Citation
Electromyogram power spectra frequencies associated with the development of chronic low back pain

Material Information

Title:
Electromyogram power spectra frequencies associated with the development of chronic low back pain
Creator:
Patel, Nirav M
Publication Date:
Language:
English
Physical Description:
xiv, 119 leaves : ; 28 cm

Subjects

Subjects / Keywords:
Backache ( lcsh )
Electromyography ( lcsh )
Power spectra ( lcsh )
Backache ( fast )
Electromyography ( fast )
Power spectra ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 113-119).
General Note:
Department of Electrical Engineering
Statement of Responsibility:
by Nirav M. Patel.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
71814897 ( OCLC )
ocm71814897
Classification:
LD1193.E54 2006m P37 ( lcc )

Full Text
CT
ELECTROMYOGRAM POWER SPECTRA FREQUENCIES ASSOCIATED
WITH THE DEVELOPMENT OF CHRONIC LOW BACK PAIN
A thesis submitted to the
University of Colorado at Denver & Health Science Center
In partial fulfillment
Of the requirements for the degree of
Master of Science in Electrical Engineering
By
Nirav M. Patel
B.E., University of Pune, 1999
2006


This thesis for the Master of Science
Degree by
Nirav M. Patel
Has been approved
By
Dr. Jan Bialasiewicz


Patel, Nirav M. (M.S., Electrical Engineering)
Electromyogram power spectra frequencies associated with the development of
chronic low back pain
Thesis directed by Chair / Associate Professor Dr. Miloje Radenkovic
Abstract
The objective of this thesis was to study the effect of static flexion on the
electromyogram (EMG) power spectra frequencies and to study the effect of sub-
failure injuries of ligament which may cause low back pain due to muscle control
dysfunction. Three experimental groups of a feline model were used with the rest
duration between sequential static load periods was set to 5, 10, and 20 min,
corresponding to load to rest ratio of 2:1, 1:1 and 1:2 respectively. The reflex
electromyographic activity from the multifidus muscles and supraspinous ligament
displacement was recorded during the 6 loading periods each of 10 min duration and
over 7 h of rest following the load rest cycles. Signals were band-pass filtered (5-
500Hz) and stored with a sampling rate of 1 kHz. The EMG signal was split in
segments of 1 s for each trial and Fast Fourier Transform (FFT) of each such segment
was computed with 50 % overlap. Median frequency (MF) was computed and the
method of moving averages was implemented to smooth the data. Spectral analysis
showed a very fast drop in median frequencies in the group subjected to load to rest
ratio of 2:1, suggesting severe neuromuscular disorder development while in the
groups subjected to load to rest ratio of 1:1 and 1:2 did not show neuromuscular
dysfunction. The results suggest that cumulative micro-trauma resulting in sub-failure
injuries of the ligaments and embedded mechanoreceptors may be the cause of
corrupted muscle response pattern produced by the neuromuscular control unit.
This abstract accurately represents the content of the candidates thesis. I recommend
its publication.
Signed
Dr. Miloje Radenkovic
in


DEDICATION
I dedicate this thesis to my parents who always encourage me to pursue knowledge.


ACKNOWLEDGEMENT
I would like to thank Dr. Miloje Radenkovic for his recommendation to work as a
research assistant at University of Colorado Health Sciences Center, as well for his
guidance, valuable comments and criticism on the manuscript of this thesis. I would
like to express my deepest gratitude to my supervisor Dr. Moshe Solomonow, the
Director of Bioengineering Division and Musculoskeletal Disorders Research
Laboratory, for direction, suggestions and unfailing support he has given me during
all stages of this study. In particular I thank Dr. Solomonow for mentoring me in the
research involved with this thesis. I would like to thank Dr. Jan Bialasiewicz for
reviewing my thesis.
I also wish to thank Dr. Zhou for his explanations with regards to the research
experiments and data acquisition procedures; Dr. Lu for showing me the surgical
procedures required for the experiments and detailed explanation of the anatomy of a
cats spine and Dr. Arabadzhiev for having patience with me and answering my
questions regarding the physiological background required for this thesis. I would like
to thank Ms. Pamela Dunham for her help with regards to administrative functions at
the laboratory.
Finally, words alone cannot express the thanks I owe to Palak Patel, my wife, for her
patience, encouragement and assistance.


CONTENTS
Figures.......................................................................x
Tables........................................................................xiv
Chapter
1. Introduction.............................................................1
2. Physiological Background.................................................5
2.1 Lower Back Anatomy and Physiology......................................5
2.1.1 Tendons and Ligaments...............................................5
2.1.2 Vertebral Column....................................................7
2.1.3 Intervertebrae Joints...............................................8
2.1.4 Spinal Cord.........................................................9
2.1.5 Multifidus Muscles..................................................9
2.1.6 Mechanoreceptors...................................................10
2.2 Motor Unit Recruitment................................................10
2.3 Electromyography......................................................12
2.3.1 Frequency parameters of the myoelectric signal.....................13
3 Rationale and Objectives................................................16
3.1 Spinal Stabilizing feedback control system............................16
4. Methods and Procedures..................................................19
vi


4.1 Electromyogram Data Acquisition, Processing and Storage.............19
4.1.1 Factors Influencing EMG Signal....................................20
4.1.2 Differential pre-amplifier........................................22
4.1.3 Signal Conditioning and Amplification.............................22
4.2 Discrete Fourier Transform..........................................23
4.2.1 Radix-2 FFT Algorithms............................................24
4.3 Spectral Analysis of deterministic signals..........................31
4.3.1 Windowing.........................................................32
4.3.2 Zero Padding......................................................34
4.3.3 Overlap processing................................................35
4.4 Preparation.........................................................35
4.5 Instrumentation.....................................................36
4.6 Experimental Protocol...............................................37
4.7 Data Analysis.......................................................40
5. Results...................................................................42
5.1 EMG Responses.......................................................55
5.1.1 5 Minute Rest.....................................................55
5.2.2 10 Minute Rest....................................................62
5.1.3 20 Minute Rest....................................................68
5.2 Statistical Analysis............................................... 74
vii


6.
Summary and Conclusions
76
Appendix
A. Tables....................................................................83
A. 1 Mean normalized MF values of the first loading period (0-10min) for the
group subjected to 5 min of rest between each loading period..........83
A.2 Mean normalized MF values of the second loading period (15-25min) for
the group subjected to 5 min of rest between each loading period......84
A.3 Mean normalized MF values of the third loading period (30-40min) for the
group subjected to 5 min of rest between each loading period..........85
A.4 Mean normalized MF values of the fourth loading period (45-55min) for
the group subjected to 5 min of rest between each loading period......86
A.5 Mean normalized MF values of the fifth loading period (60-70min) for the
group subjected to 5 min of rest between each loading period..........87
A.6 Mean normalized MF values of the sixth loading period (75-85min) for the
group subjected to 5 min of rest between each loading period..........88
A.7 Mean normalized MF values of the first loading period (0-10min) for the
group subjected to 10 min of rest between each loading period.........89
A.8 Mean normalized MF values of the second loading period (20-30min) for
the group subjected to 10 min of rest between each loading period.....90
A.9 Mean normalized MF values of the third loading period (40-50min) for
the group subjected to 10 min of rest between each loading period.....91
A. 10 Mean normalized MF values of the fourth loading period (60-70min) for
the group subjected to 10 min of rest between each loading period.....92
viii


A. 11 Mean normalized MF values of the fifth loading period (80-90min) for the
group subjected to 10 min of rest between each loading period........93
A. 12 Mean normalized MF values of the sixth loading period (100-11 Omin) for
the group subjected to 10 min of rest between each loading period....94
A. 13 Mean normalized MF values of the first loading period (0-1 Omin) for the
group subjected to 20 min of rest between each loading period........95
A. 14 Mean normalized MF values of the second loading period (30-40min) for
the group subjected to 20 min of rest between each loading period....96
A. 15 Mean normalized MF values of the third loading period (60-70min) for
the group subjected to 20 min of rest between each loading period....97
A. 16 Mean normalized MF values of the fourth loading period (90-1 OOmin) for
the group subjected to 20 min of rest between each loading period....98
A. 17 Mean normalized MF values of the fifth loading period (120-130min) for
the group subjected to 20 min of rest between each loading period....99
A. 18 Mean normalized MF values of the sixth loading period (150-160min) for
the group subjected to 20 min of rest between each loading period....100
B. Computer Program.....................................................101
References..................................................................113
IX


FIGURES
Figure
3.1 Spinal Stabilizing feedback control system...............................16
4.1 Data Acquisition System..................................................19
4.2 Computation of N = 8 point DFT...........................................27
4.3 Flow graph of complete decimation in time decomposition of N = 8 point
DFT.....................................................................28
4.4 Flow graph of a basic two-point DFT......................................28
4.5 Bit reversed storage of input data.......................................30
4.6 Illustration of discrete Fourier transform based Fourier analysis for
continuous time signals................................................31
4.7 Commonly used window functions: rectangular, Bartlett, Hamming,
Hanning and Tukey window with N = 64 samples and Tukey window with
ratio of taper to constant sections, r = 0.25...........................34
4.8 Experimental set up showing the S shaped hook inserted around the L-4/5
supraspinous ligament of a cat with external fixation and bipolar wire
electrodes inserted to record data......................................38
4.9 Schematic representation of the experimental set-up showing the lumbar
spine with external fixation and loading apparatus. Ex-Fix: external fixator. 39
5.1 EMG responses from the 6 channels, force and displacement channel
during six 10 min constant loading period and 7 hours of recovery. The
constant loading periods subjected to 40N with 5 min of rest period
between them............................................................43
5.2 L-3/4 EMG response during six 10 min constant loading period subjected
to 5 min of rest period between them...................................44
x


5.3 Moving average MF values at lumbar level L-3/4 during 6 loading periods
of constant load........................................................44
5.4 EMG response at lumbar level L-4/5 during six 10 min constant loading
period subjected to 5 min of rest period between them..................45
5.5 Moving average MF values at lumbar level L-4/5 during 6 loading periods
of constant load........................................................45
5.6 EMG response at lumbar level L-5/6 during six 10 min constant loading
period subjected to 5 min of rest period between them..................46
5.8 EMG responses from each of the 6 channels, force and displacement
channel during six 10 min constant loading period and 7 hours of
recovery. The constant loading periods subjected to 40N with 10 min of
rest period between them................................................47
5.9 EMG response at lumbar level L-3/4 during six 10 min constant loading
period subjected to 10 min of rest period between them. 48
5.10 Moving average MF values at lumbar level L-3/4 during 6 loading periods
of constant load........................................................48
5.11 EMG response at lumbar level L-4/5 during six 10 min constant loading
period subjected to 10 min of rest period between them.................49
5.12 Moving average MF values at lumbar level L-4/5 during 6 loading periods
of constant load........................................................49
5.13 EMG response at lumbar level L-5/6 during six 10 min constant loading
period subjected to 10 min of rest period between them.................50
5.14 Moving average MF values at lumbar level L-5/6 during 6 loading periods
of constant load........................................................50
5.15 EMG responses from each of the 6 channels, force and displacement
channel during six 10 min constant loading period and 7 hours of
recovery. The constant loading periods subjected to 40N with 20 min of
rest period between them................................................51
xi


5.16 EMG response at lumbar level L-3/4 during six 10 min constant loading
period subjected to 20 min of rest period between them.................52
5.17 Moving average MF values at lumbar level L-3/4 during 6 loading periods
of constant load........................................................52
5.18 EMG response at lumbar level L-4/5 during six 10 min constant loading
period subjected to 20 min of rest period between them.................53
5.19 Moving average MF value at lumbar level L-4/5 during 6 loading periods of
constant load...........................................................53
5.20 EMG response at lumbar level L-5/6 during six 10 min constant loading
period subjected to 20 min of rest period between them..................54
5.21 Moving average MF values at lumbar level L-5/6 during 6 loading periods
of constant load........................................................54
5.22 Average MF values at lumbar level L-3/4 during 6 loading periods of
constant load for 7 preparations subjected to 5 min of rest between the
loading periods.........................................................59
5.23 Average MF values at lumbar level L-4/5 during 6 loading periods of
constant load for 7 preparations subjected to 5 min of rest between the
loading periods.........................................................60
5.24 Average MF values at lumbar level L-5/6 during 6 loading periods of
constant load for 7 preparations subjected to 5 min of rest between the
loading periods..................................................61
5.25 Average MF values of lumbar level L-3/4 during 6 loading periods of
constant load for 6 preparations subjected to 10 min of rest between each
loading period..........................................................65
5.26 Average MF values of lumbar level L-4/5 during 6 loading periods of
constant load for 6 preparations subjected to 10 min of rest between each
loading period..........................................................66
xii


