Objective To construct a quantitative diagnostic model for predicting the nature of breast nodules based on multimodal ultrasound (color Doppler flow imaging, two-dimensional grayscale ultrasound, and ultrasound elastography) and analyze its predictive value.

Methods A total of 110 patients with surgical treatment for breast nodules (135 nodules in total) in the hospital from June 2020 to January 2023 were selected as research objects. All patients received ultrasound examinations before surgery. According to the postoperative pathological results, they were divided into benign group and malignant group. The general data, color Doppler flow imaging grading, two-dimensional grayscale ultrasound features, and ultrasound elastography score were compared between the two groups. Principal component analysis (PCA) and a multivariate Logistic regression model were used to explore the risk factors affecting the nature of breast nodules and construct a risk prediction model. The receiver operating characteristic (ROC) curve was drawn to analyze the value of the model in predicting the nature of breast nodules.

Results The postoperative pathological results showed that among the 135 breast nodules, there were 73(54.07%) benign nodules and 62(45.93%) malignant nodules. The proportion of color Doppler flow imaging grades Ⅱ and Ⅲ in the malignant group was significantly higher than that in the benign group (P < 0.05). The proportions of irregular shape, non-parallel orientation, irregular margin, presence of microcalcifications, posterior echo attenuation/mixture, and abnormal lymph node morphology in the malignant group were all significantly higher than those in the benign group (P < 0.05). The proportion of elastography scores of 4 to 5 for breast nodules in the malignant group was significantly higher than that in the benign group (P < 0.05). PCA and Logistic regression analysis results showed that color Doppler flow imaging grades Ⅱ and Ⅲ (OR=2.474, 95%CI, 1.226 to 4.991), irregular shape (OR=2.120, 95%CI, 1.055 to 4.261), irregular margin (OR=4.206, 95%CI, 2.030 to 8.716), non-parallel orientation (OR=2.329, 95%CI, 1.137 to 4.768), posterior echo attenuation/mixture (OR=2.983, 95%CI, 1.398 to 6.365), microcalcifications (OR=4.615, 95%CI, 2.117 to 10.062), abnormal lymph node morphology (OR=4.194, 95%CI, 1.762 to 9.984), and elastography scores of 4 to 5 (OR=2.246, 95%CI, 1.125 to 4.481) were all significantly positively correlated with malignant breast nodules. The ROC curve showed that the area under the curve (AUC) of the Logistic regression model for predicting the nature of breast nodules was 0.906 (95%CI, 0.849 to 0.963), with a sensitivity of 0.857, a specificity of 0.868, and a Youden index of 0.725.

Conclusion The prediction model constructed based on multimodal ultrasound (color Doppler flow imaging, two-dimensional grayscale ultrasound, and ultrasound elastography) can effectively differentiate benign and malignant breast nodules. The model is reliable and stable, and conductive to reduce unnecessary puncture biopsies and surgeries.