Computer analysis is not better than human clinical evaluation when it comes to measuring breast density

1. Similar associations were observed with clinical BI-RADS and automated BI-RADS with respect to interval and screen-detected cancer.

2. Both automated and clinical BI-RADS showed similar discriminatory accuracy and mammography accuracy.

Evidence Rating Level: 2 (Good)

Study Rundown: In the United States, the standard for monitoring breast density is through the Breast Imaging Reporting and Data System (BI-RADS). There has been recent concern regarding the reproducibility of this system, as there is subjectivity in the breast density estimated by radiologists. Due to this subjectivity, it has been questioned whether the BI-RADS should be automated as opposed to clinical. The purpose of this study was to evaluate whether breast cancer risk and detection are equivalent between the clinical and automated BI-RADS density measures. Overall, the results of the study suggested that an automated system would not be superior to clinical evaluation regarding the measurement of breast density in terms of accuracy and mammography sensitivity. However, this study had several limitations. First, tomosynthesis, which is an emerging breast-density measurement technology, was not incorporated into this study. As a result, this study may not be as applicable to States using tomosynthesis.

Click to read the study, published in Annals of Internal Medicine

Relevant Reading: Breast Density Analysis Using an Automatic Density Segmentation Algorithm

In-Depth [retrospective cohort]: A case-control study was done to determine whether automated BI-RADS was superior to clinical. Study participants were selected from 2 case-control studies from within large prospective breast imaging cohorts—the San Francisco Mammography Registry and the Breast Cancer Surveillance Consortium. Primary outcomes included assessment of automated and clinical BI-RADS density on digital mammography twice between September 2006 and October 2014. Measurements also included interval and screen-detected breast cancer risk, as well as mammography sensitivity. It was found that mammography sensitivity was similar between automated and clinical BI-RADS based on category breakdown. Specifically: fatty, 93% compared to 92%; scattered fibroglandular densities, 90% versus 90%; heterogeneously dense, 82% versus 78%; and extremely dense, 63% compared to 64%, respectively. Furthermore, discriminatory accuracy was similar between both BI-RADS measurements. However, it was found that the BI-RADS measures prior to cancer diagnosis had higher discriminatory accuracy for interval-detected cancer, compared to screen-detected cancer.

Image: PD

©2018 2 Minute Medicine, Inc. All rights reserved. No works may be reproduced without expressed written consent from 2 Minute Medicine, Inc. Inquire about licensing here. No article should be construed as medical advice and is not intended as such by the authors or by 2 Minute Medicine, Inc.