Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Towards a neck injury prevention helmet for head-first impacts : a mechanical investigation Nelson, Timothy Scott Jan 5, 2011

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata


24-ubc_2012_spring_nelson_timothy.pdf [ 22.78MB ]
JSON: 24-1.0072514.json
JSON-LD: 24-1.0072514-ld.json
RDF/XML (Pretty): 24-1.0072514-rdf.xml
RDF/JSON: 24-1.0072514-rdf.json
Turtle: 24-1.0072514-turtle.txt
N-Triples: 24-1.0072514-rdf-ntriples.txt
Original Record: 24-1.0072514-source.json
Full Text

Full Text

TOWARDS A NECK INJURY PREVENTION HELMET FOR HEAD-FIRST IMPACTS: A MECHANICAL INVESTIGATION  by  TIMOTHY SCOTT NELSON  B.A.Sc., The University of British Columbia, 2002  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY  in  THE FACULTY OF GRADUATE STUDIES  (Biomedical Engineering, Specialization in Mechanical Engineering)  THE UNIVERSITY OF BRITISH COLUMBIA  (Vancouver)     December 2011  © Timothy Scott Nelson, 2011  ii Abstract  Cervical spine and spinal cord injuries (SCI) have catastrophic and permanent neurological consequences and are known to occur from head-first impacts in many activities where helmets are worn.  A particularly dangerous posture for catastrophic cervical SCI from head-first impacts occurs when the head is flexed (nodded) approximately 30 degrees downward such that the cervical spinal column becomes aligned.  In this posture, the neck reacts axially along its stiffest axis such that high forces develop over small displacements. The deceleration of the torso creates strain energy in the vertebrae beyond their tolerance.  It has been shown that increasing constraint on the head at impact places the cervical spine at greater risk of injury compared to less constraining head conditions that allow the head to rotate and translate along the impact surface.  At impact speeds near the tolerance for injury this degree of head constraint can make the difference between avoiding neck injury altogether or the development of unstable neck fractures. This thesis involves the design, construction, and testing of a mechanical head, neck, and a helmet prototype that induces horizontal motion to the head as a neck injury mitigation approach in aligned column head-first impacts.  In addition, a new in vitro cervical spine model of head-first impact was developed for testing newer 3D helmet prototypes. All testing utilized a free standing drop tower to create an experimental model of the head, neck and torso system. The head and neck model exhibited an impact response that was in good agreement with the in vitro human response and was sensitive to surface compliance and platform angle. The lower-neck, head, and impact surface were instrumented to provide estimates of impact severity.  The helmet prototype, of realistic size, mass, and inertia, showed that when the induced head motion acted to increase the obliqueness of the impact, a combined injury metric comprised of peak neck axial force and peak bending moment was reduced by 39% to 43% compared to testing without induced head motion.  These reductions in lower-neck reaction loads were achieved without significant increases, or accompanying decreases in head accelerations.  This work is being used to develop and test subsequent helmet prototypes.  iii Preface  This work was conducted on a free standing drop tower already constructed by former students in our lab, chiefly, Philip Morley.  In addition, the surrogate head presented in Chapters 2-4 was initially designed by an undergraduate student design team led by Mr. Morely.  Aspects of both of the drop tower and surrogate head were redesigned for use in this thesis.  A version of Chapter 2 has been published as: Nelson, T.S. and P.A. Cripton, A New Biofidelic Sagittal Plane Surrogate Neck for Head-First Impacts. Traffic Injury Prevention, 2010. 11(3): p. 309-319.    T.S. Nelson was jointly responsible for the original ideas behind the paper, designing and building the surrogate neck, conducting the experiments, data analysis, and presentation of the findings as well as writing and editing the original paper.  Dr. Peter Cripton was jointly responsible for the original ideas, provided supervision, and was the principal editor of the paper.  A version of Chapter 4 has been published as a peer-reviewed conference proceedings: Nelson, T.S. and P.A. Cripton. Inducing Head Motion with a Novel Helmet during Head- First Impact Can Mitigate Neck Injury Metrics: An Experimental Proof-of-Concept Investigation using Mechanical Surrogates. in International Research Council on the Biomechanics of Impacts. 2008. Bern, Switzerland.  Co-author T.S. Nelson was jointly responsible for the original ideas behind the paper, designing and building the prototype helmet, conducting the experiments, data analysis, and presentation of the findings as well as writing and editing the original paper.  Co-author Dr. Peter Cripton was jointly responsible for the original ideas, provided supervision, and was the principal editor of the paper.  A version of Chapter 4 will be submitted to an academic journal.    iv The research performed and presented in Chapter 5 was carried out according to the ethical standards approved by the Natural Sciences and Engineering Research Council of Canada (NSERC).  – “Biomechanics of Spinal Cord Injury: High Speed Experimental Investigations”, UBC CREB Number: H04-70219  A version of chapter 5 has been published in several non-peer-reviewed conference proceedings: 1. Nelson, T.S., Van Toen (née Greaves), C.Y., Jones, C.F., Street, J., Cripton, P.A. Experimental Impact to the Hybrid III Head and Cadaveric Cervical Spine with an Advanced Muscle Force Replication System. Proceedings of the 3rd Annual Workshop on Biomechanical Experiments, September 16, 2008, Bern, Switzerland.  2. Van Toen (née Greaves), C.Y., Nelson, T.S., Jones, C.F., Street, J., Cripton, P.A. Development of an in vitro model of head-first impact with a Hybrid III head, surrogate spinal cord and simulated neck muscles. Proceedings of the 36th NHTSA International Workshop on Human Subjects for Biomechanical Research, November 2, 2008, San Antonio, Texas.  3. Van Toen, C., Jones, C.F., Nelson, T.S., Street, J., Cripton, P.A. Simulation of head- first impact using cervical spine specimens, simulated neck muscles, and a Hybrid III ATD head. Proceedings of the Ohio State University's 5th Annual Injury Biomechanics Symposium, May 18-19, 2009, Columbus Ohio.  Only a portion of the joint-authored experiment has been presented in Chapter 5 of this thesis and was written solely by co-author Nelson, T.S. and edited by co-author Dr. Peter Cripton.  It is anticipated that three first-author publications will be submitted to academic journals based on this work each focusing on the core area of contribution of each of the first 3 co- authors.  Co-author Nelson, T.S. was jointly responsible for the original ideas behind the paper(s) and his contribution was mainly to do with developing the use of the Hybrid III ATD head with the cadaveric cervical spine specimens, the necessary design and construction of experimental apparatus, and the implementation and data analysis of the head  v and neck instrumentation used in the experiment.  Co-author Van Toen, C was jointly responsible for the original ideas behind the paper and her contribution was mainly to do with design and implementation and data analysis related to the muscle force replication system and cadaveric specimen preparation.  Co-author Jones, C.F. was jointly responsible for the original ideas behind the papers and her contribution was centered around the construction, implementation, and imaging of a biofidelic surrogate spinal cord and data analysis relating to the spinal cord strains, as well as significant preparation of, and imaging, of the cadaveric specimens.  Co-author Street, J is an orthopaedic surgeon who was primarily responsible for providing expertise with the post-test diagnosis of injuries through CT images and dissections.  Co-author Dr. Peter Cripton was jointly responsible for the original ideas, provided supervision, was the principal editor of the paper(s), and did participate in the physical testing.  vi Table of Contents   Abstract .................................................................................................................................... ii  Preface ..................................................................................................................................... iii  Table of Contents ................................................................................................................... vi  List of Tables .......................................................................................................................... ix  List of Figures .......................................................................................................................... x  List of Acronyms and Abbreviations ................................................................................. xiv  Acknowledgements ............................................................................................................... xv  Dedication ................................................................................................................................ 1  Chapter 1: Introduction ......................................................................................................... 1  1.1  Overview ............................................................................................................................... 1  1.2  Significance of SCI and Cervical SCI ................................................................................... 2  1.2.1  Epidemiology of Cervical SCI from Head-First Impacts ................................................. 2  1.3  The Cervical Spine – Characteristics and Mechanical properties ......................................... 4  1.3.1  Anatomical Planes and Directions .................................................................................... 4  1.3.2  Head Motions ................................................................................................................... 5  1.3.3  Cervical Spine Anatomy ................................................................................................... 5  Cervical Spine Vertebrae and Ligaments ................................................................. 6  Cervical Spine Musculature ................................................................................... 