Annals" of Biomedical Engineering, Vol. 19, pp. 303-316, 1991 Printed in the USA. All rights reserved.
0090 6964/91 $3.00 + .00 Copyright 9 1991 Pergamon Press plc
Properties of Implanted Electrodes for Functional Electrical Stimulation D e j a n P o p o v i c , Tessa G o r d o n , Victor F. R a f u s e , and Arthur P r o c h a z k a Division of Neuroscience University of Alberta Edmonton, Alta., Canada (Received 7/17/90," Revised 12/13/90)
Implanted wire electrodes are increasingly being used f o r the functional electrical stimulation o f muscles in partially paralysed patients, yet many o f their basic characteristics are poorly understood. In this study we investigated the selectivity, recruitment characteristics and range o f control o f several types o f electrode in triceps surae and plantaris muscles o f anaesthetized cats. We f o u n d that nerve cuffs are more efficient and selective (i.e., cause less stimulus spread to surrounding muscles) than intramuscular electrodes. Bipolar intramuscular stimulation was more efficient and selective than monopolar stimulation, but only if the nerve entry point was between the electrodes. Monopolar electrodes are eJ)Cicient and selective if located close to the nerve entry point, but their performance declines with distance from it. Nonetheless, for a variety o f reasons monopolar stimulation provides the best compromise in many current applications. Short duration pulses offer the best efficiency (least charge per pulse to elicit force) but high peak currents, increasing the risk o f electrode corrosion and tissue damage. Electrode size has little effect on recruitment and should therefore be maximised because this minimises current density. K e y w o r d - Electrical stimulation electrodes.
INTRODUCTION In various motor disorders, it is possible to activate dysfunctional muscles with functional electrical stimulation (FES). The application of FES to the restoration of gait was first investigated systematically in Ljubljana, Yugoslavia (2,19,20,44). FES is now used in a clinical setting in several rehabilitation centers (1,25,36) and there is a limited but growing trend for the design of devices for home use. FES systems developed for the upper extremities can restore limited grasp functions in some quadriplegic patients, and are also at a transitional stage from the laboratory to the home environment (13,34,35). Multichannel stimulators have been developed for the coordinated activation of several muscle groups either by direct stimulation of efferent Acknowledgments-This work was supported by the Alberta Heritage Foundation for Medical Research and the Canadian Medical Research Council. Address correspondenceto Dr. A. Prochazka, Division of Neuroscience, University of Alberta, Edmonton, Alta. T6G 2S2, Canada.
303
304
D. Popovic et al.
nerve fibres or by the stimulation of afferent nerve fibres, leading to synergistic reflexes (2,18,21,24,30,34,36,38,42). For effective design of FES systems it is important to understand the biomechanics of the muscles, the neurophysiological mechanisms involved in normal motor control and the functional status of the injured neuromuscular structures (44). However, it is equally important to address the following practical issues which have arisen in the course of implementing FES systems: 9 How does electrode placement affect stimulation? 9 How do monopolar and bipolar stimulation compare with regard to stimulus spread and maximal achievable force? 9 How do pulse waveform parameters affect stimulation? Current FES systems use various configurations of surface and implantable electrodes. Many types of surface electrode are available including conductive rubber patches coated with electrolyte gel, metal plates contacting the skin via thin, moist sponges, and flexible, disposable, stainless-steel mesh or rubber electrodes with selfadhesive conductive polymers (2,3,27). They are relatively easy to apply and replace, are noninvasive, and in some cases, are designed to stay attached for days and even weeks. However, they suffer from various drawbacks that include: (a) activation of skin pain receptors; (b) difficulty of reproducible positioning; (c) poor selectivity: stimulation spreads to muscles other than those targeted; (d) difficulty of fixation on moving limbs; (e) variability of electrical properties of the skin-electrode interface; (f) skin inflammation; (g) unacceptable appearance. As a result, implantable electrodes have been developed to avoid these drawbacks (e.g., 31). However, the implantable electrodes currently in use, nerve cuff, intraneural, epimysial, and intramuscular electrodes, have some drawbacks. Nerve cuffs, typically comprising short silastic tubes containing fine wires, stimulate efficiently and selectively but require open surgery for implantation (10,15,17,39,43). They may also damage the nerves they enclose. Similar problems are encountered with intraneural electrodes, which in addition are difficult to insert and stabilise (4,16,37). Epimysial electrodes sutured to muscle fascia have not yet been evaluated in humans (12). Wire intramuscular electrodes are the most commonly implanted type, and have the advantage that they may be inserted percutaneously via a canula without open surgery (6,7,13,15,22, 24,25,32,40). However, positioning and anchoring o f wire electrodes are still problematical in FES systems. There is surprisingly little information to guide the surgeon in the placement of intramuscular electrodes. Generally they are inserted as deeply as possible into the belly o f the target muscle, with or without concomitant stimulation (e.g., 40). In our study we explored various aspects of electrode configuration and placement in relation to the efficacy and selectivity of activation of muscles sharing a common nerve supply. We also studied the effect of varying electrode length and pulse parameters. The aim was to establish some basic principles which would be of general applicability to the implantation of intramuscular and epimysial electrodes for FES. METHODS
