Volume 29, Issue 3
DOI: 10.24205/03276716.2020.914
A Comparative Study of Logistic Regression Model in Color Doppler Ultrasound Diagnosis of Breast Cancer
Abstract
Objective To establish a logistic regression model based on the breast imaging report data system (BI-RADS) to evaluate the diagnostic efficiency of breast cancer risk prediction. Methods a retrospective analysis of 1660 cases of breast ultrasound image data from January to September 2011 was carried out. The image features were standardized according to BI-RADS standard, and the pathological results were taken as the gold standard. A logistic regression model was established by integrating the ultrasound image features of diagnostic value into univariate analysis, to explore the sensitivity, specificity and accuracy of the model in predicting the risk of breast cancer.
Results single factor analysis showed that 18 of the 30 ultrasound image features had statistical significance in differentiating benign and malignant breast diseases (P < 0.001). The sensitivity, specificity and accuracy of the logistic regression model based on these image features were 84.5%, 95.5% and 91.4%, respectively. The area under the working characteristic curve of the subjects was 0.964, and the prediction accuracy was 91.0%. Conclusion the logistic regression model based on the characteristics of BI-RADS ultrasound image has a good diagnostic efficacy in predicting the risk of breast cancer, which indicates that the big data of standardized breast ultrasound report can establish a clinical decision-making system for breast cancer, and assist ultrasound doctors to improve the diagnostic level.
Keywords
Ultrasound; breast image reporting data system; logistic regression model