Accelerating functional MRI using fixed-rank approximations and radial-cartesian sampling.

Area of Science:

• Magnetic Resonance Imaging
• Neuroimaging
• Biomedical Engineering

Background:

• Functional magnetic resonance imaging (fMRI) data acquisition is often limited by speed.
• Existing acceleration techniques like k-t FASTER aim to improve acquisition speed but can be further optimized.
• Rank-constrained acceleration methods offer a promising avenue for enhancing fMRI acquisition efficiency.

Purpose of the Study:

• To introduce an improved k-t FASTER technique for accelerated fMRI data acquisition.
• To demonstrate the benefits of a hybrid 3D radial-Cartesian sampling approach.
• To incorporate coil sensitivity encoding for enhanced reconstruction fidelity.

Main Methods:

• Implemented a hybrid 3D radial-Cartesian sampling strategy for fMRI.
• Utilized multicoil rank-constrained reconstruction with hard thresholding and shrinkage of singular values.
• Employed sensitivity encoding and nonuniform Fast Fourier Transform for data consistency in the k-t domain.
• Enabled variable acceleration factors using a golden ratio-based radial increment.

Main Results:

• Achieved up to R=12.5 acceleration in visual-motor task fMRI data.
• Demonstrated reduced data matrix reconstruction errors compared to the original k-t FASTER.
• Showcased the ability to tailor acceleration factors for different signal components recovery.
• Validated performance on both retrospective and prospective undersampled data.

Conclusions:

• The enhanced k-t FASTER technique significantly accelerates fMRI data acquisition.
• Minimal penalty to data quality is observed even at higher acceleration factors.
• This method offers a valuable tool for improving fMRI scan efficiency and data quality.