ThesisPDF Available

The application of data analysis methods for surface electromyography in shot putting and sprinting

Authors:

Abstract and Figures

Muscles are the key drivers in any human movement. Since the muscles generate the forces and consequently the impulses to move the athlete from one position to another, it can be useful to study the muscle activity during sports movements to help with optimisation of technique, injury prevention and performance enhancements. Due to recent advances in electromyography (EMG) technologies, muscle activity in sports movement such as shot putting and overground sprinting can now be acquired using wireless surface mount sensors. Previously the use of tethered devices restricted the movements which could be analysed. The aim of this research was to investigate data analysis methods for use with EMG. There is a need to develop an in depth understanding of what EMG data can convey by understanding muscle activations and patterns in various sports movements and techniques. The research has been implemented by conducting a literature review, a survey and experimental studies to examine EMG signals on shot putting, sprinting and to understand cross-talk. There has been significant work done in understanding the biomechanics of sprinting, with emphasis on kinematics. The literature review on muscle activities in sprinting highlighted the need for wireless devices to allow testing of athletes in ecologically valid environments, rather than on a treadmill which offers little comparison with the environment of the sprinter, and proposed that there existed a bias on the muscles studied which may have been due to technology constraints of tethered systems. The survey of biomechanists gave an insight into the sensor devices utilised, the types of experimental studies being undertaken and the specifications desired in these devices. The study of muscle activations during the glide technique in shot put delivered meaningful activation patterns which coincided with key movements in the technique and augmented previously known kinematic data and anecdotal evidence. The study on muscle activations during maximal sprinting returned similar results, the 50% threshold provided information on the higher volume of muscle activity and these bursts of activity also coincided with key kinematic events. The use of independent component analysis (ICA) was examined to reduce cross-talk during sporting movements and recreating EMG signals due incorrectly positioned electrodes. Few studies have examined ICA with myoelectric signals. This research applies ICA to EMG signals during isometric contractions; small increases in correlation were found in some cases between the output signals and the ideal signals. The data analysis methods used in this research along with the supporting studies may prove to be a vital aid in supporting practitioners, coaches and athletes in the analysis of shot putting and sprinting using muscle activations and patterns. The thresholding methods used in this work may be useful in future studies to distinguish between low and high volumes of EMG activity in sports movements. It is recommended that future studies examine the muscle activity of specific exercises and compare the activity to that of sports movements to determine which exercises are most suitable in training and for pre-activation. The ICA algorithm should be examined further, to analyse isotonic movements.
Content may be subject to copyright.
The application of data analysis methods
for surface electromyography in shot
putting and sprinting
Róisín Marie Howard
A thesis submitted to the University of Limerick in fulfilment of the requirements of the
degree Doctor of Philosophy
Supervisors: Dr. Richard Conway & Prof. Andrew J. Harrison
Department of Electronics and Computer Engineering
Submitted to the University of Limerick November 2016
i
Abstract
Title: The application of data analysis methods for surface electromyography in a shot putting and
sprinting
Author: Róisín Marie Howard
Muscles are the key drivers in any human movement. Since the muscles generate the forces
and consequently the impulses to move the athlete from one position to another, it can be
useful to study the muscle activity during sports movements to help with optimisation of
technique, injury prevention and performance enhancements. Due to recent advances in
electromyography (EMG) technologies, muscle activity in sports movement such as shot
putting and overground sprinting can now be acquired using wireless surface mount sensors.
Previously the use of tethered devices restricted the movements which could be analysed.
The aim of this research was to investigate data analysis methods for use with EMG. There is
a need to develop an in depth understanding of what EMG data can convey by understanding
muscle activations and patterns in various sports movements and techniques.
The research has been implemented by conducting a literature review, a survey and
experimental studies to examine EMG signals on shot putting, sprinting and to understand
cross-talk. There has been significant work done in understanding the biomechanics of
sprinting, with emphasis on kinematics. The literature review on muscle activities in
sprinting highlighted the need for wireless devices to allow testing of athletes in ecologically
valid environments, rather than on a treadmill which offers little comparison with the
environment of the sprinter, and proposed that there existed a bias on the muscles studied
which may have been due to technology constraints of tethered systems. The survey of
biomechanists gave an insight into the sensor devices utilised, the types of experimental
studies being undertaken and the specifications desired in these devices. The study of muscle
activations during the glide technique in shot put delivered meaningful activation patterns
which coincided with key movements in the technique and augmented previously known
kinematic data and anecdotal evidence. The study on muscle activations during maximal
sprinting returned similar results, the 50% threshold provided information on the higher
volume of muscle activity and these bursts of activity also coincided with key kinematic
events. The use of independent component analysis (ICA) was examined to reduce cross-talk
during sporting movements and recreating EMG signals due incorrectly positioned
electrodes. Few studies have examined ICA with myoelectric signals. This research applies
ICA to EMG signals during isometric contractions; small increases in correlation were found
in some cases between the output signals and the ideal signals.
The data analysis methods used in this research along with the supporting studies may prove
to be a vital aid in supporting practitioners, coaches and athletes in the analysis of shot
putting and sprinting using muscle activations and patterns. The thresholding methods used
in this work may be useful in future studies to distinguish between low and high volumes of
EMG activity in sports movements. It is recommended that future studies examine the
muscle activity of specific exercises and compare the activity to that of sports movements to
determine which exercises are most suitable in training and for pre-activation. The ICA
algorithm should be examined further, to analyse isotonic movements.
ii
Declaration
I hereby declare that the work contained within this current thesis is my own and was
completed with the council of my supervisors Dr. Richard Conway of the Department of
Electronic and Computer Engineering, University of Limerick and Prof. Andrew Harrison of
the Department of Physical Education and Sports Sciences, University of Limerick. The
work has not been submitted to any other university or higher education institution or for any
other academic award within this university. To the best of my knowledge and belief, the
thesis contains no material previously published or written by another person except where
due reference is made.
_________________________
Róisín M. Howard
_________________________
Dr. Richard Conway
________________________
Prof. Andrew J. Harrison
Date: _________________________
iii
Acknowledgements
I would like to take this opportunity to express my thanks to the following people who all
contributed in some way to the completion of this thesis:
Firstly, to my primary supervisor Dr. Richard Conway and my secondary supervisor Prof.
Andrew Harrison, for their continued support and valuable advice over the last 3 years.
The Irish Research Council and Analog Devices Inc. who funded this Ph.D. program.
Without their financial backing this research would not have been possible. I would like to
especially thank Declan McDonagh, my enterprise partner supervisor, and John O’ Donnell,
both Analog Devices employees who have shown great interest and guidance over the course
of this research.
All the technicians and staff of both the ECE Department and the PESS Department, for
always being there to give a helping hand along the way, especially to Dr. Ian Kenny for his
valuable comments.
The PESS post graduates past and present for always being there to lend an ear, help out with
pilot studies and provide valuable advice.
The athletes who volunteered their time to take part in pilot and experimental studies.
Finally, my family and friends for always being there along the way, especially to my mother
for her expert knowledge throughout.
iv
Table of Contents
Abstract ....................................................................................................................................... i
Declaration ................................................................................................................................. ii
Acknowledgements .................................................................................................................. iii
List of Figures ........................................................................................................................ viii
List of Tables ............................................................................................................................. x
List of Appendices .................................................................................................................... xi
Nomenclature & Definitions .................................................................................................. xiii
Submissions and publications arising from and related to the thesis ...................................... xvi
Chapter 1: Introduction to the Thesis .................................................................................... 1
1.1 Introduction ................................................................................................................. 3
1.2 Thesis background....................................................................................................... 3
1.3 Thesis context .............................................................................................................. 4
1.4 Thesis purpose ............................................................................................................. 5
1.5 Thesis structure ........................................................................................................... 7
Chapter 2: Background .......................................................................................................... 9
2.1 Introduction ............................................................................................................... 11
2.2 Biomechanics ............................................................................................................ 11
2.2.1 Kinematics ......................................................................................................... 12
2.2.2 Kinetics .............................................................................................................. 13
2.2.3 Electromyography .............................................................................................. 13
2.2.4 Sports Biomechanics .......................................................................................... 13
2.3 Muscles...................................................................................................................... 14
v
2.3.1 Muscle Physiology ............................................................................................. 16
2.3.2 Muscle Activity .................................................................................................. 17
2.4 Wireless Sensor Devices & Electromyography ........................................................ 20
2.4.1 Wireless Sensor Devices .................................................................................... 20
2.4.2 Electromyography .............................................................................................. 21
2.5 Signal Processing ...................................................................................................... 29
2.5.1 Processing the EMG signal ................................................................................ 29
2.6 Cross-talk .................................................................................................................. 38
2.7 Locomotion ............................................................................................................... 40
2.7.1 Walking .............................................................................................................. 40
2.7.2 Running .............................................................................................................. 42
2.7.3 Sprinting ............................................................................................................. 43
2.8 Conclusion ................................................................................................................. 46
Chapter 3: Literature Review - Muscle Actions in Sprinting .............................................. 47
3.1 Introduction ............................................................................................................... 49
3.2 Methods ..................................................................................................................... 51
3.3 Results ....................................................................................................................... 54
3.3.1 Study design and sample .................................................................................... 54
3.3.2 Muscle activations and timings in sprinting ...................................................... 58
3.3.3 EMG systems and specifications ....................................................................... 63
3.4 Discussion ................................................................................................................. 69
3.5 Conclusion ................................................................................................................. 75
vi
Chapter 4: A Survey of Sensor Devices and their use in Sports Biomechanics .................. 77
4.1 Introduction ............................................................................................................... 79
4.2 Methods ..................................................................................................................... 81
4.2.1 Statistical Analysis ............................................................................................. 82
4.3 Results ....................................................................................................................... 83
4.3.1 Participant Specific Demographics .................................................................... 83
4.3.2 Nature and Use of Sensors ................................................................................. 87
4.3.3 EMG Specific Findings...................................................................................... 90
4.4 Discussion & Implications ........................................................................................ 93
4.4.1 Participant Specific Demographics .................................................................... 93
4.4.2 Nature and Use of Sensors ................................................................................. 94
4.4.3 EMG Specific Findings...................................................................................... 95
4.5 Conclusion ................................................................................................................. 98
Chapter 5: Muscle activation sequencing of leg muscles during linear glide shot putting . 99
5.1 Introduction ............................................................................................................. 101
5.2 Methods ................................................................................................................... 104
5.2.1 Participants ....................................................................................................... 104
5.2.2 Equipment ........................................................................................................ 105
5.2.3 Test Procedure ................................................................................................. 107
