Experimental determination of compound action potential direction and propagation velocity from multi-electrode nerve cuffs

Abstract

Information extracted from whole-nerve electroneurograms, recorded using electrode cuffs, can provide signals to neuroprostheses. However, the amount of information that can be extracted from a single tripole is limited. This communication demonstrates how previously unavailable information about the direction of action potential propagation and velocity can be obtained using a multi-electrode cuff and that the arrangement acts as a velocity-selective filter. Results from in vitro experiments on frog nerves are presented.

Introduction

Devices for correcting drop-foot were the first and are still the most commonly used type of motor neuroprosthesis. Recently Hansen et al. [1] showed that the naturally occurring neural signal in a sensory nerve from the foot (sural nerve) could be used as the input source to a system that activates the tibialis anterior muscle via the peroneal nerve. In their patient, the neural signal was picked up by a tripolar cuff which produces a signal in which afferent and efferent traffic is indistinguishable, and all action potentials of all velocities ( all fibre diameters) are mixed, although there is a strong weighting of the detected signal to the larger fibres [2]. For their application, the tripole was adequate because the traffic was all afferent, and because contact with the ground is the dominant means of excitation of the sural nerve’s receptors during walking. If naturally occurring nerve signals are to be used for the detection of signals that are not dominant in the nerve traffic, it will be essential to discriminate between traffic with different propagation velocities.

We have proposed that this can be done using a multi-electrode cuff (MEC) and an array of amplifiers, arranged to give double-differential recording [3] (a technique that has been used by Farina et al. [4] for EMG recording) combined with subsequent signal processing. In this Communication, we present our first experimental evidence for the feasibility of this idea.

Fig. 1 shows a double-differential array of amplifiers. Each triplet of electrodes is connected to three amplifiers in the so-called True Tripole configuration [5], which determines the recorded signal amplitude and shape. The timing of a single fibre action potential (SFAP), recorded by each tripole, depends on the distance of the tripole from the end of the cuff (Fig. 1). Thus each tripole records the same action potential (AP) delayed by a time T, as given by

$$\text{T=d}\text{/}\text{ν}$$

where d is the electrode pitch and ν is the propagation velocity of the AP. The time delay T is visible on the recorded data and, in combination with the known electrode distance, allows calculation of velocity ν.

A method for the measurement of propagation velocity is to sum the outputs of the tripole amplifiers after introducing artificial time shifts τ. When τ matches T, the AP peaks add constructively to give the largest signal. Conversely, when using mismatched delays τ, the amplitude remains smaller. The time shift that results in the maximal amplitude corresponds to T and the sign of τ indicates the direction of propagation. Such an arrangement is, in effect, a velocity-selective filter since τ tunes the arrangement to one matched velocity. This applies to an SFAP but, because the system is linear, the response to a compound action potential (CAP) will be the sum of the responses (SFAPs) of all the active fibres. Thus, it should be possible to record selectively from different fibre populations in a compound nerve by using several channels with different time shifts τ, each tuned to a specific population.