A

vaifMsbt©

K

1

ibeviic

disw[rsm

1d

Available online at www.sciencedirect.com

Medical Engineering & Physics 31 (2009) 392–399

Optical characterization of acceleration-induced strain fields ininhomogeneous brain slices

C. Lauret, M. Hrapko, J.A.W. van Dommelen ∗, G.W.M. Peters, J.S.H.M. WismansMaterials Technology Institute, Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven, The Netherlands

Received 19 February 2008; received in revised form 17 March 2008; accepted 16 May 2008

bstract

The aim of this study was to measure high-resolution strain fields in planar sections of brain tissue during translational acceleration to obtainalidation data for numerical simulations. Slices were made from fresh, porcine brain tissue, and contained both grey and white matter as wells the complex folding structure of the cortex. The brain slices were immersed in artificial cerebrospinal fluid (aCSF) and were encapsulatedn a rigid cavity representing the actual shape of the skull. The rigid cavity sustained an acceleration of about 900 m/s2 to a velocity of 4 m/sollowed by a deceleration of more than 2000 m/s2. During the experiment, images were taken using a high-speed video camera and Von

ises strains were calculated using a digital image correlation technique. The acceleration of the sampleholder was determined using the

ame digital image correlation technique. A rotational motion of the brain slice relative to the sampleholder was observed, which may haveeen caused by a thicker posterior part of the slice. Local variations in the displacement field were found, which were related to the sulci andhe grey and white matter composition of the slice. Furthermore, higher Von Mises strains were seen in the areas around the sulci.

2008 IPEM. Published by Elsevier Ltd. All rights reserved.

orrelat

pmrlo

modbl

owsa

eywords: Brain tissue; Strain concentration; Acceleration; Digital image c

. Introduction

Annually 1.4 million people sustain a traumatic brainnjury (TBI) in the United States, of which 20% is causedy vehicle traffic accidents [1]. Although vehicles are alreadyquipped with belts and airbags, even more sophisticated pre-entive measures are needed to further reduce this number ofnjuries. The development of these measures can be based onnjury predictions with numerical head models, by simulatingrash situations.

Many numerical head models have been developed [2–7],iffering in the constitutive models used and the level of detailn the modelled geometries of the brain and the skull. Con-titutive models describe the mechanical behaviour of tissue,hich is nonlinear and visco-elastic in the case of brain tissue

8]. Moreover, brain tissue may be anisotropic and show inter-

egional variations. The quality of numerical head modelimulations depends partly on the ability of the constitutiveodel to describe this complex mechanical behaviour, and

∗ Corresponding author. Tel.: +31 40 247 4521; fax: +31 40 244 7355.E-mail address: j.a.w.v.dommelen@tue.nl (J.A.W. van Dommelen).

tssigo

350-4533/$ – see front matter © 2008 IPEM. Published by Elsevier Ltd. All rightoi:10.1016/j.medengphy.2008.05.004

ion; Heterogeneity

artly on the modelled geometry. Therefore, the constitutiveodel and the head model need to be validated in order to give

eliable and representative injury predictions. However, onlyimited experimental data exist because of the inaccessibilityf the cranium.

Pudenz and Shelden [9] measured the deformation in aacaque brain through a cranial window. Although this was

ne of the first successful strain measurements of the brainuring acceleration, only the deformation of the surface coulde observed. Furthermore, in the past two decades the use ofiving animals is being restricted due to legislations.

To validate model predictions, Brands et al. [3] used bothpen and closed cylindrical cups filled with silicon gel, whichere subjected to transient rotational acceleration. In both

etups, the gel response was measured using optical markersnd a high-speed video camera. Ivarsson et al. [10,11] studiedhe natural protection of the brain, also using gel and high-peed tracing of markers. More specifically, lateral ventricle

ubstitutes were included in this physical model to investigatef these structures give strain relief during head impact. Mar-ulies et al. [12] and Meaney et al. [13] recorded the motionf grid patterns painted on gel inside animal and human skulls

s reserved.

eering

dataimb

tcel

btdiofttibb0

iaTiwanawawmm

idrt

atbfistp

2

(lbpsnfpmwratwtsdmaRttw

[sAA

C. Lauret et al. / Medical Engin

uring angular acceleration. The overall deformation patterns a result of rotation was compared to the pathological por-rait of diffuse brain injury, as determined from animal studiesnd autopsy reports. Although gel-based setups can providensight in the global mechanical behaviour, they are unable to

imick the local brain structures like grey and white matteroundaries and the folding structure of the cortex.

