NvisionVLE Imaging System, NvisionVLE Optical Probe, NvisionVLE Inflation System

The NvisionVLE Imaging System employs swept-source optical coherence tomography (SS-OCT) for real-time, cross-sectional visualization of esophageal and other human tissue microstructures, with an integrated laser for tissue marking. An artificial intelligence-driven software enhancement, IVE, utilizes deep learning to segment and colorize key imaging features like layering and hypo-reflective structures, demonstrating high accuracy in identifying these pre-defined patterns.

The NvisionVLE Imaging System by NinePoint Medical is a high-resolution volumetric imaging tool designed for evaluating human tissue microstructure, with a specific application in esophageal tissue. It operates using swept-source optical coherence tomography (SS-OCT), also known as Optical Frequency Domain Imaging (OFDI), which, similar to ultrasound, generates images from the time delay and magnitude of signals reflected from tissue. This technology, termed Volumetric Laser Endomicroscopy (VLE), provides two-dimensional, cross-sectional, real-time depth visualization. The system also incorporates a 1470nm laser for marking areas of tissue identified during imaging. Key components include a mobile console with touch-screen interfaces, proprietary software for image acquisition and processing, a single-use sterile marking probe for endoscopic use, an inflation system for probe placement, and a probe lock accessory. A software enhancement, known as Image and Visualization Enhancements (IVE), has been incorporated to improve the visualization of common imaging features. The IVE module uses an artificial intelligence deep learning algorithm, specifically a trained artificial neural network, to segment and apply color overlays to features such as hyper-reflective surfaces, tissue layering, and hypo-reflective structures. This is intended as a convenience tool to aid image review by increasing the conspicuity of these features. Performance of the IVE software was assessed through bench testing using a sequestered dataset from a clinical registry. The software demonstrated high true positive and true negative detection fractions for identifying tissue surface (92.9% TP, 95.0% TN), layering (89.6% TP, 97.8% TN), and hypo-reflective structures (91.1% TP, 92.9% TN), with all results meeting pre-defined performance metrics. It is important to note that the system is for imaging tissue microstructure, and its effectiveness for diagnostic analysis of specific abnormalities or diseases has not been evaluated.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. Given the limitation that diagnostic analysis for specific abnormalities has not been evaluated, what is the intended clinical benefit of identifying features like 'hypo-reflective structures' or 'layering' with the IVE software?; How robust is the deep learning algorithm in the IVE software to variations in image quality or artifacts commonly encountered during in-vivo esophageal VLE procedures?; What was the methodology for generating 'ground truth' labels by human experts for training and testing the IVE algorithm, and what measures were taken to ensure inter-observer consistency?; Beyond enhanced visualization, are there plans to quantify the characteristics of the segmented features (e.g., size, depth, density of hypo-reflective structures) to aid in tissue characterization?; How does the image resolution and depth penetration of this SS-OCT system compare to other advanced endoscopic imaging modalities for esophageal assessment, such as confocal laser endomicroscopy or high-frequency ultrasound?