Diagnostic performance of artificial intelligence for coronary plaque characterization using OCT and IVUS: a systematic review and meta-analysis

Background Intravascular ultrasound (IVUS) and optical coherence tomography (OCT) are central to contemporary percutaneous coronary intervention (PCI), yet image interpretation remains time-intensive and operator-dependent. Artificial intelligence (AI), machine learning (ML), and deep learning (DL) enable automated plaque characterization and may improve consistency and scalability of imaging-guided PCI. Purpose To evaluate the diagnostic performance of AI/ML/DL-enabled IVUS and OCT for identifying high-risk coronary plaque features. Methods A systematic review and meta-analysis were performed of studies applying AI/ML/DL to intravascular imaging with a predefined reference standard. PubMed, Embase, and the Cochrane Library were searched from inception to November 2025. Pooled sensitivity, specificity, and hierarchical summary receiver operating characteristic (HSROC) curves were estimated for studies with extractable 2×2 data. Risk of bias was assessed using QUADAS-2. Results Ten studies comprising 2,632 participants were included; four contributed to quantitative synthesis and six to qualitative assessment. AI-enabled IVUS demonstrated pooled sensitivity of 80% and specificity of 80%, with strong rule-out performance (negative predictive value [NPV] 96%) and an area under the curve (AUC) of 0.81. OCT-based AI/ML/DL models showed lower sensitivity (64%) but higher specificity (91%), similarly high NPV (96%), and an AUC of 0.91. Positive predictive values were modest across modalities. Qualitative synthesis confirmed consistently high diagnostic performance, and one multimodality IVUS+OCT study reported strong discriminative accuracy (AUC 0.88). Limited lesion-level data restricted quantitative pooling. Conclusion AI-enabled intravascular imaging demonstrates promising diagnostic performance for coronary plaque characterization, particularly for excluding high-risk features. Prospective, outcome-driven validation and standardized reporting are warranted to support clinical integration into PCI workflows.