Movements Elicited by Electrical Stimulation of Muscles ... - IEEE Xplore

IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING , VOL. 12, NO. 1, MARCH 2004

1

Movements Elicited by Electrical Stimulation of Muscles, Nerves, Intermediate Spinal Cord, and Spinal Roots in Anesthetized and Decerebrate Cats Yoichiro Aoyagi, Vivian K. Mushahwar, Member, IEEE, Richard B. Stein, and Arthur Prochazka

Abstract—Electrical stimulation offers the possibility of restoring motor function of paralyzed limbs after spinal-cord injury or stroke, but few data are available to compare possible sites of stimulation, such as muscle, nerve, spinal roots, or spinal cord. The aim of this study was to establish some characteristics of stimulation at these sites in the anesthetized and midcollicular decerebrate cat. The hind limb was constrained to move in the sagittal plane against a spring load. Ventral-root stimulation only produced movements down and back; the direction moved systematically backward the more caudal the stimulated roots. In contrast, dorsal-root stimulation only produced movements up and forward. Thus, neither method alone could produce the full range of normal movements. Muscle, nerve, and intraspinal stimulation within the intermediate regions of the gray matter generated discrete, selective movements in a wide range of directions. Muscle stimulation required an order of magnitude more current. Single microwire electrodes located in the spinal gray matter could activate a synergistic group of muscles, and generally had graded recruitment curves, but the direction of movement occasionally changed abruptly as stimulus strength increased. Nerve stimulation produced the largest movements against the spring load ( 80 of the passive range of motion) and was the most reproducible from animal to animal. However, recruitment curves with nerve stimulation were quite steep, so fine control of movement might be difficult. The muscle, nerve, and spinal cord all seem to be feasible sites to restore motor function. The pros and cons from this study may be helpful in deciding the best site for a particular application, but further tests are needed in the chronically transected spinal cord to assess the applicability of these results to human patients.

%

Index Terms—Decerebration, movement, muscle, nerve, spinal cord, spinal roots, stimulation.

I. INTRODUCTION

S

PONTANEOUS neurological recovery after spinal-cord injury or stroke occurs to some extent in the first months after the onset of the disorder, but is extremely limited after the first year [1]. Minimizing the damage to the central nervous system and maximizing recovery is a continuing goal for research and treatment of these conditions. Although advances have been made in the field of central nervous system regenera-

Manuscript received June 19, 2002; revised December 11, 2003. This research was supported by the Canadian Institutes of Health Research. The work of V. K. Mushahwar was supported by the Alberta Heritage Foundation for Medical Research. Y. Aoyagi is with the Department of Rehabilitation Medicine, Kawasaki Medical School, Okayama 701-0192, Japan. V. K. Mushahwar, R. B. Stein, and A. Prochazka are with the Center for Neuroscience, University of Alberta, Edmonton, AB T6G 2S2, Canada (e-mail: [email protected]). Digital Object Identifier 10.1109/TNSRE.2003.823268

tion, a cure for spinal-cord injury or stroke does not yet exist [2], [3]. The spinal cord below the lesion, as well as the muscles and nerves, remain essentially intact in many individuals [4], though reflex transmission and the tonic state of interneuronal networks may be abnormal [5]. Thus, stimulating the surviving neurons below the lesion is another approach to restore motor function [6]. If excitation of muscles or nerves can be suitably coordinated, functional movements such as walking and grasping can be generated [7]–[9]. Motor prostheses have been developed to restore functional movement and partially overcome the paralysis of spinal-cord injury by electrical stimulation [10]–[12]. Although some clinical applications of electrical stimulation have been successful [7], [8], [13], [14], the electrical restoration of whole limb movements remains problematic [15], [16]. More recently, stimuli have been applied to the spinal cord and ventral roots as other avenues for restoring functional movement by electrical stimulation [17]–[20]. One rationale for using intraspinal stimulation is the interesting suggestion that a spinal electrode may activate a group of muscles that form “force-field primitives” [21], [22]. These primitives were posited to be basic building blocks from which more complex movements might be constructed. However, many unresolved questions remain about where and how to use electrical stimulation for restoring limb movement. To date, very few studies have compared results using different methods in the same subjects. In addition, many of the basic biomechanical and neurophysiological properties of movements produced by electrical stimulation are poorly understood. This study represents the first step in comparing these properties using electrical stimulation of muscles, nerves, ventral and dorsal roots, and the spinal cord in anesthetized and decerebrated cats. The results provide baseline information that can be tested further in chronically spinalized animals for eventual human application. Brief accounts of some of this material have appeared in [23] and [24]. II. METHODS A. Animal Preparation Ten adult cats were used in this study according to protocols approved by the University of Alberta Animal Care Committee in accordance with the guidelines of the Canadian Council of Animal Care. Anesthesia was either induced by inhalation of halothane (five cats that were decerebrated later in the experiment) or by use of sodium pentabarbital (the other five cats that remained under anesthesia until they were euthanized).

1534-4320/04$20.00 © 2004 IEEE

2


(a)

surrounding tissue, cuff or epineural electrodes were implanted. Before implantation, each nerve was stimulated with a hook electrode in order to verify its innervation. The hamstring nerve, which separates from the sciatic nerve, has three main branches [Fig. 1(a)] [26], [27]. Two branches innervate HE: semimembranosus and anterior biceps femoris. One branch innervates KF: posterior biceps femoris and semitendinosus. The femoral nerve typically had four main branches [Fig. 1(b)]; two innervated hip flexors: sartorius and rectus femoris, one innervated KE branches of the quadriceps femoris and the remaining branch contained cutaneous nerves, such as the saphenous nerve. The TB and CP nerves were exposed after an incision was made along the popliteal region and surrounding fatty tissue was removed. Electrodes were implanted on the TB just proximal to the knee after the sural nerve has split off, but before the branches to the triceps surae leave. Electrodes were implanted on the CP nerve distal to the knee, before the nerve enters the ankle dorsiflexor muscles. These nerves innervate ankle plantarflexors and dorsiflexors, respectively. Thus, six nerve electrodes were implanted and the skin was closed. C. Muscles

(b) Fig. 1. (a) Schematic view of the hamstring branch of the sciatic nerve and bipolar, epineural electrodes attached near the HE and KF branches. (b) Schematic view of the femoral nerve and epineural electrodes attached over the surface of the HF and KE branches. After passing the dorsal iliopectineal arch, the femoral divides into four branches: two innervate hip flexors, one innervates KE, and the remaining branch contains cutaneous nerves. Further details are in Section II-B.

