VIDAvision

VIDA|vision is an advanced medical imaging software that utilizes computed tomography (CT) data to aid clinicians in the diagnosis, treatment planning, and monitoring of pulmonary diseases. Its core science involves the reconstruction of 2D CT scans into interactive 3D models, applying Hounsfield unit-based analysis, and leveraging non-adaptive deep learning algorithms for automated lung segmentation and quantitative measurements.

VIDA|vision is a self-contained image analysis software package designed for the real-time interactive evaluation of CT volume datasets, reconstructing two-dimensional images into a three-dimensional format. Its primary purpose is to support physicians in the diagnosis, treatment planning, and documentation of various chest diseases, including lung cancer, asthma, COPD, and interstitial lung disease, by examining pulmonary and thoracic tissue from CT scans. The software's methodology involves analyzing pulmonary CT slice data where each voxel is measured by Hounsfield units (HU) to distinguish between air, water, tissue, and bone. It incorporates tools for 3D segmentation and isolation of sub-compartments, volumetric analysis, density evaluations, and low-density cluster analysis, combined within a dedicated workflow to automate tasks that would otherwise be tedious or error-prone if performed manually. VIDA|vision provides functionalities such as CT viewing, airway analysis, and parenchymal density analysis, including imaging of bronchial airways to assess therapy effectiveness and aid in treatment planning. It reconstructs complex bronchial branches and provides quantitative measurements, with data exportable in PDF or CSV formats. A significant advancement over its predicate device, Pulmonary Workstation 2 (PW2), is the introduction of non-adaptive deep learning-based algorithms for automatic segmentation of lung regions. Other enhancements include a distributed deployment option, a transition to the Windows operating system, an improved user interface, distinct user workflows for airway mapping and lung volume analysis, DICOM query/retrieve for data importation, expanded path planning capabilities to regions of interest in lung tissue, and more versatile reporting options. Non-clinical testing confirmed that the device met all product specifications and user needs, with lung and lobe segmentation performance validated against the predicate.

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. What specific deep learning architecture is employed for the lung segmentation, and what were its performance metrics (e.g., Dice coefficient, Jaccard index) compared to both the predicate device and ground truth annotations during non-clinical testing?; How does the software's automated quantitative analysis (e.g., volumetric, density) compare in accuracy and consistency to manual analysis performed by experienced radiologists, particularly for subtle or complex lung abnormalities?; What is the range of CT scan protocols (e.g., slice thickness, reconstruction kernels) with which VIDA
• Q2. vision maintains optimal performance for its deep learning segmentation and quantitative analyses?; Beyond automating existing manual tasks, what novel diagnostic insights or efficiencies does the introduction of deep learning and enhanced quantitative tools in VIDA
• Q3. vision offer for conditions like COPD or interstitial lung disease?; How does the 'path planning to a region of interest in the lung tissue' feature technically operate, and how has its utility been validated for guiding diagnostic or therapeutic procedures?