2001-22-0016

Stapp Car Crash Journal, Vol. 45 (November 2001), pp.

Copyright © 2001 The Stapp Association

Investigation of Head Injury Mechanisms Using Neutral Density

Technology and High-Speed Biplanar X-ray
Warren N. Hardy, Craig D. Foster, Matthew J. Mason, King H. Yang, and Albert I. King
Wayne State University Bioengineering Center

Scott Tashman
Henry Ford Hospital Bone and Joint Specialty Center

__________________________________

ABSTRACT – The principal focus of this study was the measurement of relative brain motion with respect to the skull using a
high-speed, biplanar x-ray system and neutral density targets (NDTs). A suspension fixture was used for testing of inverted,
perfused, human cadaver heads. Each specimen was subjected to multiple tests, either struck at rest using a 152-mm-diameter
padded impactor face, or stopped against an angled surface from steady-state motion. The impacts were to the frontal and
occipital regions. An array of multiple NDTs was implanted in a double-column scheme of 5 and 6 targets, with 10 mm between
targets in each column and 80 mm between columns. These columns were implanted in the temporoparietal and occipitoparietal
regions. The impacts produced peak resultant accelerations of 10 to 150 g, and peak angular accelerations between 1000 and
8000 rad/s2. For all but one test, the peak angular speeds ranged from 17 to 22 rad/s. The relative 3D displacements between the
skull and the NDTs were analyzed. The localized motions of the brain generally followed loop or figure eight patterns, with peak
displacements on the order of ± 5 mm. These results can be used to further finite-element modeling efforts.

KEYWORDS – Cadaver, head, impact, brain displacement, skull kinematics, x-ray, NDT.
__________________________________

INTRODUCTION prediction and finite-element model validation. Such

A high rate of deficit and morbidity are associated
with head injury, particularly within the automotive
environment. Nearly fifty-percent of all severe skull-
brain trauma results from automotive accidents
(Kallieris et al., 1997). Of particular concern is
hemorrhage, such as subdural hematoma (SDH), and
diffuse axonal injury (DAI). Advances in treatment
notwithstanding, injury prevention is critical. To
mitigate or prevent injury, injury mechanisms must
be understood so that a reasonable tolerance criterion
can be established.
The focus of this study is the investigation of head
injury mechanisms, specifically the measurement of
relative brain/skull kinematics. It is hypothesized that
DAI is related to local strain, shear strain, and rate of
strain. Although DAI cannot be studied using
cadavers, the mechanical response can be
determined. Neuropathological studies can help
provide injury tolerance information. Examination of
brain displacement, stretch, strain, and strain rate,
along with measurement of skull acceleration, speed,
and displacement, can lead to improved injury
a model could then be used to examine a wide variety
of impact conditions and details that would not be
possible using experimental models. This study
represents a step toward understanding brain injury.
Although strain was not measured, displacements of
the cadaver brain were compared to the kinematics of
the skull during impact.
Relative brain/skull distortions have been of interest
historically. The translucent plastic calvarium has
been used by various researchers in an attempt to
directly observe relative brain/skull phenomena. This
technique was first applied nearly fifty years ago
(Shelden et al., 1944). Subsequently, Pudenz and
Shelden (1946) illustrated motion of the Macaque
brain using a Lucite calvarium. In a later study,
Gosch et al. (1969) employed a Lexan calvarium
when impacting Rhesus monkeys. The level of
brain/skull decoupling due to entrapped intracranial
gasses is not known for these early studies.
Radiographs have been used in an attempt to observe
relative brain/skull phenomena as well. Hodgson et
al. (1966) conducted flash x-ray head impact studies

1
2 Hardy et al. / Stapp Car Crash Journal 45 (November 2001)

on anesthetized dogs. Intravascular contrast media
and lead targets were used to track the motion of the
brain. During impact, the single-film tray x-ray
system captured one image approximately at the time
of maximum compression of the head. The influence
that the target density may have had on the overall
displacement is unknown for these tests.
Shatsky (1973) described a high-speed flash x-ray
cinematography system. The system was capable of
resolving diameters of 0.3 mm and an image rate of
1000 frames per second. The image from the device
was ultimately captured on 16 mm film. Shatsky et
al. (1976) used this system to investigate in vivo
blunt head injury trauma in the sagittal plane in
primates. This approach was limited to observation of
the major vasculature.
Stalnaker et al. (1977) conducted fifteen head
impacts on human cadavers. It was found that
pressurization (vascular, CSF) improved the coupling
appreciably. In a subsequent study, Nusholtz et al.
(1984) investigated head impact in anesthetized
monkeys, deceased monkeys, and repressurized
cadavers using high-speed x-ray. Four curved lines of
neutral density radio-opaque gel were injected into
the brains of the subjects. Differential rotational
motion of the brain with respect to the skull was
found, as well as internal displacement of the brain.
Hardy et al. (1997) described the design and use of
triaxial neutral density accelerometers (NDA-3) for
the measurement of brain motion. The devices were
polyurethane foam-filled, polyester resin shells that
contained three mutually-orthogonal integrated
accelerometer dice. The units displaced 0.187 ml and
had a density of 1.07 gm/ml. Each NDA-3 was
acrylic coated, and had a 1.5-mm gold ball fixed to
the center of the bottom face. The NDA-3’s were
designed to maintain their positions with respect to
surrounding brain tissue during impact. The devices
were implanted in the brain, and their output was
compared to an extra corporal rigid body transducer
array fixed to the skull. The devices were validated
using high-speed x-ray. Tests using the NDA-3s
showed 3-to-5 mm displacement of the brain with
respect to the skull during low-speed impacts.
To further the understanding of the mechanics of
head impact, controlled impacts of inverted, aCSF-
perfused, human cadaver head preparations were
conducted as part of this study. The specimens were
attached to the impact fixtures via the neck.
Displacements of the brain were observed using
neutral density targets and a high-speed, biplanar x-
ray system. Displacements of the brain were
compared to the kinematics of the skull, which were
measured using a nine-accelerometer array.

Relative brain/skull phenomena have been examined METHODS

using acceleration measurement also. Trosseille, et al.
(1992) conducted five cadaver tests to obtain data
against which a finite element model of the head
could be validated. Accelerometers were implanted in
the cadaver brain. However, the accelerometers were
not the same density as the surrounding brain tissue,
and water, not artificial cerebral spinal fluid (aCSF),
was used for perfusion.
This study involved the use of a unique high-speed
biplanar x-ray system and neutral density technology
for the measurement of brain displacement in human
cadaver heads. Many of the methods are summarized
in Al-Bsharat et al. (1999). Usable data were
collected from ten tests using three cadavers, as
summarized in Table 1.

