Artificial Intelligence and Intravascular Imaging in Interventional Cardiology: A Scoping Review

Background : Intravascular imaging plays a key role in interventional cardiology, particularly in guiding procedures like PCI, using modalities such as optical coherence tomography (OCT) and intravascular ultrasound (IVUS). Nevertheless, traditional interpretation is still constrained by operator dependence, time consumption and inter-observer variability. Recently, automated imaging modalities to assist intravascular imaging analysis have been developed with the aim of increasing accuracy, efficiency, and reproducibility of these analyses, most notably through the emergence of artificial intelligence (AI) based algorithms. Objective : This scoping review seeks to highlight the current literature on the current state of AI in intravascular imaging in interventional cardiology focusing on the years from 2020 to 2025. Methods : Using the Joanna Briggs Institute (JBI) scoping review methodology, we performed a systematic PubMed search of English-language articles that used AI, that is, machine learning (ML) or deep learning (DL), for analyzing OCT or IVUS imaging. Screening of studies was guided by the Population–Concept–Context (PCC) framework. Four variables were extracted from each article including AI model type, imaging modality, application domain, and validation approaches. Results : Out of 19 identified studies, 16 met inclusion criteria. They were used for applications such as lumen, plaque and stent segmentation, plaque characterization, lesion assessment, and procedural prediction. Several studies expanded the use of AI to longitudinal plaque tracking and anatomical modeling. Validation metrics of Dice similarity coefficient (DSC), area under the curve (AUC), and intraclass correlation coefficient (ICC) showed excellent agreement to expert interpretation. Conclusion : Artificial intelligence has great potential for the evolution of intravascular imaging analysis in interventional cardiology. Future work will need to increase model generalizability, validate prospectively and integrate AI tools into the clinical workflow to maximize the impact of these data-driven approaches to procedural planning and patient outcomes.