Real-time 2D spatially selective MRI experiments: Comparative analysis of optimal control design methods

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Radio-Frequency (RF) Pulse Design
• Optimal Control (OC) Algorithms

Background:

• Modern MRI systems require advanced RF pulse techniques due to ultra-high magnetic fields and new coil designs.
• 2D spatially selective RF pulses are vital for clinical applications like reduced field of view imaging and MR spectroscopy.
• Numerical optimization methods are commonly used for generating these complex RF pulses.

Purpose of the Study:

• To demonstrate the efficiency of numerical optimal control (OC) algorithms for designing 2D spatially selective MRI experiments.
• To evaluate the robustness of OC algorithms against challenges like magnetic field inhomogeneity.
• To identify the most suitable OC algorithms for practical MRI applications.

Main Methods:

• Comparison of three popular OC algorithms: two gradient-based (first- and second-order) and one sequential, monotonic method.
• Application of algorithms to two experimental models: a water phantom and an in vivo human head.
• Implementation of all algorithms in MATLAB, ensuring accessibility for the MRI research community.

Main Results:

• The study evaluated the performance of different OC algorithms in designing 2D spatially selective RF pulses.
• Experimental validation was performed on both phantom and human head data.
• Analysis focused on computational speed, experimental robustness, and resulting image quality.

Conclusions:

• The sequential, monotonic OC approach and the second-order gradient-based OC approach are recommended for designing robust 2D spatially selective RF pulses in MRI.
• These methods offer a favorable balance of computational speed, experimental robustness, and image quality.
• The freely available MATLAB implementations facilitate adoption by the MRI community.