Hardy and colleagues used neutral density accelerome-ers [14] and targets [15] to measure brain motion in humanadavers via high-speed X-ray imaging during angular accel-ration. The spatial resolution of these measurements was tooow to estimate local strain fields.

Bayly et al. [16] used MRI to measure the deformation ofrain tissue induced by mild acceleration in human volun-eers. Strains of 0.02–0.05 were typical during the occipitaleceleration, and compression in anterior regions and stretch-ng in posterior regions were observed. Moreover, the motionf the brain appeared to be constrained by structures at therontal base of the skull. A drawback of volunteer tests is thathey can only be performed at a level well below the injuryhreshold. The same method was used to obtain strain fieldsn the brain of a perinatal rat [17,18]. In these experiments therain was not accelerated, but the flexible skull was indentedy 2 mm and during 21 ms. Lagrangian strains of more than.20 at strain rates exceeding 40 s−1 were observed.

The main objective of this study was to develop an exper-ment in which a realistic crash situation was mimickednd that can be used to validate numerical head models.herefore, high-resolution strain-fields were measured in

nhomogeneous, planar sections of fresh porcine brain tissueith a complex and detailed geometry, during translational

cceleration normally occurring in crash situations. The pla-ar brain samples contained both grey and white matters well as the complex folding structure of the cortex. Itas hypothesized that these inhomogeneities influence the

cceleration-induced strain pattern. Since fresh porcine tissueas used, the experiment was conducted within 6 h post-ortem in order to prevent any time-related changes of theechanical behaviour of the tissue [19].During translational acceleration, the samples were

mmersed in artificial cerebrospinal fluid (aCSF) to preventehydration of the tissue. Furthermore, in order to obtain aepresentative model of the relative motion of the brain insidehe skull during acceleration, the sample together with the

a

(c

Fig. 1. Lateral (a) and top (b) view of a brain

& Physics 31 (2009) 392–399 393

CSF layer was encapsulated in an almost rigid cavity withhe shape of the actual brain slice. The motion was recordedy a high-speed video camera and displacements and strainelds were obtained using digital image correlation. Thetrain patterns obtained were qualitatively compared withhe grey and white matter composition of the slice and theositions of the sulci.

. Materials and methods

Planar brain slices were prepared from female pig brainsDutch landrace hybrid) of 4–6 months old, obtained at aocal slaughterhouse. During transport and preparation, therains were cooled and stored in porcine based aCSF [20] torevent dehydration and to slow down the degradation andwelling process of the tissue. Sagittal slices of 4 mm in thick-ess were made using a standard slicing machine (Bizerba)rom the region about 2 cm outwards from the mid-sagittallane, see Fig. 1. This thickness was chosen as a compro-ise between the influence of inertial effects which increasesith thickness and the effect of through-thickness geomet-

ical variations for which a thin slice is desired. The medialnd lateral sides of the slices differed in grey and white mat-er distribution and in the geometry of the sulci. However,hen comparing corresponding sides of different brain slices,

he geometries and distributions of matter were qualitativelyimilar. The experimental procedure was carried out for twoifferent para-sagittal brain slices and multiple accelerationeasurements were conducted on each side of the slice with

t least 1 min duration between subsequent measurements.epetitive mechanical loading of brain tissue was found not

o affect its constitutive response in a previous study [8]. Allhe measurements were carried out at room temperature andere completed within 6 h post-mortem.In order to use a digital image correlation technique

22–24], a fine, random speckled pattern was applied on theurfaces of the slice, using matt black Enamel (Alkyd resin,irbrush email color, Revell GmbH) and an airbrush (Aztek,4305, The Testor Corporation, USA), see Fig. 2(a). The

verage dot diameter of the pattern was about 0.5 mm.The sample holder (Fig. 2(b)) consisted of three plates

125 mm ×100 mm); two outer plates of transparent poly-arbonate and a the middle plate of black polyvinylchloride

, with indicated sample region [21].

