The Reading Room

Digital breast tomosynthesis significantly outperforms diagnostic mammography in study


Shaun Lillard, Imaging Performance Partnership

A study published in the January issue of Radiology provides clinical evidence that digital breast tomosynthesis (DBT) enables greater diagnostic accuracy than standard diagnostic mammography. Led by Dr. Margarita Zuley, researchers from the University of Pittsburgh compared the findings of eight breast imaging radiologists who had retroactively reviewed 217 lesions in 182 patients.

The study found that “tomosynthesis significantly improved diagnostic accuracy for noncalcified lesions compared with supplemental mammographic views.” There is hope that evidence such as this will reduce unnecessary biopsies and lower downstream costs as more sites implement DBT. 


Supplemental mammography vs. DBT

The study’s comparison of DBT and diagnostic mammography focused on the use of supplemental mammography, which includes angled, rolled, and spot compression views of the breast beyond standard mammography images. These techniques, however, may not be overly effective, according to the study authors.

The average positive predictive value of an abnormal interpretation in the diagnostic setting with these supplemental views is less than 36%. With this in mind, the study's radiologists were told to interpret each lesion once with supplemental mammography and once with DBT.

To assess the radiologists’ ability to accurately determine the malignancy of the lesions, researchers used receiver operator characteristics analysis, which is a method for comparing the sensitivity and specificity of diagnostic tests. In that analysis, the average area-under-the-curve for DBT was .87 to only .83 for supplemental mammography (higher scores indicate better performance). In fact, each of the eight readers performed better when using DBT than supplemental mammography views. 

Additionally, DBT resulted in lower false-positive rates compared with supplemental mammography—74% instead of 85% for cases rated BI-RADS category 3 or higher and 48% instead of 57% for BI-RADS category 4 or 5 cases. Of the 217 lesions, one-third (72) were malignant, while two-thirds (145) were benign.


What’s next for DBT?

Despite the ample clinical evidence supporting DBT’s effectiveness, there are still some concerns. One of the biggest barriers to its adoption has been the lack of a specific CPT code for payment. Vendors are cautiously optimistic that CMS could fix that as soon as late summer, but it’s still unclear whether that will happen.

Vendors also continue to address greater dose concerns with DBT, as Hologic has done with its C-view software. At RSNA in Chicago in November, TK profiled a new dose reduction technique called PICCS (prior image constrained compressed sensing). PICCS is still a work-in-progress, but it was able to reduce the dose by 40% while maintaining a readable exam in phantom and cadaver studies.

Learn More

Check out the Partnership’s recent study on breast imaging centers, the Breast Imaging Outlook, for more information on how breast imaging practices are improving workflow, optimizing throughput, and investing in new technologies.