Rapid three-dimensional multiparametric MRI with quantitative transient-state imaging.

Area of Science:

• Magnetic Resonance Imaging
• Quantitative Imaging
• Biomedical Engineering

Background:

• Quantitative imaging is crucial for disease assessment and treatment monitoring.
• Non-Cartesian imaging offers advantages but requires efficient reconstruction and parameter estimation.
• Accurate multiparametric quantification is essential for robust clinical applications.

Purpose of the Study:

• To develop and validate novel methods for quantitative, transient-state multiparametric MRI.
• To assess common Non-Cartesian readout trajectories (radials and spirals) for efficient anti-aliasing.
• To propose neural network-based parameter inference incorporating proton density estimation.

Main Methods:

• Utilized 2D/3D radial and spiral Non-Cartesian trajectories.
• Implemented k-space view-sharing for efficient anti-aliasing.
• Developed neural networks for parameter inference, including proton density estimation.
• Validated results against gold standards and phantom references at 1.5T and 3T.

Main Results:

• Achieved good agreement with gold standard and phantom references across readout trajectories.
• Neural network parameter inference showed <6.58% difference compared to high-resolution dictionary methods.
• Concordance correlation coefficients >0.92 and normalized root mean squared error between 4.2-12.7% for T1 and T2.
• Demonstrated sub-millimetric isotropic resolution in vivo in under 5 minutes with <7 min reconstruction/inference times.

Conclusions:

• The 3D quantitative transient-state imaging approach enables high-resolution multiparametric tissue quantification.
• Clinically acceptable acquisition and reconstruction times are achievable.
• This method holds potential for improved disease assessment and treatment monitoring.