ORIGINAL COMMUNICATION

Recent Firsts in Cadaveric ImpactBiomechanics Research

ALBERT I. KING,1 KING H. YANG,1 AND WARREN N. HARDY2*1Wayne State University, Detroit, Michigan

2Virginia Tech, Blacksburg, Virginia, Blacksburg, Virginia

High-speed biplane x-ray and neutral density targets were used to examinebrain displacement and deformation, as well as aortic motion and deformationwithin the mediastinum, during impact. Thirty-five impacts using eight humancadaver head and neck specimens and eight impacts of the intact cadaverthorax are summarized. During impact, local brain tissue tends to keep itsposition and shape with respect to the inertial frame, resulting in relativemotion between the brain and skull and deformation of the brain. The localbrain motions tend to follow looping patterns. Similar patterns are observedfor impact in different planes, with some degree of posterior–anterior andright–left symmetry. Clinically relevant damage to the aorta was observed inseven of the thorax tests. The presence of atherosclerosis was demonstratedto promote tearing. The isthmus of the aorta moved dorsocranially duringfrontal impact and submarining loading modes. The aortic isthmus movedmedially and anteriorly during impact to the left side. Clin. Anat. 24:294–308,2011. VVC 2011 Wiley-Liss, Inc.

Key words: injury mechanisms; kinematics; deformation; brain; aorta; biplanex-ray

INTRODUCTION

The use of cadavers in impact biomechanicsresearch has been ongoing for over 60 years, start-ing in the 1940s with the work of Lissner and Gurd-jian at Wayne State University (Gurdjian and Lissner,1947). In the early days every experiment was con-sidered a ‘‘first,’’ such as determining the strainneeded to fracture a skull or studying the responseof the spine during caudocephalad acceleration, asexperienced by pilots who eject from jet aircraft.More recently, brain and aorta injury have becomeof great interest. In this article, the techniques usedto measure the relative motion of the cadaveric brainwith respect to the skull during blunt impact andreproduce traumatic rupture of the aorta (TRA) inthe cadaver are described.

Brain motion was measured using a high-speedbiplane x-ray system that captured images of thebrain at 1,000 frames-per-second via digital cam-eras. Radio-opaque neutral density targets (NDTs)were implanted in the brain to enable the visualiza-tion of brain motion. The results of head impact

experiments reported by Hardy et al. (2001, 2007)are reviewed in this article. Regarding consistentgeneration of clinically relevant TRA in the cadaver,attempts were made by several investigators overthe past 3 decades, including attempts by WayneState University in the 1980s (Cavanaugh et al.,1990). Other attempts have been made by research-ers at the University of Michigan (Culver et al.,

Abbreviations used: AC, anterior column; aCSF, artificial cere-bral spinal fluid; AIC, average (or apparent) instant center;MDA, major displacement (or deformation) axis; NDT, neutraldensity target; PC, posterior column; TRA, traumatic ruptureof the aorta.

*Correspondence to: Warren N. Hardy, VT Campus, VirginiaTech—Wake Forest Center for Injury Biomechanics, Rm. 443ICTAS building, Stanger Street, MC 0194, Blacksburg, Virginia.E-mail: whardy@vt.edu

Received 18 October 2010; Revised 18 January 2011; Accepted23 January 2011

Published online 18 March 2011 in Wiley Online Library(wileyonlinelibrary.com).DOI10.1002/ca.21151

VVC 2011 Wiley-Liss, Inc.

Clinical Anatomy 24:294–308 (2011)

1977; Robins et al., 1982) and more recently at theUniversity of Virginia (Forman et al., 2008). Thereare few publications to document these attemptsbecause it is not typical to publish negative results.High-speed biplane x-ray facilitated visualization ofaortic motion and examination of strain patternswithin the aorta. Judicious specimen positioning andloading mode selection enabled the reliable repro-duction of TRA in the intact cadaver. Results of theTRA experiments are taken from Hardy et al. (2008).

