Quantification of coronary artery plaque using 64-slice dual-source CT: comparison of semi-automatic and automatic computer-aided analysis based on intravascular ultrasonography as the gold standard

Abstract

We evaluated the feasibility of automatic computer-aided analysis (CAA) compared with semi-automatic CAA for differentiating lipid-rich from fibrous plaques based on coronary CT angiography (CCTA) imaging. Seventy-four coronary plaques in 57 patients were evaluated by CCTA using 64-slice dual-source CT. Quantitative analysis of coronary artery plaques was performed by measuring the relative volumes (low, medium, and calcified) of plaque components using automatic CAA and by measuring mean CT density using semi-automatic CAA. We compared the two plaque measurement methods for lipid-rich and fibrous plaques using Pearson's correlation. Intravascular ultrasonography was used as the goal standard for assessment of plaques. Mean CT density of plaques tended to increase in the order of lipid [36 ± 19 Hounsfield unit (HU)], fibrous (106 ± 34 HU), and then calcified plaques (882 ± 296 HU). The mean relative volumes of 'low' components measured by automatic CAA were 13.8 ± 4.6, 7.9 ± 6.7, and 3.5 ± 3.0 % for lipid, fibrous, and calcified plaques, respectively (r = -0.348, P = 0.022). The mean relative volumes of 'medium' components on automatic CAA were 12.9 ± 4.1, 15.7 ± 9.6, and 5.6 ± 4.8 % for lipid, fibrous, and calcified plaques, respectively (r = -0.385, P = 0.011). The mean relative volumes of low and medium components within plaques significantly correlated with the types of plaques. Plaque analysis using automatic CAA has the potential to differentiate lipid from fibrous plaques based on measurement of the relative volume percentages of the low and medium components.