Brain tissues, grey and white matter, are among the most compliant biological tissues, and are highly susceptible to damage, with the global annual incidence of traumatic brain injury estimated to range from 27-69 million^1,2. These injuries, caused by everyday occurrences including falls, contact sports, and road injuries, are stratified into varying degrees of severity, with mild traumatic brain injury being significantly underdiagnosed.

Damage sustained by the brain is visible in cases of moderate, severe, and some mild traumatic brain injury using clinical scanning (magnetic resonance imaging, computed tomography); however, concussion, a form of mild traumatic brain injury, requires an absence of visible damage for diagnosis^3,4.

Clinical management of patients presenting with concussion is as such informed by patient symptom presentation, without an understanding of the magnitude of microstructural damage the patient has suffered. This lack of insight contributes to significant heterogeneity in patient recovery times, with adolescents, children, and women typically taking longer to recover.

The leading hypothesis for the mechanism of this damage is shear delamination between the grey and white matter causing delamination and microtears at this interface and causing diffuse axonal injury^4,5.

Advanced imaging techniques can detect these types of damage, though are not typical in clinical practice^6,7, and further verification of this hypothesis for the damage mechanism with these methods is required. Though this damage mechanism is not well understood, concussion has been demonstrated to be a significant risk factor for the development of neurodegenerative conditions such as Parkinson’s disease, Alzheimer’s disease, and particularly chronic traumatic encephalopathy (CTE)^8,9.

Additionally, second-impact syndrome, where a patient suffers a second impact within the concussion healing period, has been associated with mortality rates of 50%¹⁰, raising questions about the transient effect of impact on the mechanical properties of brain tissues.

Characterisation of the mechanical behaviour of brain tissues at a bulk scale has increased in recent decades, on human tissues and that of many species of animals^11,12.

The compressive, tensile, shear, rheological and viscoelastic behaviours of these tissues have been described to varying degrees with differing experimental parameters, leading to a high degree of heterogeneity in published results.

The link between tissue microstructure and bulk tissue behaviour has not been well explored, with no insight into the mechanical properties of the microstructural components, such as various cell types, fibres, and extracellular matrices, which would be beneficial for the understanding of disease states and the brain’s response to injury.

Factors that have notably been overlooked in published literature are regional variance in mechanical properties within cortical grey and coronal white matter, within which homogeneity has been assumed^13,14, the fracture behaviour of brain tissues, the mechanical properties of the grey-white matter interface, and the compressibility of brain tissues, which this study aims to address.

Materials and methods

Sample preparation

Figure 1: Breakdown of the sample preparation and experimental procedure; a) Regions of ovine brain (n=35) used, b) Isolated GM and WM samples (n=614), with images of the experimental setup for unconfined compression, tension and simple shear. c) Interface samples, alongside descriptions of mode I and II fracture¹⁵, performed using the same tension and shear setup as in b). d) Experimental setup for the volumetric analysis, as well as the loading paths for viscoelastic testing, each detailed below (combined n=288). Sheep illustration adapted from Servier Medical Art¹⁶.

Fresh ovine brains were obtained from a local abattoir, refrigerated at 4℃ and kept in sealed, airtight bags to minimise tissue degradation and dehydration. Time to test post-mortem was minimised, with tests being carried out an average of 24 hours post-mortem up to a maximum of 60 hours.

Any residual meningeal tissue was removed before dissection, and during dissection, tools were sprayed with phosphate-buffered saline (PBS) to minimise adherence to tissue, while tissue was intermittently sprayed to prevent tissue dehydration, minimising absorption of PBS by the tissues.

Samples of isolated grey and white matter were excised (Fig. 1 b)), ranging in height from ~2-5 mm, with an even distribution of grey and white either side of the grey-white interface prioritised in interface samples (Fig. 1 c)).

The region of origin of each sample (Fig. 1 a)) was noted, as the lobe of origin for cortical grey matter or corona radiata white matter, or as deep brain grey matter (thalamus, basal ganglia) or corpus callosum white matter when collected from these deep brain regions.

