Collective dynamics, control and imaging

Beyond Nyquist – accelerating magnetic resonance imaging

Sebastian Kozerke

ETH Zürich


The talk will introduce and review concepts for accelerating Magnetic Resonance Imaging based on parallel signal reception, transform coding, compressive sampling and prior-driven image reconstruction. An update on data-driven methods that can account for object motion, vector field divergence and other physical prior information will be given along with explorations into nonlinear transform domains and dictionary learning for improved separation of image content and image artifacts. Examples from preclinical and clinical imaging of cardiac perfusion, metabolism, tissue microstructure and tissue mechanics will be presented leading to predictive image-guided computational modeling of the entire human heart.