5.2.4 Data Analysis ................................................................................................... 108
5.2.5 Statistical Analysis ........................................................................................... 112
5.3 Results ..................................................................................................................... 113
vii
5.4 Discussion & Implications ...................................................................................... 125
5.5 Conclusion ............................................................................................................... 131
Chapter 6: Muscle activation sequencing of the lower limbs during maximal sprinting .. 133
6.1 Introduction ............................................................................................................. 135
6.2 Methods ................................................................................................................... 137
6.2.1 Participants ....................................................................................................... 137
6.2.2 Equipment ........................................................................................................ 137
6.2.3 Test Procedure ................................................................................................. 138
6.2.4 Data Analysis ................................................................................................... 139
6.3 Results ..................................................................................................................... 142
6.4 Discussion & Implications ...................................................................................... 148
6.5 Conclusion ............................................................................................................... 152
Chapter 7: Thesis conclusions and future directions ......................................................... 153
7.1 Introduction ............................................................................................................. 155
7.2 Key findings and implications of the thesis ............................................................ 156
7.3 Constraints of the thesis .......................................................................................... 158
7.4 Directions of future research ................................................................................... 160
7.5 Conclusions ............................................................................................................. 162
viii
List of Figures
Figure 2.1 The coordinate system for human movement analysis........................................... 12
Figure 2.2 The macro and micro structure of muscles ............................................................. 15
Figure 2.3 Skeletal musculature ............................................................................................... 17
Figure 2.4 The generation of the action potential (AP) ........................................................... 18
Figure 2.5 The motor unit action potential (MUAP) recorded by the surface EMG ............... 19
Figure 2.6 Bipolar Surface Electrode....................................................................................... 23
Figure 2.7 Timeline of sensor development ............................................................................ 24
Figure 2.8 Parallel Bar Technology ......................................................................................... 25
Figure 2.9 Indwelling electrodes.............................................................................................. 26
Figure 2.10 An example of sampling ....................................................................................... 27
Figure 2.11 An example of a sine wave sampled by a 4-bit analog-to-digital converter ........ 27
Figure 2.12 An example of the input range of the ADC and the amplifier gain set too low ... 28
Figure 2.13 Sample Amplitude Analysis of EMG signal ........................................................ 30
Figure 2.14 Low Pass Filter ..................................................................................................... 32
Figure 2.15 High Pass Filter .................................................................................................... 33
Figure 2.16 An example of Fourier decomposition of a motor unit action potential (MUAP)36
Figure 2.17 Power Spectrum of the EMG signal ..................................................................... 37
Figure 2.18 SEMG sensors can pick up cross-talk from adjacent muscles ............................. 39
Figure 2.19 The walking gait cycle.......................................................................................... 42
Figure 2.20 The phases of the running gait cycle .................................................................... 43
Figure 2.21 Muscle activity across the running gait cycle....................................................... 45
Figure 3.1 The Selection Criteria Flow Chart .......................................................................... 53
Figure 3.2 Muscle activity of the lower limbs during sprinting .............................................. 62
Figure 4.1 Years in Biomechanics ........................................................................................... 83
ix
Figure 4.2: Locations of researchers ........................................................................................ 84
Figure 4.3 Human Movement Measures .................................................................................. 86
Figure 4.4: Frequency of use of sensor devices ....................................................................... 87
Figure 4.5: Sensors included in a single device ....................................................................... 89
Figure 4.6: EMG device specifications .................................................................................... 90
Figure 4.7: Software Tools ...................................................................................................... 91
Figure 4.8: EMG Systems ........................................................................................................ 92
Figure 5.1. Shot Put Test Set-Up ........................................................................................... 106
Figure 5.2. Wireless EMG electrode placement on the lower limbs ..................................... 107
Figure 5.3. Sequence of events and phases of the shot put .................................................... 110
Figure 5.4. Sample EMG signal with threshold levels .......................................................... 113
Figure 5.5. An EMG profile of the eight lower limb muscles of the male throwers during the
glide technique represented across the percentage cycle time. .............................................. 115
Figure 5.6 An EMG profile of the eight lower limb muscles of the female throwers during the
glide technique represented across the percentage cycle time. .............................................. 119
Figure 5.7 Linear Envelope Mean Curves for the eight lower limb muscles of the preferred
and non-preferred legs of the male throwers. ........................................................................ 123
Figure 5.8 Linear Envelope Mean Curves for the eight lower limb muscles of the preferred
and non-preferred legs of the female throwers. ..................................................................... 124
Figure 6.1 Static Test using 3D motion analysis ................................................................... 138
Figure 6.2 Test set up ............................................................................................................. 139
Figure 6.3 An example of the initial contact (IC) and toe-off (TO) of the left leg using
Visual3D, a 3D motion analysis software package ............................................................... 140
Figure 6.4 The EMG profile across the sprinting gait cycle .................................................. 145
Figure 6.5 The ensemble averages of each muscle analysed across the sprinting gait cycle 147
x
List of Tables
Table 3.1. Participant information from the selected 18 review papers .................................. 55
Table 3.2 Study information from the selected 18 review papers ........................................... 56
Table 3.3 Muscles Analysed using Electromyography during Sprinting ................................ 59
Table 3.4 EMG Electrode Types ............................................................................................. 63
Table 3.5 EMG Electrode Specifications................................................................................. 64
Table 3.6 EMG Systems .......................................................................................................... 65
Table 3.7 Specifications of EMG Systems .............................................................................. 67
Table 3.8 Data Analysis Steps: Filtering ................................................................................. 68
Table 3.9 Data Analysis: Amplitude Analysis ......................................................................... 69
Table 4.1 Cross tabulation of Gender, Geographical Region and Years’ Experience ............. 85
Table 4.2 Researchers’ Expertise ............................................................................................. 85
Table 4.3 Human Movement Measures ................................................................................... 86
Table 4.4 Frequency of use of sensor devices ......................................................................... 88
Table 4.5 EMG device specifications ...................................................................................... 91
Table 4.6 Software Tools ......................................................................................................... 92
Table 4.7 EMG Systems .......................................................................................................... 93
Table 5.1 The timings of key events, total throw time and glide performance data .............. 114
Table 5.2 The onset & termination times for the male throwers. .......................................... 116
Table 5.3 The onset & termination times for the female throwers ........................................ 120
Table 6.1 Stride Timing Data................................................................................................. 142
Table 6.2. The minimum threshold onset and termination times .......................................... 143
Table 6.3. The 50% maximum threshold onset and termination times ................................. 146
Table 6.4 The difference between onset and termination times ............................................ 148
xi
List of Appendices
Appendix A Ethics Information .............................................................................................. A1
A.1 Questionnaire Chapter 4 ........................................................................................ A1
A.1.1 Cover Letter ...................................................................................................... A1
A.2 EMG studies Chapters 5 - 7 .................................................................................. A3
A.2.1 Participant Consent Form ................................................................................. A3
A.2.2 Participant Information Sheet ........................................................................... A4
Appendix B Supplimentary Data ............................................................................................ B9
B.1 The Questionnaire Chapter 4................................................................................. B9
B.2 Shot Put Chapter 5............................................................................................... B19
Appendix C Thresholding Matlab Code ............................................................................... C20
C.1 Chapter 5 & Chapter 6 ........................................................................................... C20
C.1.1 Theory behind the calculation of the onset and termination timings .............. C20
C.1.2 Matlab function: emgonoff_RH ...................................................................... C21
C.1.3 Matlab function: OnOffTimePts ..................................................................... C23
Appendix D The exploration of cross-talk using Independent Component Analysis .......... D24
D.1 Introduction ............................................................................................................ D24
D.2 Background ............................................................................................................ D24
D.3 Theory of ICA ........................................................................................................ D26
D.4 Suitability of sEMG signals ................................................................................... D29
D.5 Limitations of ICA ................................................................................................. D30
D.6 Applications of ICA ............................................................................................... D30
xii
D.7 Testing the FastICA algorithm using predefined signals ....................................... D36
D.7 Methods .................................................................................................................. D38
D.7.1 Pilot test protocol ............................................................................................ D38
D.7.2 Final test protocol ........................................................................................... D39
D.7.3 Hardware ......................................................................................................... D40
D.7.4 Data analysis ................................................................................................... D41
D.7.2 Statistical analysis ........................................................................................... D47
D.8 Results .................................................................................................................... D47
D.9 Discussion .............................................................................................................. D48
D.10 Conclusion .............................................................................................................. D50
D.11 Matlab code ............................................................................................................ D51
D.7.1 Matlab function: CalvesCrossTalk ................................................................. D51
D.7.2 Matlab function: cross_corr_RH..................................................................... D58
D.12 Additional references ............................................................................................. D59
xiii
Nomenclature & Definitions
Abbreviations used for terminology throughout the thesis
2D Two Dimensional
EEG Electroencephalogram
3D Three Dimensional
EMEA Europe, Middle East & Africa
ADC Analog-Digital Convertor
EMG Electromyography
AEMG Average EMG
FP Force Platform
AMER Americas
FT Flight Time
APAC Asia Pacific
GA Gastrocnemius
BSS Blind Source Separation
GMAX Gluteus Maximus
BF Biceps Femoris
GMED Gluteus Medialis
CMJ Countermovement Jump
GPS Global Positioning Unit
CMR Common Mode Rejection
GRF Ground Reaction Force
CMRR- Common Mode Rejection Ratio
HS High Speed
CNS - Central Nervous System
HJ Height Jumped
CoM Centre of Mass
IAAF International Athletics Association
Federation
CT Contact Time
ICA Independent Component Analysis
DJ Drop Jump
iEMG Integrated EMG
DSP Digital Signal Processing
IMU Inertial Measurement Units
DWT Discrete Wavelet Transform
ISBS International Society of
Biomechanics in Sport
ECG Electrocardiogram
ISEK International Society of
Electromyography and Kinesiology
LG Lateral Gastrocnemius
MEG Magnetoencephalography
xiv
MG Medial Gastrocnemius
SENIAM Surface Electromyography for
the non-invasive assessment of muscles
MU Motor Unit
SM Semimembranosus
MUAP Motor Unit Action Potential
SNR Signal to noise ratio
MUAPT Motor Unit Action Potential Train
SOL Soleus
MVC Maximum Voluntary Contraction
SPSS Statistical Package for the Social
Sciences
PB Personal Best
ST Semitendinosus
PCA Principal Component Analysis
TA Tibialis Anterior
RF Rectus Femoris
VL Vastus Lateralis
sEMG Surface EMG
VM Vastus Medialis
Symbols used to represent variables in equations throughout the thesis
s Speech signals
x Mixture of signals, a weighted sum of speech signals
A Matrix,  are parameters that depend on the distance of the receiver from the speaker
()
Symbols used to abbreviate statistical terminology throughout the thesis
d Cohen’s D Effect Size
RMS Root mean square
p Probability
SD Standard deviation
r Correlation Coefficient
χ2 Chi Square
xv
Definitions of key terms used throughout the thesis
Braking Phase The early stance or absorption phase in running gait, when the leg in
question becomes in contact with the ground and muscles are working in a stabilisation role
until mid-stance.
