1307

Normative trajectories of quantitative MRI parameters in sub-cortical grey matter on healthy ageing

Kwok-Shing Chan^1,2, Michelle G. Jansen¹, Joukje Oosterman¹, David G. Norris^1,3,4, Christian F. Beckmann^1,2, and José P. Marques¹
¹Donders Institute for Brain, Cognition and Behaviour, Nijmegen, Netherlands, ²Department of Cognitive Neuroscience, Radboud University Medical Center, Nijmegen, Netherlands, ³The Erwin L. Hahn Institute, Essen, Germany, ⁴University of Twente, Twente, Netherlands

Synopsis

Keywords: Neurodegeneration, Aging

The effects of ageing on quantitative MRI, including R₁, R₂* and tissue susceptibility, are investigated in subcortical grey matter using a healthy cohort. General Linear Regression analysis indicates that ageing has significant impacts on these quantitative measures: the mean R₁ has inverted U-shape appearances with time, while the mean R₂* and susceptibility are closer to linear. We further studied the spatial variation in these structures and the results show that the spatial gradient of some structures (e.g. caudate and putamen) also changes with age. Normative trajectories of these parameters in subcortical grey matter associated with ageing are also investigated.

Introduction

Characterising brain development across the lifespan is essential not only to understand how our brain matures, but also to define normative space from which neurological disorders, such as Alzheimer’s and Parkinson’s diseases, can be detected and allows studies of the pathophysiology. Extensive studies demonstrated several structural changes in the brain with MRI^1–4. Meanwhile, quantitative MRI (qMRI) parameters, including R₁, R₂* and tissue magnetic susceptibility ($$$\chi$$$), are receiving more attention as surrogate biomarkers of myelination⁵ and iron concentration^6–8, providing more specific insights into the biochemical environment changes with ageing⁹. Previous research mainly investigated descriptive statistics of qMRI, for example, changes of the mean parameters across the entire cortical/subcortical structures¹⁰. Additionally, ageing and diseases may also affect the spatial distribution (gradient) of the quantitative parameters and functional connectivity^11,12. In this work, we set out to investigate the cross-sectional mean and spatial effects of ageing on qMRI parameters R₁, R₂* and $$$\chi$$$ in subcortical grey matter (GM) structures using a healthy ageing cohort.

Methods

Data acquisition
Data acquisition was performed at 3T (Siemens, Erlangen) on 295 healthy volunteers. Thirty subjects were excluded due to incidental findings and/or severe motion artefacts, resulting in 265 subjects included in the statistical analysis (range, 18-79years, mean±SD=51.2±16.8years, see Fig.1a). The imaging protocol consists of:

Whole-brain T₁ scan using MP2RAGE¹³, res=1mm iso., $$$\alpha_{1}$$$/$$$\alpha_{2}$$$=6°/5°, TI₁/TI₂=700ms/2000ms, TR/TE=6000ms/2.34ms, TA=7mins;
Bipolar 3D multi-echo GRE, TR/TE₁/$$$\Delta$$$TE/nTE=44ms/6.14ms/4ms/9, GRAPPA=3, res=0.8mm iso., $$$\alpha$$$=20°, TA=9.5min.

Data Processing
Brain extraction was performed using HD-BET¹⁴. The GRE brain masks were refined by excluding voxels with high R₂* values on the edge to improve estimation robustness. $$$\chi$$$ maps were derived using SEPIA¹⁵ with the following pipeline: ROMEO¹⁶ for total field computation, V-SHARP¹⁷ for background field removal and LP-CNN¹⁸ for dipole field inversion. The mean susceptibility value across the whole brain was used as reference. For R₂* mapping, MP-PCA denoising^19,20 was applied to the complex-valued mGRE data to improve SNR and the magnitude of the denoised data was then extracted to derive R₂* maps based on a closed-form solution²¹.

Subcortical grey matter parcellation
To facilitate data/statistical analysis in the common (MNI) space, image registration was performed between the R₁ and GRE data, and across all subjects (see Fig.1b). R₁-QSM hybrid images²² were used to deliver high-quality registration in subcortical regions. Subcortical GM parcellation was achieved by non-linear image registration²³ between the group-averaged hybrid image and the template from the MuSus-100 atlas²⁴, from which the derived transformation matrices were applied to the atlas labels. The mean qMRI (R₁, R₂* and $$$\chi$$$) values were computed for each structure. Additionally, a 1^st order 3D polynomial function was employed to fit the spatial distribution of the qMRI parameters across the mask of each subcortical GM structure so that the spatial gradients (Posterior-Anterior/Lateral-Medial/Ventral-Dorsal of the MNI coordinate) could also be studied.

