AI breast cancer risk assessment in screening ABUS ultrasound

Automated Breast Ultrasound (ABUS) is now used as a screening method, together with screening mammography/tomosynthesis, to detect breast cancer at an early stage.

The development of artificial intelligence algorithms that assess the glandular tissue component of the breast (GTC) — which mammography cannot separate within the breast parenchyma — makes it possible to estimate a woman's risk of developing breast cancer. That risk is higher the more glandular tissue the breast contains.

What we are studying

We estimate the risk of women with dense breasts who are screened with a combination of ABUS and mammography/tomosynthesis, with an emphasis on the composition of the glandular tissue, which only ultrasound can assess. We then compare the risk derived from each technique.

How the study was carried out

Since April 2026 we have performed combined screening with mammography and automated ultrasound (ABUS) in 1,200 women with dense breasts.

Images were acquired with the Hologic 3Dimensions mammography system and the GE Invenia ABUS 2.0. The mammograms were analyzed with the iCAD algorithm and the ultrasound images with the Moncad algorithm, which calculates the GTC. Risk was categorized with both techniques and then compared.

On ABUS, women were categorized according to their glandular tissue component (GTC). On mammography, they were placed in 4 risk categories based on age, breast density and findings.

Early results

In our experience so far, the ABUS risk algorithm has placed risk in a lower category than mammography, as it relies on a single factor — the glandular tissue — rather than the findings and the woman's age, which mammography takes into account.