Boosting quantification accuracy of chemical exchange saturation transfer MRI with a spatial-spectral...

Area of Science:

• Biomedical Imaging
• Magnetic Resonance Imaging
• Metabolomics

Background:

• Chemical Exchange Saturation Transfer (CEST) enables noninvasive detection of low-concentration endogenous metabolites in vivo.
• CEST imaging is challenged by a low signal-to-noise ratio (SNR) due to water signal reduction from spin transfer.
• This SNR limitation hinders accurate and reliable quantification in CEST imaging.

Purpose of the Study:

• To introduce BOOST (suBspace denoising with nOnlocal lOw-rank constraint and Spectral local-smooThness regularization), a novel spatial-spectral denoising method.
• To enhance the SNR of CEST images and improve quantification accuracy.
• To validate the method's efficiency and robustness across various conditions and magnetic field strengths.

Main Methods:

• Decomposition of noisy CEST images into a low-dimensional subspace using a global spectral low-rank prior.
• Application of a spatial nonlocal self-similarity prior to subspace-based images.
• Incorporation of spectral local-smoothness via a weighted spectral total variation constraint.

Main Results:

• BOOST demonstrated superior noise elimination compared to state-of-the-art algorithms.
• The method was validated in numerical simulations, preclinical, and clinical settings.
• Effectiveness was shown across a range of magnetic field strengths (3.0 to 11.7 T).

Conclusions:

• BOOST significantly enhances SNR and quantification accuracy in CEST imaging.
• It is a cost-effective and widely available post-processing technique.
• BOOST can be integrated into existing CEST protocols to improve the detection of subtle CEST effects.