Keynote Talk: Jerry Prince

Progress in Disentangling Anatomy and Contrast for MRI Harmonization


Synthesizing contrasts in magnetic resonance imaging (MRI) has become commonplace and quite robust since the introduction of deep networks in medical imaging. Imputation of missing contrasts is the most common use for such synthesis (sometimes called image translation). Harmonization is the more subtle adjustment of similar contrasts so that images can be processed consistently across multiple sites. This talk describes advances in harmonizing MR images by separating their anatomies from their contrasts by using disentangled latent space deep networks. The key insight we recently advanced is to exploit paired multi-contrast images that are available for all scanning sessions as a supervised part of an otherwise unsupervised scenario in which no traveling subject data are available across sites. We have since developed an underlying theoretical basis for this approach and optimized several aspects of the network architecture and its training methodology. This talk provides and overview of these developments along with results on several diverse harmonization challenges.

Speaker’s Bio

Jerry L. Prince received the B.S. degree from the University of Connecticut in 1979 and the S.M., E.E., and Ph.D. degrees in 1982, 1986, and 1988, respectively, from the Massachusetts Institute of Technology, all in electrical engineering and computer science. He worked at the Brigham and Women’s Hospital, MIT Lincoln Laboratories, and The Analytic Sciences Corporation (TASC). He joined the faculty at the Johns Hopkins University in 1989, where he is currently William B. Kouwenhoven Professor in the Department of Electrical and Computer Engineering and holds joint appointments in the Departments of Radiology, Biomedical Engineering, Computer Science, and Applied Mathematics and Statistics. Dr. Prince is a Fellow of the IEEE, Fellow of the MICCAI Society, Fellow of the AIMBE, and a member of Sigma Xi. He also holds memberships in Tau Beta Pi, Eta Kappa Nu, and Phi Kappa Phi honor societies. He was an Associate Editor of IEEE Transactions on Image Processing from 1992-1995, an Associate Editor of IEEE Transactions on Medical Imaging from 2000-2004 and is currently a member of the Editorial Board of Medical Image Analysis. Dr. Prince received a 1993 National Science Foundation Presidential Faculty Fellows Award, was Maryland’s 1997 Outstanding Young Engineer, and was awarded the MICCAI Society Enduring Impact Award in 2012. He is also co-founder of Sonavex, Inc., a biotechnology company based in Baltimore, MD. His current research interests are in image processing, computer vision, and machine learning with primary application to medical imaging; he has published over 500 articles on these subjects.