Close
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Dynamic knee loading asymmetries following anterior cruciate ligament reconstruction: methods for clinical detection
(USC Thesis Other)
Dynamic knee loading asymmetries following anterior cruciate ligament reconstruction: methods for clinical detection
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Running head: DYNAMIC KNEE LOADING ASYMMETRIES i
DYNAMIC KNEE LOADING ASYMMETRIES
FOLLOWING ANTERIOR CRUCIATE LIGAMENT RECONSTRUCTION:
METHODS FOR CLINICAL DETECTION
by
Kristamarie Anne Pratt
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(BIOKINESIOLOGY)
May 2016
DYNAMIC KNEE LOADING ASYMMETRIES ii
DEDICATION
I dedicate this dissertation to my rocks:
Momma, Daddy −o, Katrina & Stephanie…
You give me life and continue support me through it all
It’s because of you that when I get the choice to sit it out or dance,
I DANCE!!
DYNAMIC KNEE LOADING ASYMMETRIES iii
ACKNOWLEDGEMENTS
Over the past five years I embarked on a challenging, yet rewarding journey to complete
this dissertation. It was a roller coaster ride, but I am so thankful for the opportunity to study
and research a subject I am passionate about. I would like to thank the USC Division of
Biokinesiology and Physical Therapy for the generous financial support that made this
dissertation possible and for welcoming me into the Trojan BKN & PT family. Thank you to Dr.
Jacquelin Perry for her generous endowment of our program and pioneering this field of
biomechanics and physical therapy. I would also like to thank the Division Chair, Dr. Jim Gordon,
who I had a pleasure of not only learning from in class and division seminars, but also during coffee
chats post workouts. Thank you for your unwavering support and your leadership within our
division.
First and foremost, I would like to recognize my advisor, Dr. Susan Sigward. Thank you
for taking me under your wing and teaching the engineer in me how to think like a clinician, and
translate my knowledge and expertise into clinical practice. Thank you for fostering a prosperous
environment for us to develop into a well −rounded independent researchers and teachers. Thanks
for being my anchor throughout this journey, for believing in me when I did not believe in myself,
and for being my workout buddy and constant coffee date. I would also like to thank my
dissertation committee who challenged me and facilitated my transformation to the independent
researcher I am today. Thank you for your guidance and expertise in developing this dissertation:
Dr. Robert Gregor for forcing me to think outside the box and about the body as a whole system;
Dr. Robert Keim for your statistics wisdom and “how you doing?” talks; Dr. Kornelia Kulig for our
intellectual discussions on either the whiteboard or over a walk outside and for your visits to
DYNAMIC KNEE LOADING ASYMMETRIES iv
G −12’s corn hole; and Dr. George Salem for your extensive knowledge and insight in studying this
patient population.
In addition to my dissertation committee members, numerous faculty in the USC Division
of Biokinesiology and Physical Therapy have contributed to my academic and research
development over the past several years: Dr. Christopher Powers for your questions and guidance
in lab meetings and value of socialization outside the lab; Dr. Todd Schroeder for the experience
with ETScience and comedic relief; Dr. Nina Bradley for your friendly smile and chats in the
hallways; Dr. Lucinda Baker for being the mother hen of G −12; and Dr. Lori Michener and Dr. Julie
Tilson for your help in implementing analyses to translate the findings from this dissertation into
the clinic. A special thanks for the endless support and kindness of the Division of Biokinesiology
and Physical Therapy staff. Thank you David, Matt, Chad, Janet, Veronica, Jennifer, Weslie,
Virginia, Troy, Sara and Lydia for your help keeping us graduate students in check and the division
working as a well −oiled machine.
I would like to acknowledge the founders of CATZ, Jim and Dan. Thanks for keeping me
in shape over these last five years and for your amazing facility that the Human Performance Lab
is thankful to be a part of. Thank you to the staff and physical therapists at CATZ: Chris, AG,
Kevin, Maureen, Bianca and Audra, who helped recruit participants. Also a special thanks to the
PT residents, faculty and staff at Marengo and UPC faculty practices who aided in recruitment
too. I am particularly grateful for my study participants as completing my dissertation became
possible with each and every one of you.
I would like to thank my colleagues, old and new, who have been there through the ups
and downs that are an integral part of the PhD journey. They say it takes a village to raise a child,
well it takes a village to earn a PhD. Thank you to my MBRL elders for the knowledge you shared
DYNAMIC KNEE LOADING ASYMMETRIES v
and examples you set: John, Shawn, Ching, Ping, Kaiyu, Kristen, Marks, Rami, Sachi, Eugene,
Sharon, Jenny and Jo. To my past & current G −12 residents: Tulika, Michael, Jia, Jennifer, Jasmine,
Steffi and Sapna, thank you for the knowledge we exchanged, the guidance to excel, an ear to vent,
a shoulder for support and laughs to remain sane. To my fellow PhD buddy and desk neighbor,
Danielle, thanks for the chats, races we ran, and welcoming me into LA with open arms. We did
it! To iron mom Jackie, thanks for providing an escape from the basement and ray of ExPhys
sunshine. To my analytical teaching partners Yo & Jonathan thank you and don’t forget to use
colored chalk.
The Human Performance Lab and its members will always have a special place in my heart.
Thanks to my big sis Kate, big bro Gui, little bro Matt, little sis Paige and Sarah. Thanks for
everything: intellectual discussions, advice, team work, lunch dates, delicious food, sarcasm and
laughs. Go team! A big thanks to all those I had the opportunity to mentor in my time here: DPT
students, Brazilian exchange students and undergraduates. Thank you for your enthusiasm in
learning about research and providing me with the opportunity to teach you. Thank you for your
countless hours helping collect, label and process data. Thank you Olivia, Lindsay, Rachel, Maria.
I would also like to thank the 5 DPT classes who have molded me into the teacher I am today. You
taught me as much as I taught you. A special thanks to the DPT class of 2016 for welcoming into
your cohort with opens arms when I need support the most. Thank you to the Prestigious
Plantarii softball team for Tuesday nights, the ladies for all your loving energy, and my
cheerleaders, Shawna & Chelsea, for your endless hugs and unwavering support.
I am particularly grateful for my family and friends, near and far, for their love and care
over the last several years. 20+ couples were wed, ~5 couples were engaged, ~16 babies were born
and endless memories were made during the production of this dissertation. I would like to thank
DYNAMIC KNEE LOADING ASYMMETRIES vi
you all for continued support despite life’s busy −ness. Thanks to Lindsay for the early morning
runs and little Olivia to remind me about life through a child’s eyes. Thanks to Sanna for roomie
support and yogurt dates, Willbees for the cookies, and Emily for simple texts here and there just
letting me know you are always there. Thanks to the Kirkwoods for being my second family.
Antonia, thanks for moving 3,000 miles across the country with me to pursue our PhDs and for
always understanding our mutual love for NYC and the East Coast. And a finally, a big thank you
to Andrea for not only being my PhD lab mate to talk business with, but for being a really good
friend who knew when I needed a night off before I did and then would take the time off just to
be with me. Thanks for reminding me about balance in life.
I would also like to thank my Uncle Jerry, for leading by example: persevering when the
going got tough and yet somehow making us laugh amidst it all; “Hakuna matata.” Thank you to
George Boiardi for touching my life in a unique way at that lacrosse game in 2004 and inspiring
us always to follow our dreams; “well done is better than well said.” A special thanks to my west
coast family: Aunt Wendy, Uncle David and Jamie. Thank you for welcoming me into the USC
family and always being there.
Finally, I would like to thank my sisters and my parents. Katrina & Stephanie: thank you
for your love, positive thinking, listening sessions, advice and care. Mom & Dad: thank you for
being above average role models, teaching me the value of hard work and being patient as I find
my way. Thank you to my family for believing in me always and only ever being a plane ride away.
My dreams become a reality because of you. This dissertation is finally complete because of you.
Believe it as you read it!
DYNAMIC KNEE LOADING ASYMMETRIES vii
TABLE OF CONTENTS
DEDICATION .............................................................................................................................................. ii
ACKNOWLEDGEMENTS .............................................................................................................. ........ iii
TABLE OF CONTENTS .......................................................................................................................... vii
LIST OF TABLES ........................................................................................................................................ ix
LIST OF FIGURES ...................................................................................................................................... x
ABSTRACT ................................................................................................................................................ xii
CHAPTER I: OVERVIEW ......................................................................................................................... 1
CHAPTER II: BACKGROUND & SIGNIFICANCE ......................................................................... 4
STATEMENT OF THE PROBLEM ...................................................................................................... ...................... 4
LOADING STRATEGIES FOLLOWING ACL RECONSTRUCTION ........................................................... 5
QUANTIFYING DYNAMICS OF MOVEMENT ................................................................................................... 6
CLINICAL MOVEMENT ASSESSMENTS .............................................................................................................. 7
MARKER −BASED MOTION CAPTURE VERSUS INERTIAL SENSOR MOTION ANALYSIS ............ 8
CLINICAL DETECTION OF LOWER EXTREMITY LOADING IMPAIRMENTS USING INERTIAL
SENSORS ........................................................................................................................................................................... 9
SUMMARY ..................................................................................................................................................................... 10
CHAPTER III: DYNAMIC KNEE LOADING DEFICITS IN SINGLE LIMB LOADING AND
RUNNING IN INDIVIDUALS FOLLOWING ACL RECONSTRUCTION ............................ 12
ABSTRACT ..................................................................................................................................................................... 12
INTRODUCTION ......................................................................................................................................................... 13
METHODS ...................................................................................................................................................................... 16
Participants .................................................................................................................................................................................................................................. 16
Procedures .................................................................................................................... .................................................................................................................. 18
Dynamic Tasks: Single Limb Loading Test and Running............................................................................ ....................................................... 19
Data Analysis .............................................................................................................................................................................................................................. 20
Statistical Analysis .......................................................................................................... ......................................................................................................... 21
RESULTS ........................................................................................................................................................................ 22
DISCUSSION ................................................................................................................................................................. 25
DYNAMIC KNEE LOADING ASYMMETRIES viii
CHAPTER IV: ANGULAR VELOCITIES MEASURED WITH INERTIAL SENSORS
REFLECT DYNAMIC KNEE LOADING DURING SINGLE LIMB LOADING IN
INDVIDUALS FOLLOWING ACL RECONSTRUCTION ........................................................... 30
ABSTRACT .....................................................................................................................................................................30
INTRODUCTION ......................................................................................................................................................... 31
METHODS ..................................................................................................................................................................... 34
Participants .................................................................................................................................................................................................................................. 34
Procedures ..................................................................................................................................................................................................................................... 35
Single Limb Loading Test ..................................................................................................................................................................................................... 38
Data Analysis .............................................................................................................................................................................................................................. 39
Statistical Analysis .......................................................................................................... ......................................................................................................... 41
RESULTS ........................................................................................................................................................................ 42
DISCUSSION ................................................................................................................................................................. 45
CHAPTER V: DETECTING KNEE LOADING ASYMMETRIES IN THE CLINIC:
IMPLICATIONS FOR REHABILITATION FOLLOWING ACL RECONSTRUCTION .....49
ABSTRACT .................................................................................................................................................................... 49
INTRODUCTION ........................................................................................................................................................ 50
METHODS ...................................................................................................................................................................... 53
Participants .................................................................................................................................................................................................................................. 53
Procedures ..................................................................................................................................................................................................................................... 54
Single Limb Loading Test ..................................................................................................................................................................................................... 57
Data Analysis .............................................................................................................................................................................................................................. 58
Statistical Analysis .......................................................................................................... ........................................................................................................ 60
RESULTS ......................................................................................................................................................................... 61
DISCUSSION ................................................................................................................................................................. 66
CHAPTER VI: SUMMARY & CONCLUSIONS ............................................................................. 69
BIBLIOGRAPHY ....................................................................................................................................... 76
DYNAMIC KNEE LOADING ASYMMETRIES ix
LIST OF TABLES
Table 3.1
Subject characteristics of the ACLR and Control Groups (mean
± standard deviation)
17
Table 4.1
Subject demographics (mean ± standard deviation) 35
Table 4.2
Descriptive statistics for the reconstructed (ACLr) and
non −surgical(Non −Sx) limb for joint and segment variables
measured with marker −based motion capture and inertial
sensor measurement systems; Data represents mean ± standard
deviation and (range)
42
Table 4.3
Intraclass correlation coefficients (2,k) between marker −based
motion capture and inertial sensor measurements for peak knee
angular velocity and peak thigh and shank angular velocities
measured in all limbs, the reconstructed (ACLr) and
non −surgical (Non −Sx) limb.
44
Table 5.1
Subject demographics (mean ± standard deviation) 54
Table 5.2
A 2 x 2 table of data from this study for the gold standard
(marker −based knee power (KPow) ratio) and new test
(inertial sensor thigh angular velocity(ThAV) ratio
64
DYNAMIC KNEE LOADING ASYMMETRIES x
LIST OF FIGURES
Figure 3.1
Single Limb Loading Test*
(*subject in figure reproduced with permission)
20
Figure 3.2
Comparisons of (A) peak knee extensor moment (B) peak knee
power absorption, (C) peak knee angular velocity between
reconstructed (ACLr, solid line) and non −surgical (Non −Sx,
dashed line) limb during single limb loading(SLL) and running
(RUN); Data represents mean ± standard deviation; #main
effect for limb; +main effect for task; **p 0.001, *p 0.05
23
Figure 3.3
Comparisons of (A) peak knee extensor moment (B) peak knee
power absorption, (C) peak knee angular velocity between
single limb loading(SLL) and running (RUN) in the control
limb; Data represents mean ± standard deviation; **p 0.001
24
Figure 3.4
The relationship of between limb symmetry ratios during single
limb loading (SLL) and running (RUN) for (A) peak knee
power absorption and (B) peak knee angular velocity; Pearson
product −moment correlation reported; *p 0.05
25
Figure 4.1
Orientation and location of inertial sensors and markers on the
lower extremity during testing
37
Figure 4.2
The relationship between (A) peak knee angular velocities and
(B) peak thigh angular velocities measured with marker −based
motion capture and peak knee power absorption in the
reconstructed (ACLr) and non −surgical (Non −Sx) limb;
**p 0.001
43
Figure 4.3
The relationship between (A) peak knee angular velocities and
(B) peak thigh angular velocities measured with inertial
sensors and peak knee power absorption in the reconstructed
(ACLr) and non −surgical (Non −Sx) limb; **p 0.001
44
Figure 5.1
Orientation and location of inertial sensors and markers on the
lower extremity during single limb loading test
56
Figure 5.2
Single Limb Loading Test*
(*subject in figure reproduced with permission)
58
DYNAMIC KNEE LOADING ASYMMETRIES xi
Figure 5.3
The relationship between thigh angular velocity ratios
extracted from inertial sensors and the knee power symmetry
ratios calculated from marker −based motion capture. A ratio of
1 indicated that ACLr limb Non −Sx limb. A ratio 1 indicated
that ACLr limb Non −Sx limb
62
Figure 5.4
The Receiver Operating Characteristic (ROC) curve for
differentiating between asymmetrical and symmetrical knee
loading during SLL in individuals following ACLr. The ROC
curve provides a visual depiction of sensitivity and specificity
of thigh angular velocity ratio cutoff scores for detecting knee
power ratio 0.85.
63
Figure 5.5
The likelihood ratio nomogram is a graphical representation of
the probability that an individual following ACLr will have
asymmetrical knee power during a single limb loading task.
Pre −test probability was estimated at 76% (based on the
overall percentage of participants in this study with
asymmetrical loading). The red line plots the positive
likelihood ratio (LR+), used when the thigh angular velocity
(ThAV) ratio for an individual is less than or equal to 0.81. It
indicates that the individual has greater than 100% post −test
probability of having asymmetrical knee loading. The blue line
plots the negative likelihood ratio (LR −) used when an
individual exceeds the thigh angular velocity ratio of 0.81. It
indicates the individual has only a 38% post −test probability of
having asymmetrical knee loading.
65
DYNAMIC KNEE LOADING ASYMMETRIES xii
ABSTRACT
Deficits in sagittal plane knee loading continue to exist after rehabilitation in individuals
following anterior cruciate ligament reconstruction (ACLr). The persistence of these deficits
during dynamic activities such as running, cutting or jumping is of concern, particularly for those
returning to high level physical activities, as loading deficits are associated with increased risk for
re −injury. While knee loading strategies following surgery are consistently characterized by
decreased knee flexion angles and extensor moments, less is known about deficits in knee power
and rate of knee loading in this population. Understanding potential deficits in loading rate is
particularly important as individuals’ progress to more dynamic activities during rehabilitation.
Therefore, the primary purpose of this dissertation was to understand how individuals following
ACLr rapidly accommodate forces at the time they progress to running, the first dynamic loading
task during rehabilitation. The aim of the primary objective, presented in Chapter III, was to
compare dynamic loading strategies at the knee between limbs and to controls during two
dynamic tasks that require different magnitudes of loading, in individuals following ACLr, at the
time that they progress to run. Two groups of recreationally active individuals participated;
fifteen healthy controls and fifteen individuals post −ACLr (ACLR). Participants performed three
trials of over ground running and of a single limb loading task. Separate 2x2 repeated measures
analysis of variance were used to compare the effects of limb and task in the ACLR group;
independent t −tests were used to compare between ACLR and Control groups and paired t −tests
between limbs in ACLR group and between tasks in Control group. Control data indicated that
the magnitude of loading was two times greater in running than single limb loading. Decreased
magnitude and rate of loading were observed in the reconstructed compared to the non −surgical
limb during both tasks with larger between limb differences during the more demanding task,
run. The presence of loading rate deficits during the single limb loading task suggests that even
DYNAMIC KNEE LOADING ASYMMETRIES xiii
during tasks that require smaller magnitudes of loading individuals limit their rate of loading. This
may indicate specific impairments in the ability to rapidly load the knee at a time when
individuals are progressed to running during rehabilitation. While a better understanding of how
individuals dynamically engage the knee as they initiate more dynamic tasks may inform
rehabilitation strategies aimed at mitigating sagittal plane loading deficits, identification of these
deficits remain a challenge for clinicians. Difficulty detecting deficits in knee power during
dynamic tasks in individuals following ACLr may underlie the persistence of these deficits over
time. The expense, time, and expertise needed to quantify knee power deficits using the current
gold standard techniques preclude their use in the clinic. As knee joint angular velocity is used to
calculate power, the potential for the use of inertial sensors, gyroscopes, to identify power deficits
in the clinic exists. Therefore, the secondary objective of this dissertation was to determine the
feasibility of objectively quantifying knee power asymmetries during a dynamic single limb tasks
in individuals following ACLr using segment kinematics measured with inertial sensors. To do
this, twenty −one individuals following ACLr were evaluated during the single limb loading task.
The first aim of the secondary objective, presented in Chapter IV, was to determine if angular
velocities measured with inertial sensors can provide meaningful information regarding knee
power absorption during a dynamic single limb loading task. To determine the best predictors of
knee power absorption, separate stepwise linear regressions were performed using thigh, shank
and knee angular velocities calculated from marker −based motion analysis and inertial sensors,
respectively. Intraclass correlation coefficients (2,k) were used to determine concurrent validity.
