Nerve Cuff Recordings of Muscle Afferent Activity during Passive Joint Motion in a Rabbit

This study is part of an ongoing effort to determine if nerve cuff recordings from muscle afferents can provide sensory information for control of functional electrical stimulation (FES) systems that restore mobility in an otherwise paralyzed subject. The rabbit’s ankle was passively flexed and extended in the saggital plane while the activity from both the tibial and peroneal nerves was recorded. The responses of the two nerves behaved in a reciprocal (pushpull) manner. The results suggest that signals derived from muscle afferents (in non-stimulated muscles) using cuff electrodes may be useful for the control of FES systems. Introduction. Natural sensors are embedded in the muscles, tendons, joint capsules and skin from where they send information about muscle length, limb loading and joint position to the central nervous system (CNS) via peripheral nerves (Gandevia et al., 1992). Natural sensors might be an attractive alternative to artificial sensors because they are available and functioning in virtually all the patients who would use a motor prosthesis. If reliable information from these sensors can be extracted, it might be used to provide feedback in clinical FES assistive devices (Haugland and Hoffer 1994; Haugland and Sinkjær 1995). The objective of the present study was to describe neural activity regarding ankle joint kinesthesia (flexion-extension) recorded from muscle afferents using whole nerve cuff electrodes in a rabbit model of human peripheral nerves. Cuff electrodes register the compound activity within the nerve and have the property that they are most sensitive for recording from the larger axons within the nerve( Hoffer 1990). This characteristic is fortuitous for our application because the muscle afferents are among the largest fibers in mixed nerves. Part of this study has been submitted for publication (Riso et al., 1997). Methods Acute experiments were conducted on five New Zealand White adult rabbits (3-5Kg) anesthetized with an IM injection of Fentanyl/Fluanision. An incision was made on the lateral side of the left hind limb to expose the sciatic nerve approx. 3cm above the knee. The tibial and the peroneal nerves were separated proximally to create enough length to install 22mm long spiral cuff electrodes around each nerve while the sural nerve was transected to eliminate its cutaneous input. The muscles and the skin were then closed at the incision site. Two more incisions were made proximal to the ankle joint on the lateral and medial sides of the shank to expose the descending branches of the tibial and peroneal nerves that innervate the foot and the ankle. Both branches were transected to eliminate proprioceptive input from the ankle joint and foot. Finally the foot was secured in a cradle that restricted movement to rotation of the ankle joint only in the saggital plane. The cradle was coupled to a computer controlled motor shaft to produce repeatable flexion-extension movements. Nerve signals were pre-amplified, band pass filtered at 1-5 kHz and sampled at 10kHz along with ankle torque and position signals. Rectification, averaging, and all other signal processing were performed digitally. Results. The activity of the tibial and peroneal nerves during movement of the ankle joint was mainly reciprocal. During ankle flexion, the extensor muscle group was stretched and most of the recorded activity was from the tibial nerve. Conversely, during ankle extension, the flexor muscle group was stretched and the recorded activity was mainly from the peroneal nerve. The records shown in Fig. 1 are exerpted from a series of 10 repetitive movements in which the ankle was passively rotated at a velocity of 30 deg/s through a 60° excursion starting with the foot fully extended (130°); moving to the fully flexed position (70°), holding that position for 2s and then returning (30 deg/s) to the fully extended position. The unprocessed nerve activities are shown in traces b and c in the figure while the corresponding rectified and integrated nerve activities are presented as traces d and e. At the onset of the extension plateau (static position) phase of the motion as denoted by point 1 in traces a e in Fig. 1, there was ongoing activity in the peroneal nerve recording. This activity ceased abruptly as soon as the ankle began to move in the flexion direction (point 2). There was only a very weak driven activity for either nerve during the interval when the ankle was rotated from 130° until just beyond 100° (neutral). Once the flexion motion crossed the neutral position, however, a strong stretch response was evoked in the tibial nerve (point 3). The tibial nerve continued to show substantial activity during the sustained flexion position associated with the flexion plateau (point 4). Finally, this activity ceased abruptly as soon as the ankle began to be rotated in the extension direction (point 5), (similar to the effect observed with the peroneal activity at the end of the extension plateau phase). We observed that the response to ankle stretch was greater if the ankle joint was allowed to remain stationary for a protracted period. This is shown in Fig. 2 where tibial nerve responses to successive stretches have been superimposed. The two trials with early response onset resulted when the intertrial delay was 3 minutes, while the remaining eight traces are the responses when the intertrial delay was only 1s. Fig. 1 The ankle was alternately flexed and extended as indicated in trace [a]. The dotted line at 100° indicates the neutral position where there was no passive torque present at the ankle. [b,c] Nerve activity recorded from the tibial and peroneal nerves, (bandpass filtered at 1 5 kHz). [d,e] rectified and integrated nerve activity. Fig. 2 Illustrating the enhanced responses recorded during ankle rotation when a prolonged stationary rest period is introduced between the ramp trials. A stationary period of 3 minutes between stretch trials pertained for the two early tibial nerve responses. The remaining 8 responses occurred with intertrial intervals of 1second. Discussion. The recorded activity is most likely dominated by muscle spindle activity, since tendon organs are not very active in the passive muscle (Houk and Hennemann 1967). Consistent with muscle spindle firing properties (Houk et al., 1981) we observed pronounced hystereses (where there is little or no activity during the relaxation phase of the muscle stretches) in recorded activity from both the tibial and the peroneal nerves. In an FES application, the hystereses effects can be overcome by only using the activity from the tibial nerve during flexion and the activity from the peroneal nerve during extension motion. Muscle activity recorded from cuff electrodes in an FES application may only be of use when obtained from nonstimulated muscles. However, it may also be useful to investigate the case of actively shortening muscle, where the tendon organs will be active and the muscle spindles will be quiet. The history dependence shown in Fig.2 is likely related to the viscoelastic properties of the muscle/tendon system and may complicate the use of the muscle afferent responses for control purposes if there are prolonged periods with no movement. While such stationarity is unlikely for the case of FES assisted standing, we are studying this history dependence in detail so that it can be thoroughly described. Using intrafasicular recording techniques, Yoshida and Horch (1966) were able to control ankle movement in acute animal studies for limited ankle rotation velocities. Problems with migration of intrafasicular electrodes, however, may ultimately preclude that approach for chronic applications, whereas cuff recording techniques have been shown to have good recording stability (Thomsen et. al.this meeting). Conclusion. These results suggest that signals derived from muscle afferents (in non-stimulated muscles) using cuff electrodes may be useful for the control of FES systems. It would be possible to know when an innervated muscle is subjected to added stretch by the increased activity, and conversely, it would be possible to deduce that such stretch has begun to relax if there is an abrupt halt in the ongoing activity. Some rate information is also present in the responses to movement, and this may be useful in grading an appropriate correction factor during closed loop control of FES systems. References. Gandevia et al. TINS, vol.,15(2), pp. 62-65, 1992. Haugland & Hoffer.IEEE Trans. Rehab. Eng., vol., 2(1), pp. 2936, 1994. Haugland & Sinkjear, IEEE Trans. Rehab. Eng., vol. 3(4), pp. 307-317, 1995. Hoffer, In Neuromethods, vol., 15, Neurophysiol. Techniques: applications to neural systems, A.A. Boulton, Baker, and Vanderwolf, Eds., Humana Press, Clifton, N.J., 1990, pp. 65145. Houk & Henneman, J. Neurophysiol., vol. 30; pp. 466-481, 1967. Houk et al. J. Neurophysiol., vol., 46, 1981. Naples et al. IEEE Trans. Biomed. Eng., vol 35, pp. 905-916, 1988. Riso, R.R., Mosallaie, F.K., Sinkjær, T., Nerve cuff recordings of muscle afferent activity from tibial and peroneal nerves in rabbit during passive ankle motion. IEEE Trans. on Rehab. Eng., 1997 (Submitted). Yoshida & Horch IEEE Trans. Biomed. Eng., vol., 43(2), pp. 167-176, 1996. Acknowledgement This work was supported by The Danish National Research Foundation.