Researchers have introduced an artificial intelligence-based technique to detect fatty deposits, known as lipid-rich plaques, inside coronary arteries using optical coherence tomography (OCT) images. These plaques are associated with an increased risk of serious cardiac events, including heart attacks. The new method aims to help clinicians identify dangerous plaques before they rupture.
OCT is commonly used during catheter-based procedures such as opening blocked blood vessels and placing stents. While OCT produces detailed images of vessel structures, it does not typically reveal the composition of the vessel wall, which is crucial for assessing heart attack risk.
The team published their findings in Biomedical Optics Express from the Optica Publishing Group. Their approach extracts spectral information from standard OCT images and applies deep learning techniques for quantitative and automatic assessment of lipids directly from intravascular OCT scans. This method does not require changes to existing clinical OCT hardware.
"During a coronary intervention, this method could provide clinicians with additional information to support risk assessment, procedural planning and evaluation of treatment response," said Nam. "Ultimately, it has the potential to contribute to safer clinical decision making, more individualized treatment strategies and improved long-term management of patients with coronary artery disease."
Currently, identifying high-risk lipid-rich plaques in clinical practice depends largely on physician experience. To address this limitation, researchers collaborated with Jin Won Kim's team at Korea University Guro Hospital.
"Our group previously demonstrated that spectroscopic OCT can detect lipid-related optical signatures within atherosclerotic plaques," said Nam. "This new study builds on that by extending it with modern deep learning techniques to significantly improve detection accuracy and robustness."
The new AI model uses wavelength-dependent data from OCT images because different tissues interact with light differently. For example, lipid tissue absorbs and reflects light in ways distinct from fibrous tissue or calcium. The AI learns these patterns to automatically highlight regions likely containing lipid-rich tissue.
"Importantly, unlike many conventional AI systems that require experts to painstakingly label lipid regions at the pixel level - an extremely time-consuming and subjective process - our approach learns from much simpler frame-level annotations that indicate only whether lipid is present or absent," said Nam. "This substantially lowers the annotation burden and makes the method far more practical for real-world clinical use."
Validation was conducted using imaging data from a rabbit model of atherosclerosis. The AI’s predictions were compared against histopathology results obtained through lipid-specific staining to assess how accurately the system identified frames containing lipid-rich plaques.
"The results showed strong classification performance along with good spatial agreement with the pathological findings," said Nam. "Looking ahead, the same framework we applied could be extended to other intravascular or optical imaging modalities where subtle spectral or signal variations are present but underutilized."
Researchers plan further improvements in processing speed and robustness for real-time application in clinics. Additional validation studies using human coronary artery data are also planned as they work toward integrating this technology into current clinical workflows.
