In Conversation with Prof. Anno Graser on the Use of AI in Prostate Diagnostics

As a leading expert in prostate diagnostics, Prof. Anno Graser, partner at Radiologie München, evaluates the role AI plays today and in the future. In this interview, he shares which aspects are particularly beneficial for young doctors and why he is excited to be part of the ongoing development of AI.

Prof. Graser, you are one of the leading experts in prostate diagnostics in Germany. What led you to this specialty, and how has your career developed in this field?

I started working with MRI of the prostate as part of my doctoral thesis in 1999, which was 25 years ago. Since then, I have personally conducted more than 25,000 mpMRI scans of the prostate. My passion for uroradiology began during my specialist training at the University Hospital of Munich, where I gained deep insights through daily collaboration with the urology department. I still benefit from the clinical knowledge I acquired during this time. A key experience was a research stay at MSKCC in New York in 2001, where I had the privilege of working with Prof. Hedvig Hricak, one of the pioneers of prostate MRI.

Your radiology center conducts nearly 5,000 prostate MRI exams annually. What measures do you take to ensure consistent quality across all diagnoses?

All prostate MRI exams at Radiologie München are either primarily reviewed and assessed by me or, at the very least, personally approved by me. This ensures that each patient benefits from my experience. It's time-consuming but crucial for maintaining consistent quality. Our urological partners know that we prioritize the highest standards in mpMRI of the prostate, which is one of our flagship services.

The PROBASE study shows that the accuracy of MRI in detecting clinically significant prostate cancer heavily depends on the radiologist's experience. In your opinion, can AI help reduce this dependency?

At present, this question can probably be answered with "yes." The advantage of AI lies in its voxel-based analysis, particularly of diffusion measurements, which allows for highly accurate detection of hypercellular areas that an inexperienced radiologist might overlook. Furthermore, AI results can help reduce false positives, particularly in the peripheral zone, which are often found by less experienced radiologists. In the transitional zone, I see potential for further improvement in AI applications, but even here, recent developments are very promising.

How do you evaluate the role of artificial intelligence in prostate diagnostics? What are the benefits of using AI algorithms for your radiology center, and why did you choose mdprostate from mediaire?

mdprostate has already proven to be robust in clinical practice. The AI is user-friendly and fast, making it easy to integrate into our workflow. I particularly appreciate that colleagues I train in mpMRI can use the AI to test themselves. Often, beginners find that the AI performs significantly better than humans. Competing with AI is both a stimulus and a challenge. As a top expert, I value the speed and reproducibility of AI, allowing me to ensure consistent quality in my results and make my findings traceable and legally defensible.

What are the biggest economic challenges you face when integrating AI into your diagnostic workflow, and how do you address them?

Due to its speed and reproducibility, AI will make nearly every radiologist a bit better. In this situation, economic considerations take a backseat, especially since we’ve been able to negotiate very fair conditions for using AI in cooperation with mediaire. Ultimately, I can only recommend that AI be widely implemented in prostate diagnostics, as the benefits are obvious.

What developments and innovations do you expect in the near future in prostate diagnostics, and where do you see the greatest need for innovation in the industry, particularly for AI providers?

A top expert is still better than the best algorithms, and I don’t believe this will change anytime soon, as humans are simply "smarter" than the best AI in borderline cases and atypical findings. Innovations are needed to further improve specificity, particularly in the transitional zone of the prostate. I also wish to better understand exactly what the algorithm is doing, as the localization schemes don’t always perfectly match the findings. However, I am confident that we will see innovations in this area almost monthly. I am excited to be part of this thrilling development!

Solutions

*bald verfügbar 

mediaire GmbH

Möckernstrasse 63
10965 Berlin, Germany

P +49 30 286 490 67
F +49 30 286 490 66

© 2022 mediaire GmbH

WORLD PREMIERE AT THE RÖKO 2024 IN GERMANY

Experience mdprostate, the first fully-automated AI-tool including longitudinal assessment for the evaluation of prostate MRIs.

mediaire GmbH

Möckernstrasse 63
10965 Berlin, Germany

P +49 30 286 490 67
F +49 30 286 490 66

© 2022 mediaire GmbH