Contact time The time at which the foot remains in contact with the ground during a sprint
for example.
Indwelling electrodes These electrodes are inserted into the muscle under observation.
Nyquist rate The minimum sampling rate required to recover the signal from its samples
without introducing errors / aliasing.
Power Clean An Olympic weight lifting exercise; the bar is quickly pulled from the floor to
the front of the shoulders in one movement.
Pre-activation Phase The late swing phase in running gait, the muscles tense prior to
ground contact, pre-activation to brace for impact.
Propulsion Phase The late stance phase in running gait, when the leg in question is no
longer in the braking phase, the leg is moving from the mid-stance position and the muscles
are now working in a driving / propulsive role to drive the body forward until the toe leaves
the ground.
Recovery Phase The early swing phase in running gait, the toe of the leg in question has left
the ground.
Sampling Rate The number times an analog signal is measured per second in order to be
converted into digital form, measured in Hz.
Spectral Analysis A spectrum of frequencies of the signal is analysed.
xvi
Submissions and publications arising from and
related to the thesis
Peer-reviewed journal articles
Howard, R. M., Conway, R. & Harrison A. J. (2016) A survey of sensor devices: Use in
sports biomechanics. Journal of Sports Biomechanics, 15, 450461, doi:
10.1080/14763141.2016.1174289
Howard, R. M., Conway, R. & Harrison A. J. (2016) Muscle activities in sprinting: A review.
Journal of Sports Biomechanics. (In Press)
Howard, R. M., Conway, R. & Harrison A. J. (2016) Muscle activation sequencing of leg
muscles during linear glide shot putting. Journal of Sports Biomechanics. (In Press)
Journal articles in review
Howard, R. M., Conway, R. & Harrison A. J. (20XX) Exploring the Recreation of Signals
due to Misplaced Electrodes in Surface Electromyography using Independent Component
Analysis. Journal of Biomedical Signal Processing.
Peer-reviewed magazine articles
Howard, R. M. (2016) Wireless Sensor Devices in Sports Performance. IEEE Potentials, 35,
40-42, doi: 10.1109/MPOT.2015.2501679
xvii
Peer-reviewed international conference proceedings (4 page paper)
Howard, R. M., Conway, R., & Harrison, A. J. (2015). An EMG Profile of Lower Limb
Muscles during Shot Putting. Paper presented at the 33rd International Conference on
Biomechanics in Sports, Poitiers, France, 29 June 3 July 2015.
Howard, R., Healy, R., Conway, R., & Harrison, A. J. (2014). A Method Comparison of
Force Platform and Accelerometer Measures in Jumping. Paper presented at the 32nd
International Society of Biomechanics in Sport, East Tennessee State University, Johnson
City, TN, USA, 12 16 July 2014.
Healy, R., Howard, R., Kenny, I., & Harrison, A. J. (2014). A Comparison of Methods to
Examine Double and Single Leg Drop Jump Performance. Paper presented at the 32nd
International Society of Biomechanics in Sport, East Tennessee State University, Johnson
City, TN, USA, 12 16 July 2014.
Peer-reviewed international conference proceedings (1 page paper)
Howard, R. M., Conway, R., & Harrison, A. J. (2015). The use of Independent Component
Analysis on EMG Data to Explore Cross-Talk. Paper presented at the 25th Congress of the
International Society of Biomechanics, Glasgow, UK, 12 16 July 2015.
Peer-reviewed national conference proceedings (6 page paper)
Howard, R. M., Conway, R., & Harrison, A. J. (2015). An exploration of eliminating cross-
talk in surface electromyography using independent component analysis. Paper presented at
the 26th Irish Signals and Systems Conference, Carlow IT, Carlow, Ireland, 24 25 June
2015.
xviii
Howard, R., Conway, R., & Harrison, A. J. (2014). Estimation of force during vertical jumps
using body fixed accelerometers. Paper presented at the 25th IET Irish Signals & Systems
Conference, University of Limerick, Limerick, Ireland, 26 27 June 2014.
Peer-reviewed national conference proceedings (1 page paper)
Howard, R. M., Conway, R., & Harrison, A. J. (2016). The use of ICA to combat misplaced
sensors in surface EMF during a simple squat exercise: a pilot study. Paper presented at the
22nd Bioengineering in Ireland Conference, NUIG, Galway, Ireland, 22 23 January 2016.
Howard, R. M. Conway, R., & Harrison A. J. (2015). The estimation of force during vertical
jumps using accelerometers. Paper presented at the All-Ireland Postgraduate Conference in
Sports Science and Physical Education, University of Limerick, Limerick, Ireland, 23
January 2015.
1
Chapter 1: Introduction to the Thesis
He who is not courageous enough to take risks will accomplish nothing in life
Muhammad Ali
2
3
1.1 Introduction
This research outlines the application of sensor devices in sport, the signal processing and
analysis methods which were used to enhance the data gathered using these devices and an
outline of further work which can be done. The first three sections of this chapter outline the
background and context of the research, and its purposes. The last section includes an outline
of the remaining chapters of the thesis.
1.2 Thesis background
Devices which are used to measure the electrical impulses of the muscles during a
contraction, Electromyography (EMG) systems, were in the past tethered devices. Advances
with technology allowed these devices to become telemetric. However, there were still wires
attached between the sensors and the transceiver pack which was attached to the participant,
all of which restricted the types of movement and had extra interference from wires. In
recent years, wireless technologies have been developed for many sensor devices; EMG is
one such sensor device. With the advances in technology, acquiring the muscle activations
from athletes in an ecologically valid environment was possible. There was an opportunity to
research the muscle activations in track and field athletes, with an emphasis on sprinting and
exploring ways to improve the results using signal processing techniques which were tried
and tested in other disciplines such as image and audio processing.
A profile of the muscle activations in terms of timing and sequencing across an event was
missing in some cases; this was mainly due to technology constraints. This was now possible
with the wireless EMG devices. Information of this kind would be very beneficial to both the
athlete and the coach. It would help with understanding the movement in terms of what the
muscles are doing throughout, and how the impulses generated in the muscles cause all of the
movements. The types of EMG sensor devices that are generally used for these high velocity
4
movements in track and field athletics are surface mount sensors which are less invasive and
less cumbersome for participants. However, the disadvantages to these sensors are that they
are limited to superficial muscles and are susceptible to cross-talk from adjacent muscles.
There are recommendations for placements in each muscle which were designed to limit
cross-talk and achieve the cleanest signals possible, but it is known that there will still be
cross-talk with these ideal placements. By using signal processing techniques from other
disciplines it may be possible to reduce the cross-talk from the acquired signal and possibly
even recreate ideal signals which were gathered incorrectly due to misplaced electrodes.
1.3 Thesis context
In sport many sensor devices are utilised to characterise movement, improve performance
and identify injury prevention techniques. As sport is very versatile and consists of rapid
complex movements, sensor devices need to be small and cause little encumbrances. This
research focuses on movements in track and field athletics with an emphasis on sprinting.
Previously for the analysis of muscle activations in sprinting, data was acquired using
tethered devices while the athlete ran on a treadmill. Treadmills have little ecologically
validity compared to that of overground sprinting. Participants require non-invasive wireless
sensor devices so their movement can be performed with ease in an ecologically valid
environment, without wires and invasive electrodes impeding the movements. Practitioners
also see the benefit to such sensors, it allows them to be used in the participants’ ecologically
valid environment, therefore signals are more representative of the movements and set-up is
quicker. However with these advantages there are still limitations to devices. Cross-talk is a
common problem in surface mount electrodes. There is a need to deepen the knowledge of
EMG signals in terms of muscle activations and make use of signal processing algorithms to
separate the signals into their individual muscle contributions. A deeper understanding into
the EMG signal across a range of movements and contraction types will help with the
5
analysis of muscle activity. Finding the correct way to separate signals without losing
information will aid the practitioner and give them a greater insight into the workings of
muscles throughout human movement.
1.4 Thesis purpose
The purpose of this research is to allow for a greater understanding of muscle activity in shot
put and sprinting by utilising successful algorithms and data processing techniques from a
range of disciplines. The aim of this research is to provide a novel contribution to the area of
sports biomechanics in terms of EMG analysis and signal processing as follows:
1. Gathering information on the various muscles analysed during sprinting and
understanding the important muscles for sprinting in terms of sequencing and timings
of muscle activations.
2. Gathering information on the various EMG technologies and their key features used
for sEMG during sprinting.
3. Gather information from a significant population of sports biomechanists on their use
of sensor devices across a sporting application and on the features required of these
devices.
4. Creating a profile of timings and sequencing of muscle activations of the legs in shot
put during the glide technique to provide representative data on this event which was
not previously available.
5. Creating a profile of timings and sequencing of muscle activations using wireless
sEMG in overground sprinting to provide representative data on sprinting which was
previously available using tethered devices.
6
6. Comparing and contrasting the data gathered using sEMG in overground sprinting to
the data collated in the literature review.
The thesis objectives were set out to meet the aims as follows:
1. Complete an in depth review of the literature review on articles utilising sEMG
during sprinting to examine the various muscles analysed during sprinting to
highlight the important muscles for sprinting in terms of sequencing and timings
of muscle activations, and to also understand the various technologies used and
their key features.
