Synopsis

Cerebrum microstructure has been extensively investigated with diffusion-weighted MRI, but little attention has been dedicated to the microstructural characterisation of the cerebellum.

We considered an anatomical parcellation of the cerebellum and fitted a multi-compartment model to diffusion data exploiting the spherical mean technique, which provides parametric maps unconfounded by the underlying fibre orientation distribution. For each region we report average values for multi-compartment parameters (e.g. intra-neurite volume fraction and intrinsic diffusivity) and diffusion tensor metrics.

Multi-compartment metrics more specific to microstructure provide information complementary to diffusion tensor metrics in the cerebellum, thus giving new insights about microstructural correlations between regions.

Purpose

Diffusion-weighted MRI (DW-MRI) is a valid tool for in vivo investigation of neural tissue microstructure. Over the years a lot of attention has been dedicated to study the cerebrum. However, more recently, attention has also turned onto the cerebellum because of new evidences of its key role in complex functions and pathology¹.

The most widely used DW-MRI signal model is the diffusion tensor (DT), which has proven to be sensitive but not specific to microstructure and inadequate, for example, in regions of complex structure. To overcome DT limitations, new techniques, models and metrics have been recently proposed and applied to study the brain². Amongst these, the spherical mean technique³ (SMT) is particularly appealing because it can be applied to complex diffusion signal models making no a priori assumptions about the underlying fibre orientation distribution (FOD).

The aim of the present work is to characterize cerebellar microstructure using the SMT in its recently proposed two-compartment version⁴, considering a validated anatomical parcellation.

Methods

The study was carried out on high-quality images of 15 subjects from the Human Connectome Project⁵, 100 Subjects Data Release with minimal pre-processing pipeline (diffusion data: 1.25mm isotropic resolution, b=1000,2000,3000s/mm², 90 isotropically distributed directions for each shell; anatomical 3DT1-weighted data: 1.25mm isotropic resolution).

As our interest focused on the cerebellum we exploited SPM12⁶ and the SUIT toolbox⁷ to isolate this structure and to perform its segmentation on the 3DT1-weighted images co-registered to diffusion data. Images were then registered to the SUIT template and the inverse transformation was applied to the standard SUIT cerebellar atlas⁸ and to the probabilistic atlas of cerebellar peduncles⁹ (50% threshold) to bring them to native space.

We applied the freely available two-compartment SMT implementation to DW-MRI data, obtaining voxel-wise estimates of intra-neurite volume fraction (v_int), the intrinsic diffusivity (λ), the transverse and mean extra-neurite diffusivity. Unlike other models¹⁰ where a fixed λ is assumed across the brain, here it is directly estimated from data.

Region-specific average values of SMT/DTI metrics were obtained within cerebellar cortical regions, cerebellar peduncles and deep cerebellar nuclei¹¹ derived from SUIT.

Lastly, maps were registered to SUIT standard space and concatenated across subjects. We created a between-region cross-correlation matrix for each microstructural feature to assess whether between-region microstructural relationships are repeatable across subjects.

The same procedure was followed also for classical DT metrics for comparison.

Results and discussion

Figure 1 reports maps of computed metrics for a single subject. It is noteworthy that λ exhibits regional variation even within the same tissue. Hence, models where a constant λ is assumed may be not appropriate for cerebellar applications.

Boxplots in figure 2 show average v_int, λ and fractional anisotropy (FA) in each region across subjects and corresponding maps of the average v_int are displayed in figure 3. Considering grey matter v_int and FA (figure 2), both metrics show no significant difference between left and right cerebellar regions. However, unlike FA, v_int suggests lower neurite density of the vermis as compared to the two cerebellar hemispheres. Furthermore, the same comparison in the cerebellar peduncles highlights greater differences between superior cerebellar peduncles (SCPs) and inferior cerebellar peduncles (ICPs) in terms of v_int. SCPs, which project their fibres towards the cerebral cortex, show higher v_int than ICPs. Since ICPs receive inputs from the spinal cord and the medulla oblongata, this finding suggests that these neurite populations may have different microstructural properties, such as different neurite packing, average fibre diameter or myelination level.

Figure 4 shows cross-correlation matrices for v_int, λ and FA (75% absolute threshold). SMT metrics cross-correlation matrices underline new possible microstructural relationships among structures, unseen with FA, which are repeatable across subjects. A striking difference, for example, is the correlation between the dentate and interposed nuclei, respectively rows 29-30 (L-R) and 31-32, very strong for v_int and not detectable for FA. Moreover, the known strong structural connection between DCNs and cerebellar peduncles¹² emerges from the cross-correlation matrix of v_int, suggesting that functionally related structures may also share microstructural features.

Conclusions

Acknowledgements

Data were provided by the Human Connectome Project, WU-Minn Consortium (Principal Investigators: David Van Essen and Kamil Ugurbil; 1U54MH091657) funded by the 16 NIH Institutes and Centers that support the NIH Blueprint for Neuroscience Research; and by the McDonnell Center for Systems Neuroscience at Washington University.

Horizon2020-EU.3.1 CDS-QUAMRI (ref: 634541).

Serena M. Valle for valuable consultancy and help in the production of the present abstract.

References

1. D'Angelo E, Casali S. Seeking a unified framework for cerebellar function and dysfunction: from circuit operations to cognition. Frontiers in neural circuits 6 (2013): 116.

2. Ferizi U, Schneider T, Panagiotaki E, et al. A Ranking of Diffusion MRI Compartment Models with In Vivo Human Brain Data. Magnetic Resonance in Medicine. 72:1785-1792 (2014).

3. Kaden E, Kruggel F, Alexander D C. Quantitative mapping of the per-axon diffusion coefficients in brain white matter. Magnetic resonance in medicine. 75:1752–1763 (2016).

4. Kaden E, Kelm N D, Carson R P, et al. Multi-compartment microscopic diffusion imaging. NeuroImage. 139 (2016): 346–359.

5. Van Essen D C, Smith S M, Barch D M, et al. for the WU-Minn HCP Consortium. The WU-Minn Human Connectome Project: An overview. NeuroImage. 80 (2013):62-79.

6. Statistical Parametric Mapping, http://www.fil.ion.ucl.ac.uk/spm/

7. Diedrichsen J. A spatially unbiased atlas template of the human cerebellum. NeuroImage. 33.1 (2016): 127-138.

8. Diedrichsen J, Balsters J H, Flavell J, et al. (2009). A probabilistic atlas of the human cerebellum. NeuroImage. 46.1 (2009): 39-46.

9. Van Baarsen K M, Kleinnijenhuis M, Jbabdi S, et al. A probabilistic atlas of the cerebellar white matter. NeuroImage. 124 (2016): 724-732.

10. Zhang H, Schneider T, Wheeler-Kingshott C A M, et al. NODDI: practical in vivo neurite orientation dispersion and density imaging of the human brain. NeuroImage. 61.4 (2012): 1000-1016.

11. Diedrichsen J, Maderwald S, Kuper M, et al. Imaging the deep cerebellar nuclei: A probabilistic atlas and normalization procedure. NeuroImage. 54.3 (2011): 1786-1794.

12. Witter L, De Zeeuw C I. Regional functionality of the cerebellum. Current opinion in neurobiology. 33 (2015): 150-155.