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, or phase poles, are a common issue in MRI reconstruction.
• These singularities arise from inherent ambiguities in estimating auto-calibrated coil sensitivities.
• They can degrade the quality and accuracy of reconstructed MR images.

Purpose of the Study:

• To develop and validate a method for detecting and correcting phase poles.
• To improve the accuracy of coil sensitivity maps in non-linear inverse (NLINV) reconstruction.
• To enhance the reliability of MR image reconstruction from limited auto-calibration data.

Main Methods:

• Phase poles are detected by computing the curl in individual coil sensitivity maps.
• A weighted average of the curl is used for robust phase pole detection.
• Detection and correction are integrated into the NLINV algorithm's Gauss-Newton method.
• The method is also shown to correct phase poles in ESPIRiT coil sensitivity maps.

Main Results:

• The developed method reliably removes phase poles, yielding singularity-free coil sensitivity profiles.
• Accurate coil sensitivity estimation is achieved even with very small auto-calibration regions (e.g., 7x7 pixels).
• The technique was successfully evaluated on accelerated Cartesian MPRAGE brain imaging and real-time cardiac MRI.

Conclusions:

• The proposed method effectively detects and corrects phase poles in MRI reconstructions.
• This leads to improved coil sensitivity maps and more reliable MR image reconstruction.
• The approach enhances the utility of auto-calibration techniques in MRI.