10  1.4  Models to Study Cervical Spine Properties and Injuries..................................................... 12  1.5  Cervical Spine:  Mechanical Properties .............................................................................. 14  1.5.1  Nonlinear Stiffness and Neutral Zone ............................................................................ 14  1.5.2  Head-Neck-Torso System ............................................................................................... 15  1.5.3  Column Theory, Buckling, and the Cervical Spine ........................................................ 17  1.6  Cervical Spine Injuries ........................................................................................................ 19  1.6.1  Spinal Column Injuries ................................................................................................... 20  Upper Cervical Spinal Column Compressive Injuries ........................................... 20  Lower Cervical Spinal Column Compression Injuries .......................................... 23  1.6.2  Spinal Cord Injuries (SCI) .............................................................................................. 24  vii 1.6.3  Factors Affecting Injury and SCI in Head-First Impacts ................................................ 25  Impact Velocity ...................................................................................................... 25  Posture at Impact – Observational ......................................................................... 26  Alignment of the Load with the Crown of the Helmet and Axis of the Spine ....... 28  End Conditions Matter ........................................................................................... 28  Surface Conditions, Friction and Compliance (padding) ....................................... 30  Cervical Muscular Contraction .............................................................................. 30  1.7  Helmets and Head Injury .................................................................................................... 31  1.8  Helmets and Neck Injuries .................................................................................................. 31  1.9  Prior Preventative Strategies ............................................................................................... 32  1.10  Induced Head Motion in Axial Impacts .............................................................................. 36  1.11  Conclusions from Literature about Cervical Spine Injuries in Head-First Impact .............. 37  1.12  Research Questions and Objectives .................................................................................... 39  1.13  Scope ................................................................................................................................... 40  Chapter 2: A New Biofidelic Sagittal Plane Neck Model for Head-First Impacts .......... 41  2.1  Introduction ......................................................................................................................... 41  2.2  Materials and Methods ........................................................................................................ 43  2.2.1  Surrogate Neck Model Design ........................................................................................ 43  2.2.2  Methods: Flexibility Testing ........................................................................................... 51  2.2.3  Methods: Drop Testing ................................................................................................... 52  2.3  Results ................................................................................................................................. 54  2.3.1  Flexibility Testing – Effects of Follower Load .............................................................. 54  2.3.2  Drop Testing – Effects of Follower Load ....................................................................... 58  2.4  Discussion ........................................................................................................................... 62  Chapter 3: Characterization and Comparison of a New Biofidelic Sagittal Plane Surrogate Neck to the Hybrid III Head and Neck for Head-First Impacts .................... 69  3.1  Introduction ......................................................................................................................... 69  3.2  Materials and Methods ........................................................................................................ 71  3.2.1  Materials ......................................................................................................................... 71  3.2.2  Experimental Design ...................................................................................................... 72  3.2.3  Methods .......................................................................................................................... 74  3.3  Results ................................................................................................................................. 77  3.3.1  Preliminary L9 Fractional Factorial Experiment ............................................................ 77  viii 3.3.2  Hybrid III Full Factorial Comparison Experiment ......................................................... 80  3.4  Discussion ........................................................................................................................... 86  Chapter 4: An Experimental Neck Injury Prevention Helmet: Inducing Head Motion to Mitigate Neck Loading in Head-First Impacts .................................................................. 96  4.1  Introduction ......................................................................................................................... 96  4.2  Materials and Methods ........................................................................................................ 98  4.3  Results ............................................................................................................................... 108  4.4  Discussion ......................................................................................................................... 120  Chapter 5: A New Model of Head-First Impact using Aligned Cervical Spine Specimens, the Hybrid III ATD Head, and Neck Muscle Force Simulation ................. 131  5.1  Introduction ....................................................................................................................... 131  5.2  Materials and Methods ...................................................................................................... 136  5.3  Results ............................................................................................................................... 144  5.4  Discussion ......................................................................................................................... 160  Chapter 6: Discussion and Conclusions ............................................................................ 169  6.1  Overview of Findings ........................................................................................................ 169  6.2  Comparisons to Existing Findings .................................................................................... 170  6.3  Strengths and Limitations ................................................................................................. 180  6.4  Conclusions ....................................................................................................................... 187  6.5  Contributions ..................................................................................................................... 189  6.6  Applications of Research Findings ................................................................................... 190  6.7  Future Research................................................................................................................. 191  References ............................................................................................................................ 194  Appendix A Machine Drawings of SC7 Head and Neck ............................................................... 205  Appendix B Machine Drawings of Prototype Helmet ................................................................... 221  Appendix C High Speed Video Sequences for Chapter 4 Experiments ......................................... 228  Appendix D Full Instrumentation Traces for Chapter 5 Experiments ........................................... 246  ix List of Tables  Table 2-1:  Surrogate head and neck mass and inertia comparisons ...................................... 45  Table 2-2:  Surrogate head and neck center of gravity location comparisons ........................ 45  Table 2-3:  Surrogate neck model vertebral dimensions and human comparison .................. 46  Table 2-4:  Neutral zone and range of motion results from flexibility testing ....................... 57  Table 2-5:  Drop testing impact parameters ............................................................................ 60  Table 3-1:  Impact variables studied in L9 fractional factorial ............................................... 72  Table 3-2:  L9 experiment impact parameters ........................................................................ 79  Table 3-3:  Impact parameters for full factorial experiment ................................................... 84  Table 3-4:  Impact parameter comparison between SC7 and Hybrid III head-neck models .. 85  Table 4-1:  Mechanical helmet mass and inertia comparison ............................................... 101  Table 4-2:  Head and neck kinematic measurements (and range) for duplicated experimental runs ........................................................................................................................................ 109  Table 4-3:  Mean peak kinetic impact parameters (and range) by run ................................. 111  Table 4-4:  ANOVA tables for all injury metrics ................................................................. 112  Table 4-5:  Factorial ANOVA, post hoc multiple comparisons for head and neck injury metrics ................................................................................................................................... 