Table 1. Summary of Test Subjects and Conditions

Basic Head Anthropometry (cm)
Occiput to Vertex to Impact Impact Impact Relative Camera
Test Cadaver Breadth Circumference
Glabella Mentum Type Surface Area Severity System
C731-T2 731 19 25 15 56 acceleration flat piston occipital lower JCL
C731-T3 acceleration flat piston occipital higher JCL
C755-T2 755 19 25 16 55 acceleration flat piston occipital lower JCL
C755-T3 acceleration flat piston occipital lower JCL
C755-T4 acceleration flat piston occipital higher JCL
C755-T5 acceleration flat piston frontal lower JCL
C383-T1 383 20 26 14 56 deceleration angled block frontal higher VR4
C383-T2 deceleration angled block frontal higher VR4
C383-T3 deceleration angled block frontal higher VR4
C383-T4 deceleration angled block occipital higher VR4
 Hardy et al. / Stapp Car Crash Journal 45 (November 2001)
Apparatus
A special suspension fixture was used for inverted,
human cadaver head testing. This system allowed
rotation and translation of the inverted head and neck
complex. Suspending the head in this configuration
improved the ability to evacuate gasses from the
intracranial space using a perfusion system. The
fixture had two sections: a translational carriage and
a rotational cage that pivoted about a point on this
platform. The head could be held at any angle in the
horizontal or vertical plane. Torso mass and
distribution of the torso mass was simulated by
weights added to the translational and rotational
sections of the fixture. The specimen was attached to
the fixture by means of a collar on the lower end of
the rotational cage. The four pins used to fix the
specimen at the level of the second thoracic vertebra
were set into the vertebral body and both laminae,
and fiberglass was used as a potting material. The
carriage slid along case-hardened shafts via linear
ball bushings. These shafts were fixed to an
adjustable frame. Its height and fore-aft positions
could be varied based upon the anthropometry of the
test specimen. This fixture is shown in Figure 1a.
This suspension fixture facilitated either acceleration
of the stationary head by contact with a padded
linear-piston impactor, or deceleration of the moving
head by contact with an unpadded fixed surface. For
the deceleration tests, the impactor was used to
gradually accelerate the carriage and the mechanism
suspending the head, without directly contacting the
head. The head was then abruptly stopped by
contacting an angled (adjustable) acrylic block on the
suspension fixture, and the carriage was stopped
shortly thereafter by contact with cross member on
the fixture. The head was supported during the
a b
Figure 1. The inverted, head suspension fixture (a), and the biplanar x-ray system set up for a test at Henry Ford Hospital (b).
3
acceleration phase so that neck flexion/extension was
essentially eliminated prior to the impact.
The high-speed bi-planar x-ray unit is currently in
use in the Motion Analysis Laboratory of Henry Ford
Hospital, and is shown in Figure 1b with the testing
fixtures in place. The output fluorescent phosphor of
the image intensifiers is capable of 3000-Hz
response. The x-ray is a continuous, non-gated
system. The system has two sets of x-ray heads and
intensifiers mounted to a dual overlapping gantry
fixture. Relative height, angle, and distance
adjustment between the two gantries is virtually
limitless. However, the distance between each head
and image intensifier was fixed at 1.2 m in the
horizontal plane, and the angle of each pair is fixed in
the vertical plane. Because this is a biplanar system,
3-D motion of x-ray opaque objects can be
quantified. Images are captured from the rear of the
intensifiers by digital video cameras that have very
high dynamic range. The output of the image
intensifiers is monochromatic.
Two different high-speed video camera systems were
used. The older JCL cameras (JC Labs) had image
resolution of 384x240 pixels, and operated at 250 fps.
Aside from the frame-rate limitation of these
cameras, the combination of low-pixel resolution and
modest dynamic range of these cameras meant that
the NDTs were difficult to resolve in the x-ray
images. In some cases, manual enhancement of the
target locations was required prior to tracking of the
targets’ 3-D trajectories. The newer VR4 cameras
(Vision Research Phantom 4) had 512x512 pixel
resolution using a frame rate of 1000 fps. In addition,
the higher dynamic range of the VR4 cameras eased
most of the target tracking problems and typically the
specimen remained in view for the deceleration tests.
4 Hardy et al. / Stapp Car Crash Journal 45 (November 2001)
Specimen Preparation
After the head of the specimen was separated from
the body between the second and third thoracic
vertebrae, compression fittings were attached to the
carotid arteries and jugular veins. The remaining
major vessels were sutured closed. The vertebral
arteries, however, were left open and used to bleed
air from the circulatory system. The specimen was
then mounted to the rotational subassembly of the
impact fixture by setting pins in T2 and potting T2 in
fiberglass. A Tygon tube was attached to the spinal
dura at the level of T2 for perfusion purposes.
Neutral density targets, or NDTs were implanted in
the brain to image brain displacements. The current
NDTs were made from thin-walled polystyrene
tubing. The finished targets were 5-mm long, 2.5-mm
diameter cylinders. Custom fabricated tin granules,
1.9 mm in diameter were inserted into the center of
the tubing and held in place with cyanoacrylate. The
ends of the tubes were capped with thin sheets of
styrene. The density of the targets was at or below
1.5 gm/ml. These targets were designed to occupy a
minimal volume, move with the brain, and not to
lacerate the brain during the impact. A set of tin NDT
components is shown in Figure 2a.
The NDTs were implanted in the brain using a
cannula passed through trephines in the skull. This
was done using fluoroscopy. The trephines were
outfitted with thread inserts to facilitate sealing using
small screws, O-rings, and a silicone sealing
compound. A kinematics transducer array, the WSU
3-2-2-2, was fixed to the apex of the head. This
allowed three-dimensional characterization of the
generalized kinematics of the head. Three attachment
points were used to secure the device to the skull, and
the space between the mount and the skull was
stabilized using polymethyl-methacrylate.
a b
Figure 2. NDT components: end caps, tin granules, and polystyrene capsules (a), and a completed specimen preparation (b).
A completed specimen preparation is shown in
Figure 2b. It shows a fixture that is attached to the 3-
2-2-2 interface plate for pretest x-ray purposes.
The NDTs were implanted in two vertical columns as
shown in Figure 3. Typically, a column of five or six
NDTs was located in the occipitoparietal region
(labeled “p”, for posterior column) and in the
temporoparietal region (labeled “a” for anterior
column). The spacing between the centers of the
NDTs in the columns ranged from 7 to 12 mm. To
provide fixed reference points on the head, spherical
lead targets were fixed to the skull. In addition, the
infraorbital ridge and auditory meatus were marked
with lead spheres on each side of the head to define
the Frankfort plane.
Testing
As mentioned, Table 1 shows the subjects and
conditions for each test. Ten tests were conducted
using three specimens. Tests for which no usable
target data were available, such as the first test for
each of the first two specimens, were not included.
Complete anthropometry was not obtained for each
specimen, as the emphasis was on speed of
preparation to minimize the time between death and
testing. However, basic measurements of the head
were collected and are presented in Table 1.
Six tests on the first two specimens were conducted
with the moving impactor striking the stationary
head. The impact surface was a 150-mm flat disk,
padded with 50 mm of Ensolite. The four tests on the
third specimen were conducted with the moving head
striking a fixed block. The block was a 25-mm wide
and 250-mm long acrylic structure. Six of the tests
using the impactor to accelerate the head were blows
to the occipital region, and one test (C755-T5) was a
blow to the frontal region.
 Hardy et al. / Stapp Car Crash Journal 45 (November 2001)
Figure 3. Schematic of the dual-column NDT
approach showing the skull markers and the
approximate c.g. location.
Three of the four deceleration tests were blows to the
frontal region, and one test (C383-T4) was an
occipital blow. Figure 4a shows a representative
frontal blow using the impactor to accelerate the
head. Figure 4b shows a representative frontal blow
using the angled block to arrest the motion of the
moving head. Both types of tests delivered combined
translational/rotational motion of the skull, but the
tests using the angled block generated some more
obvious skull rotations that were not coincident in
time with the translational acceleration. These
deceleration tests have the added benefit of having
the cadaver head staying within the field of view of
the image intensifiers during the entire impact. The
cadaver head moved out of view rather quickly
during the impactor tests. The angled block was
inclined 45 degrees.
Figure 4a and Figure 4b also show elements of the
perfusion system. The specimens were perfused using
recirculated aCSF. The aCSF was collected in a
reservoir beneath the specimens, and pumped up to a
vessel above. A float valve apparatus maintained the
level of aCSF in this vessel. The height of the vessel
above the specimen determined the perfusion
pressure, which was maintained at 10.5 kPa (the
average of mean arterial pressure and venous return
pressure). The use of aCSF was critical because
distilled water will cause the brain to swell within the
skull, and normal saline will cause the brain to
shrink.
5
Impact parameters such as load and acceleration are
not tabulated because of the complex reactions of the
head and neck with multiple components of the
impact fixtures. Instead, the impacts are characterized
as being of slightly lower or higher severity.
However, a complete summary of the resulting head
kinematics is provided along with analyses of the
target data. This information provides complete
details of the selected impacts, and could be used for
either test design or finite-element model input for
validation purposes. In addition, it provides a basis
for interpretation of the target data. The distinction
between higher and lower impact severity is subtle,
as all of the impacts were between 2.5 and 3.5 m/s.
The impacts were performed at this low level so that
multiple tests could be conducted on a single
specimen. This was essential given the scarcity of
specimens and the inordinate amount of time required
to prepare a specimen and to set up the laboratory for
this type of test. In addition, the JCL cameras initially
used were limited to 250-fps frame rate.
Synchronization with the x-ray system was also very
important. This was accomplished by an electronic
controller designed to coordinate the data acquisition
system with the cameras, x-ray system, and impactor
firing system. Time of contact was recorded on both
the transducer data acquisition system and a data
collection system synchronized with the video
camera image capture. In this way, the transducer
data were time aligned with the x-ray data.
Data Reduction
The NDT locations were determined in three-
dimensional space via an automated image
enhancement and target-tracking algorithm that
processed the digital video information of the two
recorded oblique perspectives. This involved a
background subtraction mask, intensity profile, ray
projection, 3D auto tracking technique. This
technique looked for a bell-shaped intensity profile to
locate the targets in the planar views. The
backgrounds were then subtracted, and intersections
of rays projected from the remaining points were
found in 3-D space using the distortion correction
and spatial calibration information acquired after
each test. Fine-mesh distortion correction grids were
placed against the front of each image intensifier
after each test series, and were imaged using the
same x-ray and camera settings as those used for the
tests. Black level and white balance adjustments were
also made. For calibration, a multipoint 3D
calibration cube was placed in the field of view after
each test series, centered at the location of the
specimen at the point of contact. As mentioned, not
all targets could be tracked for all tests, and for some
6 Hardy et al. / Stapp Car Crash Journal 45 (November 2001)
tests, some targets could be tracked for only a portion
of the test. Some of the target images collected using
the JCL cameras required manual enhancement.
The initial skull and brain measurements were made
with respect to an inertial frame. The motion of the
brain with respect to the skull was found using three
of the targets fixed to the skull. These skull targets
defined an arbitrary local body-fixed basis. The
displacements of the NDTs were referred to this
basis, resulting in position data in three directions
with respect to this basis. The positions of the three
skull targets were found in anatomical coordinates
from lateral and anteroposterior x-rays taken after
instrumentation of the specimen, but prior to testing.
These x-rays were used to define the Frankfort plane,
and locate the c.g. of the head. The positions of the
skull targets were then measured in the anterior-
posterior, medial-lateral, and inferior-superior
directions. These coordinates were then used to
transform the locations of the NDTs from the
arbitrary body-fixed basis to anatomical directions of
the skull with the c.g. of the head as the origin. In this
way, the three-dimensional components of the target
motion could be analyzed within a coordinate system
that made anatomical sense, and was easy to
visualize. Within this coordinate system, positive X
was from posterior to anterior, and positive Z was
from inferior to superior. Therefore, positive Y was
from right to left. All of the relative brain/skull target
data, as well as the 3-2-2-2 kinematics, were referred
to the c.g. of the head, which was the origin of the
body-fixed frame.
The target data obtained using the JCL cameras were
filtered using a 50-Hz, fourth-order low-pass
Butterworth profile (SAE pseudo channel class 30
Hz). The data were then interpolated to 2000 Hz.
This procedure made the low-frame rate data much
easier to visualize, and did little to affect the data
a b
Figure 4. Representative acceleration, or padded impactor test (a), and deceleration, or fixed block test (b) to the frontal region.
Perfusion system connections can be seen on the left, and the aCSF reservoir can be seen below the specimen on the right.
trends and most peaks. The NDT data collected using
the VR4 cameras were filtered using a channel class
60-Hz profile. This had the effect of removing short-
duration noise, the magnitude of which was less than
0.25 mm. To eliminate the resonance of the head and
the 3-2-2-2 mount system, the kinematics data were
filtered using a channel class 180-Hz profile. This
procedure elucidated the major trends for better
graphical presentation of the data. However, channel
class 1-kHz data are also tabulated in the Appendices.
For these tests, especially those conducted using the
JCL cameras, the way the data were filtered is
critical. The filter used was an FFT implementation
of the aforementioned Butterworth profiles. Prior to
filtering, the data were expanded initially by
reflection over X = 0 and then over Y = 0, at the first
point. Similar reflections were performed at the last
point, resulting in a dataset with almost three times
the number of points as the original data. The ends of
this new dataset were then padded with hundreds of
additional points and filtered. After filtering, the
artificial points were removed, leaving a filtered
version of the original data of the original length.
This technique assured that the first point of each
curve was the same for the filtered and unfiltered data
to the hundredth of a millimeter in X, Y, and Z.
One final consideration pertains to synchronization
between the transducer and target data. Because of
the slow camera frame rate (4 ms between images) of
the JCL cameras, the instant of contact was not
necessarily recorded. This created some ambiguity
regarding time alignment of the target data, as contact
could theoretically take place nearly 4 ms prior to the
first motion detected in the image data. This is
primarily an issue for test C755-T4. Time zero was
taken as the first frame of skull motion for the target
data. Contact times for all tests were verified using
the onset of motion of the skull targets.
 Hardy et al. / Stapp Car Crash Journal 45 (November 2001) 7

