Abstract
BACKGROUND: Previous studies show the size of lesions could affect the diagnostic accuracy of contrast-enhanced ultrasound (CEUS). It is unclear whether CEUS has good diagnostic performance for lesions ≤2.0 and ≤1.0 cm. It is beneficial for the early diagnosis to explore the application of CEUS in breast lesions of different sizes. This study aims to analyze the diagnostic performance of CEUS and explore diagnostic models better suited to breast lesions of different sizes. METHODS: A total of 1,059 lesions (656 benign and 403 malignant) examined by ultrasound and CEUS with definite pathological results were included in this retrospective study and divided into training (n=847) and validation (n=212) sets. All lesions were divided into three groups according to size. Diagnostic models (M(0): all lesions; M(1): ≤1.0 cm, M(2): >1.0-2.0 cm, and M(3): >2.0 cm) were developed through logistic regression analyses of CEUS features from the training set. Diagnostic performance was evaluated using the area under the receiver operating characteristic curve (AUC) and validated in the validation set. RESULTS: The median age of patients was 45±11 years (range, 18-80 years). The AUC values of M(0) combined with the Breast Imaging Reporting and Data System (BI-RADS) in the training and validation sets were 0.921 and 0.922, respectively (P=0.893). The AUC values of M(0) combined with BI-RADS in the three groups were 0.844, 0.936 and 0.928 respectively. M(0) was less effective in diagnosing lesions ≤1.0 cm (0.844 vs. 0.921, P=0.029). The AUC of M(1) combined with BI-RADS for lesions ≤1.0 cm was higher than that of M(0) (0.893 vs. 0.844, P=0.047), and M(2) and M(3) had no statistical difference in diagnostic performance when compared with M(0) (P=0.243; P=0.246). CONCLUSIONS: The diagnostic performance of CEUS was closely related to lesion size. Establishing a new diagnostic model for lesions ≤1.0 cm can improve the CEUS diagnostic performance for breast lesions ≤1.0 cm.