Background: Doppler ultrasound (DUS) and computed tomography angiography (CTA) are the most commonly used imaging modalities for carotid disease. The aim of this study was to test the accuracy and reproducibility of CTA-derived measurements of carotid stenosis and compare them with those obtained by DUS.
Methods: Images of 100 carotid arteries of patients who underwent carotid DUS at our unit and CTA of the carotids within a 28-day period were identified retrospectively from multidisciplinary team meeting records. CTAs were assessed by 2 investigators, each using a manual and a semi-automated method. With both methods, the degree of stenosis was calculated using the North American Symptomatic Carotid Endarterectomy Trial equation and graded as mild (0-49%), moderate (50-69%), or severe (70-99%). Cohen's kappa and specificity and sensitivity for ≥50% stenosis were calculated.
Results: The interobserver agreement was moderate (κ 0.407, weighted-κ 0.517) for the manual method and good (κ 0.786, weighted-κ 0.842) for the semi-automated method. Using DUS as the gold standard, the semi-automated method had greater sensitivity (75%) and specificity (91%) in detecting clinically significant carotid artery stenosis (≥50%) than the manual one (63% and 86%, respectively). Agreement between DUS and the semi-automated method of CTA reporting was moderate (κ 0.453, 95% confidence interval [CI]: 0.320-0.586, weighted-κ 0.598, 95% CI: 0.486-0.710), whereas DUS and the manual method of CTA reporting had only fair agreement (κ 0.344, 95% CI: 0.209-0.478, weighted-κ 0.446, 95% CI: 0.315-0.577).
Conclusions: CTA tends to underestimate the degree of stenosis when compared with DUS. The semi-automated method of CTA reporting has greater reproducibility and greater agreement with DUS. These findings have practical implications when CTA is used to measure the degree of carotid stenosis in clinical practice.
Copyright © 2018 Elsevier Inc. All rights reserved.