Simulation of dynamic DTI of 3D cardiac fiber structures

In-vivo three-dimensional (3D) DTI acquisition of an entire human heart for a complete cardiac cycle is currently technically impossible. We propose an alternative to this problem by simulating DTI of dynamic cardiac fiber structures. We first model a virtual dynamic cardiac fiber structure by integrating cardiac motion information into an ex vivo referential cardiac fiber structure and then use a Monte-Carlo method to simulate the corresponding diffusion weighted (DW) images, from which the diffusion tensors are derived and the relationship between diffusion image properties and cardiac motion is analyzed. The proposed approach enables us to investigate dynamic diffusion image properties at different cardiac phases. The results show that during the diastole, the cardiac fiber orientation distribution is rather regular, the fractional anisotropy (FA) increases, and the mean diffusivity (MD) decreases, and that this trend is reversed in the systole.