Accelerated ¹ H MRSI using randomly undersampled spiral-based k-space trajectories.

Area of Science:

• Magnetic Resonance Imaging
• Spectroscopic Imaging
• Medical Physics

Background:

• Accelerated Magnetic Resonance Spectroscopic Imaging (MRSI) is crucial for reducing scan times.
• Undersampling spiral trajectories is a promising approach for acceleration.
• Developing robust reconstruction methods is key to maintaining data quality.

Purpose of the Study:

• To develop and evaluate a novel acquisition and reconstruction method for accelerated MRSI.
• To assess the performance of random SENSE+TV using undersampled spiral trajectories.
• To compare the proposed method against existing techniques for MRSI acceleration.

Main Methods:

• Developed a random undersampling spiral acquisition combined with sensitivity encoding (SENSE) and total variation (TV) regularization (random SENSE+TV).
• Evaluated the method on numerical phantoms and in vivo single-slice and 3D MRSI at 3 Tesla.
• Compared random SENSE+TV with five alternative accelerated MRSI methods.

Main Results:

• Random SENSE+TV achieved significant reductions in root-mean-square error (RMSE) for metabolite maps (NAA, creatine, choline) compared to other methods.
• Demonstrated up to 2.7x RMSE reduction in single-slice in vivo MRSI and 1.6x in 3D in vivo MRSI.
• Achieved acceleration factors of 4.5 for single-slice and 4 for 3D MRSI with comparable quality to fully sampled data.

Conclusions:

• Random SENSE+TV offers superior performance in reducing RMSE of metabolite maps compared to evaluated methods at the same scan time.
• The method enables substantial acceleration (up to 4.5-fold) while maintaining data quality comparable to fully sampled MRSI.
• Random SENSE+TV represents a significant advancement for accelerated MRSI acquisition and reconstruction.