Hydrocephalus is a brain condition characterized by enlarged ventricles, due to an excess of cerebrospinal fluid. Although it is known to affect cognition, development, gait, and mood, the impact of hydrocephalus on large-scale functional brain organization is poorly documented. Here, we present results on a single spontaneous occurrence of severe hydrocephalus observed in a 3xTgAD mouse, compared to a control cohort of 3xTgAD littermates.

Resting-state functional magnetic resonance imaging analysis, carried out under light anesthesia, revealed the remarkable presence of functional connectivity (FC) resembling the common mouse resting-state networks (RSNs). Four main components were identified in the hydrocephalic mouse, attributable to the Default Mode network, Salience network, and sensorimotor networks. Characteristic features of the RSNs in the hydrocephalic mouse were found to be well preserved, both in spatial distribution and in FC magnitude, despite the severity of the pathology.

This is the first documented case of resting-state fMRI conducted on a mouse affected by severe hydrocephalus. The surprising presence of resting-state networks was found to be comparable to littermate controls, highlighting a remarkable functional resilience in the hydrocephalic brain.


Anatomical and GE-EPI

Figure 1. Anatomical and GE-EPI images of the hydrocephalic brain; functional networks in a representative control (top panels) and hydrocephalic mouse (bottom panels) and related Z-statistics.

(A) TurboRARE anatomical images acquired with a T2 contrast denote the extent of CSF (bright) with respect to grey (light grey) and white (dark grey) matter.

(B) Single-shot single-echo gradient-echo echo-planar images acquired at 11.75 T present minimal distortions and recapitulate the features of the anatomical images.

(C) The mouse Default Mode network (DMN) presents characteristic correlation in the cingulate (red arrows) and retrosplenial cortices (yellow arrows), whilst anti-correlation is found in the insular cortex (white arrows).

(D) The Salience network (SN) displays correlation within the insula (green arrow) and secondary somatosensory cortex (cyan arrows). In addition, two sensorimotor networks were observed: (E) the anterior portion encompassing the barrel field somatosensory cortex and anterior motor cortex; and (F) the posterior portion encompassing the front and hindlimb, auditory, and visual cortices. Reference-independent components are shown as color-coded Z-scores overlaid on the respective turboRARE anatomical images. The distance from Bregma of each coronal slice is indicated in mm.

(G) Z-statistics for the ROI located in the cingulate cortex, within the DMN: controls 6.16±2.39 and hydrocephalus 10.71.

(H) Z-statistics for the ROI located in the retrosplenial cortex, within the DMN: controls 5.41±2.28 and hydrocephalus 6.85.

(I) Z-statistics for the ROI located in the insular cortex, within the SN: controls 8.99±3.10 and hydrocephalus 12.65.

(J) Z-statistics for the ROI located in the somatosensory cortex, within the SN: controls 8.36±2.13 and hydrocephalus 7.53. n = 17 controls, n = 1 hydrocephalus.


Cerebrospinal fluid (CSF) plays an important role in the regulation of the interstitial fluid of the brain parenchyma and in brain development. Compromised CSF dynamics, due either to excess production, reduced resorption, or altered flow of CSF within the ventricles, affect brain development, leading to hydrocephalus. This condition is characterized by an excess of CSF in the ventricles, impaired cognitive and physical development or, when appearing later in life, cognitive decline, gait disturbance, and urinary incontinence[1]. Congenital hydrocephalus is one of the most common developmental disorders, affecting nearly 1 in 1000 newly born babies[2].

Magnetic resonance imaging (MRI) is a non-invasive technique for diagnosis of hydrocephalus, due to its excellent soft tissue contrast and ability to resolve multiple imaging parameters[3]. Beyond anatomical imaging to resolve the extent of CSF distribution and plan for surgeries, diffusion tensor imaging studies have reported lesions in the white matter in hydrocephalic patients[4][5][6]. While the latter provides an insight into the integrity of the structural tracts across the whole brain, the use of functional (f)MRI is an attractive method to assess the functional integrity of underlying neuronal networks.