2. Create and conduct a survey of biomechanists on their use of sensor devices to
gather information on the best EMG systems and the specifications desired by
researchers and to guide the selection of the EMG system used in experimental
studies.
3. Gather a target population of athletes, who are at least national level in their
chosen discipline, who will volunteer as participants in the all the experimental
studies undertaken.
4. Create testing plans for the chosen disciplines in which participants will perform a
standardised warm up and specific movements related to the event before the
maximal trial of the event is tested.
5. Develop and test Matlab scripts and functions for offline analysis of all EMG data
gathered in experimental studies.
6. Compare the muscle activation timings between data acquired from wireless
sEMG devices and tethered sEMG device.
7
1.5 Thesis structure
This research contains a series of progressively linked studies and is structured as follows:
Chapter 2 contains the background knowledge on biomechanics, muscle physiology,
sprinting, EMG and other sensor devices, and signal processing. All of which was
necessary knowledge to complete this research.
Chapter 3 contains a comprehensive review of the literature on surface EMG in
sprinting. Information on the muscles analysed, the timings and the activity level of
the muscles across the phases of the running gait cycle and the technologies used are
discussed. Due to technology constraints and tethered devices, previously most
sprints were performed on treadmills which provided little ecologically validity; there
is the option to now analyse overground sprinting using new wireless technologies.
Chapter 4 examines the results from the conducted survey of biomechanists, focusing
on current use of electromyography (EMG) device preferences. This chapter
highlights the main findings of the survey that there is a need for a simple, low power,
multi-channel device which incorporates the various sensors into one single device.
It also emphasises the need to develop software analysis tools to accompany the
multi-channel device; providing all the basic functions while maintaining
compatibility with existing systems.
Chapter 5 provides an analysis on the muscle activation sequencing of leg muscles
during linear glide shot put technique utilising an EMG device identified in Chapter 4.
There is an absence of previous research on the muscle activations of the legs during
shot put. Since the muscles generate the impulse to move the athlete and project the
shot into the air, there is a need for information on phasic muscle activity of the event
to provide initial representative data and provide important technical information for
8
coaches. A comprehensive understanding of movement and muscle activation
patterns for coaches is essential to facilitate optimal technique throughout each of the
key phases of the event.
Chapter 6 provides an analysis on the muscle activation sequencing of leg muscles
during overground sprinting. The research design, implementation and data analysis
from Chapter 5 are applied to similar data gathered during a maximal sprint to
evaluate the use of the thresholding algorithm on different movements. A comparison
between the muscle activations of previous research from Chapter 3 and the data in
this chapter using wireless sensors was conducted.
Chapter 7 contains the discussion, implications and conclusions of the research. The
key findings of the research are summarised, the practical implications and the
constraints of the research are identified. The future directions for research in this
area are also identified.
9
Chapter 2: Background
Biomechanics is the study of structure and function of biological systems by means of the
methods of mechanics
Hatze (1974)
10
11
2.1 Introduction
This chapter provides a review of the fundamental material necessary for this research. The
purpose of this chapter was to provide an overview of biomechanics, including topics such as
kinematics, kinetics, kinesiology and electromyography (EMG). The aim of this chapter was
to provide detailed background on the musculoskeletal system, sensor devices and EMG
acquisition and analysis methods which were used in the ensuing chapters.
2.2 Biomechanics
A simple definition for biomechanics is the study of the effects of forces on living systems
(McGinnis, 2013a). The study of biomechanics of humans covers a broad range of topics.
Biomechanics spans from the inner workings of a cell, the mechanical properties of soft and
hard tissues, to the development and movement of the neuromusculoskeletal system of the
body. Mechanical factors affect the form, performance and function of the musculoskeletal
system. Human movement is a complex and highly coordinated mechanical interaction
between bones, muscles, ligaments and joints within the musculoskeletal system under the
control of the nervous system. Deviation from normal movement in motion analysis in terms
of kinematic, kinetic or electromyography (EMG) patterns can be identified and used to
evaluate neuromusculoskeletal conditions. The performer’s technique in a sporting setting is
a combination of the internal and external forces acting on the human body. These forces
determine how the various parts of the body move while the activity is being performed.
Biomechanists seek to improve sporting performances by improvements in technique or
developing new techniques. (Ethier & Simmons, 2007a; Hay, 1993a; Kamen, 2014a;
McGinnis, 2013a; Winter, 2009)
12
2.2.1 Kinematics
Figure 2.1 The coordinate system for human movement analysis
The frontal plane divides the body into anterior (front) and posterior (rear) halves. The sagittal
plane divides the body into left and right halves. The transverse plan divides the body into
superior (top) to inferior (bottom) halves. The direction in which the body is progressing
(principal horizontal direction of motion) is the X-axis, the vertical direction (orthogonal to the X-
axis) is the Y-axis and the sideways direction (right perpendicular to the X-Y plane) is the Z-axis.
Kinematics examines the motion of bodies, it encompasses speed and time. It is divided into
two parts: linear kinematics to deal with linear motion and angular kinematics to deal with
angular motion. Kinematics is concerned with details on the movement itself, to ascertain
how fast a body is moving, what distance it is moving and if it is moving at a constant speed
or is it accelerating or decelerating. Displacement (distance), velocity and acceleration are
the main parameters studied in kinematics. It can be studied in two-dimensions (2D) or
three-dimensions (3D). The human body has a special coordinate system; it is divided into 3
planes which help describe the movement relative to the ground or the direction of gravity
(see Figure 2.1). The main questions asked in the kinematics of sporting movements are:
What athlete is moving faster? What are the various ranges of motion of the joints during this
13
sporting movement? And, how do the motion patterns of these athletes differ? (R. Bartlett,
2014b; Hay, 1993a; Robertson & Caldwell, 2014; Winter, 2009)
2.2.2 Kinetics
Kinetics is the branch of biomechanics which is concerned with causes of motion or tendency
to move or change state of motion. It is primarily the recording of forces that affect motion.
Internal forces are those which are internal to the system such as the movement of one limb
causing the movement of another, and external forces are those which are external to the
system such as the ground reaction force from a runner’s heel strikes. (Caldwell, Robertson,
& Whittlesey, 2014; Hay, 1993a; Winter, 2009)
2.2.3 Electromyography
Electromyography (EMG) is a technique that is used to record and evaluate electrical activity
which is produced within the skeletal muscles. An electromyograph is used to record
changes in the electrical potential which is generated by the activated muscle cells. The
EMG signals recorded are useful in detecting medical abnormalities, the activation level of a
muscle or to analyse the biomechanics of movement in humans or animals. In the area of
sports biomechanics the EMG signal is used to analyse athletic performance and to reduce the
likelihood of sports injuries. More information on EMG can be found in Section 2.4.2.
(Basmajian & De Luca, 1985; Criswell, 2011; Kamen, 2014b; Kamen & Gabriel, 2010a;
McGinnis, 2013b; Weiss, Silver, & Weiss, 2004)
2.2.4 Sports Biomechanics
Sports biomechanists are movement analysts who study and analyse human movement
patterns in sport (R. Bartlett, 2014b). Potential benefits within sport as a result of
biomechanics include performance, technique, equipment, training, injury prevention and
rehabilitation (McGinnis, 2013a). Ultimately the goal of sports biomechanics is to improve
14
performance during a particular sporting task. Injury prevention and rehabilitation are
secondary goals; however there is a very close link between injury prevention and
performance. If biomechanists can analyse human movement patterns and know what the
optimum movement pattern is for the particular sport, then any deviations from the norm will
be a sign of an impending injury. By noticing that an injury is likely to occur, training can be
modified for the athlete, thus preventing the injury from occurring and allowing more
technique and performance related training to take precedence. (McGinnis, 2013a)
2.3 Muscles
Muscles are required for movement of the body, such as locomotion walking, running or
sprinting, the muscular action is caused by the generation of force by the muscle tissue
(Ethier & Simmons, 2007b). In the average person around 40 50% of gross body weight is
skeletal muscle (R. Bartlett, 2014b; Ethier & Simmons, 2007b). The development of tension
within the muscle is known as a contraction. There are three types of contractions muscles
can undergo while exerting forces (R. Bartlett, 2014a; Zatsiorsky & Prilutsky, 2012):
Isometric (static) contraction: The muscle maintains a constant length as tension
develops
Isotonic (dynamic) contraction: The muscle length changes as it develops tension
o Concentric (miometric) contraction: The muscle develops tension with a
shortening of the muscle length
o Eccentric (plyometric) contraction: The muscle lengthens as tension develops
Isokinetic contraction: The muscle develops tension at a constant speed while the
muscle lengthens or shortens
15
Agonist muscles are the prime movers; they undergo a concentric contraction and cause
movement. Antagonist muscles are opposing muscles, they relax when the agonists contract.
Stabilisers are the muscles which undergo an isometric contraction; they fix one bone against
the pull of the agonists to allow the bone at the other end to move freely. When agonists have
more than one function the fixators prevent undesired actions of the agonists. (R. Bartlett,
2014a).
Figure 2.2 The macro and micro structure of muscles
The composition of muscle cells, muscle fascicles, muscle fibre, myofibril, myofilaments,
sarcomere, thick and thin filaments. Source: Netter Anatomy Illustration Collection, © Elsevier,
Inc. All Rights Reserved.
16
2.3.1 Muscle Physiology
Tendons are tough, flexible bands of fibrous connective tissue which connect muscles to
bones. Muscles contain compartments of muscle fibres. Muscle fibres are 10 to 100 µm in
diameter and up to 0.3 m in length. Perimysium is the name for the connective tissue that
surrounds the bundles of muscle fibres. These bundles of muscle fibres are called muscle
fascicles. The fibrous elastic tissue surrounding a muscle is called the epimysium. Each
individual muscle fibre is enclosed by a layer of connective tissue called the endomysium.
Muscle fibres can be broken down further into clusters of individual myofibrils. The muscle
fibres are filled with a thick solution known as the sarcoplasm and enclosed in sarcolemma.
