Medical Image Registration: Toward Applicaton-Specific Analysis Approaches
Speaker: Marc Niethammer, University of North Carolina at Chapel Hill
Image registration is an important component for many medical image analysis methods. It is a means to establish spatial correspondences within or across subjects and a fundamental building block for example for atlas-building. Classical image registration approaches consider registrations between pairs of images, rely on globally smooth spatial transformations, and assume similarity in image appearances. Hence, they are not designed to address longitudinal imaging studies, to capture complex spatial transformations, or to account for expected image appearance changes as observed for example in studies of neurodevelopment or during disease progression in cancer or traumatic brain injury. This talk will therefore discuss application-specific modeling approaches to address these challenges.
Appropriate modeling is of particular importance for non-parametric deformable registration approaches, such as the popular large-displacement diffeomorphic metric mapping (LDDMM) method, which allow to capture spatially complex transformations, but may result in undesirable transformations if model assumptions do not hold. In particular, this talk will discuss approaches to model spatial and appearance changes, how to numerically solve the associated high-dimensional optimization problems, and some initial strategies to account for registration uncertainties. An outlook on challenges and opportunities in large-scale image analysis will be given.
Bio: Marc Niethammer is an Associate Professor at the University of North Carolina at Chapel Hill with a joint appointment in the Department of Computer Science and with the Biomedical Research Imaging Center (BRIC). He received his Ph.D. in Electrical and Computer Engineering from the Georgia Institute of Technology and post-doctoral training at Harvard Medical School/Brigham and Women's Hospital. His research interests lie in the areas of biomedical image analysis focusing on application-driven algorithm design for segmentation, registration, and shape analysis.
For more information, contact Prof. Guido Gerig.