ECR 2013 Rec: Does the adjunct of digital breast tomosynthesis (DBT) increase inter-reader reproducibility of two-dimensional digital mammography (2D-DM)? #SS1302 #B0687
B-0687 Does the adjunct of digital breast tomosynthesis (DBT) increase inter-reader reproducibility of two-dimensional digital mammography (2D-DM)?
G. Di Leo, L.A. Carbonaro, M. Bazzocchi, V. Londero, A. Dal Col, R.M. Trimboli, F. Sardanelli | Sunday, March 10, 10:30 – 12:00 / Room F2
Purpose: To estimate the inter-reader reproducibility of DBT added to 2D-DM in comparison to that of 2D-DM alone.
Methods and Materials: A series of 65 breasts (65 women, aged 58±9 years) underwent DBT (Giotto, IMS, Italy) as an adjunct to 2D-DM. After an agreement on how to report 2D-DM and DBT, 3 independent readers (R1, R2, R3) with >6 years of experience in 2D-DM and >1 year of experience in DBT evaluated all studies. Each reader evaluated the breast density and assigned a BI-RADS score using only 2D-DM. After 30 days, they repeated the evaluation adding DBT to 2D-DM. The inter-reader reproducibility was estimated using the quadratically weighted Cohen’s kappa.
Results: The breast density reported at 2D-DM by the most experienced reader was: 75 % in 1 (1%). Considering the B1, B2, B3, B4, and B5 BI-RADS scores, the same reader assigned the followings: 25, 0, 10, 20, and 10 using 2D-DM; 31, 1, 7, 13, and 13 using 2D-DM plus DBT. For each pair of readers, 2D-DM plus DBT resulted in a higher inter-reader reproducibility than that of 2D-DM alone: kappa value for 2D-DM ranged from 0.418 to 0.566 while that for 2D-DM plus DBT ranged from 0.656 to 0.758.
Conclusion: While the inter-reader reproducibility of 2D-DM resulted only moderate, the adjunction of DBT allowed to reach a substantial reproducibility. The use of DBT added to 2D-DM can reduce inter-reader variability and allow for more reliable readings from different radiologists.