Multi-Source Collaborative Annotation and Management Platform for CCA Images: Bridging AI Development and Clinical Diagnostics

Conventional Coronary Angiography (CCA), recognized as the gold standard for diagnosing coronary artery disease, provides critical decision-making evidence through dynamic imaging sequences. However, the complexity of coronary anatomical structures and motion artifacts during imaging render the accurate interpretation of CCA images highly dependent on clinical expertise, thereby limiting the widespread application of Computer-Aided Diagnosis (CAD) systems in this field. Despite the substantial volume of existing CCA imaging data, the scarcity of high-quality annotations-attributable to time-consuming annotating processes and inconsistent standards-severely hinders the development of artificial intelligence (AI) algorithms. To address this bottleneck, this paper proposes and constructs a multi-source information collaborative annotating and integrated management platform specifically designed for CCA imaging data. By extending the DICOM entity object model, the platform enables structured and hierarchical annotation of CCA images, incorporating a spatiotemporal registration method for multisource information integration and ensuring annotation reliability through validation by domain experts with a decade of clinical experience in CCA. Further-more, the platform embeds a modular evaluation framework with AI algorithm testing interfaces, facilitating customizable training dataset configuration and quantitative performance assessment of algorithms, thereby bridging the translational gap between CCA data resources and AI-driven coronary diagnostics. The platform also provides standardized annotation exemplars and diagnostic reference guidance for junior clinicians and radiologists, demonstrating significant potential for clinical education and research applications.