For animals under halothane, a tracheotomy was done and a tracheal tube was inserted and firmly tied to the trachea. The cats breathed spontaneously a mixture of 95% O and 5% CO mixed with halothane. The halothane-anesthetized cats also had the right jugular vein and right carotid artery cannulated. The left carotid artery was ligated. Arterial pressure was measured by connecting the carotid catheter to a pressure transducer. Five percent Dextran was given as necessary through a jugular vein cannula. Body temperature was monitored and maintained close to 37 C using a heating pad and/or lamp. The animal’s right leg and back were shaved. B. Nerves The following nerves were exposed in the right hind limb: tibial nerve (TB); common peroneal (CP) nerve; hamstring branches of the sciatic nerve supplying knee flexors (KF) and hip extensors (HE) [Fig. 1(a)]; and femoral nerve branches supplying hip flexors (HF) and knee extensors (KE) [Fig. 1(b)]. In the first two experiments, all nerves were implanted with cuff electrodes [25], but conduction in some fine nerves tended to block over the course of a long experiment. In later experiments, epineural electrodes were used for all but the CP and TB nerves and no conduction blockage occurred. All data that could have been affected by nerve conduction blockage were excluded from the analysis. After the nerves were cleared from

To stimulate muscle, bipolar intramuscular electrodes were placed close to the motor points in seven right hind limb muscles [lateral gastrocnemius (LG), tibialis anterior (TA), vastus lateralis (VL), semimembranosus anterior (SM), posterior biceps femoris (BF), sartorius (SA), and iliopsoas (IP)]. Thus, “muscle” stimulation actually involved stimulating motor nerves close to the neuromuscular junctions. A second electrode was inserted into each muscle 1–2 cm away for bipolar stimulation. Each intramuscular electrode consisted of a fine insulated wire that was inserted through a 21-gauge hypodermic needle. The electrodes were insulated except for 3 mm at the tip. Electromyograms (EMG) were also recorded with these electrodes. D. Spinal Cord The spinal cord was exposed from L4 to S3 by a laminectomy. The cat was mounted in a conventional spinal frame. To stimulate the spinal cord, 10–19 microwire electrodes were inserted through minute holes made in the dura mater with a 30-gauge dental needle on the right side of the spinal cord [20], [28]. The electrodes were stainless-steel wires (California Fine Wire, Gover City, CA) with a diameter of 30 m and 30–70 m exposed at the tip. They were inserted every 2–3 mm along the cord from L5 to S1, 0.5–2 mm from the midline and 2.5–3.5 mm deep, depending on the size of spinal cord at each level. The positions were verified histologically in four cats and most were located in intermediate gray matter locations posited to contain “movement primitives” [12], [22]. Final placement was determined by stimulation through each electrode to ensure that a consistent, clear movement could be elicited. E. Motion Analysis To monitor the movement of the joints three-dimensionally, motion tracking sensors (6D-Research, Skill Technologies, Inc., Phoenix, AZ) were placed on the skin near: (1) the hip joint;

AOYAGI et al.: MOVEMENTS ELICITED BY ELECTRICAL STIMULATION

(a)

3

ment, sensors 2 and 3 were rigidly fixed to the femur and tibia by surgical suture through holes drilled in the respective bone. All apparatus within 1 m of the sensors was devoid of magnetic metals to avoid any distortion of the electromagnetic field associated with the signals recorded from the motion sensors. The right paw was fixed in a U-shaped foot holder. The foot holder was connected to one end of a 50-cm joystick with its fulcrum 40 cm from the foot holder and approximately perpendicular to the sagittal plane in which the leg moved. The foot holder was connected to the joystick by a ball joint to allow free movement in the sagittal plane. The other end of the rod was connected to a spring. Prior to the experiment, the force to move the holder from the neutral position in the sagittal plane was measured. The mean change in force with distance in the forward, upward, backward, and downward directions of the sagittal plane was 1.4 0.1 N/cm (mean SD). None of the directions was statistically different from the others. A spring load is intermediate between the isometric state that has usually been studied and a freely moving limb. It allows movement, but against a known load, whereas unconstrained movements are made against a mixture of gravitational, inertial, and viscoelastic forces, which are generally poorly defined. The resting position of the limb was adjusted to approximate a normal standing posture. F. Experimental Protocol

(b) Fig. 2. (a) Schematic view of the experimental setup. The motion sensors were attached to the lateral surface of (1) the hip joint, (2) the lateral epicondyle of the femur, (3) the lateral malleolus of the tibia, and (4) the lateral metatarsophalangeal joint. The right paw was fixed in a foot holder. The foot holder was connected to one end of a 50-cm rod that had a joystick-like fulcrum 40 cm from the foot holder. The other end of the rod was connected to a spring. (b) Sagittal trajectory of the hindlimb generated during stimulation of posterior biceps femoris (BF) muscle. The stimulus threshold was 200 A. The stick figures illustrate the rest and extreme position when BF was stimulated with a train of 40, 1000-A pulses. The O and , respectively, represent the position of sensors 1–4 at rest and at the extreme position during stimulation. The dotted lines represent the real trajectory of the hindlimb joints during stimulation. The thick line represents the total vector (from rest to the extreme position) of the lateral metatarsal–phalangeal joint, which corresponds to sensor 4. Note that forward, upward, backward, and downward directions correspond to 0 , 90 , 180 , and 270 , respectively.

