Quantitative MRI by nonlinear inversion of the Bloch equations

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Biomedical Engineering
• Computational Imaging

Background:

• Quantitative MRI aims to derive tissue-specific parameters directly from imaging data.
• Model-based reconstruction offers a powerful approach but often requires sequence-specific implementations.
• Existing methods can be computationally intensive and lack generalizability across different pulse sequences.

Purpose of the Study:

• To develop a generic, model-based reconstruction framework for multiparametric quantitative MRI.
• To enable the framework's application with data from diverse pulse sequences without sequence-specific calibration.
• To enhance the efficiency and accuracy of quantitative parameter estimation in MRI.

Main Methods:

• Developed a generic nonlinear model-based reconstruction framework using numerical optimization.
• Combined direct sensitivity analysis and pre-computed state-transition matrices for efficient Bloch equation solving.
• Implemented and validated the framework for quantitative T1 and T2 mapping using inversion-recovery FLASH and bSSFP sequences.

Main Results:

• Direct sensitivity analysis provided accurate and stable derivative calculations.
• State-transition matrices accelerated computations by a factor of 10 compared to standard ODE solvers, maintaining accuracy.
• The framework reproduced known quantitative results for radial IR FLASH and generated accurate T1/T2 maps for the NIST phantom with IR bSSFP.
• Demonstrated feasibility in human brain imaging, noting potential influence from magnetization transfer effects.

Conclusions:

• The developed framework enables generic model-based reconstruction for quantitative MRI.
• Efficient numerical optimization tools utilizing Bloch equations are key to this advancement.
• This approach enhances the versatility and applicability of quantitative MRI techniques.