Phase-Pole-Free Images and Smooth Coil Sensitivity Maps by Regularized Nonlinear Inversion

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Image Reconstruction
• Signal Processing

Background:

• Phase singularities are a common challenge in MRI reconstruction, particularly with auto-calibrated sensitivities, stemming from inherent estimation ambiguities.
• These singularities can degrade the quality of reconstructed MR images and coil sensitivity maps.

Purpose of the Study:

• To develop and validate a method for detecting and correcting phase poles within non-linear inverse (NLINV) reconstruction of MR images and coil sensitivity maps.
• To address the issue of phase singularities in MRI data processing.

Main Methods:

• Phase poles are identified in individual coil sensitivity maps by calculating the curl at each pixel.
• A weighted average of the curl across coils is used for phase pole detection.
• Phase pole detection and correction are integrated into the iteratively regularized Gauss-Newton method within the NLINV algorithm.

Main Results:

• The developed method reliably removes phase poles in NLINV reconstructions for both accelerated Cartesian MPRAGE brain imaging and real-time radial MRI of the heart.
• NLINV with phase pole correction efficiently estimates coil sensitivity profiles without singularities, even from small auto-calibration regions (7x7).

Conclusions:

• The integrated phase pole correction method enhances NLINV reconstruction, ensuring singularity-free coil sensitivity profiles.
• NLINV is demonstrated as an efficient and reliable tool for image reconstruction and coil sensitivity estimation in demanding MRI applications.