Knee angular velocity and thigh angular velocity were identified as the best predictors of knee
power when considering angular velocities extracted from the marker −based system and from
inertial sensors, respectively. Both were strong predictors explaining nearly seventy percent of
the variance in knee power. High intraclass correlation coefficients indicated strong agreement
DYNAMIC KNEE LOADING ASYMMETRIES xiv
between measurement systems. Together, these data suggested that information from inertial
sensors positioned on the thigh provide meaningful information regarding knee power which may
be helpful in identifying deficits clinically. The second aim of the secondary objective, presented
in Chapter V, was to determine the diagnostic accuracy of using inertial sensor thigh angular
velocities to detect asymmetrical knee loading during a dynamic single limb loading task in
individuals following ACLr. Between limb ratios (reconstructed/non −surgical limb) were
calculated for knee power, calculated from marker −based motion system, and thigh angular
velocity extracted from inertial sensors. To determine the relationship between ratios, a linear
regression was performed using knee power ratios and thigh angular velocity ratios. Sensitivity
and specificity of thigh angular velocity in diagnosing asymmetrical knee loading, knee power
ratio less than 0.85, was determined using receiver operating characteristic curve analysis. Thigh
angular velocity ratios were strong predictors of knee power ratios. Thigh angular velocity ratios
could diagnose asymmetrical knee loading, knee power deficits greater than 15% when performing
the single limb loading task, with high sensitivity and specificity. These findings support the use
of cost −effective wearable sensors to objectively quantify movement in the clinic in individuals
following ACLr. Taken together, the results of this dissertation indicate that individuals following
ACLr have dynamic knee loading deficits not only in magnitude, but also in rate of loading during
dynamic tasks at the time they are progressed to running; and inertial sensor measurements, in
particular sagittal plane thigh angular velocity, can be used to identify dynamic knee loading
deficits in the clinic. Providing clinicians with a tool to identify altered loading patterns during
dynamic tasks may reduce risk for re −injury by allowing for focused attention on ameliorating
deficits in loading rate during dynamic tasks.
DYNAMIC KNEE LOADING ASYMMETRIES 1
CHAPTER I
OVERVIEW
Despite improvements in evidence based rehabilitation programs, deficits in sagittal plane
knee loading persist after rehabilitation is complete in individuals following anterior cruciate
ligament reconstruction (ACLr) (Noehren et al., 2013; Oberlander et al., 2014; Ortiz et al., 2011;
Webster et al., 2012). The persistence of these deficits during dynamic activities (i.e. running,
cutting, jumping) is of concern, particularly for those returning to high level physical activities, as
loading deficits have been associated with increased risk for re −injury (Paterno et al., 2010).
Strategies that shift the demands away from the knee extensors in the reconstructed limb during
dynamic tasks are well documented in individuals following ACLr. While these strategies are
consistently characterized by decreased knee flexion angles and extensor moments, less is known
about deficits in knee power and rate of knee loading in this population. Studies that have
reported knee power in individuals following ACLr, during running and hopping, found that
deficits in knee power are observed along with smaller deficits in knee extensor moments. These
data suggest that individuals have difficulty meeting the demands of increased velocity during
dynamic tasks (Devita et al., 1992; Orishimo et al., 2010) as power not only reflects the magnitude
(joint moment) but also the rate (joint angular velocity) of knee loading. Understanding potential
deficits in loading rate is particularly important as individuals’ progress to more dynamic
activities during rehabilitation. Given the potential long term consequences of persistent deficits,
identification and improvement of altered loading patterns prior to returning to previous
activities is important. Therefore, the primary objective of this study was to understand how
DYNAMIC KNEE LOADING ASYMMETRIES 2
individuals following ACLr rapidly accommodate forces at the time they are progressed to
running, the first dynamic loading task during rehabilitation.
While a better understanding of how individuals dynamically engage the knee as they
initiate more dynamic tasks may inform rehabilitation strategies aimed at mitigating sagittal
plane loading deficits, identification of these deficits remain a challenge for clinicians. Clinically,
it is difficult to observe sagittal plane knee loading deficits as they often coincide with smaller
alterations in joint angles and during tasks that are preformed quickly (Orishimo et al., 2010;
Salem et al., 2003). The expense, time and expertise needed to quantify knee loading deficits using
the current gold standard techniques preclude their use in the clinic. Given that knee joint angular
velocity is used to calculate power, the potential use of inertial sensors, particularly gyroscopes,
to identify power deficits in the clinic is possible. However, it is not clear if the strength of the
relationship between segment kinematics and knee power absorption during dynamic tasks will
be adequate for use in clinical testing. Moreover, it is not known if clinical testing procedures
using inertial sensors are capable of providing diagnostic accuracy that is sufficient for clinical
diagnosis of loading impairments. Therefore, the secondary objective of this study was to
determine the feasibility of objectively quantifying knee power asymmetries during a dynamic
single limb task in individuals following ACLr using segment kinematics measured with inertial
sensors.
The specific aim and associated hypotheses for the primary objective:
Specific Aim 1: To investigate dynamic knee loading (power absorption and angular velocity),
during two dynamic tasks that require different magnitudes of loading in individuals following
ACLr, at the time they progress to running (Chapter III). It was hypothesized that individuals
DYNAMIC KNEE LOADING ASYMMETRIES 3
following ACLr would exhibit decreased knee power and angular velocity in the reconstructed
limb during both tasks and that these deficits would be larger during the more demanding task.
The specific aims and associated hypotheses of the secondary objective:
Specific Aim 2: To determine if segment angular velocities measured with inertial sensors can
provide meaningful information regarding knee power absorption during a dynamic single limb
loading task in individuals following ACLr (Chapter IV). It was hypothesized that angular velocities
would be good predictors of knee power and that strong agreement between measurement
systems would be found for both joint and segment angular velocities.
Specific Aim 3: To determine if measurements of thigh angular velocity during a dynamic single
limb loading task in individuals following ACLr can be used as a proxy for knee power absorption
clinically by providing appropriate information regarding between limb deficits (Chapter V). It
was hypothesized that thigh angular velocity measurements could detect asymmetrical knee power
with high sensitivity and specificity.
DYNAMIC KNEE LOADING ASYMMETRIES 4
CHAPTER II
BACKGROUND & SIGNIFICANCE
Statement of the Problem
An estimated 250,000 individuals tear their anterior cruciate ligament (ACL) annually in
the United States with the majority being athletes 15 to 25 years old (Griffin et al., 2006).
Disruption of the ligament can occur in isolation but commonly coincides with associated damage
to other ligaments, cartilage, menisci and bone (Lohmander et al., 2007). Reconstructive surgery
is often recommended for those who experience joint instability and wish to return to higher
levels of physical activity. An estimated 130,000 of these surgeries (Buller et al., 2015) are
performed annually with the goals of restoring knee joint stability and enabling safe return to
pre −injury levels of activity (Myer, G. D. et al., 2004; Stergiou et al., 2007). Surgery is followed by
4 −6 months of rehabilitation to address joint level impairments (pain, swelling, limited range of
motion and strength), restore function and return individuals back to previous activities (Cascio
et al., 2004; Chmielewski et al., 2008; Myer, G.D. et al., 2006; Myklebust & Bahr, 2005). Return
to previous activities time line varies, but can take 6 to 12 months (Adams et al., 2012).
Despite improvements of evidence −based rehabilitation programs, individuals continue to
present with altered lower extremity loading patterns during functional activities 6 to 24 months
following ACL reconstruction (ACLr) (Delahunt et al., 2012; Di Stasi, S. L. et al., 2013b; Noehren
et al., 2013; Oberlander et al., 2014; Ortiz et al., 2011; Webster et al., 2012; White et al., 2013).
Typically, by 6 months, individuals are performing more demanding dynamic functional tasks and
returning to participation in higher levels of activities and sports. The presence of altered loading
strategies at this time is of particular concern as they are related to an increased risk for re −injury
(Paterno et al., 2010). The overall risk for a second knee injury is as high as 40% (Barber-Westin
DYNAMIC KNEE LOADING ASYMMETRIES 5
& Noyes, 2011). A recent prospective study found that 23% of athletes who exhibited
asymmetrical loading strategies at the time they return to their sport incurred a second ACL injury
(Paterno et al., 2010). Given these potential long term consequences, it is important to be able to
clinically identify and ameliorate asymmetrical loading in individuals following ACLr.
Loading Strategies following ACL Reconstruction
Altered sagittal plane knee loading following ACLr is typically characterized by a shift in
loading demands away from the knee extensors of the reconstructed limb. Biomechanically, these
deficits are commonly quantified as decreases in knee flexion angles (position of the knee) and
extensor moments (magnitude of knee joint loading) in the reconstructed knee during dynamic
tasks compared to the non −surgical or healthy control knee. Decreased knee loading is most
pronounced during portions of tasks that require eccentric control or deceleration and are
reported consistently across dynamic tasks such as running, hopping, and landing, throughout
rehabilitation (Decker et al., 2002; Ernst et al., 2000; Gokeler et al., 2010; Oberlander et al., 2013;
Orishimo et al., 2010; Salem et al., 2003). Loading deficits are frequently documented across tasks
and throughout rehabilitation and recovery. The largest differences are present during the landing
or deceleration phase of the task when the knee extensors eccentrically control knee flexion
(Ernst et al., 2000; Oberlander et al., 2013; Salem et al., 2003; Webster et al., 2004). Running is
typically introduced as early as 8 to 12 weeks following surgery as one of the first tasks during
rehabilitation that requires eccentric control of the knee extensors to rapidly accommodate forces
in the lower limb (Adams et al., 2012; Logerstedt et al., 2010; van Grinsven et al., 2010). At 3 to 5
months post −surgery, during running, significant deficits in knee extensor moments (on average
DYNAMIC KNEE LOADING ASYMMETRIES 6
35%) are observed in the reconstructed limb when compared to the non −surgical limb (Lewek et
al., 2002). Similar decreases in knee extensor moments are observed during dynamic tasks, such
as single limb hopping and both double and single limb drop vertical jumps, in individuals 4 to 12
months post −surgery. While these altered loading patterns appear to improve from 6 to 12
months post −operatively, smaller knee extensor moments in the reconstructed limb are still
observed greater than 1 year and even up to 7 years following surgery during running, drop vertical
jump and hopping tasks (Decker et al., 2002; Devita et al., 1992; Oberlander et al., 2013; Ortiz et
al., 2011). The persistence of these strategies across tasks and throughout rehabilitation suggests
that a more directed focus on sagittal plane knee loading during rehabilitation is warranted.
Quantifying Dynamics of Movement
While deficits in knee extensor moments are well documented, less attention has been
placed on the dynamics of knee loading, knee power and angular velocity. The knee plays an
integral role in force attenuation during dynamic tasks contributing nearly half of the lower
extremity’s power absorption when running 2 −5 meters per second (Schache et al., 2015) at
angular velocities of nearly 500 degrees per second (Buczek & Cavanagh, 1990). The few studies
that have quantified power absorption have reported deficits at the knee in individuals 4 to 12
months post −surgery during a single limb hop and nearly 3 years post −surgery during running. A
deficit as large as a 69% decrease in knee negative work (the integral of knee power) in the
reconstructed limb compared to a healthy control limb exists nearly 3 years after surgery (Devita
et al., 1992). In both studies, power deficits coincided with smaller deficits in knee extensor
moments. This suggests that individuals following ACLr may have difficulty meeting the
DYNAMIC KNEE LOADING ASYMMETRIES 7
demands of increased velocity during dynamic tasks (Devita et al., 1992; Orishimo et al., 2010).
Power not only reflects the magnitude (joint moment) but also the rate (joint angular velocity) of
knee loading(Winter, 2009). To date, no study has evaluated knee loading rate during dynamic
tasks in individuals following ACLr; therefore, little is known regarding potential impairments
specific to the speed at which individuals load the knee during dynamic tasks. This may be
particularly important to understand at the time individuals’ progress to running during
rehabilitation.
Clinical Movement Assessments
Currently, clinical movement assessments are generally limited to visual observation of
movement quality or standard video cameras or tablets. While these assessments are commonly
used by clinicians for detection of movement impairments, they are limited to quantification of
kinematics during slower moving tasks. Biomechanical assessments of knee power deficits find
that they are often present along with small changes in joint angles (Orishimo et al., 2010; Salem
et al., 2003). Small differences in knee joint angles are difficult to overserve and quantify clinically.
For example, during a single limb hop a 43% decrease in peak knee power absorption coincided
with only a 10 degree decrease in peak knee flexion angle in the reconstructed limb compared to
the non −surgical limb (Orishimo et al., 2010). Moreover, small differences in angles may also be
difficult to detect during dynamic tasks, using standard video capture (24 −32 frames per second)
because they usually occur in less than 200 milliseconds from initial contact. Therefore, more
sensitive objective metrics of movement quality are needed for clinical identification loading
deficits clinically during dynamic exercises and assessments.
DYNAMIC KNEE LOADING ASYMMETRIES 8
Marker −based Motion Capture versus Inertial Sensor Motion Analysis
Three −dimensional marker −based motion analysis systems that track markers placed over
anatomical landmarks with multiple cameras (250 −340Hz) and measure ground reaction forces
(1360 −1500Hz) are currently the gold standard for objectively assessing human motion. Using
this instrumentation, joint kinematics and kinetics are calculated using kinematic models and
inverse dynamics equations allowing for an estimation of individual joint loading patterns.
However, these analyses are complex, expensive and time consuming; thus, impractical in clinical
settings. Inertial sensor based wearable sensors (accelerometers and gyroscopes) can also provide
information regarding three −dimensional motion. Recent advances in wireless technology and
improvements in data storage and capture frequency capabilities make it possible to study human
movement in more detail (Dejnabadi et al., 2006; Dowling et al., 2012a; Favre et al., 2008; Fong &
Chan, 2010). The use of wearable sensors is widely established in measurement of temporal and
spatial characteristics during gait in individuals with significant movement deviations such as
those with Parkinson’s diseases (Bachlin et al., 2010; Bregou Bourgeois et al., 2014; Yang et al.,
2013). However, recent studies suggest that wearable sensors are capable of providing
meaningful information regarding segment angles and angular velocities in individuals who have
smaller movement deviations such as those who have undergone ACLr (Barrios et al., 2010;
Dowling et al., 2011; Dowling et al., 2010; Patterson et al., 2014; Tate & Milner, 2010).
The ability to capture data during movements at higher frequencies (up to 128Hz)
improves the sensitivity of inertial sensors for quantification of movement during dynamic tasks.
Despite the fact that inertial sensors are only able to directly measure kinematic variables, such as
accelerations and angular velocities of the body segments they are affixed to, they may still be able
DYNAMIC KNEE LOADING ASYMMETRIES 9
provide valuable information regarding movement impairments. Previous studies indicate that
thigh and shank angular velocities and accelerations during running or double limb landing tasks
can provide information regarding risk factors for knee injury in healthy individuals (Crowell &
Davis, 2011; Dowling et al., 2012a, 2012b; Yang et al., 2011). For example, Dowling and colleagues
found that coronal plane angular velocity of the thigh segment is correlated with knee adductor
moment during a double limb drop jump task in healthy individuals (Dowling et al., 2012a).
While the correlation was relatively weak (r 0.46), these data suggest that segment kinematics
can provide some information regarding joint kinetics during dynamic tasks. This relationship is
likely due to the fact that segment kinematics are used along with ground reaction forces to
calculate joint moments. It is possible that a stronger relationship exists between segment
angular velocity and knee joint power given that power is the product of knee joint moment and
angular velocity. Moreover, segment kinematics may be more sensitive to sagittal plane knee
power deficits as more motion is afforded at the knee in the sagittal compared to the frontal plane
at the knee during loading tasks. However, further investigation is warranted to determine how
segment kinematics relate to knee joint power in individuals following ACLr.
Clinical Detection of Lower Extremity Loading Impairments using Inertial Sensors
The presence of a relationship between variables calculated with inertial sensors and knee
power determined from marker −based motion analysis procedures does not guarantee their
clinical usefulness in identification of loading deficits following ACLr. The feasibility of
translating this information to the clinic requires the development of clinical testing procedures
and determination of diagnostic accuracy of these procedures. The process of determining
diagnostic accuracy requires a definition of the presence or absence of a deficit. Clinicians
DYNAMIC KNEE LOADING ASYMMETRIES 10
typically make comparisons between the reconstructed and non −surgical limbs by calculating a
between limb ratio. This ratio represents a level of symmetry to determine progress during
rehabilitation. A between limb ratio of 85 −90% is often used as a benchmark for progression to
more challenging activities (Adams et al., 2012; Kvist, 2004; Thomee et al., 2011). Using this
definition for the presence of a between limb deficit in knee power, diagnostic accuracy must be
determined by calculating test sensitivity, specificity, likelihood ratios and post −test probability
of inertial sensors’ measurements in identifying asymmetries in knee power. These measures are
very sensitive to the characteristics of the test population; more specifically for individuals after
ACLr, factors such as time since surgery, current functional level, physical activity experience
prior to surgery, and age can factor into the strength of these measures and the validity of using
them clinically. Running, initiated 2 to 4 months post −surgery, imposes greater demands on the
knee than previously experienced during rehabilitation (Adams et al., 2012; Logerstedt et al., 2010;
van Grinsven et al., 2010). Therefore, a focus on this population may be warranted for the
identification of impaired knee loading.
Summary
Following ACL reconstruction, individuals present with altered sagittal plane knee
loading patterns with significant deficits on the reconstructed limb that persist across tasks and
throughout recovery. These deficits, characterized by decreased knee extensor moments, describe
a strategy that limits the magnitude of knee loading. Less attention has been placed on
understanding how individuals modulate the rate of knee loading. Understanding potential
impairments, specific to the speed at which individuals load the knee during dynamic tasks, is
critical when individuals aim to return to more dynamic activities. This is particularly important
DYNAMIC KNEE LOADING ASYMMETRIES 11
to understand potential impairments at the time individuals’ progress to running during
rehabilitation. A better understanding of how individuals load their knee as they initiate more
dynamic tasks will inform rehabilitation strategies. However, identification of these deficits
remains a challenge for clinicians and may underlie their persistence. Improved wearable sensors
technology may provide a mechanism for quantification of power deficits in the clinic. Inertial
sensors can measure angular velocities, particularly, segment angular velocities. Given that knee
power is a combination of joint angular velocity and moment, segment angular velocity
measurements may be able to provide information regarding knee power clinically. A
determination of the strength of these relationships can facilitate the development of clinical
testing procedures. However, in order to translate this relationship to clinical practice, diagnostic
accuracy of these procedures must be established in this population.
