Title: Combining visual computing with machine learning to improve image guided interventions
Abstract: Bothever-increasing computing power as well as data availability benefit the
medical imaging domain these days. However, for the image-guided surgery domain, this can only be
exploited with a carefully chosen mix of technologies. At ImFusion, we combine classical
medical image processing, a real-time visual computing framework, and deep learning, in
order to develop and analyze powerful algorithms. Those in turn allow us to create
software solutions that have a meaningful impact in real clinical environments. To show how
we arrived there; thetalk starts with an overview of some of my coming-of-age R&D
projects. I then present more recent work in interventional imaging, where real-time image
feeds are analyzed on ultrasound, video, and X-Ray, for various registration and motion
compensation problems. The main architecture decisions of the ImFusion software stack
are outlined, along with a company overview. The presentation concludes by discussing general
trends and business aspects of innovation in the medical device sector.
Bio: Wolfgang Wein has worked in medical image computing research for the last 20
years, conducting numerous projects from early feasibility to product implementation. He
received his doctoral degree in 2007 from Prof. Navab at the group for Computer Aided
Medical Procedures(CAMP) at TU Munich, Germany. From 2006-2010 he worked as a
research scientist at Siemens in Princeton, NJ USA, on various interventional navigation and
medical imaging projects. In 2012, he founded the R&D lab ImFusion in Munich, which aids
medical device companies around the world to create and commercialize innovation. He also has a
teaching assignment at TU Munich.