Regular Article
Optimization of Residual Water Signal Removal by HLSVD on Simulated Short Echo Time Proton MR Spectra of the Human Brain

https://doi.org/10.1006/jmre.2001.2318Get rights and content

Abstract

Suppression of the residual water signal from proton magnetic resonance (MR) spectra recorded in human brain is a prerequisite to an accurate quantification of cerebral metabolites. Several postacquisition methods of residual water signal suppression have been reported but none of them provide a complete elimination of the residual water signal, thereby preventing reliable quantification of brain metabolites. In the present study, the elimination of the residual water signal by the Hankel Lanczos singular value decomposition method has been evaluated and optimized to provide fast automated processing of spectra. Model free induction decays, reproducing the proton signal acquired in human brain localized MR spectroscopy at short echo times (e.g., 20 ms), have been generated. The optimal parameters in terms of number of components and dimension of the Hankel data matrix allowing complete elimination of the residual water signal are reported.

Cited by (84)

  • Quantitative T<inf>1</inf>-relaxation corrected metabolite mapping of 12 metabolites in the human brain at 9.4 T

    2022, NeuroImage
    Citation Excerpt :

    1H MRSI data were reconstructed using a spatial Hanning filter before the 2D FFT, then eddy current corrected which also corrects for zeroth-order phase (Klose, 1990), and coil combined using the singular value decomposition (SVD) method (Bydder et al., 2008). Following reconstruction, data post-processing included removing residual water using the Hankel-Lanczos (HLSVD) (Cabanes et al., 2001) method using 10 decaying sinusoids in the range of 4.4–5 ppm. Data were then corrected for 1st-order phase distortion by means of linear back prediction of the missing FID points (Kay, 1988).

View all citing articles on Scopus
View full text