RESULTS For the fixed-block tests involving frontal blows

Ten head impacts were conducted using three
specimens. The tests involved both acceleration and
deceleration of the head, and combined translation
and rotation of the head. Table 2 provides a summary
of the generalized kinematics. These data are filtered
per SAE channel class 180-Hz specifications.
Graphical time histories of these data are
subsequently provided for selected tests. Table A1 in
Appendix A summarizes the peak values of channel
class 1k-Hz data.
The values that are of particular significance to these
tests are resultant linear acceleration at the c.g. of the
head, and angular acceleration and speed about the Y
axis, or in the X-Z plane. For the acceleration tests
involving impactor blows to the occiput (C731-T1
through C755-T4), the peak resultant accelerations
ranged from 21 to 28 g. The peak angular
accelerations ranged from 1227 to 2681 rad/s2
(forward flexion), and the peak angular speeds
ranged from 17 to 22 rad/s. For the frontal impactor
test (C755-T5), the peak resultant acceleration was
12 g. The peak angular acceleration was –843 rad/s2,
and the peak angular speed was –7 rad/s. The channel
class 1-kHz data summarized in Table A1 differ little
from these values for the aforementioned tests.
(C383-T1 through C383-T3), the peak resultant
accelerations ranged from 44 to 62 g. The peak
angular accelerations ranged from –1930 to –2523
rad/s2, and the peak angular speeds ranged from –20
to –22 rad/s. In general, these deceleration tests were
more severe than the impactor tests. The occipital
blow deceleration test (C383-T4) was the most severe
of all tests. The peak resultant acceleration was 108
g, and the peak angular acceleration was 7492 rad/s2.
However the peak angular speed was still only 22
rad/s. As mentioned, filtering of the data was
performed to elucidate the major trends, and to
eliminate resonance. This had a greater effect on the
peaks of the fixed-block, or deceleration tests.
Figure 5 shows the results of the first successful NDT
tracking effort in anatomical coordinates from tests
C731-T2 and C731-T3. These plots represent limited
data sets, and essentially reflect only an initial, first
peak response. These data were used by Al-Bsharat et
al. (1999) as partial validation of a finite element
brain injury model, the WSUBIM. At that time, only
the magnitude of displacement with respect to the
skull was known, as the coordinate system was
arbitrary. On these plots, the origin corresponds to
the c.g. of the head.

