Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Are inversion, posture, motion and muscle effects important to spinal alignment? Newell, Robyn 2014

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2014_spring_newell_robyn.pdf [ 10.93MB ]
Metadata
JSON: 24-1.0167424.json
JSON-LD: 24-1.0167424-ld.json
RDF/XML (Pretty): 24-1.0167424-rdf.xml
RDF/JSON: 24-1.0167424-rdf.json
Turtle: 24-1.0167424-turtle.txt
N-Triples: 24-1.0167424-rdf-ntriples.txt
Original Record: 24-1.0167424-source.json
Full Text
24-1.0167424-fulltext.txt
Citation
24-1.0167424.ris

Full Text

ARE INVERSION, POSTURE, MOTION AND MUSCLE EFFECTS IMPORTANT TO SPINAL ALIGNMENT?   by Robyn Newell  M.A.Sc., Dalhousie University, 2007 B.Eng., Dalhousie University, 2004  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Biomedical Engineering)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  April 2014  © Robyn Newell, 2014 ii  Abstract Rollover accidents are dynamic and complex events in which head contacts with the vehicle interior can cause catastrophic neck injuries through head-first impact.  Ex vivo cadaver tests are valuable for studying these mechanisms of head-first axial loading neck injuries; however, they lack a biofidelic representation of neuromuscular control, postural stability, and overall spine posture.  Computational modeling can be used to evaluate changes in the risk of neck injury under the influence of muscle forces, yet the exact muscles and levels of forces that are involved leading up to a head-first impact are unknown.  Knowing the state of the neck prior to impact is critical to improving cadaveric and computational models of neck injury.      Four human volunteer experiments were conducted to determine whether inversion, head position, muscle tensing, and dynamic motion influence the cervical spine alignment.  These four studies included: (1) static inversion, (2) muscle tensing, (3) moment generation, (4) dynamic flexion/extension.  For each experiment, cervical alignment was captured using fluoroscopy and muscle activity was captured using electromyography.    The inverted posture and muscle activations were found to be different than the upright relaxed posture and the differences depend on the position of the head (study 1).  Actively tensing the neck muscles in a free unconstrained task (study 2) and in generating flexion and extension forces with head constraint (study 3) resulted in different cervical alignment compared to the initial resting spine.  Not only do these neck muscle contractions induce postural changes, they also provide a substantial stiffening effect to the neck.  Finally, dynamically arriving at the iii  neutral position did not result in the same cervical alignment as static neutral and the alignment depended on the direction that neutral is approached from (full flexion or full extension).    These findings suggest that it may not be sufficient to replicate the upright resting posture in cadaveric and computational models of neck injury.  Adopting in vivo postures and muscle activations, relevant to head-first impact, in the laboratory may help in replicating the spectrum of injuries observed in real life rollovers, an important step toward injury prevention.     iv  Preface This project was funded by NSERC-MITACS and was conducted in collaboration with MEA Forensics Engineers & Scientists in Richmond, BC.  Mircea Oala-Florescu and Jeff Nickel were a great deal of help for many aspects of the entire project: helping with designing and fabricating the inversion device, setting up and troubleshooting data acquisition equipment, machining, and procuring materials.  A version of Chapter 2 has been published as: Newell RS, Blouin J-S, Street J, Cripton PA, Siegmund GP. Neck posture and muscle activity are different when upside down: A human volunteer study. Journal of Biomechanics 2013;46(16):2837-43.  A version of Chapter 3 has been accepted for publication as: Newell RS, Siegmund, GP, Blouin J-S, Street J, Cripton PA. Cervical vertebral realignment when voluntarily adopting a protective neck posture. Spine 2014 (in press).  A version of Chapter 4 is being prepared for submission.  Newell RS, Siegmund GP, Blouin JS, Street J, Cripton PA.  Realignment of the cervical vertebrae during neck muscle contractions.  A version of Chapter 5 is being prepared for submission.  Newell RS, Blouin J-S, Street J, Cripton PA, Siegmund GP. Dynamic motion of the cervical spine affects the alignment of the neck in the neutral position.  v  R.S. Newell was jointly responsible for the original ideas behind each paper, designing the inversion device, procuring and validating the fluoroscopic C-arm for the experiment, implementing the data acquisition set-up, designing and conducting the experiments, data analysis, presentation of the findings, and is the primary author for each journal submission.  Dr. Peter Cripton and Dr. Gunter Siegmund were jointly responsible for the original ideas, providing supervision and project support, and editing each journal submission.  Dr. Jean-Sébastien Blouin was jointly responsible for providing expertise for the experiments, providing data acquisition equipment and an ultrasound unit for the experiments, assistance with ethics approval, and editing each article.  Dr. Gunter Siegmund and Dr. Jean-Sébastien Blouin were also responsible for inserting the electromyography wires during the experiments.  Dr. John Street was jointly responsible for providing a clinical perspective, expertise for each experiment, and editing each article.    The research conducted for chapters 2-5 of this thesis was approved by the University of British Columbia’s Clinical Research Ethics Board (certificate #H07-01445) and all subjects gave their written informed consent.   vi   Table of Contents  Abstract .......................................................................................................................................... ii Preface ........................................................................................................................................... iv Table of Contents ......................................................................................................................... vi List of Tables ................................................................................................................................ xi List of Figures ............................................................................................................................. xiii List of Symbols .......................................................................................................................... xvii List of Abbreviations ............................................................................................................... xviii Acknowledgements ......................................................................................................................xx Chapter  1: Introduction ...............................................................................................................1 1.1  Overview ......................................................................................................................... 1 1.2  Cervical spine injury ....................................................................................................... 2 1.2.1  Cervical spine injury in rollover accidents ................................................................. 3 1.2.2  Cervical spine injury mechanisms .............................................................................. 4 1.2.3  Cervical spine injury models ...................................................................................... 7 1.2.3.1  Cadaver neck injury models ................................................................................ 8 1.2.3.2  Computational modeling ................................................................................... 11 1.2.3.3  Anthropomorphic test devices and human subjects .......................................... 14 1.2.4  Rollover injury prevention ........................................................................................ 16 1.3  Factors that are important to cervical spine injury ........................................................ 18 1.3.1  Mechanical factors that influence injury................................................................... 19 vii  1.3.1.1  Neck eccentricity .............................................................................................. 20 1.3.1.2  Neck curvature .................................................................................................. 22 1.3.1.3  Head alignment ................................................................................................. 24 1.3.1.4  Muscles ............................................................................................................. 25 1.3.2  Limitations to simulating cervical alignment and muscle forces ex vivo ................. 32 1.4  Measuring in vivo responses ......................................................................................... 34 1.4.1  In vivo neck kinematics ............................................................................................ 35 1.4.1.1  Kinematic parameters ....................................................................................... 37 1.4.2  Electromyography ..................................................................................................... 38 1.4.2.1  Muscle selection................................................................................................ 41 1.5  Research objectives ....................................................................................................... 44 Chapter  2: Does inversion and head orientation affect the relaxed neutral posture of the cervical spine?  .............................................................................................................................48 2.1  Introduction ................................................................................................................... 48 2.2  Methods......................................................................................................................... 50 2.2.1  Subjects ..................................................................................................................... 50 2.2.2  Conditions ................................................................................................................. 50 2.2.3  Cervical spine posture ............................................................................................... 51 2.2.4  Head orientation ........................................................................................................ 54 2.2.5  Electromyography ..................................................................................................... 56 2.2.6  Maximum voluntary contractions ............................................................................. 56 2.2.7  Data processing notes ............................................................................................... 59 2.2.8  Data and statistical analysis ...................................................................................... 59 viii  2.3  Results ........................................................................................................................... 59 2.3.1  Head and neck posture .............................................................................................. 59 2.3.2  Muscle activity .......................................................................................................... 63 2.4  Discussion ..................................................................................................................... 64 Chapter  3: Does bracing for impact result in a realignment of the cervical spine?  ............69 3.1  Introduction ................................................................................................................... 69 3.2  Methods......................................................................................................................... 70 3.2.1  Subjects ..................................................................................................................... 70 3.2.2  Electromyography ..................................................................................................... 73 3.2.3  Cervical spine alignment........................................................................................... 74 3.2.4  Data processing notes and statistical analysis ........................................................... 75 3.3  Results ........................................................................................................................... 76 3.3.1  Muscle activity .......................................................................................................... 76 3.3.2  Kinematic results ...................................................................................................... 78 3.4  Discussion ..................................................................................................................... 82 Chapter  4: Does exerting a flexion and extension force (with head constraint) alter the alignment of the spine? ................................................................................................................87 4.1  Introduction ................................................................................................................... 87 4.2  Methods......................................................................................................................... 89 4.2.1  Subjects ..................................................................................................................... 89 4.2.2  Conditions ................................................................................................................. 89 4.2.3  Electromyography ..................................................................................................... 91 4.2.4  Maximum voluntary contractions ............................................................................. 94 ix  4.2.5  Cervical spine posture ............................................................................................... 95 4.2.6  Head orientation ........................................................................................................ 97 4.2.7  Data processing notes ............................................................................................... 98 4.2.8  Data and statistical analysis ...................................................................................... 98 4.3  Results ........................................................................................................................... 99 4.3.1  Forces ........................................................................................................................ 99 4.3.2  Muscle activity .......................................................................................................... 99 4.4  Discussion ................................................................................................................... 107 Chapter  5: Does dynamic motion of the cervical spine affect the alignment of the neck in the neutral position? ..................................................................................................................117 5.1  Introduction ................................................................................................................. 117 5.2  Methods....................................................................................................................... 120 5.2.1  Subjects ................................................................................................................... 120 5.2.2  Conditions ............................................................................................................... 121 5.2.3  Cervical spine posture ............................................................................................. 123 5.2.4  Postural measures.................................................................................................... 126 5.2.5  Speed ....................................................................................................................... 127 5.2.6  Electromyography ................................................................................................... 127 5.2.7  Data processing notes ............................................................................................. 128 5.2.8  Data and statistical analysis .................................................................................... 129 5.3  Results ......................................................................................................................... 129 5.3.1  Postural measures.................................................................................................... 129 5.3.2  Muscle activity ........................................................................................................ 135 x  5.4  Discussion ................................................................................................................... 138 Chapter  6: Integrated discussion .............................................................................................145 6.1  Implications................................................................................................................. 151 6.1.1  Implications for ex vivo testing ............................................................................... 151 6.1.2  Implications for computational models of injury .................................................... 157 6.1.3  Implications for prevention and protection ............................................................. 160 6.2  Limitations and recommendations .............................................................................. 165 6.3  Contributions............................................................................................................... 174 6.4  Conclusions ................................................................................................................. 176 References ...................................................................................................................................179 Appendix A: Automatic tracking algorithm ...........................................................................199 Appendix B: Individual subject data for static inversion ......................................................218 Appendix C: Individual subject data for free tensed task .....................................................221 Appendix D: Direct linear transformation using the headband beads .................................223 Appendix E: Individual subject data for constrained tensed task ........................................226 Appendix F: Additional data for dynamic tasks .....................................................................236  xi  List of Tables  Table 1-1: Cervical spine injury classification ............................................................................... 6 Table 1-2: Anatomical attachments of several neck muscles ....................................................... 39 Table 2-1: Kinematic and EMG result .......................................................................................... 60 Table 2-2:  Vertebral translations and angles ............................................................................... 62 Table 3-1:  P-values for EMG metrics .......................................................................................... 77 Table 3-2:  P-values for kinematic metrics ................................................................................... 79 Table 4-1:  Flexion and extension force tasks test matrix ............................................................ 96 Table 4-2:  Statistical p-values for the muscle activity in the flexion and extension force tasks 100 Table 4-3:  Statistical p-values for the curvature and intersegmental angles in the flexion and extension force tasks ................................................................................................................... 103 Table 4-4: Computationally predicted moment generating capability of the neck muscles ....... 109 Table 5-1: Average speeds for C1 and head COM and p-values for the main effect of speed ... 130 Table 5-2: Statistical p-values for the kinematics in the flexion and extension directions ......... 131 Table 5-3: Statistical p-values for the muscle activity in the flexion and extension directions .. 136 Table 6-1: Summary of kinematic findings ................................................................................ 146 Table 6-2: Summary of muscle findings ..................................................................................... 147 Table 6-3: Comparison of upright relaxed group alignment to various cervical spine shapes ... 168 Table A-1: Results of the static motion out-of-plane flexion angle error at C3-C4. .................. 207 Table A-2:  Frankfort plan accuracy results ............................................................................... 214 Table A-3: Summary of all accuracy studies .............................................................................. 215 Table A-4: Summary of desired accuracy and measured errors ................................................. 217 xii  Table B-1: Individual subject vertebral translations ................................................................... 218 Table B-2: Individual subject head and vertebral angles ............................................................ 219 Table B-3: Individual EMG activities and MVC force, moments, and direction ....................... 220 Table E-1: X-Y data of all subjects for upright and inverted flexion force ................................ 230 Table E-2: X-Y data of all subjects for upright and inverted extension force ............................ 231 Table E-3: Angle data of all subjects for upright and inverted flexion force ............................. 232 Table E-4: Angle data of all subjects for upright and inverted extension force ......................... 233 Table E-5: Muscle activation data for upright and inverted flexion force .................................. 234 Table E-6: Muscle activation data for upright and inverted extension force .............................. 235 Table F-1: Individual subject vertebral angles (relative to global horizontal) ............................ 236 Table F-2: Individual subject vertebral angles (relative to Upright-Static) ................................ 237 Table F-3: Individual EMG activities in the dynamic task ......................................................... 238 Table F-4: Mean maximum muscle activity through the phases of dynamic flexion/extension 239  xiii  List of Figures  Figure 1-1: Examples of compression and compression-flexion injury mechanisms .................... 7 Figure 1-2: Examples of ex vivo tests ............................................................................................. 9 Figure 1-3: Example of a neuromuscular model for axial impact loading of the neck ................ 13 Figure 1-4: Varying eccentricity of the spine ............................................................................... 21 Figure 1-5: Relationship between load eccentricity and type of neck injury ............................... 24 Figure 1-6: Follower load and instantaneous axes of spinal motion ............................................ 27 Figure 1-7: Ex vivo tests with various end conditions .................................................................. 29 Figure 1-8: Cervical spine neuromuscular model by Chancey et al. ............................................ 31 Figure 1-9: Realignment of the cervical spine and simulation of muscle tone in an ex vivo axial impact ............................................................................................................................................ 33 Figure 1-10: Cross-sectional MRI and neck muscle identification .............................................. 42 Figure 2-1: Experimental set-up ................................................................................................... 51 Figure 2-2:  Vertebral landmarks .................................................................................................. 53 Figure 2-3:   Postural metrics derived from the fluoroscopic images. .......................................... 55 Figure 2-4:  Exemplar data for EMG processing. ......................................................................... 58 Figure 2-5: Kinematic results........................................................................................................ 61 Figure 2-6: Group average muscle activities ................................................................................ 63 Figure 3-1: Exemplar EMG data. .................................................................................................. 72 Figure 3-2: Exemplar data of vertebral motions. .......................................................................... 73 Figure 3-3:  Initial and tensed group mean EMG ......................................................................... 78 Figure 3-4: Initial and tensed group mean vertebral coordinates ................................................. 80 xiv  Figure 3-5:  Initial and tensed curvature indices ........................................................................... 81 Figure 4-1: The 50% MVC force task .......................................................................................... 91 Figure 4-2: Exemplar data flexion force task ............................................................................... 93 Figure 4-3: MVC task ................................................................................................................... 95 Figure 4-4: Postural metrics .......................................................................................................... 97 Figure 4-5: Mean muscle activations for the flexion and extension force tasks ......................... 101 Figure 4-6: Group average XY vertebral motions for the flexion and extension force tasks ..... 104 Figure 4-7:  Vertebral and intervertebral angle changes in flexion and extension force tasks ... 105 Figure 4-8: Curvature indices for the flexion and extension tasks ............................................. 106 Figure 4-9: Simplified free body diagram at the C5/C6 joint for Subject #11 ........................... 112 Figure 5-1: Subjects completing flexion and extension task inverted ........................................ 122 Figure 5-2: Cervical spine eccentricity ....................................................................................... 123 Figure 5-3: Exemplar cervical spine motion data for dynamic flexion and extension motion ... 125 Figure 5-4: Curvature index during dynamic motion compared to relaxed upright ................... 132 Figure 5-5: Angles of the head and cervical vertebrae during dynamic “neutral” ..................... 134 Figure 5-6: Mean muscle activity at neutral head posture .......................................................... 137 Figure 5-7: Potentially different force paths through a relaxed and dynamic spine ................... 139 Figure 6-1: Realignment in the constrained force task compared to Myers et al. ...................... 156 Figure 6-2: Tensed muscle activity patterns compared to literature ........................................... 160 Figure 6-3: Vertebral data of all subjects for all experimental tasks .......................................... 163 Figure 6-4: Changes in all curvature indices .............................................................................. 165 Figure A-1: Template image for normalized cross correlation ................................................... 199 Figure A-2: Example results of the normalized cross correlation .............................................. 200 xv  Figure A-3: Accuracy study of human vertebrae with optotrak markers ................................... 202 Figure A-4: Vertebral angle error compared to Optotrak for small out-of-plane motions ......... 204 Figure A-5: Vertebral angle error compared to Optotrak for large out-of-plane motions .......... 205 Figure A-6: The C3 (top) and C4 (middle) flexion angle during the dynamic motion .............. 206 Figure A-7:  Apparatus used to conduct accuracy study ............................................................ 209 Figure A-8:  Example X-ray images from the accuracy study where the vertebrae were moved around the X-ray plane and placed in various flexion angles. .................................................... 210 Figure A-9: Angle and displacement of C4 in various locations ................................................ 211 Figure A-10: Intervertebral angle and displacement of C3/C4 in various positions with varying image quality ............................................................................................................................... 212 Figure A-11: Frankfort plane validation ..................................................................................... 214 Figure C-1: Individual plots of upright tensed ............................................................................ 221 Figure C-2: Individual plots of inverted tensed .......................................................................... 222 Figure D-1: DLT rig.................................................................................................................... 224 Figure D-2: DLT rig and X-ray projection of the beads ............................................................. 224 Figure D-3: 3D projection of the beads onto the 2D image plane .............................................. 225 Figure D-4: FaroArm digitization of head landmarks ................................................................ 225 Figure E-1: Individual plots of upright flexion direction ............................................................ 226 Figure E-2: Individual plots of inverted flexion direction .......................................................... 227 Figure E-3: Individual plots of upright extension direction ........................................................ 228 Figure E-4: Individual plots of inverted extension direction ...................................................... 229 Figure F-1: Subject #3 vertebral and head angles throughout motion ........................................ 241 Figure F-2: Subject #4 vertebral and head angles throughout motion ........................................ 242 xvi  Figure F-3: Subject #5 vertebral and head angles throughout motion ........................................ 243 Figure F-4: Subject #6 vertebral and head angles throughout motion ........................................ 244 Figure F-5: Subject #7 vertebral and head angles throughout motion ........................................ 245 Figure F-6: Subject #8 vertebral and head angles throughout motion ........................................ 246 Figure F-7: Subject #9 vertebral and head angles throughout motion ........................................ 247 Figure F-8: Subject #10 vertebral and head angles throughout motion ...................................... 248 Figure F-9: Subject #11 vertebral and head angles throughout motion ...................................... 249 Figure F-10: Subject #1 muscle activities throughout motion .................................................... 251 Figure F-11: Subject #3 muscle activities throughout motion .................................................... 252 Figure F-12: Subject #4 muscle activities throughout motion .................................................... 253 Figure F-13: Subject #5 muscle activities throughout motion .................................................... 254 Figure F-14: Subject #6 muscle activities throughout motion .................................................... 255 Figure F-15: Subject #7 muscle activities throughout motion .................................................... 256 Figure F-16: Subject #8 muscle activities throughout motion .................................................... 257 Figure F-17: Subject #9 muscle activities throughout motion .................................................... 258 Figure F-18: Subject #10 muscle activities throughout motion .................................................. 259 Figure F-19: Subject #11 muscle activities throughout motion .................................................. 260  xvii  List of Symbols  Θ  Angle [degrees] F  Force [N] M  Moment [Nm] XECC  Eccentricity [mm] C.I.  Curvature Index [%] N  Number of subjects X  Position along the x-axis [mm] Y  Position along the y-axis [mm]  C1x  Position of C1 along the x-axis [mm] C7x  Position of C7 along the x-axis [mm] FFLEX  Head band force [N] WHEAD  Weight of head [kg] dFLEX  Head band force moment arm along the y-axis [mm] dCOM   Head band force moment arm along the x-axis [mm] FJOINT  Joint reaction force [N] MJOINT  Joint reaction moment [Nm] FSHEAR  Shear force [N] FCOMP  Compression force [N] MNET  Net applied moment [Nm]  xviii  List of Abbreviations  2D  Two dimensional 3D   Three dimensional AIS  Abbreviated Injury Scale ANOVA  Analysis of Variance ATD   Anthropomorphic test device CI   Confidence interval CIREN  Crash Injury Research and Engineering Network COM  Center of mass CT  Computed tomography DLT  Direct linear transformation Ecc  Eccentricity EMG   Electromyography Ext  Extended Flex   Flexed  G  Gravity I  Inverted  ICR  Instantaneous center of rotation LS   Levator scapulae MultC4  Multifidus MR  Magnetic resonance MVC  Maximum voluntary contraction xix   N.S.  Not significant OCC  Occiput RMS   Root mean square SCM  Sternocleidomastoid SsCap   Splenius capitis SsCerv  Semispinalis cervicis STH   Sternohyoid Trap   Trapezius U  Upright              xx  Acknowledgements  There are many people that I would like to thank for their contributions to this work and support through this PhD journey.  First and foremost, I would like to thank the subjects that volunteered for this research, without their contributions this work would not have been possible.    I would like to thank my supervisory committee for their support and guidance through the process of such a large endeavor; your contributions have been integral to conducting high quality research.  Thank you to Dr. Peter Cripton and Dr. Gunter Siegmund for your valuable mentorship, your patience, and your encouragement.  Peter, thanks for showing me that the glass should always be half full, a lesson that has enabled me to persist (the unofficial P in PhD!) through the tough times.  Thank you for putting up with my eternal skepticism, for all of your research and teaching mentorship, and providing opportunities for me to grow as an academic.  It has been fun, and not just “Peter” fun!  Gunter, you have taught me how to strive for perfection, thank you for being tough on me!  You have taught me a great deal about experimental design and I have enjoyed and greatly benefited from all of our academic discussions.  Thanks to Drs. Jean-Sébastien Blouin and John Street for all of your help and support over the years, your contributions and feedback have been invaluable.    Thanks to Mircea Oala-Florescu and Jeff Nickel for helping with many aspects of the work.  Your technical support and wealth of knowledge were extremely valuable.  You have also taught me that it never hurts to have strong people around to lift the heavy stuff!   xxi  Thanks to Gunter Siegmund, Jean-Sébastien Blouin, Hannah Gustafson, Claire Jones, Shawn Newell, and Jennifer Douglas for your help conducting the experiments and Chantal Percival, Amanda Percival, Justin Krieger, Hank Hsu, and Kohle Merry for helping with data processing.  Your contributions are greatly appreciated.    I would like to thank my family for their never ending support throughout not only this degree but my seemingly never-ending career as a professional student.  I absolutely could not have done this without your love, reassurance, and understanding.  Mom and Dad, thanks for being great role models and for teaching me to work hard, to believe in myself, and encouraging me to reach for the stars!  Shawn and Dana, I couldn’t ask for more supportive and caring brothers.  Nan, thanks for your never-ending support and being the strong woman that I strive to be.     Thanks to my fellow OIBG lab members that have been a big support network for me.  Claire Jones and Hannah Gustafson, thanks for being there through the hard times; I couldn’t have done this without you.  Thanks to Tim Bhatnagar, Seth Gilchrist, Tim Nelson, Carolyn Van Toen, Jake McIvor, Angela Melnyk, Jennifer Douglas, Steve Mattucci, and Emily McWalter for being a big part of the rollercoaster ride.  Thanks everyone for supporting my East Coast style and keeping it classy at the Fairview!   Thanks to my friends (especially Kate and Mel) for being there for the well needed down-time and understanding when I had to tend to the demands of a PhD.  Sharel and Mike, you have been a big part of this PhD journey, thank you for making me feel like part of your family.  Thanks to the all of the Wildcats ’97 (players and parents), Gary Kingman, and Lisa Ebel for your support xxii  over the years.  Girls, even though I’m the coach you have taught me a lot about life and have provided perspective during stressful times.    Finally, I would like to thank MEA Forensics Engineers & Scientists for the generous contributions to this work and NSERC-MITACS for their funding.      1  Chapter  1: Introduction  1.1 Overview Serious cervical spine and spinal cord injuries can result from axial impacts to the head, such as those sustained in motor vehicle and sports accidents.  More specifically, this type of injury may occur to the occupant(s) of a car rollover accident.  An occupant can be thrown head-first into the roof of a car, often while inverted, in which case the neck is required to support the torso momentum.  Past cadaveric research efforts have highlighted that the pre-impact alignment of the cervical spine is a large determinant of the type and severity of injury sustained in a head-first impact.  These past efforts have been invaluable to understanding the biomechanical factors important to injury tolerance and the anatomic components at most risk of failure in various impact configurations.  However, little information is known about what these exact postures may be before impact.  In order to effectively prevent these injuries, clinically relevant injuries must first be reliably replicated in the laboratory setting.  These efforts are fraught with challenges and, particularly in relation to rollover accidents, the laboratory does not produce the full spectrum of injuries observed in real-life.  Since pre-impact cervical alignment is critical to the injury mechanism, it is possible that, at least in part, the disparity between real-life and laboratory-induced injuries can be attributed to the inability to replicate relevant pre-impact postures.    Neuromuscular computational models can also be useful to study neck injury, but require definition of a muscular activation and control scheme and are typically not validated by human subject data relevant to rollovers.  Improving current head-first cervical spine injury models and 2  rollover models requires a better understanding of the role of muscle activation in cervical spine alignment, and in particular, spinal alignment in an upside-down configuration.  Due to the complex technical and ethical challenges faced by human volunteer research of this nature, no study has explored neck posture and muscle activation patterns in an inverted posture or when actively tensing the neck muscles in preparation for impact.     The first aim of this introduction chapter is to provide an overview of cervical spine injury (more specifically head-first axial impact neck injury in rollover accidents), past and current techniques for studying neck injury (including cadaveric and computational modeling) along with the limitations of these injury models in studying neck injury and injury prevention in rollover applications.  Second, this introduction chapter provides an overview of the mechanical factors that are known to influence the type and severity of injury during head-first impact, with a specific concentration on the importance of neck alignment and neck muscle activations.  This section presents the concept that the manner in which the pre-impact state of the neck (alignment and muscle activity) is currently modeled is generally not physiologic or based on in vivo data relevant to rollover injury.  This is followed by a section describing techniques that can be used to study the in vivo response in conditions that would be relevant to the pre-impact state.  Finally, the introduction chapter concludes with the objectives of this thesis and a brief summary of the aim of the four studies conducted for this thesis.    1.2 Cervical spine injury This section provides an overview to how major neck injuries occur in rollover accidents, followed by a description of how injury mechanisms are classified based on past ex vivo work.  3  A description of the models that are currently used to study axial impact neck injury and a summary of the gap between injuries that are observed in real-life accidents and the injuries currently being simulated in the laboratory are provided.  Finally, the importance of knowing the pre-impact spinal posture for rollover injury prevention is discussed.    1.2.1 Cervical spine injury in rollover accidents Vehicular rollover accidents (those in which the vehicle undergoes at least one quarter-turn of rotation about the longitudinal or lateral axis) account for 33% of all fatal crashes and are a significant source of major injuries [1].  Among crash conditions, rollovers account for the majority of serious cervical spine injuries [2], leading to an increased likelihood of spinal cord injury [3].  Fatal spinal injury in motor vehicle accidents typically involve the occiput-C2 joint, whereas damage to the lower cervical spine (mainly C5-C6) is more frequent in survivors [2, 4].  In motor vehicle accidents, there is a strong association between head and spine injuries [2].  In the case of a rollover, this association may be due to contact of the head with the interior of the vehicle [5, 6].  Crash Injury Research and Engineering Network (CIREN) cases of rollover occupants that have sustained neck injury have reported evidence of loading through the head (i.e. scalp lacerations from head contact with the roof) [7].   Neck injury in a rollover can happen when the roof of the car hits the ground and the occupant, who is upside-down, strikes the interior of the vehicle (typically the roof) with their head [8, 9].  It has been suggested that when contacting the roof, the head is stationary relative to the impact surface and the cervical spine is loaded from the momentum of the moving torso [10].   Devastating injuries (with or without neurological damage) such as vertebral fractures, ligamentous and disc injuries can occur to the spinal column and surrounding structures as a result of this axial impact loading.   4  Neck injuries to rollover occupants may appear to be mechanistically similar; however, the patterns of injuries to the spinal cord and vertebral column are largely heterogeneous [7, 11-14].  Understanding how specific injuries occur in a rollover is important to span applications from recreating clinically relevant injuries in a repeatable fashion in a laboratory, for diagnosing and treating cervical spine injury, and for designing injury prevention devices.    1.2.2 Cervical spine injury mechanisms The neck is mechanically complex: it comprises of seven vertebrae and more than forty muscles and it interacts with the head through the unique C1 and C2 vertebrae.  Motions and postures can be attained with remarkable redundancy in both muscles and joints.  In fact, it has been described as one of the most complex – and least understood – neuromechanical systems of the body [15].  The cervical spine can be injured by loading from all directions, yet the mechanism of injury at the individual joint or vertebral level is complex and does not always follow the overall loading directions of the head/neck/torso.  For instance, impact that induces flexion of the head does not necessarily cause localized flexion-type injuries to the individual vertebrae or joints [16].  The vertebral orientations, local anatomical features, vertebral size, and ligamentous anatomy vary throughout the spine, thus influencing localized loading paths. Individual joints can be subjected to pure loading in one direction (i.e. compression or shear), moments in one direction (i.e. flexion/extension, lateral bending, or torsion), or combined loadings.  The direction, rate, and magnitude of loading all influence the types of injuries sustained at the segmental level.    The classification of neck injuries is complicated and has changed as more is learned about injury mechanisms.  Injury classification is an important component of the diagnosis and 5  treatment of cervical spine injury.  One classification scheme described by Myers et al. [17] designates the localized injuries (as opposed to the gross motions of, and loads to, the head) based on experimental observations.  The majority of cervical injuries are thus categorized into compression, compression-flexion, compression-extension, tension, tension-extension, tension-flexion, torsion, shear, and lateral bending.  Specific clinical descriptions of injury are classified under one of these nine categories (see Table 1-1).  For instance, when sufficient force is directed axially through the anterior column of the cervical spine (i.e. through the vertebral bodies) a compression injury may follow [18-20].  Burst fractures, which are particularly hazardous to the spinal cord [21-24], are an example of a compressive injury [23] (see Figure 1-1).                 6  Table 1-1: Cervical spine injury classification The classification of cervical spine injuries based on the applied forces with experimental validation [17]. Injury Classifications  Compression Jefferson fracture Multiple atlas fracture Vertebral body compression fracture Teardrop fracture  Compression-flexion Teardrop fracture Burst fracture Wedge compression fracture Hyperflexion sprain Bilateral facet dislocation Unilateral facet dislocation  Compression-extension Hangman’s fracture Clay-shoveler’s fracture Posterior element fracture Anterior longitudinal ligamentous rupture Anterior disc rupture Horizontal vertebral body fracture Teardrop fracture  Tension Occipitoatlantal dislocation  Tension-extension Hangman’s fracture Anterior longitudinal ligamentous damage Disc rupture Horizontal fracture of vertebral body Teardrop fracture  Tension-flexion Bilateral facet dislocation Unilateral facet dislocation  Torsion Atlantoaxial rotary dislocation Unilateral atlantoaxial facet dislocation  Shear Odontiod fracture Transverse ligament rupture  Lateral bending (in combined loading) Asymmetric injury Nerve root avulsion Peripheral nerve injury  Some injuries are more difficult to classify because controversy still exists as to the exact injury mechanism.  