(220 days)
The NeuroField Analysis Suite is to be used by qualified clinical professionals for the display and storage of electrical activity of a patient's brain, including the post-hoc statistical evaluation of the human electroencephalogram (EEG) and event-related potentials (ERP).
The NeuroField Analysis Suite is a Normalizing Quantitative Electroencephalograph (QEEG) Software that can (1) execute EEG analysis and (2) conduct ERP test and ERP analysis. The NeuroField Analysis Suite is Software as a Medical Device (SaMD). The NeuroField Analysis Suite consists of two modules, the NF EEG Analysis Module and the NF ERP Module. The NF EEG Analysis Module is a separate analysis module that integrates with the Q21 EEG system by adding "Analysis", "Report", and "Tools" menu items and toolbars. It performs real-time and offline analysis functions and displays analysis results in separate windows in the UI, which can be accessed via the "Analysis" and "Reports" menu items. The NF ERP Module is a separate evoked response potential (ERP) module that can control and get data from the Q21 EEG system and performs typical ERP functions like stimulus presentation, EEG epoching, epoch averaging, reaction time, and ERP display. The NF ERP Module is a separate stand-alone application.
The NeuroField Analysis Suite, comprising the NF EEG Analysis Module and the NF ERP Module, was evaluated for performance. Here's a breakdown of the acceptance criteria and the study details:
1. Table of Acceptance Criteria and Reported Device Performance
The FDA document doesn't explicitly state acceptance criteria in terms of numerical performance metrics (e.g., specific accuracy, sensitivity, specificity values). Instead, substantial equivalence is claimed based on comparable functionality and performance to predicate devices. The "performance" reported reflects this comparative approach.
| Feature Category | Acceptance Criteria (Implied from Predicate Comparison) | Reported Device Performance |
|---|---|---|
| NF EEG Analysis Module (vs. NeuroGuide Analysis System) | ||
| EEG Analysis Functionality | Comparable re-montaging, filtering, event markers, and data analysis modes (marker-based or whole data). | Performs re-montaging, filtering, adding event markers, and analysis based upon markers or whole data. |
| Z-Score Generation | Comparable generation of Z-Scores for absolute and relative power of EEG bands/frequencies. | Generates Z-Scores for absolute power and relative power of EEG bands or individual frequencies. |
| Visualization | Comparable generation of head maps and FFT spectra with equivalent frequency resolution. | Generates head maps and FFT spectra, and with equivalent frequency resolution (0.5 Hz). |
| Reporting | Comparable tabular export and automated report generation. | Provides for tabular export and automated report generation. |
| Inverse Solution (LORETA) | Comparable calculation of inverse solution (Key Institute 2394 LORETA model) and generation of current densities/powers of ROIs using Brodmann Atlas. | Can calculate the inverse solution (using the standard LORETA Key Institute 2394 LORETA model) and generate the current densities and powers of the ROIs (Region of Interest) over the cortex using the Brodmann Atlas. |
| Supported File Formats | Comparable ability to read standard EDF and EDF+ files. | Reads standard EDF and EDF+ files, additionally supports XDF. |
| Signal/Spectrum Comparison | Comparable signals on EEG Plot and comparable spectrum/sidelobes for real and simulated data. | Shows comparable signals on the EEG Plot for the same EDF file as Neuroguide. Spectrum and sidelobes for real and simulated data are comparable. |
| Z-Score Comparison | Comparable Z-Scores for real and simulated data. | Shows comparable Z-Scores. |
| Headmap Topography | Comparable topographies over standard EEG bands. | Visualization of headmaps shows comparable topographies over the standard EEG bands. |
| NF ERP Module (vs. eVox System) | ||
| Averaged ERP Graphs | Comparable generation of averaged ERP graphs, allowing user to measure latency and magnitude. | Generates averaged ERP graphs for EEG channels, allowing user to measure latency and magnitude. |
| Session Length | Comparable customizable session length. | Allows for customizable session length. |
| Oddball Paradigm | Comparable application of auditory and visual oddball paradigms. | Applies auditory and visual oddball paradigm variations. |
| Parameter Settings | Comparable ability to set parameters for visual and auditory oddball paradigm. | Allows the user to set parameters for the visual and auditory oddball paradigm. More user-definable parameters than the predicate (e.g., visual filtering, oddball period, Go probability, customizable audio/visual stims). |
| Mathematical Correctness | Mathematically correct generated averaged ERPs. | Demonstrated mathematically correct generated averaged ERPs through mathematical validation. |
| ERP Component Elicitation | Elicitation of correct ERP components. | Analysis of real data shows it can elicit correct ERP components. |
| Stimulus Reliability | Reliable generated stimulus. | Confirmed via analysis of real data. |
| Test Reliability | Confirmed test reliability (split-half reliability). | Split-half reliability measure confirms test reliability. |
2. Sample Size Used for the Test Set and Data Provenance
- NF EEG Analysis Module: The document states that "Comparative performance evaluations showed that the same EDF file, loaded to NF-EEG and Neuroguide shows comparable signals on the EEG Plot." It also mentions "real and simulated data" for spectrum and sidelobe calculations. No specific sample size (number of EDF files or simulated data instances) is provided.