5.27 Average MF values of lumbar level L-5/6 during 6 loading periods of
constant load for 6 preparations subjected to 10 min of rest between each
loading period.........................................................67
5.28 Average MF values of lumbar level L-3/4 during 6 loading periods of
constant load for 7 preparations subjected to 20 min of rest between each
loading period.........................................................71
5.29 Average MF values of lumbar level L-4/5 during 6 loading periods of
constant load for 7 preparations subjected to 20 min of rest between each
loading period.........................................................72
5.30 Average MF values of lumbar level L-5/6 during 6 loading periods of
constant load for 7 preparations subjected to 20 min of rest between each
loading period.........................................................73
xiii


TABLES
Table
5.1 F values and Pr>F values obtained for the experimental group 6x10:05 for
6 loading trials for lumbar levels L-3/4, L-4/5, L-5/6.................74
5.2 F values and Pr>F values obtained for the experimental group 6x10:10 for
6 loading trials for lumbar levels L-3/4, L-4/5, L-5/6.................74
5.3 F values and Pr>F values obtained for the experimental group 6x10:20 for
6 loading trials for lumbar levels L-3/4, L-4/5, L-5/6.................75
xiv


1.
Introduction
Lower back pain is a common problem for millions of Americans. In 1994,
the U.S Department of Labor, Bureau of Labor Statistics determined the number of
injury or illnesses resulting from repetitive motion and overexertion as 705,800 per
year [8], According to the National Institute for Occupational Safety and Health
(NIOSH) operators, fabricators, personnel associated with service industry and
laborers are the occupational groups most susceptible to back, spine and spinal cord
injuries and disorders adding up to 60 percent of the total musculoskeletal disorders
(MSDs). Epidemiological studies shows that cyclic lumbar flexion performed by
workers in professions like construction, roofing, welding, farming, driving, etc. leads
to the development of cumulative low back disorder (CLBD) over long period of
time[53]. CLBD is characterized by low back pain, stiffness, limited range of motion
and weakness in the posterior muscles. The epidemiological data shows that number
of daily repetition, magnitude and duration of load were the factors for the
development of CLBD.
Workplace musculoskeletal injuries are classified under two categories:
idiopathic and traumatic. Idiopathic injuries are mediated through mechanical
degradation and cannot be attributed to a specific act or incident. Traumatic injuries
are associated with an incident or an action including overexertion, sudden
imbalance, pulling apart, crushing, impact, slip and fall, cut, abrasion and laceration.
1


The risk factors for MSDs are classified in four major categories [33] as
follows: genetic, morphological, psychosocial and biomechanical. Only
biomechanical and psychosocial factors allow effective control strategies. Exposures
to repetitive static and vibratory activities are known to result in musculoskeletal
disorders including soft tissue injuries. Ligaments and tendons are two connective
soft tissues that are more susceptible to injury.
Ligament injuries are classified into two categories [64]: repetitive micro-
trauma and macro-trauma. In micro-trauma soft tissue failure occurs due to exposure
to forces which are below the normal ultimate tensile strength while in macro-trauma
the forces within a ligament are sufficient to cause partial or complete rupture of its
fibers.
Scientific studies shows that cyclic loading result in creep or laxity of the
viscoelastic structures like ligaments, discs, facet capsules, etc. and was demonstrated
in animals[60]. The afferents located in the ligaments, capsule and discs are
desensitized due to laxity developed in the viscoelastic structures. As a result the
reflexive activation of the muscles is decreased and leaves the spine without
protection.
Recent studies have shown that in humans and animals, the ligamento-
muscular reflex arc stabilizes the knee, shoulder, elbow, ankle joints and spine [26],
[28], [30], [35], [48], [55], [58], [59], [62], [69]. The reflex is triggered by
mechanoreceptors in the ligaments, discs and facet capsules to the multifidus and
2


longissimus muscles, allowing the musculature and the visco-elastic tissues of the
spine to act synergistically.
One important question that arises is the duration of the rest period that is
required to recover the creep and laxity in the viscoelastic structures and restore the
reflexive muscular activity. Several studies investigated the time required to recover
from creep for various types of loads. Solomonow et al., [61] studied the reflexive
muscular activity after prolonged cyclic loading and the rest period required for
recovery in the lumbar spine. Laxity induced by 50 min of cyclic loading at 0.25Hz
was fully recovered after 7 h of rest. Another importing finding in the study was that
the muscle showed hyper excitability, generating more electromyographic response to
viscoelastic deformation then at the beginning of passive loading once the reflexive
activity of the multifidus muscles was recovered fully.
McGill and Brown [39] have studied the creep induced in the human lumbar
spine during flexion in seated position for 20 minutes which included both male and
female subjects. After 20 minutes of rest the recovery of the creep was 50 percent in
males while the recovery was faster in females. It was concluded in the study that the
viscoelastic recovery of the creep was exponential and required longer time for full
recovery. In a study of wrist ligament, Crisco et al. [15] observed that 2 h of rest did
not result in full recovery after 1 h of cyclic loading applied to the wrist ligament.
Full recovery was observed after 24 h of rest. It was demonstrated by Courville et
3


al.[14], that short rest period of load to rest ratio of 2:1 leads to CLBD after static
lumbar flexion.
Various studies show that there are changes in motor unit recruitment
depending on the required force to be generated by the muscle. The motor units are
recruited according to their size in slow voluntary contraction [27], The motor unit
recruitment and variations in firing rate contributes to the change in median frequency
(MF) of the electromyograms power density spectrum. It is a well established fact
that Electromyogram (EMG) power density frequency spectrum is associated with the
average conduction velocity of the active motor units in the muscle. It was shown by
Moritani and Muro [42], Solomonow et al.[57], that median frequency can be used as
the index to identify the changes in motor unit recruitment.
This study concentrates on the changes in motor unit recruitment strategy
associated with the development of low back pain.
The following sections includes brief description of low back anatomy and
physiology, various topics leading to the study, objectives and goals of the study,
method, experimental procedure and protocol, results, discussion and conclusion.
4


2. Physiological Background
2.1 Lower Back Anatomy and Physiology
It is very important to understand the physiological processes of the lower
back structure in order to analyze the effects of loading on viscoelastic tissue
behavior and muscle activity. It is also important to understand voluntary muscle
contraction and the interactive processes that regulate force and length of the muscle
and study the changes in recruitment of the individual motor units. This section
briefly describes the anatomy of the lower back and its structures like tendons,
ligaments, vertebral column, intervertebrae joints, spinal cord, multifidus muscle and
mechanoreceptors. It also describes the basic idea behind motor unit recruitment and
its relationship with muscle contraction.
2.1.1 Tendons and Ligaments -
Tendons attach muscle to bones. They are viscoelastic and have high tensile
strength and modulus of elasticity [31]. They bear tensile loads in the joints and are
made up of visco-elastic tissue called collagen. They help in transferring forces from
muscles to bones.
The ligaments attach one bone to another across the joint. They are made up
of elastin and collagen fibers. The supraspinous ligament is a single, long vertical
5


fibrous band attached to the tips of the spinous processes of the vertebrae from the
seventh cervical to the sacrum [18].
Ligaments and tendons are connective tissue which protects and support the
organs of the body and binds the organs together. They consist of three elements:
cells, ground substance and fibers [66], The cells are made up of mesodermal
embryonic cells, the ground substance supports and binds the cells and provides a
medium between blood and cells. The fibers provide strength and support to the
tissues.
The fibers are divided into three groups: collagen fibers, elastic fibers and
reticular fibers. The collagen fibers are made up of a protein called collagen. They are
found in bundles of tiny fibrils lying parallel to each other. The elastic fibers are made
up of a protein called elastin and they can be stretched up to 150 percent of their
normal relaxed length without breaking. Reticular fibers are made up of the protein
collagen and a coating of glycoprotein. They form a network around fat cells, nerve
fibers, and skeletal and smooth muscle cells and provide support in the walls of blood
vessels [66].
The collagenous fibers rearrange their position parallel to the axis of stress
during loading. As the loading is further increased to the point of failure the fibers
start gliding upon one another which represents the breaking of the forces that hold
the fibers together and fibers tear completely [65]. The longitudinal ligaments change
their length to adjust to the position of the vertebral column during flexion and
6


extension of the lumbar spine [65] and the lumbar spine ligaments are innervated by
neurofilament immunoreactive nerve fibers [51 ] and these fibers appear as single
nerve fibers or bundles terminating as free nerve endings.
In response to load ligament properties can be described in three different
ways: structural behavior, material behavior and viscoelastic behavior [11], [23], [64],
[72], [71], [73]. Structural behavior is the ligaments response to a mechanical load
regardless of its shape and size and material properties are used in order to compare
ligaments of different size [64], The viscoelastic behavior of ligament refers to the
ability of tissues to adapt to loading by changing either the length or their load over
time [31]. In ligaments at low loads viscous behavior dominates and at higher loads
elastic behavior dominates. This allows the ligaments to function within a large range
of loads.
2.1.2 Vertebral Column
The vertebral column is a strong, flexible structure which protects the spinal
cord, support the head and serve as a point of attachment for the ribs and muscles of
the back. It can bend anteriorly, posteriorly, laterally and it can rotate. The adult
vertebral column consists of 26 vertebrae among which there are 7 cervical vertebrae,
12 thoracic vertebrae posterior to the thoracic cavity, 5 lumbar vertebrae supporting
the lower back, 5 sacral vertebrae fused into one bone called sacrum and 4 coccygeal
7


vertebrae fused into one or two bones called the coccyx. The intervertebral discs are
located between each vertebrae from the first vertebrae to the secrum.
The study of interest is the lumbar vertebrae (L1-L5) which are the largest and
strongest in the vertebral column. They support higher amounts of body weight than
cervical and thoracic vertebrae. The load in this study is applied around the
supraspinous ligament between L-4 and L-5. The spinous processes are well adapted
for the attachment of the large back muscles [66],
2.1.3 Intervertebrae Joints
An interverbetrael joint is made up of two vertebrae with an interv ertebral disc
between them. This joint allows mobility to the spine and several studies have shown
that they also have an important role in stabilizing the lumbar spinal segments [9],
[21], [68]. The intervertebral disc bears load in axial compression, flexion and
anterior shear translation [1].
In previous studies it was shown that due to compression elicited by
prolonged sitting and standing there was change in the fluid content of the
intervertebral disc which resulted in changes of its height and viscoelastic properties
[17][67]. It was also shown that there was fluid loss in the disc by compression
elicited as a result of cyclic, vibratory or prolonged loading [2]. It was shown by
Solomomow et al.[60], that the reduced fluid content in the disc increases the laxity
8


of the intervertebrae joints which allows increase in the intervertebral motion and
causes instability.
2.1.4 Spinal Cord
The spinal cord is located in the spinal canal of the vertebral column and
conducts sensory impulses from the periphery to the brain and motor impulses from
the brain to the periphery. The gray matter around the central canal of the spinal cord
receives and integrates incoming and outgoing information [66].
2.1.5 Multifidus Muscles
The Latin word Multifidu means split in many parts. Multifidus muscles are
more superficial, thick and present more vertical fibers in the lumbar spine as
compared to the multifidus muscle in the thoracic region. As a result they are
expected to produce enough tension to ensure posterior stabilization in the lumbar
region while providing mobility [10].
The multifidus muscles are the longest and the most medially oriented
bilateral group of back muscles and act as an extensor and generates rotary
compressive and shear forces on the spinal motion segments. They originate from the
mamillary processes of the superior facet and inserts on the spinous processes and so
it can produce control capacity for rotation, abduction and extension among the
individual motion segments [18].
9


