Magnetic resonance susceptibility contrast induced by capillaries: a numerical comparison of two models.

Monte-Carlo computer simulations have proven to be very powerful tools for the analysis of the magnetization decay induced by susceptibility gradients, as well for contrast agent characterization, as for the BOLD effect allowing fMRI. A recent vasculature model containing capillaries and venules uses homogeneous magnetized cylinders as models for the vessels. This modeling is questioned by comparing results obtained from simulation results based on two different models, one using homogeneous cylinders and another taking into account the existence of red blood cells, treated as homogeneous magnetized spheres. The results show the nonequivalence of both models, with the modeling by cylinders systematically overestimating the transverse relaxation rates, and the difference increasing with the adopted value of the diffusion coefficient. The discrepancy is attributed to the dominating role, regarding relaxation, of the local magnetic field in the immediate vicinity of the capillaries, which results in the suggestion of elaborating a "mixed modeling": the analytical expressions corresponding to the homogeneous cylinder model could be used except when the spin packets are wandering in the immediate vicinity of the capillaries, where accounting for the existence of individual red blood cells (whose motion may be neglected) seems unavoidable.