Nuclei are located peripherally, under the sarcolemma and satellite cells are precursors to
skeletal muscle cells found in mature muscles. The basement membrane is a layer of
extracellular matrix material which coats muscle fibres and other cells; they attach layers of
tissue in the body. The myofibrils are strands which are braided in light and dark bands; they
consist of overlapping myosin and actin protein filaments from one Z line to the next Z line
called sarcomeres, see Figure 2.2. The actin filament is a thin filament, and the myosin
filament is the thick filament. Both filaments are negatively charged. The I bands contain
only actin, the H bands contain only myosin and the A bands is where the myosin and actin
fibres overlap. The cross-bridges are formed between actin and myosin filaments, this is
responsible for the contraction of a muscle. The actin and myosin filaments are the
contractile proteins. The muscle fibres shorten during a concentric contraction when the actin
and myosin filaments slide over each other. When the contraction is finished the filaments
return to their original positions. The number of cross-bridges formed relates to the tension
of the contraction and the contraction of a whole muscle is the sum of the singular contraction
events which occur within the sarcomeres. (R. Bartlett, 2014b; Criswell, 2011; Ethier &
17
Simmons, 2007b; Kamen, 2014b; Kamen & Gabriel, 2010a; McGinnis, 2013b; Weiss et al.,
2004; Winter, 2009; Zatsiorsky & Prilutsky, 2012)
There are more than 430 skeletal muscles in the human body (Triplett, 2016). Figure 2.3
shows the frontal and rear view of the musculoskeletal system. This research examines the
lower limb muscles.
Figure 2.3 Skeletal musculature
The frontal view (a) and rear view (b) of the musculoskeletal system (Triplett, 2016, p. 4).
2.3.2 Muscle Activity
Muscle tissue membranes are excitable. Electrical impulses can be conducted through the
membranes of muscles. The resting voltage gradient across the muscle fibre membrane is
about -90 mV, the concentration of Na+ is high outside the membrane and the concentration
18
of K+ is low outside the membrane, inside the fibre the Na+ is low and the K+ is high. The
resting membrane potential varies from person to person; it depends on factors such as the
slow twitch and fast twitch fibres and exercise training (Moss et al. 1983). An impulse from
a motor neuron is received by the muscle fibres to produce a muscular force. The Central
Nervous System (CNS) activates the motor neuron and an electrical impulse propagates down
the motor neuron to each motor endplate. The endplate region is the anatomical structure
where the nerve interfaces with the muscle (synapse). The electrical impulse reaches the
endplate region where an ionic event occurs and a muscle fibre action potential (AP) is
generated. Every segment of the muscle fibre is activated by this neural message. The AP
occurs in two phases, the NA+ permeability of the muscle fibre increases and there is an
influx of Na+ into the cell and the inside of the muscle fibre becomes positive, the second
phase occurs when there is an outflow of K+ from the muscle fibre which restores the resting
membrane potential, see Figure 2.4. (Kamen, 2014b; Kamen & Gabriel, 2010a)
Figure 2.4 The generation of the action potential (AP)
The changes in membrane permeability to NA+ and K+ ions describes the generation of the action
potential (Kamen & Gabriel, 2010a, p. 6).
19
The Motor Unit (MU) is responsible for the recruitment of the muscle through the nervous
system. A MU is one motor neuron and all the muscle fibres innervated by that motor neuron.
The summed electrical activity of all muscle fibres activated within the MU is the Motor Unit
Action Potential (MUAP). See Figure 2.5, the MUs are denoted as αA and αB, the AP from
muscle fibres 1 5 can also be seen,  is the sum of the APs from the muscle fibres
activated by αA (1, 4, 5) and  is the sum of the APs from the muscle fibres
activated by αB (2, 3). Recruitment is the orderly addition of MUs to increase the force of a
contraction. In a contraction the first motor units to fire tend to be the smallest, known as
Type I motor units. The motor units are recruited by increasing size, as the contraction
increases there is an orderly recruitment of the larger motor units. More force is added to the
contraction as a result of the larger motor units beginning to fire. (Basmajian & De Luca,
1985; Kamen, 2014b; Kamen & Gabriel, 2010a; McGinnis, 2013b; Moritani, Stegemen, &
Merletti, 2004; Weiss et al., 2004; Winter, 2009)
Figure 2.5 The motor unit action potential (MUAP) recorded by the surface EMG
The MUAP is composed of the sum of all APs (Kamen & Gabriel, 2010a, p. 11)
The strength of a contraction can be increased in two ways. If the MUs which are firing
increase their rate of firing the force of a contraction will increase. The second way to
increase the force of a contraction is if additional MUs commence firing. A sequence of
20
stimuli is sent from the CNS to maintain a muscle contraction, the MUs are repeatedly
activated and a train of MUAP is produced. Myopathic Recruitment is when there is an
increased number and early recruitment of motor unit action potentials for the strength of the
contraction. (Kamen, 2014b; Kamen & Gabriel, 2010a; Moritani et al., 2004; Weiss et al.,
2004)
2.4 Wireless Sensor Devices & Electromyography
Accelerometers, gyroscopes and inertial measurement units (IMU) are an important part of
human movement analysis. In recent years, there has been a large emphasis on physical
activity and health monitoring and the use of accelerometers for physical activity monitoring
has become popular (Mathie, Coster, Lovell, & Celler, 2004; Yang & Hsu, 2010).
2.4.1 Wireless Sensor Devices
Accelerometers are very commonly used in gait measures to quantify physical activity.
Accelerometers and gyroscopes have been utilised and validated in running gait analysis
studies (Norris, Anderson, & Kenny, 2014). Such studies have derived both coach oriented
and research oriented kinematic parameters. They can acquire data on the segment
accelerations and whole body accelerations depending on sensor placement, however various
methodologies have been utilised by previous researchers and further investigation of the
methodologies are needed in the future (Norris et al., 2014). It is possible to calculate
different kinematic parameters from accelerometer data such as joint angles at particular
phases of a movement, speed and distance values (J. P. Alexander et al., 2016; Alonge,
Cucco, D'Ippolito, & Pulizzotto, 2014; Djurić-Jovičić, Jovičić, & Popović, 2011; Higginson,
2009; Yang & Hsu, 2010). As with any activity it is not always guaranteed that the
orientations of joints are consistent in the initial plane. Observing a high jump, for example,
the legs are initially perpendicular to the ground, during the flight they move into a position
21
parallel to the ground. If the trace from an accelerometer placed on the shin of this athlete is
observed the trace in one axis may not account for the true movement. A simple
countermovement jump may provide a clearer picture. During this movement an
accelerometer placed at the centre of gravity will measure the vertical component of the
jump, however as the participant squats down to generate momentum to jump into the air
they tilt their torso forward and the vertical axis in the accelerometer is no longer pointing in
the vertical direction. When the orientations change the accelerometer no longer aligns with
the global access (Howard, Conway, & Harrison, 2014; Howard, Healy, Conway, &
Harrison, 2014). The resultant acceleration of all three axes can be calculated but it will not
provide an accurate enough representation of the acceleration in the vertical axis (Howard,
Conway, et al., 2014; Howard, Healy, et al., 2014); further analysis is needed in order to use
the rotation values from the gyroscope to rotate the axes accordingly during the movement.
The use of gyroscopes or magnetometers is thus used to understand the rotation of the device
or its orientation in space. This added functionality can be used to re-orientate the signal to
the global coordinate system. Their use in physical activity measure and in sports
biomechanics increased in recent years (Howard, Conway, & Harrison, 2016). However, due
to various collection and processing methods used with accelerometers between-study
comparisons are not possible; limitations exist with their validity and comparability (Cain,
Sallis, Conway, Van Dyck, & Calhoon, 2013; Pedisic & Bauman, 2014).
2.4.2 Electromyography
An overview of EMG was given in Section 2.2.3. An outline of the electrodes used for EMG
analysis is given in this section. There are two types of electrodes which can be used to
measure the muscle activity, surface mount electrodes and indwelling electrodes:
22
2.4.2.1 Surface Electromyography
SEMGs are a non-invasive form of measuring the electrical impulses during a muscle
contraction. The electrodes are easy to apply and are placed on the surface of the skin over
the muscle belly. Due to the non-invasive nature of SEMG, the encumbrance of the sensor
for the subject being tested is reduced. The disadvantages however, are that the surface
mount electrodes are limited to superficial muscles and are subject to cross-talk as it is
difficult to isolate the individual muscle activity. (De Luca, Kuznetsov, Gilmore, & Roy,
2012; De Luca & Merletti, 1988; Howard, Conway, & Harrison, 2015; Kamen, 2014b;
Kamen & Gabriel, 2010b; Winter, Fuglevand, & Archer, 1994)
2.4.2.2 Preparation of skin & electrode placement
Recommendations in Surface Electromyography for the non-invasive assessment of muscles
(SENIAM) outline that the skin should be prepared in such a way as to increase the signal to
noise ratio (SNR), and to create better EMG recordings (SENIAM, 2016). The impedance of
the skin should be reduced to < 50 Ω. The normal impedance across the skin surface is 50
kΩ. Techniques for preparing the skin vary throughout the papers but the general approach is
to shave any hair in the area and cleanse using an alcohol solution. Gels are used to create
better conduction in some studies. SENIAM recommends pre-gelled surface electrodes with
the wrist or ankle as the reference electrode point. The orientation of the sensors should be
parallel to the muscle fibres. (R. Bartlett, 2014a; Basmajian & De Luca, 1985; Criswell,
2011; Hermens, 1999; Hermens, Freriks, Disselhorst-Klug, & Rau, 2000; Kamen & Gabriel,
2010b; Merletti & Hermens, 2004)
23
Figure 2.6 Bipolar Surface Electrode
Shown here is the bipolar surface electrode arrangement. The individual action potentials from 1
& 2 are shown in the first plot and the resultant signal is the differential, 3, shown in the second
plot.
2.4.2.3 Electrodes
Various materials can be used for the electrodes; Silver/Silver Chloride (Ag/AgCl) is the
most commonly used electrode material due to its connectivity properties. The most common
arrangement of electrodes is in a bipolar fashion (see Figure 2.6). This is also known as a
single differential arrangement. A double differential arrangement has also become popular
(see section 2.4.2.4). A signal electrode is known as mono-polar. The electrode shape can be
square or oval, a 10 mm diameter for circular electrodes is recommended. Either a circular
disk of diameter R or a square electrode with length of R will have roughly the same
coverage area. An inter-electrode distance of 20 mm is recommended to reduce cross-talk.