DYNAMIC KNEE LOADING ASYMMETRIES 12
CHAPTER III
DYNAMIC KNEE LOADING DEFICITS IN SINGLE LIMB LOADING AND RUNNING IN
INDIVIDUALS FOLLOWING ACL RECONSTRUCTION
Abstract
Well documented deficits in sagittal plane knee loading, during dynamic tasks indicate that
individuals limit the magnitude of knee loading following anterior cruciate ligament
reconstruction (ACLr). However, it is not known how they modulate the rate at which they load
their knee which is important for performance of dynamic tasks. The purpose of this chapter was
to compare dynamic loading strategies at the knee between limbs and to controls during two
dynamic tasks that require different magnitudes of loading, in individuals following ACLr, at the
time that they progress to run. Two groups of recreationally active individuals participated;
fifteen healthy controls and fifteen individuals post −ACLr (ACLR). Participants performed three
trials of over ground running and of a single limb loading task. Separate 2x2 repeated measures
analysis of variance were used to compare the effects of limb and task in the ACLR group;
independent t −tests were used to compare between ACLR and Control groups, and paired t −tests
to compare between limbs in ACLR group and between tasks in Control group. Control data
indicated that the magnitude of loading was more than two times greater in running than single
limb loading. Decreased magnitude and rate of loading were observed in the reconstructed
compared to the non −surgical limb during both tasks with larger between limb differences during
the more demanding task, run. The presence of loading rate deficits during the single limb loading
task suggests that despite the lower magnitude of loading, individuals limit their rate of loading.
This may indicate underlying impairments in the ability to rapidly load the knee at a time when
individuals are progressed to running during rehabilitation.
DYNAMIC KNEE LOADING ASYMMETRIES 13
Introduction
Despite improvements in evidence −based rehabilitation programs, deficits in sagittal
plane knee loading can persist 6 to 24 months after surgery in individuals following anterior
cruciate ligament reconstruction (ACLr) across a variety of tasks (Noehren et al., 2013;
Oberlander et al., 2014; Roewer et al., 2011; Webster et al., 2004). Specifically, individuals adopt
strategies that shift the demands away from the knee extensors of the reconstructed limb (Barber-
Westin & Noyes, 2011; Di Stasi, S. et al., 2013a; Paterno et al., 2010; White et al., 2013). These
strategies are well documented across tasks and through recovery. They are characterized by
decreases in knee flexion and extensor moments in the reconstructed limb when compared to a
healthy control or non −surgical limb (Devita et al., 1992; Lewek et al., 2002; Orishimo et al., 2010).
Knee extensor moment deficits as large as 35% are observed 6 to 15 months post −surgery during
the deceleration portion of dynamic tasks including, running, landing and hopping, as the knee
extensors act to eccentrically control knee flexion (Ernst et al., 2000; Oberlander et al., 2013; Salem
et al., 2003; Webster et al., 2004). This is of concern as the presence of asymmetrical loading is
thought to contribute to risk for re −injury (Chmielewski, 2011; Dai et al., 2012; Paterno et al.,
2010).
While many studies describe a strategy that limits the magnitude of sagittal plane knee
loading (knee extensor moments) during dynamic tasks following ACLr, less attention has been
focused on potential deficits in loading rate. Studies reporting knee joint power, which considers
the magnitude and rate of loading, describe decreases in knee extensor moments along with larger
deficits in knee power absorption. For example, at 7 months post −surgery, individuals exhibited
a 6% decrease in knee extensor moment along with a 43% decrease in knee power absorption in
the reconstructed limb compared to the non −surgical limb during a single leg hop test (Orishimo
DYNAMIC KNEE LOADING ASYMMETRIES 14
et al., 2010). Given that power considers not only knee moments but also knee angular velocity,
the larger deficits in power suggest that individuals following surgery limit the rate at which they
load their knee. Deficits in dynamic knee loading (knee joint power and potentially knee angular
velocity) in this population are of particular concern as individuals following ACLr are returning
to sports and dynamic activities such as running, cutting, and single or double limb landings that
involve attenuation of forces nearly 3 −30 times body weight within 10 −40 milliseconds of initial
contact (Decker et al., 2003; Weinhandl, 2010; Zhang et al., 2000). As such, an understanding of
how individuals modulate the rate of loading and their ability to rapidly accommodate forces
during dynamic tasks following surgery is needed in this population.
Running is typically one of the first tasks introduced in rehabilitation (as early as 8 to 12
weeks following surgery) that requires eccentric control of the knee extensors to rapidly
accommodate of forces during single limb support(Adams et al., 2012; Logerstedt et al., 2010; van
Grinsven et al., 2010). During running, ground reaction forces can be as much as 2 −2.9 times body
weight(Nilsson & Thorstensson, 1989) whereas during exercises considered precursors to
running, step up and step down, ground reaction forces are only 1 −1.3 times body weight
(Baumgart et al., 2015; Harty et al., 2011). Given these demands, it is important to understand if
deficits in knee loading rate exist at the time individuals following ACLr initiate running
programs. Power absorption deficits during running are observed in individuals nearly three years
following surgery and as large as a 69% decrease in knee negative work in the reconstructed limb
compared to a healthy control limb (Devita et al., 1992). Despite an emphasis on early progression
to running to introduce more dynamic loading in ACL rehabilitation, deficits in the ability to
rapidly accommodate forces persist (Adams et al., 2012; Logerstedt et al., 2010).
DYNAMIC KNEE LOADING ASYMMETRIES 15
The extent to which dynamic knee loading is limited during early rehabilitation, as
individuals progress to more dynamic tasks, has not been characterized. It is possible that, in
response to the increased demands, during early running individuals adopt a more protective
strategy that limits not only the magnitude of loading, seen as decreased knee extensor moments
(Devita et al., 1992; Kuenze et al., 2014; Lewek et al., 2002), but also the rate of knee loading. Given
that both the magnitude and rate of loading are increased during running, it may be difficult to
determine if impairments specifically related to the ability to rapidly accommodate forces, loading
rate, exist. The rate at which an individual can load their surgical knee may be constrained by
magnitude of loading demands. Therefore, analyses of power and rate of loading during running
and dynamic tasks with smaller magnitude of loading demands is needed to gain insight into
individuals ability to load the knee at faster velocities.
The purpose of this study was to investigate ability to rapidly accommodate forces at the
knee during dynamic tasks, with an emphasis on power absorption and angular velocity, in
individuals following anterior cruciate ligament reconstruction. Our analysis focused on the
phase of rehabilitation when individuals progress to running protocols during treatment. To do
this, comparisons were made between limbs and to healthy control limbs, and between two
dynamic tasks; running and a dynamic single limb loading task designed to require a lower
magnitude of knee loading. It was hypothesized that knee extensor moments, power absorption,
and sagittal plane knee angular velocity would be greater during running than the single limb
loading task, and no differences would be observed between the control limbs and the
non −surgical limbs for any variable. In addition, it was hypothesized that individuals following
ACLr would exhibit decreased knee extensor moments, power absorption, and sagittal plane knee
angular velocity in the reconstructed limb when compared to non −surgical limbs in both tasks.
DYNAMIC KNEE LOADING ASYMMETRIES 16
More specifically, smaller dynamic knee loading (knee power absorption and knee angular
velocity) deficits would be observed in the reconstructed knee during the less demanding single
limb loading task when compared to running. Furthermore, it was hypothesized that between
limb sagittal plane knee loading symmetries during the more demanding task, running, would be
related to symmetries during the less demanding task single limb loading, in individuals following
ACLr. In particular, individuals who had larger between limb asymmetries in knee power
absorption during running would have larger asymmetries during single limb loading despite
differences in movement demands.
Methods
Participants
An a priori sample size analysis was performed based on pilot data of six subjects to
determine the appropriate sample size to provide more than 80% power at an alpha level of 0.05.
Two groups of 15 individuals (n=30), between the ages of 16 and 39, who were recreationally active
participated; 15 individuals who had previously undergone ACL reconstruction, and 15 healthy
controls (Table 3.1). The ACLR group consisted of individuals (7 females) who had primary
unilateral ACL reconstruction (10 right and 5 left) using either a bone −patellar −tendon −bone
(BTB) autograft or allograft, approximately 4.6 ± 1.4 months prior to testing. At the time of
participation subjects in the ACLR group were attending physical therapy and had initiated a
running progression within 1.5 months of testing. The Control group consisted of individuals (9
females) with no history of ACL injury or knee surgery, and were matched to the ACLR group
based on age, and activity level (evaluated using Cincinnati Sports Activity questionnaire)
DYNAMIC KNEE LOADING ASYMMETRIES 17
(Barber-Westin et al., 1999). Age (p=0.946), body mass (p=0.216) body height (p=0.070), and
activity level (p=0.858) did not differ between groups.
Subjects in the ACLR group were excluded from the study if they: (1) were not cleared by
physical therapist to perform the study tasks, or (2) had prior ACL injury and knee surgery on the
contralateral limb. One subject in the ACLR group had prior ACL surgery on the ipsilateral limb
approximately 18 months prior to the current injury and surgery, and two subjects in the ACLR
group had ipsilateral partial tears of their ACL with no surgery two and eight years prior to the
current injury and surgery. All three individuals were included in the study as their previous
injury was on the ipsilateral limb as the current injury and surgery. Subjects in either group were
excluded from participation if they reported any of the following: (1) concurrent pathology or
morphology that could cause pain or discomfort during physical activity, and (2) any physical,
Table 3.1. Subject characteristics of the ACLR and Control Groups
(mean ± standard deviation)
ACLR (n = 15) Control (n = 15)
Age (year) 23.4 ± 7.3 23.6 ± 2.3
Body mass (kg) 71.3 ± 12.4 66.6 ± 7.7
Body height (cm) 174.2 ± 9.2 168.1 ± 9.9
Activity Level^ 91 ± 8.9 92 ± 11.1
Time after surgery (months) 4.6 ± 1.4 NA
Graft Type
8 BTB autograft
7 allograft
NA
^ACLR group: activity level prior to surgery; Control group: current activity
level
DYNAMIC KNEE LOADING ASYMMETRIES 18
cognitive or other condition that may impair subjects’ ability to perform the tasks proposed in
this study.
Procedures
Testing took place at University of Southern California’s Human Performance Laboratory
located at the Completive Athletic Training Zone, Pasadena CA. All procedures were explained
to each subject and informed consent was obtained as approved by the Investigational Review
Board at the University of Southern California Health Sciences Campus. Parental consent was
obtained for all subjects under the age of 18.
After consenting to participate, the subject’s age, height, weight, tibia length, dominant
limb (defined as leg with which subject would kick a ball), knee medical history, and physical
activity were recorded. Physical activity prior to knee injury for the ACLR group, and current
activity level for the Control group were determined using the Cincinnati Sports Activity Scale to
match activity and athletic level between groups (Barber-Westin et al., 1999). Recreational
athlete was defined as Level I or II on the Cincinnati Sports Activity scale.
Prior to testing, participants were asked to warm −up on a stationary bike for five minutes.
Next, reflective markers were placed on first and fifth metatarsals, end of distal second toes,
medial and lateral malleoli, medial and lateral epicondyles of femurs, greater trochanters, posterior
superior iliac spines, iliac crests, and L5 −S1 junction. In addition, reflective markers secured on
rigid plates, tracking marker clusters, were secured bilaterally on participant’s thigh, lower leg
and heel of their shoe by the examiner. All markers were removed after a static calibration trial
was collected except the rigid plates, pelvis markers, and distal toe markers which remained on
during testing. Kinematic data and ground reaction force data were collected using a
DYNAMIC KNEE LOADING ASYMMETRIES 19
marker −based, 11 camera motion capturing system (Qualysis Inc., Gothenberg, Sweden) at a
250Hz and force platforms at 1500Hz (Advanced Mechanical Technologies, Inc., Newton, MA,
USA) or a 14 camera motion capturing system (BTS Bioengineering Corp., Milan, Italy) at 340Hz
and force platforms at 1360Hz (BTS Bioengineering Corp., Milan, Italy).
Dynamic Tasks: Single Limb Loading Test and Running
During testing, each participant performed two tasks: dynamic single limb loading (SLL)
running (RUN). For SLL, subjects were instructed to stand on both feet on a single force platform
in front of a target (Figure 3.1). Tape was placed on the ground on an adjacent force platform as a
target for foot placement. The distance from the target was normalized to each individual as a
distance equal to the length of their tibia. Subjects were asked to leap forward to the target
location onto a single limb and to lower as far they can and then return to the start position. For
each trial they were asked to perform three consecutive repetitions. Participants performed single
limb loading trials alternating between limbs beginning with the non −surgical limb in the ACLR
group and dominant limb in the Control group. A trial was considered acceptable when it
contained presence of a distinct flight phase, maintenance of balance throughout the task and
complete foot placement on the target force platform. The presence of a flight phase was
considered as criteria for a successful trial to avoid double limb support. It was determined by the
absence of forces on either force platform prior to foot contact on the target force platform.
Practice trials were allowed for subjects to become familiar with the task. After completing
dynamic single limb loading subjects were instructed to run at their self −selected speed over 15
meter distance (RUN). Speed was calculated using laser timing gates (Brower IRD −T175; Brower
Timing Systems, Draper UT, USA) placed five meters apart in the capture area. RUN trials were
DYNAMIC KNEE LOADING ASYMMETRIES 20
accepted if the subject’s speed was within ±5% interval and complete foot placement on the force
platforms. Participants performed three trials on each limb of each task.
Figure 3.1: Single Limb Loading Test*
(*Subject figure reproduced with permission)
Data Analysis
Three −dimensional marker coordinates (Qualysis Inc. Tracking Manager, BTS
Bioengineering Corp SMARTtracker) were reconstructed and in combination with force and
anthropometric data were used to calculate joint kinematics, kinetics and energetics (Visual
3D
TM
, Version 4.8, C −Motion, Inc., Rockville, MD, USA). Coordinate data was low −pass filtered
using a fourth order zero −lag Butterworth filter with a 12Hz cut −off frequency. The local
coordinate systems of body segments were derived from the standing calibration trial. Lower
extremity segments were modeled as a frusta of cones, while the pelvis was modeled as a cylinder.
Six degrees of freedom of each segment were calculated by transforming the triad of markers to
DYNAMIC KNEE LOADING ASYMMETRIES 21
the position and orientation of each segment during the standing calibration trial. Euler angles
with the following order of rotations were used to calculate joint kinematics: flexion/extension,
abduction/adduction, and internal/external rotation. Joint angles were expressed as movement
of the distal segment relative to the proximal segment. Kinematics, anthropometrics, and ground
reaction forces were used in standard inverse dynamics equations to calculate internal net joint
moments. Net joint power was calculated as the product of joint moment and joint angular
velocity. All kinetic and energetic data were normalized to body mass. To represent average
healthy control movement, an average between control limbs was calculated for all variables. Data
obtained from Visual3D
TM
(Version 4.8, C −Motion Inc., Rockville, MD, USA) were exported and
analyzed using a customized MATLAB® program (Version R2014b, The MathWorks, Matick,
MA, USA).
All dependent variables were analyzed during the deceleration phase. Consistent with
previous studies, deceleration was defined as the time from initial contact to maximal knee flexion
(Sigward et al., 2012). Peak knee power absorption, peak knee angular velocity, and peak knee
extensor moment were identified during both tasks. To characterize between limb differences in
peak knee power absorption and peak knee angular velocity, a between limb symmetry ratio was
calculated for each task within the ACLR group by dividing the reconstructed limb values by the
non −surgical limb values. The average of three trials of each limb (ACL reconstructed,
non −surgical, and control) was used for analysis in both tasks.
Statistical Analysis
A mixed −factor multivariate analysis of variance (MANOVA) was performed to compare
differences in the variables of interest between the reconstructed (ACLr), non −surgical (Non −Sx),
DYNAMIC KNEE LOADING ASYMMETRIES 22
and control (CTRL) limbs during single limb loading and running. Three dependent variables
were analyzed: peak knee power absorption, peak knee angular velocity, and peak knee extensor
moment. The independent variables were task (SLL, RUN) and limb (ACLr, Non −Sx, CTRL).
Based on the significant MANOVA (p 0.001) separate repeated measures ANOVAs (task x limb)
were performed for post hoc comparison between limbs and tasks in the ACLR group and
independent t −tests were performed to compare between the non −surgical limb (ACLR group)
and control limb. Paired t −tests were used to compare between tasks in the Control group.
One −tailed Pearson product −moment correlations were performed to determine if between limb
ratios of knee power and angular velocity during running and single limb loading were related. A
strong correlation was defined as a correlation greater than 0.75 and a moderate correlation was
defined as a correlation between 0.50 and 0.75. Statistical analyses were performed using PASW
software (version 18, SPSS, Inc., Chicago, IL) with a significance level of 0.05.
Results
A significant task (SLL, RUN) by limb (ACLr, Non −Sx) interaction was observed for peak
knee extensor moment (p 0.01) (Figure 3.2A). Subjects exhibited significantly greater between
limb differences (ACLr and Non −Sx) in knee extensor moments during RUN (p 0.001) than SLL
(p 0.001). Task x limb interactions did not reach significance for peak knee power absorption (p
0.065) or knee angular velocity (p 0.084). Main effects for task and limb were observed for all
dependent variables. When collapsed across limb greater peak knee extensor moment (p 0.001;
Figure 3.2A), power absorption (p 0.001; Figure 3.2B) and angular velocity (p 0.002; Figure
3.2C) were observed during RUN compared to SLL. When collapsed across task, the Non −Sx
DYNAMIC KNEE LOADING ASYMMETRIES 23
limb exhibited greater peak knee extensor moment (p 0.001; Figure 3.2A), power absorption
(p 0.001; Figure 3.2B) and angular velocity (p 0.001; Figure 3.2C) compared to ACLr limb.
Figure 3.2: Comparisons of (A) peak knee extensor moment (B) peak knee
power absorption, (C) peak knee angular velocity between reconstructed
(ACLr, solid line) and non −surgical (Non −Sx, dashed line) limb during single
limb loading(SLL) and running (RUN); Data represents mean ± standard
deviation; #main effect for limb; +main effect for task; **p 0.001, *p 0.05
DYNAMIC KNEE LOADING ASYMMETRIES 24
For the control limb, significantly greater peak knee extensor moment (p 0.001; Figure
3.3A), and power absorption (p 0.001; Figure 3.3B) were observed during RUN compared to SLL.
There was no significant difference in peak knee angular velocity (p 0.062; Figure 3.3C) between
tasks. No significant differences were observed between control and non −surgical limbs for both
tasks and all variables.
Figure 3.3: Comparisons of (A) peak knee extensor moment (B) peak knee power
absorption, (C) peak knee angular velocity between single limb loading(SLL) and
running (RUN) in the control limb; Data represents mean ± standard deviation;
**p 0.001
DYNAMIC KNEE LOADING ASYMMETRIES 25
The between limb symmetry ratios for knee power absorption (r 0.577; p 0.012; Figure
3.4A) and knee angular velocity (r 0.647; p 0.004; Figure 3.4B) during single limb loading were
moderately correlated with power absorption and angular velocity ratios during running,
respectively.
Figure 3.4: The relationship of between limb symmetry ratios during single limb
loading (SLL) and running (RUN) for (A) peak knee power absorption and (B)
peak knee angular velocity; Pearson product −moment correlation reported;
*p 0.05
Discussion
Consistent with previous studies, significant knee loading deficits were observed in the
reconstructed limb during running (Devita et al., 1992; Kuenze et al., 2014; Lewek et al., 2002).
