ABSTRACT:Objective:Carotid ultrasound (US) is a screening test for patients with transient ischemic attacks (TIAs) or stroke who then undergo Digital Subtraction Angiogram (DSA) or Magnetic Resonance Angiography (MRA). Gold standard DSA is invasive with inherent risks and costs. MRA is an evolving technology. This study compares reliability of MRA and US modes with DSA in determining degree of internal carotid artery stenosis.Methods:A five year retrospective analysis of 140 carotid arteries from patients who had carotid US and DSA, and possibly Magnetic Resonance Angiography was undertaken. Recorded US parameters were peak systolic velocity (PSV), end diastolic velocity (EDV), and ICA/CCA peak systolic velocity ratio. The MRA and DSA parameters used NASCET technique for measuring stenosis. Statistical analysis included ROC curves and Kappa computation.Results:US grading of carotid stenosis can be made more reliable by choosing appropriate parameters. The best combination of sensitivity and specificity for stenosis > 70% in our hospital was seen at PSV > 173cm/s (sensitivity 0.87, specificity 0.8, Positive Predictive Value (PPV) 0.70, Negative Predictive Value (NPV) 0.93, kappa 0.64 and weighted kappa 0.71). MRA kappa was 0.78, (sensitivity 0.75, specificity 1.0, PPV 1.0, NPV 0.85).Conclusions:US parameters should be validated in each centre. At best, US can only approximate the accuracy of DSA, probably due to inherent limitations of this modality. Magnetic Resonance Angiography has a perfect specificity and PPV but this technique needs to be standardized. Simultaneous use of MRA and US for screening increases sensitivity to over 0.9 without compromising specificity in > 70% stenosis.