The radiology data company that turned a 30-year safety principle into software running quietly inside the hospital.
Pictured: a logo, photographed at rest. It does not look like much. Neither does the box in the server room that keeps your CT scan from leaving the building - which is rather the point.
Somewhere in a hospital basement, between the MRI and the vending machine, a small piece of software is reading every CT scan as it comes off the scanner. It is not diagnosing anything. It is counting radiation. Then it reports that number - cleanly, securely, on the hospital's own terms - to the people who decide whether the dose was reasonable. That software is the Alara Medical Imaging Gateway, and it is the reason this company exists.
Alara Imaging calls itself "healthcare's private compute platform." Stripped of the slogan, the idea is unfashionably simple: the most sensitive data in medicine - your imaging - should be measured where it already lives, not shipped off to a cloud and hoped for. Alara built the local node that makes that possible: real-time, two-way data flow between the hospital and the cloud, with the computation happening at the edge.
"Make healthcare safer." A mission statement short enough to fit on the box - which, conveniently, is roughly where the product lives.
Here is the uncomfortable fact that started everything. Two patients can walk into two hospitals, get the same scan, and leave with radiation doses that differ by an order of magnitude. The image looks the same. The risk does not. For decades this variation went largely unmeasured, because measuring it meant pulling structured data out of imaging systems that were never designed to give it up.
Radiologists have a phrase for the ideal: ALARA - As Low As Reasonably Achievable. It is the principle every trainee learns and every department aspires to. The trouble with aspirations is that they are notoriously hard to audit. You cannot lower what you do not count, and counting, it turns out, was the entire problem.
The company is named after a safety rule. Most startups name themselves after a Latin verb and a vowel they bought at auction.
Then there is the second problem, the one that makes the first one expensive. The Centers for Medicare and Medicaid Services decided that CT radiation dose should become a national quality measure - something hospitals report on, the way they report infection rates. Good policy. But policy without plumbing is just a deadline. Someone had to build the secure pipe between the scanner and the report, in thousands of hospitals, without turning patient data into a liability. That someone was not lining up.
Alara did not start in a garage. It started in a research laboratory at the University of California, San Francisco, where Rebecca Smith-Bindman had spent roughly three decades running the Radiology Outcomes Research Laboratory and documenting, in NIH-funded study after study, exactly how much unnecessary radiation patients were absorbing. She helped author the CMS-endorsed measure to standardize CT doses. Then she did the thing academics rarely do: she co-founded a company to ship it.
Career spent operating and investing in data-driven, machine-learning companies. Runs strategy, operations and the business side of a clinical mission.
UCSF professor; ~30 years leading radiology outcomes research. Co-author of the CMS CT radiation-dose measure. The clinical conscience of the company.
Imaging informatics physician. Bridges the gap between what radiology data should do and what hospital systems actually allow.
The bet was counterintuitive. Instead of selling another dashboard, Alara would become a CMS measure steward - the official party responsible for a national quality measure - and give hospitals the software to comply. The plan even includes providing the radiation-dose measure software to US hospitals at no charge. A startup whose flagship deliverable is partly a public good is not a normal pitch. It is, however, a very effective way to get every hospital to install your gateway.
They spent about four years building the gateway before scaling it. In startup years, that is roughly a geological era - and exactly how long it takes to be trusted with a hospital's data.
The Alara Medical Imaging Gateway is, technically, a local node that lives inside the health system. It speaks the native languages of medical imaging - DICOM and HL7 - and moves data both directions between the facility and the cloud in real time. The clever part is what it refuses to do: it does the computation at the edge, so the sensitive imaging data does not have to leave the building to be counted.
Edge node for real-time, bidirectional data flow between hospital and cloud. The informatics layer radiology never got around to building.
Calculates CMS quality measures - starting with CT radiation dose - securely and cost-effectively. Alara is the measure steward.
Connects DICOM and HL7 sources to cloud and AI workloads, so imaging data finally talks to everything else.
HITRUST certified, SOC 2 clean, HIPAA ready. Computation stays at the edge by design, not by promise.
Most healthcare software wants all of your data in its cloud. Alara's whole argument is that it wants as little of it as reasonably achievable.
A timeline, assembled from public filings and press. Read top to bottom; the dots are yellow because the navy ran out of cheer.
Alara Imaging incorporated in San Francisco, spun out of UCSF radiology outcomes research.
Raises early funding to build the gateway in earnest. Roughly four quiet years of engineering follow.
Former President & CMO of Radiology Partners, the largest US radiology practice, lends operational weight.
The healthcare security gold seal. Translation: hospitals' compliance teams stop saying no.
Alara aligns with ACR on radiology quality measures - the field's professional home turf.
Jill Spear (ex-Volpara Health) joins as VP Operations; David Mezzoprete as VP Business Development.
A startup can claim anything. The more interesting signal is who agrees to put their name next to yours. Alara's partner list reads less like a vanity page and more like a coalition: Bayer collaborating to standardize CT radiation doses, NVIDIA on imaging AI at the edge, UCSF as the founding academic root, the American College of Radiology on measures, and Qaelum on dose-management technology - which is notable, because Qaelum could just as easily be a competitor.
Standardizing CT doses across the country could prevent tens of thousands of cancers a year. That is not a feature. That is the entire reason the box is in the basement.
Most healthcare-technology companies sell you software and ask for your data. Alara inverted the trade. It wants to be the secure layer that lets hospitals and health-tech companies work together transparently, while keeping the imaging data where it belongs. The radiation measure is the wedge; the larger ambition is to be the private compute platform on which a lot of clinical AI eventually runs - at the edge, under the hospital's control.
That is a slower business than a viral app. It is also a sturdier one. When your differentiator is trust - HITRUST, SOC 2, HIPAA, a measure stewardship granted by CMS - the moat is made of paperwork and patience, which happen to be the two things competitors find hardest to fake. The senior hires from Volpara Health and Radiology Partners suggest Alara is moving from "research project that works" to "company that scales."
The fastest way to make every hospital install your software is to make part of it a public good. Generosity, it turns out, is also distribution strategy.
Tomorrow, the argument gets bigger. As AI moves into radiology, the question of where the data is computed - hospital or cloud - stops being technical and becomes the whole ballgame. A platform that already runs at the edge, already trusted, already wired into DICOM and HL7, is well positioned for that fight. Alara is betting that the future of medical AI is private, local, and measured. It built the box for that future early.
Go back to that quiet machine between the MRI and the vending machine. Before Alara, the radiation it now counts was simply unmeasured - a number nobody owned, a risk nobody scored. Today the same scan produces a figure that travels, securely, to the people who can act on it, and the data never leaves the building to do it. The hospital looks identical. The accountability does not.
Alara Imaging did not invent the principle. As Low As Reasonably Achievable is older than every founder in the building. What Alara did was give the principle a place to run, a way to be counted, and a standard hospital cannot quietly ignore. The slogan promised to make healthcare safer. The box in the basement is the part that actually does it.