Knee moment deficits in the current study are similar to those described in a previous study
assessing individuals 3 to 5 months post −surgery (Lewek et al., 2002) and those approximately 3
years post −surgery (Devita et al., 1992; Kuenze et al., 2014). In addition to smaller knee extensor
moments, individuals following ACLr demonstrated significantly smaller peak power absorption
in the reconstructed knee than the non −surgical knee during running. Similarly, peak knee
DYNAMIC KNEE LOADING ASYMMETRIES 26
angular velocities were also smaller in the reconstructed limb during running. Together, between
limb differences in these variables indicate that individuals not only modulate the magnitude of
knee loading, but also the rate at which they accommodate loads during running.
Running is one of the first dynamic tasks introduced during rehabilitation and is a more
demanding task, especially when compared to other activities or exercises performed at this time
during rehabilitation. Running involves nearly 3 −4 times larger knee extensor moments than
walking or single limb squatting and subsequently places greater demand on the knee extensors
to eccentrically control deceleration (Noehren et al., 2013; Willson & Davis, 2008). Between task
comparisons in this study suggest that the single limb loading task assessed in this study was less
demanding than running. When considering the Control group data; the magnitude of loading,
characterized by peak knee extensor moments, was 2.2 times greater in RUN than SLL. While
power absorption was also 2.3 times greater in RUN, the absence of differences in angular velocity
suggests that the rate of loading demands do not differ significantly between tasks.
Understanding how individuals following ACLr modulate the rate of loading during a task in
which the magnitude of loading is substantially lower provides further insight into the nature of
dynamic knee loading impairments at this time.
As hypothesized, larger between limb differences were observed in knee extensor
moments, power absorption, and angular velocity during RUN than SLL. It is important to note
that the interaction between limb and task only reached significance for knee extensor moments
and trended toward significance in power absorption and angular velocity. When considering
the between limb difference in peak knee extensor moment, the difference observed during RUN
(1.2 Nm/kg) was 2 times greater than that observed during SLL (0.6 Nm/kg). Similarly, when
DYNAMIC KNEE LOADING ASYMMETRIES 27
considering between limb differences in knee power absorption, RUN (7 W/kg) had a 1.7 times
greater difference than SLL (4.1 W/kg). Between limb difference in knee angular velocity during
SLL (106 deg/s) was 1.4 times greater than the between limb difference during RUN (73 deg/s).
Despite the substantially smaller magnitude of loading during SLL, individuals continued to
exhibit deficits in the rate of knee loading. These findings suggest that individuals following ACLr
may have underlying impairments related to the rate of knee loading at this point in rehabilitation
regardless of magnitude of loading demands. Furthermore, these impairments may translate
across tasks as the power absorption and angular velocity ratios during RUN and SLL are related.
Those who had larger deficits in power absorption and knee angular velocity during RUN also
had larger asymmetries during SLL.
The presence of sizeable deficits in dynamic knee loading across tasks, specifically knee
power and angular velocity, is of concern as dynamic activities and sports require substantial knee
power absorption and angular velocities. The knee plays an integral role in force attenuation
during dynamic tasks. Knee flexion contributes to nearly 40 and 55% of the lower extremity’s
power absorption when running 2 meters per second and 5 meters per second, respectively
(Schache et al., 2015). During a drop land, the knee contribution to lower extremity power
absorption is nearly 56% (Moolyk et al., 2013). Furthermore, dynamic tasks require force
attenuation across the knee at high angular velocities; nearly 455 degrees per second and 600
degrees per second during running and landing, respectively (Buczek & Cavanagh, 1990; Decker
et al., 2003). Therefore, restoration of proper dynamic knee loading may be imperative for those
looking to return to high level activities. Deficits in knee power absorption and angular velocity
observed in the current study highlight the importance of considering the rate of loading during
early rehabilitation.
DYNAMIC KNEE LOADING ASYMMETRIES 28
From these data, we are unable to determine what factors underlie impairments in loading
rate. Muscle function was not assessed in these individuals; however, it is possible that impaired
quadriceps strength or more importantly, rate of force development in the reconstructed limb may
limit their ability to meet the demands of increased velocity during dynamic tasks. Limitations in
loading rate may also reflect a reluctance or fear of loading as the individuals in this study were
only introduced to running within the last month. Study participants may not have developed a
level of confidence and comfort in dynamic knee loading tasks regardless of the magnitude. While
further research is needed to determine factors that underlie dynamic knee loading deficits, these
data suggest that incorporating exercises for which the magnitude of loading is lower relative to
running may allow for more focused attention on the speed of movement. This type of practice
may not only increase individuals’ confidence in loading their knee, but also lead improvements
in the rate of force attenuation in their knee extensors. Further research is needed to understand
the effects of interventions targeting speed of movement on the restoration of knee power and rate
of loading across dynamic tasks.
Interestingly, no differences were observed between the non −surgical limb in individuals
following ACLr and the healthy control limb in the Control group in knee power absorption,
angular velocities, and extensor moments. This indicates that the non −surgical limb acts like a
healthy control limb during the tasks evaluated. In this population, performance on the
reconstructed limb is commonly compared to the non −surgical limb (Gokeler et al., 2010; Lewek
et al., 2002; Orishimo et al., 2010; Webster et al., 2004; Xergia et al., 2013). However, recently
some researchers argued that ACLr surgery leads to the development of altered movement
patterns that effect the non −surgical limb and therefore it may not represent a control reference
(Chmielewski, 2011; Di Stasi, S. et al., 2013a; Wright et al., 2011). While interlimb compensations
DYNAMIC KNEE LOADING ASYMMETRIES 29
are well documented in double limb landing tasks and walking, it is not as evident in single limb
tasks (Decker et al., 2002; Devita et al., 1998; Devita et al., 1992). Findings from this study suggest
that during these single limb tasks the non −surgical limb may be used as a representative healthy
limb for comparison.
DYNAMIC KNEE LOADING ASYMMETRIES 30
CHAPTER IV
ANGULAR VELOCITIES MEASURED WITH INERTIAL SENSORS REFLECT
DYNAMIC KNEE LOADING DURING SINGLE LIMB LOADING IN INDIVIDUALS
FOLLOWING ACL RECONSTRUCTION
Abstract
Difficulty detecting deficits in knee power during dynamic tasks in individuals following anterior
cruciate ligament reconstruction (ACLr) may underlie the persistence of these deficits over time.
The expense, time, and expertise needed to quantify knee power deficits using the current gold
standard techniques preclude their use in the clinic. As knee joint angular velocity is used to
calculate power, the potential for the use of inertial sensors, gyroscopes, to identify power deficits
in the clinic exists. The purpose of this chapter was to determine if angular velocities measured
with inertial sensors can provide meaningful information regarding knee power absorption
during a dynamic single limb loading task in twenty −one individuals following ACLr. To
determine the best predictors of knee power absorption, separate stepwise linear regressions
were performed using thigh, shank and knee angular velocities calculated from marker −based
motion analysis and inertial sensors, respectively. Intraclass correlation coefficients (2,k) were
used to determine concurrent validity. Knee angular velocity and thigh angular velocity were
identified as the best predictors of knee power when considering angular velocities extracted
from the marker −based system and from inertial sensors, respectively. Both were strong
predictors explaining nearly seventy percent of the variance in knee power. High intraclass
correlation coefficients indicated strong agreement between measurement systems. Together,
these data suggest information from inertial sensors positioned on the thigh provide meaningful
information regarding knee power which may be helpful in identifying deficits clinically.
DYNAMIC KNEE LOADING ASYMMETRIES 31
Introduction
Individuals following ACL reconstruction (ACLr) continue to present with altered
sagittal plane knee loading patterns 6 to 24 months post −surgery as they perform more demanding
functional tasks and return to participation in higher levels of physical activities and sports
(Noehren et al., 2013; Oberlander et al., 2014; Roewer et al., 2011; Webster et al., 2004). The
presence of altered loading strategies at this time point is of particular concern as they are related
to an increased risk for re −injury.
A recent prospective study found that 23% of athletes who
exhibited asymmetrical knee loading at the time they return to their sport incurred a second ACL
injury (Paterno et al., 2010). While objective measures of function are used to determine and
individuals readiness to return to play(Abrams et al., 2014; Logerstedt et al., 2010), they do not
appear to be sensitive to altered knee joint loading (Orishimo et al., 2010). Given the potential
long −term consequences of persistent deficits, identification and improvement of altered loading
patterns prior to returning to previous activities is critical.
Altered sagittal plane knee loading strategies following ACLr are characterized
biomechanically as decreases knee power absorption, angular velocities, and extensor moments
in the reconstructed knee when compared to non −surgical knee and healthy controls. As seen in
Chapter III and previous studies, they are typically observed during portions of dynamic tasks
that require eccentric control or deceleration(e.g. running, hopping, landing, single limb loading)
(Ernst et al., 2000; Oberlander et al., 2013; Salem et al., 2003; Webster et al., 2004). Knee power
represents the combination of magnitude (knee moment) and rate(knee angular velocity) of knee
loading (Winter, 2009); and the presence of deficits in power across tasks is of particular concern
as dynamic activities and sports require rapid accommodation of forces. The knee plays an
integral role in force attenuation and contributes to nearly 40 −55% of lower extremity’s power
DYNAMIC KNEE LOADING ASYMMETRIES 32
absorption at speeds up to 450 −600 degrees per second of knee flexion as seen in Chapter III and
previous studies (Buczek & Cavanagh, 1990; Dai et al., 2012; Decker et al., 2003; Moolyk et al.,
2013; Schache et al., 2015).
Currently, identification of deficits in knee power absorption during dynamic tasks
requires the gold standard marker −based three −dimensional motion analysis technology. During
marker −based motion analysis, marker positions are recorded at 250 −340 Hz, in conjunction with
ground reaction force data, collected at 1360 −1500 Hz from force platforms; and together are used
to calculate knee moments, knee power absorption and angular velocities. However, these
analyses are complex, expensive, and time consuming, thus impractical in clinical settings.
Two −dimensional video assessment using traditional video cameras or tablets recording at 24 −32
frames per second are becoming popular among clinicians for detection of movement
impairments, but are limited to quantification of kinematics during tasks performed at slower
speeds. Unfortunately, large knee power absorption deficits often coincide with small differences
in joint angle making them difficult to observe clinically (Orishimo et al., 2010; Salem et al., 2003).
For example, during landing a 43% decrease in peak power absorption corresponded with only a
10 degree decrease in peak knee flexion angle in the reconstructed limb compared to the
non −surgical limb (Orishimo et al., 2010). This difference in angle may be particularly difficult to
detect during more dynamic tasks as individuals go through nearly 30 −50 degrees of flexion in less
than 200 milliseconds. The persistence of altered loading following ACL reconstruction may be
related to difficulty in identifying loading deficits clinically. More sensitive metrics are needed
for clinical identification loading deficits during dynamic exercises and assessments.
Recent advances in wireless capabilities and data storage in wearable technology make
inertial sensors more affordable and practical for movement assessments outside a motion analysis
DYNAMIC KNEE LOADING ASYMMETRIES 33
laboratory (Delahunt et al., 2012; Dowling et al., 2012b; Favre et al., 2008; Fong & Chan, 2010).
Inertial sensors comprised of an accelerometer, gyroscope and magnetometer, are typically affixed
to body segments and used to quantify segment angular velocities or accelerations. Findings from
previous studies indicate that thigh and shank angular velocities and accelerations can provide
information regarding movement, during double limb landing or running tasks, and can be used
to identify knee injury risk factors in healthy individuals (Dowling et al., 2012a, 2012b; Greene et
al., 2010; Patterson et al., 2014; Yang et al., 2011). However, it is not known if segment kinematic
data can provide valuable information regarding sagittal plane knee loading without ground
reaction force data. Moreover, it is unknown if these measurements are valid outside healthy
populations in individuals with known movement deficits such as those following ACLr or
particularly in other dynamic tasks in individuals following ACLr.
Despite that kinematics are not the only variables considered in the calculation of power,
it is possible that kinematic variables, specifically joint and segment angular velocities, can
adequately reflect knee power absorption during dynamics tasks. Knee, as well as, thigh and
shank angular velocities not only quantify how fast a segment rotates, but also provide a
preliminary understanding of neuromuscular control during eccentric control of dynamic tasks.
Individuals may employ subtle changes in joint kinematics and concurrently, segment kinematics
to alter forces and decrease knee loading. Small differences in knee angular velocity are found to
be related to differences in peak ground reaction forces during a stop −jump task (Yu et al., 2006),
and subtle differences in thigh and shank angular velocities in the coronal plane are related to
alterations in knee adductor moments during both single and double limb landing (Dowling et
al., 2012a). As segment angular velocities are factored into joint angular velocities and subsequent
joint power calculations, it stands to reason that these variables are related. However, it is
DYNAMIC KNEE LOADING ASYMMETRIES 34
unknown how well they are related and if they can accurately reflect known deficits in knee
power in individuals following ACLr. Therefore, the aim of this study was to determine if segment
angular velocities measured with inertial sensors could provide meaningful information regarding
knee power absorption during a dynamic single limb loading task. To do this, the extent of the
relationship between joint and segment angular velocities measured with marker −based motion
analysis systems and knee power was established. For clinical translation, concurrent validity of
using inertial sensors to measure segment kinematics was assessed and the relationship between
knee power and angular velocities measured with inertial sensors was confirmed. It was
hypothesized that angular velocities would be good predictors of knee power and that strong
agreement between measurement systems would be found for both joint and segment angular
velocities.
Methods
Participants
Twenty −one individuals (12 females) who had primary unilateral ACL reconstruction (11
right and 10 left) using a bone −patellar −tendon −bone (BTB) autograft, allograft or hamstring
autograft approximately 5.1 ± 1.5 months prior to testing participated (Table 4.1). All subjects
reported that they were recreationally active prior to their injury (evaluated using Cincinnati
Sports Activity questionnaire) (Barber-Westin et al., 1999). At the time of participation subjects
were actively attending physical therapy and had initiated a running progression within 2 months
of testing.
DYNAMIC KNEE LOADING ASYMMETRIES 35
Table 4.1. Subject demographics (mean ± standard deviation)
ACLR (n = 21)
Age (year) 28.8 ± 11.2
Body mass (kg) 69.7 ± 13.1
Body height (cm) 170.9 ± 9.9
Time after surgery (months) 5.1 ± 1.5
Graft Type
10 BTB autograft
3 hamstring autograft
8 allograft
Subjects were excluded from the study if they: (1) were not cleared by physical therapist
to perform the functional activities, (2) had prior ACL injury and knee surgery on the contralateral
limb, (3) concurrent pathology or morphology that could cause pain or discomfort during physical
activity, and (4) any physical, cognitive or other condition that may impair subjects’ ability to
perform the tasks proposed in this study. An a priori power analysis performed using pilot data
from 6 subjects determined that 14 subjects would provide more than 80% power at the alpha
level of 0.05.
Procedures
Testing took place in the Human Performance Laboratory of the Division of Biokinesiology
and Physical Therapy at University of Southern California located at the Completive Athletic
Training Zone, Pasadena CA. All procedures were explained to each subject and informed
consent was obtained as approved by the Investigational Review Board at the University of
Southern California Health Sciences Campus. Parental consent was obtained for all subjects
under the age of 18.
DYNAMIC KNEE LOADING ASYMMETRIES 36
After consenting to participate, the subject’s age, height, weight, tibia length, knee medical
history, and physical activity prior to injury were recorded. To ensure individuals met the criteria
of being a recreational athlete, physical activity prior to knee injury was determined using the
Cincinnati Sports Activity Scale (Barber-Westin et al., 1999). Recreational athlete was defined as
Level I or II on the Cincinnati Sports Activity scale.
Prior to testing, participants were asked to warm −up on a stationary bike for five minutes.
Reflective markers were placed on first and fifth metatarsals, end of distal second toes, medial and
lateral malleoli, medial and lateral epicondyles of femurs, greater trochanters, posterior superior
iliac spines, iliac crests and L5 −S1 junction. In addition, reflective markers secured on rigid plates,
tracking marker clusters, were secured bilaterally on participant’s thigh, lower leg and heel of
their shoe by the same examiner. All markers were removed after a static calibration trial was
collected except the rigid plates, pelvis markers and distal toe markers which remained on during
testing.
Inertial sensors were placed on the mid lateral thighs and shanks with the X −axis aligned
superior −inferior, bilaterally. Care was taken to align the X −axis of the thigh sensors with the
greater trochanter and lateral epicondyle of the femur, and X −axis of the shank sensors with the
lateral epicondyle and lateral malleoli (Figure 4.1). For testing the position of the inertial sensors
coincided with the position of the tracking marker clusters therefore, they were affixed to the
rigid plates firmly using elastic Velcro® straps and tape.
DYNAMIC KNEE LOADING ASYMMETRIES 37
Figure 4.1: Orientation and location of inertial sensors and markers on the lower
extremity during testing
Kinematic data and ground reaction force data were collected using a marker −based, 11
camera motion capturing system (Qualysis Inc., Gothenberg, Sweden) at a 250Hz and force
platforms at 1500Hz (Advanced Mechanical Technologies, Inc., Newton, MA, USA) or a 14 camera
motion capturing system (BTS Bioengineering Corp., Milan, Italy) at 340Hz and force platforms
at 1360Hz (BTS Bioengineering Corp., Milan, Italy). Inertial data was collected using four inertial
sensors equipped with a tri −axial accelerometers, tri −axial gyroscopes and tri −axial
magnetometers (Opal brand, APDM Inc., Portland, OR, USA) simultaneously with kinematic and
ground reaction force data. The primary variable of interest from the inertial sensors, angular
velocity, was measured using the gyroscope. While direct measurements from the accelerometer
Y
X
DYNAMIC KNEE LOADING ASYMMETRIES 38
and magnetometer were not used for analysis, they remained active throughout data collection to
increase the accuracy of the gyroscope measurements using APDM’s proprietary algorithm. The
range for the gyroscope in the X − and Y −axes is ±34.9 rad/s and the Z −axis is ±26.8 rad/s. The
gyroscope’s noise density in the X − and Y −axes is 0.81 mrad/s/ Hz and 2.2mrad/s/ Hz for the
Z −axis. Inertial data was recorded at 128Hz using Motion Studio software (APDM Inc., Portland,
OR, USA). Data was wirelessly streamed from all four sensors directly to the computer using the
“Robust Synchronized Streaming” mode. During this mode of data collection information was
streamed from multiple synchronized sensors directly to the computer. Data was buffered on the
sensors to prevent data loss in the case of wireless interruptions.
Single Limb Loading Test
During testing, participants performed a dynamic single limb loading (SLL) task on each
limb. For this task subjects were instructed to stand on both feet on a single platform in front of
a target. Tape was placed on the ground on an adjacent force platform as a target for foot
placement. The distance to the target was normalized to each individual as a distance equal to
the length of their tibia. Subjects were asked to leap forward to the target location onto a single
limb and to lower as far they could and then return to the start position. For each trial they were
asked to perform three consecutive repetitions. Participants performed single limb loading trials
alternating between limbs beginning with the non −surgical limb. A trial was considered accepted
when it contained presence of a distinct flight phase, maintenance of balance throughout the task
and complete foot placement on the target force platform. The presence of a flight phase was
considered as criteria for a successful trial to avoid instances of double limb support. It was
determined by the absence of forces on either force plate prior to foot contact on the target force
DYNAMIC KNEE LOADING ASYMMETRIES 39
platform. Practice trials were allowed for subjects to become familiar with the task. Participants
performed three trials on each limb.
