O-84 An observer performance study comparing the interpretation of full-field digital mammography with digital breast tomosynthesis

Abstract

Objective: The purpose of the current study was to compare the diagnostic accuracy of two-view full-field digital mammography (FFDM) with co-registered two-view digital breast tomosynthesis (DBT). Methods: The ethics committee approved trial evaluating a Hologic Selenia Dimensions machine recruited women recalled after their initial breast screening using analogue film-screen mammography, scheduled to undergo further breast assessment following informed consent. Radiologists fulfilling the professional criteria for National Health Service Breast Screening Programme’s radiology practice reviewed all images using the Royal College of Radiologists Breast Group classification for mammographic features 1–5 (M1 = normal, M2 = benign, M3 = probably benign, M4 = suspicious, M5 = malignant). Results: The study included bilateral mammograms of 526 women (age range 47–73 years) with 119 (22.6%) histologically proven primary breast cancers assessed by 5 readers. Using receiver operating characteristic (ROC) test, the analysis yielded statistically significant difference (p = 0.0001) between FFDM and DBT. The area under the ROC curve (AUC) was 0.968 ± 0.007 for DBT compared to AUC of 0.913 ± 0.015 for FFDM, demonstrating DBT interpretation was superior. Conclusion: Patients recalled after routine screening mammography will benefit from DBT as an additional technique for diagnostic workup.