MATERIALS AND METHODS

Measurement of Cadaveric Brain Motion

Speculation abounds as to how the brain mightshift inside the skull during head impact leading tosurface contusions from this relative sliding againstthe rough internal surfaces of the skull. Yet, knowl-edge of brain motion within the skull is critical tounderstanding the mechanisms of head injury. Theearly efforts of Hodgson et al. (1966), Shatsky(1973), Stalnaker et al. (1977), and Nusholtz et al.(1984) that employed a variety of high-speed x-raytechniques were reviewed by Hardy et al. (2001)None of these studies was able to provide accuratethree-dimensional quantitative brain motion data.Not until a high-speed biplane system was developedat the Herrick-Davis Motion Analysis Lab of HenryFord Hospital in Detroit, during the tenure of Dr. EricL. Radin as director of the Bone and Joint Center,were these data obtainable. It was at that time aone-of-a-kind device that took many years to per-fect, as evidenced by the articles by Al-Bsharat et al.(1999) and Hardy et al. (2001), came to fruition.Eventually, the proper combination of image intensi-fiers and digital video cameras resulted in the abilityto track NDTs to within 0.1-mm accuracy in threedimensions using stereo-photogrammetric methods.

Unembalmed human cadaver1 head and neckcomplexes were tested in an inverted configurationand were repressurized using artificial cerebral spinalfluid (aCSF). Seventeen cadavers (seven male and10 female) were used in 62 tests. The average age,stature, and mass were 77 years, 167 cm, and 77kg, respectively. Inverting the specimens improvedthe effectiveness of the perfusion process. Accumu-lation of compressible gases within the cranial vaulttends to decouple the brain from the skull, resultingin exaggerated motion of the brain relative to theskull during impact. Therefore, care must be takento evacuate the intracranial space of any gas. Toachieve aCSF infusion, each neck was dissected toexpose the common carotid arteries and internal jug-ular veins. Compression fittings were attached toeach vessel to facilitate perfusion of the cerebralvasculature. Substantial tissue was removed fromthe neck, leaving only enough material to inhibitdesiccation of the cervical ligaments. Any remaining

compromised vasculature was ligated, with theexception of the vertebral arteries, which were leftpatent to assist in elimination of intracranial gasses.The vertebral column was disarticulated between thethird and fourth thoracic vertebrae before suspend-ing each specimen in the impact apparatus. The thirdthoracic vertebra was separated from the remainingcolumn, while preserving the associated length ofspinal dura. Approximately 2 cm of spinal cord wasexcised from the dural sheath, and a barbed fittingwas attached to the sheath to provide adequate per-fusion of the brain. These attachments were con-nected to an aCSF reservoir that was suspendedabove the specimens such that appropriate baselineintracranial pressure was obtained via a gravity-feedapproach.

The skull was instrumented with a nine-acceler-ometer array composed of linear accelerometers tomeasure its generalized three-dimensional kinemat-ics, specifically linear and angular acceleration(Padgaonkar et al., 1975) and a tetrahedral versionof the array (Hardy, 2002). Intracranial pressurewas monitored by two cranial pressure transducersinserted into approximate coup and contrecoup loca-tions through trephines in the skull. These trephineswere sealed using a unique system consisting ofminiature electrical connectors trapped between o-rings and secured via threaded inserts set in theskull.

To visualize brain motion, special radio-opaqueNDTs were fabricated from 1.9 mm tin granulesplaced inside thin-walled polystyrene cylinders hav-ing an outside diameter of 2.5 mm and a length of 5mm. Each NDT had an overall density at or below1.5 gm/ml and was designed to maintain its positionrelative to surround brain tissue during impact with-out damaging the tissue or influencing its motion.Holes were drilled in the skull to allow insertion of a3-mm diameter cannula for the placement of NDTsin two vertical columns, as shown in Figure 1, or theinsertion of a cannula fixture having five tubes withbeveled ends for the insertion of a cluster of targets,as shown in Figure 2. Typically, a column of five orsix NDTs was located in the occipitoparietal region(labeled ‘‘PC,’’ for posterior column) and in the tem-poroparietal region (labeled ‘‘AC’’ for anterior col-umn). The spacing between the centers of the NDTsin the columns ranged from 7 to 12 mm. For theclusters, there were a total of seven targets thatoccupied roughly 4 ml of tissue. For each cluster,five of the NDTs were in a single plane and there wasone NDT above and one below the plane. There weretwo clusters implanted in each head tested.

