V5med Lung AI

V5med Lung AI is computer-aided detection software that assists radiologists in detecting pulmonary nodules (4-30 mm diameter) in chest CT scans of asymptomatic populations. The software displays AI-identified regions of interest alongside original CT images to support radiologist decision-making without replacing their clinical judgment.

V5med Lung AI uses a Deep Convolutional Neural Network (CNN) algorithm for nodule detection, accepts multi-vendor CT images from PACS/RIS/scanner, and outputs DICOM GSPS annotations. Compared to the primary predicate (AVIEW), it supports a broader nodule range (4-30 mm vs. 3-20 mm), works with multiple CT scanner vendors rather than Siemens-only, and accepts both contrast and non-contrast images.

Software testing followed FDA's General Principles of Software Validation (January 2002), Guidance for Device Software Functions (June 2023), and IEC 62304:2006/Amd 1:2015 for medical device software life cycle processes. Testing included unit tests, system integration tests, and standalone performance validation on phantom and clinical data.

The device performs the same clinical function as the predicate—CAD-assisted pulmonary nodule detection in asymptomatic populations. Clinical reader study data showed V5med Lung AI achieved superior AUC (0.830 aided vs. 0.734 unaided) and reduced reading time by 13%, demonstrating equivalent or better performance. Differences in nodule size range, scanner compatibility, and image type support are broader than the predicate but align with the secondary predicate (ClearRead CT) and do not introduce new safety concerns.