For example, bilateral facet dislocations are classified as both tension-flexion and compression-flexion injuries [17, 25].  These injuries can occur when the major injuring vector is posterior to the apex of the head, while the head and neck are flexed forward [26, 27] (see Figure 1-1).  Some have suggested that localized flexion at the intervertebral level can produce bilateral facet dislocations, even if the impact is anterior to, or aligned with, the apex of the head [28, 29].  These dislocations have also been achieved by applying slow axial loads to cadaveric spines with 7  a rotational constraint at each end of the cervical spine [30, 31].  Indeed, no clear consensus appears to exist as to how exactly bilateral facet dislocations are sustained in vivo.  The heterogeneity of cervical injury patterns and a lack of understanding of neck mechanics at impact have largely hindered efforts to standardize clinical injury nomenclature [32, 33].  To gain a better understanding of the injury mechanisms in the context of a rollover, and ultimately prevent them, it is imperative to consistently simulate clinically relevant injuries in a laboratory setting.     Figure 1-1: Examples of compression and compression-flexion injury mechanisms Illustrations of how three types of compression and compression-flexion injuries (wedge fractures, burst fractures, and bilateral facet dislocation) are thought to occur in the cervical spine. Reproduced from Nahum, A.M. and Melvin, J.W., 2002. Accidental injury: Biomechanics and prevention, Springer Science+Business Media, Inc., New York, NY. with permission from Springer [34].  1.2.3 Cervical spine injury models Cadaveric and computational models have been important to studying cervical spine injury and exploring potential injury prevention approaches.  Anthropomorphic test devices (ATDs) offer 8  another avenue for studying neck loading in the rollover environment.  However, as will be discussed in this section, limitations inherent to these techniques somewhat preclude their direct application to the context of a rollover.  Human volunteer experiments have been important for capturing the in vivo response to a rollover environment, yet comprehensive data sets are largely lacking.     1.2.3.1 Cadaver neck injury models A variety of cadaveric models have been used to study axial neck injury, ranging from smaller segmental units to full body cadavers; each model has added to the body of knowledge on the factors that influence injury.  Full cadavers (see Figure 1-2) have been useful in studying the interaction of the neck with the head and torso, typically with soft tissues (such as passive musculature) left intact [35, 36].  Some researchers have tested cadaveric neck segments, with and without a head, enabling easier manipulation of the end conditions of the spine [23, 27, 30, 37-41].  To visualize the spinal column, these tests are often conducted with the soft tissues removed.  Others have conducted experimental impacts on smaller segments of the cervical spine [42, 43] providing the advantage of controlling and isolating the direct effects of loading parameters, but such models lack interactions with the rest of the spine and head, as well as spinal curvature.  Materials testing machines [27, 37-41] (see Figure 1-2) provide a method of controlling the positions of the spine while applying controlled loading.  Pendulum and linear impacts [44, 45] applied to whole cadavers also allow for realistic impact energy to be imparted, yet it is difficult to make direct measurements of loading through the spine and cadavers are typically in the prone or supine position.  Drop tests are useful for simulating realistic inverted loading from free fall and the energy imparted by the torso momentum [16, 28, 35, 36, 46]; 9  however, it is difficult to control the loading parameters and spinal posture (cables are often needed to pre-align the head and neck).                     Figure 1-2: Examples of ex vivo tests Experimental impact, induced by a materials testing machine, on a head and neck cadaver specimen (left) and inverted drop test with a full cadaver (right). Image on the left Reprinted with permission from SAE International Paper No.  892436: Pintar, F.A., Yoganandan, N., Reinartz, J., Sances, A.J., Harris, G., and Larson, S.J., 1989. Kinematic and anatomical analysis of the human cervical spinal column under axial loading. SAE Technical Paper No. 892436 [37].  Image on the right reprinted with permission from SAE International Paper No.  831616: Nusholtz, G., Huelke, D., Lux, P., Alem, N., and Montalvo, F., 1983. Cervical Spine Injury Mechanisms. SAE Technical Paper No. 831616 [36].  Cadaver testing has been valuable to understanding the mechanics of axial loading injury; yet, a disparate spectrum of injuries is observed in ex vivo specimens compared to real-life rollover 10  occupants [11, 12].  One of the biggest limitations of cadaveric testing is the difficulty in simulating muscular responses prior to, and during, impact.  Even efforts to simulate musculature are not often based on in vivo data relevant to a rollover scenario and individual muscles are not represented.  Moreover, an occupant in a rollover can sustain neck injuries from head contact with vehicle structures when they are upside-down [9, 47], while tests with cadaver neck specimens have traditionally been conducted in upright [18, 19, 39] or prone/supine [44, 45] configurations.  Even drop tests performed upside-down have made attempts to preserve the upright “neutral” posture [28, 36].  The effect of muscle activations and whole body inversion, which are often neglected, could be critical elements to improving cadaveric models for rollover applications.  Previous experiments that have tried to create clinically relevant cervical spine injuries in cadavers (via head-first impact) have had difficulties consistently replicating some of the injuries observed clinically [16, 27, 36, 44, 45, 48].  If effective evaluation of injury prevention devices is to be achieved, it is imperative that these injuries can be reliably recreated in ex vivo models.  Injury in the context of rollover accidents is particularly challenging as the exact application of force, and thus the mechanism of injury, is never truly known.  The direction and magnitude of forces applied to the head and spine and mechanisms of injury in a rollover can only be deduced from the injury patterns reported in real-life accidents and evidence of impact with the vehicle structures.    The vast set of previous cadaver tests have mostly been aimed at studying axial head-first impact injuries in general, and have not always had the particular focus of a rollover application.  11  However, comparing the relative rates of particular injury patterns from head-first impact cadaver tests with those of injured rollover occupants may help in understanding what mechanical factors relevant to rollovers are missing from current ex vivo testing.  For instance, a high proportion of major neck injuries observed in rollovers are flexion or flexion-compression injuries and unilateral and bilateral facet dislocations [11, 49].  Even when bilateral facet dislocations are created in a cadaver model, they are not always clinically relevant.  Fractures near the facet joints (e.g. lateral mass) (as opposed to the superior facets “jumping” over the inferior ones) are more common in rollover occupants than in cadavers [11].  Furthermore, real-world rollovers tend to have higher distributions of lower cervical spine injuries than are seen in cadaveric testing [11, 12], where 65.2% of vertebral fractures are from C5 to C7 compared to 39% in ex vivo specimens [11].  Many mechanical factors could explain why certain injuries are more difficult to replicate in the laboratory.  As will be discussed further in Section 1.3, relevant pre-positioning of the cervical alignment and muscle simulation (or lack of) in the laboratory could be major contributing factors to this disparity, and thus may need to be more accurately modeled in ex vivo experiments.        1.2.3.2 Computational modeling Cadaveric tests have been critical to identifying the mechanical factors (such as the critical loads and postures) important to axial neck injury.  However, cadaveric tests are time intensive, expensive, and measuring localized strains and loads is challenging.  Furthermore, muscle tissue is typically removed because it inhibits direct visualization of the vertebral column during injury simulation.  Even when muscle tissue is left intact on the specimen [39, 44, 45], this only simulates the passive characteristics of muscle behavior during impact.   Not only does 12  computational modeling help to address the aforementioned problems, it provides an opportunity to tune mechanical parameters that may be important to neck injury (such as inertial and stiffness contributions) [50] and prevention strategies (friction and direction of impact) [51].    Several neuromuscular models have been useful for studying axial neck loading injury in tension [52, 53] and compression [51], and ligament and disc strains in “near-vertical” impacts [54] (in seated, helmeted helicopter pilots).  Hu et al. [51] proposed that decreasing impact velocity, reducing padding thickness (to limit head pocketing), increasing padding stiffness, decreasing the head-surface coefficient of friction, and changing the angle of impact can all reduce the neck injury risk in rollover accidents with a computational model (see Figure 1-3).  Notably, they also predicted that active musculature almost doubled the risk of cervical spine injury in a rollover scenario.  These are novel and dramatic findings, yet the validity of these findings depends on the simulation of realistic muscle activation schemes.  Indeed, a notable limitation to current axial impact muscle models is that the muscle activation patterns are not based on, or validated to, measured in vivo responses.    13   Figure 1-3: Example of a neuromuscular model for axial impact loading of the neck Finite element computational cervical spine model (which includes 23 pairs of cervical muscles) for evaluating the factors that influence the risk of cervical spine injury during axial head-first impacts in rollover crashes. Adapted from Hu, J.W., Yang, K.H., Chou, C.C., and King, A.I., 2008. A numerical investigation of factors affecting cervical spine injuries during rollover crashes. Spine 33 (23), 2529-2535, with permission from Lippincott Williams & Wilkins [51].  Neuromuscular modeling has also been used to explore the influence of active and/or passive muscles on kinematic responses in simulations of vehicular impacts in the frontal, lateral, and rear directions [55-63]; and to compare the predicted effects to those observed in volunteer subjects.  Some neuromuscular modeling work, which was validated using in vivo kinematics, has investigated joint loading and the effect of head position in aircraft combat maneuvers [64].  However, this model applies to fighter aircraft pilots (who tend to be a select population) wearing helmets while being subjected to large superior-inferior G-forces.  Indeed, unlike 14  cervical spine loading in other directions of automobile impact (e.g. frontal), axial impact models are often not validated with in vivo kinematics.  Many computational injury models of axial impact have been validated with the mechanical response of cadaveric intervertebral segments or ex vivo drop tests [51, 54, 65-68] (notably those conducted at Duke University [28, 46, 69]).  As detailed in the last section, these cadaveric tests do not necessarily simulate an upside-down posture or the effects of muscle.  Although it is not possible to obtain in vivo data during injurious loading, data concerning the in vivo response leading up to impact would prove useful as inputs to, and validation for, computational models.  Furthermore, axial neck injury models have often been generic, usually based on the 50th percentile male and/or derived from a single cadaveric specimen [52, 65, 67, 68, 70].  Between-subjects variation is typically not addressed in modeling neck injury; therefore, the pre-impact response of a variety of individuals is also necessary.  1.2.3.3 Anthropomorphic test devices and human subjects Several types of models have been used to study full-body occupant kinematics during rollovers, or the head-first impact portion of the rollover. However, each of these has limitations predominantly associated with a lack of data with which to design or validate the model.  In addition to computer simulations and cadaver models, rollover simulations also include Hybrid III ATDs [71, 72].  For instance, a study of the kinematics of an anthropometric test device in relation to roof excursion in rollovers concluded that higher neck loads result from alignment of the head, neck and torso at impact [10].  One of the major limitations of this ATD approach is that the dummy does not represent an occupant response in a rollover situation [49].  ATDs are stiffer (i.e. transmit higher loads through the cervical spine) in the axial and bending directions 15  compared to live humans [73] and cadavers [30, 38, 41] (up to 50 times larger axial neck stiffness in unconstrained axial loading [30]) and may not simulate a human’s motion in a rollover environment [72, 74].  Similarly, it has been shown that the neck and shoulder complex of the ATD model is too stiff and lacks movement of the neck compared to a human response to the initiation of a rollover accident [75, 76].    Studies of human subject occupant kinematics and muscle responses have been valuable to interpreting the biofidelity of the ATD biomechanical response in frontal, rear, and lateral automobile impacts [73, 77-83].  In contrast, very few studies have performed dynamic rollover simulations using human subjects [84, 85] (typically with limited subject group size), and the majority of these have only analyzed the initial phases of the rollover prior to inversion of the occupant’s body [74-76, 86].  Head excursion, a measure of head-to-roof clearance in the inverted posture, has been studied with human subjects in a static environment [71, 72].  However, there has been no specific attention to cervical spine alignment in this posture.  In a limited number of publications it has been shown that there is a substantial amount of neck muscle activity in simulations of rollover conditions [74, 75, 86].  In fact, volunteer studies of the initiation of a rollover (a full rotation was not simulated), demonstrated that occupants had sufficient time to activate the sternocleidomastoid and trapezius muscles [76, 86].  One volunteer subjected to a lateral roll (testing commenced inverted and stabilized, then rapidly rotated through less than 1 rad) noted: “At the initiation of a dynamic roll, I experienced an involuntary tensing of the neck muscles but flexion bending of the head is not indicated on the video” [84].  Although anecdotal (and the volunteer was the primary author of the study), this provides further motivation for the necessity of understanding musculature in rollover occupant behavior.   16  Rollovers involve complex motions where the occupant is exposed to centripetal acceleration (typically 3-4 times the acceleration of gravity) [87].  Images of volunteers during steady state rolling indicate that occupants may have a variety of responses to a rollover environment.  They may resist the centripetal acceleration and maintain a forward gaze [74, 84, 88], have an active response to the lack of headroom (i.e. “ducking”) [89], or neck flexion induced by the seatbelt [90].  In the case of lower roll rates the neck may be under gravity-induced tension [91] or, in the case of higher roll rates, the occupant’s neck may be compressed by the vehicle roof [92].  The exact in vivo neck alignment and muscular response of an occupant during a rollover is still largely unknown.  In fact, the neck realignment resulting from purely being inverted or actively anticipating impact is yet to be quantified.    1.2.4 Rollover injury prevention Several devices have been proposed to prevent neck injuries in rollover accidents, including novel roof designs [67] and seat-mounted airbags [93].  Heudorfer et al. [93] proposed a rollover protective mechanism whereby an airbag deploys from the car seat head rest (behind the occupant’s head) and forces the occupant’s head into flexion.  They termed this a “rollover-protected” position and used crash test dummies (as mentioned earlier, they are documented for their lack of biofidelity in rollovers) to assess the effectiveness of the device.  They reported that axial compressive neck forces were reduced by 280%; yet, this assumed the major loading vector is above the head.  Rollover occupants often have evidence of impact to the forehead and face [11, 12, 14] and flexion and flexion-compression injury mechanisms represent a large proportion of serious neck injuries in rollover accidents (reported to be as high as 90% of cases where the injury mechanism was identified) [49].  Although it is impossible to deduce the exact loading 17  mechanism in real-life rollovers, these flexion and flexion-compression injuries highlight the possibility of axial impact loading through the flexed and/or aligned head and spine.    Halldin et al. [67] proposed another protective design for a vehicle roof during a rollover: a surface between the occupant’s head and the roof that would move independently of the roof, forcing the head and neck into a flexed position.  Theoretically, this allows the head and neck to escape the loading of the torso; thus reducing the risk of injury.  Numerical simulations of this device yielded promising results (lower loads and stresses through the cervical spine); yet, the model was missing key elements such as active and passive musculature.  Moreover, before inducing the head and neck into a flexed posture the authors implicitly assumed the neck alignment would be similar to the upright relaxed posture, an assumption that is yet to be conclusively validated, in the rollover context, in vivo.    In an inverted head-first impact the head often acts as a constraint, while the momentum of the torso loads the cervical spine predominantly in the axial direction [28, 46].  There is a greater risk of cervical spine injury when the neck is perpendicular to the impact plane and the apex of the head is impacted [28].  Myers et al. [30] impacted cervical spine specimens with the resting lordosis preserved and did not produce injury with an unconstrained spine.  However, they produced mid-cervical compression fractures in a fully constrained spine and produced lower-cervical bilateral facet dislocations in a rotationally constrained spine.  In similar experiments, by impacting an angled platen to induce head rotation, Nightingale et al. [28] demonstrated that the neck is capable of “escaping” serious injury induced from the torso (which still has considerable momentum).  Thus, diverting the head and neck from the incoming torso momentum may mean 18  the difference between catastrophic injury and no injury at all [94].  Cadaveric models have enabled researchers to better understand these mechanical factors that are important to preventing axial neck injury.  However, the exact initial conditions prior to any real-world head-first impact in a rollover are largely unknown.  To be able to design prevention devices based on these theories, a better understanding of the initial alignment of the spinal column and the stiffening capability of muscles is critical.    1.3 Factors that are important to cervical spine injury Although limited in their application to rollover accidents, the ex vivo cadaveric and computational modeling described above has provided evidence that several mechanical factors influence the mechanism and severity of injury.  The magnitude and direction of the loading vector, and the alignment of the neck, head, and torso, have been identified and well documented as important factors.  This section provides an overview of mechanical factors influential in determining the mechanism of neck injury one is likely to sustain in a head-first impact.  The importance of cervical alignment prior to axial head-first impact is highlighted, particularly cervical eccentricity (i.e. the anterior/posterior distance between the top and bottom of the cervical spine) and cervical curvature.  Recent evidence from computational modeling that active musculature is capable of influencing the risk and type of injury is also discussed.  This section concludes with a summary of how cervical alignment and muscle forces are, in general, currently being achieved in non-physiologic manner in the laboratory.    19  1.3.1 Mechanical factors that influence injury Although axial loading has been highly researched using cadavers, critical thresholds for injury are difficult to estimate.  This is mainly due to the large number of conditions that have been varied, or not controlled for, in these experiments, all of which have the potential to influence injury.  There is evidence that the higher the speed of impact the more likely serious cervical spine injury will occur.  Head and neck specimens (plus 16 kg of simulated torso mass) dropped from 0.54-0.61 m (approximately 3.2 m/s at impact) onto a rigid surface are reported to consistently sustain spinal column disruption without skull fracture [16].  It has been estimated that at head impact velocities above 4.2 m/s, there is a 50% risk of serious neck compression injury (injuries also included some fractures to the upper thoracic and basilar skull) [95].  Poor correlations exist between the velocity at impact and the force imparted at impact during both drop tests and pendulum impacts, and neither are strong predictors of neck injury [95].  For instance, Nusholtz et al. [44] produced fractures of C6-C7 (with ligament ruptures and disc involvement) in a specimen with only 1,800 N of impact, whereas Alem et al. [45] impacted a specimen with 16,000 N in a similar manner and did not observe neck injury.  In general, pure compressive injuries are sustained under higher axial loads than those required to produce injury in compression-flexion or compression-extension.  Myers and colleagues [30] produced compression fractures at an average loading of 4,810 N and bilateral facet dislocations at an average loading of 1,720 N, purely by altering the rotational constraint on the spine.  Indeed, it is difficult to use force as a predictor of injury, since critical force levels are highly dependent on other mechanical factors.    20  Surface padding [28, 66], impact angle [28], degree of constraint [26, 30], torso orientation [35, 36], and coefficient of friction between the impact surface and head [51, 65] also influence the risk of and type of injury.  Regardless of the other conditions present at impact, the forces that develop throughout the spine are highly dependent on the orientation of the spine, both overall and at the intervertebral level.  The wide variation in reported critical forces is largely due to differences in the pre-impact alignment of the spine, with respect to both the positions of the ends of the cervical spine (i.e. eccentricity) and the intrinsic curvature of the spine [96].      1.3.1.1 Neck eccentricity The eccentricity of the cervical spine is typically defined as the difference in the anterior/posterior alignment of the superior and inferior ends of the spine, i.e. the top and bottom of the cervical segment.  The mechanism of axial loading injury has been reported to be influenced by changes in the eccentricity of the cervical spine [18-20, 41].   Pintar et al. [27] found axial loads exceeding 1850 N and bending moments exceeding 62 Nm predict a 25% risk of major hyperflexion injuries; however, the ability to predict injury with force or moment was not strong.  Indeed, the authors proposed that spinal eccentricity (occipital condyles to T1) was the most significant factor in producing hyperflexion type injuries.  Using a rigid test fixture, Maiman et al. [20] pre-aligned cervical spines with 15° forward head flexion and manipulated the eccentricity of the spine by moving the occipital condyles (with respect to T1) from 0.5 cm (posterior) to 10.2 cm (anterior) (see Figure 1-4).  Average eccentricities of -5 mm, 1 mm, 23 mm, and 53 mm are reported to result in compression-extension, vertical compression, compression-flexion, and hyperflexion injuries, respectively [19, 20] (see Table 1-1 for examples of these types of injuries).  Thus, injuries to the spinal column are sensitive to the overall spinal 21  eccentricity.  Although eccentricity is a major factor in the risk and type of injury, the cervical spine usually also has a natural curvature (i.e. lordosis) that greatly affects the load path through the spine.      Figure 1-4: Varying eccentricity of the spine Manipulation of the eccentricity of the spine – defined by the horizontal distance of the top of the spine (occipital condyles) relative to the bottom (T1) – to study the effects of anterior (left) and posterior (right) eccentricities and an aligned column (center) on the cervical injury.  Adapted from Maiman, D.J., Yoganandan, N., and Pintar, F.A., 2002. Preinjury cervical alignment affecting spinal trauma. Journal of Neurosurgery 97 (1), 57-62 with permission from the Journal of Neurosurgery: Spine [20].   22  1.3.1.2 Neck curvature The eccentricity of the cervical spine (i.e. end conditions) is not sufficient to prescribe the expected mechanism of injury.  The localized positions of each cervical vertebra determines the path of load transmission throughout the cervical column at each level.  When the cervical joints are flexed forward (or kyphosis is present) or extended backward (or lordosis is present), the column tends to bend away from the direction of the load.  In contrast, when the spinal column is aligned, the column typically cannot escape the path of the load; therefore, the aligned column tends to be stiffer and can withstand higher loads before failure [18, 39, 45].  In this case, the vertebrae (instead of the soft tissues) absorb larger proportions of the energy being imparted to the column, and it is often the vertebrae themselves that fracture in this load scenario.  In general, extension injuries are observed in cadaveric specimens with a slight lordosis and flexion injuries are observed in those with kyphosis [18].  Aligned ex vivo columns have produced mid- and lower-column compression type injuries (wedge and burst fractures) [18, 23, 97], and anterior alignments (>15 mm of eccentricity) have resulted in lower-column ligamentous flexion injuries (without bony involvement) [18].  Compressive neck injuries, such as burst fractures, have also been produced experimentally in spines with the lordotic curvature retained [28, 36, 39, 45, 97].  This likely suggests that, even if there is curvature in the spine, the load may only have to align with portions of the cervical curvature.  Nonetheless, the straightened spine (with no lordosis or kyphosis) is commonly accepted as the stiffest and most vulnerable posture of the osteo-ligamentous spine.       Even though eccentricity has been manipulated in cadaveric testing, direct measures of curvature have not been documented.  Curvature is not necessarily independent of eccentricity; since 23  adjacent vertebrae do not move independently, it can be difficult to separate the effects of each on injury.  An example of this is illustrated in Figure 1-4, where the manipulation of eccentricity also induces changes in the curvature of the spine.  Localized eccentricities, induced by changes to the neck curvature, do not have to be large to influence the loading path through the cervical spine.  The balance point of a cervical segment is the point where compressive forces do not induce rotation and has been reported to be 0.5-1 cm anterior to the posterior longitudinal ligament in lower cervical segments (three vertebrae and two discs) [42].  Even small posterior shifts of the loading vector (beyond the balance point) induce joint rotation.  Furthermore, small posterior shifts in the loading vector can also reverse the vertebral body strain from compressive to tensile and increase the compressive strain on the lateral masses [42].   Toomey et al. [96] described a linear relationship between the resultant force in the sagittal plane and eccentricity for two series of cadaveric drop tests (one set with straightened curves and one set with lordosis retained).  However, the relationships were weak, particularly in the extension direction, and potentially reflect variations in curvature that were not accounted for.    Thus, to fully understand the likelihood of specific injuries, the localized positions of the vertebrae throughout the entire column are needed.  In essence, as highlighted in Figure 1-5, the position of the loading vector relative to the column itself is a major determinant of the injury likely to be produced [94].  This has important implications for the risk of spinal cord injury: burst fractures and bilateral facet dislocations are frequently associated with spinal cord injuries [2] while wedge compression fractures are recognized as more clinically benign [98].     24   Figure 1-5: Relationship between load eccentricity and type of neck injury Changes in the type of injury (from compression-extension, to compression, to compression-flexion) produced as the eccentricity of the loading vector moves anteriorly.  Reproduced from Winkelstein, B.A. and Myers, B.S., 1997. The biomechanics of cervical spine injury and implications for injury prevention. Medicine and Science in Sports and Exercise 29 (7 Suppl), S246-55 with permission from Wolters Kluwer Health [94].  1.3.1.3 Head alignment The angle and position of the head are other important factors to cervical injury mechanism.  Maiman et al. [39] demonstrated that cadaveric specimens with pre-extension (to 25°) that were impacted at the hairline experienced extension type injuries (disruption of the anterior longitudinal ligament), whereas impacts slightly anterior to the vertex of pre-flexed (to 25°) heads resulted in flexion type injuries (such as dislocations and disruption of the posterior ligaments). Axial impacts to the vertex of specimens with neutral head positions (0°) resulted in mostly compression type injuries (i.e. burst and teardrop fractures).  Although the curvatures and eccentricities were not documented, the impact loads and energies reported for the neutral head posture were approximately double those reported for pre-flexed and pre-extended heads.  In real-world motor vehicle accidents, extension-tension loading caused by impact to the lower 25  facial regions result in a larger proportion of upper cervical spine injuries, while extension-compression loading caused by impact to the higher facial regions result in lower cervical spine injuries [2, 4], also highlighting the importance of head position.    1.3.1.4 Muscles Active neck musculature has largely been ignored in cadaveric testing of cervical spines, as axial loading injury is reported to occur faster than it takes for muscles to react [65].  It is estimated to take approximately 54-87 ms to elicit a reflexive muscle contraction and another 60 ms to fully contract the neck muscles [99], whereas injury has been reported to typically occur within 20 ms of head impact [46, 100].  Yet, a rollover can be a relatively long duration event (roll rates of 200-300 deg/sec are not uncommon [101] and pre-roll events often occur prior to trip initiation) during which an occupant may have time to react to an impending impact to the head.  These muscle activations prior to impact may stiffen the vertebral column and change the forces that develop through the column during impact.     Simulation of only a few muscles (splenius capitis, semispinalis capitis, longus capitis) in ex vivo specimens stabilizes the upper cervical spine (i.e. reduces the range of motion and neutral zone) [102].  Forces have typically been applied at 70-110 N to simulate overall muscle tone [35] and 140 N [103] and 200 N [104] to simulate compressive spinal preload in cadaver testing.  These ranges are typically to simulate the overall spinal compression in a relaxed posture [105].  Even these low levels of passive muscle simulation in axial loading impact have been shown to influence the rate of injury [35].  Thus, it is plausible that higher levels of muscle activation would further stiffen the spine and have a significant effect on neck injuries.   26  The mechanical loading of the cadaveric neck in drop tests can also be influenced by a follower load [100] (see Figure 1-6) that simulates overall muscular compressive forces.  Since this method applies forces along the approximate centers of rotation of the spinal column, this method does not offer significant rotational stability.  Indeed, muscle forces applied near the center of rotation violate physiological loading conditions in the in vivo spine [106], particularly since the joint centers of rotation shift during active spinal motions [107] (Figure 1-6).  As with other techniques of simulating overall muscle tone, the follower load does not account for the various attachment sites of muscles and the various directions of forces that muscles apply.            27   Figure 1-6: Follower load and instantaneous axes of spinal motion The simulation of a follower load on a cadaveric cervical spine (left) where the ideal follower load passes through the instantaneous center of rotation (ICR) of each segment (circles).  For comparison, the in vivo ICR of each segment (right) are highlighted through dynamic extension (blue = full extension) and flexion (red = full flexion). Left image reproduced from Patwardhan, A.G., Havey, R.M., Ghanayem, A.J., Diener, H., Meade, K.P., Dunlap, B., and Hodges, S.D., 2000. Load-carrying capacity of the human cervical spine in compression is increased under a follower load. Spine 25 (12), 1548-54 with permission from Lippincott Williams & Wilkins [108].  Right image reproduced from Anderst, W., Baillargeon, E., Donaldson, W., Lee, J., and Kang, J., 2013. Motion path of the instant center of rotation in the cervical spine during in vivo dynamic flexion-extension implications for artificial disc design and evaluation of motion quality after arthrodesis. Spine 38 (10), E594-E601 with permission from Lippincott Williams & Wilkins [107].  28  Loading of the cervical spine, and subsequent injuries, are highly sensitive to the segment stiffness and the constraints on the ends of the spine.  In cadaveric models, constraining the ends of the spine renders the spine more susceptible to compressive type injuries.  Without constricting the occipital-C1 joint, Hodgson and colleagues [26] were not able to reproduce compressive injuries.  Myers et al. [30] demonstrated that by releasing the constraint on the spine they could not produce injury because the spine deflected away from the load path (Figure 1-7).  Although the loading rate was slow, they were able to consistently produce lower cervical bilateral facet dislocations by rotationally constraining the upper cervical spine, and mid-cervical spine compression injuries by fully constraining the spine.  By stiffening the lower cervical spine (C6 to T1) with a rod (albeit non-physiologic) and constraining the upper cervical spine, Bauze and Ardran [31] obtained unilateral and bilateral facet dislocations.  Clinically relevant hyperflexion injuries have historically been particularly difficult to replicate ex vivo [109], particularly with locked facets [27], perhaps confirming the important role of active and passive musculature.      29   Figure 1-7: Ex vivo tests with various end conditions Cadaver tests with unconstrained (left), rotationally constrained (center), and fully constrained (right) end conditions. Adapted from Myers, B.S., McElhaney, J.H., Richardson, W.J., Nightingale, R.W., and Doherty, B.J., 1991. The influence of end condition on human cervical spine injury mechanisms. SAE Technical Paper No. 912915 391-9, reprinted with permission from SAE International Paper No. 912915 © 1991 SAE International [30]  A column that is sufficiently slender will buckle (i.e. become unstable) under loads that are less than the failure load of the material it is comprised of [110]; this form of instability is termed Euler buckling.  A segmented column provides a more mechanically complex description of the cervical spine; yet, the concept of instability is still important and is determined by joint stiffness and column geometry.  Under this comparison, changes in the intersegmental stiffness significantly change the axial buckling mechanics of the neck [50].  In fact, computational simulations applying 4000 N of compressive load while increasing the cervical joint stiffness by a factor of eight resulted in a “structurally stable column” [50].  Osteoligamentous spines buckle even from the weight of the head [111]; however, healthy spines do not buckle under normal in 30  vivo conditions.  This means that active musculature is primarily responsible for regulating stability in the spinal column [110].    Spine modeling has suggested that neck musculature can also affect the load sharing between the muscles and ligamentous spine [112] and has the potential to protect the cervical ligaments from injury [56].  Modeling of pre-impact muscle tensing has predicted that the neck muscles could generate compressive preloads as high as 800-1400 N throughout the spine [52].  Numerical simulations of a pre-impact tensed cervical spine performed by Dibb et al. [57] demonstrated that the muscles are capable of imparting significant spinal pre-loads: up to 40% of the reported tolerance of the neck in compression [28].  They suggested that loads of this magnitude (up to 1023 N in the adult model) would drastically influence the risk of spinal injury during compressive loading.  The magnitude of forces across the cervical joints increased caudally, and although it was not explicitly explored, this suggests that tensed muscle activations would be capable of shifting the location of compressive injury.  In fact, Brolin et al. [54] reported that computational simulation of active musculature predicts changes to the risk of ligamentous injury and is capable of shifting the location of injury in axial loading of the neck.  Simulating head-first impact in a rollover environment, Hu et al. [51] also proposed that active muscle forces may increase the risk of cervical spine fracture.  Research on high-performance aircraft pilots exposed to axial G-forces highlights the influence of overall spinal posture on the mechanical advantage of muscles and advocates for the protective capabilities of active neck musculature [113].    31  Furthermore, visual inspection of a computational model with active “bracing” of the muscles indicates that some level of spinal realignment may be present (see Figure 1-8), although realignment was not specifically quantified [52].  Thus, not only do active neck muscle contractions have the ability to influence the stiffness of the cervical column prior to injury, it is likely they can alter the cervical spine alignment in a significant way.  Yet, there is currently a lack of knowledge of how muscle control alters the posture and stiffness of the cervical spine [74].  Figure 1-8: Cervical spine neuromuscular model by Chancey et al.  Neuromuscular model of the cervical spine under gravity (left), in the relaxed state (center), and the tensed state (right) prior to tensile loading.  Reproduced by permission of the STAPP Association: Chancey, V.C., Nightingale, R.W., Van Ee, C.A., Knaub, K.E., and Myers, B.S., 2003. Improved estimation of human neck tensile tolerance: Reducing the range of reported tolerance using anthropometrically correct muscles and optimized physiologic initial conditions. Stapp Car Crash Journal 47 135-153 [52].  32  1.3.2 Limitations to simulating cervical alignment and muscle forces ex vivo Even though numerous activities have been shown to induce considerable changes in cervical alignment [114-128], there is a paucity of data describing the realignment of the spine in conditions prior to an axial impact.  These pre-impact alignments are critical for cadaveric and computational modeling of injury.  The osteoligamentous cadaver cervical spine is inherently unstable and reportedly withstands only approximately 10 N of compression before buckling [111].  In fact, this “critical” load is insufficient to even support the head for ex vivo testing; therefore, simulation of the overall muscle tone is required to support the weight of the head and achieve a desired head and spinal alignment.  Attempts to study axial injury mechanisms of an “aligned” posture (i.e. cervical lordosis removed) in an osteoligamentous cadaver have also been achieved with rigid attachment of the spine and/or head to a testing fixture [37, 41] (see Figure 1-2).  Combinations of cables, pulleys, and springs have also been used to apply forces directly to the head [18, 44, 46, 97, 129] (see Figure 1-9), which can create non-physiologic tensile forces in the spine [130].  A similar challenge exists in inverted drop tests of ex vivo specimens, whereby cables are needed to counterbalance the gravity-induced extension of the head [16, 28, 35, 36, 46].  The head/C1 joint extends in a non-physiologic manner in the osteoligamentous spine and thus also needs to be stabilized in cadaveric testing [130].  In the in vivo environment, supporting the weight of the head, stabilizing the joints, and aligning the spine are physiologically achieved by muscle forces.    33   Figure 1-9: Realignment of the cervical spine and simulation of muscle tone in an ex vivo axial impact Pulley-weight system used to manipulate the cervical alignment and simulate muscle tone prior to an axial impact test on a cadaveric head and neck.  Reprinted with permission from SAE International Paper No. 952722 © 1995 SAE International, Pintar, F., Yoganandan, N., Voo, L., Cusick, J., Maiman, D.J., and Sances, A., 1995. Dynamic characteristics of the human cervical spine. SAE Technical Paper No. 952722 [18].  Currently in the laboratory, overall changes to the posture of the cervical spine are achieved in non-physiological ways.  The overall effect of muscle tone can be simulated to change the end positions of the spine.  However, in vivo the muscle forces required to adopt these positions are achieved by a complex network of muscles, none of which are directly perpendicular to the spine.  In fact, we are not capable of solely manipulating our head position or the ends of our 34  cervical spine (i.e. eccentricity) without applying other forces throughout the spine.  These additional muscle forces likely realign the curvature of the spine or further stabilize the cervical joints.  These subsequent changes to the cervical alignment are yet to be accurately simulated in a cadaver model.  Furthermore, the extent to which cervical realignment happens in vivo in conditions relevant to a head-first impact are not fully known.  Thus, a data set describing in vivo cervical realignment, under the conditions of whole body inversion, and with activation of musculature, may help in defining the factors most important to pre-impact alignment.  1.4 Measuring in vivo responses An in vivo data set of vertebral and muscular responses, in the context of pre-impact in a rollover environment, could be used to improve and validate current injury models and advance injury prevention strategies.  Although neck posture and muscle activity influence the resulting injury, there are currently no in vivo data describing these parameters immediately prior to a head-first impact.  Since a primary aim of several proposed rollover protection devices is to actively change the neck posture [67, 93], the initial posture of the head and neck is critical.  Yet, it is currently unclear what an upside-down occupant’s posture would be or if actively tensing the muscles in anticipation for an impact may have the potential to further alter the spinal posture.  In vivo measurements during conditions that could be present in a rollover, such as whole body inversion and active bracing with muscle contractions, may help in filling the gap between real-life injury and those simulated in a laboratory setting.  Therefore, this section provides a summary of techniques for capturing and quantifying in vivo neck alignment and muscle activations.  35  1.4.1 In vivo neck kinematics Radiographic image analysis is the gold standard for static posture measurement and motion tracking for in vivo spine biomechanics [131-135].  Due to the complexity of the spinal column structures and the overlying muscle and soft tissue, particularly in the neck, external motion tracking of the head and neck are not sufficiently accurate [136].   Discrepancies between vertebral angles determined using external surface markers versus radiographic landmarks have been reported to be between 4.8° [137] and 10.8° [132].  At the extremes of cervical spine sagittal plane motion, skin movement with respect to bone has been reported to be as much as 3 mm [137].  X-ray fluoroscopy is an effective and highly accurate method for studying the complex relative motion of the underlying bony structures of a joint [138].  The complex interplay of neuromuscular control and neck vertebral kinematics has been comprehensively described in animals such as the cat [139, 140] and rhesus monkey [141], but a detailed description is generally lacking in humans.     There are several options for measuring three-dimensional (3D) motion of the cervical spine, including combinations of planar and bi-planar fluoroscopy, computed tomography (CT) and magnetic resonance (MR) imaging modalities.  Radiostereometric analysis techniques use a bi-planar X-ray configuration to collect offset pairs of images in which bony landmarks are identified and used to derive 3D kinematics.  Radiostereometry has been shown to have 2 mm and 1.5° accuracy in the lumbar spine [133].  One drawback to this bi-planar technique is the difficulty in precisely identifying the same bony landmark in separate views [142].  Accuracy can be improved to 0.18 mm, which would be acceptable for this study, by implanting radio-opaque beads into the spine [143]; however, this is highly invasive and limited to a post-surgical 36  patient population [144].  In addition, bi-planar fluoroscopy more than doubles the subject radiation dose compared to that of a sagittal configuration, due to the anterior location of radiosensitive organs such as the thyroid [145].  Registration techniques exist to combine dynamic two-dimensional (2D) images from a planar or biplanar fluoroscopes with subject-specific, static, 3D data sets from either CT or MR scans [146-148].  