Citation
Comparative analysis of the effects of cyclic loading on spinal stability of the lumbar region

Material Information

Title:
Comparative analysis of the effects of cyclic loading on spinal stability of the lumbar region
Creator:
Ben-Masaud, AbdAllah
Publication Date:
Language:
English
Physical Description:
xvi, 100 leaves : illustrations ; 28 cm

Thesis/Dissertation Information

Degree:
Master's ( Master of Science)
Degree Grantor:
University of Colorado Denver
Degree Divisions:
Department of Electrical Engineering, CU Denver
Degree Disciplines:
Electrical engineering

Subjects

Subjects / Keywords:
Spine -- Instability ( lcsh )
Lumbosacral region ( lcsh )
Lifting and carrying ( lcsh )
Lifting and carrying ( fast )
Lumbosacral region ( fast )
Spine -- Instability ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 95-100).
General Note:
Department of Electrical Engineering
Statement of Responsibility:
by AbdAllah Ben-Masaud.

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:
436222407 ( OCLC )
ocn436222407
Classification:
LD1193.E54 2009m B46 ( lcc )

Full Text
COMPARATIVE ANALYSIS OF THE EFFECTS OF CYCLIC LOADING ON
SPINAL STABILITY OF THE LUMBAR REGION
by
AbdAllah Ben-Masaud
B.S., University of Colorado Denver, 2007
A thesis submitted to the
University of Colorado Denver
in partial fulfillment
of the requirements for the degree of
Masters of Science
Electrical Engineering
2009


This thesis for the Masters of Science
degree by
AbdAllah Ben-Masaud
has been approved
by
Date


Ben-Masaud, AbdAllah (M.S., Electrical Engineering)
Comparative Analysis of the Effects of Cyclic Loading on Spinal Stability of the
Lumbar Region
Thesis directed by Professor Hamid Fardi
ABSTRACT
The effects of cyclic lumbar loading were observed with specific regard to
neuromuscular neutral zones (NNZ) and the associated lumbar stability tracked from
recorded EMG responses of deep-tissue multifidus muscles. The aim of the study
was to obtain a quantitative assessment of the negative impact of cyclic loading for
comparative contrast of results from differing load magnitudes. Two groups of in
vivo feline specimens were loaded for 10 minutes at 0.25Hz cycles, one group having
20N peaks and the other 60N, followed by ten minutes of rest. This pattern was
repeated 5 more times for a culmination of 60 minutes of work with 60 minutes rest.
Each specimen was observed for 7 hours following the last work period to track
recovery progress. The first two responses at the start of the protocol were averaged
for an initial baseline control response. Following the completion of the prescribed 2
hour work/rest period, instability in the lumbar region was apparent for as much as 5
hours due to the cumulative damages of loading, indicating a greater potential for
injury within that region. This was justified by the assessment of increased creep and
NNZ along with decreased mean absolute voltage (MAV) and median frequency
(MF). All trends shifted gradually toward baseline values in the absence of loading.
The displacement NNZ began to exponentially decrease along with creep nearly
reaching baseline by the 7tfi hour of recovery. The tension NNZ however
dramatically decreased assuming baseline values by the 3rd to 4th hour of recovery and
continued to decrease beyond baseline for the remaining recovery period.
Conversely, the reduced MAV at the end of the work period exponentially increased,
reaching baseline values by the 3rd to 4th hour of recovery and continuing to increase
beyond baseline thereafter. MF behavior mirrored MAV trends. It is apparent that
immediately after the 2 hour work period the lower back was subject to instability due
to viscoelastic laxity and underperforming muscle support. By the 3rd to 4th hour of
rest however, hyperexcitability induced within the stabilizing musculature stiffened
the lumbar response to movement in overcompensation for the increased risk of
injury. The 60N study saw a significant increase in the level of compensation effort
versus the 20N, implying that greater load magnitudes result in greater sustained
damage to the loaded tissues and accordingly a higher level of instability.


This abstract accurately represents the content of the candidates thesis. I recommend
its publication.
Signed
Hamid Fardi


To Arezou
Thanks for keeping my nose to the grindstone ...
W


ACKNOWLEDGEMENT
All praise and thanks is due to Allah, who is capable of anything.
Thanks to Dr. Solomonow for his guidance and support in this endeavor. Doc. S is
the man! Along with Dr. Lu and Dr. Zhou, they not only gave me a spot in the lab
but also showed me the ropes and made sure I was on track.
To my advisor Prof. Hamid Fardi, thanks for assisting me through my undergrad and
graduate career. I appreciate the help and direction. Prof. Tim Lei, thanks for taking
the time to read my report and sit on my thesis committee.
Thanks Debbie! You did half my work for me plus verified the other half, Ill never
forget that. Not to mention youre a good cook. Middle Eastern food, yum! Jimmy
and Brook, thanks for walking me through everything I needed to do and mentoring
me. I built off all your guys work.
Thanks Brad for your advice and encouragement and pushing me to be better.
Big mama Maureen, thanks for taking care of me, I know I cant take care of myself.
Youre a really good person.
Everyone else in the lab, thanks!
Family, thanks!
Friends, thanks!
Zouzi-Q, thanks!


TABLE OF CONTENTS
FIGURES.....................................................................xi
TABLES......................................................................xv
Chapter
1. Introduction to Study....................................................1
1.1 Loading Types and Dangers:...............................................1
1.2 Clinical Disorders and Epidemiology:.....................................2
1.3 Focus of Study and Overview:.............................................3
1.4 Statement of Intent:.....................................................5
1.5 Hypothesis:..............................................................6
1.6 Significance of Study:...................................................7
2. Anatomy of the Lower Back Region.........................................9
2.1 Body Planes:.............................................................9
2.2 Lumbar Spine:............................................................9
2.3 Vertebrae:..............................................................10
2.4 Ligaments:..............................................................12
2.5 Intervertebral Discs:...................................................13
2.6 Paraspinal Muscles:.....................................................14
2.7 Mechanoreceptors:.......................................................15
2.8 Spinal Feed-back Control:...............................................16
vii


2.9 Action Potentials:
17
2.10 Motor Unit Recruitment and Firing Rate:................................18
2.11 Electromyography and Median Frequency:.................................20
2.12 Cytokines and Neutrophils:.............................................22
3. Procedure...............................................................24
3.1 Preparation:............................................................24
3.2 Instrumentation:........................................................26
3.3 Protocol:...............................................................27
4. Methods.................................................................30
4.1 EMG:....................................................................30
4.2 Creep:..................................................................31
4.3 Neuromuscular Neutral Zones:............................................32
4.4 Normalized Peak Mean Average Voltage:...................................36
4.5 Median Frequency:.......................................................37
4.6 Modeling:...............................................................40
4.6.1 DNNZ................................................................40
4.6.2 TNNZ................................................................41
4.6.3 PMAV................................................................42
4.6.4 MF..................................................................44
viii
4.7 Curve Fitting:
44


4.8 Statistical Variance:................................................45
5. Results..............................................................47
5.1 20NDNNZ:.............................................................48
5.1.1 Stretch Phase...................................................49
5.1.2 Relaxation Phase................................................51
5.2 60N DNNZ:..........................................................52
5.2.1 Stretch Phase...................................................53
5.2.2 Relaxation Phase................................................54
5.3 20NTNNZ:.............................................................55
5.3.1 Stretch Phase...................................................56
5.3.2 Relaxation Phase................................................57
5.4 60NTNNZ:.............................................................59
5.4.1 Stretch Phase...................................................60
5.4.2 Relaxation Phase................................................61
5.5 20NMAV:..............................................................63
5.6 60NMAV:..............................................................67
5.7 20NMF:...............................................................70
5.8 60NMF:...............................................................73
5.9 20N Creep:...........................................................76
5.10 60N Creep:.........................................................78
IX


6. Discussion
80
6.1 Spinal Instability:....................................................81
6.1.1 Increased Unprotected Neutral Zones...............................81
6.1.2 Deficient Muscle Response.........................................82
6.2 Compensation Mechanism:...............................................84
6.3 Comparison of Load Magnitude:..........................................86
6.4 Comparison with Previous Studies:......................................90
6.5 Transition to Human Models:............................................91
7. Conclusion.............................................................93
REFERENCES.................................................................95
x


LIST OF FIGURES
Figure
1.1 Viscoelastic creep developed from a single load cycle.....................5
2.1 Body planes (SEER 2008)...................................................9
2.2 Vertebral levels of the spine (SEER 2008)............................... 10
2.3 Oblique view of vertebral orientation and structures (Wikipedia 2009)... 11
2.4 Vertebral processes (Gray's Anatomy Fig. 93).............................11
2.5 Median sagittal cross-section of lumbar vertebral section showing
interconnecting ligament groups (Gray's Anatomy Fig. 301)...............12
2.6 Intervertebral viscoelastic discs (ADAM 2008)........................... 13
2.7 Paraspinal muscles embedded sagittally parallel to the spinal column (All
About Back and Neck Pain 2007).......................................... 14
2.8 Spinal feedback control system in the lower lumbar level (Solomonow et al.
2003)................................................................... 17
2.9 Motor unit representation showing an efferent axon connecting to muscle
fibers (Mosby 2009)..................................................... 19
2.10 Muscle twitch summation affected by firing rate toward tetanus (Marieb
2008)...................................................................20
2.11 Neutrophil concentrations in two ligament samples, the tissue on the left was
loaded whereas the control on the right was left unloaded (adapted from Le
2007)..............................................................23
3.1 Experimental setup of the lumbar level for the feline specimen...........25
xi


3.2 Stainless steel wire electrodes..........................................27
3.3 Experimental protocol applied to each specimen at respective load........29
4.1 Full 9 hour experimental EMG response for 6 channels.....................30
4.2 Flexion and extension phase patterns from the neutral position repeated in
cyclic work (O'Connor 2005)..............................................32
4.3 Full EMG signals during successive stages of signal processing finishing with
the consolidated down-sampled 800 point signal used in NNZ tracking......34
4.4 Zoomed-in representation of EMG stages during signal processing..........34
4.5 Typical threshold plot displaying the single cycle EMG for three channels and
the associated displacement and load plots...............................35
4.6 MAV represents total contraction of muscle over time.....................37
4.7 0.5 sec EMG extraction at peak tension for PSD analysis..................37
4.8 Zero padding the targeted EMG selection in SigmaPlot.....................38
4.9 Tukey trapezoidal window transform chosen with r=0.25 taper (green)......38
4.10 Raw MF data for all three channels during recovery.......................40
4.11 Smooth MF trend after running through 3 point moving average.............40
5.1 Composite DNNZ trend for 8 specimens at 20N during recovery period.......48
5.2 Percent change of 20N DNNZ stretch phase from baseline during recovery
period...................................................................50
5.3 Percent change of 20N DNNZ relaxation phase from baseline during recovery
period...................................................................51
xii


5.4 Composite DNNZ trend for 9 specimens at 60N during recovery period......52
5.5 Percent change of 60N DNNZ stretch phase from baseline during recovery
period..................................................................53
5.6 Percent change of 60N DNNZ relaxation phase from baseline during recovery
period..................................................................54
5.7 Composite TNNZ trend for 8 specimens at 20N during recovery period......55
5.8 Percent change of 20N TNNZ stretch phase from baseline during recovery
period..................................................................57
5.9 Percent change of 20N TNNZ relaxation phase from baseline during recovery
period..................................................................58
5.10 Composite TNNZ trend for 9 specimens at 60N during recovery period......59
5.11 Percent change of 60N TNNZ stretch phase from baseline during recovery
period..................................................................61
5.12 Percent change of 60N TNNZ relaxation phase from baseline during recovery
period..................................................................62
5.13 Composite MAV trend for 8 specimens at 20N during recovery period.......63
5.14 Percent change of 20N MAV stretch phase from baseline during recovery
period..................................................................66
5.15 Composite MAV trend for 9 specimens at 60N during recovery period.......67
5.16 Percent change of 60N MAV stretch phase from baseline during recovery
period..................................................................69
5.17 Composite MF trend for 8 specimens at 20N during recovery period........70
xiii


5.18 Percent change of 20N MF stretch phase from baseline during recovery period
72
5.19 Composite MF trend for 9 specimens at 60N during recovery period......73
5.20 Percent change of 60N MF stretch phase from baseline during recovery period
75
5.21 (A) 8 specimen average peak displacement illicited along the supraspinous
ligament during 20N cyclic loading and recovery periods
(B) corresponding creep accumulation during loading and recovery......76
5.22 (A) 9 specimen average peak displacement illicited along the supraspinous
ligament during 60N cyclic loading and recovery periods
(B) corresponding creep accumulation during loading and recovery......78
6.1 Consecutive 16 sec single cycle EMG response along L4-5...............83
6.2 DNNZ phases vs. creep for 20N cyclic loading..........................88
6.3 DNNZ phases vs. creep for 60N cyclic loading..........................88
xiv


LIST OF TABLES
Table
5.1 Function parameters for the fitted curves of 20N DNNZ model..........48
5.2 P values for the independent variables of the 2-way ANOVA test for 20N DNNZ
data and the results of the piece-wise tests........................48
5.3 Function parameters for the fitted curves of 60N DNNZ model..........52
5.4 P values for the independent variables of the 2-way ANOVA test for 60N DNNZ
data and the results of the piece-wise tests........................52
5.5 Function parameters for the fitted curves of 20N TNNZ model..........55
5.6 P values for the independent variables of the 2-way ANOVA test for 20N TNNZ
data and the results of the piece-wise tests........................55
5.7 Function parameters for the fitted curves of 60N TNNZ model..........59
5.8 P values for the independent variables of the 2-way ANOVA test for 60N TNNZ
data and the results of the piece-wise tests........................59
5.9 Function parameters for the fitted curves of 20N MAV model...........63
5.10 P values for the independent variables of the 1-way ANOVA test for 60N MAV
data and the results of the piece-wise tests........................64
5.11 Function parameters for the fitted curves of 60N MAV model..........67
5.12 P values for the independent variables of the 1-way ANOVA test for 60N MAV
data and the results of the piece-wise tests........................68
5.13 Function parameters for the fitted curves of 20N MF model...........70
xv


5.14 P values for the independent variables of the 1-way ANOVA test for 20N MF
data and the results of the piece-wise tests.......................71
5.15 Function parameters for the fitted curves of 60N MF model..........73
5.16 P values for the independent variables of the 1-way ANOVA test for 60N MF
data and the results of the piece-wise tests.......................74
5.17 P values for the independent variables of the 1-way ANOVA test for 20N peak
displacement data and the results of the piece-wise tests..........77
5.18 P values for the independent variables of the 1-way ANOVA test for 60N peak
displacement data and the results ofthe piece-wise test............79
xvi


1. Introduction to Study
1.1 Loading Types and Dangers:
Cyclic loading is a form of work involving repeated flexion and extension
motion of a joint over prolonged periods of time. It can be exemplified by dock
workers who repeatedly bend down to pick up boxes and then straighten to load them
onto a vehicle. Each full set of bending down and returning up is one cycle of the
loading. Conversely, static loading involves the prolonged holding of a flexed muscle
state, as seen by farmers who bend down to tend to agriculture close to the ground
and maintain that hunched over posture as they work. Any prolonged, strenuous
work takes a toll on the body, especially when focused on the lower back, and heavy
loading is among the greatest potentially injuring work. Lower back disorders remain
as the most common work-related musculoskeletal problem found in ergonomic
studies of blue-collar employees.25 Cyclic loading specifically is cited as increasing
the risk of low back injury by 10-fold and may be more damaging to the lower back
due to the greater strain applied with each additional loading cycle.38,26,31 Years of
epidemiological research have shown that the effects of loading, day in and day out,
can weaken the health and strength of the body, and cyclic loading is that much more
exhaustive to the body due to the repeated exertion compiled with multiple flexion
and extension cycles.2 With each successive cycle, more and more damage to the
lower back is done.
1


