Abstract

The F-wave is a recurrent discharge of an antidromically activated motor neurone and has received little attention in the FES literature. It though may have important influence on the design of neural prosthesis. It is present in the myoelectric signal from the 16Hz stimulated tibialis anterior muscle with a mean latency (F_start) of 35ms and an end interval (F_end) of 55ms. Since the F-wave is normally used for diagnostics it is measured at very low frequency and latency is evaluated manually. The elevated frequency used for FES purpose together with the need to analyse long term effects has created a need for an automatically method to characterise F-wave. Such a method to determine F_start and F_end has been developed and tested in 5*20 trials on five neurological normal male subjects. Results show a low intra-subject standard deviation. With the hypothesis that latency do not change for each subjects during five sessions, the low standard deviation is proof of robustness of this method. Higher inter-subjective standard deviation values implies that an F-wave latency period has to be calculated for each subject.

1. Introduction

F-waves are a recurrent discharge of an antidromically activated motor neuron. They follow the M wave and occur in 0-5% of the stimuli in each motor unit [1]. It has been demonstrated that F-waves registered after stimulation were preferentially generated by the fastest conducting axons [2].

These findings suggest that the F-waves may be elicited in motor-neurones with different depolarisation threshold, but primarily in larger and faster nerve fibres, with a lower threshold of depolarisation. [3].

The variability of the configuration and the low amplitude of F waves are the consequence of the infrequent backfiring and of the fact that the potential is composed by a few motor units only [4,5]. The myoelectrical signal from a stimulated muscle is thus strongly influenced by the F-waves in a manner that is difficult to distinguish from volitional contraction except that the F-waves will occur in a limited interval after the stimulation only.

This study regards electrical stimulation on the anterior tibial muscle (TA). In literature there has only been found one articles regarding the F-wave during FES of TA[6].

Many clinical studies on F-wave during FES have been done, but this is one of the first works on the F-wave registered during FES of TA looking for the application on a neuroprosthesis with a stimulation interval of 60ms.

2. Methods

A system for Myo-electrical Controlled Functional Electrical Stimulation (MeCFES), has been used which is capable of recording myoelectrical signals (MES) from the same muscle as it stimulates [7,8].

The signal is sampled at 2.5 kHz and divided into blocks of 60ms, such that block starts after each stimulation impulse. The first 10ms of each block is saturated by the M-wave and thus discarded. Each block of comprise 125 samples (50ms) of MES.

In the Figure 1 it is possible to see two of the registered signals.

3

2

1

Figure 1. The MES contains three components: 1) End of M-wave (first 10ms of the muscle contraction signal are not registered). This can be reduced by applying a comb filter to the signal. 2) Randomly occurring F-wave 3) Eventually volitional signal components (registered together with the stimulation, shown as dashed)

F-waves were present in all the muscles of the legs, with a percentage of 60 % in TA and, in this muscle, H-reflex was not found [Jušic et al., 1995].

Voluntary component shown is not usually present during tests where the subject is relaxing the TA.

It is difficult to reduce the F-wave by filtering, because it is similar in frequency to the volitional component and since they are generated by the same motor units.

The difference between them is that F-wave has a delay and could be characterised by its latency period.

The F-wave can be considered like a disturbance in the control of a myoelectrical controlled prosthesis, because it can be confused by the control system with a voluntary signal.

An experimental protocol with twenty trials of constant stimulation has been adopted.

Five male healthy subjects have been tested.

Characteristics of subjects:

· age = from 22 to 34 years

· height = from 178 to 198 cm

· weight = from 65 to 90 kg

Five sessions have been done on every subjects.

Subjects were asked to relax their leg during stimulation and they were helped by a visual feedback of the F-wave level.

The stimulation parameters were:

· current intensity = from 17 to 24 mA

· pulse rate = 16.6 Hz

· pulse shape = biphasic with interpulse interval

· pulse width = 300 µs

Among every trials of 30s there are pauses of 30s.

The electrodes configuration is represented in Figure 2.

Stimulation electrodes

MES electrodes

Reference electrode

Figure 2 Electrode placement on the TA

It has been individuated the F-wave latency period for each trial with algorithms implemented with MATLAB™.

It has been calculated the standard deviation (STD) of the entire trial of 500 blocks.

If the M-wave and noise should be constant for all blocks, it could be possible to see a not null STD curve only in the F-wave latency period.

Due to the fatigue and noise, it is not possible, but it could be said that a STD curve risen after 30ms [10] could be joined to the F-wave presence.

All minimum values after 30ms have been calculated.

Figure 3

Two of these minimum values, plotted in Figure 3 on the STD curve, identify the starting and the end of the F-wave latency period.

The program could make this choice.

It builds a new curve making a low-pass FIR filtering and a resampling of the STD curve previously calculated.

On this new curve are individuated only two minimum values that focalise the zone where the extremes of the F-wave interval fall.

Starting from them, the program can find on the first curve the two real extremes.

