摘要目的:探讨基于多模态超声特征联合炎性复合指标构建的列线图对乳腺影像报告和数据系统（BI-RADS）4类结节良恶性的诊断价值。方法:纳入2021年1月至2024年3月于西北妇女儿童医院就诊并治疗的102例乳腺BI-RADS 4类结节女性患者作为研究对象，根据术后病理检查结果确定结节良恶性。术前1周内，采用迈瑞resona7型超声诊断仪对每位乳腺BI-RADS 4类结节患者进行常规超声、剪切波弹性成像和超声造影检查，记录超声特征。根据实验室检查结果，计算并记录中性粒细胞与淋巴细胞比值（NLR）、血小板与淋巴细胞比值（PLR）、系统性炎症免疫指数（SII）和系统性炎症反应指数（SIRI）。结果:病理检查结果显示恶性结节患者42例，良性结节患者60例。超声显示相较于良性组，恶性组结节边界、内部回声、Alder血流分级、硬环征、增强后面积有无扩大、周围放射状增强和灌注缺损比较差异有统计学意义（ χ2值分别为5.83、9.31、5.20、10.25、6.99、7.24、4.90， P<0.05）。相较于良性组，恶性组的NLR和SIRI显著增高，差异有统计学意义（ t值分别为3.13和4.33， P<0.05）。采用多因素Logistics回归分析，表明结节边界模糊（X1）、Alder血流分级2~3级（X3）、出现硬环征（X4）、增强后面积扩大（X5）、周围放射状增强（X6）及 NLR≥2.29（X8）、 SIRI≥0.91（X9）为乳腺BI-RADS 4类结节诊断为恶性结节的独立危险因素（ P<0.05），基于这些因素，建立Logistics回归模型为Logit（P）= -3.217 + 0.291 X1 + 1.051 X3 + 0.354 X4 + 0.360 X5 + 0.524 X6 + 0.513 X8 + 0.726 X9。构建列线图模型，校准曲线和临床决策曲线评估结果显示该模型有较好的准确度和较高的预测价值。ROC曲线分析结果表明列线图模型预测乳腺BI-RADS 4类结节良恶性的曲线下面积为0.925（0.877~0.974），灵敏度为88.6%，特异性为86.1%。 结论:多模态超声联合炎性复合指标NLR、SIRI构建的列线图模型在乳腺BI-RADS 4类结节良恶性的鉴别诊断中有一定应用价值。

abstractsObjective:To explore the diagnostic value of the nomogram constructed based on multi-modal ultrasound features combined with inflammatory complex indicators for benign and malignant nodules in breast imaging reporting and data system (BI-RADS) .Methods:102 female patients with breast BI-RADS type 4 nodules who were treated in Northwest Women’s and Children’s Hospital from Jan. 2021 to Mar. 2024 were included as study objects, and benign and malignant nodules were determined according to postoperative pathological examination results. Conventional ultrasound, shear-wave elastography and contrax-ultrasound were performed in each breast BI-RADS patient with Class 4 nodules using Mindray resona7 ultrasonic diagnostic instrument within 1 week before surgery to record the ultrasonic characteristics. According to the results of laboratory examination, neutrophil to lymphocyte ratio (NLR) and platelet to lymphocyte ratio (NLR), PLR, the systemic immune inflammation index (SII), and the systemic inflammatory response index (SIRI) were calculated and recorded.Results:There were 42 cases of malignant nodules and 60 cases of benign nodules. Compared with the benign group, there were significant differences in nodule boundary, internal echo, Alder blood flow grading, hard ring sign, area expansion after enhancement, peripheral radial enhancement and perfusion defect in the malignant group ( χ2 value was 5.83, 9.31, 5.20, 10.25, 6.99, 7.24, 4.90, P < 0.05). Compared with the benign group, NLR and SIRI in the malignant group were significantly increased, with statistical difference (t-value 3.13 and 4.33, P < 0.05). The results of multi-factor Logistics regression analysis show that breast BI-RADS were characterized by blurred nodule boundary (X1), Alder blood flow grade 2-3 (X3), hard ring sign (X4), enlarged area after enhancement (X5), peripheral radial enhancement (X6), NLR≥2.29 (X8), SIRI≥0.91 (X9) were independent risk factors for malignant nodules ( P < 0.05). Based on these factors, the Logistics regression model was established as Logit (P) = -3.217 + 0.291X1 + 1.051X3 + 0.354X4 + 0.360X5 + 0.524X6 + 0.513X8 + 0.726X9. The results of calibration curve and clinical decision curve evaluation showed that the model had good accuracy and high predictive value. ROC curve analysis results showed that the area under the curve of the nomogram model for predicting benign and malignant nodules of category 4 breast BI-RADS was 0.925 (0.877-0.974), the sensitivity was 88.6%, and the specificity was 86.1%. Conclusion:The nomogram model constructed by multi-modal ultrasound combined with inflammatory complex index NLR and SIRI has certain application value in the differential diagnosis of benign and malignant breast BI-RADS 4 nodules.