1.2 Clinical Disorders and Epidemiology:
The lower back region, encompassing the five lumbar vertebrae, functions as
a load-bearing support system for most of the upper body weight and stress along the
spinal column. The lumbar musculoskeletal system plays a part in most movements,
almost guaranteed in activities utilizing the full body.9 It follows that the lower back
is naturally at high risk to disorders due to the higher volume of strain put on that
region; this is especially the case when prolonged cyclic or static loading is a daily
exercise.4 Back pain as a result of injury is considered the second most common
neurological ailment in the United States and is the most common cause for job
related injury. Medical and health concerns for low back pain generate a $50 billion
annual consumer industry in America alone.1
One such disorder known as Cumulative Trauma Disorder (CTD), which
when developed in the lower back, is a type of disorder characterized through
epidemiological studies by low back pain, stiffness, loss in range of motion, and
weakness.35,40, 12,20 The development of this disability can be attested to the effects
of creep induced in the adjoining viscoelastic tissues and intervertebral disks through
loading.5,615 Laborers in fields that employ high work activity that involve
prolonged static or cyclic loading were found to be at a greater risk for incurring
CTD.4 The higher the magnitude of carried load, longer length of sequential work,
and shorter time taken to recuperate can exacerbate spinal instability and accelerate
the advent of a lower back injury or disorder such as CTD.21,16 It was also
demonstrated that the more repetitions of loading applied resulted in greater damage,
signifying cyclic loading as more potentially hazardous than static.17,37 Heavier loads
and longer durations without sufficient rest can become disastrous to the lower back.
While this may seem intuitive, conclusive epidemiological observations need
biomechanical experimentation as proof. Further research can also address the
questions of why such injuries occur and how one can prevent them.
2


1.3 Focus of Study and Overview:
Back injury often stems as the stabilizing interaction orchestrated between the
passive and active components within the neuromuscular structure of the lower back
is compromised. Such instability can be wrought through heavy labor as a
malfunction or miscommunication of the collaborating biomechanical systems. The
lumbar spine is responsible to burden tremendous loads superimposed from the upper
body weight, and in cats the L4-5 level in particular generally experiences the greatest
loads and is exposed to the greatest range of motion within the lumbar section,7
Skeletal assemblage and interconnecting ligaments, tendons, and discs comprise the
passive structural architecture to the lumbar stability matrix while the adjacent
musculature imparts dynamic components.30
The passive elements are acted upon by the active elements allowing for
guided spinal mobility. This mobility is controlled by the same dynamic elements
employing a neural subsystem which monitors kinematics and proprioception through
embedded mechanoreceptors that relay necessary feedback and prompt support. As
the spine is moved from a rest position, the ligaments holding the separate vertebrae
together are strained, thus compressing mechanoreceptors embedded within the
tissues. The deformation of the mechanoreceptors stimulates a neuromuscular
response from deep tissue muscles in the lower back. Muscular contractions are
induced which provide antagonistic resistance to stabilize the spinal motion and
maintain cohesion. This activity can be interpreted through electromyography
(EMG) signal processing. EMG represents a muscular contraction by tracking the
electrical potential generated in the muscle from all the involved neurological signals
that induce contraction.52
In a prone, rest position the spine is labeled as being in a neuromuscular
neutral zone (NNZ). This means that no stabilizing muscular intervention is applied
to the lumbar section of the spine, it is in fact able to passively maintain stability.
3


Entering a flexion or an extension shift, the ligaments and other connective tissues
begin to stretch, and at a certain extent, muscle contractions initiate around the spine
for the purpose of spinal stabilization. The onset of muscular intervention is
stimulated when a certain level of stretch is achieved in the viscoelastic structures,
such as the interconnecting ligaments, while the spine is bending, facilitated by the
embedded mechanoreceptors. The amount of stretch in the ligaments over time can
be measured as displacement in millimeters or internal tension in Newtons. Initiation
of EMG within the stabilizing muscles of the lumbar region marks the end of the
NNZ, describing either the displacemental or tensile threshold within the ligaments
where their passive connective forces alone are no longer sufficient to secure spinal
stabilization.
Being viscoelastic in nature, connective tissues, such as ligaments, succumb to
the tendency toward creep, a phenomenon where a solid material can move or deform
when subjected to stresses.50 The application of tension upon a viscoelastic substance
will promote greater stretch displacement over time. In the body, viscoelastic tissues
are composed of collagen fibers, and under tensile stress these fibers may break
advancing the entire tissues capacity to stretch further. The broken fibers can no
longer add their tensile support to the composite tissue when stressed and the
remaining fibers therefore stretch further under the same load. The effects of creep
are cumulative with each successive loading period. A single cycle of constant load
application resulting in viscoelastic creep is shown in Figure 1.1 along with the
transient recovery post-loading. The figure shows in the load plot a constant
magnitude of load applied to a viscoelastic tissue. The amount of elicited
displacement is tracked simultaneously in the displacement plot. While load is
maintained, greater displacement is achieved, referring to a longer and longer amount
of stretch. Creep in viscoelastic tissue is exponentially recovered with the cessation
of the load but this requires a much longer duration of time than the loading period.
4


Figure 1.1 viscoelastic creep developed from a single load cycle
As work continues and creep accumulates, the ligaments within the lower back
become more and more lax, and this has the potential to interfere with the
neuromuscular feedback driven by the mechanoreceptors that respond to tension
thresholds within the same ligaments. As the neural subsystem coordinates the level
of spinal stability, a dysfunction therein may lead to possible injury through negligent
responses to spinal perturbations. Being as immuno-repair of collagen fibers takes
much longer to restore pre-loading conditions, short-term compensation initiated from
other subsystems is required in attempt to avoid injury while the effects of creep are
countermanded.30
1.4 Statement of Intent:
This study intends to explore the impact of cyclic loading on spinal stability
with respect to lower back injury and disorder. Spinal stability is designated as being
the collective maintenance of spinal structural integrity, allowance of mobility, ability
to bear loads, and the avoidance of pain or injury.19 Exposing the feline L4-5
5


supraspinous ligament to light (20 Newton) and heavy (60 Newton) cyclic
loading, post-work instability will be analyzed through the viscoelastic creep found
within connective tissues of the spine, the neuromuscular gradient in the spine and
established NNZs, and the amplitude and frequency parameters of the EMG of the
stabilizing musculature in the lumbar region. Results of the two varying weight tests
will be compared to ascertain the consequences of load magnitude on spinal stability.
1.5 Hypothesis:
Due to trauma incurred from 60 cumulative minutes of cyclic loading to the
lower back, damage and discord to associated musculature, connective tissues, and
neural subsystems will be pronounced for many hours resulting in the prognosis of
instability of the spinal column directly after the cessation of the work period
associated with the established high risk of injury as concurred by:
1. Occurrence of creep in the maximal displacement of the
supraspinous ligament and adjacent viscoelastic tissues taking
several hours to recover back to preload conditions, suggesting
micro-damage sustained to the collagen fibers and possible damage
to embedded mechanoreceptors
2. Enlargement of NNZs after the cessation of work and requirement
of several hours to exponentially recover preloaded baseline
conditions, which signifies dampened response sensitivity to
perturbations along the spine resulting in delayed stabilizing muscle
contractions
3. Decreased maximal EMG amplitudes, analyzed through the mean
absolute voltage (MAV) of the EMG, within the multifidus and
other paraspinal muscle groups initially after work showing weaker
contractions and less muscular support
6


4. Similar changes within the EMG median frequency (MF),
shadowing MAV behavior, and giving insight to derecruitment of
the muscle motor units
This all accumulates to show a physiologically weakened state due to
instability and will be apparent in both 20N and 60N test data. However, the 20N
group would have changes in these instability parameters of a magnitude much less
than the 60N, if not nearly unchanging. The 60N case will allude to greater damage
incurred with higher percent changes from baseline and more pronounced travel over
the recovery period. Statistically significant variance in NNZ, MAV, and MF away
from baseline data should be more apparent in the 60N case.
1.6 Significance of Study:
The study of the effects of cyclic loading of the lower lumbar level of a feline
model at 20 and 60 Newton loads is a continuation of previous studies that concern
the damaging effects of different work agendas at differing load magnitudes. How
does cyclic work compare to static, and how do light, moderate, or heavy loads
facilitate changes in the involuntary response of the body? As a comparative
analysis, it is hoped that quantitative data gleamed from working conditions in a
hypothetical lab environment can be extracted to postulate feasible improvements in
the natural working environment, thus improving human health and the quality of
employment. Specifically, work/rest schedules for high intensity labor firms based
off this studys findings would be beneficial to avoid losses and damages occurred
through some of the severe low back injuries, such as CTD among others, that plague
the arenas of manufacturing, construction, shipping, and agriculture. There are even
applications within sports medicine and athletic training.
As an analysis of biomechanical systems and responses, such experimentation
broadens the understanding of how our bodies work and is invaluable to the further
7


development of medical knowledge. Such experimentation also has been applied to a
more encompassing control model of the neuromuscular feedback loop.
Specifically, this study maintained the findings of previous work, describing
the inherent instability caused by extensive creep in the viscoelastic structures by
skewing the perception of mechanoreceptors embedded therein. The damaging
effects of cyclic loading can lead the body to be less aware of the instability
threatening the spine, and compensation is delayed by as much as 4 hours, before the
muscles are induced into a state of hyperexcitability. Furthermore, the justification
that heavier loading sequences have greater potential for damage is quantitatively
displayed.
8


2. Anatomy of the Lower Back Region
It is important to understand the physiological mechanisms involved in this
study; the spinal column and specifically the lumbar region. A synopsis of the
anatomical structures and physiological relationships follows, describing the
referential planes of the body, the lumbar spine and individual vertebrae, viscoelastic
structures therein, associated paraspinal muscles, and neural receptors. Also is
included a description of spinal control feedback as well as action potentials, motor
unit recruitment, firing rate, and EMG analysis through its amplitude and frequency.
Sagittal Plane
2.1 Body Planes:
As illustrated in Figure 2.1 the body is
anatomically described in three major planes. The
sagittal plane divides the body in the left and right
or sinister and dexter sides respectively. The
coronal plane designates front and back or ventral
and dorsal sides. The transverse plane divides top
and bottom or superior and inferior.43,48
Figure 2.1 body planes (SEER 2008)
2.2 Lumbar Spine:
The target of this study involves the lumbar back and the associated symptoms
that are accrued through heavy workloads and strain. The susceptibility of the lumber
region to injury and disorder is directly related to its position in the body and its
heavily relied upon functions.36 The entire spine along the back is composed of
different sections each containing specific vertebrae. The lumbar region follows
9


Cervical curve
sequentially after the cervical and thoracic
spinal levels and contains the last 5
mobile vertebrae in the spine. The sacral
and coccygeal sections that follow after
have vertebrae that are all fused together
and are anchored to the pelvis. Figure 2.2
shows the vertebral levels in succession
along the spine. Located above the pelvis
and beneath the rib cage, the lumber level
not only is a foundation for support of the
upper torso but also is a fulcrum to
mobility, allowing such movements as
posterior, anterior, or lateral extension and flexion and axial rotation.45 It is
responsible not only for holding the body together but also enabling physical activity,
and because of this coupled functionality, it begs the fact that the grunt of wear and
tear from labor would target this section the most.
2.3 Vertebrae:
The larger loads imposed on the individual lumbar vertebrae dictate their
larger size in proportion to other vertebrae along the spine. The bulk of the vertebral
size is in the anterior section, designated the vertebral body. This portion constitutes
a flat base to the vertebra allowing for sequential stacking of vertebrae upon which
the entire spinal column is compiled. Articulated in this way, vertebrae work in
cohesion to compose the spinal column as the strong pillar of support which defines
its function. Mobility is also allowed due to the detachment of each vertebra from
one another. See Figure 2.3 for an image of a typical arrangement of vertebra along
the spine.
Vertebral Column
Cervical vertebrae
Thoracic vertebrae
Lumbar vertebrae
Sacrum
Coccygealvertebrae
Thoracic curve
Lumbar curve
Sacral curve
Figure 2.2 vertebral levels of the spine
(SEER 2008)
10


Each individual movement
artifact on the spine, however, must
be secured marginally to maintain
collective spinal integrity. The
second portion of the vertebra is
the posterior
attachment known as
the vertebral arch
which helps to limit
the range of motion of
the spine while also
creating a spinal canal
to house the collection
Superior articular process
Spinous process
Posterior tubercle of
transverse process
Nucleus pulposus
Disc annulus
Vertebral body
Anterior tubercle of
transverse process
foramen transversium
Figure 2.3 oblique view of vertebral orientation and structures
(Wikipedia 2009)
of nerves in the spinal cord. These continuous canals, the vertebral foramina, initiate
at the first cervical vertebra after the skull, C1, and connect inferiorly until the last
lumbar vertebra before the pelvis, L5. This composite tract gives a space for the
spinal cord to run
Transverse process
Superior urticular
process
Injertor articular
protest
Mamillary process
Accessary process
Figure 2.4 vertebral processes (Grays Anatomy Fig. 93)
lengthwise along the
body, protecting it in a
honey enclosure, while
also permitting axons to
extend to and from the
spinal cord in respective
nerve roots. Also associated
with the vertebral arch are 7
processes which extend outward
from the vertebral body. See
11