In particular, functional connectivity (FC), estimated in a paradigm-free setting, allows for the imaging of several resting-state networks (RSNs) in parallel across the brain. This provides a comprehensive representation of the functional parcellation of healthy and diseased tissue.

To date, the extent of FC across RSNs in hydrocephalic conditions has been marginally investigated. Patients with idiopathic normal pressure hydrocephalus were shown to have decreased DMN activity relative to healthy controls[7]. However, the same DMN activity was counterintuitively increased with symptom severity. A more recent clinical study on resting-state fMRI (rs-fMRI) found disrupted interhemispheric FC in hydrocephalic patients compared to healthy subjects[8]. Currently, there are no established biomarkers for the functional impact of hydrocephalus[8].

Preclinical imaging, using dedicated high-field MRI systems, offers a unique translational platform to image the rodent brain, using analogous protocols as in human. Recent interest in rodent imaging has focused on FC imaging, using rs-fMRI, with the motivation to understand the functional reorganization taking place in murine models of brain disorders ranging from Alzheimer’s disease[9][10][11][12] to autism and depression[13][14].

However, previous research on hydrocephalus has mainly focused on anatomical measures of white matter integrity and myelination[15][16]. To date, there is a lack of research investigating functional consequences in this condition. Preclinical studies on hydrocephalic rodents would contribute to the understanding of the RSNs arrangements in hydrocephalus, and they may help shed a light onto RSNs changes in clinical cases of hydrocephalus[7][8].


The spontaneous occurrence of a hydrocephalic mouse, in a cohort of 3xTg Alzheimer’s disease model (3xTgAD[17]) mice, offers a unique opportunity to investigate the functional consequences of this rare condition in a preclinical setting, at the individual level. We report the observation of remarkably preserved FC in a single case study of a hydrocephalic mouse brain, despite the severity of this case.

Results & Discussion

The hydrocephalic mouse belonged to a cohort of transgenic mice used in a resting-state study on early stages of cerebral amyloidosis. No evident behavioral differences could be observed with respect to gait, feeding, grooming, or mood. Stable physiological parameters were observed for all the animals without a noticeable difference, ensuring optimal recording conditions (Fig. S1): heart rate 336.5±42.3 (Fig. S1A) and blood oxygenation mean 95±4.4 (Fig. S1B).

Anatomical images acquired for the hydrocephalic brain showed a broad expansion of the CSF volume, extending to 44% of the total brain volume (Fig. 1A). Cortical thickness was reduced to a range of 0.5 – 0.8 mm, compared to 0.9 – 1.5 mm in the control animals. The GE-EPI images acquired at high-field presented minimal geometric distortion, ensuring precise mapping of the functional networks (Fig. 1B).

We referred to rules defined in[18] to identify relevant mouse RSNs in the ICA. Four plausible anatomically-relevant components for RSNs were found in the hydrocephalic mouse (Fig. 1C, D, E, F, bottom panels). In the control cohort, 4.14±1.01 plausible components could be identified. The representative components in one control animal are shown as reference (Fig. 1C, D, E, F, top panels).

The incidence map for the four RSNs shows a 70–100% co-occurrence of the major anatomical regions within each RSN investigated, across all control subjects (Fig. S2). This indicates that individual-level RSNs were stable and spatially converged within the control cohort. One network was found to overlap with the cingulate and retrosplenial cortex, together with patterns of anti-correlations with the anterior parietal cortex (Fig. 1C, S2A), therefore, showing a spatial distribution corresponding to the rodent DMN.

The second network identified presented FC extent attributable to the Salience network (SN), specifically, overlapping with the insular and secondary somatosensory cortex (Fig. 1D, S2B). The full extent of the DMN (Fig. 1C, S2A) and SN (Fig. 1D, S2B) are shown together with the 3-dimensional extension of the CSF in both controls (top panels) and hydrocephalic (bottom panels) mice (Fig. S3A, B).