Table 2. Summary of Channel Class 180-Hz Head Kinematics

C731 C731 C755 C755 C755 C755 C383 C383 C383 C383
Test
T2 T3 T2 T3 T4 T5 T1 T2 T3 T4
Impact type accel. accel. accel. accel. accel. accel. decel. decel. decel. decel.
Impact area occipital occipital occipital occipital occipital frontal frontal frontal frontal occipital
X 20 27 21 23 27 3 5 6 7 90
Linear -6 -11 -8 -4 -5 -12 -46 -35 -53 -104
c.g. Y 2 4 3 3 3 3 6 11 16 20
accel. -4 -6 -7 -7 -6 -2 -2 -3 -4 -18
(g) Z 5 8 10 8 6 3 20 15 17 19
-9 -12 -14 -15 -18 -4 -41 -25 -32 -34
Resultant (g) 21 27 22 24 28 12 62 44 62 108
HIC15ms 13 21 17 22 31 5 47 34 68 164
X 206 375 121 98 89 130 395 524 633 2129
-228 -361 -342 -329 -298 -54 -407 -1025 -1214 -3338
Angular
Y 1227 1695 1882 1995 2681 488 2592 1885 1594 7492
accel.
(rad/s/s) -805 -1051 -574 -638 -568 -843 -1930 -2216 -2523 -6438
Z 59 257 60 150 183 118 875 1859 2065 10364
-373 -569 -204 -126 -169 -396 -523 -910 -1879 -9424
X 0 2 0 0 0 1 1 4 3 5
-1 -2 -3 -3 -2 -1 -3 -5 -6 -5
Angular
Y 18 22 18 19 17 4 4 2 4 22
speed
(rad/s) 0 0 -1 -2 -4 -7 -20 -22 -22 -8
Z 4 4 0 1 0 0 3 3 5 4
-1 0 -1 0 -1 -2 -1 -4 -3 -8
 8 Hardy et al. / Stapp Car Crash Journal 45 (November 2001)
C731-T2
-12
C.G.
0
36 24 12 -12
Inferior
12
Z (mm)
24
36
Superior
48
60
0 0
Anterior X (mm)
Figure 5. Results of the first successful NDT tracking efforts in anatomical coordinates from test
C731-T2 (top), and test C731-T3 (bottom).
The head is inverted, and facing left. That is, the X
direction is positive from right-to-left on the plots,
and the Z direction is positive from top-to-bottom.
Therefore the top of the head is toward the bottom of
the graphs, and the back of the head is toward the
right side of the graphs. All of the subsequent
displacement plots are oriented as just described, and
match the schematic representation given in Figure 3.
Each of the small curves represents displacement of
the brain with respect to the skull as measured at an
NDT implant location. The above data are included
primarily for continuity and completeness, as they do
not provide a complete picture of brain displacement.
However, the beginnings of small figure-eight shapes
and loops can be seen. This is significant because
subsequent testing yielded the same type of response,
but with enough information to gain insight into the
reasons for the shapes, and the nature of the different
paths at different points within the brain. Therefore,
subsequent tests are described in detail.
Comparison of Brain Displacement Response to
Skull Kinematics
Figure 6 through Figure 25 compare the resulting
brain displacement response to the skull kinematics.
The brain displacement responses in the X-Z plane
are presented first, followed by the pertinent skull
kinematics time histories such as resultant linear
acceleration components, angular acceleration and
speed components, and rotation in the X-Z plane.
Selected NDT displacement time histories are then
provided. If available, targets from the “four corners”
are selected for analysis, since many of the effects are
most pronounced at these locations. This usually
means NDT-a1 and NDT-a5 or NDT-a6 from the
C731-T3
-12
C.G.
0
-24 -36 -48 -60 36 24 12 -12 -24 -36 -48 -60
Inferior
12
Z (mm)
24
36
Superior
48
60
Posterior Anterior X (mm) Posterior
anterior target column (on the left of the plots), and
NDT-p1 and NDT-p5 or NDT-p6 from the posterior
target column (on the right).
Some of the key displacement paths are labeled with
directional arrows to show the primary motions.
When a path is described by more than one arrow, the
first arrow is marked with a dot. Since many of the
results are similar between tests, not every test is
described in detail to avoid repetition. Specifically,
the kinematics and displacement time histories for
tests C755-T4 and C383-T2 are not presented.
Figure 6 shows the brain displacement results for test
C755-T2. A primary observation is the large degree
of similarity between these data and the initial data
shown in Figure 5. The figure eight and loop patterns
are intriguing. Each of the NDT curves shows a small
(about 1 mm) initial displacement. This is followed
by a larger displacement curve that follows an
essentially counter-clockwise path for locations near
the c.g. (top of the plot), and a clockwise path for the
locations near the apex of the head (bottom of the
plot). As the directions of the paths change, this trend
reverses. Further, for the locations near the apex of
the head (bottom of the plot), the primary direction of
motion is more aligned with the Z axis (vertical on
the plot). That is, the figure eight and loop patterns
are longer and flatter along the Z axis. This local long
axis of the NDT patterns is referred to as the major
displacement axis, similar to the description of an
ellipse. Not all NDT displacement patterns have a
major axis. The major displacement axis rotates in
this manner from more aligned with the X axis of the
skull (horizontal on the plot) near the c.g. to more
aligned with the Z axis near the apex of the head for
 Hardy et al. / Stapp Car Crash Journal 45 (November 2001)
the targets in both the anterior (left) and posterior
(right) columns. The magnitude of the displacements
also seems to lessen slightly when progressing from
the c.g. to the apex of the head. The displacement
patterns do not return to zero because the head moved
out of the field of view of the image intensifiers
during the test, but the trend back toward the initial
position can be seen.
Figure 7 provides time histories of the head
kinematics as measured by the nine-accelerometer
array, for test C755-T2. The component linear
accelerations at the head c.g. are grouped together in
the top-left plot. The angular acceleration
components are grouped together in the top-right
plot. The angular speeds and skull rotation in the X-Z
plane (about the Y axis) are below the linear and
angular accelerations respectively. Figure 8 provides
the X and Z direction time histories of the
displacements at NDT locations a1, a5, p1, and p5
(i.e., “a” indicating anterior, or left column, and 1
being close to the c.g. of the head, or top of the plot).
To relate these displacements to the motion of the
skull, they are compared to the angular speed about
the Y axis. After forward flexion begins, (positive Y-
axis angular speed, near 10 ms), positive X-axis
displacement of the NDT-a1 and NDT-p1 locations
begins. This is accompanied by minus Z-axis
displacement. This can be envisioned as the brain
lagging the skull motion. As the skull rotation begins
to slow, the brain displacement continues slightly. As
the skull rotation continues to slow, the brain
displacement at these locations returns to its initial
position (near 38 ms), and deforms past its initial
position. After rotation of the skull ceases, the brain
begins the return to its initial position. These
observations are consistent throughout the
experiments, and are described in more detail in
association with the C383 series of tests.
Figure 9 shows the displacement data for test C755-
T3. The first significant observation is that the initial
NDT locations are the same as those for test C755-
T2. This provides confidence in the ability of the
targets to maintain their position with respect to
surrounding brain tissue, and in the effectiveness of
the perfusion system. The second significant
observation is that the alignment of the major
displacement axes is similar to the patterns of test
C755-T2. The magnitudes and directions along the
major displacement axes are similar to those of test
C755-T2 as well. However, the directions along the
minor axes of displacement are largely the reverse of
the previous test, giving the displacements a more
open appearance.
9
Figure 10 provides the kinematics for test C755-T3,
and Figure 11 shows the X and Z direction
displacements for selected targets. The trends for this
test are essentially the same as those for test C755-
T2, with the minor exception that the displacements
at the locations near the apex and c.g. of the head
oppose each other slightly earlier for test C755-T3. It
is important to note that the X and Z direction
displacements oppose each other between the anterior
and posterior target columns for positions in the
middle of each column as well, where the major
displacement axes are aligned with the Z axis. Tests
C755-T2 and C755-T3 were conducted using the
same cadaver, and both were occipital impacts. Both
tests produced visually similar results.
Figure 12 shows the displacement data for test C755-
T4. The beginning of the data for each NDT location
is missing, so the initial positions appear shifted
when they in fact are not. The characteristics of the
displacement shapes for this test are nearly identical
to those of test C755-T2. This suggests that the
measured displacements are not random effects, but
reflect recurring physical phenomena associated with
the response of the brain to impact.
Figure 13 shows the displacement data for test C755-
T5, which was a frontal blow delivered by the
impactor. Given the initial orientation and position of
the head with respect to the impactor, the initial head
rotation was forward flexion. This forward head
rotation was significantly longer than the initial
backward head rotation experienced during the
occipital blows. Figure 14 provides the kinematics
for this test, and Figure 15 provides the X and Z
direction displacements for selected targets.
Figure 16 shows the displacement data from the first
test conducted on the third specimen, test C383-T1.
This is the first test using the new VR4 cameras. Test
C383-T1 is a deceleration test in which the moving
head struck a fixed-block inclined 45 degrees.
Although these data are from a frontal blow, the
shapes look very much like those from C755-T2,
which is an occipital impact. However, closer
inspection reveals that the directions of the
displacements are reversed, as expected. It is
interesting to note that many of the figure-eight
patterns of both of these tests essentially transcribe a
counterclockwise loop followed by a clockwise loop.
Figure 17 provides the kinematics data for test
C383-T1. Since the specimen remained largely in the
field of view for much longer than during the
10 Hardy et al. / Stapp Car Crash Journal 45 (November 2001)
impactor tests, brain displacement data were obtained
for much greater duration. Because of this, not only
was the starting phase of head rotation captured, but
the stopping phase was captured as well. Another
difference between this type of test and the impactor
tests is that the angular speed during the impactor
tests is roughly haversine, while during the fixed-
block tests angular speed has a distinct plateau. This
start-plateau-stop response is evident in the angular
acceleration and angular speed plots shown in Figure
17. Another effect evident in these data is small
forward rotation of the head upon contact with the
fixed block. The head motion then becomes
backward rotation for the majority of the test. This is
similar to the small initial forward head rotation seen
during occipital-blow impactor tests.
Figure 18 provides the X and Z direction
displacements for selected targets. The initial forward
rotation of the head was captured. The brain
displacements are again compared to angular speed
of the skull. Initially the skull experiences slight
forward rotation, and then begins the dominant
motion, which is rearward rotation. The brain
displacements lag this initial skull motion. As speed
(and magnitude) of the rearward rotation of the skull
increases (point “a” in Figures 16, 17 and 18), the
brain displacements reverse direction in response to
this motion, but lag the motion of the skull. As
angular speed of the skull continues to increase at a
decreasing rate the brain displacements peak (point
“b”) and begin to reduce. As the angular speed of the
skull plateaus (point “c”), the brain has essentially
relaxed to its initial shape. As the skull rotation
begins to slow (point “d”), the brain continues on,
displacing in the opposite direction. These
displacements peak about the time the skull stops
rotating (point “e”), and relax back to the initial
positions after the skull has stopped rotating (end of
the traces). It is possible that these displacements
may oscillate for some time after the skull stops
rotating. Figure 18 also shows that the displacements
at the selected NDT positions toward the apex and
c.g. of the head tend to oppose each other throughout
the test.
Figure 19 shows the displacement data for test C383-
T2. Similar to test C383-T1, this is a frontal blow,
deceleration test. Like the results from the C755
series of tests, these similar tests produced similar
displacement patterns within the C383 test series.
Like the C755 series, the initial positions of the
NDTs are the same from one test to the next. The
direction of the displacements and the relationship of
the major displacement axes follow the same trends
as test C383-T1. One minor difference is in the more
open nature of the loops.
Figure 20 shows the displacement data for test C383-
T3. It is visually similar to the results from test
C383-T2 shown in Figure 19. The NDT locations
marked with an asterisk in Figure 20 could not be
tracked for as long as the other targets on the plot.
The effects shown in Figures 20, 21, and 22 for test
C383-T3 are nearly identical to those described for
test C383-T1, and do not require additional
description.
Figure 23 shows the displacement data for test C383-
T4, which is an occipital blow. This is the final test in
which the head was decelerated against the fixed
block. The displacement data follow all of the same
trends as discussed for all of the previous tests. The
patterns are reversed because the head motion is
reversed. As shown in Figure 24, there is no plateau
of angular speed, and the displacement time-histories
appear much like those of the impactor tests. Figure
25 shows the associated brain displacements. Initially
the skull experiences slight rearward rotation, and
then begins the dominant motion, which is forward
rotation. The brain displacements lag this initial skull
motion. As speed (and magnitude) of the forward
rotation of the skull increases (point “a” in Figures
23, 24 and 25), the brain displacements reverse
direction in response to this motion, but lag the
motion of the skull. As angular speed of the skull
peaks, the brain tends toward its initial configuration.
The brain then does return to its initial configuration
as the skull rotation slows (point “c”), and then
displaces beyond its initial configuration as the head
rotation stops (point “d”). The brain finally tends
back towards its initial configuration (end of traces).
 Hardy et al. / Stapp Car Crash Journal 45 (November 2001) 11

C.G.

