乳腺小肿块灰阶和彩色多普勒超声特征的logistic回归分析  

Logistic Regression Analysis of the Features of Small Breast Masses on Gray Scale and Color Dopper Ultrasound

作　　者：顾秀娟[1] 散志华[1] 孙心平[1] 李建国[2] 

机构地区：[1]北京市垂杨柳医院功能科,北京市100022 [2]北京大学人民医院超声科

出　　处：《中国超声医学杂志》2011年第3期209-212,共4页Chinese Journal of Ultrasound in Medicine

摘　　要：目的评价并筛选鉴别乳腺小肿块(≤2 cm)的灰阶和彩色多普勒超声特征指标。方法术前观察71个乳腺小肿块的灰阶和彩色多普勒超声及三维能量多普勒血流特征,将单因素分析得出的与乳腺恶性肿块有关的变量进入回归统计,利用1ogistic逐步回归分析筛选出鉴别乳腺良恶性小肿块有意义的指标并建立诊断恶性可能性公式。结果乳腺良恶性肿块在纵横比、微小钙化、高回声晕、毛刺征、腋窝淋巴结肿大、阻力指数、血管指数、穿入型血流、中心型血流、环绕型血流等方面存在显著性差异(P<0.05),经过logistic逐步回归,得出以下方程:恶性可能性=1/(1+e^(-z)),其z=-4.248+2.276×血管指数+3.736×穿入型血流+3.266×纵横比+3.068×腋窝淋巴结肿大。当乳腺小肿块纵横比≥0.77,出现穿入型血流,血管指数≥1.65条/cm^3,伴腋窝淋巴结肿大时恶性危险可能性0.99。结论以超声特征诊断乳腺恶性小肿块的logistic回归模型有助于鉴别乳腺良恶性病变。Objective To evaluate and screen out the specific characters of small breast masses on gray scale and color Doppler ultrasonography. Methods Seventy one small breast masses were studied with gray scale firstly ,then with color Doppler ultrasonography and three-dimensional color power angiography before surgery,and were analyzed via multivariate logistic regression. Results The significant characters to differentiate malignant from benign were height/width ratio（ H/W）, micro calcification, hyperechoic halo, spicule margin, axillary lymph nodes, resistance index, vascular index（VI）, penetrating vessel, central vessel, peripherl vascular type（P〈0.05）, after logistic regression analysis,a fomula were worked out： Probability of malignancy - 1/（1 ＋ e-Z）, of which z =- 4. 248 ＋ 2. 276 × VI＋ 3. 736 X penetrating vessel＋3. 266 × H/W ＋ 3. 068 ×axillary lymph nodes. If small breast masses with H/W≥0.77, penetrating vessel, VI ≥ 1.65 strip /cm3 and axillary lymph nodes, the probability of malignancy was 0.99. Conclusions The Logistic model is helpful to differentiate benign from malignant breast lesions.

关 键 词：超声检查 乳腺小肿块 LOGISTIC模型 

分 类 号：R445.1[医药卫生—影像医学与核医学] R737.9[医药卫生—诊断学]