Assessing internal carotid artery stenosis with a semiautomated computed tomography angiography tool and duplex ultrasound

Duplex ultrasound (DUS) and computed tomography angiography (CTA) are both used as first-line noninvasive methods to investigate patients for internal carotid artery (ICA) disease. Although manual assessment of CTA is well established, semiautomated vessel analysis programs have yet to prove their clinical benefit. We compared one such vessel analysis program (TeraRecon, Foster City, Calif) with DUS.

A total of 85 arteries in 50 patients (35 men, 15 women; mean age, 73 ± 10 years) were eligible for comparison with the North American Symptomatic Carotid Endarterectomy Trial method. Duplex scanning comprised stenosis estimation based on (1) the intrastenotic and distal ICA diameter measurements on color-coded imaging (CCI), (2) the application of German Society for Ultrasound in Medicine (DEGUM) criteria (intrastenotic peak systolic velocity [PSV] ≥ 2 00 cm/s indicates 50% stenosis; intrastenotic PSV ≥ 300 cm/s together with a PSV of ≥ 50 cm/s in the distal ICA indicates 70% stenosis), and (3) the application of the University of Washington stenosis criteria (≥ 50% stenosis is indicated by PSV >125 cm/s and end-diastolic velocity <140 cm/s; ≥ 80% stenosis is indicated by PSV >125 cm/s and end-diastolic velocity ≥ 140 cm/s). The minimum and maximum diameters at the stenosis site and at the distal reference ICA were automatically measured with the CTA vessel analysis tool. In addition, automated tracking generated corresponding cross-sectional areas at these two sites. Angiographic stenosis was then calculated using the minimum diameter (CTAmin), the average of the minimum and maximum diameters (CTAavg), and the areas (CTAarea) at both sites.

Compared with duplex CCI, the three CTA modalities exhibited only a moderate agreement in terms of regression analysis (R(2) = 0.41-0.54) and Bland-Altman analysis (the standard deviation of the stenosis differences was >20%). In terms of sensitivity, specificity, positive predictive value, negative predictive value, and accuracy, DEGUM stenosis graduation was best balanced by duplex CCI (50% stenosis: 100%, 93%, 85%, 100%, 95%; 70% stenosis: 71%, 100%, 100%, 97%, 98%) followed by CTAarea (50% stenosis: 80%, 73%, 54%, 90%, 75%; 70% stenosis: 66%, 94%, 55%, 96%, 96%). University of Washington stenosis was best balanced by duplex CCI followed by CTAarea.

CTA analysis with a semiautomated vessel analysis tool provides variable results. Large discrepancies between methods in the degree of reported stenosis must be taken into consideration when CTA and DUS are used for clinical purposes. The semiautomated software tools need further improvements.