Combining CT and fluoroscopy achieves accurate and precise 3D kinematics; however, the radiation dose from CT to subjects is high (approximately 4.6 mSv of effective dose - the equivalent of 55 conventional chest X-rays [149]).  Unlike CT, MR scanning does not involve additional radiation exposure; however, it is difficult to define cortical bone surfaces in MR images and there is poor agreement between kinematics derived from MR images and conventional X-ray techniques [150].  Our pilot work (unpublished) with MR and single plane fluoroscopy porcine cervical spine specimens proved to be time intensive and yielded insufficient repeatability.  Single-plane fluoroscopy provides a good balance between time efficiency, accuracy, and most importantly, limiting the radiation dose to the subjects.  Frobin et al. proposed a technique to describe vertebral sagittal plane joint angles that is almost uninfluenced by X-ray magnification and distortion errors [151].  In this method, landmarks on the cervical vertebral bodies are identified in fluoroscopic images to define the corners of the vertebral contour.  A disadvantage to this approach is that it only uses minimal information about the vertebral body (i.e. four corners) and manually identifying of these four corners is onerous and generally not repeatable.  Other techniques using automatic image registration have been proposed to improve on reliability and decrease analysis time [152-157].  Automatic registration allows for less user interaction (time saving), is more robust (gives the same results regardless of the user), and has 37  the potential to use more visual information from the vertebral body (beyond only four corner points).  An automatic tracking algorithm for spinal kinematics, proposed by Bifulco et al. [157], was chosen for the studies of this thesis.  This algorithm uses normalized cross-correlation to match a template of each vertebra to the fluoroscopic image of interest.  This technique yields maximum root mean squared errors of 0.2° and 0.3 mm in the lumbar spine [157].  Fluoroscopic videos are typically recorded at 30 frames per second; thus, automatic registration allows the vertebral translations and angles to be detected through time in a time efficient and accurate manner.    1.4.1.1 Kinematic parameters One of the main goals of measuring realignment of the cervical spine is to evaluate which factors affect alignment, through statistical comparisons.  The kinematic measures chosen for this thesis strike a balance between this and providing useful data sets for improving cadaveric and computational models.  There are numerous methods of describing cervical spine kinematics since there are multiple joints present in the neck, each of which is capable of three-dimensional rotation and translation.  Reporting the angles and translations of each vertebral joint provides thorough, albeit large, data sets on the spinal motions.  Thus, combinations of the absolute (relative to the global horizontal) vertebral motions and intervertebral motions are reported throughout this thesis.    Metrics of overall spinal realignment are also relevant to the context of neck injury and would be beneficial to future ex vivo and computer modeling.  As discussed earlier, changes in eccentricity and curvature have been reported to influence neck injury; therefore, measures of these are also 38  reported throughout this thesis.  Eccentricity, the horizontal offset of the top and bottom of the spine, has typically been reported as the distance between the occipital condyles and T1.  Since a clinical C-arm was used for this work, these are not easily measured (the shoulders tend to block the view of the thoracic spine); thus, the horizontal distance between the cranial (C1) and caudal (C7) vertebrae are used as a measure of eccentricity.  There are numerous ways to clinically describe spinal curvature (such as Cobb angles and the posterior tangent method), thirteen of which are reviewed in detail by Vrtovec et al. [158].  A measure termed the curvature index, which compares the relative length of the arc and chord lengths of the cervical spine, was chosen due to its intuitive nature and because Klinich et al. [159] previously used this metric to describe a large database of adults in a seated automotive posture.   1.4.2 Electromyography Muscle activations likely influence neck injury (as discussed previously); however, simulating musculature in ex vivo tests is challenging.  Passive and pre-impact activation of muscles are approximated using cables and springs, and reflex mediated activations are yet to be modeled in a cadaver.  One of the main limitations to date is the lack of information regarding the selection of important muscles and the magnitude of forces involved.    One of the primary roles of the muscles in the cervical spine is to support load and provide stability in conjunction with the ligaments in the spine [160].  Active musculature drives spinal motion; yet, the exact motion is complex and is also partly determined by ligaments and gravity [161, 162].  In general terms the muscles can be categorized by their attachment points (see 39  Table 1-2) and comprehensive anatomical descriptions of the muscles have been provided by Kamibayashi et al. [163]   Table 1-2: Anatomical attachments of several neck muscles The general anatomical attachments of numerous neck muscles relative to the vertebral column, head, shoulder girdle, and rib cage as described by Kamibayashi et al. [163] Muscles linking the skull with the shoulder girdle Muscles linking the skull with the vertebral column Muscles that link the vertebral column with the rib cage Muscles that link the scapula and vertebral column • sternocleidomastiodeus • trapezius Long dorsal muscles • splenius capitis/cervicis • semispinalis capitis/cervicis • longissimus capitis/cervicis • scalenus anterior • scalenus medius • scalenus posterior • rhomboideus major and minor • levator scapulae  Suboccipital muscles • rectus capitis posterior major and minor • obliquus capitis inferior and superior    Ventral muscles • Longus capitis • Longus colli • Rectus capitis anterior minor • Rectus capitis lateralis    It is difficult to surmise the functional roles of the neck muscles based on anatomical dissections and attachment points, primarily due to the complexity of the joints they surround and the intricacy of the muscle composition (i.e. fiber type and architecture) [161].  A single muscle can be capable of multiple tasks [163]; thus, studying the patterns of activation during a variety of cervical postures and normal ranges of motion is helpful in identifying the specialized functions of individual muscles [139].  Direct measurements of force in live human muscles are currently not feasible; thus, measuring a proxy of muscle force is necessary.   Electromyography (EMG) is valuable for studying relative activations and patterns of recruitment of muscles.  The 40  relationship between the electrical signal that a muscle emits and the force it applies is complex and depends on various factors such as the muscle orientation, cross sectional area, length, and velocity.  Thus, estimating specific muscle forces is currently only possible by exploring EMG signals in conjunction with advanced computational models.   Inverse dynamics models are driven by the measured joint kinematics and can be used to predict joint forces and muscle contributions [164].  Anderst et al. [164] proposed one such approach while offering the advantage of simulating subject-specific models.  Without the direct input of muscle patterns, this required the assumption of minimal muscle co-contractions throughout a dynamic flexion/extension task.  Forward dynamics models are driven by measured muscle activations and can be used to predict joint kinematics and joint contact forces.  However, the neck muscles and joint kinematics are highly redundant; therefore, these models do not generate unique solutions.  Grouping the muscles or applying optimization routines can help manage this redundancy inherent to various joints [165].  Optimization methods in the neck include minimizing parameters such as fatigue (using muscle stress as a proxy) [52], joint reaction moments [166], individual muscle activations [64], and compressive forces [105] or maximizing muscle force [57].  Models of muscle tensing can also be tuned to allow for head equilibrium under the effects of gravity [167].  Chancey et al. [52] proposed a model in which the objective function maximized the total muscle forces and minimized fatigue, while establishing dynamic equilibrium and maintaining head posture.  Models such as this can use muscle activation levels (i.e. 0%-100% of maximum activation) as inputs to calculate the magnitude of neck muscle forces, as well as the force at each cervical level.    41  Similarly, “EMG assisted” techniques provide an opportunity to use EMG activations (normalized to MVC tasks) to drive generic [168] or a subject-specific [169] spinal models.  Although there are inherent limitations to this approach (such as the need to accurately describe each subject’s geometrical data), EMG driven models allow individual muscle activation patterns and co-contractions to be simulated.  Surface and indwelling EMG recordings have been used to drive isometric muscle models, with various head positions where joint moments were validated with isometric force measurements [170].  However, fine wire EMG data sets for static and dynamic tasks are generally unavailable for axial injury simulations [52].  The neck is particularly challenging to model; thus, the combination of EMG and vertebral kinematics may provide the basis for a more detailed approach to modeling spinal behavior.     1.4.2.1 Muscle selection To access accurate recordings of specific musculature, particularly the deep muscles, fine wire (indwelling) electrodes were selected.  The cervical spine consists of a large number of cervical muscles that are located anterior, lateral, and posterior to the spinal column.  Although EMG recordings from a large number of muscles would be useful, indwelling insertion is time consuming and invasive.  To reduce experimental time and subject discomfort, eight muscles were selected for this thesis work.  These muscles were the sternohyoid (STH, with surface electrodes), sternocleidomastoid (SCM), trapezius (Trap), levator scapulae (LS), splenius capitis (SPL), semispinalis capitis (SsCap), semispinalis cervicis (SsCerv), and multifidus (Mult) (Figure 1-10).    42   Figure 1-10: Cross-sectional MRI and neck muscle identification  A cross-sectional MR image of the mid cervical spine identifying the location of seven neck muscles (SCM, LS, Trap, SPL, SsCap, C4 Mult, and SsCerv). Reproduced from Siegmund, G.P., Blouin, J.S., Brault, J.R., Hedenstierna, S., and Inglis, J.T., 2007. Electromyography of superficial and deep neck muscles during isometric, voluntary, and reflex contractions. Journal Biomechanical Engineering 129 (1), 66-77, with permission from ASME [171].    These eight muscles are thought to play a large role in head and neck motion and force generation.  Vasavada et al. [172] estimated that the SsCap and SPL are major extensors, while SCM is a major contributor to the maximal flexion moment.  Conley et al. [173] summarized similar functional roles and suggested SsCerv and Mult may play a larger role in postural stability than SsCap and SPL.  Interestingly, work by Blouin et al. suggests that deep (SsCerv 43  and Mult) and superficial (SCM, Trap, LS, SPL, and SsCap) muscles have a common neural drive and these muscle layers may not act completely independent of one another during isometric contractions [174].  Trapezius has previously been described as being involved with neck extension and lateral bending [175].  Some have suggested that Trap plays a larger role in manipulating and stabilizing the scapula than the head [176].  It has been suggested that the LS primarily acts on the shoulder and scapula and that its role in motion of the cervical spine is negligible [161].  Although LS engagement in generating isometric forces with the head in the postero-lateral direction has been demonstrated [171, 174], computational modeling suggests this muscle has lateral and extension moment generating capability in the lower cervical spine.  The role of SPL is complex and distinct from other neck muscles, and ambiguity exists as to its role in stabilization and focal isometric force generation in the neck [174].  The function of SPL appears to be highly subject specific [171, 174] and seemingly contradictory, since its role has been described in lateral bending [177], extension [178, 179], antero- and postero-lateral bending [174], and head rotation (ipsilateral splenius may function as a rotator while contralateral splenius functions as an extensor) [179].    Dynamic motion in the extension direction has been reported to involve SPL, SsCap, SsCerv, and Mult, with contributions from LS [113, 171].  SCM and STH play a large role in flexion, and straightening of the cervical lordosis during dynamic flexion is reported to occur with deactivation of SPL, SsCap, SsCerv, and Mult [161]. The exact contribution of LS, SPL and Trap to dynamic motion is not particularly clear as a variety of individual responses have been noted [171].    44  Seven of these eight selected muscles (all but STH) have previously been accessed with fine-wire electrodes [171, 174] and, as described above, they have all been identified as major generators of isometric forces in the cervical spine and major contributors to spine posture and motion.  Thus, they were selected to obtain a better understanding of the contribution of muscle activations to the realignment of the cervical spine.  1.5 Research objectives To improve our understanding of head-first cervical spine injury, particularly in the context of rollover accidents, and to be able to consistently replicate and study these injuries in the laboratory requires a better understanding of the cervical alignment and muscle activations prior to an impact.  Identifying how the cervical spine can realign under various conditions, particularly in an upside-down configuration (a condition present in a rollover accident), is important if we ultimately want to prevent neck injury in rollover accidents.  Therefore, the objective of this body of work was to analyse segmental cervical spine kinematics (via fluoroscopy) and neck muscle activity (via electromyography) during various quasi-static and dynamic conditions, performed in both the right-side-up and upside-down conditions.  These conditions include: (1) static inversion, (2) neck muscle tensing without head constraint, (3) neck muscle tensing with head constraint, (4) dynamically passing through the neutral neck position.  Although these conditions were selected to represent those that could be present (to some extent) in a rollover environment, they were also selected to better understand how the cervical spine may realign prior to an impact.  More specifically, the main focus of these experiments was to quantify alignment changes and muscle response in the near neutral posture (where head and 45  cervical alignment are assumed to be in their upright, relaxed configurations), a posture that is directly applicable to current ex vivo and computational approaches to studying neck injury.    The overarching question of this thesis is: of whole body inversion (inversion), head position (posture), cervical motion (motion) and muscle activations (muscles), which factor(s) significantly influence the cervical spine posture?  To answer this, the human volunteer experiments consisted of four different ways of mechanically perturbing the cervical spine and observing which of these four factors are capable of producing cervical spine realignment.  These four research studies, each with a specific objective (outlined below), are presented in detail in Chapters 2 through 5 of this thesis.  Objective of study 1:  Does whole body inversion and head orientation affect the relaxed neutral posture of the cervical spine? (Chapter 2)  Postural alignment of the cervical spine is important to the study of cervical spine injury in head-first impact.  The cervical spine alignment in the inverted posture of human subjects, as well as the influence of neck muscle activation on this alignment, is currently unknown.  In this study (Chapter 2), the role of gravity on cervical alignment is explored by turning subjects upside-down, both in a relaxed position and with the head flexed to achieve a level gaze.  Objective of study 2: Does bracing for impact result in a realignment of the cervical spine? (Chapter 3)  46  It is currently unclear what the in vivo muscular response and pre-alignment of the cervical column would be while bracing for an impending impact.  In this study (Chapter 3), the effect of actively tensing the neck muscles (without head constraint) on cervical realignment is explored.  In this task, subjects tense their neck and shoulder muscles to stiffen and simultaneously draw in their neck to simulate a protective response in preparation for a head-first impact (both upright and inverted).  Objective of study 3: Does exerting a flexion/extension force (with head constraint) alter the alignment of the spine? (Chapter 4)  The activation of cervical spine muscles has the ability to stiffen and realign the cervical column, yet it is unclear to what extent this occurs in vivo.  In this study (Chapter 4), the role of actively tensing the neck muscles (with head constraint) on cervical realignment is explored when subjects generate 50% of their maximal voluntary contractile forces in the flexion and extension directions (both upright and inverted).  Objective of study 4: Does dynamic motion of the cervical spine affect the alignment of the neck in the neutral position? (Chapter 5)  Motion of the neck prior to impact might be expected in a dynamic environment such as a rollover.  This motion (particularly in the inverted configuration) may reposition the spine in a way that is dissimilar to upright resting posture (a posture often used in cadaveric testing).  In this study (Chapter 5), dependency of the “neutral” alignment on prior neck motion is explored 47  through active dynamic flexion and extension.  The alignment of the cervical spine as the head and neck dynamically pass through the neutral position (as determined by neck eccentricity) is compared to the alignment in the upright resting posture, for both upright and inverted configurations.    48  Chapter  2: Does inversion and head orientation affect the relaxed neutral posture of the cervical spine? 1  2.1  Introduction Rollovers are dynamic and complex events in which serious neck injuries, such as vertebral fractures and spinal cord injuries, can be caused by head contact with the vehicle interior [5, 13].  A common mechanism of neck injury in rollovers occurs when the vehicle roof contacts the ground and the upside-down occupant continues head-first into the roof at their original velocity [8, 9].  Cadaver studies have shown that the axial force that develops in the neck from such inertial loading generates a spectrum of spine fractures and other injuries [28, 35, 36, 103].  Heterogeneous neck injury patterns also occur in real-world rollovers [13]; yet, there is some disparity between the neck injuries observed in rollover accidents and cadaver experiments [11].    A number of studies have shown that the overall alignment of the head, neck, and torso has an effect on the neck injury mechanism observed [20, 26, 28, 44, 45].  Loss of the neck’s lordosis (e.g. by flexing the spine forward) can increase susceptibility to compressive loads and thus burst fracture patterns of spinal column injury [18, 37, 97], often with concomitant spinal cord injury [21, 22].  Although posture is known to influence the neck injury sustained in a head-first impact, cadaveric specimens are often tested upright or inverted using the neck’s upright resting   1 Chapter 2 has been published.  Newell RS, Blouin J-S, Street J, Cripton PA, Siegmund, GP. Neck posture and muscle activity are different when upside down: A human volunteer study. Journal of Biomechanics 2013;46:2837-43.  Permission to use content granted from Elsevier.  49  (lordotic) posture.  In fact, the actual configuration of the cervical spine immediately before an inverted head-first impact remains unknown.    Ex vivo neck models of head-first impact lack neuromuscular control and the associated postural stability.  Some cadaver tests use osteo-ligamentous spines, whereas others have left the muscle tissues intact to capture a passive muscle response [36, 39, 45].  Overall muscle tone has been simulated with cables and springs, typically to restrain the head and maintain a particular cervical posture for axial testing [97, 103, 104].  Computational modeling has suggested that muscle activation influences neck injury risk in axial loading [54] and may increase the neck fracture risk in rollovers [51].  However, as with the cadaver experiments discussed above, muscle activation schemes are typically not based on human subject data.  Indeed, the level of muscle contraction prior to an impact is unknown, and there is little understanding of how this muscle control alters the posture and stiffness of the cervical spine during a rollover [74].  The general assumption of existing head-first cervical spine impact models, likely due to the paucity of data, is that the upright and inverted necks have the same posture.  However, if inversion influences neck posture, the initial conditions currently used in modeling head-first injury may not adequately replicate rollover conditions.  Muscle activation levels while inverted may also be important inputs to these cadaveric and computational models.  Therefore, the aim of this study was to measure and compare the neuromuscular activity and cervical spine posture in upside-down configurations versus a neutral resting upright configuration.   We hypothesized that, compared to the upright posture, inversion would change the head angle, neck curvature, and neck eccentricity (horizontal distance between C1 and C7), and increase neuromuscular 50  activity; and further that these differences would depend on whether subjects were relaxed or maintaining a level gaze.    2.2  Methods 2.2.1 Subjects Eleven asymptomatic subjects (6 females and 5 males, 33 ± 7 years, average ± standard deviation) participated in this study.  Subjects were excluded if they had a history of neck injury or pain, or known muscle, nerve, or balance problems.  The average group height was 172.9 ± 6.7 cm, their weight was 70.0 ± 15.7 kg, and their neck and head circumferences were 34.3 ± 4.2 cm and 56.4 ± 3.2 cm, respectively.  One subject did not participate in the fluoroscopy portion of the experiment and another subject did not participate in the EMG portion.  Subjects gave their written informed consent and the study was approved by the University of British Columbia’s Clinical Research Ethics Board.  2.2.2 Conditions Subjects were seated in a bucket seat (36 Series - Intermediate 20° Layback, Kirkey Racing Fabrication Inc., St. Andrew’s West, ON), secured with a 75-mm wide 5-point harness (RCI Racer’s Choice Inc., Tyler, TX) and held statically in one upright and two inverted postures (Figure 2-1). Subjects were instructed to adopt three postures: 1) upright and relaxed (U-R), 2) inverted and relaxed (I-R), and 3) inverted and looking forward (I-F).  The inversion trials were not randomized (I-R first, then I-F) and subjects were given about 30 seconds to experience the sensation of inversion before the experiment started.   51    Figure 2-1: Experimental set-up Left: Subject secured in the inversion device, with the C-arm positioned for a sagittal view of the neck.  Center: Head band with beads used to track the motion of the Frankfort plane in the X-ray.  Right: Subject performing MVC tasks while wearing a helmet attached to a load cell. Reproduced from Newell, R.S., Blouin, J.-S., Street, J., Cripton, P.A., and Siegmund, G.P., 2013. Neck posture and muscle activity are different when upside down: A human volunteer study. Journal of Biomechanics 46 (16), 2837-2843. with permission from Elsevier [180].  2.2.3 Cervical spine posture A fluoroscopic C-arm (OEC 9400, GE) was used to capture sagittal plane images of the cervical vertebrae at 30 Hz.  Each subject was exposed to a total of 12 seconds of fluoroscopy and on average a total effective dose of 0.0067 mSv for this experiment (a typical transatlantic flight is 0.03-0.045 mSv [181]).  Fluoroscopic images were corrected for distortion [182] and vertebral motions were tracked in the images using an automatic tracking algorithm based on Bifulco et al. [157].  Briefly, the image gradient was calculated and normalized cross-correlation was used to 52  match a template of each vertebra to the fluoroscopic image of interest.  Anatomical landmarks were identified for each vertebral body (Figure 2-2) in the template images and the tracking algorithm was used to find the angle and displacements of the vertebral bodies and landmarks. Accuracy studies using cadaveric cervical vertebrae (see Appendix A) yielded maximum root mean squared (RMS) errors of 0.6° and 0.2 mm (Bifulco et al. [157] reported values of 0.2° and 0.3 mm in the lumbar spine).  Several images were tracked to ensure the subject remained static, and one image near the end of each trial (at the mid-point of the EMG window – see Section 2.2.5 below) was used for the statistical comparisons.       53   Figure 2-2:  Vertebral landmarks The vertebral body outlines of C1, C2 and C3 from the lateral X-ray image for one subject.   Vertebral landmarks (white dots) were identified (based on Klinich et al. [159]) and used to define the anatomical coordinate system for each vertebral body.  The origin of each vertebral coordinate system was located at the mid-point between the two inferior points of the vertebral body.  Vertebral bodies C4 through C7 (not illustrated) were defined in the same way as C3. Reproduced from Newell, R.S., Blouin, J.-S., Street, J., Cripton, P.A., and Siegmund, G.P., 2013. Neck posture and muscle activity are different when upside down: A human volunteer study. Journal of Biomechanics 46 (16), 2837-2843. with permission from Elsevier [180].  Cervical spine posture was quantified with three measures: the eccentricity (XECC), the curvature index (C.I.), and the C7 angle (θC7).  Eccentricity is the horizontal distance between the mid-superior point of C1 and the mid-inferior point of C7 vertebral bodies (positive = C1 anterior to 54  C7) (Figure 2-3).  The curvature index is defined as the percentage difference between the arc and chord lengths of the spine (Figure 2-3) [159], where the arc length is the sum of the straight line segments passing through the superior and inferior mid-points of adjacent vertebral bodies from C2 to C7 and the chord length is the distance from the top of the dens to the inferior mid-point of the C7 vertebral body.  A higher curvature index indicates more curvature.  The angle of the C7 vertebral body is with respect to the true horizontal and is an indication of the overall neck orientation (Figure 2-3, positive = flexion).  2.2.4 Head orientation Head orientation was measured using a bead array (six-2mm beads) that was fastened to the head but projected into the field of view of the fluoroscope (Figure 2-1).  The Frankfort plane angle (ΘF, RMS error = 0.74º, see Appendix A) is defined by the right and left tragus and the mid-point of the right and left inferior orbits and was measured relative to the true horizontal (Figure 2-3) (positive = flexion).  Prior to data collection, the beads and Frankfort plane landmarks were digitized using a FaroArm (B08-02, Lake Mary, FL) and the initial angle offset between the beads and the Frankfort plane was established.  Angular motion of the Frankfort plane was thus equivalent to subsequent angular motion of the beads.      55   Figure 2-3:   Postural metrics derived from the fluoroscopic images. Curvature index (CI, top left), eccentricity (XECC, top right), C7 angle (θC7, bottom left), and Frankfort plane angle (θF, bottom right).  A positive eccentricity, a positive C7 angle, and a negative Frankfort plane angle are shown in the diagrams. Reproduced from Newell, R.S., Blouin, J.-S., Street, J., Cripton, P.A., and Siegmund, G.P., 2013. Neck posture and muscle activity are different when upside down: A human volunteer study. Journal of Biomechanics 46 (16), 2837-2843. with permission from Elsevier [180].   56  2.2.5 Electromyography EMG activity was measured using both surface and indwelling electrodes.  Surface electrodes (Ambu Blue Sensor, Ambu A/S, Ballerup, Denmark) were placed on the skin superficial to the left sternohyoid (STH) muscle [171].  Indwelling fine-wire electrodes were inserted into the left sternocleidomastoid (SCM), trapezius (Trap), levator scapulae (LS), splenius capitis (SPL), semispinalis capitis (SsCap), semispinalis cervicis (SsCerv), and multifidus (MultC4) muscles.  The indwelling electrodes consisted of pairs of PFA-coated Stainless Steel, 0.0055” diameter wire (A-M Systems, Inc., Sequim, WA) with 1 mm exposed wire and 2-3 mm spacing between the exposed ends of the two wires.  The wires were inserted under ultrasound guidance into the center of each muscle belly at the C4/5 level [174].  Since the cross-section of the trapezius muscle at C4/C5 is typically only a few millimeters thick, these wires were inserted near the C5/C6 level.  The EMG signals were amplified, band-pass filtered (wire: 50-1000 Hz; surface: 30-1000 Hz) and sampled at 2 kHz.  The RMS of the muscle activity was calculated for a 500 ms window during the trial.  A trigger pulse (emitted by the image acquisition card) enabled the fluoroscopic image of interest to be synchronized with the middle of this 500 ms window.  Each muscle’s activation was normalized to the maximum RMS activity recorded for that muscle in a maximum voluntary contraction (MVC).      2.2.6 Maximum voluntary contractions Seated subjects were secured to a rigid backboard while wearing a snug skateboard helmet.  The helmet was attached to a 6-axis load cell (45E15A-U760, JR3, Inc., Woodland, CA, nominal horizontal accuracy: ±2.5N) above the subject’s head (Figure 2-1).  Isometric MVCs were performed with verbal encouragements [183] and real-time visual feedback of force/moment 57  magnitude and direction.  With a neutral head posture, MVCs were executed in seven directions: flexion, extension, left lateral bending, two 45° oblique combinations (flexion/left lateral bending, extension/left lateral bending), and right and left axial rotations.  Each contraction was three seconds long and repeated twice.  The EMG signals were amplified and filtered as described above and both the load cell and EMG were sampled at 2 kHz.  A 500 ms window centered on the maximum force was used to calculate the RMS for each muscle’s EMG in all seven directions (Figure 2-4).  The maximum RMS value for each muscle, regardless of direction, was used for normalization.         58   Figure 2-4:  Exemplar data for EMG processing.  The 500 ms window (gray dashed) containing the maximum RMS force in the lateral bending direction for the MVC task (top left).  The RMS of the EMG signal for the muscle of interest (Trap) was calculated in this 500 ms window in all seven force and moment directions.  The maximum RMS value, regardless of direction, was identified for the Trap (the left lateral bending direction for this case).  For the static data, a 500 ms window (gray dashed) near the end of the trial is used to calculate the RMS of the Trap EMG traces for the three static conditions (U-R top right, I-R center right, I-F bottom right) and are normalized to the MVC trial (% MVC).  Note that all of the static EMG traces are on the same µV scale as the MVC signal. Reproduced from Newell, R.S., Blouin, J.-S., Street, J., Cripton, P.A., and Siegmund, G.P., 2013. Neck posture and muscle activity are different when upside down: A human volunteer study. Journal of Biomechanics 46 (16), 2837-2843. with permission from Elsevier [180].   59  2.2.7 Data processing notes All cervical vertebrae and beads were visible in the 23cm field of view except for C1 in two subjects while inverted.  To obtain eccentricity for these subjects, a linear regression was derived between the horizontal location of the mid-superior points of C1 and C2 from the other subjects (C1 = 0.996*C2 – 5.9 mm; R2=0.99).  The neck posture measures of one subject were omitted because the C7 vertebra was occluded by the shoulders in one image and another subject did not have an EMG recording for MultC4.       2.2.8 Data and statistical analysis All head/neck posture variables and EMG signals were analyzed using Matlab (R2010b, MathWorks Inc., Natick, MA).  The kinematic metrics and EMG were compared across the three static conditions (Upright-Resting, Inverted-Resting and Inverted-Forward) using repeated-measures ANOVA (Statistica v10, StatSoft, Inc., Tulsa, OK).  The EMG data were not normally distributed and had unequal variances; therefore a log-normal transformation of the data was applied to compare each muscle’s response across the three conditions.  For all postural and EMG variables, a Dunnett’s test was used to compare the inverted conditions to the upright relaxed condition.  Statistical significance was set at p=0.05 and data are presented as means (± standard deviation).  The number of subjects used in each comparison is indicated in Table 2-1.  2.3 Results 2.3.1 Head and neck posture Neck vertebral postures were different in both inverted conditions compared to the relaxed upright condition (Table 2-1, Figure 2-5).  Neck curvature was greater when inverted and relaxed 60  than when upright (CI = 3.0% vs. 1.4%, p<0.01, Figure 2-5).  Compared to upright, eccentricity was 14.6 mm (±15.4) more posterior in the inverted-relaxed condition and 14.1 mm (±15.3) more anterior in the inverted-forward condition (p≤0.02). C7 was flexed 5.4º (±6.1) more for the inverted-forward condition than the upright-relaxed condition (p=0.01) whereas the Frankfort plane was extended 18.1º (±12.6) more in the inverted-relaxed condition than in the upright-relaxed condition (p<0.01).   Average positions and angles at each vertebral level are summarized in Table 2-2.    Table 2-1: Kinematic and EMG result Mean (and standard deviation) values for the kinematic and EMG metrics for the subject group (N = number of subjects) for all three conditions (U-R = upright relaxed, I-R = inverted relaxed, and I-F = inverted forward).  * p-value < 0.05 compared to the upright-relaxed condition (U-R)  N U-R I-R I-F Kinematics  C.I. (%) 9 1.4 (1.0) 3.0 (1.4)* 1.3 (0.8) XECC (mm) 9 3.5 (11.2) -11.1 (12.6)* 17.6 (12.7)* ΘC7 (deg) 9 22.3 (8.3) 25.3 (6.5) 27.7 (8.2)* ΘF (deg) 10 -7.1 (5.5) -25.2 (10.5)* -6.2 (8.2) EMG (RMS %MVC)  STH 10 0.9 (0.5) 2.6 (1.6)* 7.2 (4.6)* SCM 10 0.5 (0.4) 3.0 (2.0)* 6.7 (3.6)* LS 10 2.2 (4.3) 12.6 (32.4) 7.6 (15.2) MultC4 9 1.9 (3.4) 12.7 (19.9)* 2.5 (4.9) SsCerv 10 0.8 (0.8) 6.2 (9.6)* 1.0 (0.8) SsCap 10 1.1 (1.2) 20.3 (40.5)* 1.2 (1.1) SPL 10 0.5 (0.3) 3.4 (4.0)* 1.4 (2.0) Trap 10 1.5 (1.6) 10.4 (24.0) 4.4 (5.2) 61   Figure 2-5: Kinematic results Average (solid circle) and single data points (open circles) for all subjects for the curvature index (top left), eccentricity (top right), C7 angle (bottom left), and Frankfort plane angle (bottom right) for the three static conditions: upright-relaxed (U-R), inverted-relaxed (I-R), and inverted-forward (I-F) (positive=flexion/anterior, negative=extension/posterior). *p<0.05 compared to the upright-relaxed condition. Reproduced from Newell, R.S., Blouin, J.-S., Street, J., Cripton, P.A., and Siegmund, G.P., 2013. Neck posture and muscle activity are different when upside down: A human volunteer study. Journal of Biomechanics 46 (16), 2837-2843. with permission from Elsevier [180]. 62   Table 2-2:  Vertebral translations and angles Mean (and standard deviation) values for the vertebral translations and angles for the subject group for all three conditions (U-R = upright relaxed, I-R = inverted relaxed, and I-F = inverted forward).   The vertebral translations are the mid-inferior points of C1 through C7 and are described relative to C7 (positive x = anterior, positive y = superior) and the vertebral angles (as determined by the inferior vertebral corners) are described relative to the true horizontal (positive θ = flexion).      U-R I-R I-F Translations (mm)  XC1 9.7 (10.6) -1.7 (13.1) 21.5 (12.9) XC2 12.2 (8.2) 4.8 (9.9) 21.5 (8.5) XC3 11.7 (7.1) 8 (8) 18.8 (6.7) XC4 10.6 (5.9) 9.8 (6.5) 16.2 (5.3) XC5 9 (4.3) 9.7 (4.5) 12.7 (4.1) XC6 5.9 (2.5) 6.8 (2.4) 8 (2.6) XC7 0 (0) 0 (0) 0 (0) YC1 106 (10.7) 108.3 (13.5) 106.6 (13.1) YC2 84.9 (8.7) 87.8 (8.9) 85.6 (10) YC3 67.1 (6.9) 69.7 (6.6) 67.7 (7.9) YC4 50.2 (5) 51.9 (4.7) 50.4 (5.3) YC5 33.8 (3.3) 34.7 (3) 33.7 (3.3) YC6 17.3 (1.9) 17.4 (1.6) 17.1 (1.7) YC7 0 (0) 0 (0) 0 (0) Angles (deg)  ΘC1 -14.5 (5.4) -22.7 (8.4) -3.8 (9.2) ΘC2 13.1 (9.3) 1.6 (8.3) 20.6 (9.2) ΘC3 12.5 (8.5) 1.7 (7.5) 19.8 (8.7) ΘC4 12.5 (8.8) 4.8 (7.4) 18.3 (7.2) ΘC5 14.9 (8) 10.9 (8.3) 20.1 (5.9) ΘC6 18 (7) 18.2 (7) 22.8 (5.5) ΘC7 22.3 (8.4) 25.3 (6.5) 27.7 (8.2)      63  2.3.2 Muscle activity While upright and relaxed, subjects had minimal muscle activity, ranging from a mean group response of 0.5% MVC in SPL and SCM to 2.2% MVC in LS (Table 2-1, Figure 2-6).  When subjects were inverted and relaxed, there was increased activity in six muscles (multiple p<0.05), most notably in the three deep cervical muscles: SsCap, MultC4, and SsCerv increased to 20.3% MVC, 12.7% MVC, and 6.2% MVC, respectively.  When subjects were in the inverted-forward posture, a different pattern of muscle activity emerged with increased activity in SCM and STH compared to when upright (multiple p<0.01).     Figure 2-6: Group average muscle activities Mean and standard deviation RMS EMG activity of all subjects for the 8 neck muscles normalized to % MVC for the three conditions:upright-relaxed (gray), inverted-relaxed (left, black), and inverted-forward (right, black).    *p<0.05 compared to the upright-relaxed condition.  Reproduced from Newell, R.S., Blouin, J.-S., Street, J., Cripton, P.A., and Siegmund, G.P., 2013. Neck posture and muscle activity are different when upside down: A human volunteer study. Journal of Biomechanics 46 (16), 2837-2843. with permission from Elsevier [180]. 64  Subject-specific data for vertebral displacements, vertebral angles, and muscle amplitudes can be found in Appendix B (see Table B-1 to Table B-3).  2.4 Discussion The aim of this study was to determine whether vertebral alignment and neck muscle activity are different between a relaxed upright posture and two inverted postures: relaxed and looking forward.  When inverted and relaxed, the curvature index increased and the eccentricity was more posterior than when upright.  The head center of gravity is normally slightly anterior to the occipital condyles [184], creating a static moment pulling the head into flexion when upright.  When inverted (and relaxed) this geometry should tend to extend the head, consistent with the extended Frankfort angle observed here.  When subjects were instructed to look forward (mirroring the upright Frankfort plane angle), the eccentricity moved anteriorly and the C7 vertebral angle flexed forward.  While inverted, the posterior muscles (SPL, SsCerv, SsCap, MultC4) were most active when relaxed, whereas the anterior muscles (SCM, STH) were most active when looking forward.  Since the inverted tasks were not randomized, we cannot rule out habituation as a partial explanation for the differences in the posterior muscle activities; however, these patterns of muscle activity are consistent with a task requiring head flexion.  These changes in posture and EMG activations indicate that the in vivo alignment of the neck vertebrae and muscles in inverted subjects differs considerably from that of upright.   A common neck injury mechanism in rollovers is axial loading from head contact with the vehicle roof [14], often when the vehicle is upside-down [9, 47].  Previous axial impact tests of cadaver necks, however, have used an upright configuration [18, 19, 39], a prone/supine 65  configuration [44, 45] or an inverted configuration with the neck positioned in an upright, neutral posture [28, 36].  Although valuable to understanding axial loading injury mechanisms, these cadaver tests do not fully represent the spectrum of neck injuries observed in real-life rollovers [11, 12].  Our data suggest that the inverted neck differs from the upright neck in curvature (I-R), head angle (I-R), eccentricity (I-R and I-F), and C7 angle (I-F).  These differences with inversion may be important given that prior cadaver studies have shown that injury mechanisms depend on neck posture.    Initial head and neck posture may also be important when designing injury prevention devices for rollover crashes.  For example, one proposed car roof design attempts to reduce neck injury risk by moving the head and neck forward at roof contact during a rollover crash [67].  A novel seat-mounted airbag also attempts to flex the head and neck forward [93]. Our data suggest that changes in neck posture due to inversion should also be considered when studying neck injury or prevention devices for rollovers.   Neck muscles have generally been ignored in cadaver tests [24, 28, 41] and computational models [65-68] used to study axial neck injury and prevention.  The assumption has been that there is little or no muscle activation prior to impact and that reflex muscle activation would occur after the injury.  However, during a rollover an occupant may have sufficient time to anticipate and react to an impending head impact.  Indeed, we observed changes in muscle activity in response to inversion alone.  This active muscle contraction likely preloads the inverted and posturally susceptible neck, thus influencing spinal column stiffness and the compressive forces imparted at impact.  Hu et al. [51] simulated rollover-type impacts with a 66  neuromuscular neck model and suggested that maximally active muscles increase the risk of neck fracture.  This model included 23 pairs of neck muscles, where the active muscle state was modeled as 100% of maximal activation and was not based on in vivo data.  Although the muscle activations recorded in the present study were below 100%, our findings suggest that it may be important to include neck muscles in head-first injury models.   Neck muscle activation levels were not only different when upright and inverted, they were considerably different between subjects.  Between-subjects variability has been reported previously for neck muscles such as the splenius capitis, particularly for a novel task [171, 174, 176].  Although inverted subjects were instructed to relax (I-R) and to look forward (I-F), many subjects found being inverted uncomfortable; thus the large range of activation levels may reflect a lack of familiarity with being inverted.  Sizable between-subjects variability was also observed for the postural metrics, but similar variability was observed in the upright and inverted postures.  Similar standard deviations have been reported previously for an upright automotive seated posture: ±7° for the Frankfort plane and ±1.6% for the curvature index [159].  Knowledge of the range of postures while inverted is critical for designing and testing neck injury prevention devices.  Previous cadaver research has suggested that even small changes in eccentricity of the spine or loading vector, on the order of ±10 mm, can change the injury mechanism [18, 41].  Our eccentricity measurements had standard deviations up to 12.7 mm within a single posture. This between-subjects variability suggests that testing a limited number of conditions (i.e. only one eccentricity) may not simulate the response of different people.    67  These experimental observations are in keeping with observations in clinical practice, in which the heterogeneity of injury patterns continues to preclude standardized classification.  Despite apparent mechanistic similarities in rollovers, the patterns of vertebral column and spinal cord injury are widely varied.  In clinical practice, spine surgeons have attempted to standardize descriptive nomenclature and treatment algorithms for both atlanto-axial and sub-axial cervical spine injuries [32, 33]. The heterogeneity of injury patterns and a lack of understanding of neck mechanics at impact have largely frustrated these efforts.  We believe that the findings of this study will enhance clinicians’ appreciation of injury mechanics, which is critical to developing more cohesive and standardized surgical treatment algorithms.  This study was conducted using static inverted postures, and their direct applicability to the wide variety of dynamic conditions of a rollover is not yet known.  Images of volunteers in steady-state rolls indicate that occupants may resist the centripetal acceleration and maintain a forward gaze [74, 84, 88] (similar to the I-F condition).  However, occupants may also have an active response to the lack of headroom (i.e. “ducking”) [89] or neck flexion induced by the seatbelt [90], and these responses were not simulated in the current study.  Rollovers involve complex ballistic motions, often with 3-4 g of centripetal acceleration [87], whereby the occupant’s neck  may be compressed by the vehicle roof [92].  The aim of the current study was to perturb and compare the neck response from compression to tension induced by gravity only, perhaps simulating lower roll rates [91].  Further work is needed under dynamic conditions to better understand the occupant response in more realistic rollover environments.  In spite of the limitations, these results provide a more realistic in vivo data set that can be used to improve and validate current ex vivo and computational models for studying rollovers.   68  In summary, compared to an upright relaxed condition, subjects increased their EMG activity and changed their vertebral posture when inverted.  The exact pattern of changes depended on whether the subjects were relaxed or looking forward while inverted.  In general, the inverted spine was more curved when relaxed and tended to be more flexed when looking forward.  These data suggest that there may be muscle activity and a realignment of the cervical vertebrae prior to an inverted head-first impact in a real-world rollover crash.  These differences with inversion, along with the between-subjects variability observed, mirror clinical observations of heterogeneous cervical spine injury patterns and offer new avenues of research in injury prevention and treatment.    69  Chapter  3: Does bracing for impact result in a realignment of the cervical spine? 1   3.1 Introduction A common catastrophic neck injury mechanism in rollovers is axial loading from a head impact with the roof of the car [11, 14].  Many of these injuries are thought to occur when the vehicle is inverted [9, 47] and the occupant ‘dives’ head-first into the vehicle roof.  Attempts to recreate these neck injuries in cadaver experiments have produced some – but not all – of the injuries seen in real-world rollover crashes [11, 13].  