(392 days)
The HemoScreen is a point-of-care (POC) automated hematology analyzer intended for the enumeration and classification of the following parameters in capillary and venous whole blood (K2EDTA anticoagulated): WBC, RBC, HCT, MCV, MCH, MCHC, RDW, PLT, MPV, NEUT%, NEUT#, LYMP%, LYMP#, MONO%, MONO#, EO%, EO#, BASO%, and BASO#. The HemoScreen is for in vitro diagnostic use in clinical laboratories and/or POC settings for adults and children at least 2 years of age.
HemoScreen is a point of care (POC), automated hematology analyzer that provides 20 common CBC parameters, including a 5-part leukocyte (WBC) differential, in capillary and venous whole blood samples. The HemoScreen analyzer (reader) is a tabletop device that is designed to use with a disposable reagent Cartridge. In addition to the Cartridge, the system includes a disposable Sampler with two glass capillaries which is used to collect the blood sample and then transfer it to the Cartridge.
Once the Cartridge is inserted into the reader, there are no further procedural steps; blood is expelled from the capillaries (Sampler) into the reagent compartments (Cartridge). The reader then mixes the blood sample with the reagents by alternately pressing compressible portions of the Cartridge, eventually causing the suspension of cells to flow into the microfluidic chamber. Cells flowing in the microfluidic chamber focus into a single-cell plane due to a patented physical phenomenon known as viscoelastic focusing.
The reader then captures images of the focused cells and analyzes them in real time using machine vision algorithms. When analysis is complete, the results are displayed to the user on the reader's touch screen and may be printed to an adjacent printer or exported to a USB flash drive. The Cartridge is ejected by the analyzer after analysis, and can then be safely disposed of, as the reagents and blood sample remain within the Cartridge.
The basic staining and microscopic image analysis performed by HemoScreen closely resembles the traditional blood smear and the hemocytometer counting chamber. Leukocytes are classified based on their staining properties and morphology, whereas absolute counts are obtained by counting the cells contained in a chamber of predetermined volume. Test results are obtained within six (6) minutes and the results are saved.
Quality Control: Commercial 3-level liquid quality controls, PIX-CBC Hematology Controls, are recommended for use with the HemoScreen. These controls cover all the tested parameters and are sampled the same way whole blood is sampled.
Software: The HemoScreen software displays an intuitive, simple-to-use user interface that is operated via the touch screen. The software is responsible for operating the device, performing the measurements, and recording the results.
The provided text describes the HemoScreen Hematology Analyzer's performance and the study conducted to demonstrate its substantial equivalence. The key change evaluated was the introduction of direct fingerstick sampling, compared to the previously cleared indirect fingerstick sampling method.
Here's a breakdown of the requested information based on the provided text:
1. Table of Acceptance Criteria and Reported Device Performance
The document does not explicitly state "acceptance criteria" in a numerical or categorical format for each parameter. Instead, it relies on correlation and regression analysis to show agreement between the new direct sampling method and the previously cleared indirect sampling method. The implicit acceptance criteria are likely based on acceptable Passing-Bablok regression parameters (slope and intercept ideally close to 1 and 0, respectively) and high Pearson Correlation coefficients.
However, the reported performance is provided as follows:
Performance of HemoScreen (Direct Fingerstick) vs. HemoScreen (Indirect Fingerstick) for 20 Parameters
| Parameter | N | Result Range (Indirect) | Intercept [95% CI] | Slope [95% CI] | Pearson Correlation |
|---|---|---|---|---|---|
| WBC (10³/μL) | 42 | 4.54-12.52 | -0.216 [-0.893, 0.412] | 1.028 [0.955, 1.108] | 0.978 |
| RBC (10⁶/μL) | 42 | 4.14-5.78 | 0 [-0.401, 0.498] | 0.994 [0.881, 1.078] | 0.97 |
| HGB (g/dL) | 42 | 11.84-17.12 | -0.749 [-2.348, 1.097] | 1.048 [0.913, 1.156] | 0.969 |
| HCT (%) | 42 | 35.76-50.11 | -1.77 [-7.048, 3.527] | 1.032 [0.901, 1.157] | 0.967 |
| MCV (fL) | 42 | 76.6-93.84 | 1.013 [-2.325, 3.887] | 0.988 [0.954, 1.025] | 0.994 |
| MCH (pg) | 42 | 24.67-32.98 | -0.191 [-0.907, 0.605] | 1.007 [0.981, 1.032] | 0.998 |
| MCHC (g/dL) | 42 | 32.2-35.83 | -1.519 [-4.767, 1.859] | 1.047 [0.949, 1.143] | 0.961 |
