We caught up with Günther Brueggenwerth and Arndt Schmitz from Bayer at the Festival of Genomics in January to talk about Radiomics, its promises and uses, and about Bayer’s newly established LifeHub UK.
FLG: Could you give us an introduction of yourselves and the work you do?
Günther: I’m Günther Brueggenwerth and I’m a radiologist in my background. Since October 2019, I’ve been leading the radiology section of the newly opened LifeHub in Reading, UK where we are focusing on developing new artificial intelligence algorithms for the radiology diagnostic world, but also for oncology and other areas.
Arndt: I’m Arndt Schmitz and I’m a biochemist by training. I’m a Senior Technology Expert in the Clinical Operations department at Bayer, and I’m very interested in this role and in anything that drives forward our processes to better serve patients.
FLG: What was your motivation to do what you do?
Günther: I was working as a general radiologist in hospital environment, but moving to the industry and being able to develop products is certainly a challenge that I thought at the time would be really great to do. Since 2002, I’ve been in research and development at Bayer. Now, there is a path to more development with newer techniques such as artificial intelligence, and this exciting opportunity has been my motivation to move forward with AI solutions in the UK.
Arndt: I agree. Even today it’s still a thrill and there’s a lot of excitement to tackle very challenging projects and hopefully to bring them to result, whilst at the same time advancing health care in general. It’s very rewarding.
FLG: What is radiomics?
Günther: Radiomics is a part of radiology where we are analysing classical imaging modalities such as computed tomography (CT), magnetic resonance imaging (MRI) or positron emission tomography (PET) with advanced statistical methods that we never had before. This is very special as now we can analyse features on a level that cannot be seen with the naked eye. This means analysing at pixel level, finding correlations between sharpness and between any histograms made in correlation to prognostic outcome and diagnosis of a disease. That’s the new chance we are having now with radiomics.
FLG: What success rates is radiomics seeing in correctly diagnosing patients, and what does it mean for patients?
Günther: Great value has been shown with the radiomics approach. For example, in differentiating benign from a malignant disease, making a prognosis on how long a patient might live, making a prognosis on what would be the best treatment, or finding a correlation to the genomic profile. However, we are still at the very early stages with radiomics and it’s currently not established in the clinical routine. Not yet. For patients, there’s no further risk as this is not an invasive method and it’s only the analysis of these images that is new.
FLG: How does genomics help in radiomics?
Arndt: Without genomics, there would be much less accuracy with radiomics alone because we don’t only need the images acquired , we also need genomic annotations. If you have a large cohort of training data and have the image combined with its molecular genetic annotation, only then can you provide these algorithms and in turn can help develop these tests further.
FLG: Are there any challenges that you face in your work?
Arndt: As Günther mentioned, it’s early stages for this field. There’s a lot to learn, and with LifeHub UK, we are currently establishing a new ecosystem of collaborations in this field. We’re having lots of coffees with different potential partners and listening carefully to the community, seeing what a good pilot would be to run.
Günther: Another challenge is that, as I mentioned before, the radiomics approach is not yet established in the clinical routine setting. This is something we’re working on, especially for imaging modalities like MRI where it’s been shown that the imaging features coming out of these procedures aren’t so stable between different MRI modalities from different sites. The standardisation of this method is in its early days, and is would be the challenge to solve when bringing it into the clinical routine.
FLG: What do you think is the most exciting thing happening in genomics right now?
Arndt: I was a Postdoc in the USA when the first human genome was sequenced and published, and today just a few decades down the line, there are cohorts of hundreds of genomic sequences being published. There are loads of genomes being sequenced every day, and I think this is very exciting. It allows a lot more research to be carried out and has such a true impact on clinicians and patients.
FLG: Why have you decided to participate at the Festival of Genomics this year?
Arndt: I was honoured to speak at the Festival last year and it was easy to decide to come back. I love this event for the UK genomics and clinical community with a strong presence of the NHS, but also for all other regions attending. It’s great to get a view of the field and for networking, and we also thought in our humble opinion, that we have a nice case study to present and provoke some thought.
Günther: I think it’s a great opportunity for me to introduce the newly established LifeHub UK and to speak to others about any interests in partnering with us. Also, having new ideas and connecting the radiology world with the genomics world and developing new AI algorithms is all for the benefit of the patients.