The testing apparatus included a suspension fix-ture supporting the inverted head, which could be ei-ther struck by an impactor or accelerated by thesame impactor until it was stopped by a rigid sur-face. The specimen was attached to a subassemblyof the apparatus that was free to translate androtate in a single plane. The testing apparatus,including the perfusion system, is shown with aspecimen outfitted with an American football helmetin Figure 3. The second thoracic vertebra was passedthrough an aperture in an aluminum cup, and pinned

1This work was carried out in accordance with the practicesoutlined by the Willed Body Program of the Wayne State Uni-versity School of Medicine, Department of Anatomy.

295Recent Firsts in Cadaveric Research

3. Results

The peak sled acceleration was 5.18 g, simulating a1.98 m/s (4.46 mph) rear-end impact. Fig. 2 shows sledacceleration and velocity as a function of time. Theimpact was initiated 97 ms after data collection wasstarted. The A-P (x) and S-I (z) acceleration time tracesof the head and T1 are shown in Fig. 3. There is a 24 msdelay in the onset of thex component of T1 accelerationrelative to that of the sled. The x acceleration of the

head was further delayed 80 ms relative to thex accel-eration of T1. It is observed that there was no significanttime lag between the initiation of the acceleration of thesled andz acceleration of the head and that of T1.

Fig. 2. Sled acceleration (solid) and velocity (dotted).

Fig. 3. (a) Thorax (solid) and head center of gravity (HDCG) (dotted)acceleration in the A–P direction. (b) Thorax (solid) and head center ofgravity (HDCG) (dotted) acceleration in the I–S direction.

Fig. 1. (a) Sequential X-ray images from a high-speed X-ray videocamera of test HFH19 (250 frames/s). (b) Sequential images from anordinary high-speed video camera of test HFH19 (250 frames/s).

F. Luan et al. / Clinical Biomechanics 15 (2000) 649–657 651

Horizontally, it is obvious from the X-ray images(Fig. 1) that the lower cervical vertebrae moved fasterthan the upper ones during the impact. This is becausethe crash pulse acts to drive the seat toward the torso.Shear forces must be transmitted to C7 by the thoraxbefore they can be transmitted to the levels superior toit. The forward translation of C7 with respect to C6 actsto produce shear forces on the soft tissues (such as theintervertebral disc and the ligaments) between C6 andC7.

After the soft tissues between C6 and C7 became taut,the forward translation proceeded up to the next leveluntil the shear forces reached the head. As demonstratedin Fig. 3(a), the x component of T1 acceleration startedabout 80 ms prior to the x component of the head ac-celeration. Thus, the lower cervical vertebrae translatedmore than the head. As a result, there was a loss ofcervical lordosis and/or the formation of an S-shapedcurve of the neck. Since shear forces activate the for-ward motion of the vertebrae and the head, large sheardeformations are to be expected in the soft tissues of thespine. As shear forces can only be transmitted to adja-cent levels after soft tissues have become taut, it takessome time for the head to start moving in the horizontaldirection. This compliance is a possible explanation forthe Ôhead lagÕ phenomenon.

In the vertical direction, the thorax, neck and head allexperienced an upward translation at approximately thesame time (Fig. 3(b)). This upward trunk motion wasprobably due to the straightening of the kyphotic tho-racic spine. Hence, an upward motion is transmittedfrom the thorax to the head while compression in theaxial direction of the neck was generated early inthe impact. This compressive force can be transmitted tothe head without delay because it does not require thetightening of soft tissue fibers. Due to the heavy torsomass, the trunk could only move a limited distance asthe head continued to move upward. Thus, tension inthe neck was generated later in the impact.

Fig. 4 shows the history of the angle of rotation of thehead and neck relative to a global coordinate system, asmeasured from high-speed X-ray images. As expected,each individual cervical vertebra and the head were inextension for the first 240 ms. However, upon closerobservation, the extension angle of the lower vertebralbodies (C4–C6) increased more rapidly than that of theupper vertebrae (C1–C3) and the head. The head did notshow significant extension until 180 ms; this was about40 ms later than the onset of the upper cervical motion.

