K-Means Clustering Algorithm-Based Detection of Carotid Atherosclerotic Plaque Using Contrast-Enhanced Ultrasound Images

Abstract

The aim of this study was to investigate the diagnostic value of contrast-enhanced ultrasound (CEUS) based on the K-means clustering (KMC) algorithm for the vulnerability of carotid atherosclerotic plaque (CAA). In this study, 90 patients with CAA were enrolled into a control group (group A) and an experimental group (group B). The angiography method and KMC-based ultrasound detection were applied to diagnose the CAA patients from the two groups, respectively. The results showed that the sensitivity, specificity, and positive predictive value of patients from group B (92.3%, 90.1%, and 94.8%) for diagnosing CAA were obviously higher than those of patients from group A (81.4%, 88.6%, and 75.3%) ( $P<0.05$ ). The detection rate of patients from group B (83%, 85%) was dramatically higher than that of patients from group A (65%, 71%) in terms of artery bifurcation and CAA ( $P<0.05$ ). Besides, patients from group B were more satisfied with their diagnostic method than group A ( $P<0.05$ ). In conclusion, the ultrasound detection method based on KMC had high sensitivity, specificity, and accuracy in the detection of CAA. In addition, ultrasound detection was better than angiography in the diagnosis of plaque in different parts, and it was worthwhile to apply the ultrasound detection method based on KMC in clinical practice.