Introduction

Myelin and iron play critical roles in maintaining structural integrity and normal function of the brain. Histology has shown heterogeneous distribution of myelin and iron organized in layers at various cortical depths. In vivo imaging of these fine-scale structures at the mesoscopic scale (<0.5 mm), approaching that of histology, can potentially improve characterization of neurological diseases, such as Alzheimer’s disease¹, multiple sclerosis² and epilepsy³, and, if so, lead to more effective treatment planning. In this study, we quantified layer-specific R₂* and magnetic susceptibility (χ), both of which are sensitive to myelin and iron, using 0.3x0.3x0.4 mm³ T₂*-weighted (T₂*w) MRI at 7 T. A navigator-based motion and B₀ correction method was used to improve the robustness of high-resolution MRI data⁴.

Methods

Data acquisition and image reconstruction
Experiments were performed on a 7 T MRI scanner (Magnetom, Siemens) with a 32-channel head RF receiver array (Nova Medical). Eleven healthy subjects were recruited with signed consent under an IRB-approved protocol. High-resolution dual-echo T₂*w data were acquired with 3D GRE. To correct for head motion and B₀ fluctuation, a series of volumetric navigator images was obtained using a multi-shot 3D EPI readout at a shorter echo time than that of the T₂*w acquisition⁴. The T₂*w data were acquired with the following parameters: resolution=0.3x0.3x0.4 mm³, TE1/TE2/TR=18/39/74 ms, flip angle=14°, bandwidth=54 Hz/pixel, FOV=240x180x32 mm³, SENSE rate=2x1, scan time=35.5 mins. Spatial and temporal resolution of the navigator were 5x5.6x3.2 mm and 0.4 s, respectively. Subjects were asked to stay relaxed during the scans.
T₂*w complex images were reconstructed with either correction of motion and spatially linear B₀ fluctuation (Correction mode)^4,5 or only correction of global average B₀ fluctuation (Non-correction mode). T₂*w magnitude image was calculated as the mean of magnitude images across echoes. Quantitative susceptibility mapping (QSM) was performed on the T₂*w complex data using the JHU/KKI QSM toolbox^6–8 including the following steps: Laplacian phase unwrapping ⁹, combined LBV¹⁰ and VSHARP¹¹ for background field removal, and dipole inversion regularized by a learned proximal convolutional neural network (LP-CNN)¹².
Data analysis
Blinded review of T₂*w images was performed by two experienced image readers. ROIs corresponding to superficial, middle (Line of Gennari), and deep layers were manually segmented in the primary visual cortex based on T₂*w image intensity. R₂* was quantified using nonlinear least square complex fitting. Potential effect of cortical orientation on R₂* and χ was investigated in each ROI using a mixed-effect linear model of y=a₁cos(2θ)+a₂sin(2θ)+c with subject representing the random effect on a₁, a₂, and c. Here, θ denotes the angle between normal direction of cortical ribbon and B₀, and was calculated using the LAYNII toolbox¹³.

Results

T₂*w magnitude and susceptibility examples in Fig. 1A demonstrate delineation of anatomical details that necessitate high spatial resolution. Correction of motion and B₀ fluctuation increased the image quality and delineation of cortical layers, as shown in Figs. 1B and 2. Reducing resolution renders the line of Gennari difficult to observe both visually (Fig. 3) and quantitatively (Fig. 4). At the subject level, ROI-averaged R₂* values are distinct at different depths in the primary visual cortex (Fig. 4A). This effect was reduced or diminished without correction and at lower spatial resolution. Similarly, lower resolution reduced χ difference between cortical depths and led to more negative values (Fig. 4B), potentially due to partial volume effect from whiter matter region. Unlike R₂*, the ROI-averaged χ distributions exhibit more overlap between different depths. At the group level, a cortex-to-field orientation dependency in the form of cos(2θ) was observed in χ with a peak-to-peak difference of 0.02 ppm (p-values < 1x10^-5) (Fig. 5) but not in R₂* (p-values > 0.2).

Discussion and Conclusion

In this study, R₂* and χ at different layers in the primary visual cortex were investigated with high-resolution MRI at 7 T, enabled by a navigator-based motion- and B₀-corrected GRE sequence. Consistent with prior work, distinct R₂* and χ were observed between layers. A significant orientation effect (~ 0.02 ppm) was found in cortical χ with sources to be determined in future studies. In vivo high-resolution imaging of R₂* and χ, which are related to myelin and iron distribution, can potentially contribute to novel biomarkers sensitive to subtle early-stage pathological changes in neurological disorders and degeneration.

Acknowledgements

This work was supported in part by the intramural research program of NINDS, NCRR and NIBIB (P41EB031771) and the startup funding of J.L.

References

1. Kenkhuis, B. et al. 7T MRI allows detection of disturbed cortical lamination of the medial temporal lobe in patients with Alzheimer’s disease. Neuroimage Clin 21, 101665 (2019).
2. Kilsdonk, I. D. et al. Increased cortical grey matter lesion detection in multiple sclerosis with 7 T MRI: a post-mortem verification study. Brain 139, 1472–1481 (2016).
3. Bartolini, E. et al. Ultra-High-Field Targeted Imaging of Focal Cortical Dysplasia: The Intracortical Black Line Sign in Type IIb. American Journal of Neuroradiology (2019).
4. Liu, J., van Gelderen, P., de Zwart, J. A. & Duyn, J. H. Reducing motion sensitivity in 3D high-resolution T2*-weighted MRI by navigator-based motion and nonlinear magnetic field correction. Neuroimage 206, 116332 (2020).
5. Gretsch, F., Marques, J. P. & Gallichan, D. Investigating the accuracy of FatNav-derived estimates of temporal B0 changes and their application to retrospective correction of high-resolution 3D GRE of the human brain at 7T. Magn Reson Med 80, 585–597 (2018).
6. Bao, L., Li, X., Cai, C., Chen, Z. & van Zijl, P. C. M. Quantitative Susceptibility Mapping Using Structural Feature Based Collaborative Reconstruction (SFCR) in the Human Brain. IEEE Trans Med Imaging 35, 2040–2050 (2016).
7. van Bergen, J. M. G. et al. Quantitative Susceptibility Mapping Suggests Altered Brain Iron in Premanifest Huntington Disease. AJNR Am J Neuroradiol 37, 789–796 (2016).
8. Li, X. et al. Multi-atlas tool for automated segmentation of brain gray matter nuclei and quantification of their magnetic susceptibility. Neuroimage 191, 337–349 (2019).
9. Schofield, M. A. & Zhu, Y. Fast phase unwrapping algorithm for interferometric applications. Opt. Lett., OL 28, 1194–1196 (2003).
10. Zhou, D., Liu, T., Spincemaille, P. & Wang, Y. Background field removal by solving the Laplacian boundary value problem. NMR in Biomedicine 27, 312–319 (2014).
11. Wu, B., Li, W., Guidon, A. & Liu, C. Whole brain susceptibility mapping using compressed sensing. Magn Reson Med 67, 137–147 (2012).
12. Lai, K.-W., Aggarwal, M., van Zijl, P., Li, X. & Sulam, J. Learned Proximal Networks for Quantitative Susceptibility Mapping. arXiv:2008.05024 [cs, eess] (2020).
13. Huber, L. (Renzo) et al. LayNii: A software suite for layer-fMRI. NeuroImage 237, 118091 (2021).