Sample orientation was additionally recorded to the outer brain surface for cortical grey matter and coronal white matter, or in the case of the corpus callosum to the orientation of axonal fibres, and in deep brain grey matter, to the closest ventricular surface. Between dissection and testing, samples were stored at room temperature in sealed containers and were sprayed with PBS prior to removal to minimise adhesion to the containers.

Unconfined compression

Samples were placed on a base plate, adjusting crosshead position to achieve maximal contact between samples and the top platen (Fig. 1 b)) without preload, noting this grip-to-grip separation as sample height, which was used to calculate crosshead displacement (δ= ε∙h ) and speed (δ= ε∙h ) to apply a strain of -50% at 25% /s, before returning to the initial position. Here, ε  is applied strain, h  is sample height, and ε  is applied strain rate. Force-displacement curves were produced for each sample through loading and unloading phases.

Tension

Samples were first fixed to the top platen, before being lowered and fixed onto the base plate, in both cases using a cyanoacrylate adhesive, maintaining maximum sample-platen contact and a load-free state during adhesive curing. A strain of 50% was applied at 25% /s before returning to the initial position, with samples inspected throughout the tests for separation from base plate or platen, or for connectivity between adhesive at sample top or bottom, with these incidences being considered failures and being excluded from further analysis.

Simple shear

Samples were fixed between the left platen and the right base plate in a custom-built rig using a cyanoacrylate adhesive, allowing the glue to harden and fixing the base plate in place, such that the sample was not subject to any deformation (Fig. 1 b)). Sample height (distance between left platen and right base plate) was measured using a Vernier Calliper, and used to calculate the applied displacement (δ=tanγ∙h ) and velocity (δ=tanγ∙h ) to apply 50% (0.5 rads) simple shear strain (γ ) at a shear strain rate (γ ) of 25% /s (0.25 rads/s) from the provided formulae. Again, force was measured over displacement through loading and unloading, and samples were monitored for separation from either fixed side throughout each test.

Volumetric compressibility

Compressive and tensile loading was carried out as above, with extensometer cameras recording orthogonal cross-sectional area measurements (Fig. 1 d). A semi-automated MATLAB script was used to process these cross-sections into normalised volume measurements at the initial position and peak displacement, from which sample volume change in loading was calculated.

Viscoelasticity

Compression, tension, and shear testing was performed as above at three load rates, 25% /s, 100% /s, and 500% /s, holding samples at ± 50% strain for 120 s, described in Fig. 1 d).

Fracture

For interface samples, one tissue type was fixed to the base plate, and another to the platen, with the interface being centred between the two. Videos recording was used to identify fracture initiation points and propagation pathways (Fig. 1 c)). Sample preparation for mode I (tensile) fracture testing was the same used in tensile testing, with an applied strain of 300%, while for mode II (shear) fracture, methods were the same as in shear testing, with an applied shear strain of 140% (1.4 rads).

Results

Figure 2: Regional behaviour of fresh GM and WM in unconfined compression and tension, and in simple shear. The region of interest in each column is illustrated, with each plot comparing the behaviour of GM and WM in the region and loading mode at hand. Data is presented as mean (indicated by markers), and standard deviation (indicated by surrounding colour bands), with figure legends denoting the sample numbers used for each matter type, the matter types indicated in the compression subplots, with purple curves conventionally indicating GM, and green curves conventionally indicating WM.

Initially, due to logistical constraints, frozen tissue (stored at 20℃) was used in compression and tensile tests, after one or two freeze-thaw cycles, which was discontinued once a regular supply of fresh tissue was established, where we found that the freeze-thaw process significantly degrades the stiffnesses of both grey and white matter in both tension and compression, rendering them inappropriate for use in experimental characterisation. All subsequent testing was therefore carried out solely with fresh tissue.

Fig. 2 presents stress-strain curves for each region of grey and white matter in unconfined compression, tension and shear to 50% at 25% /s. Both grey and white matter are strain stiffening in compression with white matter typically being stiffer than grey throughout, with little significant difference within the grey or white matter.

