Chronic headaches are a well-documented complaint of shunted hydrocephalic patients ¹. However, it is also one of the signs of shunt malfunction. The current gold standard for determining shunt efficiency is to place an intracranial pressure monitor into the lateral ventricle, but it is a highly invasive procedure.

Cranial compliance (inverse of viscoelasticity) deficiency may cause headaches in some chronically shunted patients with functioning shunts (often with slit or smaller than normal ventricles) ^2-3.This study aims to use a novel, non-invasive imaging technique, Magnetic Resonance Elastography (MRE) ⁴, to investigate the role of brain viscoelasticity in the pathophysiology and symptoms in pediatric hydrocephalus.

A total of 14 shunt-dependent patients (age 15-37, median age 23) who developed hydrocephalus as infants and suffer from chronic headaches were selected. Patients with abnormally large ventricles were excluded. Imaging data from volunteers with no medical history of neurological issues were also collected (N=12). T1W images and MRE data were acquired on a 3T Philips MRI. MRE was performed by inducing a mechanical wave at 30Hz, transmitted through the zygomatic arches.

Phase images were unwrapped using a Laplacian-based phase unwrapping algorithm ⁵ and de-noised using complex dualtree wavelets ⁶ with overlapping group sparsity thresholding ⁷. Tissue shear modulus was then measured using Algebraic Helmholtz Inversion ⁸. Images were motion and distortion corrected using MRE magnitude and field maps respectively. To ensure low vibration amplitude areas were excluded, data was masked > 40% of the maximum amplitude. Image segmentation was performed on high-resolution T1W images to produce CSF, white and gray matter masks using FSL’s FAST tool on brain-extracted images (BET). Using the CSF component, the ventricles were split into frontal, occipital and temporal horns (Fig. 1) and volumetric figures for each segment was calculated. Subjects were then normalized to MNI152 template space using FNIRT. MNI structural atlas masks were used to calculate G* in the frontal, occipital, parietal and temporal lobes (Fig. 2).

All statistical calculations were performed using IBM SPSS statistics package v22. An independent-samples t-test was performed to compare G* values in controls and patients and p-values were FDR corrected for multiple comparisons. Within the patient group, Pearson’s product-moment correlation was calculated to determine the relationship between ventricular volume and G*.

Group comparison between the controls and patients showed a significant decrease in G* in whole gray matter (1054.2 (SD=89.5) Pa vs 824.6 (SD=212.8) Pa, respectively, p=0.0013) and white matter (1407.82 (SD=111.3) Pa vs 1099.33 (SD=262.86) Pa, respectively, p =0.0001) (Fig.4). Similarly for structural lobes, comparisons between controls and patients showed the patient group had decreased viscoelasticity in the frontal (p=0.0321), occipital (p =0.0027), parietal (p =0.0002) and temporal lobes (p=0.0032).

Within the patient group, an inverse correlation was found between G* and ventricular volume, specifically, the right parietal lobe G* with right occipital horn volume (p =0.003, R2=0.5369) (Fig. 5).

Few studies have investigated the effect of hydrocephalus on the viscoelasticity of the brain. A study of normal pressure hydrocephalus (NPH) patients found viscoelasticity to be reduced by 20% in patients compared to controls ⁹. A separate study on NPH patients pre and post-shunt placement found patients’ brain viscoelasticity to be lower than healthy controls pre-surgery, but no difference was found within the patient group pre and post-surgery ¹⁰.

This study found patients to have lower viscoelasticity compared to healthy controls, indicating a marked change in brain biomechanics within treated chronically shunted patients. Additionally, an inverse correlation between regional ventricular volume and viscoelasticity was found, indicating regional effects of ventricular dilation on regional brain viscoelasticity.

This study demonstrates that viscoelasticity is altered in hydrocephalus patients and indicates that it may play a role in clinical symptoms, such as headaches, in chronically shunted patients. MRE is shown to be an effective, non-invasive tool to determine regional viscoelastic differences within the brain and may be a powerful diagnostic tool in hydrocephalus. Further ongoing work includes investigating the relationship between viscoelasticity and headache severity along with other imaging methods such as DTI and MRICP. Once developed, these techniques may assist neurosurgeons in the future to diagnose chronically shunted patients and for guiding the development of alternative or supplementary therapies in hydrocephalus.