Joost Batenburg about AI Leiden style: 3D images and ensuring AI belongs to everyone

Joost Batenburg is a mathematician and computer scientist who works to build bridges to other disciplines. He hopes to bring intelligent software to fields where it can make a difference. Conversely, he also seeks connections to the disciplines that are needed to make AI a success.

Take the legal sector. Batenburg: ‘AI makes it possible to quickly search through legal texts and put together smart contracts. At the same time, you still need legal and ethical experts to check that our rules and regulations are strong enough to keep all these technical possibilities on the right track.’

'I want AI not only to make the government more efficient, but also to help citizens'

One problem in AI is that the focus often lies on what is technically possible, with too little attention for interaction with humans, says Batenburg. ‘Our society is becoming increasingly data driven. Systems are becoming more business-like. Citizens get municipal letters written by a computer. I want AI not only to make the government more efficient, but also to help citizens. I want the whole population to see the opportunities of AI, that no one need be afraid of it. This is a good challenge for our university because we – also in collaboration with partners in Zuid-Holland such as Delft University of Technology and Erasmus University Rotterdam – already do so much research on the legal and ethical aspects of AI.’

Great potential for looking inside objects

Leiden is well-known for computational imaging, the branch of artificial intelligence that includes things like robotics. ‘This form of AI can, for example, recognise people on camera images and photographs; the kind of software Google uses to distinguish images. This technology also has great potential for looking inside objects, which is something we work on a lot at Leiden University. Ariane Briegel uses it to look inside cells, and Boudewijn Lelieveldt inside his patients.’

Ever smarter scanner

‘Between an MRI scanner taking shots and a doctor studying the resulting image, an image reconstruction programme has been at work. This process of image reconstruction is what my algorithmic research focuses on.’ Joost Batenburg is not only a bridge builder for AI; he also contributes to further improving the technology. 

Artificial intelligence is very suitable for identifying patterns and assessing data in large data sets, like medical scans, says Batenburg. ‘How much air is there in this lung? Is this a tumour or not?’ The next step is to create as complete an image as possible with a minimum of measurements. ‘This makes the process faster and less taxing for the patient. The software that translates the scanner’s measurements into images has to fill in the missing information itself, using known data.’ 

No need to wait for MRI results

Medical scanners may be faster, but they still search for standard patterns in the same way. Batenburg works on creating an algorithm that makes a scanner so smart it can adjust to what it sees in real time. ‘If it sees anything suspicious, it will look more carefully. And it could even do so in real time: if the algorithm becomes so fast it can present 3D images during a scan, an expert can look too and immediately zoom in where necessary, and patients no longer have to wait for MRI results.’

Text: Rianne Lindhout
Photo: Patricia Nauta