Investigation of Human Cardiac Fiber Architecture with Multiple diffusion MRI schemes at Multi-Scales based on Simulation Data

Background: Diffusion tensor imaging (DTI), diffusion spectrum imaging (DSI) and Q-ball imaging (QBI) are currently three main diffusion MRI (dMRI) schemes available for non-invasive investigation of cardiac fiber architecture. Although DSI and QBI have undoubtedly greater potential to reveal complex cardiac fiber structures than DTI, it however remains unclear to which level and at which scale they provide more gain for investigating cardiac fiber structure.Method This work attends to provide a quantitative description of cardiac fiber architecture derived from different schemes at various scales. Due to the limit of the spatial resolution of clinical MRI scanner and with the absence of the ground-truth, it is difficult to give the accurate description. To deal with this issue, we simulate firstly DTI, DSI and QBI of a cardiac fiber model with the structure a priori known at different scales, and then the estimation accuracy, the diffusion metrics and the helix and transverse angles of cardiac fiber obtained by different schemes at different scales are calculated. Results The results show that although DSI and QBI can distinguish multiple fiber orientations, they are readily to generate false positive and false negative fibers which influence therefore the estimation accuracy. When there are multiple fiber orientations in one voxel, the diffusion anisotropy detected by DTI is higher than DSI and QBI, the range of helix and transverse angle decreases with increasing of the scales, and that detected by DSI is larger than DTI and QBI. Conclusion The results showed that the proposed dMRI simulator provides a valuable tool for simulating realistic DW images of whole human hearts, which can be used as the gold standard to study the fiber structures of the heart.