Impact of Obtaining a Digital Breast Tomosynthesis (DBT) Spot Compression View on Assessment of Equivocal DBT Findings.

Abstract

BACKGROUND. A recently introduced digital breast tomosynthesis (DBT) device allows acquisition of DBT spot compression views with a small paddle during DBT acquisition. OBJECTIVE. The purpose of this study was to evaluate the impact on diagnostic performance of obtaining a DBT spot compression view for assessment of equivocal DBT findings. METHODS. This retrospective study included 102 women (mean age, 60 years) in whom a DBT spot compression view was obtained to characterize an equivocal finding on DBT at the performing radiologist's discretion. The DBT examinations were performed from December 14, 2018, to December 18, 2019. Two fellowship-trained breast radiologists and one breast imaging fellow, who were aware of the location of the equivocal lesions, independently reviewed the examinations. Readers first assigned a BI-RADS category using standard DBT views and then immediately assigned a category using the DBT spot compression view. BI-RADS categories 2 and 3 were considered negative, and categories 4A and greater were considered positive. Histology and at least 1 year of imaging follow-up served as the reference standard. Intrareader agreement for one reader and interreader agreement among all readers were evaluated with kappa coefficients. Diagnostic performance was compared between DBT with and DBT without spot compression views by use of McNemar tests. RESULTS. Intrareader agreement increased from 0.43 to 0.72, and interreader agreement increased from 0.21 to 0.45 on the basis of kappa coefficients for DBT without and with spot compression views. Eighteen cancers were present. Compared with standard DBT views, DBT spot compression views yielded significantly increased accuracy for all three readers (75% vs 90%, 74% vs 94%, 72% vs 94%); significantly increased specificity for all three readers (69% vs 90%, 75% vs 94%, 68% vs 93%); and significantly increased sensitivity for one reader (67% vs 94%) without significant change in sensitivity for the two other readers (89% vs 100%, 100% vs 89%). Radiation dose was 1.97 mGy for the DBT spot compression view versus 1.78-1.81 mGy for standard DBT craniocaudal and medio-lateral oblique views. CONCLUSION. Use of the DBT spot compression view increased intrareader agreement, interreader agreement, and diagnostic accuracy (primarily owing to improved specificity); the supplemental dose for the spot compression view was slightly higher than that for a standard DBT view. CLINICAL IMPACT. DBT spot compression may help characterize equivocal DBT findings, reducing further workup for benign findings.