Figure 2.4 for an illustration of the various processes attached to the vertebral arch.
The main purpose of these processes is to serve as grips where tendons, muscles, or
other connective tissues attach for positive reinforcement. Also, some of these
processes are paired and constitute as interlocks to oppose a matching pair of
processes of adjacent vertebrae. The two superior articular processes each have an
embedded concave receptacle, known as the facet joint, whereby the pair of inferior
articular processes in the vertebra directly above fit. This definitively limits the
degree of motion of any particular vertebra with respect to another by direct contact.
In the lumbar region, transverse rotation is completely prevented because of this.
2.4 Ligaments:
Ligaments are the
tissues that connect and hold
the spinal column together,
attaching to the walls and
processes of each vertebra.
Ligaments are bundles of
collagen fibers that provide
cushion and viscoelastic
cohesion to the body parts
which they connect. In the
case of the spine, many sets
of ligament line both the anterior and posterior walls of the vertebral bodies and
arches to hold each vertebra in dynamically stiff position much like cables along a
suspension bridge. The supraspinous ligament is a very thick rope-like ligament that
runs along the back of the spine, connecting the prominent apices of each spinous
process from the 7th cervical vertebra down to the sacrum.56 Notice the supraspinous
Figure 2.5 median sagittal cross-section of lumbar vertebral
section showing interconnecting ligament groups
(Grays Anatomy Fig. 301)
12


ligament on the right most side of Figure 2.5 attached to the successive spinous
processes of the stacked vertebrae. This ligament is of specific interest to this
experiment because it is comparably thick and the first ligament accessible dorsally
making it the most readily analyzable tissue to examine the viscoelastic properties of
the spine and lower back during cyclic loading. The nature of its position and
function dictates that it stretches first during flexion and undergoes the greatest
displacement changes of spinal ligaments. As such, it compromises a portion of
ligament-muscular reflex, triggering muscular support by establishing NNZs.
called the annulus fibrosus and spongy core called the nucleus pulposus, the discs are
more viscous than elastic in nature and not only isolate the individual vertebrae for
the purpose of spinal mobility but also act as shock absorbers and deterrents to
excessive motion. The shape of the discs in the lumbar level are responsible for
lumbar lardosis, which is the tendency for the lumbar region to assume a concave
alignment, as opposed to the convex nature of the cervical and thoracic levels. These
opposing alignments in the spinal levels due to the differing intervertebral disc shapes
create the characteristic S-shape that the entire spinal column assumes. Being
viscoelastic in nature, they are subject to the same effects of creep as the ligaments
undergo during loading.6 This results in the discs flattening out and being less
Another viscoelastic structure
integral to the spine is the
intervertebral disc, a flexible spacer in
between each vertebral body making up
one fourth of the entire length of the
spine as shown in Figure 2.6.
Composed of an tougher outer shell
2.5 Intervertebral Discs:
Figure 2.6 intervertebral viscoelastic discs
(ADAM 2008)
13


resistive to applied force. Such a condition could result in discs slipping or popping
out of place.
2.6 Paraspinal Muscles:
The muscles accompanying the spine are called paraspinal muscles and are
responsible for the initiation of movement along the spinal column through muscular
contractions. Having short movement arms and high compressive force, the
paraspinal muscles contract antagonistically to apply a stabilizing hold that draws in
vertebral displacements out of bounds and actively retains spinal solidarity during
""".iv movement. The multifidus muscle group is of particular concern
-^flV to ^is experiment as it acts acutely as a stabilizing lower back
jm Wr mechanism, embedded directly into the mamillary processes of the
if lumbar vertebrae. The mamillary processes are extensions out
from the superior articular processes in each lumbar vertebra.
Multifidus fasciculi extend from one process to processes of
vertebrae two to four spaces above, essentially creating an
overlapping network along the spine interconnecting all
vertebrae. Being a deep tissue muscle, the multifidi tend to be
slow twitch, functionally enabling continuous contractions for
long periods of time without experiencing fatigue. This lends
it to be a prominent proponent of stabilization along the
lumbar region, and therefore the EMG activity of the
multifidus is an applicable signal in the determination of the
pattern of stabilization played by the paraspinal muscles.
Figure 2.7 shows the multifidus muscles positioned laterally
adjacent to the spine running the length of all vertebrae.
Figure 2.7 paraspinal
muscles embedded
sagittally parallel to the
spinal column
(All About Back and
Neck Pain 2007)
14


2.7 Mechanoreceptors:
In order for musculature and active organs to apply a robust stabilization
scheme upon the spine, the central nervous system provides cognitive direction to the
level of intervention requisite per situation. The nervous system acts as a highway for
communication to and from the brain and all extremities. Therefore feedback
pertaining to the conditions of the internal surrounding within the body, the physical
manifestation of efferent signal, is pivotal to the assertion of valid responses. Sensory
receptors provide this feedback to the central nervous system, and are integrated
within muscles, organs, joints, or skin as the end node of an afferent axon.
Mechanoreceptors in particular relay information regarding mechanical
pressure and stretch, giving proprioceptive, kinematic, and tactile representations
concerning their receptive fields. There exist many specific nerve endings each
attuned to a specific mechanical reactivity. For example, numerous types of
mechanoreceptors have been categorized, each group having its own force or stretch
tolerances, some responding to high or low magnitudes of displacement, others to
sustained pressure or frequency dependent vibrations. Responding independently per
individual characteristics, a gradient of information gives a generalized composite
view.
Ligaments, tendons, connective tissue, and joints usually contain multiple
types of mechanoreceptors, and because of this are deemed sensory organs apart from
their structural responsibilities, as they facilitate the transmission of valuable sensory
feedback about their respective internal states. The supraspinous ligament has been
found to contain type II and type III mechanoreceptors,33 specifically Pacinian
corpuscles and Golgi organs,19 which respond to fast-adapting and slow-adapting
pressures and extreme changes in load.55,53 These mechanoreceptors are attuned
mainly to gross changes in tension, and therefore instigate stabilizing activity from
the paraspinal muscles based on internal tension thresholds. The muscles themselves
15


also have numerous receptors within their fibers, the most prominent being the
muscle spindle and the Golgi tendon organ, which give both proprioceptive and
kinesthetic data. The result of which, in the low back, gives a network of continuous
afferent information concerning the displacemental or tensile state within a specific
lumbar organ used to appropriate the stabilizing activity of paraspinal muscle
response.
2.8 Spinal Feed-back Control:
All of the elements of the body cooperate as a self-regulating, closed-loop
feedback system. As the spine forms a foundation for the facilitation of posture and
movement, the bone structure, paraspinal muscles, intervertebral tendons and discs,
and associated mechanoreceptors must all work in conjunction with the central
nervous system for comprehensive effectiveness. The spine must be stiff yet dynamic
in order to accomplish all the tasks required of it for environmental interaction and
must be robust in order to avoid injury and pain. Active muscle contractions induce,
dampen, and counter motion along the spinal infrastructure. Passive viscoelastic
components including ligaments, tendons, and discs deform under pressure and the
slack and flexibility of these tissues permit a healthy range of spinal motion due to
such perturbations. The elastic quality of these tissues also works to return and hold
mobile vertebrae in their original configuration. Both muscles and ligaments contain
sensors that relay proprioceptive and kinesthetic information through afferent axons
to the central nervous system. The brain can then conduct muscular actions and
reactions responsively through efferent neural stimuli. Oftentimes, the brain is
bypassed with immediate reactionary neural loops about the spine or contained within
the opposing musculature. A block diagram for the full control system of the spinal
lumbar area is displayed in Figure 2.8.
16


Force
Length
Velocity
Acceleration
etc .
Figure 2.8 spinal feedback control system in the lower lumbar level (Solomonow et al. 2003)
2.9 Action Potentials:
A muscular contraction begins with the initiation of an action potential along
the axon of the motor neuron which in turn relays to the muscle and induces
shortening along the muscle fibers. An action potential is a self propagating
electrochemical transmission through an axon of the nervous system. When a nerve
ending is stimulated, its membrane begins to energize through the opposing diffusion
of sodium and potassium ions across the cellular gradient increasing the voltage
potential at that site. If the membrane potential increases past a critical threshold,
usually 15mV higher than the resting membrane potential, a runaway condition
activates whereby positive feedback generates a wave of increasing membrane
potentials down the length of the axon. Typically, the resting membrane potential of
a nerve is -70mV and action potentials are generated when that potential positively
exceeds a threshold of -55mV.45,49 Greater external stimuli affect a larger receptive
field, activating more neurons and firing more action potentials. Greater summation
of action potentials ensures the signal down not decay at the neural head or soma.
17


If a mechanoreceptor is properly stimulated, the action potential fired relays
feedback to the central nervous system. This is known as afferent communication,
sensory information collected at peripheral neural receptors and transmitted to the
spine and up to that brain. Axons also carry action potentials back to muscles to
instigate a response in much the same way; traveling in the reverse direction, this is
known as efferent communication. When an efferent action potential firing down a
motor neuron reaches a muscle, a chemical response is elicited. Instead of sodium or
potassium, however, calcium ions diffuse across cell membranes within the muscle,
thus driving the shortening of myosin filaments that make up the individual myofibril
of a muscle fiber. This induces a muscular contraction.
2.10 Motor Unit Recruitment and Firing Rate:
Each muscle within a body is controlled by multiple independent signal-
carrying efferent axons emanating from the spinal cord. As a result, each axon
commands only a percentage of the muscle fibers within a single muscle, and the
action potentials from each axon trigger the contraction of only its associated muscle
fibers. The connection of axon to muscle fiber is known as innervation; the more
muscle fibers to axon, the higher the innervation ratio. The grouping of one efferent
axon and its innervated muscle fibers is known as a motor unit. The muscle fibers
within a motor unit contract simultaneously as a unit when prompted by the shared
axon in an all-or-none scheme. Therefore, the number of fibers that a motor unit
innervates and the size of those fibers can signal the strength of a single contraction
emanated from that motor unit, and hence its inherent function. Fewer, smaller fibers
of a motor unit signify fine-tuned control, while many, larger fibers would show
ability for greater force output. In this way, the composite effects of the entire muscle
contraction can be regulated by allocating which motor units to activate.
18


Figure 2.9 gives a representation
of a motor unit involving one axon
innervating a bundle of muscle
fibers. Having smaller motor units
with smaller innervation ratios, a
muscle is usually identified as
being slow twitch. Slow twitch
muscles, as opposed to fast twitch
muscles, tend to metabolize
oxygen instead of lipids and
carbohydrates. This allows for
steady, weaker muscle
contractions that dont incur
fatigue. Such muscle groups are
found in the body to provide functions that require indefinite activity. The multifidus
muscle group is such an example as it is needed to be continuously able to exert
steady force in order to maintain proper spinal stability in the lower back. As greater
displacement or force outputs are required of the muscle, more motor units can
become activated. Motor units are recruited in this way during a muscle contraction,
smallest to largest, in a process called orderly recruitment.
Another way to increase the magnitude of a muscular contraction is by
increasing the firing rate of each motor unit. The firing rate corresponds to the rate of
successive action potentials to a motor unit. Each impulse stimulates a single twitch,
but if impulses are sent in series, the compiling twitches can become a sustain
contraction. As more impulses are sent rapidly to the muscle via the motor unit, a
greater composite contraction is elicited from the muscle fibers. This occurs
essentially as subsequent twitches are built upon each other before being allowed to
dissipate. Muscle tension continues to increase with a greater rate of transmitted
Motor neuron
Figure 2.9 motor unit representation showing an
efferent axon connecting to muscle fibers
(Mosby 2009)
19


action potentials until maximal muscle tension is acquired; this limit is known as
tetanus. Figure 2.10 shows the effects of increasing firing rate on total tensile output
toward the achievement of tetanus.
s f \
c o *\ a AAA A"e aye / \
1/. c .X /v.
Stimuli t t t r mm {tmmmm
(1} Tyvitcn (2) Summation (3) Incomplete tetancs (4) Complete tetanus
I'unfusedl (Vised)
Figure 2.10 muscle twitch summation affected by firing rate toward tetanus (Marieb 2008)
2.11 Electromyography and Median Frequency:
Spatio-temporal summation of the action potentials from all the active motor
units causing a muscular activity contributes to the electrical response of the muscle
seen in the EMG signal.32 The more motor units involved and the higher the firing
rate attributes to a greater number of action potentials developed, and their spatio-
temporal summation contributes to greater resulting EMG amplitude. Along this vein
of reason, larger developed EMG signals correspond directly to stronger respective
muscular contractions.
The speed at which an action potential propagates down a nerve is called the
conduction velocity. This property is determined in muscle groups by the size of the
motor unit. Each motor unit has an individual axon diameter based on the functions it
plays in the body. Motor units with larger diameters, and in accordance faster twitch,
have higher conduction velocities while slow twitch motor units have smaller axon
diameters and lower conduction velocity. It follows if a physical activity solicits
more muscle involvement, the conduction velocity is increased as larger motor units
are recruited. The conduction velocity has also been linked to the EMG median
frequency. This is termed the frequency that equally divides the power of a signal in
half. Hence, as the larger motor units enter the active motor unit pool during a
20


muscular contraction, more action potentials are included in the composite EMG
signal and it is expected that these additional signals would represent higher
frequencies. As such, the total power spectral density of the EMG would encompass
more power on the right-hand side of the spectrum relative to these higher
frequencies, thus shifting the median frequency to right-hand side of the spectrum,
which is otherwise a numerical increase. In this manner, conduction velocity and
median frequency were shown to be linearly proportional.34,39,18 Therefore, median
frequency can be tracked over time as a measure of total motor unit recruitment
within a muscle, and is a preferred parameter being not as sensitive to noise as other
representative models.41
Fatigue occurs as larger motor units exhaust energy supplies and drop out
from muscle activity through orderly derecruitment. In order to counteract the
resulting drop in force, remaining motor units increase their firing rate. A rising
mean EMG magnitude coupled with a falling median frequency would represent a
muscle experiencing fatigue. For this experiments purposes, fatigue was not
considered to be an impacting factor because the level of activation from the passive
cyclic loading was relatively low. This is most conclusively displayed with the
parallel patterns exhibited by the simultaneous trends of peak MAV and MF of the
EMG. As they never counteracted each other, the advent of fatigue is not included.
What can be gleamed from this is the connection that by observing the
behavior of the median frequency of the EMG signal, the behavior of the conduction
velocity and thus the motor unit recruitment is proportionally observed. This attests
to the behavior of the level of activity of the entire muscle and the changes it incurs in
response to different loading conditions.
21


2.12 Cytokines and Neutrophils:
Cytokines are a molecular body that helps facilitate communication between
cells, much like hormones or neurotransmitters, and are typically transported through
the bloodstream. Their appearance within a cellular matrix can signal the presence of
harmful foreign bodies or damaged tissue and increase or decrease immunological
responses. The family of cytokines is vast, composed mainly of peptides or
glycoproteins, and individual types of cytokines can vary in their signaling
functionality engendered by what cell-surface receptors are embedded in their cells
walls. A nominal classification of cytokines is based off the understanding of which
type of cells a cytokine directly affects.
The establishment of an immune response is critically related to the work of
cytokines. Usually cytokines are secreted when a pathogen is encountered by an
immuno-regulating cell. Like a marker indicating the location of the pathogen, and
often subsequent phagocytosis thereof, other immuno-regulating cells rush to that
area. Cytokines can also stimulate the controlled production and cessation of more
cytokines within the infected area, thus drawing more and more immune cells.
Certain established cytokines, known as endocrine cytokines, are maintained in the
blood stream and can readily diffuse to distant regions in the body when needed.51
Cytokines can also catalyze metabolization of pathogens by attaching themselves to
the appropriate antibodies.
Neutrophils are the most abundant white blood cell, normally circulating
through the blood system to congregate around areas of inflammation or bacterial
infection.54 Unlike the larger bodied monocytes, neutrophils have short lifespans of
one to two days but metabolize bacteria or dead tissue similarly through
phagocytosis. Due to their greater volume in the blood stream, a foreign body is
much more likely to face a neutrophil first; neutrophils are drawn to infected sites
22


through chemotaxis, or the sensitivity of chemical gradients permeated by cytokines,
where they internalize targeted microorganism, microbes, or particles.
The injured, broken collagen fibers of the ligaments sustained during loading
must first be metabolized by neutrophils before new tissues can be rebuilt. Drawn by
the inflammation within the injured region, cytokines subsequently draw an influx of
neutrophils to flood the region. Two feline tissue samples are shown in Figure 2.11.
The image on the left, Cat 3-5-02, has undergone loading and has a much higher
concentration of neutrophils as a result. The neutrophils are stained purple.
Conversely, the control specimen shown on the right shows less tissue damage and
less neutrophil concentration. The loaded tissue shows damage sustained from the
loading sequence, many cells and blood vessels appeared severed, which explains the
flood of neutrophils compared to the control.
Figure 2.11 neutrophil concentrations in two ligament samples, the tissue on the left was loaded
whereas the control on the right was left unloaded (adapted from Le 2007)
23


