(302 days)
The EasyCell is intended to locate and display images of white cells, red cells, and platelets acquired from fixed and stained peripheral blood smears and assists a qualified technologist in conducting a WBC differential, RBC morphology evaluation, and platelet estimate using those images. For in vitro diagnostic use only. For professional use only.
The EasyCell automatically locates and presents images of blood cells on peripheral smears. The operator reviews the suggested classification of each white cell according to type and may manually change the suggested classification of any cell. The operator can characterize red cell morphology and estimate platelets based on observed images. The EasyCell is intended to be used by skilled operators, trained in the use of the device and in the identification of blood cells.
This document describes the performance characteristics and acceptance criteria for the EasyCell Automated Cell-Locating Device (K092116).
1. Table of Acceptance Criteria and Reported Device Performance
The acceptance criteria are implicitly derived from the "Correlation between Reference Method and Test Method" and "Sensitivity and Specificity" tables provided in the submission. The device is deemed acceptable if its performance is "good" or "equivalent" to the reference method. Specific quantitative targets for correlation coefficients, sensitivity, and specificity should have been defined as acceptance criteria but are not explicitly stated as such. Based on the provided data, the device meets these implicit criteria.
| Metric | Acceptance Criteria (Implied) | Reported Device Performance (Full Sample) | Reported Device Performance (100 vs. 200 cells) |
|---|---|---|---|
| WBC Differential (Correlation Coefficient r) | ≥ 0.90 for Neutrophils, Lymphocytes, Eosinophils; ≥ 0.80 for Monocytes | Neutrophil: 0.99 | |
| Lymphocyte: 0.98 | |||
| Monocyte: 0.93 | |||
| Eosinophil: 0.97 | Neutrophil: 0.96 | ||
| Lymphocyte: 0.95 | |||
| Monocyte: 0.83 | |||
| Eosinophil: 0.93 | |||
| WBC Differential (Slope) | Close to 1.00 | Neutrophil: 1.00 | |
| Lymphocyte: 1.00 | |||
| Monocyte: 0.85 | |||
| Eosinophil: 0.95 | Neutrophil: 0.99 | ||
| Lymphocyte: 1.00 | |||
| Monocyte: 0.83 | |||
| Eosinophil: 0.99 | |||
| WBC Differential (Intercept) | Close to 0.00 | Neutrophil: 0.39 | |
| Lymphocyte: 0.88 | |||
| Monocyte: 0.59 | |||
| Eosinophil: 0.066 | Neutrophil: -0.15 | ||
| Lymphocyte: 0.73 | |||
| Monocyte: 0.72 | |||
| Eosinophil: -0.01 | |||
| Efficiency (% agreement) | Not explicitly specified, assumed to be high | Overall: 84% | 200 cells: 84% |
| 100 cells: 83% | |||
| Sensitivity | Not explicitly specified, assumed to be high | Overall: 91% | 200 cells: 91% |
| 100 cells: 90% | |||
| Specificity | Not explicitly specified, assumed to be high | Overall: 72% | 200 cells: 72% |
| 100 cells: 70% | |||
| Platelet Estimate Accuracy | "Good agreement" (Cohen's kappa) | "Good agreement" between methods (Cohen's kappa statistic calculated, but value not provided) | N/A |
| Red Blood Cell Morphology Accuracy | "Good agreement" (e.g., >90%) | >90% agreement between methods | N/A |
| Between Run Precision | Equivalent to Reference Method | "Test Method has equivalent precision to the Reference Method." | N/A |
2. Sample Size Used for the Test Set and Data Provenance
- Sample Size: A total of 304 specimens were collected and analyzed. This sample included:
- 155 specimens from normal (healthy) subjects.
- 149 specimens from subjects with specific disease conditions.
- Data Provenance: The data was collected and analyzed at three sites. The country of origin is not explicitly stated, but the submission is to the US FDA, implying clinical sites within the US or compliant with US standards. The study design is prospective, as specimens were collected and analyzed for this method comparison.
3. Number of Experts Used to Establish the Ground Truth for the Test Set and Qualifications of Those Experts
- Number of Experts: The ground truth (Reference Method) for the test set was established by two trained technologists at each of the three sites. This totals 6 experts for the initial reading, though the independent analysis by two technologists per site suggests independent assessments that were likely compared or adjudicated.
- Qualifications of Experts: The experts are described as "trained technologist" or "skilled operators, trained in the use of the device and in the identification of blood cells." Specific experience levels (e.g., years of experience or board certification) are not provided.
4. Adjudication Method for the Test Set
The document states that slides were "randomly selected, blinded and read by two technologists at each site." It does not explicitly mention an adjudication method (e.g., 2+1 or 3+1). It is implied that the readings from the two technologists at each site formed the "Reference Method" for comparison, but how discrepancies between these two readers were resolved (if at all) is not detailed.
5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study
An MRMC study, comparing human readers with AI assistance versus without AI assistance, was not explicitly detailed in the provided text. The study design focuses on comparing the EasyCell (with human operator review) to a manual reference method. The "Test Method" described involves an operator reviewing and potentially reclassifying the suggested classifications from the EasyCell. Therefore, the reported performance is already that of the human-in-the-loop system, not "without AI assistance." As such, an effect size of how much human readers improve with AI vs. without AI assistance is not presented.
6. Standalone (i.e., algorithm only without human-in-the-loop performance) Study
A standalone performance study of the algorithm without human-in-the-loop review was not conducted or reported. The device description explicitly states: "The operator reviews the suggested classification... and may manually change the suggested classification of any cell." The "Test Method" in the performance study is defined as "EasyCell using Examiner Review." This indicates that all reported performance metrics include the human element of review and potential reclassification, making it a human-in-the-loop system, not a standalone algorithm.
7. Type of Ground Truth Used
The ground truth used for comparison was expert consensus / manual microscopic examination. Specifically, it is referred to as the "Reference Method," which is described as "a manual differential microscopic examination of the peripheral blood slides by a trained technologist."
8. Sample Size for the Training Set
The sample size for the training set is not provided in the given excerpts. The document describes the "ANN's trained to distinguish between classes of white blood cells" but does not give details about the data used for this training.
9. How the Ground Truth for the Training Set was Established
The method for establishing the ground truth for the training set is not explicitly stated in the provided excerpts. It can be inferred that it would involve expert classification of blood cells, similar to the reference method for the test set, but specific details are absent.
§ 864.5260 Automated cell-locating device.
(a)
Identification. An automated cell-locating device is a device used to locate blood cells on a peripheral blood smear, allowing the operator to identify and classify each cell according to type. (Peripheral blood is blood circulating in one of the body's extremities, such as the arm.)(b)
Classification. Class II (performance standards).