In tension, grey matter is typically initially stiffer than white matter, with some variation in the transition strain to a stress plateau and softening region, while white matter typically transitions earlier than grey, with a less significant drop in stiffness. There is insignificant variation within white matter generally, and within cortical grey matter, from which deep brain grey matter behaves distinctly.

Both grey and white matter exhibit bilinear-type behaviour in each region in simple shear, each softening below ~5% shear strain, with little variance here between cortical grey matter, though deep grey matter is distinct, while heterogeneity evident not only between coronal white matter and the corpus callosum, but additionally within the corona radiata.

As previously discussed, cortical grey matter and coronal white matter are often grouped in literature, differentiated from deep brain regions of grey and white matter. Fig. 3 combines the coronal and cortical regions in Fig. 2, contrasted with the deep brain grey and corpus callosum white matter, due to the significant homogeneity observed in these regions.

Figure 3: Regional variance in mechanical behaviour of cortical grey and coronal white matter, contrasted to the deep brain grey matter, and the corpus callosum, with these regions indicated on an ovine brain cross-section.

Figure 4 a) shows that the tangent moduli of both grey and white matter are rate dependent in the strain ranges tested, with the exception of grey matter in tension, experiencing higher magnitude stress values at higher load rates, though peak stress at 50% strain was not significantly affected by load rate. 

Figure 4: a) Behaviour of GM and WM in loading at 25%, 100%, and 500% /s to ± 50% strain. b) Stress relaxation behaviour of GM and WM at each of these strain rates for both tension and compression, shown as residual stress magnitude, with bar charts highlighting relevant statistically significant differences at 1 s, 5 s, and 120 s, where colour denotes the strain rate, and bar fill denotes the matter type.

The relaxation behaviour detailed in Fig. 4 b) shows that both tissues experienced stress relaxation of more than 80% over 120 s. The magnitude of this relaxation was found to be initially rate dependent for grey matter, with a higher magnitude of stress relaxation in samples loaded at higher rates, though as time approached 120 s, this behaviour faded, the relaxation for each rate converging over time, while in white matter, this phenomenon was seen at each time point.

Figure 5: Volume change measurements for grey and white matter in compression and tension, with representative images used in measurement shown, with statistically significant differences indicated. Volume change is here defined as ΔV/V0 , the difference between loaded volume and initial volume, divided by the initial volume.

Fig. 5 presents the volume change which occurred in both tissues in unconfined compression and tension, with grey matter being significantly more volumetrically compliant than white matter in both loading modes, and both tissues experiencing a higher magnitude of volume change in tension than compression.

Figure 6: Traction-displacement curves for mode I and II fracture for grey, white, and interface samples, as well as the total fracture energy for each sample type for each fracture mode.

The traction-displacement curves in Fig. 6 show that the fracture initiation stress, or nucleation stress, is similar between grey and white matter in mode I and II, while being significantly lower for interface samples in both fracture modes. White matter is more ductile than grey matter in both modes, while interface samples are similarly brittle to grey matter samples in mode I, and similarly ductile to white matter samples in mode II. These variances in ductility are reflected in the fracture energies for each sample type, where white matter has the highest fracture energy in each mode, while grey matter fracture energy is only higher than that of interface samples in mode I.

Figure 7: Interface samples before and during mode I and mode II fracture, demonstrating tissue delamination not only at the grey-white interface, but also within the bulk grey and white matter.

Examination of the recordings of mode I and II fracture testing of grey-white interface samples revealed that tissue fracture was initiated and/or propagated in the bulk grey and white matter in several samples, not solely occurring at the boundary between the tissues, as shown in Fig. 7, motivating the isolated grey and white matter fracture testing in Fig. 6.

Discussion

Demonstrating the degrative effect of freeze-thaw cycles on the mechanical properties of brain tissues, rendering them inappropriate for mechanical characterisation, contradicts studies finding negligible differences in tissue behaviour after freezing^17,18, though these studies may have overlooked the degradation due to considering brain tissues as homogenous.

