Nonrigid motion correction in 3D using autofocusing with localized linear translations.

Area of Science:

• Medical Imaging
• Biophysics

Background:

• Magnetic Resonance (MR) scans are susceptible to motion artifacts due to long scan times.
• Existing methods for nonrigid motion correction are computationally intensive and challenging for real-time reconstruction.

Purpose of the Study:

• To develop a computationally efficient autofocusing algorithm for reducing motion artifacts in MRI.
• To enable practical online reconstruction of MR images affected by complex body motion.

Main Methods:

• Approximating complex motions as linear translations on a small spatial scale.
• Implementing a localized gradient-entropy metric for motion correction.
• Utilizing novel "Butterfly" navigators for intrinsic translational motion measurement.
• Applying the correction scheme to free-breathing abdominal patient studies.

Main Results:

• Demonstrated reduction in artifacts caused by complex, nonrigid motion in abdominal MR scans.
• Successful approximation of linear motion paths for image voxels using multichannel navigator data.
• Validation of the practical autofocusing algorithm in real-world patient studies.

Conclusions:

• The proposed localized autofocusing method offers an effective and practical solution for motion artifact reduction in MRI.
• The use of "Butterfly" navigators and linear motion approximation facilitates efficient online reconstruction.
• This technique shows promise for improving the quality of abdominal MR imaging in free-breathing conditions.