http://informahealthcare.com/jmt
ISSN: 0309-1902 (print), 1464-522X (electronic)
J Med Eng Technol, 2014; 38(1): 1–4
! 2014 Informa UK Ltd. DOI: 10.3109/03091902.2013.844207
INNOVATION
MR-based measurement of spinal cord motion during flexion of the
spine: implications for intradural spinal cord stimulator systems
S. Viljoen1, C. A. Smittkamp2, B. D. Dalm1, S. Wilson1, C. G. Reddy1, G. T. Gillies3, and M. A. Howard III*1
Department of Neurosurgery, 2Department of Radiology, University of Iowa Hospitals and Clinics, Iowa City, IO 52242, USA, and 3Department of
Mechanical and Aerospace Engineering, University of Virginia, Charlottesville, VA 22904, USA
Abstract
Keywords
This study develops a means of delivering electrical stimuli directly to the pial surface of the
spinal cord for treatment of intractable pain. This intradural implant must remain in direct
contact with the cord as it moves within the spinal canal. Therefore, magnetic resonance
imaging was used to measure the movement of the spinal cord between neutral and flexedback positions in a series of volunteers (n ¼ 16). Following flexion of the back, the mean change
in the pedicle-to-spinal cord dorsal root entry zone distance at the T10-11 level was (8.5 6.0)
mm, i.e. a 71% variation in the range of rostral-caudal movement of the spinal cord across all
patients. There will be a large spectrum of spinal cord strains associated with this observed
range of rostral-caudal motions, thus calling for suitable axial compliance within the electrode
bearing portion of the intradural implant.
Dorsal root entry zone, intradural implants,
neuromodulation, spinal cord, spinal cord
stimulator
1. Introduction and background
A limiting factor in the efficacy of epidural spinal cord
stimulation for the treatment of intractable pain is the inability
of the standard devices to selectively modulate the targeted
neural fibres without inadvertent stimulation of neighbouring
non-targeted structures, e.g. the dorsal nerve rootlets. This is a
consequence of the shunting effects of the relatively high
conductivity CSF located between the epidural electrodes and
the spinal cord surface, migration of the leads following
implantation, the formation of epidural scar tissue and several
other such problems. The result is limited therapeutic efficacy
in up to half of all patients having a standard epidural stimulator
[1]. The Human Spinal Cord Modulation System (HSCMS) is
being developed to improve this situation. It is an intradural
device [2] placed via durotomy directly onto the pial surface of
the spinal cord, thus avoiding the electrical shunting effects of
the CSF and lead migration issues and enabling more focused
delivery of the stimulus currents into the targeted regions [3].
The design features of the wired version of the implanted
component of this device are shown in Figure 1. The electrode
bearing surface at the bottom of the device makes contact
with the dorsal surface of the spinal cord. It consists of a thin
membrane (0.6 mm thick) of silicone into which are
embedded six stimulating electrodes. The individual leads
from the electrodes are configured in loops which join to form
a single lead cable that exits the spinal canal. This cable is
held in a stable position relative to the spinal canal by a
*Corresponding author. Email: [email protected]
History
Received 24 April 2013
Revised 3 September 2013
Accepted 8 September 2013
titanium strap (termed the Oya Strap) that is secured to the
lateral margins of the boney spinal canal. The dura is closed
underneath the Oya Strap, forming a water-tight seal that
prevents CSF leak. The restoring forces in the compressed
lead loops inside the subdural space create a gentle pressure
on the electrode membrane [4] (515 mm Hg, which is within
the range of normal intrathecal pressures) to keep it in contact
with the spinal cord as it moves within the spinal canal. It is
important that this membrane remains stably in place and
does not lift off the surface during rostral-caudal displacement
of the cord or slide laterally from its midline position, the
latter to prevent potentially harmful contact with the dorsal
root entry zones (DREZ).