Statistical analysis
The general linear model (GLM) analysis was conducted to investigate the impact of ageing on qMRI parameters in various structures (ROI). The design matrix comprised four regressors: Age, Age², Sex, and (left-/right-)Hemisphere:
$$qMRI_{mean/\Delta(A-P)/\Delta(L-M)/\Delta(V-D)}=\beta_{constant}+\beta_{Age}Age+\beta_{Age^2}Age^2+\beta_{Sex}Sex+\beta_{Hemisphere}Hemisphere[Eq.1]$$
The utility of Age² as a regressor is to consider the inverted-U shape effect observed in our R₁ data which was also shown previously²⁵. Z-statistics for each regressor was computed by converting the t-statistic of GLM using fsl_glm²⁶. In the following normative modelling analysis, only the structures with a |z-score| greater than 3.1 on age-related effects (Age or Age²) were studied, and the data were corrected for the sex-specific and inter-hemispherical effects based on the GLM results. Further, Gaussian process regression (GRP)²⁷ was used to compute the normative trajectory on the corrected qMRI parameters as a function of age.

Results and Discussion

The group-averaged qMRI parameters maps and the parcellation results are shown in Fig.2.

The GLM results are shown in Fig.3, where the significances of the regressors are expressed in z-statistic. All means of the qMRI parameters show age-dependency except in nucleus accumbens (NAcc)(Fig.3a). The strong gradient in the P-A direction of the caudate R₁ (reproducing, in a continuous manner, the findings in ¹¹) and susceptibility are also clearly highlighted. Among the three qMRI parameters, R₁ has a consistently strong Age² effect (Fig.3b). Sex had a relatively small impact on all qMRI (Fig. 3c), whereas inter-hemispherical differences are seen in the P-A direction of R₁ and in various susceptibility metrics (Fig. 3d).

The GPR analysis of the mean qMRI parameters is shown in Fig.4 highlighting the varying rates and shapes of these quantities and the increasing variance with ageing.

The GPR analysis of the spatial gradient qMRI parameters is shown in Fig.5. Clearly, gradients have very different variances which are likely related to the size, shape and segmentation quality of individual structures (e.g. NAcc). Further investigation is required to confirm these findings.

Conclusions

Most results on the mean qMRI parameters reproduce the previous findings on discretised age ranges. To the best of our knowledge, the spatial pattern variations have only been partially previously reported on ¹¹. Future work will focus on: (i)interpreting the meaning of such gradients and whether they can be related to variations of the mean of different substructures (e.g. thalamus); (ii)extending the analysis to white matter and cortical GM; and (iii)evaluate the impact of cognitive markers in explaining the age-related variance.

Acknowledgements

This work is part of the Marie Curie “Initial Training Networks” action from the European Union with the project reference “FP7-PEOPLE-2013-ITN”. CFB gratefully acknowledges funding from the Wellcome Trust Collaborative Award in Science 215573/Z/19/Z and the Netherlands Organization for Scientiﬁc Research Vici Grant No. 17854.

References

1. Bethlehem RAI, Seidlitz J, White SR, et al. Brain charts for the human lifespan. Nature 2022;604:525–533 doi: 10.1038/s41586-022-04554-y.

2. Terribilli D, Schaufelberger MS, Duran FLS, et al. Age-related gray matter volume changes in the brain during non-elderly adulthood. Neurobiol Aging 2011;32:354–368 doi: 10.1016/j.neurobiolaging.2009.02.008.

3. Scahill RI, Frost C, Jenkins R, Whitwell JL, Rossor MN, Fox NC. A Longitudinal Study of Brain Volume Changes in Normal Aging Using Serial Registered Magnetic Resonance Imaging. Arch Neurol-chicago 2003;60:989–994 doi: 10.1001/archneur.60.7.989.

4. Ramanoël S, Hoyau E, Kauffmann L, et al. Gray Matter Volume and Cognitive Performance During Normal Aging. A Voxel-Based Morphometry Study. Front Aging Neurosci 2018;10:235 doi: 10.3389/fnagi.2018.00235.

5. Lutti A, Dick F, Sereno MI, Weiskopf N. Using high-resolution quantitative mapping of R1 as an index of cortical myelination. Neuroimage 2014;93:176–188 doi: 10.1016/j.neuroimage.2013.06.005.