The third and fourth networks corresponded, respectively, to the anterior and posterior sensorimotor networks (Fig. 1E, F; S2C, D), including overlaps with the anterior motor and somatosensory (barrel field) cortex (Fig. 1E, S2C), and the posterior somatosensory (front and hindlimb, auditory, and visual) cortex (Fig. 1F, S2D). Only marginal clusters were observed within sub-cortical networks in the hydrocephalic mouse, while a fifth component, presenting a strong sub-cortical basis, could be observed in 8 of the 17 mice comprising the control cohort. We concluded that the RSN found in the hydrocephalic mouse corresponded to the counterparts found in the control cohort.

To validate these qualitative findings, reference components shown in figure 1 were regressed within the hydrocephalus scan or across all scans in the control animals. Z-statistics, denoting FC strength within a RSN, were extracted using two ROIs located within the DMN and two within the SN (Fig. 1G, H, I, J). Specifically, FC within the cingulate cortex and retrosplenial cortex, two key elements of the DMN (red and yellow arrow in Fig. 1C, respectively), ranged 6.16±2.39 and 5.41±2.28 in the control cohort, while it reached 10.71 and 6.85 in the hydrocephalic mouse (Fig. 1G, H). With respect to the SN, the magnitude of FC within the insula and somatosensory cortex (green and cyan arrow in Fig. 1D, respectively) ranged 8.99±3.10 and 8.36±2.13 in the control group and 12.65 and 7.53 in the hydrocephalic mouse (Fig. 1I, J).

In summary, network strength in the hydrocephalic mouse was found to be within ±1 standard deviation of the estimate in the control group in three of the four ROIs, and ±2 standard deviations in the fourth. We conclude that RSN within the hydrocephalic mouse presented FC strength on par with those found in animals from the control cohort.

This case report showed that, despite a major excess of CSF, the hydrocephalic mouse brain presented a surprising functional resilience. The intrinsic FC organization resembled the characteristics of four RSNs commonly present in the rodent brain, including DMN, SN, and sensorimotor networks. The spatial distribution and the magnitude of the RSNs have been found to be comparable to the littermate controls. Both the DMN and SN have been identified as major hubs underlying neuropathologies and neurodevelopment in human[19][20]. The strong functional resilience to structural insults exhibited, in this case, departs from observations in human reports, where both these networks have been shown to be affected[7][8].

This first observation, made in a spontaneous hydrocephalus case, offers a rare glimpse into the functional organization of the mouse brain following a severe insult. This was made possible with ultra-high field magnets and advanced cryoprobe receiver coils allowing for high sensitivity of the resting fMRI signal. Moreover, previously optimized anesthesia[21] and preprocessing protocols[18] provide additional sensitivity for the robust and reliable detection of RSNs at the individual level in rodents. This is exemplified by the high reproducibility of the RSNs detected at the individual level in the control cohort.


We present here the first fMRI investigation into a spontaneous case of hydrocephalus in a mouse. Results highlight the preservation of cortical functional networks despite substantial ventricular enlargement. This report, beyond being a rare scientific curiosity, provides the first insight into RSNs organization following a severe brain insult and could represent an innovative approach for future research aiming to investigate functional resilience in severe pathological conditions.


The major limitation of this study is the reliance on a single observation, due to the spontaneous nature of the occurrence. Dedicated animal models presenting increased incidence of hydrocephalus would contribute to understanding the full extent of the functional resilience to brain insults, and the parameters associated with it, in rodents[22]. Preclinical functional imaging, in targeted mouse models, would offer a unique opportunity to investigate underlying mechanisms in functional resilience in hydrocephalus.


All applicable international, national, and/or institutional guidelines for the care and use of animals were followed. All procedures performed in studies involving animals were in accordance with the ethical standards of the Institutional Animal Care and Use Committee (A*STAR Biological Resource Centre, Singapore, IACUC #171203). The complete raw dataset for this study can be found on Project ID: Mouse_rest_3xTG.

18 animals (3-month-old males; 1 hydrocephalic mouse and 17 controls, 26–30 g) were triple-transgenic mice (3xTgAD, Jackson Laboratory, Bar Harbor, Maine, USA) bred in-house. Animals were prepared as in[11]. Briefly, anesthesia was induced with 4% isoflurane (50%–50% medical air and Oxygen mixture) to allow the animals to be endotracheally intubated. They were then positioned on an MRI-compatible cradle and artificially ventilated at 90 breaths per minute (Kent Scientific Corporation, Torrington, Connecticut, USA).