Therefore, the lead loops of the device must be designed
such that they accommodate all of the relative motion
between the implant on the spinal cord surface and the lead
exit point which is at a fixed position at the dural exit site
under the Oya strap. Moreover, the membrane must conform
to the shape of the dorsal surface of the spinal cord in order
for the electrodes to make uniform direct contact with the pial
surface. This means that the radius of curvature of the
membrane must closely approximate that of the oval-shaped
spinal cord. It should also subtend as large an angle as
possible of the dorsal arc, while leaving a small buffer zone
on either side that separates the lateral edges of the membrane
from the dorsal rootlets. These requirements call for knowledge of (1) the spinal cord’s thoracic morphology and (2) its
axial motion within the spinal canal as a function of spinal
flexion. The first point was addressed via direct image-based
determination of the inter-DREZ dorsal arc lengths found
20
13
J Med Eng Technol Downloaded from informahealthcare.com by Miss Sandy Dalgleish on 12/16/13
For personal use only.
1
J Med Eng Technol Downloaded from informahealthcare.com by Miss Sandy Dalgleish on 12/16/13
For personal use only.
2
S. Viljoen et al.
Figure 1. Artist’s rendering of the HSCMS device. The 1.5 cm long
electrode-bearing silicone membrane at the bottom of the device is
placed onto the pial surface of the spinal cord. The leads from the
electrodes loop upwards, three per side, to meet and exit as a bundle
through the titanium Oya Strap which serves as the fixation structure for
the device, attaching it to the surrounding vertebra. A dural seal (not
shown) encloses the lead loops and membrane within the dura and allows
for leak-free passage of the lead bundle through the exit port on the Oya
Strap. An implantable pulse generator (not shown) is located in a nearby
subcutaneous pocket. The lead bundle connects to it and conveys the
electrical stimuli to the electrodes.
across a large number (n ¼ 50) of patients [5]. The present
study addresses the second point through implementation of
an experimental protocol aimed at measuring the flexiondriven rostral-caudal movement of the spinal cord.
2. Materials and methods
2.1. Patient position within the magnet
In order to allow for spinal flexion during imaging, a 1.5 T
Magnetom Espree (Siemens, Erlangen, Germany) magnet was
utilized to take advantage of its larger bore size, which was
designed to accommodate bariatric patients. Informed consent
was obtained from a total of 16 healthy volunteers ranging in
age from 23–58. Each volunteer was first imaged in a supine
neutral position and then imaged in a maximal attainable flexed
position. To obtain the maximal flexion of the spine, patients
were given three basic positioning instructions. The first was to
rotate their hips/pelvis backwards towards the gantry as far as
possible to remove the lumbar lordosis and straighten the
lumbar spine. The second was to curl their upper back, neck and
head forward so that their shoulders were as close to their knees
as possible. The third instruction was then to tuck their chin
down as close to their chest as possible. While attaining this
flexed position in the bore, a variety of foam wedges and
pillows were utilized for added support so that the patient could
remain as still as possible during image acquisition. Maximal
flexion was limited by volunteer flexibility in 14 of the patients.
In only two patients was flexion limited by MR bore size. Not
surprisingly, these were also the two tallest patients with
heights of 2.03 and 1.98 m (60 800 and 60 600 ).
2.2. Imaging protocol
Each volunteer had a vitamin E capsule taped to their midline
lower thoracic spine for help in level localization. A sagittal
HASTE sequence was performed initially as a localizer both for
vertebral level counting and identification of a more focal field
of view centred over the region of lowest thoracic spinal nerves
and the conus medullaris. In order to acquire anatomic images
with enough resolution to accurately measure intervals
J Med Eng Technol, 2014; 38(1): 1–4
between spinal nerve dorsal root entry zones, a CISS sequence
was selected for its high spatial resolution. Although this is a
highly T2-weighted sequence, acquisition time still required
2 min 5 s. This length of time initially caused too much motion
degradation during flexed imaging to make accurate measurements. The use of the pillows and foam wedges provided just
enough support for volunteers to remain still and in a state of
quiet respiration for the required 2 min duration. This
minimized breathing-related artifacts which, like the residual
motions induced by the cardiac cycle, otherwise limit the
imaging resolution that can be obtained. Hence, the measurements we have are technically an average during quiet
respiration. All neutral and flexed sequences were obtained
utilizing
TR ¼ 4.35 ms,
TE ¼ 2.18 ms,
slice
thickness ¼ 0.8 mm, matrix size ¼ 192 192, one acquisition per
average, 192 phase encoding steps, field of view ¼ 200 mm and
a 70 flip angle.
