Niessen emphasizes the need for trust and collaboration in AI

Which AI applications are the most promising? How can developers leverage data collection for deep learning? What predictions can we make about AI in radiology? These and other questions were addressed by panelists in a webinar organized on January 12.

In this edited Q&A article from the webinar, Niessen talks about the challenges of effective algorithm development and areas of particular interest to him. He is a professor of biomedical image analysis at Erasmus MC in Rotterdam and at the Delft University of Technology, founder and scientific manager of the AI ​​company Quantib (see the report of January 24) and director of the platform of IA from the European Organization for Research in Biomedical Imaging. He was president of the Medical Image Computing and Computer Assisted Interventions Society from 2016 to 2019.

Philip Ward, editor of, interviews Wiro Niessen from Rotterdam during the January 12 AI webinar.

Q: What are the hot topics in AI in radiology?

A: A hot topic is “trustworthy AI” and how the technology will bring value to clinical practice. If you translate that into what needs to be done on the imaging computing side, we need to develop algorithms that are generalizable and explainable, meaning they can be applied in different contexts and conditions, and explained to people who use them to enable responsible use. in clinical practice.

Trustworthy AI comes from transparency. It should be clear how the algorithms were developed and evaluated, and this knowledge should be available to people using the tool in their decision making. Such an approach can help avoid a disconnect between AI-generated results and clinicians’ reluctance to fully accept those results.

There is also a growing need for collaboration with clinicians as it is the clinical end user for whom AI should be of value. We need to establish the value of AI together – joint development of AI is the way forward.

Q: What hopes and expectations do you have for AI in the coming year?

A: A few years ago, everyone expected a lot from AI, but it took longer than expected for it to be introduced on a large scale. However, I have seen the power of this technology in so many applications. By bringing together daily routine clinical processes, data and advanced algorithms, we will build more and more reliable AI solutions that really make an impact.

I think by the end of this year we will have more examples of the true impact of AI technology. Even then, we won’t be close to reaching the full potential of AI in medicine or radiology, but in doing so, we will take very important steps and show that this is the path by which we will improve the quality of diagnoses and prognoses.

Q: What specific projects are you most excited about in 2022?

A: We would like to address a key question: how does nature and nurture contribute to patient outcomes?

The AI ​​tools that we use in radiology to make imaging a more quantitative aid to radiologists are also very powerful in investigating this fundamental life science question. In addition to deep learning networks for imaging data, we are also building deep learning networks to link genetic and omic data (genomics, proteomics, metabolomics, metagenomics, etc.) to patient outcomes. I am extremely interested in analyzing imaging, pathology, genetic and environmental data collected over a lifetime and using deep learning to find relationships between these and outcomes of relevant health, in order to better predict these.

Q: How do you think the pandemic will affect the evolution and implementation of AI over the next 12 months?

A: Of course, the pandemic has had a huge impact on so many people, but there will be real benefits if we can learn from the pandemic. One for our field is that people are increasingly aware that access to data is essential to respond quickly to change and to better treat patients.

Many countries are building infrastructure to better reuse routine clinical data for research and innovation. Good access to data is also essential for the development and validation of AI. This part of the pandemic response is actually going to support the development of AI methods.

If we can use the data collected in hospitals to make better diagnoses and prognoses, everyone will benefit. We need to respect privacy laws, but we also need to build a system where we increasingly give our data to research and create frameworks that make it easier for researchers to access the data to develop new tools. This is in the public interest and part of building a better ecosystem for AI in healthcare, and in particular in radiology.

Q: How many researchers do you have working specifically on AI at Erasmus MC?

A: In total, we have about 50 AI researchers, including scientific staff, post-docs, PhDs and Master of Science students, but these people work in multidisciplinary teams with AI and clinical supervisors, so that the extended team comprises about 100 people.

There is a lot of interest in funding AI research in the Netherlands. We are going to set up two new AI labs with 20 PhD students in the next ten years. These are public/private collaborative programs between industry and academia to ensure that the research we do will translate into products that will reach the patient.

Q: In a previous talk, you referred to “data collection is for life”. Can you explain what you meant by that?

A: We know that many diseases have their origins in youth or early life. Why do some people age in good health and why do others develop diseases? In order to answer these questions, we need to understand what happens throughout life. Effective prevention programs are only possible if we better understand the risk factors. So I think a combination of data collection across the lifespan combined with the power of AI-powered data analytics tools will provide new insights and make our approach to health more proactive.

Q: Another phrase you used is “Anything you can do, AI can do better”. Do you believe that?

A: This is not my own expression. A number of machine learning people have claimed that clinicians will become obsolete because of technology, but we now know that statement is not true. We increasingly see that human intelligence and current AI technology are very complementary.

I disagree with “Anything you can do, AI can do better”. Instead, I would say, “Some things you can do, AI can do better.”

We need to develop AI tools that complement clinicians for the benefit of the patient. This is the process we need to work on together in 2022 and beyond. Through better interaction between radiologist and AI, we will adopt AI techniques to improve clinical practice. I don’t foresee the radiologist becoming obsolete.

Editor’s Note: You can view a recording of the and webinar, “AI Trends in 2022: Clinical Radiology”, on demand via this link. Other AI webinars will take place on February 10 and April 6. The series is produced by Brian Casey, editor of, and sponsored by Bayer.

Copyright © 2022

About Hector Hedgepeth

Check Also

Study investigates cause of lymphopenia as seen in COVID-19 patients

In a recent study published on Research Square* preprint server, researchers explored the underlying mechanism …