摘要
目的应用Logistic回归模型对乳腺实性结节的超声特征进行分析,筛选出对其良恶性鉴别诊断具有统计学意义的特征变量。方法对95例乳腺结节患者共106个实性结节进行常规二维超声成像和剪切波弹性成像,记录其常规超声声像图表现及弹性模量值特征,所有结节均经手术或穿刺活检获得病理结果,建立Logistic回归模型,对各个特征指标进行分析评价。结果经过Logistic回归分析,选取进入回归方程的特征变量有:形态、边界、RI、内部回声、后方回声衰减和弹性模量值六个指标。结论 Logistic回归分析能有效地筛选出对乳腺实性结节良恶有鉴别意义的指标。
Objective By Logistic regression model,to analyze the ultrasonic characters of breast solid nodule.Screened out of their differential diagnosis of benign and malignant characters with statistically significant variables.Methods 95 patients with 106 breast solid nodules for conventional two-dimensional ultrasound imaging and shear wave elasticity imaging,recorded the performance of conventional sonographic features and elastic modulus.All the nodules were confirmed by surgical pathologyand and established Logistic regression model to analyze the various characters of indicators.Results The characters of selected variables into the regression equation are： Morphology,boundary,internal echo,RI,rear echo and elastic modulus.Conclusions Logistic regression analysis can effectively filter out the meaningful benign and malignant characters of breast solid nodules.
出处
《齐齐哈尔医学院学报》
2012年第18期2449-2450,共2页
Journal of Qiqihar Medical University