Data Analysis
Three −dimensional marker coordinates (Qualysis Inc. Tracking Manager, BTS
Bioengineering Corp. SMARTtracker) were reconstructed and in combination with force data
were used to calculate joint kinematics, kinetics and energetics (Visual 3D
TM
, Version 4.8,
C −Motion, Inc., Rockville, MD, USA). Coordinate data was low −pass filtered using a fourth order
zero −lag Butterworth filter with a 12Hz cut −off frequency. The local coordinate systems of body
segments were derived from the standing calibration trial. Lower extremity segments were
modeled as a frusta of cones, while the pelvis was modeled as a cylinder. Six degrees of freedom
of each segment were calculated by transforming the triad of markers to the position and
orientation of each segment during the standing calibration trial. Euler angles with the following
order of rotations were used to calculate joint kinematics: flexion/extension,
abduction/adduction, and internal/external rotation. Joint angles were expressed as movement
of the distal segment relative to the proximal segment. Kinematics, anthropometrics and ground
reaction forces were used in standard inverse dynamics equations to calculate internal net joint
moments. Net joint power was calculated as the product of joint moment and joint angular
velocity. All kinetic and energetic data were normalized to body mass. Segment angular velocities
measured with the marker −based motion analysis system were calculated with respect to the
global coordinate system. Data obtained from Visual3D
TM
were exported and analyzed using a
customized MATLAB® program (Version R2014b, The MathWorks, Matick, MA, USA).
DYNAMIC KNEE LOADING ASYMMETRIES 40
Signals from the inertial sensors placed on the thighs and shanks were used to measure
thigh and shank angular velocity, respectively. Angular velocities, a direct output of the
gyroscope, in the Z −plane of the sensor were chosen to represent sagittal plane movement (Figure
4.2). Thigh and shank angular velocities were negated on the subjects’ right side to coincide with
the global coordinate system where knee flexion involved negative rotation of the proximal thigh
and positive rotation of the proximal shank segment from vertical (Figure 4.2). Angular velocity
data was low −pass filtered using a fourth order zero −lag Butterworth filter with a 15Hz cut −off
frequency. Knee angular velocity was calculated as the sum of thigh and shank angular velocities
at each time point with positive rotation representing knee flexion. All dependent variables were
identified during the deceleration phase; defined as the time between heel strike and peak knee
flexion. Identification of events from inertial sensors was made from segment angles during the
single limb task. Angles were calculated from segment angular velocity by converting the output
from radians to degrees per second by multiplying by 180 and dividing by π and then integrating
these data using the cumtrapz function in MATLAB®. Heel strike and toe off were determined
by identifying local minimums of shank angle time series graphs. Peak knee flexion angle was
determined by identifying the minimum thigh angle between heel strike and toe off. Customized
MATLAB® programs were used to identify variables of interest extracted from inertial sensors.
Peak knee power absorption was identified using the marker −based motion capture
system and peak knee, thigh and shank angular velocities in sagittal plane were identified using
both the marker −based system and inertial sensors. The average of three trials (middle repetition
of each trial) of each limb (ACL reconstructed (ACLR), non −surgical (Non −Sx)) were used for
analysis in both systems.
DYNAMIC KNEE LOADING ASYMMETRIES 41
Statistical Analysis
To determine the best predictors of knee power absorption, stepwise linear regression was
performed using data calculated from marker −based motion capture. Peak power absorption was
the dependent variable and peak knee, thigh and shank angular velocities were the independent
variables. To determine the best predictors of knee power absorption using angular velocities
measured with inertial sensors, a separate stepwise linear regression was performed. Peak power
absorption was the dependent variable and peak knee, thigh and shank angular velocities derived
from inertial sensors were independent variables. For both regression models, data from ACLr and
Non −Sx limbs were considered together as initial multiple linear regression analysis that included
limb as an independent variable determined that limb had no significant effect on the
relationships; marker −based (p 0.118), inertial sensors (p 0.072). Therefore data presented
below represents combined data from both limbs. One −tailed Pearson product −moment
correlations were used to quantify the strength of the relationship between knee power
absorption and angular velocities in both measurement systems. A strong correlation was defined
as a correlation greater than 0.75 and a moderate correlation was defined as a correlation between
0.50 and 0.75.
Further, to quantify the level of agreement between measurement systems, concurrent
validity was determined using intraclass correlation coefficients (ICC) (2,k). For clinical
measurements agreement between measurement systems should exceed 0.90 to ensure reasonable
validity. Statistical analyses were performed using PASW software (version 18, SPSS, Inc.,
Chicago, IL) with a significance level of 0.05.
DYNAMIC KNEE LOADING ASYMMETRIES 42
Results
Descriptive statistics for 21 participants can be found in Table 4.2. When considering
angular velocities extracted from marker −based motion analysis, in a stepwise regression model,
knee angular velocity (p 0.001; R
2
0.707) was the only variable to enter the regression model;
explaining 71% of the variance in peak knee power absorption during SLL. Knee angular velocity
(r 0.845, p 0.001; Figure 4.2A) and thigh angular velocity (r 0.796, p 0.001; Figure 4.2B)
were positively and strongly correlated with peak knee power absorption. Shank angular velocity
(r 0.639; p 0.001) was moderately correlated with peak knee power absorption. Positive
correlations indicated that faster angular velocities were related to greater peak knee power
absorption (Figure 4.2).
Table 4.2: Descriptive statistics for the reconstructed (ACLr) and non −surgical(Non −Sx)
limb for joint and segment variables measured with marker −based motion capture and inertial
sensor measurement systems; Data represents mean ± standard deviation and (range)
ACLr Limb Non-Sx Limb
Marker-based Inertial Sensor Marker-based Inertial Sensor
Knee Power Absorption
(W/kg)
5.8 ± 3.5
(1.3-15.4)
NA
9.2 ± 2.4
(5.6-14.7)
NA
Knee Angular Velocity
(deg/s)
328.0 ± 97.5
(154.4-521.2)
326.5 ± 100.6
(154.0-508.7)
420.1 ± 81.7
(249.6-543.3)
410.5 ± 84.55
(197.1-532.5)
Thigh Angular Velocity
(deg/s)
156.0 ± 62.9
(48.7-276.0)
152.0 ± 67.2
(39.7-264.0)
210.0 ± 52.53
(108.0-308.0)
207.0 ± 54.0
(75.3-291.0)
Shank Angular Velocity
(deg/s)
205.2 ± 53.8
(124.0-327.0)
195.1 ± 53.2
(119.8-301.3)
224.9 ± 33.79
(147.8-284.6)
221.5 ± 34.8
(125.7-270.7)
DYNAMIC KNEE LOADING ASYMMETRIES 43
Figure 4.2: The relationship between (A) peak knee angular velocities and (B)
peak thigh angular velocities measured with marker −based motion capture and
peak knee power absorption in the reconstructed (ACLr) and non −surgical
(Non −Sx) limb; **p 0.001
When considering segment variables extracted from inertial sensors in a stepwise
regression model, thigh angular velocity (p 0.001; R
2
0.660) was the only variable to enter the
regression model; it explained 66% of the variance in peak knee power absorption during SLL.
Peak knee angular velocity (r 0.806, p 0.001; Figure 4.3A) and thigh angular velocity (r 0.812,
p 0.001; Figure 4.3B; Equation 4.1) were strongly correlated and shank angular velocity (r
0.596; p 0.001) was moderately correlated with peak knee power absorption (Figure 4.3). Joint
and segment angular velocities were positively correlated with peak knee power absorption
indicating faster velocities were related to larger peak knee power absorption.
4 .1 0 .042 ∗ 0 .087
DYNAMIC KNEE LOADING ASYMMETRIES 44
Figure 4.3: The relationship between (A) peak knee angular velocities and (B)
peak thigh angular velocities measured with inertial sensors and peak knee power
absorption in the reconstructed (ACLr) and non −surgical (Non −Sx) limb; **p
0.001
High intraclass correlation coefficients (ICC 0.90) indicated strong agreement between
measurement systems for both thigh and shank angular velocities in the sagittal plane during SLL
(Table 4.3).
Table 4.3: Intraclass correlation coefficients (2,k) between marker −based
motion capture and inertial sensor measurements for peak knee angular
velocity and peak thigh and shank angular velocities measured in all limbs, the
reconstructed (ACLr) and non −surgical (Non −Sx) limb.
ALL Limbs ACLr Limb Non-Sx Limb
Knee Angular Velocity 0.978** 0.989** 0.950**
Thigh Angular Velocity 0.967** 0.947** 0.973**
Shank Angular Velocity 0.962** 0.95** 0.978**
** Indicates significance; p < 0.001
DYNAMIC KNEE LOADING ASYMMETRIES 45
Discussion
The primary purpose of this study was to determine if segment angular velocities
measured with inertial sensors can provide meaningful information regarding knee power
absorption during a single limb loading task performed by individuals following ACLr. Chapter
III and previous studies demonstrated that individuals following ACLr exhibit deficits in dynamic
knee loading that are challenging to detect in the clinic (Ernst et al., 2000; Orishimo et al., 2010).
Findings from this study provide the foundation for using inertial sensors to detect altered knee
loading without the presence of force plates or marker −based motion system during this single
limb loading task. This is important as the knee plays an essential role in force attenuation during
high level dynamic tasks (Buczek & Cavanagh, 1990; Dai et al., 2012; Moolyk et al., 2013) that
these individuals are hoping to return to without a risk of re −injury.
The strong relationship between peak knee power absorption and knee angular velocity
in the marker −based system is not surprising given that power is calculated as the product of knee
angular velocity and sagittal plane moment (Winter, 2009). When all marker −based motion
analysis variables were considered together, sagittal plane peak knee angular velocity was the best
predictor of knee power absorption explaining nearly 70% of the variance during SLL. After
accounting for the effects of knee angular velocity, segment angular velocities did not add any
additional information. The strength of the relationships suggests that angular velocity alone can
provide meaningful information regarding knee power without marker derived kinematics and
ground reaction forces. Thigh and shank angular velocities, measured with a marker −based
motion analysis, were strongly and moderately related to peak knee power absorption,
respectively. Knee angular velocity is a combination of thigh and shank angular velocities
(Winter, 2009); therefore, an increase in thigh or shank angular velocity would coincide with an
DYNAMIC KNEE LOADING ASYMMETRIES 46
increase in knee angular velocity and subsequently knee power. This relationship likely precluded
their inclusion in the prediction model as they did not add any additional information after knee
angular velocity was considered. However, the strength of segment angular velocity relationships
suggests segment kinematics, in particular thigh angular velocities, may be useful in
characterizing knee power absorption deficits during this single limb loading task. It is not clear
why shank angular velocity only had a moderate relationship to knee power. The range of angular
velocities at the shank was smaller than those observed at the knee, but of similar magnitudes.
Findings from this study suggest that the motion at the thigh is more directly related to knee
flexion during this task. The instructions to perform the task encourage individuals to lower
themselves as far as they can. This may have resulted in large hip flexion angles increasing the
contribution of the thigh to knee flexion. However, future work is needed to determine if thigh
and shank kinematics are reflective of hip and ankle kinematics, respectively.
Calculation of segment angular velocities using marker −based motion capture systems is
not feasible for clinical use. However, the results of this study suggest that segment angular
velocities measured with inertial sensors and marker −based motion analysis systems provide
similar information supporting the use of inertial sensors in the clinic. The agreement between
the marker −based system and inertial sensors was high, with ICCs ranging between 0.94 and
0.989, when measuring knee and segment angular velocities. While inertial sensors are a direct
measurement of segment angular velocities and marker −based measurements involve calculations
from marker positions, strong intraclass correlation coefficients confirm that both methods may
be used to quantify thigh and shank angular velocities. In addition, knee angular velocities that
involve calculations when measured with inertial sensors and marker −based system, also had
strong intraclass correlation coefficients. Together, these data confirm that direct measurements
DYNAMIC KNEE LOADING ASYMMETRIES 47
from the gyroscope of inertial measurement devices and calculated joint measurements provide a
feasible alternative for marker −based motion analysis systems.
When inertial sensor variables were considered together, sagittal plane peak thigh
angular velocity was the best predictor of knee power absorption explaining 66% of the variance
during single limb loading. After accounting for the effects of thigh angular velocity, knee and
shank angular velocities did not add any additional information. It is not surprising that
correlations of similar magnitude were found between knee power and angular velocities
measured with inertial sensors and marker −based motion analysis. Knee (r 0.806) and thigh (r
0.812) angular velocities were strongly correlated with knee power. The correlation was slightly
higher for thigh angular velocity; as a result, it was determined to be the stronger predictor of knee
power when measured with inertial sensors. The similarities in the strength of these correlations
suggest both variables could provide similar information about knee power. Using direct output
of a single sensor on the thigh would be more practical for clinical use rather than calculating joint
kinematics from sensors on the thigh and shank. The strength of these relationships exceeded
previously reported relationships between coronal plane thigh and shank angular velocities and
knee adductor moments during single and double limb drop lands (Dowling et al., 2012a). It is
not surprising that segment angular velocity was more related to joint power than moments given
how they are calculated. Moreover, this single limb loading task requires significant sagittal plane
motion at the knee whereas frontal plane motion is limited. The testing procedures described in
the current study allow for exploration of sagittal plane loading deficits commonly observed in
individuals following ACLr.
DYNAMIC KNEE LOADING ASYMMETRIES 48
The findings of this study are promising for clinical translation as they indicate the direct
measurement from inertial sensors are highly related to knee power, and only sensors placed on
the thighs are needed to predict knee power and the ability to rapidly accommodate forces in the
lower limb during this single limb loading task. Angular velocity measurements with inertial
sensors provide meaningful information about an individual’s ability to accommodate forces on
their limb following ACL reconstructive surgery during phases of tasks that are too quick for our
eyes and traditional video recorders to capture. It is likely that thigh angular velocity measured
with inertial sensors is highly sensitive to difference in power observed between limbs or changes
over time. The regression equation indicated that a 0.042 deg/s change in thigh angular velocity
coincides with a 1 W/kg change in knee power absorption (Figure 4.4B). Interestingly, limb did
not influence the relationship between sagittal plane angular velocities and knee power
absorption during this task despite the presence of between limb differences in angular velocities
and knee power at this time point in rehabilitation. This suggests that the non −surgical limb may
be used for comparison to assess knee loading asymmetries in the clinic, as seen commonly in
assessment of rehabilitation progression (Kvist, 2004; Myer, G.D. et al., 2011; Thomee et al., 2011).
While these findings set the foundation for quantifying knee power with angular velocity
measurements extracted from inertial sensors, they are limited to the single limb task assessed in
this study. It is not clear if similar relationships exist during other dynamic tasks such running
or hopping. For application of these data to the clinic, further work is needed to determine the
sensitivity and specificity of these measures for identifying altered knee loading.
DYNAMIC KNEE LOADING ASYMMETRIES 49
CHAPTER V
DETECTING KNEE LOADING ASYMMETRIES IN THE CLINIC: IMPLICATIONS FOR
REHABILITATION FOLLOWING ACL RECONSTRUCTION
Abstract
Individuals following ACLr present with significant deficits in knee power during portions of
dynamic tasks that require eccentric control or deceleration (Chapter III). The inability to
quantify these knee loading deficits clinically without the use of marker −based motion analysis
may underlie their persistence. The strong relationship between thigh angular velocity measured
with inertial sensors and knee power absorption calculated with marker −based motion analysis
suggest that inertial sensors may be used to quantify dynamic knee loading asymmetries in the
clinic (Chapter IV). Therefore, the purpose of this chapter is to determine the diagnostic accuracy
of using inertial sensor thigh angular velocities to detect asymmetrical knee loading during a
dynamic single limb loading task in twenty −one individuals following ACLr. Between limb ratios
(reconstructed/non −surgical limb) were calculated for knee power, calculated from the
marker −based system, and thigh angular velocity, extracted from inertial sensors. To determine
the relationship between ratios, a linear regression was performed using knee power and thigh
angular velocity ratios. Sensitivity and specificity of thigh angular velocity in diagnosing
asymmetrical knee loading, knee power ratio less than 0.85, was determined using receiver
operating characteristic curve analysis. Thigh angular velocity ratios were strong predictors of
knee power ratios. Thigh angular velocity ratios could diagnose asymmetrical knee loading, knee
power deficits greater than 15% when performing the single limb loading task, with high
sensitivity and specificity. These findings support the use of cost −effective wearable sensors to
objectively quantify movement clinically in this population of individuals following ACLr.
DYNAMIC KNEE LOADING ASYMMETRIES 50
Introduction
Individuals following ACL reconstruction (ACLr) continue to present with altered
sagittal plane knee loading patterns 6 to 24 months post −surgery as they progress to participation
in higher levels of physical activities and sports (Noehren et al., 2013; Oberlander et al., 2014;
Roewer et al., 2011; Webster et al., 2004). The presence of altered loading strategies at this time
point is of particular concern as they are related to an increased risk for re −injury.
A recent
prospective study found that 23% of athletes who exhibited asymmetrical knee loading at the
time they return to their sport sustained a second ACL injury (Paterno et al., 2010). These data
suggest that clinical identification and restoration of asymmetrical sagittal plane knee loading
may be particularly important following ACLr.
Biomechanically, asymmetrical sagittal knee loading following ACLr is characterized by
decreases in knee extensor moments and power absorption in the reconstructed knee compared
to non −surgical and healthy control knees during dynamic tasks that require eccentric control or
deceleration as seen in Chapter III and previous studies (Ernst et al., 2000; Oberlander et al., 2013;
Salem et al., 2003; Webster et al., 2004).
Deficits in knee power, the product of magnitude (knee
moment) and rate(knee angular velocity) of knee loading, across a variety of tasks is of particular
concern as dynamic activities and sports are performed quickly requiring substantial knee power
(Winter, 2009). Difficulty rapidly attenuating forces during dynamic tasks is seen during
rehabilitation as individuals begin running. At approximately 5 months post −surgery deficits in
knee power absorption and knee angular velocity are as large as 37% and 21%, respectively
(Chapter III). While this may not be surprising given the demands of running, similar deficits are
also present at this time during a less demanding dynamic single limb loading task (Chapter III).
These data suggest that individuals following ACLr may have impairments related to the rate of
DYNAMIC KNEE LOADING ASYMMETRIES 51
loading irrespective of the magnitude of loading demands. Similar impairments are also observed
during running at 3 years post −surgery indicating that deficits in loading rate may persist long
term if not addressed (Devita et al., 1992).
Clinically, the presence and magnitude of such deficits are often difficult to detect; thus,
challenging clinicians’ ability to recognize and address them. Assessment of movement quality is
typically made subjectively with visual observation, although, a recent rise in use of video has
made assessments of movement more objective. However, large knee loading deficits often
coincide with smaller changes in kinematics making them difficult to detect not only visually, but
also with current video technology (Orishimo et al., 2010; Salem et al., 2003). For example, during
deceleration of a single limb hop, a 43% decrease in peak power absorption corresponded with
only a 10 degree decrease in peak knee flexion angle in the reconstructed limb when compared to
the non −surgical limb(Orishimo et al., 2010). Small differences in joint excursions may be
particularly difficult to observe during dynamic tasks as individuals go through nearly 30 −50
degrees of flexion in less than 200 milliseconds. Even with the use of current two −dimensional
video assessment, the standard video capture rates (24 −32 frames per second) restrict the ability
to accurately quantify differences in joint excursions during tasks performed at faster speeds.