Inferior

Superior

Anterior Posterior

Figure 6. Sagittal plane brain displacement patterns at the NDT locations for test C755-T2, in
which the head was struck in the occipital region by a padded impactor. The arrows show the
direction of displacement for selected targets.
12 Hardy et al. / Stapp Car Crash Journal 45 (November 2001)

C755-T2 Linear X C755-T2 Angular X

3000
Y Y
60
Z Z

2000
40

Acceleration (r/s/s)
Acceleration (g)

1000
20
0

0
15 30 45 60 75 90 105 120 15 30 45 60 75 90 105 120

-1000
-20

-2000
-40

-3000
-60

0 0
Time (ms) Time (ms)

C755-T2 Angular X C755-T2 Angular Y

Y

75
30

50
20

Displacement (deg)
25
10
Speed (r/s)

0
0

15 30 45 60 75 90 105 120 15 30 45 60 75 90 105 120

-25
-10

-50
-20

-75
-30

0 0
Time (ms) Time (ms)

Figure 7. The head kinematics from test C755-T2 (occipital): Linear acceleration components (upper left), angular acceleration
components (upper right), angular speed components (lower left), and Y-axis rotation (lower right).

C755-T2 X NDT-a1 C755-T2 Z NDT-a1

NDT-a5 NDT-a5
12

12

NDT-p1 NDT-p1
NDT-p5 NDT-p5
8

8
Displacement (mm)

Displacement (mm)
4

4
0

15 30 45 60 75 90 105 120 15 30 45 60 75 90 105 120

-4

-4
-8

-8
-12

-12

0 0
Time (ms) Time (ms)

Figure 8. Displacement time histories for NDT locations a1, a5, p1, and p5 in the X (left) and Z (right) directions for test
C755-T2 (occipital).
 Hardy et al. / Stapp Car Crash Journal 45 (November 2001) 13

C.G.

Inferior

Superior

Anterior Posterior

Figure 9. Sagittal plane brain displacement patterns at the NDT locations for test C755-T3, in
which the head was struck in the occipital region by a padded impactor. The arrows show the
direction of displacement for selected targets.
14 Hardy et al. / Stapp Car Crash Journal 45 (November 2001)

C755-T3 Linear X C755-T3 Angular X

3000
Y Y
60
Z Z

2000
40

Acceleration (r/s/s)
Acceleration (g)

1000
20
0

0
15 30 45 60 75 90 105 120 15 30 45 60 75 90 105 120

-1000
-20

-2000
-40

-3000
-60

0 0
Time (ms) Time (ms)

C755-T3 Angular X C755-T3 Angular Y

Y

75
30

50
20

Displacement (deg)
25
10
Speed (r/s)

0
0

15 30 45 60 75 90 105 120 15 30 45 60 75 90 105 120

-25
-10

-50
-20

-75
-30

0 0
Time (ms) Time (ms)

Figure 10. The head kinematics from test C755-T3 (occipital): Linear acceleration components (upper left), angular acceleration
components (upper right), angular speed components (lower left), and Y-axis rotation (lower right).

C755-T3 X NDT-a1 C755-T3 Z NDT-a1

NDT-a5 NDT-a5
12

12

NDT-p1 NDT-p1
NDT-p6 NDT-p6
8

8
Displacement (mm)

Displacement (mm)
4

4
0

15 30 45 60 75 90 105 120 15 30 45 60 75 90 105 120

-4

-4
-8

-8
-12

-12

0 0
Time (ms) Time (ms)

Figure 11. Displacement time histories for NDT locations a1, a5, p1, and p5 in the X (left) and Z (right) directions for test
C755-T3 (occipital).
 Hardy et al. / Stapp Car Crash Journal 45 (November 2001) 15

C.G.

Inferior

Superior

Anterior Posterior

Figure 12. Sagittal plane brain displacement patterns at the NDT locations for test C755-T4, in
which the head was struck in the occipital region by a padded impactor.
16 Hardy et al. / Stapp Car Crash Journal 45 (November 2001)

C.G.

Inferior

Superior

Anterior Posterior

Figure 13. Sagittal plane brain displacement patterns at the NDT locations for test C755-T5, in
which the head was struck in the frontal region by a padded impactor. The arrows show the
direction of displacement for selected targets.
 Hardy et al. / Stapp Car Crash Journal 45 (November 2001) 17

C755-T5 Linear X C755-T5 Angular X

3000
Y Y
60
Z Z

2000
40

Acceleration (r/s/s)
Acceleration (g)

1000
20
0

0
15 30 45 60 75 90 105 120 15 30 45 60 75 90 105 120

-1000
-20

-2000
-40

-3000
-60

0 0
Time (ms) Time (ms)

C755-T5 Angular X C755-T5 Y

Y

75
30

50
20

Displacment (deg)
25
10
Speed (r/s)

0
0

15 30 45 60 75 90 105 120 15 30 45 60 75 90 105 120

-25
-10

-50
-20

-75
-30

0 0
Time (ms) Time (ms)

Figure 14. The head kinematics from test C755-T5 (frontal): Linear acceleration components (upper left), angular acceleration
components (upper right), angular speed components (lower left), and Y-axis rotation (lower right).

C755-T5 X NDT-a5 C755-T5 Z NDT-a5

NDT-p1 NDT-p1
12

12

NDT-p6 NDT-p6
8

8
Displacement (mm)

Displacement (mm)
4

4
0

15 30 45 60 75 90 105 120 15 30 45 60 75 90 105 120

-4

-4
-8

-8
-12

-12

0 0
Time (ms) Time (ms)

Figure 15. Displacement time histories for NDT locations a5, p1, and p5 in the X (left) and Z (right) directions for test
C755-T5 (frontal).
18 Hardy et al. / Stapp Car Crash Journal 45 (November 2001)

C.G.

a, c, d e
Inferior

Superior

Anterior Posterior

Figure 16. Sagittal plane brain displacement patterns at the NDT locations for test C383-T1, in which the
frontal region of the moving head struck a fixed block. The arrows show the direction of displacement for
selected targets. Points “a” through “e” correspond to those shown in Figures 17 and 18.
 Hardy et al. / Stapp Car Crash Journal 45 (November 2001) 19

C383-T1 Linear C383-T1 Angular X

3000
X
60
Y Y
Z Z

2000
40

Acceleration (r/s/s)
1000
Acceleration (g)
20

0
0

15 30 45 60 75 90 105 120 15 30 45 60 75 90 105 120

-1000
-20

-2000
-40

-3000
-60

0 0
Time (ms) Time (ms)

C383-T1 Angular X
C383-T1 Angular Y

75
Y
30

50
20

Displacement (deg)
25
10
Speed (r/s)

0
0

15 30 45 60 75 90 105 120 15 30 45 60 75 90 105 120

-25
-10

-50
-20

-75
-30

a b c d e 0
0
Time (ms) Time (ms)

Figure 17. The head kinematics from test C383-T1 (frontal): Linear acceleration components (upper left), angular acceleration
components (upper right), angular speed components (lower left), and Y-axis rotation (lower right). Points “a” through “e”
correspond to those in Figures 16 and 18.

C383-T1 X NDT-a1 C383-T1 Z NDT-a1

NDT-a6 NDT-a6
12

12

NDT-p1 NDT-p1
NDT-p6 NDT-p6
8

8
Displacement (mm)

Displacement (mm)
4

4
0

15 30 45 60 75 90 105 120 15 30 45 60 75 90 105 120

-4

-4
-8

-8

a b c d e
-12

-12

0 0
Time (ms) Time (ms)

Figure 18. Displacement time histories for NDT locations a1, a6, p1, and p6 in the X (left) and Z (right) directions for test
C383-T1 (frontal). Points “a” through “e” correspond to those in Figures 16 and 17.
20 Hardy et al. / Stapp Car Crash Journal 45 (November 2001)

C.G.

Inferior

Superior

Anterior Posterior

Figure 19. Sagittal plane brain displacement patterns at the NDT locations for test C383-T2, in
which the frontal region of the moving head struck a fixed block.
 Hardy et al. / Stapp Car Crash Journal 45 (November 2001) 21

C.G.

Inferior

Superior

Anterior Posterior

Figure 20. Sagittal plane brain displacement patterns at the NDT locations for test C383-T3, in
which the frontal region of the moving head struck a fixed block. The arrows show the direction of
displacement for selected targets. * Indicates NDT locations for which the tracking was incomplete.
22 Hardy et al. / Stapp Car Crash Journal 45 (November 2001)

C383-T3 Linear X C383-T3 Angular X

3000
Y Y
60
Z Z

2000
40

Acceleration (r/s/s)
Acceleration (g)

1000
20
0

0
15 30 45 60 75 90 105 120 15 30 45 60 75 90 105 120

-1000
-20

-2000
-40

-3000
-60

0 0
Time (ms) Time (ms)

C383-T3 Angular X
C383-T3 Angular Y

75
Y
30

50
20

Displacement (deg)
25
10
Speed (r/s)

0
0

15 30 45 60 75 90 105 120 15 30 45 60 75 90 105 120

-25
-10

-50
-20

-75
-30

0 0
Time (ms) Time (ms)

Figure 21. The head kinematics from test C383-T3 (frontal): Linear acceleration components (upper left), angular acceleration
components (upper right), angular speed components (lower left), and Y-axis rotation (lower right).

