Application Of Logistic Regression Analysis In The Diagnosis Of Small Breast Nodules

Objective:To analyze the ultrasonographic features of benign and malignant breast nodules with diameter?1cm,and to establish a logistic regression model based on ultrasonography to improve the accuracy of ultrasonography in small breast nodules.Methods:A retrospective analysisi of the a total of 344 small nodules were diagnosed by surgical resection and confirmed by pathology,including 265 benign small nodules and 79 small malignant nodules.The ultrasonic features of breast small nodules and the age of patients was collected,the ultrasonic features including internal echo,internal echo uniformity,shape,boundary,margin,anteroposterior and transverse diameter ratio?A/T?,calcification,posterior acoustic,blood flow grading,blood flow resistance index,calcification,hyperechoic halo,ipsilateral axillary lymph nodes and the patient of age.The ultrasonogra-phic features of small breast nodules and ipsilateral axillary lymph nodes were used to diagnose the malignant nodules of breast.All the collected ultrasonic features and age as independent variables,pathology as the dependent variable,using binary Logistic regression based on maximum likelihood estimation in partial forward stepwise regression filter out the most significant risk factors,then to establish regression model and evaluation of regression coefficients by wald?² test,use likelihood-ratio test to estimate the goodness-of-fit of the regression model.The receiver operating characteristic curve?ROC?is used to evaluate the regression model and preoperative ultrasonography diagnostic efficacy.Results:The ultrasonography of breast small nodules that low internal echo,rear echo attenuation,irregular shape,A/T?0.70 showed the sensitivity was 94.9%,8.9%,64.6%,78.5%,and the specificity was14.3%,96.6%,81.1%,69.1%.The ultrasonography of breast small nodules that fuzzy boundary,irregular margin,A/T?1,microcalcifica-tion,hyperechoic halo,ipsilateral axillary lymphadenectasis,adler flow grade?-?,resistance index?1,the sensitivity range of was 29.1%-43%,the specificity range was 90.2%-98.1%.And all the ultrasonographic features in the?² test were statistically significant?P<0.05?.Logistic regression analysis showed that the ultrasonographic features of the regression model were internal echo uniformity,shape,margin,aspect ratio²?A/T?0.70?,flow grade,flow resistance index,hyperechoic halo,microcalcification,ipsilateral axillary lymphadenectasis.Establishment of regression model[Logit?P?=-5.993+2.075×Internal echo uniformity+1.18×shape+1.668×margin+1.467×aspect ratio²+2.529×flow grade+2.087×f-low resistance index+4.951×hyperechoic halo+2.105×microcalcification+3.434×ipsilateral axillary lymphadenectasis+1.390×age].Use likelihoo-d ratio test to estimate the goodness-of-fit of the regression model was statistically significant??²=111.454,P=0.000?.According to the proba-bility value P=1/[1+e^z],Z is substituted,if P>0.5 predicted as malignant nodules,P?0.5 prediction for benign nodules,use the logistic regression model to evaluate the 344 small breast nodules in this study,the correct rate was 94.2%.The correct rate of preoperative ultrasonography was87.2%.The area under the curve?AUC?of logistic regression model and preoperative ultrasonography were 0.918,0.793,P<0.05,95%confidenc-e interval were?0.873,0.963?,?0.726,0.859?.Indicating that the model has a high value in predicting the properties of small nodules of mammary gland.Conclusion:1.The internal echo uniformity,shape,margin,A/T?0.70,flow grade,flow resistance index,hyperechoic halo,microcalcification,ipsilateral axillary lymphadenectasis were found in the ultrasonographic features of breast nodules in differential diagnosis of small breast nodules has a certain significance.2.The accuracy of the logistic regression model and the area under the ROC curve were higher than the preoperative ultrasound,and had higher clinical practicability.