Accuracy of synthetic 2D mammography compared with conventional 2D digital mammography obtained with 3D tomosynthesis Academic Article Article uri icon


MeSH Major

  • Pelvic Organ Prolapse
  • Transitional Care
  • Urogenital Abnormalities


  • © American Roentgen Ray Society. OBJECTIVE. The purpose of this study was to compare the diagnostic performance of synthetic 2D imaging generated from 3D tomosynthesis (DBT) with traditional 2D full-field digital mammography (FFDM) by use of the most up-to-date software algorithm in an urban academic radiology practice. MATERIALS AND METHODS. The records of patients undergoing screening mammography with DBT, synthetic 2D imaging, and FFDM between August 13, 2014, and January 31, 2016, were retrospectively collected. The cohort included all biopsy-proven breast cancers detected with screening mammography during the study period (n = 89) and 100 cases of negative or benign (BI-RADS category 1 or 2) findings after 365 days of follow-up. In separate sessions, three readers blinded to outcome reviewed DBT plus synthetic 2D or DBT plus FFDM screening mammograms and assigned a BI-RADS category and probability of malignancy to each case. The diagnostic performance of each modality was assessed by calculating sensitivity and specificity. Reader performance was assessed by ROC analysis to estimate the AUC of the likelihood of malignancy. RESULTS. No statistically significant difference was found in diagnostic accuracy (sensitivity, specificity, positive predictive value, or negative predictive value) between DBT plus synthetic 2D mammography and DBT plus FFDM. There was no statistically significant difference between the AUC of DBT plus synthetic 2D mammography and the AUC of DBT plus FFDM for any reader. CONCLUSION. DBT plus synthetic 2D mammography performs as well as and not worse than DBT plus FFDM in measures of diagnostic accuracy and may be a viable alternative for decreasing radiation dose without sacrificing diagnostic performance.

publication date

  • June 2019



  • Academic Article


Digital Object Identifier (DOI)

  • 10.2214/AJR.18.20520

PubMed ID

  • 30917028

Additional Document Info

start page

  • 1406

end page

  • 1411


  • 212


  • 6