An artificial intelligence (AI) algorithm can identify high-risk coronary plaques on optical coherence tomography (OCT) images and predict adverse outcomes, according to findings from the PECTUS-AI study.¹ The research suggests that AI-assisted analysis of the entire imaged vessel provides superior prognostic value compared to manual analysis of a target lesion alone.
Coronary thin-cap fibroatheromas (TCFA) are known precursors to adverse cardiovascular events, but their identification is time-consuming and subject to interobserver variability. The OCT-AID algorithm is a deep learning model designed for automated full-vessel segmentation of OCT images, enabling standardised and reproducible plaque characterisation.
PECTUS-AI was a secondary analysis of the prospective, observational PECTUS-obs study.² It included 414 patients who had presented with myocardial infarction (MI) and subsequently underwent OCT imaging of all non-culprit lesions with a fractional flow reserve (FFR) >0.80.
OCT images were analysed for the presence of TCFA by both an independent core laboratory (CL-TCFA) and the OCT-AID algorithm (AI-TCFA). The primary outcome was a composite of death from any cause, non-fatal MI, or unplanned revascularisation at 2-year follow-up.
At the patient level, AI-TCFA was identified in 34.5% of patients, compared to 30.0% for CL-TCFA. The presence of AI-TCFA within a target lesion was significantly associated with the primary outcome (hazard ratio [HR] 1.99; 95% confidence interval [CI] 1.02–3.90; p=0.04). In contrast, the association for CL-TCFA was not statistically significant (HR 1.67; 95% CI 0.84–3.30; p=0.14).
A key finding emerged when the AI algorithm analysed the complete OCT pullback. In this analysis, the presence of an AI-TCFA anywhere in the imaged segment showed a much stronger association with the primary outcome (HR 5.50; 95% CI 1.94–15.62; p<0.001). This comprehensive approach also demonstrated a high negative predictive value of 97.6%.
These findings indicate that AI-based OCT analysis offers a standardised and rapid method for identifying patients at increased risk of future cardiovascular events. The ability of the AI to comprehensively evaluate an entire vessel segment—a task that is impractical to perform manually on-site—appears to provide significant incremental prognostic value. This supports the concept that plaque vulnerability is a systemic process rather than being confined to discrete, visually identified lesions. The high negative predictive value may be particularly useful for risk stratification following MI.
The study authors note that future validation studies are required to confirm these findings and to assess the algorithm's performance and utility in a real-time clinical setting.
This study was funded by the Dutch Research Council (NWO), Abbott Vascular, the Dutch Ministry of Economic Affairs and Climate Policy (EZK), and Health∼Holland.
References
1. Volleberg RHJA, Luttikholt TJ, van der Waerden RGA, et al. Artificial intelligence-based identification of thin-cap fibroatheromas and clinical outcomes: the PECTUS-AI study. Eur Heart J 2025;46:5032–41. https://doi.org/10.1093/eurheartj/ehaf595
2. Mol JQ, Belkacemi A, Volleberg RH, et al. Identification of anatomic risk factors for acute coronary events by optical coherence tomography in patients with myocardial infarction and residual nonflow limiting lesions: rationale and design of the PECTUS-obs study. BMJ Open 2021;11:e048994. https://doi.org/10.1136/bmjopen-2021-048994
Disclaimer: The information presented in this article is for educational purposes only. Any quotes included reflect the opinions of the individual quoted, and do not necessarily reflect the views of the publisher. The publisher does not guarantee the accuracy or completeness of the content and accepts no responsibility for any errors, or any consequences arising from its use.