Search Results

The NvisionVLE Imaging System is indicated for use as an imaging tool in the evaluation of human tissue microstructure, including esophageal tissue microstructure, by providing two-dimensional, cross-sectional, real-time depth visualization and may be used to mark areas of tissue. The software provides segmentation and display of common imaging features, including hyper-reflective surface, layering, and hypo-reflective structures.

The NvisionVLE® Imaging System is intended to provide an image of tissue microstructure. The safety and effectiveness of this device for diagnostic analysis (i.e. differentiating normal versus specific abnormalities) in any tissue microstructure or specific disease has not been evaluated.

The NinePoint Medical NvisionVLE® Imaging System is a high-resolution volumetric imaging system based on optical coherence tomography (OCT). In an analogous fashion to ultrasound imagery, OCT images are formed from the time delay and magnitude of the signal reflected from the tissue of interest. The NvisionVLE Imaging System employs an advanced form of OCT known as sweptsource OCT (SS-OCT), or Optical Frequency Domain Imaging (OFDI), in combination with a scanning optical probe to acquire high-resolution, cross-sectional, real-time imagery of tissue called Volumetric Laser Endomicroscopy (VLE).

In addition to the imaging capability, the device provides a means of marking areas of tissue with an additionally integrated 1470nm laser. The ability to create temporary laser marks directly on tissue enables a clinician to place visual reference marks on tissue regions of clinical interest immediately following their identification via VLE. The device consists of the following five main components and accessories: (i) a mobile NvisionVLE Console with an integrated computer and two touch-screen interfaces; (ii) proprietary NvisionVLE Software used to acquire, process, and visualize VLE images; (iii) a single-use, sterile NvisionVLE Marking Probe that is inserted through the working channel of an endoscope; (iv) a single-use, sterile NvisionVLE Inflation System that is used to inflate the Marking Probe's balloon to facilitate placement; and (v) a Probe Lock Accessory to prevent longitudinal motion of the Marking Probe within the endoscope.

The purpose of this 510(k) submission is to add an artificial intelligence software tool referred to as Image and Visualization Enhancements (IVE) to the previously cleared, predicate NvisionVLE Imaging System (K153479). The IVE software module allows enhanced visualization (segmentation and colorized display) of the following commonly observed image features (also referred to as IVE features): (1) hyper-reflective surface, (2) layering and (3) hypo-reflective structures. The segmentation algorithm was developed using an artificial intelligence machine learning technique known as deep learning. Here, an artificial neural network was trained with manually labelled examples of each feature and then locked for realtime inference on new image data acquired by the device. Display of each feature can be toggled via the user interface, where a respective color overlay is presented. The default display of the IVE features is disabled and the standard VLE image data displayed per the cleared NvisionVLE Imaging System. Segmentation of these structures are based on existing image features, and IVE simply increases the conspicuity via the color overlays, thus aiding image review. It is a convenience tool and a resource for the clinician and as such, it does not alter the standard of care or the role of the physician in reviewing and assessing images generated by the system.

The provided text describes the acceptance criteria and a study proving the device meets those criteria. Here's a breakdown of the requested information:

1. Table of Acceptance Criteria and Reported Device Performance:

The document states that "The target true positive and true negative detection fractions were prospectively set." However, the specific target values for these fractions are not explicitly listed in numerical form in the provided text. Instead, it states that the observed results exceeded their target value with a significance level a < 0.05.

Here's the table of the reported device performance:

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

• Sample Size:
  • Hypo-Reflective Structures: 192 positive ROIs and 266 negative ROIs (Total 458 ROIs)
  • Layering: 229 positive ROIs and 225 negative ROIs (Total 454 ROIs)
  • Tissue Surface: 253 positive ROI and 220 negative ROIs (Total 473 ROIs)
• Data Provenance: Data was obtained from a "sequestered subset (~40%) of the 1000-patient 18-site NvisionVLE Clinical Registry study." This indicates the data is from a prospective clinical registry. The country of origin is not explicitly stated, but the registration of the device with the FDA for a US company (NinePoint Medical, Inc. based in Bedford, Massachusetts) implies that the clinical study data is likely from the United States or at least includes data from the US.

3. Number of Experts Used to Establish the Ground Truth for the Test Set and Qualifications:

• Number of Experts: Not explicitly stated, but referred to as "trained experts."
• Qualifications of Experts: They were "trained experts in clinical interpretation of VLE imagery." Specific details like "radiologist with 10 years of experience" are not provided.

4. Adjudication Method for the Test Set:

• The document states, "Assessment was performed by comparing ground truth data generated by human observers, with the software-based detection of the IVE features." It does not specify an adjudication method such as 2+1 or 3+1 for resolving discrepancies among multiple human observers. It refers to "trained experts" in the plural, suggesting more than one, but no detail on how consensus was reached or disagreements were handled is provided.

5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study Was Done, and Effect Size:

• No, an MRMC comparative effectiveness study that assesses how much human readers improve with AI vs. without AI assistance was not explicitly stated or described. The study focused on the performance of the algorithm itself in detecting features against a human-generated ground truth, rather than human performance with and without the AI. The document describes the IVE as a "convenience tool" that "does not alter the standard of care or the role of the physician."

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

• Yes, a standalone study was done. The performance testing "evaluated the ability of the software to detect each IVE feature... by comparing ground truth data generated by human observers, with the software-based detection of the IVE features." This describes the algorithm's performance independent of human-in-the-loop interaction for the quantitative metrics presented.

7. The Type of Ground Truth Used:

• Expert Consensus: The ground truth consisted of "ROIs labeled by trained experts in clinical interpretation of VLE imagery."

8. The Sample Size for the Training Set:

• The document states, "an artificial neural network was trained with manually labelled examples of each feature." However, the sample size for the training set is not explicitly provided in the given text.

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

• The ground truth for the training set was established through "manually labelled examples of each feature." It can be inferred that these labels were created by experts, similar to how the ground truth for the test set was established.

Ask a specific question about this device