Dependence of R2* Bias on Through-Voxel Frequency Dispersion and Gradient Echo Train in High-Resolution 3D R2* Mapping

Gunther Helms¹, Peter Dechent¹

An empirical model for the influence of through-voxel gradients on log regression of R2* was derived from simulations. This advocates trains of many gradient echoes that start early and are short compared to local frequency dispersion, that is, use of non-selective high-resolution 3D acquisitions. The general trade-off is between statistical error of R2* and sensitivity to bias. For 1mm resolution at 3T, excessive bias can be confined to small orbito-frontal and temporo-basal regions, whereas correction of bias is unreliable. High-resolution R2* mapping of (almost) the whole brain seems feasible.