Asymmetric susceptibility tensor imaging.

Area of Science:

• Medical Imaging
• Neuroimaging
• Biophysics

Background:

• Susceptibility Tensor Imaging (STI) is a powerful MRI technique for visualizing microscopic tissue structures.
• Current STI methods often assume a symmetric susceptibility tensor, which may not fully represent the underlying physics.
• Investigating the impact of relaxing this symmetry constraint is crucial for improving STI accuracy.

Purpose of the Study:

• To explore the effects of a symmetry constraint on susceptibility tensor imaging.
• To derive the relationship between MRI frequency shift and the magnetic susceptibility tensor without assuming symmetry.
• To evaluate the performance of asymmetric tensor reconstruction in STI.

Main Methods:

• Developed a theoretical framework for asymmetric susceptibility tensor reconstruction.
• Acquired gradient echo phase data from postmortem mouse brain and kidney samples.
• Applied both symmetric and asymmetric tensor reconstructions, followed by fiber tracking.
• Conducted simulations with varying noise levels to assess reconstruction accuracy.

Main Results:

• Asymmetric reconstruction demonstrated reduced noise and streaking artifacts compared to symmetric methods.
• Improved image contrast and more complete fiber tracking were observed with asymmetric reconstruction.
• Simulations showed better mean squared error and angular difference for asymmetric reconstruction under noisy conditions.
• Decomposition of the asymmetric tensor confirmed its underlying symmetry, with asymmetry attributed to noise and artifacts.

Conclusions:

• While the physical susceptibility tensor is symmetric, asymmetric reconstruction effectively suppresses noise and artifacts.
• This leads to more accurate estimation of the susceptibility tensor in susceptibility tensor imaging.
• Asymmetric reconstruction offers a promising approach for enhancing the quality and reliability of STI.