6. Stüber C, Morawski M, Schäfer A, et al. Myelin and iron concentration in the human brain: A quantitative study of MRI contrast. Neuroimage 2014;93:95–106 doi: 10.1016/j.neuroimage.2014.02.026.

7. Ghadery C, Pirpamer L, Hofer E, et al. R2* mapping for brain iron: associations with cognition in normal aging. Neurobiol Aging 2015;36:925–932 doi: 10.1016/j.neurobiolaging.2014.09.013.

8. Pontillo G, Petracca M, Monti S, et al. Clinical correlates of R1 relaxometry and magnetic susceptibility changes in multiple sclerosis: a multi-parameter quantitative MRI study of brain iron and myelin. Eur Radiol 2022:1–10 doi: 10.1007/s00330-022-09154-y.

9. Wang C, Martins-Bach AB, Alfaro-Almagro F, et al. Phenotypic and genetic associations of quantitative magnetic susceptibility in UK Biobank brain imaging. Nat Neurosci 2022;25:818–831 doi: 10.1038/s41593-022-01074-w.

10. Seiler A, Schöngrundner S, Stock B, et al. Cortical aging – new insights with multiparametric quantitative MRI. Aging Albany Ny 2020;12:16195–16210 doi: 10.18632/aging.103629.

11. Drori E, Berman S, Mezer AA. Mapping microstructural gradients of the human striatum in normal aging and Parkinson’s disease. Sci Adv 2022;8:eabm1971 doi: 10.1126/sciadv.abm1971.

12. Oldehinkel M, Llera A, Faber M, et al. Mapping dopaminergic projections in the human brain with resting-state fMRI. Elife 2022;11:e71846 doi: 10.7554/elife.71846.

13. Marques JP, Kober T, Krueger G, Zwaag W van der, Moortele PFV de, Gruetter R. MP2RAGE, a self bias-field corrected sequence for improved segmentation and T1-mapping at high field. Neuroimage 2010;49:1271–1281 doi: 10.1016/j.neuroimage.2009.10.002.

14. Isensee F, Schell M, Pflueger I, et al. Automated brain extraction of multisequence MRI using artificial neural networks. Hum Brain Mapp 2019;40:4952–4964 doi: 10.1002/hbm.24750.

15. Chan K-S, Marques JP. SEPIA—Susceptibility mapping pipeline tool for phase images. Neuroimage 2021;227:117611 doi: 10.1016/j.neuroimage.2020.117611.

16. Dymerska B, Eckstein K, Bachrata B, et al. Phase unwrapping with a rapid opensource minimum spanning tree algorithm (ROMEO). Magnet Reson Med 2020 doi: 10.1002/mrm.28563.

17. Li W, Wu B, Liu C. Quantitative susceptibility mapping of human brain reflects spatial variation in tissue composition. Neuroimage 2011;55:1645–1656 doi: 10.1016/j.neuroimage.2010.11.088.

18. Lai K-W, Aggarwal M, Zijl P van, Li X, Sulam J. Medical Image Computing and Computer Assisted Intervention – MICCAI 2020, 23rd International Conference, Lima, Peru, October 4–8, 2020, Proceedings, Part II. Lect Notes Comput Sc 2020;12262:125–135 doi: 10.1007/978-3-030-59713-9_13.

19. Veraart J, Novikov DS, Christiaens D, Ades-aron B, Sijbers J, Fieremans E. Denoising of diffusion MRI using random matrix theory. Neuroimage 2016;142:394–406 doi: 10.1016/j.neuroimage.2016.08.016.

20. Does MD, Olesen JL, Harkins KD, et al. Evaluation of principal component analysis image denoising on multi-exponential MRI relaxometry. Magnetic resonance in medicine 2019;81:mrm.27658-3514 doi: 10.1002/mrm.27658.

21. Gil R, Khabipova D, Zwiers M, Hilbert T, Kober T, Marques JP. An in vivo study of the orientation-dependent and independent components of transverse relaxation rates in white matter. NMR in biomedicine 2016;29:1780–1790 doi: 10.1002/nbm.3616.

22. Zhang Y, Wei H, Cronin MJ, He N, Yan F, Liu C. Longitudinal data for magnetic susceptibility of normative human brain development and aging over the lifespan. Data Brief 2018;20:623–631 doi: 10.1016/j.dib.2018.06.005.

