What exactly does the IMAGINATION project involve, and what problem does it aim to tackle?
IMAGINATION is investigating how we can make better use of artificial intelligence in the imaging of patients with gliomas, a group of malignant brain tumours. MRI scans are frequently used for these patients. AI can help to extract information from these images, for example regarding the type of tumour and the future course of the disease. However, an AI model that works well from a technical perspective is not automatically suitable for day-to-day patient care. Large datasets are required, and we need to know whether such a model will also work reliably for patients from other hospitals. As part of IMAGINATION, we are therefore pooling imaging data and specialist knowledge from three centres in the Netherlands and Germany. We are investigating various scientifically validated AI methods and standardising the analysis of MRI images across the centres. Ultimately, we want to find out where AI actually offers added value for the care of patients with glioma.
Many people associate AI with great opportunities, but also with risks. In your opinion, what challenges and opportunities does the use of AI in medical practice present?
The opportunities are enormous. As a radiologist or nuclear medicine specialist, you are constantly reviewing ever-increasing volumes of increasingly complex image data. AI can help to recognise patterns, automate measurements and potentially reveal information that is difficult for the human eye to detect. But we must also remain critical. An algorithm may work very well in the hospital where it was developed, but less well with other patients, scanners or imaging protocols. That is precisely why large, diverse data sets are so important. Furthermore, aspects such as data protection, transparency and accountability naturally play a role, but fortunately there are an increasing number of laws and regulations in this area in Europe. This enables us to ensure that data is used securely.
Furthermore, I do not see AI as a replacement for doctors. The interesting question is rather how we can combine the strengths of AI with the knowledge and skills of doctors so that patients ultimately receive better care.
The IMAGINATION project will run until September 2027. What has the project achieved so far?
We have now achieved some quite concrete results. First of all, we have jointly developed a method for consistently capturing, processing and archiving image data from the three centres. This is essential if one wishes to apply AI reliably to multicentre data.
Furthermore, we surveyed patients, relatives and healthcare professionals from the Netherlands and Germany on the use of AI in this context. The interview study showed that patients are predominantly positive towards AI, whilst professionals with more experience of AI are more critical. Furthermore, a discrepancy was particularly evident between the capabilities of AI and clinical requirements. Not all available AI tools address a meaningful research question or deliver clinically relevant results. End-users (doctors, nursing staff, patients and relatives) should be more closely involved in the process.
We have also made great strides technologically. We have selected and implemented various AI software packages. Through analyses of the data collected, we are increasingly moving from development towards the actual assessment of clinical added value.
IMAGINATION brings together partners from Germany and the Netherlands. Why is cross-border collaboration important in this area?
It is, in fact, essential for medical AI. Gliomas are relatively rare, and reliable AI requires a large amount of data. Through the collaboration of three centres from the Netherlands and Germany, we are not only obtaining a larger, but above all a more diverse, dataset. The latter is perhaps even more important. Hospitals in the Netherlands and Germany sometimes operate in different ways, which makes the dataset more diverse than if we were to include only Dutch or only German centres. An algorithm developed solely on the basis of data from a single centre or healthcare system may work well there, but could be far less reliable elsewhere. It is precisely through cross-border collaboration that we can investigate whether our methods remain robust under different conditions.
Furthermore, the collaboration offers an excellent opportunity to take a look behind the scenes at each other’s operations. Where do differences in working practices stem from, and what are their advantages and disadvantages? This also forces us to critically examine our own working practices. Ultimately, we aim to contribute to more consistent care in a border region where a five-minute drive can sometimes mean a patient ends up in a different healthcare system. Care should not depend on which side of the border a patient lives on.
What message would you like to pass on to other healthcare organisations that also wish to collaborate across borders?
My most important piece of advice is: invest at least as much in collaboration as you do in technology. Different countries and hospitals have different systems, procedures and regulations. Whilst this can sometimes present practical and administrative challenges, it is helpful to establish clear agreements in such cases. At IMAGINATION, we’ve found that straightforward communication and trust between partners are absolutely vital. When this foundation is in place, differences between the institutions actually become a strength. You bring together different areas of expertise and perspectives to develop solutions that work not only within your own hospital, but for people in the healthcare systems of both countries.
The IMAGINATION project is being carried out as part of the Interreg VI Deutschland-Nederland programme and is co-funded with €509,345.62 by the European Union, the MWIKE NRW and the Provincie Gelderland.