394 C. Lauret et al. / Medical Engineering & Physics 31 (2009) 392–399

fication

(aCarpfditst

anaborsdvs(vo

Fs

feclmmdiaoctTautrctam

Fig. 2. (a) A brain slice with the applied speckled pattern and a magni

PVC), each with a thickness of 5 mm. A cavity with thectual shape of the slice was created in the middle plate byNC (Computer Numerical Control) milling, but with andditional space of 2 mm around the brain slice for a sur-ounding aCSF layer. Moreover, the thickness of the middlelate was 1 mm larger than the thickness of the slice, allowingor natural aCSF layers. For the CNC milling, a numericalescription of the sample contour was obtained from a digitalmage of the slice. To prevent the brain tissue from adhering tohe inner surface of the plates, they were coated with a Teflonpray. Because of gravity, the slice rested on the bottom ofhe cavity.

The experimental setup is schematically shown in Fig. 3nd consisted of an air spring (Amoteck, type Deamv) con-ected to the sampleholder. The sampleholder sustained ancceleration of about 900 m/s2 to a velocity of 4 m/s followedy a deceleration of more than 2000 m/s2. The accelerationf the sampleholder was measured using digital image cor-elation of a patch on the surface of the sampleholder. Theampleholder was initially moved in the posterior–anteriorirection. Digital images were collected with a high-speedideo camera (Phantom v9.0, Vision Research Inc., New Jer-

ey) at a frame rate of 1600 Hz. The 8-bit grey-scale images1440 × 952 pixels) were cross-correlated (Aramis software,5.4.1-5, GOM Optical Measuring Techniques mbH) with anptimized facet size of 15 × 13 pixels.

ig. 3. Schematic illustration of the experimental setup, with an airspring,ampleholder and high-speed video camera.

tasisMlsii

ε

wd

m

of this pattern; (b) the sampleholder with a milled out sample cavity.

Digital image correlation (DIC) is an optical methodor displacement and strain measurements, widely used inngineering and scientific research. However, reported appli-ations of DIC for the evaluation of biological materials areimited because of practical challenges like the necessity ofaintaining hydration [25], whereas surfaces are normallyatted to prevent reflections in the images. Images of the

eformed stage are compared with the undeformed referencemage or the previous image and, from that, the displacementnd deformation are calculated. This is based on recognitionf surface structures. If the object has only a few surfaceharacteristics, surface preparation is needed, like applica-ion of a random pattern that deforms with the object itself.he reference image is divided into a regular grid (facets)nd then matched to the consecutive images, based on thenique grey level pattern of each facet. The displacement ofhe middle point of each facet is determined, with the accu-acy depending on the facet size. Enamel paint was used toreate a speckled pattern because of its good adherence tohe wet surface of the tissue. Although with this technique,

successful pattern correlation was obtained, the accuracyay be less than as reported for specimens that do not need

o be hydrated (accuracy of 0.1%, Aramis, GOM [26]). Theccuracy of the image correlation was investigated for thispecific experimental situation. Two experiments were donen which a known rigid body translation was applied to theampleholder containing a tissue slice. The maximum Von

ises strain of the surface obtained for this rigid body trans-ation was found to be 0.4%. For the equivalent Von Misestrain field, the out-of-plane strains were computed assumingncompressibility of the tissue. This equivalent strain measures defined as:

¯ =√

32εd : εd, (1)

ith ε the logarithmic strain tensor and the superscript “d”enoting the deviatoric part.

After the images were cross-correlated, the rigid bodyovement of the cavity was eliminated from the displace-

C. Lauret et al. / Medical Engineering

Ftm

msdt

Fri2

hdaecTo

3

3

uinformation was obtained from either the lateral or themedial side of the slice. The results are shown from boththe lateral and medial sides of one slice, at the most pro-

ig. 4. Displacement, velocity and acceleration of the sampleholder duringwo experiments in which images from the lateral side (dashed line) and

edial side (solid line) of the same sample are obtained.

ent field by minimizing the sum of the displacements of thepeckled pattern of the patch, between the reference and theeformed stage [26]. The resulting displacement field showshe relative movement of the tissue with respect to the sample-

ig. 5. Relative displacements of the lateral side of the brain slice withespect to the sampleholder, with an initial posterior–anterior acceleration,ndicated by vector a, with a displacement vector magnification factor of.2. Areas indicated by A show almost no relative displacement.

n

FioFriit

& Physics 31 (2009) 392–399 395

older. Because of this larger relative motion, smaller, localisplacement variations were not visible. Therefore this rel-tive (mostly rotational) displacement field of the tissue wasliminated in the same way as the rigid body motion of theavity, but now the speckled pattern of the tissue was used.he local displacement field shows the internal displacementsf the tissue with respect to the rest of the slice.

