Search Results

The ADVIA 2120 and ADVIA 2120i with autoslide are quantitative, automated hematology analyzers that provide the following information for in vitro diagnostic use in clinical laboratories:

• A complete blood count (CBC) consisting of WBC, RBC, Hgb, CN-Free Hgb, Calculated Hgb, MCV, Hct, MCH, MCHC, CHCM, RDW, HDW, CH, Plt, MPV.
• A leukocyte differential count consisting of Neut (%/#), Lymph (%/#), Mono (%/#), Eos (%/#) Baso (%/#), LUC (%/#).
• A reticulocyte analysis consisting of Retic (%/#), MCVg, MCVr, CHCMg, CHCMr, CHg, CHr.
• A nucleated red blood cell count consisting of NRBC(%/#).
• Enumeration of the total nucleated cell (TNC) count and RBC count for pleural, peritoneal, and peritoneal dialysis (PD) specimens.
  Note: Above measurands are determined (in whole blood, pleural, peritoneal, or peritoneal dialysis specimens with K2 and/or K3 EDTA anti-coagulants).
• Quantitative determination of blood cells in Cerebrospinal Fluid (CSF) consisting of WBC, RBC, Neut (%/#), Lymph (%/#), Mono (%/#), MN (%/#),PMN (%/#).
  In addition, the system provides the added capability to automatically prepare and stain high quality blood smears on a microscope slide.

The ADVIA 2120/210i Hematology systems with Auto slide are an integrated option of a Hematology analyzers with complete blood cell count, leukocyte differential cell count, reticulocyte analysis capability, nucleated red blood cell count, quantitative determination of blood cells in Cerebrospinal Fluid (CSF), enumeration of the total nucleated cell (TNC) count and RBC count for pleural, peritoneal, and peritoneal dialysis (PD) specimens and a slide stainer designed to provide reflexive slide making/staining without user intervention based upon pre-selected, user-definable criteria.
The ADVIA 2120/210i Hematology systems with Auto slide consists of the following: an analytical module that aspirates, dilutes, and analyzes whole blood samples; an auto sampler that automatically mixes, identifies, and presents the samples for processing; a computer workstation that controls the instrument, provides primary user interface with the instrument and manages the data produced by the instrument; a printer that optionally generates reports based on the instrument results and an auto slide module that prepares a wedge smear from a drop of blood, places it on a microscope slide and stains the slide in accordance with Wright, Wright-Giemsa and May-Grnwald Giemsa Staining techniques.

The provided text is a 510(k) Summary for the ADVIA® 2120/2120i Hematology auto-analyzers. This document focuses on demonstrating substantial equivalence to a predicate device, which means the new device is as safe and effective as a legally marketed device. It does not describe a study that proves the device meets specific acceptance criteria in the way you might expect for a novel AI device with a defined set of performance metrics.

Instead, the submission shows the new device with an ARM9 CPU board performs similarly or identically to the predicate device (the ADVIA 2120/2120i with the current CPU board) across various specifications. The "acceptance criteria" here are essentially the established performance characteristics of the predicate device, and the "study" is the comparison data presented to support substantial equivalence.

Here's an attempt to answer your questions based on the provided text, acknowledging that some information you requested (like AI-specific details, ground truth establishment for a training set, and expert adjudication for a test set) are not relevant or present in this type of FDA submission for a hardware/firmware upgrade to an existing analyzer.

1. A table of acceptance criteria and the reported device performance

The acceptance criteria are implied to be the performance characteristics of the predicate device. The "reported device performance" is the expectation that the new device (AVIA 2120/2120i with ARM9 CPU) will exhibit identical performance.

• RBC (10⁶/μL): 0.0 to 7.0 (Max Deviation: 0.1)
• HGB (g/dL): 0 to 22.5 (Max Deviation: 0.2 or 2.0%)
• PLT (10³/μL): 5.0 to 3500 (Max Deviation: 5.0 or 5.0%)
• %RETIC: 0.2 to 24.5 (Max Deviation: 5.0%)
• CN-free HGB (g/dL): 1 to 22.5 (Max Deviation: 0.3 or 3.0%)
• CSF WBC (cells/µL): 0 to 50 (Max Deviation: 5)
• CSF RBC (cells/µL): 50 to 5000 (Max Deviation: 10%)
• BF TNC (10³/µL): 0 to 50 (Max Deviation: 5)
• BF RBC (10⁶/µL): 50 to 1500 (Max Deviation: 10%) | Same |
  | Within-Run Precision| (See detailed table in original text; e.g., WBC: Nominal 7.5, SD 0.2, CV 2.66%) | Same |
  | Carryover | Less than or equal to 1% | Same |
  | Physical/Electrical | (Various detailed specifications for Temperature, Humidity, Noise, Weight, Dimensions, Vacuum/Pressure, Reaction Chamber Temp, Power Pack Temp, Light Intensities, Power Supply Voltages, Sample Mode Volumes, Throughput, Sample Capacity, Tube Sizes/Types, Barcode Reader functionality) | Same |
  | Data Management | TDC version 9 or higher, Database storage, Review/edit, User-defined reports/ranges, Bi-directional communication, QC features, ILQC programs, User assistance | Same |
  | Consumables/Reagents| CBC TIMEPAC Baso HGB RBC/PLT Defoamer CN-Free CBC TIMEPAC; DIFF TIMEPAC, Perox 1, 2, 3, Perox Sheath, autoRetic, EZ KLEEN, Sheath/Rinse, CSF | Same |
  | Calibrators | ADVIA OPTIpoint, ADVIA SETpoint | Same |
  | Controls | ADVIA TESTpoint Low/Normal/High, Retic Low/High, 3-in-1 Abnormal1/Normal/Abnormal2 | Same |

2. Sample sizes used for the test set and the data provenance (e.g., country of origin of the data, retrospective or prospective)

The document lists performance specifications (linearity, precision) but does not specify the sample sizes or data provenance (country, retrospective/prospective) used to establish these predicate device performance characteristics. The context is that the new device's performance is expected to be identical to the established predicate performance, implying that these performance metrics have been previously validated and are being maintained.