C383-T3 X NDT-a1 C383-T3 Z NDT-a1

NDT-a6 NDT-a6
12

12

NDT-p6 NDT-p6
8

8
Displacement (mm)

Displacement (mm)
4

4
0

15 30 45 60 75 90 105 120 15 30 45 60 75 90 105 120

-4

-4
-8

-8
-12

-12

0 0
Time (ms) Time (ms)

Figure 22. Displacement time histories for NDT locations a1, a6, p1, and p6 in the X (left) and Z (right) directions for test
C383-T3 (frontal).
 Hardy et al. / Stapp Car Crash Journal 45 (November 2001) 23

C.G.

c
Inferior

b d
a

Superior

Anterior Posterior

Figure 23. Sagittal plane brain displacement patterns at the NDT locations for test C383-T4, in
which the occipital region of the moving head struck a fixed block. The arrows show the direction of
displacement for selected targets. Points “a” through “d” correspond to those in Figures 24 and 25.
24 Hardy et al. / Stapp Car Crash Journal 45 (November 2001)

C383-T4 Linear C383-T4 Angular

12000
120 X X
Y Y
Z Z

8000
80

Acceleration (r/s/s)
Acceleration (g)

4000
40
0

0
15 30 45 60 75 90 105 120 15 30 45 60 75 90 105 120

-4000
-40

-8000
-80

-12000
-120

0 0
Time (ms) Time (ms)

C383-T4 Angular X C383-T4 Angular Y

Y

75
30

50
20

Displacement (deg)
25
10
Speed (r/s)

0
0

15 30 45 60 75 90 105 120 15 30 45 60 75 90 105 120

-25
-10

-50
-20

a b c d
-75
-30

0 0
Time (ms) Time (ms)

Figure 24. The head kinematics from test C383-T4 (occipital): Linear acceleration components (upper left), angular acceleration
components (upper right), angular speed components (lower left), and Y-axis rotation (lower right). Points “a” through “d”
correspond to those in Figures 23 and 25.

C383-T4 X NDT-a1 C383-T4 Z NDT-a1

NDT-a6 NDT-a6
12

12

NDT-p1 NDT-p1
NDT-p6 NDT-p6
8

8
Displacement (mm)

Displacement (mm)
4

4
0

15 30 45 60 75 90 105 120 15 30 45 60 75 90 105 120

-4

-4
-8

-8
-12

-12

0
a b c d 0
Time (ms) Time (ms)

