Target audience: Physicians and scientists interested in cerebral tissue properties.

Introduction: In this study, two imaging modalities sensitive to biophysical tissue properties, magnetic resonance elastography (MRE) and arterial spin labeling (ASL), were used to compare the mechanical properties of brain tissue with average regional perfusion. Previous studies showed significant differences in mechanical properties and capillary density between white and gray matter in human brains: white matter was observed to exhibit higher stiffness than gray matter (1), while average perfusion was higher in gray matter than in white matter (2). Fluid transport may influence the viscoelastic properties of the in vivo human brain (3). We therefore investigated whether the magnitude of the complex shear modulus |G*| of deep gray matter correlates with brain perfusion by analyzing both parameters in the putamen, globus pallidus, hippocampus, thalamus, nucleus accumbens and amygdala. Each quantified volume per subject was split into sub-volumes corresponding to the locations of all six regions of interest. |G*| and CBF were averaged across each region. Group analysis was based on mean values for each region and subject.

Methods: 14 healthy male subjects were investigated in a 3T MRI scanner by 3D anatomical scans (MPRAGE), multi-frequency MRE (MMRE) and ASL. The following parameters were used for MMRE: 40 transversal slices of 2 mm isotropic resolution, TE=82ms, TR=8.49s, 30, 40, 50Hz vibration frequency, 8 acquisitions over a wave cycle; for the pseudo-continuous ASL sequence (4): label duration 1.5s, post-label delay=1.2s, 120 volumes (60 label, 60 control) with 32 ascending transversal slices of 2.5 mm isotropic resolution (10% gap), TE=18ms, TR=4.7s. All images were post-processed by SPM12 for correcting motion and distortion artifacts before parameter quantification. The volumes were then normalized to fit the dimensions of the tissue probability maps and the Neuromorphometrics atlas of brain regions in MNI152 space to allow for inter-subject statistics. Voxels in the vicinity of cerebrospinal fluid (CSF) were eliminated to avoid the influence of partial volume effects on the analysis. Six subcortical regions of interest (ROI) were chosen for their high ratio of voxels remaining after this partial volume filter. MMRE data processing was based on MDEV inversion as described in detail in (5), which provides a lumped parameter of viscoelasticity, |G*| in kPa. For quantification of the cerebral blood flow (CBF, in ml/100g/min) control-label pairs in ASLtbx (6) were used.

Results: When considering all analyzed regions we observed no correlation between |G*| and CBF (r=0.65, p=0.16). However, the situation changed by deselecting the striatum (represented by the putamen and nucleus accumbens) which had marked higher |G*| values than the other four regions [(1.49±0.08_vs_1.10±0.08)kPa] as well as more CBF [(48±3_vs_37±3)ml/100g/min]. Figure 1 shows that |G*| and CBF are reciprocally correlated in all regions but the striatum (r=-0.77, p=0.04). Pooling data of both hemispheres, the reciprocal correlation between |G*| and CBF improves to (r=-0.97, p=0.02). A decrease of |G*| with increased CBF is observed in striatal regions (putamen vs. n. accumbens: [(1.58±0.25 vs. 1.41±0.11)kPa], [(45±10 vs. 50±10)ml/100g/min]) which resembles the reciprocal correlation of |G*| with CBF found in other deep gray matter regions. Intriguingly, the two clusters in the |G*|-CBF space (hippocampus, thalamus, g. pallidus, amygdala vs. putamen, n. accumbens) reveal distinct mechanical-vascular properties in deep gray matter regions which are not perceivable on either of both parameters axes.

Discussion: We separated the striatum from other deep gray matter regions in our correlation analysis which is justified by its distinct microstructure and neurological function. As the primary recipient of cortical input and distributor of excitatory and inhibitory impulses among the other subcortical regions, the striatum might be considered to be the “control center” of neuronal activity in deep gray matter regions. Within the striatum, medium-spiny neurons make up 95% of the neuronal population (about 100 million). They are covered in large and extensive dendritic trees, suggesting an especially tight network of dendrites. This might serve as an explanation for higher stiffness as well as higher average perfusion to cover the oxygen requirement of a high neuronal density within the striatum.

Conclusion: An inverse correlation of tissue stiffness (|G*|) and average perfusion (CBF) is seen in deep gray matter when accounting for structurally and functionally distinct areas. In the |G*|-CBF space, globus pallidus, hippocampus, thalamus and amygdala clearly cluster from putamen and nucleus accumbens which indicates their anatomical differences in network density and vasculature. These differences in the microstructure between the striatum and other deep gray matter regions are not perceivable by MRE or ASL alone.