Figure 4 Solid curve is the filtered block.

In Figure 4 the solid curve is the filtered block. The two circles are the two minimum values found on it and the dot curve is the original STD curve, with its minimum values.

Sometimes the program falls in the calculation of some of the two minimum values represented by the circles, but it gives to the operator the percentage of falls and the number of the trial where the extremes have not been automatically calculated.

After the data processing, the operator can choose not calculated extremes. This choice could be classified “semi-automatic”, because the operator has to manually choose one or two of the minimum values on the STD curve, but the program automatically does the identification of minimum values.

3. Results and discussion

The data processing gives results reported in Table 1.

	F-wave starting				F-wave end
	Mean	STD	Min	Max	Mean	STD	Min	Max
S30	38.6	0.6	34.8	41.2	56.0	0.7	53.2	58.0
S51	35.2	0.8	32.8	37.2	52.2	1.4	48.8	57.2
S52	35.4	0.8	32.8	38.0	53.4	0.4	51.6	57.2
S53	35.0	0.5	33.2	39.6	54.9	0.9	52.4	58.0
S54	32.6	0.4	31.2	34.0	50.2	1.0	47.2	54.4

TOT	35.3	2.1	31.2	41.2	53.3	2.3	47.2	58.0

Table 1 Start intervals for the F-wave and the end interval where F is decayed. Mean, standard deviation and extreme values are given.

The first column indicates the number of subject and TOTal means values referring to all the subjects.

In the four columns are reported the Mean, STD, Minimum and Maximum values of all trials done on each subject and, in TOT row, on all the subjects.

The hypothesis is that latency values do not change for a subject during the five sessions. Effort has been put into maintaining electrode positions between sessions.

The STD values found for each subject are low indicating good robustness of the procedure F_Start and F_end.

STD values found for all the subjects are higher. In the idea to use latency values to control a neuroprosthesis, an F-wave latency period has to be found for each patient. A study on F-wave latency of TA thougn using another electrode configuration reports latencies from 34-50ms [10] and we conclude that our findings are in the right magnitude. The findings have verified by spot checking the recorded signal from the subjects.

4. Conclusion

We have developed an automatic method for determination of the signal interval where F-waves are present after each stimulation pulse. This is useful for automatic analysis of myoelectric signals with the scope of characterisation of the F-wave level.

Acknowledgments

This work has been sponsored by the EU-project NeuralPro with equipment and methods developed in University Twente, The Netherlands and University College London, England with support from the EU-project NEUROS.

The data processing has been done at Bioengineering Centre, Fond. Don C.Gnocchi & Politecnico di Milano as part of the graduation project of first author.

References

[1] E. Stålberg and B. Falck, “Clinical Motor Nerve Conduction Studies,” Methods in Clinical Neurophysiology, vol. 4, pp. 61-80, 1993.

[2] Maciel Nobrega J.A., Mastrocola Manzano G., Ferreira Novo N., Monteagudo P.T. “F waves and condution velocities range”. Electromyogr. Clin. Neurophysiol., 40: 327-329, 2000.

[3] J. J. Vatine and B. Gonen, “Behavior of F-response and determination of actively involved motoneurons,” Electromyogr Clin Neurophysiol, vol. 36, pp. 349-55, 1996.

[4] Feasby T.E. and Brown W.F. “Variation of motor unit size in the human extensor digitorum brevis and thenar muscles”. J. Neurol. Neurosurg. Psychiat., 37: 916-926, 1974.

[5] Yates S.K. and Brown W.F. “Charateristics of the F-response: a single motor unit study”. J. Neurol. Neurosurg. Psychiat., 42: 161-170, 1979

[6] P. Mela, R. Thorsen, H. J. Hermens, P. A. Huijing, and P. H. Veltink, “Volitional And Reflex Contributions To Force During Isometric Contractions Elicited By Electrical Stimulation,” Medical

& Biological Engineering and Computing., Submitted 2001.

[7] R. Thorsen, “Restoration of Hand Function in Tetraplegics Using Myoelectrically Controlled Functional Electrical Stimulation of the Controlling Muscle,” Ph.D Thesis, Electronics Institute, The Technical University of Denmark, Lyngby, 1997.

[8] Thorsen R., Spadone R., Ferrarin M. “A pilot study of myoelectrically controlled FES of upper extremity”. IEEE Trans. Neural. Syst. Rehabil. Eng. 9(2): 161-8, Jun 2001.

[9] A. Jusic, R. Baraba, and A. Bogunovic, “H-reflex and F-wave potentials in leg and arm muscles,” Electromyogr Clin Neurophysiol, vol. 35, pp. 471-8, 1995.

[10] M. R. Alavian Ghavanini, S. Samadzadeh, and A. Alavian Ghavanini, “Normal values of F wave in upper extremities of 50 healthy individuals in Iran,” Electromyogr Clin Neurophysiol, vol. 38, pp. 305-8, 1998.

An Automatic Procedure For Determination Of F-Wave Latency And End Intervals