| RDW (%) | 42 | 11.64-14.16 | -0.25 [-0.876, 0.266] | 1.019 [0.977, 1.068] | 0.991 |
| PLT (10³/μL) | 42 | 142.4-399.6 | 3.646 [-16.602, 21.01] | 0.963 [0.894, 1.051] | 0.979 |
| MPV (fL) | 42 | 9.18-13.46 | 0.572 [-1.036, 1.837] | 0.949 [0.832, 1.113] | 0.92 |
| NEUT (10³/μL) | 42 | 2.19-7.95 | -0.221 [-0.531, 0.062] | 1.049 [0.995, 1.109] | 0.984 |
| LYMP (10³/μL) | 42 | 1.51-4.17 | -0.153 [-0.562, 0.251] | 1.075 [0.925, 1.273] | 0.953 |
| MONO (10³/μL) | 42 | 0.27-0.86 | -0.087 [-0.256, 0.023] | 0.996 [0.796, 1.358] | 0.825 |
| EO (10³/μL) | 42 | 0.04-0.53 | -0.021 [-0.032, -0.005] | 1.017 [0.945, 1.081] | 0.988 |
| BASO (10³/μL) | 42 | 0.01-0.07 | 0.003 [-0.01, 0.012] | 1.225 [0.842, 1.659] | 0.598 |
| NEUT (%) | 42 | 43.75-68.8 | -2.615 [-11.774, 4.007] | 1.05 [0.941, 1.194] | 0.957 |
| LYMP (%) | 42 | 19.95-45.05 | -1.418 [-4.219, 1.404] | 1.073 [0.983, 1.162] | 0.969 |
| MONO (%) | 42 | 3.95-11.25 | -2.108 [-3.977, -0.95] | 1.147 [0.985, 1.459] | 0.823 |
| EO (%) | 42 | 0.7-7.3 | -0.27 [-0.429, -0.028] | 1.029 [0.968, 1.097] | 0.987 |
| BASO (%) | 42 | 0.1-0.7 | -0.075 [-0.3, 0.15] | 1.5 [1, 2] | 0.614 |
2. Sample size and Data Provenance
- Test Set Sample Size: 42 subjects.
- Data Provenance: The study described is a "prospective clinical study" involving "matched fingerstick collections." The document does not explicitly state the country of origin, but the company (PixCell Medical Technologies, Ltd.) is based in Israel. Given the FDA submission, it's likely the study was conducted to meet US regulatory requirements, but the specific location isn't provided.
3. Number of Experts and Qualifications for Ground Truth
The document does not mention the use of experts to establish a "ground truth" in the traditional sense of consensus reading for images. The study compares two different methods of sampling with the same device (HemoScreen) – direct vs. indirect fingerstick. The "ground truth" for the comparison is essentially the measurement obtained by the indirect sampling method, which was previously cleared. There are no mentions of radiologists, pathologists, or similar experts reviewing cases to establish a ground truth for a diagnostic AI.
4. Adjudication Method for the Test Set
Not applicable. This was a method comparison study for quantitative measurements, not a diagnostic AI study requiring adjudication of expert interpretations.
5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study
No, a multi-reader multi-case (MRMC) comparative effectiveness study was not conducted. This study focused on comparing two sampling methods for a highly automated hematology analyzer, not on human readers' interpretive performance with or without AI assistance.
6. Standalone (Algorithm Only) Performance
The HemoScreen device itself is described as using "machine vision algorithms" to analyze images of cells. Therefore, the reported performance metrics (Pearson correlation, Passing-Bablok regression) represent the standalone performance of the device's algorithms in quantifying blood parameters. The study specifically compared the input method (direct vs. indirect fingerstick) to the device, not an AI assisting a human.
7. Type of Ground Truth Used
The ground truth used for performance comparison was the quantitative measurements obtained from the HemoScreen device itself, using the indirect fingerstick sampling method. This method was previously cleared (K180020) and served as the reference for evaluating the new direct sampling method. It's a "device-to-device" or "method-to-method" comparison, rather than comparison to a clinical expert consensus, pathology, or outcomes data.
8. Sample Size for the Training Set
The document does not explicitly describe a "training set" for the "machine vision algorithms." The algorithms are inherent to the device's operation. It's possible the algorithms were developed and validated internally during the device's initial design and prior K180020 clearance, but this specific submission focuses on validating the direct fingerstick sampling method rather than training new algorithms or updating existing ones. Thus, the sample size for a training set is not provided in this document.
9. How Ground Truth for Training Set Was Established
As mentioned above, information regarding a specific "training set" and its ground truth establishment is not provided in this document. The existing HemoScreen device's core technology and algorithms were already cleared under K180020. This submission primarily validates a new sample collection modality for the same device and its existing algorithms.
§ 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.”