目的 通过描述乳腺肿块的超声特征，探索其在乳腺癌筛查中对乳腺癌的预测价值。方法 选取2023年1—12月参与广东省适龄妇女乳腺癌免费筛查项目，行双侧乳腺超声检查，并发现有乳腺单发肿块的妇女3 811例作为研究对象。通过广东省妇幼信息系统收集患者的基本信息、乳腺超声检查和病理学诊断信息。使用logistic回归评估超声图像特征和乳腺癌的关系，再通过R软件构建列线图预测模型。结果 乳腺癌和非乳腺癌妇女在年龄、初潮年龄、文化程度、是否绝经、是否哺乳方面的差异有统计学意义（P <0.001）。乳腺癌与非乳腺癌妇女在肿块大小、形态、方向、边缘后方回声、有无钙化及血流信号等超声特征方面的差异有统计学意义（P <0.001）。乳腺导管原位癌和乳腺浸润癌患者的超声特征在肿块大小和形态方面的差异有统计学意义（P <0.05）。多因素logistic回归分析结果显示，超声肿块大小> 2 cm(OR=2.423,95%CI:1.838～3.181)、形态不规则（OR=2.361,95%CI:1.827～3.046）、边缘不清晰（OR=2.142,95%CI:1.609～2.840）、有钙化灶（OR=1.676,95%CI:1.260～2.216）和有血流信号（OR=2.305,95%CI:1.717～3.077）是乳腺癌的独立预测变量（P <0.05），列线图模型预测乳腺癌的灵敏度69.15%，特异度为77.85%，曲线下面积（Area Under the Curve,AUC）为0.796(95%CI:0.770～0.823)。结论 乳腺肿块大小、形态、边缘、钙化灶和血流信号的超声特征对乳腺癌具有较高预测价值，可为乳腺癌筛查和诊治提供依据。

Objective To describe ultrasonic features of breast masses and explore the predictive value of ultrasonic features in breast cancer screening. Methods From January 2023 to December 2023, 3 811 women who participated in the free breast cancer screening project in Guangdong Province, underwent breast ultrasound examination, and being diagnosed with single breast lump were recruited as research subjects. Data were collected through Guangdong Maternal and Child Information System including their general information, the results of breast ultrasound examination and pathological diagnosis. Logistic regression was used to evaluate the association between ultrasound image characteristics and breast cancer, and then nomogram prediction model was constructed by R software. Results There were significant differences in age, menarche age, education level, reaching menopause and breastfeeding or not between women with or without breast cancer(P < 0.001). The contrast of ultrasound mass features showed that there were significant differences between women with and without breast cancer in terms of mass size, shape, direction, echo behind the edge,calcifying or not and blood flow signals(P < 0.001). There were significant differences in the size and shape of ultrasound mass features, between ductal carcinoma in situ and invasive carcinoma of breast(P < 0.05). Logistic regression analysis showed that tumor size > 2 cm(OR = 2.423, 95% CI: 1.838 ～ 3.181), irregular shape(OR = 2.361, 95% CI: 1.827 ～ 3.046), unclear edge(OR = 2.142,95% CI: 1.609 ～ 2.840), calcification(OR = 1.676, 95% CI: 1.260 ～ 2.216) and blood flow signal(OR = 2.305, 95% CI: 1.717 ～ 3.077)were independent predictors of breast cancer. The sensitivity and specificity of the nomogram prediction model was 69.15% and77.85%, respectively, and the Area Under the Curve(AUC) was 0.796(95% CI: 0.770 ～ 0.823). Conclusion The ultrasonic characteristics of breast mass size, shape, edge, calcification and blood flow signal have great potential to predict breast cancer, which provide evidence for breast cancer screening, diagnosis and treatment.