Additionally, without the use of force platforms, inferences regarding joint loading from joint
position data are limited. More sensitive objective metrics are needed for clinical identification
and quantification of these deficits during dynamic tasks.
Three −dimensional marker −based motion analysis systems, that combine high speed
motion and ground reaction force data to calculate joint kinematics and kinetics, are currently the
gold standard for objectively assessing human motion. However, these analyses are complex and
time consuming thus impractical in the clinic. More recently, wearable inertial sensors have been
DYNAMIC KNEE LOADING ASYMMETRIES 52
used to quantify human motion. These sensors, often comprised of accelerometers, gyroscopes
and/or magnetometers, are capable of collecting kinematic data (linear acceleration and angular
velocity) at high capture rates (greater than 120 frames per second) in multiple planes. Recent
studies suggest that these outputs are sensitive enough to detect movement impairments in
populations with orthopedic impairments (Patterson et al., 2014). Specifically, Chapter IV
demonstrated that a strong association exists between thigh angular velocity measured using
inertial sensors and knee power absorption determined from gold standard motion analysis
during a single limb loading task in individuals following ACLr. Peak thigh angular velocity
predicted nearly 66% of the variance in knee power in individuals 4 to 6 months post −surgery
(Chapter IV). The strength of this relationship suggests that inertial sensors have the potential
to provide valuable information about knee loading deficits, specifically knee power, without the
presence of force platform data.
While a strong correlation was found between peak thigh angular velocities and knee
power absorption during single limb loading, it is not known if the strength of this relationship
is sensitive enough to discriminate between symmetrical and asymmetrical loading deficits
(Chapter IV). Traditionally, comparisons are made between the reconstructed limb and
non −surgical limb during rehabilitation to determine progress. These comparisons are often
represented by a between limb ratio representing a level of symmetry. Generally speaking,
reaching a ratio of 85 −90% is often used as a standard to allow progression to more challenging
activities (Kvist, 2004; Myer, G.D. et al., 2011; Thomee et al., 2011). It is important to determine
the sensitivity and specificity in diagnosing asymmetrical knee power in individuals following
ACLr during a single limb loading task using inertial sensors’ angular velocity measurements
before applying these methods to clinical practice. Therefore, the aim of this study was to
DYNAMIC KNEE LOADING ASYMMETRIES 53
determine if measurements of thigh angular velocity during a dynamic single limb loading task
could be used as a proxy for knee power absorption clinically by providing appropriate
information regarding between limb deficits. First, it was determined if between limb ratios of
angular velocity were related to a between limb ratios of power in individuals post −ACLr during
a single limb loading task. Using the same data, discriminative accuracy was determined by
calculating the area under the receiver operating characteristic curve, the sensitivity and
specificity, and likelihood ratios in diagnosing asymmetrical knee loading (between limb knee
power ratio 85%) with thigh angular velocity ratios. It was hypothesized that thigh angular
velocity measurements extracted from inertial sensors could detect asymmetrical knee power
with high sensitivity and specificity.
Methods
Participants
Twenty −one individuals (12 females) who had primary unilateral ACL reconstruction (11
right and 10 left) using a bone −patellar −tendon −bone (BTB) autograft, allograft or hamstring
autograft approximately 5.1 ± 1.5 months prior to testing participated (Table 5.1). All subjects
reported that they were recreationally active prior to their injury (evaluated using Cincinnati
Sports Activity questionnaire) (Barber-Westin et al., 1999). At the time of participation subjects
were actively attending physical therapy and had initiated a running progression within 2 months
of testing.
DYNAMIC KNEE LOADING ASYMMETRIES 54
Table 5.1. Subject demographics (mean ± standard deviation)
ACLR (n = 21)
Age (year) 28.8 ± 11.2
Body mass (kg) 69.7 ± 13.1
Body height (cm) 170.9 ± 9.9
Time after surgery (months) 5.1 ± 1.5
Graft Type
10 BTB autograft
3 hamstring autograft
8 allograft
Subjects were excluded from the study if they: (1) were not cleared by physical therapist
to perform the functional activities, (2) had prior ACL injury and knee surgery on the contralateral
limb, (3) concurrent pathology or morphology that could cause pain or discomfort during physical
activity, and (4) any physical, cognitive or other condition that may impair subjects’ ability to
perform the tasks proposed in this study. Two subjects had prior ACL surgery on the same limb
approximately four and thirty years prior to the current injury and surgery, two subjects had prior
ACL surgery on the opposite limb approximately three and four years prior to the current injury
and surgery, and three subjects had partial tears of their ACL with no surgery two, eight and
fifteen years prior to the current injury and surgery. All individuals were included in the study.
Procedures
Testing took place in the Human Performance Laboratory of the Division of Biokinesiology
and Physical Therapy at University of Southern California located at the Completive Athletic
Training Zone, Pasadena CA. All procedures were explained to each subject and informed
consent was obtained as approved by the Investigational Review Board at the University of
DYNAMIC KNEE LOADING ASYMMETRIES 55
Southern California Health Sciences Campus. Parental consent was obtained for all subjects
under the age of 18.
After consenting to participate, the subject’s age, height, weight, tibia length, knee medical
history, and physical activity prior to injury were recorded. To ensure individuals met the criteria
of being a recreational athlete, physical activity prior to knee injury was determined using the
Cincinnati Sports Activity Scale (Barber-Westin et al., 1999). Recreational athlete was defined as
Level I or II on the Cincinnati Sports Activity scale.
Prior to testing, participants were asked to warm −up on a stationary bike for five minutes.
Reflective markers were placed on first and fifth metatarsals, end of distal second toes, medial and
lateral malleoli, medial and lateral epicondyles of femurs, greater trochanters, posterior superior
iliac spines, iliac crests and L5 −S1 junction. In addition, reflective markers secured on rigid plates,
tracking marker clusters, were secured bilaterally on participant’s thigh, lower leg and heel of
their shoe by the same examiner. All markers were removed after a static calibration trial was
collected except the rigid plates, pelvis markers and distal toe markers remained which remained
on during testing.
Inertial sensors were placed on the mid lateral thighs and shanks with the X −axis aligned
superior −inferior, bilaterally. Care was taken to align the X −axis of thigh sensors with the greater
trochanter and lateral epicondyle of the femur, and the X −axis of shank sensors with the lateral
epicondyle and lateral malleoli. For testing the position of the inertial sensors coincided with the
position of the tracking marker clusters, therefore, they were affixed to the rigid plates firmly
using elastic Velcro® straps and tape (Figure 5.1).
DYNAMIC KNEE LOADING ASYMMETRIES 56
Figure 5.1: Orientation and location of inertial sensors and markers on the
lower extremity during single limb loading test
Kinematic data and ground reaction force data were collected using a marker −based, 11
camera motion capturing system (Qualysis Inc., Gothenberg, Sweden) at a 250Hz and AMTI force
platforms at 1500Hz (Advanced Mechanical Technologies, Inc., Newton, MA, USA) or a 14 camera
motion capturing system (BTS Bioengineering Corp., Milan, Italy) at 340Hz and force platforms
at 1360Hz (BTS Bioengineering Corp., Milan, Italy). Inertial data was collected using four inertial
sensors equipped with a tri −axial accelerometers, tri −axial gyroscopes and tri −axial
magnetometers (Opal brand, APDM Inc., Portland, OR, USA) simultaneously with kinematic and
ground reaction force data. The primary variable of interest from the inertial sensors, angular
velocity, was measured using the gyroscope. While direct measurements from the accelerometer
Y
X
DYNAMIC KNEE LOADING ASYMMETRIES 57
and magnetometer were not used for analysis, they remained active throughout data collection to
increase the accuracy of the gyroscope measurements using APDM’s proprietary algorithm. The
range for the gyroscope in the X − and Y −axes is ±34.9 rad/s and the Z −axis is ±26.8 rad/s. The
gyroscope’s noise density in the X − and Y −axes is 0.81 mrad/s/ Hz and 2.2mrad/s/ Hz for the
Z −axis. Inertial data was recorded at 128Hz using Motion Studio software (APDM, Inc. Portland,
OR, USA). Data was wirelessly streamed from all four sensors directly to the computer using the
“Robust Synchronized Streaming” mode. During this mode of data collection, information was
streamed from multiple synchronized sensors directly to the computer. Data was buffered on the
sensors to prevent data loss in the case of wireless interruptions.
Single Limb Loading Test
During testing, participants performed a dynamic single limb loading (SLL) task on each
limb (Figure 5.2). For this task subjects were instructed to stand on both feet on a single platform
in front of a target. Tape was placed on the ground on an adjacent force platform as a target for
foot placement. The distance to the target was normalized to each individual as a distance equal
to the length of their tibia. Subjects were asked to leap forward to the target location onto a single
limb and to lower as far they could and then return to the start position. For each trial they were
asked to perform three consecutive repetitions. Participants performed single limb loading trials
alternating between limbs beginning with the non −surgical limb. A trial was considered accepted
when it contained presence of a distinct flight phase, maintenance of balance throughout the task
and complete foot placement on the target force platform. The presence of a flight phase was
considered as criteria for a successful trial to avoid instances of double limb support. It was
determined by the absence of forces on either force plate prior to foot contact on the target force
DYNAMIC KNEE LOADING ASYMMETRIES 58
platform. Practice trials were allowed for subjects to become familiar with the task. Participants
performed 3 trials on each limb.
Figure 5.2: Single Limb Loading Test*
(*Subject figure reproduced with permission)
Data Analysis
Three −dimensional marker coordinates (Qualysis Inc. Tracking Manager, BTS
Bioengineering Corp. SMARTtracker) were reconstructed and in combination with force data
were used to calculate joint kinematics, kinetics and energetics (Visual 3D
TM
, Version 4.8,
C −Motion, Inc., Rockville, MD, USA). Coordinate data was low −pass filtered using a fourth order
zero −lag Butterworth filter with a 12Hz cut −off frequency. The local coordinate systems of body
segments were derived from the standing calibration trial. Lower extremity segments were
modeled as a frusta of cones, while the pelvis was modeled as a cylinder. Six degrees of freedom
of each segment were calculated by determined transforming the triad of markers to the position
DYNAMIC KNEE LOADING ASYMMETRIES 59
and orientation of each segment during the standing calibration trial. Euler angles with the
following order of rotations were used to calculate joint kinematics: flexion/extension,
abduction/adduction, and internal/external rotation. Joint angles were expressed as movement
of the distal segment relative to the proximal segment. Kinematics, anthropometrics and ground
reaction forces were used in standard inverse dynamics analysis to calculate internal net joint
moments. Net joint power was calculated as the product of joint moment and joint angular
velocity. All kinetic and energetic data were normalized to body mass. Data obtained from
Visual3D
TM
were exported and analyzed using a customized MATLAB® program (Version
R2014b, The MathWorks, Matick, MA, USA).
Signals from the inertial sensors placed on the thighs and shanks were used to
measurement thigh and shank angular velocity, respectively. Angular velocities, a direct output
of the gyroscope, in the Z −plane of the sensor were chosen to represent sagittal plane movement
(Figure 5.1). Thigh and shank angular velocities were negated on the subjects’ right side to
coincide with the global coordinate system where knee flexion involved negative rotation of the
proximal thigh and positive rotation of the proximal shank segment from vertical (Figure 5.1).
Angular velocity data was low −pass filtered using a fourth order zero −lag Butterworth filter with
a 15Hz cut −off frequency. All dependent variables were identified during the deceleration phase;
defined as the time between heel strike and peak knee flexion. Identification of events using
inertial sensors was made from segment angles during the single limb task. Angles were
calculated from segment angular velocity by converting the output from radians to degrees per
second, multiplying by 180, and dividing by π. Then segment angular velocities were integrated
using the cumtrapz function in MATLAB®. Heel strike and toe off were determined by identifying
local minimums of shank angle time series graphs. Peak knee flexion angle was determined by
DYNAMIC KNEE LOADING ASYMMETRIES 60
identifying the minimum thigh angle between heel strike and toe off. Customized MATLAB®
programs were used to identify variables of interest extracted from the inertial sensors.
The following variables were identified. Peak knee power absorption was calculated using
the marker −based motion capture system. Peak thigh angular velocities were identified using
measurements from inertial sensors. Knee power ratios were calculated by dividing peak knee
power absorption in the reconstructed limb by the non −surgical limb values. Similarly, thigh
angular velocity ratios were calculated by dividing peak thigh angular velocities in the
reconstructed limb by the non −surgical limb values. A ratio of 1 indicated that the reconstructed
limb and non −surgical limb were equal in value. A ratio less than 1 indicated that the
reconstructed limb had smaller values than the non −surgical limb. The average of three
repetitions (middle repetition of each trial) was used for analysis.
Statistical Analysis
To determine the relationship between knee power and thigh angular velocity ratios a
linear regression was performed using peak knee power absorption ratios calculated using
marker −based motion capture data and peak thigh angular velocities ratios measured using
inertial sensor data. The knee power ratio was the dependent variable and thigh angular velocity
ratio was the independent variable.
To determine the optimal threshold of thigh angular velocity ratio for diagnosis of knee
loading asymmetries, receiver operating characteristic (ROC) curve analysis was performed. The
area under the ROC curve (AUC) and 95% confidence intervals where calculated to represents
the probability that angular velocity ratio can discriminate between symmetrical and
asymmetrical knee power. Asymmetrical knee power was defined as a deficit greater than a 15%
DYNAMIC KNEE LOADING ASYMMETRIES 61
in knee power absorption in the reconstructed limb when compared to the non −surgical limb or
a knee power ratio less than 0.85. The AUC values ranged from 0 to 1, with an AUC of 1 indicating
100% probability that a given angular velocity ratio can discriminate between symmetrical and
asymmetrical knee power. In the case of a significant AUC, the cut −off point of angular velocity
ratio for distinguishing between individuals with and without asymmetrical knee power at the
highest sensitivity and specificity was identified. To facilitate interpretation and utilization of
the single limb loading test clinically, positive and negative likelihood ratios (LR+, LR −) were
calculated to characterize the value of thigh angular velocity measurements for identifying
asymmetrical loading in individuals following ACLr. A likelihood nomogram was used to
determine the probability that an individual similar to the participants of our study would have
asymmetrical loading using the established thigh angular velocity ratio threshold (Fetters &
Tilson, 2012). Statistical analyses were performed using PASW software (version 18, SPSS, Inc.,
Chicago, IL) with a significance level of 0.05.
Results
Knee power ratios (0.62 ± 0.29) ranged from 0.15 to 1.25 (Figure 5.3). Thigh angular
velocity ratios (0.73 ± 0.24) ranged from 0.21 to 1.22 (Figure 5.3). Between limb ratios for peak
thigh angular velocity (p 0.001; R
2
0.664) explained 66.4% of the variance in knee power ratios
(p 0.001; Equation 5.1, Figure 5.3). Thigh angular velocity ratios were positively correlated with
knee power ratios indicating that larger angular velocity ratios are related to greater knee power
ratios.
DYNAMIC KNEE LOADING ASYMMETRIES 62
5.1 0.97 ∗ 0.09
Figure 5.3: The relationship between thigh angular velocity ratios extracted from
inertial sensors and the knee power symmetry ratios calculated from
marker −based motion capture. A ratio of 1 indicated that ACLr limb Non −Sx
limb. A ratio 1 indicated that ACLr limb Non −Sx limb; **p 0.001
Of the 21 subjects 16 were categorized as having asymmetrical knee power (knee power
ratio less than 0.85). ROC curve analysis revealed significant AUC (0.90; p 0.008; 95%
Confidence Interval 0.765 to 1.000) for the use of thigh angular velocity ratios to discriminate
between asymmetrical and symmetrical knee power (Figure 5.4).
DYNAMIC KNEE LOADING ASYMMETRIES 63
Figure 5.4: The Receiver Operating Characteristic (ROC) curve for differentiating
between asymmetrical and symmetrical knee loading during SLL in individuals
following ACLr. The ROC curve provides a visual depiction of sensitivity and
specificity of thigh angular velocity ratio cutoff scores for detecting knee power
ratio 0.85.
Thigh angular velocity ratios less than or equal to 0.811 would classify the individual as
performing the SLL task with asymmetrical knee power at 81.2% sensitivity and 100% specificity
(Table 5.2). For these data, 100% specificity indicates that the inertial sensor measurements
resulted in no false positives; meaning that all of subjects who had symmetrical power based on
the gold standard (marker −based motion capture) were identified symmetrical using inertial
sensors. 81.2% sensitivity indicates that inertial sensor measurements resulted in a few false
negatives; meaning that all subjects who had asymmetrical power as indicated by the gold
standard, 81% (13/16) were identified as asymmetrical using inertial sensor measurements (Table
5.2). Only 3 of the 16 were considered symmetrical using inertial sensors when the gold standard
DYNAMIC KNEE LOADING ASYMMETRIES 64
test deemed them asymmetrical (Table 5.2). The positive likelihood ratio (LR+) was infinite and
negative likelihood ratio (LR −) was 0.188. Overall, 76.2% of the participants in this study had
asymmetrical knee power. Using the overall prevalence rate of asymmetry in this study (76.2 %)
as the estimated pre −test probability along with the calculated likelihood ratios, it was
determined that the post −test probability for having asymmetrical knee loading for a participant
who had thigh angular velocity ratio less than or equal to 0.811 was greater than 100% (Figure 5.5)
and for those who had a thigh angular velocity ratio greater than 0.811 was 38% (Figure 5.5).
Table 5.2: A 2 x 2 table of data from this study for the gold standard (marker −based knee power (KPow)
ratio) and new test (inertial sensor thigh angular velocity (ThAV) ratio).
Gold Standard:
Marker −Based Motion Analysis
Asymmetrical
(KPow Ratio < 0.85)
Symmetrical
(KPow Ratio ≥ 0.85)
Total
New Test:
Inertial
Sensors
Asymmetrical
(ThAV Ratio < 0.81)
13
true positive
0
false positive
13
Symmetrical
(ThAV Ratio ≥ 0.81)
3
false negative
5
true negative
8
Total 16 5 21
DYNAMIC KNEE LOADING ASYMMETRIES 65
76% = Overall probability of
asymmetrical loading
LR+ = infinite LR- = 0.188
Probability of asymmetrical knee
power based on thigh angular
velocity ratio:
= greater than 100% if had ThAV
ratio less than or equal to 0.81
= 38% if participant had ThAV
ratio greater than 0.81
Figure 5.5: The likelihood ratio nomogram is a graphical representation of the
probability that an individual following ACLr will have asymmetrical knee power
during a single limb loading task. Pre −test probability was estimated at 76%
(based on the overall percentage of participants in this study with asymmetrical
loading). The red line plots the positive likelihood ratio (LR+), used when the
thigh angular velocity (ThAV) ratio for an individual is less than or equal to 0.81.
It indicates that the individual has greater than 100% post −test probability of
having asymmetrical knee loading. The blue line plots the negative likelihood ratio
(LR −) used when an individual exceeds the thigh angular velocity ratio of 0.81. It
indicates the individual has only a 38% post −test probability of having
asymmetrical knee loading.
