A fast MR fingerprinting simulator for direct error estimation and sequence optimization

Area of Science:

• Quantitative Magnetic Resonance Imaging (qMRI)
• Medical Imaging Physics
• Biomedical Engineering

Background:

• Magnetic Resonance Fingerprinting (MRF) is a powerful quantitative MRI technique for simultaneous tissue property mapping.
• Optimizing MRF sequences requires accurate modeling of undersampling and field imperfections, which is computationally intensive.
• Current optimization methods are often impractical due to high computational demands.

Purpose of the Study:

• To introduce a fast MRF simulator for efficient sequence optimization.
• To enable direct estimation of quantitative errors and improve robustness against artifacts.
• To develop optimized MRF sequences with reduced scan times and improved accuracy.

Main Methods:

• Development of a fast MRF simulator to model aliased images under realistic scan conditions (undersampling, system imperfections).
• Evaluation of the simulator's performance and speed via simulations and in vivo experiments.
• Application of the fast simulator within an MRF optimization framework for sequence design.

Main Results:

• The fast MRF simulator achieved a 158x reduction in processing time compared to conventional methods.
• Simulations closely matched in vivo MRF data.
• Optimized sequences produced artifact-free T1 and T2 maps with equivalent accuracy but shorter scan times.
• The simulator enabled feasible direct estimation of undersampling errors during optimization.

Conclusions:

• The fast MRF simulator significantly reduces computational cost for MRF optimization.
• Optimized MRF sequences are robust against undersampling and field inhomogeneity, improving image quality and efficiency.
• This approach facilitates the development of practical and high-performance quantitative MRI techniques.