Accelerated whole-brain multi-parameter mapping using blind compressed sensing.

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Medical Physics
• Biomedical Engineering

Background:

• Accelerating MRI acquisition is crucial for efficient patient scanning and reducing motion artifacts.
• Current compressed sensing (CS) methods face limitations in handling complex signal variations and motion.
• Multiparameter mapping provides comprehensive tissue characterization but is often time-consuming.

Purpose of the Study:

• To introduce and validate a novel Blind Compressed Sensing (BCS) framework.
• To demonstrate BCS feasibility for high-resolution, whole-brain T1ρ and T2 mapping.
• To assess BCS performance at high acceleration factors (R).

Main Methods:

• BCS models magnetization evolution using a jointly estimated dictionary and sparse coefficients from undersampled data.
• Utilizes a large set of non-orthogonal bases to capture complex signal characteristics.
• Employs sparse coefficients for reduced degrees of freedom, minimizing artifacts at high R.

Main Results:

• Achieved mean square errors within 0.1% for T1ρ and T2 maps up to R=10 in retrospective undersampling.
• BCS demonstrated superior robustness to patient-specific motion compared to other CS techniques.
• Prospective 3D imaging suggested an acceleration factor of 8 with reasonable reconstruction quality.

Conclusions:

• BCS significantly reduces scan time for whole-brain multiparameter mapping.
• The framework yields minimal artifacts and enhanced robustness to motion-induced signal changes.
• BCS outperforms existing CS and principal component analysis-based techniques for accelerated MRI mapping.