Accuracy of dual-source CT in the characterisation of non-calcified plaque: use of a colour-coded analysis compared with virtual histology intravascular ultrasound

Abstract

Non-invasive assessment of plaque volume and composition is important for risk stratification and long-term studies of plaque stabilisation. Our aim was to evaluate dual-source computed tomography (DSCT) and colour-coded analysis in the quantification and classification of coronary atheroma. DSCT and virtual histology intravascular ultrasound (IVUS-VH) were prospectively performed in 14 patients. 22 lesions were compared in terms of plaque volume, maximal per cent vessel stenosis and percentages of fatty, fibrous or calcified components. Plaque characterisation was performed with software that automatically segments luminal or outer vessel boundaries and uses CT attenuation for a colour-coded plaque analysis. Good correlation was found for per cent vessel stenosis in DSCT (53+/-13%) and IVUS (51+/-14%; r(2) = 0.70). Mean volumes for entire plaque and non-calcified atheroma were 68.5+/-33 mm(3) and 56.7+/-30 mm(3), respectively, in DSCT and 60.8+/-29 mm(3) and 55.8+/-26 mm(3), respectively, in IVUS. Mean percentages of fatty, fibrous or calcified components were 28.2+/-6%, 53.2+/-9% and 18.7+/-13%, respectively, in DSCT and 29.9+/-5%, 55.3+/-12% and 14.4+/-9%, respectively, in IVUS-VH. Significant overestimation was present for the entire plaque and the volume of calcified plaque (p = 0.03; p = 0.0004). Although good correlation with IVUS was obtained for the entire plaque (r(2) = 0.76) and non-calcified plaque volume (r(2) = 0.84), correlation proved very poor and insignificant for percentage plaque composition. Interclass correlation coefficients for non-calcified plaque volume and percentages of fatty, fibrous or calcified components were 0.99, 0.99, 0.95 and 0.98, respectively, and intraclass coefficients were 0.98, 0.93, 0.98 and 0.99, respectively. We found that using Hounsfield unit-based analysis, DSCT allows for accurate quantification of non-calcified plaque. Although percentage plaque composition proves highly reproducible, it is not correlated with IVUS-VH.

Figure 1

(a) Bland–Altman analysis comparing non-calcified plaque volume in dual-source CT (DSCT) and intravascular ultrasound (IVUS). (b–d) Comparison of percentage plaque composition between DSCT and IVUS-VH. The x-axis denotes the average of DSCT and IVUS, and the point of intersection with the y-axis indicates the bias of DSCT. Dotted lines show one standard deviation of the bias.

Figure 2

(a) Correlation of non-calcified plaque volume in dual-source CT (DSCT) and intravascular ultrasound (IVUS). (b–d) Correlation of percentage plaque composition in DSCT with IVUS-VH.

Figure 3

Virtual plaque histology in dual-source CT (DSCT) and intravascular ultrasound (IVUS). In a curved planar reconstruction (a), the arrow indicates a non-calcified plaque in segment 6 of the left anterior descending coronary artery. (b) Software automatically performs segmentation of luminal and outer wall vessel boundaries. Contrast the standard axial CT image (c) and colour-coded overlay with the grey-scale IVUS (e) and IVUS-VH (f). Despite a satisfactory match of plaque components, IVUS-VH excels at higher spatial resolution and reveals a greater level of complexity.