Medical imaging AI is not the model
It’s the data preparation
An Israeli, Lithuanian and South African walk into a bar in Dubai.
It sounds like the beginning of a story - and it is.
Asta and Chris were my guests on the podcast recently.
They are based in Dubai and they are building a startup that provides imaging from radiology scans to train AI models.
Their twist ?
They provide the images as ‘infrastructure as a service ‘ for countries.
In the US, it’s almost impossible to get non-US data and outside the US, its almost impossible to get US data. Countries have their own privacy regulation and country level data security requirements.
And that’s where Medsyntra come in.
Chris is a radiology software veteran and Asta is a entrepreneur. And because they’re not locked into the US healthcare system mentality - they decided to develop a country level platform for AI developers.
In Episode 52 of Life Sciences Today, we sat down to talk about what it actually takes to build global medical imaging infrastructure for clinical AI.
Medsyntra aggregates and de-identifies millions of imaging studies across radiology, pathology, oncology, and women’s healthcare — ethically sourced, sovereignty-respecting, and annotation-ready. That’s harder than it sounds.
Chris and Asta don’t sugarcoat where the tradeoffs bite.
We covered the questions most AI teams avoid: Why do models trained on one population quietly degrade when deployed in another? What does data sovereignty actually cost you when you’re trying to build global intelligence? And is “ethical data” a real standard — or a marketing layer over bad practices?
One thread I kept coming back to: the difference between assembling a dataset and building infrastructure. Chris and Asta crossed that line early, and the decisions they made at that fork still shape how Medsyntra operates today.
We also got into annotation — the bottleneck nobody budgets for — and the anti-patterns Asta sees repeated across the medical AI ecosystem, often by teams that should know better.
If you’re building AI for clinical use, or advising companies that are, this episode will reframe where you put your attention.
🎧 Listen now
Visit Medsyntra here
About me
I’m an entrepreneur, writer, host of Life Sciences Today podcast, ex-pharmatech founder, father of 4 and founder of OpenCRO.
OpenCRO specializes in threat modeling for FDA Cyber.
42 MedTech companies. 14 FDA clearances. 0 rework requests on threat model submissions.
You have threats to your patients, your clinical data, your reputation.You shouldn’t be discovering your cybersecurity gaps during FDA review.
We find them first, document them the way FDA reviewers expect — and deliver to you a cost-effective, prioritized countermeasure plan.
Fixed price. Fixed outcome.
Book a call to see if you’re a good fit
Not sure yet? Sign up for the 2026 CEO Audit — 60 minutes on your operation, unlimited access to my patterns library.