How accurate are volumetric software programs when compared to breast MRI?

Radiologists are in need of accurate methods for assessing a patient’s percent breast density (PBD) from a 2D mammogram. Are today’s volumetric breast density software programs up to the challenge?

Jennifer A. Harvey, MD, from the department of radiology and medical imaging at the University of Virginia Health System in Charlottesville, Va., and colleagues explored this question in a recent study published in Academic Radiology.

“Volumetric-based methods are highly reliable and should theoretically improve breast cancer risk prediction over area-based methods,” the authors wrote. “Volumetric breast density is likewise associated with breast cancer risk, but the degree is variable. Concern has been raised, however, about the ability of these widely available programs to accurately measure volumetric breast densities from 2D mammographic images.”

To test the accuracy of modern volumetric software programs, Harvey et al. retroactively studied data from 110 women who had undergone both a breast MRI exam and full-field digital mammography. The mean patient age was 52.5 years, and the mean interval between mammogram and MRI was 166 days. The team tested two commercially available software programs to obtain PBD measurements from the patients’ 2D mammographic images and were able to judge the accuracy of those programs by comparing the numbers to measurements obtained from the breast MRI exams.

Overall, the two programs earned high marks for their calculations, though some discrepancies were present. PBD was “not different” between MRI and one of the volumetric software programs. Meanwhile, there was a “small but significant difference present in the discrepancy” between MRI and the other program. In addition, the authors found that both programs “slightly underestimated” total breast volume (TBV), but “there was relatively good agreement” between fibroglandular volume (FGV) values derived from the MRI exams and the two software programs.

“The present study further solidifies prior research that volumetric mammographic breast density software calculations can effectively estimate FGV and TBVs to determine an accurate measurement of PBD,” the authors wrote, noting that the algorithms could still be improved “to reduce the overestimation and the underestimation of density measurements that are most pronounced at higher breast density values.”

“This and other studies support the validity of using automated software programs to quantify mammographic breast density in clinical use,” they added.