Analytical performance bounds for multi-tensor diffusion-MRI

Area of Science:

• Neuroimaging
• Diffusion MRI
• Computational anatomy

Background:

• Accurate estimation of white matter (WM) fiber orientation is crucial for understanding brain structure and function.
• Crossing fibers in WM present a significant challenge for traditional diffusion MRI analysis.
• The Multi-Tensor Model (MTM) offers a more robust approach to modeling complex WM architecture.

Purpose of the Study:

• To investigate the impact of MR acquisition parameters on estimating WM fiber orientation and clinical parameters within crossing fiber regions using the MTM.
• To develop and apply a methodology for optimizing these acquisition parameters for improved estimation precision.

Main Methods:

• Computation of the Cramér-Rao Bound (CRB) for the MTM and key clinical parameters like Fractional Anisotropy (FA).
• Development of an approximate closed-form formula for the Fisher Information Matrix for multi-coil, multi-shell diffusion MRI acquisitions.
• Generalization of FA and mean diffusivity concepts to the multi-tensor model.

Main Results:

• Demonstrated that CRB can guide scan time reduction while maintaining high estimation precision.
• Showcased how increasing the number of acquisition coils can compensate for fewer diffusion gradient directions.
• Analyzed the influence of b-value and Signal-to-Noise Ratio (SNR), revealing quadratic error variance reduction with SNR and non-unique optimal b-values.

Conclusions:

• Emphasized the critical importance of selecting appropriate MR acquisition parameters, particularly for analyzing crossing fiber areas.
• Presented a CRB-based methodology for the optimal tuning of acquisition parameters to enhance the reliability of diffusion MRI analyses.