Surface EMG Signal is Less Gaussian at Lower Contraction Levels

yakzephyrAI and Robotics

Nov 24, 2013 (3 years and 8 months ago)

129 views

Surface EMG Signal is Less Gaussian at Lower
Contraction Levels



Ali H. Al
-
Timemy
1
, Guido Bugmann
1
, Nick Outram
2,

Javier Escudero
2

and Kianoush Nazarpour
3

1
Centre for Robotics and Neural Systems, University of Plymouth, PL4 8AA, UK

2
Signal Processing and Multimedia Communication (SPMC), University of Plymouth, UK

3
Institute of Neuroscience, School of Medical Sciences, Newcastle University, UK

E
-
mail
:
ali.ali@plymouth.ac.uk

Introduction


-
Surface Electromyogram (sEMG)
signal represents the electrical
activity of closely spaced muscles.



-
EMGs represents a summation of
the electrical activity passing
through the muscle and reflects the
level of muscular activity.



-

EMG is used for prosthetic control,
assistive device control because of
its non invasiveness as well as ease
of use.


Background




Gaussian density function can model the EMG PDF [
1
].




But during low intensity isometric contractions, the PDF of the sEMG signal is more peaked
near zero than the Gaussian distribution (super
-
Gaussian) [
2
].



Negentropy

analysis of the EMGs showed that the non
-
Gaussianity

level of the sEMG signal
depends on the muscular contraction level [
3
].




Bicoherence analysis of the EMGs showed that EMGs are highly non
-
Gaussian at low and
high levels of force while being in the maximum
Gaussianity

at the mid
-
level of MVC
[
4
].




But
Hussian

et al. [
5
] used the bicoherence index and showed that the sEMG signal
becomes less Gaussian with increased walking speed force.




We revisited this problem and investigated the suitability of the EMG bicoherence
index to characterize the non
-
Gaussianity

level of sEMG.

Acknowledgments

-
This work is supported by the Iraqi Ministry of
higher Education and Scientific Research /
Cultural and scholarship Directorate Scholarship.


-

KN’s work is supported by
T
he Leverhulme
Trust and The Medical Research Council, UK.

References

[1]E.A. Clancy and N.
Hogan
, “
Probability

density of the
surface

electromyogram

and
its

relation
to

amplitude

detectors
,”
IEEE
Transactions

on

Biomedical

Engineering
,
vol. 46, (no.
6), pp. 730
-
739, 1999.

[2]I. Hunter, R.
Kearney
, and L. Jones, “
Estimation

of the
conduction

velocity

of
muscle

action

potentials

using

phase

and
impulse

response

function

techniques
,”
Medical

and
Biological

Engineering

and Computing
, vol. 25, (no. 2), pp. 121
-
126, 1987.

[3]K. Nazarpour, A.R.
Sharafat
, and
S.M.P.

Firoozabadi
, “
Application

of
Higher

Order

Statistics

to

Surface

Electromyogram

Signal

Classification
,”
IEEE
Transactions

on

Biomedical

Engineering
,

vol. 54, (no. 10), pp. 1762
-
1769, 2007.

[4]P.A.
Kaplanis
, C.S.
Pattichis
,
L.J.

Hadjileontiadis
, and S.M. Panas, “Bispectral analysis of surface EMG,” in Mediterranean Electrotechnical Conference MELECON, 2000.

[5]M.S.
Hussain
,
M.B.I.

Reaz
, F.
Mohd

Yasin
, and M.I.
Ibrahimy
, “
Electromyography

signal

analysis

using

wavelet

transform

and
higher

order

statistics

to

determine

muscle

contraction
,”
Expert
Systems
, vol. 26, (no. 1), pp. 35
-
48, 2009
.



Conclusions




Kurtosis analysis results
confirm that the EMG
PDF is less Gaussian at
low muscle contraction
levels.



This feature of the
sEMG signal has proved
effective in the control of
myoelectric prostheses
[
3
].

Method


1. Participants.


-
Four right
-
handed subjects (two females and two males)
participated in the study. The study was approved by the
local ethics committee at Newcastle University.



2. Pre
-
Processing


-

Participants controlled a myoelectric cursor by making
isometric contractions of right upper
-
limb muscles with the
arm strapped to the chair armrest and the hand restrained in
a glove which was fixed to the chair (Fig.1). The cursor and
targets were displayed on a computer screen positioned in
front of the subject.


We recorded surface EMG signals from
abductor
policis

brevis

(APB) and
flexor carpi
radialis

(FCR) muscles. EMG
was first amplified (gain 1K


10K) and high
-
pass filtered at
30 Hz (
Neurolog

NL824/820,
Digitimer
) before sampling at
10 KHz (PCI
-
6071E, National Instruments).


At the start of the experiment, subjects were informed
of the structure of the task (Fig. 2) Control signals
were computed by smoothing the rectified EMG to
determine the cursor position along a 1D vertical axis.


Visual feedback was provided using the computer
monitor.


3.

Data Analysis:

We investigated the
bispectrum

and kurtosis of
the EMG measurements to explore the features
of the EMG signals recorded in different force
level conditions.

Fig5.
Bispectrum

plot for FCR muscle of 4 subjects
with standard deviation


Results



-
In all four subjects and for both muscles,
kurtosis values monotonically decreased
when the contraction level increased
consistent with the predictions of the
Central Limit Theorem.


-
However, in contrast to [
4
] and [
5
], we did not
observe any modulation of
the bicoherence
index
with respect to muscle activity.



Fig4

Kurtosis plot of FCR muscle for 4 subjects with
standard deviation

0
10
20
30
40
50
60
70
80
90
100
110
3
3.5
4
4.5
5
5.5
6
6.5
7
7.5
8
Force level % of MVC
Kurtosis


Subject1
Subject2
Subject3
Subject4
0
10
20
30
40
50
60
70
80
90
100
110
-1000
-500
0
500
1000
1500
2000
2500
3000
3500
Force level % of MVC
Sg


Subject1
Subject2
Subject3
Subject4
Fig3
. The subject performing 50% MVC with the visual feedback



Move
(1 s)

Hold
(
3

s)

Relax

Start

50 % MVC

40 % MVC

30 % MVC

20 % MVC

10 % MVC


Fig
2 . The
myoelectric
-
controlled interface.

Fig. 1 Hand fixation for the experimental protocol


We recorded the EMGs in six blocks each subject for
each muscle in a range of the force levels

Institute of Neuroscience