Link to Source: Github, Paper, Online Web Application
Summary: AI-powered software for cardiac IVUS analysis with automated segmentation and gating capabilities.
AIVUS-CAA is an application designed for deep learning-based intravascular ultrasound (IVUS) image processing in coronary artery anomalies. The software enables clinicians and researchers to inspect IVUS images frame-by-frame, manually draw or automatically segment lumen contours, and automatically detect cardiac phases through intelligent gating algorithms. Built with Python and featuring GPU acceleration support, the application calculates detailed metrics including lumen area, circumference, and elliptic ratio for each frame, with automatic saving and comprehensive report generation.
The software employs advanced algorithms that analyze both image-derived metrics (pixel-wise correlation and blurriness) and vector-based contour measurements to identify resting phases of the cardiac cycle. It features an intuitive interface with keyboard shortcuts, interactive gating capabilities with zoom and pan functions, and the ability to export data in NIfTi format for machine learning model training. Published in Computer Methods and Programs in Biomedicine (2025), AIVUS-CAA represents a significant advancement in making sophisticated IVUS analysis accessible to clinicians working with coronary artery disease.