AngioWaveNet

AngioWaveNet is an AI/machine learning software system that enhances the visibility of blood vessels in X-ray coronary angiographic images. It is intended for use by qualified physicians to aid in the analysis and interpretation of coronary angiograms during diagnostic procedures by improving the clarity of vascular structures in angiographic cines.

AngioWaveNet uses a spatio-temporal enhancement processing (STEP) method employing an encoder-decoder neural network architecture that processes multiple contiguous frames from angiographic cines. It operates as a DICOM node supporting 512×512 and 1024×1024 resolution images in 8 or 16-bit formats, requires Windows 10/11 Professional on Intel i7 processors with 32GB RAM, and handles anonymization, storage, retrieval, and cloud-based processing tasks.

The software complies with NEMA PS 3.1-3.20 2024a Digital Imaging and Communications in Medicine (DICOM) standard. Software documentation followed FDA guidance for BASIC level documentation including risk management, requirements specification, architecture diagrams, design specifications, and verification/validation testing.

AngioWaveNet demonstrates substantial equivalence to the predicate ClariCT.AI through: (1) identical regulatory classification (Class II, 21 CFR 892.2050) and intended purpose of enhancing medical image visualization through deep-learning processing; (2) comparable technological approach using pre-trained deep-learning models on DICOM-formatted images; (3) clinical testing showing 100% processing success, neutral-to-positive clinical decision impact (mean Likert 3.23/5), and 99.4% improvement in visualization ease across 31 patient cases with 3,211 assessed tasks; and (4) similar system compatibility, hardware requirements, and DICOM conformance standards.