目的： 探讨超声黏弹性成像技术在乳腺肿瘤良恶性鉴别中的应用价值。 方法： 连续纳入2023年2月至2023年8月期间，上海交通大学医学院附属瑞金医院收治的经手术病理证实为乳腺肿瘤的717例患者，其中471例为恶性，246例为良性。所有患者均在治疗前进行乳腺超声检查，包括灰阶超声、超声应变弹性成像、超声剪切波弹性成像、超声黏性成像。超声黏弹性成像技术包括测量肿瘤及其周围组织的黏性系数、频散系数和剪切波弹性模量、应变比等4组参数，以4组参数中的较佳预测指标，分别构建多种预测模型，包括黏性系数单变量模型、频散系数单变量模型、黏性组合模型（Shell/T-Vmean+Shell/T-Dmean）、剪切波单变量模型、应变单变量模型、乳腺影像报告和数据系统（Breast Imaging Reporting and Data System， BI-RADS）、BI-RADS联合黏性组合模型，评估每种模型在乳腺肿瘤良恶性鉴别中的效能。 结果： 超声黏弹性成像的黏性系数、频散系数、弹性模量及应变比等参数均可有效地区分乳腺良恶性肿瘤，其中肿瘤边缘2 mm区域与瘤体的参数比值Shell/T-Vmean、Shell/T-Dmean、Shell/T-Emean、Strain Ratio A为较佳预测指标，曲线下面积分别为0.742、0.745、0.726、0.705，而BI-RADS模型预测乳腺肿瘤良恶性的0.822。将Shell/T-Vmean、Shell/T-Dmean分别与BI-RADS分类联合时，受试者操作特征曲线的曲线下面积高达0.895（95%CI为0.868～0.917），高于BI-RADS。 结论： 超声黏弹性成像的黏弹性参数中，肿瘤边缘2 mm区域与瘤体的黏性系数、频散系数及弹性模量均值比为关键诊断指标；Shell/T-Vmean、Shell/T-Dmean联合BI-RADS后，可为术前无创精准诊断提供了新策略。

Objective To evaluate the application value of ultrasonic viscoelastic imaging technology in differentia-ting benign and malignant breast tumors. Methods A total of 717 patients with breast tumors confirmed by surgical patho-logy were consecutively enrolled at Ruijin Hospital, Shanghai Jiao Tong University School of Medicine between February 2023 and August 2023, including 471 malignant and 246 benign cases. All patients underwent breast ultrasound examinations before treatment, including grayscale ultrasound, ultrasound strain elastography, ultrasound shear wave elastography, and ultrasound viscoelastic imaging. Ultrasound viscoelastic imaging technology included measuring four sets of parameters of the tumor and its surrounding tissues: viscosity coefficient, dispersion coefficient, shear wave elastic modulus, and strain ratio. Using the optimal predictive indicators from the four parameter groups, multiple prediction models were established, including single-variable models (viscosity coefficient, dispersion coefficient, shear wave, strain), a combined viscoelastic model (Shell/T-Vmean + Shell/T-Dmean), a Breast Imaging Reporting and Data System (BI-RADS) model, and a combined model integrating BI-RADS with viscoelastic parameters. The effectiveness of each model in differentiating benign and malignant breast tumors was evaluated. Results Parameters including viscosity coefficient, dispersion coefficient, elastic modulus, and strain ratio from ultrasound viscoelastic imaging could effectively distinguish benign and malignant breast tumors. Among them, the ratios of the tumor margin (2 mm region) to the tumor itself—Shell/T-Vmean, Shell/T-Dmean, Shell/T-Emean, and Strain Ratio A—were optimal predictive indicators, with areas under the curve (AUCs) of 0.742, 0.745, 0.726, and 0.705, respectively. The BI-RADS model for predicting benign and malignant breast tumors achieved an AUC of 0.822. When Shell/T-Vmean and Shell/T-Dmean were respectively combined with BI-RADS classification, the receiver operating characteristic (ROC) curve's AUC reached 0.895 (95% CI: 0.868–0.917), which was higher than that of BI-RADS alone. Conclusion Among the viscoelastic parameters of ultrasound viscoelastic imaging, the average ratios of viscosity coefficient, dispersion coefficient, and elastic modulus between the 2 mm tumor margin region and the tumor body are key diagnostic indicators. The combination of Shell/T-Vmean and Shell/T-Dmean with BI-RADS provides a new stra-tegy for noninvasive preoperative precision diagnosis.