119  Table 5-1:  Cervical spine specimen information ................................................................. 136  Table 5-2:  Impact parameters for the five drop tests ........................................................... 147    x List of Figures  Figure 1-1: Anatomical reference planes .................................................................................. 4  Figure 1-2: Head motions provided by cervical spine .............................................................. 5  Figure 1-3: Human skull and cervical spine ............................................................................. 6  Figure 1-4:  Typical cervical vertebra superior view ................................................................ 7  Figure 1-5: Typical cervical functional spinal unit ................................................................... 7  Figure 1-6: The upper cervical vertebrae, atlas and axis .......................................................... 9  Figure 1-7:  Ligaments of a typical functional spinal unit ...................................................... 10  Figure 1-8:  Neck musculature cross section at C4 level ........................................................ 11  Figure 1-9:  Force deformation characterization of the cadaveric cervical spine ................... 15  Figure 1-10:  Head and lower-neck axial forces for cadaveric head and neck drop tests. ...... 16  Figure 1-11:  Jefferson fractures of the atlas .......................................................................... 21  Figure 1-13:  C2 ring fracture classification ........................................................................... 22  Figure 1-14:  Compressive cervical spine injury classification based on eccentricity ........... 24  Figure 1-15: Neutral and aligned head and neck postures ...................................................... 27  Figure 1-16:  The Leatt Brace™ neck brace for mountain biking and motorcycle markets .. 33  Figure 1-17:  Football neck collars ......................................................................................... 34  Figure 1-18:  Schematic of helmet-shoulder coupling device (left) and prototype version (right). ..................................................................................................................................... 35  Figure 2-1:  New head and neck model on drop tower ........................................................... 43  Figure 2-2:  Schematics of as-built model showing lateral and frontal views ........................ 47  Figure 2-3:  Range of motion for fully compressed and fully distracted conditions .............. 49  Figure 2-4:  Surrogate neck vertebral dimensions related to human anatomy and segmental stiffness ................................................................................................................................... 51  Figure 2-5:  Schematic of surrogate head and neck on drop tower ........................................ 53  Figure 2-6:  Drop testing pre-impact postures and flexibility testing motion endpoints ........ 54  Figure 2-7:  Upper and lower surrogate neck flexibility testing curves ................................. 55  Figure 2-8:  Segmental surrogate neck flexibility curves ....................................................... 56  Figure 2-9:  Drop testing temporal head-neck forces ............................................................. 58  Figure 2-11:  Drop testing neck compression ......................................................................... 61  Figure 2-12:  Drop testing intervertebral angles ..................................................................... 62  xi Figure 3-1:  Schematic of head and SC7 neck model on drop tower ..................................... 71  Figure 3-2:  Impact variables studied in the L9 fractional factorial ....................................... 73  Figure 3-3:  L9 experiment lower-neck axial force main effects ............................................ 77  Figure 3-4:  L9 experiment sagittal lower-neck sagittal moment main effects ...................... 78  Figure 3-5:  Impact variables studied in the Hybrid III full factorial ..................................... 80  Figure 3-6:  Peak lower-neck axial force – Hybrid III full factorial experiment .................... 82  Figure 3-7:  Peak lower-neck sagittal moment – Hybrid III full factorial experiment ........... 82  Figure 3-8:  Full factorial experiment temporal head-neck forces and head accelerations .... 83  Figure 4-1:  Mechanical head, neck, and helmet .................................................................. 100  Figure 4-2:  Schematic of full factorial experiment .............................................................. 103  Figure 4-3:  Drop testing schematic ...................................................................................... 107  Figure 4-4:  Coordinate systems for head and neck instrumentation .................................... 108  Figure 4-5:  Head rotation and horizontal translation ........................................................... 110  Figure 4-6:  Factorial results – lower-neck axial force ......................................................... 113  Figure 4-7:  Factorial results – lower-neck sagittal moment ................................................ 114  Figure 4-9:  Factorial results – peak resultant head CoG acceleration ................................. 116  Figure 4-10:  Factorial results – Head Injury Criterion ........................................................ 117  Figure 4-11:  Temporal head and neck axial force development comparison run 12 EPT vs run 18 NH ............................................................................................................................. 121  Figure 4-12:  Temporal lower-neck axial force and sagittal moment comparsion run 12 EPT vs run 18 NH ......................................................................................................................... 122  Figure 4-13:  Helmet-shoulder coupling load path conceptual schematic ............................ 125  Figure 5-1:  The Saari et al. model of head-first impact ....................................................... 134  Figure 5-2:  Testing schematic .............................................................................................. 138  Figure 5-3:  Schematic of muscle force vertebral anchoring ................................................ 139  Figure 5-4:  Photographs of model pre-impact ..................................................................... 142  Figure 5-5:  Coordinate systems for instrumentation and data analysis ............................... 143  Figure 5-6:  Free body diagram to resolve moments at C7/T1 ............................................. 144  Figure 5-7:  Representative temporal plot of lower-neck forces, moments, and head accelerations .......................................................................................................................... 146  Figure 5-8:  High speed video frames for specimen H1220 ................................................. 149  xii Figure 5-9:  H1220 kinetic analysis plot ............................................................................... 150  Figure 5-10:  High speed video frames for specimen H1221 ............................................... 151  Figure 5-11:  H1221 kinetic analysis plot ............................................................................. 152  Figure 5-12:  High speed video frames for specimen H1222 ............................................... 153  Figure 5-13:  H1222 kinetic analysis plot ............................................................................. 154  Figure 5-14:  High speed video frames for specimen H1223 ............................................... 155  Figure 5-15:  H1223 kinetic analysis plot ............................................................................. 156  Figure 5-16:  High speed video frames for specimen H1224 ............................................... 158  Figure 5-17:  H1224 kinetic analysis plot ............................................................................. 159  Figure A-1:  SC7 head, neck, and helmet prototype assembly……………………………..205  Figure A-2:  SC7 head assembly .......................................................................................... 206  Figure A-3:  Skull cap ........................................................................................................... 207  Figure A-4:  Head tube ......................................................................................................... 208  Figure A-5: Motion guide inner ............................................................................................ 209  Figure A-6:  Head steel mass ................................................................................................ 210  Figure A-7:  C0 occiput – as built ......................................................................................... 211  Figure A-8:  SC7 neck assembly .......................................................................................... 212  Figure A-9:  C1 vertebra – as built ....................................................................................... 213  Figure A-10:  C2 vertebra – as built ..................................................................................... 214  Figure A-11;  C3 vertebra – as built ..................................................................................... 215  Figure A-12:  C4 vertebra – as built ..................................................................................... 216  Figure A-13:  C5 vertebra – as built ..................................................................................... 217  Figure A-14:  C6 vertebra – as built ..................................................................................... 218  Figure A-15:  C7 vertebra – as built ..................................................................................... 219  Figure A-16:  T1 vertebra – as built ...................................................................................... 220  Figure B-1:  Prototype helmet full assembly ........................................................................ 221  Figure B-2:  Prototype helmet top cap sheet 1 of 2 – preformed flat state ........................... 222  Figure B-3:  Prototype helmet top cap sheet 2 of 2 – formed state ...................................... 