Novel insights into in-vivo diffusion tensor cardiovascular magnetic resonance using computational modeling and a histology-based virtual microstructure.

Purpose: To develop histology-informed simulations of diffusion tensor cardiovascular magnetic resonance (DT-CMR) for typical in-vivo pulse sequences and determine their sensitivity to changes in extra-cellular space (ECS) and other microstructural parameters.

Methods: We synthesised the DT-CMR signal from Monte Carlo random walk simulations. The virtual tissue was based on porcine histology. The cells were thickened and then shrunk to modify ECS. We also created idealised geometries using cuboids in regular arrangement, matching the extra-cellular volume fraction (ECV) of 16-40%. The simulated voxel size was 2.8 × 2.8 × 8.0 mm for pulse sequences covering short and long diffusion times: Stejskal-Tanner pulsed-gradient spin echo, second-order motion-compensated spin echo, and stimulated echo acquisition mode (STEAM), with clinically available gradient strengths.

Results: The primary diffusion tensor eigenvalue increases linearly with ECV at a similar rate for all simulated geometries. Mean diffusivity (MD) varies linearly, too, but is higher for the substrates with more uniformly distributed ECS. Fractional anisotropy (FA) for the histology-based geometry is higher than the idealised geometry with low sensitivity to ECV, except for the long mixing time of the STEAM sequence. Varying the intra-cellular diffusivity (D ) results in large changes of MD and FA. Varying extra-cellular diffusivity or using stronger gradients has minor effects on FA. Uncertainties of the primary eigenvector orientation are reduced using STEAM.

Conclusions: We found that the distribution of ECS has a measurable impact on DT-CMR parameters. The observed sensitivity of MD and FA to ECV and D has potentially interesting applications for interpreting in-vivo DT-CMR parameters.