New Faculty: Renata Raidou
Renata is delighted to be back at TU Wien, where she did her Post-Doc. In her research, she tries to integrate Visual Analytics into clinical decision making.
Renata Raidou joined our research unit Computer Graphics in March this year as an assistant professor. She investigates and designs Visualization and Visual Analytics strategies, which address biological and medical applications. Her research focuses on the interface between Visual Analytics, Image Processing, and Machine Learning, with a strong focus on medical applications.
What brought you to TU Wien Informatics and Vienna?
I was already at TU Wien before, and I was also in the group where I am now. So practically, I knew what I was signing up for. And I was already happy with what I had experienced in the past. For TU Wien, I can say that it is a very high-ranking university with a tremendous reputation. This very old university has an excellent standing in the national and international research scene. What matters to me is also the high-quality education they provide, not just the research. The group where I am now, the visualization group, is part of the computer graphics research group. It is one of the oldest, maybe even the most senior group in visualization, and it has a history of 25 to 30 years in this domain. It has also been a cradle of well-known researchers, as many people in the community have passed through here.
The leader of the group, Edi Gröller is one of the most inspiring people and researchers that I have met, and also one of the most supportive colleagues and bosses I have had. So, I was delighted to go back to the group. We are also very diverse in terms of topics: I do biomedical visualization, but other people do networks and information visualization. I find it very interesting because you can always learn from other topics, even when they are not close to yours. We have diverse personalities: shy people, talkative people, and crazy people, which I find very attractive. There is also this family-like environment in the group and the close interaction with students. Vienna, too, played a significant role in my decision. As a cultural and social hotspot, it is one of the best living spaces in the world, and it is evident that when you get an offer to come here, you are like, “sure, I have already packed”. It is so much better than any other city that I have lived in so far.
Can you explain what exactly Medical Visualization and Visual Analytics are?
It is a scientific field that takes advantage of our vision and perception to increase our understanding and insight into complex data. Biomedical visualization concerns itself with biomedical data and processes. We want to gain insight into how the human body works (or doesn’t work). Visual analytics has emerged in the last few decades and has been applied to many biomedical scenarios. It concerns itself with combining automated analysis techniques, such as machine learning or data mining, and interactive visualization in such a way that you can support reasoning and decision-making. It mostly looks into very complex and extensive data sets. You need to find ways to integrate the strengths of the computer, like speedy calculations, complex algorithms, automated approaches like machine learning with human strengths. You need a human expert who can bring in their knowledge, expertise, and mental models concerning patients in the particular biomedical visualization case.
Traditionally, biomedical visualization initially started with volume rendering, with application domains such as radiotherapy for cancer. Nowadays, it has moved on to a vast spectrum of data and applications. We have a giant galaxy of medical information like genomic data, cellular data, tissue formation, multi-modal images, simulations, models that tell you how physiological processes in our body work, data from population studies. And all this creates an enormous volume of medical information, which goes beyond this traditional 3D-rendering of, for example, CT data. We still do that, but we mainly look into all this vast “jungle” of data.
How does the collaboration with medical personnel work?
There are biomedical visualization frameworks for patients themselves and other frameworks created for clinical researchers. Clinicians have neither the time nor the technical knowledge to start interacting with the data and looking at everything. They need to have some support and visual enhancement of the data to quickly decide since they have about one minute to inspect the data. We have to take care of and respect this fact. I cannot create a super complex framework, needing one thousand interactions before the doctor can decide. On the other hand, we can find clinical researchers and maybe some industrial partners more concerned with understanding the data, confirming some hypotheses throughout a more extended period. In this case, we have more flexibility to develop more complex frameworks.
I have worked with my clinical collaborators for many years in the past. You need a very long time to feel comfortable with the domain and understand what they do. From their point of view, they do not know what visualization does and what we are doing. You need this period of mutual understanding and learning the language of each other. Visual analytics approaches are complex and are rarely adopted in clinical praxis. With my long-time collaboration with European and American colleagues, we managed to integrate visual analytics into clinical decision making, which is an unusual example of going so far into clinical practice.
How would you describe your work in 90 seconds?
I am an expert in biomedical visual analytics, and I investigate and design visual analytics approaches, especially concerning complex biomedical data. My focus is on applications such as prostate cancer radiotherapy. My work focuses on the P4 direction of medicine, which stands for personalized, predictive, preventive and participatory medicine. It indicates that the diagnosis or treatment is tailored to one specific individual. Also, risk factors in this treatment can be identified early, and you can address them before they even show up. Finally, the patients are actively involved in the whole treatment process.
I try to provide clinical researchers with new possibilities to create a more targeted and robust P4 patient treatment through the use of visual analytics systems. Recently, we started looking at how we can apply physicalization concepts to anatomical education: Creating physical objects from the data, where instead of using expensive 3D printing, you can use inexpensive and accessible material like paper and foils. Physicalization engages people in looking, interacting and even playing with their data. Through this edutainment process, we want to engage laypeople or school children to look more into anatomy and to learn more about how our body looks like.
What are the challenges in your research area?
The first challenge I would like to mention is related to the data or the applications we work with. There is a need of stopping doing old stuff. Traditionally, visualization was meant for diagnosis and treatment. Nowadays, we have this massive amount of data and the opportunity to move towards new directions, like helping doctors and clinical researchers to predict and prevent diseases. We need to put together a considerable amount of heterogeneous data coming from large populations. There are many sub-challenges because you have to put together all this information, make the data available and accessible to the users in a digestible manner, and help them make sense of the data. Also, reliable visualizations play an essential role. Visual analytics relies heavily on machine learning approaches, and clinical users don’t know whether they can rely on them. Questions like “Who is the patient going to sue?—the doctor or the engineer or the system designer? ” have to be addressed before integrating visual analytics systems into clinical practice.
Finally, patient communication is becoming more relevant: a doctor tries to inform the patient about their condition or treatment. Here, visualization can be of great value. We live in an era of fake news concerning health, as seen in this COVID pandemic. Medicine and biology require profound knowledge a lot of people do not have. It is a big challenge to help people understand how these complex issues like vaccinations work and make this scientific information available to everybody. Biomedical visualization has a great potential in supporting the dissemination of more accurate information to the general public.
What makes you happy in your work?
Small things make me happy in my work. First of all, if I manage to be creative throughout my work makes me happy. I realized throughout the years that I need to work on topics that are both societally impactful and relevant for industry. In biomedical visualization, you create a framework anticipating and keeping in mind that it is clinically and industrially adopted. Most applications do not reach this point, so this would be the happiest moment for me. Apart from this, being part of a perfect working environment like I am in now is very important. Especially teaching students is the most gratifying part of my work. Staying in university and interacting with students who bring freshness and enthusiasm to my work is vital. Sometimes they come up with the craziest and super creative stuff, and you also get into this enthusiastic trail with them, and this is the part of my work that makes me most happy.