DYNAMIC KNEE LOADING ASYMMETRIES 66
Discussion
The primary purpose of this study was to determine if segment angular velocities
measured with inertial sensors could appropriately detect knee loading asymmetries in
individuals following ACLr performing the single limb loading task. Previous studies and Chapter
III demonstrated that individuals following ACLr exhibit dynamic knee loading deficits that can
be challenging to detect in the clinic (Ernst et al., 2000; Oberlander et al., 2013; Orishimo et al.,
2010; Webster et al., 2004). However, it is important identify asymmetrical loading as the knee
plays an important role in force attenuation during high level dynamic tasks (Buczek & Cavanagh,
1990; Dai et al., 2012; Moolyk et al., 2013) that these individuals are anticipating to return to
without a risk of re −injury. It may be particularly important to identify dynamic knee loading
deficits when individuals are progressed to running during rehabilitation as it represents a
progression to highly demanding dynamic tasks following surgery.
This study supports the use of thigh angular velocities extracted from inertial sensors for
clinical detection of knee power asymmetries in individuals following ACLr using the described
testing procedure. ROC curve analysis determined that thigh angular velocity ratios are able to
discriminate between asymmetrical and symmetrical knee power with high specificity (100%)
and sensitivity (81.2%). A between limb ratio in thigh angular velocity of 0.811 was determined to
be the critical cut −off for determining asymmetry in knee power that is greater than 15%. In the
context of the population tested, clinical interpretation of these data suggests that this test can
serve as a good diagnostic tool for identifying knee loading asymmetries using the described
testing procedures and single limb loading task. Given that the positive likelihood ratio was
calculated as infinite, the post-test probability analysis indicates that if an individual has a
between limb angular velocity ratio less than or equal to 0.81 indicating asymmetry, they are
DYNAMIC KNEE LOADING ASYMMETRIES 67
highly likely to have less power in their reconstructed knee (Figure 5.5). With a negative
likelihood ratio of 0.188 the post −test probability analysis indicates that if an individual has a
between limb angular velocity ratio greater than 0.81, indicating symmetry, there is only a 38%
probability they are actually asymmetrical with respect to knee power (Figure 5.5). These
findings are promising as it demonstrates the strong potential of using inertial sensor
measurements to identify knee loading asymmetries in the clinic.
The implications of these findings are exciting as they establish procedures for examining
dynamic knee loading in the clinic using inertial sensors. While previous studies and Chapter IV
determined that segment angular velocities measured with inertial sensors can provide objective
information regarding movement quality in individuals following ACLr (Patterson et al., 2014),
this is the first to translate these findings for use in the clinical setting. A need for more objective
information regarding joint mechanics in clinical decision −making is underscored by the inability
of current functional testing to identify mechanical deficits (Barber-Westin & Noyes, 2011;
Orishimo et al., 2010). Interpretation of current functional assessments including, distance
hopped or time to task completion, is limited to overall limb function. Furthermore, completion
of such tasks can be accomplished with compensatory patterns that increase the demands on the
hip and ankle to accomplish the overall goal (Orishimo et al., 2010; Salem et al., 2003). The current
testing procedure provides information specific to the knee during a functional single limb
loading task.
The procedures established in this study are particularly suited to detect the dynamic
loading deficits present at the time when patients are progressing to running (Chapter III).
Identification of decreased knee power and angular velocity may be most important at this time
DYNAMIC KNEE LOADING ASYMMETRIES 68
as running is typically one of the first dynamic tasks introduced during rehabilitation. The single
limb loading task used in the current study is appropriate for assessment of dynamic loading
deficits at this time because it requires rapid deceleration and high knee angular velocities, but
with much smaller demands with respect to the magnitude of loading when compared to running
(Chapter III). While a typical clinical cut off of 85% was used to define between limb symmetry,
a determination of when individuals should meet this standard for the single limb loading task
cannot be made using these data. However, the previously established relationship between knee
power asymmetries during this single limb task and running suggest that there may be some value
in using this test to determine readiness to initiate running. While only clinical accuracy of
identifying knee power asymmetries larger than 15% was established, the strong relationship
between angular velocity and power ratios suggests that the angular velocity ratio may be able to
provide a reasonable estimation of knee power ratio using the prediction equation (Equation 5.1).
The current testing paradigm can be applied to individuals 4 to 6 months following ACLr who
are beginning to run again after surgery. Further work is needed to establish the diagnostic
accuracy of such testing during other dynamic tasks, such as running, cutting or drop landing, to
allow for a broader application of these technologies.
DYNAMIC KNEE LOADING ASYMMETRIES 69
CHAPTER VI
SUMMARY AND CONCLUSIONS
This dissertation is the first to identify knee loading deficits related to the rate at which
individuals’ following anterior cruciate ligament reconstruction (ACLr) load their reconstructed
knee, and to offer a tool capable of identifying these deficits in the clinic. The contributions of
these studies to the understanding and detection of sagittal plane knee loading deficits are
particularly important because these deficits persist long after rehabilitation is complete despite
concentrated efforts to restore proper knee mechanics. Their persistence is concerning as these
individuals return to more dynamic activities, such as running, cutting and jumping, because these
deficits are associated with increased risk for re −injury. Re−injury rates as high as 40%
underscore the importance of addressing these factors (Barber-Westin & Noyes, 2011).
Chapter III is the first study to focus on the dynamics of knee loading by considering knee
power and angular velocity during dynamic tasks. Previous studies in this field highlight knee
loading deficits following ACL surgery, characterized by a decrease in the magnitude of loading
(knee extensor moment) in the reconstructed knee; however, understanding potential deficits in
loading rate is especially important as individuals’ progress to more high level physical activities.
This dissertation examined knee loading dynamics in individuals following ACLr who had
recently (within 1 month) progressed to running as a part of their rehabilitation. Running is one
of the first dynamic tasks introduced in rehabilitation, as early as 8 to 12 weeks post −surgery
(Adams et al., 2012; Logerstedt et al., 2010; van Grinsven et al., 2010), that requires individuals to
rapidly accommodate forces at the knee that are greater in magnitude than the exercises that
proceed them. Given that the rate at which an individual can load their reconstructed knee may
be constrained by magnitude of loading demands, comparisons were made between running and
DYNAMIC KNEE LOADING ASYMMETRIES 70
a single limb loading task. The single limb loading task required a similar rate of loading, but a
50% reduction in knee extensor moments (magnitude of loading). The non −surgical limb
represented a good control based on the absence of differences between the non −surgical limb and
the control limb across tasks and variables. Comparisons between limbs and tasks in individuals
following ACLr revealed deficits in power absorption (a combination of magnitude and rate of
knee loading), and angular velocity (rate of knee loading) during both tasks, in the reconstructed
compared to non −surgical knee. Interactions that trended toward significance suggest that
between limb deficits were larger for both variables during the more demanding task, running.
The findings in Chapter III add to the current understanding of sagittal plane loading deficits
following ACLr by highlighting the presence and level of impairments in ability to rapidly
accommodate loads at the knee. Given that previous studies have consistently described deficits
in knee extensor moments across tasks and throughout rehabilitation (Decker et al., 2002; Ernst
et al., 2000; Gokeler et al., 2010; Oberlander et al., 2013; Orishimo et al., 2010; Salem et al., 2003),
it is not surprising that smaller knee power absorption was observed in the reconstructed knee.
Large between limb differences in power observed during running would be expected given that
they recently progressed to running. During rehabilitation, running represents the introduction
of a more demanding training stimulus in both magnitude and rate of knee loading. However, the
persistence of deficits in knee power during running up to three years following surgery (Devita
et al., 1992) suggests that it may be difficult to resolve these impairments over time. In this
dissertation, the presence of decreased power and angular velocity in a task in which the demands
of loading were reduced by over 50% was somewhat surprising; as a criterion based progression
to running often requires that individuals be able to perform single limb loading tasks without
pain, swelling or obvious movement deviations. However, it is reasonable to believe that
DYNAMIC KNEE LOADING ASYMMETRIES 71
individuals following ACLr are not exposed to more dynamic tasks prior to the initiation of
running as typical single limb exercises introduced in early rehabilitation, step ups and single limb
squats, are performed at slower speeds. This lack of experience may underlie the impairments
specifically related to the ability to load the knee rapidly. The persistence of these deficits over
time may be due to the lack of appropriate dynamic training.
From these data, we are unable to determine what factors underlie impairments in loading
rate; muscle function, strength or rate of force development, and fear or reluctance to load quickly
were not assessed in these individuals. In addition, it is not known if impairments in loading rate
would be observed during tasks with magnitude of loading demands smaller than the single limb
loading task, or if practice and continued exposure to the experimental tasks may have improved
loading rate. However, the correlation between deficits observed during running and during the
single limb loading task suggests that there is a carryover between tasks. The single limb loading
task used in this study was designed to mimic the sagittal plane loading mechanics utilized during
deceleration phase of running. Perhaps incorporating this task or similar exercises, for which the
magnitude of loading is lower relative to running, may allow for more focused attention on the
speed of movement during rehabilitation. This type of practice may not only increase individuals’
confidence in loading their knee, but also lead improvements in the rate of force attenuation in
their knee extensors.
Another factor that could contribute to the persistence of loading impairments is the
difficulty detecting or quantifying deficits in knee power during dynamic tasks. Impairments in
knee loading are challenging to identify clinically as they occur during dynamic tasks performed
quickly, and coincide with subtle changes in joint kinematics (Orishimo et al., 2010; Salem et al.,
DYNAMIC KNEE LOADING ASYMMETRIES 72
2003). Visual observation and video recording applications commonly used in the clinic are
limited to quantification of kinematics during tasks performed at slower speeds. The expense,
time, and expertise needed to quantify knee power deficits using the current gold standard motion
analysis techniques preclude their use in the clinic. Chapter IV aimed to determine if inertial
sensors could provide a feasible alternative for clinical detection of loading impairments. Despite
improvements in wireless capabilities, larger data storage and higher capture rates, inertial
sensors are still limited to quantification of segment kinematics without important information
regarding ground reaction forces. Previous studies have used inertial sensors on the thigh and
shank segments to calculate knee joint angles (Dowling et al., 2011; Patterson et al., 2014);
however, errors associated with these calculations limit their application in this population. On
the other hand, the potential use of inertial sensor gyroscopes, to identify power deficits in the
clinic exists given that segment angular velocities (direct output of gyroscope)inertial sensors)
and resulting joint angular velocities are used to calculate joint power.
Using the single limb loading task as described in Chapter III, in Chapter IV a relationship
between segment angular velocities and knee power measured with marker −based motion
analysis system was determined in individuals following ACLr who recently progressed to
running. These relationships quantified in Chapter IV suggest that segment kinematics can
provide meaningful information regarding loading without force platform data. For clinical
translation, concurrent validity of using inertial sensors to measure segment kinematics was
assessed and the relationship between knee power and angular velocities measured with inertial
sensors was confirmed. With strong agreement between measurements systems for thigh and
shank angular velocities it was not surprising that thigh and shank angular velocities measured
with the inertial sensors were also related to knee power measured with marker −based motion
DYNAMIC KNEE LOADING ASYMMETRIES 73
capture. In fact, thigh angular velocity was found to be the strongest predictor of knee power
during this single limb loading task, explaining 66% of the variance in knee power. Limb did not
influence this relationship; therefore supporting the use of these testing procedures for
assessment of the surgical and non −surgical knees despite the presence of between limb
differences in both variables. This relationship was stronger than previously explored
relationships between coronal segment angular velocities and frontal plane knee moments
(Dowling et al., 2012a); and, strong enough to support the use of sensors placed on the thigh
segments for predicting knee power in individuals following ACLr during this single limb loading
task. The results presented in Chapter IV set the foundation for clinical identification of dynamic
knee loading deficits.
In order to translate findings from Chapter III and IV into the clinic, Chapter V focused
on quantifying the diagnostic accuracy of inertial sensor thigh angular velocity measurements in
identifying asymmetrical dynamic knee loading (knee power) during a single limb loading task in
this same clinical population. Data from Chapters III and IV supported the use of the non −surgical
limb as a reference. Therefore, diagnostic accuracy was determined using a typical clinical
comparison, between limb ratios. Generally speaking, reaching a ratio of 85 −90% is required for
progression to more challenging activities (Kvist, 2004; Myer, G.D. et al., 2011; Thomee et al., 2011).
Therefore, a between limb knee power ratio less than 0.85 was classified as asymmetrical knee
loading. Receiver operating characteristic curve analysis revealed that a thigh angular velocity
ratio less than 0.81 can diagnose asymmetrical knee loading with high sensitivity and specificity.
Furthermore, high positive and low negative likelihood ratios and post −test probability confirmed
the discriminative accuracy of inertial sensor thigh angular velocity in diagnosing asymmetrical
knee power in this population.
DYNAMIC KNEE LOADING ASYMMETRIES 74
The implications of these findings in Chapter V are exciting as they define procedures for
examining dynamic knee loading in the clinic using inertial sensors. Establishing the relationship
between segment angular velocities measured with inertial sensors and knee power derived from
marker −based motion analysis procedures is important; but it does not directly translate into
clinical use (Chapter IV). Chapter V described procedures that can be reproduced in the clinic
using technology that is more cost −effective and easier to implement than the gold standard
techniques used to quantify knee power. With the established likelihood ratios, clinicians can
use these procedures with relative confidence that they are receiving accurate information
regarding their patients’ knee loading capabilities. The need for this type of information is
highlighted by the inability of current functional testing to identify joint level impairments
(Barber-Westin & Noyes, 2011; Orishimo et al., 2010). Interpretation of current functional
assessments including distance hopped or time to task completion is limited to overall limb
function. Furthermore, completion of such tasks can be accomplished with compensatory
patterns that increase the demands on the hip and ankle to accomplish the overall goal (Orishimo
et al., 2010; Salem et al., 2003). The current testing procedure provides important information for
clinical decision −making specific to the knee during a functional single limb loading task.
This dissertation highlighted impairments in the ability to rapidly accommodate forces at
the knee at time individuals following ACLr are progressing to running. However, future work
can investigate the extent to which dynamic loading impairments exist and progress across
rehabilitation; as well as the effects of specific training on amelioration of these deficits. A better
understanding of the factors that underlie these deficits will inform the development of
appropriate interventions. This dissertation also provided a strong foundation for the use of
inertial sensors to objectively quantify movement in the absence of force platforms and expensive
DYNAMIC KNEE LOADING ASYMMETRIES 75
motion capture systems in this population. Expansion of these studies could have significant
impact on clinical decision −making and rehabilitation of individuals with knee injuries. Future
studies including individuals at other time points post −surgery will allow for more expanded
clinical use. Furthermore, work can be done to adapt these procedures to other functional tests
used in clinical decision −making. Finally, the potential exists to use the inertial sensor
measurements established in this dissertation for knee loading feedback during rehabilitation
exercises in the clinic or at home. Providing this movement feedback could lead to improvements
in individuals’ ability to rapidly accommodate forces during dynamic tasks following surgery.
DYNAMIC KNEE LOADING ASYMMETRIES 76
BIBLIOGRAPHY
Abrams, G. D., Harris, J. D., Gupta, A. K., McCormick, F. M., Bush-Joseph, C. A., Verma, N. N., . .
. Bach, B. R., Jr. (2014). Functional Performance Testing After Anterior Cruciate
Ligament Reconstruction: A Systematic Review. Orthop J Sports Med, 2(1),
2325967113518305. doi:10.1177/2325967113518305
Adams, D., Logerstedt, D. S., Hunter-Giordano, A., Axe, M. J., & Snyder-Mackler, L. (2012).
Current concepts for anterior cruciate ligament reconstruction: a criterion-based
rehabilitation progression. J Orthop Sports Phys Ther, 42(7), 601-614.
doi:10.2519/jospt.2012.3871
Bachlin, M., Plotnik, M., Roggen, D., Giladi, N., Hausdorff, J. M., & Troster, G. (2010). A
wearable system to assist walking of Parkinson s disease patients. Methods Inf Med, 49(1),
88-95. doi:10.3414/ME09-02-0003
Barber-Westin, S. D., & Noyes, F. R. (2011). Factors used to determine return to unrestricted
sports activities after anterior cruciate ligament reconstruction. Arthroscopy, 27(12), 1697-
1705. doi:10.1016/j.arthro.2011.09.009
Barber-Westin, S. D., Noyes, F. R., & McCloskey, J. W. (1999). Rigorous statistical reliability,
validity, and responsiveness testing of the Cincinnati knee rating system in 350 subjects
with uninjured, injured, or anterior cruciate ligament-reconstructed knees. Am J Sports
Med, 27(4), 402-416.
Barrios, J. A., Crossley, K. M., & Davis, I. S. (2010). Gait retraining to reduce the knee adduction
moment through real-time visual feedback of dynamic knee alignment. J Biomech, 43(11),
2208-2213. doi:10.1016/j.jbiomech.2010.03.040
Baumgart, C., Schubert, M., Hoppe, M. W., Gokeler, A., & Freiwald, J. (2015). Do ground
reaction forces during unilateral and bilateral movements exhibit compensation
strategies following ACL reconstruction? Knee Surg Sports Traumatol Arthrosc.
doi:10.1007/s00167-015-3623-7
Bregou Bourgeois, A., Mariani, B., Aminian, K., Zambelli, P. Y., & Newman, C. J. (2014). Spatio-
temporal gait analysis in children with cerebral palsy using, foot-worn inertial sensors.
Gait Posture, 39(1), 436-442. doi:10.1016/j.gaitpost.2013.08.029
Buczek, F. L., & Cavanagh, P. R. (1990). Stance phase knee and ankle kinematics and kinetics
during level and downhill running. Med Sci Sports Exerc, 22(5), 669-677.
Buller, L. T., Best, M. J., Baraga, M. G., & Kaplan, L. D. (2015). Trends in Anterior Cruciate
Ligament Reconstruction in the United States. Orthop J Sports Med, 3(1),
2325967114563664. doi:10.1177/2325967114563664
Cascio, B. M., Culp, L., & Cosgarea, A. J. (2004). Return to play after anterior cruciate ligament
reconstruction. Clin Sports Med, 23(3), 395-408, ix. doi:10.1016/j.csm.2004.03.004
Chmielewski, T. L. (2011). Asymmetrical lower extremity loading after ACL reconstruction:
more than meets the eye. J Orthop Sports Phys Ther, 41(6), 374-376.
doi:10.2519/jospt.2011.0104
DYNAMIC KNEE LOADING ASYMMETRIES 77
Chmielewski, T. L., Jones, D., Day, T., Tillman, S. M., Lentz, T. A., & George, S. Z. (2008). The
association of pain and fear of movement/reinjury with function during anterior cruciate
ligament reconstruction rehabilitation. J Orthop Sports Phys Ther, 38(12), 746-753.
doi:10.2519/jospt.2008.2887
Crowell, H. P., & Davis, I. S. (2011). Gait retraining to reduce lower extremity loading in
runners. Clin Biomech, 26(1), 78-83. doi:10.1016/j.clinbiomech.2010.09.003
Dai, B., Butler, R. J., Garrett, W. E., & Queen, R. M. (2012). Anterior cruciate ligament
reconstruction in adolescent patients: limb asymmetry and functional knee bracing. Am J
Sports Med, 40(12), 2756-2763. doi:10.1177/0363546512460837
Decker, M. J., Torry, M. R., Noonan, T. J., Riviere, A., & Sterett, W. I. (2002). Landing
adaptations after ACL reconstruction. Med Sci Sports Exerc, 34(9), 1408-1413.
doi:10.1249/01.MSS.0000027627.82650.1F
Decker, M. J., Torry, M. R., Wyland, D. J., Sterett, W. I., & Richard Steadman, J. (2003). Gender
differences in lower extremity kinematics, kinetics and energy absorption during
landing. Clin Biomech, 18(7), 662-669. doi:Doi 10.1016/S0268-0033(03)00090-1
Dejnabadi, H., Jolles, B. M., Casanova, E., Fua, P., & Aminian, K. (2006). Estimation and
visualization of sagittal kinematics of lower limbs orientation using body-fixed sensors.
