Model-based frequency-and-phase correction of ¹H MRS data with 2D linear-combination modeling.

Area of Science:

• Magnetic Resonance Imaging
• Spectroscopy
• Biomedical Engineering

Background:

• Frequency-and-phase correction (FPC) is crucial for improving MR spectroscopy data quality by mitigating variations between transients.
• Traditional FPC methods, such as spectral registration, face limitations, especially at low signal-to-noise ratios (SNR).

Purpose of the Study:

• To introduce and evaluate a novel model-based FPC method that directly integrates FPC into a 2D linear-combination model (2D-LCM).
• To compare the performance of this 2D-LCM based FPC against traditional spectral registration followed by 1D-LCM for estimating frequency-phase drifts and metabolite levels.

Main Methods:

• Development of a synthetic dataset simulating in-vivo MR spectroscopy data with controlled noise and frequency-phase variations.
• Comparison of 2D-LCM with traditional spectral registration and 1D-LCM using metrics like frequency/phase/amplitude errors and Cramér Rao lower bounds (CRLBs).
• Validation of the proposed method on publicly available in-vivo MR spectroscopy data.

Main Results:

• The 2D-LCM method accurately estimates and corrects frequency-and-phase variations directly from uncorrected data.
• Metabolite amplitude estimates derived from 2D-LCM were as accurate, precise, and certain as those from conventional methods.
• The 2D-LCM approach demonstrated superior performance in low-to-very-low SNR conditions.

Conclusions:

• Model-based FPC integrated with 2D-LCM is a feasible technique with significant potential for enhancing metabolite level estimation in MR spectroscopy.
• This method is especially beneficial for low-SNR scenarios, such as those encountered with long echo times (TEs) or strong diffusion weighting.