Since the lower vertebra for each pair of adjacentvertebrae extends more than that of the upper vertebra,all the intervertebral segments were loaded in flexion forthe first 100 ms. Due to the limited field of view of theX-ray system, the motion of the cervical vertebrae canonly be completely monitored for about 140 ms after theimpact was initiated. As demonstrated in Fig. 4, the

extension angles for the lower cervical vertebrae reachedtheir peaks earlier due to the constraint of the trunk. Atthe same time, extension angles of the upper cervicalvertebrae continued to increase. At this stage, the con-straint of the trunk begins to lead the lower neck intoextension, while the upper neck is still in flexion. Thisphenomenon is consistent with the X-ray images shownin Fig. 1.

Fig. 5 shows the head rotation measured by the an-gular velocity sensor and those calculated from the nine-accelerometer array. In both methods, the head rotationwas integrated from angular velocity data either ob-tained directly from the angular velocity sensor or froman integration of the calculated angular acceleration,using the data from the nine linear accelerometers.However, the 3-2-2-2 data indicated that there was ashort period of time (from about 100 to 200 ms) when

Fig. 4. Global rotation of the head and each cervical vertebra (C1–C6).

Fig. 5. Head rotation about the lateral axis (3-2-2-2: solid, sensor:dotted) (extension:positive) (The vertical line indicates the time whenthe head contacted the headrest).

652 F. Luan et al. / Clinical Biomechanics 15 (2000) 649–657

column target on the aorta is used as the referencepoint for each plot. The black arrows indicate thedirection of impact. The full scale in both directionsfor both axes is 75 mm for both plots.

Strain experienced by the aorta was calculatedusing the finite element method (LS-Dyna solver)and the motion of the lead markers as measured bythe high-speed x-ray system. Table 3 catalogs thepeak average longitudinal tensile strain for eachavailable center column target location for each test.The center target locations range from 0 to 8, withthe 0 target being at the level of the left subclavianartery, typically. Missing entries reflect either a loca-tion in which a target was not placed, or for which atarget could not be tracked reliably. The bold type-face in the table indicates peak average strain foreach test. The gray shading indicates the target

location nearest to the primary aortic damage. Thisgives some indication of the strain measured at thelevels of the observed TRA, although the TRA loca-tions do not correspond directly to the target loca-tions, and the relative target locations are approxi-mate between specimens. For many tests, the levelof the peak longitudinal strain is coincident or neigh-boring a level of observed TRA.

DISCUSSION

Measurement of Cadaveric Brain Motion

Better understanding of head injury will lead tomore effective treatment and prevention. Brain dis-placement and deformation during impact have beenexamined in repressurized human cadaver head and

Fig. 20. Horizontal perspective of the mediastinalmotion of the aorta for the side impact of the arm, TestXR4: Sagittal: S, superior; A, anterior; L, left. Scale is(mm).

Fig. 19. Sagittal perspective of the mediastinalmotion of the aorta for the frontal impact shoveling,Test XR2: S, superior; A, anterior; L, left. Scale is(mm).

TABLE 3. Peak Average Longitudinal Tensile Strain Responses (No Data for Test XR3)

Shoveling Side impact Submarining Combined

Center target XR1 XR2 XR4 XR5 XR6 XR7 XR8

0 1 – 0.072 0.158 – – –1 0.088 0.350 0.022 0.067 0.041 0.166 –2 0.075 0.301 0.073 0.042 0.032 0.127 –3 0.093 0.188 0.093 0.013 0.026 0.154 0.0814 0.021 0.356 0.076 0.022 0.088 0.095 0.0755 0.050 0.644 0.137 0.130 – 0.062 0.0736 0.107 – – 0.153 0.009 0.056 0.0877 0.131 – – – 0.001 0.067 0.0778 – – – – – – 0.074

Bold typeface indicates peak longitudinal strain for each test.Gray shading indicates location nearest primary aortic damage (listed as Tear #1 in Tables 3 and 4).

