Abstract
BACKGROUND: Ultrasound (US) is the primary modality used to evaluate thyroid nodules (TNs). However, the US features of a portion of benign TNs may suggest malignancy, leading to decreased diagnostic efficacy. Previous studies have reported that contrast-enhanced ultrasound (CEUS) has good efficacy in determining the malignancy of TNs. However, guidelines for thyroid CEUS remain lacking. The purpose of this study was to conduct single-factor and multifactor analysis to assess the value conventional US combined with CEUS in diagnosing TNs. METHODS: In this study, a cohort of 871 patients (345 males and 526 females) with 1,039 TNs was enrolled, and TNs were classified into a benign or malignant group based on the pathological results. TN echogenicity, echo level, aspect ratio, margin regularity and smoothness, extrinsic invasion, presence and degree of calcification, acoustic halo, lesion boundary, enhancement direction, level and distribution, peripheral circular enhancement, and TN texture after enhancement were included by single-factor analysis. A binary logistic regression model was constructed, with statistically significant indices serving as the independent variables. Receiver operating characteristic (ROC) curves and forest plots were used to evaluate the model effectiveness in differentiating benign and malignant TNs. RESULTS: The US and CEUS characteristics of benign and malignant TNs were different. An aspect ratio ≥1, poorly defined boundaries on enhancement, and hypoenhancement were risk factors for malignant TNs (area under the ROC curve =0.891; 95% confidence interval: 0.868-0.913; P<0.001). The specificity, sensitivity, and Youden index of the model in diagnosing malignant TNs were 76.4%, 88.0%, and 0.643, respectively. CONCLUSIONS: An aspect ratio ≥1, unclear boundaries after enhancement, heterogeneous enhancement, centripetal enhancement, hypoenhancement, and stiffness of the TN after enhancement were identified as indicators of malignant TNs.