2.1.6 Mechanoreceptors
There are five different types of receptors in human body [66] and they are
mechanoreceptors, thermoreceptors, nociceptors, photoreceptors and chemoreceptors.
The mechanoreceptors sense mechanical pressure or stretching [18]. Thermoreceptors
detect temperature change, nociceptors detect pain, photoreceptors detect light and
chemoreceptors detect chemicals in the mouth, nose and body fluids.
In the supraspinous ligament only type II and type III mechanoreceptors are
found [51]. Type II mechanoreceptors are mostly fast adapting and signal the
initiation and termination of a stimulus to the central nervous system while type III
receptors signals during extreme range of motion and load. Thus type II
mechanoreceptors may provide continuous signals of changes in ligament strain
which results in joint stability by continuous adjustments in muscular activity. Free
nerve ending present in the supraspinous ligament provide long lasting information
relative to the deformation of the tissues in which they are embedded, as well as pain
in case of tissue damage.
2.2 Motor Unit Recruitment
There are five interactive processes that regulate force and length of the
muscle during voluntary muscle contraction [54]. Two of these processes are the
recruitment of individual motor units and the increase in action potential firing rate of
the motor units.
10


Recruitment is the process by which motor units are selectively activated. At
the beginning of the muscle contraction only a few motor units are active and more
and more motor units are activated gradually as the need of force increases. It has
been shown that the order of motor unit recruitment progresses from smaller motor
units to larger ones as the force requirement increases. This process of recruitment
during voluntary contraction is generally referred to as the Size Principal [27], [41].
The motor units begin firing action potentials at an initial rate when they are
first recruited and as more force is required more motor units are recruited. The initial
firing rate of these newly recruited motor units is higher than the firing rate of the
motor units that were already active. Once all motor units are recruited, additional
force can be generated only by increasing the firing rate of all motor units.
In general, the smaller motor units that are recruited during the lower muscle
force levels, have low innervation ratios and provide small force increments upon
activation. The muscle fibers of these small units have slow rising, long lasting twitch
tensions. They are red in appearance and are resistant to fatigue. Motomeurons of the
units that are recruited during the higher force levels have larger innvervation ratios
and provide larger force increments, but they fatigue relatively fast. They have larger
diameters and relatively high conduction velocities. The muscle fibers of these units
are fast twitch and they are white in appearance.
When the contraction is sustained for a relatively long period, there will be
decrease in firing rate since the faster motor units will fatigue relatively fast [7], Since
11


the conduction velocity of the faster motor units is higher then the average, the effect
of their fatigue is a decrease in average conduction velocity and thus a decrease in
median frequency.
2.3 Electromyography
The muscle fiber or the muscle cell is the structural unit of contraction. The
muscle fibers contract as a group and each group is supplied by a single nerve fiber or
axon. Thus a motor unit consists of the nerve cell body, the axon, its terminal
branches and all the muscle fibers supplied by these branches [7], The myoneural
junction where the axonal branches terminate on a muscle fiber is also called motor
endplate. The motor unit is the functional unit of striated muscle since it causes the
contractions of the muscle fibers in the group simultaneously. The nervous impulses
pass through the nerve axon and are supplied to all the muscle fibers in that group.
Once an impulse reaches the endplate, the contraction spreads over the fiber resulting
in a brief twitch. A minute electrical potential in the mV range is generated during the
twitch.
Spatio-temporal summation of the motor unit action potentials (MUAPs) of all
the active motor units constitutes the electromyogram of the muscle [50],
Electromyography is the recording of this electrical activity associated with
contracting muscle, by means of using surface electrodes, needle electrodes or wire
electrodes. The surface electrodes consists of a silver disc shaped detection surface
12


which is adhered to the skin and it measures surface EMG. Needle electrodes are
inserted through the skin into the muscle. An insulated wire is passed through the
needle. The tip of the wire is not insulated and it serves as the detection surface for
recording intra muscular EMG. Wire electrodes consists of a flexible, small diameter,
non-oxidizing insulated wire which is inserted into the muscle. The tip of the wire is
not insulated like the needle electrode [6].
The structure of myoelectric signals is statistically different from random
noise and is non-linear in nature [43], [19]. As a result deterministic signal model can
be used to analyze the EMG signal.
2.3.1 Frequency parameters of the myoelectric signal
The spectrum of the myoelectric signal undergoes compression as a function
of time during sustained muscular contraction. This compression has been shown by
Stulen et al. [63], to be related to the decreasing conduction velocity of the muscle
fibers. The relationship between the power density spectrum of the myoelectric signal
and the conduction velocity of action potentials along the muscle fiber was shown by
a mathematical model in a study by Lindstrom et al.[36]. To derive the power density
spectrum S(f), Fourier transform may be used in this model which is given below:
S(f) = G
K V ,
(2.1)
Where G rational function of frequency f which denotes the shape of the spectrum
v Conduction velocity and
13


d Inter electrode distance of the bipolar electrodes.
The decrease in conduction velocity compresses the spectrum, which can be
seen in the above expression, since the conduction velocity inversely scales the
frequency components of the spectrum. Also, an increase in conduction velocity
would expand the spectrum of the myoelectric signal. Increase in the conduction
velocity would indicate the shift of the power spectrum towards higher frequencies
and thus recruitment of fast motor units. Decrease in the conduction velocity would
indicate the shift of the power spectrum towards lower frequencies indicating fatigue
or change in recruitment of the motor units. In various studies it was shown that EMG
power spectrum may be sensitive specifically to motor unit recruitment [42], [57],
Since all frequencies are scaled by the same factor in the power density
spectrum given by equation 2.1, a frequency shift of the myoelectric signal may be
observed by any characteristic frequency [63], Among various possibilities for the
characteristic frequency are the mean and the median frequencies. The mean
frequency (Fmean) is the average frequency as given below:
ac
F =^-------------- (2.2)
|S(f)df
0
The median frequency (MF) is the frequency at which the power spectrum of
the EMG signal is divided into two regions, each representing equal amounts of
power. This frequency is defined by the following expression:
14


(2.3)
MF oc 1 00
jS(f)df= \S(f)df=-.\S(f)df
0 MF ^ 0
Frequency shift in the myoelectric signal can also be tracked by monitoring
the ratio of the RMS-value of the high frequency components to the low frequency
components. The value of the frequency chosen to divide the spectrum of the
myoelectric signal into two regions may be set initially to any characteristic
frequency, e.g. the mean or the median frequency and the ratio parameter R can be
computed as shown below:
1/2
(2.4)
Where, fchar is the characteristic frequency.
In a study, Stulen et al.[63], concluded that the ratio parameter is the most
sensitive to changes in the conduction velocity but the least reliable when compared
with mean frequency and median frequency. This study also indicated that the MF is
the least sensitive to noise and concluded that from a theoretical point of view, the
median frequency is the preferred parameter for tracking shifts in the spectrum of the
myoelectric signal during changes in conduction velocity [63].
15


3.
Rationale and Objectives
3.1 Spinal Stabilizing feedback control system
The spinal stabilizing system consists of three subsystems [46]: spinal
column, spinal muscles and neuromuscular control unit as shown in the figure below.
Figure 3.1 Spinal Stabilizing feedback control system.
The spinal column has two functions: structural and transducer [47], In the
structural function it provides mechanical stability and the transducer function
16


consists of generating feedback signals describing spinal posture, motions, loads, etc.
This transducer signals are generated by the mechanoreceptors embedded in the
ligaments. Depending on the feedback from the muscle spindles, golgi tendon organs
of the muscles and mechanoreceptors the neuromuscular control unit generates
muscle response pattern to activate and coordinate the spinal muscles in order to
provide mechanical stability.
The stability of spine can be defined as the ability of the spine to limit patterns
of displacement under loading conditions so as not to damage the spinal cord and the
associated structures [68]. Thus it prevents incapacitating deformity or pain due to
structural changes in spine.
The neuromuscular control unit provides for adequate and overall mechanical
stability of the spine. If the structural function of the spinal column is compromised
due to injury then the muscular stability is increased to compensate the loss. This
study concentrates on the effects if the transducer function of the ligaments of the
spinal column is compromised.
It is important to mention that a feline model is used in this study. Although
significant differences may exist between humans (biped, 5 lumbar vertebrae, gravity
vector parallel to the spine) and cats (quadruped, 7 lumbar vertebrae, gravity vector
perpendicular to the spine), the behavior of the viscoelastic laxity and reflexive
muscular activity is expected to be similar [22]. The scientific literature indicates that
animal models are appropriate for the skeletal muscle investigations, and one
17


advantage of using anesthetized animals is the ability to dissect involved muscle
groups and to control the stimulation [70], such that system identification procedures
could be used to gain insight on the functions and properties of the system under
various normal and abnormal conditions.
18


4.
Methods and Procedures
4.1 Electromyogram Data Acquisition, Processing and Storage
The path of digital signal processing, as shown in the figure, from its source to
its final presentation includes transducers to measure the signal, a data acquisition
system, digital processing and final storage of the processed data.
Signal
Source
Figure 4.1 Data Acquisition System.
The EMG signal is measured by applying electrodes to the skin surface or
invasively within the muscle. In this study bipolar wire electrodes were used. The
19


electrodes acts as sensors and after differential pre-amplification the measured signal
is fed to the data acquisition system (DAC).
The DAC performs signal conditioning like further differential amplification
providing high input impedance, filtering, isolation, sample and hold, current to
voltage conversion, linearization, etc. The output from the signal conditioning block
is given to the analog to digital converter (ADC) where the analog voltage is
converted to digital signal that is fed to the personal computer (PC) for further
processing and storage. Oscilloscope is used to display real time analog signal
collected from the sensors.
4.1.1 Factors Influencing EMG Signal
The EMG signal can be affected by several external factors which can change
its shape and characteristics. The amplitude of the EMG signal varies from the uV to
the low mV [7]. The amplitude and frequency of the EMG depends on the following
factors [25]:
1. The timing and intensity of muscle contraction.
2. The type of electrode used like surface or intramuscular and the electrode
placement like the distance of electrode from active muscle area.
3. In case of surface EMG, the quality of contact between the electrode and the
skin.
20


4. The properties of the overlying tissue for example the thickness of the
overlying skin and adipose tissue in case of surface EMG.
5. Physiological cross talk. Local electrode can pick up EMG signal from
neighboring muscles.
6. The amplifier properties.
7. External noise like noisy electrical environment (noisy machines, instruments,
etc.) and direct interference of the power hum caused by incorrect grounding
of other external devices. Ambient noise can also affect the EMG signal.
In this study the timing and intensity of muscle contraction is desired. The
variability of the EMG due to remainder of the factors can be reduced by using
identical signal conditioning parameters within subjects like same electrodes and
amplifier, by reducing noise and filtering the EMG signal and ensuring consistency in
the quality of contact between the electrode and the skin in case of surface
measurement. Moreover the ground or reference electrode is placed away from the
recording electrode. Usually an electrically unaffected area is selected, such as
gluteus muscle.
The following sections outlines the factors influencing the characteristic of the
EMG signal and describes the techniques used in this study to minimize their affect
and increase the accuracy of the EMG signal.
21


4.1.2 Differential pre-amplifier
The goal with EMG measurement is to minimize the noise and increase the
signal to noise ratio. In this study bipolar electrode arrangement was used with a
differential pre-amplification. The differential amplifier suppresses the common
mode signal between the electrode pair and amplifies only the differential voltage. In
this study each electrode pair constituted the input to a differential amplifier with a
110 db Common Mode Rejection Ratio (CMRR), a gain of up to 200,000 and input
impedance in the range of mega ohms.
Correlated signals common to both electrodes like interference from power
sources and electromagnetic devices as well as the EMG signal from more distant
muscles are suppressed by differential amplifier. The signals from the muscle tissue
close to the electrodes are not correlated and are amplified by the differential
amplifier [25].
4.1.3 Signal Conditioning and Amplification
The bipolar arrangement with differential pre-amplification increases the
spatial resolution and makes it possible to measure full effective bandwidth of the
EMG signal. The power density spectra of the EMG contains most of its power in the
frequency range of 5-500 Hz. High pass filtering is required because movement
artifacts are comprised of low frequency components, typically below 10 Hz. Low
pass filtering is required to remove high frequency components to avoid aliasing [25].
22


