Computer-aided image analysis algorithm to enhance in vivo diagnosis of plaque erosion by intravascular optical coherence tomography

Abstract

Background: Recent reports show that plaque erosion can be diagnosed in vivo using optical coherence tomography in patients with acute coronary syndrome. However, quantitative optical coherence tomographic image criteria for computeraided diagnosis of plaque erosion have not been established. Methods and Results: A total of 42 patients with acute coronary syndrome caused by plaque erosion were included. Plaque erosion was identified according to the previously established optical coherence tomography criteria. Both optical properties and morphological features of the focal-eroded region as well as erosion-adjacent region were analyzed using a custom-designed computer algorithm. Noneroded fibrous plaques remote from the erosion site within the same vessel were used as controls. Eroded plaques have significantly lower surface intensity (P<0.001), lower region of interest intensity (P<0.001), lower surface normalized SD (P<0.001), lower region of interest normalized SD (P<0.001), higher optical attenuation (P<0.001), larger tissue protrusion area (P<0.001), and greater surface roughness (P<0.001) whencompared with control plaques. Erosion-adjacent regions also have lower region of interest normalized SD (P=0.008), higher attenuation (P<0.001), and greater surface roughness (P=0.005). Using a logistic regression model built on the quantitative features, plaque erosion can be accurately classified against intact fibrous plaques. There was low inter-and intraobserver variability associated with the algorithm-assisted quantitative assessment. Conclusions: Plaque erosion has distinctive optical properties and morphological features when compared with noneroded fibrous plaques. Quantitative image analysis may enhance diagnostic accuracy for plaque erosion in vivo.