3. Procedure
3.1 Preparation:
The protocol administered in this study was approved by the Institutional
Animal Care and Use Committee (IACUC). Two preparations were designated, a
20N series and a 60N series. The 20N series contained 8 adult feline preparations
where as the 60N series contained 9. In both series the same work/rest sequence was
implemented with respect to time.
In-vivo cats were anesthetized with alpha-chloralose proportional to their
individual weights (60mg/kg). This specific anesthesia is unique in that under
specific dosages can prevent all neural activity from filtering across the brain stem
without inhibiting reflexive neuromuscular activity within the body. Cognitive
awareness and control of the torso and appendages is temporarily paralyzed, but the
animal is stable and physiologically awake.8,28,21 This allows for clean collection of
reflexive neuromuscular activity bereft of voluntary interaction while satisfying
humane ethics for animal experimentation.
A straight incision was made along the spinal path from the thoracic to sacral
level penetrating only dermal layers and exposing the intact dorsolumbar fascia for
the entire lumbar length. Posterior processes for each of the seven feline lumbar
vertebrae were located and marked for reference. A stainless steel S-shaped hook
was inserted between the L-4 and L-5 vertebrae penetrating beneath the supraspinous
ligament. The specimen was then placed dorsally prone in a rigid stainless steel
frame and connected to life-support systems including:
24


an automated respirator
an intravenous drip pump supplying saline fluid
a non-invasive oral sensor with display of the cat vitals
a heat pad and heat lamp to help maintain stable body temperature
External fixators
were positioned on the
posterior processes of the L-
1 and L-7 vertebrae so as to
isolate movement along the
full lumbar level. The entire
frame was then adjusted for
vertical alignment with the
hydraulic actuator. A
schematic of the feline
preparation is shown in
Figure 3.1.
Six pairs of stainless
Ex-Fix
Load
Cell

1SL SSL
E
HjHf o mrr
v l\ J
4 5
Ex-Fix
mm
-r^---ft---
> 6 M 7
Load
CcU [-1
Ex-Fix jSL SSL \ I x mm Ex-Fix
Figure 3.1 experimental setup of the lumbar level for the
feline specimen
(A) spinal orientation during rest with fixators at LI and L7
(B) applied cyclic loading with stainless steel hook at L4-5
steel fine-wire electrodes were inserted 6 to 8 mm to the right of the spinal midline
into to the adjacent multifidus muscle of each lumbar section L-l/2, L-2/3, L-3/4, L-
4/5, L-5/6, L-6/7, each collecting an individual signal from that level for a total of 6
channels. With another ground electrode inserted into a gluteus muscle, this system
of sensors collected EMG data. Sterile gauze cloths were placed over the open
incision of the specimen to induce clotting and were maintained moist with de-
ionized water for the duration to simulate a natural moist internal state and to keep
internal tissues from drying.
At this point, the specimen would be ready for initiation of the computerized
protocol which will be discussed further under the protocol section. Along the entire
duration under lab supervision of the experiment, records of the specimens internal
25


body temperature, blood-oxygen saturation (SP02), heart rate, and fluid intake were
recorded every 15 minutes.
3.2 Instrumentation:
The instrumentation used for this experimental procedure assisted and
monitored the vitals of the specimen, implemented the desired experimental protocol,
and collected resulting data from the specimen. The life-support systems included:
Harvard Inspira Advanced Safety Ventilator that controlled and monitored
respiration of the specimen
Flo-Gard Volumetric Infusion Pump connected intravenously to the specimen
providing a regulated 1 OmL/kg/hr of saline fluid
Rad-5 Pulse Oximeter that monitored SP02 levels and heart rate
Gaymar heating pad to maintain internal temperatures that tend to drop due to
sedation
WelchAllyn anal thermometer to record internal temperature.
A desktop computer was used as an interface to the Bionix 858 Material
Testing System controller that ran the linked mobile actuator connected to the
specimen through the MTS station manager software FlexTest SE ver. 3.5C 1808,
which implemented the block diagram protocol. Triggered by the FlexTest software,
another desktop computer collected and stored output data in memory at a 1000 Hz
sampling rate via the program Easyest LX ver. 1.1. Vertical displacement and
loading rate were each recorded off the actuator and load cells continuously along
with the EMG data. The stored data sets were formatted and converted to more
commonplace file types for transfer and further analysis after the completion of the
experiment.
26


Bipolar wire electrodes shown in Figure
3.2 were implemented in this experiment for their
better ability to focus on signal reception of
specific regions without nearby muscle crosstalk
during both muscle contraction and rest. Although
more invasive than surface electrodes, wire
Figure 3.2 stainless steel wire
electrodes
electrodes allow for the recording of signals from
deep muscles at greater resolution without the
conduction of other superficial signals or ion movements on the skin. Wire electrodes
are also more appropriate due to their greater flexibility and lower impedance. Each
electrode pair was the input to an independent differential amplifier with a 1 lOdB
common mode rejection ratio, gain capability of up to 200,000, and a band pass filter
in the range of 20-500Hz. Attenuation of frequencies beyond the band-pass region
filtered unwanted movement artifacts below 20Hz and high frequency noise
phenomena above 500Hz. This was acceptable because only approximately 5% of
multifidi EMG power was found higher than 500Hz.16 A notch filter was placed at
60Hz to eliminate a large artifact of noise that occurred due to a large mechanical
noise signature at this frequency. The recorded EMG data from the electrodes was
sampled at 1000Hz, double the Nyquist frequency, for storage in the adjoined
computer and simultaneously diverted to a set of oscilloscopes whereby EMG activity
could be observed in real time.
3.3 Protocol:
Two sets of cats were tested under two experimental protocols for this
research study. The first set, containing 8 feline specimens, was run under successive
work and rest periods of 20N peak load cycles. The second set, containing 9 feline
preparations, followed the same protocol but with 60N peak load cycles. The loading
27


sequence consisted of blocks of cyclic work actualized by tension developed
vertically perpendicular to the ventral feline orientation at the L-4/5 connection.
Loading occurred at 0.25 Hz sinusoidal cycles oscillating from latent tension to
respective 20N or 60N applied force.
The feline preparation would initially be primed at an individually calibrated
pretension of 1N applied prior to each cycle of loading so as to standardize all
magnitudes of work dependent on differing individual physiology with a controlled
equivalent instigation of loading. The preparation would then be loaded for 10
minutes. It should be noted the first two single cycles of this first loading period
would be used for baseline data to be compared to post work recovery data as they
signify the response of an un-worked model. A rest period for 10 minutes at no-
loading was allotted after each full 10 minutes of loading to allow rest alleviating any
fatigue and cumulative damage, which would then be followed by the next block of
loading and so on for 2 hours. At the end of the complete loading sequence, the
specimen would have been subjected to a total of 6 loading blocks for an accumulated
1 hour of work and 1 hour of rest.
Following the loading sequence, a 7 hour recovery sequence at no-loading
examined the specimens biological response post-work. Immediately after the final
rest period, or ten minutes of rest after the final loading, a single 0.25 Hz cycle at
respective load was applied to elicit displacement, tension, and EMG responses. The
actual loading took 4 seconds but was recorded in a 16 second window to observe any
previous or following EMG signatures. The next single cycle test in like manner
occurred 20 minutes after the previous, or half an hour post-work. The third occurs
half an hour after the previous, or one full hour post-work. The final six tests follow
at each successive hour interval; one test every hour for six more hours. Figure 3.3
displays the protocol as a block diagram.
28


Baseline
Load/Rest Period
7 Hours Recovery Period
f = 0.25 Hz
BrBa
L-ji l-ji , l_g
RrR2, -,R5
T1,T2,...,T9
N,
n2
n3
n4.n5....n9
Baseline Establishment
Loading Periods (10 min)
Rest Periods (10 min)
Single Cycle Test During Recovery Period
Recovery Period 1 (10 min)
Recovery Period 2 (20 min)
Recovery Period 3 (30 min)
Recovery Periods (1 hr)
Figure 3.3 experimental protocol applied to each specimen at respective load
29


4. Methods
4.1 EMG:
Applying a load to the L4-5 supraspinous ligament, the induced reflexive
electromyographic signals within the stabilizing multifidus musculature were
recorded for each mid-lumbar channels, L3-4, L4-5, and L5-6. The EMG activity of
a typical full experiment is shown in Figure 4.1.
5
E
S
E
X
i m
>>!> -- + + 44** + +!
i*4H Ml 1HMHE + BA+T
.A .A A A A A ,a .a 5
?
1/6 1/6 0 10 20 30 40 50 60 70 80 90 100 1101/6 1/2
Time (hr)
Recovery Period
Figure 4.1 full 9 hour experimental EMG response for 6 channels
The muscle activity represented by the EMG signal is plotted for all 3
channels shown in successive order vertically where EMG3 is the EMG from the L3-
4 lumbar level, etc. Each channel gives the very first two 16 second cycles of the
work period for initial unloaded baseline perspectives. Immediately following is the
full 2 hour work period where six 10 minute blocks of cyclic loading were applied
interspersed with 10 minutes non-loading rest. The last nine signals are the hourly
test cycles from the 7 hour recovery period, charting internal damage compensation.
30


The parallel displacement charts and load charts are also given chrono-
synchronously. The displacement graphs show the accumulation of creep over time
whereas the applied load is maintained at a constant, which in this case is seen as
20N.
4.2 Creep:
The driving factor behind the lower back injuries addressed in this study can
be ascribed to the property of the connective tissues in the spinal region to exhibit
creep when stressed. As mentioned, creep is the tendency of a material to stretch to
greater lengths each time it is pulled. Creep can compile with multiple loading
sessions and usually follows an inverse exponential curve peaking at a maximal
stretch capacity determined by internal characteristics. In effect, the material is
getting more lax as it is being worked. This is usually due to micro-damage being
afflicted in the material because of the strain from loading. Microscopic fissures are
induced as the molecular bonds unifying the material break. In the case of the
viscoelastic tissues associated with the spinal region, this micro-damage accrues
through the breakdown of the collagen fibers that form its structure. These tissues
can then repair themselves post work, reforming the connective collagen fibers as an
immunological response associated with inflammation. Viscoelastic tissues require
much more time to recover from creep than to accumulate it, because it takes longer
to heal than damage the broken collagen fibers.
Creep at a given point in time can be mathematically described as:
Creep = Lx ~ Lbase x 100% ... (Equation 4.1)
^base
31


Lbase is the baseline displacement under a specific load and is recorded before
any prior loading. This is a latent response taken as a representative of the
natural state of the tissue.
is the peak displacement reached per work period driven at the same load
used for the baseline.
4.3 Neuromuscular Neutral Zones:
Neuromuscular neutral zones
(NNZ) designate ranges along phases of
flexion or extension where muscular
intervention is not necessary for
stability. The integrity of the system is
held together solely by its own rigidity.
For example, the natural posture of a
subject exemplifying a straight spinal
orientation requires no activity from the
multifidus musculature for stability.
The viscoelastic tissues and latent
tensions of the internal connections can FiSure 41'flexion and ^tension phase patterns
from the neutral position repeated in cyclic work
resist the strain of gravity to maintain (OConnor 2005)
this posture. Thus at this posture the lower lumbar back is within an NNZ as no
muscle activity is needed. See Figure 4.2 for flexion and extension phases of the
lumbar level in the sagittal plane. The neutral zone continues as the posture begins to
enter a flexion phase, bending forward. With the continued absence of stabilizing
EMG from lower back musculature, stabilization is maintained until at a certain angle
off center where the forward vector of gravitational pull is great enough to
compromise the unbiased stability of the spine and connecting tissues. At this point,
32


muscle activity kicks in to ensure the spine is held together and posture is maintained
thus defining the limit of the NNZ. Returning from the bent position back to a
vertical posture, EMG will again stop at some angle, whereby the spinal stability can
be again held without muscular intervention. The body enters another NNZ. For
each complete contraction of a muscle, two NNZs are found, at the initiation of
contraction and at the cessation. These two phases of NNZ are termed stretch and
relaxation respectively.
Two types of NNZs must be designated, relating to the kinesthetic factors
which are exerted by muscles. A muscle contraction produces both displacement and
force. Depending upon the muscle structure and type, one effect may be favored over
another. For example, optical muscles whose primary function is to move the eye
may favor displacement whereas intestinal muscles that squeeze solids along the
digestive tract may favor force. As such, the neutral zones of a muscle have both a
displacement and a force property. The same NNZ of a muscle relates to a threshold
for both displacement and tension, which begs specification, displacement NNZ
(DNNZ) or tension NNZ (TNNZ). Mechanoreceptors, such as the Golgi organ,
embedded in the ligaments and viscoelastic tissues monitor tension and displacement
along a flexion or extension phase, relaying this information as a part of the lower
lumbar control mechanism so that stabilizing EMG can be initiated when thresholds
are sensed. The supraspinous ligament, being prone to stress present in static or
cyclic work cycles of the back, is one such mechanism for monitoring stability.
NNZs are determined from a specific muscular contraction by comparing the
initiation of stimuli to the advent of EMG response. The amount of mechanical
stretch in millimeters or force in Newtons at which EMG is instigated marks the
corresponding displacement or tension limit. The range from zero up to that value is
thusly determined as the NNZ, being that no EMG was recorded. Hence the system
within that range was inherently stable without external support from the muscles.
33


A selectively down-sampled EMG signal was used for NNZ determination.
The 16,000 point original EMG signal was band-pass fdtered between 20-500Hz with
a notch fdter at 60Hz. Then it was run through a 40 point moving outlier extractor
that right-aligned peak positive values and center-aligned peak negative values for
every 40msec period resulting in an 800 point alternating signal. These processing
steps are shown in Figure 4.3 and Figure 4.4.
HHlMl

Figure 4.3 full EMG signals during successive stages of signal processing finishing with the
consolidated down-sampled 800 point signal used in NNZ tracking
4 50 4 55 4 60 4 65 4 70 4 75
Time
Figure 4.4 zoomed-in representation of EMG stages during signal processing
34


The exact time at which EMG starts and stops for each channel was recorded
as two phases of the same neutral zone. Average voltage values prior to load
initiation were accredited as resting muscle potential and the advent of three times the
magnitude of this resting value was considered the start of EMG. Similarly, EMG
finished when the voltage dropped and remained 3 times below the resting potential.
Figure 4.5 shows how each NNZ is referenced to the EMG for each channel from a
single recovery cycle. The cross-referenced displacement and force value recorded at
the corresponding times signified the DNNZ and TNNZ for each phase per signal.
DNNZs and TNNZs were plotted over time for each channel and averaged
according to applied weight, 20N or 60N, to give a general population trend.
E 12.5
25 30 35 40 45 50 55 60 65 70 75
Time
Figure 4.5 typical threshold plot displaying the single cycle EMG for three channels and the
associated displacement and load plots
35