Figure 25. Displacement time histories for NDT locations a1, a6, p1, and p6 in the X (left) and Z (right) directions for test
C383-T4 (occipital). Points “a” through “d” correspond to those in Figures 23 and 24.
 Hardy et al. / Stapp Car Crash Journal 45 (November 2001)
DISCUSSION
The brain displacement measurements made in this
study represent a significant improvement over the
techniques employed in earlier studies, and are
essentially the first of their kind. Both general skull
kinematics and local brain displacements were
measured and compared to each other; the brain
displacement output resulting from the skull
kinematics input. It is important to note that the
relationship between the brain displacements and
potential brain injury are unknown.
Brain Displacement and Skull Motion
These tests show that when the human cadaver head
is subjected to impact, the brain undergoes
displacement with respect to the skull. For these low-
severity tests, the relative displacement between the
brain and skull are on the order of +/- 5mm, as
illustrated graphically.
The displacements of the brain with respect to the
skull have been described in relation to the overall
motion of the skull. The angular speed was found to
be the most convenient measure for basis of
comparison. One key observation is that the
displacements of the brain seem to lag the motion of
the skull. Another important observation is that the
brain seems to return to its initial configuration.
It is important to note that all of the brain
displacements are described with respect to a moving
reference frame. This reference frame is the body-
fixed basis of the skull. This coordinate system
translates and rotates with the skull. The relative
brain/skull displacements must be observed with this
moving coordinate system in mind.
Experimental Error and Limitations
The methods of this study are laborious,
computationally intensive, and not without some
error. The greatest specimen-preparation concerns
were related to coupling between the brain and the
skull. The primary issue was the evacuation of gasses
from the intracranial space. The specimens were
obtained and prepared very quickly after death in an
effort to avoid the effects of hemolysis, accumulation
of gasses, and tissue degradation. No air was visible
on the test x-rays, and the perfusion device was left in
operation throughout testing (approximately eight
hours per series). Elimination of air was the primary
impetus for conducting the tests in an inverted
configuration. The effects of baseline intracranial
pressure (ICP) were unknown, but ICP was closely
regulated. The effects of small air bubbles,
particularly those trapped beneath the tentorium, will
25
be investigated using modeling. The effects of large
amounts of air are to be demonstrated in the future by
deactivating the perfusion device at the end of a test
series for some specimens, and then conducting an
additional test.
The greatest data processing concerns were related to
measurement of skull target locations in anatomical
coordinates from pretest x-rays. Error in this
measurement does not introduce relative errors
between sequential points within the displacement
path at given target locations, nor does it introduce
relative errors between points along different paths. It
causes absolute errors in initial position and path
direction. This error was checked by processing a few
NDT locations for test C383-T1 using two very
different sets of skull targets. One set of targets
consisted of three in a triangle on the right side of the
skull. The other set consisted of the auditory meatus
target on the left side of the skull and both
infraorbital ridge targets (which were actually rotated
around the lateral aspect of the zygoma). The results
of this comparison are shown in Figure B1, in
Appendix B. The degree of agreement between the
two approaches illustrates the integrity of these data.
Another potential source of error is the manual
enhancement of the target locations for the C755 test
series, and in the tracking and computational
techniques for all tests. One way to assess these
errors is to look at the distance between targets on the
skull. Variations in the distance between targets used
to define the rigid-body motion of the skull would
give an indication of error due to measurement
limitations and error due to flexure of the skull. The
average distances between two skull targets used for
each test, along with standard deviations and standard
errors, are presented Table B1 in Appendix B. The
standard errors range from 0.99 to 2.14 percent for
the C755 tests, and from 0.19 to 0.37 percent for the
C383 tests. The larger errors experienced during the
C755 tests are due to the manual enhancement
approach and the lower resolution of the JCL
cameras used for these tests. These errors are
considered insignificant.
The filtering techniques used to disclose the major
trends for the C755 test series also can introduce
error. Data for representative targets from tests C755-
T2 and T3 are presented in Figures B2 and B3
respectively, in various stages of processing. In
Figures B2 and B3 the top plots are the raw data, the
middle plots are the filtered data, and the bottom
plots are filtered and interpolated data. The columns
on the left of Figures B2 and B3 are sagittal-
perspective plots of motion at the NDT locations, and
26 Hardy et al. / Stapp Car Crash Journal 45 (November 2001)
the columns on the right are displacement-time
histories of these motions. These plots show that the
effects of signal processing are minimal with regard
to the start of the data, particularly in the X direction.
This indicates that the initial X-direction responses
are not the result of filtering, and shows that any
miniscule changes in the initial Z-direction responses
are below the measurement resolution and are
therefore meaningless. While the jitter and some of
the higher-frequency displacements are removed by
this procedure, the important trends are preserved and
are easier to visualize. It is important to note that the
data from the C383 test series were processed
differently, but exhibit trends similar to those seen in
the filtered C755 series data, which reinforces
confidence in the C755 test series data. Those
engaged in modeling efforts can follow the same
filtering regimes and get the same results.
A much greater effect than any filtering artifacts
would be the inexactness of the contact time
determination and subsequent data synchronization
for the C755 test series. This error is related to the
slow frame rate of the JCL cameras used for the
C755 test series, and cannot be determined exactly.
However, the synchronization errors between the
brain displacement data and the skull kinematics data
could range from 0 ms to 4 ms, with the displacement
data beginning too late.
The effects of target size, mass, and density are
unknown, but the NDTs were made as small as
possible, with their density as close to that of brain
tissue as possible. Table C1 in Appendix C provides
the initial positions in three dimensions of the NDTs
with respect to the c.g. of the head specimens. The
initial positions changed little between tests of the
same specimen. The average standard deviation of
the initial positions in three directions for specimens
C755 and C383 are 0.36 mm and 0.56 mm
respectively. The larger difference for specimen
C383 is expected because different sets of skull
targets were used for different tests of this specimen.
Since the NDTs returned to the same initial location
between tests on a given specimen, it is thought that
the NDTs were capable of maintaining their position
with respect to surrounding brain tissue. Since the
brain displacement patterns were similar for similar
tests, even using different specimens, it is thought
that the data are robust. In addition, at autopsy the
NDTs were found to have been enveloped by the
brain tissue, and the implant cannula tracks had
essentially disappeared. The effects of small changes
in target geometry and mass will be investigated
using finite element modeling.
Future Research
The translational inputs used in this current study are
relatively small, and the tests essentially represent
combined translation and rotation of the head. It is
therefore not possible to completely assess the effects
of translation using the results of this study. Future
studies are planned in which the relative effects of
head translation and rotation of varying severity can
be better analyzed separately. It is also planned to
study the effects of head translation and rotation in
different directions of impact.
The classic intracranial pressure study by Nahum et
al. (1977), and the more recent study by Trosseille et
al. (1992) are frequently referenced. Because of
technical limitations, pressure measurements could
not be made as part of this study. However, the
current protocol calls for the use of cranial pressure
transducers (CPTs), and the results will be reported in
the future.
Due to technical limitations, this current study could
not be designed to measure shear strain or principal
strain, although these measurements were an initial
goal of the study. Future plans include determining
stretch between targets using the data generated from
this study, and placing targets within the brain such
that the full strain tensor can be developed at various
locations.
McGowan, et al. (1999) examined the use of
magnetic resonance imaging (MRI) for disclosing
DAI within the brain. The magnetic transfer ratio
(MTR) technique was applied using miniature swine.
The swine were subjected to high-rates of angular
acceleration through coronal rotation of the head. The
MTR technique was said to be sensitive to damaged
myelin, which is associated with DAI. Relating the
development of DAI to mechanical input using the
MTR technique is a potential future direction of this
research effort.
CONCLUSION
Understanding of head injury will lead to more
effective treatment and prevention. Relative motion
between the brain and skull has been measured in the
human cadaver head using neutral density targets and
a high-speed, biplanar x-ray system. This motion has
been described with respect to the body-fixed basis of
the skull, which both translates and rotates with the
movement of the skull. Considerable attention was
paid to testing the specimens shortly after death and
to the evacuation of air from the intracranial space.
Varying levels of resultant linear and angular
 Hardy et al. / Stapp Car Crash Journal 45 (November 2001)
acceleration were produced during low-speed,
occipital and frontal impacts. Within this limited
series of tests:
• A unique system for three-dimensional
measurement of displacement within the human
cadaver head under impact conditions has been
developed.
• When the human cadaver head is subjected to
impact, the brain undergoes displacement
relative to the skull, which for these tests is on
the order of +/- 5 mm.
• For a given specimen, the initial neutral density
target locations do not change between tests,
providing confidence that the targets maintain
their position with respect to surrounding brain
tissue.
• No accumulation of gas within the intracranial
space was visible on the high-speed radiographs
during testing, suggesting a well-coupled system.
• The displacements observed during these tests
can be described as either figure-eight patterns or
tight loops.
• The displacement patterns seen during multiple
tests of a given specimen or during comparable
tests of different specimens were visually
similar, reinforcing confidence in the
observations.
ACKNOWLEDGMENTS
This work was conducted under the auspices of the
Bioengineering Center at Wayne State University and
was supported in part by the Centers for Disease
Control and Prevention, National Center for Injury
Prevention and Control Grant No. R49/CCR503534-
12. We also acknowledge the use of equipment
purchased with the support of the Alliance of
Automotive Manufacturers and Honda R.&D. Co.,
Inc. The efforts of the Bioengineering Center staff are
deeply appreciated, especially those of Dr. Ruhan
Iskender. The efforts of the staff of the Motion
Analysis Laboratory at the Henry Ford Hospital Bone
and Joint Specialty Center, in particular William
Anderst, are also appreciated. The authors wish to
thank Svein Kleiven of the Royal Institute of
Technology, Stockholm, for his valuable assistance.
27
REFERENCES
Al-Bsharat, A.S., Hardy, W.N., Yang, K.H., Khalil,
T.B., King, A.I., and Tashman, S. (1999)
Brain/skull relative displacement magnitude due to
blunt head impact: New experimental data and
model. Proc. 43rd Stapp Car Crash Conference, pp.
321-332. Society of Automotive Engineers,
Warrendale PA.
Gosch, H.H., Gooding, E., and Schneider, R.C.
(1969) Distortion and displacement of the brain in
experimental head injuries. Surgical Forum
20:425-426.
Hardy, W.N., Foster C.D., King, A.I., and Tashman
S. (1997) Investigation of brain injury kinematics:
Introduction of a new technique. Crashworthiness,
Occupant Protection and Biomechanics in
Transportation Systems AMD 225:241-254.
Hodgson, V.R., Gurdjian, E.S., and Thomas, L.M.
(1966) Experimental skull deformation and brain
displacement demonstrated by flash x-ray
technique. Journal of Neurosurgery 25:49-52.
Kallieris, D., Rizzetti, A., and Mattern, R. (1997).
Some observations to the skull-brain trauma. Proc.
AGARD Conference 597, pp. 1-4. Advisory Group
for Aerospace Research and Development,
Neuilly-Sur-Seine, FR.
McGowan J.C., McCormack T.M., Grossman R.I.,
Mendonca R., Chen X.H., Berlin J.A., Meaney
D.F., Xu B.N., Cecil K.M., McIntosh T.K., and
Smith D.H. (1999) Diffuse axonal pathology
detected with magnetization transfer imaging
following brain injury in the pig. Magnetic
Resonance in Medicine 41(4):727-733.
Nahum, A., Smith, R., and Ward, C. (1977)
Intracranial pressure dynamics during head impact.
Proc. 21st Stapp Car Crash Conference, pp. 337-
366. Society of Automotive Engineers,
Warrendale, PA.
Nusholtz, G.S., Lux, P., Kaiker, P.S., and Janicki,
M.A. (1984) Head impact response - Skull
deformation and angular accelerations. Proc. 28th
Stapp Car Crash Conference, pp. 41-74. Society of
Automotive Engineers, Warrendale, PA.
28 Hardy et al. / Stapp Car Crash Journal 45 (November 2001)
Pudenz, R.H. and Shelden, C.H. (1946) The Lucite
calvarium - a method for direct observation of the
brain. II. Cranial trauma and brain movement.
Journal of Neurosurgery 3:87-505.
Shatsky, S.A. (1973) Flash x-ray cinematography
during impact injury. Proc. 17th Stapp Car Crash
Conference, pp. 361-376. Society of Automotive
Engineers, Warrendale, PA.
Shatsky, S.A., Alter, W.A. III, Evans, D.E.,
Armbruster, V., and Clark, G (1976) Traumatic
distortions of the primate head and chest:
Correlation of biomechanical, radiological and
pathological data. Proc. 18th Stapp Car Crash
Conference, pp. 351-381. Society of Automotive
Engineers, Warrendale, PA.
Shelden, C.H., Pudenz, R.H., Restarski, J.S., and
Craig, W.M. (1944) The Lucite calvarium - a
method for direct observation of the brain. I. The
surgical and Lucite processing techniques. Journal
of Neurosurgery 1:67-75.
Stalnaker, R.L., Melvin, J.W., Nusholtz, G.S., Alem,
N.M., and Benson, J.B. (1977) Head impact
response. Proc. 21st Stapp Car Crash Conference,
pp. 303-335. Society of Automotive Engineers,
Warrendale PA.
Trosseille, X., Tarriere, C., Lavaste, F., Guillon F.,
and Domont, A. (1992) Development of a F.E.M.
of the human head according to a specific test
protocol. Proc. 36th Stapp Car Crash Conference,
pp. 235-253. Society of Automotive Engineers,
Warrendale, PA.
 Hardy et al. / Stapp Car Crash Journal 45 (November 2001) 29

APPENDIX A SAE channel-class 1-kHz 3-2-2-2 kinematics.

Table A1. Summary of cc 1000 Hz Head Kinematics
C731 C731 C755 C755 C755 C755 C383 C383 C383 C383
Test
T2 T3 T2 T3 T4 T5 T1 T2 T3 T4
Impact type accel. accel. accel. accel. accel. accel. decel. decel. decel. decel.
Impact area occipital occipital occipital occipital occipital frontal frontal frontal frontal occipital
X 23 28 22 24 28 25 8 11 11 1395
Linear -9 -11 -10 -5 -6 -16 -53 -42 -94 -1664
c.g. Y 3 5 4 3 5 6 12 27 28 791
accel. -5 -7 -8 -8 -8 -3 -3 -13 -8 -581
(g) Z 6 8 10 9 6 3 21 15 17 179
-10 -13 -15 -15 -19 -7 -45 -31 -38 -321
Resultant (g) 23 28 23 25 29 25 68 54 100 1669
HIC15ms 13 21 17 22 32 5 50 40 73 36695
X 304 419 284 181 319 404 424 2374 677 68553
-380 -428 -427 -418 -428 -314 -800 -1884 -1974 -57338
Angular
Y 1326 1746 1935 2063 2782 1336 3401 1978 4752 96117
accel.
(rad/s/s) -972 -1167 -755 -733 -733 -2947 -2312 -3668 -4984 -82444
Z 185 353 208 253 703 640 939 2238 2529 66270
-475 -640 -296 -242 -423 -676 -564 -1510 -2362 -73924
X 0 2 0 0 0 1 1 4 3 15
-1 -2 -3 -3 -2 -1 -3 -5 -6 -5
Angular
Y 18 22 18 19 17 4 5 2 4 28
speed
(rad/s) 0 0 -1 -2 -4 -7 -20 -22 -22 -26
Z 4 4 0 1 0 0 3 3 5 8
-2 0 -1 -1 -1 -2 -1 -4 -3 -23

APPENDIX B Error analyses.

