The accuracy of digital breast tomosynthesis compared with coned compression magnification mammography in the assessment of abnormalities found on mammography

Abstract

To compare the diagnostic accuracy of the digital breast tomosynthesis (DBT) with coned compression magnification mammography (CCMM). The study design included two reading sessions completed by seven experienced radiologists. In the first session, all readers read bilateral standard two-view mammograms and a CCMM view of the lesion before giving a combined score for assessment. In the second session, readers read bilateral standard two-view mammograms plus one-view DBT. The two reading sessions of the experiment were separated by at least 2 weeks to reduce the chance of reader memory of the images read in the previous session from influencing the performance in the subsequent session. Three hundred and fifty-four lesions were assessed and receiver-operative characteristic (ROC) analysis was used to evaluate the difference between the two modes. For standard two-view mammography plus CCMM, the area under the curve (AUC) was 0.87 [95% confidence interval (CI): 0.83-0.91] and for standard two-view mammography plus DBT the AUC was 0.93 (95% CI: 0.91-0.95). The difference between the AUCs was 0.06 with p-value of 0.0014. Two-view mammography with one-view DBT showed significantly improved accuracy compared to two-view mammography and CCMM in the assessment of mammographic abnormalities. These results show that DBT can be used effectively in the further evaluation of mammographic abnormalities found at screening and in symptomatic diagnostic practice.