Efficient high-resolution RF pulse design applied to simultaneous multi-slice excitation.

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Optimal Control Theory
• Pulse Sequence Design

Background:

• Traditional radiofrequency (RF) pulse design methods rely on gradient and quasi-Newton approaches.
• These conventional methods often exhibit slow convergence, limiting their efficiency in complex pulse design scenarios.

Purpose of the Study:

• To develop a more efficient and faster RF pulse design method.
• To improve the convergence rate of optimal control-based RF pulse design.

Main Methods:

• Implementation of a globally convergent trust-region conjugate gradient-Newton (CG-Newton) method.
• Incorporation of exact second-order information into the optimization process.
• Application to designing RF pulses for single-slice and simultaneous multi-slice (SMS) excitation.

Main Results:

• The proposed trust-region CG-Newton method demonstrates an improved convergence rate compared to traditional approaches.
• Successful application in designing RF pulses for both single-slice and SMS excitation.
• Validation through phantom and in vivo experiments on a 3T MRI scanner.

Conclusions:

• The novel optimal control method offers a flexible and highly efficient approach for RF pulse design.
• This method significantly enhances convergence speed, making it suitable for advanced MRI applications like SMS excitation.
• Experimental validation confirms the practical utility and effectiveness of the developed technique.