(2) the lateral epicondyle of the femur; (3) the lateral malleolus of the tibia; and (4) the lateral surface of the foot holder corresponding to the lateral metatarsophalangeal joint of the foot [Fig. 2(a)]. To avoid skin slippage or displacement during move-

The following measurements were made. 1) Passive range of motion of the right hind limb in the sagittal plane was measured and recorded through the 6D-Research system. 2) The muscles were stimulated individually and the kinematics were recorded through the 6D-Research system. Bipolar stimulation was used and the stimulus consisted of trains of monophasic, cathodic pulses at 50 pulses/s lasting 0.8 s. Each stimulus pulse had a duration of 0.3 ms. The stimulation current needed to produce a threshold muscle movement was determined for each muscle using visual observation and was increased in steps up to five times threshold. 3) Nerves were stimulated individually in a bipolar manner through epineural or cuff electrodes over a similar range of intensities and the EMG and kinematics were recorded. 4) The spinal cord was also stimulated with monopolar, intraspinal electrodes. The reference electrode [nine-strand stainless steel Cooner wire (California Fine Wire, Grover City, CA)] was embedded within the back muscles. Stimulus levels were used up to six times threshold: up to 500 A in the anesthetized state because of the high thresholds in this state. This level is higher than recommended [29] and higher than would be used in chronic, unanesthetized animals, but was needed to generate a full recruitment curve because of the gradual recruitment with intraspinal stimulation [Fig. 6(b)]. 5) In five animals, the right dorsal and ventral roots from L4 to S1 were exposed. Dorsal roots and ventral roots were sequentially cut peripherally. The proximal stump of the cut dorsal roots and the distal stump of the cut

4


ventral roots were then stimulated by hook electrodes at intensities up to five times threshold. 6) In some animals, to induce maximal movement in the forward, upward, backward, and downward directions from muscle, nerve, and spinal cord, various stimulus combinations were given. For example, to induce the maximal upward movement by nerve stimulation, a synergistic group of flexor nerves (CP, KF, and HF) was stimulated together at a supramaximal level [see also Fig. 5(a)]. After decercats) the anesebration at an intercollicular level ( thetic was discontinued and parts (2)–(4) were repeated for comparison. Muscle and nerve were generally stimulated while the animal was coming out of anesthesia, but spinal stimulation was typically delayed more than 1 h after decerebration until reflexes were well developed. G. Data Acquisition Kinematic data recorded through the 6D-Research system were digitized at a rate of 30 Hz. During stimulation, the trajectories of the four motion sensors in the sagittal plane were measured using the 6D-Research motion analysis software. Then, the direction and distance of the paw movement vector were calculated based on the trajectory produced by sensor 4 that was placed near the metatarsophalangeal joint. Calculations were done using custom-written programs in Matlab (Math Works, Inc., Natick, MA). The vectors shown in the polar plots of Figs. 2, 3, and 5 represent the movement of the paw from the rest position to the extreme position produced by stimulation. III. RESULTS The lengths of the thigh, shank, and foot were 9.4 0.4, 10.9 0.6, 9.8 0.4 cm (mean S.D.). In the rest position of the hind limb, the hip was flexed by 30 9 , the knee was flexed by 69 9 and the ankle was plantarflexed by 19 16 in a uniform stiffness field [see Section II and Fig. 2(a)]. These values are similar to those found in a quietly standing animal [30]. As an example of the motion analysis, Fig. 2(b) demonstrates the trajectory of the hip, knee, ankle, and paw (near the lateral metatarsophalangeal joint) when BF was stimulated with four times threshold (T). The distance of the paw movement during stimulation was 6.4 cm from the rest to the extreme position. The direction was 143 with respect to the forward direction. In the same way, vectors induced by stimulation of muscles, nerves, spinal cord, ventral, and dorsal roots were analyzed and plotted in polar coordinates. When the stimuli were repeated (not shown), the movements were very reproducible. A. Trajectories Elicited by Stimulating Muscles, Nerves, Spinal Cord, and Roots Fig. 3(a)–(d) show typical results elicited from one animal. Movement vectors (from rest to the maximal excursion) were generated by intramuscular electrodes [Fig. 3(a)] and epineural electrodes [Fig. 3(b)] with stimulus amplitudes ranging from 1.2–6T. Stimulation of individual muscles and nerves produced

Fig. 3. (a) and (b) Muscles and nerves were stimulated with trains of 40 pulses up to six times threshold (T). Lines radiating from the center of the polar graphs represent the vectors from rest to extreme positions during stimulation. All data were obtained from a halothane-anesthetized animal. The directions were quite consistent in all cases (see also Table I and Fig. 7). Abbreviations: lateral gastrocnemius (LG), tibialis anterior (TA), vastus lateralis (VL), semimenbranosus anterior (SM), posterior biceps femoris (BF), sartorius (SA), and iliopsoas (IP), tibial (TB), common peroneal (CP), knee extensor (KE), hip extensor (HE), knee flexor (KF) and hip flexor (HF). (c) and (d) Ventral and dorsal roots were stimulated at levels up to five times threshold. Note that stimulating the ventral roots produced consistent movements at all stimulus levels (mainly down and backward). The direction rotated backward as more caudal roots were stimulated. Stimulating the dorsal roots produced movements up and forward; the direction rotated backward as more caudal roots were stimulated. Data in (c) and (d) were obtained from the same animal as in (a) and (b) after decerebration. (e) All vectors were produced by intraspinal stimulation up to about 5T in a different, anesthetized animal. The directions produced by intraspinal stimulation included forward, upward, backward, and downward movements.

