CT angiography with volume rendering for quantifying vascular stenoses: in vitro validation of accuracy.

OBJECTIVE The purpose of this study was to evaluate the accuracy of CT angiography with volume rendering for quantifying vascular stenoses in vitro. MATERIALS AND METHODS Vascular models with three degrees of stenosis (33%, 67%, and 83%) were imaged at three orientations to the axial plane (parallel, perpendicular, or 45 degrees ) using helical CT with 2-mm collimation and two pitches (1 or 2), two reconstruction intervals (1 or 2 mm), and two scan times (.75 or 1 sec). Diameter and percentage of stenosis were measured from volume renderings using full width at half maximum. Images were measured in two planes whenever resolution varied with direction. Statistical analysis was performed using analysis of variance. RESULTS Mean absolute error of the measured percentage of stenosis was 7% (range, 0-27%). The actual percentage of stenosis and vessel orientation had the most significant effects on accuracy (p < .001). The measured percentage of stenosis was significantly less accurate with phantoms parallel to the axial plane than with other orientations (p < .001). Mean absolute error in the measured percentage of stenosis was 4% when the parallel-to-the-axial-plane orientation was excluded. Overlapping (1-mm) reconstructions were significantly more accurate than 2-mm reconstructions (p < .05) and direction of measurement significantly affected accuracy (p < .05), but these effects were secondary. CONCLUSION CT angiography with volume rendering can accurately quantify vascular stenoses, but it is less accurate for vessels in the axial plane. With 2-mm collimation, vessel characteristics have greater effects on accuracy than do acquisition parameters.