In our latest interview, we spoke with Merel Huisman, a cardiovascular and musculoskeletal radiologist at Radboud University Medical Center in Nijmegen, Netherlands and an EuSoMII Board Member. Huisman has a passion for artificial intelligence, spanning almost a decade, even appearing on television in the Netherlands to discuss AI in the healthcare space. Furthermore, Huisman is also an Associate Editor for the AI-focused journal “Radiology: Artificial Intelligence”.
What is your background/experience with artificial intelligence and what first attracted you to the topic?
Six years ago, I attended an AI-themed workshop at my current institution, which sparked my interest, especially since I was one of the few clinicians present. During my Ph.D., I completed a second master’s in clinical epidemiology, where I saw overlaps with data science and the need for interdisciplinary collaboration. Recognizing the potential and challenges, I made it my mission to invest in advancing the field and inspiring others to embrace AI. Along the way, I got involved in several national and international journals, committees, and societies, and I founded the EuSoMII Young Club – now one of the most active ESR Young Clubs and the only interdisciplinary one.
What are the biggest challenges to AI adoption in clinical practice?
The products that make a real difference have just started to develop due to the recent advent of generative AI. The biggest need for radiology practices is to manage increasing volumes and complexity of imaging, creating a demand for reporting and workflow tools. First-generation narrow AI tools often lack sound evidence and a solid business case, preventing broader clinical uptake for good reasons.
Give us an example (or an educated guess) of what you think AI will be able to do in 3 years. What about in 10 years?
In three years, AI will significantly enhance workflow efficiency by automating routine tasks in the preparation phase of reading (e.g., summarizing relevant clinical information and automated hanging protocols regardless of the DICOM header) and part of the reporting (automated measures and tables). Large language models will save substantial time, and autonomous reporting of simpler studies will become more common. In ten years, AI will play a pivotal role in personalized medicine thereby hopefully preserving healthcare access and maintaining adequate resource management in times of escalating demand.
As a radiologist, what is your advice for younger colleagues wanting to dive into the topic of Artificial Intelligence, Machine/Deep Learning, and/or Radiomics? What are your tricks for staying up to date in this fast-evolving field?
For younger colleagues, I recommend pursuing interdisciplinary education whenever possible, such as combining clinical training with data science or epidemiology. For those past the training phase, there are many good resources managed by societies and journals, such as the ESR Masterclass in AI series. Engaging in research projects and pilot studies in your department provides the necessary practical experience.
Journal blogs, podcasts, videos, editorials, and strategic use of social media are a less time-consuming but effective way to stay up to date. I personally also stay up to date by both interacting with my network and by giving presentations, which I always have to update shortly before I speak given the fast pace of development.
