Efficient estimation of propagator anisotropy and non-Gaussianity in multishell diffusion MRI with micro-structure...

Area of Science:

• Diffusion MRI physics and signal processing.
• Quantitative imaging biomarkers.
• Neuroimaging analysis.

Background:

• The Ensemble Average Propagator (EAP) in diffusion MRI provides insights into tissue microstructure.
• Propagator Anisotropy (PA) and Non-Gaussianity (NG) are key EAP features, originally developed within the Mean Apparent Propagator diffusion MRI (MAP-MRI) framework.
• Efficient and accurate computation of these metrics is crucial for clinical translation.

Purpose of the Study:

• To reformulate Propagator Anisotropy (PA) and Non-Gaussianity (NG) within the Micro-Structure adaptive convolution kernels and dual Fourier Integral Transforms (MiSFIT) framework.
• To compare the performance of MiSFIT-based PA and NG with the original MAP-MRI methods.
• To evaluate accuracy, reliability, and computational efficiency.

Main Methods:

• Analytical reformulation of PA and NG indices within the MiSFIT framework.
• Visual comparison of index maps.
• Quantitative assessment using numerical simulations and synthetic data.
• Test-retest reliability study on the MICRA dataset.
• Computational time evaluation.

Main Results:

• Visual analysis showed similarity between MiSFIT and MAP-MRI derived indices.
• MiSFIT demonstrated improved accuracy against synthetic ground truth data.
• MiSFIT exhibited higher test-retest reliability in most white matter regions.
• MiSFIT achieved up to two orders of magnitude reduction in computational time.

Conclusions:

• PA and NG can be reliably and efficiently computed using the MiSFIT framework.
• MiSFIT offers a more accurate and computationally efficient alternative for calculating key diffusion MRI metrics.
• These advancements support the potential integration of diffusion MRI into routine clinical settings.