fairly distinct and reproducible preferred directions. Preferred directions for each muscle and nerve stimulation were always within 30 with stimuli from just above threshold to supramaximal levels. As expected, stimulation of the TB nerve and the LG muscle, which is innervated by the TB nerve, showed similar preferred directions. Also, CP and TA, KE and VL, HE and SM, KF and BF, HF, SA, and IP showed almost identical preferred directions. The movements were also consistent between animals (Table I). Movement vectors from stimulating L4-S1 ventral roots from the same animal after decerebration ranged from 149 to 360 [Fig. 3(c)], whereas the vectors from stimulating L4-S1 dorsal roots ranged from 20 to 138 [Fig. 3(d)]. Neither ventral nor


5

TABLE I

MEAN

6 S.D. OF DIRECTION OF MOVEMENT BEFORE AND AFTER DECEREBRATION

(a)

(b)

Fig. 4. Comparison of maximal directional change according to the level of stimulus intensity. Movements generated by muscle, nerve, and ventral root stimulation were elicited under anesthesia and those by dorsal root stimulation were elicited after decerebration. Directions generated by muscle, nerve, ventral, and dorsal root stimulation were relatively constant from minimal to maximal stimulation 20 , whereas directions generated through intraspinal electrodes in both anesthetized and decerebrate states often showed abrupt changes and were significantly larger than the other four types of stimulation 0.01 .

(< )

(p

>

> >

>

that the stimuli in the intermediate region of the spinal cord occasionally spread directly to –motoneurons. When interneurons were primarily stimulated (e.g., latency 4 ms), the jitter and coefficient of variation were relatively large, suggesting that transmission to motoneurons was rather variable due to the extra synapses involved. IV. DISCUSSION Our aim was to compare some biomechanical and neurophysiological characteristics of electrical stimulation through muscles, nerves, spinal cord, ventral, and dorsal spinal roots in the anesthetized or decerebrate cat. We studied the selectivity of

8


muscle activation, recruitment characteristics, maximal movements, movement directions, consistency of responses, and required stimulus intensities. A. Selectivity Stimulating specific nerves or muscles could evoke distinct, reproducible movements in one of six distinct directions, regardless of stimulus intensity (Figs. 3 and 7). Biceps femoris, a biarticular muscle, is well known to have separate nerve branches with different mechanical actions [27], [31], [32]. Thus, we could separately stimulate nerve branches to the HE and KF. In this experiment, we also found that the femoral nerve largely innervates hip flexors and KE separately. For example, the biarticular muscle, rectus femoris is supplied by separate nerve branches that innervate hip flexor and knee extensor compartments separately. For the artificial control of limb movements through peripheral nerve stimulation, it is obviously desirable if motor nerves divide into branches that selectively control movement at one joint, as was seen here for the femoral and hamstring nerves, rather than activating uniand biarticular muscle groups nonselectively. Neither ventral nor dorsal root stimulation was able to induce movements in all directions (Fig. 3, Table I). A ventral root contains all the –motoneurons from one spinal segment [33]. Since most lumbosacral segments contain both flexor and extensor –motoneurons, it was initially surprising that VR stimulation was unable to induce flexor movements. However, the extensor (antigravity) muscles are generally stronger than the flexor muscles, so coactivation would result in a net downward (extensor) movement. The L4 spinal root primarily innervates hip flexor muscles and the L5 spinal root is composed of fibers of the femoral nerve that innervate hip flexor and knee extensor muscles [34], [35]. At the resting position of the limb, L5 ventral root stimulation produced movements directed somewhere between forward (hip flexor) and downward (knee extensor) [Fig. 3(c)]. The L6 spinal root is mainly composed of motoneurons innervating HE, KE, and ankle flexor muscles [34], [35]. The L7 spinal root is mainly composed of motoneurons innervating HE, KE, and ankle flexor and extensor muscles [34], [35]. The S1 spinal root is mainly composed of hamstring and TB motoneurons [34], [35]. Thus, broadly speaking the biomechanical results obtained by stimulating spinal roots are consistent with the anatomical distribution of extensor –motoneurons. Stimulating the dorsal roots generated flexor movements. Sherrington made the classic observation that the flexor reflex in the decerebrate cat consists of hip, knee, and ankle flexion combined with extensor muscle relaxation [31]. Flexor reflexes could be elicited by stimulating skin or mixed nerves. Although skin stimulation and muscle stimulation can also produce crossed extension and stretch reflexes [31], the ipsilateral flexor component seemed to predominate with stimulation of whole dorsal roots in this experiment. More interestingly, the movement vectors also rotated backward as root stimulation moved from L4 to S1 [Fig. 3(d)]. Forward movements generated by stimulating L4 and L5 dorsal roots were very similar to those generated by IP and SA stimulation, whose –motoneurons are located primarily in the L4 and L5 segments [34], [36]. In addition to hip flexor movements, stimulation of the L4 and L5