The performance of electrodes in activating muscle was evaluated with reference to maximal muscle force elicited with supramaximal tetanic nerve stimulation. We
Implanted FES Electrodes
305
chose the ankle extensor muscles for testing purposes for the following reasons: (a) they are large muscles in physical contact with each other over their entire length; (b) they are innervated by branches of one and the same nerve (tibial), (c) there is a well-established procedure for limb fixation and dissection, which ensures that the forces in the muscles may be monitored accurately without contamination from other hind limb muscles; (d) muscle length can be systematically controlled. Five adult cats were studied. The animals were anaesthetized with an initial intraperitoneal dose of pentobarbital (50 mg/kg) and anaesthesia was maintained with subsequent doses given intravenously as required. The trachea was cannulated to allow for artificial ventilation if needed. The triceps surae muscle group was exposed and the medial gastrocnemius (MG), lateral gastrocnemius (LG), and plantaris (Pl) tendons were isolated for attachment to force transducers (Grass Model FT10). Care was taken not to disturb the blood supply to these muscles or the physical contact between them. Triphasic silastic cuff electrodes (39) were implanted around the sciatic nerve for the recording of neural potentials, and around the MG nerve for electrical stimulation. An epimysial electromyographic (EMG) electrode (15) was sewn to the MG muscle fascia in two cats. Intramuscular stainless steel stimulating electrodes (Cooner wire, AS631) were implanted into the belly of the MG muscle at four positions (Fig. 1) using a 26-gauge needle. Wires with bared lengths of 1, 5, and 10 m m were used. Position 1 was just proximal to the MG nerve entry point and position 2 just distal to it. Positions 3 and 4 were approximately 30 and 60 m m distal to the nerve entry point respectively. The
LGf o r cMG eforce ~
~
M~GEM,.-,/
~
P l a n t a r i s ~
U
4
3
f SO'ram
I1"
2
Intramuscularelectrodes FIGURE1. Schematic
(MG)via
of the arrangement to stimulate cat medial gastrocnemius muscle either a nerve cuff electrode or intramuscular electrodes implanted at the following locations with respect to the MG nerve entry (motor) point: (a) just proximal, (b) just distal, mm distal, Isometric forces exerted by and plantaris muscles were separately monitored. MG electromyogram (EMG) was recorded with an epimysial patch electrode. Nerve cuffs were used to record neurograms (compound action potentials: CAPs) either from the sciatic nerve, when stimulation was via nerve cuff, or the MG nerve, when stimulation was via the intramuscular electrodes.
(d)60mmdistal.
MG,lateralgastrocnemius(LG)
theMG
(c) 30
306
D. Popovic et al.
skin wound was sutured shut to restore the limb to as near-normal as possible. A reference electrode, consisting either of a 21-gauge hypodermic needle inserted subcutaneously along the midline of the back or of a moistened pad pressed onto the cat's back was used as the indifferent electrode for m o n o p o l a r stimulation. The animals were supported in a prone position in a stereotaxic frame. Hip, knee, and ankle were immobilized with metal pins. Body temperature was maintained with a feedback-regulated heating pad. The tendons of the M G a n d / o r LG and plantaris were each tied to the force transducers with stout surgical thread. The length of the M G muscle was adjusted for m a x i m u m twitch and tetanic force. Experimental Protocol
Electrical stimulation was performed with an isolated, battery-powered stimulator which delivered feedback-controlled, monophasic, constant-current pulses. Between pulses no current flowed in the stimulating electrodes. Pulse amplitude (0.5-50 mA), pulse width (5-500 ~s) and interpulse interval (10 ms-1 s) were independently controlled. All of these characteristics were monitored and checked with a Tektronix 5111 oscilloscope and current probe. In one of the experiments asymmetrical biphasic constant-current pulses were used: the results obtained were qualitatively identical to the characteristics observed with monophasic stimulation, though thresholds were generally somewhat higher (there was no time delay between the two phases of the pulse). The effectiveness and selectivity of stimulation of MG muscle was determined from the isometric force monitored in MG, LG and plantaris muscles, from compound action potentials (CAPs) recorded with nerve cuffs on the sciatic and MG nerves, and from E M G recorded with epimysial electrodes on M G muscle. The signals were amplified and digitized with an LSI-11 computer which also received trigger signals from the isolated stimulator. Trials included: 1. Short trains of stimulus pulses (5/s, 10/s, or 100/s) through the MG electrode to determine maximal M G contractile force. 2. M o n o p o l a r and bipolar intramuscular stimulation with different configurations. 3. Stimulation with current-controlled pulses modulated in amplitude 4. Intramuscular stimulation with electrodes of varying active length 10 mm).