A medetomidine bolus (0.05 mg/kg s.c., Dormitory, Elanco, Greenfield, Indiana, USA) was administered, together with muscle relaxant (Pancuronium Bromide, Sigma-Aldrich Pte. Ltd., Singapore), followed by a maintenance infusion (0.1 mg/kg/h s.c.), while isoflurane was reduced to 0.5%. Functional scans were acquired 20 min following bolus to ensure physiological stability of the animals.

The animal temperature was maintained at 37°C. Physiological parameters were recorded with a MRI-compatible oximeter (Kent Scientific Corporation, Torrington, Connecticut, USA): heart rate (HR) measured in beats per minute (bpm) and oxygen saturation (SpO2, measured in %). Data were acquired on an 11.75 T (Bruker BioSpin MRI, Ettlingen, Germany) equipped with a BGA-S gradient system, a linear volume resonator coil for transmission and a 2×2 phased-array cryogenic surface receiver coil.

Images were acquired using Paravision 6.0.1 software. An anatomical reference scan was acquired using a spin-echo turboRARE sequence: field of view (FOV) = 17×11 mm2 (adapted from 17×9 mm2 in control mice to account for enlarged brain), FOV saturation slice masking non-brain regions, number of slices = 28, slice thickness = 0.35, slice gap = 0.05 mm, matrix dimension (MD) = 200×100, repetition time (TR) = 2750 ms, echo time (TE) = 30 ms, RARE factor = 8, number of averages = 2. Rs-fMRI was acquired using a gradient-echo echo-planar imaging (GE-EPI) sequence with the same geometry as the anatomical: MD = 90×60, TR = 1000 ms, TE = 15 ms, flip angle = 50°, volumes = 600, bandwidth = 250 kHz. Field inhomogeneity was corrected using MAPSHIM protocol.

Images were processed using a protocol optimized for the mouse[18]. Images underwent motion correction (mcflirt, FMRIB Software Library v5.0, fsl.fmrib.ox.ac.uk), automatic brain masking (bet), smoothing with a 0.45 mm2 kernel (susan), and a 0.01 Hz high-pass filter (fslmaths). Control animals were registered to a representative anatomical scan drawn from within the control cohort (antsIntroduction, Advanced Normalization Tools,. Lesions in the hydrocephalus brain precluded normalization to the same reference space, instead, an analysis of the hydrocephalus data was carried out in its native space. Within-subject spatial independent component analysis (ICA, melodic) was estimated with automatic dimensionality estimation.

Automated nuisance removal (FIX) was carried out using a classifier trained on an existing dataset acquired previously in-house. Reference components derived from a representative animal were regressed into individual scans using a dual regression framework in order to obtain individual-level representations of these components[23]. Z-statistic, a parameter indicative of RSN strength, was estimated in the dual regression analysis and was extracted using regions-of-interest (ROI). ROIs were selected to be representative of major anatomical regions within the respective RSN. Descriptive statistics are given as mean ±1 standard deviation.

Funding Statement

This work was supported by the University of Manchester and A*STAR Research Attachment Programme (ARAP), which is co-funded through the University of Manchester, Faculty of Biology, Medicine and Health Doctoral Academy, and Singapore Bioimaging Consortium (SBIC), A*STAR, Singapore. Animal work was supported by Laboratory of Bio-optical Imaging (LBOI), SBIC A*STAR, Singapore.


The authors would like to thank Richard Clayton for critical reading of the manuscript.

Conflict Of Interest

The authors declare no conflicts of interest.

Ethics Statement

All applicable international, national, and/or institutional guidelines for the care and use of animals were followed. All procedures performed in studies involving animals were in accordance with the ethical standards of the Institutional Animal Care and Use Committee (A*STAR Biological Resource Centre, Singapore, IACUC #171203 ).

No fraudulence is committed in performing these experiments or during processing of the data. We understand that in the case of fraudulence, the study can be retracted by ScienceMatters.