Quantitative Study On Diagnostic Indexes Of Ultrasound Breast Imaging Reporting And Data System(Fifth Edition)

Objectives Logistic regression model was used to analyze the application value of ultrasound diagnostic indexes in the fifth edition of the ultrasound breast imaging reporting and data system(BI-RADS-US)in the diagnosis of benign and malignant breast tumors.Based on this,we establishment a scoring and categorizing method of the BI-RADS-US system(3-5 class),to provide an objective and effective evaluation criteria for ultrasound doctors.Methods We reviewed and analyzed 401 patients who have sonographic features of breast nodules with the descriptors of breast imaging reporting and data system linked to age factors.The pathological examination was the diagnostical golden standard in order to build Logistic regression,which can analysis perform to predict significant variables.A scoring system and categorizing method was developed based on the results of the Logistic regression model and the weighting of each factor.In addition,243 cases of breast tumor were selected as the test group,the breast lesions were classified according to the scoring system,we will calculate each type of positive predictive value and compare with the theorized risk range of malignancy to observe the consistency between them.And then evaluate the diagnostic efficacy of the new scoring and categorizing method.Results 1.The results of multiple regression analysis were as followed:age(?40years),morphology(irregular shape),orientation(not parallel),margin(notcircumscribed),echo pattern(heterogeneous)and microcalcifications.2.According to the results of the regression model and the weighting of each factor,a scoring and categorizing method was established,BI-RADS 3,4a,4b,4c,5 categories are corresponding to score 6,7-8,9-15,16-22,?23.3.Base on BI-RADS,the positive predictive value of the model cases was 0%,2.67%,16.18%,90.74%,100%.The positive predictive value of the test cases was 0%,4.17%,21.43%,84.85%,100%.The positive predictive value of the study cases was 0%,3.25%,17.70%,88.51%,100%.4.The area under the ROC(receiver operating characteristic)curve of test cases was0.947.With 14 points as the Cut-off value,the sensitivity was 90.10%,specificity was90.14%,Youden index was 0.80,the accuracy rate was 90.12%.Conclusions Logistic regression model was established based on sonographic features and formulate the initial scoring system.This scoring system and categorizing method based on the sonographic features with different weightings can be of great use to the objective classification of BI-RADS-US and help clinicians predict risk of breast cancer.