For surface recording the filtering of the EMG in the band of 10-350 Hz is preferred.
In case of intramuscular recording the high frequency cut-off is increased and a band
pass filter of 10-450 Hz is required while for needle recording a bandwidth of 10-
1,500 Hz is required. The minimal sampling rate using the Nyquist criteria should be
at least twice the highest frequency cut-off of the band pass filter, e.g., if a bandpass
filter of 5-500 Hz was used, the minimal sampling rate employed to store the signal in
the computer should be 1000 Hz (2 x 500) or higher. Notch filtering to remove power
line (A/C) noise components like 60 Hz should be avoided since EMG signal has
large signal contributions at these and neighboring frequencies.
To optimize the resolution of the digitizing equipment amplification is
necessary [25]. Amplifiers should have adjustable gains of between 100-5000 to
maximize the signal to noise ratio of the EMG signal during each recording.
4.2 Discrete Fourier Transform
The analysis of EMG signal in frequency domain involves discrete Fourier
transform (DFT) of the digitized EMG signal to study its frequency spectrum. Median
frequency of the power density spectrum is used as a parameter of study here, which
provides a useful measure of the EMG frequency spectrum.
The discrete Fourier transform of the data sequence x(n) is,
x(k) = ^x(nW k = 0, 1, ..., N-1 (4.1)
rt~ 0
23


-A2*/)
Where, WN = e /N [45], The inverse discrete Fourier transform (IDFT) is,
x{n) = ]~Y,X(kW^, = 0, 1, ..., N-1 (4.2)
In equations (4.1) and (4.2), both x (n) and X (k) may be complex. DFT and
IDFT involve generally same type of computations. It can be noted from equation
(4.1) that for each computation of Fourier coefficient x (k), N complex
multiplications (4N real multiplications) and N-l complex additions (4N-2 real
additions) are required. Thus for all N values of DFT requires N2 complex
multiplications and N2 N complex additions. As a result procedures that reduce
the number of multiplications and additions are necessary. The computation of DFT
can be done efficiently by exploiting the following special properties of the
quantities :
1. Symmetry property: W^+N'2 =-JFy
2. Periodicity property: W*N =
The algorithms which exploit these two basic properties of the phase factor
are collectively known as fast Fourier transform (FFT) algorithms.
4.2.1 Radix-2 FFT Algorithms
In order to increase efficiency in computing DFT it is necessary to decompose
the DFT computation into successively smaller DFT computations. Algorithms which
24


decomposes the sequence x (n) into successive smaller subsequences are called
decimation-in-time algorithms. It can be illustrated by considering the case of
computing N = 2V point DFT. The N-point data sequence is separated into two N/2-
point data sequences /, () and J2{n), consisting of the even-numbered and odd-
numbered samples of x (n), respectively, that is,
fl(n) = x(2n) (4.3)
fl(n) = x(2n + l), s ll 0 K>\ 1 (4.4)
The N-point DFT can be expressed as,
x{k) = Yx(n)]VN, k=Q 1, , N-\
n even n odd
(NA i
= £x(2 r)W+ £x(2r + l)^1)* (4.5)
r=0 r=0
But W = WN/2, substituting this, equation (4.5) becomes,
{"/U Hh
X(k)=
r=0 r=0
= Fx(k) + KF2(k), k = 0, 1, ..., N (4.6)
Where Fx(k) and F2(k) are the N/2- point DFTs of the sequences fx(r) and f2(r),
respectively.
25


Since Fx(k) and F2(k) are periodic with period N/2 and in addition the phase
factor W+N'2 = -W^ the equation (4.6) can be expressed as,
X(k) = Fl(k) + WkF1(k),
k = o, i,
(4.7)
a
k +
N
= Fx(k)-WkNF2(k),
N
k = 0, 1, ..., --1
2
(4.8)
Now for the direct computation of Fx(k) andF2(k), each requires (N/2)2
complex multiplications. Moreover there are N/2 additional complex multiplications
to compute W^F2(k). Hence the computation of X (k) requires ^^ + ^2 comP^ex
multiplication. This results in a reduction of the number of multiplications from (N)2
to^2/ + N/} which is about a factor of 2 for large N.
Since N is equal to a power of 2, N/2 is even. Two N/4-point DFTs are
obtained by breaking each of the sums of equation (4.6) and the N/2-point DFTs are
computed as indicated below,
[N/M


£/,wA2 = I/,(2/)^?2+ Z/.(2/+1) 0*
r=0
1=0
1=0
Or
F,(k)= £/,(2/)'4+<2 £/i(2/ + 1)<*
/ 4
(4.9)
1=0
1=0
And similarly,
26


(4.10)
(NAU (%}>
F2(k) = £/2(2/) /=0 7=0
This decimation of the data sequence is repeated again and again until the
sequences are reduced in to one-point sequences. For ./V = 2V, this decimation can be
performed v = log2 N times and thus the total number of complex multiplications is
reduced to(J^)log2 N This is illustrated in figure 4.2 for eight-point DFT.
x(O)
x(4)
*(2)
*(6)
*0)
x(5)
x(3>
x(7)
X(0>
*0)
X(2)
*<3)
X X{5)
X(6)
X{7)
Figure 4.2 Computation of N = 8 point DFT.
27


Stage 1
Stage 2
Stage 3
JT{0)
*(1)
X{2)
XO)
X{4)
*(5)
X(6)
XO)
Figure 4.3 Flow graph of complete decimation in time decomposition of N = 8 point
DFT.
Figure 4.4 Flow graph of a basic two-point DFT.
28


The input data is stored in bit -reversed order so that the computation may be
done in place so that the result of the newly computed array is stored in the same
storage locations as the original array [45]. This bit reversal is illustrated in figure 4.5
29


Decimation 1
Decimation 2
Memory Address
(decimal) (binary)
0 000
1 00 1
2 010
3 Oil
4 1 00
5 10 1
6 110
7 111
Natural
Order
Bit-reversed
Order
(a)

(0 0 0) (0 0 0) - (0 0 0)
(0 0 1) (1 0 0) - (1 0 0)
(0 10) (0 0 1) (0 10)
(0 1 1) (101) (11 0)
(10 0) (01 0) (0 0 1)
(10 1) (1 10) -* (10 1)
(1 10) * (0 1 1) (0 1 1)
(111) - (111) (111)
(b)
Figure 4.5 Bit reversed storage of input data.
30


4.3 Spectral Analysis of deterministic signals
The mathematical tools for spectral analysis for a deterministic signal models
are the Fourier series and Fourier Transforms as described in the previous section.
The steps for DFT-based Fourier analysis for continuous time signals [38] are
illustrated in figure 4.6.
sc(t) xe(0 x(n)
Figure 4.6 Illustration of discrete Fourier transform based Fourier analysis for
continuous time signals.
As illustrated in figure 4.6 the continuous signal is first digitized and a
sequence of samples is obtained. These samples are grouped in a fixed length (N)
data segment and windowed for further processing. Then the spectrum is computed at
the desired set of frequencies and its DFT is implemented using some efficient
implementation like radix-2 FFT algorithms.
31


4.3.1 Windowing
Collection of finite duration data segment is equivalent to multiplying the
signal by a rectangular function of unit amplitude with the duration equal to the
length of the segment. This rectangular function is called a rectangular window. The
window function can be any arbitrary finite-duration sequence.
The rectangular window function results in discontinuities at the ends of the
data segments and gives rise to side lobes in the Fourier transform which is also
called spectral leakage. The effects of spectral leakage can be reduced by multiplying
the data segment with some function that smoothly reduces the signal to zero at the
end points. This process of tapering the data segment smoothly to zero by a function
is called windowing.
The window function does not eliminate the leakage entirely. They only
change the shape of the leakage. The height of the side lobe is diminished at the
expense of a wider main lobe. Many different windows have been proposed over
time, each with its own advantage and disadvantage. Some are more effective for
specific types of signals types such as random or sinusoidal. Some improve the
frequency resolution while some improve the amplitude accuracy. The selection of
window function depends on specific application. The most commonly used window
functions are defined below [45].
32


Rectangular:
H'() = 1, 0 < n < N \
Bartlett:
w(n) =
2 n
N-\
0 < n <
N-\
2
N -1
n
< n < N -1
Hanning:
w(n) = -
r
1 cos
2m
N-\
0 < n < N -1
Hamming:
w(n)
0.54-0.46 cos
^ 2m '
vA^I /
0 < n < N -1
Tukey:
w{n) =
1 ( 2n n 1
1 + cos
2 - l r N-1 J
1,
2
1 (2n 2n n -1 A
1 + COS -71
2 { r r N -1 )
<^(A^-l)+l
2

Figure 4.7 illustrates various windows functions.
(4.11)
(4.12)
(4.13)
(4.14)
-1) (4.15)
33


Time domain
Figure 4.7 Commonly used window functions: rectangular, Bartlett, Hamming,
Hanning and Tukey window with N = 64 samples and Tukey window with ratio of
taper to constant sections, r = 0.25.
4.3.2 Zero Padding
Zero padding is used for many purposes including augmentation of the
sequence length so that the length N can be made power of 2 and radix-2 FFT
algorithm can be used. Generally zeros are padded at the end of segmented data.
However zero padding does not increase the resolution of the spectrum but provides a
better display by providing a high density spectrum [38],
34


4.3.3 Overlap processing
The end points of the signal are attenuated in the calculation of the spectrum
due to windowing functions. As a result more averages must be taken to get a good
statistical representation of the spectrum. Overlap processing is a feature which can
recover lost data. Each frame of finite length data segment is overlapped by selecting
frame overlap length N0.
4.4 Preparation
Twenty adult cats (weight 3.81kg 0.44) were used in this study. They were
anesthetized with chloralose (60 mg/kg), according to the protocol approved by the
Institutional and Use Committee (IACUC). Chloralose is an anesthetic which does
not inhibit muscle activity. A booster injection was given whenever the depth of
anesthesia was insufficient by testing eye reflexes. The surgical procedure composed
of exposing the lumber fascia by dissecting the skin over the lumber spine. The
vertebral processes were marked to guide during the placement of electrodes. The S
shaped hook was inserted around the supraspinous ligament between L-4 and L-5.
The cat was placed on a rigid stainless steel frame and fixed for EMG electrodes
insertion. A static load was applied on the S shaped hook during the work period. The
temperature of the feline model was maintained by a heating pad. A gauze pad soaked
with saline was applied over the incision during the experiment to prevent the
exposed tissue from drying. The pulse rate, functional oxygen saturation of arterial
35


hemoglobin (Sp02) reading was monitored by Rad-5 Handheld Pulse Oximeter.
Ventilation was given by a ventilator (Inspira AVS). The pulse rate, ventilation rate,
Sp02 reading and temperature was recorded every half hour.
4.5 Instrumentation
The following instrumentation was applied:
1) The lumber spine was first isolated by means of two external fixators applied
to L-l and L-7 posterior processes. The external fixation was intended to limit
the elicited flexion to the lumbar spine and to prevent interaction of thoracic
and sacral/pelvic structures.
2) Six pairs of fine stainless steel wire EMG electrodes were inserted in the right
L-l/2, L-2/3, L-3/4, L-4/5, L-5/6, L-6/7 multifidus muscles.
3) A ground electrode was inserted into the gluteus muscles.
Each electrode pair constituted the input to a differential amplifier. The
differential amplifier constituted of 110-db common mode rejection ratio, a gain up to
200,000 and a band pass filter with a range of 5-500 Hz. High pass filtering is
required because movement artifacts are comprised of low frequency components,
typically below 10 Hz. Low pass filtering is required to remove high frequency
components to avoid aliasing [25]. The EMG signal measured by the electrodes was
stored in computer after digitizing it by an analog to digital card with a sampling rate
of 1 kHz. The EMG was also monitored simultaneously on an oscilloscope.
36