. Results

.1. Strain fields

Two brain slices, obtained from different brains, weresed and 17 measurements were conducted in total. Field

ounced moments in time. The displacements are presented

ig. 6. Local displacement fields of the lateral side of the brain slice with annitial posterior–anterior acceleration, and with a vector magnification factorf 3.1. (a) Obtained after correction for a rigid body rotation of the results inig. 5(a) by 1.5◦ (counter-clockwise) and (b) obtained after correction for aigid body rotation of the results in Fig. 5(b) by 0.6◦ (clockwise). The areasndicated by A show the opposite local displacement of the tissue, areasndicated by B, C, D and E show the influences of the inhomogeneities onhe local displacement field.

3 eering

abTpsw

asbTmwcwrmtrltps

Fwttr

idtfbsttficwvMip

and acceleration of the sampleholder of which the corre-sponding motion relative to the sampleholder and the localdisplacement fields of the medial side of the same slicediscussed previously, can be seen in Fig. 8. These results cor-

96 C. Lauret et al. / Medical Engin

s vector fields, plotted on an image of the correspondingrain slice, taken before spraying a pattern on the surface.he vectors were scaled corresponding to the largest dis-lacement value present in each stage. Vectors are onlyhown for locations where a successful image correlationas obtained.The dashed line in Fig. 4 shows the displacement, velocity

nd acceleration of the sampleholder of which the corre-ponding global, rotational motion of the lateral side of therain tissue relative to the sampleholder can be seen in Fig. 5.he clockwise relative motion in Fig. 5(a) corresponds to theaximum velocity of the sampleholder to the right (12 ms),hereas the counter-clockwise relative motion in Fig. 5(b)

orresponds to a second positive acceleration after the back-ard motion has halted (23 ms). Because the center of

otation is in the middle of the slice, these relative displace-ents are gradually increasing towards the edge. However,

wo regions can be observed with almost no displacementelative to the sampleholder, indicated by A. In Fig. 6, theocal displacement patterns are visualized after eliminating

he tissue rotation relative to the sampleholder, from the dis-lacement field. A local displacement pattern parallel to theulcus can be observed, indicated by B. Just below the hor-

ig. 7. Equivalent Von Mises strains of the lateral side of the brain slice,ith F indicating Von Mises strains of 11% around the vertical sulcus on

he posterior side, and G the Von Mises strains of 0–2% in the area aroundhe horizontal sulcus. Only for coloured regions, digital image correlationesults were obtained.

Fsmb(vrti

& Physics 31 (2009) 392–399

zontal sulcus on the border of grey and white tissue, theisplacement field changes sign over a relatively small dis-ance, indicated by C. However, this last pattern cannot beound in Fig. 6(b), as here a more local center of rotation cane seen at that position. The vertical sulcus on the posterioride and the small sulcus above the midline both influencehe displacement field locally, indicated by D and E, respec-ively. From the displacement field, the logarithmic straineld with respect to the initial configuration has been cal-ulated. In Fig. 7, the equivalent Von Mises strain is shown,here strains of 11% can be seen in the regions around theertical sulci, indicated by F. In addition, Fig. 7(a) shows Vonises strains between 0 and 2% in the region around the hor-

zontal sulcus, indicated by G. However, this pattern is lessronounced in Fig. 7(b).

The solid line in Fig. 4 shows the displacement, velocity

ig. 8. Situation at 12.5 ms. (a) Relative displacement field of the medialide of the brain slice with respect to the sampleholder and with a vectoragnification factor of 1.8. Initial posterior–anterior acceleration indicated

y vector a. Areas indicated by H show almost no relative displacement.b) Local displacement field of the medial side of the brain slice and with aector magnification factor of 2.7, obtained after correction for a rigid bodyotation of the results in (a) by 1.8◦ (clockwise). Areas indicated by H showhe opposite internal displacement; areas I and J show the influences of thenhomogeneities on the local displacement field.

eering & Physics 31 (2009) 392–399 397

rltaopattpcaomsLgrwuM

Fig. 9. Equivalent Von Mises strains of the medial side of the brain slice,with K indicating Von Mises strains of 0–2%, L indicating strains of up to14% in the grey matter areas, M indicating the area with Von Mises strainsof more than 17% in the area of the airbubble and N also showing higherVon Mises strains, around a thicker part of the brain slice. Only for coloured

