Artificial Intelligence-Based Analysis of Coronary Atherosclerotic Plaque on Intravascular Ultrasound: A Systematic Review and Meta-Analysis.
Authors
Affiliations (3)
Affiliations (3)
- Cardiovascular Imaging Research Center, Rajaie Cardiovascular Institute, Tehran, Iran.
- Hamadan University of Medical Sciences, Hamadan, Iran.
- College of Human Medicine, Michigan State University, East Lansing, Michigan, USA.
Abstract
Coronary atherosclerotic plaque detection is essential for the diagnosis and management of coronary artery disease. Although intravascular ultrasound (IVUS) enables detailed plaque assessment, its clinical use is limited by time-consuming interpretation and observer variability. Artificial intelligence (AI)-based methods offer a promising approach to automate IVUS plaque detection. This systematic review and meta-analysis evaluated the diagnostic performance of AI models for IVUS-based coronary plaque detection. A comprehensive literature search was conducted across major databases from inception to December 2025. Data extraction and risk of bias assessment were performed independently using the PROBAST-AI tool. Pooled sensitivity and specificity were estimated using bivariate random-effects models, and sources of heterogeneity were explored through meta-regression. Certainty of evidence was assessed using the GRADE framework. Among 10 included studies (six for meta-analysis), the pooled sensitivity and specificity were both 0.99 (95% CI, 0.93-1.00 and 0.96-1.00, respectively), with an area under the summary receiver operating characteristic curve of 1.00 (95% CI, 0.99-1.00). Moderate between-study heterogeneity was observed (generalized I<sup>2</sup> = 46.1%), affecting both sensitivity and specificity. Meta-regression analyses identified sample size as a significant contributor to heterogeneity. Risk of bias assessment revealed a moderate overall risk, and the certainty of evidence was rated as moderate. AI-based IVUS analysis shows promising performance for automated plaque and calcification assessment in research settings. However, limitations related to study design, sample size, and reporting transparency remain. Future research should prioritize large, multicenter, prospective studies and strict adherence to TRIPOD + AI guidelines to support clinical translation.