nerve cuff electrode or width. (1, 5, and
RESULTS When bipolar intramuscular electrodes straddled the nerve entry point they were as effective at activating MG as a nerve cuff. In the experiment of Fig. 2, the intramuscular electrodes were in positions 1 (cathode) and 4 (anode) of Fig. 1, straddling the nerve entry point. The MG nerve cuff was about 20 m m proximal to the nerve entry point. Intramuscular stimulation recruited all M G motor units as evidenced by the fact that retrograde sciatic CAPs and M G forces were identical in amplitude to those elicited by supramaximal nerve stimulation (Figs. 2A, B, E, and F). Similar results were obtained in three other cats. M G nerve stimulation with 10 ~s pulses increasing f r o m 3 to 8 m A (Fig. 3A) produced maximal MG motoneuronal recruitment, reflected in plateaus in the sciatic
Implanted FES Electrodes
307
Bipolar intramuscular stimulation +10
MG nerve stimulation 410 ~ /
t
A
Sciatic neurogram
mV 0
mV 0 [s
10 +10
mV 0 ~ 10 5O
B Sciatic neurogram
/ -10 L
1 ms
mV 0 ~
E
1
MG force
rns
lOO/sstim
loL 50F
1ms F
MG force
lOO/sstirn
N 10/8 stim 0
0 ~
10ms
10ms
FIGURE 2. Data showing that bipolar intramuscular stimulation (left) can activate MG as effectively as supramaximal nerve stimulation (right). Intramuscular electrodes were located at positions 1 (cathode) and 4 (anode) of previous figure. (a), (b), (c): CAPs from sciatic and MG nerves. (d): MG EMG recorded with epimysial electrode. (e), (f): MG force elicited by stimulus trains at 10 and 100/s respectively.
CAP and MG force, without stimulus spread to other muscles. Above 9 mA, plantaris muscle was activated, and the sciatic CAP amplitude quickly increased, reflecting a spread of current to nerve branches neighbouring that of MG, presumably with recruitment of sensory as well as motor nerves. By contrast, stimulation via bipolar intramuscular electrodes (Fig. 3B) required more current and had a smaller range over which stimulation remained selective. Results were similar in the three cats tested in this way. If the bipolar electrodes did not straddle the nerve entry point, full recruitment of the muscle required very strong stimulation, and could not be achieved without activation of surrounding muscles (Fig. 4). In the best of three bipolar configurations (Fig. 4A: electrode positions 2,4) 80% of maximal force (elicited by nerve stimulation) was reached at 8 ~s pulse width, but no further recruitment occurred, even when pulse width was increased by an order of magnitude (force actually declined slightly, which we attribute to muscle fatigue). Electrode configurations 2,3 and 3,4 produced at most 30~ and 20~ of maximal force respectively. In contrast, monopolar stimulation was less position-dependent, though it could not match the selectivity obtained with good bipolar placement. Figure 4D shows that 100~ recruitment was achieved with monopolar stimulation at position 2 close to the nerve entry point. However, the increase in force was quite nonlinear, showing a twotier profile (see also Fig. 7 below) and full recruitment was accompanied by spread to LG muscle. As the electrode was shifted distally (E and F) the initial force plateau changed, spread occurred earlier and full recruitment required greater pulse widths. Similar results were obtained in two other cats.
308
D. Popovic et al. l O / s MG nerve stimulation
A
(10 p.s pulses) 1.6.
