Startups focused on fighting cancer is a topic we keep coming back to, and I find the founders behind these companies are some of the smartest and most driven that I encounter.

Today’s startup is no exception. Meet Praxium…

– Martin

Praxium wants to bring AI cancer detection to the rest of the world

Sadly, none of us have to go too far from our own lives to find someone affected by breast cancer, either directly or indirectly. 

While treatments and technology to achieve better outcomes for breast cancer patients have improved significantly over the years, access to these advances isn’t equally distributed around the world.

With this in mind, new startup Praxium has developed a “copilot” to increase radiologists' accuracy, productivity, and diagnostic confidence even if they don’t happen to work in a well-resourced healthcare system.

“We're trying to make it economically and technologically feasible for every country in the world to offer their citizens breast cancer screening, with a really affordable, accessible technology,” says co-founder and CEO Nelson Jordan.

Praxium’s offering consists of two parts. The first is a web app for viewing and engaging with images in the DICOM (Digital Imaging and Communications in Medicine) file format used in medical imaging. This provides an easily accessible way of analysing mammograms.

But paired with that is what Jordan hopes will make Praxium a must-buy for healthcare providers in the startup’s target territories: AI breast cancer detection from mammograms.

He says the tech is built on an open-source framework, adding pre-processing and fine-tuning to achieve what he says brings the automated detection up to the level of a human radiologist. The tech has been trained on North American data and tested on open-source Chinese and Vietnamese mammography datasets.

Praxium hopes this tech will prove compelling in low- and middle-income countries that don’t have the level of expertise that patients in countries like the UK can draw on.

Jordan says that while breast screening specialists in the UK might read a minimum of 5,000 mammograms per year, in many other countries they might not have anywhere near that level of experience.

“The vast majority of those radiologists will be generalists, which will mean that they have to focus on not only different anatomical regions, but also different imaging modalities as well,” says Jordan.

“They might be using mammograms, X-rays, CTS, ultrasounds, but they might also be doing it in different parts of the body, so they just never get good enough to be able to pick up these cancers that are a little bit rarer, the edge cases, the ones that aren't immediately visible, or that might be small and new.”

In addition to raising the standard of breast cancer detection in their target markets, Jordan hopes Praxium can help address a global problem - a widespread shortage of radiologists.

The story so far

Jordan previously ran a web3 startup focused on decentralised collaboration. He found his way into Praxium when his future co-founder, Nickolas Ivanov, was looking for someone to build a cancer detection startup with. 

As Jordan tells it, Ivanov couldn’t find anyone suitable on the Antler venture builder programme he was taking part in, so convinced Antler to invite Jordan to join.

“They said yes, with the caveats that I had to pass all of the admissions process that a normal person on the accelerator would have. So I had an intense couple of days of meeting all the partners and going through that process.”

Ivanov has a background in AI development and previously co-founded a startup building an education platform for breast cancer radiologists in low-and middle-income countries. This gave him strong insights into the needs of Praxium’s target market.

“He was speaking to them all the time, interviewing them, understanding their pain points, understanding the type of training that they needed to see, their workflows, really just understanding why they were being failed and underserved by the current solutions,” says Jordan. 

“And the answer is, a lot of the solutions just aren't aimed at them. They're not tailored to them. They're aimed at radiologists in high-income countries. That's where you can do a deal as a potential software provider, worth hundreds of thousands, or even millions. 

“Nobody has taken the time to look at these more fragmented markets abroad and think that there's more than enough volume there to support a massive business.”

Since starting work on Praxium late last year, the pair have made good progress. Jordan says that with the AI product ready for use and the web-based DICOM viewer coming along well, they’re now starting to work with their first paying customers.

“They’re design partners who are also willing to put their hands in their pocket to fund the problem, because they see immense value in our solution.”

Praxium was recently announced as part of Praetura Ventures’ first cohort for its Praeseed programme, which could lead to the firm investing in the startup.

Subscribe to Premium to read the rest.