Fm

Fai

C. Lauret et al. / Medical Engin

espond to a maximum velocity of the sampleholder to theeft (at 12.5 ms). In Fig. 8(a), a rotational motion of the brainissue with respect to the sampleholder can be seen, with threereas with almost no displacement, indicated by H. The centerf rotation was not located in the middle region of the sam-le, as it was for the lateral side. In the top part of the cavitysmall air bubble was present, influencing the relative, rota-

ional motion, as well as the local displacements. Althoughhe medial side had less pronounced inhomogeneities com-ared with the lateral side, still influences of the substructureould be observed. For example, arrows I and J indicate tworeas where local changes in the direction of the vector fieldccur. Fig. 9 shows the equivalent Von Mises strains of theedial side of the brain slice. In the central region of the

lice, Von Mises strains of 0–2% can be seen, indicated by K.indicates areas with Von Mises strains of up to 14% in the

rey matter areas. These areas correspond with grey matteregions, see Fig. 8. M indicates the area where the air bubble

as present, which induced apparent Von Mises strains ofp to 17%. Finally, the area indicated by N also shows Vonises strains up to 17% as well.

regions, digital image correlation results were obtained.

ig. 10. (a) Displacement, velocity and acceleration of the sampleholder and (b) relative number of facets exceeding a certain Von Mises strain at the time ofaximum velocity, for 14 measurements conducted on sample 1 for the lateral (black) and medial (grey) side.

ig. 11. (a) Evolution of Von Mises strain in time, averaged for all measurements of each side of sample 1 and (b) Von Mises strain for sample 1 (open symbols)nd sample 2 (closed symbols) at the maximum velocity corresponding to a certain fraction of facets versus the velocity of the sampleholder. The medial sides indicated in black and the lateral side is grey.

3 eering

3

tsFwFtacaTprMotfVrasivdal

4

dware2Ivbttm

ovtaadmtli

Mtuasepata

bostEwattsh[ipbietpittwimr

esdiIisa

A

g

98 C. Lauret et al. / Medical Engin

.2. Reproducibility

The data of 17 measurements were used in order to inves-igate the reproducibility of the setup and the influence of theeverity of the acceleration pulse on the strain levels obtained.or this, two different samples were used, and a distinctionas made between the medial and lateral side of the samples.ig. 10(a) presents the displacement, velocity and accelera-

ion histories of sample 1, and shows the repeatability of thectuation of the experimental setup. The relative number oforrelated facets that exceed a certain Von Mises strain levelt the time of maximum velocity is displayed in Fig. 10(b).his excludes peak Von Mises strains calculated for a smallercentage of facets, possibly caused by incorrect image cor-elation. Fig. 11(a) represents the progression of the Von

ises strain in time averaged for all the experiments donen sample 1 (except for the experiment with lower accelera-ion magnitude) and corresponding to a fixed percentage ofacets. After the first acceleration peak has been reached, theon Mises strain obtained for 10% of the facets remains in the

ange of 0.08–0.1. A recovery to zero strain occurred after thecceleration event and before the next experiment. Fig. 11(b)hows the sustained Von Mises strain at the moment of max-mum velocity corresponding to a fixed percentage of facetsersus the maximum velocity applied. One measurement wasone with a significantly lower acceleration level. Figs. 10(b)nd 11(b) show that the tissue in this measurement sustainedower Von Mises strains compared to the other measurements.

. Discussion and conclusions

Strain fields were measured in porcine brain tissue slices,uring translational acceleration. Although the sampleholderas accelerated translationally, a rotation of brain tissue rel-

tive to it was found. This may have been induced by a smallotation of the sampleholder during the onset of the accel-ration pulse. However, this rotation was limited to about◦, compared to a maximum translation of about 23 mm.nstead, the rotational displacement field could also be due toariations in slice thickness. Thickness variations may haveeen caused by the slicing procedure. Furthermore, duringhe experiment the lower part of the sample was resting onhe sampleholder, which could have influenced the relative

otion of the tissue.The internal displacement patterns showed an influence

f geometrical heterogeneities such as sulci and constitutiveariations between the surrounding grey and white matter onhe local displacement field. Sulci are geometrical boundariesnd the opposite surfaces may slide relative to each other. Inddition, sulci are always surrounded by grey matter whichiffers in mechanical properties from its neighbouring white

atter. Cloots et al. [27] have shown the potential impor-

ance of geometrical heterogeneities for the stress or strainevels that occur in brain tissue during severe intertial load-ng with numerical simulations. Bayly et al. [16] presented