223  Figure B-4:  Prototype helmet side support .......................................................................... 224  Figure B-5:  Prototype helmet side support gusset ............................................................... 225  Figure B-6:  Prototype helmet motion deployment guide - right ......................................... 226  xiii Figure B-7:  Prototype helmet motion deployment guide - left ............................................ 227  Figure C-1:  High speed video frames with post-head-contact time for run 1 ..................... 228  Figure C-2:  High speed video frames with post-head-contact time for run 2 ..................... 229  Figure C-3:  High speed video frames with post-head-contact time for run 3 ..................... 230  Figure C-4:  High speed video frames with post-head-contact time for run 4 ..................... 231  Figure C-5:  High speed video frames with post-head-contact time for run 5 ..................... 232  Figure C-6:  High speed video frames with post-head-contact time for run 6 ..................... 233  Figure C-7:  High speed video frames with post-head-contact time for run 7 ..................... 234  Figure C-8:  High speed video frames with post-head-contact time for run 8 ..................... 235  Figure C-9:  High speed video frames with post-head-contact time for run 9 ..................... 236  Figure C-10:  High speed video frames with post-head-contact time for run 10 ................. 237  Figure C-11:  High speed video frames with post-head-contact time for run 11 ................. 238  Figure C-12:  High speed video frames with post-head-contact time for run 12 ................. 239  Figure C-13:  High speed video frames with post-head-contact time for run 13 ................. 240  Figure C-14:  High speed video frames with post-head-contact time for run 14 ................. 241  Figure C-15:  High speed video frames with post-head-contact time for run 15 ................. 242  Figure C-16:  High speed video frames with post-head-contact time for run 16 ................. 243  Figure C-17:  High speed video frames with post-head-contact time for run 17 ................. 244  Figure C-18:  High speed video frames with post-head-contact time for run 18 ................. 245  Figure D-1: Full instrumentation output for specimen H1220 ............................................. 246  Figure D-2: Full instrumentation output for specimen H1221 ............................................. 247  Figure D-3: Full instrumentation output for specimen H1222 ............................................. 248  Figure D-4: Full instrumentation output for specimen H1223 ............................................. 249  Figure D-5: Full instrumentation output for specimen H1224 ............................................. 250       xiv List of Acronyms and Abbreviations  AMFR – advanced muscle force replication  ANOVA – analysis of variance  COR – center of rotation  EPT – extension posterior translation – the name of one of the helmet motion escapes that cause head posterior translation with extension rotation  FAT – flexion anterior translation – the name of the helmet motion escapes that cause head anterior translation with flexion rotation  Hybrid III – the most widely used crash test dummy for automotive safety research which is endorsed by NHTSA and developed by General Motors  HIII – Hybrid III (see above)  HIIIAMFR – see HIII and AMFR above  L9 – a fractional factorial experimental design consisting of 9 unique experimental runs  NH – no helmet – tests performed without a head-motion-inducing helmet  NZ – neutral zone  ROM – range of motion  SC7 – sagittal compressive 7 vertebrae  - the name given to the self-designed crash test dummy neck designed and used throughout this thesis.   xv Acknowledgements   I would like to thank my wife Kristy for her never-ending support and encouragement.  I would like to thank my daughter Amber, who was the best thesis-baby I could have imagined, for really providing extra motivation to finish up.  I offer my sincere gratitude to my friends and family who have supported and encouraged me throughout this journey.  I always felt so much more grounded after a visit home.  I would like to thank Dr. Peter Cripton for his encouragement, tolerance, optimism, and reminders not to put life on hold as I enjoyed several personal milestones along the way.  I would like to thank the 5 anonymous men and women who donated their bodies to science and participated in this study.  I would like to thank my research committee, Dr. Peter Cripton, Dr. Tom Oxland, and Dr. Antony Hodgson for their guidance.  I would like to thank my examination committee, Dr. Douglas Romilly, Dr. Karim Khan, and Dr. John Bolte IV for their input and recommendations regarding this thesis.  I am grateful to fellow students and colleagues Amy Saari and Phil Morley for getting me started, Markus and Roland in the machine shop who were always up for the challenge, Robyn Newell, Claire Jones, and Carolyn Van Toen for setting such high standards, James Boak for great times on conference road trips and especially in Switzerland.  I would like to acknowledge funding provided by the British Columbia Innovation Council Innovation scholarship, UBC for their PhD Tuition award, the UBC Industry Liaison Office for project support, and ICORD for travel grants.    xvi Dedication  Luck:  I have had it many times when I needed it.  Like many Canadians, I grew up with a passion for multiple action-sports that carried huge risks that eventually changed my life through injury, but of the type where I recovered.  I got lucky numerous times and got up from some crashes where I didn’t then understand how I avoided injury.  More specifically, getting thrown over the bars of motocross and mountain bikes and landing head-first.  While I understand better now how I could have avoided injury, I also know that much more about just how lucky I was, and everyone else is who has gotten up from one of these crashes, to still be walking.  This body of work is dedicated to all of those who weren’t so lucky to get up from their crashes and suffered spinal cord injury, but who kept on going nonetheless.  When the PhD process got tough for me, I looked to you for inspiration; as a reminder of how lucky I am, as a reminder of why I was doing it, and a reminder of what real adversity is.  It has been a privilege to work on this project and it is my sincere hope that this body of work contributes to helping others get up from their crashes.    1 Chapter 1: Introduction  1.1 Overview  Spinal cord injuries (SCI) are medically devastating events that usually leave victims partially or completely paralyzed below the level of the injury. The vast majority of SCI are presently irreversible and they occur most often in young men [1].  Approximately half of all SCI occur in the 7 uppermost vertebrae of the spine known as the cervical spine or colloquially as the neck [2].  While the cervical spine can be injured and SCI incurred through a number of different loading vectors, a particular class of cervical spine injuries that can have immediate and catastrophic consequences are axial compressive type injuries.  These injuries are most likely to occur when the cervical spine is aligned and subjected to compressive load from a force delivered to the crown of the head.  This aligned posture occurs when the head is flexed (nodded downwards) approximately 30 degrees and has been shown to be the primary cause of catastrophic cervical spine injuries causing quadriplegia in both football [3] and hockey [4].  It can occur from an inverted fall onto one’s head or a head-first impact with another athlete or an inanimate object and can occur in a wide range of activities under certain conditions including: automotive rollovers, motorcycles, bicycles, football, hockey, equestrian, diving, falls from heights, among others.  By the nature of these injury environments, the large majority occur in the presence of an engineered interface between the head and the contact surface such as a helmet or an automobile roof. Despite advances in protective equipment in sport and passive safety systems for automobiles, the cervical spine remains unprotected against axial compressive injuries despite there being a clear need for such protection.  As will be described in detail, a summary of the literature on cervical spine injuries from head-first impacts suggests that a possible prevention strategy exists by modifying the conditions between the head and impact surface upon impact to keep the head moving along the impact surface during the impact. The overall goal of this dissertation research was to design, manufacture, and evaluate a helmet prototype that induces horizontal motion to the head upon impact as a neck injury mitigation strategy in these highly dangerous aligned column impacts to the crown of the helmeted head.  Towards this goal, a mechanical head and neck was designed, built, and characterized in order to test the prototype helmet.  In addition, a 2nd model of head-first impact using cadaveric cervical spines was developed for testing future 3D helmet prototypes.  2 1.2 Significance of SCI and Cervical SCI  SCI are a significant problem throughout the world with worldwide incidences estimated at between 10.4 and 83 per million inhabitants per year [5].  About 1,100 people suffer from SCI each year in Canada [6] and about 12,000 new cases occur each year in the USA [2].  Prevalence estimates show that in 2009, in Canada there were approximately 41,000 persons living with SCI [6] while in the USA the number was approximately 311,000 [2]. The cost of SCI relates directly to the severity of neurological impairment and thus varies greatly according to the location of the spinal cord lesion along the spinal column.  The higher the lesion occurs in the cervical spine, the more costly and more severe the injury which is why cervical SCI are particularly devastating.  Canadian data shows that the lifetime healthcare costs for a person with SCI range between $1.25 and $25 million dollars [6].  In the USA, the average yearly health care and living expenses costs that are directly attributable to spinal column injury range from USD $17,139 for Incomplete Motor Functional at any Level to $146,645 for High Tetraplegia (C1-C4).  These costs do not consider the first year costs of treatment, which range from $244,562 to $829,843 respectively.  Over the lifetime of an injured 25 year old, the estimated lifetime costs range from USD $729,560 to $3,273,270 [2].  