Stephen Aylward, MICCAI Fellow 2019

Stephen Aylward is the Global Lead for Strategic Applied Research in the Healthcare and Life Sciences, Medical Devices Team at NVIDIA, specializing in AI-enabled medical technologies. His collaborations focus on combining hardware acceleration with open-source software for the discovery and support of next generation medical AI devices. He is also an adjunct professor in computer science at the University of North Carolina at Chapel Hill (UNC), chair of the MONAI advisory board, and an advisory board member for the Vanderbilt Institute for Surgery and Engineering and for Johns Hopkins Laboratory for Computational Sensing and Robotics. Previously, he was senior director of Strategic Initiatives at Kitware and a tenured professor in radiology at UNC. (Source: NVIDIA)

Stephen was elected to the MICCAI Society Board of Directors in 2013 and served as the Board Treasurer from 2016 to 2021. He was named a MICCAI Fellow in 2019 for his work in promoting open-source software and his contributions to the field of medical image analysis in addition to his outstanding service to the MICCAI community. He currently serves on the MICCAI 2025 Organizing Committee as the Industry Engagement Co-Chair.

Thank you for connecting with us Stephen and congratulations on your achievements!

Q. When and how did you get involved with the MICCAI Society? 

A. I was at the first MICCAI in 1998, and I consider that to be the start of my joining the Society.  Even though the Society wasn’t officially formed until 2004, my friendships and collaborations with the MICCAI community began at that first conference.   I was surrounded by the folks who shaped our field, as well as the MICCAI Society: Ron Kikinis, Tina Kapur, Eric Grimson, Russ Taylor, Sandy Wells, Max Viergever, Nicholas Ayach, Jim Duncan, and many others.  As a recent PhD graduate under Steve Pizer, I had heard these names and was basing my work on the algorithms and applications they had developed.  To get to interact with them, was phenomenal.  And that camaraderie and thought leadership continue to this day.

My involvement in MICCAI has been relatively continuous since that time and has featured quite a bit of exploration.   As a part of the development of ITK, we held the first-ever MICCAI tutorial, featuring ITK, in 2003, and I have helped on numerous workshops since, with particular interest in the series of workshops on ultrasound that began in 2017 and continue to this year.  I also started and led the Young Scientist Publication Award for 10 years and was a Society board member for 7 years and the Treasurer for 5 years.  As treasurer, I worked closely with Janette Wallace and Johanne Langford to engage Dekon to bring financial and organization stability to the MICCAI conference.  I also helped to start the SIG initiative.   This is in addition to over 20 main conference publications and many more workshop publications.

Q. What is your main area of research? How has your focus evolved over time?

A. A major focus of my career has been open science, collaborating with researchers around the world to develop and disseminate open-source software and open-access data.  I get amazing satisfaction from the impact that our open-source efforts (e.g., ITK, CMake, 3D Slicer, and now MONAI) have had on research and clinical products and practice.

Regarding research, it also has always been very collaborative. Early in my career I was working with Dr. Etta Pisano, an amazing researcher and mammographer, on the introduction of digital mammography into clinical practice, featuring novel breast cancer risk assessment methods and visualization techniques.  However, for much of my career, I was focused on vascular segmentation with Dr. Elizabeth Bullitt, a neurosurgeon who was also an excellent programmer.  She was an outstanding collaborator and mentor.   Perhaps our best-known work involves her pioneering use of high-field MRI to quantify the tortuosity of vasculature with and adjacent to tumors as an indicator of malignancy and an early indicator of response to therapy.  At Kitware, much of my research focused on point-of-care ultrasound and the analysis of ultrasound RF for a variety of in-field trauma assessment tasks, ranging from detecting intra-abdominal bleeding to increased intracranial pressure associated with traumatic brain injury.