C

& Physics 31 (2009) 392–399

RI data with indications that there were strain concentra-ions at the interface of grey and white matter, but could notnambiguously identify these effects. Moreover, Graham etl. [28] supported these indications by observations of moreevere local injury near these boundaries. The highest strainsxperienced by 10% of the correlated surface area of the sam-le was found to be in the range of 0.08–0.1 for the typicalcceleration level applied in these experiments. Furthermore,his strain level was found to depend on the severity of thecceleration pulse.

Enamel was used for the application of a speckled patternecause of its relatively good adherence to the wet surfacef the tissue. Narmoneva et al. [29], used Enamel to apply apeckled pattern on cartilage. A microscopic examination ofhe cartilage sample surface showed that the water-resistantnamel was attached well to the surface. Moreover, the paintas found not to penetrate into the sample, and the dot size

nd shape were preserved during the experiment. Althoughhe digital image correlation software can correlate the pat-ern present on the brain slices, the pattern is less fine than in atandard experiment where the specimen does not need to beydrated and for which an accuracy of 0.1% can be achieved26]. Therefore, the accuracy of the correlated regions wasnvestigated for this specific experimental situation. For thisurpose, two experiments were done in which a known rigidody translation was applied to the sampleholder contain-ng a tissue slice similar to the samples used in the actualxperiment. The displacement obtained with the correlationechnique was found to be 4.8% smaller than the applied dis-lacement. This deviation is believed to be due to inaccuracyn scaling to actual coordinates, in combination with reflec-ions of the wet surface and the motion of the sample relativeo the lights. Moreover, the pattern density on the grey matteras lower than on the white matter because of a difference

n adherence capacity of the two tissues with Enamel. Theaximum Von Mises strain of the surface obtained for this

igid body translation was found to be 0.4%.The medial and lateral sides of the slices differed consid-

rably in grey and white matter distribution and geometry ofulci. Both sides were measured, however, this could not beone in the same experiment. However, strain levels obtainedn repeated experiments were found to be well reproducible.n future studies, the setup could be extended to a system thats able to obtain field information from both sides of the sliceimultaneously. Furthermore, the setup could be modified topply rotational accelerations as well.

cknowledgements

This work was conducted within the APROSYS Inte-rated Project supported by the European Commission.

onflict of interest

None declared.

eering

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[[

C. Lauret et al. / Medical Engin

eferences

[1] Langlois JA, Rutland-Brown W, Thomas KE. Traumatic brain injuryin the United States: emergency department visits, hospitalizations,and deaths. Technical report, Centers for Disease Control and Pre-vention, National Center for Injury Prevention and Control, Atlanta,USA; 2004.

[2] Claessens MHA, Sauren F, Wismans JSHM. Modelling of the humanhead under impact conditions: a parametric study. In: Proceedings ofthe 41th Stapp Car Crash Conference, number SAE 973338; 1997. p.315–28.

[3] Brands DWA, Bovendeerd PHM, Wismans JSHM. On the potentialimportance of non-linear viscoelastic material modelling for numeri-cal prediction of brain tissue response: test and application. Stapp CarCrash J 2002;46:1–19.

[4] Brands DWA, Peters GWM, Bovendeerd PHM. Design and numericalimplementation of a 3-d non-linear viscoelastic constitutive model forbrain tissue during impact. J Biomech 2004;37:127–34.

[5] Willinger R, Kang H, Diaw B. Three-dimensional human head finite-element model validation against two experimental impacts. AnnBiomed Eng 1999;27:403–10.

[6] Zhang L, Yang KH, King A. Comparison of brain responses betweenfrontal and lateral impacts by finite element modelling. J Neurotrauma2001;18:21–30.

[7] Kleiven S, von Holst H. Consequences of brain size following impactin prediction of subdural hematoma with numerical techniques. In: Pro-ceedings of the international conference on the biomechanics of impact,IRCOBI, Isle of Man; 2001. p. 161–72.