E
B
1.2-
/
...... t.l.l----_==~l~ .............
13.,.
o
0090 6964/91 $3.00 + .00 Copyright 9 1991 Pergamon Press plc
Properties of Implanted Electrodes for Functional Electrical Stimulation D e j a n P o p o v i c , Tessa G o r d o n , Victor F. R a f u s e , and Arthur P r o c h a z k a Division of Neuroscience University of Alberta Edmonton, Alta., Canada (Received 7/17/90," Revised 12/13/90)
Implanted wire electrodes are increasingly being used f o r the functional electrical stimulation o f muscles in partially paralysed patients, yet many o f their basic characteristics are poorly understood. In this study we investigated the selectivity, recruitment characteristics and range o f control o f several types o f electrode in triceps surae and plantaris muscles o f anaesthetized cats. We f o u n d that nerve cuffs are more efficient and selective (i.e., cause less stimulus spread to surrounding muscles) than intramuscular electrodes. Bipolar intramuscular stimulation was more efficient and selective than monopolar stimulation, but only if the nerve entry point was between the electrodes. Monopolar electrodes are eJ)Cicient and selective if located close to the nerve entry point, but their performance declines with distance from it. Nonetheless, for a variety o f reasons monopolar stimulation provides the best compromise in many current applications. Short duration pulses offer the best efficiency (least charge per pulse to elicit force) but high peak currents, increasing the risk o f electrode corrosion and tissue damage. Electrode size has little effect on recruitment and should therefore be maximised because this minimises current density. K e y w o r d - Electrical stimulation electrodes.
INTRODUCTION In various motor disorders, it is possible to activate dysfunctional muscles with functional electrical stimulation (FES). The application of FES to the restoration of gait was first investigated systematically in Ljubljana, Yugoslavia (2,19,20,44). FES is now used in a clinical setting in several rehabilitation centers (1,25,36) and there is a limited but growing trend for the design of devices for home use. FES systems developed for the upper extremities can restore limited grasp functions in some quadriplegic patients, and are also at a transitional stage from the laboratory to the home environment (13,34,35). Multichannel stimulators have been developed for the coordinated activation of several muscle groups either by direct stimulation of efferent Acknowledgments-This work was supported by the Alberta Heritage Foundation for Medical Research and the Canadian Medical Research Council. Address correspondenceto Dr. A. Prochazka, Division of Neuroscience, University of Alberta, Edmonton, Alta. T6G 2S2, Canada.
303
304
D. Popovic et al.
nerve fibres or by the stimulation of afferent nerve fibres, leading to synergistic reflexes (2,18,21,24,30,34,36,38,42). For effective design of FES systems it is important to understand the biomechanics of the muscles, the neurophysiological mechanisms involved in normal motor control and the functional status of the injured neuromuscular structures (44). However, it is equally important to address the following practical issues which have arisen in the course of implementing FES systems: 9 How does electrode placement affect stimulation? 9 How do monopolar and bipolar stimulation compare with regard to stimulus spread and maximal achievable force? 9 How do pulse waveform parameters affect stimulation? Current FES systems use various configurations of surface and implantable electrodes. Many types of surface electrode are available including conductive rubber patches coated with electrolyte gel, metal plates contacting the skin via thin, moist sponges, and flexible, disposable, stainless-steel mesh or rubber electrodes with selfadhesive conductive polymers (2,3,27). They are relatively easy to apply and replace, are noninvasive, and in some cases, are designed to stay attached for days and even weeks. However, they suffer from various drawbacks that include: (a) activation of skin pain receptors; (b) difficulty of reproducible positioning; (c) poor selectivity: stimulation spreads to muscles other than those targeted; (d) difficulty of fixation on moving limbs; (e) variability of electrical properties of the skin-electrode interface; (f) skin inflammation; (g) unacceptable appearance. As a result, implantable electrodes have been developed to avoid these drawbacks (e.g., 31). However, the implantable electrodes currently in use, nerve cuff, intraneural, epimysial, and intramuscular electrodes, have some drawbacks. Nerve cuffs, typically comprising short silastic tubes containing fine wires, stimulate efficiently and selectively but require open surgery for implantation (10,15,17,39,43). They may also damage the nerves they enclose. Similar problems are encountered with intraneural electrodes, which in addition are difficult to insert and stabilise (4,16,37). Epimysial electrodes sutured to muscle fascia have not yet been evaluated in humans (12). Wire intramuscular electrodes are the most commonly implanted type, and have the advantage that they may be inserted percutaneously via a canula without open surgery (6,7,13,15,22, 24,25,32,40). However, positioning and anchoring o f wire electrodes are still problematical in FES systems. There is surprisingly little information to guide the surgeon in the placement of intramuscular electrodes. Generally they are inserted as deeply as possible into the belly o f the target muscle, with or without concomitant stimulation (e.g., 40). In our study we explored various aspects of electrode configuration and placement in relation to the efficacy and selectivity of activation of muscles sharing a common nerve supply. We also studied the effect of varying electrode length and pulse parameters. The aim was to establish some basic principles which would be of general applicability to the implantation of intramuscular and epimysial electrodes for FES. METHODS
