Clinical High-Resolution 3D-MR Spectroscopic Imaging of the Human Brain at 7 T. Issue 4 (April 2020)

• Record Type: Journal Article
• Authors: Hingerl, Lukas; Strasser, Bernhard; Moser, Philipp; Hangel, Gilbert; Motyka, Stanislav; Heckova, Eva; Gruber, Stephan; Trattnig, Siegfried; Bogner, Wolfgang
• Is Part Of: Investigative radiology. Volume 55:Issue 4(2020)
• Publication Date: 2020-04
• Subjects: MR spectroscopic imaging -- concentric ring trajectory -- 7 Tesla -- 3D whole-cerebrum -- whole-brain -- neuroinflammation -- multiple sclerosis -- glioma -- glycine -- glutamine; Diagnosis, Radioscopic -- Periodicals; Radiology, Medical -- Periodicals; 616.0757
• DOI: 10.1097/RLI.0000000000000626
• Languages: English
• ISSNs: 0020-9996

Abstract

Abstract : Objectives: Available clinical magnetic resonance spectroscopic imaging (MRSI) sequences are hampered by long scan times, low spatial resolution, strong field inhomogeneities, limited volume coverage, and low signal-to-noise ratio. High-resolution, whole-brain mapping of more metabolites than just N -acetylaspartate, choline, and creatine within clinically attractive scan times is urgently needed for clinical applications. The aim is therefore to develop a free induction decay (FID) MRSI sequence with rapid concentric ring trajectory (CRT) encoding for 7 T and demonstrate its clinical feasibility for mapping the whole cerebrum of healthy volunteers and patients. Materials and Methods: Institutional review board approval and written informed consent were obtained. Time-efficient, 3-dimensional encoding of an ellipsoidal k-space by in-plane CRT and through-plane phase encoding was integrated into an FID-MRSI sequence. To reduce scan times further, repetition times were shortened, and variable temporal interleaves were applied. Measurements with different matrix sizes were performed to validate the CRT encoding in a resolution phantom. One multiple sclerosis patient, 1 glioma patient, and 6 healthy volunteers were prospectively measured. For the healthy volunteers, brain segmentation was performed to quantify median metabolic ratios, Cramér-Rao lower bounds (CRLBs), signal-to-noise ratios, linewidths, and brain coverage among all measured matrix sizes ranging from a 32 × 32 × 31 matrix with 6.9 × 6.9 × 4.2 mm 3 nominal voxel size acquired in ~3 minutes to an 80 × 80 × 47 matrix with 2.7 × 2.7 × 2.7 mm 3 nominal voxel size in ~15 minutes for different brain regions. Results: Phantom structures with diameters down to 3 to 4 mm were visible. In vivo MRSI provided high spectral quality (median signal-to-noise ratios, >6.3 and linewidths, <0.082 ppm) and fitting quality. Cramér-Rao lower bounds were ranging from less than 22% for glutamine (highest CRLB in subcortical gray matter) to less than 9.5% for N -acetylaspartate for the 80 × 80 × 47 matrix (highest CRLB in the temporal lobe). This enabled reliable mapping of up to 8 metabolites ( N -acetylaspartate, N -acetylaspartyl glutamate, total creatine, glutamine, glutamate, total choline, myo -inositol, glycine) and macromolecules for all resolutions. Coverage of the whole cerebrum allowed visualization of the full extent of diffuse and local multiple sclerosis-related neurochemical changes (eg, up to 100% increased myo -inositol). Three-dimensional brain tumor metabolic maps provided valuable information beyond that of single-slice MRSI, with up to 200% higher choline, up to 100% increased glutamine, and increased glycine in tumor tissue. Conclusions: Seven Tesla FID-MRSI with time-efficient CRT readouts offers clinically attractive acquisition protocols tailored either for speed or for the investigation of small pathologic details and low-abundant metabolites. This can complement clinical MR studies of various brain disorders. Significant metabolic anomalies were demonstrated in a multiple sclerosis and a glioma patient for myo -inositol, glutamine, total choline, glycine, and N -acetylaspartate concentrations. Abstract : Supplemental digital content is available in the text.