The S shaped hook inserted around the L-4/5 supraspinous ligament was
connected to the vertical actuator of the Bionix Material Testing System (MTS Inc.,
Minneapolis, MN, USA.). A load cell located at the bottom end of the crosshead of
the MTS system measured the load applied to the hook. The load was applied through
the MTS actuator with a computer controlled closed loop loading system operating in
a load control mode. The loading and the vertical displacement were continuously
monitored and sampled along with the EMG signal.
4.6 Experimental Protocol
The experiments were divided into three groups and were subjected to the
protocol as described below. The initial conditions in all the three groups of
experiments were standardized by a pre-tension of 1 N to the supraspinous ligament
[20]. A static load of 40 N applied to the lumbar spine via the S shaped stainless steel
hook to elicit a moderate lumbar flexion. The static load was applied for 6 working
periods. Each working period of 10-min was followed by a rest period of different
duration. The first group was subjected to rest duration of 5-min between each work
period (6 x 10:5 group, N = 7). The second group was subjected to a rest duration of
10-min between each work period (6 x 10:10 group, N = 6). While the third group
was subjected to a rest duration of 20-min between each work period (6 x 10:20
group, N = 7). The EMG signal, the vertical displacement and the load were recorded
during the working periods. The 6 working periods were followed by 7-h of recovery
37


with the spine in a neutral/rest position. Nine 8-s loading tests were performed during
the following of 7 h of recovery. This was obtained by a linear increase in tension to
40 N over 6 s followed by 2 s of constant load. This linear increase in load over 6 s
was used in order to avoid possible damage to the ligaments due to a sudden or fast
stretch [49], In each of the three experimental groups, the respective load was also
applied with a 6-s ramp in the initial loading of the six 10-min working periods.
The EMG, load and supraspinous ligament displacement data were stored in a
PC with a sampling rate of 1000 Hz for subsequent analysis.
Figure 4.8 Experimental set up showing the S shaped hook inserted around the L-4/5
supraspinous ligament of a cat with external fixation and bipolar wire electrodes
inserted to record data.
38


Load
Cell
Ex-Fix
ISL SSL
1\ N 1\
djbto
1\ 1\
Ex-Fix
mm

1
Load
Figure 4.9 Schematic representation of the experimental set-up showing the lumbar
spine with external fixation and loading apparatus. Ex-Fix: external fixator.
Two external fixators were used to isolate the lumbar spine and one was applied
to the L1 posterior spinal process while the other one was applied to the L7 process.
The external fixation was intended to limit the elicited flexion on the lumbar spine
and to prevent interaction of thoracic and sacral/pelvic structures.
39


4.7 Data Analysis
The digitized EMG signal data was stored in PC in the form of ASCII files,
each containing data of 1-min. Each file stored 6 EMG channels, a load channel and a
displacement channel. The EMG signal data was subsequently filtered through a high
pass second order Butterworth filter with cut off frequency of 20 Hz. In order to make
sure that the load was fully applied, the first 2 s of the constant load was discarded
and the analysis was performed over the following 1 s.
One second windows of the resultant EMG data, from L-3/4, L-4/5 and L-5/6,
was first multiplied by a Tukey window [45], because it did not attenuate any
significant part of the signal compared with Hamming or Hanning windows. The ratio
of taper to constant sections was set to 0.25. Once the EMG was windowed, zero
points were added to yield 1024 points of data to result in a spectral resolution of
0.9766 Hz in the final analysis. The fast Fourier transform of the resultant data was
then taken, and the power density spectrum was calculated [45]. Additional
calculations identified the Median Frequency of the resultant data from the following
equation,
mf oo oo
]p(f)c/f = \P(f)df = y2 \P(f)df (4.16)
0 mf 0
Where P (f) is the power density spectrum and f is the frequency.
Thus the median frequency is the frequency which divides the power
spectrum in half.
40


The attenuation of the signal at the beginning and end of the window, in the
calculation of the spectrum was avoided by overlap processing. The window was
moved half a second and the next 1 s of EMG data was analyzed.
The curve of all the MFs over the six 10 min constant load periods was
smoothed by taking moving average of 100 points with a shift of 39 points and
normalized with respect to the average of the first 100 points computed for the first
10 min constant load. All the corresponding normalized moving average MF data
were pooled together and the mean and standard deviation (SD) values were
computed for each of the muscles of the three lumbar levels investigated.
41


5.
Results
In general, the recorded data demonstrated that the EMG activity decreased
gradually during the six 10 min constant load period for all the preparations exposed
to different rest periods between 10 min of constant loadings.
Figure 5.1, 5.8 and 5.15 show typical EMG responses from each of the 6
channels as well as tension and displacement for the six 10 min constant load period
and 7 hours of recovery for each of the experimental groups (6x 10:5, 6 x 10:10, 6 x
10:20), respectively.
42


10-16-2003 (6x10/5)
0
2
LU
0
2
LU
0
2
LU
0
2
0
2
0
2
Hi |i|' |i , H 1 1 t t k i
+++ff
ILM u,, , ,
.PI r 1 tttt
-U4i, li || i
n w
10 20 30 40 50 60 70 80 1/6 1/2 1 2 3 4 5 6 7
Time (min) Time (hr)
Figure 5.1 EMG responses from the 6 channels, force and displacement channel
during six 10 min constant loading period and 7 hours of recovery. The constant
loading periods subjected to 40N with 5 min of rest period between them.
Figures 5.2,5.3, 5.4, 5.5, 5.6 and 5.7 shows the EMG response and respective
median frequency curves of different loading at lumbar levels L-3/4, L-4/5, L-5/6
respectively, where the preparation was subjected to 5 min of rest period between
each constant loading periods.
43
L-4/5 L-4/5 L-6/7 L-5/6 L-4/5 L-3/4 L-2/3 L-1/2


6x1 0:05 1016CH3
Figure 5.2 L-3/4 EMG response during six 10 min constant loading period subjected
to 5 min of rest period between them.
6x10:05 1016CH3
Figure 5.3 Moving average MF values at lumbar level L-3/4 during 6 loading periods
of constant load.
44


6x1 0:05 1 01 6CH4
Figure 5.4 EMG response at lumbar level L-4/5 during six 10 min constant loading
period subjected to 5 min of rest period between them.
I
£
6x10:05 1016CH4
01 23456789 10
Time (min)
Figure 5.5 Moving average MF values at lumbar level L-4/5 during 6 loading periods
of constant load.
45


6x10:05 1016CH5
>
E,
O
2
LU
hAli ki Ji 1- IL
Y if
0 20 40 60 80
Time (min)
Figure 5.6 EMG response at lumbar level L-5/6 during six 10 min constant loading
period subjected to 5 min of rest period between them.
6x10:05 1016CH5
Figure 5.7 Moving average MF values at lumbar level L-5/6 during 6 loading periods
of constant load.
46


10 30 2002 (6 x 10 / 10)
Ik-ZZ. 1L 4 U ,1 ,J| ]k Jl Jl
f r >* ^TT+Tfffff
jfchjLjL^_^_^k_ _j^i_ _^li_ _JIl_ _^b1_
1 jW^~~~^ r t "if Pi
jLd J^ jyi
^P ^P W ^T
! I...............................................II-M---n--II-II--1'-11--1,-,
m m jog |t^ n -|r 7J- ~w~ ~w~ jj- w~ ~jj" "X"
W w ^
' ' I ' I ' ' I '.....I--1 1 II II-1 I-11-1 I-II-'-1
lliL^u M A *to ^ d A. Jl lk Vk
PPW W^ WW W ^ "V PP "^p
*> 2 T
E
8
i i 11 11 11 ir
<1 |p |* h* M- r f" LC r r (1 ,/l ,f\ A /i ,/i a ,/i n,
0 1 0 20 3 0 4 0 5 0 6 0 7 0 8 0 9 0 1C I r )0 110 mmui 1/6 1/2 1 2 3 4 5 6 7
Time (m)
Time (hr)
Figure 5.8 EMG responses from each of the 6 channels, force and displacement
channel during six 10 min constant loading period and 7 hours of recovery. The
constant loading periods subjected to 40N with 10 min of rest period between them.
Figures 5.9,5.10, 5.11, 5.12, 5.13 and 5.14 shows the EMG response and
respective median frequency curves of different loading at lumbar levels L-3/4, L-4/5,
L-5/6 respectively, where the preparation was subjected to 10 min of rest period
between each constant loading periods.
47
L-4/5 L-4/5 L-6/7 L-5/6 L-4/5 L-3/4 L-2/3 L-1/2


6x1 0:1 0 1 0 3 0 C H 3
2.0 -r
1 .5 -
*1.5 11 i i i i i
0 20 40 60 80 100
T im e (min)
Figure 5.9 EMG response at lumbar level L-3/4 during six 10 min constant loading
period subjected to 10 min of rest period between them.
220
210
I 200
£
<
p
190
180
170
6x10:10 1030CH3

0-10
20-30
40-50
60-70
80-90
100-110


*4;..:


- -4----
'4**-
160
4 5 6
Time (min)
10
Figure 5.10 Moving average MF values at lumbar level L-3/4 during 6 loading
periods of constant load.
48


6x1 0:1 0 1 0 3 0 C H 4
Figure 5.11 EMG response at lumbar level L-4/5 during six 10 min constant loading
period subjected to 10 min of rest period between them.
6x10:10 1030CH4
200
195 -
_ 190
£
f 185 -
< 180 -
P
^ 175 -
170 -



0-10
20-30
40-50
60-70
80-90
100-110

V "
' ,4
. /
Xx
165
4 5 6
Time (min)
10
Figure 5.12 Moving average MF values at lumbar level L-4/5 during 6 loading
periods of constant load.
49


6x10:1 0 1030CH5
2.0
-1 5 -
-2.0 A--------------------1------------------1------------------1------------------r
0 20 40 60 80 100
Tim e (min)
Figure 5.13 EMG response at lumbar level L-5/6 during six 10 min constant loading
period subjected to 10 min of rest period between them.
6x10:10 1030CH5
Figure 5.14 Moving average MF values at lumbar level L-5/6 during 6 loading
periods of constant load.
50


10-23-2003 (6x10/20)
I oU
§ -iFr LU 11 ! i " ** ^ ** ** ^ 1* m
i :HI i -HL I I < | f I I 1 M I i i MUM ^llf|4Ml*4Nlg ''' i'''' i'' i i1 '' m ''' i' ''' i' '' i''1' i 1 11 11'''' i''' i i v ii- !( ii' if' 1 ii 1 *'ir~ir"
i :Sjl i -,;HL 1 1 1 H Nt i
t* M 1 N INI 1 I 3
I -Unn?
1 :|L 1 -'HL
1 20 a. CA Q 0 '' n 16s .G Q fl riAAAAAAAAA s
50 t Z """" 5* 25 3 n n n n/i/i/i/i/i/i/i/i/i *
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 1/6 1/2 1 2 3 4 5 6 7
Time (mm) Time (hr)
Figure 5.15 EMG responses from each of the 6 channels, force and displacement
channel during six 10 min constant loading period and 7 hours of recovery. The
constant loading periods subjected to 40N with 20 min of rest period between them.
Figures 5.16,5.17, 5.18, 5.19, 5.20 and 5.21 shows the EMG response and
respective median frequency curves of different loading at lumbar levels L-3/4, L-4/5,
L-5/6 respectively, where the preparation was subjected to 20 min of rest period
between each constant loading periods.
51
L-4/5 L-6/7 L-5/6 L-4/5 L-3/4 L-2/3 L-1/2


6x1 0:20 1 0 2 3 C H 3
2.0
0 20 40 60 80 100 120 140 160
Time (min)
Figure 5.16 EMG response at lumbar level L-3/4 during six 10 min constant loading
period subjected to 20 min of rest period between them.
6x10:20 1023CH3
Figure 5.17 Moving average MF values at lumbar level L-3/4 during 6 loading
periods of constant load.
52