306 King et al.

neck specimens using high-speed biplane x-ray andNDTs. Brain motion patterns can be interpreted visu-ally with respect to angular speed. As the headbegins to rotate, local brain tissue tends to keep itsposition and shape with respect to the inertial frame,creating relative brain displacement and deforma-tion. As the head rotation slows, reaches steadystate, or changes direction, the brain motion exceedsthat of the skull. Additional observations regardingthe relationships between head kinematics andresulting relative motion of the brain with respect tothe skull include that displacement and deformationin regions closer to the c.g. of the head are less thanfor regions of the brain farther from the c.g. duringimpact in the median plane. Further, peak averagemaximum principal strain decreases with increasinglinear acceleration in the direction of impact, increas-ing coup pressure, and increasing coup pressurerate. Similarly, peak average maximum shear straindecreases with increasing linear acceleration in thedirection of impact, increasing coup pressure, andincreasing coup pressure rate. However, no strainparameters were found to vary significantly withangular acceleration. Linear and angular accelerationof the head are reduced with use of a helmet. How-ever, the presence of a helmet did not influence theextent of relative brain displacement and angularspeed was not reduced significantly with use of ahelmet.

Traumatic Rupture of the Aortain the Cadaver

Historically, the generation of clinically relevantTRA in a cadaver has been difficult to achieve. Asearly as 1893, Rindfleisch (1893) considered stretchdeformation of the aorta to be responsible for injury.Based on the techniques developed by Hardy et al.(2006), it was possible to produce such ruptures incadavers in seven of eight tests. The results are clini-cally relevant because they are transverse and fitthe description of injuries seen in autopsy, includingassociation of the site of rupture with the presenceof atherosclerotic plaque (Viano et al., 1978). In theautomotive crash environment, rib fractures arecommon in TRA cases because the mechanism ofTRA is the imposition of traction on the aorta throughcompression of the thorax. Primarily dorsocranialmotion was generated during shoveling impacts andthe simulation of submarining. This demonstratesthat contact resulting in superior and posteriormotion of the heart and aortic arch related to themotion of the diaphragm and sternum, resulting inTRA. Motion of the aorta to the right side of thespecimens away from the impact was generated forthe lateral blows. In addition, the aorta moved ante-riorly during the side impacts. The heart and aorticarch are tethered to the sternum and can moveanteriorly with the sternum. The results of Melvin etal. (1998) and Cavanaugh et al. (2005) reinforce theimportance of anterior motion of the sternum, andthe importance of limiting chest compression andclavicle motion. Both axial elongation (longitudinalstretch) of the aorta and tethering of the descending

thoracic aorta by the parietal pleura are central tothe generation of TRA.

CONCLUSION

This article summarizes results of recent cadaverstudies that produced biomechanical results thatheretofore have not been achieved. The brain motionstudy is of major significance to the understanding ofthe mechanism of injury to the brain. Large motionsnear the center of the brain due to angular rotation ofthe head are seen while this motion is less near theskull. As for linear acceleration, the motion is consid-erably less, implying that the injury mechanism isquite different and is the result of a dynamic pressurewave that is generated during blunt impact. Recentreports of mTBI among returning veterans exposedto explosive blasts further support the pressure-wavehypothesis for brain injury. More research is neededto study this hypothesis as it seems that veterans donot have to undergo large head accelerations of anykind to sustain an mTBI. As for TRA, automotiveengineers can now try to tailor their vehicle design soas to avoid the types of impact loading on the chestthat can cause the aorta to rupture. However, addi-tional research is still required to investigate otherpossible loading modes which may contribute to TRA.

REFERENCES

Al-Bsharat AS, Hardy WN, Yang KH, Khalil TB, King AI, Tashman S.1999. Brain/skull relative displacement magnitude due to blunthead impact: New experimental data and model. In: Proceed-ings of 43rd Stapp Car Crash Conference. Warrendale, PA: Soci-ety of Automotive Engineers. p 321–332.

Cavanaugh JM, Koh SW, Kaledhonkar SL, Hardy WN. 2005. An anal-ysis of traumatic rupture of the aorta in side impact sled tests.SAE Technical Paper No. 2005-01-0304. Society of AutomotiveEngineers, Warrendale, PA.

Cavanaugh JM, Walilko TJ, Malhotra A, Zhu Y, King AI. 1990. Bio-mechanical response and injury tolerance of the thorax in twelvesled side impacts. Stapp Car Crash J 34:28–38.

