We discuss recent advances in diffeomorphic image registration and related correspondence and shape matching problems. Diffeomorphic image registration is a classical, ill-posed, non-linear, non-convex, inverse problem with numerous applications in imaging sciences. It typically involves an infinite number of unknowns, which, upon discretization, results in high-dimensional, ill-conditioned systems. Image registration poses significant numerical challenges. We will showcase state-of-the-art techniques in scientific computing to tackle these challenges, highlight new theoretical developments, and discuss challenging application scenarios that require tailored formulations to obtain plausible solutions.

Modelling and complexity issues on large deformations for shape ensembles

Alain Trouvé (Centre de Mathématiques et Leurs Applications)

Optimal transport for diffeomorphic registration

François-Xavier Vialard (University Paris-Dauphine)

A Lagrangian Framework for Fast and Flexible Diffeomorphic Image Registration

Lars Ruthotto (Department of Mathematics and Computer Science, Emory University)

Statistically-constrained Robust Diffeomorphic Registration

Aristeidis Sotiras (University of Pennsylvania)

Non-parametric registration of medical image data using Schatten-q-Norms

Kai Brehmer (Institute of Mathematics and Image Computing, University of Lübeck)

Machine Learning Approaches for Deformable Image Registration

Marc Niethammer (University of North Carolina at Chapel Hill)

GPU Based Geodesics of Image Time Series

Benjamin Berkels (RWTH Aachen University)

CLAIRE: A parallel solver for constrained large deformation diffeomorphic image registration

Andreas Mang (Department of Mathematics, University of Houston)

Organizers:

George Biros (Institute for Computational Engineering and Sciences, University of Texas at Austin)

Andreas Mang (Department of Mathematics, University of Houston)

Keywords:

computer vision, image registration, inverse problems, nonlinear optimization