23. Avants BB, Tustison NJ, Song G, Cook PA, Klein A, Gee JC. A reproducible evaluation of ANTs similarity metric performance in brain image registration. Neuroimage 2011;54:2033–2044 doi: 10.1016/j.neuroimage.2010.09.025.

24. He C, Guan X, Zhang W, et al. Quantitative susceptibility atlas construction in Montreal Neurological Institute space: towards histological-consistent iron-rich deep brain nucleus subregion identification. Brain Struct Funct 2022:1–23 doi: 10.1007/s00429-022-02547-1.

25. Yeatman JD, Wandell BA, Mezer AA. Lifespan maturation and degeneration of human brain white matter. Nature communications 2014;5:4932 doi: 10.1038/ncomms5932.

26. Jenkinson M, Beckmann CF, Behrens TEJ, Woolrich MW, Smith SM. FSL. Neuroimage 2012;62:782–790 doi: 10.1016/j.neuroimage.2011.09.015.

27. Rutherford S, Kia SM, Wolfers T, et al. The normative modeling framework for computational psychiatry. Nat Protoc 2022;17:1711–1734 doi: 10.1038/s41596-022-00696-5.

28. Erramuzpe A, Schurr R, Yeatman JD, et al. A Comparison of Quantitative R1 and Cortical Thickness in Identifying Age, Lifespan Dynamics, and Disease States of the Human Cortex. Cereb Cortex 2020;31:1211–1226 doi: 10.1093/cercor/bhaa288.

29. Li Y, Sethi SK, Zhang C, et al. Iron Content in Deep Gray Matter as a Function of Age Using Quantitative Susceptibility Mapping: A Multicenter Study. Front Neurosci-switz 2021;14:607705 doi: 10.3389/fnins.2020.607705.

30. Treit S, Naji N, Seres P, et al. R2* and quantitative susceptibility mapping in deep gray matter of 498 healthy controls from 5 to 90 years. Hum. Brain Mapp. 2021;42:4597–4610 doi: 10.1002/hbm.25569.

Figures

Figure 1: (a) Study population demographic and the median (red line) and IQR (box) of the participants with respect to sex are shown in the boxplots. 30 subjects (10%) were excluded from the analysis due to poor data quality and incidental findings. (b) Data registration procedures. The qMRI maps from all subjects were registered to the MNI space defined in the MuSus-100 atlas. Hybrid images combining the R₁ and 𝜒 maps were used to ensure high-quality registration because of the enhanced subcortical structure contrasts.

Figure 2: (a) Subcortical GM parcellation was performed by using a two-step approach. By using the original and a left-right flipped version of the atlas template, image coregistration was then conducted separately and the final labels were defined as the intersection of both registration steps. This procedure is to minimise the registration bias by considering only the voxels that have high certainty inside the mask in the analysis. (b) Group-averaged qMRI parameter (R₁, R₂* & $$$\chi$$$) maps.

Figure 3: A summary heatmap of GLM z-statistics, showing on the x-axis are the 11 ROIs while the y-axis represents 3 panels (Top: R₁; Middle: R₂*; Bottom: χ). Most of the subcortical GM structures demonstrate a strong quadratic effect on mean R₁, whereas mean R₂* and χ show more significant linear relationship with age. Fewer structures show age dependencies on the gradient change on qMRI parameters compared to the results of the means. Sex differences in general do not introduce extra bias in the GLM. It can also be seen that some GM structures show hemispherical R₁ and χ differences.

Figure 4: Normative trajectories of the mean qMRI parameters in ageing. Only structures with a |z-score|>3.1 in GLM are shown. The normative trajectories of R₁ have an inverted-U shape, which agrees with a previous study on cortical GM²⁸. The trajectories of χ are similar to the corresponding R₂* trajectories. The distributions of the R₂* and χ generally agree with previous studies^29,30 despite different models being used to illustrate the ageing effects.

Figure 5: Normative trajectories of spatial gradients of qMRI parameters in ageing. Only data with a |z-score|>3.1 are shown. Previous reports have shown gradient differences between younger and older adults using discretised ROIs along the principal structural axes of the putamen and caudate R₁. The z-statistics of our GLM analysis also indicates significant age-related gradient differences in these structures. Further investigation is required to validate these findings.

Proc. Intl. Soc. Mag. Reson. Med. 31 (2023)

1307

DOI: https://doi.org/10.58530/2023/1307