dorsal roots also activated TA. As L6, L7, and S1 dorsal roots were stimulated, SA, TA, and then BF were primarily activated. Since –motoneurons of TA and BF are located primarily in the L6/L7 and L7/S1 segments, respectively, [34], [36], stimulating each dorsal root appears to have excitatory synaptic connections mainly to flexor –motoneurons at the same spinal segmental level. It may also have inhibitory synaptic actions on extensor –motoneurons in the same spinal segmental level. Afferent information is conveyed to –motoneurons in other spinal segments, but presumably with less strength and additional synaptic delays. Extensor muscles such as VL, LG, and SM were also activated in accordance with the anatomical distribution of their –motoneurons. However, the level of extensor muscle activation was significantly smaller than that for flexor muscles. Intraspinal stimulation was able to induce movements in all directions in the anesthetized cat. However, vectors produced by the intraspinal electrodes showed large changes in direction with varying stimulus amplitudes, which was very different from stimulation of muscles, nerves, and spinal roots. In this study, we targeted intermediate areas that have been suggested to produce movement primitives [22], so stimulation in the ventral horn may produce different results. B. Stimulus Intensity, Recruitment Properties, and Movement Distance As described previously [37]–[39] during isometric contractions, thresholds for muscle stimulation were an order of magnitude higher than for nerve stimulation. Thresholds to elicit movement by spinal, ventral, or dorsal root stimulation were also much lower than those for muscular stimulation 0.001 and were not significantly different from thresh0.05 . Low stimulation levels olds for nerve stimulation are obviously desirable in terms of power consumption in a wearable device and to provide selectivity in stimulating only the desired tissue. Electrode breakage due to electrochemical reactions depends on the current density and charge transfer, rather than the absolute currents. With careful placement of electrodes, we feel that these quantities can be kept below the well known limits at all the sites tested [29]. Muscle stimulation tends to produce a relatively gradual increase in the twitch contractile force of muscle, compared to that of nerve, as the stimulus intensity is increased [40]. Our results obtained from actual movements showed the same tendency, unless the electrodes were close to the nerve entry zone. The graded control of force and movement is needed to achieve precise artificial control of limb trajectories. The recruitment properties of muscle stimulation are influenced by two factors: electrode position and muscle volume. Electrodes placed close to the nerve entry point produced the greatest force but may also have the steepest recruitment [38], [41]. Recruitment curves tend to show sudden steps or increments that are attributable to the sequential activation of terminal nerve branches [38]. Therefore, the position of the stimulating point is a critical factor that determines the recruitment curve of muscle stimulation and the recruitment curves for muscle were more variable from experiment to experiment in spite of careful placement of electrodes by a skilled person.


Stimulation of nerves, ventral roots, and dorsal roots produced the steepest recruitment curves. Interestingly, intraspinal stimulation showed the most gradual recruitment (for a detailed description and muscle recruitment with intraspinal stimulation, see [42]). In this study, we primarily targeted intermediate areas of the spinal gray matter suggested to include movement primitives [22], but stimuli can spread up to about 1 mm at the maximal stimulus intensities we used [17]. Individual motoneuron pools are less than 1 mm in diameter [12], [36], [43]. The chance of spreading to motoneuronal and interneuronal areas that may be dependent on the state of the animal probably explains the variability. However, as mentioned earlier, significantly lower stimulus amplitudes are recommended for intraspinal microstimulation which will lead to much lower levels of stimulus spread. Nerve stimulation produced the longest distance of movement (Fig. 5). It consistently produced more than 80% of the passive range of motion in the upward, backward, and downward directions, but less than 70% in the forward direction. This is because the nerves to IP, one of the main muscles producing forward movement were not stimulated in these experiments. IP is innervated from the lumbar plexus in the cat [26], whereas in human it is innervated by both the femoral nerve and lumbar plexus [33]. Muscle and intraspinal stimulation produced movements covering about 75% of the distance of the passive range of motion on average, indicating that not all the synergistic motor units were activated. Overall, since nerve, muscle, and intraspinal stimulation produced distances 60 of the passive range of motion in all directions, these methods of stimulation have the potential to produce functional, whole limb movements using stimulation of appropriate synergies. This has also been shown previously in normal cats implanted with intramuscular and intraspinal microelectrodes [20]. C. Perspectives on Clinical Application to Restore Functional Movements Muscle: Application of electrical simulation to persons with paraplegia has been used for the restoration of standing, walking, hand grasp, bladder, and respiratory control [7]–[9], [44]. Intramuscular and surface stimulation, which are relatively noninvasive, are the most commonly used methods. Disadvantages of these approaches include the high currents required to stimulate the motor units and the technical difficulty in placing the electrodes on or very close to the nerve entry point, as our data indicate. These electrodes are then exposed to high mechanical stresses and motion, which can result in lead breakage and electrode migration [15], [45]. Moreover, the electrodes become encapsulated, elevating stimulus thresholds to higher levels and in some cases causing neuromuscular damage in the long run [15], [46]. To overcome these difficulties, recent work has shifted toward neural stimulation. Nerve: Peripheral nerve stimulation also has some drawbacks. Large nerve trunks typically contain axons that innervate several different muscle groups, so selectivity can be a problem [47]. However, we found that flexor and extensor muscles of the hip, knee, and ankle joints can be activated selectively by electrodes placed close to selected nerve branches. The connecting wires were tunneled subcutaneously to place the