3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts (e.g., radiologist with 10 years of experience)

This is not applicable to this submission. The "ground truth" for hematology analyzers is typically established through reference methods, calibrated controls, and comparison to established, validated manual methods, not through expert consensus in the way an imaging AI algorithm's ground truth might be. The document focuses on the technical specifications and equivalence of a hardware component change.

4. Adjudication method (e.g., 2+1, 3+1, none) for the test set

Not applicable. This type of submission does not involve adjudication of diagnostic decisions as would be relevant for an AI algorithm.

5. If a multi-reader multi-case (MRMC) comparative effectiveness study was done, If so, what was the effect size of how much human readers improve with AI vs without AI assistance

Not applicable. This is for a hematology auto-analyzer, not an AI-assisted diagnostic tool for human readers.

6. If a standalone (i.e. algorithm only without human-in-the-loop performance) was done

The device itself is a standalone automated hematology analyzer. The submission is not for a new algorithm, but for a hardware (CPU board) and associated software upgrade to an existing analyzer. The performance characteristics presented are those of the entire automated system.

7. The type of ground truth used (expert consensus, pathology, outcomes data, etc)

Not explicitly stated for the predicate device's original validation. However, for hematology analyzers, "ground truth" for parameters like cell counts, hemoglobin, etc., is typically established using:

• Reference methods: Highly accurate and precise laboratory methods.
• Calibrated materials: Use of control materials with known values traceable to international standards.
• Manual microscopy or other established techniques: For differential counts or specific cell morphology, comparison to manual review by highly trained laboratorians.

8. The sample size for the training set

Not applicable. This device does not use a "training set" in the context of machine learning. It's a change to a pre-programmed analytical instrument.

9. How the ground truth for the training set was established

Not applicable, as there is no "training set" in the AI sense.

Ask a specific question about this device

The RET-He parameter on the Sysmex® XE-2100, Automated Hematology Analyzer, determines the hemoglobin of reticulocytes for in vitro diagnostic use in clinical laboratories.

The XE-2100 is an automated hematology analyzer previously cleared by the FDA. The RET-He parameter determines the hemoglobin of the reticulocytes. (Note: XE-Pro and RET Master are required to obtain results described.)

Here's an analysis of the acceptance criteria and the study proving the device meets them, based on the provided text:

1. Table of Acceptance Criteria and Reported Device Performance

2. Sample Size Used for the Test Set and Data Provenance

The provided text does not explicitly state the sample size used for the test set or the data provenance (e.g., country of origin, retrospective or prospective). It only mentions that "Studies were performed to evaluate the equivalency of the RET-He parameter to the predicate method."

3. Number of Experts Used to Establish Ground Truth and Their Qualifications

The provided text does not mention using experts to establish ground truth or their qualifications. The study focused on comparing the new device's output (RET-He) to a predicate device's output (CHr).

4. Adjudication Method for the Test Set

The provided text does not describe an adjudication method. Since the study compared a device's output to a predicate device's output, and not to a human-established ground truth, adjudication would likely not be relevant in the same way it is for image interpretation tasks.

5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study Was Done

No, a Multi-Reader Multi-Case (MRMC) comparative effectiveness study was not done. This type of study is more common for diagnostic devices where human interpretation plays a significant role and AI aims to augment that interpretation. This submission concerns a hematology analyzer, where the device performs the measurement without human-in-the-loop interpretation.

6. If a Standalone (Algorithm Only Without Human-in-the-Loop Performance) Was Done

Yes, the study described is effectively a standalone (algorithm only without human-in-the-loop performance) study. The RET-He parameter is determined by the Sysmex XE-2100 analyzer using its internal methodology, and its performance was evaluated by comparing its results directly to those of the predicate device (Bayer Advia 120). There is no indication of human intervention in interpreting the results from either device for the purpose of this comparison.

7. The Type of Ground Truth Used

The ground truth used in this study was the output of the predicate device (CHr parameter on the Bayer Advia 120 Hematology System). The study aimed to demonstrate "excellent correlation" and "equivalent performance" of the new RET-He parameter to this established predicate method.

8. The Sample Size for the Training Set

The provided text does not specify a sample size for any training set. This type of device (hematology analyzer) likely has its algorithms and parameters established during its initial development and validation, rather than through a separate "training set" in the context of machine learning. The submission focuses on the performance comparison of the developed device.

9. How the Ground Truth for the Training Set Was Established

Since no specific "training set" is mentioned in the context of this 510(k) submission, the question of how ground truth for it was established is not applicable based on the provided text. The device's methodology for deriving reticulocyte parameters is described ("using the reticulocyte forward scattered light signals from the reticulocyte measurement channel and a proprietary Sysmex calculation equation"), implying a deterministic process rather than a machine learning model trained on a ground-truthed dataset.

Ask a specific question about this device