This article is more than 5 years old. Information may no longer be current.
Breast imaging technique shows promise for reducing unnecessary biopsies
Click Here to Manage Email Alerts
Analysis of breast masses using quantitative, three-compartment breast imaging combined with mammography radiomics may decrease unnecessary breast biopsies, according to a prospective study published in Radiology.
As many as 60% of women who receive mammograms over 10 years have at least one abnormal result, according to study background, and the recall rate for additional diagnostic imaging and biopsies after abnormal mammography findings that prove to be benign is as high as 10%.
“The callback rate with mammography is much higher than ideal,” Karen Drukker, PhD, research associate professor in the department of radiology at University of Chicago, said in a press release. “There are costs and anxiety associated with recalls, and our goal is to reduce these costs but not miss anything that should be biopsied.”
The new three-compartment breast (3CB) imaging technique measures the water, lipid and protein composition of tissue throughout the breast, with the goal of identifying biological signatures for tumors.
Drukker and colleagues obtained dual-energy craniocaudal and mediolateral oblique mammograms immediately before biopsy from 109 women (mean age, 51 years; range, 31-85) with Breast Imaging Reporting and Data System (BI-RADS) category 4 or 5 breast masses between 2012 and 2017. Thirty-five masses were invasive cancer and 74 were benign. Eleven women in the study had microcalcifications — six malignant and five benign — as a secondary finding.
The researchers derived 3CB images from the mammograms, using a within-image phantom to calculate the three quantitative compartments of water, lipid, and protein thickness at each pixel from the attenuation at high or low energy. They automatically segmented masses and extracted features from the low-energy mammograms and quantitative compartment images.
Researchers distinguished between malignant and benign masses with each of the following: water and lipid-protein composition images (quantitative 3CB analysis); mammography radiomics, which uses artificial intelligence algorithms to analyze features and patterns in images; or a combined image assessment of both.
Positive predictive value of biopsy performed at the highest level of sensitivity served as the primary performance metric.
The researchers found that quantitative 3CB analysis and mammography radiomics provided added benefit in estimating the probability of malignancy.
The quantitative 3CB analysis suggested nine fewer unnecessary benign biopsy of the 74 benign masses than conventional digital mammography, but incorrectly eliminated biopsy of one of 35 invasive cancers (P < .001 for both).
The mammography radiomics approached avoided 28 benign biopsies and the combined approach avoided 38 benign biopsies, both of which were significant improvements over conventional diagnostics (P < .001 for both). However, they on average misclassified one of 35 cancers ((P < .001).
Thus, the combination had a positive predictive value of 49% (95% CI, 36.5-58.9) with a sensitivity of 97% (95% CI, 90.3-100) compared with a positive predictive value of 32.1% (95% CI, 23.9-41.3) for conventional diagnostic digital mammography with 100% sensitivity.
The combined approach recommended 35.8% fewer total biopsies — or 39 of the 109 women — than conventional diagnostic digital mammography, at the cost of a 2.9% (1 of 35) reduction in sensitivity (P < .001).
There were statistically significant improvements in positive predictive value for mammography radiomics vs. quantitative 3CB analysis (34 of 82 vs. 34 of 99; P = .006) and the combined approach vs. mammography radiomics (34 of 70 vs. 34 of 82; P = .006).
A separate evaluation revealed area under the curves for differentiating between malignant and benign masses of 0.76 (95% CI, 0.66-0.95) for quantitative 3CB analysis, 0.8 (95% CI, 0.72-0.88) for mammography radiomics and 0.86 (95% CI, 0.78-.0.92) for the combined approach.
Limitations of the study included use of in-house methods to acquire 3CB images and mammography radiomics, as well as the modest size of the data set.
planned on how the combined approach might guide radiologists’ decisions. Moreover, researchers plan to study the combined approach using digital breast tomosynthesis.
Drukker said in the press release. “Combining 3CB image analysis with mammography radiomics, the reduction in recalls was substantial.” – by Jennifer Byrne
Disclosures: Drukker reports no relevant financial disclosures. Please see the study for all other authors’ relevant financial disclosures.
Click Here to Manage Email Alerts