Abstract
The feasibility of selective recordings of motor signals from the corticospinal tract was investigated. In cats under Ketamine anesthesia, the motor cortex around the cruciate sulcus was stimulated with a tungsten microelectrode. A pulse train was used to evoke muscle twitches in various parts of the forelimb to identify the corresponding regions of the motor cortex. Once a cortical area was identified, a single stimulus pulse was applied without moving the stimulation electrode. The evoked potentials descending through the lateral corticospinal track (LCST) were recorded from the cervical spinal cord with silicone substrate multi-contact electrodes. Multiple acquisitions were averaged to eliminate the afferent activity from the recordings. Selectivity of the electrode between various sets of traces recorded during stimulation of different points in the motor cortex was calculated. The preliminary data show that significant levels of selectivity can be obtained both epidurally and intradurally.
Introduction
The lateral corticospinal tract (LCST)
is the principal motor pathway, in human, which originate in the frontal lobe
of the cerebral cortex. The fibers of the LCST decussate at the junction of the
spinal cord and medulla, and descend in the dorsolateral portion of the lateral
column of the spinal cord white matter. These fibers primarily control distal
limb muscles and they are the targeted group of nerve fibers for recording in
this project. The lateral corticospinal track becomes close to the surface
below medulla in the white matter and makes up a large portion of the spinal cord
at the cervical level. The main objective of this study to demonstrate the
feasibility of selective recordings from the LCST fibers at the cervical level
with multi-contact electrodes placed over the surface of the cord intradurally or epidurally.
Methods
Experimental Set-up
Five
adult cats (2.6-4.0 kg) were used in this study. Anesthesia was induced with
Xylazine (0.8 mg/kg, IM) and Ketamine (20 mg/kg, IM). Atropine (0.044 mg/kg,
IM) and Dexamethazone (2 mg/kg, IM) were administered
early in the experiment to prevent secretions in the airways and CNS edema,
respectively. The trachea was intubated and the cat
was ventilated mechanically. The femoral vein and artery were catheterized for
administration of drugs and monitoring the arterial blood pressure. Anesthesia
was maintained by intravenous administration of Ketamine using an infusion pump
(10 mg/kg/hr). The cat was placed in a stereotaxic frame and the body
temperature was maintained between 37-39°C
using a regulated heating pad. ECG, rectal temperature, and expiratory
CO2 (3-5%) were monitored throughout the procedure. The spinous
process of the first thoracic vertebra was clamped to stabilize the cervical
spinal column in a horizontal position. Dorsal laminectomy (C3-C7) and
craniotomy were performed to expose the spinal cord and the left or right motor
cortex around the cruciate sulcus.
Spinal
Cord Recording Procedure
One
of the three electrode designs with multiple contacts (Fig. 1) was placed
either epidurally or intradurally
between the cervical spinal roots C5 and C6. In the cases of electrode design
3, which had two sets of contacts along the cord, only the waveforms recorded
from the contacts placed at C5/C6 border were used in
this analysis. The recording contacts were connected to the positive inputs of
multi-channel Grass amplifiers. The negative inputs of the amplifiers were
connected to the reference electrodes on each side. A tungsten microelectrode
(0.1 MW) was inserted at
selected points into the motor cortex at a depth of 1500 mm from the pia surface. A pulse train (PW=0.2 ms,
f=330 Hz, duration=45 ms) was applied to evoke muscle twitches in various parts
of the front limb and occasionally in the hind limb, and thereby identify the
corresponding areas of the cortex. Once a location was identified, a single
pulse (PW=0.2 ms, I=400 mA) was applied to evoke volleys in the descending lateral corticospinal
tract (LCST). In some experiments, the cortical stimulation train did not
elicit muscle twitches later during the experiment without a substantial change
in the recorded amplitudes. In these cases, the spinal cord recordings were
continued by allowing at least a millimeter between the cortical stimulation
sites to assure that different groups of fibers were activated within the LCST.
A hundred to 256 acquisitions were averaged for each stimulus site, using
spike-triggered-averaging, depending on the background neural activity level.
Peak-to-peak amplitudes of the largest volleys that occurred within 2
milliseconds after the stimulus were used to form a measurement vector for each
set of acquisition during stimulation of a unique point in the motor cortex.
The selectivity indices for each vector (Eq. 2) and the overall selectivity for
the given electrode design and implantation-method (Eq. 3) were calculated as
described below.
Recording contacts
Design 1
Design
2
Design 3
Figure 1. Multi-contact
electrode designs used for epidural and intradural recordings of the cortically
evoked volleys from the LCST. The separation between the reference
contacts were 14 mm in designs 1 and 2, and 20 mm in design 3. The
inter-contact separation was 0.8 mm in design 1 and about 1.8 mm in design 3.
The contacts in design 2 were placed along diagonal lines within a rectangular
area of 4.5 mm (longitudinal) by 5 mm (radial). The exposed contact area of
each contact was approximately 250mm x 250mm.
