Objective      To assess the clinical value of digital breast tomosynthesis (DBT) in the diagnosis of breast cancer.       Methods      Seven hundreds and fifty-nine women taking bilateral mammography in our department from October 2014 to March 2015 were recruited in this study. They all received full-field digital mammography (FFDM) and DBT. Then 3 doctors independently made the diagnosis of each examination by reading FFDM and FFDM plus DBT. The diagnostic capacities of FFDM and FFDM combined with DBT were compared with pathological diagnosis as the gold standard.       Results      Among these 759 participants

555 were diagnosed as benign and normal

and 204 were determined as being malignant by pathology. There were statistical differences in breast diseases (Z=-6.594

P0.000 1)

and in malignant diagnoses (Z=-8.082

P0.000 1) between FFDM and FFDM plus DBT. The sensitivity

specificity

Youden index

positive value (PPV)

and negative predictive value (NPV) were 67.6%

96.2%

0.638

86.8%

and 89%

respectively

for FFDM

and were 80.9%

97.1%

0.78

91.2%

and 93.3% respectively for FFDM plus DBT. Receiving operating characteristic (ROC) curve analysis showed that the area under the curve (AUC) of ROC was 0.899 for FFDM and was increased to 0.969 for FFDM plus DBT. And there was statistical difference between their AUC (Z=7.126

P0.001). But

the average glandular dose (AGD) was significantly higher in DBT than FFDM (craniocaudal view: Z=-27.005

P0.000 1; mediolateral oblique view: Z=-27.514

P0.000 1).       Conclusion      FFDM plus DBT is superior to FFDM in the diagnosis of breast cancer

but the AGD of DBT is higher. Therefore

DBT should be used for differential diagnosis of suspicious lesions

but not for general investigation.