Fast 3D-MRSI using sparse acquisition and 4D compressed sensing reconstruction.

Area of Science:

• Medical Imaging
• Metabolic Research
• Spectroscopy

Background:

• Magnetic Resonance Spectroscopic Imaging (MRSI) is crucial for metabolic research but traditional methods are too slow for hyperpolarized (HP) agents.
• Rapid signal decay and non-renewable magnetization of HP agents like [1-¹³C]pyruvate limit traditional MRSI applications.
• Existing fast MRSI techniques often compromise spectral bandwidth and are prone to aliasing.

Purpose of the Study:

• To develop and validate an accelerated MRSI method for HP applications.
• To preserve broad spectral bandwidth and weak metabolite detectability without aliasing.
• To overcome limitations of techniques like echo-planar spectroscopic imaging (EPSI).

Main Methods:

• Implementation of a sparsely sampled 3D-MRSI pulse sequence with high reduction ratios.
• Development of a 4D compressed sensing (CS) reconstruction algorithm for undersampled data.
• In vivo experiments using hyperpolarized [1-¹³C]pyruvate in rat kidneys on a 3T MRI scanner.

Main Results:

• High-quality MRSI reconstructions achieved at acceleration factors up to R=32 (3.125% sampling).
• Low reconstruction errors (nRMSE < 4×10^-3) and high similarity (SSIM > 0.95) even at high undersampling.
• Successful resolution of lactate, alanine, pyruvate, and bicarbonate distributions in rat kidneys with high fidelity.

Conclusions:

• Sparse MRSI acquisition combined with 4D-CS reconstruction enables rapid, high-fidelity HP ¹³C-MRSI.
• Acquisition time is reduced up to 32-fold, enhancing dynamic metabolic studies.
• The method improves feasibility for preclinical and future clinical applications.