Model Separates Benign, Malignant Breast Disease
Superior model includes logistic regression and radiologists' assessments
MONDAY, April 13 (HealthDay News) -- A logistic regression model combined with radiologists' assessments is better than either alone in discriminating between benign and malignant mammography findings, according to research published in the April issue of the American Journal of Roentgenology.
Jagpreet Chhatwal, Ph.D., of the University of Wisconsin School of Medicine and Public Health in Madison, and colleagues constructed two logistic regression models using mammography features and demographic factors from more than 48,000 mammography examinations of more than 18,000 patients. Outcomes from state cancer registries were matched to the data as the reference standard. Model 1 contained a variety of mammographic features and demographic factors, and Model 2 contained these variables plus radiologists' Breast Imaging Reporting and Data System (BI-RADS) assessment categories.
The investigators found that at 90 percent specificity, Model 2 had a significantly better sensitivity than the radiologists and Model 1 (90 percent versus 82 and 83 percent, respectively). At 85 percent sensitivity, Model 2 also had significantly better specificity (96 percent) compared to the radiologists and Model 1 (88 and 87 percent, respectively).
"In conclusion, we found that our logistic regression models (Model 1 and Model 2) can effectively discriminate between benign and malignant lesions. Furthermore, we have found that the radiologist alone or the logistic regression model incorporating only mammographic and demographic features (Model 1) are inferior to Model 2, which incorporates the model, the features, and the radiologist's impression as captured by the BI-RADS assessment categories," the authors write.