IEEE Trans Biomed Eng, 53(7), 1385-1393. doi:10.1109/TBME.2006.873678
Delahunt, E., Sweeney, L., Chawke, M., Kelleher, J., Murphy, K., Patterson, M., & Prendiville, A.
(2012). Lower limb kinematic alterations during drop vertical jumps in female athletes
who have undergone anterior cruciate ligament reconstruction. Journal of Orthopaedic
Research, 30(1), 72-78. doi:10.1002/jor.21504
Devita, P., Hortobagyi, T., & Barrier, J. (1998). Gait biomechanics are not normal after anterior
cruciate ligament reconstruction and accelerated rehabilitation. Med Sci Sport Exer, 30(10),
1481-1488.
Devita, P., Hunter, P. B., & Skelly, W. A. (1992). Effects of a functional knee brace on the
biomechanics of running. Med Sci Sports Exerc, 24(7), 797-806.
Di Stasi, S., Myer, G. D., & Hewett, T. E. (2013a). Neuromuscular training to target deficits
associated with second anterior cruciate ligament injury. J Orthop Sports Phys Ther, 43(11),
777-792, A771-711. doi:10.2519/jospt.2013.4693
Di Stasi, S. L., Logerstedt, D., Gardinier, E. S., & Snyder-Mackler, L. (2013b). Gait patterns differ
between ACL-reconstructed athletes who pass return-to-sport criteria and those who
fail. Am J Sport Med, 41(6), 1310-1318.
Dowling, A. V., Favre, J., & Andriacchi, T. P. (2011). A wearable system to assess risk for anterior
cruciate ligament injury during jump landing: measurements of temporal events, jump
height, and sagittal plane kinematics. J Biomech Eng, 133(7), 071008. doi:10.1115/1.4004413
Dowling, A. V., Favre, J., & Andriacchi, T. P. (2012a). Characterization of thigh and shank
segment angular velocity during jump landing tasks commonly used to evaluate risk for
ACL injury. J Biomech Eng, 134(9), 091006. doi:10.1115/1.4007178
DYNAMIC KNEE LOADING ASYMMETRIES 78
Dowling, A. V., Favre, J., & Andriacchi, T. P. (2012b). Inertial sensor-based feedback can reduce
key risk metrics for anterior cruciate ligament injury during jump landings. Am J Sports
Med, 40(5), 1075-1083. doi:10.1177/0363546512437529
Dowling, A. V., Fisher, D. S., & Andriacchi, T. P. (2010). Gait modification via verbal instruction
and an active feedback system to reduce peak knee adduction moment. J Biomech Eng,
132(7), 071007. doi:10.1115/1.4001584
Ernst, G. P., Saliba, E., Diduch, D. R., Hurwitz, S. R., & Ball, D. W. (2000). Lower extremity
compensations following anterior cruciate ligament reconstruction. Phys Ther, 80(3), 251-
260.
Favre, J., Jolles, B. M., Aissaoui, R., & Aminian, K. (2008). Ambulatory measurement of 3D knee
joint angle. J Biomech, 41(5), 1029-1035. doi:10.1016/j.jbiomech.2007.12.003
Fetters, L., & Tilson, J. (2012). Evidence Based Physical Therapy. United States of America: F. A.
Davis Company.
Fong, D. T., & Chan, Y. Y. (2010). The use of wearable inertial motion sensors in human lower
limb biomechanics studies: a systematic review. Sensors (Basel), 10(12), 11556-11565.
doi:10.3390/s101211556
Gokeler, A., Hof, A. L., Arnold, M. P., Dijkstra, P. U., Postema, K., & Otten, E. (2010). Abnormal
landing strategies after ACL reconstruction. Scand J Med Sci Sports, 20(1), e12-19.
doi:10.1111/j.1600-0838.2008.00873.x
Greene, B. R., McGrath, D., O'Donovan, K. J., O'Neill, R., Burns, A., & Caulfield, B. (2010).
Adaptive estimation of temporal gait parameters using body-worn gyroscopes. Conf Proc
IEEE Eng Med Biol Soc, 2010, 1296-1299. doi:10.1109/IEMBS.2010.5626400
Griffin, L. Y., Albohm, M. J., Arendt, E. A., Bahr, R., Beynnon, B. D., Demaio, M., . . . Yu, B. (2006).
Understanding and preventing noncontact anterior cruciate ligament injuries: a review
of the Hunt Valley II meeting, January 2005. Am J Sports Med, 34(9), 1512-1532.
doi:10.1177/0363546506286866
Harty, C. M., DuPont, C. E., Chmielewski, T. L., & Mizner, R. L. (2011). Intertask comparison of
frontal plane knee position and moment in female athletes during three distinct
movement tasks. Scand J Med Sci Sports, 21(1), 98-105. doi:10.1111/j.1600-0838.2009.01022.x
Kuenze, C., Hertel, J., Weltman, A., Diduch, D. R., Saliba, S., & Hart, J. M. (2014). Jogging
biomechanics after exercise in individuals with ACL-reconstructed knees. Med Sci Sports
Exerc, 46(6), 1067-1076. doi:10.1249/MSS.0000000000000217
Kvist, J. (2004). Rehabilitation following anterior cruciate ligament injury: current
recommendations for sports participation. Sports Med, 34(4), 269-280.
Lewek, M., Rudolph, K., Axe, M., & Snyder-Mackler, L. (2002). The effect of insufficient
quadriceps strength on gait after anterior cruciate ligament reconstruction. Clin Biomech,
17(1), 56-63.
DYNAMIC KNEE LOADING ASYMMETRIES 79
Logerstedt, D. S., Snyder-Mackler, L., Ritter, R. C., Axe, M. J., & Godges, J. J. (2010). Knee
stability and movement coordination impairments: knee ligament sprain. J Orthop Sports
Phys Ther, 40(4), A1-A37. doi:10.2519/jospt.2010.0303
Lohmander, L. S., Englund, P. M., Dahl, L. L., & Roos, E. M. (2007). The long-term consequence
of anterior cruciate ligament and meniscus injuries: osteoarthritis. Am J Sports Med, 35(10),
1756-1769. doi:10.1177/0363546507307396
Moolyk, A. N., Carey, J. P., & Chiu, L. Z. (2013). Characteristics of lower extremity work during
the impact phase of jumping and weightlifting. Journal of Strength and Conditioning Research,
27(12), 3225-3232. doi:10.1519/JSC.0b013e31828ddf19
Myer, G. D., Ford, K. R., & Hewettt, T. E. (2004). Rationale and clinical techniques for anterior
cruciate ligament injury prevention among female athletes. J Athl Train, 39(4), 352-364.
Myer, G. D., Paterno, M., Ford, K., Quatman, C., & Hewett, T. (2006). Rehabilitation after
anterior cruciate ligament reconstruction: criteria-based progression through the return-
to-sport phase. J Orthop Sport Phys, 36, 385 - 402.
Myer, G. D., Schmitt, L. C., Brent, J. L., Ford, K. R., Barber Foss, K. D., Scherer, B. J., . . . Hewett,
T. E. (2011). Utilization of modified NFL combine testing to identify functional deficits
in athletes following ACL reconstruction. J Orthop Sports Phys Ther, 41(6), 377-387.
doi:10.2519/jospt.2011.3547
Myklebust, G., & Bahr, R. (2005). Return to play guidelines after anterior cruciate ligament
surgery. Br J Sports Med, 39(3), 127-131. doi:10.1136/bjsm.2004.010900
Nilsson, J., & Thorstensson, A. (1989). Ground reaction forces at different speeds of human
walking and running. Acta Physiol Scand, 136(2), 217-227. doi:10.1111/j.1748-
1716.1989.tb08655.x
Noehren, B., Wilson, H., Miller, C., & Lattermann, C. (2013). Long-term gait deviations in
anterior cruciate ligament-reconstructed females. Med Sci Sports Exerc, 45(7), 1340-1347.
doi:10.1249/MSS.0b013e318285c6b6
Oberlander, K. D., Bruggemann, G. P., Hoher, J., & Karamanidis, K. (2013). Altered landing
mechanics in ACL-reconstructed patients. Med Sci Sports Exerc, 45(3), 506-513.
doi:10.1249/MSS.0b013e3182752ae3
Oberlander, K. D., Bruggemann, G. P., Hoher, J., & Karamanidis, K. (2014). Knee mechanics
during landing in anterior cruciate ligament patients: A longitudinal study from pre- to
12 months post-reconstruction. Clin Biomech, 29(5), 512-517.
doi:10.1016/j.clinbiomech.2014.03.014
Orishimo, K. F., Kremenic, I. J., Mullaney, M. J., McHugh, M. P., & Nicholas, S. J. (2010).
Adaptations in single-leg hop biomechanics following anterior cruciate ligament
reconstruction. Knee Surg Sports Traumatol Arthrosc, 18(11), 1587-1593. doi:10.1007/s00167-
010-1185-2
Ortiz, A., Olson, S., Trudelle-Jackson, E., Rosario, M., & Venegas, H. L. (2011). Landing
mechanics during side hopping and crossover hopping maneuvers in noninjured women
DYNAMIC KNEE LOADING ASYMMETRIES 80
and women with anterior cruciate ligament reconstruction. PM R, 3(1), 13-20.
doi:10.1016/j.pmrj.2010.10.018
Paterno, M. V., Schmitt, L. C., Ford, K. R., Rauh, M. J., Myer, G. D., Huang, B., & Hewett, T. E.
(2010). Biomechanical measures during landing and postural stability predict second
anterior cruciate ligament injury after anterior cruciate ligament reconstruction and
return to sport. Am J Sports Med, 38(10), 1968-1978. doi:10.1177/0363546510376053
Patterson, M. R., Delahunt, E., Sweeney, K. T., & Caulfield, B. (2014). An ambulatory method of
identifying anterior cruciate ligament reconstructed gait patterns. Sensors (Basel), 14(1),
887-899. doi:10.3390/s140100887
Roewer, B. D., Di Stasi, S. L., & Snyder-Mackler, L. (2011). Quadriceps strength and weight
acceptance strategies continue to improve two years after anterior cruciate ligament
reconstruction. J Biomech, 44(10), 1948-1953. doi:10.1016/j.jbiomech.2011.04.037
Salem, G. J., Salinas, R., & Harding, F. V. (2003). Bilateral kinematic and kinetic analysis of the
squat exercise after anterior cruciate ligament reconstruction. Arch Phys Med Rehabil, 84(8),
1211-1216. doi:10.1016/s0003-9993(03)00034-0
Schache, A. G., Brown, N. A., & Pandy, M. G. (2015). Modulation of work and power by the
human lower-limb joints with increasing steady-state locomotion speed. J Exp Biol, 218(Pt
15), 2472-2481. doi:10.1242/jeb.119156
Sigward, S. M., Pollard, C. D., & Powers, C. M. (2012). The influence of sex and maturation on
landing biomechanics: implications for anterior cruciate ligament injury. Scand J Med Sci
Sports, 22(4), 502-509. doi:10.1111/j.1600-0838.2010.01254.x
Stergiou, N., Ristanis, S., Moraiti, C., & Georgoulis, A. D. (2007). Tibial rotation in anterior
cruciate ligament (ACL)-deficient and ACL-reconstructed knees: a theoretical
proposition for the development of osteoarthritis. Sports Med, 37(7), 601-613. doi:Doi
10.2165/00007256-200737070-00004
Tate, J. J., & Milner, C. E. (2010). Real-time kinematic, temporospatial, and kinetic biofeedback
during gait retraining in patients: a systematic review. Phys Ther, 90(8), 1123-1134.
doi:10.2522/ptj.20080281
Thomee, R., Kaplan, Y., Kvist, J., Myklebust, G., Risberg, M. A., Theisen, D., . . . Witvrouw, E.
(2011). Muscle strength and hop performance criteria prior to return to sports after ACL
reconstruction. Knee Surg Sports Traumatol Arthrosc, 19(11), 1798-1805. doi:10.1007/s00167-
011-1669-8
van Grinsven, S., van Cingel, R. E., Holla, C. J., & van Loon, C. J. (2010). Evidence-based
rehabilitation following anterior cruciate ligament reconstruction. Knee Surg Sports
Traumatol Arthrosc, 18(8), 1128-1144. doi:10.1007/s00167-009-1027-2
Webster, K. E., Feller, J. A., & Wittwer, J. E. (2012). Longitudinal changes in knee joint
biomechanics during level walking following anterior cruciate ligament reconstruction
surgery. Gait Posture, 36(2), 167-171. doi:10.1016/j.gaitpost.2012.02.004
DYNAMIC KNEE LOADING ASYMMETRIES 81
Webster, K. E., Gonzalez-Adrio, R., & Feller, J. A. (2004). Dynamic joint loading following
hamstring and patellar tendon anterior cruciate ligament reconstruction. Knee Surg Sports
Traumatol Arthrosc, 12(1), 15-21. doi:10.1007/s00167-003-0400-9
Weinhandl, J. T. M. J., M.; O’Connor, K. M. (2010). Gender Comparisons Between Unilateral
and Bilateral Landings. J Appl Biomech, 26, 444-453.
White, K., Logerstedt, D., & Snyder-Mackler, L. (2013). Gait Asymmetries Persist 1 Year After
Anterior Cruciate Ligament Reconstruction. Orthop J Sports Med, 1(2), 2325967113496967.
doi:10.1177/2325967113496967
Willson, J. D., & Davis, I. S. (2008). Lower extremity mechanics of females with and without
patellofemoral pain across activities with progressively greater task demands. Clin
Biomech, 23(2), 203-211. doi:10.1016/j.clinbiomech.2007.08.025
Winter, D. A. (2009). Biomechanics and motor control of human movement (4th ed.). Hoboken, N.J.:
Wiley.
Wright, R. W., Magnussen, R. A., Dunn, W. R., & Spindler, K. P. (2011). Ipsilateral graft and
contralateral ACL rupture at five years or more following ACL reconstruction: a
systematic review. J Bone Joint Surg Am, 93(12), 1159-1165. doi:10.2106/JBJS.J.00898
Xergia, S. A., Pappas, E., Zampeli, F., Georgiou, S., & Georgoulis, A. D. (2013). Asymmetries in
functional hop tests, lower extremity kinematics, and isokinetic strength persist 6 to 9
months following anterior cruciate ligament reconstruction. J Orthop Sports Phys Ther,
43(3), 154-162. doi:10.2519/jospt.2013.3967
Yang, S., Mohr, C., & Li, Q. (2011). Ambulatory running speed estimation using an inertial
sensor. Gait Posture, 34(4), 462-466. doi:10.1016/j.gaitpost.2011.06.019
Yang, S., Zhang, J. T., Novak, A. C., Brouwer, B., & Li, Q. (2013). Estimation of spatio-temporal
parameters for post-stroke hemiparetic gait using inertial sensors. Gait Posture, 37(3), 354-
358. doi:10.1016/j.gaitpost.2012.07.032
Yu, B., Lin, C. F., & Garrett, W. E. (2006). Lower extremity biomechanics during the landing of
a stop-jump task. Clin Biomech, 21(3), 297-305. doi:10.1016/j.clinbiomech.2005.11.003
Zhang, S. N., Bates, B. T., & Dufek, J. S. (2000). Contributions of lower extremity joints to
energy dissipation during landings. Med Sci Sports Exerc, 32(4), 812-819.
Abstract (if available)
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Gait flexibility following anterior cruciate ligament reconstruction: implications for early gait retraining
PDF
Altered gait mechanics reduce sagittal plane knee loading following ACL reconstruction
PDF
Limb loading asymmetry in individuals following anterior cruciate ligament reconstruction: an indication of nonuse phenomenon
PDF
Whole body mechanics of running turn maneuvers: relationship to injury and performance
PDF
Biomechanical analysis of knee joint loading in females who have undergone anterior cruciate ligament reconstruction
PDF
Development of postural strategies for the control of forward momentum
PDF
Quantification of lower extremity dynamic capability: implications for anterior cruciate ligament injury and change of direction ability
PDF
Development of a movement performance assessment to predict ACL re-injury
PDF
The biomechanical changes associated with the transition from shod to barefoot running
PDF
The influence of hip muscle performance on postural stability & ankle biomechanics: implications for ankle injury
PDF
Biomechanical determinants of patellofemoral joint forces
PDF
Influence of sagittal plane trunk posture on lower extremity biomechanics during running
PDF
The influence of tibiofemoral kinematics and knee extensor mechanics on patellar tendon stress: a comparison of persons with and without patellar tendinopathy
PDF
The structure of kinematic variability during hopping: a window into motor control factors in overuse injury
PDF
The neuromechanical adaptations to Achilles tendinosis
PDF
The influence of hip muscle performance on lower extremity biomechanics and neuromuscular activation: implications for anterior cruciate ligament injury
PDF
Trunk neuromechanics during turning: a window into recurrent low back pain
PDF
Trunk control during dynamic balance: effects of cognitive dual-task interference and a history of recurrent low back pain
PDF
The influence of patellofemoral joint loading on patella strain and patella water content in females with patellofemoral pain
PDF
Control and dynamics of turning tasks with different rotation and translation requirements
Asset Metadata
Creator
Pratt, Kristamarie Anne
(author)
Core Title
Dynamic knee loading asymmetries following anterior cruciate ligament reconstruction: methods for clinical detection
School
School of Dentistry
Degree
Doctor of Philosophy
Degree Program
Biokinesiology and Physical Therapy
Publication Date
02/17/2016
Defense Date
12/08/2015
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
ACL injury,ACL rehabilitation,biomechanics,knee,OAI-PMH Harvest,Rehabilitation,Running,wearable sensors
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Sigward, Susan M. (
committee chair
), Gregor, Robert (
committee member
), Keim, Robert (
committee member
), Kulig, Kornelia (
committee member
), Salem, George J. (
committee member
)
Creator Email
kristamarie.pratt@gmail.com,kristamp@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c40-208563
Unique identifier
UC11278353
Identifier
etd-PrattKrist-4111.pdf (filename),usctheses-c40-208563 (legacy record id)
Legacy Identifier
etd-PrattKrist-4111.pdf
Dmrecord
208563
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Pratt, Kristamarie Anne
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
ACL injury
ACL rehabilitation
biomechanics
wearable sensors