Automated breast density on par with visual assessment

Twitter icon
Facebook icon
LinkedIn icon
e-mail icon
Google icon
 - Doctor reviewing mammogram
Rebecca Zuurbier, MD, Sibley Memorial Hospital, examines a digital mammogram of a dense breast and points to a potential cancer.
Source: National Cancer Institute

Ten states have passed breast density legislation, and another 19 and the federal government are considering bills. Thus, breast density measurement is becoming increasingly important in practice. Automated volumetric breast density tools may provide an objective way for practices to reliably measure density, according to a study published in the June issue of Clinical Radiology.

Dense breasts can pose multiple problems in the screening process; they can obscure calcifications or lesions, which may reduce sensitivity and also pose a risk factor for breast cancer. Thus, J. M. Seo, from the department of radiology at Samsung Medical Center in Seoul, and colleagues devised a study to compare automated volumetric breast density (VBD) measurement with visual assessment according to the Breast Imaging Reporting and Data System (BI-RADS).

The researchers reviewed 193 consecutive screening mammograms obtained between October and November 2011. Two radiologists and a radiology resident conducted density analysis according to BI-RADS categories, and then re-read the studies four weeks later to evaluate intra-observer agreement. They also recorded factors that may impact density assessments, such as tube current, tube voltage, recorded breast thickness, breast volume, VBD, manufacturer, target and filter.

The radiologists’ assessments were compared to VBD as evaluated by an automated, volumetric system.

Of the 193 mammograms, the BI-RADS density category assessed by three radiologists and density grade (VDG), as measured by the automated system, agreed in 134 cases. A total of 54 of the disagreement cases were overscored using VDG and five were underscored.

The study identified target material as a factor that may affect the difference between the BI-RADS density category and VDG.

Seo and colleagues concluded by emphasizing “reasonable agreement” between the automated system and visual assessment in most cases. They indicated the need for more dedicated and larger studies to translate to the technology into practice.