The performance of electrodes in activating muscle was evaluated with reference to maximal muscle force elicited with supramaximal tetanic nerve stimulation. We
Implanted FES Electrodes
305
chose the ankle extensor muscles for testing purposes for the following reasons: (a) they are large muscles in physical contact with each other over their entire length; (b) they are innervated by branches of one and the same nerve (tibial), (c) there is a well-established procedure for limb fixation and dissection, which ensures that the forces in the muscles may be monitored accurately without contamination from other hind limb muscles; (d) muscle length can be systematically controlled. Five adult cats were studied. The animals were anaesthetized with an initial intraperitoneal dose of pentobarbital (50 mg/kg) and anaesthesia was maintained with subsequent doses given intravenously as required. The trachea was cannulated to allow for artificial ventilation if needed. The triceps surae muscle group was exposed and the medial gastrocnemius (MG), lateral gastrocnemius (LG), and plantaris (Pl) tendons were isolated for attachment to force transducers (Grass Model FT10). Care was taken not to disturb the blood supply to these muscles or the physical contact between them. Triphasic silastic cuff electrodes (39) were implanted around the sciatic nerve for the recording of neural potentials, and around the MG nerve for electrical stimulation. An epimysial electromyographic (EMG) electrode (15) was sewn to the MG muscle fascia in two cats. Intramuscular stainless steel stimulating electrodes (Cooner wire, AS631) were implanted into the belly of the MG muscle at four positions (Fig. 1) using a 26-gauge needle. Wires with bared lengths of 1, 5, and 10 m m were used. Position 1 was just proximal to the MG nerve entry point and position 2 just distal to it. Positions 3 and 4 were approximately 30 and 60 m m distal to the nerve entry point respectively. The
LGf o r cMG eforce ~
~
M~GEM,.-,/
~
P l a n t a r i s ~
U
4
3
f SO'ram
I1"
2
Intramuscularelectrodes FIGURE1. Schematic
(MG)via
of the arrangement to stimulate cat medial gastrocnemius muscle either a nerve cuff electrode or intramuscular electrodes implanted at the following locations with respect to the MG nerve entry (motor) point: (a) just proximal, (b) just distal, mm distal, Isometric forces exerted by and plantaris muscles were separately monitored. MG electromyogram (EMG) was recorded with an epimysial patch electrode. Nerve cuffs were used to record neurograms (compound action potentials: CAPs) either from the sciatic nerve, when stimulation was via nerve cuff, or the MG nerve, when stimulation was via the intramuscular electrodes.
(d)60mmdistal.
MG,lateralgastrocnemius(LG)
theMG
(c) 30
306
D. Popovic et al.
skin wound was sutured shut to restore the limb to as near-normal as possible. A reference electrode, consisting either of a 21-gauge hypodermic needle inserted subcutaneously along the midline of the back or of a moistened pad pressed onto the cat's back was used as the indifferent electrode for m o n o p o l a r stimulation. The animals were supported in a prone position in a stereotaxic frame. Hip, knee, and ankle were immobilized with metal pins. Body temperature was maintained with a feedback-regulated heating pad. The tendons of the M G a n d / o r LG and plantaris were each tied to the force transducers with stout surgical thread. The length of the M G muscle was adjusted for m a x i m u m twitch and tetanic force. Experimental Protocol
Electrical stimulation was performed with an isolated, battery-powered stimulator which delivered feedback-controlled, monophasic, constant-current pulses. Between pulses no current flowed in the stimulating electrodes. Pulse amplitude (0.5-50 mA), pulse width (5-500 ~s) and interpulse interval (10 ms-1 s) were independently controlled. All of these characteristics were monitored and checked with a Tektronix 5111 oscilloscope and current probe. In one of the experiments asymmetrical biphasic constant-current pulses were used: the results obtained were qualitatively identical to the characteristics observed with monophasic stimulation, though thresholds were generally somewhat higher (there was no time delay between the two phases of the pulse). The effectiveness and selectivity of stimulation of MG muscle was determined from the isometric force monitored in MG, LG and plantaris muscles, from compound action potentials (CAPs) recorded with nerve cuffs on the sciatic and MG nerves, and from E M G recorded with epimysial electrodes on M G muscle. The signals were amplified and digitized with an LSI-11 computer which also received trigger signals from the isolated stimulator. Trials included: 1. Short trains of stimulus pulses (5/s, 10/s, or 100/s) through the MG electrode to determine maximal M G contractile force. 2. M o n o p o l a r and bipolar intramuscular stimulation with different configurations. 3. Stimulation with current-controlled pulses modulated in amplitude 4. Intramuscular stimulation with electrodes of varying active length 10 mm).
