Profile / Health Tech / Radiology
Sirona Medical
San Francisco's quiet attempt to rip out radiology's tangled software stack and replace it with a single browser tab.
Sirona, in the wild. Logo + ambition.
San Francisco's quiet attempt to rip out radiology's tangled software stack and replace it with a single browser tab.
A radiologist with two monitors and three coffees opens Citrix. Then opens the PACS. Then the worklist. Then the dictation tool. Then a separate browser for priors. Six tabs in, the patient still hasn't been read.
This is the part of medicine that nobody photographs for the press release. It is also the exact problem Sirona Medical has spent seven years quietly trying to delete. The company doesn't build a better viewer or a cleverer AI. It builds an operating system - one tab, one workflow, one radiologist - and tries to convince an entire profession that the seven-tab status quo was never normal in the first place.
The pitch is unfashionably simple. The execution, of course, is not.
Hospitals run their imaging on a Frankenstein of PACS, RIS, dictation, voice-recognition, viewer, worklist, AI plug-ins and a Citrix wrapper holding it all together with tape. Each tool was acquired separately, integrated by someone who has since left, and patched on Tuesdays. The radiologist sits at the end of that supply chain and pays the cognitive tax.
Sirona's founder, Cameron Andrews, watched the tax get paid up close. His grandfather was a private-practice radiologist and founded the company that is now Rayus Radiology. Sunday dinners involved more software complaints than they probably should have.
Andrews went to Stanford. Studied biology and computer science. Did graduate work at Stanford's AIMI - the Center for Artificial Intelligence in Medicine & Imaging. Then spent three years at Lux Capital evaluating AI-in-medicine startups, which, in his telling, mostly convinced him that the industry was building features on top of a broken foundation. So in 2018, age twenty-five, he started building a foundation.
Most radiology AI companies are selling a roof. Sirona is laying a slab.
The bet Sirona made was specific and, at the time, slightly heretical: that the whole radiology stack - PACS, viewer, worklist, reporter, AI - belonged in a single cloud-native application, accessible through a browser, deployable on top of (or instead of) the old on-prem mess. No fat client. No Citrix. No twelve-month integration project to add one new vendor.
This sounds obvious in 2026. In 2018 it sounded like a Stanford kid who hadn't read enough HL7 specs. Andrews kept building.
The product is called Unify. It bundles a diagnostic viewer, a worklist, the reporting engine ("Pixel-Powered Reporter"), and AI - including AI-generated impressions and voice-driven commands - into one experience. It runs on AWS, is SOC 2 certified, and is HIPAA-compliant in the way that lets enterprise compliance officers sleep.
The mothership. Viewer, worklist, reporter, AI - one tab, cloud-native.
Reporting tied to the pixels. Voice-driven. AI impressions baked in.
FDA 510(k) cleared. PET-CT, MIP, MPR, image fusion - the heavy stuff.
Many sites, many systems, one queue. The thing teleradiology groups dream about.
Years and dollars. The boring stuff that turns out not to be boring.
Walk a radiologist through Unify and the demo is mostly about what's missing. No tab-switching. No swivel-chair integrations. The viewer talks to the report; the report talks to the AI; the AI talks to the worklist. When the radiologist dictates "no acute findings" the impression generator is already drafting. When the next case loads, the priors load with it.
The platform is designed to overlay existing on-prem PACS/RIS, not replace them on day one. This is the kind of strategic patience that wins enterprise sales and bores conference panels. It also explains why the company has shipped real installations instead of demos.
We didn't build features. We deleted them. Most of what radiology software does is reproduce decisions made when imaging files lived on physical disks.
Sirona's investor list is not crowded with tourists. Lux Capital wrote the first check and kept writing. 3L Capital joined for Series B. Avidity Partners led the $42M Series C in November 2024 - $118M cumulative, which in radiology IT is real money.
The other proof is regulatory. In October 2025 Sirona's Advanced Imaging Suite received FDA 510(k) clearance, giving the platform Class II medical device functionality for PET-CT, MIP generation, MPR and image fusion. Translation: the FDA reviewed the math and said yes.
The 2025 leadership reshuffle is its own signal. Ken Kaufman, with two decades in healthcare IT and a stint leading PACS sales at McKesson, took the CEO seat in January. Cameron Andrews moved into the Founder & President role - which, in startup grammar, usually means "still obsessing over product." Dr. Curtis Langlotz, who runs Stanford's AIMI, joined the board, lending the kind of academic credibility that helps when you are arguing with radiology department chairs.
The rest of the org is roughly 150 people, weighted toward engineering and clinical product. The technologies stack reads like a modern SaaS company that happened to wander into a regulated industry: React, TypeScript, GraphQL, Python, FastAPI, AWS, Salesforce. Slack and Atlassian for the humans. The mundane-stack-of-a-serious-company look.
If you press Sirona's public messaging hard enough, what comes out is not a feature list. It's a posture toward the work itself - that radiology is a craft, that radiologists are knowledge workers, and that any software which forces a board-certified physician to alt-tab through five vendors before reading a chest CT has, in some small way, failed them. Unify is built to give the radiologist back their attention.
That is a quieter mission than most healthcare AI pitches. It also happens to be the one that scales, because attention is the only thing in radiology that does not scale by itself.
The job is to make the software disappear. If the radiologist notices the software, we've lost.
The macro picture is unkind. CT and MRI volumes climb every year. The pipeline of new radiologists doesn't. The math forces a choice: read faster or accept longer turnaround times. Faster, in practice, means software has to do more of the work that used to be ambient cognitive load.
Sirona's position is that you can't bolt that capability onto a 1998 architecture. You have to start in the cloud and build outward. Whether the industry agrees fast enough to matter to Sirona's S-1 is a different question. The technical bet, at least, no longer looks heretical.
In the Sirona version of this scene, she opens one tab. The worklist is already prioritized. The viewer loads the priors next to the new study. She dictates. The impression drafts itself. She edits two words. Signs. Next case.
The coffee is still hot.