Bunkerhill Health received FDA clearance on October 8, 2025, for Bunkerhill MAC, the first-ever AI algorithm designed to detect and quantify mitral annular calcification, or MAC, on routine, non-gated chest CT scans — the kind of scan patients already receive for unrelated reasons, such as lung cancer screening.
A Heart Problem Hiding in Plain Sight
Mitral annular calcification is calcium buildup in the ring-shaped structure that anchors the heart’s mitral valve. It’s often found incidentally on imaging, and studies have linked it to higher cardiovascular mortality and a greater risk of complications during structural heart procedures. The catch is that MAC is easy for radiologists to overlook, simply because it isn’t the condition the scan was ordered to look for. A chest CT ordered to screen for lung cancer, for instance, captures the heart in the same images but radiologists reading it are typically focused elsewhere.
Turning an Existing Scan Into a Second Diagnosis
That’s precisely the opportunity Bunkerhill MAC is built around: opportunistic screening, or using imaging that’s already being done and already being paid for to catch a second, unrelated disease. Rather than requiring an additional scan or a new referral, the algorithm analyzes the same chest CT data already captured during routine screening and flags MAC when it’s present, giving physicians a chance to catch a cardiovascular risk factor that might otherwise go unnoticed for years.
How the Algorithm Was Built and Tested
The algorithm was developed and tested using data from Bunkerhill’s multi-institutional research consortium, which includes more than 25 academic medical centers. For this specific FDA clearance, data from 7 consortium sites were used, including Emory University, Thomas Jefferson University, and UCSF. That multi-site foundation is intended to ensure the tool performs consistently across different scanners, patient populations, and imaging protocols rather than being narrowly tuned to a single hospital’s equipment.
Part of a Broader AI Platform
Bunkerhill MAC runs within the company’s Carebricks platform, which allows health systems to apply both large language models and FDA-cleared algorithms across patient data in order to automate appropriate next steps — for example, automatically flagging a cardiology referral when MAC is detected. The idea is to close the loop between detection and action, rather than simply generating a finding that sits in a radiology report without a clear follow-up pathway.
The Case for Opportunistic Screening — and Its Limits
Cardiologists describe opportunistic screening as one of the most cost-effective applications of AI in medicine, precisely because the imaging has already been ordered and paid for. Finding a second disease hiding in a scan taken for something else means no added cost, no added radiation exposure, and no added appointment — just more value extracted from data that already exists.
Skeptics raise a more practical concern, however: detecting MAC doesn’t, by itself, change treatment for most patients. Cardiology societies still need to define the clinical pathway that should follow a positive flag — whether that means a referral to a cardiologist, ongoing monitoring, or in many cases nothing at all. Without that defined pathway, critics warn, the tool risks generating anxiety-inducing incidental findings that patients and even some physicians won’t know how to act on, undercutting some of the efficiency gains opportunistic screening is supposed to deliver.
What’s Next for Opportunistic AI Screening
Bunkerhill’s clearance adds mitral annular calcification to a growing list of conditions that AI tools can now opportunistically detect in scans ordered for other purposes, and it puts pressure on cardiology societies to issue clearer guidance on how clinicians should respond to a MAC flag before the tool reaches widespread use. If that clinical pathway gets defined clearly — distinguishing which patients need referral versus monitoring versus no action — Bunkerhill MAC could become a template for how opportunistic AI screening moves from a clever use of existing data into an accepted part of routine care. Without that guidance, health systems adopting the tool may find themselves managing a stream of incidental findings with no consistent protocol for what to do next. In the meantime, the clearance itself demonstrates that regulators are willing to approve AI tools built specifically around the idea of mining existing imaging for additional diagnostic value, rather than requiring every new AI application to be tied to a dedicated scan ordered for that purpose alone. That precedent could matter well beyond cardiology, since the same non-gated chest CT scans used for lung cancer screening already capture other anatomy, such as bone density and early signs of emphysema, that opportunistic algorithms could eventually be trained to flag as well. For now, though, Bunkerhill MAC’s success will likely be measured not just by how many cases of mitral annular calcification it finds across the seven consortium sites and beyond, but by whether cardiology societies move quickly enough to give physicians a clear answer when a patient’s scan comes back positive.
Photo: Tumisu / PIXABAY via Pixabay
