Search Results

The Sight OLO is a quantitative multi-parameter automated hematology analyzer intended for in vitro diagnostic use in screening capillary or venous whole blood samples collection tubes, or fingertip samples collected using the Sight OLO test kit micro-capillary tubes.

When used with the Sight OLO cartridge, the Sight OLO utilizes computer vision algorithms to enumerate the following CBC parameters in whole blood: WBC, RBC, HCT, MCV, MCH, MCHC, RDW, PLT, NEUT%/#, LYMPH %/#, MONO %/#, EOS%/#, and BASO%/#.

The Sight OLO is indicated for use by clinical laboratories to identify and classify one or more of the formed elements of blood in children 3 months and above, adolescents and adults.

The Sight OLO device is a computer vision based platform for blood analysis. The platform combines computer-vision algorithms for image processing to identify and quantify blood components (e.g., red blood cells) and their characteristics (e.g., cell volume) in an automated fashion. Using dedicated staining, the proposed platform provides complete blood count analysis. The Sight OLO is a compact device, designed to be automated and simple to operate, to enable rapid testing and analysis. The Sight OLO consists of a scanning and analyzing device and a CBC test kit, including a disposable cartridge and sample preparation tools. The disposable cartridge containing the blood sample is loaded into the device through the loading slot. The device is operated through the touch screen interface.

The Sight OLO provides complete blood count information with 5-part differentials for white blood cell types. Specifically, the CBC parameters measured by the Sight OLO are listed below and include: WBC, RBC, HGB, HCT, MCV, MCH, MCHC, RDW, PLT, NEUT%/#, LYMPH %/#, MONO %/#, EOS%/# and BASO%/#. In addition, the Sight OLO signals specific WBC abnormal cases by flagging the sample.

The original text provided is a 510(k) Premarket Notification from the FDA regarding the Sight OLO device. It details the device's technical specifications, indications for use, and performance data used to establish substantial equivalence to a predicate device.

Here's an analysis of the acceptance criteria and study data based on the provided text:

Acceptance Criteria and Reported Device Performance

The core of the acceptance criteria in this submission appears to be demonstrating substantial equivalence to a predicate device (Sight OLO K190898) and a reference device (Sysmex XN-Series Hematology Analyzer, K112605) for various blood parameters and flagging capabilities. The performance is assessed primarily through method comparison studies and flagging studies.

Table of Acceptance Criteria (Implied) and Reported Device Performance:

The document explicitly states that "all measurands met the prespecified acceptance criteria for correlation, bias, slope, intercept (and the 95% two-sided confidence interval (CI) around the slope and intercept)" for the method comparison. For flagging, it states the "overall flagging capabilities of the Sight OLO device met the predefined acceptance criteria for both sensitivity and specificity." While the exact numerical criteria for each parameter's correlation, slope, intercept, and bias are not explicitly listed as "acceptance criteria" but rather the results that met them, the table below reflects the reported performance that demonstrates meeting these criteria.

Note on MCHC, BASO% and BASO# Correlation: The correlation coefficients for MCHC, BASO%, and BASO# are notably lower than other parameters (0.687, 0.658, 0.646 respectively). However, the document states "all measurands met the prespecified acceptance criteria," implying these values were acceptable for the purpose of demonstrating substantial equivalence. This could be due to factors like the analytical measuring range of these parameters or inherent variability.

Study Details:

1. Sample sizes used for the test set and the data provenance:

   • Method Comparison Study:

     • Sample Size: A total of 700 residual clinical K2EDTA whole blood samples. The 'N' column in the method comparison results table shows slight variations (e.g., 662 for WBC, 674 for RBC), indicating samples might have been excluded for specific parameter analysis for various reasons (e.g., insufficient volume, issues during analysis).
     • Data Provenance:
       • Country of Origin: Three (3) US sites.
       • Retrospective or Prospective: The samples were "residual clinical" samples and re-run with the updated algorithm, suggesting they were previously collected, indicating a retrospective approach to the sample acquisition for the re-run study, though the initial collection for K190898 might have been prospective. The text says, "The samples previously collected in K190898 were re-run with the updated algorithm of the subject device."
       • Sample Characteristics: Included normal and pathological samples (e.g., acute inflammation, various anemias, leukemias etc.). Covered an age range of 3 months to 94 years old, with 32% pediatric samples (3M-21Y). 365 males (52%) and 335 females (48%).

   • Flagging Study:

     • Sample Size: Over 200 samples.
     • Data Provenance:
       • Country of Origin: 3 clinical study sites (location not explicitly stated but implied US given other elements of the submission).
       • Retrospective or Prospective: "The samples previously collected in K190898 were re-run with the updated algorithm of the subject device," indicating a retrospective re-analysis with the new algorithm.

2. Number of experts used to establish the ground truth for the test set and the qualifications of those experts:

   • Flagging Study (Ground Truth for Flagging):
     • Number of Experts: Two primary qualified morphology examiners (Reader A and Reader B), with a third qualified morphology examiner (arbitrator) used in case of disagreement. So, a total of 3 experts potentially involved for each discordant case.
     • Qualifications: Referred to as "qualified morphology examiners." Specific experience in years or board certification is not detailed, but their "qualification" is stated.

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

   • Flagging Study (for Ground Truth of Manual Microscopy): A 2+1 adjudication method was used. "Two qualified morphology examiners evaluated one of the three blood films... The third blood film was saved for reading by a third qualified morphology examiner (i.e., arbitrator) in the event that there was disagreement between Reader A and Reader B."

4. 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:

   • No MRMC comparative effectiveness study involving human readers improving with AI assistance is described in this document. The studies presented are primarily a comparison of the device's performance to a reference device (Sysmex) and to manual microscopy (for flagging), not a human-AI teamed performance study. The device is an automated analyzer, not an AI-assisted human reading tool in the sense of an MRMC study.

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

   • Yes, the performance data presented for the Sight OLO device is for its standalone performance. The device is described as an "automated hematology analyzer" that "utilizes computer vision algorithms to enumerate" parameters. The listed performance metrics (correlation, bias, sensitivity, specificity) reflect the device's output independently.

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

   • Method Comparison Study: The ground truth was established by a reference device, the Sysmex XN-Series Hematology Analyzer. This is a common method for validating new automated hematology analyzers, where an established, cleared device serves as the comparator.
   • Flagging Study: The ground truth was established by expert consensus (manual light microscopy combined with adjudication by "qualified morphology examiners"). This is often considered the gold standard for morphological assessment of blood cells.

7. The sample size for the training set:

   • The document does not specify the sample size for the training set. It focuses on the performance data for the updated algorithm applied to previously collected samples. The exact details of the original training data for the computer vision algorithms are not provided within this summary for this specific 510(k).

8. How the ground truth for the training set was established:

   • The document does not explicitly state how the ground truth for the training set was established. It describes the "minor modifications to analysis algorithms" made to increase actionable results and improve flagging specificity/reduce invalidation. It also mentions that "the device is a computer vision based platform for blood analysis." However, the methodology for establishing the ground truth for the training of these computer vision algorithms is typically proprietary and not detailed in this type of summary. It is generally assumed to involve expert-labeled data, similar to the "qualified morphology examiners" used for the flagging study's ground truth, but this is not confirmed in the provided text for the training set itself.

Ask a specific question about this device