A denoising method for multidimensional magnetic resonance spectroscopy and imaging based on compressed sensing

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Nuclear Magnetic Resonance (NMR) Spectroscopy
• Signal Processing

Background:

• t1-noise is a common artifact in 2D Magnetic Resonance (MR) experiments, originating from various instabilities like temperature fluctuations, field drifts, and motion.
• This noise significantly impacts the quality of both MR spectroscopy and imaging data.
• Existing methods may not sufficiently address the pervasive nature of t1-noise across diverse MR applications.

Purpose of the Study:

• To introduce and validate a novel post-processing method for attenuating t1-noise in 2D MR data.
• To improve the signal-to-noise ratio (SNR) in various 2D MR experiments without compromising data integrity.
• To demonstrate the method's versatility across different MR techniques and sample types.

Main Methods:

• Developed Compressed Sensing Multiplicative (CoSeM) denoising, a post-processing technique for 2D MR data.
• CoSeM involves reconstructing data from undersampled k-space, iteratively masking and reconstructing data to generate multiple representations.
• Averaging selected representations based on a noise reduction criterion to achieve denoising.

Main Results:

• CoSeM processing demonstrated 2-3 fold increases in SNR across synthetic data, 2D solid and solution state NMR, 2D localized MRS of live brains, and 2D abdominal MRI.
• The method effectively reduced t1-noise without introducing biases, false peaks, or spectral/image blurring.
• Quantitative linearity was preserved, enabling reliable T1 inversion-recovery MRI mapping.

Conclusions:

• CoSeM denoising is a robust and effective post-processing method for reducing t1-noise in 2D MR experiments.
• The technique offers significant SNR improvements and maintains data accuracy, making it valuable for various MR applications.
• CoSeM enhances the quality and reliability of MR spectroscopy and imaging data.