C383-T1
0

40 35 30 25 20 15 10 5
5

Inferior
10
Z (mm)

15
20

Superior
25
30

0
Anterior X (mm) Posterior

Figure B1. Comparison of NDT paths calculated using two very different sets of skull
markers, which illustrates the robustness of the tracking and computational techniques.

Table B1. Estimate of Tracking and Computational Errors

Using Distance Between Skull Targets
Average 3D Distance Standard Deviation Standard Error
Cadaver Test
Between Targets (mm) (mm) (%)
T2 22.74 0.25 1.11
755 T3 21.27 0.46 2.14
T5 22.93 0.23 0.99
T1 35.00 0.12 0.35
383 T3 34.98 0.13 0.37
T4 131.41 0.25 0.19
30 Hardy et al. / Stapp Car Crash Journal 45 (November 2001)

APPENDIX B Continued.

C755-T2 Raw C755-T2 Raw NDT-a1 X

NDT-a1 Z

-6

6
NDT-a2 X
NDT-a2 Z
NDT-a3 X
NDT-a3 Z

4
0

Displacement (mm)
35 30 25 20 15 10 5 -5
Inferior

2
6
Z (mm)

12

0
8 16 24 32 40 48 56 64

18

-2
Superior

24

-4
30

-6
0 0
Anterior X (mm) Posterior Time (ms)

C755-T2 Filtered C755-T2 Filtered NDT-a1 X

NDT-a1 Z
-6

6
NDT-a2 X
NDT-a2 Z
NDT-a3 X
NDT-a3 Z

4
0

Displacement (mm)
35 30 25 20 15 10 5 -5
Inferior
2
6
Z (mm)

12

8 16 24 32 40 48 56 64
18

-2

Superior
24

-4
30

-6

0 0
Anterior X (mm) Posterior Time (ms)

C755-T2 Interpolated C755-T2 Interpolated NDT-a1 X

NDT-a1 Z
-6

NDT-a2 X
NDT-a2 Z
NDT-a3 X
NDT-a3 Z
4
0

Displacement (mm)

35 30 25 20 15 10 5 -5
Inferior
2
6
Z (mm)

12

8 16 24 32 40 48 56 64
18

-2

Superior
24

-4
30

-6

0 0
Anterior X (mm) Posterior Time (ms)

Figure B2. Various signal-processing steps for three different NDT locations from test C755-T2: raw data (top), filtered data
(middle), and filtered and interpolated data (bottom). The time-history data are shown on the right.
 Hardy et al. / Stapp Car Crash Journal 45 (November 2001) 31

APPENDIX B Continued.

C755-T3 Raw C755-T3 Raw NDT-a3 X

NDT-a3 Z

6
NDT-a4 X
6

35 30 25 20 15 10 5 -5 NDT-a4 Z

12

4
Displacement (mm)
Inferior

18

2
Z (mm)

24

0
8 16 24 32 40 48 56 64

30

-2
Superior

36

-4
42

-6
0 0
Anterior X (mm) Posterior Time (ms)

C755-T3 Filtered C755-T3 Filtered NDT-a3 X

NDT-a3 Z
6

6
NDT-a4 X
35 30 25 20 15 10 5 -5 NDT-a4 Z
12

4
Displacement (mm)

Inferior
18

2
Z (mm)

24

8 16 24 32 40 48 56 64
30

-2

Superior
36

-4
42

-6

0 0
Anterior X (mm) Posterior Time (ms)

C755-T3 Interpolated C755-T3 Interpolated NDT-a3 X

NDT-a3 Z
6

NDT-a4 X
35 30 25 20 15 10 5 -5 NDT-a4 Z
12

4
Displacement (mm)

Inferior
18

2
Z (mm)

24

8 16 24 32 40 48 56 64
30

-2

Superior
36

-4
42

-6

0 0
Anterior X (mm) Posterior Time (ms)

Figure B3. Various signal-processing steps for two different NDT locations from test C755-T3: raw data (top), filtered data
(middle), and filtered and interpolated data (bottom). Time-history data are shown on the right.
32 Hardy et al. / Stapp Car Crash Journal 45 (November 2001)

APPENDIX C Initial target locations for selected tests.

Table C1. NDT Initial Positions With Respect to the c.g. of the Head in Millimeters
NDT C755-T2 C755-T3 C755-T5 Average Std. Dev. C383-T1 C383-T3 C383-T4 Average Std. Dev.
X 13.60 12.95 - 13.28 0.46 27.23 27.65 28.44 27.77 0.61
a1 Y -31.48 -31.88 - -31.68 0.28 -39.06 -39.55 -38.61 -39.07 0.47
Z -3.66 -3.47 - -3.57 0.13 12.49 13.37 12.88 12.91 0.44
X 13.28 13.16 - 13.22 0.08 25.61 25.80 26.33 25.91 0.37
a2 Y -33.67 -34.09 - -33.88 0.30 -40.00 -39.87 -39.32 -39.73 0.36
Z 7.22 7.55 - 7.39 0.23 20.23 21.21 20.56 20.67 0.50
X 13.08 12.88 - 12.98 0.14 26.21 25.74 26.59 26.18 0.43
a3 Y -33.44 -34.88 - -34.16 1.02 -39.69 -40.08 -39.16 -39.64 0.46
Z 22.51 22.48 - 22.50 0.02 27.16 28.14 27.61 27.64 0.49
X 12.62 12.52 13.43 12.86 0.50 25.87 25.17 25.89 25.64 0.41
a4 Y -33.43 -34.16 -34.09 -33.89 0.40 -40.85 -40.94 -39.53 -40.44 0.79
Z 27.99 27.75 27.55 27.76 0.22 34.85 35.75 35.40 35.33 0.45
X 12.86 13.18 13.80 13.28 0.48 24.81 23.69 24.24 24.25 0.56
a5 Y -34.42 -35.55 -35.18 -35.05 0.58 -40.47 -39.98 -38.40 -39.62 1.08
Z 39.51 39.23 39.18 39.31 0.18 42.02 43.18 42.38 42.53 0.59
X - - - 23.40 22.29 23.20 22.96 0.59
a6 Y - - - -41.71 -41.39 -39.72 -40.94 1.07
Z - - - 49.50 50.38 49.69 49.86 0.46
X -26.34 -26.67 -26.10 -26.37 0.29 -42.57 -42.56 -40.67* -42.57 0.01
p1 Y -27.86 -27.76 -27.37 -27.66 0.26 -32.34 -33.38 -35.79* -32.86 0.74
Z -2.77 -2.50 -2.40 -2.56 0.19 0.35 -0.48 -2.84* -0.07 0.59
X -30.09 -29.83 -29.29 -29.74 0.41 -43.56 -43.47 -42.79 -43.27 0.42
p2 Y -28.00 -27.57 -27.59 -27.72 0.24 -32.86 -33.80 -33.17 -33.28 0.48
Z 7.02 7.50 7.54 7.35 0.29 9.89 8.90 9.56 9.45 0.50
X -33.80 - -32.65 -33.23 0.81 -43.43 -43.52 -42.16 -43.04 0.76
p3 Y -29.53 - -29.20 -29.37 0.23 -35.45 -36.32 -35.85 -35.87 0.44
Z 18.03 - 18.57 18.30 0.38 15.62 15.07 16.63 15.77 0.79
X -34.29 -33.72 -33.24 -33.75 0.53 -44.98 -45.36 -44.93 -45.09 0.24
p4 Y -29.90 -28.68 -29.47 -29.35 0.62 -36.17 -36.98 -36.46 -36.54 0.41
Z 26.97 27.38 27.86 27.40 0.45 24.99 24.18 25.73 24.97 0.78
X -37.22 -36.15 -35.95 -36.44 0.68 -45.60 -46.94 -45.80 -46.11 0.72
p5 Y -30.60 -29.61 -30.14 -30.12 0.50 -37.89 -38.54 -38.24 -38.22 0.33
Z 38.31 38.45 38.95 38.57 0.34 36.03 35.17 36.91 36.04 0.87
X - -36.57 -36.47 -36.52 0.07 -44.88 -46.63 -45.37 -45.63 0.90
p6 Y - -31.09 -31.89 -31.49 0.57 -37.49 -37.99 -37.31 -37.60 0.35
Z - 48.16 48.25 48.21 0.06 46.49 45.63 46.83 46.32 0.62
Average 0.36 0.56

*Note: This location was not included in the average calculation because it was not the same target as
NDT-p1 in the other two tests. It was a neighboring target (p0), tracked only during test C383-T4.