Search Results

hemoFISH Masterpanel is a qualitative nucleic acid hybridization assay performed directly on blood culture samples identified as positive by a continuous monitoring blood culture system that demonstrates the presence of organisms as determined by Gram stain and is intended for the identification of the following species / families:

Gram-positive: Staphylococcus spp., Staphylococcus aureus, Streptococcus spp.
Gram-negative: Enterobacteriaceae, Escherichia coli, Klebsiella pneumoniae

The hemoFISH Masterpanel is indicated as an aid in the diagnosis of bacteremia and results should be used in conjunction with other clinical and laboratory findings. Positive hemoFISH Masterpanel results do not rule out coinfection with organisms not included in the hemoFISH Masterpanel is not intended to monitor treatment for bacteremia.

Subculturing of positive blood cultures is necessary to recover organisms for susceptibility testing and epidemiological typing, to identify organisms in the blood culture that are not identified by the hemoFISH Masterpanel, and for species determination of Staphylococcus spp., Streptococcus spp., and Enterobacteriaceae that are not identified by the hemoFISH Masterpanel.

The hemoFISH Masterpanel identifies bacteria within 30 minutes directly from positive blood cultures.

The methodology is based on classical fluorescence in-situ hybridization (FISH) combined with the usage of fluorescently labeled molecular beacons.

A solution of fluorescently labeled molecular DNA beacons is dispensed on fixed and perforated cells prepared from a blood culture. The hybridization is carried out at 52°C for 10 minutes. Submerging the slide into a bath containing a stop solution ceases the reaction. Adding a drop of Mounting Medium to each field prevents fading of fluorescence. After applying a cover slip, the slide is ready for examination using fluorescence microscopy.

The hemoFISH Masterpanel is a qualitative nucleic acid hybridization assay designed for the rapid identification of specific bacterial species/families directly from positive blood culture samples. It utilizes fluorescence in-situ hybridization (FISH) with molecular DNA probes.

Here's an analysis of the acceptance criteria and the study proving the device meets them:

1. Acceptance Criteria and Reported Device Performance

The provided document doesn't explicitly list "acceptance criteria" in a separate table with specific numerical thresholds for sensitivity and specificity that the device must meet. Instead, it presents the results of its performance studies, implying that these results are considered acceptable for substantial equivalence.

However, based on the "Performance Characteristics" section, we can infer the de-facto acceptance criteria by examining the reported performance and the overall conclusion of substantial equivalence. The key performance metrics are Sensitivity (Positive Agreement) and Specificity (Negative Agreement) against routine laboratory methods.

Here's a table summarizing the reported device performance, which the FDA implicitly accepted as meeting the criteria for substantial equivalence:

Table 1: Reported Device Performance (Monomicrobial Cultures, Prospective and Spiked Samples Combined)

Note: The numbers in parentheses for each target agent represent the number of samples identified as such by the "Routine Identification" (ground truth).

2. Sample Size and Data Provenance

• Test Set Sample Size:

  • Clinical Studies (Monomicrobial): 609 prospective blood culture bottles + 61 in-house spiked (monomicrobial) blood culture bottles = 670 monomicrobial blood cultures (combined in Table 3).
  • Clinical Studies (Polymicrobial): 55 polymicrobial prospective blood culture bottles (analyzed separately in Table 4).
  • Analytical Inclusivity: 120 representative reference strains.
  • Analytical Specificity: 215 strains representing clinical relevant species and/or species selected based on sequence similarities.
  • Limit of Detection (LoD): Not specified for the number of strains, but "≥19/20 replicates (≥95%) were positive at 10^3 CFU/mL" for each tested species.
  • Reproducibility: 90 tests per strain (3 slides/strain tested by 2 different operators on 2 x five consecutive days).

• Data Provenance: The document does not explicitly state the country of origin for the clinical data. It mentions "3 clinical laboratory studies." Given the submitter's location (Düsseldorf, Germany), it's highly probable that the clinical studies were conducted in Germany or Europe, though this is not confirmed. The data appears to be prospective for the main clinical method comparison study, with additional in-house spiked samples (retrospective/controlled laboratory setting).

3. Number of Experts and Qualifications for Ground Truth

The document does not specify the number of experts or their qualifications involved in establishing the ground truth for the clinical test set. The ground truth for the clinical studies was established by "routine laboratory methods" (e.g., culture, biochemical tests, mass spectrometry, or other conventional microbiology methods), which are presumed to be performed by qualified laboratory personnel following established protocols.

4. Adjudication Method for the Test Set

The document does not mention any specific adjudication method (e.g., 2+1, 3+1) for discordant results between the hemoFISH Masterpanel and the routine laboratory methods. The "routine identification" is presented as the reference standard.

5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study

No, a multi-reader multi-case (MRMC) comparative effectiveness study was not conducted. This device is a diagnostic assay, and the performance evaluation focused on its agreement with standard laboratory identification methods, not on how human readers (e.g., clinicians or microbiologists) improve their diagnostic performance with or without AI assistance. The interpretation of the device results is "Visual by fluorescence microscopy" and is a direct output from the assay, not an interpretation by a human of an AI output.

6. Standalone (Algorithm Only) Performance

The study primarily evaluates the standalone performance of the hemoFISH Masterpanel assay. The "visual by fluorescence microscopy" interpretation refers to a direct reading of the assay's output, not an algorithm's output. There isn't an "algorithm only without human-in-the-loop performance" in the typical sense of AI-driven image analysis. The device itself performs the detection (hybridization and fluorescence), and the "visual" part is the human reading the output. This is a traditional IVD, not an AI/ML-driven device in the sense of image interpretation.

7. Type of Ground Truth Used

The primary ground truth used for the clinical studies was "routine identification" based on standard laboratory methods, which typically involve microbial culture and subsequent identification techniques (e.g., Gram stain, biochemical tests, mass spectrometry, gene sequencing).

For analytical studies:

• Analytical Inclusivity/Specificity: Determined using "representative reference strains" with known identities.
• LoD: Spiked blood culture bottles with known reference strains.

8. Sample Size for the Training Set

The document does not specify an explicit "training set" sample size in the context of machine learning. As a traditional IVD (fluorescence in-situ hybridization assay), this device does not typically undergo a "training" phase like an AI/ML model. The design and validation of the probes (the core "method" of the device) are based on molecular biology principles and analytical testing, not iterative learning from a large dataset.

9. How Ground Truth for the Training Set Was Established

Since there is no "training set" in the context of machine learning for this traditional IVD, the concept of establishing ground truth for a training set does not apply. The development of the hemoFISH Masterpanel involved designing specific DNA probes to target ribosomal RNA sequences of the target organisms. The "ground truth" during this development (analogous to a training phase) would have involved:

• Bioinformatics: In silico analysis of ribosomal RNA sequences to design highly specific and inclusive probes.
• Analytical validation: Testing probe performance against a wide range of known bacterial strains (as mentioned in the Analytical Inclusivity and Specificity sections) to confirm their binding specificity and sensitivity. These analytical studies are part of the device development and validation, not a separate "training set" as understood in AI/ML.

Ask a specific question about this device