DeepHealth, the AI imaging subsidiary of RadNet, announced on June 25, 2026 that it had received two new FDA 510(k) clearances for its Breast Suite platform, adding a tool that flags cardiovascular risk from a routine screening mammogram and a second feature that compares a patient’s current scan against her own prior exams to catch new or changed lesions faster.
Reading heart risk into a breast cancer scan
The first clearance covers Breast Arterial Calcification Assessment, or BAC Assessment, a tool that runs automatically on standard screening mammograms already being taken for breast cancer detection and identifies calcium deposits in the breast arteries, which research has linked to elevated cardiovascular disease risk. According to DeepHealth’s announcement, the tool demonstrated more than 90% sensitivity and more than 88% specificity for identifying arterial calcifications in clinical performance testing, and it works across both dense and non-dense breast tissue — a distinction that matters because dense tissue has historically made abnormalities, of any kind, harder for both radiologists and algorithms to spot on a mammogram.
Why breast arterial calcification is a big deal for women’s cardiovascular care
Cardiovascular disease remains a leading cause of death among women, and screening mammography is one of the few imaging exams a large share of women undergo routinely regardless of any cardiac symptoms. By extracting a cardiovascular signal from a scan that is already being performed for another purpose, DeepHealth is positioning BAC Assessment as a way to surface heart-disease risk in patients who might not otherwise be flagged for cardiac workup, without requiring any additional test, appointment or radiation exposure — the tool works from imaging data that already exists in a patient’s chart.
The second clearance: catching change over time
The companion clearance adds prior-exam integration to what DeepHealth now markets as Mammo Dx, software previously known as ProFound Pro. The upgraded version automatically processes a patient’s current mammogram alongside her previous studies, tracking known lesions and distinguishing genuinely new findings from ones already on record, according to DeepHealth’s press release and reporting by AuntMinnie. The company’s stated goal is improving cancer detection rates while reducing unnecessary recalls — the callbacks for additional imaging that generate anxiety and cost for patients whose "new" finding turns out, on closer comparison, to be unchanged from a prior scan.
Part of a broader breast-imaging suite
Both features join an expanding Breast Suite product line that already includes Mammo Dx, Breast Density assessment, Safeguard Review, Risk Assessment, a DeepHealth Viewer, Mammography Insights and, as of a separate clearance in July 2026, an AI-powered breast ultrasound tool. DeepHealth’s parent, RadNet, operates a large network of outpatient imaging centers, giving the company a built-in distribution channel to deploy new Breast Suite features at scale once cleared, rather than needing to build hospital relationships from scratch the way an independent AI startup would.
Questions from radiologists and patient advocates
Radiologists broadly welcome tools that reduce recall rates, since unnecessary callbacks are a well-documented source of patient anxiety and downstream cost in breast screening programs. But some in the field have raised a scope question specific to BAC Assessment: a mammography-derived cardiovascular flag sits outside a radiologist’s traditional area of expertise and workflow, raising practical issues about who is responsible for acting on a positive finding — the reading radiologist, the ordering primary care physician, or a cardiologist who was never part of the original referral chain. Patient advocates have separately pushed for clarity on how an incidental cardiovascular finding gets communicated to a patient who came in expecting only a breast cancer screening result.
What’s next
DeepHealth has not published a timeline for rolling BAC Assessment and the upgraded Mammo Dx out across RadNet’s own imaging centers versus licensing them to outside hospital systems, but the company’s clearance announcement frames both as immediately available additions to the existing Breast Suite license. The near-term test will be whether ordering physicians and cardiologists build the referral pathways needed to act on an unexpected cardiovascular flag surfaced from a breast cancer screening exam, and whether the prior-exam comparison tool measurably moves recall rates at scale once deployed across a large screening population. Because RadNet operates its own chain of imaging centers, the company is also positioned to generate real-world performance data faster than a standalone AI vendor selling into third-party hospital systems, which could either accelerate broader adoption of both tools or invite scrutiny over whether results from RadNet’s own centers generalize to other patient populations and equipment.
Photo: Kampus Production / PEXELS via Pexels
