Item

Abstract

Purpose/Introduction: In SENSE-accelerated [1] Magnetic Resonance Spectroscopic Imaging (MRSI), lipid signal originating in the skull region is often found in the center of the brain due to voxel bleeding and/or residual spatial aliasing. By direct optimization of the Spatial Response Function (SRF) on an overdiscrete spatial grid [2,3] a better spatial specificity and higher SENSEacceleration can be achieved. In this work, we introduce anatomical spatial prior knowledge to further improve the suppression of residual lipid artifacts. Subjects and Methods: Using Cartesian k-space sampling with fourfold SENSE acceleration (R=2x2 in APxRL), a FIDLOVS [4] MRSI data set of a transversal brain slice of a healthy volunteer was acquired at 7T with a 32 channel head coil. A 200x160mm2 FOV with voxel size 10x10x10mm3 was chosen, and eddy current correction was perfomed using a non water-suppressed reference. The MRSI reconstruction operator F is calculated by minimizing where SRF optimization is regularized by the noise level for every pixel π [7]. E is the encoding matrix and T the Gaussian SRF target functions, both with a ζ2=9-fold overdiscretization in real space [2]. The new diagonal matrix A contains spatial weights and introduces a prioritization of SRF optimality across the FOV depending on the coarse tissue types f (fat), b (brain) and o (outside the object) (Fig.1). Results: With higher priority of SRF optimality on fat than on brain, a drop of lipid signal intensity in representative spectra is observed (Fig.2B). The effective spatial resolution decreases only slightly for moderate weighting factors f/b<10 (Table 1) and stays well below the value of 1.85 achieved with conventional Hamming filtering. In the brain region, a decrease of lipid signal can be achieved with spatial prior knowledge (f/b =10, Fig.3). Discussion/Conclusion: Adapted target-driven, overdiscretized reconstruction largely reduces lipid artifacts in accelerated FIDLOVS MRSI through direct optimization of the SRF [2]. We demonstrate that a further suppression of any residual lipid artifacts in 1H brain spectra can be achieved by assigning a moderately elevated priority to SRF optimality in the region of subcranial lipids of 1H MRSI.