Thomas Kirchner1, Lukas Eisenring1, Klaas P. Pruessmann1, Anke Henning1, 2
1Institute for Biomedical Engineering, University and ETH Zurich, Zurich, Switzerland; 2Max Planck Institute for Biological Cybernetics, Tuebingen, Germany
As a consequence of voxel bleeding and residual spatial aliasing in SENSE-accelerated in vivo 1H MRSI, lipid signal originating in the skull region is often found in spectroscopic imaging voxels in the brain. To a certain extent, those artifacts can be contained by target-driven overdiscretized reconstruction with direct optimization of the Spatial Response Function (SRF) compared to conventional SENSE with Hamming filtering. In this work, we demonstrate that an even better spectral quality can be achieved by exploiting spatial prior knowledge to specify spatially selective priority for SRF optimality.