Automated breast ultrasound: fewer recalls, same cancers found

Design

Observational multi-registry comparison

Population

25,328 supplemental screening ultrasounds in women with dense breasts, 27 US facilities

Primary outcome

Abnormal interpretation, follow-up, biopsy and cancer detection rates

Effect

Abnormal 14.9% vs 19.2%; short-interval follow-up 3.0% vs 14.6%

This observational study used three US Breast Cancer Surveillance Consortium registries, covering 27 facilities from 2017 to 2020. It compared 21,110 handheld and 4,218 automated breast ultrasound examinations used as supplemental screening in women with dense breasts. It was published in July 2026.

Automated ultrasound was associated with fewer abnormal interpretations (14.9% vs 19.2%), many fewer recommendations for short-interval follow-up (3.0% vs 14.6%) and fewer false-positive initial assessments. Biopsy recommendations, positive predictive value and cancer detection did not differ apparently.

The short-interval follow-up difference is large and may reflect how readers manage uncertainty with a full volumetric dataset rather than a handful of saved images. Facilities were not randomised to one method, so case mix and reader experience may contribute. For Indian practice, where handheld ultrasound is the default supplemental tool and operator time is scarce, the result supports automated systems as a reasonable alternative where they are available, not a reason to stop handheld scanning.

• Consider automated breast ultrasound as an alternative to handheld scanning for supplemental dense-breast screening.
• Audit your own short-interval follow-up rate; handheld rates here were about one in seven.
• Use prior comparisons consistently to reduce probably-benign follow-up recommendations.
• Keep handheld ultrasound for targeted problem-solving and biopsy guidance.

Why it matters

False positives are the main cost of supplemental ultrasound, and this suggests a way to reduce them.

Don't overread it

This was observational; facilities chose their method, so the comparison is not randomised.

The statistics, in plain English

The cancer detection rate did not differ apparently, but a study of this size may not detect a small difference in a rare outcome. The recall differences are large and precise.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.