Published literature exploring the mechanical characterisation of brain tissues is highly heterogenous¹², with the results presented here¹¹ for tension, compression, and shear, and their viscoelastic behaviours aligning highly with some studies, and running contrary to others. This can to a large extent be attributed to the variation in experimental methods in the literature, in particular with regards to strain rates in loading, tissue preparation methods, and an assumption of homogeneity within grey and white matter, or within 'brain tissue' as a whole.

The apparently insignificant variation of mechanical properties within the cortex and corona radiata supports their assumed homogeneity in literature, though the variation observed within the corona radiata in shear warrants further investigation, ideally through the development of an understanding of the mechanical properties of the microstructural components of both grey and white matter.

The observed viscoelastic behaviour demonstrates an increase in the tangent moduli of both tissues, with the exception of grey matter in tension, in clinically relevant strain ranges, an effect which is observed to compound at higher strain rates in literature.

The loading rate dependence of relaxation magnitude seen in grey and white matter, and the transient evolution of this phenomenon in grey matter, suggest differing mechanisms of stress relaxation between the tissues. This could have significant implications for sustained loading, such as the swelling and increased pressure typical post-concussion, where the speed of loading may have no effect on the long-term relaxation of grey matter, while it may for white matter.

The demonstrated volume changes of grey and white matter in both tension and compression have significant implications for in vitro and in silico models of the brain, where tissue incompressibility is a common assumption. Incompressible models are unable to accurately capture the behaviour of compressible tissues under increased pressure and tissue swelling due to breakdown of the blood brain barrier or haemorrhage, limiting their insight into the transient effect of impacts on the brain and the mechanisms driving second impact syndrome.

Additionally, the heterogeneity of compressibility between tissue types and loading modes is notable and warrants further investigation into this behaviour. It should be noted in characterising this behaviour that brain tissues should be considered poroviscoelastic, and that the cerebrospinal fluid should be considered in further characterising and modelling this tissue compressibility.

The complex fracture initiation and propagation behaviour observed in the novel mode I and II fracture testing of grey-white interface samples challenges the clinical hypothesis, showing that damage is not limited to this interface, which again will have significant implications for computational models of damage at or around this interface.

Though fracture was initiated in and/or propagated into the grey and white matter from their interface, isolated grey and white matter were found to have nucleation stresses significantly higher than the interface, suggesting heterogeneity of the mechanical properties of grey and white matter approaching this interface, once more motivating investigation into the heterogeneity of the microstructure of grey and white matter, and the corresponding effects on their mechanical behaviours.

Further detail and in-depth discussion on this work is available in its published form at Annals of Biomedical Engineering (https://doi.org/10.1007/s10439-025-03877-x). 

Author: Conal Sheridan.

References

1) S. L. James et al., ‘Global, regional, and national burden of traumatic brain injury and spinal cord injury, 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016’, Lancet Neurol., vol. 18, no. 1, pp. 56–87, Jan. 2019, doi: 10.1016/S1474-4422(18)30415-0.

2) M. C. Dewan et al., ‘Estimating the global incidence of traumatic brain injury’, J. Neurosurg., vol. 130, no. 4, pp. 1080–1097, Apr. 2019, doi: 10.3171/2017.10.JNS17352.

3) J. P. Guenette, M. E. Shenton, and I. K. Koerte, ‘Imaging of Concussion in Young Athletes’, Neuroimaging Clin. N. Am., vol. 28, no. 1, pp. 43–53, Feb. 2018, doi: 10.1016/j.nic.2017.09.004.

4) F. Alisafaei, Z. Gong, V. E. Johnson, J.-P. Dollé, D. H. Smith, and V. B. Shenoy, ‘Mechanisms of Local Stress Amplification in Axons near the Gray-White Matter Interface’, Biophys. J., vol. 119, no. 7, p. 1290, Oct. 2020, doi: 10.1016/j.bpj.2020.08.024.

5) L. Pankatz et al., ‘Adverse Outcome Following Mild Traumatic Brain Injury Is Associated with Microstructure Alterations at the Gray and White Matter Boundary’, J. Clin. Med., vol. 12, no. 16, p. 5415, Aug. 2023, doi: 10.3390/jcm12165415.

