Robust phase unwrapping for Dixon magnetic resonance imaging multiresolution pixel clustering and local surface...

Area of Science:

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

Background:

• Water-fat separation is vital for clinical MRI.
• The Dixon technique faces challenges like water-fat swapping due to magnetic field inhomogeneity and low SNR.
• Existing methods struggle with robustness in complex scenarios.

Purpose of the Study:

• To develop a robust phase-unwrapping method for improved water-fat separation in two-point Dixon MRI.
• Introduce the multiresolution pixel clustering and local surface fitting (MrCLOSE) algorithm.
• Enhance the reliability of water-fat separation in clinical MRI.

Main Methods:

• Constructing a multiresolution pyramid from in-phase and out-of-phase complex MRI data.
• Performing hierarchical phase unwrapping using pixel clustering and local surface fitting (CLOSE) within a coarse-to-fine framework.
• Evaluating MrCLOSE against B0-NICEbd, ROMEO, Graph Cut, and original CLOSE using simulations and in vivo datasets.

Main Results:

• MrCLOSE demonstrated robust phase unwrapping across various SNRs, disconnected regions, and phase changes (up to 10π).
• In simulations, MrCLOSE significantly reduced error ratios compared to CLOSE, especially at higher thresholds (0.67% vs. 43.71%).
• In vivo studies showed a substantially lower water-fat swap ratio for MrCLOSE (3.38%) compared to other methods (e.g., ROMEO 96.34%, Graph Cut 67.61%).

Conclusions:

• The integration of multiresolution strategies into CLOSE (MrCLOSE) enhances robustness and accuracy in phase unwrapping.
• MrCLOSE offers a promising solution for phase-related MRI applications, particularly for clinical water-fat separation.
• This method addresses key limitations of current Dixon MRI techniques.