Optimal Control-Based Generic Framework for Radiofrequency Pulse Design in MRI.

Area of Science:

• Medical Imaging
• Biophysics
• Computational Science

Background:

• Optimal design of radiofrequency (RF) pulses is crucial for Magnetic Resonance Imaging (MRI) performance.
• Existing methods often face limitations in incorporating hard constraints and minimizing pulse amplitude.
• Need for flexible and efficient tools for designing advanced MRI RF pulses.

Purpose of the Study:

• To propose a general framework for the optimal design of RF pulses in MRI using optimal control theory.
• To develop a Python-based numerical implementation for practical RF pulse design.
• To enhance pulse design by incorporating hard constraints, using wavelet coefficients, and minimizing peak amplitude.

Main Methods:

• Utilized optimal control theory and a state-of-the-art nonlinear optimization solver (IPOPT).
• Developed a Python package for RF pulse design, incorporating hard constraints and wavelet-based pulse representation.
• Implemented an innovative approach for minimizing pulse peak amplitude.

Main Results:

• The framework effectively incorporates hard constraints into the RF pulse optimization process.
• Wavelet coefficient representation proved highly effective for pulse design.
• Optimized pulses demonstrated significant improvements in the energy/peak power trade-off and overall performance compared to existing solutions.

Conclusions:

• The proposed framework offers a flexible and efficient solution for optimal RF pulse design in MRI.
• The method successfully addresses short T2 selective excitation and B1-robust problems with reduced power and energy.
• The open-source Python package enables external users to solve diverse RF pulse design challenges.