Logistic回归分析在超声多因素鉴别乳腺良恶性肿瘤中的应用

Application value of Logistic regression analysis in differential diagnosis of malignant and benign breast tumors by ultrasonic multiple factors

目的探讨Logistic回归分析在超声多因素鉴别乳腺良恶性肿瘤中的应用价值。方法选取经手术病理证实的乳腺恶性肿瘤患者119例,良性肿瘤患者47例,分析其声像图特征,包括乳腺肿块内部和后方回声、边缘特征、血流分布、收缩期峰值血流速度(PSV)及阻力指数(RI)。对所有患者的声像图特征行单因素分析,将有统计学差异的指标作多因素Logistic回归分析。结果超声诊断乳腺恶性肿瘤111例,误诊8例,良性肿瘤44例,误诊3例,其对恶性肿瘤诊断的准确性、特异性及敏感性分别为93.4%、93.6%及93.3%,阳性预测值和阴性预测值分别为97.4%和84.6%。单因素分析显示所有观察指标在乳腺良恶性肿瘤的鉴别诊断中差异均有统计学意义(P<0.05)。多因素分析显示边缘毛刺、微钙化、血流分布、阻力指数及纵横比>1与乳腺恶性肿瘤有显著相关性(P<0.05)。结论超声多因素分析对乳腺良恶性肿瘤的鉴别诊断有重要价值。

Objective To investigate the application value of Logistic regression analysis in differential diagnosis of benign and malignant breast tumors by ultrasonic muhiple factors. Methods One hundred and nineteen patients with inalignant tumors and 47 patients with benign tumors confirmed by surgery and pathology were enrolled in this study. Their sonography features were analyzed, including internal and posterior echo, peripheral features, blood distribution, peak systolic velocity （PSV） and resistance index（RI）. The single factor analysis was performed on the sonography features of all the patients, the multiple factors Logistic regression analysis was performed on indexes with significant difference. Results There were 111 cases of malignant breast tumors diagnosed by ultrasound, 8 cases were misdiagnosed, and 44 cases of benign breast tumors diagnosed by ultrasound, 3 cases were misdiagnosed. The accuracy, specificity and sensitivity of uhrasonography in diagnosis of malignant tumors were 93.4% , 93.6% and 93.3% , respectively, the positive predictive value and negative predictive values were 97.4% and 84.6%. Single factors analysis showed there was significant difference of all the indexes between benign and malignant tumors （P 〈 0.05 ）. Multiple factors analysis showed the edge burr, microcalcification, blood distribution, RI and vertical / horizontal ratio 〉 1 were related with malignant breast tmnors （P 〈 0.05）. Conclusion Ultrasonic muhiple factors analysis has important value in differential diagnosis of benign and malignant breast tumors.