Multi-echo dipole inversion for magnetic susceptibility mapping

Area of Science:

• Medical Imaging
• Biophysics
• Computational Science

Background:

• Quantitative Susceptibility Mapping (QSM) reconstructs tissue magnetic susceptibility from MRI phase data.
• Standard QSM involves solving multiple ill-posed inverse problems, including background field removal and field-to-source inversion.
• Multi-echo MRI acquisitions complicate QSM due to errors introduced during intermediate field estimation.

Purpose of the Study:

• To develop a novel multi-echo field-to-source forward model for QSM that bypasses the need for separate field estimation.
• To introduce a general underestimation correction step to recover regularized susceptibility sources.
• To improve the accuracy and reduce artifacts in QSM reconstruction from multi-echo MRI data.

Main Methods:

• A multi-echo field-to-source forward model was proposed, directly processing multi-echo phase data.
• A general underestimation correction step was developed for susceptibility source recovery.
• The proposed methods were validated on the QSM Challenge 2.0 datasets and in vivo human brains.

Main Results:

• The multi-echo forward model and correction step achieved state-of-the-art results on the QSM Challenge 2.0 datasets across all metrics.
• Experiments in vivo demonstrated agreement between the multi-echo model and standard single field-to-source models.
• The proposed approach showed compatibility with various dipole inversion methods.

Conclusions:

• The novel multi-echo forward model eliminates the need for multi-echo phase data fitting, achieving top performance on QSM Challenge 2.0.
• The correction step enables partial recovery of underestimated low-frequency susceptibility distributions.
• This method offers a robust and versatile advancement for QSM reconstruction.