Joint reconstruction of image and motion in MRI: implicit regularization using an adaptive 3D mesh

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Computational Science

Background:

• Magnetic resonance imaging (MRI) is susceptible to motion artifacts from physiological processes like breathing and cardiac beating.
• Existing joint image and motion reconstruction methods often require regularization, such as Tikhonov's method, to address ill-posed optimization problems.
• Tikhonov regularization's effectiveness can be limited by the need for careful selection of the regularization parameter, especially in complex 3D scenarios.

Purpose of the Study:

• To develop an adaptive, implicit regularization method for motion artifact correction in MRI.
• To introduce subject-specific, spatially varying smoothness constraints for motion modeling.
• To improve the accuracy and convergence rate of joint image and motion reconstruction.

Main Methods:

• Proposing an adaptive implicit regularization technique based on solving for motion at key points forming a mesh.
• Developing a practical algorithm for automatic mesh generation.
• Comparing the proposed method against the Tikhonov method using both simulated (in silico) and real (in vivo) data.

Main Results:

• The proposed adaptive method demonstrated a superior convergence rate compared to the Tikhonov method.
• Both simulated and in vivo experiments showed improved accuracy in reconstructed images.
• Enhanced precision was observed in the reconstruction of motion parameters.

Conclusions:

• The adaptive implicit regularization method offers a more robust and accurate approach to correcting motion artifacts in MRI.
• Subject-specific, spatially varying smoothness constraints enhance the reliability of motion modeling.
• This technique represents a significant advancement in improving the quality of MRI data acquisition and analysis.