Navigator-Free EPI Ghost Correction With Structured Low-Rank Matrix Models: New Theory and Methods

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Medical Imaging
• Image Reconstruction

Background:

• Structured low-rank matrix models aid calibrationless MRI reconstruction.
• These models have been extended for navigator-free echo-planar imaging (EPI) ghost correction.

Purpose of the Study:

• To analyze theoretical limitations of existing low-rank models for EPI ghost correction.
• To propose and evaluate improved formulations for navigator-free EPI using parallel imaging constraints.

Main Methods:

• Theoretical analysis of structured low-rank matrix optimization for EPI data.
• Incorporation of image-domain or k-space domain parallel imaging side information.
• Investigation of nonconvex low-rank matrix regularization.

Main Results:

• Identified inherent non-uniqueness in uniform subsampled EPI low-rank problems without constraints.
• Demonstrated effective elimination of ghost artifacts in phantom and in vivo data.
• Achieved superior performance compared to state-of-the-art methods across various acquisition schemes.

Conclusions:

• Navigator-free EPI ghost correction requires additional constraints beyond basic low-rank modeling.
• Integrating parallel imaging methods significantly enhances EPI ghost artifact removal.
• Proposed methods show robust performance in single-channel and accelerated multi-channel acquisitions.