Phase constraints in parallel Magnetic Resonance Imaging (MRI) allow improved reconstruction by exploiting that the image is real-valued. However, in practice, high-frequency phase variations may occur and may lead to artifacts. Here, we present a formulation of phase-constrained parallel MRI as a nonlinear inverse problem and show how it is related to a linear inverse problem subject to a rank penalty. This method is robust and efficient, and allows artifact-free reconstruction in these cases.

This is poster number 14 in Poster Session

Authors:

H. Christian M. Holme (University Medical Center Göttingen)

Sebastian Rosenzweig (University Medical Center Göttingen)

Martin Uecker (University Medical Center Göttingen)

Keywords:

image reconstruction, inverse problems, magnetic resonance imaging, nonlinear optimization