Incorporating the effect of white matter microstructure in the estimation of magnetic susceptibility in ex vivo mouse brain

Purpose: To extend quantitative susceptibility mapping to account for microstructure of white matter (WM) and demonstrate its effect on ex vivo mouse brain at 16.4T. Theory and Methods: Previous studies have shown that the MRI measured Larmor frequency also depends on local magnetic microstructure at the mesoscopic scale. Here, we include effects from WM microstructure using our previous results for the mesoscopic Larmor frequency (Omega) over bar (Meso) of cylinders with arbitrary orientations. We scrutinize the validity of our model and QSM in a digital brain phantom including (Omega) over bar (Meso) from a WM susceptibility tensor and biologically stored iron with scalar susceptibility. We also apply susceptibility tensor imaging to the phantom and investigate how the fitted tensors are biased from (Omega) over bar (Meso). Last, we demonstrate how to combine multi-gradient echo and diffusion MRI images of ex vivo mouse brains acquired at 16.4T to estimate an apparent scalar susceptibility without sample rotations. Results: Our new model improves susceptibility estimation compared to QSM for the brain phantom. Applying susceptibility tensor imaging to the phantom with (Omega) over bar (Meso) from WM axons with scalar susceptibility produces a highly anisotropic susceptibility tensor that mimics results from previous susceptibility tensor imaging studies. For the ex vivo mouse brain we find the (Omega) over bar (Meso) due to WM microstructure to be substantial, changing susceptibility in WM up to 25% root-mean-squared-difference. Conclusion: (Omega) over bar (Meso) impacts susceptibility estimates and biases susceptibility tensor imaging fitting substantially. Hence, it should not be neglected when imaging structurally anisotropic tissue such as brain WM.