The USA cost data combined with the worldwide incidence rates above puts the annual costs of worldwide cervical spinal cord injuries well into the billions of dollars.  Furthermore, these costs are only those directly associated with injury treatment and specialized medical care afterwards.  There is additional financial cost to society when one considers that the major demographic is young males [1] and that 62% of those who suffer an SCI remain unemployed throughout the majority of their lives [6]. Those living with cervical SCI can certainly lead meaningful, happy, and productive lives but the real cost of cervical SCI cannot be monetized; the loss of quality of life is significant and the expected lifespan for those living with SCI is shorter than the non-injured population [2]. 1.2.1 Epidemiology of Cervical SCI from Head­First Impacts  Approximately 55% of SCI occur in the cervical spine region [2].  Estimating the proportion of cervical SCI incurred from head-first impacts (HFI) is challenging as many larger epidemiological studies lack information about the injury mechanism.  However, smaller populations of athletes can provide some perspective: taking football as an example, sixty five percent of 220 spinal cord injured football players over the 25 year period 1977 and 2001 were  3 injured during events associated with head-first impact such as tackling or tackling head down [7].  Although rule changes implemented in 1976 to prevent heads-down tackling in football greatly reduced injury incidence rates, since then they have changed very little and head-down impacts are still regularly occurring [8].  The number of heads-down hits in football has been estimated observationally at approximately 20 per team, per game.  This was extrapolated, along with SCI data to make the estimate that at the high school level, one case of quadriplegia was occurring per 251,000 head-down contacts [8].  The rarity of these injuries is consistent with other studies as well.  Incidence rates of catastrophic spinal injuries in high school level football and hockey have been estimated at 0.68 and 2.56 per 100,000 participants respectively [9]. In recreational skiing and snowboarding, the incidence rate of cervical spine fractures (with or without SCI) was estimated as 0.79 and 0.21 per 100,000 days of skiing or snowboarding respectively, at the Whistler/Blackcomb ski area in Canada [10].  A study analyzing European motocross racers over a 12 year period, where each race lasts approximately 20-30 minutes, found 3 cervical spine fractures with accompanying SCI out of 1500 recorded accidents that occurred over an estimated 66,000 hours of riding time [11].  In another observational study looking at SCI from mountain biking over a 13 year period in British Columbia, of the 107 spinal injuries that presented, 79 of them, and 31 of the 43 SCI (72%), occurred in the cervical region.  Of the 31 cervical SCI, 9 (29%) resulted in complete paralysis below the level of injury.  This study also confirmed that head-first impacts from an “over the bars” type of fall with primary impact to the head was the most common injury mechanism [12].  Overall it seems clear that while the incidence rate for these injuries is low, they are regularly occurring with high costs, devastating consequences, and no cure.    4 1.3 The Cervical Spine – Characteristics and Mechanical properties  This section describes the cervical spine anatomy using anatomical reference systems and terminology. 1.3.1 Anatomical Planes and Directions  In order to describe the anatomy of the cervical spine, first some anatomical definitions are required.  Figure 1-1 shows anatomical reference planes.  A sagittal plane divides the body into left and right portions; the central one being referred to as the mid-sagittal plane.  Similarly a coronal plane (also termed frontal) divides the body into front and rear portions while a transverse plane partitions the body into upper and lower portions.  Again referring to Figure 1-1, the direction towards the head is termed superior while inferior means away from the head.  The direction towards the front of the body is called anterior while towards the rear is posterior.  The direction meaning towards the mid-sagittal plane is medial while away from this plane is lateral.  These terms will be used throughout the thesis.   Figure 1-1: Anatomical reference planes Sagittal, Coronal, and Transverse planes of the human body.  Image obtained from Wikimedia Commons.    5 1.3.2 Head Motions  The four main head motions that the cervical spine allows are shown in Figure 1-2.  The cervical spine is by far the most mobile section of the column.  In a recent study to quantify range of motion, young adult male volunteers moved their heads through 70 degrees of flexion, 69 degrees of extension, 42 degrees of lateral bending (average left and right), and 80 degrees axial rotation [13].  The lateral bending and axial rotation values are for a single direction and are nearly symmetric from left to right side.  Flexion and extension bending occur within the sagittal plane and lateral bending and axial rotation are coupled motions due to the geometry of the cervical anatomy [14].   Figure 1-2: Head motions provided by cervical spine The four major physiologic head motions: flexion, extension, lateral bending, and axial rotation.  Graphic from Huelke and Nusholtz, 1986 [15] and used with permission from John Wiley and Sons, Inc.  1.3.3 Cervical Spine Anatomy   The cervical spine is comprised of several bones, joints, muscles and ligaments.  From a structural standpoint its two main roles are to allow for movement of the head and to protect the  6 delicate spinal cord which connects the brain to the peripheral nervous system throughout the body.  Figure 1-3: Human skull and cervical spine The 7 cervical vertebrae comprising the human cervical spine and the first thoracic vertebra.  The posture shown is the natural lordotic curvature.  Image modified from Wikimedia Commons provided by Patrick J Lynch, medical illustrator. Cervical Spine Vertebrae and Ligaments  The cervical spine comprises the uppermost 7 bony vertebrae of the human spinal column.  In the neutral anatomical posture a healthy cervical spine has a lordotic curvature which is defined to mean that the convex side of the curvature is anterior and is shown in Figure 1-3.  The bony vertebrae are named C1 to C7 moving inferiorly.  C1 interacts superiorly with the occipital condyles at the base of the skull and C7 interacts inferiorly with the first vertebra of the thoracic spine, T1.  The cervical vertebrae C3 to C7 are considered “typical” because while their size increases moving inferiorly, they have common features.  Figure 1-4 shows a typical cervical vertebra.  Adjacent cervical vertebrae articulate with each other through three joints.  The largest is the fibrous intervertebral discs, on the anterior portion of the column, which connect adjacent articulating bodies and the other two, on the posterior portion of the column, and equidistant from the mid-sagittal plane, are the diarthrodial zygapophysial joints also called “facet” joints.  The connection of two vertebrae together are referred to as a functional spinal unit as shown in Figure 1-5.  The orientation of the two bilateral zygapophysial joints located in the posterior portion of the column guide and restrict the motion allowed between adjacent  7 vertebrae.  The intervertebral discs, in the anterior portion of the column serve to connect adjacent vertebrae (ligament) and also to act as shock absorbers.  The discs are comprised of an outer annulus fibrosis that surrounds a central nucleus pulposis which has the consistency of a fluid-like gel.  In the lumbar spine it has been shown that healthy discs exhibit hydrostatic pressure [16].  This has not been proven in the cervical spine although it is assumed that the morphologically similar cervical discs also exhibit hydrostatic pressure [17].    Figure 1-4:  Typical cervical vertebra superior view Superior view showing anatomical features of a typical cervical vertebra.  Image adapted from Gray’s Anatomy for Students, 2004 [18] and used with permission from Elsevier.   Figure 1-5: Typical cervical functional spinal unit Two adjacent cervical vertebrae comprising a typical cervical functional spinal unit.  Each functional spinal unit has three articulating joints consisting of the Interbody joint and two Zygapophysial joints.  Image adapted from Gray’s Anatomy for Students, 2004 [18] and used with permission from Elsevier.   8 The large range of motion in the cervical spine is attributed to the higher ratio of intervertebral disc to vertebral body thickness, the large surface area of the articulating facet joints, and their relatively flat anterior-posterior slope in the sagittal plane [19].  The flatter sloped contact surface combines with a looser articular capsule to promote rotation and allow for a greater deal of flexion and extension than other spinal regions.  The sloped orientation of the joints dictates that axial rotation of the head is coupled with lateral bending of the cervical spine [20].  Another anatomical feature unique to the cervical region is the presence of bi-lateral uncinate processes that project superiorly on the lateral borders of the vertebral bodies.  These are referred to by some as uncovertebral joints as there are joint capsules in some people [19]. They are thought to act as guides to accommodate the larger degree of relative intervertebral anteroposterior shear translation observed in the cervical spine compared to other spinal regions. At the upper cervical spine there are two “atypical vertebrae” with unique anatomy called the atlas (C1) that articulates superiorly with the skull and inferiorly with the axis (C2) that are shown in Figure 1-6.  The main role of the atlas is to articulate with and support the skull via the articular facets on its superior surface.  The only substantial degree of relative motion allowed between the skull and C1 is flexion-extension or nodding.  Dense ligaments as well as musculature secure the occipital condyles in the articular facets on C1.  The atlas can be thought of as a “washer” that moves with and supports the head.  The most evident motion of this is axial rotation of the head which involves the head and C1 pivoting about the odontoid process (dens) of C2 which projects superiorly from C2 to articulate with the facet on the posterior margin of the anterior ring of the atlas.  