9

electrodes close to the target muscles, which involves fairly extensive surgery, but the use of injectable microstimulators [39] may eliminate this problem. Nerve stimulation did produce the largest movements and they were relatively constant and reliable over the short duration of these experiments. In addition, the stimulus current required to activate nerve axons by stimulation with epineural or cuff electrodes is much lower than that required for intramuscular stimulation [Fig. 6(a)]. The main disadvantage of nerve stimulation demonstrated by our data is the relatively narrow range of stimulation intensity from threshold to maximal movements [i.e., steep recruitment, Fig. 6(b)], which can also lead to variability in the outcome [39]. This may be a problem for whatever type of epineural electrode is used, if small relative movements occur during limb movements. In addtion, we did not test nerve stimulation in the fully awake animal, so reflexes could confound its effects in clinical applications. Spinal Cord: When stimulation pulses were delivered through intraspinal electrodes, movements were activated with low stimulus intensity (Fig. 6). Large jitter and coefficient of variation suggested that identical motoneurons were not necessarily recruited with each stimulus pulse (Fig. 9). One favorable aspect is that muscle fatigue may be reduced compared to muscle and nerve stimulation (e.g., [42]). The map of movement vectors demonstrated that there were regions within the spinal cord from which single muscles or synergistic muscle groups can be activated [20]. This is not surprising because individual –motoneuronal pools exist as clusters along lumbo-sacral segments [34], [36], [48]. However, movement recruitment curves generated by intraspinal stimulation were rather variable depending on the electrode location and the direction of movement could change dramatically at the stimulus intensities used. Presumably, intraspinal stimuli activated large numbers of interneurons, whose state may have varied, as well as activating neighboring motoneuron pools in the spinal cord. Overall, the data obtained in this study using cats did not correspond well to data measuring isometric force by stimulating frog spinal cord in terms of “primitives” [22]. This issue will be discussed in detail elsewhere. In regard to clinical application, generating reproducible and selective movements may be a challenging problem for intraspinal stimulation targeting movement primitive locations. However, single chronically implanted microwires in the ventral horn can generate reproducible, synergistic movements of the whole limb sufficient for weight support or forward swing in the normal cat [20]. Intraspinal stimulation may, therefore, provide whole limb synergies from a small and contained implanted area. Roots: Since ventral roots are composed of mixed nerve groups, selectivity is relatively poor compared to single nerve or muscle stimulation and mainly extensor movements are produced. Thus, stimulation of ventral roots clinically may not induce good functional cyclic movement such as walking, although it has been used to restore leg powered cycling for exercise [19], [49]. Our results suggest that dorsal root stimulation, generating flexor movements through reflex pathways is not a good candidate to restore a full range of functional movements, though such movements can be useful for initiating swing. Furthermore, some improvement of motor function has

10


been noted in paraplegic people with dorsal column stimulation [50]. Dorsal roots may be a good location for recording sensory signals to control electrical stimulation [51], [52]. Fatigue is an important issue that has not been studied here. A priori, stimulating the dorsal roots or interneurons in the intermediate gray matter of the spinal cord may reduce fatigue by recruiting small, fatigue-resistant motoneurons before large, fatigable ones [21]. The other methods will tend to excite motoneurons in the opposite order. However, in persons with chronic spinal cord injury, virtually all motor units have converted to fast, fatigable types [53]. The conversion can be reversed to some extent by stimulating 1–2 h/day [4]. Only the units that are activated by stimulation will be modified to become more fatigue resistant. The issue of fatigue needs further examination in chronic studies. Acute studies are not adequate because of the long-term changes that take place after injury and stimulation. V. CONCLUSION There is unlikely to be one best stimulatation site for all functional-electrical stimulation (FES) applications, but the results of the present study should help in deciding the best site for a given clinical application. The state-dependence of responses to electrical stimulation highlights the need for information from chronic studies to supplement the present acute data. REFERENCES [1] J. F. Ditunno Jr, V. Graziani, and A. Tessler, “Neurological assessment in spinal cord injury,” Adv. Neurol., vol. 72, pp. 325–33, 1997. [2] O. Behar, K. Mizuno, S. Neumann, and C. J. Woolf, “Putting the spinal cord together again,” Neuron, vol. 26, pp. 291–3, 2000. [3] F. P. Girardi, S. N. Khan, F. P. Cammisa Jr, and T. J. Blanck, “Advances and strategies for spinal cord regeneration,” Orthop. Clin. North Amer., vol. 31, pp. 465–472, 2000. [4] R. B. Stein, T. Gordon, J. Jefferson, A. Sharfenberger, J. F. Yang, J. T. de Zepetnek, and M. Belanger, “Optimal stimulation of paralyzed muscle after human spinal cord injury,” J. Appl. Physiol., vol. 72, pp. 1393–1400, 1992. [5] C. S. Sherrington, The Integrative Action of the Nervous System. New Haven, CT: Yale Univ. Press, 1906. [6] W. T. Liberson, H. J. Holmquest, D. Scott, and M. Dow, “Functional electrotherapy: Stimulation of the peroneal nerve synchronized with the swing phase of the gait of hemiplegic patients,” Arch. Phys. Med. Rehab., vol. 42, pp. 101–105, 1961. [7] A. Kralj, T. Bajd, and R. Turk, “Enhancement of gait restoration in spinal injured patients by functional electrical stimulation,” Clin. Orthop., pp. 34–43, 1988. [8] P. H. Peckham and G. H. Creasey, “Neural prostheses: Clinical applications of functional electrical stimulation in spinal cord injury,” Paraplegia, vol. 30, pp. 96–101, 1992. [9] A. Prochazka, M. Gauthier, M. Wieler, and Z. Kenwell, “The bionic glove: An electrical stimulator garment that provides controlled grasp and hand opening in quadriplegia,” Arch. Phys. Med. Rehab., vol. 78, pp. 608–614, 1997. [10] A. R. Kralj and T. Bajd, Functional Electrical Stimulation: Standing and Walking After Spinal Cord Injury. Boca Raton, FL: CRC, 1989. [11] R. B. Stein, P. H. Peckham, and D. Popovic, Neural Prostheses: Replacing Motor Function After Disease or Disability. New York: Oxford Univ. Press, 1992. [12] J. K. Chapin and K. A. Moxon, Neural Prostheses for Restoration of Sensory and Motor Function. Boca Raton, FL: CRC, 2000. [13] R. Dai, R. B. Stein, B. J. Andrews, K. B. James, and M. Wieler, “Application of tilt sensors in functional electrical stimulation,” IEEE Trans. Rehab. Eng., vol. 4, pp. 63–72, June 1996. [14] M. Wieler, R. B. Stein, M. Ladouceur, M. Whittaker, A. W. Smith, S. Naaman, H. Barbeau, J. Bugaresti, and E. Aimone, “Multicenter evaluation of electrical stimulation systems for walking,” Arch. Phys. Med. Rehab., vol. 80, pp. 495–500, 1999.