For instance, rollovers tend to have a higher proportion of C5 to C7 fractures (65.2% of all vertebral fractures) than seen in cadaver testing (39% of fractures) [11, 12].  This disparity between the lab and real-world injuries may stem from simplifying assumptions regarding neck posture and pre-impact muscle activity [17]; consequently, in vivo experiments are needed to elucidate the state of the neck prior to neck loading.  Foster et al. [11] noted that certain cervical spine injuries, particularly bilateral facet dislocations, are common in real-world rollovers but are difficult to replicate in ex vivo experiments.  They proposed that pre-impact muscular bracing could preload the spine and help explain this discrepancy.   Muscle-related loading of any sort has largely been ignored in head-first impact   1 Chapter 3 has been accepted for publication.  Newell RS, Siegmund, GP, Blouin J-S, Street J, Cripton PA. Cervical vertebral realignment when voluntarily adopting a protective neck posture Spine 2014 (in press).  70  cadaver testing because reflex latencies are many times longer than the time from head contact to neck injury [46].  During a rollover, however, there is sufficient time between the onset of a roll  and first roof contact to activate the neck muscles [76, 86].  Computer simulations have shown that maximal muscle activation before axial head impacts increases the risk of cervical spine fractures [51], although the level of muscle activation during actual bracing is currently unknown.  Furthermore, neck alignment at head impact is known to influence the type of neck injury, yet neck posture changes due to bracing are yet to be quantified.  Thus, to better understand and ultimately prevent neck injury in rollovers, it is important to understand how muscle activity changes the neck posture and initial loading state.    The aims of this study were to measure neck muscle activity and differential vertebral alignment between the relaxed and tensed (i.e. braced) states in both upright and inverted conditions.  We hypothesized that both bracing and inversion would alter muscle activation levels, cervical spine curvature, and horizontal neck position.  3.2 Methods 3.2.1 Subjects Eleven subjects (6 females and 5 males) between 19 and 45 years old with no history of neck, muscle, nerve or balance problems participated in this study.  Average height, weight, neck circumference and head circumference were 172.9 ± 6.7 cm, 70.0 ± 15.7 kg, 34.3 ± 4.2 cm and 56.4 ± 3.2 cm, respectively.  One subject declined the fluoroscopy portion of the experiment and another subject declined the EMG portion.  The study was approved by the University of British Columbia’s Clinical Research Ethics Board and all subjects gave their written informed consent. 71   Subjects were seated in a custom-built inversion device, secured with a 5-point harness, and held in upright (U) and inverted (I) postures while neck muscle activity and fluoroscopy were recorded.  In both orientations, subjects were instructed to tense their neck and shoulder muscles to stiffen and simultaneously draw in their neck to simulate a protective response in preparation for a head-first impact.  During each trial, subjects started relaxed (Initial) and proceeded to tense and then sustain the contraction (Tensed), resulting in two states for each orientation (U-Initial, U-Tensed, I-Initial, and I-Tensed) (Figure 3-1 and Figure 3-2).    72   Figure 3-1: Exemplar EMG data. Exemplar EMG (SCM, Mult, and LS) from a single subject with muscle contractions beginning at 0.5 s (gray shaded region).  The mid-point of the 200 ms resting EMG window (black box) was used to calculate the RMS % MVC of the Inverted-Initial state (Initial, 0.15-0.35 s) and the 200 ms of sustained muscle contraction (black box) was used to calculate the RMS % MVC of the Inverted-Tensed state (Tensed, 1.9-2.1 s).  73   Figure 3-2: Exemplar data of vertebral motions.   The translations of the mid-inferior point of C7 and mid-superior point of C1 were measured for both the Initial (black) and Tensed (gray) states (positive x = anterior, positive y = superior).  Statistical analyses were performed on the horizontal displacements (motion in the x-direction).     3.2.2 Electromyography EMG activity was measured using surface electrodes for the sternohyoid (STH) and indwelling electrodes for the sternocleidomastoid (SCM), trapezius (Trap), levator scapulae (LS), splenius capitis (SPL), semispinalis capitis (SsCap), semispinalis cervicis (SsCerv), and multifidus (MultC4) muscles on the left side only.  Surface electrodes (Ambu Blue Sensor, Ambu A/S, Ballerup, Denmark) were placed superficial to the left sternohyoid (STH) muscle.  Indwelling electrodes (PFA-coated Stainless Steel, 0.0055” diameter wire; A-M Systems, Inc., Sequim, WA) were inserted under ultrasound guidance into the center of each muscle belly at the C5/C6 74  level (Trap) and C4/C5 (remaining muscles) [174].  EMG signals were band-pass filtered (wire: 50-1000 Hz; surface: 30-1000 Hz) and acquired at 2000 Hz. For each trial, the root mean square (RMS) of each signal was calculated for a 200 ms window, centered on the fluoroscopic images (Figure 3-1), corresponding to steady-state equilibrium of the neck (defined by less than 1.5 mm and 1 degree of head motion over 200 ms).  The first window captured the muscle activity during a relaxed (Initial) state, prior to the initiation of the muscle contractions (active bracing started at 0.5 seconds into the trial).  The second window captured the sustained contraction (Tensed), at a time point during the muscle contractions where fluctuations in neck posture were minimal (about 1.5 s after initiation of muscle contraction).  The activation of each muscle was normalized by the maximum RMS activity observed during 3 s maximum voluntary isometric contractions (MVCs) from the neutral posture into flexion (0°), extension (180°), left lateral bending (90°), anterolateral bending (45°), posterolateral bending (135°), and bilateral axial rotations.    3.2.3 Cervical spine alignment A fluoroscopic C-arm (OEC 9400, GE, Salt Lake City, UT) was used to capture sagittal plane images of the cervical vertebrae at 30 Hz.  Each video sequence (containing both initial and tensed states) was 2.5 s long and was repeated twice in both orientations (upright and inverted), resulting in 10 s of total fluoroscopy.  The motions of the vertebral bodies (and anatomical landmarks) were analyzed using a custom tracking algorithm [157] and the repetition with the clearest fluoroscopy images were chosen for analysis.  Two single images were selected from each video sequence (corresponding to the EMG windows), representing the Initial and Tensed postures under steady-state equilibrium. 75  Changes in cervical spine alignment were quantified using three metrics: a curvature index (CI) and the horizontal position of the cephalad (C1x) and caudal vertebrae (C7x).  The four main corners of each vertebral body were identified as anatomical landmarks and the mid-points were located between the two superior corners and the two inferior corners of each vertebra (with the exception of C2 where the two inferior corners and the tip of the dens were identified).  CI is the ratio of the arc length connecting the superior and inferior midpoints of all seven vertebral body to the chord connecting the superior midpoint of C1 to the inferior midpoint of C7 [159]. A higher CI indicates increased curvature.  Vertebral body translations were analyzed for the superior midpoint of C1 and inferior midpoint of C7 in both the initial and tensed states (Figure 3-2) and were measured perpendicular to the direction of gravity (positive = anterior).  Torso position was assumed to stay constant across trials (the torso was constrained with the 5-point harness).  Head orientation in the sagittal plane was measured relative to the true horizontal using an array of six 2-mm beads that were strapped to the head and projected into the fluoroscope’s field of view.  The Frankfort plane (ΘF) of the head is defined by the right and left tragus and the mid-point of the right and left inferior orbits (positive = extension) and was measured relative to the true horizontal.  The location of the beads relative to the Frankfort plane was measured and thus bead motion in the X-ray enabled head orientation to be tracked throughout a trial.   3.2.4 Data processing notes and statistical analysis One subject flexed their head forward when tensed whereas all other subjects maintained similar Frankfort plane angles when tensed.  This subject was deemed an outlier and removed from the 76  statistical analysis and group averages, but included in the individual data sets.  C1 was not fully captured in two subjects; thus, a linear regression was used to predict the horizontal location of the mid-superior points of C1 and C2 and data from the other subjects (C1 = 0.98*C2 – 5.91 mm; R2 = 0.99).  One subject was missing C7 in the Inverted-Tensed condition; therefore, this subject was excluded from the statistical analysis for C7x and CI (C1x and ΘF were available).  One subject was missing a recording for MultC4 for all conditions.  The number of subjects used in each statistical comparison is shown in Table 3-1 and Table 3-2.  Separate two-way repeated-measures ANOVAs (Statistica v10, StatSoft, Inc., Tulsa, OK) were used to compare postural and muscle variables: the main effects were tensed (Initial, Tensed) and orientation (Upright, Inverted).  A log-normal transformation was performed on the EMG data since they were not normally distributed and had unequal variances.  For all significant interactions, post-hoc Bonferroni tests were used to compare across the Initial versus Tensed states: U-Initial versus U-tensed and I-Initial versus I-Tensed.  Between-subjects variability was quantified using 95% confidence intervals about the mean (back transformed for presentation of the EMG data).  3.3 Results 3.3.1 Muscle activity Activity levels of all muscles were significantly higher when tensed than their initial state, regardless of orientation (Table 3-1 and Figure 3-3).  Trap and SPL increased the most when tensed and inverted (mean increases of 36.2% and 22.7%, respectively) and MultC4 increased the least (mean increases of 4.3% while upright and 6.3% while inverted).  The between-subjects 77  variability in each muscle’s activation level was large, with confidence intervals exceeding 20% in many conditions and reaching as high as 68% in Trap (Inverted-Tensed) (Figure 3-3).    Table 3-1:  P-values for EMG metrics Number of subjects (N) and p-values for the EMG metrics for the main factors (Orientation = Upright versus Inverted; Tensed = Initial versus Tensed), and interaction effects.  The post-hoc comparisons of Upright-Initial (U-I) versus Upright-Tensed (U-T) and Inverted-Initial (I-I) versus Inverted-Tensed (I-T) are reported for significant interactions (bold = significant). EMG N Main and Interaction Effects Post-hoc Comparisons Orientation Tensed Interaction U-I vs U-T I-I vs I-T STH 9 0.00 0.00 0.02 0.00 0.00 SCM 9 0.00 0.00 0.00 0.00 0.00 LS 9 0.01 0.00 0.36 - - MultC4 8 0.54 0.02 0.62 - - SsCerv 9 0.11 0.00 0.55 - - SsCap 9 0.18 0.00 0.48 - - SPL 9 0.01 0.00 0.94 - - Trap 9 0.00 0.00 0.95 - -    78   Figure 3-3:  Initial and tensed group mean EMG The group mean (±95% confidence intervals) RMS values (normalized to % MVC) of the EMG activities for all 8 muscles in the initial (black) and tensed (gray) states for the upright (left) and inverted (right) conditions.   3.3.2 Kinematic results The horizontal positions of C1 and C7 were influenced by tensing; however, the effect depended on orientation (see interactions in Table 3-2).  C1x was 10.6 mm more anterior when tensed than when relaxed in the inverted condition (Figure 3-4); whereas, there was no significant change when upright. During tensing, C7 moved anteriorly in both orientations but with different magnitudes: 5.6 mm upright and 9.5 mm inverted (Figure 3-4).  Tensing also increased the curvature index by 1.1% (Figure 3-5); yet, there was no difference in the Frankfort plane angle 79  with inversion or with tensing (Table 3-2).   As with the muscle response, there was considerable subject variation in the vertebral alignment (see confidence intervals for the vertebral position, Figure 3-4).    Table 3-2:  P-values for kinematic metrics Number of subjects (N) and p-values for the kinematic metrics for the main factors (Orientation = Upright versus Inverted; Tensed = Initial versus Tensed), and interaction effects.  The post-hoc comparisons of Upright-Initial (U-I) versus Upright-Tensed (U-T) and Inverted-Initial (I-I) versus Inverted-Tensed (I-T) are reported for significant interactions (bold = significant). Kinematics N Main and Interaction Effects Post-hoc Comparisons Orientation Tensed Interaction U-I vs U-T I-I vs I-T C1x 9 0.04 0.01 0.03 1.00 0.01 C7x 8 0.28 0.04 0.02 0.00 0.00 CI 8 0.19 0.03 0.82 - - ΘF 9 0.31 0.31 0.17 - - C1x = horizontal displacement of C1, C7x = horizontal displacement of C7, CI = curvature index, ΘF = Frankfort plane angle    80   Figure 3-4: Initial and tensed group mean vertebral coordinates  Group average (±95% confidence intervals) of the X-Y coordinates of the vertebrae (mid-superior point of C1 and mid-inferior points of C1-C7 in descending order) for the initial (black) and tensed (gray) states for both upright (left) and inverted (right).  For visualization, the ‘initial’ C7 mid-inferior point is the origin (0,0).  81   Figure 3-5:  Initial and tensed curvature indices Individual subject (open circles) and group mean (closed circles) of the curvature index for the initial and tensed states for the upright (left) and inverted (right) conditions.  C7 was missing for I-Tensed for one subject; thus, the data for this subject (diamonds) was omitted from the group averages.                                                                                           Individual subject muscle levels and vertebral motions are presented in Appendix C (see Figure C-1 and Figure C-2).    82  3.4 Discussion We observed that neck muscle tensing, mimicking a protective drawing-in of the head, caused significantly greater curvature and (orientation-dependent) anterior motion of the cervical spine.  Moreover, we observed these changes with muscle activation levels averaging about 10% MVC when upright and about 20% MVC when inverted. These inversion and muscle-induced changes in vertebral alignment indicate that the in vivo state of the neck may differ considerably from its initial alignment prior to a forewarned head impact such as can occur during a rollover crash.  Cadaver experiments (which lack active musculature) tend to report high rates of ligamentous and disc injuries (not commonly identified in rollover occupants) and vertebral body fractures, whereas rollover occupants have a high proportion of shear type bony injuries (i.e. facet fractures) [11, 12].  Computational modeling has suggested that neck musculature can affect the load sharing between the muscles and ligamentous spine [112] and has the potential to protect the cervical ligaments from injury [56]; thus, the lack of musculature in cadaveric testing could help explain disparities in mechanistic spine loading.  Cables have been used to simulate musculature in cadaver tests, but the selection of muscles and the applied force levels to simulate pre-impact awareness or bracing were not based on in vivo data [35, 103, 104].  Specimens constrained by passive muscles exhibit higher rates and severities of injury [35], and compressive preloads affect the loading biomechanics of the spine [100].  In contrast, computational studies have reported that active neck muscles can also change axial impact injury mechanics, although the active muscle state was modeled as maximally active [51, 54].  When asked to stiffen their neck to simulate preparation for a head-first impact, our subjects activated their neck muscles well below maximum.  Moreover, although the eight muscles significantly 83  increased, the activation levels were not consistent across all muscles: the more superficial muscles had a more pronounced increase in activity in the tensed condition, whereas the deeper muscles had activations under 15% MVC.  Consequently, the actual muscle activity present during pre-impact bracing may be more accurately represented as somewhere between passive and fully active.    Chancey, et al. [52] modeled a more realistic muscle activation scheme in a pre-injury tensed state using optimization techniques.  Although this model evaluated tensile neck injury, their tensed activation predicted the neck muscles could generate compressive preloads as high as 800-1400 N throughout the cervical spine [52].  Their muscle activation patterns were similar to those of the eight muscles measured herein, albeit their average magnitudes were two (I-T) and four (U-T) times higher than the levels observed in the current study.  Adjusting for our average decrease in activation magnitudes compared to Chancey et al. [52], we estimate the total overall tensed muscle force to be approximately 425 N upright and 850 N inverted.  While we do not know whether our subjects’ contraction levels represent those that would occur during an actual rollover, our data, coupled with the data from Chancey et al. [52], suggest that sub-maximal and muscle-specific control schemes may more accurately represent ‘tensed’ muscle states in vivo.    The ability of the neck muscles to alter the alignment of the cervical spine before a head-first impact has largely been overlooked in cadaver drop tests [11], and this oversight may explain why some rollover neck injury patterns are underrepresented in cadaver tests.  When inverted and tensed, our subjects had an average of 20 mm of horizontal displacement at the top of the neck and 6 mm of initial eccentricity (C1 relative to C7, upright), shifting the line of action of the 84  head impact force forward an average of 26 mm relative to the line of action of the torso’s inertia.  Given that eccentricities (occipital condyles to T1) of -5 mm, 1 mm, 23 mm, and 53 mm have been reported to result in compression-extension, vertical compression, compression-flexion, and hyperflexion injuries, respectively [19, 20], this anterior vertebral shift due to bracing could change the type of injury caused by a head-first impact relative to that expected for no bracing.    A high proportion of cervical spine injuries observed in rollovers are flexion or flexion-compression injuries [11, 49]  and bilateral facet dislocations [17]. The latter are thought to require a larger flexion moment than injuries that do not involve facet dislocation [17].  Experimentally, bilateral facet dislocations have been generated in cadaveric spines with rotational constraint to the top of the spine [30], with stiffening of the lower spine [31], and with head protrusion [129].  If a rollover occupant was to brace for impact, pulling their neck forward relative to the torso, and the head were to pocket in a structure in the car (i.e. the roof), this could potentially create a constraint on the top of the neck that has a forward alignment.  The accompanying neck muscle activation could also stiffen the lower spine or provide rotational restraint to the upper spine.  This ‘protrusion’ of the muscle-stiffened spine might be an important mechanical change to the neck and offers motivation for exploring the role of forward translation of the neck (without head flexion) in shearing and flexion-moment type injuries observed in rollovers.  Cadaveric and computational models of head-first impact are rarely tested in the tensed configuration we observed (neck pulled forward, curvature increased, compressive load on the spine, and head angle constant).  Therefore, current neck injury models may not represent the loading characteristics present when someone braces during a rollover. 85  Between-subjects variation is generally not addressed in modeling neck injury or simulating injury prevention strategies; yet, this variability is critical to studying and preventing neck injury.  Although our subjects increased the curvature of their spine on average, two subjects straightened their spine during tensing.  The consequences and relevance of these curvature variations are not clear as curvature indices have not been documented in previous cadaver testing.  These average changes in the postural and muscle responses with bracing may be useful for determining pre-impact alignment of a cadaveric test; however, each individual’s response is important to consider.  Injurious loading of the spine is complex and depends on many postural and muscular factors and there were no clear patterns of motor strategies that could explain the postural findings of each subject.  Although it is unclear what distribution of occupants actively reacts in real-world rollovers, this varied response of the muscles and spine curvature in the tensed state highlights the variability with which people may respond to an injury scenario.      For this study, subjects were instructed to tense their neck and shoulder muscles to stiffen and simultaneously draw in their neck to simulate preparation for a head-first impact.  Each subject’s response depended on their interpretation of the verbal instructions. The responses were simulated without any actual threat and were performed in a quasi-static environment; thus, these results should be interpreted carefully.  It is not clear whether an occupant can elicit this coordinated “tensed” response or if the magnitude of these changes would be influential in a rollover environment.  The aim of this study was to capture the posture and muscle state in a condition that may exist immediately before a headfirst diving impact and future experiments are needed to understand what further cervical alignment changes (if any) occur in a dynamic rollover simulation.  Nonetheless, the application of these findings extends beyond the context of 86  a rollover as it provides evidence that an active response of tensing the neck muscles is capable of generating sizable changes in cervical spine alignment that are potentially relevant to catastrophic neck injury mechanics.     In summary, the muscle-specific activity levels in this human subjects study – albeit sub-maximal – were substantial and resulted in measurable changes in cervical spine alignment.  The mean tensed response was to shift the cervical spine forward relative to the torso and to increase the cervical spine curvature.  These results indicate that tensing the neck muscles can change the alignment of the spine and potentially alter the loads experienced by the neck during an axial head impact.      87  Chapter  4: Does exerting a flexion and extension force (with head constraint) alter the alignment of the spine?  4.1  Introduction Cadaver testing has been key to understanding the injury thresholds and vulnerable spine postures associated with axial impact neck injury [18-20, 27, 30, 37, 39, 41, 44-46, 97, 129]. To visualize the vertebral column during injury, ex vivo models are often osteoligamentous.  The osteoligamentous cadaver cervical spine is inherently unstable and, without artificial muscle support, it is only able to support approximately 10 N (less than the weight of the head) in compression before buckling [111].  Researchers have dealt with this problem by applying direct support forces to the head [18, 44, 46, 97, 129] or simulating overall muscular compressive forces (i.e. a follower load) [100, 108].  However, these approaches can create non-physiological forces in the spine [106, 130] and do not represent the complex neuromuscular control and actuation systems that exist in vivo.  Rigid test fixtures [37, 41] have been used to manipulate the overall neck angle of ex vivo cervical specimens without regard to how the curve of the spine may respond to in vivo muscle forces.  Although neck muscles have the ability to stiffen and stabilize the cervical column [52, 57, 102, 105, 110], the extent of vertebral realignment (i.e. changes to the overall cervical curvature and individual intersegmental angles) due to neck muscle activation in vivo remains unclear.    Simulating musculature is critical to both understanding and ultimately preventing neck injuries; however, simulating in vivo muscle forces is difficult for two main reasons. First, with over 20 88  pairs of muscles, the neck is functionally redundant and the same net force can be achieved with different combinations of muscles. Thus the pattern of muscle activation, and therefore the load state in the neck, cannot be deduced simply by knowing the net force output.  Second, the cervical alignment changes that occur with neck muscle activation may vary depending on the specific pattern of neck muscles used to generate a given amplitude and direction of force, particularly in different orientations (e.g. upright and inverted) relevant to axial neck injury [180]. Therefore, in vivo data are needed to characterize the pattern and amplitude of muscle activation during specific tasks.  Intersegmental stiffness influences the buckling mechanics of the neck under axial loading [50, 185]; therefore, simulating pre-impact joint stiffness relevant to real-life injury scenarios (both in cadaver and computational models) is critical to studying and preventing neck injuries.  Intervertebral stiffness of the cervical spine is very low near the neutral posture, for which each level of the spine is within the so-called neutral zone.  Within this zone the spine offers very little resistance to sagittal plane rotation [186, 187].  Applying large anterior or posterior forces to a cadaver spine would cause it to bend significantly.  Yet this is not necessarily the case in the in vivo spine where the muscles can provide the intersegmental stability required to maintain overall neck posture.  Quantifying the global and intervertebral motions under muscle contractions in an isometric task may help to elucidate the magnitude of the effective stiffness and stability that muscles are able to provide to the osteoligamentous spinal column.    The aim of this study was to measure the muscle activation patterns and resulting realignment of the cervical spine during neck muscle contraction tasks in the neutral posture.  The specific 89  hypotheses were that actively contracting the neck muscles to produce flexion and extension forces with the head (in the neutral posture) would result in realignment of the cervical spine (changes to the overall cervical curvature and individual intersegmental angles), and that muscle activation strategies and realignment patterns would be orientation-dependent (i.e. upright versus upside-down).  4.2  Methods 4.2.1 Subjects Eleven human subjects (6 females and 5 males) participated in this study and the average (± standard deviation) age was 33 ± 7 years.  Subjects were excluded if they had a history of neck injury or pain, or known muscle, nerve, or balance problems.  The average group height was 172.9 ± 6.7 cm, their weight was 70.0 ± 15.7 kg, and their neck and head circumferences were 34.3 ± 4.2 cm and 56.4 ± 3.2 cm, respectively.  One subject did not participate in the fluoroscopy portion of the experiment and another subject did not participate in the EMG portion.  Subjects gave their written informed consent and the study was approved by the University of British Columbia Clinical Research Ethics Board.  4.2.2 Conditions All subjects performed controlled neck muscle contractions to produce target forces in flexion (pushing forward with the head) and extension (pushing backward with the head) in two orientations (upright and inverted) while fluoroscopy of the cervical vertebrae and electromyography of the neck muscles were recorded.  For the force generation trials subjects were statically held in the seated position using a custom inversion device and a 5-point harness.  90  Subjects performed muscle contractions against resistance to 50% of their maximum force obtained in maximum voluntary contraction (MVC) trials (i.e. 50% of the maximum extension and 50% of the maximum flexion forces).  To achieve these 50% target forces, subjects wore a tight-fitting headband that was attached via a chain to the rigid portion of the inversion device.  The chain was attached to the front of the headband that was placed just above the eyes for the extension force, and attached to the back of the headband that was placed just above the external occipital protuberance for the flexion force.  A load cell (Sensotec Model 31, 250lb full scale, Sensotec Inc., Columbus, OH) was connected to the chain to measure force (see Figure 4-1).  A monitor placed directly in front of the subject provided visual feedback of, and a target for, the applied force (see Figure 4-1).  Subjects were instructed to follow a target that ramped linearly to 50% of their MVC over 5 seconds and then to hold that force for one second. The displayed target was circular with a ±10 N tolerance. Fluoroscopy and EMG recordings were analyzed in the relaxed (Initial) and 50% MVC (Tensed) conditions within each trial. Two trials were acquired for each combination of orientation (U-upright, I-inverted) and force direction (flexion, extension).  Trial order was randomized.    91   Figure 4-1: The 50% MVC force task Subject secured in the inversion device (wearing a CT apron to reduce radiation exposure) and following the target forces displayed on the monitor (top).  Image of the chain attached to the back of the head (to generate a flexion force) with the load cell mounted between the head and the frame (bottom).  The C-arm (bottom) is positioned to capture the changes in cervical alignment.    4.2.3 Electromyography A combination of surface (Ambu Blue Sensor, Ambu A/S, Ballerup, Denmark) and indwelling (PFA-coated Stainless Steel, 0.0055” diameter wire; A-M Systems, Inc., Sequim, WA) electrodes were used to measure EMG activity for eight neck muscles: sternohyoid (STH), 92  sternocleidomastoid (SCM), trapezius (Trap), levator scapulae (LS), splenius capitis (SPL), semispinalis capitis (SsCap), semispinalis cervicis (SsCerv), and multifidus (MultC4) muscles.  Indwelling electrodes were inserted under ultrasound guidance into the center of each muscle belly at the C5/C6 level (Trap) and C4/C5 level (remaining muscles) [174].   EMG signals were band-pass filtered (indwelling: 50-1000 Hz; surface: 30-1000 Hz), acquired at 2000 Hz, and normalized to maximum voluntary contractions (MVCs).  For each target force trial, the root mean square (RMS) of each EMG signal was calculated over two 200 ms windows per trial. The first window captured the muscle activity during a relaxed (Initial) state (before the onset of the contraction) and the second window captured the sustained target force contraction (Tensed) (Figure 4-2). Within the available data, the window location was selected to minimize motion (clear fluoroscopic images) and force fluctuations. A trigger pulse (emitted by the image acquisition card) enabled the fluoroscopic image of interest to be synchronized with the middle of the 200 ms windows.   93   Figure 4-2: Exemplar data flexion force task Exemplar data (subject #11, upright orientation) for the initial and tensed states for the flexion contraction forces, with an example EMG (SCM) signal (µV, middle) and the force (N) that was slowly ramped and held at a target force of 50% of the MVC flexion force (bottom).  The 200 ms windows used to calculate the RMS EMG are highlighted in gray (the first window was selected before any force occurred and the second during the sustained target force). Fluoroscopy was recorded from 0 sec to 6 sec; the two fluoroscopic images were selected to correspond with the point of no force (at 0 sec) (Initial) and during the sustained target force (at 6 sec) (Tensed).  The outlined vertebral bodies are illustrated at these two time points. 94  4.2.4 Maximum voluntary contractions For the MVC trials, seated subjects were secured to a rigid backboard while wearing a snug skateboard helmet (Pro-tec, San Diego, CA).  The helmet was attached to a 6-axis load cell (45E15A4-U760, JR3, Inc., Woodland, CA, nominal horizontal accuracy: ±2.5N) above the subject’s head (see Figure 4-3).  Isometric MVCs were performed with verbal encouragements [183] and real-time visual feedback of force/moment magnitude and direction.  With a neutral head posture, MVCs were executed in seven directions: flexion, extension, left lateral bending, two 45° oblique combinations (flexion/left lateral bending, extension/left lateral bending), and right and left axial rotations.  Each contraction was three seconds long and repeated twice. A 200 ms window, centered on the maximum force that was exerted, was used to calculate the RMS for each muscle’s EMG in the seven directions.  The maximum RMS value for each muscle, regardless of direction, was used for normalization.  The maximum forces in the flexion and extension directions were recorded, and provided target forces for the 50% MVC contraction trials.    The MVCs were conducted in a constrained manner (i.e. limited head motion in all directions, Figure 4-3) to provide a basis for eliciting maximum contractions of each muscle; however, the target task was performed in a less constrained manner.  Because the forces were generated through a chain, small motions were permitted in the anterior/posterior direction (Figure 4-1).  From (unpublished) pilot work, when people were turned upside-down they came slightly off the seat and the distance between the seat and the top of the head tended to increase (presumably because the entire spine is under tension and there is some slack in the 5-point harness).  To accommodate this vertical motion, we selected a less constrained condition for the target force 95  tasks because we did not want a rigid mount to influence (or restrict) the spinal realignment, by placing the neck in compression while upside-down.      Figure 4-3: MVC task A subject seated and secured while generating MVCs with the skateboard helmet on, which is rigidly attached to the load cell above.    4.2.5 Cervical spine posture Fluoroscopic images of the sagittal plane of the cervical spine were recorded at 30 Hz.  Each video sequence (containing both initial and tensed states) was 6 s long, thus each subject was exposed to a total of 48 s of fluoroscopy (Table 4-1) (receiving an average effective dose of 0.025 mSv).  The vertebral bodies were identified in the fluoroscopic images, and two single 96  images were selected from each video sequence (corresponding to the EMG widows), representing the Initial and Tensed postures (Figure 4-2).   Table 4-1:  Flexion and extension force tasks test matrix The time, number of repetitions, and total fluoroscopy exposure time for the flexion and extension force tasks.  Task Position Time per trial (sec) Number of repetitions Total fluoroscopy exposure time (sec) Flexion force Upright 6 2 12 Inverted 6 2 12 Extension force Upright 6 2 12 Inverted 6 2 12 Total Exposure Time (sec)    48   The curvature index (C.I.) was measured to provide an overall quantification of cervical realignment.  The curvature index was defined as the ratio of the arc (a straight line between the mid-superior point of C1 and the mid-inferior point of C7) and chord (a line passing through the mid-superior and mid-inferior points of each vertebral body) lengths of the cervical spine [159] (Figure 4-4).  A higher curvature index indicated an increased curvature.  To evaluate the magnitude of neck stiffness provided by the muscles while producing the target forces, the intervertebral angular motions at each cervical vertebral level (e.g. ΘC4/C5) were also measured.  The angle of each vertebral body was defined as a line between the two inferior corners of the vertebral body.  The absolute angles (i.e. ΘC7) were measured relative to the true horizontal (positive = extension) and the intervertebral angles were measured relative to the adjacent 97  vertebra (Figure 4-4).  The cervical eccentricity was measured by the horizontal distance between the mid-inferior point of C7 and the mid-superior point of C1 (Figure 4-4) and was measured to verify that the cervical spine remained in the near neutral position.      Figure 4-4: Postural metrics The curvature index (C.I., left), the eccentricity (XECC, center), the C7 angle (ΘC7, right), and an example intervertebral angle (ΘC4/C5, right).  4.2.6 Head orientation Head orientation (position and angle) were measured to verify that the head remained in a near neutral position.  Head orientation in the sagittal plane was measured relative to the true horizontal using an array of six 2-mm diameter beads that were strapped to the head and projected into the fluoroscope’s field of view.  Prior to data collection, a FaroArm (B08-02, Lake Mary, FL) was used to digitize the location of the beads relative to anatomical landmarks on the head.  The Frankfort plane (ΘF) of the head was defined by the right and left tragus and the mid-point of the right and left inferior orbits (positive = extension) and was measured relative to the 98  true horizontal.  The head center of mass (COM) was defined as 3.13 cm superior to the Frankfort plane and 0.83 cm anterior to the mid-point of the right and left tragus [188].  A direct linear transformation was used to relate the 3D position of the bead array and the head anatomical landmarks with the 2D X-ray image plane (see Appendix D for details).  Because the bead array was fastened to the head, they moved together as a rigid body. Since the bead array was visible in the fluoroscope images, the orientation of the Frankfort plane and the x and y positions of the head COM could be calculated for each image even though the head itself was not in the image.   4.2.7 Data processing notes C1 was not fully captured in two subjects while inverted; thus, a linear regression was used to predict the horizontal location of the mid-superior points of C1 and C2 and data from the other subjects (C1 = 0.98*C2 – 5.91 mm; R2 = 0.99); however, the C1 angle was missing. C7 was occluded for one subject for the upright flexion trial; therefore, eccentricity and the C7 angle were missing.  The activity for MultC4 was missing for one subject for the inverted flexion trial.  The number of subjects in each statistical comparison is shown in Table 4-2  and Table 4-3.  4.2.8 Data and statistical analysis Separate two-way repeated-measures ANOVAs (Statistica v10, StatSoft, Inc., Tulsa, OK) were used to compare postural (curvature and intersegmental angles) and muscle variables for both flexion and extension, where the main effects were tensed (Initial, Tensed) and orientation (Upright, Inverted).  The EMG data were not normally distributed and had unequal variances; thus, a log-normal transformation was performed on these data.  Post-hoc Bonferroni tests were 99  used to compare significant interactions across the Initial versus Tensed states: U-Initial versus U-tensed and I-Initial versus I-Tensed.  EMG and postural data are presented as mean and 95% confidence intervals (95% CI, mean - t-statistic * standard error, mean + t-statistic * standard error) (back transformed for presentation of the EMG data).    4.3 Results 4.3.1 Forces The average MVC forces (± standard deviation) in the flexion and extension were 113.8 ±52.4 N and 188.8 ±102.3 N, respectively.  The mean sustained 50% flexion force was 58.9 ± 35.7 N (51.8 ± 8.3 % of MVC target force) upright and 58.4 ± 36 N (51.2 ± 11.7 % of MVC target force) inverted.  The mean sustained 50% extension force was 78.0 ± 48.7 N (42.6 ± 12.2 % of MVC target force) upright and 81.5 N ± 48.6 N (44.8 ± 15.4 % of MVC target force) inverted.  4.3.2 Muscle activity The activities (expressed as % MVC) of all the muscles were significantly higher in the tensed state compared to the initial state for both the flexion and extension force tasks (except MultC4 in flexion) (multiple p<0.05, Table 4-2 and Figure 4-5).  The largest activity in tensed flexion was in the STH muscle both upright (45.6% MVC, 95% CI, 33.6% MVC, 61.7% MVC) and inverted (60.7% MVC, 95% CI, 41.5% MVC, 88.7% MVC).  The largest activity in extension was the SPL upright (38.7% MVC, 95% CI, 22.8% MVC, 65.9% MVC) and LS inverted (50.8% MVC, 95% CI, 31.1% MVC, 82.8% MVC).  An interaction was found (Tensed x Orientation) for SCM, STH and Trap in flexion (Table 4-2).  The increase in activation of STH (p=0.006) and 100  SCM (p=0.031) with the tensed extension task was only found to be significant while upright (Table 4-2).  Table 4-2:  Statistical p-values for the muscle activity in the flexion and extension force tasks Number of subjects (N) and p-values for the EMG metrics for the main factors (Orientation = Upright versus Inverted; Tensed = Initial versus Tensed), and interaction effects.  The post-hoc comparisons of Upright-Initial (U-I) versus Upright-Tensed (U-T) and Inverted-Initial (I-I) versus Inverted-Tensed (I-T) are reported for significant interactions (bold = significant). EMG N Main and Interaction Effects Post-hoc Comparisons Orientation Tensed Interaction U-I vs U-T I-I vs I-T Flexion       STH 10 0.000 0.000 0.001 0.000 0.000 SCM 10 0.002 0.000 0.004 0.000 0.000 LS 10 0.061 0.018 0.714 - - MultC4 9 0.651 0.194 0.660 - - SsCerv 10 0.057 0.006 0.278 - - SsCap 10 0.127 0.019 0.097 - - SPL 10 0.440 0.001 0.338 - - Trap 10 0.238 0.005 0.048 0.000 0.000 Extension       STH 10 0.000 0.006 0.001 0.000 1.000 SCM 10 0.002 0.031 0.006 0.012 1.000 LS 10 0.003 0.000 0.238 - - MultC4 10 0.287 0.001 0.930 - - SsCerv 10 0.446 0.000 0.505 - - SsCap 10 0.795 0.000 0.821 - - SPL 10 0.235 0.000 0.410 - - Trap 10 0.442 0.000 0.999 - -   101   Figure 4-5: Mean muscle activations for the flexion and extension force tasks Flexion (left) and extension (right) group mean (±95% confidence intervals) RMS values (normalized to % MVC) of the EMG activities for all 8 muscles in the initial (black) and tensed (dark gray) states for the upright (top) and inverted (bottom) conditions.  As a reference, the group mean activations of each muscle in flexion (left) and extension (right) MVC tasks are presented (light gray columns). Note that none of the MVC group means are 100% because not all subjects activated these muscles maximally in the flexion or extension directions.  102  4.2.3 Postural changes On average, the eccentricity shifted anteriorly from 20.4 mm (95% CI, 13.0 mm, 27.8 mm) to 37.4 mm (95% CI, 30.3 mm, 44.6 mm) in the tensed flexion and shifted posteriorly from 12.9 mm (95% CI, 5.8 mm, 20.1 mm) to -4.4 mm (95% CI, -13.8 mm, 4.9 mm) in the tensed extension tasks (Figure 4-6).  The motion of the head COM accounted for the majority of the eccentricity shift, as the COM also shifted by similar magnitudes in the tensed state in both flexion (17.3 mm anterior) and extension (19.8 mm posterior).  The Frankfort plane angle was 1.9° for Upright-Initial (95% CI, -1.3°, 5.2°), -1.0° for Upright-Tensed (95% CI, -5.1°, 3.1°), 5.9°  for Inverted-Initial (95% CI, 2.9°, 9.0°), and 2.6° for Inverted-Tensed (95% CI, -1.6°, 6.7°) (Figure 4-7).    The curvature index significantly increased by 1.2% (95% CI, 0.3%, 2.1%) in the extension task between the tensed and initial states, regardless of orientation (Table 4-3 and Figure 4-8).  The flexion task resulted in an orientation dependent increase in curvature index: the index increased by 0.6% (95% CI, 0.05%, 1.0%) when inverted (Table 4-3 and Figure 4-8).  Spinal realignment with muscle contraction was observed at the head/C1 (flexion of 5.0° upright, 95% CI, 1.5°, 8.5°), the C6/C7 (mean extension of 2.4°, 95% CI, 1.1°, 3.6°) and C1/C2 (4.4° more flexed, 95% CI, 2.1°, 6.6°) levels for the flexion force (Table 4-3 and Figure 4-7).  In producing an extension force, cervical realignment was significant at all vertebral levels except C2/C3 and C6/C7 (Table 4-3 and Figure 4-7).  The magnitude of vertebral motion depended on orientation for C3/C4 (3.0° more extended upright, 95% CI, 1.9°, 4.0° and 1.7° more extended inverted, 95% CI, 0.6°, 2.8°) and C4/C5 (4.0° more extended upright, 95% CI, 2.6°, 5.5° and 2.5° more extended inverted, 95% CI, 1.0°, 4.0°) joint levels. 103  Plots and data of individual subject vertebral positions and muscle activations are provided in Appendix E (see Figure E-1 to Figure E-4 and Table E-1 to Table E-6).  Table 4-3:  Statistical p-values for the curvature and intersegmental angles in the flexion and extension force tasks Number of subjects (N) and p-values for the curvature index (C.I.) and intersegmental angles (i.e. ΘC3/C4) for the main factors (Orientation = Upright versus Inverted; Tensed = Initial versus Tensed), and interaction effects.  The post-hoc comparisons of Upright-Initial (U-I) versus Upright-Tensed (U-T) and Inverted-Initial (I-I) versus Inverted-Tensed (I-T) are reported for significant interactions (bold = significant). Kinematics N Main and Interaction Effects Post-hoc Comparisons Orientation Tensed Interaction U-I vs U-T I-I vs I-T Flexion       C.I. 9 0.439 0.152 0.045 1.000 0.030 ΘFrank/C1 8 0.005 0.018 0.033 0.007 1.000 ΘC1/C2 8 0.030 0.003 0.723 - - ΘC2/C3 10 0.079 0.264 0.381 - - ΘC3/C4 10 0.902 0.355 0.493 - - ΘC4/C5 10 0.004 0.973 0.073 - - ΘC5/C6 10 0.398 0.855 0.205 - - ΘC6/C7 9 0.793 0.002 0.572 - - Extension       C.I. 9 0.637 0.017 0.957 - - ΘFrank/C1 8 0.036 0.026 0.084 - - ΘC1/C2 8 0.046 0.013 0.936 - - ΘC2/C3 10 0.069 0.771 0.124 - - ΘC3/C4 10 0.192 0.009 0.020 0.000 0.003ΘC4/C5 10 0.180 0.002 0.036 0.000 0.002 ΘC5/C6 10 0.079 0.003 0.134 - - ΘC6/C7 9 0.249 0.113 0.706 - -  104   Figure 4-6: Group average XY vertebral motions for the flexion and extension force tasks Flexion (left) and extension (right) group average (±95% confidence intervals) of the X-Y coordinates of the vertebrae (mid-superior point of C1 and mid-inferior points of C1-C7 in descending order) and head COM (open circles) for the initial (black circles) and tensed (gray circles) states for both upright (top) and inverted (bottom) orientations.  For visualization, the ‘initial’ C7 mid-inferior point is the origin (0,0).      105   Figure 4-7:  Vertebral and intervertebral angle changes in flexion and extension force tasks The mean difference (and 95% confidence intervals of the difference) in the vertebral (left, head through C7) and intervertebral (right) angles (C1/C2 through C6/C7) for the tensed flexion (top) and extension (bottom) contractions compared to the initial state (i.e. zero angle means the angle was the same as initial) for the upright (black) and inverted (gray) trials.  Ranges reported for the neutral zone of cadaveric cervical joints (i.e. region around neutral posture where the passive cervical joints provide little resistance to rotation)[189] are also plotted as a reference (clear boxes). 106    Figure 4-8: Curvature indices for the flexion and extension tasks Individual subject (open circles) and group mean (closed circles) curvature indices in the initial and tensed states for the upright (left) and inverted (right) conditions for the flexion (top) and extension (bottom) tasks. * significant differences between initial and tensed.    107  4.4 Discussion The goal of this study was to measure the muscle activity and cervical spine realignment during sustained flexion and extension neck muscle contractions in a “near neutral” head and neck posture.  