- Data Provenance: Not explicitly stated, but the use of "real and simulated data" suggests a mix. "Same EDF file" implies existing data, which could be retrospective.
- NF ERP Module:
- "Mathematical validation of NF-ERP was performed." - No specific sample size mentioned, as this is a theoretical assessment.
- "Analysis of real data shows that the NF-ERP can elicit correct ERP components, the generated stimulus is reliable, and the split-half reliability measure confirms test reliability." - No specific sample size (number of cases/patients) is provided for this "real data" analysis.
- Data Provenance: "Real data" is used, but country of origin or whether it's retrospective/prospective is not specified.
3. Number of Experts Used to Establish the Ground Truth for the Test Set and Qualifications
The document does not mention the use of experts to establish a "ground truth" for the comparative performance evaluations of either the EEG or ERP modules. The comparison is primarily against the output of the predicate devices.
4. Adjudication Method for the Test Set
Not applicable, as no external "ground truth" established by experts or adjudication process is described for the test set. The comparison is directly between the outputs of the NeuroField Analysis Suite and the predicate devices (Neuroguide or eVox).
5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study was done
No, a multi-reader multi-case (MRMC) comparative effectiveness study evaluating human readers' improvement with AI vs. without AI assistance was not conducted or described in this document. The study focuses on the standalone performance and comparison to predicate devices, not human-in-the-loop performance.
6. If a Standalone (Algorithm Only Without Human-in-the-Loop Performance) was Done
Yes, a standalone performance evaluation was done for both modules.
- NF EEG Analysis Module: Directly compared its outputs (EEG plots, spectrums, sidelobes, Z-scores, headmaps) against those of the Neuroguide Analysis System.
- NF ERP Module: Underwent mathematical validation to confirm the correctness of averaged ERPs and analysis of real data to demonstrate correct ERP component elicitation, stimulus reliability, and test reliability using split-half reliability.
7. The Type of Ground Truth Used
The "ground truth" for the performance evaluations is implicitly the output or established functionality of the predicate devices (Neuroguide for EEG and eVox for ERP). For the NF ERP module, mathematical correctness and consistency with known physiological responses (correct ERP components, reliable stimulus) served as the "ground truth" for its internal validation. There's no mention of pathology, outcomes data, or external expert consensus as a ground truth.
8. The Sample Size for the Training Set
The document does not provide any information regarding a training set size. This suggests that the NeuroField Analysis Suite, as a QEEG and ERP analysis software, likely relies on established algorithms and mathematical models rather than machine learning models that require extensive training data.
9. How the Ground Truth for the Training Set Was Established
Not applicable, as no training set or machine learning model is described. The device appears to be based on deterministic algorithms for signal processing and statistical evaluation of EEG/ERP data.
§ 882.1400 Electroencephalograph.
(a)
Identification. An electroencephalograph is a device used to measure and record the electrical activity of the patient's brain obtained by placing two or more electrodes on the head.(b)
Classification. Class II (performance standards).