[15] A. Prochazka, “Comparison of natural and artificial control of movement,” IEEE Trans. Rehab. Eng., vol. 1, pp. 7–17, Mar. 1993. [16] E. C. Field-Fote, “Spinal cord control of movement: Implications for locomotor rehabilitation following spinal cord injury,” Phys. Ther., vol. 80, pp. 477–484, 2000. [17] V. K. Mushahwar and K. W. Horch, “Proposed specifications for a lumbar spinal cord electrode array for control of lower extremities in paraplegia,” IEEE Trans. Rehab. Eng., vol. 5, pp. 237–243, Sept. 1997. [18] C. Tai, A. M. Booth, C. J. Robinson, W. C. de Groat, and J. R. Roppolo, “Isometric torque about the knee joint generated by microstimulation of the cat L6 spinal cord,” IEEE Trans. Rehab. Eng., vol. 7, pp. 46–55, Mar. 1999. [19] D. N. Rushton, N. D. Donaldson, F. M. Barr, V. J. Harper, T. A. Perkins, P. N. Taylor, and A. M. Tromans, “Lumbar root stimulation for restoring leg function: Results in paraplegia,” Artif. Organs, vol. 21, pp. 180–182, 1997. [20] V. K. Mushahwar, D. F. Collins, and A. Prochazka, “Spinal cord microstimulation generates functional limb movements in chronically implanted cats,” Exp. Neurol., vol. 163, pp. 422–429, 2000. [21] E. Bizzi, M. C. Tresch, P. Saltiel, and A. d’Avella, “New perspectives on spinal motor systems,” Nat. Rev. Neurosci., vol. 1, pp. 101–108, 2000. [22] S. F. Giszter, F. A. Mussa-Ivaldi, and E. Bizzi, “Convergent force fields organized in the frog’s spinal cord,” J. Neurosci., vol. 13, pp. 467–491, 1993. [23] Y. Aoyagi, V. Mushahwar, R. B. Stein, and A. Prochazka, “Movement synergies elicited by intraspinal microstimulation compared to stimulation of muscles, nerves and roots in the cat,” in Proc. Int. Functional Electrical Stimulation Soc., Cleveland, OH, 2001, pp. 120–122. [24] R. B. Stein, Y. Aoyagi, V. Mushahwar, and A. Prochazka, “Limb movements generated by stimulating muscle, nerve and spinal cord,” Arch. Ital. Biol., vol. 140, pp. 273–281, 2002. [25] J. A. Hoffer, R. B. Stein, M. K. Haugland, T. Sinkjaer, W. K. Durfee, A. B. Schwartz, G. E. Loeb, and C. Kantor, “Neural signals for command control and feedback in functional neuromuscular stimulation: A review,” J. Rehab. Res. Dev., vol. 33, pp. 145–157, 1996. [26] J. E. Crouch, Text-Atlas of Cat Anatomy. Philadelphia, PA: Lea & Febiger, 1969. [27] T. M. Hamm, W. Koehler, D. G. Stuart, and S. V. Noven, “Partitioning of monosynaptic Ia excitatory post-synaptic potentials in the motor nucleus of the cat semimembranosus muscle,” J. Physiol. (Lond.), vol. 369, pp. 379–398, 1985. [28] A. Prochazka, V. K. Mushahwar, and D. B. McCreery, “Neural prostheses,” J. Physiol., vol. 533, pp. 99–109, 2001. [29] W. F. Agnew and D. B. McCreery, “Considerations for safety with chronically implanted nerve electrodes,” Epilepsia, vol. 31, pp. S27–S32, 1990. [30] J. M. Macpherson, “Strategies that simplify the control of quadrupedal stance. I. Forces at the ground,” J. Neurophysiol., vol. 60, pp. 204–217, 1988. [31] C. S. Sherrington, “Flexion-reflex of the limb, crossed extension reflex and reflex stepping and standing,” J. Physiol. (Lond.), vol. 40, pp. 28–121, 1910. [32] C. M. Chanaud, C. A. Pratt, and G. E. Loeb, “Functionally complex muscles of the cat hindlimb. II. Mechanical and architectural heterogenity within the biceps femoris,” Exp. Brain Res., vol. 85, pp. 257–270, 1991. [33] W. Kahle, H. Leonhardt, and W. Platzer, Color Atlas and Textbook of Human Anatomy. Chicago, IL: Thieme, 1978. [34] G. J. Romanes, “The motor cell columns of the lumbo-sacral spinal cord of the cat,” J. Comp. Neurol., vol. 94, pp. 313–63, 1951. [35] V. G. Vanderhorst, H. de Weerd, and G. Holstege, “Evidence for monosynaptic projections from the nucleus retroambiguus to hindlimb motoneurons in the cat,” Neurosci. Lett., vol. 224, pp. 33–6, 1997. [36] V. G. Vanderhorst and G. Holstege, “Organization of lumbosacral motoneuronal cell groups innervating hindlimb, pelvic floor, and axial muscles in the cat,” J. Comp. Neurol., vol. 382, pp. 46–76, 1997. [37] A. Branner, R. B. Stein, and R. A. Normann, “Selective stimulation of cat sciatic nerve using an array of varying-length microelectrodes,” J. Neurophysiol., vol. 85, pp. 1585–1594, 2001. [38] D. Popovic, T. Gordon, V. F. Rafuse, and A. Prochazka, “Properties of implanted electrodes for functional electrical stimulation,” Ann. Biomed. Eng., vol. 19, pp. 303–316, 1991. [39] K. Singh, F. J. Richmond, and G. E. Loeb, “Recruitment properties of intramuscular and nerve-trunk stimulating electrodes,” IEEE Trans. Rehabil. Eng., vol. 8, pp. 276–285, Sept. 2000. [40] P. H. Gorman and J. T. Mortimer, “The effect of stimulus parameters on the recruitment characteristics of direct nerve stimulation,” IEEE Trans. Biomed. Eng., vol. 30, pp. 407–414, July 1983.