For smaller muscles the inter-electrode distance should not exceed ¼ of the muscle fibre
length. The position at which the nerve enters the muscle is the motor point. The EMG
signals at this point are not accurate. (Kamen, 2014b; Kamen & Gabriel, 2010b; Merletti &
Hermens, 2004)
24
2.4.2.4 Parallel bar & double differential electrodes
Figure 2.7 Timeline of sensor development
Delsys EMG technologies and software solutions development timeline
(http://www.delsys.com/about-delsys/innovation/)
Delsys sensors first came to the market in 1979 (see Figure 2.7). Extensive research and
testing has been done on the sensor design and they are thought to achieve the best signal
quality and also help reduce cross-talk (Basmajian & De Luca, 1985; De Luca, 1979; De
Luca & Forrest, 1972; De Luca & Merletti, 1988; Roy, De Luca, & Schneider, 1986).
Electrodes have been designed in the form of parallel bars (10 mm long and 1 mm wide) with
an inter-electrode distance of 10 mm. Additional advantages to this type of electrode are that
it is small and lightweight and thus is not obtrusive to the subject, and the spacing between
the electrodes is large enough so that when skin sweats an electrical shortening path is not
created (De Luca, 2003). There are single (see Figure 2.8 (a)) and double (see Figure 2.8 (b))
differential sensors. The double differential sensor differs to the single differential sensor by
having three electrodes rather than two, each separated by 10 mm and performing a two-stage
subtraction (Delsys, 2016).
25
Figure 2.8 Parallel Bar Technology
The single differential sensor (a) and double differential sensor (b), adapted from DeLuca (2003)
2.4.2.5 Indwelling Electrodes
There are two types of indwelling electrodes, needle electrodes and fine-wire electrodes (see
Figure 2.9), these electrodes are inserted into the muscle under observation. This form of
EMG measurement is more precise in locating muscle activity; there is also the added benefit
of getting access to deeper muscle tissue and thereby reducing cross-talk from other muscle
responses. While there are benefits of fine-wire EMG measurements, it is still an invasive
technique and poses minor discomfort for the subject. Due to this invasive nature of the
electrodes it is less likely to obtain ethical approval for routine use and sEMG is more
commonly found in studies on non-pathological subjects. The proficient use of indwelling
electrodes requires significantly more training than that required for surface electrode use. It
only offers a small detection area and cannot be repositioned once inserted. Accurate
placement of these electrodes is more difficult than with surface electrodes. Indwelling
electrodes also contain higher frequency content than the surface EMG electrodes; as the
electric currents propagate through the muscle tissue the amplitude and frequency content are
reduced, thus the surface electrodes record lower frequency content. (Kamen, 2014b; Kamen
& Gabriel, 2010b; Moritani et al., 2004)
26
Figure 2.9 Indwelling electrodes
Examples of a) Needle and b) fine-wire electrodes (Kamen & Gabriel, 2010b, pp. 60, 64)
2.4.2.6 Recording the EMG signal
2.4.2.6.1 Sampling Frequency
To correctly reconstruct the signal, the sampling rate must be greater than twice the
frequency of the highest component of the signal. According to SENIAM recommendations
sampling should be between 1-2 kHz as the SEMG signal components are within the range of
10-500 Hz. Most studies use a 1 kHz sampling frequency with some cases using a smaller
sampling frequency which could cause aliasing or distortion if the sampling frequency is too
low (DeLuca, 2003; Lathi & Green, 2014). Aliasing occurs if signals are sampled below the
Nyquist frequency. This causes a distortion in the signal. Samples are incorrectly detected as
lower frequencies if they are above that of the sampling frequency and the signal cannot be
reconstructed. Figure 2.10 shows three signals with different frequencies all sampled at the
same rate. The first signal has a high sampling rate to allow reconstruction, the second has a
just enough points to allow reconstruction and the sampling rate of the third signal is too low
to allow correct reconstruction. The sampling frequency should be at least 1 kHz, if not 2
kHz, for any activity which involves such rapid movements like sprinting so as not to cause
27
distortion of the signal. (R. Bartlett, 2014a; Criswell, 2011; Hermens, 1999; Kamen, 2014b;
Kamen & Gabriel, 2010b; Merletti & Hermens, 2004)
Figure 2.10 An example of sampling
a) This signal has a good sampling rate to allow reconstruction
b) This signal has just enough points to allow a reconstruction
c) The sampling rate of this signal is too low to allow correct reconstruction
2.4.2.6.2 Amplifier
SEINAM recommends a 12-bit or a 16-bit analog-to-digital converter (ADC). The resolution
is 2n levels, n being the number of bits. Figure 2.11 outlines a 4-bit ADC; there are 24 = 16
binary levels. A higher resolution increases the number of levels and decreases the
quantisation error.
Figure 2.11 An example of a sine wave sampled by a 4-bit analog-to-digital converter
The resolution of the converter is 4-bits; if the number of bits was increased a smoother, more
accurate digital waveform would be created. Adapted from Kamen and Gabriel (2010b)
(a)
(b)
(c)
28
The gain (the ratio of the output to input voltage) should be set with the intension that the
amplitude of the signal is matched to the range of the ADC. Figure 2.12 shows an example
where the gain is set too high and clipping of the signal occurs. A typical gain used is 1000.
(R. Bartlett, 2014a; Criswell, 2011; Hermens, 1999; Kamen, 2014b; Kamen & Gabriel,
2010b; Merletti & Hermens, 2004; Winter, 2009)
Figure 2.12 An example of the input range of the ADC and the amplifier gain set too low
The amplifier gain is set too low for the resolution of the ADC, clipping will occur and some of
the signal will be lost
A common technique, differential amplification, is used when signals are picked up from the
body. Power line interference is a common unwanted signal. To eliminate this signal, which
is known as a common mode signal, a differential amplifier is used to reject the common
signal at the amplifiers inputs, detectable differences are then amplified. This is known as
Common Mode Rejection (CMR). The Common Mode Rejection Ratio (CMRR) is the
ability of an amplifier to amplify differential signals over common signal. A high CMMR
(90 140dB) is typically considered adequate for suppressing extraneous electrical
interference. Figure 2.6 shows the arrangement of amplifiers which is used to create the
bipolar surface EMG and the output signals in monopolar- and in bipolar arrangement. (R.
Bartlett, 2014a; Criswell, 2011; Kamen, 2014b; Kamen & Gabriel, 2010b; Merletti &
Hermens, 2004; Winter, 2009)
29
2.4.2.6.3 Input Impedance of amplifier
A typical input impedance of > 100 MΩ is seen in the articles reviewed. According to the
SENIAM recommendations, the impedance of the skin should be reduced to < 50 Ω. Very
high input impedance is necessary to reduce the loading at the skin-electrode contact point.
When the electrodes are attached, a circuit is formed between the amplifier and the muscle.
Due to the fact that the amplifier draws current, the potential difference between the
recording electrodes is decreased. This means that the voltage recorded by the amplifier is
less than the actual magnitude. High input impedance is critical to make sure the larger
proportion of voltage drop is across the amplifier rather than the body. (R. Bartlett, 2014a;
Criswell, 2011; Hermens, 1999; Kamen, 2014b; Kamen & Gabriel, 2010b; Merletti &
Hermens, 2004; Winter, 2009)
2.5 Signal Processing
Signal processing is the enabling technology for the generation, transformation, and
interpretation of information. With advances in digital hardware, digital signal processing
(DSP) has grown exponentially since its emergence in the 1970s. Applications of these
techniques are now prevalent in such diverse areas as biomedical engineering, acoustics,
sonar, radar, seismology, speech communication, telephony, nuclear science, image
processing and many others. (Lathi & Green, 2014)
2.5.1 Processing the EMG signal
The acquisition and analysis of the sEMG signal is very important in making sure that results
are valid and reliable. Many standards have been developed in reporting sEMG signals. The
SENIAM guidelines outline standards for reporting sEMG data during the acquisition stage.
There are also standards for the data analysis stage outlined by the International Society of
Electromyography and Kinesiology (ISEK). The data analysis of raw EMG signals is crucial.
30
(Basmajian & De Luca, 1985; Hermens, 1999; Hermens et al., 2000; Kamen, 2014b; Kamen
& Gabriel, 2010c; Merletti, 1999; Weiss et al., 2004)
2.5.1.1 Amplitude Analysis
2.5.1.1.1 Peak to Peak Amplitude
The peak-to-peak amplitude measurement is useful in measuring M-waves and H-reflex’s.
The M-wave is the maximum amplitude of the EMG signal; it is the maximum EMG activity
the muscle is capable of producing. It is measured from the negative to the positive peak.
The H-reflex is a reflex action of the muscle after electrical stimulation; it is smaller in
magnitude than the M-wave. When a low intensity electrical stimulus is applied to the
peripheral motor nerve the H-reflex is evoked. (Kamen, 2014b)
Figure 2.13 Sample Amplitude Analysis of EMG signal
The raw EMG signal is the unprocessed signal acquired from the EMG system, a DC offset is
applied to the raw EMG signal and then it is full wave rectified, a 10Hz low pass Butterworth filter
is applied then to achieve a linear envelope, finally the signal is integrated and the maximum value
is the iEMG.
2.5.1.1.2 Rectification & Removal of DC offset
Full-wave rectification is a technique used to get the absolute value of all parts of the EMG
signal; it is generally one of the first post-processing steps performed on the raw EMG signal,
31
see Figure 2.13. The polarity of the negative part is inverted so the signal is super imposed
on the positive side. Half wave rectification removes the negative part of the signal
completely and only keeps the positive values. The AP of a resting muscle tends to sit at -90
mV; another initial step in the post processing of the raw EMG signal is to remove the DC
offset and return the EMG signal to the 0V level. (Kamen, 2014b; Winter, 2009)
2.5.1.1.3 Root Mean Square Amplitude
The Root Mean Square (RMS) Amplitude is calculated on the raw EMG signal; see equation
2.1. Since the raw EMG signal has both positive and negative values the average may yield a
value of zero. To prevent this, the values are made positive by squaring each amplitude
component. The next step is to take the square root of the average (mean) of the squared
values.



