Tensor-Based Method for Residual Water Suppression in ¹H Magnetic Resonance Spectroscopic Imaging.

Area of Science:

• Medical Imaging
• Spectroscopy
• Data Analysis

Background:

• Magnetic Resonance Spectroscopic Imaging (MRSI) signals are frequently contaminated by residual water and artifacts, hindering accurate metabolite quantification.
• Effective residual water suppression is crucial for reliable analysis of MRSI data.

Purpose of the Study:

• To introduce and evaluate a novel tensor-based method for suppressing residual water in MRSI signals.
• To enhance the accuracy and efficiency of metabolite quantification in MRSI.

Main Methods:

• A third-order tensor was constructed by stacking Löwner matrices of MRSI voxel spectra.
• Canonical polyadic decomposition was applied to extract and remove the water component from the MRSI signals.

Main Results:

• The proposed tensor-based method demonstrated effective water suppression on both simulated and in-vivo MRSI data.
• The method successfully suppressed residual water simultaneously across all voxels in the MRSI grid.

Conclusions:

• The tensor-based Löwner method offers superior performance in residual water suppression compared to the Hankel singular value decomposition (SVD) method.
• This technique prevents water suppression failures in individual voxels, improving overall MRSI data quality.