Reproducibility of serial whole-brain MR spectroscopic imaging
A A Maudsley1, C Domenig, S Sheriff
1Department of Radiology, University of Miami School of Medicine, Miami, FL 33136, USA. AMaudsley@med.miami.edu
NMR in Biomedicine
|September 25, 2009
Summary
This study evaluated the reproducibility of volumetric proton Magnetic Resonance Spectroscopic Imaging (MRSI) for brain metabolite analysis. Results show reliable intra-subject reproducibility, comparable to single-voxel methods, enabling sensitive longitudinal studies.
Area of Science:
- Neuroimaging
- Biophysics
- Medical Physics
Background:
- Proton Magnetic Resonance Spectroscopic Imaging (MRSI) is crucial for non-invasively quantifying brain metabolites.
- Assessing the reproducibility of volumetric MRSI is essential for its clinical application in longitudinal studies.
- Lipid suppression and automated processing are key factors in improving MRSI data quality.
Purpose of the Study:
- To evaluate the reproducibility of serial measurements using volumetric proton MRSI at 3 Tesla with lipid suppression.
- To assess the reliability of metabolite quantification using automated processing and registration techniques.
- To compare the reproducibility of voxel-based and region-averaged analyses.
Main Methods:
- Volumetric proton MRSI data were acquired from two subjects over five time points at 3 Tesla with inversion-recovery lipid suppression.
- Fully-automated processing, including rigid registration, was applied to the acquired data.
- Coefficients of variance (COV) were calculated for individual voxels and atlas-defined brain regions for key metabolites (NAA, Cr, Cho) and their ratios.
Main Results:
- Volumetric MRSI provided data of sufficient quality for analysis over 70 +/- 6% of the brain volume.
- Median voxel-based COVs were 6.2% for N-acetylaspartate, 7.2% for creatine, and 9.7% for choline.
- Mean COVs after averaging over regional brain areas were significantly lower: 3.5% (NAA), 3.7% (Cr), and 5.2% (Cho).
Conclusions:
- Longitudinal volumetric MRSI studies with post-acquisition registration offer intra-subject reproducibility comparable to single-voxel MRS.
- The technique enables increased sensitivity through averaging over larger tissue volumes, supporting its use in clinical research.
- Spatial distribution of COV values correlates with known magnetic field inhomogeneity, validating the analysis approach.
