Clinical performance of Siemens digital breast tomosynthesis versus standard supplementary mammography for the assessment of screen-detected soft-tissue abnormalities: a multi-reader study

Abstract

To compare the diagnostic accuracy of standard screening images plus single-view digital breast tomosynthesis (DBT), using Siemens DBT equipment, with standard screening images plus supplementary mammographic views in non-calcific, screen-detected mammographic abnormalities. Participants were unselected women aged 50-69 years recalled within a population-based European breast screening programme for assessment of soft-tissue mammographic abnormalities. Supplementary mammographic views (SMVs) and DBT were performed in all cases. A range of equipment was used for screening and supplementary mammography, but all DBT examinations were performed using the Siemens Mammomat Inspiration. A retrospective multi-reader study including 238 cases for whom either histology or at least 2 years' follow-up was available was performed with eight suitably accredited UK breast screening personnel reading all cases under both conditions, with temporal separation. Readers were blinded to case outcomes and findings from other examinations. Diagnostic accuracy using receiver operating characteristic (ROC) analysis was compared between screening plus SMV images and screening plus DBT images. The study was powered to detect a 3% inferiority margin in diagnostic accuracy between methods. The final sample with complete data available for analysis included 195 benign cases (1,560 reads) and 35 malignant cases (280 reads). The DBT method yielded a slightly higher area under the curve (AUC) value than the SMV method (0.870 versus 0.857), but the difference was not statistically significant (p=0.4890), indicating that the methods have equivalent accuracy. Siemens DBT demonstrates equivalent diagnostic accuracy according to ROC curve analysis when used in place of SMVs in screen-detected soft-tissue mammographic abnormalities.