摘要
目的通过对最大径线≤15 mm的乳腺内小结节彩色多普勒超声图像的分析,探讨其声像图特征性表现,运用Logistic回归分析筛选鉴别小乳癌敏感声像图特征,并对其危险度进行排序,以提高超声诊断小乳癌的准确率.方法收集经手术病理证实的166个最大径线≤15 mm乳腺小结节病例,进行回顾性分析,对其声像图特征进行统计学赋值,并进行二分类Logistic回归分析,建立Logistic回归模型,依据OR值对各特征的危险度进行排序.结果二分类Logistic回归分析显示9个超声特征引入Logistic回归模型方程,分别为:Cooper韧带、浅筋膜和腺体模糊中断(9.182)、结节周边成角、呈锯齿样改变(7.675)、含有砂粒状微小钙化(7.471)、纵横比大于1(4.776)、毛刺(3.982)、无侧方回声失落(2.277)、后方回声有衰减(1.861)、内部回声不均匀(1.640)、形态不规则(1.097).结论以超声特征诊断乳腺恶性病变的Logistic回归模型有助于鉴别乳腺内良恶性小结节病变,在小乳癌的诊断中具有较大的鉴别意义.
Objective To investigate the features by analyzing the sonograms of small breast nodules（≤15 mm）.The points for differential diagnosis of breast cancer through Logistic regression analysis were selected and risk rank was disposed to improve the accurate rate of differential diagnosis of small breast cancer by Ultrasound.Methods 166 cases of small breast nodules（≤15 mm）confirmed by pathological examination were collected and retrospectively analyzed,a Logistic model for predicting breast malignancy on the basis of ultrasonographic features was obtained and the risk degrees were ranked through OR values.Results The Logistic regression analysis demonstrated that 9 ultrasonic features were enrolled in the Logistic equation,Indistinct and discontinue of cooper ligament,super fascia and lobule of breast（9.182）,angular margin（7.675）,microcalcification（7.471）,shape taller than wide（4.776）,spicular sign（3.982）,absence of lateral shadow（2.277）,posterior attenuation（1.861）,uneven echogenicity（1.640）,irregular shape（1.097）.Conclusion The Logistic model of sonographic features is helpful to differentiate the benign small nodules from the malignant ones,and has significant value in differential diagnosis of small breast cancer.
出处
《昆明医学院学报》
2010年第12期60-65,共6页
Journal of Kunming Medical College