4.4 Normalized Peak Mean Average Voltage:
The EMG recorded in muscles is seen as an AC electric signal. The spatio-
temporal summation of the action potentials within the muscle resembles dispersed
pulses, positively or negatively charged, imposed along a base resting oscillation.
Spasms occur with fatigue and damage occurring in a muscle, where motor units
clench up to offset the lack of force attributed to broken or underperforming fibers
and therefore should not be accredited in our analysis for sustained stability control.
This makes quantitative analysis of total EMG strength difficult due to the
subjectivity of accepting responsive EMG output versus routine spasmodic activity.
Peak EMG amplitude only accounts for single twitch pulses and does not consider
sustained activity with both positive and negative fields. However, representing the
EMG as a positively compiling signal, EMG from the same muscle can be compared
over time in regard to peak contractile effort.
The EMG signal was fully rectified into the positive domain and then run
through a 200 point moving average. The rectification, mean, and moving average
produced a smoothed parabolic signal denoted as the mean average voltage (MAV).
A center-aligned moving average assured that there was no integration of a time
delay. Peak MAV values were acquired for each signal and plotted with respect to
time for each channel. Normalized about the initial baseline value, composite curves
averaging all specimens for each channel was attained. Normalization eliminated
actual EMG voltage values by scaling all data sets respective to a baseline value of 1
such that the composite MAV behavior would be devoid of any individual
physiological variations. In regard to any test specimen, the magnitude of an EMG
signal occurring in a muscular response differs based off individual characteristics
such as size, strength, age, gender, and so on. Thus, through normalization, such
inconclusive factors would not be allowed to undermine the composite average by
favoring one data set over another. Figure 4.6 shows the MAV signal in red
36


overlaying its comparable EMG signal. The peak value for MAV represented the
scale of the elicited muscular contraction as a numerical grade to be compared with
other single cycle responses over time.
2.5 3.0 3.5 4.0 4.5 5.0 5.5 60 6.5 7.0 7.5
Time
Figure 4.6 MAV represents total contraction of muscle over time
4.5 Median Frequency:
Median frequency is tied proportionally to motor unit recruitment. To find the
median frequency, a 0.5 second window of EMG was extracted centered about the
peak tension. Focused on a small portion of the total EMG signal corresponding to a
relatively planar force magnitude, it is assumed that this represents a stationary signal
by piece-wise observations. Figure 4.7 shows this selection of the 0.5 sec window.
Time
Figure 4.7 0.5 sec EMG extraction at peak tension for PSD analysis
37


As shown in Figure 4.8 the sampled data vector composing the targeted signal
portion was zero padded as prerequisite for spectral analysis in order to expedite the
Fast Fourier Transform.23
1 t-EMG 201 12-0-1/L2) 13-(L2/L3) 14-(L3/14) 15-0-4/L5) 16KIS/L6) 17-(L6/17) I30-F-201
4.7470 -0.0639 0.6917 0.4121 -0.2479 -0.0481 0.0762 0.0000
4.7480 6.0000e-3 0.2084 -0.0982 0.0182 0.0579 -0.0387 1.9531
4.7490 5.6000e-3 0.7209 -0.4413 0.0718] -0.0822 0.0524 3.9063
4.7500 -0.0124 -0.2315 -6.OOOC0-4 -0.1412 | -0.1250 0.0611 5.8594
4.7510 9.20000-3 0.0314 0.4583 0.0559 5.4CCO0-3 6.40000-3 P, 7.8125
4.7520 5.6000e-3 -0.0387 -0.1393 0.2472 0.0613 0.0X4 9.7656
4.7530 -6.00000-4 0.0322 -0.0390 : 0.0312 7.3000e-3 -0.0240 11.7188
4.7540 -0.0138 6.4000e-3 -0.0392 -0.0647 J 3.9000e-3 0.0242 13.6719

4.7550 8.9000e-3 0.0238 -0.0413 0.0233 0.0132 0.0230 15.6250
4.7560 0.0103 0.0139 0.0153 0.0113 0.0209 4.2000e-3 > 17.S781
4.7570 0.0100 1.50000-3 -0.0320 ' -140000-3 0.0216 0.0X7
19.5313
4.7580 0.0342 0.0331 -0.0323 3.00000-3 0.0200 7.60000-3
4.7590 7.00000-3 1.90000-3 -0.0601 -0.0379 7.4000-3 -0.0245 21.4844
4.7600 -0.0532 3.80000-3 -0.0501 -4 SOOOe-3 0 0204 0.0123 23.4375
4.7610 -7.50000-3 0.3991 -0.0672 -9.6000e-3 j 4.10000-3 * no00e-3 25.3906
4.7620 0.0676 1.4596 -0.2207 -0.0826 6.2000e-3 -0.017S p 27.34X
4.7630 0.1604 1.6347 -0.0403 -0.1264 8.01X00-3 -0.0171 29.2969
4.7640 -0.0185 -1.5112 0.4961 0.0453 1.7000e-3 0.0358 31.2500
4.7650 -0.2184 -1.7151 0.1075 -0.2286 -2.3000e-3 0.0741 33.2031
4.7660 0.0256 0.5762 -0.2332 -0.1104 5.00000-3 -0.1358 35.1563
4.7670 0.0332 -0.1704 -0.1395 0.3530 5.10000-3 -0.0759 37.1094
131-C.3/L4) 132-0-4/15) 133-(LS/L6) M*P5D(L3/M>
0.0000 0.0000 0.0000
0.0000 0.0000 0.0000
0.0000 0.0000 0.0000
-0.4413 0.0713 -0.0622
-6.0000e-4 -0.141? -0.1250
0.4530 0.0559 5.40000-3
-0.1393 ! 0.2472 0.0613
-0.0390 ; 0.0312 7.3X00-3
-0.0392 1 -0.06471 [ 3.9000e-3
-0.0413 0.0233 , 0.0182
0 0158 0.0113 : 0.0209
-0.0320 -1.40000-3 j 0.0216
-0.0323 3 00000-3 ; 0.0200
-0.0601 -0.0379 j 7.4000e-3
-0.0501 -4 5000e 3 0.0204
-0.0672 -9.60000-3 j 4.1X00-3
-0.2207 -0.0826; 6.2000e-3
1.7424e-
0.012
0.016
8.45210-
0.238
0.461
3.722
0.710
6.113
10.677
1.574
2.802
35.509
0.146
24.585
16.423
16.002
Figure 4.8 zero padding the targeted EMG selection in SigmaPlot
A Tukey trapezoidal transform was superimposed upon the zero-padded vector in
order to minimize spectral leakage at the edges of the window without losing any
signal gain.
Frequency-Domain Window
Baseband-Equivalent Frequency in Hz
Figure 4.9 Tukey trapezoidal window transform chosen with r=0.25 taper (green)
38


The Tukey window was chosen as it was a tradeoff between the rectangular
and Hanning windows, confronting EMG signals with low, disparate amplitudes. The
signal amplitude was not attenuated due to the window as the trapezoidal taper at the
sides were attuned to the zero-padded sections. Attenuation was avoided as it would
be problematic where the EMG is already small, within the in millivolt range. With
an angled taper r value chosen at 0.25, the Tukey window can be described as:
(2n n-\ \
1 + cos n
l r N-1 JJ
w(n) = s 1,
1 1 + cos (in In n-l }

2 { r r N-1 J
n<^{N-l)+l
^(N-l)+\ N-^(N-l) ... (Equation 4.2)
The power spectral density was computed with a Fast Fourier Transform of
the selected windowed signal. Baseline noise spectra were computed in the same
manner from an adjacent 0.5 second window prior to the initiation of loading. The
noise signal was subtracted from the power spectral density of the targeted window to
filter out residual noise harmonics. The frequency that split the total power spectral
density in half was the designated median frequency. Median frequencies were found
for each data set during recovery and arrayed versus time. The stream of median
frequencies over time for each channel was run through a three point moving average
shown in Figure 4.10 and Figure 4.11, smoothing out the curve. Moving averages
were implemented throughout this experiment in avoidance of additional time delays
solicited through gross signal transforms.
39


0 100 200 300 400 500 600 0 100 200 300 400 500 600 0 100 200 300 400 500 600
Time(min)
Figure 4.10 raw MF data for all three channels during recovery
L3-4 L4-5 L5-6
Time(min)
Figure 4.11 smooth MF trend after running through 3 point moving average
4.6 Modeling:
4.6.1 DNNZ
The pooled data for the displacement NNZs was modeled for each lumbar
level using a typical exponential formula matching the transient nature of viscoelastic
tissue decay. Showing asymptotic regression, the model is expected to fall more
sharply along the y-axis initially and gradually decreases decent with increase along
the x-axis until leveling off along the asymptotic plane. Physiologically, this exhibits
the tendency for viscoelastic tissues, like most tissues in the body, to show more
40


dramatic rates of recovery initially after accrued damage as opposed to slower rates of
recovery hours later. This is also indicative of the long period of time required for
any full recuperation back to original conditions. The model used for DNNZ data is
described as:
D0 is the displacement offset of the asymptote. Theoretically this value
would be the baseline limit for either stretch or release phases.
Dm represents the maximum displacement magnitude accrued off the
baseline.
rDl is the time constant indicating the rate the exponential decay with respect
to time.
rr is the time at the onset of the recovery period or 120 (min)
4.6.2 TNNZ
The model for the tension NNZs is much like the DNNZs due to the similar
effects of tension and displacement on ligamento-muscular feedback. In fact,
DNNZs and TNNZs are dependent of one another and as a result and directly
related. The TNNZ model differs only in an added term that allows for a second
exponential curvature as per the tendency of the tension tracking. This is accounted
for in the beginning of the TNNZ data, where tension was found to continue to
increase during the first three time measurements and then thereafter decay as
expected toward baseline. The model is expressed as:
DNNZ = D0+Dm
(Equation 4.3)
V
J
( J~T' \ ( Jllz. \
TNNZ = T0+(t-rr)TL e rn +TU e r
(Equation 4.4)
\ J \ J
41


As with the previous model for displacement, T0 represent the offset that the
exponential decay approaches.
Tl magnitude increased from initial intercept to peak
Tm is the magnitude of tension shifted from the asymptote to peak
rri represents the time constant for the initial rising exponential
tT2 represents the time constant relating to exponential decay toward the
asymptote
Tr is the time at the onset of the recovery period or 120 (min)
4.6.3 PMAV
In order to model the peak MAV, two equations are employed that split the 7
hour recovery time period. The data is tracked with an initial minimal decrease
before increasing with recovery past baseline. It was found that a compensation
mechanism is triggered in the muscles of the back after the advent of physical trauma
which elicits over-excitation in the muscle responses in order to promote spinal
stabilization. Strain and fatigue accumulated throughout cyclic work leaves the
lumbar region of the spine in a weakened state, and in order to counter the structural
laxity post-work, stabilizing muscles overcompensate compared to baseline. EMG
magnitude is greater to accompany the faster response times in the shortening of
NNZs. The compensation, however, is not introduced immediately post-work, it
usually kicks in after 2 to 4 hours of rest.
In order to model this behavior, an equation much like that used for the
NNZs is used for the initial period of recovery prior to the initiation of the
compensation mechanism. The model tracks the MAV datas tendency to decrease
initially and then to rise back toward baseline. Two exponential terms are added to an
offset, establishing upper and lower bounds. Notice that one of the terms is
42


subtracted from 1, allowing the model to eventually increase with exponential decay
toward and upper asymptote.
After the introduction of the compensation mechanism, a new exponential
term is added to the model for the remaining data points. This term is imposed upon
the existing model, shifting the limit of decay to a higher value, in accordance with
higher excitability of muscles and greater EMG response. The time at which the
compensation mechanism is triggered is designated rd and was adjusted for each
model to maximize the fit. The entire model is described as:
MAV = \
( l-Tr \
P + P 1 0 ^ 1 L e Tp' + PM S! 1 1
v ) v /
( t-T, \ ( t~Tf ^ ( <-Tr \
P + P 1 0 T 1 L e Vpi + PM 1 e !pl k. i + e Tn
v ) \ ) \ /
... (Equation 4.5)
P0 is again the initial offset or baseline that the model decays towards.
PL is the magnitude of initial MAV decay adjusted from intercept
PM is the magnitude of the of the shift from the asymptote
PH represents the magnitude of the shift due to compensation
xpx is the time constant relating to the initial decrease in MAV after work
tp2 is the time constant relating to the increase of MAV back towards
baseline
rp3 is the time constant that affects the integration of the compensation
rd represents the time where the compensation mechanism begins
rr is the time at the onset of the recovery period or 120 (min)
43


4.6.4
MF
The model for the median frequency is exactly like that of the PMAV. This is
due to the expectation of the median frequency to follow the MAV proportionately
under the lack of muscle fatigue. The entire model is described:
MF = \
Fo+K
/ t-r. \ f *-r, ^
e Tf' + Fm 1 -e Tfi
V ) y
( t-Tr \ ( ~Tr \
+ FL e lF' + fm l-e Tf2 1 +
\ J \ J v )
t < T,

(Equation 4.6)
F0 is again the initial offset or baseline that the model decays towards.
Fl is the magnitude of initial MAV decay adjusted from intercept
Fm is the magnitude of the of the shift from the asymptote
Fh represents the magnitude of the shift due to compensation
rn is the time constant relating to the initial decrease in MAV after work
tF2 is the time constant relating to the increase of MAV back towards
baseline
rF3 is the time constant that affects the integration of the compensation
rd represents the time where the compensation mechanism begins
rr is the time at the onset of the recovery period or 120 (min)
4.7 Curve Fitting:
Using the model constraints described for each focus of study, DNNZ, TNNZ,
MAV, MF, a fitted curve was found that represents the trend of each data pool over
time per channel using the Levenberg-Marquardt algorithm through the Matlab
computer program. The resulting equation tracked the data points as closely as
possible given the predetermined model definition. This resulting curve is
44


superimposed on the data plots shown in the Results section and the applied variable
constants used are given in accompanying tables. R2 values were given to represent
the goodness of fit. Higher values approaching 1 signify a model that fits the data
points more closely.
4.8 Statistical Variance:
Statistical significance comparing data points within a specific pool was found
using the JMP 7 computer program. DNNZ, TNNZ, MAV, MF, and peak
displacement data sets were analyzed and conformed to normal distribution bell
curves in order perform a repeated measures analysis of variance (ANOVA).
Maintaining time and phase as independent variables, the ANOVA checked for
significant interaction between data across all cats, all channels. A null hypothesis,
an assumption that some data points within the set would be significantly different
than others, was crosschecked against Type 1 errors. Type 1 errors are the error in
rejecting the null hypothesis when it should be accepted, meaning that data points are
truly not significant from each other. The probability of Type 1 errors are given as a
p value.57 In short, a low p value means there is low probability of committing a
Type 1 error, or to say that the data does in fact contain significant variance amongst
itself. A p value threshold was set to a standard of p<0.05 for significance, p values
must be less than 0.05 to be acceptable. Pair-wise comparisons were used to reveal
any significant variance between all data points within the set to their baseline value
according to the LS mean Students t test.
For statistical calculation of data significance concerning DNNZ and TNNZs,
a 2-way ANOVA was run holding both time and phase independent. Having two
independent variables also called for the necessity of analyzing the mathematical
cross product of the two in order to ensure no cross dependency. In short, as all the
data is being analyzed for statistical variance, it must be ensured that time and phase
45


variables actually behave independently of each other. The stretch phase trend does
not influence the relaxation and vice versa. All specie samples were randomized and
given a p value as well as with each independent variable, i.e. time, phase, and the
cross between them in the JMP 7 program. All variable p vales were recorded along
with the significant data points.
For MAV, MF, and peak displacement data sets, a 1-way ANOVA was
applied instead of a 2-way due to the lack of the phase variable. MAV analysis
scrutinizes the peak absolute EMG per load, only one per cycle, and therefore doesnt
take into account any stretch or release phase. Therefore, there is neither an
independent phase variable nor a time-phase cross. Only time is held independent.
The same is true for MF and peak displacement.
46


5. Results
Data was collected for 8 specimens under 20N load experiments and 9 specimens
under 60N loading. The extracted EMG response signals during the 7 hours recovery
period for each specimen were analyzed in regards for DNNZ, TNNZ, MAV, and MF
behavior. The resulting analyses were plotted using SigmaPlot and averaged across
all cats within each load specific population for a generalized trend. Curve fitted
models were applied to each of these trends as well as statistical analysis of variance.
Peak displacement from the supraspinous ligament stretch per load cycle was also
extracted and used to formulate load specific creep trends as well.
Figure 5.1, Figure 5.4, Figure 5.7, Figure 5.10, Figure 5.13, Figure 5.15,
Figure 5.17, and Figure 5.19 show the DNNZ, TNNZ, MAV, and MF plots and
fitted curves for both the 20N and 60N cases. Stretch versus relaxation phase plots
are given within each of the NNZ sections as well. Creep accumulation results over
time are given per population afterwards.
47