Overall, actively generating flexion and extension forces in the neutral head posture resulted in realignment of the cervical spine, as indicated by changes in curvature index and intersegmental angles.  We also found that the muscle activation and realignment patterns were orientation-dependent.  We hypothesized that voluntarily applying flexion and extension forces with the head would result in significant changes in the alignment of the cervical spine.  Extension forces caused realignment of the mid-cervical spine, resulting in a significantly higher curvature index compared to that of the initial spinal alignment.  By activating the neck extensor muscles we expect this curvature to increase further since the extensor muscles span the concave portion of the lordotic cervical spine.  The opposite might be assumed in the flexion direction; however, a significant change in curvature was not present for the flexion force in the upright condition, and an increased curvature index was observed in the inverted case.  Differences in realignment when producing flexion and extension forces might be explained by the moment generating capabilities of the cervical muscles estimated by Oi et al. [175] (see Table 4-4).  Unlike the extensors, the net moment produced by the flexor muscles varies along the length of the spine with the largest moments occurring at the lower cervical spine (C5/C6 through C7/T1).  Dynamic flexion is initiated by the lower cervical spine [162]; thus, the initial flexion force provided by the muscles pulls the lower cervical spine into flexion.  Being constrained to the neutral position (i.e. little head motion) seems to inhibit, on average, the upper cervical spine 108  moments required to straighten a lordotic curvature.  C5 has been described as the “pivot” vertebra in curve reversal [29]; therefore, straightening of the spine likely requires the net resultant force vector of the activated muscles to be anterior to the C5 vertebra.  Mayoux-Benhamou et al. [161] demonstrated that when subjects tried to straighten their spine (by head nod that was restricted to the upper cervical spine) the extensor muscles were deactivated.  Although levator scapulae and trapezius are not often described as major extensors (they also play a large role in neck lateral bending and shoulder stabilization [161, 171, 174, 176]), we observed moderate activation of these muscles (possibly from bracing the shoulders against the harness straps).  Levator scapulae and trapezius run laterally and posteriorly to the spinal column, providing column stability during flexion moments, which could in turn be preventing the spine from straightening.  Other mechanical factors may have influenced the different realignment strategies in both extension and flexion, such as inherent anatomical asymmetry of the posterior and anterior osteoligamentous vertebral column.  However, the different muscle activation patterns of these eight cervical muscles are likely playing an important role in determining the subsequent spinal realignment under flexion and extension loading.           109  Table 4-4: Computationally predicted moment generating capability of the neck muscles  Computational model results for moment generating capacity of the neck muscles in the upper cervical spine (Occ/C1) and lower cervical spine (C7/T1) reported by Oi et al. [175] (positive is extension).  This model was simulated in the neutral posture with maximum muscle contractions. Muscle Flexion (Nm) Extension (Nm)  Occ/C1 C7/T1 Occ/C1 C7/T1 STH* -15 -7 - - SCM 7 -18 - - LS - - - 6 MultC4 - - - - SsCerv - - - 6 SsCap - - 23 6 SPL - - 3 6 Trap - - 18 13 Total Moment -11 -29 49 47 * Values were reported for hyoid muscles (sternohyoid, sternothyroid, and omohyoid)   We have demonstrated that the muscle contractions necessary to apply global forces with the head also have the ability to change the curvature of the neck; yet, the curvature index is a global cervical spine measure and does not include C1.  Therefore, it is also useful to explicitly explore the realignment of each individual joint of the cervical spine.  When producing a flexion force, the intervertebral angles at the top of the cervical spine were extended (head/C1 and C1/C2) and the bottom of the cervical spine was flexed (C6/C7) but there were no significant changes in the mid-cervical spine intervertebral angles, relative to the initial spinal alignment.  During extension contraction, the top of the spine was flexed (head/C1 and C1/C2), while the intervertebral angles of the mid-cervical spine (C3/C4 through C5/C6) were extended relative to the initial state.  This realignment of the spine is caused by muscle forces; however, the exact contribution of each muscle to this spinal realignment is complex.  Most notably, the moment arm of the sternocleidomastoid from the cervical spinal column changes with spinal level and has its 110  greatest flexion moment generating capability at C6/C7 [190].  Since the sternocleidomastoid acts as an extensor in the upper cervical spine (see Table 4-4), this might offset the flexion generating capabilities of the other flexors (i.e. sternohyoid) [175].   In addition, the origins and insertions of the neck extensors influence their moment generating capabilities.  For instance, semispinalis capitis inserts at the occipital bone (at the nuchal line) and originates at the transverse processes of the thoracic spine and C7 and the articular processes of C4 to C6 [191].  While activation of semispinalis capitis could globally extend the head and neck, forces exerted at the insertions of semispinalis capitis in the mid cervical spine may act to stabilize the curvature and prevent further extension at the intervertebral level.  Thus, although the primary goal of the neck muscles in this experiment was to generate target forces with the head, increased activations of multiple muscles (and their various attachment points) influenced intersegmental rotations that varied in magnitude throughout the cervical column.  The average intervertebral rotations observed in our study emphasize that the rotational stiffness of the muscle-activated in vivo neck is higher than the ex vivo osteoligamentous spine.  The cadaveric spine is highly flexible in the sagittal plane and we would therefore expect the ex vivo spine to bend significantly under the loads measured herein (approximately 60 N in flexion and 80 N in extension).  Yet, even when applying large forces our subjects were able to maintain a near neutral head and neck posture.  For example, the net moment observed at the C5/C6 level in upright flexion for subject #11 was approximately 7.9 Nm (see Figure 4-9A).  Moroney et al. [192] reported the average stiffness coefficients (i.e. the inverse of flexibility) of intact mid/lower cervical spine ex vivo specimens to be approximately 0.43 Nm/deg in flexion and 0.73 Nm/deg in extension.  If we assume that this entire net moment was resisted by the joint (see 111  Figure 4-9B), with no muscle stabilization, we would expect large segmental rotation (on the order of 18° in extension, based on the reported stiffness coefficients) at the C5/C6 joint.  However, the actual C5/C6 joint motion for subject #11 was 0.8°.  In fact, experimental moment-rotation curves derived from the cervical spine often do not reach values approaching this estimated moment of 7.4 Nm: the critical bending moment (that which will result in mechanical failure of the joint) for spinal segments of the cadaveric cervical spine is approximately 7 Nm at 26° (mid cervical spine) and 12 Nm at 24° (lower cervical spine) [98].  The influence of combined loading may also explain this discrepancy between the in vivo stiffness with muscle activation and ex vivo stiffness reported for cadavers [193, 194]; cervical spine rotational stiffness is not well quantified with high combined joint loads (compressive and shear).  The overall in vivo passive neck stiffness (i.e. without active muscle contraction) of young adults is reported to be less than 0.05 Nm/deg in the sagittal plane around the neutral posture (head and torso were within 10°) [195].  This is a considerably lower stiffness than the active in vivo measurements in the current study.  Since in general there was little observed joint rotation, we conclude that even though some muscles are generating these large moments, some are also providing the column stability to resist large intervertebral rotations.         112   Figure 4-9: Simplified free body diagram at the C5/C6 joint for Subject #11 A simplified summary of the external forces exerted on the C5/C6 joint level from the weight of the head (WHEAD) and applied force (FFLEX) at their respective distances (dCOM and dFlex) (A).The applied net moment (MNET) and forces (FSHEAR and FCOMP) that, in the absence of muscles (B), would need to be balanced by the joint reaction moments (MJOINT) and forces (FJOINT 1 and  FJOINT 2). This model assumes (simplistically) that the head band force (FFLEX) acts through the head COM, the weight of the head (WHEAD) is approximately 4 kg [184], the weight of the cranial cervical spine is negligible, and the center of rotation (segmental balance point) is at the mid-inferior point of C5.  Under these assumptions, we can determine the approximate change in joint reaction forces and moments (with WHEAD = 4 kg, FFLEX = 65.0 N, dFLEX = 128.2 mm, initial dCOM = 43.5 mm and tensed dCOM = 55.1 mm).   Overall, the stiffening effect of neck muscles in extension, particularly at the ends of the cervical spine, is highly apparent in this study (as there is little angular motion).  Oi et al. [175] suggests that higher compressive forces in the lower cervical spine provide a greater degree of stability and limit joint motion.  Yet, in the flexion task performed by our subjects, the neck appeared to have substantial stabilization in the upper and mid-cervical regions.  Furthermore, the average 113  joint rotations observed at all levels were within the reported neutral zone for cadaveric cervical spine segments.  These findings have important implications for ex vivo testing, where not only is overall muscle tone necessary but individual muscle forces and intersegment stiffness may play a large role in the simulation of injury.  Basic Euler buckling theory (of a straight and non-segmented column) suggests that the critical load on the spine is proportional to the rotational stiffness (i.e. Nm/deg); thus, it is conceivable that mechanical loading in an injury scenario could be influenced by changes in effective stiffness.  Although the exact contributions of muscles to increases in stiffness are not measurable herein, this provides motivation (and in vivo data) to further study this effect with cadaveric or computational models.  Based on the orientation-dependent findings of this data set, inversion may result in a different alignment and muscle induced stiffness compared to upright.  Specifically, changes in the angles of the head/C1 (flexion), C3/C4 (extension), and C4/C5 (extension) joints depended on orientation in the contraction tasks.  For example, tensing the neck muscles in extension resulted in a localized extension at the C4/C5 joint of 4° upright and 2.5° inverted.  The balance point of a lower cervical spine segment (0.5-1 cm anterior to the posterior longitudinal ligament) lies within the stiffest axis of the segment [42].  Thus, orientation dependent changes in the intervertebral angles reflect deviations of individual joints from the alignment of the stiffest axis.  To cause changes in the intervertebral angles in the sagittal plane, the line of action of the net forces applied to the segment need to be anterior to (causing flexion) and posterior to (causing extension) the balance point of the segment [42, 196].  Therefore, orientation-dependent segmental rotations also indicate changes in the net effect of muscle forces when upside-down.  A significant interaction (for orientation and tensed factors) was also present with the 114  sternohyoid, sternocleidomastoid and trapezius; this could in part explain the intersegmental differences detected while inverted.  We removed some of the main effects of gravity with the headband and chain set-up.  The point of force application was slightly above the occipital protuberance for flexion and slightly above the orbital rim for extension; thus providing some passive gravity compensation (although initial forces on the chain were on average 1.6 N).  In spite of this, inversion clearly played a role in some aspects of the realignment of the spine under active muscle forces.  Thus, the interplay between gravity, head orientation, and muscle response is important to obtaining desired stiffness responses, particularly at certain joint levels, in an inverted configuration.    The main muscles used to produce the target flexion force with the head were sternocleidomastoid and sternohyoid, both activating (on average) to approximately 50% MVC.  Conversely, splenius and levator scapulae were largely recruited to produce the target extension force (mean EMG recordings were approximately 40% MVC), with semispinalis capitis, semispinalis cervicis, and multifidus also contributing (around 20% MVC).  These findings are partially consistent with the preferred directions of the muscles measured during the 100% MVC trials.  The majority of subjects maximally activated their sternocleidomastoid and sternohyoid in the flexion direction, and the majority activated multifidus, semispinalis capitis, semispinalis cervicis in the extension directions (splenius, levator scapulae and trapezius were more involved in lateral bending).  Interestingly, the average activations of the deep extensor muscles (multifidus and semispinalis cervicis) in the target task were less than 25% of the group average activations in the MVC extension task (even though the forces were 50%).  Maximum forces were produced using a rigidly mounted helmet (a fixed condition) whereas the target forces were 115  produced using a headband and chain (a relatively free, less stable condition that allows for some spinal realignment and possibly additional stabilizing muscle activity).  The preferred direction of activation of the neck muscles is complex and may change as increased forces are applied to the head [174, 176], and greater increases of muscle activity levels are reportedly required for generating the proportional increases in moments [197].  However, the disproportionately lower activations in the target tasks compared to MVC could reflect a trade-off between force generation and joint stability.    Our goal was to observe spinal realignment with isometric contractions (i.e. the motion of the head and neck were somewhat limited by the headband and chain); however, due to compliance in the headband the spine and head did move.  Although changes in the Frankfort plane angle and motions of the head and top of the spine were relatively small (group mean horizontal translations were on the order of 2 cm), postural changes of the head and neck can alter the moment-generating capacities of the neck muscles (particularly the extensors) [172].  In order to reduce the motion of the head and neck during the contractions, the head and headband were fitted with a slight pre-tension in the chain (1.6 N on average).  However, these small translations of the head and spine may have influenced the magnitude of the intersegmental rotations and muscle patterns observed with muscle tensing.      The aim of this study was to expose the cervical spine to active flexion and extension forces applied to the head and to study the subsequent vertebral realignments induced by these contractive muscle forces.  The muscle activations were significantly higher in the target force generation compared to their initial activation levels and, on average, the muscle contractions 116  resulted in realignment of the cervical spine.  In the flexion task there was an increase in curvature for the inverted case only; whereas in the extension task there were increases in curvature in both orientations.  Both tasks resulted in significant changes in intervertebral alignment at some cervical levels, several of which were orientation-dependent.  The mean intersegmental motions were all within ranges reported for the neutral zone of cadaveric cervical joints.  These relatively small intersegmental rotations, which occurred when the overall neck moment was large, suggest that active musculature provides considerable resistance to joint rotation in the in vivo neck.  This ability of the neck muscles to both realign the cervical column and stiffen the neck is likely to alter the load path through the spine and influence the risk of injury during axial loading, and thus should be considered in future cadaveric and computational models exploring axial neck injury.     117  Chapter  5: Does dynamic motion of the cervical spine affect the alignment of the neck in the neutral position?   5.1  Introduction Knowing the posture of the cervical spine immediately prior to a head-first impact is critical to studying and preventing cervical spine injuries.  To identify vulnerable postures and design prevention approaches with ex vivo impact testing, the spine is often manipulated into various static postures to study the resulting injuries [18, 20, 26, 36, 39, 41, 45, 97, 129].  One such postural manipulation consists of changing the overall neck angle by setting the relative horizontal offset positions of the ends of the spine (i.e. eccentricity) prior to impact.  Yet, parameters such as spinal eccentricity are set without further adjustments to the alignment of the rest of the cervical column, such as the curvature.  Catastrophic injury to the cervical spine typically occurs in a dynamic environment, and thus, motion of the spine could be present prior to injury.  However, the individual vertebral orientations are purely based on the joint stiffness of the cadaveric spine with no regard to the forces that could be present from motion or muscle activations.  Indeed, efforts to pre-position the eccentricity of a cadaver spine prior to impact implicitly assume: (1) the vertebral alignment and head angle are equivalent at the same eccentricity in a neutral resting and dynamic condition, (2) spine alignment does not depend on the direction from which it approached this eccentricity, and (3) muscle forces required to pre-position the spine have negligible effects on the spinal alignment.  These three assumptions are yet to be fully verified in an in vivo population; thus, it is important to quantify what differences, if any, exist between resting and dynamic alignment.   118  One common posture used in cadaver tests is the so-called neutral posture, in which the spine and each intervertebral joint are approximately in the middle of their ranges of motion and the spine is in the classic lordotic neutral alignment.  When placing the spine into this neutral posture, the main assumption is that the head and cervical alignment should be in their upright, relaxed, resting configurations.  Indeed, adjacent vertebrae do not move independently of each other [118]; however, the relative contributions of individual joints (C2/C3 through C6/C7) to the overall neck (C2/C7) range of motion have been shown to change throughout active flexion and extension of the spine [198].  Furthermore, intervertebral range of motion measured from X-rays of subjects holding a static position of full neck flexion and extension do not represent the intervertebral angles measured at these extremes of motion using fluoroscopy during dynamic motion [116].  We therefore expect that for a given eccentricity, the vertebral angles and cervical curvature would not the same for upright resting neutral and when the spine dynamically arrives to this neutral position; however, this is yet to be confirmed.   There is reason to believe that the state of the spine does in fact depend on the history of where the spine has been (i.e. from an extended or flexed posture) as “hysteresis” has been demonstrated (i.e. the flexion and extension motion envelopes differ) in dynamic flexion and extension of the head and neck [199].  Motion at the lower cervical spine is reported to lag head motion, and intervertebral angles are not the same for identical head angles in either direction (flexion versus extension) [199].  Not only does this mean that the lower cervical spine joint angles cannot be predicted based on the head angle, but the spinal posture also depends on the direction of motion.  Although head and intervertebral angles have, in general, been shown to depend on the direction of motion, the exact alignment of the entire spine as it dynamically 119  passes through the same neutral neck eccentricity in flexion and extension has not explicitly been described.  Injuries can happen in a dynamic environment (such as a rollover accident), where someone could be actively moving their head and neck in the flexion or extension direction immediately prior to impact.  Furthermore, postural hysteresis is yet to be demonstrated when upside-down, an orientation that is highly relevant to inverted cadaveric drop testing [16, 28, 35, 36, 46] and a rollover environment [8, 9].  Knowledge of the dependency of the alignment on the direction of motion is needed to test ex vivo specimens in postures relevant to a dynamic condition.    Finally, since muscles are primarily responsible for dynamic motion of the head and neck, sizable muscle forces may be present as the neck passes through its neutral position.  Cadaveric testing of cervical spines demonstrates a directional dependency in the intervertebral joint angle versus moment curve [200].  Specifically, in the “neutral zone” of spinal motion, there is a wide variety of intervertebral angles that can be achieved with very small changes in the applied moment [201].   In the in vivo case, this applied moment would be induced by both the weight of the head and the muscle activity required in performing these motions.  Descriptions of muscle patterns (both upright and inverted) in conjunction with vertebral positions may help in understanding the role of musculature in dynamic motion.  Indeed, in vivo studies regarding the muscle activation patterns in dynamic flexion/extension, particularly of the deep neck muscles, are sparse [161, 171].    The evaluation of prevention strategies needs to be conducted on cadaveric spines placed in head and neck postures (with appropriate muscle forces) relevant to the injury scenario being 120  simulated.  The overall aim of this study was to assess the assumptions inherent to simulating a resting postural alignment.  Specifically, we propose that for a neutral cervical eccentricity: (1) the vertebral alignment and head angle are not the same in dynamic and resting conditions, (2) spine alignment depends on the direction from which the spine approached this eccentricity, and (3) muscle forces are not negligible.  To confirm this we compared the “neutral” cervical alignment under resting and dynamic (toward flexion and toward extension) conditions, in both upright and inverted orientations.  We also set out to compare the muscle activation patterns in resting neutral and as the neck passes through neutral in active flexion and extension, both upright and inverted.  Our specific hypotheses were that the alignment of the spine and muscle activations would be different in resting and dynamic “neutral” and that the difference would depend on the pre-neutral direction of motion.    5.2  Methods 5.2.1 Subjects Eleven human subjects (6 females and 5 males) participated in this study and the average (± standard deviation) age was 33 ± 7 years.  Subjects were excluded if they had a history of neck injury or pain, or known muscle, nerve, or balance problems.  The average group height was 172.9 ± 6.7 cm, their weight was 70.0 ± 15.7 kg, and their neck and head circumferences were 34.3 ± 4.2 cm and 56.4 ± 3.2 cm, respectively.  One subject did not participate in the fluoroscopy portion of the experiment and another subject did not participate in the EMG portion.  Subjects gave their written informed consent and the study was approved by the University of British Columbia Clinical Research Ethics Board.  121  5.2.2 Conditions The subjects were seated in a bucket seat (36 Series - Intermediate 20 Degree Layback, Kirkey Racing Fabrication Inc., St. Andrew’s West, ON) and secured with a 75-mm wide 5-point harness (RCI Racer’s Choice Inc., Tyler, TX) (Figure 5-1).  Subjects were instructed to sit in a relaxed, upright posture (Upright-Relaxed) for 2 seconds.  Subjects were then instructed to perform a dynamic flexion and extension of their head and necks (Figure 5-1) in both the upright and inverted orientation.  Subjects were instructed to start in full extension, continue to full flexion, reverse at three seconds, and end in full extension at six seconds.  They were given audible beeps to help with the timing of the task (one beep to start the trial, one to reverse motion, and another to stop motion).  Subjects were given a practice run while seated, both upright and inverted, to become familiar with the timing and were encouraged to follow the frame of the inversion device (in front of and above their head, see Figure 5-1) with their eyes to minimize out of plane motions.  During the tasks, cervical spine posture was captured with fluoroscopy and muscle activity was recorded using electromyography.       122   Figure 5-1: Subjects completing flexion and extension task inverted Two upside-down subjects at the time point of full head/neck extension (left) and full head/neck flexion (right).  The subjects were secured in the inversion device (left), with the C-arm positioned for a sagittal view of the neck (left), and a head band with beads was used to track the motion of the Frankfort plane in the X-ray (right).  The Upright-Relaxed position provided a measure of resting neutral, where subjects were instructed to relax in a comfortable position for 2 seconds.  This position was used to compare postural measures of the head and spine as it dynamically passed through this neutral position in the flexion/extension task.  During the flexion/extension task, we defined this “neutral” neck posture to correspond to the posture of the spine when the eccentricity (horizontal displacement between the top of C1 and the bottom of C7, Figure 5-2) of the neck was closest to that measured 123  in Upright-Relaxed. The kinematic and muscle activities were assessed in the Upright-Relaxed task and as subjects passed through the neutral neck position in both directions (flexion and extension), in both orientations (upright and inverted): Upright-Flexion, Upright-Extension, Inverted-Flexion, and Inverted-Extension.   Figure 5-2: Cervical spine eccentricity The eccentricity of the spine (XECC) was used to define the neutral position of the spine, where the cervical spine passed through “neutral” when the eccentricity was closest to that measured in Upright-Relaxed.     5.2.3 Cervical spine posture A fluoroscopic C-arm (OEC 9400, GE) (Figure 5-1) was used to capture sagittal plane images of the cervical vertebrae at 30 Hz (with a 23 cm field of view). Fluoroscopic images were corrected for distortion [182], and vertebral motions were tracked in the images using an automatic tracking algorithm based on the work of Bifulco et al. [157] (maximum static root mean squared 124  errors of 0.6° and 0.2 mm).  The task was repeated twice both upright and upside-down, resulting in a total effective dose of 0.0139 mSv, and the repetition with the clearest fluoroscopy images (i.e. the most vertebrae were visible) was selected for analysis.  The vertebral bodies were tracked throughout the entire motion (Figure 5-3); however, analysis was only performed on the postural metrics at the neutral neck position.   125   Figure 5-3: Exemplar cervical spine motion data for dynamic flexion and extension motion Vertebral motions (C1 through C7) of Subject #3 during dynamic flexion (top) and extension (bottom) of the head and neck.  The four phases of motion plotted are from full extension to neutral (top, left), from neutral to full flexion (top, right), from full flexion to neutral (bottom, right), and from neutral to full extension (bottom, left).  Vertebral motions are plotted starting from black and progressing to gray through time.   126  5.2.4 Postural measures The postural measures included the curvature index (C.I.), the absolute vertebral angles, and the head angle.  The curvature index was defined as the percentage difference between the arc and chord lengths of the spine [159], where the arc length is the sum of the straight line segments passing through the superior and inferior mid-points of each adjacent vertebral body from C2 to C7 and the chord length is the straight-line distance from the top of the dens to the inferior mid-point of C7 vertebral body.  A higher curvature index indicates more curvature.  The absolute angle of each vertebral body is a line connecting the anterior and posterior corners of the inferior edge of each vertebral body and was calculated with respect to the true horizontal (positive = extension).    Head orientation in the sagittal plane was measured relative to the true horizontal using an array of six 2-mm beads that were strapped to the head (Figure 5-1) and projected into the fluoroscope’s field of view.  Prior to data collection, a FaroArm (B08-02, Lake Mary, FL) was used to digitize the location of the beads and anatomical landmarks on the head.  The Frankfort plane (ΘF) of the head was defined by the right and left tragus and the mid-point of the right and left inferior orbits (positive = extension) and was measured relative to the true horizontal.  The head center of mass (COM) was defined as 3.13 cm superior to the Frankfort plane and 0.83 cm anterior to the mid-point of the right and left tragus [188].  A direct linear transformation was used to relate the 3D position of the bead array and the head anatomical landmarks with the 2D X-ray image plane (see Appendix D for details).  Since the bead array was fastened to the head, they moved together as a rigid body. Since the bead array was positioned to be visible in the 127  fluoroscope images, the orientation of the Frankfort plane and the x and y positions of the head COM could be calculated for each image even though the head itself was not in the image.   5.2.5 Speed Subjects were given practice trials, both upright and upside-down, so they could perform the motions at comfortable, consistent speeds.  The motion of the mid-superior point of C1 (i.e. top of the neck) was used to confirm subjects were passing through the neutral neck position at consistent speeds (speeds, in m/s, were calculated as the distance travelled from 100 ms before to 100 ms after passing through the neutral posture).  To assess the speed of head motion, the motion of the center of mass of the head was also calculated (in m/s) over the same time interval.  5.2.6 Electromyography The EMG activity was measured for 8 neck muscles, using a combination of surface (Ambu Blue Sensor, Ambu A/S, Ballerup, Denmark) and fine-wire electrodes (pairs of PFA-coated Stainless Steel, 0.0055” diameter wire; A-M Systems, Inc., Sequim, WA) on the left side only.  Surface electrodes were placed superficial to the sternohyoid (STH), and indwelling fine-wire electrodes were inserted (with ultrasound guidance) into the left sternocleidomastoid (SCM), trapezius (Trap), levator scapulae (LS), splenius capitis (SPL), semispinalis capitis (SsCap), semispinalis cervicis (SsCerv), and multifidus (MultC4) muscles.  EMG signals were band-pass filtered (wire: 50-1000 Hz; surface: 30-1000 Hz) and acquired at 2000 Hz.   Muscle activities were normalized to the levels elicited during maximum voluntary contraction trials (MVC).  With a neutral head posture, MVCs were executed in seven directions: flexion, 128  extension, left lateral bending, two 45° oblique combinations (flexion/left lateral bending, extension/left lateral bending), and right and left axial rotations. The maximum RMS value for each muscle, regardless of direction, was used for EMG normalization.  Each contraction was three seconds long and repeated twice.  The EMG signals were amplified and filtered as described above and both the load cell and EMG were sampled at 2 kHz.  A 20 ms window centered on the maximum force was used to calculate the root mean square (RMS) for each muscle’s EMG in all seven directions.  The RMS of the muscle activity during the dynamic flexion and extension motions and the relaxed upright condition was calculated using a 20 ms moving window during the trial and normalized to MVC.  A trigger pulse (emitted by the image acquisition card) enabled the fluoroscopic image of interest to be synchronized with the EMG data.  The muscle activity of each subject was assessed as their neck passed through the neutral position (as defined by eccentricity) and the 20 ms EMG window selected corresponded to 10 ms before and after this position.    5.2.7 Data processing notes The first cervical vertebra (C1) was not visible for one subject (Subject #5) in the inverted dynamic trial.  The eccentricity could not be determined without C1; therefore, this subject was excluded from the statistical analyses.  Two subjects (Subjects #1 and #2) were pilot subjects and completed the trials at double the speed (two complete motions in the allotted time versus one).  Some of their data are provided in Appendix F; however, they were not included in the statistical 129  analysis or group averages.  Muscle data during the dynamic motion was missing for MultC4 for one subject and SCM for another.    5.2.8 Data and statistical analysis A one-way repeated-measures ANOVA was used to compare the head and vertebral angles and EMG activities in the dynamic and relaxed neutral conditions.  The EMG data were not normally distributed and had unequal variances; thus, a log-normal transformation was performed on these data.  Post-hoc Dunnett’s tests were used to compare each dynamic trial (Upright-Flexion, Upright-Extension, Inverted-Flexion, and Inverted-Extension) to the Upright-Relaxed trial (statistical significance was set at p=0.05). EMG data and the vertebral angles are presented as mean and 95% confidence intervals (95% CI, mean - t-statistic * standard error, mean + t-statistic * standard error) (back transformed, i.e. remove the log transformation, for presentation of the EMG data).  5.3 Results 5.3.1 Postural measures The average speeds of the top of C1 and the head center of mass are summarized in Table 5-1.         130  Table 5-1: Average speeds for C1 and head COM and p-values for the main effect of speed The average (± standard deviation) speeds for the top of C1 and the head COM passing through neutral in the four directions (Upright-Flexion, Upright-Extension, Inverted-Flexion, Inverted – Extension).   Speed (m/s) Upright-Flexion Upright-Extension Inverted-Flexion Inverted - Extension Main Effect p-value C1 0.05 (0.02) 0.04 (0.02) 0.03 (0.01) 0.04 (0.01) 0.034 COM 0.07 (0.02) 0.07 (0.02) 0.05 (0.01) 0.06 (0.01) 0.005   The eccentricity of the spine was not different (p=0.74) between any of the five conditions, including the Upright-Relaxed (which had a mean eccentricity of 3.67 ±11.4 mm) (see Table 5-2).  The curvature index as the neck passed through neutral was significantly higher in Upright-Flexion (4.2%, 95%CI: 2.0-6.4%), Upright-Extension (3.7%, 95%CI: 1.5-5.9%), and Inverted-Extension (3.4%, 95%CI: 1.3-5.6%) compared to Upright-Relaxed (1.9%, 95%CI: 0.7-3.1%) (see Figure 5-4 and Table 5-2).            131  Table 5-2: Statistical p-values for the kinematics in the flexion and extension directions Number of subjects (N) and p-values for the kinematic metrics (eccentricity, curvature index, and angles) for the main effect and interaction effects.  The post-hoc comparisons of Upright-Flexion (U-F), Upright-Extension (U-E), Inverted-Flexion (I-F), and Inverted-Extension (I-E) are reported for significant interactions versus Upright-Relaxed (U-R) (bold = significant). Kinematics N Main Effect U-F vs U-R U-E vs U-R I-F vs U-R I-E vs U-R XECC 8 0.740 - - - - C.I. 8 0.004 0.002 0.015 0.750 0.048 ΘF 8 0.006 0.114 0.065 0.130 0.001 ΘC1 8 0.006 0.010 0.035 1.000 0.310 ΘC2 8 0.003 0.003 0.013 0.836 0.048 ΘC3 8 0.001 0.008 0.004 0.881 0.007 ΘC4 8 0.442 - - - - ΘC5 8 0.009 0.005 0.058 0.852 0.077 ΘC6 8 0.002 0.002 0.015 0.817 0.060 ΘC7 8 0.000 0.000 0.001 0.094 0.001 *θ = angle, C.I. = curvature index, XECC = eccentricity  132   Figure 5-4: Curvature index during dynamic motion compared to relaxed upright The mean (solid circles) and individual (open circles) curvature indices as subjects passed through the neutral position during dynamic flexion (left) and extension (right), both upright (top) and inverted (bottom). *significant differences between relaxed and dynamic conditions.  133  The angle of the Frankfort plane was significantly different during inverted dynamic extension compared to the upright resting (16.0° more extended, 95% confidence interval: 4.9°, 27.1°, p=0.0008) (Figure 5-5 and Table 5-2).  In the upright configuration, the top of the cervical spine (C1 through C3) was more extended and the bottom of the spine (C5 to C7) was more flexed during the dynamic flexion (multiple p<0.05) (Figure 5-5 and Table 5-2).  The same was true for upright extension, except that C5 was not significantly different.  During Inverted-Extension, C2 and C3 were more flexed (p=0.0476 and p=0.007, respectively) while C7 was 7.8° more extended (95% confidence interval, ±3.5°, p=0.0007) (Figure 5-5 and Table 5-2).  No differences were found for inverted flexion.  A summary of each subject’s angle data at neutral are presented in Appendix F (see Table F-1 and Table F-2).    Vertebral angle differences in flexion compared to extension were typically present throughout the entire motion (note the asymmetrical pattern of motion in the two directions, particularly near the middle of the motions in Figure 5-3).  Individual subject plots over the entire motion are also presented in Appendix F (see Figure F-1 to Figure F-9).  134   Figure 5-5: Angles of the head and cervical vertebrae during dynamic “neutral” The mean difference (and confidence intervals of the difference) in the absolute angles of the head and each cervical vertebra (C1 through C7) in the neutral head position during flexion (left) and extension (right) for the upright (top) and inverted (bottom) conditions compared to the Upright-Relaxed state (i.e. zero angle means the posture was the same as Upright-Relaxed).  Significant differences between each dynamic trial and the Upright-Relaxed condition are indicated with a star.   135  5.3.2 Muscle activity The mean group activities (expressed as % MVC) of the muscles at the point of passing through neutral were significantly larger than the resting muscle activities in the Upright-Relaxed trial, although this depended on the direction of motion and orientation (Figure 5-6 and Table 5-3).  While passing through neutral in flexion only the STH was higher upright (p=0.0001) and STH and SCM were higher inverted (multiple p<0.05) compared to the relaxed trial.  Although statistical differences exist, these muscle activities were all less than 5% MVC.  Overall in extension, the activities increased for the extensor muscles (MultC4, SsCerv, SsCap, SPL) when upright, with SsCerv (2.5% MVC, 95% confidence interval, 1.2% MVC, 5.0% MVC, p=0.0036) having the highest mean activation.  All the muscles had higher activation levels during inverted extension: ranging from 2.2% MVC (confidence interval: 0.9% MVC - 5.5% MVC) for SPL to 6.7% MVC (95% confidence interval, 2.2% MVC, 20.1% MVC) for LS (multiple p<0.05).  A summary of each subject’s EMG data at neutral is in Appendix F (see Table F-3).  Individual subject EMG activities throughout the entire motion are also in Appendix F (see Figure F-10 to Figure F-19).         136  Table 5-3: Statistical p-values for the muscle activity in the flexion and extension directions Number of subjects (N) and p-values for the EMG metrics for the main effect and interaction effects.  The post-hoc comparisons of Upright-Flexion (U-F), Upright-Extension (U-E), Inverted-Flexion (I-F), and Inverted-Extension (I-E) are reported for significant interactions versus Upright-Relaxed (U-R) (bold = significant). EMG N Main Effect U-F vs U-R U-E vs U-R I-F vs U-R I-E vs U-R STH 9 0.000 0.000 0.989 0.000 0.000 SCM 8 0.000 0.071 0.904 0.000 0.000 LS 9 0.007 0.775 0.226 0.123 0.002 MultC4 8 0.007 0.377 0.024 0.256 0.002 SsCerv 9 0.000 0.986 0.004 0.852 0.000 SsCap 9 0.000 0.460 0.005 0.304 0.000 SPL 9 0.002 0.295 0.017 0.287 0.000 Trap 9 0.025 0.439 0.222 0.148 0.005 137   Figure 5-6: Mean muscle activity at neutral head posture The % MVC group mean (and confidence intervals) RMS muscle activity (20 ms window) for the Upright-Relaxed trial (relaxed) and at the point when subjects passed through the neutral head posture in the flexion (left) and extension (right) directions (dynamic), for both upright (top) and inverted (bottom) (* p-value is less than 0.05 compared to Upright-Relaxed) 138  5.4 Discussion The aim of this study was to evaluate whether the “neutral” alignment of the cervical spine and muscle activations are different in a dynamic task versus a relaxed, resting task.  The alignment of the cervical spine and corresponding muscle activity at the neutral neck eccentricity during flexion and extension motions were different than that of relaxing.  Furthermore, the differences depended on the direction of dynamic motion, particularly when upside-down.    Compared to the upright relaxed neutral, the curvature of the spine was greater in all of the dynamic neutral tasks except inverted flexion. Thus, we confirmed that at identical neck eccentricities (i.e., at identical overall neck postures) the relaxed and dynamic spines do not necessarily have the same posture. This finding could have important implications for cadaveric testing of cervical spines.  The mechanism of axial loading-related injury is highly sensitive to the eccentricity of the spine [18, 41]; yet, the intrinsic cervical alignment can determine the path of loading throughout the column (see Figure 5-7).  Indeed, the straightened spine is aligned along its stiffest axis and tends to produce mid- and lower-column compression type injuries (such as wedge and burst fractures) [18, 23, 97].  We found that the curvature of the spine was higher in all dynamic instances except Inverted-Flexion.  In general, vertebrae cranial to C4 were more flexed while the vertebrae caudal to this joint were more extended.  In fact, on average the C7 vertebra was almost 10° more flexed during upright flexion, a magnitude which would likely influence the forces and moments the spinal column would experience during head-first impact.  Therefore, even though the same eccentricity was present for all relaxed and dynamic neutral conditions studied, the intrinsic alignment of the column may be drastically different in a dynamic instance.  139    Figure 5-7: Potentially different force paths through a relaxed and dynamic spine The vertebral alignment for Subject #11 for Upright-Relaxed (left) and Upright-Flexion (right), where the two conditions have equivalent eccentricities (XECC).  If a force were imparted to the spine from head contact (i.e. head-first impact) the differences in head COM positions, Frankfort plane angles (ΘF), and vertebral orientations in the two cases would result in a different loading path through the cervical column.    The head and C1 appear to be moving, to a certain extent, independent of one another through the neutral position during the dynamic trials (Figure 5-5).  We found that the head angle lags the neck eccentricity when passing through neutral in upright flexion and leads the neck when passing back through neutral in extension (both upright and inverted).  In contrast, on average 140  the head angle and neck eccentricity arrive at neutral at the same time during inverted flexion.  Differences in the upper cervical spine and the head may be important to neck injury as motion constraint on the head and the atlanto-axial joint (C1/C2) appears to play an important role in axial compressive neck injury [26, 30, 35, 40].  Furthermore, impacts to cadaveric specimens with pre-extended and pre-flexed heads (to magnitudes of 25°) have been reported to result in extension and flexion type injuries to the neck while axial impacts to the vertex of a neutrally aligned head mostly result in compression type neck injury [39].  Thus, knowing where the spine has been prior to approaching neutral might be important to determining the head and upper cervical spine angles at impact.    While the differences between the neutral postures approached from flexion and extension point to differences in the way muscles apply forces to the head and spine when passing through neutral, the differences between upright and inverted (i.e. orientation-dependent) demonstrate that gravity plays a large role in the postural and muscular interactions.  Nightingale et al. [16] reported that they attempted to replicate an anatomically neutral configuration of the neck and head in inverted cadaver drop tests.  Specifically, they aimed to preserve the neutral head configuration (i.e. no antero-posterior rotation of the head) and spinal posture (lordosis maintained, and the upright C7/T1 angle preserved) by tethering the head.  Our data demonstrate that in the case where dynamic head and neck motion is present prior to injury (as is likely in the case in many sports and motor vehicle accident related injuries), the effect of both inversion and direction of motion on cervical alignment and head posture need to be considered.  141  Muscle activation levels when moving through neutral were generally higher than when upright and relaxed, but the magnitude of these differences depended on the motion direction (flexion and extension) and subject orientation (upright and upside-down). Co-activation of the flexor and extensor muscles in the Inverted-Extension task is particularly interesting. The head’s center of gravity is located anterior to the atlanto-occipital joint and thus gravity pulls the head into extension when inverted and relaxing (see Chapter 2). To maintain an inverted neutral head posture requires activation of the neck flexors (STH and SCM), which also flexes the upper cervical spine and generates an anterior eccentricity (see Chapter 2).  During extension motion through neutral, activation of the flexor muscles may act to control head and neck speed against the pull of gravity and thus contribute to the changes observed in cervical alignment. The peak muscle activations throughout the phase of motion from fully flexed to neutral and then from neutral to fully extended shows that, on average, STH decreases while LS, Mult, SsCerv increase after passing this transition point (see Appendix F, Table F-4).  The simultaneous activation of extensor muscles, indicates a co-contraction that may be related to a transition from controlled head descent against gravity immediately before reaching neutral and lifting the head against gravity immediately after head descent.     