracy, precision, clinical sensitivity) are all described as evaluations of "The White IRIS" which implies the full, ready-for-use device. The technology involves "color image analysis" and classifies cells "presumptively," suggesting an underlying algorithm, but there's no mention of a distinct training phase or dataset with its own associated ground truth establishment.
9. How the Ground Truth for the Training Set Was Established
- As no separate training set is explicitly mentioned or analyzed in the provided text, information on how its ground truth was established is not available. The document focuses on the validation of the final device-with-human-in-the-loop system against the established predicate method.
§ 864.5220 Automated differential cell counter.
(a)
Identification. An automated differential cell counter is a device used to identify one or more of the formed elements of the blood. The device may also have the capability to flag, count, or classify immature or abnormal hematopoietic cells of the blood, bone marrow, or other body fluids. These devices may combine an electronic particle counting method, optical method, or a flow cytometric method utilizing monoclonal CD (cluster designation) markers. The device includes accessory CD markers.(b)
Classification. Class II (special controls). The special control for this device is the FDA document entitled “Class II Special Controls Guidance Document: Premarket Notifications for Automated Differential Cell Counters for Immature or Abnormal Blood Cells; Final Guidance for Industry and FDA.”