Hybrid data fidelity term approach for quantitative susceptibility mapping.

Area of Science:

• Medical Imaging
• Geophysics
• Computational Science

Background:

• Quantitative Susceptibility Mapping (QSM) is crucial for analyzing magnetic field variations.
• Traditional QSM methods often use L2-norm, which struggles with noisy data and artifacts.
• L1-norm offers robustness but can degrade performance in high signal-to-noise ratio (SNR) regions.

Purpose of the Study:

• To introduce a novel hybrid data fidelity (HD-QSM) approach for QSM.
• To leverage the strengths of both L1 and L2 norms for improved susceptibility map reconstruction.
• To enhance accuracy, reduce artifacts, and increase computational efficiency in QSM.

Main Methods:

• Developed a two-stage method using L1-norm followed by L2-norm initialization.
• Employed linear susceptibility inversion with total variation regularization.
• Utilized the Alternating Direction Method of Multipliers (ADMM) for functional optimization.
• Incorporated spatially variable weights to enhance reconstruction quality.

Main Results:

• HD-QSM achieved high-quality quantitative reconstructions with excellent structural definition.
• The method demonstrated significant noise reduction and effective suppression of streaking artifacts.
• Achieved first place in the lowest RMS error category at the 2019 QSM Reconstruction Challenge.
• Demonstrated comparable results to nonlinear methods with superior computational efficiency.

Conclusions:

• The proposed HD-QSM method provides robust and accurate QSM reconstructions.
• It outperforms current state-of-the-art methods in terms of accuracy and efficiency.
• HD-QSM offers a promising advancement for quantitative susceptibility mapping applications.