5.1
20N DNNZ
E
E,
N
C
0)
E
8
is
CL
U>
b
16
14
12
10
8
6
4
2
0
L3-4 L4-5 L5-6
O Release Stretch n = 8 20N n = 8 20N T n = 8 20N
Si I h Hi,,. TTtt, -i-if Hi,
-e-~44 "T ii T Inn
? T
0 100 200 300 400 500 600 0 100 200 300 400 500 600 0 100 200 300 400 500 600
Time (min)
Figure 5.1 composite DNNZ trend for 8 specimens at 20N during recovery period
Table 5.1 function parameters for the
fitted curves of 20N DNNZ model
( \
A+ A/
v /
Stretch Phase
Ch. L34 L45 L56
A 3.55 3.004 3.488
A 2.993 3.355 2.287
rDl 128.8 206.3 121.9
R2 0.9872 0.9849 0.9702
Relaxation Phase
Ch. L34 L45 L56
A 2.497 3.375 2
A 7.106 6.42 7.555
r D\ 518.3 411.4 515.5
R2 0.9916 0.9758 0.9799
Table 5.2 p values for the independent
variables of the 2-way ANOVA test for 20N
DNNZ data and the results of the piece-wise
tests
Ch. L34 L45 L56
R2 0.82 0.76 0.79
spec. <0.001 <0.001 <0.001
time <0.001 <0.001 <0.001
phase <0.001 <0.001 <0.001
t/p cross 0.3595 0.7686 0.0479
Student's t (+) 1-9 (+) 1-9 (+) 1-9
Table 5.1 gives the appropriate numerical values found when fitting the pre-
determined function to the DNNZ data set for each channel L3-4, L4-5, and L5-6
where tr = 120. The goodness of fit R2 percentage is also given. This percentage
48


alerts to how closely the derived equation actually matches the data; values of 0.9 or
higher are considered a good fit for this studys purposes. As can be seen, in both
stretch and relaxation phases, every derived curve fit well to the data. The vales of
Table 5.1 are associated with the curves displayed in Figure 5.1.
The results of the 2-way ANOVA for DNNZ can be found Table 5.2. Each p
value and the data points significantly deviated from baseline are given along with an
R value for the entire data set. In this case, the higher the R value, the more
consistent all the specimens behaved.
The statistical values for the 20N DNNZ case confirm that there is
significance within the trend and that there is not any correlation between time and
phase. The results from the Students t test are given in Table 5.2 as well. The (+)
symbol denotes positive significance and the numbers are the test numbers during
recovery that are significant versus baseline. 9 test cycles were performed during the
7 hour recovery period and are labeled 1 through 9 in successive order. Thus Table
5.2 shows that the Students t test of the 20N DNNZ gives all 9 values during
recovery as positively significant above the baseline value. This indicates that the
displacement NNZs did not return to the initial baseline values after 7 hours of
inactivity.
5.1.1 Stretch Phase
The composite baseline values for 8 feline specimens found for the 20N
DNNZ case was found to be (2.6836 1.0633)mm, (1.9710 0.5459)mm, and
(2.0630 0.6589)mm respectively for the L3-4, L4-5, and L5-6. These are average
DNNZ values at the initiation of work, without any previous activity. Off the block,
the supraspinous ligament at the L4-5 level can stretch 2.6836mm give or take
1.0633mm before stabilizing EMG in the L3-4 level is required. L4-5 EMG begins at
a ligament stretch length of 1.9710mm give or take 0.5459mm and so on.
49


After the prescribed 2 hour cyclic work/rest period, the DNNZ values were
found have increased significantly to (6.4157 2.2615)mm, (6.5150 1.8285)mm,
and (5.7783 1,5987)mm respectively. The supraspinous ligament had accumulated
creep to the point where around 6mm give or take 2mm of stretch was required in
order to trigger stabilizing multifidus EMG in the three channels. On average, the
DNNZ increased by 210.3% after those 2 hours of cyclic work. Simply put, the
muscles are turning on given a much greater stretch in the ligament. A much greater
physical effort is required to induce muscular stabilization.
As the specimen is left to rest, the DNNZ decreases steadily over time. At the
end of 7 hour recovery period, the final DNNZ measurement was
(3.5338 1.2566)mm, (3.31751 1.0684)mm, and (3.60001 1.6028)mm. This still
has an average of 70.66% above baseline for all channels. Figure 5.2 graphs the
percent change the DNNZ data compared to baseline and the statistical significance.
It is clearly shown that that baseline was never significantly reached in the stretch
phase.
400
300
si
^ 200
z
£ 100
E
8
rc 0
Q.

a -100
-200
0 100 200 300 400 500 600 0 100 200 300 400 500 600 0 100 200 300 400 500 600
Time (min)
Figure 5.2 percent change of 20N DNNZ stretch phase from baseline during recovery period
L3-4
L4-5
L5-6
50


5.1.2 Relaxation Phase
Baseline was found to be (4.3160 2.5852)mm, (4.7219 1.6123)mm, and
(4.6410 1.3427)mm respectively for the channels L3-4, L4-5, and L5-6. As the
ligament is relaxing, the stabilizing muscles for each lumbar level stop contracting at
their respective value of stretch.
After the 2 hour work/rest period, the DNNZs were found to be
(9.510011.5616)mm, (9.80001 1.5713)mm, and (9.35501 1.7756)mm respectively.
This is an average increase of 150.3% post-work. The muscles are deciding to off
much sooner than they usually do each time the same cyclic motion is performed,
leaving a much greater portion of the activity unguarded.
As in the case of the stretch phase, the relaxation phase of the DNNZ also
steadily decreases back to baseline during recovery. After the full 7 hour period, the
DNNZs were found at (5.85881 1.9900)mm, (5.69381 1.8984)mm, and
(5.41861 1.6735)mm which is still significantly 38.29% above baseline on average
for all channels. Figure 5.3 gives the DNNZ relaxation phase data in percentage
variations off baseline and the accompanying statistical significance.
L3-4 L4-5 L5-6
Time (min)
Figure 5.3 percent change of 20N DNNZ relaxation phase from baseline during recovery period
51


5.2
60N DNNZ:
L3-4
L4-5
L5-6
0 100 200 300 400 500 600 0 100 200 300 400 500 600 0 100 200 300 400 500 600
Time (min)
Figure 5.4 composite DNNZ trend for 9 specimens at 60N during recovery period
Table 5.3 function parameters for the
fitted curves of 60N DNNZ model
D0 + Dm
<-Tr \
td\
V J
Stretch Phase
Ch. L34 L45 L56
Do 3.799 5.369 2.658
dm 9.825 7.267 8.928
*D\ 255.2 154.3 320.7
R2 0.9973 0.9854 0.9789
Relaxation Phase
Ch. L34 L45 L56
Do 0 0 0
dm 17.46 17.13 17.26
^ D\ 958.6 925.7 1160
R2 0.9687 0.9597 0.9896
Table 5.4 p values for the independent
variables of the 2-way ANOVA test for 60N
DNNZ data and the results of the piece-wise
tests
Ch. L34 L45 L56
R2 0.77 0.75 0.8
spec. <0.001 <0.001 <0.001
time <0.001 <0.001 <0.001
phase <0.001 <0.001 <0.001
t/p cross 0.605 0.6324 0.6188
Student's t (+) 1-6 (+) 1-6 (+) 1-7
Table 5.3 gives the model values used in the fitted curves for each channel of
the 60N DNNZ case. Table 5.4 gives the statistical results of the 2-way ANOVA and
subsequent piecewise tests results. The DNNZs returned to baseline between the 4th
52


and 5th hours of recovery for lumbar level 3-4 and 4-5. L5-6 required more time,
restoring original baseline condition after the 5th hour.
5.2.1 Stretch Phase
The baseline DNNZ stretch phase for the 60N load conditions were found to
be (6.1237 1,8358)mm, (6.0134 2.6282)mm, and (5.5606 2.2910)mm for the
respective lumbar channels L3-4, L4-5, L5-6. As with the 20N case, the DNNZs are
found at a much higher value after the 2 hour work period having values of
(13.7789 4.4588)mm, (13.0067 + 4.1958)mm, and (12.0678 3.8875)mm, an
average of 146.5% above baseline. Again, almost like the mechanoreceptors have
been numbed, greater displacement thresholds are required to illicit muscular
response. Within the recovery period where no work is being demanded, the
DNNZs immediately begin to decrease over time reaching values of
(5.9100 2.1104)mm, (6.1044+ 2.5270)mm, and (5.2689 2.0296)mm at the end of
the 7 hour period. Overall, this marks a 0.5049% difference from baseline. Figure
5.5 displays the journey of the DNNZ recovery as a percentage and those recovery
values found to be significant from baseline.
L3-4 L4-5 L5-6
Time (min)
Figure 5.5 percent change of 60N DNNZ stretch phase from baseline during recovery period
53


5.2.2 Relaxation Phase
The relaxation phase follows suit, having baseline values of
(9.905212.3693)mm, (9.658412.372l)mm, and (9.75541 2.4866)mm. After the
work period, the DNNZs were at (17.233313.3744)mm, (16.933313.2659)mm,
and (17.25671 3.2341 )mm, 80.79% above baseline, and thereafter fell to
(11.133314.9266)mm, (11.161114.9167)mm, and (12.22001 3.9512)mm by the 7th
hour, 19.03% above baseline. Figure 5.6 shows the DNNZ relaxation phase percent
recovery and significance.
L3-4 L4-5 L5-6
Time (min)
Figure 5.6 percent change of 60N DNNZ relaxation phase from baseline during recovery period
54


5.3 20N TNNZ:
L3-4 L4-5 L5-6
Time (min)
Figure 5.7 composite TNNZ trend for 8 specimens at 20N during recovery period
Table 5.5 function parameters for the
fitted curves of 20N TNNZ model
( r JzLl >
T0+(t-Tr)TL e rn + tm e Tjl
v J v y
Stretch Phase
Ch. L34 L45 L56
To 3.971 2.5 4.459
tl 0.1629 0.01665 0.08634
T 1 M 3.252 4.767 0.5205
tt\ 22.35 54.73 34.36
rT2 202.8 361.9 35
R2 0.943 0.9796 0.7919
Relaxation Phase
Ch. L34 L45 L56
To 3.211 1.164 1.5
tl 0.06729 0.03613 0.06909
T 1 M 10.03 12.14 8.877
XT\ 176.3 268.7 143.3
TT2 176.8 269.2 294.9
R2 0.9194 0.9888 0.9758
Table 5.6 p values for the independent
variables of the 2-way ANOVA test for 20N
TNNZ data and the results of the piece-wise
tests
Ch. L34 L45 L56
R2 0.73 0.75 0.71
spec. <0.001 <0.001 <0.001
time <0.001 <0.001 <0.001
phase <0.001 <0.001 <0.001
t/p cross 0.3891 0.9391 0.4026
Student's t (+) 1-3 (+)1-4 (+) 2-3
55


Table 5.5 gives the model values used in the fitted curves for each channel of
the 20N TNNZ case. The model used for the tension NNZs contained an extra term
than the displacement NNZs, requiring a few more variables to be accounted for.
Table 5.6 gives the statistical results of the ANOVA and subsequent piecewise tests
for the same data. The data set for the 20N TNNZ was found to be not normal as it
did not fit bell curve distribution. To run an ANOVA test, the applied data set must
model normal distribution fit. It has been shown that a simple mathematical
transform over a data set does not notably change the results of an ANOVA test.
Therefore, the entire TNNZ data set was transformed using a basic root function. The
square root of each value was taken. As this newly transformed data set was
sufficiently normal, an ANOVA test was run. The results show that the TNNZs
display significance according to the Students t test only during the first 4 test times,
comprising the first 2 hours after loading. The TNNZs quickly return to baseline
thereafter, even slightly but not significantly dropping beneath the baseline.
5.3.1 Stretch Phase
In the case of the 20N stretch phase, the TNNZ baselines were found at
(4.9583 2.3279)N, (3.83561 1.6002)N, and (4.1423 2.1688)N for the respective
L3-4, L4-5, and L5-6 channels. So around 4N of tension give or take about 2N, EMG
from the multifidus musculature was prompted in response to increasing tension
applied to the supraspinous ligament. Immediately after the work/rest period, the
TNNZs were found at (7.3243 5.6633)N, (7.25751 4.5650)N, and
(5.043313.4730)N which is 74.28% above the baseline. The TNNZs continued to
follow a general increasing trend within the first hour of recovery, peaking 84.70%
above baseline at (8.15141 4.5243)N, (7.24131 4.1496)N, and (5.49831 3.4813)N
after a half hour of rest and (7.60141 5.2148)N, (6.997514.2396)N, and
(5.86331 3.7512)N at the first hour marker, a 70.09% increase from baseline.
Thereafter, the TNNZs decreased dramatically, returning to baseline after two hours
56


of recovery. The TNNZ had final values of (3.8638 1.9241)N, (3.772512.0745)N,
and (4.7143 3.9511)N at the end of the total 7 hour monitored recovery period.
This is 2.425% offset just above baseline. Figure 5.8 gives the percent change during
stretch phase and statistical significance.
L3-4 L4-5 L5-6
300
200
5
o^-
N 100
I 0
C
<1)
H
-100
-200
0 100 200 300 400 500 600 0 100 200 300 400 500 600 0 100 200 300 400 500 600
Time (min)
Figure 5.8 percent change of 20N TNNZ stretch phase from baseline during recovery period
5.3.2 Relaxation Phase
In the relaxation phase, the TNNZ baselines for the L3-4, L4-5, and L5-6
channels are (9.67871 5.8525)N, (8.703014.7350)N, and (7.9441 14.1453)N
respectively. As with the stretch phase, the first three test cycles done within the first
hour of recovery showed an increasing trend. 10 minutes after work, the TNNZ
values were found at (12.172916.1687)N, (12.761315.4880)N, and
(10.158314.6193)N, 129.3% above baseline. A half hour with the recovery period,
the TNNZs were (13.36861 5.2529)N, (12.96751 5.4480)N, and
(11.445014.1939)N, 127.7% above baseline. And an hour after the work/rest period,
the TNNZs were (13.53001 5.2517)N, (12.85001 5.8149)N, and
(11.17501 5.5785)N, 132.4% above baseline. The TNNZs started decreasing
rapidly over time even surpassing baseline. The TNNZs were measured at
(7.796316.0993)N, (6.838714.8497)N, and (5.504313.0135)N at the end of the
57


recovery period, 5.307% below baseline. Figure 5.9 gives the percent change and
significance for the entire recovery period.
Time (min)
Figure 5.9 percent change of 20N TNNZ relaxation phase from baseline during recovery period
58