6x1 0:20 1 023CH4
>
E.
O
2
LU
-3 -
-4 -
-5 H-----------1----------1-----------1----------1----------1----------1----------1----------
0 20 40 60 80 100 120 140 160
Time (min)
Figure 5.18 EMG response at lumbar level L-4/5 during six 10 min constant loading
period subjected to 20 min of rest period between them.
6x10:20 1023CH4
Figure 5.19 Moving average MF value at lumbar level L-4/5 during 6 loading periods
of constant load.
53


6x1 0 :20 C H 5
3
2 -
Time (min)
Figure 5.20 EMG response at lumbar level L-5/6 during six 10 min constant loading
period subjected to 20 min of rest period between them.
6x10:20 1023CH5
Figure 5.21 Moving average MF values at lumbar level L-5/6 during 6 loading
periods of constant load.
54


Typical EMG response to static loading demonstrates a gradual decrease in
subsequent loading cycles. Median frequency decreases during each 10 min of static
loading. The median frequency also decreases with subsequent loading cycles
demonstrating less and less recruitment of motor units. During the recovery, the peak
displacement gradually decreased demonstrating that the viscoelastic tissues were
recovering towards their resting dimensions.
5.1 EMG Responses
5.1.1 5 Minute Rest
Seven preparations were exposed to six loading periods of 10 min each with 5
min of rest between them. The average median frequency values of the EMG
spectrum of the multifidus muscles demonstrate a decrease in value for the three
lumbar levels over the period of six constant loading cycles of 10 minutes each with 5
minutes of rest between them.
The average median frequency values decrease, from 195.62 Hz to 163.31 Hz
for lumbar level L-3/4, 193.41 Hz to 166.69 Hz for lumbar level L-4/5, 184.36 Hz to
165.45 Hz for lumbar level L-5/6, for the first 10 min loading. This shows a decrease
of about 84, 87 and 90 percent for L-3/4, L-4/5, and L-5/6 respectively for first
loading cycle.
55


During the second loading cycle, the average median frequency values
decrease from 185.5 Hz to 152.94 Hz for lumbar level L-3/4, 187.93 Hz to 140.28 Hz
for lumbar level L-4/5, 168.72 Hz to 154.11 Hz for lumbar level L-5/6. It shows a
recovery of the average median frequency value of about 96 percent, 97 percent and
92 percent for L-3/4, L-4/5, and L-5/6 respectively during the 5 min rest period after
the first loading. At the end of the second loading the results show an overall decrease
of about 79, 74 and 84 percent for L-3/4, L-4/5, and L-5/6 respectively from average
median frequency values at the beginning of first loading cycle.
During the third loading cycle, the average median frequency values decrease
from 170.7 Hz to 145.81 Hz for lumbar level L-3/4, 169.14 Hz to 133.33 Hz for
lumbar level L-4/5, 151.21 Hz to 144.83 Hz for lumbar level L-5/6. It shows a
recovery of the average median frequency value of about 87, 88 and 83 percent for L-
3/4, L-4/5 and L-5/6 respectively during the 5 min rest period after the second
loading. The average median frequency values at the beginning of the first loading
decreases to 76, 71 and 80 percent for L-3/4, L-4/5, and L-5/6 respectively at the end
of third loading cycle.
The average median frequency values during the fourth loading cycle decrease
from 168.10 Hz to 145.5 Hz for lumbar level L-3/4, 162.27 Hz to 133.85 Hz for
lumbar level L-4/5, 149.03 Hz to 145.21 Hz for lumbar level L-5/6. It shows a
recovery of the average median frequency value of about 86, 84 and 82 percent for L-
3/4, L-4/5 and L-5/6 respectively during the 5 min rest period after the third loading.
56


The average median frequency values at the beginning of the first loading decreases
to 75, 71 and 80 percent for L-3/4, L-4/5, and L-5/6 respectively at the end of fourth
loading cycle.
The average median frequency values during the fifth loading cycle decrease
from 160.52 Hz to 145.76 Hz for lumbar level L-3/4, 152.61 Hz to 135.43 Hz for
lumbar level L-4/5, 147.7 Hz to 146.07 Hz for lumbar level L-5/6. It shows a
recovery of the average median frequency value of about 83, 80 and 81 percent for L-
3/4, L-4/5 and L-5/6 respectively during the 5 min rest period after the fourth loading.
The average median frequency values at the beginning of the first loading decreases
to 75, 72 and 81 percent for L-3/4, L-4/5, and L-5/6 respectively at the end of fifth
loading cycle.
The average median frequency values during the sixth loading cycle decrease
from 158.75 Hz to 146.11 Hz for lumbar level L-3/4, 154.17 Hz to 139.97 Hz for
lumbar level L-4/5, 148.87 Hz to 146.59 Hz for lumbar level L-5/6. It shows a
recovery of the average median frequency value of about 82, 80 and 82 percent for L-
3/4, L-4/5 and L-5/6 respectively during the 5 min rest period after the fifth loading.
The average median frequency values at the beginning of the first loading decreases
to 76, 74 and 81 percent for L-3/4, L-4/5, and L-5/6 respectively at the end of sixth
loading cycle.
The normalized average median frequency values at lumbar levels L-3/4, L-
4/5 and L-5/6 for six loading periods are tabulated as shown in tables A. 1 A.6.
57


Figure 5.22 5.24 shows the average median frequency values during six loadings
separated by 5 min of rest between them for lumbar levels L-3/4, L-4/5 and L-5/6
respectively.
The average median frequency decreased during the loading periods with
trends described by bi exponential fitting curve. The bi-exponential equation used is
indicated below:
y(t) = a.e~b' +c.e-dl (5.1)
Where, y (t) average median frequency value at time t
a, b, c and d parameter of the model
The average median frequency values were fitted with non-linear models. The
curves for the experimental group 6x10:05 for lumbar levels L-3/4, L-4/5 and L-5/6
are shown in figure 5.22, 5.23 and 5.24 respectively.
58


6x10:05 L-3/4
Figure 5.22 Average MF values at lumbar level L-3/4 during 6 loading periods of
constant load for 7 preparations subjected to 5 min of rest between the loading
periods.
59


6x10:05 L-4/5
Figure 5.23 Average MF values at lumbar level L-4/5 during 6 loading periods of
constant load for 7 preparations subjected to 5 min of rest between the loading
periods.
60


6x10:05 L-5/6
Figure 5.24 Average MF values at lumbar level L-5/6 during 6 loading periods of
constant load for 7 preparations subjected to 5 min of rest between the loading
periods.
The average median frequency decreased gradually for the first trial for
lumbar levels L-3/4, L-4/5 and L-5/6 for the group 6x10:05. The second trial after the
rest period shows that there was recovery in median frequency values as compared to
the end portion of first trial. In second trial the values decreases exponentially during
the loading period. The third, fourth, fifth and sixth trial shows corrupted muscle
response pattern since the curves shows fast saturation. In the third trial the median
61


frequency values goes to the saturation levels with in about 5 min of the loading
period and for forth, fifth and sixth trials the values saturates with in 4 min of the
loading. Thus it indicates that there was increased activity during the first, second and
first 4 min portion of the rest of the trials.
5.2.2 10 Minute Rest
Six preparations were exposed to six loading periods of 10 min each with 10
min of rest between them. The average median frequency values of the EMG
spectrum of the multifidus muscles demonstrate a decrease in value for the three
lumbar levels over the period of six constant loading cycles.
The average median frequency value decrease from 228.09 Hz to 213.4 Hz at
lumbar level L-3/4, 214.45 Hz to 198.7 Hz at lumbar level L-4/5, 198.21 Hz to
185.07 Hz at lumbar level L-5/6 for the first 10 min loading. This shows a decrease of
about 93, 93 and 93 percent for L-3/4, L-4/5, and L-5/6 respectively during first
loading cycle.
During the second loading cycle, the average median frequency value
decrease from 219.42 Hz to 210.69 Hz for lumbar level L-3/4, 205.03 Hz to 186.94
Hz for lumbar level L-4/5, 185.25 Hz to 174.45 Hz for lumbar level L-5/6. It shows a
recovery of the average median frequency value of about 96 percent, 96 percent and
94 percent for L-3/4, L-4/5, and L-5/6 respectively during the 10 min rest period after
the first loading. At the end of the second loading the results show an overall decrease
62


of about 92, 87 and 88 percent for L-3/4, L-4/5, and L-5/6 respectively from average
median frequency values at the beginning of first loading cycle.
During the third loading cycle, the average median frequency values decrease
from 216.95 Hz to 189.44 Hz for lumbar level L-3/4, 199.7 Hz to 186.06 Hz for
lumbar level L-4/5, 177.04 Hz to 169.04 Hz for lumbar level L-5/6. It shows a
recovery of the average median frequency value of about 95, 93 and 89 percent for L-
3/4, L-4/5 and L-5/6 respectively during the 10 min rest period after the second
loading. The average median frequency values at the beginning of the first loading
decreases to 83, 87 and 86 percent for L-3/4, L-4/5, and L-5/6 respectively at the end
of third loading cycle.
The average median frequency values during the fourth loading cycle decrease
from 219.51 Hz to 191.62 Hz for lumbar level L-3/4, 196.03 Hz to 180.11 Hz for
lumbar level L-4/5, 172.22 Hz to 167.39 Hz for lumbar level L-5/6. It shows a
recovery of the average median frequency value of about 96, 91 and 87 percent for L-
3/4, L-4/5 and L-5/6 respectively during the 10 min rest period after the third loading.
The average median frequency values at the beginning of the first loading decreases
to 84, 84 and 85 percent for L-3/4, L-4/5, and L-5/6 respectively at the end of fourth
loading cycle.
The average median frequency values during the fifth loading cycle decrease
from 217.93 Hz to 192.51 Hz for lumbar level L-3/4, 192.84 Hz to 177.73 Hz for
lumbar level L-4/5, 171.72 Hz to 164.07 Hz for lumbar level L-5/6. It shows a
63


recovery of the average median frequency value of about 95, 90 and 87 percent for L-
3/4, L-4/5 and L-5/6 respectively during the 10 min rest period after the fourth
loading. The average median frequency values at the beginning of the first loading
decreases to 84, 83 and 83 percent for L-3/4, L-4/5, and L-5/6 respectively at the end
of fifth loading cycle.
The average median frequency values during the sixth loading cycle decrease
from 217.02 Hz to 201.66 Hz for lumbar level L-3/4, 188.03 Hz to 178.85 Hz for
lumbar level L-4/5, 169.94 Hz to 169.15 Hz for lumbar level L-5/6. It shows a
recovery of the average median frequency value of about 95, 88 and 86 percent for L-
3/4, L-4/5 and L-5/6 respectively during the 10 min rest period after the fifth loading.
The average median frequency values at the beginning of the first loading decreases
to 88, 83 and 86 percent for L-3/4, L-4/5, and L-5/6 respectively at the end of sixth
loading cycle.
The normalized average median frequency values at lumbar levels L-3/4, L-
4/5 and L-5/6 for six loading periods are tabulated as shown in tables A.7 A. 12.
Figure 5.25 5.27 shows the normalized average median frequency values during six
loadings separated by 10 min of rest between them at lumbar levels L-3/4, L-4/5 and
L-5/6 respectively.
64


6x10:10 L-3/4
Figure 5.25 Average MF values of lumbar level L-3/4 during 6 loading periods of
constant load for 6 preparations subjected to 10 min of rest between each loading
period.
65


6x10:10 L-4/5
Figure 5.26 Average MF values of lumbar level L-4/5 during 6 loading periods of
constant load for 6 preparations subjected to 10 min of rest between each loading
period.
66


6x10:10 L-5/6
Figure 5.27 Average MF values of lumbar level L-5/6 during 6 loading periods of
constant load for 6 preparations subjected to 10 min of rest between each loading
period.
The trends of the average median frequency in the group with 10 min rest
periods show slow transition to saturation levels as compared to the group with 5 min
rest periods. This can be easily seen for lumbar levels L-3/4 and L-4/5.
67


