Jaap Kroes: "AI enables a more humane healthcare system."
Nijmegen tops the international rankings of cities using artificial intelligence to fight breast cancer and other diseases, right next to universities such as Harvard and Stanford. “Research, business and government come together here in a unique way, says Jaap Kroes, AI team leader at Screenpoint Medical. The Nijmegen-based medtech company is specialized in developing technology that allows breast cancer scans to be analyzed faster and can be analyzed more efficiently. "The amount of mistakes by AI decreases. With proper research and the right regulation, we can work on technology that allows us to
find breast cancer faster and more often, and hopefully in the future doctors will regain the time and space they need for their patients."
"One of the first real applications of AI in the medical world is in mammograms,the scans we use in detecting breast cancer," Kroes says. "That's because there have been all kinds of population studies all over the world for thirty years, so that data has been collected very consistently. In addition, the question with breast cancer is very clear; 'is there breast cancer yes or no?' With lung cancer, for example, it is more complicated. A lung is closer to all kinds of organs and because of that, there is more chance of 'bycatch'. All kinds of things happen in a breast, too, but because there aren't as many diseases that show up there, a tumor is relatively easier to find there."
Needle in a haystack
"Out of the thousands of women who get tested and have a mammogram, only 3 or 4 - fortunately - actually have breast cancer. But radiologists do have to check them all equally carefully, and it's not easy to spot a tumor on a scan." Finding breast cancer thus remains searching for a needle in a haystack. Previously, doctors did this by eye, but with technology from Screenpoint Medical, they can now have the scans analyzed by an AI system. "Our system already analyzed millions of scans, from women of different countries and with different backgrounds. Based on this data, it then comes up with an opinion or
conclusion on the scan entered. It’s up to the physician how to move forward with this output."
Kroes is in the lead of Screenpoint Medical's AI team. "I'm looking with the team at improvement of current products, but we also explore new interesting directions outside radiology, for example. We recently completed the MARBLE project, in which we developed software to compare women's current mammograms with their mammograms from previous scans [https://www.diagnijmegen.nl/news/marble_kickoff/]. A change between the two scans, may indicate something is going on there. Together with the Radboud University and the Radboudumc, and with support from EFRO, we expanded our technology, which in the future will allow us to provide even better support."
FDA-cleared
According to Kroes, Nijmegen is a world leader in the application of AI technology,
mainly in healthcare. "The Netherlands is hugely advanced in the field of breast cancer research and treatment. We were one of the first countries in the 1990s to establish a nationwide breast cancer screening program. Nijmegen is home to the Radboudumc, but also, for example, the national expertise center for breast cancer research (LRCB).
So here you can find companies, research, the clinical side - as well as the connection with the population. That's really unique."
Screenpoint Medical is not the only company in Nijmegen working with AI on the early detection of types of cancer. "The floor above us houses Thirona, they are specialized in AI technology for lung and eye cancer scans. Together with them, we are one of the few in the Netherlands that are FDA-cleared in this field. This means that we are may also roll out our technology in the US. It is really special to find two companies with that certification here in the same building on the Heijendaal campus."
Are we too strict with AI?
It sometimes seems as though humans are allowed to make mistake after mistake, but what about AI? "Twenty years ago AI was still a funny little thing you would come across once at a trade show somewhere. 5 years ago it was booming, everyone wanted something to do with it. Now we are in a phase where it is seen more and more as an application, really as a product. I think that is a positive development, but we still have to remain cautious. Because when confidence is misplaced or gets out of hand, then there is a greater risk of mistakes. You see this for example in ChatGPT, where people copy-paste texts without any check."
Still, AI generally makes fewer mistakes than humans. "People, including doctors, can be distracted, tired, or have things on their minds. As a patient, you can suffer from that. AI is controllable and measurable, and is not guided by prejudice or discrimination - with humans, you’re not always sure. Still, we are stricter with AI than on humans. Just look at the ‘benefits affair’ or self-driving cars: when things go wrong, all confidence is lost. Understandably so, because when it goes wrong, the consequences are huge. It may not sound very exciting, but that is precisely why good regulation and legislation at this stage are so important."
Staying critical in the AI revolution
"In other countries, the adoption of AI in population research is moving faster than in the Netherlands," notes Kroes. "For example, our products are already being used in Sweden, Denmark and Spain. One reason for this is that everything is locally or regionally organized. After a successful pilot in the Copenhagen region, we were able to move on to the next region in Denmark. In the Netherlands, the introduction of this type of technology is organized nationally, so it sometimes takes longer, but nevertheless, our country is still playing a leading role in this."
Kroes thinks that, with the right laws and regulations, AI can actually contribute to making healthcare more humane. "We should not use everything unchecked, but we should dare to apply of what we know works. From that you get better decisions and more personalized healthcare. If we apply AI intelligently, it also creates more space again for the
real contact between doctor and patient and more tailored care. And if you have the opportunity of using AI for a second opinion, and possibly save someone's life, wouldn’t it almost be unethical not to do so?”