6) S. S. G. Brown, K. Dams-O’Connor, E. Watson, P. Balchandani, and R. E. Feldman, ‘Case Report: An MRI Traumatic Brain Injury Longitudinal Case Study at 7 Tesla: Pre- and Post-injury Structural Network and Volumetric Reorganization and Recovery’, Front. Neurol., vol. 12, p. 631330, May 2021, doi: 10.3389/fneur.2021.631330.

7) D. T. Vo et al., ‘Diffuse axonal injury: a case report and MRI findings’, Radiol. Case Rep., vol. 17, no. 1, pp. 91–94, Jan. 2022, doi: 10.1016/j.radcr.2021.10.006.

8) B. L. Brett, R. C. Gardner, J. Godbout, K. Dams-O’Connor, and C. D. Keene, ‘Traumatic Brain Injury and Risk of Neurodegenerative Disorder’, Biol. Psychiatry, vol. 91, no. 5, pp. 498–507, Mar. 2022, doi: 10.1016/j.biopsych.2021.05.025.

9) J. Mez et al., ‘Clinicopathological Evaluation of Chronic Traumatic Encephalopathy in Players of American Football’, JAMA, vol. 318, no. 4, pp. 360–370, Jul. 2017, doi: 10.1001/jama.2017.8334.

10) S. D. Stovitz, J. D. Weseman, M. C. Hooks, R. J. Schmidt, J. B. Koffel, and J. S. Patricios, ‘What Definition Is Used to Describe Second Impact Syndrome in Sports? A Systematic and Critical Review’, Curr. Sports Med. Rep., vol. 16, no. 1, p. 50, Feb. 2017, doi: 10.1249/JSR.0000000000000326.

11) C. Sheridan and J. Concannon, ‘Mechanical Characterisation of Grey, White, and Interface Behaviour in the Brain’, Ann. Biomed. Eng., vol. 53, no. 12, pp. 3400–3414, Dec. 2025, doi: 10.1007/s10439-025-03877-x.

12) S. Budday, T. C. Ovaert, G. A. Holzapfel, P. Steinmann, and E. Kuhl, ‘Fifty Shades of Brain: A Review on the Mechanical Testing and Modeling of Brain Tissue’, Arch. Comput. Methods Eng., vol. 27, no. 4, pp. 1187–1230, Sep. 2020, doi: 10.1007/s11831-019-09352-w.

13) S. Budday et al., ‘Mechanical characterization of human brain tissue’, Acta Biomater., vol. 48, pp. 319–340, Jan. 2017, doi: 10.1016/j.actbio.2016.10.036.

14) W. Kang, Q. Li, L. Wang, Y. Zhang, P. Xu, and Y. Fan, ‘Systematic analysis of constitutive models of brain tissue materials based on compression tests’, Heliyon, vol. 10, no. 18, p. e37979, Sep. 2024, doi: 10.1016/j.heliyon.2024.e37979.

15) J. Brunet, B. Pierrat, and P. Badel, ‘Review of Current Advances in the Mechanical Description and Quantification of Aortic Dissection Mechanisms’, IEEE Rev. Biomed. Eng., vol. 14, Jan. 2020, doi: http://dx.doi.org/10.1109/RBME.2019.2950140.

16) Sheep. Accessed: Apr. 15, 2025. [Image]. Available: https://smart.servier.com/smart_image/sheep/

17)  A. Tamura, S. Hayashi, I. Watanabe, K. Nagayama, and T. Matsumoto, ‘Mechanical Characterization of Brain Tissue in High-Rate Compression’, J. Biomech. Sci. Eng., vol. 2, no. 3, pp. 115–126, 2007, doi: 10.1299/jbse.2.115.

18)  P. Devananthan, J. Zwirner, P. Docherty, B. Ondruschka, and N. Kabaliuk, ‘Effect of Freezing on Mechanical Properties of Ovine Brain Tissue’, Feb. 21, 2025, Social Science Research Network, Rochester, NY: 5139433. doi: 10.2139/ssrn.5139433.