Accuracy of dual-source computed tomography in quantitative assessment of low density coronary stenosis--a motion...

We assessed the accuracy and reproducibility of non-calcified plaque quantification as simulated by a low-density stenosis in vessel phantoms using diameter and area measures, as well as the influence of vessel size and motion on quantification accuracy in dual-source computed tomography (DSCT).

Four phantoms (2, 2.5, 3, and 4 mm in luminal diameter) made from a radiopaque Lucite (126 +/- 23 Hounsfield units, HU) simulating a fixed radiolucent concentric coronary stenosis (7 +/- 2 HU, 50% luminal narrowing) were connected to a cardiac motion simulator. Stenosis quantification was based on area and diameter measurements. All measurements were highly reproducible (all ICC > or =0.95, p < 0.001).

The mean measured degree of stenosis was 38.0 +/- 11.7% for a single diameter measurement, resulting in a mean relative error of 22.0 +/- 18.7%, decreasing with increasing phantom size (31.9 +/- 22.1%; 25.2 +/- 20.9%; 16.3 +/- 12.8%; 14.5 +/- 11.4%; for 2-, 2.5-, 3-, and 4-mm phantoms, respectively; p < 0.0001). Measurement accuracy significantly increased to 13.3 +/- 13.9% by using area measurement (p < 0.0001). The degree of stenosis was not significantly different when comparing a motioned image with an image at rest.

DSCT enables highly reproducible quantification of low density stenosis, but underestimates the degree of stenosis, especially in small vessels. Area-based measurements reflect the true degree of stenosis with higher accuracy than diameter.