Dense ligaments, called the transverse ligaments of the atlas prevent anterior motion of the atlas relative to the axis.  C1 has bilateral articular facets on its inferior surface that articulate with flat bilateral facets on the superior side of C2.  The majority of axial rotation of the head occurs at the C1/C2 joint.  The odontoid process also secures the alar ligaments which connect to the medial borders of the occipital bone on the skull and act to limit the degree of maximum head axial rotation as well as assist the transverse ligaments in preventing anterior motion of the atlas relative to the axis.   9  The cervical column has a great number of fibrous ligaments that connect the vertebrae together.  The ligaments which attach two vertebrae to comprise a typical functional spinal unit are shown in (Figure 1-7) the main ligaments are the anterior and posterior longitudinal ligaments which are on the anterior and posterior portions of the vertebral bodies respectively.  Thus the posterior longitudinal ligament is inside the vertebral canal.  In addition there are bilateral inter transverse ligaments which connect the transverse processes of adjacent vertebrae.  Similarly, the ligamentum flavum connect the lamina of adjacent vertebrae.  At each zygapophysial (or facet) joint there are capsular ligaments which adjoin the superior articular process of an inferior vertebra to the inferior articular process of a superior vertebra.  These capsular ligaments are somewhat loose to allow the motion that occurs at the zygapophysial  Figure 1-6: The upper cervical vertebrae, atlas and axis The atlas and axis vertebrae.  top left: superior view of atlas, top right: superior view of atlas and axis, bottom left: superior view of axis, bottom middle: posterior view of axis, bottom right: posterior view of atlas and axis with transverse and alar ligaments.  Image adapted from Gray’s Anatomy for Students, 2004 [18] and used with permission from Elsevier.  10 joints.  At the posterior part of the column, the interspinous ligaments connect the spinous processes of adjacent vertebrae and at the very posterior margin of the spinous processes, the supraspinous ligament runs the length of the spinal column and connects all of the spinous processes together.  Figure 1-7:  Ligaments of a typical functional spinal unit Lateral (left) and superior (right) views of a typical cervical functional spinal unit.  Ligaments shown are the anterior longitudinal ligament (A), anterior (B) and posterior (C) annulus fibrosus which together comprise the intervertebral disc, posterior longitudinal (D), intertransverse (E), facet capsular (F), ligamentum flavum (H), and the interspinous and supraspinous (I) where the supraspinous runs along the most posterior margins of the spinous processes.  Image and description adapted from Panjabi et. al., 1975 [21] and used with permission from Elsevier Limited. Cervical Spine Musculature  Muscles make up the bulk of the human neck and their anatomy is very complex.  It is not necessary in this thesis to present all of the neck muscles that act on the head and neck although some of the more major muscles are discussed in Chapter 5 so they are introduced briefly here.  There are at least 25 pairs of superficial muscles that act on the cervical vertebrae and the head without including the paraspinal muscles that connect vertebrae usually across multiple levels along the spine [22].  The moment arms that muscles have are not constant throughout the range of motion and also individual muscles have an effect upon other muscles’ lines of action due to their overlapping nature [22].  Many muscles act along similar lines of action such that contraction of more than one muscle can produce the same movement.  In addition a broad  11 muscle such as the trapezius which connects to the clavicle, the spinous processes, the skull, and the scapula, can act to produce multiple different motions depending on which portion of it is activated.  Neck muscles are activated in complex and concerted patterns that produce multidirectional movements [23].  Neck (and back) muscles are usually divided into groups such as superficial and deep muscle groups.  They are divided into groups as flexors that act to flex the head and extensors which extend the head.  Figure 1-8 shows a cross section of the human neck taken at the C4 level.  As is evident from the figure, the musculature is responsible for the majority of the cross section.  In the figure, anterior is upwards and it is evident that the cervical spine has more musculature around the posterior region than the anterior.  The flexor muscles labelled in Figure 1-8 are the sternocleidomastoid, the longus colli and longus capitus.  The extensor muscles are the levator scapulae, splenius capitis, seminspinalis capitis, semispinalis cervicus, C4 multifidus, and the trapezius.   Figure 1-8:  Neck musculature cross section at C4 level Deep and Superficial neck musculature.  SCM-Sternocleidomastoid, LS-Levator Scapulae, Trap- Trapezius, SsCap – Semispinalis Capitis, SsCerv – Semispinalis Cervicus, C4 Mult- C4 Muttifidus, Lng Coll – Longus Colli, Lng Cap – Longus Capitus.  Image adapted from Siegmund et al., 2007 [24] and used with permission from the American Society of Mechanical Engineers (ASME).  12 The role that muscles play in the cervical spine during head-first impacts is still unclear and will be discussed below.  In general, how muscles affect the human neck response is still a topic of intense research.  Due to the highly over-constrained nature, overlapping lines of action, complicated muscle contraction patterns of both agonist and antagonist muscle groups, and wide difference in properties from passive to active contraction, there are certainly many discoveries still to be made.  1.4 Models to Study Cervical Spine Properties and Injuries  The cervical spine has been studied using many different models and the suitability of a particular model largely depends upon what the researcher is studying.  In order to study non- injurious motions and characteristics of the cervical spine, volunteer testing is widely performed.  These studies are used to quantify measures such as strength of the neck, the relative contribution of individual muscles to given moments, the pressure in the cervical intervertebral discs, range of motion in different directions, and even kinematic patterns of inertial loadings that are below the tolerance to injury.  For obvious ethical reasons, studies of injurious scenarios must be conducted with other types of models.  This is the central challenge to the study of injury biomechanics.  In order to study and prevent injuries, a combination of indirect approaches along with engineering judgment must be used. For compressive loading as occurs in head-first impacts, model types such as full cadavers, cadaveric head and neck specimens, isolated but full cadaveric necks, cadaveric segments of the neck, finite element head-neck models, rigid multi-body dynamics models, anthropometric test devices (ATDs) and other mathematical representations have all been used to answer a wide range of questions. Cadaveric models have the advantage of real anatomy and also with proper hydration and freezing, the key mechanical properties of unembalmed bone and ligament do not change significantly over time compared to just post-mortem.  Muscles however, are a “live” tissue that have an innervated passive and active contraction and cannot be included in cadaveric models and must instead be simulated in some fashion.  This is the main limitation of cadaveric models, although they are also very time consuming, expensive, difficult to obtain, and are of course only an approximation to a live human.  In addition, since there is wide biological variability among humans, cadaveric models require large sample sizes to achieve statistical significance.  Despite  13 their limitations they are a necessary part of injury biomechanics and in particular they are used to validate other types of physical models such as ATDs and computational models. Computational models such as rigid body and finite element models overcome some of the limitations of cadaveric models.  In particular they do not suffer from biologic variability, and while they may be time consuming to initially develop and validate, once completed a large number of simulations can be studied that simply would not be feasible with cadaveric models.  Given the complexity of the cervical spine and the large number of variables that affect its response, computational models can play a role.  Computational models can also include the effects of musculature and in general it is much easier to simulate muscles computationally than physically as both passive and active nonlinear stiffnesses can be incorporated.  The central challenge to computational models of human injury is validation. ATDs, also called dummies, are another approximation to the live human.  Simulating the impact response of a complex biological system with a robust mechanical equivalent presents a design challenge and usually means that many different dummies, or components of dummies, are required for specific loading scenarios.  The word biofidelity refers to the concordance of a dummy response to the expected human response in a given impact and is usually defined kinematically.  For instance, the most widely used crash test dummy, the Hybrid III, was originally developed by General Motors to study high speed frontal and rear-end collisions [25].  Its neck stiffness was designed so that the head kinematics in a frontal impact were within corridors based upon earlier volunteer and cadaveric testing [26].  While the Hybrid III neck response is adequate for its intended purpose, it is not biofidelic for simulating low-speed rear- end collisions or side impact collisions.  Specialized necks have been designed for low speed rear impacts called BioRID [27, 28] to evaluate passive safety systems such as active seatbacks designed to prevent whiplash injuries.  Similarly, side impact dummies have been designed to help develop side air bags as well as improved door design to provide increased protection [29]. The principal disadvantage that ATDs have is the degree of their biofidelity.  Because they need to be robust enough to avoid damage in sequential testing, they are made of various alloys and rubber elements and can develop forces and moments far higher than would be present in humans.  Thus before an ATD can be used as a predictive tool of injury, Injury Assessment Reference Values (IARVs) [30] must be determined from subjecting the ATD to an impact of severity equivalent to that which would create injury in humans.  