[8] Hrapko M, van Dommelen JAW, Peters GWM, Wismans JSHM. Themechanical behaviour of brain tissue: large strain response and consti-tutive modelling. Biorheology 2006;43(5):623–36.

[9] Pudenz RH, Shelden CH. The lucite calvarium: a method for directobservation of the brain. J Neurosurg 1946;3:487–505.

10] Ivarsson J, Viano DC, Lovsund P, Aldman B. Strain relief from thecerebral ventricles during head impact: experimental studies on naturalprotection of the brain. J Biomech 2000;33:181–9.

11] Ivarsson J, Viano DC, Lovsund P. Influence of the lateral ventriclesand irregular skull base on brain kinematics due to sagittal plane headrotations. J Biomech Eng, Trans ASME 2002;124:422–31.

12] Margulies SS, Thibault LE, Gennarelli TA. Physical model simulationsof brain injury in the primate. J Biomech 1990;23:823–36.

13] Meaney DF, Smith DH, Schreiber DI, Bain AC, Miler RT, Ross DT,et al. Biomechanical analysis of experimental diffuse axonal injury. JNeurotrauma 1995;12:689–94.

14] Hardy WN, Foster CD, King AI, Tashman S. Investigation of braininjury kinematics: introduction of a new technique. Crashworthiness,

[

[

& Physics 31 (2009) 392–399 399

Occupant Protection and Biomechanics in Transportation SystemsAMD 1997;225:241–54.

15] Hardy WN, Foster CD, Mason J, Yang KH, King A, Tashman S. Inves-tigation of head injury mechanisms using neutral density technologyand high-speed biplanar X-ray. Stapp Car Crash J 2001;45:337–68.

16] Bayly PV, Cohen TS, Leister EP, Ajo D, Leuthardt EC, Genin GM.Deformation of the human brain induced by mild acceleration. J Neu-rotrauma 2005;22:845–56.

17] Bayly PV, Ji S, Song SK, Okamoto RJ, Massouros P, Genin GM. Mea-surement of strain in physical models of brain injury: a method basedon harp analysis of tagged magnetic resonance images (mri). J BiomechEng, Trans ASME 2004;126:523–8.

18] Bayly PV, Black EE, Pedersen RC, Leister EP, Genin GM. In vivoimaging of rapid deformation and strain in an animal model of traumaticbrain injury. J Biomech 2006;39:1086–95.

19] Garo A, Hrapko M, van Dommelen JAW, Peters GWM. Towards areliable characterization of the mechanical behaviour of brain tissue:the effects of post-mortem time and sample preparation. Biorheology2007;44:51–8.

20] Heinritzi K, Schillinger S. Untersuchung des liquor cerebrospinalisbei gesunden und zentralnervos erkrankten schweinen. Tierarztl Prax1996;24:359–67.

21] Welker W, Johnson JI, Noe A. Comparative mammalian brain col-lections, http://www.brainmuseum.org. Technical report, Universityof Wisconsin, Michigan State and National Museum of Health andMedicine; 2007.

22] Gonzalez RC, Wintz P. Digital image processing. Amsterdam:Addison-Wesley Publishing Company; 1983.

23] Chu TC, Ranson WF, Sutton MA, Peters WH. Applications of digital-image-correlation techniques to experimental mechanics. Exp Mech1985;25:232–44.

24] Sutton MA, Cheng M, Peters WH, Chao YJ, McNeill SR. Application ofan optimized digital correlation method to planar deformation analysis.Image Vis Comput 1986;4:143–50.

25] Zhang D, Arola DD. Applications of digital image correlation to bio-logical tissues. J Biomed Opt 2004;9:691–9.

26] GOM. Aramis user manual v5. 4. 1-en-Rev-A-2005-09-27; 2005.27] Cloots RJH, Gervaise HMT, van Dommelen JAW, Geers MGD.

Biomechanics of traumatic brain injury: influences of the mor-phologic heterogeneities of the cerebral cortex; Ann Biomed Eng2008;36:1203–15.

28] Graham DI, Adams JH, Nicolle JAR, Maxwell WL, Gennarelli TA. Thenature, distribution and causes of traumatic brain injury. Brain Pathol1995;5:397–406.

29] Narmoneva DA, Wang JY, Setton LA. Nonuniform swelling-inducedresidual strains in articular cartilage. J Biomech 1999;32:401–8.