Simultaneous Multi-Surface Fitting for Vessel Wall Layer Delineation

Intravascular optical coherence tomography (OCT) provides high-resolution images of arterial cross-sections, offering rich information on vessel morphology and disease state. The detection of the media layer is crucial for improving quantitative analyses, including plaque characterization. Manual annotation of vessel layers for analysis and quantification, however, is laborious and requires substantial expertise. We therefore sought to automatically delineate the intima-media and media-adventitia layers simultaneously even when the delineated features are not fully visible, as is often seen with atherosclerotic disease. We validated our approach with a 105-frame image sequence from a single patient with severe atherosclerosis. Delineations automatically generated by the algorithm matched annotations by human experts in most regions; the media thickness deviated on average by 0.038 mm (standard deviation: 0.029 mm). These promising results could improve characterization of vessel wall plaque, quantitative assessment of patient health, and longitudinal monitoring of disease progression.