Thus if the human  14 tolerance to injury has already been determined in terms of an energy equivalent quantity, such as impact velocity, then the IARV for the dummy can be determined, after which point it can be used as a predictive tool in the same loading modality.  If an ATD is used for a loading modality in which it was not intended, which often occurs out of necessity, or if IARVs have not yet been determined, then it can only be used in a comparative sense to assess the relative difference in severity between impacts.  While ATDs are not perfect, they play a vital role in the development and verification of passive safety systems such as those found in automobiles.  Computational models of the ATDs are used for early safety system development but ultimately the constructed safety device must be tested with physical ATDs.  1.5 Cervical Spine:  Mechanical Properties  The cervical spine, from a mechanical perspective, is a segmented and curved beam- column.  It is an extremely complex structure due its anatomy, coupled motions, and the fact that it is comprised of viscoelastic materials with non-linear stiffnesses.  Adding to the complexity are the great number of muscles acting on the neck and connecting to elsewhere on the column itself or on the head.  1.5.1 Nonlinear Stiffness and Neutral Zone  All of the joints throughout the cervical spine have lax connections to each other over small displacements in all available degrees of freedom [31].  On a force vs. displacement, or moment vs. rotation graph, this region of near infinite compliance is called the “neutral zone”.  This nonlinearity is very common in biological materials and was first named in the spinal context when studying the mechanical properties of lumbar functional spinal units [32].  Soft tissues also exhibit viscoelastic characteristics such as rate-sensitive stiffnesses, hysteresis on unloading, and creep/relaxation effects.  In cadaveric testing, it has been observed that these viscoleastic effects are minimized after a joint or tissue is loaded and unloaded several times at which point it becomes “preconditioned”. The aligned cervical spine as a structure has a nonlinear force deformation curve when loaded dynamically in axial compression.  The nonlinearity has the characteristics of a very low initial stiffness, sometimes referred to as the “toe region” which is attributed to compression of the intervertebral discs.  After the discs are compressed, the spinal stiffness increases  15 significantly as the bony vertebrae are compressed.  A study by Pintar et al. [33] characterized this force deformation response using 20 cadaveric head and neck specimens that were compressed to failure at loading rates of 2.5 m/s to 8 m/s using a materials testing machine.  There was wide variability in the failure compressions, forces, stiffnesses, and injuries but the tests were used to produce a corridor for axial response that is shown in Figure 1-9.  Despite the variability, the mean response shows that the toe region has a stiffness near 33 N/mm up until a mean compression of 12 mm, after which the mean linear stiffness became 555 ± 333 N/mm.  The average failure compression was 18 ± 3 mm and occurred with a mean compression force of 3340 ± 1387 N.  Figure 1-9:  Force deformation characterization of the cadaveric cervical spine Nonlinear dynamic compressive force- deformation response of the aligned osseoligamentous cadaveric cervical spine from 20 specimens.  The dashed line shows the mean stiffness while the solid lines show the response corridor.  Graph produced using the data presented in Pintar et al., 1995 [33].   1.5.2 Head­Neck­Torso System   The cervical spine is part of a system of connected bodies namely the head and torso.  When loaded in a head-first impact, the head and neck are initially out-of-phase with each other.  A consequence of this is that when force is developing at the head, it will not initially be transmitted through the neck, until after a period of initial near zero-stiffness deformation occurs.   16 Impact experiments show that it takes approximately 2 ms from when the head makes contact with a rigid impact surface, until force develops at the lower-neck for impact speeds near to 3 m/s [34-36]. While it is unknown at this point, this time lag is likely rate-dependent. The degree that the head and neck axial load development are in-phase is affected by the compliance of the impact surface.  A plot of two cadaveric cervical head and neck specimen drop tests against an angled platform which moved the initial point of head contact anterior to the vertex is shown below in Figure 1-10.  These plots show that the impact platen force (head force) has an overall shape that is bimodal, but against the rigid surface, the two modes are much more distinct from each other compared to the padded impact surface.  Against the rigid surface, the 1st mode of head loading occurs without any force developing at the lower-neck until late in the 1st mode.  Against the padded surface, there is still a lag in the neck force development compared to the head force but the 1st mode of head loading occurs over a longer duration, associated with the compression of the padding, such that the neck loading becomes much more in-phase with the 1st mode of head force.  This lag is attributed to the low initial stiffness of intervertebral discs which in bending is referred to as the neutral zone [37].  The posture of the spine, which will be discussed shortly, affects the degree of coupling between the head and torso by stiffening the neck [38].  One can imagine the extreme where if the neck was rigid the head and torso could be considered a single body.   Figure 1-10:  Head and lower-neck axial forces for cadaveric head and neck drop tests. The temporal axial head and lower-neck forces from cadaveric head and neck specimen drop tests at 3.1 m/s impact speeds with a 15 degree angled impact platform moving the point of contact anterior to the head vertex.  On the left shows an impact against a rigid surface and the right shows against a padded surface. Images adapted from Nightingale et al., 1997[39] and used with permission from Wolters Kluwer Health.  17 1.5.3 Column Theory, Buckling, and the Cervical Spine   In the loading response of columns, the critical load is defined as the maximum load that the column can carry before it rapidly changes configurations to a lower energy state, i.e. buckling.  Variables that influence this critical load are the end conditions of the column, the length of the column, its bending stiffness, and the eccentricity of the load.  It is noteworthy that while buckling is normally considered a structural failure, it does not always involve material failure, and as this will be discussed shortly, the phenomenon has been observed in experimental head-first impacts with in vitro cervical spine specimens [34].  It was shown that the critical load of the osseoligamentous cadaveric cervical spine (without musculature) buckled under only 10.5 N of force, which is about one fifth to one quarter the weight of the average human head [40].  One experimental method of providing further stabilization of the cadaveric cervical spine, or more specifically to increase its critical buckling load, is the compressive follower load concept [41].  The central idea behind this strategy is to apply a compressive load that is guided near the approximate center of rotation [42, 43] at each vertebral level such that the compression is applied in a manner that minimizes any shear forces or moments which would act to bend or buckle the cervical spine. A follower load is thus a specific type of preload and is distinguished from the general term “preload” by the virtue of its application through the approximate centers of rotation.  The follower load method has been shown to achieve gains in stability with in vitro cervical spine specimens as evidenced by preventing buckling under 250 N of applied compression, which is almost 6 times the weight of the average human head, and 25 times the critical load observed without a follower load [41]. While a significant improvement in stability, this load magnitude of 250 N was modest compared to loads predicted to exist in vivo [44] which presumably do not cause buckling due to the increased bending resistance of the column provided by active and passive musculature.  The increased critical buckling force from a follower load observed in cadaveric specimens is somewhat counterintuitive to engineering column theory in that normally, if one preloaded a column, that column could carry less additional load before it would buckle.  This discrepancy from simple column theory is likely related to the nonlinear axial and bending stiffness, as well as segmented nature, of the cervical spine.  The follower load acts to compress the spine through its more compliant toe region such that the stiffness increases through further deformation.  18 The in vivo cervical spine carries approximately 75 to 150 N of compression in static postures as calculated through a combination of experimental disc pressure measurements and disc cross sectional area [16, 45, 46].  Neutral posture was associated with the lowest disc pressures while extension postures produced the highest pressures [45].  A mathematical model of the muscles crossing the C4 level showed that the cervical musculature can create compressive spine loads as high as 1164 N [44] during an isometric extension contraction to resist a mean voluntary moment of 29.7 Nm.  It is also evident that the in vivo cervical spine can carry loads far exceeding this from anecdotal accounts such as observing dancers and acrobats perform headstands (with no hands) thus forcing their neck to withstand nearly the full mass of their bodies.  In addition studies have been conducted with African and Nepalese porters who carry loads either directly on their heads or with head-supported namlo straps who supported 70% - 183% of their body weight on their heads [47, 48] and thus subjected their necks to far higher compressions than the 10.5 N critical load measured by Panjabi [40].  The Duke University cadaveric model of head-first impact using full cervical spine and head specimens impacting at various angled and padded surfaces showed that the cervical spine can buckle in response to axial impact.  In some drops, transient 2nd order buckling modes preceded injury and a 1st order extension buckling mode [34].  