[41] T. Cameron, F. J. Richmond, and G. E. Loeb, “Effects of regional stimulation using a miniature stimulator implanted in feline posterior biceps femoris,” IEEE Trans. Biomed. Eng., vol. 45, pp. 1036–1043, Aug. 1998. [42] V. K. Mushahwar and K. W. Horch, “Muscle recruitment through electrical stimulation of the lumbo-sacral spinal cord,” IEEE Trans. Rehabil. Eng., vol. 8, pp. 22–29, 2000. [43] , “Selective activation of muscle groups in the feline hindlimb through electrical microstimulation of the ventral lumbo-sacral spinal cord,” IEEE Trans. Rehabil. Eng., vol. 8, pp. 11–21, Mar. 2000. [44] D. N. Rushton, Neuroprostheses, Neuromodulators and Rehabilitation. London, U.K.: Brit. Soc. Rehab. Med., 1997. [45] M. F. Somers, Spinal Cord Injury: Functional Rehabilitation. Norwalk: Appleton & Lange, 1992. [46] A. Scheiner, J. T. Mortimer, and U. Roessmann, “Imbalanced biphasic electrical stimulation: Muscle tissue damage,” Ann. Biomed. Eng., vol. 18, pp. 407–425, 1990. [47] G. G. Naples, J. T. Mortimer, and T. G. H. Yuen, “Overview or peripheral nerve electrode design and implementation,” in Neural Prostheses. Fundamental Studies, W. F. Agnew and D. B. McCreery, Eds. Englewood Cliffs, NJ: Prentice-Hall, 1990, pp. 107–145. [48] W. J. W. Sharrard, “The distribution of the permanent paralysis in the lower limb in poliomyelitis. A clinical and pathological study,” J. Bone Joint Surg., vol. 37B, pp. 540–558, 1955. [49] N. Donaldson, D. Rushton, and T. Tromans, “Neuroprostheses for leg function after spinal-cord injury,” Lancet, vol. 350, p. 711, 1997. [50] R. Davis and S. E. Emmonds, “Spinal cord stimulation for multiple sclerosis: Quantifiable benefits,” Stereotact. Funct. Neurosurg., vol. 58, pp. 52–58, 1992. [51] G. E. Loeb, M. J. Bak, and J. Duysens, “Long-term unit recording from somatosensory neurons in the spinal ganglia of the freely walking cat,” Science, vol. 197, pp. 1192–1194, 1977. [52] A. Prochazka, J. A. Stephens, and P. Wand, “Muscle spindle discharge in normal and obstructed movements,” J. Physiol. (Lond.), vol. 287, pp. 57–66, 1979. [53] R. Burnham, T. Martin, R. B. Stein, G. Bell, I. Maclean, and R. Steadward, “Skeletal muscle fiber type transformation following spinal cord injury,” Spinal Cord, vol. 35, pp. 86–91, 1997.

Yoichiro Aoyagi was born in Kyoto, Japan in 1968. He received the M.D. degree from the Kyoto Prefectural University of Medicine, Kyoto, in 1993 and the Ph.D. degree from the Centre for Neuroscience, University of Alberta, Edmonton, AB, Canada, in 2002. He completed his medical internship training in Yokosuka U.S. Naval Hospital, Yokosuka, Japan in 1994 and his residency training in the Department of Rehabilitation Medicine, Keio University School of Medicine, Keio, Japan, in 1998. He is currently an Assistant Professor with the Department of Rehabilitation Medicine, Kawasaki Medical School, Okayama, Japan. His research interests include FES, sensorimotor neurophysiology, and higher cortical function.

11

Vivian K. Mushahwar (M’97) received the B.S. degree in electrical engineering from Brigham Young University, Provo, UT, in 1991 and the Ph.D. degree in bioengineering from the University of Utah, Salt Lake City, in 1996. She received postdoctoral training at Emory University, Atlanta, GA, and the University of Alberta, Edmonton, AB, Canada. She is currently as Assistant Professor with the Department of Biomedical Engineering, University of Alberta. Her research interests include identification of spinal-cord systems involved in locomotion, development of spinal-cord-based neuroprostheses, and incorporation of motor control concepts in functional neuromuscular stimulation. Dr. Mushahwar is a member of IEEE Engineering in Medicine and Biology Society, IFESS, New York Academy of Sciences, and the Society for Neuroscience.

Richard B. Stein is currently a Professor of physiology and neuroscience at the University of Alberta, Edmonton, AB, Canada, and the Codirector of the Rehabilitation Neuroscience Group. His research interests include the control of movement such as the organization of spinal circuitry for walking, the role of reflexes in modulating behavior, properties of muscles and sensory receptors relevant to motor control, and the replacement of function by FES after motor disorders such as spinal-cord injury and stroke. He has authored nearly 300 scientific articles, excluding abstracts. Dr. Stein won the Centennial Medal of the International Tesla Foundation in 1998, the Medal of Honour of the Canadian Medical Association in 1999, and the Kaplan Research Prize from the University of Alberta in 2001. He has been President of the Canadian Physiological Society and the Canadian Association for Neuroscience.

Arthur Prochazka received the B.Eng. and M.Sc. degrees from the University of Melbourne, Melbourne, Australia, in 1967 and 1970, respectively, and the Dr.rer.nat. degree in neurophysiology from the University of Ulm, Ulm, Germany, in 1973. He spent two years as a postdoctoral fellow at with Monash University, Melbourne, Australia, and from 1976 to 1986, he was a Lecturer and Senior Lecturer at St. Thomas’s Hospital Medical School, University of London, London, U.K. Since 1986, he has been a Professor of physiology at the University of Alberta, Edmonton, AB, Canada. His research interests include sensorimotor control and neuroprostheses.