5.4 60N TNNZ:
L3-4 L4-5 L5-6
Time (min)
Figure 5.10 composite TNNZ trend for 9 specimens at 60N during recovery period
Table 5.7 function parameters for the
fitted curves of 60N TNNZ model
( -^0 f T_r_ '
1 + e Tt' + TM e Tfl
V ) V y
Stretch Phase
Ch. L34 L45 L56
T0 0 5.808 0
tl 0.1808 0.01 0.01
T 1 M 31.48 20.72 23.09
Tn 20.26 20 20
TT2 279 215 298.2
R2 0.9901 0.9174 0.9443
Relaxation Phase
Ch. L34 L45 L56
To 13.64 15.38 25.84
tl 0.4331 0.283 0.3744
T 1 M 30.19 25.65 20.26
XT\ 115.1 132.6 92.59
TT2 120.1 132.1 92.61
R2 0.971 0.9697 0.9706
Table 5.8 p values for the independent
variables of the 2-way ANOVA test for 60N
TNNZ data and the results of the piece-wise
tests
Ch. L34 L45 L56
R2 0.67 0.69 0.75
spec. <0.001 <0.001 <0.001
time <0.001 0.0058 <0.001
phase <0.001 <0.001 <0.001
t/p cross 0.7685 0.8779 0.1997
Student's t (-) 6-9 (-) 7-9 (-) 6-9
59


Table 5.7 gives the model values used in the fitted curves for each channel of
the 60N TNNZ case. Table 5.8 gives the statistical results of the ANOVA and
subsequent piecewise tests for the same data. Like in the 20N case, the 60N TNNZ
data set was first applied to a square root transform before being run through the
ANOVA. Due to the variability of the tested specimens and high standard deviations,
the TNNZ data points do not show significance initially before restoring baseline
value in the 60N preparations; however recovery continues exceeding baseline
negatively during the last three recorded recovery hours. Considering the general
trend of the data, it can be assumed that the TNNZs could further drop below
baseline conceivably for many more hours if the designated recovery duration was
extended.
5.4.1 Stretch Phase
The baseline TNNZ values for the stretch phase in the 60N case were
(21.3345 10.1360)N, (20.3266+ 13.5144)N, and (18.8598 12.6383)N for the L3-
4, L4-5, and L5-6 lumbar channels respectively prior to work. After the work period,
the values were found at (31.2067 19.2246)N, (28.1867 21.1513)N, and
(23.8833 21.8356)N, a 58.04% increase from baseline, but were not statistically
significant above baseline. During the recovery period, the TNNZs decreased over
time surpassing the baseline significantly by the 4th hour and continued to decrease
having final recorded values of (7.62561 7.9591)N, (9.2311 10.0697)N, and
(5.1522 5.2459)N at the 7th hour marker. Being 67.30% statistically significant
below baseline, it should be noted that the TNNZs did not level off entirely, and
even further declension can be assumed after the allotted 7 hours of recovery. Figure
5.11 gives the percent change off baseline and statistical significance graphically.
60


L3-4
L4-5
L5-6
Time (min)
Figure 5.11 percent change of 60N TNNZ stretch phase from baseline during recovery period
5.4.2 Relaxation Phase
As for the relaxation phase of the same case, the baselines were determined to
be (37.1704 16.7594)N, (33.2673 15.4208)N, and (35.7330 17.6079)N. After
the work period, the first hour of recovery marked an increasing trend in the TNNZs.
The first test cycle at ten minutes recovery had TNNZ values of
(43.6467 16.4923)N, (41.0700 18.5771)N, and (46.3756 16.6564)N, 41.16%
above baseline. After a half hour at the second test cycle, the TNNZs were
(47.4778 11.2453)N, (42.1567 17.6367)N, and (47.9137 14.5664)N, 54.64%
above baseline, and finally after an hour (46.1944 11.0660)N,
(43.0267 12.6708)N, and (48.19131 8.8286)N, 57.25% above baseline. Thereafter,
the TNNZs decreased dramatically, shadowing the decrease exhibited in the stretch
phase, and significantly falling below baseline by the 5th hour of recovery. The final
TNNZ values were found to be (20.4756+ 13.1202)N, (22.79331 15.4991)N, and
(28.84001 15.3330)N by the end of the 7 hour recovery, 27.64% below baseline.
Figure 5.12 graphs the percentage tracking over the recovery period and associated
significant values.
61


L3-4 L4-5 L5-6
Time (min)
Figure 5.12 percent change of 60N TNNZ relaxation phase from baseline during recovery
period
62


Normalized MAV (mV)
5.5 20N MAV
L3-4
L4-5
L5-6
Figure 5.13 composite MAV trend for 8 specimens at 20N during recovery period
Table 5.9 function parameters for the fitted curves of 20N MAV model
t L56
-4.983
5.758
/ A ( t-T, \
Po + Pl e Tp' + PM 2 I I
\ J l 7
Ch.
n
PL
L34
-5.811
6.482
L45
-4.435
5.116
Pm 7.068 5.484 6.162
tp\ 102.4 66.97 100.2
T P2 120 75.02 111
( _'-rO ( <~*r >
+ PL e Tpi + Pm l-e rp2 + I -t
v ) V V
Ch. L34 L45 L56
Ph 0.003 0.003 0.001
TP 3 83.91 56.95 100
Tj 290 279.2 290
R2 0.7871 0.7527 0.7697
r/>3
t >TJ
63


Table 5.10 p values for the independent variables
of the 1-way ANOVA test for 60N MAV data and
the results of the piece-wise tests
Ch. L34 L45 L56
R2 0.4 0.49 0.45
spec. 0.0122 <0.001 0.0027
time 0.041 0.0675 0.0541
Student's t none none none
Table 5.9 gives the model values used in the fitted curves for each channel of
the 20N MAV case. The first model pertaining to time before rd was the data set
excluding the last four data points and the variable values obtained were then used to
fit the second model for time after rd to the last four data points excluding the first 5.
The resulting independent curves were then spliced seamlessly at the case specific rd
time intersection. The overall R2 value is then given considering the entire conjoined
fitted curve. Table 5.10 gives the statistical results of the ANOVA and subsequent
piecewise tests for the same data.
The Students t test in the 20N MAV focus showed no significant variance
away from baseline at any time during recovery. This is also confirmed by the high p
values given for time around the threshold p=0.05 meaning a high probability of
making a Type 1 error, or assuming there is variance in the data when there isnt. It
should also be noted that for 3 feline specimens within the 8 subjects accepted as test-
worthy, MAV was seen to drastically increase mid-recovery only to fall back in the
following hours. The strength of those signal responses might have been incongruent
when compared to the previous and following responses elicited from the same
specimen but were not irregular responses independently. And although offsetting
the composite average as can be seen in the disparity created at the 6th and 7th test
points in Figure 5.13, those data points were not throw out for fear of subsequently
depriving the other 20N results and possibly contaminating the study. This explains
the lower R2 values found.
64


The baseline for the 20N MAV case were found at (1 0)mV, (1 0)mV, and
(1 0)mV respectively for the lumbar channels L3-4, L4-5, and L5-6. The
uniformity of these baselines is due to the fact that MAV data for each individual
specimen is normalized before being averaged. Thus all baselines across the board
are scaled to 1, and the rest of the individual data set subsequently scaled
proportionally as well. The purpose of this is in response to the huge disparity that
different feline subject can exhibit through their level of EMG magnitude nominally
produced. The amount of voltage produced by regular muscle contractions per cat
can vary greatly, and in this way, just because one specimen has much greater EMG
magnitude per response, other comparative specimens' data would not be unduly
outweighed when averaged together. All cats are put on the same scale so to speak,
big, small, strong, or weak.
MAV charts the peak absolute EMG, so the baseline values represent the
strength of the muscle contraction when the body is cold. One can infer that stronger
muscle contractions compared to this baseline are elicited for greater demand of
stabilization along the spine. After the 2 hour work/rest period. MAV dropped to
(0.6780 0.2215)mV, (0.6865 0.3056)mV, and (0.7729 0.1945)mV. The
musculature was not contracting as strongly, in fact, they were contracting 28.75%
less than baseline. For the first hour of recovery, MAV continues to drop, have MAV
values of (0.6052 0.1313)mV, (0.6747 0.2869)mV, and (0.7506 0.1766)mV at
the first hour marker, a 32.31 % decline from baseline. After the first hour however,
the MAV jumped up back up toward baseline having a final value of
(1.1681 0.5567)mV, (0.9651 0.3592)mV, and (1.1222 0.3163)mV after the end
of the allotted recovery period. This is an 8.514% increase above baseline. In the
20N case, the variances between the data points over time compared with the original
baseline were not deemed statistically significant; however the data trend is clear,
showing compensation in all channels as recovery progresses post-work. Figure 5.14
65


shows the change of PMAV during recovery as percent variations from baseline and
the statistically significance values, or rather lack thereof.
L3-4 L4-5 L5-6
Time (min)
Figure 5.14 percent change of 20N MAV stretch phase from baseline during recovery period
66


5.6 60N MAY:
L3-4 L4-5 L5-6
Figure 5.15 composite MAY trend for 9 specimens at 60N during recovery period
Table 5.11 function parameters for the fitted curves of 60N MAY model
( t-T, ^ ( ~Tr \
P + P 1 0 T 1 L e Xpi + PM 1 e Xpl
V ) \ /
Ch. L34 L45 L56
n -4.613 -4.285 -3.999
pL 5.358 5.122 4.822
Pm 5.878 5.413 5.273
TP\ 63.6 38.36 59.08
T p2 75.35 45.98 75.05
f f JsLl ' f t-T, \
+ Pl e Tp' + Pm l-e rp2 + (t-rr)PH e Tp}
\ V v )
Ch. L34 L45 L56
Ph 0.00263 0.00185 0.003642
TP3 400 400 400
384 .3 316.7 285.9
R2 0.9915 0.9873 0.9977
67


Table 5.12 p values for the independent variables of
the 1-way ANOVA test for 60N MAV data and the
results of the piece-wise tests
Ch. L34 L45 L56
R2 0.65 0.78 0.59
spec. <0.001 <0.001 <0.001
time <0.001 0.0053 <0.001
Student's t (-) 3, (+) 9 none (+) 8-9
Table 5.11 gives the model values used in the fitted curves for each channel
of the 60N MAV case. Table 5.12 gives the statistical results of the ANOVA and
subsequent piecewise tests for the same data. The Students t test shows the MAV of
the L3-4 level at the first hour after recovery being significant below baseline and
then the MAV at the 7th hour as significantly above baseline. L4-5 has no significant
points away from the baseline while L5-6 shows the last two hours of recovery as
being significantly above baseline.
MAV data for the 60N case were all normalized about the baseline, making
lmV the baseline standard for the entire set. Post-work, the MAV dropped to
(0.7410 0.3324)mV, (0.8301 0.7615)mV, and (0.8308 0.6547)mV, 19.94%
below baseline. As with the 20 N case, the first hour of recovery had MAV
decreasing, with values of (0.6424 0.2707)mV, (0.6607 0.4342)mV, and
(0.6594 0.4679)mV at the 1st hour marker, 34.59% below baseline. The MAV then
drastically increased, exceeding significantly above baseline by the 7th hour of
recovery. The numerical values were found to be (1.4907 0.8602)mV,
(1.3929 0.9084)mV, and (1.7320 1.2293)mV, a 53.85% increase from baseline.
Figure 5.16 shows the percent change off baseline and significant values.
68


Time (min)
Figure 5.16 percent change of 60N MAY stretch phase from baseline during recovery period
69


5.7 20N MF
L3-4 L4-5 L5-6
Figure 5.17 composite MF trend for 8 specimens at 20N during recovery period
Table 5.13 function parameters for the fitted curves of 20N MF model
( t-Tr > f '
+ Fl e + fm 1 e TF1 t <
v > V y
Ch. L34 L45 L56
-270.9 -285.8 -186.2
fl 491.3 484.7 371.3
fm 500 500 383.9
Tfx 125.1 97.41 38.71
TF2 130 100 38.71
( ~Tr ^ ' t-Tr \
+ fl e If' + fm l-e tFi + I "I
\ 7 K J
Ch. L34 L45 L56
F 0.13 0.2 0.02398
TF2 80.16 55.49 356.6
Td 341 274.6 400
R2 0.9564 0.9716 0.9219
( JzIl \
\

t > r.
70


Table 5.14 p values for the independent variables
of the 1-way ANOVA test for 20N MF data and
the results of the piece-wise tests
Ch. L34 L45 L56
R2 0.69 0.74 0.8
spec. <0.001 <0.001 <0.001
time 0.2683 0.0081 0.1499
Student's t (-) 2,4 (-) 1 -5 (-) 1
Table 5.13 gives the model values used in the fitted curves for each channel
of the 20N MF case. Median frequency has the same model as mean absolute voltage
as it should shadow the same trend when fatigue is not a factor. Table 5.14 gives the
statistical results of the ANOVA and subsequent piecewise tests for the same data.
Like MAV, a 1-way ANOVA was applied to MF. The Students t test shows
significance at the 2nd and 4th tests for the L3-4 level, significance for all points up
and including the 5th test for L4-5, and significance at only the first data point during
recovery for L5-6. The high p values associated with the independent time constant
for L3-4 and L5-6 indicate weak significance within those levels. The points found to
be significantly variable from baseline for those channels are probably not, however
L4-5 has a lower p value exhibiting confidence in the accuracy of the Students t test
for that channel.
Median frequency baselines for the 20N case were (231.8115 22.1829)Hz,
(221.1914 16.3660)Hz, and (199.7768 17.9968)Hz. As the median frequency is
directly proportional to motor unit recruitment, when MF increases, motor unit
recruitment increases, activating more muscle percentage. MF should shadow the
same trend of the same MAV in the absence fatigue. In that vein, MF values post-
work are (220.4241 15.6830)Hz, (198.4863 30.5077)Hz, and
(184.5703 22.6006)Hz, 8.001% below baseline and increase along the same trend as
the MAV, showing the specimens in this experiment didnt go through fatigue. The
MF for each channel had statistical significant data points within the first three hours
of recovery below baseline, but MF gradually increased back to baseline over,
71


however never reaching it. MF values leveled off right below baseline by the 4th hour
of recovery and remained relatively constant thereafter having final values of
(227.4170123.1476)Hz, (213.0127121,3280)Hz, and (200.75331 16.0352)Hz at the
end of the 7 hours of recovery, 1.483% below baseline. Figure 5.18 shows the
percent change during recovery and statistical significance.
L3-4 L4-5 L5-6
10
5 -
>%
o
c
(1)
D
cr
c
ro
0)
2
-10
-15
-20 -
-25
n = 8
20N
UTl'

n = 8
20N

f
Statistically significant from baseline
Weak significance, high p value
JL-
n = 8
20N
0 100 200 300 400 500 600 0 100 200 300 400 500 600 0 100 200 300 400 500 600
Time (min)
Figure 5.18 percent change of 20N MF stretch phase from baseline during recovery period
72


5.8 60N MF
L3-4 L4-5 L5-6
Figure 5.19 composite MF trend for 9 specimens at 60N during recovery period
Table 5.15 function parameters for the fitted curves of 60N MF model
r JsLl ^
F0+FL e T" + fm 1 e TF1
V y v /
Ch. L34 L45 L56
F0 -182 -128.7 -129.3
Fl 390 319.9 314.6
FM 400 333.7 333.4
Tpx 38.46 170 130.9
tf2 40 172.6 141
( f t-Tr N r Jzh. N
+ fl e Tf' + fm l-e TF1 + I 't t?
\ v ) V >
Ch. L34 L45 L56
F 0.02726 0.1215 0.05862
TF3 122.3 207.3 187.3
275 334.1 291.4
R2 0.9919 0.9963 0.993
73


Table 5.16 p values for the independent variables of the
1-way ANOVA test for 60N MF data and the results of
the piece-wise tests
Ch. L34 L45 L56
R2 0.76 0.75 0.79
spec. <0.001 <0.001 <0.001
time 0.1228 0.0254 <0.001
Student's t (+) 6-9 (+) 8-9 (-) 2, (+)
Table 5.15 gives the model values used in the fitted curves for each channel
of the 60N MF case.
74


