Brainomix launched in 2010 as a spin-out from the University of Oxford. Its stroke software e-Stroke harnesses AI to analyse brain scans and flag up blockages. The technology has been designed to enable clinicians to make better life-saving decisions.
Stroke Rehab Times sat down with Chief Marketing and Business Development Officer, Jeff Wyrtzen, to find out more about how the technology is helping to transform stroke treatment in the UK and beyond.
What was the idea behind the development of e-Stroke?
When a stroke patient is admitted to the hospital, they get a CT scan. Most physicians use the ASPECTS score to assess how severe the stroke may be.
The problem is that that assessing the APECTS score is not very straightforward and you get a lot of variation. One doctor may say its eight out of 10 and another may say it’s six. As you would treat a patient differently depending on whether it’s a six or an eight, the score really has consequences.
Co-founders Michalis Papadakis and Professor Alastair Buchan of the University of Oxford had the idea to leverage artificial intelligence to standardise the way in which those brain scans are assessed. We first commercialised e-Aspects [one of three e-Stroke modules] in the UK in 2016.
So how did you actually develop the software and train the AI?
This was led by CTO Eric Greveson who has been involved in the development of the software from the start.
We used these giant datasets from brain scans of stroke patients in the UK, Germany and elsewhere to detect the ASPECTS score. After thousands of these, the software became more and more accurate.
Jeff Wyrtzen
So it’s still improving all the time?
Yes, absolutely. Our software has processed more than half a million brain scans. And as we’ve expanded into other modules, these have become more robust over time as well.
Who actually uses e-Stroke in the clinical setting?
Anyone who works within the stroke pathway can benefit from the software. The beauty of e-Stroke is that it’s seamlessly installed in the cloud or physically on a server. It’s designed to be incorporated into whatever stroke processes are already in place.
Let’s say a patient comes in, they get scanned at a CT machine and the image is then sent to the physicians. Our software will be automatically incorporated into that pathway.
No one needs to press a button, it just automatically processes the image and generates results within one to two minutes.
Then the doctor, nurse or radiologist will see it on their mobile phone, iPad or desktop and then can make decisions based on what they see.
Has it been pretty straightforward to get clinicians on board with the AI technology?
Yes, and we’ve seen a big uptick in the last couple of years, especially in the UK, where there’s been a more broader embrace of AI software.
If you go back to when we first launched, people were a bit nervous about AI. Doctors would say, ‘is this going to replace me?’ What’s been realised over time is that it’s not even remotely intended to be a replacement. It’s meant to strengthen and complement the work that’s being done.
I guess that’s especially important when you’re dealing with stroke…
Absolutely. Even a doctor who’s very well trained can save a couple of minutes. Every minute that’s lost with a stroke, that’s 1.9 million neurons. So, truly, every minute counts.
Have you had good feedback from the clinicians who’ve been using it?
Yes, absolutely. Last week, there was a study presented at the World Stroke Congress by Dr Kiruba Nagaratnam, Clinical Lead for Stroke Medicine at the Royal Berkshire Hospital.
The research looked at a year period before e-Stroke was installed and then the year afterwards. They measured how quickly a patient came into hospital and was then referred to go to another hospital for thrombectomy and then the outcomes.
With e-Stroke, there was a faster treatment time and they also saw a tripling of the number of patients who were functionally independent that 90 days. So those were really robust and promising results.
Where in the NHS has e-Stroke been introduced so far?
We’re in close to 70 hospitals here so far.
Last September, we were awarded a prestigious NHSX AI grant, which has allowed us to deploy the software to five different stroke networks in the UK totalling nearly 40 hospitals.
The data that we collect over three years will be regularly assessed to show what impact the software is having. So it’s things like speed of treatment, number of patients and treatment outcomes.
We’ll probably see more than 100,000 patients assessed over that time and if the results are as positive, hopefully we’ll get the green light to deploy the software, not only across more UK hospitals, but across other countries as well.
You already have a growing international footprint…
We’ve active in about 30 countries right now. In Poland, we successfully bid for what was essentially a national tender and we deployed the software across 17 of the largest hospitals in the country.
We also recently signed a distributor agreement with Wallaby Medical in China. We’ll be working with them over the next five years to distribute our software to more than 10,000 hospitals there.
Around 40 per cent of all strokes in the world occur in China, so you can imagine the scope of the opportunity.
What else do you have planned for the months and years ahead?
We have FDA clearance for one of our modules, but we’re looking to get clearance for the rest of the e-Stroke platform.
We already have a really good network of academic collaborators in the US, such as the Mayo Clinic, UCLA and Emory in Atlanta. We’re looking to leverage those relationships and have a really good presence in the US market.
Additionally, we’re the process of a new round of investment. Once we have that funding fully closed, we’ll be looking at some other indications.
In lung, for example, there’s a similar unmet need for AI to interpret images that can help standardise that assessment and help physicians make better treatment decisions.