5.1.3 20 Minute Rest
Seven preparations were exposed to six loading periods of 10 min each with
20 min of rest between them. The average median frequency values of the EMG
spectrum of the multifidus muscles demonstrate a gradual decrease in value for the
three lumbar levels over the period of six constant loading cycles.
The average median frequency value decrease from 209.32 Hz to 176.56 Hz at
lumbar level L-3/4, 177.57 Hz to 155.3 Hz at lumbar level L-4/5, 198.04 Hz to
173.77 Hz at lumbar level L-5/6 for the first 10 min loading. This shows a decrease of
about 85, 88 and 88 percent for L-3/4, L-4/5, and L-5/6 respectively during first
loading cycle.
During the second loading cycle, the average median frequency value
decrease from 206.34 Hz to 185.43 Hz for lumbar level L-3/4, 177.53 Hz to 154.12
Hz for lumbar level L-4/5, 188.94 Hz to 168.91 Hz for lumbar level L-5/6. It shows a
recovery of the average median frequency value of 100 percent, 100 percent and 95
percent for L-3/4, L-4/5, and L-5/6 respectively during the 20 min rest period after the
first loading. At the end of the second loading the results show an overall decrease of
about 90, 88 and 86 percent for L-3/4, L-4/5, and L-5/6 respectively from average
median frequency values at the beginning of first loading cycle.
During the third loading cycle, the average median frequency values decrease
from 206.69 Hz to 163.89 Hz for lumbar level L-3/4, 178.58 Hz to 155.74 Hz for
lumbar level L-4/5, 184.77 Hz to 163.69 Hz for lumbar level L-5/6. It shows a
68


recovery of the average median frequency value of about 100, 100 and 94 percent for
L-3/4, L-4/5 and L-5/6 respectively during the 10 min rest period after the second
loading. The average median frequency values at the beginning of the first loading
decreases to 80, 89 and 83 percent for L-3/4, L-4/5, and L-5/6 respectively at the end
of third loading cycle.
The average median frequency values during the fourth loading cycle decrease
from 204.09 Hz to 163.48 Hz for lumbar level L-3/4, 178.85 Hz to 151.23 Hz for
lumbar level L-4/5, 177.04 Hz to 163.51 Hz for lumbar level L-5/6. It shows a
recovery of the average median frequency value of about 100, 100 and 89 percent for
L-3/4, L-4/5 and L-5/6 respectively during the 20 min rest period after the third
loading. The average median frequency values at the beginning of the first loading
decreases to 80, 86 and 83 percent for L-3/4, L-4/5, and L-5/6 respectively at the end
of fourth loading cycle.
The average median frequency values during the fifth loading cycle decrease
from 203.36 Hz to 164.16 Hz for lumbar level L-3/4, 176.84 Hz to 150.14 Hz for
lumbar level L-4/5, 175.99 Hz to 161.11 Hz for lumbar level L-5/6. It shows a
recovery of the average median frequency value of about 100, 100 and 89 percent for
L-3/4, L-4/5 and L-5/6 respectively during the 20 min rest period after the fourth
loading. The average median frequency values at the beginning of the first loading
decreases to 80, 85 and 82 percent for L-3/4, L-4/5, and L-5/6 respectively at the end
of fifth loading cycle.
69


The average median frequency values during the sixth loading cycle decrease
from 194.99 Hz to 161.35 Hz for lumbar level L-3/4, 172.72 Hz to 147.68 Hz for
lumbar level L-4/5, 173.83 Hz to 155.67 Hz for lumbar level L-5/6. It shows a
recovery of the average median frequency value of about 96, 98 and 88 percent for L-
3/4, L-4/5 and L-5/6 respectively during the 20 min rest period after the fifth loading.
The average median frequency values at the beginning of the first loading decreases
to 78, 84 and 79 percent for L-3/4, L-4/5, and L-5/6 respectively at the end of sixth
loading cycle.
The normalized average median frequency values at lumbar levels L-3/4, L-
4/5 and L-5/6 for six loading periods are tabulated as shown in tables A. 13 A. 18.
Figure 5.28 5.30 shows the normalized average median frequency values during six
loadings separated by 20 min of rest between them at lumbar levels L-3/4, L-4/5 and
L-5/6 respectively.
70


6x10:20 L-3/4
Figure 5.28 Average MF values of lumbar level L-3/4 during 6 loading periods of
constant load for 7 preparations subjected to 20 min of rest between each loading
period.
71


6x10:20 L-4/5
Figure 5.29 Average MF values of lumbar level L-4/5 during 6 loading periods of
constant load for 7 preparations subjected to 20 min of rest between each loading
period.
72


6x10:20 L-5/6
Figure 5.30 Average MF values of lumbar level L-5/6 during 6 loading periods of
constant load for 7 preparations subjected to 20 min of rest between each loading
period.
The experimental group with 20 min rest period shows a muscle response
pattern in coordination with the force characteristics since all the trials show similar
trend and do not reach saturation. Moreover there is increased activity in all the six
trials as compared to the groups with 5 and 10 min rest periods.
73


5.2 Statistical Analysis
In order to determine if the trial number or the number of previously applied
working periods and the rest duration (5, 10 or 20 minutes) are the statistically
significant factors for the change in median frequency, two-way repeated measures
ANOVA (experiment procedure (3 levels) and trial (6 levels)) was performed using
MATLAB Statistics Toolbox (version 5). The number of working periods and the rest
duration effects were tested. Summary of the results obtained for the 5 min, 10 min
and 20 min rest durations for lumbar levels L-3/4, L-4/5 and L-5/6 are presented in
table 5.1, 5.2 and 5.3 respectively.
Tested Effects L-3/4 L-4/5 L-5/6
F value Pr> F F value Pr > F F value Pr > F
Rest Duration 5.7 <0.0001 5.16 <0.0001 0.76 0.7943
Trial Number 96.88 <0.0001 117.4 <0.0001 50.42 <0.0001
Rest duration Trial Number 0.48 1 0.68 1 0.26 1
Table 5.1 F values and Pr>F values obtained for the experimental group 6x10:05 for
6 loading trials for lumbar levels L-3/4, L-4/5, L-5/6.
T ested Effects L-3/4 L-4/5 L-5/6
F value Pr > F F value Pr > F F value Pr > F
Rest Duration 1.6 0.0351 0.97 0.5065 0.68 0.8768
Trial Number 15.25 <0.0001 24.92 <0.0001 26.27 <0.0001
Rest duration Trial Number 0.25 1 0.03 1 0.1 1
Table 5.2 F values and Pr>F values obtained for the experimental group 6x10:10 for
6 loading trials for lumbar levels L-3/4, L-4/5, L-5/6.
74


Tested Effects L-3/4 L-4/5 L-5/6
F value Pr > F F value Pr > F F value Pr > F
Rest Duration 7.91 <0.0001 4.53 <0.0001 4.73 <0.0001
Trial Number 21.49 <0.0001 8.79 <0.0001 41.48 <0.0001
Rest duration Trial Number 0.15 1 0.06 1 0.26 1
Table 5.3 F values and Pr>F values obtained for the experimental group 6x10:20 for
6 loading trials for lumbar levels L-3/4, L-4/5, L-5/6.
Based on the results of the statistical analysis both the number of working
periods (or the trial number) and the rest duration between each working periods
proved to be the affective factors for the change in the median frequency of the EMG
spectrum. Moreover there is no interactive effect of the two factors for all the
experimental groups.
For distinguishing among the statistically different trials in each experiment a
Tukey test was performed. The over all acceptable significance level of differences
for all statistical tests was set at P < 0.05. The Tukey test managed to distinguish the
first, the second and the rest of the working periods for the 6x10:05 group, while
revealing significant differences between first and rest of the working periods for
6x10:10 group for lumbar levels L-4/5 and L-5/6. It could not show significant
difference among trials for lumbar level L-3/4 for 6x10:10 group as well as for the
group 6x10:20 for lumber levels L-3/4, L-4/5, L-5/6.
75


6. Summary and Conclusions
Since the late 1950s, the creep and reflexive muscle activity behavior and
their recovery have been examined under several different conditions including
constant displacement and cyclic loading. Several hypotheses have attempted to
explain various clinical reports and research studies on the behavior of chronic low
back pain including a hypothesis of ligament subfailure injuries leading to muscle
control dysfunction by Dr. Punjabi [46]. This study analyzed the constant loading
conditions leading to ligament subfailure injuries and its effects on the median
frequency of the EMG spectrum due to corruption in the feedback loop in the spinal
stabilizing system.
Previous studies have shown that the viscoelastic creep loading increases the
laxity of the intervertebral joint causing loss of stability of the spine resulting in the
injury to the spine and low pack pain [1], [2] and [24], It was shown by Courville et
al. [14] that short rest period of 2:1 load to rest ratio leads to CLBD while longer rest
periods are necessary for preventing development of CLBD. The results of this study
confirm these previous findings for load to rest ratio of 2:1, 1:1 and 1:2.
It has been shown in previous studies that the frequency component of the
EMG power spectrum can be used as an indicator of the motor unit recruitment [42],
[57], Moreover it was shown by Solomonow et al.[57], that the changes in the firing
rate of the active motor units did not affect the median frequency of the EMG power
76


spectrum. It also showed a linear increase in median frequency with motor unit
recruitment. It has been indicated in previous studies that the average conduction
velocity of the active muscle motor units was highly correlated to twitch torque of the
active fibers and it was higher for fast twitch fibers then for slow twitch fibers [3]. It
was indicated by Arendt-Nielsen et al. [5], that there is linear relationship between the
mean power frequency of the EMG spectrum and muscle fiber conduction velocity.
Summing up all these findings it can be concluded that a shift in the power spectrum
to higher frequencies would indicate an increase in the average conduction velocity
and thus activation of faster motor units. This shift in the EMG power spectral
frequencies toward higher frequencies is represented by increase in median frequency
of the spectrum. Also the shift in the power spectrum to lower frequencies (decrease
in median frequency of the spectrum) would indicate a decrease in the average
conduction velocity and thus deactivation of fast motor units or a change in
recruitment pattern.
The literature also points out that the development of peripheral fatigue results
in reduction of conduction velocity of the motor unit action potential [34], [40], [37],
This reduction in conduction velocity due to fatigue results in reduction of median
frequency of the EMG power spectrum. Therefore two separate processes are
responsible for reduction in median frequency of the spectrum: recruitment changes
and fatigue.


It was seen in this study that the median frequency of the EMG spectrum
decrease during the loading trials of 10 min which indicates that during working
periods there is de-recruitment of faster motor units. The occurrence of fatigue may
also be the cause of this shift in the frequency content of the EMG power spectrum
towards lower frequencies. The faster motor units may be fatigued and stop firing
resulting in frequency characteristics of slower units. However median frequency
trend of gradual decrease, then increase and again gradual decrease in its values can
be seen in the first trials of all the experimental groups. All the trials of the
experimental group with load to rest ratio of 1:2 and first three trials of the
experimental group with load to rest ratio of 1:1 also show this type of trend. This is
more suggestive of recruitment changes then fatigue.
In this study the load applied to the ligament was 40 N which is intermediate
and it was distributed non-equally among 6 inter-vertebral sections. Thus the force
required to maintain the stability of the spine was even lower. Moreover total working
period was 1 h while in a study by Christensen et al. [13], [12], the muscle fatigue
was not seen during a whole day in EMG signal when applied loads were between
7% and 33% of the maximal voluntary contraction. In this study there was passive
loading and the level of the muscle contraction was relatively low since the activation
was reflexive. Solomonow et al., [60], stated that muscular activity diminished as a
result of desensitizing of the mechanoreceptors within the viscoelastic tissue of the
spine due to creep induced in it even before fatigue of the musculature. In fact they
78


