"We use AI in everyday clinical routine."
May 2024
Simon Steppacher specializes in hip joint preservation in adolescents and young adults at the Department of Orthopedic Surgery and Traumatology of the Inselspital. As his patients’ lives yet lie ahead of them, he uses artificial intelligence for detailed insights into pre-surgery imaging for the best outcomes.
Prof. Steppacher, can you tell us about your clinical work?
Hip deformities can either be present from birth or develop during growth spurt (in girls mostly 11-13, in boys 12-14 years). The impaired biomechanics of the hip leads to increasing pain and early joint degeneration. Our goal as surgeons is to restore optimal biomechanics to alleviate the pain and preserve the hip joint.
For my research, I follow patients up for 20 to 30 years to find out who profited the most. Locomotion is one of basics for our quality of live and I’m happy to see them come back without pain. I also like the manual aspect of my work, the mechanical job or sawing and changing angles, the blades and screws.
Hip functionality is a very three-dimensional problem.
What is the most challenging aspect of these surgeries?
Orthopedics is an image-based medicine and hip functionality is a very three-dimensional problem. Yet, current imaging is mostly two-dimensional. With the help of artificial intelligence, we aim to close this gap and improve diagnostics.
To correct the hip morphology and stop degeneration, it is important that the joint cartilage is still of good quality. Modern magnetic resonance imaging (MRI) allows to evaluate biochemical composition of cartilage and detect early degeneration. Today, you can only get 3D information by manually segmenting hundreds of MR slices – a task far too laborious to be done before every surgery. AI can distinguish bone or cartilage within seconds.
Since we started working with it in 2017 as part of a research project, we have developed and validated an AI tool that allows to create 3D models of hip joints with information on cartilage quality, a tool we now use in everyday clinical routine.
This new knowledge can help us understand how exactly hip deformities differ from healthy hips and which approach during surgery will improve patient results.
Do you as a doctor feel challenged by AI?
I have been in orthopedics for 17 years and there have been many advancements in software we rely on. I think AI is a hot topic and I am glad to work together with AI experts from the personalized medicine group.
Today, AI is mostly applied for image segmentation, but it could soon suggest treatment options and predict their success. This is essentially what doctors are doing now: we evaluate imaging, diagnose, propose treatment, and tell the patient what to expect. AI has the potential to do all of this, and it can play a role in the treatment itself.
This is probably what people are afraid of. We need to be aware that an AI is trained with the correct data. Training with rare deformities is difficult as we don’t have sufficient image examples of these. But I think we need AI for 3D segmentation, and it could even provide us with “early warnings” of deteriorating cartilage. And we need the human to verify if the results are plausible. So, I have no worries that the software gets autonomous.
AI has the potential to evaluate imaging, diagnose, propose treatment, and tell the patient what to expect.
Why have you chosen the Inselspital as a hip surgeon?
The Department of Orthopedic Surgery at the Inselspital is world-famous in the field of hip deformity correction. In the 1980s and 1990s, various hip deformities and their correction surgeries were first described and developed here that are today still state-of-the-art. The hip symposium Bern is an international recognized specialist training on hip deformity surgery. We have a great wealth of experience and can ask patients 30 years after surgery if they still have their natural hip joint to evaluate outcome.
In addition, we surgeons have a very good interdisciplinary exchange on technology-heavy aspects of our work with ARTORG, sitem, and our radiology right on the doorstep, a unique strength here in Bern.
The Department of Orthopedic Surgery at the Inselspital is world-famous in hip deformity correction.
After completing medical school at the University of Bern in 2005, Prof. Dr. med. Simon Steppacher dedicated himself to orthopedic research for three years, including a one-year tenure supported by a Swiss National Science (SNF) scholarship in Boston at Tufts University and Harvard Medical School. His research centered on the development and validation of computer methods for calculating three-dimensional cup orientation in hip arthroplasty. He completed his clinical training at the Inselspital across the Departments of Orthopaedic Surgery, Emergency Medicine, and Hand Surgery. In 2013 and 2014, he allocated additional time to orthopedic research, focusing on evaluating the long-term outcomes of joint-preserving hip surgery at the Inselspital. Simon Steppacher achieved board certification as an orthopedic surgeon and consultant in 2017, subsequently assuming the role of senior consultant within the hip team since 2019.
Throughout his research career, Prof. Steppacher collaborated closely with computer scientists and radiologists to advance hip imaging and pioneer new computer methods to better understand hip biomechanics. Serving as the head of the hip research team, he secured a research grant from the SNF Foundation to develop and validate AI-powered algorithms for automatic segmentation of hip MRI. These AI-driven tools facilitate enhanced 3D imaging of the hip, improve the understanding of hip biomechanics, refining surgical decision-making, and ultimately improving the outcomes of hip joint-preserving surgery.
Project
Automatic and patient-specific 3D MRI model of hip cartilage and labrum
Publications
- Ruckli AC, Schmaranzer F, Meier MK, Lerch TD, Steppacher SD, Tannast M, Zeng G, Burger J, Siebenrock KA, Gerber N, Gerber K. Automated quantification of cartilage quality for hip treatment decision support. Int J Comput Assist Radiol Surg. 2022 Nov;17(11):2011-2021. doi: 10.1007/s11548-022-02714-z. Epub 2022 Aug 17. PMID: 35976596; PMCID: PMC9515031.
- Ruckli AC, Nanavati AK, Meier MK, Lerch TD, Steppacher SD, Vuilleumier S, Boschung A, Vuillemin N, Tannast M, Siebenrock KA, Gerber N, Schmaranzer F. A Deep Learning Method for Quantification of Femoral Head Necrosis Based on Routine Hip MRI for Improved Surgical Decision Making. J Pers Med. 2023 Jan 12;13(1):153. doi: 10.3390/jpm13010153. PMID: 36675814; PMCID: PMC9862886.
- https://www.youtube.com/watch?v=RtxtPqExJRg