nerve cuff electrode or width. (1, 5, and
RESULTS When bipolar intramuscular electrodes straddled the nerve entry point they were as effective at activating MG as a nerve cuff. In the experiment of Fig. 2, the intramuscular electrodes were in positions 1 (cathode) and 4 (anode) of Fig. 1, straddling the nerve entry point. The MG nerve cuff was about 20 m m proximal to the nerve entry point. Intramuscular stimulation recruited all M G motor units as evidenced by the fact that retrograde sciatic CAPs and M G forces were identical in amplitude to those elicited by supramaximal nerve stimulation (Figs. 2A, B, E, and F). Similar results were obtained in three other cats. M G nerve stimulation with 10 ~s pulses increasing f r o m 3 to 8 m A (Fig. 3A) produced maximal MG motoneuronal recruitment, reflected in plateaus in the sciatic
Implanted FES Electrodes
307
Bipolar intramuscular stimulation +10
MG nerve stimulation 410 ~ /
t
A
Sciatic neurogram
mV 0
mV 0 [s
10 +10
mV 0 ~ 10 5O
B Sciatic neurogram
/ -10 L
1 ms
mV 0 ~
E
1
MG force
rns
lOO/sstim
loL 50F
1ms F
MG force
lOO/sstirn
N 10/8 stim 0
0 ~
10ms
10ms
FIGURE 2. Data showing that bipolar intramuscular stimulation (left) can activate MG as effectively as supramaximal nerve stimulation (right). Intramuscular electrodes were located at positions 1 (cathode) and 4 (anode) of previous figure. (a), (b), (c): CAPs from sciatic and MG nerves. (d): MG EMG recorded with epimysial electrode. (e), (f): MG force elicited by stimulus trains at 10 and 100/s respectively.
CAP and MG force, without stimulus spread to other muscles. Above 9 mA, plantaris muscle was activated, and the sciatic CAP amplitude quickly increased, reflecting a spread of current to nerve branches neighbouring that of MG, presumably with recruitment of sensory as well as motor nerves. By contrast, stimulation via bipolar intramuscular electrodes (Fig. 3B) required more current and had a smaller range over which stimulation remained selective. Results were similar in the three cats tested in this way. If the bipolar electrodes did not straddle the nerve entry point, full recruitment of the muscle required very strong stimulation, and could not be achieved without activation of surrounding muscles (Fig. 4). In the best of three bipolar configurations (Fig. 4A: electrode positions 2,4) 80% of maximal force (elicited by nerve stimulation) was reached at 8 ~s pulse width, but no further recruitment occurred, even when pulse width was increased by an order of magnitude (force actually declined slightly, which we attribute to muscle fatigue). Electrode configurations 2,3 and 3,4 produced at most 30~ and 20~ of maximal force respectively. In contrast, monopolar stimulation was less position-dependent, though it could not match the selectivity obtained with good bipolar placement. Figure 4D shows that 100~ recruitment was achieved with monopolar stimulation at position 2 close to the nerve entry point. However, the increase in force was quite nonlinear, showing a twotier profile (see also Fig. 7 below) and full recruitment was accompanied by spread to LG muscle. As the electrode was shifted distally (E and F) the initial force plateau changed, spread occurred earlier and full recruitment required greater pulse widths. Similar results were obtained in two other cats.
308
D. Popovic et al. l O / s MG nerve stimulation
A
(10 p.s pulses) 1.6.
E
B
1.2-
/
...... t.l.l----_==~l~ .............
13.,.
o
