Optical coherence tomography for carotid artery plaque: a standardised imaging and analysis methodology

BACKGROUND: Stroke remains a leading cause of morbidity and mortality, with carotid atherosclerosis representing a major contributor. High-resolution evaluation of plaque characteristics is critical for risk stratification and therapeutic guidance. However, reliable methodologies for assessments are limited. Optical coherence tomography (OCT) provides superior resolution, yet standardised protocols for carotid artery imaging are lacking. OBJECTIVE: This study aimed to develop and validate a standardised methodology for the acquisition and analysis of carotid artery plaque using OCT imaging. METHODS: Sixty-four patients with internal carotid artery stenosis from the HERALD trial who underwent paired OCT imaging were included. A standardised non-occlusive protocol was employed to accommodate carotid anatomy. Key plaque features-minimum fibrous cap thickness (FCT), maximum lipid arc, calcification, minimum lumen area, intraplaque vessels, dissection and cholesterol crystals-were assessed by three independent, blinded observers. Fibrous cap volume (FCV) was calculated using an AI-driven dual-coordinate framework. Statistical analyses included intraclass correlation coefficients (ICC) for continuous variables, Fleiss' kappa for categorical variables, Bland-Altman analysis for inter-observer agreement and Pearson correlation to assess relationships between variables. RESULTS: Analysis of 134 OCT pullbacks demonstrated excellent inter-observer reproducibility for minimum FCT (ICC=0.901), maximum lipid arc (ICC=0.812), and minimum lumen area (ICC=0.909). All observers achieved excellent intra-observer consistency. Bland-Altman analysis confirmed high agreement. Agreement for qualitative features was excellent for cholesterol crystals (κ=0.957), intraplaque vessels (κ=0.955), and dissections (κ=0.988) and moderately high for calcification (κ=0.848). FCV measurement using the AI framework showed good concordance with manual segmentation (mean Dice coefficient=0.73). A weak positive correlation was observed between changes in FCV and minimum FCT (r=0.296, p=0.024). CONCLUSIONS: This study establishes a comprehensive and standardised OCT methodology for carotid plaque imaging, demonstrating excellent reproducibility in quantifying key morphological and high-risk features.