Culver RH, Bender M, Stalnaker RL, Melvin J. 1977. Feasibility ofinvestigating the mechanisms of aortic trauma using high-speedcineradiography. Report UM-HSRI-77–53.

Gurdjian ES, Lissner HR. 1947. Deformations of the skull in head injuryas studied by the stresscoat technic. Am J Surg 73:269–281.

Forman J, Stacey S, Evans J, Kent R. 2008. Posterior accelerationas a mechanism of blunt traumatic injury of the aorta. J Biomech41:1359–1364.

Foster CD, Hardy WN, Yang KH, King AI, Hashimoto S. 2006. High-speed seatbelt pretensioner loading of the abdomen. Stapp CarCrash J 50:27–51.

Hardy WN. 2002. Instrumentation in experimental design. In:Nahum AM, Melvin JW, editors. Accidental Injury: Biomechanicsand Prevention. 2nd Ed. New York: Springer-Verlag. p 12–39.

Hardy WN, Foster CD, Mason MJ, Yang KH, King AI, Tashman S.2001. Investigation of head injury mechanisms using neutraldensity technology and high-speed biplanar x-ray. Stapp CarCrash J 45:337–368.

Hardy WN, Mason MJ, Foster CD, Yang KH, King AI. 2006. Compari-son of intracranial pressure response to cadaver head kinemat-ics. J Biomech 49(S1):S155.

Hardy WN, Mason MJ, Foster CD, Shah CS, Kopacz JM, Yang KH,King AI, Bishop J, Bey M, Anderst W, Tashman S. 2007. A studyof the response of the human cadaver head to impact. StappCar Crash J 51:17–80.

307Recent Firsts in Cadaveric Research

Hardy WN, Shah CS, Mason MJ, Kopacz JM, Yang KH, King AI, VanEe CA, Bishop JL, Banglmaier RF, Bey MJ, Morgan RM, DiggesKH. 2008. Mechanisms of traumatic rupture of the aorta andassociated peri-isthmic motion and deformation. Stapp CarCrash J 52:233–265.

Hodgson VR, Gurdjian ES, Thomas LM. 1966. Experimental skull de-formation and brain displacement demonstrated by flash x-raytechnique. J Neurosurg 25:49–52.

Melvin JW, Baron K. J, Little WC, Gideon TW. 1998. Biomechanicalanalysis of Indy race car crashes. In: Proceedings of 42nd StappCar Crash Conference. Warrendale, PA: Society of AutomotiveEngineers. p 247–266.

Nusholtz GS, Lux P, Kaiker PS, Janicki MA. 1984. Head impactresponse—Skull deformation and angular accelerations. In: Pro-ceedings of 28th Stapp Car Crash Conference. Warrendale, PA:Society of Automotive Engineers. p 41–74.

Padgaonkar AJ, Krieger KW, King AI. 1975. Measurement of angularacceleration of a rigid body using linear accelerometers. J ApplMech 42:552–556.

Robins DH, Leham RJ, Nusholtz GS, Melvin JW, Benson JB, Stal-naker RL, Culver R. 1982. Quantification of thoracic responseand injury: the gathering of data. Report DOT-HS-4-00921.

Rindfleisch E. 1893. Zur entstehung and heilung des aneurysma dis-secans aortae. Virchows Archiv fur Pathologische Anatomie undPhysiologie und fur Klinische Medizin 131:374–378.

Shatsky SA. 1973. Flash x-ray cinematography during impactinjury. In: Proceedings of 17th Stapp Car Crash Conference.Warrendale, PA: Society of Automotive Engineers. p 361–376.

Stalnaker RL, Melvin JW, Nusholtz GS, Alem NM, Benson JB. 1977.Head impact response. In: Proceedings of 21st Stapp Car CrashConference. Warrendale PA: Society of Automotive Engineers.p 303–335.

Viano DC. 1978. Evaluation of biomechanical response and potentialinjury from thoracic impact. Aviat Space Environ Med 49:125–135.

Voigt GE, Wilfert K. 1969. Mechanisms of injuries to unrestraineddrivers in head-on collisions. In: Proceedings of 13th Stapp CarCrash Conference. Warrendale, PA: Society of Automotive Engi-neers. p 295–313.

308 King et al.