Self-Navigated, Retrospective, Data-Consistent Motion Correction for MPnRAGE

Area of Science:

• Medical Imaging
• Neuroimaging
• Biomedical Engineering

Background:

• Motion artifacts are a significant challenge in 3D radial T1-weighted MRI, potentially compromising image quality and diagnostic accuracy.
• Existing motion correction methods may lack automation or struggle with diverse motion types and severities.

Purpose of the Study:

• To extend and automate a data-consistent, self-navigated motion-correction method for 3D radial T1-weighted imaging.
• To evaluate the method's effectiveness across various motion scenarios and its impact on test-retest reliability of neuroimaging measures.

Main Methods:

• Incorporated rigid-body motion into the forward model, solving for motion parameters to maximize data consistency.
• Tested the automated method on diverse datasets, assessing image quality improvements and effects on cortical thickness reliability.
• Utilized error-based weighting and fine-scale timing resolution for enhanced motion correction.

Main Results:

• Achieved significant image quality improvements across a wide range of motion types, salvaging previously unusable scans.
• Demonstrated substantial enhancement in test-retest reliability for cortical thickness measures in pediatric subjects.
• Reduced average coefficient of variation for cortical thickness from 2.73% to as low as 0.79% with the correction method.

Conclusions:

• The automated motion correction method is effective for T1-weighted radial data, offering both qualitative and quantitative benefits.
• The technique's fine-scale timing resolution and error-based weighting are particularly advantageous for motion-prone populations or studies requiring high sensitivity to small effect sizes.