2.3. Measurement algorithm
Imaging was obtained in the coronal plane. Three-dimensional multiplanar reconstruction software was utilized on a
Carestream PACS station to aid in measurement. The T10 and
T11 nerve roots were identified. A cranial caudal measurement was made in a plane parallel to the spinal canal between
the DREZ of T10 and T11. (The exact position of the entry
zones was confirmed by assessing sequential axial images to
identify the most cranial aspect of the nerve originating from
the spinal cord.) As shown in Figure 2, the difference between
this measurement on the neutral and flexed images is a
measure of spinal cord contraction/expansion along the
rostral-caudal axis. Next, a cranial caudal measurement was
made from the DREZ of the T10 nerve root along the same
plane as the prior measurement, to the level of a plane
orthogonal to the spinal canal at the level of the inferior T10
pedicles. The latter were selected as a reference point of the
bony canal inside of which the spinal cord moves. The
difference between these measurements represents cord
movement within the bony canal. An example of a coronal
image on which the relevant anatomical features are identified
is shown in Figure 3.
Lastly, a cranial caudal measurement of the change in
conus tip position was made. In order to accurately accomplish this, the position of the conus tip was first identified on
the neutral images with reference to a landmark within the
bony spinal canal at the same cranial-caudal level. This
landmark was then identified on flexed imaging and a cranial
caudal measurement was made from that level to the level of
the new conus position. This measurement also represents
movement of the spinal cord within the canal.
3. Results
The spinal cord should move rostrally during flexion and
should lie in its most caudal location when the patient is in the
neutral position. The measured change in the pedicle-to-spinal
cord DREZ distance across all patients between the neutral
and flexion positions ranged from 1.9–18.0 mm, with a mean
and standard deviation of 8.5 6.0 mm. The inter-DREZ
distance across all patients between the neutral and flexion
positions ranged from 2.0–6.7 mm, with a mean and
Measurement of spinal cord motion
DOI: 10.3109/03091902.2013.844207
3
mean) are very large, 71%, 74% and 64%, respectively. This
reflects the wide variability in the capacity of individual
subjects to maximally flex the spine, as well as possible intersubject variability in spinal cord mechanical characteristics.
J Med Eng Technol Downloaded from informahealthcare.com by Miss Sandy Dalgleish on 12/16/13
For personal use only.
4. Discussion
Figure 2. The white line running transversely is at the base of the
pedicle. The yellow vertical line (on the right) represents the distance
from there to the DREZ. The change in that length represents how much
the cord slides during flexed and neutral positions. The green vertical
line (on the left) indicates the inter-DREZ distance and represents how
much the cord stretches.
Figure 3. Coronal image on which the nerve-root sheath exit zones are
identified by orange arrows (dark arrows, three on each side), the pedicle
locations are marked by blue arrows (light arrows, three on the left side),
and two of the pedicle cross-sections are outlined in blue (the two light
traces on the right).
standard deviation of 3.5 2.6 mm. The mean and standard
deviation for the rostral-caudal conus movement was found to
be 6.4 4.1 mm within an overall range of 1.1–11.4 mm. The
fractional variations in these findings (standard deviation/
Several previous studies have reported both gross anatomical
and image-based measurements of spinal cord motions and
related mechanical parameters, some of which are relevant to
the HSCMS design issues discussed above. For instance,
Figley and Stroman [6] found that the amplitude of the
cardiac-driven pulsation of the anterior-posterior (A/P) component of bulk cord movement in 10 immobile patients was
0.60 0.34 mm in the cervical and upper thoracic region.
However, as found in their subsequent study on eight
immobile patients [7], this value was 50.10 mm in the
lower thoracic region (T4/T5 and below), with excursions of
the same size along the left/right axis, thus setting the lower
limits on the size of bulk cord motions in those directions.
Much larger changes (millimetre-scale) in the A/P position of
the spinal cord relative to the dura were found in patients who
purposely changed posture or carried out flexion/extension
manoeuvres during imaging [8,9]. See also the reviews of
Harrison et al. [10] and Cox [11]. Furthermore, it has long
been known that flexion and extension of the spine produces
centimetre-scale axial stretching motions of the spinal cord
and dura along the rostral-caudal (R/C) direction relative to
the other structures forming the spinal canal [12]. However,
that work was carried out on fixed-tissue samples from
cadavers and there are no reported equivalent series of in vivo
measurements that use a similar protocol. While all of these
studies have provided very useful background information, an
effort focusing specifically on determination of the axial
component of spinal cord motion within the region targeted
for HSCMS implantation was needed to complete the picture.