Muscle activation levels in our subjects changed throughout their flexion and extension motions (See Figure F-10 to Figure F-19 in Appendix F).  We compared activations between relaxed and dynamic cases over 20 ms windows at the instant of passing through the neutral neck position.  However, these windows did not capture the full activation patterns that occur leading up to (or after passing through) neutral.  Furthermore, the large confidence intervals observed predominantly reflect the varied muscle responses between individual subjects.  For instance, the 142  confidence intervals for the multifidus activity during Inverted-Extension ranged from 0.6 % MVC to 20.6% MVC.  Indeed, inspection of the individual responses reveals that four subjects had less than 1% MVC whereas four had activations above 10% MVC (one subject’s was as high as 62.5% MVC).  A study by Anderst et al. [164] described the muscle activation patterns in subject-specific modeling of dynamic flexion/extension, but indicated their data could not be verified since in vivo data are lacking. Thus, comparing our individual muscle activation schemes to a computational model throughout the entire motion, such as that described by Anderst et al. [164], may be useful for further understanding the differences between the relaxied and dynamic cases.   We chose neck eccentricity as the criterion defining the neck’s neutral position; however, we recognize that other criteria could be used to define a neutral position.  Eccentricity is relevant to our study because it has been reported (or controlled for) in cadaveric testing of catastrophic neck injury mechanisms [18, 41].  Others have used head position and orientation to define a neutral position when evaluating intervertebral angles [199] in the cervical spine and muscle activation levels [161] during dynamic motion.  Indeed, a separate analysis using head angle as the criterion to define neutral revealed significant differences in vertebral angles (see individual subject data in Appendix F).  Thus, although the exact posture and alignment differences depend on the criteria chosen to define neutral, our data nonetheless demonstrate that the Upright-Resting posture does not necessarily represent a dynamic posture and consideration of prior motion is important.    143  Voluntary flexion and extension can be achieved in several ways.  For instance, flexion of the head can be achieved by nodding the head and/or bending the neck.  The method subjects choose will affect the patterns of segmental motions throughout the dynamic task [202].  We did not want to influence a subject’s preference for completing the task; thus, subjects were given minimal instructions on how to move their head and neck. However, subjects were instructed to start in full extension.  Starting in full extension requires sustained muscle contractions to combat gravity, particularly in the inverted case, and could have influenced the spinal curvature.  Furthermore, the results of our study may reflect motion restriction provided by the 5-point harness and the speed that the trial was to be completed (6 seconds to complete one full cycle of flexion and extension).  The center of mass of the head moved at a slightly lower average speed in the Inverted-Flexion task; the slower head speeds could partly explain why kinematic differences were not found when comparing this trial to upright and relaxed.  Thus, further research may be warranted from an initially flexed posture, or without a 5-point harness constraint, or at various (controlled) speeds to confirm these findings.     In summary, we found that the cervical spine is more curved while moving through neutral compared to the relaxed upright posture.  More specifically, portions of the upper cervical spine are more extended and portions of the lower cervical spine are more flexed, although the exact differences between dynamic and relaxed postures depend on both the direction of motion and on body orientation (upright or upside-down).  Low muscle activation levels were measured as the head passed through neutral (most notably, all muscles were activated during inverted extension).  Based on these findings, we propose that at the same eccentricity: (1) neither the vertebral alignment (curvature and vertebral angles) nor the head angle are equivalent in a resting 144  and dynamic condition, (2) spine alignment does depend on the direction from which it approached this eccentricity (particularly upside-down), and (3) low-level muscle activity is present during dynamic motion (particularly when inverted).  These findings are important to ex vivo testing where the head and neck are statically (and neutrally) pre-positioned prior to impact, often with negligible muscle forces, and suggest that current cadaveric head-first impact tests may not reflect a dynamic injury environment.     145  Chapter  6: Integrated discussion  The overarching goal of this thesis was to evaluate how several factors influence the alignment of the cervical spine, including whole body inversion, muscle activations, and dynamic motions.  These factors were explored in a series of four experiments: (1) static whole body inversion (with two head orientations), (2) actively tensing the neck muscles, (3) slowly ramping forces in the flexion and extension direction in the neutral head angle, and (4) dynamically flexing and extending the head and neck through the neutral position.  Inversion, head angle, muscle activations, and prior motion were all found to influence the alignment of the cervical spine (see Table 6-1) in the “near neutral” position.  The alignment of the spine varied a significant amount when exposed to these four tasks, for both average responses and for individual subjects.  Each of the tasks explored in this thesis also clearly demonstrate that musculature plays an important role in realignment of the cervical spine.  Significantly higher activations were found for numerous muscles in all of the static and quasi-static conditions, compared with the neutral relaxed position, as outlined in Table 6-2.            146  Table 6-1: Summary of kinematic findings A summary of the tasks, conditions, and orientations that influence alignment of the spine  Experimental Task Curvature index Horizontal motion of ends of the spine Head angle C7 vertebral angle Other vertebral angles (C1 to C6)† Joint angles† Inverted Relaxed Increased Ecc. Posterior Ext N.S.   Inverted Forward N.S. Ecc. Anterior N.S. Flex   Upright tensed free Increased C1 N.S. C7 anterior Ecc. N.S. N.S. Flex   Inverted tensed free Increased C1 anterior C7 anterior Ecc. N.S. N.S. Flex   Upright flexion constrained N.S. Ecc. Anterior Flex Flex All Flex Occ/C1 Ext C1/C2 Ext C6/C7 Flex Inverted flexion constrained Increased Ecc. Anterior Flex Flex All Flex C1/C2 Ext C6/C7 Flex Upright extension constrained Increased Ecc. Posterior N.S. N.S. C1Ext C2 Ext C3 Ext C4 Ext C5 Ext Occ/C1 Flex C1/C2 Flex C3/C4 Ext C4/C5 Ext C5/C6 Ext Inverted extension constrained Increased Ecc. Posterior N.S. N.S. C1 Ext C2 Ext C3 Ext C4 Ext C5 Ext C6 Ext Occ/C1 Flex C1/C2 Flex C3/C4 Ext C4/C5 Ext C5/C6 Ext Upright dynamic flexion  (at neutral) Increased Ecc. N.S.* N.S. Flex C1 Ext C2 Ext C3 Ext C5 Flex C6 Flex C7 Flex  Inverted dynamic flexion  (at neutral) N.S. Ecc. N.S.* N.S. N.S. N.S.  Upright dynamic extension  (at neutral) Increased Ecc. N.S.* N.S. Flex C1 Ext C2 Ext C3 Ext C6 Flex C7 Flex   Inverted dynamic extension  (at neutral) Increased Ecc. N.S.* Flex Flex C2 Ext C3 Ext C7 Flex  Flex = flexed, Ext = extended, N.S. = not significant, Ecc. = eccentricity * eccentricity was used to define the neutral posture and was not an outcome measure † although data exist, not all data were statistically tested and are thus omitted from sections of the table.  147  Table 6-2: Summary of muscle findings A summary of muscles that were found to have significantly higher activation than the initial resting condition for all tasks, conditions, and orientations.  STH SCM LS MultC4 SsCerv SsCap SPL Trap Inverted Relaxed X X  X X X X  Inverted Forward X X       Upright tensed free X X X X X X X X Inverted tensed free X X X X X X X X Upright flexion constrained X X X  X X X X Inverted flexion constrained X X X  X X X X Upright extension constrained X X X X X X X X Inverted extension constrained   X X X X X X Upright dynamic flexion  (at neutral) X        Inverted dynamic flexion  (at neutral) X X       Upright dynamic extension (at neutral)    X X X X  Inverted dynamic extension (at neutral) X X X X X X X X     148  Although all of the conditions and tasks revealed differences in the neck alignment and muscle response, there are several key findings worth highlighting:   The importance of inversion: First, sizable changes in the alignment of the spine were observed by simply inverting a person: specifically, the curvature, eccentricity, and lower neck angle.  The tendency of the head to extend when inverted induces a moment about the contact point between the head and C1, which tends to pull the rest of the cervical spine posteriorly.  This results in an increase in curvature and a posterior eccentricity of the cervical spine.  These curvature changes were mostly isolated to the upper c-spine, i.e. a posterior translation of the upper cervical spine with no significant angular change to the C7 vertebral body.  This cervical realignment when upside-down and relaxing was also accompanied by an increased response in the extensor muscles.  Therefore, any future efforts at simulating inverted neck injury may need to consider the cervical realignment and muscle activities that are present in the inverted relaxed neck.   The importance of head angle: Second, pulling the head forward to maintain a level gaze involved an increase of approximately 7% MVC of the flexor muscles (STH and SCM).  This muscle response provides the necessary moment to counterbalance the tendency of the head to extend backward.  The muscle activation of the flexors when inverted results in the forward translation of the upper cervical spine and flexion of the C7 vertebra.  This underlying flexor muscle activity was also found to be important in the free and constrained (flexion direction) tasks, as STH and SCM had significant interactions for tensing and inversion.  This means that the flexor muscle activities with large force generation are higher when someone is upside-down (at least when a neutral head posture is maintained).  Cables have been used to counterbalance 149  the head extension of a cadaveric head in inverted drop tests [16, 28, 35, 36, 46].  Our findings highlight the importance of the SCM and STH in maintaining a neutral head position upside-down.  Thus, any future efforts at simulating neck injury upside-down, either ex vivo or computationally, should consider the fact that maintaining a forward head posture via muscle activations has a significant effect on the underlying posture of the spine.    The importance of muscles: Third, tensing the neck muscles (in both a free and constrained head condition) realigned the spine in a way that could influence the mechanism of axial impact injury.  Although the free, constrained flexion, and constrained extension tensing tasks resulted in changes in the alignment of the cervical spine, the specific postural responses were different in all three cases.  For instance, inspection of Table 6-1 reveals that curvature does not increase with upright constrained flexion; thus, the curvature increase in the free task is likely due to contraction of the extensor muscles.  Additionally, constraining the head appears to limit the motion of the bottom of the spine, whereas the spine translated forward in both orientations of the free condition. The motion of pulling the neck down toward the shoulders (as opposed to pulling forward or backward) generates the forces necessary to also pull the bottom of the neck forward.  The muscle responses observed in vivo were also task-dependent.  The free tensing task appeared to be a co-contraction of both the flexors and extensor muscles, whereas the constrained tasks revealed a more isolated activation of the flexors and extensors.  The alignment and muscle activation differences in these separate tasks are relevant to both cadaveric impact injury.  Previous cadaveric tests have involved horizontal forces that are applied to the head to achieve a desired alignment in a cadaveric specimen prior to impact.  This work suggests that without force resistance, this constrained response may not be possible in a free head condition.  150  Thus, the postural characteristics and muscle forces involved prior to head-first impact will depend on the exact injury scenario one is trying to simulate.    Furthermore, the constrained task provided a more detailed way of quantifying the forces the spinal muscles are capable of generating, studying the directional focus of muscles, and the capability of the muscles to alter rotational joint stability.  When the muscles produced 50% of maximal forces in the flexion and extension directions, there was a significant realignment of the spine.  Yet, with the applied moments, the intersegmental motions were an order of magnitude smaller than would be expected (for even small applied loads) in an osteoligamentous joint [192, 193] and those reported for the passive in vivo neck [195].  This indicates that the neck muscles provide substantial rotational stiffness to the in vivo neck, an effect that is currently not being simulated in cadaveric models of axial impact injury.    The importance of motion: Finally, dynamic motion of the spine induces a different “neutral” alignment compared to the upright resting posture of the spine.  In other words, the alignment of the cervical spine depends on which direction the head and neck approached that position from.  In general, the curvature of the spine was increased when dynamically passing through neutral.  Furthermore, the angles of isolated vertebrae and the head depended on the direction of motion and orientation of the subject.  These differences with inversion and prior motion further emphasize the fact that, in a neutral position, the cervical column curvature and muscle forces are not always in their upright, relaxed state.  If the aim is to replicate axial impact neck injury in the context of a dynamic environment, motion-dependent postural and muscle effects need to be considered.  151  6.1 Implications The findings of this body of work have numerous implications as it provides quantification of how the cervical spinal alignment can change under various neutral conditions and the muscle patterns associated with realignment.  This provides novel in vivo data to improve future injury models and emphasizes the importance of considering pre-impact muscle forces and alignment in injury prevention approaches.  The findings reported in this thesis demonstrate that current attempts at studying cervical spine injury ex vivo or computationally have largely been neglecting the effects of inversion.  This is a configuration that is important to head-first impact, and is particularly central to rollover research.  This work also suggests that muscle contractions can affect the underlying posture and stiffness of the neck and that this in vivo response is not being simulated accurately in existing injury models.  Finally, this work advocates that current injury models are only representing the posture and muscle forces of the resting cervical spine; these need to be considered in ex vivo axial impact tests attempting to recreate the injuries applicable to a dynamic environment, such as a rollover accident.  As is discussed in detail in the following sections, the influence of all four of these factors, along with the large variability in subject responses, may be important to replicating clinically relevant injuries in the laboratory or when computationally modeling head-first impact.    6.1.1 Implications for ex vivo testing The primary motivation for this work was to answer one fundamental question: what is the alignment of the spine prior to a head-first impact?  Although the subject responses reported herein are not captured in the moments leading up to an actual impact, this thesis work provides the necessary initial steps toward answering this question.  Overall alignment, neck eccentricity, 152  and curvature (although not quantified) have been manipulated in cadaveric testing efforts; however, these are not based on live human data sets in postures relevant to injury scenarios.  This work provides the novel in vivo data that are needed to describe the alignment of the spine in several configurations that may be relevant to injury scenarios.  Moreover, this work identifies potentially important factors that influence the pre-alignment of the neck before injury.  To date, a number of cadaveric tests have been performed with the head and spine upright [18, 19, 39], or inverted with the upright relaxed posture maintained [28, 36].  This work clearly demonstrates that the upright relaxed posture is not representative of the variety of “near neutral” postures people adopt upside-down, particularly when muscle activations are involved.  Deviations of the spine from the neutral upright posture result in a dramatic decrease in the spine’s ability to support high loads [113].  Since the moment arms of the muscles change throughout the neck range of motion, the mechanical advantage of the neck muscles appears to be the most effective in the neutral posture [113].  Simulating these alignment changes and muscle activations in a more realistic manner may expose some of the causes of the apparent disparity between real-life neck injuries and the cadaveric injuries produced in laboratory experiments.  Whole body inversion (along with head orientation), sustained muscle forces, and dynamic motion were all found to influence the spinal posture in a notable way.  Three of the main kinematic measures used to evaluate changes in alignment were curvature index, eccentricity, and C7 angle.  Large changes in eccentricity occurred during the isometric contractions, when inverted and relaxed, and when inverted and looking forward (changes ranged from 14 mm to 18 mm).  Even small changes in the eccentricity of the cervical spine (occipital condyles to T1), on the order of ±1 cm, can be enough to change the mechanism of axial impact injury [18-20, 41].  153  The unconstrained bracing task elicited the largest mean horizontal translation of the top and bottom of the cervical spine (10.7 mm and 9.5 mm, respectively).  These changes in the end conditions of the spine (if present before impact) are large enough to influence the mechanism of injury caused by a head-first impact.    Compared to the upright relaxed posture, the largest average change in curvature index occurred when the upright spine dynamically passed through the neutral position, both during flexion and extension (1.8% higher).  Generating isometric forces in the flexion and extension directions while inverted also induced high changes in curvature index (1.2%).  Changes in curvature have not been precisely documented in cadaveric testing of cervical spines in the context of injury.  As a reference, curvature indices reported for various age groups are 1.7% for young (18-24 years), 2.12% for mid-aged (35-44 years) and 3.51% for older (62-74 years) subject groups [159]; these differences are of a comparable magnitude to some of the differences detected in this study.  Thus, although it is difficult to speculate on the exact effect these curvature changes would have on cadaveric testing, they appear to be of sizable magnitude compared to the range of curvatures in the general population.     The largest average postural changes to the bottom of the neck (as measured by C7 relative to the true horizontal) occurred when subjects passed through the neutral head posture during dynamic upright flexion and inverted extension (both were approximately 8° more flexed than upright relaxed).  Changes to the C7 vertebral angle likely reflect changes between the bottom of the neck and torso since the magnitude of changes are in agreement with the range of motion of the C7/T1 joint (reported to be 6.0° on average [98]).  This is under the assumption that the torso 154  remains at the same angle, a reasonable assumption since subjects were constrained by a 5-point harness.  Flexing the C7 vertebra in a cadaver up to 8° would almost certainly result in a different transmission of the force between the torso and neck.  Since adjacent vertebrae do not move independently of each other, this also indicates alterations in the position of the rest of the neck versus the torso, likely producing anterior eccentricities at various levels. Compression injuries, such as burst and wedge compression fractures, happen in higher proportions in cadaver testing versus real-world rollovers [11].  Compressive burst factures often result with small (less than 1 cm from mid-vertebral body) anterior eccentricities of neck loading [17].  Wedge fractures are thought to have larger (greater than 1 cm) anterior eccentricities with a larger flexion moment component to the spinal loading [17].  Furthermore, Foster et al. [11] has noted that cadaveric drop tests (which lack active musculature) have tended to produce more upper cervical spine injuries than are seen in real-world rollover accidents.  With near neutral head and torso positions, the flexed realignment of C7 observed in several tasks in this thesis work would likely render the lower spine more susceptible to anterior loading and flexion bending moments.  This anterior flexion of the bottom of the cervical spine in vivo could explain the increased incidence of flexion-type injuries in rollover occupants compared to ex vivo.    This thesis provides evidence that the effects of whole body inversion should be considered in laboratory experiments of serious neck injury.  Neck muscles are often ignored in ex vivo testing or simulated with small levels of force.  These data suggest that neck muscles are capable of manipulating the alignment of the spine, both in a free condition and an isometrically constrained condition.  Beyond overall alignment changes (i.e. eccentricity and curvature), these activations also have the ability to cause localized rotations in the intervertebral joints in the cervical spine.  155  Moreover, a substantial stiffening effect of muscles on the overall neck was noted, an effect that is not currently simulated in cadavers (which have a large neutral zone).  The data herein provide muscle activation schemes, and activation levels relative to maximal isometric contractions, that could be used to improve simulation of the neck stiffening effect of muscles in the laboratory.    The realignment of the spine and neck stiffening observed in the muscle contraction tasks may help in understanding why some injuries are difficult to reproduce in ex vivo testing.  For instance, manipulation of ex vivo specimens is necessary to create facet dislocations in the laboratory.  This is achieved by realigning the cervical column (into a protrusion posture) and stiffening the lower cervical spine (with a rod in the spinal canal) [31] or by providing rotational constraint to the upper cervical spine [30].  Although these studies have been able to replicate facet dislocations in the laboratory, it is often in a non-physiologic way.  Pintar et al. [27] did not simulate musculature and noted that they were unable to produce locked facets and observed little evidence of shearing translation – common clinical findings in rollover neck injuries [11].   An interesting observation is that the alignment of the cervical spine when subjects were generating flexion moments appeared to be visually similar to the illustrations provided by Myers et al. [30] for cadaveric tests of rotationally constrained spines.  As an example, the anterior protrusion, forward flexion of C7, rotational constraint on the head, and spinal realignment of subject #11 appeared to be grossly similar to the alignments illustrated by Myers et al. [30] (see Figure 6-1).  Indeed, the typical changes in eccentricity with applied axial force described in the rotationally constrained specimens (approximately 30 mm) were of the same order of magnitude as that measured with flexion muscle contractions in vivo (approximately 20 156  mm).  It is plausible that flexor muscle contractions may be sufficient to render the spine susceptible to lower cervical spine dislocations.     Figure 6-1: Realignment in the constrained force task compared to Myers et al. The realignment of the cervical spine for one subject in the constrained tensing task (initial, left and tensed, right).  These realignments appear to be similar to that illustrated for axial loading of fully constrained (left) specimens, which resulted in mid-cervical spine compressive injuries and rotationally constrained (right) specimens, which produced lower cervical spine bilateral facet dislocations.  Bottom images adapted from Myers, B.S., McElhaney, J.H., Richardson, W.J., Nightingale, R.W., and Doherty, B.J., 1991. The influence of end condition on human cervical spine injury mechanisms. SAE Technical Paper No. 912915 391-9 [30], reprinted with permission from SAE International Paper No. 912915 © 1991 SAE International. 157  Finally, dynamic motion prior to impact may be influencing the posture of the spine in a way that is also not being simulated in the laboratory.  The posture of the spine when dynamically passing through “neutral” was different upright compared to upside-down.  Interestingly, the head/C1 joint was almost 10° more extended inverted than it was upright.  This joint also behaved differently between inverted and upright in the isometric force generation tasks, in which head motion was restricted.  The articulation of the occiput and C1 is an important component in load transmission from the head through the spine.  By restricting motion at this joint, researchers have been able to influence the resulting spinal injury [26, 35].  Rollovers in particular are dynamic events during which occupants often are moving within the vehicle and the cervical spine is almost certainly not in its upright, neutral, resting posture.  Simulating the alignment and the stiffness of this joint (ex vivo) in a way that incorporates the current findings (in vivo) could be of particular importance to replicating injury modes observed in real-life dynamic injuries.  Thus, this dependency of the head/neck alignment on both motion and inversion needs to be considered in modeling a dynamic injury scenario.      6.1.2 Implications for computational models of injury Even if the desired postural characteristics of the neck for a cadaveric drop tests are known, reproducing a specific alignment can be challenging.  As has been demonstrated in this thesis, these alterations in alignment are often caused by active musculature.  In addition, although the muscles appear to have substantial neck stiffening ability, it is not always clear which muscles to simulate (and to what level of force) in an ex vivo test.  The EMG activities provide evidence of which muscles may be important to a particular posture and the relative activations of each muscle required to generate these alignments.  However, the exact magnitude of force each 158  muscle provides is not yet clear.  Each neck muscle applies forces at a distance from the cervical column and these moment arms change with spinal level and with alterations in head and spine posture [172].  These moment arms affect the moment generating capabilities of each muscle at the various intervertebral levels.  Furthermore, EMG is only a relative measure of muscle activity, and does not directly indicate the force each muscle is producing.  To better understand the forces associated with these measured muscle activations, a detailed numerical simulation that models the variation in muscle moment arms along the spine and the muscle fiber dynamics (i.e. how activation is related to force) is required.    Attempts at modeling axial neck injury have been valuable to understanding the potential ability of muscle to change injury risks and shift the location of injury [51, 52, 54].  Yet, the exact pattern of muscle activations required for these simulations are often based on optimization routines instead of in vivo data.  The EMG data provided in this thesis provide a means for improving existing or future models specific to axial injury.  For instance, voluntary bracing of the neck increased the activity levels of all eight neck muscles; however, the activation levels remained sub-maximal and varied between muscles.  This is contrary to several computational models that have assumed the neck muscles to be at 100% activation prior to injurious loading [54, 167] (see Figure 6-2), a condition that is likely not physiologic.  The extensor muscles are stronger than the flexors [105, 175, 203, 204]; thus, maximal activations of the muscles in a computational model cause extension and downward motion of the head.  To combat this undesired effect, modelers have used head restraints [54, 167] or muscle-specific activation schemes [52].  In contrast, when the subjects in this thesis work tensed their muscles, they pulled their head and neck forward.  The constrained task findings indicate that when subjects generated 159  large forces solely with the extensor muscles, this caused a posterior translation of the head and neck, with the opposite effect in the flexion direction.  Comparing the muscle activation patterns of the unconstrained and constrained tensing tasks indicates co-contraction of anterior and posterior muscles is more physiologic than activating only one set of muscles (see Figure 6-2).  The patterns observed in the free and constrained tasks in this thesis work suggest that even the most sophisticated models that currently exist may generally be overestimating the levels of contractions associated with pre-impact bracing.  These results also indicate that the co-contraction of flexors is important to maintain an equilibrium position.  Thus, smaller extensor activations and flexor co-contractions should be considered when simulating pre-impact muscle activation in computational neck injury models.    Chancey et al. [52] proposed a model in which the objective function maximized the total muscle forces and minimized fatigue, while establishing dynamic equilibrium and maintaining head posture.  They reported that simulating maximal activations (100%) of the neck muscles caused the spine to buckle.  Although the in vivo superficial muscles (Trap, SPL, LS, SCM, and STH) in the free bracing task were approximately a third of the activation values reported by Chancey et al. [52], the deeper muscles (SsCap, SsCerv and Mult) were higher than those reported for the model.  Because these muscles insert at various levels throughout the cervical column, they may be providing the stability required to maintain a stable near-neutral posture. These in vivo data suggest that lower activations of the superficial muscles, and higher activations of the deeper muscles, may be a more realistic activation scheme than has previously been modeled.    160   Figure 6-2: Tensed muscle activity patterns compared to literature The pre-impact “tensed” muscle activity patterns reported in computational models by Brolin et al. [54], Halldin et al. [167] and Chancey et al. [52] (note: semispinalis is described by Halldin et al.; however, it is not clear if this includes both capitis and cervicis).  The group mean activations from the constrained flexion force (left), the constrained extension force (center), and the free tensed (bracing) task (right) for both upright (Up) and inverted (Inv) configurations.    6.1.3 Implications for prevention and protection Several designs have been proposed to prevent neck injuries in rollover accidents, including novel roof designs [67] and seat-mounted airbags [93].  Since the main aim of these devices is to actively change the neck posture, knowledge of the initial posture of the head and neck is critical.  161  This thesis has demonstrated that the pre-impact alignment of the spine is sensitive to whole body inversion, changes in head orientation, and whether the head/neck has dynamically been extended or flexed.  It is plausible that any of these factors could be present in some form in a dynamic rollover.  In addition, based on the results herein, actively tensing the muscles (either with the head free or constrained against a vehicle structure) in anticipation for an impact may have the potential to further alter the spinal posture.    Simulating a biofidelic neck stiffness is important for assessing the potential of neck injury using an ATD neck in a car rollover study [90], particularly while inverted.  In a rollover injury scenario, the head has the potential to pocket in a structure (i.e. the car roof) and someone may “isometrically” activate neck flexor or extensor muscles.  The Hybrid III neck has been reported to be stiffer axially than cadaveric necks [30, 38]; yet, comparisons to in vivo stiffness are largely lacking.  In fact, independent researchers have recently proposed prototype ATD necks designed to incorporate pre-flexion (10 to 30 degrees) and increased flexibility (three times less stiff in extension and flexion) for application in a rollover [205-207].  Not only are comparable in vivo data generally lacking, but based on the results of this work, inversion may result in a different alignment and muscle-induced stiffness compared to upright.  Thus, the interplay between gravity, head orientation, and muscle response is important to obtaining desired stiffness response, particularly at certain joint levels, in an inverted configuration.  The failure mode of the spine is sensitive to the end conditions (i.e. the top and bottom of the spine) [17], and the risk of injury is reduced if the neck can escape the momentum of the incoming torso [94].  Moving the head and neck out of the load path of the torso and increasing 162  curvature may decrease the risk of injury.  It is possible that some of the postures measured in this subject group would actually be protective for head-first impact.  However, cadaveric specimens and computational models of head-first impact have not been tested in the many different configurations observed in this study.  For instance, the free tensed task resulted in an average neck configuration in which the neck pulled forward, curvature increased, and head angle remained constant.  To my knowledge, no cadaveric test has been conducted with this posture while incorporating the preload of active musculature.  Therefore, current neck injury models may not accurately represent the loading characteristics present when someone is actively tensing their neck muscles, thus impeding efforts to develop protective devices.  Not only were inversion, head position, muscle bracing, and dynamic motion important to the average response of the subject group, they were also important to the individual responses.  For instance, the individual changes in vertebral angles for the various tasks relative to upright spanned almost 60° in the upper cervical spine (Figure 6-3).  In the lower cervical spine, some of the individual responses extended well beyond the group average ranges of motion that these vertebrae reached in the extreme positions of dynamic motion.    163   Figure 6-3: Vertebral data of all subjects for all experimental tasks The absolute vertebral angle data (C1 through C7) of every subject for all tasks/conditions (except for upright relaxed).  The data are all relative to the upright relaxed trial for that subject (i.e. an angle above or below 0 means that the angle of that vertebra was different than the angle of that same vertebra measured in the upright relaxed condition).  As a reference for the overall range of motion of these vertebrae, the group average cervical angles for full flexion and full extension (measured during the dynamic motion trial) are also plotted (white bars).    Large standard deviations and confidence intervals were also observed for the muscle activations in all of the experimental conditions.  The large variability observed in both the muscular and postural neck responses suggests that a variety of active responses and intervertebral alignments 164  may need to be considered in ex vivo testing and when modeling neck injury and evaluating prevention strategies.  In fact, inspection of the individual curvature index responses in all the experimental conditions and tasks (see Figure 6-4) reveals that the responses may fit into sub-groups.  For instance, subjects #2, #4, #5, and #7 do not change their curvature indices by a large amount in the majority of the conditions/tasks, whereas subjects #3, #6, #9, and #10 appear to have large curvature increases to several of the conditions/tasks.  The first group tended to have a “straighter” spine initially, which could be a reason for this apparent grouping.  Sex could also play a role, as there are mostly females in the first group and mostly males in the second.  Other factors could be important, such as muscle strength, height, and age.  Although identifying the reasons for these groupings was beyond the scope of this thesis, the apparent grouping does suggest that there may be more than one “typical” response that needs to be considered in the context of injury prevention.  165   Figure 6-4: Changes in all curvature indices The upright resting (solid black line) curvature index (C.I.) for each subject (subject numbers are indicated) and the curvature index measured in all tasks/conditions (subjects are arranged according to the magnitudes of curvature changes).  The vertebral outlines of each subject in the upright resting configuration are illustrated along with the subject’s sex.   1=Inverted-Relaxed       5=Upright-Flexion (constrained tensed)   9=Upright-Flexion (dynamic) 2=Inverted-Forward       6=Inverted-Flexion (constrained tensed)  10=Upright-Extension (dynamic) 3=Upright-Tensed (free)       7=Upright-Extension (constrained tensed)  11=Inverted-Flexion (dynamic) 4=Inverted-Tensed (free)       8=Inverted-Extension (constrained tensed)  12= Inverted-Extension (dynamic)  6.2 Limitations and recommendations Perhaps the most important limitation to this work is the static nature of the tests completed.  It is not clear if rollover occupants react to a rollover by adopting any of the postures studied as there are no studies of head/neck/torso alignment in actual rollovers.  From published images of 166  statically held, steady state rolls, and free-fall drops of human volunteers, subjects appear to adopt a variety of responses: head extended backward under gravity, forward gaze, slight lateral bending, and forward-flexion [74, 84, 88].  Based on these limited data, it is believed that the conditions used in this study reflect some responses in a subset of rollover conditions, but do not capture the full variety of postures assumed in real-life rollovers.   Similarly, rollovers involve complex ballistic motions, often with 3-4g of centripetal acceleration [87], whereby the occupant’s neck may be compressed against the vehicle roof [92]  and it is not known if a voluntary (as opposed to reflexive) response to these accelerations is possible. This work was an attempt to create an initial study of static inversion to understand the changes in the mechanics of the neck by inducing tension due to gravity.  The active muscle responses in this thesis work may reflect a coordinated response that is possible under the forces involved in lower roll rates (i.e. 180 deg/sec) where an occupant’s neck may be under tensile forces equivalent to the weight of the head (approximately 40 N) [91].  Although these findings may be somewhat limited in the context of real-world rollovers they do apply to a near neutral posture; therefore, they are directly applicable to the postures used in current cadaveric drop tests and computational models for understanding axial neck injury.      The subject group in this thesis represents a sample of the population between the ages 19-45 years old, with an equal distribution of males and females and no history of neck injury.   The group’s mean postural responses reported in this thesis may reflect the age and sex distribution of this particular subject group.  The passive flexion/extension range of motion of the cervical spine decreases with age and there are age dependent differences between sexes [99, 117, 208].  167  Hypomobility (decreased motion) in the cervical spine joints has also been demonstrated for subject groups with evidence of degeneration, disc complications, or previous whiplash trauma [209].  Klinich et al. [159] described the average curvature index for a large population (both genders, various statures, and a large age range) in an upright automotive seated posture to be 2.4% (with a standard deviation of 1.6%).  This was higher than the average curvature index for subjects in the current study (upright relaxed was 1.4%), consistent with a younger population with taller than average females [159].  Takeshima et al. [123] categorized a healthy population based on cervical spine shapes (Table 6-3) and our subject group appears to be most similar to group A (lordotic) with a slight tendency toward a straighter posture based on average Cobb angles [210]; two subjects had a global cervical curvature (C2-C7) of less than 4° (classified as “straight”) [211].  The larger angle of C1 may reflect a slight extension of the head [114] or the subjects being in a seated position [115] (with a 5-point harness).  Thus, in addition to the possible effects of age and sex, the fact that subjects were seated and restrained could have also influenced the postural findings.            168  Table 6-3: Comparison of upright relaxed group alignment to various cervical spine shapes The inclination of each vertebral joint level and Cobb angle reported by Takeshima et al. [123] for five subjects groups (and number of subjects, n) with different alignments (lordotic, straight, kyphotic, s-shape with upper lordosis and lower kyphosis, and s-shape with upper kyphosis and lower lordosis) compared to the group average in this thesis.    Level Group A Lordotic (n=48) Group B Straight (n=53) Group C Kyphotic (n=40) Group D S-shape (upper lordosis) (n=36) Group E S-shape (lower lordosis) (n=27) Current subject group* C1 -7.4 -1.1 -2.5 -5.5 -4.9 -16.7 C2/C3 9.9 16.3 23.5 16.9 15.5 14.3 C3/C4 13.5 16.3 21.2 20.2 13.2 15.8 C4/C5 17.0 16.0 17.1 21.4 11.3 16.4 C5/C6 19.5 14.6 11.7 17.4 12.9 18.3 C6/C7 22.8 16.3 10.8 15.7 18.0 22.5 Cobb 17.5 2.8 -16.5 -1.6 4 11.6 * Disc inclination and Cobb angle were calculated as reported by Takeshima et al. [123]   EMG is a valuable technique for identifying muscles that are activated in a particular task and, since it is normalized, it is also useful for studying the relative activations of the muscles.  However, there are several inherent limitations and assumptions when using EMG.  Ultrasound guidance was used to identify the muscle components and verify that the tip of the needle was near the belly of the compartment.  The tips of the wires were constructed with hooks to grab the muscle fibers and wires with Teflon coating were selected to reduce the friction between the neck tissues and wire.  The depth of insertion was measured when the wires were removed to confirm that wires were at a depth consistent with the depth of the muscle.  Inspection of individual recordings throughout the entire data collection appeared to be consistent with the sensors remaining in place.  However, we did not explicitly confirm that the wires remained in 169  their respective muscle compartments throughout the four experiments.  Furthermore, the localized recordings of the muscles (i.e. fine-wire) may only represent the activity with a part of the muscle [176].  Thus, although ultrasound guidance was used to consistently guide the wires toward the center of each muscle’s belly, the EMG findings of this thesis may reflect activations of the recording area only.     Another challenge with EMG is providing a meaningful context to the voltages recorded from within the muscles.  Normalization is a method that allows muscles and subjects to be compared to each other (regardless of electrode spacing or anatomical differences); however, the magnitudes of the reported muscle activations are influenced by the normalization technique and spinal posture (i.e. static versus dynamic or maximal versus sub-maximal) [212-216].  There is currently a lack of consensus as to which technique should be used for neck muscle normalization [212].  In this study, subjects maximally pushed in 7 different directions with the aim of eliciting a maximal contraction of the neck muscles of interest.  Exerting isometric forces against a rigid surface [171, 174, 177] or resisting torques [176] in various directions is useful for isolating spatial tuning of individual muscles.  Pushing in various directions is reportedly highly reliable for EMG measurements, and reliability improves with visual feedback (which was provided to the subjects in the current study) [217, 218].  Yet, it is uncertain whether subjects were actually able to maximally contract each of the 8 neck muscles in these directions as the majority of people are unfamiliar with using their heads to perform maximum exertions [212].  Activations well above those measured during maximum voluntary contractions (i.e. 257% MVC), which were associated with pain and injury, have been described in pilots during air combat [219].  Muscle activations higher than 100% MVC have also been noted in subjects 170  exposed to forward sled accelerations (with normalization techniques similar to those presented here) [171, 220].  Indeed, each muscle (with the exception of SPL) had activations above MVC in several subjects and tasks of this thesis.  Approximately 3% of all reported individual RMS values in this thesis were above 100% MVC, although no subjects were injured.  This may have inflated the size of the muscle effects reported in this work.  However, even though some of the recordings may not be relative to true maximums, they still provide a measure of activation relative to MVC tasks that were standardized and consistent among muscles and subjects.  Future research should both focus on addressing the limitations inherent to these present studies and extending these findings to more relevant injury scenarios.  Each of the four sub-experiments provides novel data for validating and extending computational and experimental models for injury biomechanics and basic neck physiology studies.  However, as with all human subject experimentation, some of the tasks could be improved.  The tensed task involved subjects drawing their neck in and activating their neck muscles in an attempt to simulate someone preparing for an impact.  This was not a realistic simulation of the moments leading up to a real-life impact.  Subjects did not experience the centripetal accelerations or the effects of free fall that may be present in a real rollover.  Furthermore, there was no true element of fear or startle which could have resulted in a muscular response that is reflexive instead of voluntary or the use of their arms (such as lifting them above their heads).  