At NVIDIA, I am now working with amazing researchers around the world on combining AI and physics to accelerate the simulation of physiological processes such as respiratory and cardiac motion as well as the flow of blood and the perfusion of tissues.  These simulations are meant to increase the fidelity of digital twins of patients, to aid in pre-operative planning and intra-operative guidance, particularly when highly dexterous, minimally invasive surgical robotics are involved.  This work is also feeding into a larger initiative at NVIDIA, Isaac for Healthcare.  That initiative is producing a framework that enables healthcare developers to create digital twins of their AI-enabled robots and sensors and to then use AI-generated patients in virtual environments to safely teach new skills to their AI systems.  In these virtual environments, an arbitrary number and diversity of clinical conditions can be simulated to create robust AI systems that need much less real-world training prior to deployment.  Furthermore, NVIDIA is producing the real-time robotic control and sensor processing hardware needed to train and deploy those AI-enabled systems in clinical environments.

Q. You spent nearly a decade as a tenured Associate Professor of Radiology at the University of North Carolina before moving into industry leadership roles at Kitware and now NVIDIA. What motivated your transition from academia to industry?

A. This path was largely driven by my desire to maximize the impact I could have via open-source software and collaboration.  At UNC, my lab was one of the teams participating in the initial development of ITK.  As ITK progressed, I saw the massive impact it had on the field.  Prior to ITK, it was expected that most grad students would write their own image processing libraries as needed to conduct their research.  Open-source was largely a new and under-appreciated concept.  With ITK, a community grew, ideas and code were shared, and the pace of research and development was massively accelerated.  Kitware was instrumental in the development and quality of ITK, and under the leadership of Will Schroeder, Kitware was pursuing a mission of enabling innovation via open science.  I knew that I could have greater impact on the field by joining that team and sharing in that mission, rather than focusing on my own research. NVIDIA is a continuation and expansion of that same desire to have impact via open science and enabling innovation. NVIDIA has a proven track record of promoting open science via MONAI and a multitude of other open-source frameworks.  Combining open science with accelerated hardware has an exponential effect on the pace of research and development. We are solving larger and more challenging problems than ever before, and that work is moving into clinical practice faster than ever before.  Once again, as I did with ITK and Kitware, I feel that I am at the right place, at the right time, with NVIDIA.

Q. Can you share some of the most exciting or transformative developments you've witnessed in medical imaging and applied clinical research during your career?

A. The adoption of open science by the AI community is one of the most influential evolutions that I have seen in the field.   The sharing of data, code, and publications has enabled AI to advance faster than any other technology that I have touched. To learn about the latest research, go to ArXiv.  To find relevant clinical data, begin looking at The Cancer Imaging Archive. To get the latest AI algorithms, turn to MONAI.   And in all that you do, contribute back to those and similar open science foundations.

Q. What does the future hold for medical image research and clinical applications?

A. The future of healthcare is digital twin technologies.  These technologies will span multiple scales: from smart hospitals that monitor and adapt resource availability to the ebb and flow of patients to surgical robots that use intra-operative data to continuously update the patient’s digital twin and refine treatment plans based on long-term projections of patient’s outcome based on the patient’s history and lifestyle.

Q. What advice would you give to early-career scientists who are looking to build their career, whether in academia or in industry?

A. First, figure out what everyone else is doing, and then do something slightly different. Think about “what’s next”.  For example, after everyone succeeds at AI for image segmentation and registration, what’s next?   After AI can write a radiological report, what’s next?

Second, play the long game: research is a team sport, and everyone can be a winner.  Like any sport, it takes practice and persistence – it will take years to succeed.  By creating a collaborative team who share your drive and focus, the chances of winning are greatly improved.  Also, research is not a zero-sum game, you don’t succeed by others failing.  Everyone can win, and helping others win very often helps you along your own path to success.

Q. Congratulations to you and the MICCAI 2025 Organizing Committee for planning such an exciting conference in Daejeon, South Korea this coming September. What are you looking forward to at MICCAI 2025?

A. I think we are going to see even broader and deeper involvement from industry, and particularly from startups.  I think the energy, clinical focus, and innovative spirit inherent in most healthcare startups make them outstanding members of the MICCAI community.  MICCAI’s CLINNICAI, Startup Village, industry keynotes, and many mentorship opportunities for entrepreneurs and new scientists are going to play an even more prominent role at MICCAI 2025, providing great opportunities to inspire and advance new ideas throughout and across academia and industry.