Title: Image Registration and Deep Neural Network Representations

Abstract:
Image registration is an important component for many medical image analysis methods to establish spatial correspondences. Over the last years many deep learning approaches to image registration have been proposed. In this talk, I will give a brief overview of how these deep learning approaches go beyond more classical optimization-based registration approaches. In particular, I will emphasize how these approaches motivate significantly simpler registration formulations and how they can motivate a different way of parameterizing deep neural networks.