Furthermore, rollover occupants may have an active response to the lack of headroom (i.e. “ducking”) [89] or neck flexion induced by the seatbelt [90], and these responses were not simulated in the present studies.  If a simulation of a rollover environment is desired, capturing the occupant interaction with a safety belt and the roof 171  and incorporating dynamic parameters relevant to rollovers and the element of surprise will be important.   Furthermore, we explored four ways of manipulating alignment; however, there are a myriad of ways this could be explored.  A study of the muscle activity and orientation dependent differences in the “aligned” posture would be of particular interest to injurious situations.  Aligning the cervical spine column places the neck into a vulnerable position for catastrophic injury.  Torg et al. [221] suggested that head-first tackles in football with slight head flexion (into the “aligned” posture) results in the majority of cervical fractures, dislocations, and subluxations.  The aligned posture, thought to occur at approximately 15 to 30 degrees of head flexion, typically straightens the cervical lordosis and (when the occipital condyles are over T1) aligns the column into its stiffest axis [18].  Compression-flexion modes of injury have been observed in the lower cervical spine in an ex vivo aligned posture [37].  However, these ex vivo tests were conducted by forcing the osteo-ligamentous spine into an aligned posture of approximately 10 to 20 degrees of column pre-flexion.  Inspection of the muscle activities of the subjects during the dynamic task at the phase of motion from neutral head posture to fully flexed (see Appendix F, Figure F-10 to Figure F-19) reveals substantial neck muscle involvement for some subjects, particularly in the inverted case.  Several subjects had activations of STH on the order of 100% of their MVC while flexing upside-down.  Activation of several extensors was also evident in the flexion phase.  Although the activity levels of individual subjects were varied, flexion into an aligned spine does not appear to be typically achieved (especially while inverted) without considerable muscle activity.  Future studies should look in more detail at the muscle forces involved in arriving at, and maintaining, an aligned cervical column. 172  The shoulder interaction with the 5-point harness likely influenced both the postural and muscular outcomes of this thesis work.  For safety reasons, it would be difficult to conduct a study such as this without the 5-point harness; therefore, it would be useful to explore the degree to which this harness affects the outcome of this or any future study.  Furthermore, the thoracic spine is difficult to view in sagittal X-rays; even though subjects were strapped down, the thoracic spine could have realigned during the simulated tasks.  Tracking the torso motions with external markers may reveal relevant changes in torso position with respect to the neck.  In pure rollover (restrained) occupants, where neck injuries are sustained from roof impacts, 26% also sustain serious injuries to the thoracic spine [12].  The neck-torso interface is an important contributing factor to injury in the cervical spine [35, 36] and accurate tracking of the top of the thoracic spine is important to simulating kinematics in models of cervical spine motion [57].     The objective of the next phase of this project should be to perform a simulation of a rollover car accident while measuring both cervical spine kinematics and neck muscle EMG.  Trip-over rollover accidents account for 57% of passenger car rollovers and occur when the vehicle tires strike an obstruction or contact soil while moving laterally [222].  It is for this reason the inversion device used in this study was designed to rotate the subject laterally to attain an upside-down configuration.  Guards were built to prevent full rotation of the device and to lock it into a static inverted position.  However, these guards could easily be removed to allow a full rotation.  To enable the experimenter to easily rotate the subject, the device was also designed with detachable weights that counterbalance the weight of the subject.  These weights enable a balanced dynamic rotation and with the aid of motors, would allow a speed-controlled rotation.  To simulate a rollover accident, the inversion frame could also be accelerated sideways down a 173  track where it would come to a stop and commence a turnover motion (similar to a vehicle hitting a curb or soil boundary while traveling sideways and then flipping over).   Ideally, the extraction of muscle forces from EMG activations would provide experimenters with the forces required to simulate musculature and in vivo like pre-test realignment of a cadaveric spine prior to impact tests.  Calculating the net muscle forces from EMG data, even when normalized, is not trivial.  It might be possible to estimate muscle forces using average geometrical data and a computational model; however, this level of analysis was not the focus of this thesis.  Previous computational models, such as that described by Chancey et al. [52], have used muscle activation levels (0%-100%) as inputs to calculate the level of neck muscle forces, as well as the force at each cervical level.  To enable this to be performed by others, this thesis includes normalization forces (in Newtons), normalization direction, and EMG RMS levels (as a percent of maximum contraction) for each muscle and each subject for all experimental tasks and conditions.    Since only two anterior muscles were measured (STH and SCM) future studies may wish to explore other major flexor muscles, particularly in an upside-down task. Vasavada et al. [172]  reported that the longus capitis and colli as well as the scalenus anterior contribute to the flexion moment of the neck.  In addition, other muscles contribute to the dynamic flexion and extension of the cervical spine, such as longus capitis and colli and scalenus medius and anterior [113].  Therefore, the activation patterns reported herein may also be absent of muscles that are important to this motion. At some point in the four experiments, all eight muscles revealed contributions to head and neck posture.  The large variability in subject responses would, at least 174  in part, be explained by individual muscle strategies.  However, strong correlations did not appear to exist between neck realignment and the eight muscles recorded.  Future work should explore the possible role of other muscles in neck realignment as attempts to correlate posture and muscles would be largely beneficial to cadaveric and computational modeling of injury.   The 2D nature of the X-ray system limited the ability to conduct motions in directions other than the sagittal plane or to measure any out-of-plane motions.  With the complex and often coupled 3D motion of the cervical spine, planar analysis is prone to large errors due to X-ray projection principles and out-of-plane vertebral motion.  These out-of-plane vertebral motions occur parallel to the measurement plane of the system and thus are near impossible to measure with a single X-ray projection.  In sagittal plane evaluation of the lumbar spine, out-of-plane motion can result in errors as high as 292 mm in identifying rotational axes [223].  Pilot tests with the C-arm used in this study indicated that intervertebral angle error induced by large out-of-plane motions was as high as 3°.  When studying a dynamic event with rapid variations of acceleration direction (such as a rollover simulation) in which subject motion cannot be constrained to one plane, 3D motion capture is essential.  Therefore, the necessity for measuring 3D motion of the cervical spine should be considered for future work.    6.3 Contributions The data obtained for this thesis are a comprehensive in vivo data set of synchronized neck muscle response and cervical spine alignment in multiple tasks, both in right-side-up and upside-down postures.  These unique data can be used to improve and validate current and future cadaveric and computational cervical spine models, both under physiologic scenarios and in the 175  context of preparing for an impact.  A better understanding of the relationship between muscle action and spinal alignment will aid in elucidating the mechanics of cervical spine and spinal cord injuries, and evaluating potential prevention strategies for these injuries.    This is the first (in vivo) human subject study to: • Invert human subjects while measuring alignment and muscle activity during a variety of experimental tasks • Combine fluoroscopy and indwelling EMG for isometric force generation and dynamic flexion and extension motions • Quantify neck realignment and in vivo muscle patterns with tensed neck muscles • Explore intervertebral stiffness changes under isometric muscle forces  • Compare the relaxed and dynamic position of the spine at neutral neck eccentricity  This work also provided the basis for the following methodological development for future studies: • Design and construction of an inversion device • Development and approval of an ethics protocol for live subject fluoroscopy and indwelling EMG • A technique for measuring alignment changes while developing isometric forces  • Development of a custom-programmed automatic image tracking algorithm for cervical spine fluoroscopy • Extensive validation of the dynamic X-ray kinematics and tracking algorithm  176  The ultimate goal of this research focus is to capture a person’s active response in a dynamic rollover environment.  Thus, although these tasks were aimed at exploring several factors that influence alignment, they also provided the basis to develop methods for future rollover testing.  Therefore, the methods developed and detailed in this thesis are themselves contributions to the fields of injury biomechanics, accident reconstruction, and basic cervical spine biomechanics.  6.4 Conclusions Alteration of the cervical alignment is known to influence the resulting injuries from a head-first axial impact.  By changing the overall neck alignment, previous researchers have documented a dependence of clinical injuries on neck alignment.  Cadaveric testing offers an avenue of changing the neck alignment in order to study the link between the posture and subsequent injuries.  Ex vivo testing also offers an avenue for exploring and validating promising neck injury prevention devices.  The postural inputs to these cadaveric models are important parameters for studying and preventing neck injuries.  Yet, little attention has previously been paid to the manner in which live humans can manipulate their spinal posture.  Similarly, computational models allow researchers to manipulate the spinal postures and, more importantly, simulate passive and active musculature.  But, as with ex vivo models, researchers are lacking realistic kinematic and muscular inputs to these computational models, precluding relevant simulations and providing challenges to validating model assumptions.  The goal of this thesis was to understand the mechanical factors important to spinal realignment.  Furthermore, the aim was to provide novel data sets of in vivo cervical spine postures that are more relevant to head-first impact injury than have previously existed.   177  In general, orientation (upright versus upside-down), head posture, active muscle contractions, and dynamic motion all had significant influences on spinal alignment.  These findings indicate that simulation of upright relaxed posture is not sufficient for replicating the spinal alignment in all situations leading up to a head-first impact.  In fact, all of the factors listed above were found to significantly change at least one metric of spinal alignment.  Extension of the relaxed head when upside-down increased the curvature of the spine, while a forward-looking gaze pulled the lower spine into flexion.  Orientation in general had a large effect on the alignment of the spine throughout other tasks; however, the differences from the upright relaxed posture were highly task dependent. Most notably, inversion tended to magnify changes in the upper cervical spine.  Actively generating muscle forces in a free “unconstrained” condition resulted in increases in curvature and forward translations of the neck.  Actively generating forces with the neck muscles in a slow ramping task (with head constraint) also resulted in increases in curvature.  Specific changes in intervertebral angles depended on the direction of forces and, although significant, the small magnitude of the angle changes highlighted the degree of neck stiffness provided by the neck muscles.  Finally, dynamic muscle-driven flexion/extension motions of the head and neck revealed that the dynamic and relaxed neutral postures are not the same and that the posture depends on the direction of motion.    Throughout all of these four tasks, muscle activities were present to some degree, with small activations during the resting trials and higher activations during the force and motion generating tasks.  The activations of these eight muscles, albeit task specific, highlight the importance of muscle forces in simulating a biofidelic head and neck for injury research.    178  Finally, large variations in kinematic and muscular responses were observed at the individual subject level in all four experiments.  This attests to the fact that it may be critical to simulate not only an average response but the variety of responses that individuals may adopt prior to a head-first impact.      179  References 1. Conroy, C., Hoyt, D.B., Eastman, A.B., Erwin, S., Pacyna, S., Holbrook, T.L., Vaughan, T., Sise, M., Kennedy, F., and Velky, T., 2006. Rollover crashes: predicting serious injury based on occupant, vehicle, and crash characteristics. Accident Analysis & Prevention 38 (5), 835-42. 2. Yoganandan, N., Haffner, M., Maiman, D.J., Nichols, H., Pintar, F.A., Jentzen, J., Weinshel, S.S., Larson, S.J., and Sances Jr., A., 1989. Epidemiology and injury biomechanics of motor vehicle related trauma to the human spine. SAE Technical Paper No. 892438.  3. O'Connor, P.J. and Brown, D., 2006. Relative risk of spinal cord injury in road crashes involving seriously injured occupants of light passenger vehicles. Accident Analysis & Prevention 38 (6), 1081-6. 4. Blackmore, M.E., Goswami, T., and Chancey, C., 2012. Cervical spinal injuries and risk assessment, injury and skeletal biomechanics. D.T. Goswami, ISBN: 978-953-51-0690-6, DOI: 10.5772/50588. Available from: http://www.intechopen.com/books/injury-and-skeletal-biomechanics/cervical-spinal-injuries-and-risk-assessment 5. Bedewi, P.G., Godrick, D.A., Digges, K.H., and Bahouth, G.T., 2003. An investigation of occupant injury in rollover: NASS-CDS analysis of injury severity and source by rollover attributes. In Proceedings of the International Technical Conference on the Enhanced Safety of Vehicles. Nagoya, Japan. 6. Rains, G.C. and Kanianthra, J.N., 1995. Determination of the significance of roof crush on head and neck injury to passenger vehicle occupants in rollover crashes. SAE Technical Paper no. 950655. 1-11. 7. Mandell, S.P., Kaufman, R., Mack, C.D., and Bulger, E.M., 2010. Mortality and injury patterns associated with roof crush in rollover crashes. Accident Analysis & Prevention 42 (4), 1326-31. 8. Moffatt, E.A., Cooper, E.R., Orlowski, K.F., Marth, D.R., Croteau, J.J., and Carter, J.W., 2003. Matched-pair rollover impacts of rollcaged and production roof cars using the controlled rollover impact system (CRIS). SAE Technical Paper No. 2003-01-0172.  9. Raddin, J., Cormier, J., Smyth, B., Croteau, J., and Cooper, E., 2009. Compressive neck injury and its relationship to head contact and torso motion during vehicle rollovers. SAE Technical Paper No. 2009-01-0829.  10. Bahling, G.S., Bundorf, R.T., Moffatt, E.A., and Orlowski, K.F., 1995. The influence of increased roof strength on belted and unbelted dummies in rollover and drop tests. Journal of Trauma 38 (4), 557-63. 180  11. Foster, J.B., Kerrigan, J.R., Nightingale, R.W., Funk, J.R., Cormier, J.M., Bose, D., Sochor, M.R., Ridella, S.A., Ash, J.H., and Crandall, J., 2012. Analysis of cervical spine injuries and mechanisms for CIREN rollover crashes. In Proceedings of the International Research Council on Biomechanics of Injury (IRCOBI) Conference. Dublin, Ireland. 12. Bambach, M.R., Grzebieta, R.H., McIntosh, A.S., and Mattos, G.A., 2013. Cervical and thoracic spine injury from interactions with vehicle roofs in pure rollover crashes. Accident Analysis and Prevention 50 34-43. 13. Hu, J., Lee, J., Yang, K., and King, A., 2005. Injury patterns and sources of non-ejected occupants in trip-over crashes: A survey of NASS-CDS database from 1997 to 2002. Annual Proceedings of the Association for the Advancement of Automotive Medicine 49 119-32. 14. Ridella, S. and Eigen, A.M., 2008. Biomechanical investigation of injury mechanisms in rollover crashes from the CIREN database. In Proceedings of the International Research Council on Biomechanics of Injury (IRCOBI) Conference. Bern, Switzerland. 15. Winters, J.M. and Peles , J.D., 1990. Neck muscle activity and 3-D head kinematics during quasi-static and dynamic tracking movements, in Multiple Muscle Systems: Biomechanics and Movement Organization, J.W. Winters, S., Editor, Springer-Verlag, New York. 16. Nightingale, R.W., McElhaney, J.H., Richardson, W.J., Best, T.M., and Myers, B.S., 1996. Experimental impact injury to the cervical spine: relating motion of the head and the mechanism of injury. The Journal of Bone & Joint Surgery. American volume 78 (3), 412-21. 17. Myers, B.S. and Winkelstein, B.A., 1995. Epidemiology, classification, mechanism, and tolerance of human cervical spine injuries. Critical Reviews in Biomedical Engineering 23 307-409. 18. Pintar, F., Yoganandan, N., Voo, L., Cusick, J., Maiman, D.J., and Sances, A., 1995. Dynamic characteristics of the human cervical spine. SAE Technical Paper No. 952722.  19. Yoganandan, N., Pintar, F., Gennarelli, T., Eppinger, R., and Voo, L., 1999. Geometric effects on the mechanism of cervical spine injury due to head impact. In Proceedings of the International Research Council on Biomechanics of Injury (IRCOBI) Conference. Spain. 20. Maiman, D.J., Yoganandan, N., and Pintar, F.A., 2002. Preinjury cervical alignment affecting spinal trauma. Journal of Neurosurgery 97 (1), 57-62. 21. Chang, D.G., Tencer, A.F., Ching, R.P., Treece, B., Senft, D., and Anderson, P.A., 1994. Geometric changes in the cervical spinal canal during impact. Spine 19 (8), 973-980. 181  22. Pintar, F.A., Yoganandan, N., Maiman, D.J., and Sances, A.J., 1995. Cervical spinal bony injury and the potential for cord injury. In Proceedings of the CDC Injury Prevention through Biomechanics Symposium. Detroit, MI. 23. Bates-Carter, J., Ching, R.P., and Tencer, A.F., 1995. Transient changes in canal geometry during axial burst fracture of the cervical spine. In Proceedings of the 5th Injury prevention through biomechanics symposium. Detroit, MI. 24. Carter, J.W., Mirza, S.K., Tencer, A.F., and Ching, R.P., 2000. Canal geometry changes associated with axial compressive cervical spine fracture. Spine 25 (1), 46-54. 25. Allen, B.L., Ferguson, R.L., Lehmann, T.R., and Obrien, R.P., 1982. A mechanistic classification of closed, indirect fractures and dislocations of the lower cervical spine. Spine 7 (1), 1-27. 26. Hodgson, V.R. and Thomas, L.M., 1980. Mechanisms of cervical spine injury during impact to the protected head. SAE Technical Paper No. 801300.  27. Pintar, F., Voo, L.M., Yoganandan, N., Cho, T.H., and Maiman, D.J., 1998. Mechanisms of hyperflexion cervical spine injury. In Proceedings of the International Research Council on Biomechanics of Injury (IRCOBI) Conference. Goeteborg, Sweden. 28. Nightingale, R.W., McElhaney, J.H., Camacho, D.L.A., Kleinberger, M., and Myers, B.S., 1997. The dynamic responses of the cervical spine: buckling, end conditions and tolerance in compressive impacts. SAE Technical Paper No. 973344.  29. Penning, L., 1995. Kinematics of cervical spine injury. A functional radiological hypothesis. European Spine Journal 4 (2), 126-32. 30. Myers, B.S., McElhaney, J.H., Richardson, W.J., Nightingale, R.W., and Doherty, B.J., 1991. The influence of end condition on human cervical spine injury mechanisms. SAE Technical Paper No. 912915. 391-9. 31. Bauze, R.J. and Ardran, G.M., 1978. Experimental production of forward dislocation in the human cervical spine. Journal of Bone and Joint Surgery - British Volume 60-B (2), 239-45. 32. Oner, F.C., 2009. Spinal injury classification systems, in Neurotrauma and critical care of the spine, J. Jallo and A.R. Vaccaro, Editors, Thieme Medical Publishers, Inc., New York, NY. 33. Vaccaro, A.R., Hulbert, J., Patel, A.A., Fisher, C., Dvorak, M., Lehman, R.A., Anderson, P., Harrop, J., Oner, F.C., Arnold, P., Fehlings, M., Hedlund, R., Madrazo, I., Rechtine, G., Aarabi, B., and Shainline, M., 2007. The subaxial cervical spine injury classification system. Spine 32 (21), 2365-2374. 182  34. McElhaney, J.H., Nightingale, R.W., Winkelstein, B.A., Chancey, V.C., and Myers, B.S., 2002. Biomechanical aspects of cervical trauma, in Accidental Injury: Biomechanics and Prevention, A.M. Nahum and J.W. Melvin, Editors, Springer Science+Business Media, Inc., New York, NY. 35. Yoganandan, N., Sances, A.J., Maiman, D.J., Myklebust, J.B., Pech, P., and Larson, S.J., 1986. Experimental spinal injuries with vertical impact. Spine 11 (9), 855-860. 36. Nusholtz, G., Huelke, D., Lux, P., Alem, N., and Montalvo, F., 1983. Cervical spine injury mechanisms. SAE Technical Paper No. 831616.  37. Pintar, F.A., Yoganandan, N., Reinartz, J., Sances, A.J., Harris, G., and Larson, S.J., 1989. Kinematic and anatomical analysis of the human cervical spinal column under axial loading. SAE Technical Paper No. 892436.  38. Yoganandan, N., Sances, A., Jr.;, and Pintar, F., 1989. Biomechanical evaluation of the axial compressive responses of the human cadaveric and manikin necks. Journal of Biomechanical Engineering 111 (3), 250-5. 39. Maiman, D.J., Sances, A., Myklebust, J.B., Larson, S.J., Houterman, C., Chilbert, M., and Elghatit, A.Z., 1983. Compression injuries of the cervical spine: A biomechanical analysis. Neurosurgery 13 (3), 254-260. 40. McElhaney, J.H., Doherty, B.J., Paver, J.G., Myers, B.S., and Gray, L., 1988. Combined bending and axial loading responses of the human cervical spine. SAE Technical Paper No. 881709. 21-28. 41. McElhaney, J.H., Paver, J.G., McCrackin, H.J., and Maxwell, G.M., 1983. Cervical spine compression responses. SAE Technical Paper No. 831615.  42. Pintar, F.A., Yoganandan, N., Pesigan, M., Reinartz, J., Sances, A., Jr.;, and Cusick, J.F., 1995. Cervical vertebral strain measurements under axial and eccentric loading. Journal of Biomechanical Engineering 117 (4), 474-8. 43. Carter, J.W., Ku, G.S., Nuckley, D.J., and Ching, R.P., 2002. Tolerance of the cervical spine to eccentric axial compression. Stapp Car Crash Journal 46 441-59. 44. Nusholtz, G.S., Melvin, J.W., Huelke, D.F., Alem, N.M., and Blank, J.G., 1981. Response of the cervical spine to superior-inferior head impact. SAE Technical Paper No. 811005.  45. Alem, N., Nusholtz, G., and Melvin, J., 1984. Head and neck response to axial impacts. SAE Technical Paper No. 841667.  46. Nightingale, R.W., McElhaney, J.H., Richardson, W.J., and Myers, B.S., 1996. Dynamic responses of the head and cervical spine to axial impact loading. Journal of Biomechanics 29 (3), 307-18. 183  47. Viano, D.C., Parenteau, C.S., Gopal, M.M., and James, M.B., 2009. Vehicle and Occupant Responses in a Friction Trip Rollover Test. SAE Technical Paper No. 2009-01-0830.  48. Saari, A., 2006. Spinal cord deformation during axial impact injury of the cervical spine. M.A.Sc thesis, University of British Columbia,  49. Herbst, B.F., 1998. Fidelity of anthropometric test dummy necks in rollover accidents. International Technical Conference on the Enhanced Safety of Vehicles 1998 2093-2097. 50. Nightingale, R.W., Camacho, D.L., Armstrong, A.J., Robinette, J.J., and Myers, B.S., 2000. Inertial properties and loading rates affect buckling modes and injury mechanisms in the cervical spine. Journal of Biomechanics 33 (2), 191-7. 51. Hu, J.W., Yang, K.H., Chou, C.C., and King, A.I., 2008. A numerical investigation of factors affecting cervical spine injuries during rollover crashes. Spine 33 (23), 2529-2535. 52. Chancey, V.C., Nightingale, R.W., Van Ee, C.A., Knaub, K.E., and Myers, B.S., 2003. Improved estimation of human neck tensile tolerance: Reducing the range of reported tolerance using anthropometrically correct muscles and optimized physiologic initial conditions. Stapp Car Crash Journal 47 135-153. 53. Van Ee, C.A., Chasse, A.L., and Myers, B.S., 2000. Quantifying skeletal muscle properties in cadaveric test specimens: effects of mechanical loading, postmortem time, and freezer storage. Journal of Biomechanical Engineering 122 (1), 9-14. 54. Brolin, K., Hedenstierna, S., Halldin, P., Bass, C., and Alem, N., 2008. The importance of muscle tension on the outcome of impacts with a major vertical component. International Journal of Crashworthiness 13 (5), 487-498. 55. van der Horst, M., Thunnissen, J., Happee, R., van Haaster, R., and Wismans, J., 1997. The influence of muscle activity on head-neck response during impact. SAE Technical Paper No. 973346.  56. Brolin, K., Halldin, P., and Leijonhufvud, I., 2005. The effect of muscle activation on neck response. Traffic Injury Prevention 6 (1), 67-76. 57. Dibb, A.T., Cox, C.A., Nightingale, R.W., Luck, J.F., Cutcliffe, H.C., Myers, B.S., Arbogast, K.B., Seacrist, T., and Bass, C.R., 2013. Importance of muscle activations for biofidelic pediatric neck response in computational models. Traffic Injury Prevention 14 Suppl S116-27. 58. Meijer, R., van Hassel, E., Broos, J., Elrofai, H., van Rooij, L., and van Hooijdonk, P., 2012. Development of a multibody human model that predicts active and passive human behaviour. In Proceedings of the International Research Council on Biomechanics of Injury (IRCOBI) Conference. Dublin, Ireland. 184  59. van der Horst, M.J., 2002. Human head neck response in frontal, lateral and rear end impact loading: Modelling and validation. PhD thesis, Eindhoven University of Technology,  60. Vasavada, A.N., Brault, J.R., and Siegmund, G.P., 2007. Musculotendon and fascicle strains in anterior and posterior neck muscles during whiplash injury. Spine 32 (7), 756-65. 61. Hedenstierna, S., 2008. 3D finite element modeling of cervical musculature and its effect on neck injury prevention. PhD thesis, Royal Institute of Technology, Stockholm, Sweden. 62. Deng, Y. and Fu, J., 2002. Simulation and identification of the neck muscle activities during head and neck flexion whiplash. SAE Technical Paper No. 2002-01-0017.  63. Ejima, S., Ono, K., Kaneoka, K., and Fukushima, M., 2005. Development and validation of the human neck muscle model under impact loading. In Proceedings of the International Research Council on Biomechanics of Injury (IRCOBI) Conference. Prague, Czech Republic. 64. Mathys, R. and Ferguson, S.J., 2012. Simulation of the effects of different pilot helmets on neck loading during air combat. Journal of Biomechanics 45 (14), 2362-7. 65. Camacho, D.L.A., Nightingale, R.W., and Myers, B.S., 1999. Surface friction in near-vertex head and neck impact increases risk of injury. Journal of Biomechanics 32 (3), 293-301. 66. Camacho, D.L.A., Nightingale, R.W., and Myers, B.S., 2001. The influence of surface padding properties on head and neck injury risk. Journal of Biomechanical Engineering 123 (5), 432-439. 67. Halldin, P.H., Brolin, K., Kleiven, S., von Holst, H., Jakobsson, L., and Palmertz, C., 2000. Investigation of conditions that affect neck compression-flexion injuries using numerical techniques. SAE Technical Paper No. 2000-01-SC10.  68. Yang, K.H., Zhu, F., Luan, F., Zhao, L., and Begeman, P.C., 1998. Development of a finite element model of the human neck. SAE Technical Paper No. 983157.  69. Nightingale, R.W., Richardson, W.J., and Myers, B.S., 1997. The effects of padded surfaces on the risk for cervical spine injury. Spine 22 (20), 2380-2387. 70. Hu, J.W., Ma, C.S., Yang, K.H., Chou, C.C., King, A.I., Mccoy, R.W., and Chen, R.J., 2007. Development of a finite element model for simulation of rollover crashes. Proceedings of the ASME International Mechanical Engineering Congress and Exposition 16 423-432. 185  71. Bahling, G.S., Bundorf, R.T., Kaspzyk, G.S., Moffatt, E.A., Orlowski, K.F., and Stocke, J.E., 1990. Rollover and drop tests - The influence of roof strength on injury mechanics using belted dummies. SAE Technical Paper No. 902314.  72. Moffatt, E.A., Cooper, E.R., Croteau, J.J., Parenteau, C., and Toglia, A., 1997. Head excursion of seat belted cadaver, volunteers and Hybrid III ATD in a dynamic/static rollover fixture. SAE Technical Paper No. 973347.  73. Seeman, M.R., Muzzy III, W.H., and Lustick, L.S., 1986. Comparison of human and Hybrid III head and neck dynamic response. SAE technical Paper No. 861892.  74. Lai, W., Ewers III, B., Richards, D., Carhart, M., Newberry, W., and Corrigan, C.F., 2005. Evaluation of human surrogate models for rollover. SAE Technical Paper No. 2005-01-0941.  75. Adamec, J., Praxl, N., Miehling, T., Muggenthaler, H., and Schonpflug, M., 2005. The occupant kinematics in the first phase of a rollover accident - Experiment and simulation. In Proceedings of the International Research Council on Biomechanics of Injury (IRCOBI) Conference. Prague, Czech Republic. 76. Muggenthaler, H., Adamec, J., Praxl, N., and Schonpflug, M., 2005. The influence of muscle activity on occupant kinematics. In Proceedings of the International Research Council on Biomechanics of Injury (IRCOBI) Conference. Prague, Czech Republic. 77. Mertz, H.J. and Patrick, L.M., 1967. Investigation of the kinematics and kinetics of whiplash. SAE Technical Paper #670919.  78. Mertz, H.J., Neathery, R.F., and Culver, C.C., 1973. Performance requirements and characteristics of mechanical necks, in Human impact response:  measurement and simulation, W.F. King and H.J. Mertz, Editors, Plenum Press, New York,pp. 263-288. 79. Prasad, P., Kim, A., and Weerappuli, D.P.V., 1997. Biofidelity of anthropomorphic test devices for rear impact. SAE Technical Paper #973342.  80. Culver, C.C., Neathery, R.F., and Mertz, H.J., 1972. Mechanical necks with humanlike responses. SAE Technical Paper #720959.  81. Schneider, L.W., Foust, D.R., Bowman, B.M., Snyder, R.G., Chaffin, D.B., Abdelnour, T.A., and Baum, J.K., 1975. Biomechanical properties of the human neck in lateral flexion. SAE Technical Paper #751156.  82. Ewing, C.L., Thomas, D.J., Lustick, L., Muzzy, W.H., Willems, G.C., and Majewski, P., 1978. Effect of initial position on the human head and neck response to +Y impact acceleration. SAE Technical Paper #780888.  83. Davidsson, J., Lövsund, P., Ono, K., Svensson, M.Y., and Inami, S., 1999. A Comparison between volunteer, BioRID P3 and Hybrid III performance in rear impacts. In 186  Proceedings of the International Research Council on Biomechanics of Injury (IRCOBI) Conference. Sitges, Spain. 84. Friedman, D. and Chng, D., 1998. Human subject experiments in occupant response to rollover with reduced headroom. SAE Technical Paper No. 980212.  85. Friedman, K. and Friedman, D., 1998. Upper interior head, face and neck injury experiments. In Proceedings of the Enhanced Safety of Vehicles (ESV). Windor, Ontario. 86. Yamaguchi, G.T., Carhart, M.R., Larson, R., Richards, D., Pierce, j., Raasch, C.C., Scher, I., and Corrigan, C.F., 2005. Electromyographic activity and posturing of the human neck during rollover tests. SAE Technical Paper No. 2005-01-0302.  87. Orlowski, K.F., Bundorf, R.T., and Moffatt, E.A., 1985. Rollover crash tests - The influence of roof strength on injury mechanics. SAE Technical Paper No. 851734.  88. Moffatt, E., Hare, B., Hughes, R., Lewis, L., Iiyama, H., Curzon, A., and Cooper, E., 2003. Head excursion of restrained human volunteers and Hybrid III dummies in steady state rollover tests. Annual Proceedings of the Association for the Advancement of Automotive Medicine 47 445-65. 89. Paver, J.G., Friedman, D., Carlin, F., Bish, J., Caplinger, J., and Rohde, D., 2008. Rollover crash neck injury replication and injury potential assessment. In Proceedings of the International Research Council on Biomechanics of Injury (IRCOBI) Conference. Bern, Switzerland. 90. Bidez, M.W., Cochran, J.E., Jr., King, D., and Burke, D.S., 3rd, 2007. Occupant dynamics in rollover crashes: influence of roof deformation and seat belt performance on probable spinal column injury. Annals of Biomedical Engineering 35 (11), 1973-88. 91. Ashby, B., Lai, W., Carhart, M., Newberry, W., Weaver, B., and Corrigan, C., 2007. Compressive neck preloading during the airborne phase of vehicle rollover. SAE Technical Paper No. 2007-01-0377.  92. Moffatt, E.A. and James, M.B., 2005. Headroom, roof crush, and belted excursion in rollovers. SAE Technical Paper No. 2005-01-0942.  93. Heudorfer, B., Breuninger, M., Karlbauer, U., Kraft, M., and Maidel, J., 2005. Roofbag - A concept study to provide enhanced protection for head and neck in case of rollover. In Proceedings of the International Technical Conference on the Enhanced Safety of Vehicles. Washington, DC. 94. Winkelstein, B.A. and Myers, B.S., 1997. The biomechanics of cervical spine injury and implications for injury prevention. Medicine and Science in Sports and Exercise 29 (7 Suppl), S246-55. 187  95. Viano, D.C. and Parenteau, C.S., 2008. Analysis of head impacts causing neck compression injury. Traffic Injury Prevention 9 (2), 144-52. 96. Toomey, D.E., Yang, K.H., Yoganandan, N., Pintar, F.A., and Van Ee, C.A., 2013. Toward a more robust lower neck compressive injury tolerance - An approach combining multiple test methodologies. Traffic Injury Prevention 14 (8), 845-852. 97. Yoganandan, N., Sances, A.J., Pintar, F.A., Maiman, D.J., Reinartz, J., Cusick, J.F., and Larson, S.J., 1990. Injury biomechanics of the human cervical column. Spine 15 (10), 1031-1039. 98. White, A.A. and Panjabi, M.M., 1990. Clinical biomechanics of the spine. 2 ed, JB Lippincott, Philadephia. 99. Foust, D.R., Chaffin, D.B., Snyder, R.G., and Baum, J.K., 1973. Cervical range of motion and dynamic response and strength of cervical muscles. SAE Technical Paper #730975. 285-308. 100. Saari, A., Dennison, C.R., Zhu, Q., Nelson, T.S., Morley, P., Itshayek, E., Oxland, T.R., and Cripton, P.A., 2013. Compressive follower load influences cervical spine kinematics and kinetics during simulated head-first impact in an in vitro model. Journal of Biomechanical Engineering 135 (11), 111003-111003. 101. Gopal, M., Baron, K., and Shah, M., 2004. Simulation and testing of a suite of field relevant rollovers. SAE Technical Paper No. 2004-01-0335.  102. Kettler, A., Hartwig, E., Schultheiss, M., Claes, L., and Wilke, H.J., 2002. Mechanically simulated muscle forces strongly stabilize intact and injured upper cervical spine specimens. Journal of Biomechanics 35 (3), 339-46. 103. Saari, A., Itshayek, E., and Cripton, P.A., 2011. Cervical spinal cord deformation during simulated head-first impact injuries. Journal of Biomechanics 44 (14), 2565-2571. 104. Ivancic, P.C., 2012. Atlas injury mechanisms during head-first impact. Spine 37 (12), 1022-9. 105. Moroney, S.P., Schultz, A.B., and Miller, J.A.A., 1988. Analysis and measurement of neck loads. Journal of Orthopaedic Research 6 (5), 713-720. 106. Kim, K. and Kim, Y.H., 2008. Role of trunk muscles in generating follower load in the lumbar spine of neutral standing posture. Journal of Biomechanical Engineering 130 (4), 041005. 107. Anderst, W., Baillargeon, E., Donaldson, W., Lee, J., and Kang, J., 2013. Motion path of the instant center of rotation in the cervical spine during in vivo dynamic flexion-extension implications for artificial disc design and evaluation of motion quality after arthrodesis. Spine 38 (10), E594-E601. 188  108. Patwardhan, A.G., Havey, R.M., Ghanayem, A.J., Diener, H., Meade, K.P., Dunlap, B., and Hodges, S.D., 2000. Load-carrying capacity of the human cervical spine in compression is increased under a follower load. Spine 25 (12), 1548-54. 109. Roaf, R., 1960. A study of the mechanics of spinal injuries. The Bone and Joint Journal 42-B (4), 810-823. 110. Crisco, J.J. and Panjabi, M.M., 1990. Postural biomechanical stability and gross muscular architecture in the spine, in Multiple Muscle Systems: Biomechanics and Movement Organization, J.W. Winters, S., Editor, Springer-Verlag, New York. 111. Panjabi, M.M., Cholewicki, J., Nibu, K., Grauer, J., Babat, L.B., and Dvorak, J., 1998. Critical load of the human cervical spine: an in vitro experimental study. Clinical Biomechanics 13 (1), 11-17. 112. Van Ee, C., Nightingale, R., Camacho, D., Chancey, V., Knaub, K., Sun, E., and Myers, B., 2000. Tensile properties of the human muscular and ligamentous cervical spine. SAE Technical Paper No. 2000-01-SC07.  113. Coakwell, M.R., Bloswick, D.S., and Moser, R., 2004. High-risk head and neck movements at high G and interventions to reduce associated neck injury. Aviation Space and Environmental Medicine 75 (1), 68-80. 114. Harrison, D.E., Harrison, D.D., Janik, T.J., Holland, B., and Siskin, L.A., 2001. Slight head extension: does it change the sagittal cervical curve? European Spine Journal 10 (2), 149-53. 115. Black, K.M., McClure, P., and Polansky, M., 1996. The influence of different sitting positions on cervical and lumbar posture. Spine 21 (1), 65-70. 116. Anderst, W.J., Lee, J.Y., Donaldson, W.F., 3rd, and Kang, J.D., 2013. Six-degrees-of-freedom cervical spine range of motion during dynamic flexion-extension after single-level anterior arthrodesis: comparison with asymptomatic control subjects. The Journal of Bone & Joint Surgery. American volume 95 (6), 497-506. 117. Dvorak, J., Panjabi, M.M., Novotny, J.E., and Antinnes, J.A., 1991. In vivo flexion/extension of the normal cervical spine. Journal of Orthopaedic Research 9 (6), 828-834. 118. Penning, L., 1977. Normal movements of the cervical spine. American Journal of Roentgenology 130 317-326. 119. Lind, B., Sihlbom, H., Nordwall, A., and Malchau, H., 1989. Normal range of motion of the cervical spine. Archives of Physical Medicine and Rehabilitation 70 (9), 692-5. 189  120. Liu, F., Cheng, J., Komistek, R.D., Mahfouz, M.R., and Sharma, A., 2007. In vivo evaluation of dynamic characteristics of the normal, fused, and disc replacement cervical spines. Spine 32 (23), 2578-84. 121. Miller, J.S., Polissar, N.L., and Haas, M., 1996. A radiographic comparison of neutral cervical posture with cervical flexion and extension ranges of motion. Journal of Manipulative Physiological Therapeutics 19 (5), 296-301. 122. Ordway, N.R., Seymour, R.J., Donelson, R.G., Hojnowski, L.S., and Edwards, W.T., 1999. Cervical flexion, extension, protrusion, and retraction. A radiographic segmental analysis. Spine 24 (3), 240-7. 123. Takeshima, T., Omokawa, S., Takaoka, T., Araki, M., Ueda, Y., and Takakura, Y., 2002. Sagittal alignment of cervical flexion and extension: lateral radiographic analysis. Spine 27 (15), E348-55. 124. Reitman, C.A., Mauro, K.M., Nguyen, L., Ziegler, J.M., and Hipp, J.A., 2004. Intervertebral motion between flexion and extension in asymptomatic individuals. Spine 29 (24), 2832-43. 125. Penning, L., 1992. Acceleration injury of the cervical spine by hypertranslation of the head. Part I.  Effect of normal translation of the head on cervical spine motion:a radiological study. European Spine Journal 1 7-12. 126. Bible, J.E., Biswas, D., Miller, C.P., Whang, P.G., and Grauer, J.N., 2010. Normal functional range of motion of the cervical spine during 15 activities of daily living. Journal of Spinal Disorders & Techniques 23 (1), 15-21. 127. Palumbo, M.A., Hulstyn, M.J., Fadale, P.D., O'Brien, T., and Shall, L., 1996. The effect of protective football equipment on alignment of the injured cervical spine. Radiographic analysis in a cadaveric model. The American Journal of Sports Medicine 24 (4), 446-53. 128. Sherbondy, P.S., Hertel, J.N., and Sebastianelli, W.J., 2006. The effect of protective equipment on cervical spine alignment in collegiate lacrosse players. The American Journal of Sports Medicine 34 (10), 1675-9. 129. Ivancic, P.C., 2012. Head-first impact with head protrusion causes noncontiguous injuries of the cadaveric cervical spine. Clinical Journal of Sport Medicine 22 (5), 390-396. 130. Panjabi, M.M., Miura, T., Cripton, P.A., Wang, J.L., Nain, A.S., and DuBois, C., 2001. Development of a system for in vitro neck muscle force replication in whole cervical spine experiments. Spine 26 (20), 2214-9. 131. Wu, S.K., Kuo, L.C., Lan, H.C., Tsai, S.W., Chen, C.L., and Su, F.C., 2007. The quantitative measurements of the intervertebral angulation and translation during cervical flexion and extension. European Spine Journal 16 (9), 1435-44. 190  132. Descarreaux, M., Blouin, J.S., and Teasdale, N., 2003. A non-invasive technique for measurement of cervical vertebral angle: report of a preliminary study. European Spine Journal 12 (3), 314-9. 133. Pearcy, M.J., 1985. Stereo radiography of lumbar spine motion. Acta Orthopaedica Scandinavica Supplementum 212 1-45. 134. Hanley, E.N., Matteri, R.E., and Frymoyer, J.W., 1976. Accurate roentgenographic determination of lumbar flexion-extension. Clinical Orthopaedics and Related Research (115), 145-8. 135. Ono, K., Kaneoka, K., Wittek, A., and Kajzer, J., 1997. Cervical injury mechanism based on the analysis of human cervical vertebral motion and head-neck-torso kinematics during low speed rear impacts. SAE Technical Paper No. 973340.  136. Johnson, G.M., 1998. The correlation between surface measurement of head and neck posture and the anatomic position of the upper cervical vertebrae. Spine 23 (8), 921-7. 137. Wu, S.K., Lan, H.H., Kuo, L.C., Tsai, S.W., Chen, C.L., and Su, F.C., 2007. The feasibility of a video-based motion analysis system in measuring the segmental movements between upper and lower cervical spine. Gait Posture 26 (1), 161-6. 138. Fernandez, J.W. and Pandy, M.G., 2006. Integrating modelling and experiments to assess dynamic musculoskeletal function in humans. Experimental Physiology 91 (2), 371-82. 139. Richmond, F.J., Thomson, D.B., and Loeb, G.E., 1992. Electromyographic studies of neck muscles in the intact cat. I. Patterns of recruitment underlying posture and movement during natural behaviors. Experimental Brain Research 88 (1), 41-58. 140. Keshner, E.A., Statler, K.D., and Delp, S.L., 1997. Kinematics of the freely moving head and neck in the alert cat. Experimental Brain Research 115 (2), 257-66. 141. Corneil, B.D., Olivier, E., Richmond, F.J., Loeb, G.E., and Munoz, D.P., 2001. Neck muscles in the rhesus monkey. II. Electromyographic patterns of activation underlying postures and movements. Journal of Neurophysiology 86 (4), 1729-49. 142. Ishii, T., Mukai, Y., Hosono, N., Sakaura, H., Fujii, R., Nakajima, Y., Tamura, S., Sugamoto, K., and Yoshikawa, H., 2004. Kinematics of the subaxial cervical spine in rotation in vivo three-dimensional analysis. Spine 29 (24), 2826-31. 143. Anderst, W.J., Vaidya, R., and Tashman, S., 2007. A technique to measure three-dimensional in vivo rotation of fused and adjacent lumbar vertebrae. Spine Journal  144. Leivseth, G., Kolstad, F., Nygaard, O.P., Zoega, B., Frobin, W., and Brinckmann, P., 2006. Comparing precision of distortion-compensated and stereophotogrammetric Roentgen analysis when monitoring fusion in the cervical spine. European Spine Journal 15 (6), 774-9. 191  145. ICRP, 2007. The 2007 recommendations of the international commission on radiological protection. ICRP publication 103. Annals of the ICRP 37 (2-4), 1-332. 146. Moro-oka, T., Hamai, S., Miura, H., Shimoto, T., Higaki, H., Fregly, B., Iwamoto, Y., and Banks, S., 2007. Can magnetic resonance imaging–derived bone models be used for accurate motion measurement with single-plane three-dimensional shape registration? Journal of Orthopaedic Research 25 867-872. 147. Thistlethwaite, P., Methodology for practical image-based, dynamic, 3D kinematic analysis of the lumbar spine. PhD thesis, University of Bern, Switzerland,  148. Anderst, W.J., Baillargeon, E., Donaldson, W.F., Lee, J.Y., and Kang, J.D., 2011. Validation of a noninvasive technique to precisely measure in vivo three-dimensional cervical spine movement. Spine 36 (6), E393-E400. 149. Biswas, D., Bible, J.E., Bohan, M., Simpson, A.K., Whang, P.G., and Grauer, J.N., 2009. Radiation exposure from musculoskeletal computerized tomographic scans. The Journal of Bone & Joint Surgery. American volume 91 (8), 1882-9. 150. Hill, A.M., McGregor, A.H., Wragg, P., Brinckmann, P., and Burton, A.K., 2005. The assessment of cervical spine kinematics: A comparison of iMR and conventional radiographic techniques. Journal of Back and Musculoskeletal Rehabilitation (18), 29-35. 151. Frobin, W., Leivseth, G., Biggemann, M., and Brinckmann, P., 2002. Sagittal plane segmental motion of the cervical spine. A new precision measurement protocol and normal motion data of healthy adults. Clinical Biomechanics 17 (1), 21-31. 152. Lam, S.C.B., McCane, B., and Allen, R., 2009. Automated tracking in digitized videofluoroscopy sequences for spine kinematic analysis. Image and Vision Computing 27 (10), 1555-1571. 153. Reinartz, R., Platel, B., Boselie, T., van Mameren, H., van Santbrink, H., and Romeny, B.T., 2009. Cervical vertebrae tracking in video-fluoroscopy using the normalized gradient field. Medical Image Computing and Computer-Assisted Intervention - Miccai 2009, Pt I, Proceedings 5761 524-531. 154. Wong, A., Dunk, N.M., and Callaghan, J.P., 2009. A systematic approach to feature tracking of lumbar spine vertebrae from fluoroscopic images using complex-valued wavelets. Computer Methods in Biomechanics and Biomedical Engineering 12 (5), 607-616. 155. Wong, K.W., Luk, K.D., Leong, J.C., Wong, S.F., and Wong, K.K., 2006. Continuous dynamic spinal motion analysis. Spine 31 (4), 414-9. 156. Zheng, Y.L., Nixon, M.S., and Allen, R., 2004. Automated segmentation of lumbar vertebrae in digital videofluoroscopic images. IEEE Transactions on Medical Imaging 23 (1), 45-52. 192  157. Bifulco, P., Cesarelli, M., Romano, M., Allen, R., and Cerciello, T., 2009. Vertebrae tracking through fluoroscopic sequence: a novel approach. International Federation for Medical and Biological Engineering (IFMBE) Proceedings 25 619-622. 158. Vrtovec, T., Pernus, F., and Likar, B., 2009. A review of methods for quantitative evaluation of spinal curvature. European Spine Journal 18 (5), 593-607. 159. Klinich, K., Ebert, S., Van Ee, C., Flannagan, C., Prasad, M., Reed, M., and Schneider, L., 2004. Cervical spine geometry in the automotive seated posture: Variations with age, stature, and gender. SAE Technical Paper No. 2004-22-0014.  160. Nightingale, R.W., Winkelstein, B.A., Van Ee, C.A., and Myers, B.S., 1998. Injury mechanisms in the pediatric cervical spine during out-of-position airbag deployments. Association for the Advancement of Automotive Medicine 153-164. 161. Mayoux-Benhamou, M.A., Revel, M., and Vallee, C., 1997. Selective electromyography of dorsal neck muscles in humans. Experimental Brain Research 113 (2), 353-60. 