Moreover, from a biophysical perspective, the findings
presented here can also be used to yield the ratio of the spinal
cord’s mean stretch-to-mean axial movement over a full
flexion cycle: 3.5 mm/8.5 mm &40%. That is, on average
across all patients, it required 1 mm of net axial displacement
of the cord to stretch it 0.4 mm in length. However, as
discussed above, there were large variations between patients
and this makes it difficult to draw generalized conclusions
about the mechanics of the spinal cord in relation to certain
aspects of its interaction with the HSCMS. For instance,
estimates of the net strain within its parenchyma will depend
strongly on the assumptions made regarding how the spinal
cord’s stretch is distributed over its length, as that will be
governed at least in part by the anisotropies in its structure in
a given patient. In the worst case, for some patients, high
levels of parenchymal strain could mean that the HSCMS
electrodes might slide across the spinal cord’s surface unless
the membrane is compliant enough to accommodate this
component of the cord’s axial motion. On the other hand, in
those cases where the stretching component is more localized
within a given region of the typically 45 cm long cord, the
strain would likely be smaller, foreseeably 1% or less, thus
resulting in little relative motion between the HSCMS
electrodes and the spinal cord. A large animal (ovine)
J Med Eng Technol Downloaded from informahealthcare.com by Miss Sandy Dalgleish on 12/16/13
For personal use only.
4
S. Viljoen et al.
in vivo trial is now underway to investigate chronic
interactions between the HSCMS and the spinal cord surface
and histologic examination of explanted samples for evidence
of tissue damage will be carried out. (The preliminary data
from an earlier acute (non-survival) ovine in vivo study [3]
showed no evidence of tissue disruption.)
Lastly, from a HSCMS design perspective, it is now clear
that the lead loops of the devices implanted in some patients
must be sufficiently large to permit a total rostral-caudal range
of motion of up to 2 cm for the cord/membrane relative to the
Oya Strap fixation point, i.e. 1 cm rostral and 1 cm caudal from
the neutral position. HSCMS prototypes with loop area
&160 mm2, when placed on a custom-designed silicone
surrogate spinal cord arrangement [13,14] can accommodate
axial movement of the surrogate cord over most of that range
of displacements without lift-off of the membrane at either the
rostral and caudal extremes of displacement. Careful bench
top studies now underway will evaluate prototype devices with
both larger and smaller loop areas and membrane lengths, to
test for potential lift-off or other movement of the HSCMS as a
function of spinal cord displacement and rotation. However,
beyond these biomechanical design considerations, the natural
scar formation processes will also play a role in stabilizing the
position of the electrode-bearing membrane on the surface of
the spinal cord. A separate series of in vivo studies, now also
underway, are investigating that point. Lastly, other future
studies will include extending the number and age range of
volunteers in the imaging protocol discussed above, to broaden
the scope of our findings.
5. Conclusions
MR-based measurements of the rostral-caudal motions of the
spinal cord during flexion of the spine were made in a series of
16 volunteers. The results indicated that there were large
variations, &70%, in the magnitude of that motion from person
to person, with implications for the design of the intradural
components of the HSCMS. In particular, there will be a
spectrum of spinal cord strains associated with flexion-driven
motion of the cord. Having suitable axial compliance within the
electrode bearing portion of the HSCMS will reduce the risk of
potential irritation of the pial surface by the HSMCS in those
patients where the intraparenchymal strains are large enough to
induce some sliding motion of the electrodes. In patients with
small levels of strain, there would be little relative motion
between cord and HSCMS, meaning that there would be small
risk of any skidding of the electrodes on the pial surface. Also,
the net axial travel of the spinal cord relative to the Oya Strap
fixation device of the HSCMS is within the range that can be
accommodated without lift-off of the electrode bearing portion
of the HSCMS for some of the devices that were preliminarily
tested. More definitive and systematic evaluations are
underway.
Acknowledgements
The first two authors have made equal contributions to this
work and are therefore designated as co-first authors.
We thank Hiroyuki Oya, MD, Hiroto Kawasaki, MD,
W. R. Smoker, MD, and H. Chen for several useful
J Med Eng Technol, 2014; 38(1): 1–4
discussions and technical assistance. We also thank R.