2.1
The RMS amplitude applies to periodic signals like sinusoids as well as noise so it is a
commonly accepted measure of amplitude. The RMS amplitude is useful for EMG signals as
they are complex waveforms; unlike sinusoids they are non-repeating, the RMS offers an
unambiguous measure of amplitude with physical significance. Some typical window sizes
used are 20 ms, 50 ms and 100 ms. For this technique there is no need for rectification of the
raw EMG signal. (Robertson et al. 2014b)
2.5.1.1.4 Filtering
There are a wide range of filters available when using signal processing to analyse different
datasets. Commonly, for EMG analysis, a high pass filter is used to remove motion artefact
and a low pass filter is used to create a linear envelope. Removing motion artefact is often
performed using a band pass filter which is a combination of a low pass and high pass filter.
32
A low pass filter is used to reject high frequency components; it allows low-frequency
components pass through and attenuates the high frequency components above that of the
cut-off frequency, see Figure 2.14. The cut-off frequency of the low pass filter should be
typically 500Hz, half that of the sampling frequency. In cases where indwelling electrodes
are used higher frequency components are registered therefore the cut-off frequency is
greater, roughly 1 kHz. A linear envelope is created using a low pass filter which has as an
input a full-wave rectified EMG signal; see Figure 2.13. This is a type of moving average as
it follows the trend of the EMG signal. It is used to measure the volume of the activity of the
EMG signal. An anti-aliasing low pass filter is used to restrict the bandwidth of the signal to
satisfy the sampling theorem. This filter is therefore used before the signal is sampled.
(Kamen, 2014b; Kamen & Gabriel, 2010c; Winter, 2009)
Figure 2.14 Low Pass Filter
The pass band ranges from the lowest frequency component to the cut-off frequency; from here
frequencies that are higher than the cut-off frequency will be attenuated. In the transition band
some frequencies will get through and no high frequency components will pass through in the stop
band.
A high pass filter is used to reject low frequency components; it allows high-frequency
components pass through and attenuates the low frequency components below that of the cut-
off frequency, see Figure 2.15. Commonly a high pass filter with a cut-off frequency of 10
Hz should be used to remove unwanted low frequency components. However, often a cut-off
33
frequency between 10-20 Hz is used to remove frequency components due to movement
artefacts.
Figure 2.15 High Pass Filter
Initially frequencies that are lower than the cut-off frequency will be attenuated. In the stop band
no low frequency components will pass through and in the transition band some frequencies will
get through. The pass band ranges from the cut-off frequency to the highest frequency component.
Notch filters with cut-off frequencies between 50 60 Hz can be used; however a significant
proportion of the EMG signal is eliminated in these cases. Notch filters are used to attenuate
radio frequency activity from lights or other equipment which may affect the EMG signals.
These filters are not being recommended. (Kamen, 2014b; Kamen & Gabriel, 2010c)
2.5.1.1.5 Integrated EMG
The integrated EMG (iEMG) is the area under the curve of the linear envelope of the EMG
signal; see Figure 2.13. A sum of the total muscle activity over a period of time is calculated.
This technique is useful for quantifying the amount of muscle activity or signal energy; it is
important for quantitative EMG relationships such as EMG vs. muscular force. The iEMG
can be calculated over the entire contraction performed; here the total iEMG value from the
curve is used. It can also be performed whereby the signal is integrated and reset after a fixed
time interval or after a particular voltage is reached (Kamen, 2014b). This can be useful in
34
analysing specific phases of a movement whereby the EMG vs force can be compared at
different points in the movement. (Kamen, 2014b; Kamen & Gabriel, 2010c; Winter, 2009)
2.5.1.2 Frequency Analysis
Previous methods were computed in the time domain, where data are expressed as a sum of
sinusoids when converted into the frequency domain. Many frequency analysis techniques
have been used in the analysis of EMG signals to assess muscle fatigue. Both physiological
and non-physiological information can be gathered, such as firing rates of MUs. It can also
be used to removing noise. By computing the number of turning points in peaks per unit time
or the number of times the signal crosses zero, an estimate of the frequency content of the
EMG signal can be calculated. Fourier analysis methods can also be used to estimate the
frequencies that make up the EMG signal. The Fourier series allows the representation of a
signal in terms of sinusoids or complex exponentials (see equations 2.2 and
2.3). The DC component is represented by the component, the amplitude of each
cosine and sine term are represented by and, and the frequency of each term is
represented by. (De Luca, 2003; Lathi & Green, 2014)
   
 2.2
   2.3
Taking an MUAP as an example signal, an infinite sum of sinusoids derived from the Fourier
series can reconstruct the MUAP (see Figure 2.16). Only 10 sinusoids are shown here for
illustrative purposes however the sum of these sinusoids provides a close to accurate
reconstruction of the MUAP.
35
2.5.1.2.1 Fourier analysis
The Fourier Transform is used to calculate the frequency spectrum of the EMG signal, which
is simply a histogram of the amplitudes of each sinusoid at each frequency. However, it
requires a large amount of computations and processing power. The Fast Fourier Transform
(FFT) was developed to reduce the processing time and number of computations. The
original sequence is split into smaller segments and the number of additions and
multiplications required is reduced, reducing the overall number of computations. Fourier
analysis assumes that the signal is stationary, i.e. an isometric contraction performed while
acquiring EMG data can undergo Fourier analysis. During isotonic contractions there are
changes occurring such as muscle force, length and contraction speed. This is an issue for the
FFT as it assumes that the frequency spectrum doesn’t vary over time.
A short-time Fourier Transform (STFT) is an option for isotonic contraction. The STFT
needs the signal to be stationary within the running window, thus the spectral analysis for
isotonic contractions are more effective using an STFT (Bigliassi et al., 2014). The power
spectral density can be calculated from the frequency spectrum of EMG signal. It is the
squared frequency spectrum; see Figure 2.17. Statistical variables such as mean and median
can be applied to the power spectrum of the EMG signal to assess muscle fatigue and analyze
MU recruitment. Wavelet based methods are another frequency domain analysis procedure
which can be used with isotonic contractions. (De Luca, 2003; Lathi & Green, 2014)
36
Figure 2.16 An example of Fourier decomposition of a motor unit action potential (MUAP)
The original MUAP is shown in red. The blue superimposed signal is the mathematical sum of the
sinusoids shown above. To reconstruct the red signal exactly, an infinite number of sinusoids
would be required. (De Luca, 2003)
2.5.1.2.2 Mean & Median Frequency
Mean frequency (MNF) and median frequency (MDF) are traditional measures of the
frequency content of the EMG signal, Figure 2.17. The MNF is simply the average; it is
calculated by determining the frequency at which the average power of the power spectrum
exists. The MDF is calculated by determining the frequency that divides the power spectrum
in two regions having the same amount of power. They are the most popular frequency
37
domain features used in the assessment of fatigue. Typically in unfatigued muscles, the MNF
or MDF in the spectrum is between 50 80 Hz. As muscles begin to fatigue the fast-twitch
fibres drop out causing a shift in the frequency spectrum to the left. The MNF and MDF are
thus lowered. There is less variability in the MNF as the SD of the MNF is lower than the
SD of the MDF; therefore it is a better measure than the MDF. (Hermens, Vonbruggen,
Baten, Rutten, & Boom, 1992; Kamen, 2014b; Kamen & Gabriel, 2010c)
Figure 2.17 Power Spectrum of the EMG signal
The bulk of the energy of the sEMG signal is between 10 500 Hz, the maximum frequency of the
sample EMG signal is about 50 Hz, the median frequency is about 169 Hz and the mean frequency
is about 188 Hz
2.5.1.2.3 Normalisation
If comparisons are sought between EMG data with variability such as trials, muscle groups
and participants, the EMG needs to be normalised. The most common technique is to
calculate the EMG value during a maximum voluntary isometric contraction (MVIC); all
other EMG values are then displayed as a percentage of the MVIC (Albertus-Kajee, Tucker,
Derman, Lamberts, & I., 2011). However, the criticism of this technique is the fact that it is
difficult to ensure the MVIC is truly maximal. Another technique is to use the maximum
recorded value during an isotonic contraction and normalise the signal to this. An example of
38
this would be using the mean of a select few of the peak amplitudes recorded in the BF and
RF during the fastest sprint, and normalising the corresponding BF datasets and RF datasets
to these mean amplitudes. The EMG data recorded could then be represented as a percentage
of the fastest sprint. It is also possible to normalise using the mean or peak EMG measured
during a specific task; this is perfectly acceptable if inter-individual variability is the aim, but
not when comparing between trials, individuals and different studies. (Albertus-Kajee et al.,
2011; Ball & Scurr, 2008, 2011, 2013; Burden, 2010; Kamen, 2014b; Kamen & Gabriel,
2010c)
2.6 Cross-talk
The main disadvantage of sEMG is its limitation to use on superficial muscles and the effect
of cross-talk. Cross-talk is the interference which is picked up by the EMG electrode from
adjacent muscles. An example of how sEMG sensors may pick up cross-talk from adjacent
muscles is shown in Figure 2.18. There are two main factors that contribute to the amount of
cross-talk signal detected, the distance between the electrodes on the sEMG sensor and the
position of the sensor on the muscle (De Luca et al., 2012). The amount of cross-talk also
depends on the size of the muscle. In smaller muscles there are location constraints due to
the size of the sensor in proportion to the size of the muscle to be analysed. The presence of
cross-talk would therefore be more dominant in the smaller muscles as the sensors would be
located near the adjacent muscles as a result of their size. During a movement where a
greater force is output by the muscles the presence of cross-talk would be greater and have an
effect on the EMG-force curves (Kuriki et al., 2012). As more MUs are contributing to the
movement to provide a greater force not only will this be occurring in the muscle being
analysed but also in the adjacent muscles which are also contributing to the movement.
39
Figure 2.18 SEMG sensors can pick up cross-talk from adjacent muscles
Muscle A and Muscle B are shown here. There are 5 EMG sensors located around the surface of
the skin. EMG 2 will pick up the best signal from Muscle A, EMG1 and EMG3 would pick up
cross-talk from adjacent muscles. Similarly EMG4 and EMG would pick up cross-talk from
adjacent muscles but predominantly the signal from Muscle B. A better placement to acquire the
signal from Muscle B would be a position between EMG 4 and EMG 5.
