The disruption of iron homeostasis in MS lesions is an important factor leading to chronic oxidative stress and contributing to progressive disease (1-3). Lesion iron quantification is therefore highly desired for studying neurodegeneration and for developing novel therapeutics targeting iron-related oxidative injury in MS. Quantitative Susceptibility Mapping (QSM) is a reliable method for mapping magnetic sources in the brain including iron (4-6). However, accurate iron quantification in the white matter (WM) is confounded by the coexistence of iron and myelin lipids, both of which contribute to tissue susceptibility (5). The objective of this study was to develop a clinically feasible method for absolute lesion iron quantification by integrating absolute myelin mapping using the recently developed Fast Acquisition using Spiral Trajectory and T2prep (FAST-T2) sequence (7) with QSM.

Separating iron and myelin contributions in QSM. Two sources in WM are the primary contributors to its total susceptibility: 1) iron with $$$\chi_{Fe}$$$ = 1.32 ppb [Fe] and 2) myelin phospholipids with $$$\chi_{PL}$$$ = $$$(\chi_A sin^2 \theta + \chi_I) V_{PL}$$$ , where $$$\chi_A$$$ = -90 ppb is the susceptibility anisotropy of a phospholipid molecule (8), $$$\theta$$$ is the angle between local WM fiber direction and B₀ and can be estimated using diffusion tensor imaging (DTI), $$$\chi_I$$$ = -16 ppb is the isotropic volume susceptibility of phospholipid and $$$V_{PL}$$$ is the phospholipid volume fraction which is proportional to myelin water content (MWC). In this study, the proportionality constant was estimated using the phospholipid density of 0.25 g/cm³ and the phospholipid to myelin water weight ratio of 1.9 (9).

Fast brain MWC mapping at 3T. FAST-T2 can provide whole brain myelin water fraction (MWF) mapping in 4 min at 3T (7). Absolute MWC can then be computed from MWF by referencing the tissue signal to that of CSF (~100% water) after correcting for differences in T1 and coil sensitivity as demonstrated recently at 1.5T (10). For 3T imaging, correction for RF flip angle inhomogeneity is also necessary. Accordingly, FAST-T2 was modified to perform rapid spiral Look-Locker imaging and the effective flip angle $$$\alpha$$$ was computed using the equation: 1/T_1,LL = 1/T₁ - ln(cos($$$\alpha$$$))/TR, where T_1,LL is the time constant of the saturation recovery signal.

Imaging experiment. Eight healthy volunteers (31 years ± 9) and three MS patients (35 years ± 4) were imaged at 3T using QSM (0.6x0.6x2.5 mm³, 10 min), MWC mapping (1.2x1.2x5 mm³, 4 min FAST-T2 using 6 geometric TEs, 5 min T1 mapping using 7 saturation recovery times between 0.2 and 10 sec, 2 min flip angle mapping), and DTI (1.9x1.9x2.5 mm³, 6 min). QSM maps were computed using the MEDI algorithm (11). MWF maps were extracted using a spatially constrained non-linear least squares algorithm (7). DTI images were first corrected for field inhomogeneity induced geometric distortion and then co-registered with QSM and MWC images using FSL software.

All scans were completed successfully. Figure 1 shows an example of MWC, DTI and QSM maps obtained from a healthy subject, illustrating the much reduced WM/GM contrast in the QSM map after the contribution from myelin is removed. The observed increased QSM homogeneity is more consistent with the expectation that iron is evenly distributed over the whole brain, except in the basal ganglia. In healthy volunteers (n=8), the average [Fe] in whole brain WM was 22.2 ± 2.1 mg/kg, which was similar to values obtained by mass spectrometry in post mortem brains (6). In MS patients (n=3), the average [Fe] was 26.0 ± 5.9 mg/kg. Figure 2 shows an example of myelin and iron maps from a MS patient with acute disease, demonstrating the ability of the proposed method to quantify both demyelination and iron activity within a lesion.Our preliminary data have demonstrated the feasibility of the developed absolute lesion myelin and iron quantification method in the in vivo brain. The combined scan time of the QSM/MWC/DTI acquisitions (currently ~27 min) can be reduced by utilizing parallel imaging in QSM and accelerating T1 mapping acquisition through compressed sensing and the use of much shorter saturation time (max. saturation time of 10 sec was used in this study to obtain a reliable estimate of CSF T1 (~4 sec), which will not be necessary in subsequent studies). Future work will be focused on evaluating the accuracy and reproducibility of the developed method using animal models of MS and ex vivo human brains.