Alternative Anisotropy Measure for diffusion MRI
A new paper entitled Apparent propagator anisotropy from single‐shell diffusion MRI acquisitions has been published in Magnetic Resonance in medicine. The work is written by members of the LPI (Santiago Aja-Fernández, Antonio Tristán-Vega) in collaboration with CUBRIC, University of Cardiff, UK (D.K. Jones). In the work, we propose a method to effectively calculate apparent Propagator Anisotropy (APA) from single shell diffusion MRI. The method avoids the calculation of the whole EAP by assuming the diffusion anisotropy is roughly independent from the radial direction. With such an assumption we achieve closed-form expressions for the measures using information from one single shell. At the same time, the closed forms provided make the method computationally efficient and robust.
Methods are implemented in MATLAB and they can be downloaded from www.lpi.tel.uva.es/AMURA
Purpose: The apparent propagator anisotropy (APA) is a new diffusion MRI metric that, while drawing on the benefits of the ensemble averaged propagator anisotropy (PA) compared to the fractional anisotropy (FA), can be estimated from single‐shell data.
Theory and Methods: Computation of the full PA requires acquisition of large datasets with many diffusion directions and different b‐values, and results in extremely long processing times. This has hindered adoption of the PA by the community, despite evidence that it provides meaningful information beyond the FA. Calculation of the complete propagator can be avoided under the hypothesis that a similar sensitivity/specificity may be achieved from apparent measurements at a given shell. Assuming that diffusion anisotropy (DiA) is nondependent on the b‐value, a closed‐form expression using information from one single shell (ie, b‐value) is reported.
Results: Publicly available databases with healthy and diseased subjects are used to compare the APA against other anisotropy measures. The structural information provided by the APA correlates with that provided by the PA for healthy subjects, while it also reveals statistically relevant differences in white matter regions for two pathologies, with a higher reliability than the FA. Additionally, APA has a computational complexity similar to the FA, with processing‐times several orders of magnitude below the PA.
Conclusions: The APA can extract more relevant white matter information than the FA, without any additional demands on data acquisition. This makes APA an attractive option for adoption into existing diffusion MRI analysis pipelines.