Shurig and colleagues of Evergreen Medical Technologies
LLC for skillful fabrication of the prototype HSCMS devices
and for the illustration used in Figure 1. AMS subject
classification: 92C10; 92C.
Declaration of interest
Authors Viljoen, Dalm, Reddy, Gillies and Howard may
receive patent royalties from any commercial licensing of the
HSCMS intellectual properties that might be negotiated by
their respective institutions.
References
1. Eldabe, S., Kumar, K., Buchser, E., and Taylor, R.S., 2010, An
analysis of the components of pain, function, and health-related
quality of life in patients with failed back surgery syndrome treated
with spinal cord stimulation or conventional medical management.
Neuromodulation, 13, 201–209.
2. Howard III, M.A., Utz, M., Brennan, T.J., Dalm, B.D., Viljoen, S.,
Jeffery, N.D., and Gillies, G.T., 2011, Intradural approach to
selective stimulation in the spinal cord for treatment of intractable
pain. Journal of Applied Physics, 110, 044702.
3. Flouty, O., Oya, H., Kawasaki, H., Kawasaki, H., Wilson, S.,
Reddy, C.G., Jeffery, N.D., Brennan, T.J., Gibson-Corley, K.N.,
Utz, M., Gillies, G.T., and Howard III, M.A., 2012, A new device
concept for directly modulating spinal cord pathways: Initial in vivo
experimental results. Physiological Measurement, 33, 2003–2015.
4. Oliynyk, M.S., Gillies, G.T., Oya, H., Wilson, S., Reddy, C.G., and
Howard III, M.A., 2013, Dynamic loading characteristics of an
intradural spinal cord stimulator. Journal of Applied Physics, 113,
026103.
5. Viljoen, S., Dalm, B.D., Reddy, C.G., Wilson, S., Smittkamp, C.A.,
Gillies, G.T., and Howard, III, M.A., Optimization of intradural
spinal cord stimulator designs via analysis of thoracic spine imaging
data. Journal of Medical and Biological Engineering, 33, 193–198.
6. Figley, C.R., and Stroman, P.W., 2007, Investigation of human
cervical and upper thoracic spinal cord motion: Implications for
imaging spinal cord structure and function. Magnetic Resonance in
Medicine, 58, 185–189.
7. Figley, C.R., Yau, D., and Stroman, P.W., 2008, Attenuation of
lower-thoracic, lumbar, and sacral spinal cord motion: Implications
for imaging human spinal cord structure and function. AJNR
American Journal of Neuroradiology, 29, 1450–1454.
8. Takiguchi, T., Yamaguchi, S., Tezuka, M., and Kitajima, T., 2009,
Measurement of shift of the cauda equina in the subarachnoid space
by changing position. Regional Anesthesia and Pain Medicine, 34,
326–329.
9. Lee, R.A., van Zundert, A.A.J., Botha, C.P., Lataster, L.M.A., van
Zundert, T.C.R., van der Ham, W.C.J.M., and Wieringa, P.A., 2010,
The anatomy of the thoracic spinal canal in different postures.
Regional Anesthesia and Pain Medicine, 35, 364–369.
10. Harrison, D.E., Cailliet, R., Harrison, D.D., Troyanovic, S.J., and
Harrison, S.O., 1999, A review of biomechanics of the central
nervous system – part I: Spinal canal deformations resulting from
changes in posture. Journal of Manipulative and Physiological
Therapeutics, 22, 227–234.
11. Cox, J.M., 2000, A review of biomechanics of the central nervous
system – part I: Spinal canal deformations caused by changes in
posture (comment). Journal of Manipulative and Physiological
Therapeutics, 23, 211–217.
12. Reid, J.D., 1960, Effects of flexion-extension movements of the
head and spine upon the spinal cord and nerve roots. Journal of
Neurology, Neurosurgery and Psychiatry, 23, 214–221.
13. Howard III, M.A., Utz, M., Brennan, T.J., Dalm, D.D., Viljoen, S.,
Kanwal, J.K., and Gillies, G.T., 2011, Biophysical attributes of an
in vitro spinal cord surrogate for use in developing an intradural
neuromodulation system. Journal of Applied Physics, 110, 074701.
14. Oya, H., Howard III, M.A., Shurig, R., and Gillies, G.T., 2012,
Spinal canal surrogate for testing intradural implants. Journal of
Medical Engineering & Technology, 36, 407–410.