Purpose: To assess the agreement between digital breast tomosynthesis (DBT) breast density (BD) assessment made using Quantra™ and fifth edition BI-RADS®.
Materials and methods: This board approved study involved BD assessment of 234 women undergoing DBT investigation. BD estimation was performed from the raw DBT images using Quantra™ 3 (v.2.1.1, Hologic, Bedford MA). BI-RADS® assessment was performed using prior digital mammograms displayed simultaneously with 2D images synthesized from DBT by three radiologists using the fifth edition BI-RADS® (A, B, C, D). Kappa (к) was used to assess inter-reader agreement, agreement between Quantra™ and each reader, as well as the majority report of all readers. Receiver Operating Characteristic (ROC) analysis was used to assess the performance of Quantra™ in reproducing the majority BI-RADS® assessment. Data was then grouped into a two-category scale [almost entirely fatty and scattered fibroglandular tissue (A-B) versus heterogeneously dense and extremely dense (C-D)], and a further analysis performed.
Results: Inter-reader agreement varied from fair [0.38 (95%CI: 0.30-0.46)] to substantial [0.68 (95%CI: 0.61-0.75)] on a four-category scale and substantial [0.70 (95%CI: 0.61-0.78)] to almost perfect [0.85 (95%CI: 0.78-0.92)] on a two-category scale. Using the majority report, the agreement between BI-RADS® and Quantra™ was 0.68 (95%CI: 0.59-0.75) on a four-category scale and 0.86 (95%CI: 0.79-0.93) on a two-category scale. Quantra™ distinguished BI-RADS® A-B from C-D with 97.1% sensitivity and 83.1% specificity.
Conclusion: Data demonstrate moderate to substantial agreement in BD assessment between fifth edition BI-RADS® and Quantra™.
Keywords: Dense tissue; Mammography; Measurement; ROC; Radiology; Two-category scale.
Copyright © 2016 Elsevier Ltd. All rights reserved.