162. Bogduk, N. and Mercer, S., 2000. Biomechanics of the cervical spine. I: Normal kinematics. Clinical Biomechanics 15 (9), 633-48. 163. Kamibayashi, L.K. and Richmond, F.J., 1998. Morphometry of human neck muscles. Spine 23 (12), 1314-23. 164. Anderst, W.J., Donaldson, W.F., Lee, J.Y., and Kang, J.D., 2013. Subject-specific inverse dynamics of the head and cervical spine during in vivo dynamic flexion-extension. Journal of Biomechanical Engineering 135 (6),  165. Barbenel, J.C., 1983. The application of optimization methods for the calculation of joint and muscle forces. Engineering in Medicine 12 (1), 29-33. 166. Toosizadeh, N. and Haghpanahi, M., 2011. Generating a finite element model of the cervical spine: Estimating muscle forces and internal loads. Scientia Iranica B 18 (6), 1237-1245. 167. Halldin, P., Hedenstierna, S., Brolin, K., and von Holst, H., 2006. Finite element analysis of the effects of head-supported mass on neck responses. KTH, School of Technology and Health (STH), Neuronic Engineering: Stockholm, Sweden. 168. Thelen, D.G., Schultz, A.B., Fassois, S.D., and Ashton-Miller, J.A., 1994. Identification of dynamic myoelectric signal-to-force models during isometric lumbar muscle contractions. Journal of Biomechanics 27 (7), 907-19. 169. Cholewicki, J., McGill, S.M., and Norman, R.W., 1995. Comparison of muscle forces and joint load from an optimization and EMG assisted lumbar spine model: towards development of a hybrid approach. Journal of Biomechanics 28 (3), 321-31. 193  170. Netto, K.J., Burnett, A.F., Green, J.P., and Rodrigues, J.P., 2008. Validation of an EMG-driven, graphically based isometric musculoskeletal model of the cervical spine. Journal of Biomechanical Engineering 130 (3), 031014. 171. Siegmund, G.P., Blouin, J.S., Brault, J.R., Hedenstierna, S., and Inglis, J.T., 2007. Electromyography of superficial and deep neck muscles during isometric, voluntary, and reflex contractions. Journal Biomechanical Engineering 129 (1), 66-77. 172. Vasavada, A.N., Li, S., and Delp, S.L., 1998. Influence of muscle morphometry and moment arms on the moment-generating capacity of human neck muscles. Spine 23 (4), 412-22. 173. Conley, M.S., Meyer, R.A., Bloomberg, J.J., Feeback, D.L., and Dudley, G.A., 1995. Noninvasive analysis of human neck muscle function. Spine 20 (23), 2505-12. 174. Blouin, J.S., Siegmund, G.P., Carpenter, M.G., and Inglis, J.T., 2007. Neural control of superficial and deep neck muscles in humans. Journal of Neurophysiology 98 (2), 920-8. 175. Oi, N., Pandy, M.G., Myers, B.S., Nightingale, R.W., and Chancey, V.C., 2004. Variation of neck muscle strength along the human cervical spine. Stapp Car Crash Journal 48 397-417. 176. Keshner, E.A., Campbell, D., Katz, R.T., and Peterson, B.W., 1989. Neck muscle activation patterns in humans during isometric head stabilization. Experimental Brain Research 75 (2), 335-44. 177. Vasavada, A.N., Peterson, B.W., and Delp, S.L., 2002. Three-dimensional spatial tuning of neck muscle activation in humans. Experimental Brain Research 147 (4), 437-48. 178. Gabriel, D.A., Matsumoto, J.Y., Davis, D.H., Currier, B.L., and An, K.N., 2004. Multidirectional neck strength and electromyographic activity for normal controls. Clinical Biomechanics 19 (7), 653-8. 179. Takebe, K., Vitti, M., and Basmajian, J.V., 1974. The functions of semispinalis capitis and splenius capitis muscles: an electromyographic study. The Anatomical Record 179 (4), 477-80. 180. Newell, R.S., Blouin, J.-S., Street, J., Cripton, P.A., and Siegmund, G.P., 2013. Neck posture and muscle activity are different when upside down: A human volunteer study. Journal of Biomechanics 46 (16), 2837-2843. 181. De Beer, D.P., 2002. Occupational radiation protection in workplaces involving exposure to natural radiation. In Proceedings of the International Conference on Occupational Radiation Protection: Protecting Workers Against Exposure to Ionizing Radiation. Geneva, Switzerland. 194  182. Brainerd, E.L., Baier, D.B., Gatesy, S.M., Hedrick, T.L., Metzger, K.A., Gilbert, S.L., and Crisco, J.J., 2010. X-Ray reconstruction of moving morphology (XROMM): Precision, accuracy and applications in comparative biomechanics research. Journal of Experimental Zoology Part a-Ecological Genetics and Physiology 313A (5), 262-279. 183. Gandevia, S.C., 2001. Spinal and supraspinal factors in human muscle fatigue. Physiological Reviews 81 (4), 1725-1789. 184. Yoganandan, N., Pintar, F.A., Zhang, J.Y., and Baisden, J.L., 2009. Physical properties of the human head: Mass, center of gravity and moment of inertia. Journal of Biomechanics 42 (9), 1177-1192. 185. Voo, L. and Liu, Y.K., 1994. Buckling analysis - A modal method for estimating cervical spine injury potential, in Head and Neck Injuries in Sports, ASTM STP 1229, E.F. Homer, Editor, American Society for Testing and Materials, Philadelphia. 186. Panjabi, M., Dvorak, J., Duranceau, J., Yamamoto, I., Gerber, M., Rauschning, W., and Bueff, H.U., 1988. 3-dimensional movements of the upper cervical spine. Spine 13 (7), 726-730. 187. Panjabi, M.M., Summers, D.J., Pelker, R.R., Videman, T., Friedlaender, G.E., and Southwick, W.O., 1986. Three-dimensional load-displacement curves due to forces on the cervical spine. Journal of Orthopaedic Research 4 (2), 152-61. 188. Beier, G.S., M.; Schuller, E.; Spann, W., 1979. Determination of physical data of the head: Center of gravity and moments of inertia of human heads. Institute of Forensic Medicine, University of Munich: Munich. 189. Panjabi, M.M., Nibu, K., and Cholewicki, J., 1998. Whiplash injuries and the potential for mechanical instability. European Spine Journal 7 (6), 484-92. 190. Ackland, D.C., Merritt, J.S., and Pandy, M.G., 2011. Moment arms of the human neck muscles in flexion, bending and rotation. Journal of Biomechanics 44 (3), 475-86. 191. Drake, R.L., Vogl, W., and Mitchell, A.W.M., 2005. Gray's anatomy for students, Elsevier Inc., Philadelphia. 192. Moroney, S.P., Schultz, A.B., Miller, J.A., and Andersson, G.B., 1988. Load-displacement properties of lower cervical spine motion segments. Journal of Biomechanics 21 (9), 769-79. 193. Shea, M., Edwards, W.T., White, A.A., and Hayes, W.C., 1991. Variations of stiffness and strength along the human cervical spine. Journal of Biomechanics 24 (2), 95-107. 194. Bell, K., 2013. Development of a physiologic in-vitro testing methodology for assessment of cervical spine kinematics. PhD thesis, University of Pittsburgh, Pittsburgh, PA. 195  195. McGill, S.M., Jones, K., Bennett, G., and Bishop, P.J., 1994. Passive stiffness of the human neck in flexion, extension, and lateral bending. Clinical Biomechanics 9 (3), 193-8. 196. Cripton, P.A., Bruehlmann, S.B., Orr, T.E., Oxland, T.R., and Nolte, L., 2000. In vitro axial preload application during spine flexibility testing: towards reduced apparatus-related artefacts. Journal of Biomechanics 33 (12), 1559-1568. 197. Schuldt, K. and Harms-Ringdahl, K., 1988. EMG/moment relationships in neck muscles during isometric cervical spine extension. Clinical Biomechanics 3 (2), 58-65. 198. Anderst, W.J., Donaldson, W.F., Lee, J.Y., and Kang, J.D., 2013. Cervical motion segment percent contributions to flexion-extension during continuous functional novement in control subjects and arthrodesis patients. Spine 38 (9), E533-E539. 199. Anderst, W.J., Donaldson, W.F., Lee, J.Y., and Kang, J.D., 2013. Cervical spine intervertebral kinematics with respect to the head are different during flexion and extension motions. Journal of Biomechanics 46 (8), 1471-1475. 200. Wilke, H.J., Claes, L., Schmitt, H., and Wolf, S., 1994. A universal spine tester for in vitro experiments with muscle force simulation. European Spine Journal 3 (2), 91-7. 201. Panjabi, M.M., 1992. The stabilizing system of the spine. Part II. Neutral zone and instability hypothesis. Journal of Spinal Disorders 5 (4), 390-6. 202. Marin, F., Hoang, N., Aufaure, P., and Ho Ba Tho, M.C., 2010. In vivo intersegmental motion of the cervical spine using an inverse kinematics procedure. Clinical Biomechanics 25 (5), 389-96. 203. Vasavada, A.N., Li, S., and Delp, S.L., 2001. Three-dimensional isometric strength of neck muscles in humans. Spine 26 (17), 1904-9. 204. Vasavada, A.N., Danaraj, J., and Siegmund, G.P., 2008. Head and neck anthropometry, vertebral geometry and neck strength in height-matched men and women. Journal of Biomechanics 41 (1), 114-21. 205. Friedman, D., Paver, J., and McGuan, R., 2012. Design, development and validation of rollover dummy injury measures. In Proceedings of the International Crashworthiness Conference. Milan, Italy. 206. Paver, J. and Friedman, D., 2012. Is BFD a hyperflexion injury or compression with localized bending injury or both? In Proceedings of the International Crashworthiness Conference Paper No. 2012-111. Milano, Italy. 207. Paver, J.G., Friedman, D., Mattos, G., and Caplinger, J., 2010. The development of IARV’s for the Hybrid III neck modified for dynamic rollover crash testing. In Proceedings of the International Crashworthiness Conference. Washington, DC. 196  208. Dvorak, J., Antinnes, J.A., Panjabi, M.M., Loustalot, D., and Bonomo, M., 1992. Age and gender related normal motion of the cervical spine. Spine 17 (10 Suppl), S393-S398. 209. Dvorak, J., Panjabi, M.M., Grob, D., Novotny, J.E., and Antinnes, J.A., 1993. Clinical validation of functional flexion/extension radiographs of the cervical spine. Spine 18 (1), 120-127. 210. Cot, P., Cassidy, J.D., YongHing, K., Sibley, J., and Loewy, J., 1997. Apophysial joint degeneration, disc degeneration, and sagittal curve of the cervical spine - Can they be measured reliably on radiographs? Spine 22 (8), 859-864. 211. Grob, D., Frauenfelder, H., and Mannion, A.F., 2007. The association between cervical spine curvature and neck pain. European Spine Journal 16 (5), 669-78. 212. Sommerich, C.M., Joines, S.M., Hermans, V., and Moon, S.D., 2000. Use of surface electromyography to estimate neck muscle activity. Journal of Electromyography & Kinesiology 10 (6), 377-98. 213. Burden, A., 2010. How should we normalize electromyograms obtained from healthy participants? What we have learned from over 25 years of research. Journal of Electromyography & Kinesiology 20 (6), 1023-35. doi 10.1016/j.jelekin.2010.07.004. Epub 2010 Aug 10. 214. Jensen, C., Vasseljen Jr, O., and Westgaard, R.H., 1996. Estimating maximal EMG amplitude for the trapezius muscle: On the optimization of experimental procedure and electrode placement for improved reliability and increased signal amplitude. Journal of Electromyography and Kinesiology 6 (1), 51-58. 215. Schuldt, K. and Harms-Ringdahl, K., 1988. Cervical spine position versus e.m.g. activity in neck muscles during maximum isometric neck extension. Clinical Biomechanics 3 (3), 129-36. 216. Queisser, F.B., R.; Seidel, H., 1994. Control of positioning the cervical spine and its application to measuring extensor strength. Clinical Biomechanics 9 (157-161),  217. Burnett, A., Green, J., Netto, K., and Rodrigues, J., 2007. Examination of EMG normalisation methods for the study of the posterior and posterolateral neck muscles in healthy controls. Journal of Electromyography and Kinesiology 17 (5), 635-641. 218. Netto, K.J. and Burnett, A.F., 2006. Reliability of normalisation methods for EMG analysis of neck muscles. Work 26 (2), 123-30. 219. Green, N.D. and Brown, L., 2004. Head positioning and neck muscle activation during air combat. Aviation Space and Environmental Medicine 75 (8), 676-80. 197  220. Siegmund, G.P., Blouin, J.S., Carpenter, M.G., Brault, J.R., and Inglis, J.T., 2008. Are cervical multifidus muscles active during whiplash and startle? An initial experimental study. BMC Musculoskelet Disord 9 80. 221. Torg, J.S., Vegso, J.J., Oneill, M.J., and Sennett, B., 1990. The epidemiologic, pathological, biomechanical, and cinematographic analysis of football-induced cervical spine trauma. The American Journal of Sports Medicine 18 (1), 50-57. 222. Parenteau, C., Gopal, M., and Viano, D., 2001. Near and far-side adult front passenger kinematics in a vehicle rollover. SAE Technical Paper No. 2001-01-0176.  223. Harvey, S.B. and Hukins, D.W., 1998. Measurement of lumbar spinal flexion-extension kinematics from lateral radiographs: simulation of the effects of out-of-plane movement and errors in reference point placement. Medical Engineering & Physics 20 (6), 403-9. 224. Cerciello, T., Bifulco, P., Cesarelli, M., Romano, M., and Allen, R., 2009. Automatic vertebra tracking through dynamic fluoroscopic sequence by smooth derivative template matching. In Proceedings of the Information Technology and Applications in Biomedicine. Larnaca, Cyprus. 225. Cerciello, T.B., P.; Cesarelli, M.; Paura, L.; Romano, M.; Pasquariello, G.; Allen, R., 2010. Noise reduction in fluoroscopic image sequences for joint kinematics analysis. IFMBE Proceedings 29 323-326. 226. Cerciello, T., Romano, M., Bifulco, P., Cesarelli, M., and Allen, R., 2011. Advanced template matching method for estimation of intervertebral kinematics of lumbar spine. Medical Engineering & Physics 33 (10), 1293-1302. 227. Crawford, N.R. and Dickman, C.A., 1997. Construction of local vertebral coordinate systems using a digitizing probe. Technical note. Spine 22 (5), 559-63. 228. Trott, P.H., Pearcy, M.J., Ruston, S.A., Fulton, I., and Brien, C., 1996. Three-dimensional analysis of active cervical motion: the effect of age and gender. Clinical Biomechanics 11 (4), 201-206. 229. Malmstrom, E.M., Karlberg, M., Fransson, P.A., Melander, A., and Magnusson, M., 2006. Primary and coupled cervical movements: the effect of age, gender, and body mass index. A 3-dimensional movement analysis of a population without symptoms of neck disorders. Spine 31 (2), E44-50. 230. Leivseth, G., Brinckmann, P., Frobin, W., Johnsson, R., and Stromqvist, B., 1998. Assessment of sagittal plane segmental motion in the lumbar spine. A comparison between distortion-compensated and stereophotogrammetric roentgen analysis. Spine 23 (23), 2648-55. 198  231. Bey, M.J., Kline, S.K., Tashman, S., and Zauel, R., 2008. Accuracy of biplane x-ray imaging combined with model-based tracking for measuring in-vivo patellofemoral joint motion. Journal of Orthopaedic Surgery and Research 3 38. 232. Kwon, Y.-H. DLT Method.  1998  [Accessed April 2012]; Available from: http://kwon3d.com/theory/dlt/dlt.html#refs.  199  Appendix A: Automatic tracking algorithm  X-ray tracking technique The vertebrae were tracked in the fluoroscopy images with a custom-programmed algorithm in Matlab (R2010b, MathWorks Inc., Natick, MA) that uses a normalized cross-correlation technique [157, 224-226].  Each vertebra is cropped from a template image (image-T), for instance the upright resting posture (see Figure A-1).  This template is used to find the location and angle of the vertebra in an image of interest (image-I).  The noise inherent to fluoroscopy imaging is reduced by applying a 2D edge detector (first order derivative of 2D Gaussian function) prior to template matching.  This effectively blurs the images and highlights the edges (i.e. area with the largest changes in intensity) (Figure A-1).    Figure A-1: Template image for normalized cross correlation The template image (image-T) of each vertebra (i.e. C4) is selected from a clear image (left).  An edge detector is used to reduce X-ray noise and highlight prominent vertebral features (right).   200   A normalized cross-correlation is used to search the image of interest (image-I) for the best match with the template image (image-T).  The location of the best match determines the X-Y location of the vertebra (see Figure A-2).    Figure A-2: Example results of the normalized cross correlation An example of the correlation output from the normalized cross correlation.  The X-Y plane represents the image plane, and the Z-axis represents the strength of the correlation at that location (red represents the strongest correlation, blue the weakest correlation).  The red peak in the middle was the best match for this particular search.    201  The template image is rotated through a series of angles (ranging anywhere from 50° counter-clock-wise to 50° clock-wise, in increments of 1°.  The optimum correlation is found throughout the 1° increments.  The search narrows around the optimum angle found in the first coarse search and then a finer search is performed.  The final overall optimum angle is used to determine the angular orientation of the vertebra in the image.    Accuracy with Out-of-Plane Motion The purpose of this pilot study was to evaluate the image tracking algorithm described above.  The subject protocol for the entire thesis was a combination of static and dynamic tasks; therefore, this pilot study evaluated the vertebral angle and displacement accuracy, both statically and dynamically.  The first objective was to determine the accuracy and precision of the tracking algorithm with images of static vertebrae and how the error was affected when the cervical vertebral column was placed in out-of-plane positions (lateral bending and axial rotation).  The second objective was to determine the accuracy in a dynamic task, also with out-of-plane motions.     To evaluate the tracking algorithm, dried human vertebrae were imaged in several out-of-plane static postures and several dynamic trials (see Figure A-3).  The automatic tracking program was used to analyze the sagittal plane angle (i.e. flexion/extension) between C3 and C4 as the vertebrae were placed in different positions.  To obtain an independent measure of the position and intervertebral motions of C3 and C4, optoelectronic markers (Optotrak Certus, NDI, 3D marker accuracy = 0.1 mm, resolution = 0.01 mm) were rigidly attached to these two vertebral bodies and recorded throughout the static and dynamic trials.  To quantify the static positions and 202  intervertebral motions in all three anatomical planes [227], the pose of each vertebral body was described in a coordinate system defined relative to the fluoroscope image intensifier surface (using Euler angles).  The X-ray and optoelectronic marker methods were compared.  The error in the flexion angle was defined as the discrepancy between the two techniques (Optotrak results minus X-ray results, i.e. bias).  The accuracy was defined as the mean difference (error) between the two techniques and the precision was defined as the standard deviation of the differences.  The error was also quantified as the root mean squared error (RMSE).  RMSE is similar to the standard deviation, but is not distributed about the mean (therefore it inherently includes bias offset and will be larger than standard deviation).     Figure A-3: Accuracy study of human vertebrae with optotrak markers The dissected human vertebrae (C1 through C7) were placed in the image plane and Optotrak markers were attached to C3 and C4 (left). X-rays (right) were taken of the vertebrae as they were moved around the image plane.   203    Studies on human subject neck range of motion have reported that when performing cervical flexion and extension motions, the coupled motions of lateral and axial bending are less than 5° on average (for ages of 20 through to 49) [228, 229].  Therefore, to assess the effect of out-of-plane motion, small (less than 5°) and large (between 5° and 10°) out-of-plane motions were created and the accuracy and precision were assessed.  The dynamic trial was conducted for out-of-plane motions within 10°.     Based on the findings of this pilot study, it was concluded that the automatic tracking technique is appropriate for a static in-plane experiment (see Figure A-4 and Table A-1).  However, this technique may not have acceptable error (up to 3°) in experiments that include out-of-plane motions (see Figure A-5).  The flexion error in the dynamic trials was found to be up to 5° when there were large out-of-plane, coupled motions, particularly when axial rotation approached 10° (see Figure A-6).   204   Figure A-4: Vertebral angle error compared to Optotrak for small out-of-plane motions Flexion angle error (Optotrak minus X-ray angles) of C3 and C4 for small amounts of out-of-plane motion (less than 5 degrees).  Top: Mean (solid line) and standard deviation (dashed line) for each observation (dots).  Bottom: the out-of-plane motion (lateral bending=black, axial rotation=gray) as captured by the Optotrak.      205   Figure A-5: Vertebral angle error compared to Optotrak for large out-of-plane motions Flexion angle error (Optotrak minus X-ray angles) of C3 and C4 for large amounts of out-of-plane motion (between 5 and 10 degrees).  Top: Mean (solid line) and standard deviation (dashed line) for each observation (dots).  Bottom: the out-of-plane motion (lateral bending=black, axial rotation=gray) as captured by the Optotrak.        206   Figure A-6: The C3 (top) and C4 (middle) flexion angle during the dynamic motion The C3 (top) and C4 (middle) flexion angle during the dynamic motion as tracked by the Optotrak and X-ray techniques (+ive flexion, -ive extension).  The out-of-plane motion in the lateral bending and axial rotations (bottom) were captured with the Optotrak.             207   Table A-1: Results of the static motion out-of-plane flexion angle error at C3-C4.   Trials consisted of lateral bending and axial rotation and combinations of the two.  The lateral bending and axial rotation of the proximal and distal vertebrae are summarized.    Optotrak Static Positions X-ray error Static Motion C3/C4 lateral bending angle (+R, -L) (degrees) C3/C4 axial rotation angle (+L, -R) (degrees) Mean out-of-plane flexion error in degrees (C3/C4) Lateral bend to right 4.74 / 4.38 1.40 / -0.84 -0.20 / -0.95 Lateral bend to left -8.17 / -7.00 -0.73 / 2.72 0.58 / 1.83 Axial rotation to right -0.43 / -3.26 -6.95 / -6.13 -1.38 / -0.67 Axial rotation to left -2.31 / 0.24 6.63 / 7.06 0.67 / -0.72 Right lateral bending + Right axial rotation 8.66 / 3.67 -7.04 / -10.09 -0.23 / -0.94 Left lateral bending + Right axial rotation -6.4 / -10.03 -10.03 / -5.93 -3.18 / -2.15 Flexion + Right lateral bend 7.56 / 5.62 -0.53 / -4.15 -1.10 / -0.76 Flexion + Left lateral bend -5.97 / -6.71 -3.68 / -0.48 -0.14 / 0.50 Close to neutral 3.47 / 2.28 -0.63 / -2.28 0.20 / -0.17 Close to neutral -1.41 / -1.92 -1.92 / -1.35 -0.65 / 0.12 Close to neutral 1.04 / 2.05 0.06 / -1.84 0.55 / 0.61 Close to neutral 0.99 / -1.05 -1.87 / -1.59 -0.11 / 0.71   Accuracy around the image plane The purpose of this pilot study was to evaluate the image tracking algorithm used to evaluate both the absolute and intervertebral angle and displacements in different locations of the image intensifier.  A template of each vertebral body is used to find a match with the image of interest using the automatic tracking algorithm.  The projection of a vertebral body can change as the body is moved around in the image plane (since an X-ray is projected as a cone).  Thus, the objective was to determine the accuracy and precision of the tracking algorithm in static vertebral positions and how the error was affected when the cervical vertebral column was place in various locations in the image plane.       208  Two human vertebrae were isolated from a cadaveric spine; all tissues were removed leaving the bone exposed.  The C3 and C4 vertebrae were mounted to a wooden wedge, which was in turn mounted to a precision translational device (see Figure A-7).  The translational device was mounted parallel to, and at a distance of 20 cm from, the image intensifier.  A wooden peg board was used for attaching the translational device, allowing the vertebrae to be moved in different locations of the C-arm image.  Levels were used to ensure the peg board was parallel to the image intensifier.  To simulate the effects of soft tissue in the image, bovine meat was placed between the vertebrae and the image intensifier.  The vertebrae were translated along the translational device in 5 mm increments and fluoroscopy images were captured at every 5 mm increment.  The translational device was moved from the middle to the top and then the bottom of the peg board so that the vertebra could be imaged at the top, middle and bottom of the image intensifier. The 5 mm increments were repeated for all three locations (top, middle, and bottom).  The vertebrae were also placed in various flexion angles, ranging from 0 to 90° (see Figure A-8). There were also several tests conducted to evaluate changes in the contrast and brightness (i.e. changing the kV and mA settings of the generated X-ray). The automatic tracking program was used to analyze the flexion angle and displacement of C4 in different static positions.  To obtain an independent measure of the position and motions of C3 and C4, a digitizing arm (FaroArm, B08-02, Lake Mary, FL, accuracy = ±0.3 mm) was used to digitize the position of each vertebral body.  The X-ray and FaroArm methods were compared; the error was defined as the discrepancy between the two measurement techniques (FaroArm results minus X-ray results) (Figure A-9).  The movement of the individual vertebrae was captured as was the relative motions of the C3/C4 intervertebral joint.  These vertebrae were fused together, thus the 209  intervertebral angle should ideally remain constant.  The accuracy was defined as the mean difference (error) between the two techniques and the precision was defined as the standard deviation of the differences (errors).  The error was also quantified as the root mean squared error (RMSE).  The accuracy results are summarized in Figure A-9 and Figure A-10.       Figure A-7:  Apparatus used to conduct accuracy study  The dissected human vertebrae (C3 and C4) were rigidly attached to each other (i.e. the intervertebral angle and displacement were constant).  The vertebrae were moved around the image plane of the image intensifier through various angles. Meat was used to simulate the effects of soft tissue in the image.            210   Figure A-8:  Example X-ray images from the accuracy study where the vertebrae were moved around the X-ray plane and placed in various flexion angles.      211   Figure A-9: Angle and displacement of C4 in various locations Angle and displacement error (FaroArm results minus X-ray results) of C4 for images in different locations of the image intensifier plane.  Top: Mean (solid line) and standard deviation (dashed line) for each angle observation (dots).  Bottom: Mean (solid line) and standard deviation (dashed line) for each displacement observation (dots).      212   Figure A-10: Intervertebral angle and displacement of C3/C4 in various positions with varying image quality Mean intervertebral angle (top) and intervertebral displacement (bottom) error (solid line) of C3/C4 images and standard deviation (dashed line) for each angle observation (dots).  The vertebrae were moved in different flexion positions, different locations of the image intensifier plane, and the contrast and brightness were altered.         213  Frankfort plane accuracy The skull of a skeleton was used to assess the accuracy of the landmark identification algorithm.  The skeleton skull was place in front of the C-arm (22 cm from the surface of the image intensifier).  The headband and bead array were placed on the skeleton in the same fashion as the human subjects for this study (see Figure A-11).  Landmarks were marked on the skeleton head and the FaroArm was used to digitize these landmarks (similar to the human subject experimental protocol).  The skeleton was then imaged with the C-arm (see Figure A-11).  The skeleton was then shifted in the extension direction, consistent with someone extending their head.  The skeleton was then re-digitized with the FaroArm, and another C-arm image was taken in this configuration.  This was repeated in the flexion direction and one more time in the neutral configuration.  The landmark algorithm was used to calculate the location of the Frankfort plane in the image plane.  Using the re-digitization with the FaroArm, the 3D motions of the head are known, thus the movement of the calculated Frankfort plane can be verified.     214   Figure A-11: Frankfort plane validation Right: Plastic skull with headband and beads attached was used to verify the Frankfort plane angle calculation.  Left: C-arm image of the plastic skull with the beads visible for tracking.  The RMSE of the Frankfort plane calculation was found to be 0.74°.  This accuracy measure was found even with out-of-plane motions ranging from 0.21° to 3.11° (see Table A-2).  Considering the distance between the beads in the images, this error is approximately equal to 2 pixels of error.  This error would also be the equivalent of approximately 1.5 mm error in identifying the height of the mid-orbital point.    Table A-2:  Frankfort plan accuracy results Frankfort plane angle (for neutral, extended, and flexed positions) from the X-ray and FaroArm techniques, with the error and out-of-plane motion summarized. Motion of head Frankfort plane from X-ray (deg) Frankfort plane from FaroArm (deg) Difference (deg) Out of plane angles (deg) Neutral 9.38 10.19 0.81 -0.21/-0.44 Extension 24.52 25.51 0.99 -3.11/-0.92 Flexion -4.68 -4.58 -0.10 0.31/2.29 215  Summary of all accuracy studies A summary of all accuracy, precision, and RMSE results are provided in Table A-3.   Table A-3: Summary of all accuracy studies The overall accuracy and precision of vertebral flexion angle, vertebral displacement, and Frankfort plane angle.   The number of accuracy tests completed (including flexion/extension positions, extreme locations in the image intensifier, up to 10° out-of-plane motions and dynamic motions) and whether the images were corrected for distortion is indicated.    Accuracy Precision # Measures RMSE Distortion Corrected* Vertebral Angle Moving around II (°)a -0.02 0.61 28 0.60 Yes Vertebral Displacement (mm)b 0.07 0.16 28 0.18 Yes Vertebral Angle (Small out-of-plane motion)  (°)b -0.03 0.58 10 0.51 No Vertebral Angle (Large out-of-plane motion) (°)b -0.55 1.26 14 1.33 No Dynamic Angle (Small out-of-plane motion) (°)b -0.02 0.55 142 0.55 No Dynamic Angle (Large out-of-plane motion) (°)b -0.74 0.81 33 1.09 No Intervertebral Angle (°)a -0.05 0.57 45 0.56 Yes Intervertebral Displacement (mm)a 0.08 0.19 13 0.20 Yes Frankfort plane (°)b 0.57 0.58 3 0.74 Yes a No change: the measurement of interest or never changed; the error was defined as the offset from the initial posture b Technique comparison: The measurement of interest was compared to a another technique (Optotrak or FaroArm) *Distortion grid recordings were not taken for some of the tests; however these tests were conducted near the center of the image intensifier where distortion is a minimum.     Comparing accuracy with other techniques Bifulco et al. [157] used a similar normalized cross-correlation algorithm to track lumbar spines in X-rays; they reported root mean squared errors of vertebral location to be 0.2° and 0.3 mm.  A vertebral digitization technique, called Distortion Compensated Roentgen Analysis (DCRA), described by Frobin et al. [151], has been used to describe cervical spine motion and alignment.  An error study on this technique reported errors (standard deviation), from inter- and intra-observer tests, of less than 2° for rotations and 5% for translations motion (approximately 0.7 mm for a 15 mm vertebra) [151].  Leivseth et al. [144] evaluated the accuracy of DCRA with radiostereometric analysis (RSA) in the cervical spine and found standard deviations of 2.4° and 216  0.78 mm for angles and displacements, respectively.  They also reported inter-observer standard deviations of 1.8° for angles and 0.65 mm for displacements.    NOTE: The negative values in the accuracy column of Table A-3 imply that the X-ray algorithm, on average, tends to underestimate the angle slightly.  Since all accuracy tests were performed with the vertebral bodies facing the same direction, the X-ray algorithm detects less flexion than there was (on average).  This may be a result of the optimization used in the algorithm, it may be because there are stronger features on one side of the vertebral body that dominate the correlation, or it may be that it is an artifact of X-ray projection.  Leivseth et al. [230] compared the DCRA technique with a gold standard, RSA, in the lumbar spine and found a negative bias in their measurements as well (RSA minus DCRA).  It could be that X-rays analysis is susceptible to “under-estimating” rotational angles because it is a projection of an object (with a width on the order of 1-2 cm) onto a planar surface.    Results in the context of literature Previous cadaver research has suggested that even small changes in the eccentricity of the spine, on the order of ±1 cm, can change the mechanism of injury [18, 19, 41].  Therefore, it would be desirable to detect changes in vertebral displacement of 10 mm.  Since the spine, on average, is 107 mm from C7 to C1 [204], this eccentricity could also be induced purely by rotation of the C7 vertebral angle by approximately 5°.  Axial impact tests on cadaveric head and neck specimens  have been conducted with changes to the neck and head postures on the order of 5° [44, 45].  Tests have also been conducted while changing the impact vector from the vertex of the head to locations from 1.5 cm through to 10 cm posterior to the vertex [95].  If the head height is 217  assumed to be 197 mm (average for a male) [204], a change in the impact vector of approximately 1.5 cm could result from a change in Frankfort plane angle of 4.4°.  The desired minimum detectable changes in kinematic measures for this study are based on the overall effect sizes of injuries reported in previous cadaveric research (see Table A-4).  A general rule for the accuracy required to measure a desired effect should ideally be an order of magnitude smaller [231].   Table A-4: Summary of desired accuracy and measured errors The desired minimum detectable changes, the ideal accuracy, and measurement error for various kinematic parameters.  The resolution is the expected change in the metric based on 1 pixel of movement (i.e. 1 pixel of motion of the corner of a vertebra 25 mm wide would result in 0.6° of reported rotation). Measure Minimum detectable change desired Ideal accuracy Measurement Error (RMSE) Resolution (per pixel) Vertebral angle (°) 5 0.5 0.6 0.57 to 0.95a Vertebral displacement (mm) 10 1 0.18 0.25 Frankfort plane angle (°) 5 0.5 0.74 0.33b a Assuming a 15 mm to 25 mm range for the width for a vertebral body [204] b Calculated from knowing the beads from the headband were 44 mm apart 218  Appendix B: Individual subject data for static inversion Table B-1: Individual subject vertebral translations  The vertebral translations (mid-inferior points of C1 through C7) for each subject for the Upright-Relaxed, Inverted-Relaxed, and Inverted-Forward conditions described relative to C7 (positive x = anterior, positive y = superior).  219  Table B-2: Individual subject head and vertebral angles Frankfort plane and vertebral angles (determined by the inferior vertebral corners) for each subject for the Upright-Relaxed, Inverted-Relaxed, and Inverted-Forward conditions described relative to the true horizontal (positive θ = flexion).        220  Table B-3: Individual EMG activities and MVC force, moments, and direction RMS EMG activity (% MVC) for Upright-Relaxed, Inverted-Relaxed, and Inverted-Forward for each subject for each muscle, maximum MVC force/moment recorded in all seven directions, and MVC direction where each muscle’s maximum activation was recorded.  221  Appendix C: Individual subject data for free tensed task   Figure C-1: Individual plots of upright tensed Upright X-Y coordinates (in mm) and EMG activities (% MVC) for each subject in the initial (black) and tensed (gray) states.  The change in curvature index (ΔCI) from the sustained muscle contractions is indicated and subjects are arranged according to the magnitude of curvature change (from lowest to highest).   222    Figure C-2: Individual plots of inverted tensed Inverted X-Y coordinates (in mm) and EMG activities (% MVC) for each subject in the initial (black) and tensed (gray) states.  The change in curvature index (ΔCI) from the sustained muscle contractions is indicated and subjects are arranged according to the magnitude of curvature change (but numbered as in Figure A3-1).    223  Appendix D: Direct linear transformation using the headband beads A direct linear transformation (DLT) rig was constructed with 17 beads at two different distances from the surface of the image intensifier (see Figure D-1).  The locations of the beads and the center of the image intensifier were digitized.  Then the X-ray image of the beads was recorded (see Figure D-2).  By knowing the 3D location of multiple beads and the projection of them onto the X-ray plane, equations [1]-[3] were used to solve for the DLT parameters for the X-ray system [232].    [1] ݑ െ ݋ݑ ൌ ܿሾ11ݎሺݔ െ ݋ݔሻ ൅ 12ݎሺݕ െ ݋ݕሻ ൅ 13ݎሺݖ െ ݋ݖሻሿ [2]       ݒ െ ݋ݒ ൌ ܿሾ21ݎሺݔ െ ݋ݔሻ ൅ 22ݎሺݕ െ ݋ݕሻ ൅ 23ݎሺݖ െ ݋ݖሻሿ [3]             െ݀ ൌ ܿሾ31ݎሺݔ െ ݋ݔሻ ൅ 32ݎሺݕ െ ݋ݕሻ ൅ 33ݎሺݖ െ ݋ݖሻሿ  Where x, y and z are the 3D locations of a single bead, xo, yo, and zo are the 3D position of the projection center (focal point of the X-ray beam), all of which are expressed in millimeters.  The principal point of the image plane coordinates are uo, vo and d is the distance between the image plane and the focal point of the X-ray beam, all of which are expressed in pixels.  The coefficients of the rotation matrix and the scaling factor between the 3D coordinate system and the 2D image plane are r11 through r33 and c, respectively.   Once the DLT parameters are derived, these can be used to back project the 2D image positions of the headband beads (in pixels) into 3D space (in millimeters), where we know the 3D locations of these beads (from the FaroArm digitization) (see Figure D-3).  The relationship between the 3D locations of the beads with respect to anatomical landmarks on the head is also known (also determined from the FaroArm digitization) (see Figure D-4).  This means that any subsequent motions of the bead on the X-ray plane can be used to determine the 3D motions of the anatomical landmarks on the head.  224   Figure D-1: DLT rig The DLT rig with 17 beads, was used to solve for the DLT parameters for the projection of the X-rays onto the image intensifier surface.   Figure D-2: DLT rig and X-ray projection of the beads The 3D position of each bead was known (left) and the 2D position of each bead could be identified in the X-ray image (right). 225   Figure D-3: 3D projection of the beads onto the 2D image plane The position of the headband beads in 3D space and the X-ray focal point are collinear (dashed black lines are two examples of this collinear relationship) with the 2D projection of those beads onto the X-ray image plane.   Figure D-4: FaroArm digitization of head landmarks The 3D positions of the anatomical landmarks on the head and beads were determined using a FaroArm. 226  Appendix E: Individual subject data for constrained tensed task  Figure E-1: Individual plots of upright flexion direction Upright flexion X-Y coordinates (in mm) and EMG activities (% MVC) for each subject in the initial (black) and tensed (gray) states.  The change in curvature index (ΔCI) from the sustained muscle contractions is indicated and subjects are arranged according to the magnitude of curvature change (from lowest to highest).  The sustained force (as a percentage of MVC) is indicated on the EMG plots (vertical gray line). 227   Figure E-2: Individual plots of inverted flexion direction Inverted flexion X-Y coordinates (in mm) and EMG activities (% MVC) for each subject in the initial (black) and tensed (gray) states.  The change in curvature index (ΔCI) from the sustained muscle contractions is indicated and subjects are arranged according to the magnitude of curvature change (from lowest to highest).  The sustained force (as a percentage of MVC) is indicated on the EMG plots (vertical gray line).  228   Figure E-3: Individual plots of upright extension direction Upright extension X-Y coordinates (in mm) and EMG activities (% MVC) for each subject in the initial (black) and tensed (gray) states.  The change in curvature index (ΔCI) from the sustained muscle contractions is indicated and subjects are arranged according to the magnitude of curvature change (from lowest to highest).  The sustained force (as a percentage of MVC) is indicated on the EMG plots (vertical gray line).  229   Figure E-4: Individual plots of inverted extension direction Inverted extension X-Y coordinates (in mm) and EMG activities (% MVC) for each subject in the initial (black) and tensed (gray) states.  The change in curvature index (ΔCI) from the sustained muscle contractions is indicated and subjects are arranged according to the magnitude of curvature change (from lowest to highest).  The sustained force (as a percentage of MVC) is indicated on the EMG plots (vertical gray line).  230  Table E-1: X-Y data of all subjects for upright and inverted flexion force The vertebral translations (mid-inferior points of C1 through C7) for each subject for the flexion conditions (upright initial, upright tensed, inverted initial, and inverted tensed) described relative to C7 (positive x = anterior, positive y = superior).  231  Table E-2: X-Y data of all subjects for upright and inverted extension force The vertebral translations (mid-inferior points of C1 through C7) for each subject for the extension conditions (upright initial, upright tensed, inverted initial, and inverted tensed) described relative to C7 (positive x = anterior, positive y = superior).  232  Table E-3: Angle data of all subjects for upright and inverted flexion force Frankfort plane and vertebral angles for each subject for the flexion conditions (upright initial, upright tensed, inverted initial, and inverted tensed) described relative to the true horizontal (positive θ = extension).      233   Table E-4: Angle data of all subjects for upright and inverted extension force Frankfort plane and vertebral angles for each subject for the extension conditions (upright initial, upright tensed, inverted initial, and inverted tensed) described relative to the true horizontal (positive θ = extension).    234  Table E-5: Muscle activation data for upright and inverted flexion force RMS EMG activity (% MVC) for the flexion conditions (upright initial, upright tensed, inverted initial, and inverted tensed) for each muscle, the force generated (in N and as % MVC), and MVC direction where each muscle's maximum activation was recorded.  235  Table E-6: Muscle activation data for upright and inverted extension force RMS EMG activity (% MVC) for the extension conditions (upright initial, upright tensed, inverted initial, and inverted tensed) for each muscle, the force generated (in N and as % MVC), and MVC direction where each muscle's maximum activation was recorded.   236  Appendix F: Additional data for dynamic tasks Table F-1: Individual subject vertebral angles (relative to global horizontal) Frankfort plane and vertebral angles (determined by the inferior vertebral corners) for each subject for the Upright-Flexion, Inverted-Flexion, Upright-Extension, and Inverted-Extension conditions described relative to the true horizontal (positive θ = extension).   237  Table F-2: Individual subject vertebral angles (relative to Upright-Static) Frankfort plane and vertebral angles (determined by the inferior vertebral corners) for each subject for the Upright-Flexion, Inverted-Flexion, Upright-Extension, and Inverted-Extension conditions described relative to the Upright-Static trial (positive θ = extension).    238   Table F-3: Individual EMG activities in the dynamic task RMS EMG activity (% MVC) for Upright-Flexion, Inverted-Flexion, Upright-Extension, and Inverted-Extension for each subject for each muscle (20 ms widow at neutral eccentricity).   239  Table F-4: Mean maximum muscle activity through the phases of dynamic flexion/extension The % MVC group mean peak RMS muscle activity (maximum 20 ms moving window throughout the phase) during the four phases of dynamic flexion and extension for both upright and inverted.  Upright Inverted Muscle (%MVC) Fully Extended to Neutral Neutral to Fully Flexed Fully Flexed to Neutral Neutral to Fully Extended Fully Extended to Neutral Neutral to Fully Flexed Fully Flexed to Neutral Neutral to Fully Extended STH 15.5 3.2 4.6 5.7 12.3 57.3 29.0 9.8 SCM 3.2 6.1 8.5 6.2 15.9 6.0 15.8 21.2 LS 4.2 10.6 24.5 11.4 19.4 9.7 13.7 51.7 MultC4 7.4 9.3 38.5 23.2 15.8 8.0 26.9 41.8 SsCerv 3.9 9.2 10.0 10.6 24.5 12.9 20.3 59.7 SsCap 9.6 13.4 12.5 21.4 14.3 23.8 25.7 19.2 SPL 11.2 14.5 18.8 10.4 16.8 14.9 19.2 21.1 Trap 20.1 19.4 22.0 20.0 38.2 28.9 46.4 45.6        240  Figures F-1 through F-10: Individual subject vertebral and head angles throughout motion Verterbal angles are plotted versus head angles for upright (left) and inverted (right) throughout the motion from full extension toward full flexion (black) and from full flexion toward full extension (black dashed).  As a reference the angles in the Upright-Relaxed trial were subtracted from the head and vertebral angles; thus, the point (0,0) is the Upright-Relaxed posture.  Extension is positive and all angle data were filtered at 1Hz [199].  Note the changes in scales for different subjects and vertebral levels.                        241   Figure F-1: Subject #3 vertebral and head angles throughout motion 242   Figure F-2: Subject #4 vertebral and head angles throughout motion 243   Figure F-3: Subject #5 vertebral and head angles throughout motion 244   Figure F-4: Subject #6 vertebral and head angles throughout motion 245   Figure F-5: Subject #7 vertebral and head angles throughout motion 246   Figure F-6: Subject #8 vertebral and head angles throughout motion 247   Figure F-7: Subject #9 vertebral and head angles throughout motion 248   Figure F-8: Subject #10 vertebral and head angles throughout motion 249   Figure F-9: Subject #11 vertebral and head angles throughout motion 250  Figure F-10 through 19: Individual subject muscle activities throughout motion EMG recordings (20 ms moving window; % MVC) for upright (left) and inverted (right) throughout the motion from full extension toward neutral (first dashed line) to full flexion (second dashed line) and from full flexion to neutral (third dashed line) toward full extension.  As a reference the head angles (black) and overall neck angle (gray) (with the Upright-Relaxed data subtracted) are plotted also over time.  Note the changes in scales for different subjects and different muscles.     251   Figure F-10: Subject #1 muscle activities throughout motion 252   Figure F-11: Subject #3 muscle activities throughout motion 253   Figure F-12: Subject #4 muscle activities throughout motion  254   Figure F-13: Subject #5 muscle activities throughout motion  255    Figure F-14: Subject #6 muscle activities throughout motion 256   Figure F-15: Subject #7 muscle activities throughout motion 257   Figure F-16: Subject #8 muscle activities throughout motion 258   Figure F-17: Subject #9 muscle activities throughout motion 259   Figure F-18: Subject #10 muscle activities throughout motion 260   Figure F-19: Subject #11 muscle activities throughout motion   

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0167424/manifest

Comment

Related Items