Table 5.16 gives the statistical results of the ANOVA and subsequent
piecewise tests for the same data.
For the 60N MF case, baseline was found to be (206.8142 20.7779)Hz,
(196.0720 16.8848)Hz, and (193.5764 17.1022)Hz for the respective channels L3-
4, L4-5, and L5-6. Post-work MF values were much higher with respect to baseline
in the 60N than the 20N, however being 1.736% below baseline at values of
(207.8993 33.7460)Hz, (191.1892 35.8339)Hz, and (185.32991 14.7601)Hz.
Only in the L5-6 was there any statistical significance below baseline, however in all
channels, MF increased significantly above baseline by the 5th hour of recovery. The
final MF was recorded at (218.9670122.5392)Hz, (211.69701 20.5240)Hz, and
(202.9080+ 17.2687)Hz, 6.879% above baseline. Figure 5,20 shows the percent
change of the MF during recovery referenced from baseline.
L3-4 L4-5 L5-6
S'
c
CT
<1)
c
ro
T3
20
15
10
5
0
-5
-10
-15
-20
-25
0 100 200 300 400 500 600 0 100 200 300 400 500 600 0 100 200 300 400 500 600
n = 9
20N

Statistically significant from baseline
Weak significance, high p value
n = 9
20N
*
Time (min)
Figure 5.20 percent change of 60N MF stretch phase from baseline during recovery period
75


5.9 20N Creep:
By maintaining the same load weight per experimental protocol, the
displacement by each work-cycle was recorded to track the creep elicited on the
targeted L4-5 supraspinous ligament and surrounding tissue over time. The creep
within each specimen followed an inverse increasing exponential curve until maxing
at an individual threshold limit during the work-loading periods. During the recovery
period, the displacement elicited at each test hour showed the creep slowly track back
down toward the initial baseline. Each specimens response within the weight
brackets were averaged over time for non-specific representation of feline lumbar
creep given a certain loading scheme. Figure 5.21 (A),(B) shows the averaged peak
displacement and creep plots for the 20N group.
Figure 5.21 (A) 8 specimen average peak displacement illicited along the supraspinous ligament
during 20N cyclic loading and recovery periods
(B) corresponding creep accumulation during loading and recovery
76


The 20N set had a baseline peak displacement of 6.24mm and a maximal
stretch of 12.8mm which is 99% creep off baseline at the end of the work period. At
the start of recovery, the peak displacement was found at 10.56mm, 63.95% creep
and by the 7 hour peak displacement was 7.67mm, 18.96% creep, still significantly
above baseline. Creep never statistically reached baseline throughout the entire 7
hour recovery period; 7 hours of rest is not sufficient to recovery from 1 hour of
cumulative cyclic work. Based off the linear regression, 8 total hours would have
seen peak displacement return to baseline. Only peak displacement values were
statistically analyzed through a 1-way ANOVA but these findings are analogous to
the creep plot as well. Table 5.17 gives the statistical results for the 20N peak
displacement Students t test
Table 5.17 p values for the independent variables of the
1-way ANOVA test for 20N peak displacement
data and the results of the piece-wise tests
R2 0.87
spec. <0.001
time <0.001
Student's t (+) 1-9
77


5.10 60N Creep:
Figure 5.22(A),(B) shows the averaged peak displacement and creep plots for
60N group.
Figure 5.22 (A) 9 specimen average peak displacement illicited along the supraspinous ligament
during 60N cyclic loading and recovery periods
(B) corresponding creep accumulation during loading and recovery
The 60N set had a baseline of 10.94mm and a maximal stretch of 19.76mm at
the end of the work period, a 77.23% creep above baseline. During recovery, peak
displacement traveled from 18.15mm with a creep percentage of 62.78% to 14.53mm
with creep of 30.2%. Even after 7 hours of immobile rest, the original pre-loaded
baseline status is not recovered. Extrapolated regression shows creep returning to
baseline by as much as 4 more hours after the 7 hour recovery period. Table 5.18
shows the results of the statistical analysis for the 60N peak displacement.
78


Table 5.18 p values for the independent variables of the
1-way ANOVA test for 60N peak displacement
data and the results ofthe piece-wise test
R2 0.96
spec. <0.001
time <0.001
Student's t (+) 1-9
A prolonged rest period with much longer duration than the working period is
necessary to return the lumbar physiology back the initial condition. Furthermore, the
creep magnitude in both the 20N and 60N exceed 70% of the initial stretch
displacement after only 2 hours of cyclic work. Continuous cyclic work from both
the light and heavy loads taxed the tissues significantly. Such a dramatic change to
the response of the ligament connecting the vertebrae lends to understanding of how
debilitating accrued damage can be from cyclic work.
79


6. Discussion
The intent of this study was of an exploratory and comparative nature.
Previous studies have investigated the extent of static and cyclic loading on the lower
lumbar spinal region of feline models. Defining 20N, 40N, and 60N loading
protocols as light, moderate, and heavy loads for a feline specimen, this study
concludes the series of experimentation for both loading patterns for all weights. The
tolls of extended cyclic loading on the lumbar region for the 20N and 60N cases were
analyzed for evaluation against the other studies.45,37,19
Concurring with the previous work, after two hours of work a weakened
physical state is established in the lumbar region due to the compiled loading
enabling the failing of passive and active spinal stabilizing systems to meet pre-work
baseline conditions. Neuromuscular neutral zones for both displacement and tension
were found to have significantly increased while EMG mean absolute voltage and
EMG median frequency decreased. These factors all agree to indicate a loss of
overall stability due to subpar performance from stabilizing paraspinal musculature,
specifically the multifidus muscles. Such instability was apparent up to three to four
hours immediately after the cessation of the loading segment, after which an apparent
compensation mechanism commenced to drive the insufficient NNZs, MAV, and
MF devices back to the original baseline status and beyond as a reaction to the
evident weakened physical state. In many cases, compensation beyond baseline
continued for the remainder of the designated recovery period without termination,
indicating further recovery as necessary beyond the monitored 7 hour period.
80


6.1 Spinal Instability:
6.1.1 Increased Unprotected Neutral Zones
As a result of loading to the lumbar region, laxity is induced in the
viscoelastic tissues that form a retaining structure for the spinal column. While these
passive tissues, including ligaments, tendons, and discs, are responsible more for
spinal integrity rather than rigidity, they do house significant mechanoreceptors that
relay neural feedback of proprioception and kinematics that trigger activity in the
stabilizing musculature; it follows that such laxity becomes problematic with the
internal spinal stability control. When these viscoelastic tissues are strained to a
certain threshold, the mechanoreceptors are deformed and activate to signal an
involuntary response. In most cases this results in deep-tissue muscle activity that
arrests the spinal column, reinforcing its structural cohesion and guiding movement.
The laxity engendered from the cyclic loading is caused by the nature of the
viscoelastic tissues to exhibit creep, and this, as a result, disrupts the communication
between the mechanoreceptors and the stabilizing musculature. As the viscoelastic
tissues are burdened to capacity, individual collagen fibers that compose the tissue
break down. With more and more collagen fibers breaking, the entire tissue stretches
to longer displacements. Like a rubber band that has become limp from extended use,
greater amounts of displacement are necessary to obtain the same tensile thresholds.
As a result, the necessary deformation to the nerve ending of the mechanoreceptors
are developed with greater degrees of extension or flexion, thus shifting integrated
physiological systems out of phase with one another and delaying nominal responses
as a result.
This is clearly demonstrated in the elevated DNNZ and TNNZ values post
work. Greater amounts of displacement or tension applied to the supraspinous
ligament are required to encourage muscular reaction. Remembering that fatigue was
not apparent, the blame for such a disparity can only be attributed to the damage
81


sustained from the hard labor. And as this damage is definitively connected to the
extent of creep sustained, it is clear that creep is disturbing the feedback harmony
dictated through the somatosensory elements within the spinal lumbar region.
The damages associated with creep illustrate a deficiency of the viscoelastic
net that houses all vertebrae and limits the extent of internal mobility. A deficiency
therein allows for greater unsanctioned movement and less stability. More muscular
intervention is needed to maintain proper spinal function and to protect against injury.
However, the delayed response times from stabilizing muscles is counter intuitive to
this as less muscular support is in fact allocated which only contributes further to the
risk of injury.
6.1.2 Deficient Muscle Response
Decreased levels of EMG, as displayed with sub-par MAV and MF levels also
indicate weakened muscle responses post work. This could be due to the elevated
neutral zones, signaling the body to apply less muscular pressure, or could to an
artifact from the sustained loading. Evaluating single cycle tests, the first load
applied to a specimen tends to activate a much greater EMG response compared to
subsequent loads afterwards. Muscles seem to react with greater force to the first
stimuli, whereas similar stimuli administered after adequate rest elicit dampened
responses. The muscles seem to become more acclimated with the burden and adjust
their response accordingly. Figure 6.1 illustrates this behavior, displaying three
single cycle responses from the same feline specimen as a reaction to the same 20N
load before any previous loading was employed. Similar with the work period and
recovery period, each complete load cycle had a 4 second duration. The single cycles
were monitored over a 16 second window within 10 minutes rest in between, more
than adequate time to recover from any creep or fatigue sustained.
82


Time (sec)
Figure 6.1 consecutive 16 sec single cycle EMG response along L4-5
The first EMG response to loading can be seen in SCI. After 10 minutes rest,
the second single cycle response can be seen in SC2 and so on. It is evident that each
subsequent response is not as pronounced and does not last as long. Not only is the
entire signal duration shortening, coinciding with an increase in NNZs, but the signal
amplitude is falling as well. From direct observation, it is reasonable that the peak
MAV is decreasing. Like an athlete warming up, the same is seen as the responses
become more dulled and controlled in order to improve performance. This tendency
can also be seen post work, where overall EMG is much less compared to initial
baseline tests; the physiology seems to maintain a responsive mode that expects
successive loading.
This indicates that the body physiologically recognizes work being applied
and compromises its own stability in order to increase work output and efficiency.
Instability can lead to injury but so can over-stability. If the back muscles are too
83


rigorous in their attempt to stiffen the spine, this could lead to pulling of muscles or
ripping of connective tissues. This is the main reason one stretches before attempting
laborious activity; to warm up as well as loosen up. It has been noted in
epidemiological studies that injuries or accidents occur more frequently toward the
end of work when people are inattentive and tiredness masks the level of instability in
the low back; in effect they take for granted the accumulated deficiencies that
accompany heavy labor, however 47% of the first episode of back pain happens at
the beginning of work.5,42,11 Over time, the body becomes used to the level of work
demanded of it. Such work anticipation must be a cognitive compromising of the
internal stability network by restraining the muscular counter effort during movement
which would explain drops in MAV and MF over time.
The inherent problem therein lies in the fact that such depletion of muscular
activity leaves the subject much more prone to injury, especially as work continues
over time. Furthermore, in post work scenarios, when the body starts to cool down
over all and relax, the cessation of work does not indicate an immediate return to full
stability. This is seen in the continued decent of MAV and MF during the first hour
after work and continued increase of TNNZ likewise. This becomes a time of high
risk after the work period is completed as the damaging effects of the cyclic loading
are most prominent and leaves the lumbar spinal area most prone to injury with the
cessation in overall muscle exertion. Many physicians experience such cases where a
patient claims the full work day went well but shortly thereafter, while bending down
to retrieve a lunch pail or to tie ones shoes, something in the back popped.
6.2 Compensation Mechanism:
Allowed to rest fully, the effects of creep begin to recover immediately. In a
parallel fashion, the displacement neutral zones steadily shrink back to baseline over
the recovery period. The speed of such recovery is generally limited to the rate to
which the collagen fibers can be repaired, thus stiffening the viscoelastic tissues and
84


Full Text
38. Solomonow, M., B.H. Zhou, R. Barrata, Y. Lu, and M. Harris, Biomechanics
of increased exposure to lumbar injury caused by cyclic loading: Part 1. Loss
of reflexive muscular stabilization. Spine 24, no. 7 (1999): 2426-2443.
39. Solomonow, M., C. Baten, J. Smit, R. Baratta, H. Hermens, R. D'Ambrosia,
and H. Shoji. Electromyogram power spectra frequencies associated with
motor unit recruitment strategies. Journal of Applied Physiology 68, no. 3
(1990): 1177-1185.
40. Solomonow, M., R.V. Baratta, B.H. Zhou, E. Burger, A. Zieske, and A.
Gedalia. Muscular dysfunction elicited by creep of lumbar viscoelastic
tissues. Journal of Electromyography and Kinesiology 13, no. 4 (2003): 381
396.
41. Stulen, F.B. and C.J. DeLuca. Frequency Parameters of the Myoelectric
Signal as a Measure of Muscle Conduction Velocity. IEEE Transactions on
Biomedical Engineering 28, no. 7 (1981): 515-523.
42. Varma, K.M.K. First episode in low back pain. M.Ch. Orth, thesis,
University of Liverpool, 1987
43. Wesley, N. Anatomical Planes of the Body. (1999).
http://home.comcast.net/~wnor/terminologyanatplanes.htm (accessed March
3, 2009).
44. Whitmore, M., S. Li, and L. Huang. LPD lipopolyplex initiates a potent
cytokine response and inhibits tumor growth Gene Therapy 6, no. 11 (1999):
1867-1875
45. Youseff, Jimmy. Neuromuscular Neutral Zones Response to Static Lumbar
Flexion. Masters thesis, University of Colorado Denver, 2008.
46. Paraspinal Muscles. All About Back and Neck Pain (2007).
http://www.allaboutbackpain.com/html/files links/spine anatomy muscle.ht
ml (accessed March 3, 2009).
47. Divisions of the Skeleton. SEER Web-based Training Modules (2008).
http://training.seer.cancer.gov/module anatomv/unit3 5 skeleton divisions.ht
ml# (accessed March 3, 2009).
99


48. Planes of the Body. SEER Web-based Training Modules (2008).
http://training.seer.cancer.gov/module anatomy/unitl 3 terminology2 planes
.html (accessed March 3, 2009).
49. Action potential. Wikipedia (2009).
http://en.wikipedia.org/wiki/Actipn_pptential (accessed April 16, 2009).
50. Creep (deformation). Wikipedia (2009).
http://en.wikipedia.org/wiki/Creep (deformation) (accessed March 3, 2009).
51. Cytokine. Wikipedia (2009).
http://en.wikipedia.org/wiki/Cytokine (accessed March 3, 2009).
52. Electromyography. Wikipedia (2009).
http://en.wikipedia.org/wiki/Electromyography (accessed March 3, 2009).
53. Mechanoreceptor. Wikipedia (2009).
http://en.wikipedia.org/wiki/Mechanoreceptor (accessed April 21, 2009).
54. Neutrophil granulocyte. Wikipedia (2009).
http://en.wikipedia.org/wiki/Neutrophil granulocyte (accessed March 3,
2009).
55. Pacinian corpuscle. Wikipedia (2009).
http://en.wikipedia.org/wiki/Pacinian corpuscle (accessed April 21, 2009).
56. Supraspinous ligament. Wikipedia (2009).
http://en.wikipedia.org/wiki/Supraspinous ligament (accessed March 3,
2009).
57. Type 1 error. Wikipedia (2009).
http://en.wikipedia.org/wiki/Type 1 error (accessed March 3, 2009).
58. Vertebra. Wikipedia (2009).
http://en.wikipedia.org/wiki/Vertebra (accessed March 3, 2009).
100