AVIEW

AVIEW is a comprehensive image processing software that utilizes DICOM CT data and deep learning algorithms for quantitative analysis of pulmonary and cardiac features, aiming to enhance diagnostic support for clinicians. Its methodology focuses on automated segmentation, detailed characterization of lung nodules (including kernel conversion for LAA analysis and Lung-RADS classification), and coronary artery calcium scoring to aid in disease assessment and management in adult patients.

AVIEW is an advanced medical image processing software designed for quantitative analysis of CT thoracic and cardiac datasets in adult patients. Its primary purpose is to support physicians in diagnosis, follow-up evaluation, and documentation by providing comprehensive tools for image segmentation of pulmonary and cardiac structures, including lungs, lobes, airways, fissures, and coronary arteries. The software facilitates the characterization of lung nodules, including measurements of size, volume, density (HU values), mass, and volume doubling time, and aids in their classification using Lung-RADS. Additionally, it quantifies coronary artery calcification by calculating Agatston, volume, and mass scores, and assesses CAC risk based on patient demographics. AVIEW operates on DICOM 3.0 images and offers on-premises and cloud-based deployment with mobile and web browser access. The software employs deep-learning algorithms for automatic segmentation of anatomical structures like lungs, lobes, and airways, and for calcium area segmentation in coronary arteries. A key scientific advancement is its ability to convert sharp kernel CT images to soft kernel images, enhancing the quantitative analysis of low attenuation areas (LAA) in the lung; this kernel conversion was validated using 96 datasets. AVIEW integrates with third-party CAD systems for automated nodule detection and utilizes models like the Brocks model for lung cancer probability. It supports follow-up studies through automated nodule matching and provides tools for analyzing fissure completeness (validated using 129 subjects), air trapping index (via inspiration/expiration registration), and airway wall measurements. Comprehensive non-clinical performance testing, including unit, system, and specific algorithm validation, confirmed the device's intended functionality and reliability.

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. How does the AVIEW software's kernel conversion algorithm for low attenuation area (LAA) analysis compare in accuracy and consistency to established methods using native soft kernel CT acquisitions, particularly across different scanner manufacturers and low-dose protocols?; What is the quantified impact of integrating AVIEW's automated lung nodule characterization and Lung-RADS classification on inter-reader variability and diagnostic workflow efficiency in a clinical lung cancer screening setting?; Can the deep learning algorithms used for segmentation of pulmonary structures (lungs, lobes, airways) and coronary arteries maintain their performance across diverse patient populations with varying degrees of pathology (e.g., severe emphysema, extensive fibrosis, or high calcium burden)?; What specific validation studies were conducted to assess the accuracy of the Brocks model integration for predicting lung cancer malignancy, and how does its performance compare to standalone implementations of the model?; How does the software's performance in fissure completeness analysis, validated with 129 subjects, correlate with clinically relevant outcomes, such as predicting collateral ventilation or guiding choices for bronchoscopic lung volume reduction?