In vivo NMR spectral parameter estimation: a comparison between time and frequency domain methods

Area of Science:

• Magnetic Resonance Spectroscopy
• Biophysical Chemistry
• Medical Imaging Analysis

Background:

• Accurate estimation of in vivo NMR spectral parameters is crucial for quantitative analysis.
• Various methods exist, but their performance and robustness under different conditions require thorough evaluation.

Purpose of the Study:

• To compare the accuracy and precision of different in vivo NMR spectral parameter estimation techniques.
• To assess the feasibility and robustness of nonlinear fitting of time-domain FID data.

Main Methods:

• Monte Carlo simulations of 31P and 13C in vivo NMR experiments.
• Comparison of nonlinear fitting in the time domain (NLTD), nonlinear fitting in the frequency domain (NLLM, NLFM), and linear prediction with singular value decomposition (LPSVD).
• Assessment of bias and statistical uncertainties for spectral parameters.

Main Results:

• For 31P, NLTD, NLLM, and NLFM showed comparable results, while LPSVD introduced significant bias in peak amplitudes.
• For 13C, only NLTD and NLFM successfully recovered the glycogen peak.
• Simulations indicated 256 data points are sufficient for both 31P and 13C.
• NLTD demonstrated feasibility and robustness on human 31P and 13C data.

Conclusions:

• Nonlinear fitting of time-domain FID data (NLTD) is a reliable method for in vivo NMR spectral parameter estimation.
• NLTD and NLFM are superior for detecting specific metabolites like glycogen in 13C spectroscopy.
• The study validates NLTD as a robust technique applicable to human in vivo NMR data.