Computer-Assisted Analysis of Heterogeneity on B-Mode Imaging Predicts Instability of Asymptomatic Carotid Plaque

BACKGROUND: Computerized assessment of plaque echogenicity by B-mode ultrasonography has demonstrated that the gray-scale median (GSM) pixel intensity of the entire plaque predicts future ischemic stroke in patients with symptomatic carotid stenosis, but not those with asymptomatic stenosis. This study investigated whether plaque heterogeneity (i.e., the distribution of pixel intensities) could predict the instability of asymptomatic plaque. METHODS: By comparison with carotid endarterectomy specimens and the GSM values of known tissues on B-mode images, the GSM values for blood, lipid, muscle/fibrous tissue, and calcification were determined. Then we estimated the percent area of each tissue component for 297 asymptomatic plaques causing 40-99% carotid artery stenosis in 250 patients, and monitored the incidence of atherothrombotic cerebral infarction due to carotid stenosis during follow-up. RESULTS: Eight infarcts occurred during a follow-up period of 22 +/- 15 months. Plaques in the top tertile for the percent area of lipid-like echogenicity (p < 0.05) and in the lowest tertile for calcification (p = 0.06) showed an association with future infarction according to Kaplan-Meier analysis. This association remained significant after adjustment for the severity of carotid stenosis (hazard ratio 4.4 for lipid-like and 0.24 for calcification-like component, both p < 0.05) according to Cox proportional hazards analysis. CONCLUSIONS: The distribution of pixel intensities in carotid plaque on B-mode ultrasonography can be employed to predict instability of asymptomatic plaque and possibly to select patients for interventional procedures. A large-scale investigation will be needed to confirm that estimating the percentage of plaque components relative to the total plaque area can predict ischemic stroke.