Contact impedance values were betwen 7-20 kW
at 1 kHz.
Data
Analysis
Several definitions have been proposed
as a measure of selectivity for multi-contact recordings of peripheral nerve
activity [1-3]. The selectivity index of Christensen et al. was modified here for the spinal cord: A vector was formed
for each cortical stimulation site by taking the peak-to-peak amplitude of the
evoked compound action potentials from each one of the contacts of the spinal
electrode. These multi-dimensional vectors were normalized such that each has a
unit length. The Euclidean distance between these vectors was calculated using
the following equation:
(Eq. 1)
for all i
and j combinations. N is the
dimension of the vectors, i.e. the number of the contacts used.
The average
distance between a given vector i and
others determined the selectivity of the electrode for a particular vector (or
cortical stimulation site) represented by i
(Eq. 2). The overall selectivity of the electrode for a subset of stimulation
sites was found by averaging the selectivity indices for individual sites
within the set (Eq. 3).
(Eq.
2)
where Z
is the number of stimulation sites in the subset.
(Eq.
3)
Results
A set of raw traces recorded with design 2 following stimulation of a unique point in the motor cortex is shown in Figure 2. The first spike to the left is the stimulus artifact. The main neural component arrives around at time equals 2 ms and it is followed by smaller components with larger delays. Notice that the large neural component is recorded at different amplitudes by various contacts.
In Table 1, the implantation method, the type of the electrode used, and the site of implantation are shown for all the procedures in all cats. The mean of the maximum selectivity values for two, three, and four vector discriminations were 21±15%, 14±10%, and 11±8% with the extradural electrodes (n=3) and 26±10%
, 17±7%, and 15±6% with the intradural electrodes (n=4). These two sets of selectivity values were significantly different when the data from the same type of electrodes (excluding electrode design 3) with similar implantation techniques were compared using a paired t test (t=0.013). The selectivity indices obtained with extradural implantation of design 3 were 22±17%, 14±11%, and 13±10% (n=2). The selectivity varied from trial to trial. The maximum and minimum selectivities for two, three, and four vector discrimination were 51%, 34%, 29% (Design 2, intradural implant) and 7%, 5%, and 4% (Design 1, extradural implant), respectively.
|
|
Cat No |
Implant Method |
Electrode Design |
Implant Site |
|
1 |
1 |
Extradural |
1 |
C5/C6 |
|
2 |
2 |
Intradural |
1 |
C5/C6 |
|
3 |
2 |
Intradural |
2 |
C5/C6 |
|
4 |
3 |
Extradural |
1 |
C5/C6 |
|
5 |
3 |
Extradural |
2 |
C5/C6 |
|
6 |
3 |
Intradural |
2 |
C5/C6 |
|
7 |
4 |
Extradural |
3 |
C5/C7 |
|
8 |
5 |
Extradural |
3 |
C4/C6 |
|
9 |
5 |
Intradural |
1 |
C5/C6 |
Table 1: Implantation methods, electrode
designs, and implantation sites that are used in each procedure.
Conclusions
The results of this study show that selective recording from the spinal cord surface (without penetrating the pia matter) is feasible with multi-contact electrodes. Furthermore, placing the electrode extradurally causes only a small reduction in selectivity. A method that utilizes multiple parameters from the recorded waveforms to form the measurement vectors may improve the calculated selectivity values and reduce the variation. This requires further analysis of the recorded signals. The large variation in the selectivity values can also be due to the fact that the location of the stimulation points in the motor cortex change from trial to trial thereby resulting a variation in the activation pattern of the fibers in the cord.
The findings of this study suggest that one can extract multi-channel voluntary motor signals from the corticospinal tract with non-penetrating electrodes from the cervical spinal cord. Such an electrode may serve as a Spinal Cord–Computer Interface in high level spinal cord injury patients.
References
Figure 2. Raw traces acquired with electrode design 2 from
C5/C6 using a stimulus amplitude of 400 mA.
[1] Christensen,
P.R., Y. Chen, K.D. Strange, K. Yoshida, and J.A. Hoffer,
“Multi-channel recordings from peripheral nerves: 4. Evaluation of selectivity
using mechanical stimulation of individual digits”, Proceedings of the Second Annual IFESS Conference and Neural
Prosthesis: Motor Systems 5, Burnaby, British Bolumbia,
Canada, 1997.
[2] Sahin, M. and Dominique M.
Durand, "Selective recordings with a multi-contact nerve cuff
electrode", 18th Annual
International Conference of the IEEE Eng. in Med. and Biol. Soc.,
Amsterdam, Netherlands, 1996.
[3] Struijk,
J.J., M.K. Haugland, and M. Thomsen, “Fascicle
selective recording with a nerve cuff electrode”, Annual International Conference of the IEEE Eng. in Med. and Biol. Soc,
Acknowledgments
The project was supported by a grant (SBI-9909-2) from
Christopher Reeve Paralysis Foundation.
|
|
|
|
|
|
|
|