showed that recalibrating the load/displacement after two hours of loading (two
repetitions of 50 min of continuous cyclic loading with 10 min of rest between
loadings), a high level of EMG discharge could be restored. If muscle fatigue was the
source of decreasing EMG, this could not be possible since in previous studies it was
shown that the amplitude of the extra cellular potential of a single nerve fiber depends
on the length of the depolarized zone [16] and its length increased while its amplitude
decreased in later stages of fatigue [29], [4], As a result it is assumed that fatigue in
the active motor units may not be the major factor in the shift of the frequency
spectrum of the EMG signal towards lower frequencies.
Development of laxity in the ligament results in reduction in its tension during
subsequent loading trials. This result in reduction in feedback signal via the
mechanoreceptors within the ligament, producing a weaker muscle response pattern
from the neuromuscular control unit which is indicative of lower force requirements
when the force requirements is same for all the trials. This will further result in lower
excitation of motor units producing a decrease in median frequency of the spectrum.
It was shown by Woo et al. [72], that changes in stress and motion
significantly changed the tissue properties as well as mass in the case of ligaments of
a rabbit. The ligament became more robust with increase in stress and mobility.
Humans are more active then a feline model which might suggest a difference in
tissue strength in sustaining loading conditions. However the neuromuscular system
of the feline and the human are similar [22], In a previous study on a feline model it
79


was shown that loads of 20 N and 40 N for load to rest ratio of 1:1 (10 min of flexion
with 10 min of rest sessions repeated 6 times) did not result in delayed hyper
excitability, which represents the muscles reaction to the development of an acute
inflammation in the micro damaged viscoelastic structures [52], [56], while load of 60
N resulted in delayed hyper excitability. Pervious literature also suggests that length
of loading period, amount of repetition, rest duration between working periods are
some of the factors affecting in the development of neuromuscular dysfunction with
their difference in magnitude among humans and feline models [14], [52].
This study indicated that for load to rest ratio of 2:1 the median frequency
values drop to constant level within 2 min of loading for third, fourth, fifth and sixth
trials while it was not the case for load to rest ratio of 1:1 and 1:2. The fast drop in
median frequencies suggested a corrupted (reduced) muscle response pattern which
resulted in significantly lower excitation of motor units via the reflex loop. During the
experiment, the most effective rest period was the load to rest ratio of 1:2 where the
median frequency values of the EMG spectrum decreased gradually and the trend was
similar for all the trials.
Previous studies have established that cumulative load exposure and tissue
overload in the lumbar spine leads to degeneration and injury of the spine and damage
to soft tissues, such as ligaments [33], [32]. The static loading causes subfailure injury
of the spinal ligaments and output distortion from the mechanoreceptors embedded in
the ligaments. The mechanoreceptors generate corrupted transducer signals which are
80


fed back to the neuromuscular control unit resulting in spatial and temporal mismatch
between the normally expected and the corrupted received signal [46], Moreover
laxity in the viscoelastic structures of the lumbar spine desensitizes the
mechanoreceptors within and causes loss of reflexive forces from the multifidus
muscles [60], Thus laxity in the ligament in subsequent trials also contributes to the
corrupted transducer signals. As a result the neuromuscular control unit generates a
corrupted muscle activation pattern affecting the spatial and temporal coordination of
each spinal muscle. This leads to corrupted feedback to the control unit via
mechanoreceptors and tendon organs of the muscles further corrupting the response
of the neuromuscular control unit resulting in the subfailure injury of the spinal
ligaments, mechanoreceptors and muscles. The abnormal stresses and strains
produced by corrupted muscle response results in inflammation of the spinal tissues
and over time back pain may develop [46].
Following the results it emerges that for the experimental group with load to
rest ratio 2:1, the rest duration of 5 min is not sufficient for recovery from the macro
damage caused on the mechanoreceptors in the ligaments. The first and the second
trial in this group the trend show that median frequency values decrease gradually but
during the rest of the trials the output of more and more mechanoreceptors are
corrupted and as a result the transducer signal generated by them are significantly
corrupted. Due to this corrupted feed back signal the muscle response pattern
81


generated by the neuromuscular control unit results in abnormality in motor unit
recruitment.
In case of experimental group with load to rest ratio of 1:2 there is enough rest
time for recovery from the macro damage on the mechanoreceptors. The
neuromuscular control unit generates normal muscle response pattern resulting in
normal motor unit recruitment in each of the six loading periods and as a result the
trend of the EMG power spectra median frequency is same for all the trials.
In conclusion this study provided experimental evidence showing the effect of
different rest periods on the motor unit recruitment activity during the development of
cumulative neuromuscular disorder. It also provides experimental evidence
supporting the hypothesis that the ligament subfailure injuries may lead to muscle
control dysfunction resulting further in low back pain.
82


Appendix
A. Tables
6 x 10:05 First Loading Time Normalized MF L-3/4 Normalized MF L-4/5 Normalized MF L-5/6
Mean SD Mean SD Mean SD
0.97 1 0 1 0 1 0
1.43 0.97 0.04 0.99 0.02 0.97 0.03
1.76 0.96 0.06 0.98 0.03 0.95 0.05
2.22 0.95 0.08 0.97 0.05 0.93 0.06
2.54 0.93 0.09 0.96 0.06 0.93 0.07
2.87 0.93 0.09 0.97 0.07 0.92 0.08
3.32 0.93 0.09 0.98 0.08 0.92 0.09
3.65 0.92 0.08 0.97 0.09 0.91 0.1
3.97 0.92 0.08 0.96 0.09 0.91 0.1
4.43 0.92 0.09 0.96 0.09 0.91 0.09
4.76 0.92 0.08 0.96 0.1 0.9 0.09
5.22 0.92 0.08 0.96 0.11 0.9 0.1
5.54 0.91 0.08 0.95 0.1 0.89 0.1
5.87 0.9 0.09 0.94 0.1 0.88 0.11
6.32 0.89 0.08 0.93 0.1 0.88 0.12
6.65 0.89 0.08 0.91 0.1 0.88 0.12
6.97 0.89 0.08 0.89 0.09 0.87 0.12
7.43 0.89 0.07 0.88 0.09 0.88 0.12
7.76 0.88 0.07 0.86 0.08 0.87 0.11
8.22 0.87 0.08 0.85 0.07 0.87 0.12
8.54 0.84 0.08 0.83 0.08 0.86 0.12
8.87 0.82 0.08 0.82 0.1 0.86 0.12
9.32 0.82 0.07 0.83 0.13 0.86 0.12
9.65 0.83 0.06 0.85 0.14 0.88 0.13
9.97 0.84 0.05 0.87 0.13 0.9 0.14
Table A.l Mean normalized MF values of the first loading period (0-10min) for the
group subjected to 5 min of rest between each loading period.
83


6 x 10:05 Second Loading Time Normalized MF L-3/4 Normalized MF L-4/5 Normalized MF L-5/6
Mean SD Mean SD Mean SD
15.97 0.96 0.1 0.97 0.1 0.92 0.08
16.43 0.94 0.08 0.93 0.13 0.89 0.11
16.76 0.91 0.07 0.91 0.14 0.88 0.12
17.22 0.88 0.06 0.88 0.13 0.87 0.12
17.54 0.86 0.07 0.85 0.1 0.86 0.12
17.87 0.86 0.09 0.82 0.09 0.85 0.11
18.32 0.85 0.08 0.81 0.1 0.84 0.1
18.65 0.84 0.08 0.8 0.11 0.84 0.11
18.97 0.83 0.1 0.78 0.11 0.84 0.12
19.43 0.83 0.11 0.77 0.1 0.84 0.12
19.76 0.82 0.11 0.77 0.11 0.84 0.12
20.22 0.82 0.11 0.76 0.11 0.83 0.12
20.54 0.81 0.11 0.75 0.11 0.83 0.11
20.87 0.81 0.11 0.75 0.11 0.83 0.11
21.32 0.8 0.11 0.75 0.11 0.82 0.11
21.65 0.8 0.1 0.75 0.11 0.82 0.11
21.97 0.79 0.1 0.75 0.11 0.82 0.11
22.43 0.79 0.09 0.75 0.11 0.83 0.11
22.76 0.79 0.1 0.75 0.11 0.84 0.12
23.22 0.79 0.11 0.74 0.11 0.85 0.13
23.54 0.79 0.11 0.74 0.12 0.85 0.13
23.87 0.79 0.11 0.73 0.12 0.85 0.12
24.32 0.79 0.11 0.73 0.13 0.84 0.12
24.65 0.79 0.11 0.73 0.12 0.84 0.12
24.97 0.79 0.11 0.74 0.12 0.84 0.13
Table A.2 Mean normalized MF values of the second loading period (15-25min) for
the group subjected to 5 min of rest between each loading period.
84


6 x 10:05 Third Loading Time Normalized MF L-3/4 Normalized MF L-4/5 Normalized MF L-5/6
Mean SD Mean SD Mean SD
30.97 0.87 0.11 0.88 0.19 0.83 0.09
31.43 0.85 0.12 0.84 0.22 0.8 0.11
31.76 0.83 0.1 0.81 0.2 0.8 0.12
32.22 0.81 0.1 0.78 0.18 0.8 0.12
32.54 0.79 0.09 0.75 0.15 0.8 0.12
32.87 0.77 0.1 0.73 0.14 0.79 0.12
33.32 0.76 0.1 0.72 0.14 0.79 0.12
33.65 0.76 0.09 0.72 0.14 0.8 0.12
33.97 0.76 0.09 0.71 0.16 0.8 0.12
34.43 0.76 0.1 0.7 0.17 0.8 0.12
34.76 0.76 0.1 0.7 0.17 0.8 0.12
35.22 0.76 0.1 0.71 0.15 0.8 0.12
35.54 0.76 0.1 0.72 0.14 0.8 0.12
35.87 0.76 0.1 0.71 0.15 0.8 0.12
36.32 0.76 0.1 0.71 0.16 0.8 0.12
36.65 0.76 0.1 0.71 0.15 0.8 0.12
36.97 0.76 0.1 0.71 0.15 0.8 0.12
37.43 0.76 0.1 0.71 0.15 0.8 0.12
37.76 0.76 0.1 0.71 0.15 0.8 0.12
38.22 0.76 0.09 0.72 0.14 0.81 0.12
38.54 0.76 0.09 0.72 0.14 0.8 0.12
38.87 0.76 0.1 0.72 0.14 0.8 0.12
39.32 0.75 0.1 0.71 0.15 0.8 0.12
39.65 0.76 0.1 0.71 0.16 0.8 0.12
39.97 0.76 0.1 0.71 0.16 0.8 0.12
Table A.3 Mean normalized MF values of the third loading period (30-40min) for the
group subjected to 5 min of rest between each loading period.
85


6 x 10:05 Fourth Loading Time Normalized MF L-3/4 Normalized MF L-4/5 Normalized MF L-5/6
Mean SD Mean SD Mean SD
45.97 0.86 0.12 0.84 0.15 0.82 0.11
46.43 0.83 0.13 0.79 0.16 0.8 0.12
46.76 0.81 0.13 0.75 0.16 0.8 0.11
47.22 0.79 0.12 0.73 0.15 0.8 0.11
47.54 0.77 0.1 0.71 0.15 0.8 0.12
47.87 0.75 0.1 0.71 0.15 0.8 0.12
48.32 0.75 0.1 0.71 0.14 0.8 0.12
48.65 0.75 0.1 0.71 0.16 0.8 0.12
48.97 0.75 0.1 0.71 0.15 0.8 0.12
49.43 0.75 0.1 0.71 0.16 0.8 0.12
49.76 0.75 0.1 0.71 0.15 0.8 0.12
50.22 0.75 0.1 0.71 0.15 0.8 0.12
50.54 0.75 0.1 0.72 0.14 0.8 0.12
50.87 0.75 0.1 0.72 0.16 0.8 0.12
51.32 0.75 0.1 0.73 0.2 0.8 0.12
51.65 0.75 0.1 0.73 0.23 0.8 0.12
51.97 0.75 0.1 0.72 0.21 0.8 0.12
52.43 0.75 0.1 0.71 0.17 0.8 0.13
52.76 0.75 0.1 0.71 0.15 0.8 0.13
53.22 0.75 0.09 0.71 0.15 0.8 0.13
53.54 0.75 0.09 0.7 0.16 0.8 0.12
53.87 0.75 0.1 0.71 0.16 0.8 0.12
54.32 0.75 0.1 0.71 0.16 0.8 0.13
54.65 0.75 0.09 0.71 0.15 0.8 0.13
54.97 0.75 0.1 0.71 0.15 0.8 0.12
Table A.4 Mean normalized MF values of the fourth loading period (45-5 5min) for
the group subjected to 5 min of rest between each loading period.
86