SENIAM recommends an inter-electrode distance of 20 mm (Hermens et al., 2000), however
De Luca et al. (2012) observed reduced cross-talk using a 10 mm inter-electrode distance. It
was found that the amplitudes of the sEMG signal using 10 mm, 20 mm, and 40 mm inter-
electrode distances were not significantly different but as the inter-electrode distance
increased, the amplitude of the cross-talk signal increased more than the signal from the
muscle in question and the optimum inter-electrode distance to reduce cross-talk
contamination was 10 mm (De Luca et al., 2012). It was observed that cross-talk is more
affected by the inter-electrode distances when double differential systems are used (De Luca
et al., 2012). Similarly, Farina, Merletti, Indino, Nazzaro, and Pozzo (2002) observed that the
cross-talk signal increased as the inter-electrode distance increased. In single differential
electrodes the differences in cross-talk across 10 mm to 40 mm inter-electrode distances
showed a mean difference of about 2% (Farina et al., 2002). Cross-talk from double
differential signals was much smaller than for single differential signals, however when the
40
inter-electrode distance increased from 10 mm to 20 mm the cross-talk almost doubled
(Farina et al., 2002). Several articles have discovered that the best position to locate the
sensor is in the middle of the muscle belly (Basmajian & De Luca, 1985; De Luca, 1997; De
Luca, Gilmore, Kuznetsov, & Roy, 2010), like the position of EMG2 in Figure 2.18. Placing
the sensor at the perimeter of the muscle surface will result in a greater detection of cross-talk
from adjacent muscles rather than one placed in the centre of the muscle surface (Basmajian
& De Luca, 1985; De Luca, 1997; De Luca et al., 2010; De Luca et al., 2012). The amplitude
and frequency content of signals gathered using large inter-electrode distances for single
differential systems are less affected by the electrode location (Farina et al., 2002).
2.7 Locomotion
The human musculoskeletal system (locomotor system) gives the human body the ability to
move using its skeletal systems and muscles. Locomotion is also known as gait: it is the
biphasic forward propulsion of the human body’s centre of gravity through the alternate
movements of different segments of the body. Gait patterns are characterised by differences
in limb movement patters, speed, forces and changes in the foot ground contact. Walking is
the slowest form of gait naturally occurring to humans. Due to the fact that a gait cycle is
repetitive and cyclical, analysis can be done on a number of different gait cycles. (R. Bartlett,
2014b; Ethier & Simmons, 2007c)
2.7.1 Walking
Walking is a fundamental skill whereby the movement advances by lifting and setting down
each foot in turn. While walking there is a single-support phase: one foot remains in contact
with the ground, and a double-support phase: both feet will be in contact with the ground at
the same time. The walking gait cycle can be defined as touch down or initial contact (IC) of
one foot to the subsequent IC of the same foot or, the toe-off (TO) of one foot to the next TO
41
of the same foot. The walking gait can be divided into two phases: the stance phase and the
swing phase. The stance phase begins at touchdown and ends at TO for one foot. The stance
phase accounts for greater than 50% of the gait cycle, due to the double-support phase which
occurs once at the beginning of the stance phase and once at the end of the stance phase
(Novacheck, 1998). The mid-stance phase occurs when the foot is flat on the ground. The
swing phase begins from the point the toe leaves the ground and ends at the subsequent
touchdown; the foot will not be in contact with the ground during the swing phase. The
double-support phase occurs at touchdown of one foot and just before TO of the opposite
foot. (R. Bartlett, 2014b; Ethier & Simmons, 2007c)
The gait cycle can be described as follows from left foot touchdown to subsequent left foot
touchdown (see Figure 2.19 ):
left foot IC
right foot TO
left foot mid-stance
right foot IC
left foot TO
right foot mid-stance
left foot IC
42
Figure 2.19 The walking gait cycle
The phases have been defined with respect to the left leg of the figure. The cycle begins and ends
with the initial contact (IC) of the left leg. The stance phase is defined as the first IC until the toe-
off (TO). The swing phase is defined as the TO until the next IC.
2.7.2 Running
Running is a fundamental skill like walking; it is also a cyclical activity where each running
stride follows the previous in a continuous pattern. The gait cycle, which is similar to
walking, generally starts and finishes at the IC of the same leg or at the TO of the same leg.
Running is faster than walking and unlike the walking gait cycle both feet will not be in
contact with the ground at the same time (double support phase), instead twice during the gait
cycle both feet will be airborne (Novacheck, 1998). This is known as the double float and
occurs once at the beginning and once at the end of the swing phase. There are two phases,
stance and swing, like walking. The phases can be broken down further into the braking
phase (early stance), the propulsion phase (late stance), the recovery phase (early swing) and
the pre-activation phase (late swing). The stance phase accounts for less than 50% of the gait
cycle in running, the time at which TO occurs depends on the speed, less time is spent in
stance as the speed increases (Novacheck, 1998). Figure 2.20 shows the various phases
across the running gait cycle. (R. Bartlett, 2014b; Ethier & Simmons, 2007c)
43
Figure 2.20 The phases of the running gait cycle
The phases have been defined with respect to the right leg of the figure. The cycle begins and
ends with the initial contact (IC) of the right leg. The stance phase is defined as the first IC until
the toe-off (TO). The swing phase is defined as the TO until the next IC.
The gait cycle can be described as follows:
right foot IC
right foot mid-stance
right foot TO
left foot IC
left foot mid-stance
left foot TO
right foot IC
2.7.3 Sprinting
Sprinting is a high velocity cyclical activity, an extreme form of running. Generally as speed
increases, running becomes sprinting and the IC changes from the heel to the ball of the foot
(Novacheck, 1998). The running gait cycle defined above also applies to sprinting, and again
the time in stance phase is reduced. World class sprinters could TO around 22% of the gait
44
cycle (Novacheck, 1998). Sprinting in Track and Field Athletics ranges from the 60 m
(indoors) sprint to the 400 m sprint. There are four stages associated with a sprint, the drive
(the athlete is driving low out of the blocks and accelerating), the transition (the athlete is
accelerating and moving into an upright body position), the maximal velocity (the athlete has
reached maximum velocity and is running in an upright body position) and the maintenance
(the athlete is maintaining maximum velocity in an upright body position through to the
finish line).
Muscle activations and their patterns are useful to consider during sprinting. With the
advances in technology the analysis during sprinting has been simplified and proved more
accurate, with less interference due to wired data logging systems. However, limitations still
arise when measuring the electrical impulses during sprinting, this movement can produce
considerable skin artefact and electrode movement, particularly during specific phases of the
gait cycle when the skin is stretched (Kamen, 2014b). EMG can be used for normal muscle
function studies, muscle activity in complex sports, rehabilitation movements, fatigue studies
and many more. Patterns of muscle activity across the running gait cycle can be identified
using EMG; see Figure 2.21, this type of figure may be useful in comparing variations across
muscle activity in the running gait cycle and in other sporting techniques to enable speciality
of training.
45
Figure 2.21 Muscle activity across the running gait cycle
The timing of muscle activity across the running gait cycle for various lower limb muscles, image
adapted from Kamen and Gabriel (2010d)
46
2.8 Conclusion
This chapter provided detailed background on various areas of biomechanics with an
emphasis on sports biomechanics, sensor devices and EMG. With the wide range of sensors
available to practitioners there is an opportunity to examine the culture around various sensor
devices, their use amongst the sports biomechanics community and their applications in sport
biomechanics. With the recent advances in technology especially into wireless sensor
devices there is also an opportunity to examine further EMG during sprinting. A
comprehensive review of current literature and expectations of future research is necessary to
highlight the use of EMG in sprinting. These recent advances into wireless EMG will also be
useful in other sporting activities to provide analysis and a profile of muscle activity where
previously due to technology constraints this was not possible. Finally the main limitation of
sEMG is cross-talk; this is an area where more analysis needs to be undertaken to discover if
it is possible to minimise cross-talk using advanced signal separation methods.
47
Chapter 3: Literature Review - Muscle Actions in
Sprinting
The more that you read, the more things you will know.
The more that you learn, the more places you'll go.”
Dr. Seuss
Howard, R. M., Conway, R. & Harrison A. J. (2016) Muscle activities in sprinting: A
review. Journal of Sports Biomechanics. (In Press)
48
49
3.1 Introduction
In sports biomechanics, EMG analysis provides important information on muscle activity
which may be useful in optimising performance or reducing the likelihood of sports injuries
(Ditroilo et al., 2011; Nummela, Rusko, & Mero, 1994; Paul & Wood, 2002). This is crucial
for athletes such as sprinters, as speed increases the likelihood of injury are greatly increased
(Higashihara, Ono, Kubota, Okuwaki, & Fukubayashi, 2010; Schache, Dorn, Blanch, Brown,
& Pandy, 2012; Yu et al., 2008). Sports performance monitoring for injury prevention is
very important for athletes and their coaches as potentially the risk of injury may be increased
with an increase in speed and due to muscle fatigue. Identification of the specific effects of
fatigue on muscle activation may provide important insights about specific injury
mechanisms in sprinting (Thelen, Chumanov, Best, Swanson, & Heiderscheit, 2005; Yu et
al., 2008). Utilising EMG to provide information on muscle activity can be useful in
examining changes across increases in speed or muscle fatigue. Many features of the EMG
signal have been associated with fatigue or speed, especially the amplitude of the EMG
signal. Of particular importance are the average EMG (AEMG): the average amplitude of the
rectified EMG signal and the integrated EMG (iEMG): the total accumulated activity of the
muscle. An increase in either the AEMG or iEMG has been reported to be associated with an
increase in muscular fatigue (Nummela et al., 1994; Nummela, Vuorimaa, & Rusko, 1992),
while also having a positive association with increasing running speeds (Chumanov,
Heiderscheit, & Thelen, 2007; Higashihara et al., 2010).
While many studies have examined applications of EMG in gait, relatively few have
examined muscle activity in sprinting. This could be due to the many challenges associated
with gathering accurate EMG data in sprinting. The demands of sprinting require EMG data
to be acquired in an unobtrusive way, therefore the EMG sensor design needs to minimise
encumbrances on the athlete during sprinting. Any change in the way in which an athlete
50
normally performs a sprint could result in unreliable data being gathered. To reduce
discomfort and avoid invasive procedures, the majority of isotonic movements are analysed
using sEMG. With advances in technology, sEMG measurements have evolved from
tethered systems to data loggers (wireless telemetry) and more recently, to fu