It is important to note that the cadaveric drops did not simulate the stabilizing effect of musculature.  In order to further study this observed buckling, the same authors created a sagittal plane rigid multibody dynamics model using MADYMO software consisting of a mass representing the head and seven ellipsoids as vertebrae [49].  Adjacent vertebrae were connected via rotational springs having non-linear directional bending stiffnesses that were based upon those measured from cadaveric flexibility testing [50].  No contact parameters were assigned such that at extreme motions the vertebrae could rotate “through” each other, but up until this point the model could offer useful phenomenological insights into dynamic buckling.  The phenomenological model was parametric to look at the effects of vertebral mass, segmental bending stiffness, mass moment of inertia, and loading rate on buckling.  The baseline model parameters were based upon the cadaveric osseoligamentous spines average values in their previous studies.  The vertebral mass and bending stiffness were then scaled at 0.125, 0.25, 0.5, 1, 2, 4, and 8 times the baseline.  The vertebral inertia was scaled at 0.1, 1, and 10 times the baseline and the loading rate was adjusted  19 (with constant baseline parameters) by ramping the applied acceleration to the head over a timeframe 0 to 10 ms such that a constant load of 4000N was applied.  Under quasi-static loading the baseline model buckled at less than 8 N which was in good agreement with the critical load determined experimentally by Panjabi [40] and in dynamic loading the baseline model recreated the observed buckling patterns observed experimentally.  The parametric model showed that vertebral mass and intervertebral stiffness had the largest effect on buckling patterns.  Increasing vertebral mass and decreasing stiffness promoted 2nd order buckling modes at a given loading rate.  Of interest was that when the intevertebral stiffnesses were scaled to a factor of 8 from the baseline, the critical load of the column became greater than 4000 N and no buckling (1st or 2nd order) was observed. It remains unknown whether or not the in vivo human cervical spine buckles under dynamic compressive loading.  As described earlier, it can be seen that the cervical musculature plays a great role in stabilizing the neck, i.e. axial loading without buckling, in isometric loading situations.  It has also been observed in other human joints that muscles increase the rotational stiffness across a joint by more than an order of magnitude from passive to active contraction [51].  While it still remains to be determined in the neck, in the case where an athlete has full neck muscle contraction such as during a spear-tackle in football, it stands to reason that the active musculature could stiffen the cervical spine by more than a factor of 8, as in the multibody dynamics model, and avoid buckling altogether.  1.6 Cervical Spine Injuries  The cervical spine, due to its complexity and the number of different tissues, can sustain many different injuries from multiple injury mechanisms.  Almost all neck injuries are caused from a force delivered to the head, whether from indirect (inertial) or direct contact.  All of the articulating surfaces in the spine can be injured.  Intervertebral discs, spinal ligaments and tendons, along with bony vertebrae can sustain a wide array of injuries some of which cause neurological damage.  This thesis is concerned with a limited but important loading condition and the reader is cautioned that the injuries presented here represent only a small portion of the overall spectrum of cervical spine injuries and relate to those associated with head-first impacts or more specifically an inferiorly directed force delivered to the crown of the head (helmet) when the head is flexed forward such that the spine is aligned.  This injury mechanism has been shown  20 to be the leading cause of catastrophic neck injuries causing permanent paralysis in both youth and professional sport [52].  1.6.1 Spinal Column Injuries  Spinal column disruptions fall into three broad categories: fractures, dislocations, and fracture-dislocations.  A fracture is defined as any deformation or damage to a bony structure (vertebra).  A dislocation involves no bony damage but instead one or more vertebrae are displaced such that there is a change in the normal anatomical alignment between vertebrae.  A partial dislocation is referred to as a subluxation.  A fracture-dislocation involves both bony fracture and displacement of one of more vertebrae from their normal anatomical position.  These injuries are further classified as either stable or unstable depending on if the damage to the structure of the column will allow for further neurological impairment without intervention [53]. It is important to note that SCI do not occur in all spinal column injuries, even those from aligned column axial impacts. It is not necessary for this thesis to include a full classification of the myriad of different injuries that can occur to the cervical spine but rather to briefly present some common injuries that occur in head-first impacts and classification schemes used by clinicians and terminology as used in this thesis. Upper Cervical Spinal Column Compressive Injuries  In head-first impacts the C1 and C2 vertebrae can be fractured in one or multiple places.  The C1 vertebra can be fractured in compression as the occipital condyles, which articulate with the lateral masses of C1, cause the ring-shaped vertebra to burst outwards radially.  This is called a Jefferson fracture which is further classified into three types depending on where the fracture occurs which are shown in Figure 1-11.  The classic 4-part burst fracture that was originally diagnosed by Jefferson is a Type II fracture.    21  Figure 1-11:  Jefferson fractures of the atlas Jefferson Fractures of the Atlas.  Left: Type I – fracture confined to a single arch and does not extend across the equator of the atlas.  Middle: Type II – fracture involved both arches and crossed the equator.  There could be two or more fragments.  Right: Type III – fracture primarily to a lateral mass with fracture extending onto one arch only[54].  Image adapted from Landells and Van Peteghem, 1988 [54] and used with permission from Wolters Kluwer Health.   The C2 or axis vertebra is also vulnerable to fracture.  The odontoid process, or dens, is the most common fracture site of C2 although the mechanism is generally agreed to be hyperextension or hyperflexion [55].  Although the odontoid is not commonly injured in head- first impacts, compressive forces that also cause hyper-physiologic head rotation can cause odontoid fractures sometimes in conjunction with Jefferson fractures or lower cervical injuries. The ligaments that connect the dens to the skull are stronger than the bony dens structure and thus when the head is subjected to extreme motions, the ligaments can cause varying fractures to the odontoid [54].  The most commonly used classification for odontoid fractures describes three main types [56] that are shown in Figure 1-12 however three subtypes (of the previous type II) were later added for better clarification and improved treatment [55].   Figure 1-12:  C2 odontoid fracture classification The most widely used odontoid fracture classification as originally defined by Anderson and D’Alonzo[56].  Image adapted from Grauer et al.2005 [55] and used with permission from Elsevier Limited.   22 Aside from odontoid fractures, a wide array of fractures to the ring of the axis can also occur [57-60].  A bi-lateral fracture to the pedicles of the axis is termed traumatic spondylolisthesis of the axis [59].  Historically this was referred to as a “Hangman’s Fracture” as the bony insult is similar to that observed in judicial hangings that apply a rapid distraction (tension) of the spine with extension moment although the mechanism of injury is different.  This fracture can occur in isolation or in conjunction with other fractures.  A classification scheme of ring fractures to the axis vertebra has been developed (Figure 1-13).  The Effendi classification scheme for ring fractures of the axis classifies three fracture types based upon the degree of dislocation of the C2 vertebral body as well as the anatomical positioning of the C2-C3 facet joint [58].  Type 1 is a stable fracture characterized by minimal displacement of C2 with C2-C3 disc space normal and stable. Type II fractures are unstable and characterized by displacement of the anterior fragment of the C2 vertebral body in extension, flexion, or anterior translation but the C2-C3 facet joints are in correct anatomic position.  Type III is characterized by the anterior portion of the body in flexion and in addition the C2-C3 facet joint dislocated.  These injuries have been reported to occur in motor vehicle collisions and often accompany facial lesions suggesting a hyperextension mechanism or extension in conjunction with a compressive vector [58].   Figure 1-13:  C2 ring fracture classification The Effendi [58] classification scheme for fractures of the ring of the axis. The images show a portion of the occiput along with the C1, C2, and C3 vertebrae.  Image reproduced from Effendi et al., 1981 [58] and used with permission and copyright © from the British Editorial Society of Bone and Joint Surgery.    23 Lower Cervical Spinal Column Compression Injuries  One of the more detailed classification schemes for indirect lower cervical spine (C3-C7) injuries was developed by Allen et al.[61].  This method of classification was developed in hopes of universal adoption such that researchers and clinicians alike could communicate their research more easily.  The system was developed retrospectively based on 165 observed clinical observations.  Ideally for a case to be considered, the posture of the head and neck at time of injury, the direction and location of force, and a detailed description of the accident had to be available.  However, all of this information was not always available for each incident.  Injuries with similar radiographic evidence of injury were grouped, and then arranged into “a continuous spectrum of anatomic damage”.  For cases where multiple injuries occurred, the more severe injury was classified as the major occurring and the lesser(s) as associated injuries.  From here, mechanisms of injury were postulated, and then these mechanisms of injury were called phylogenies where each phylogeny is named according to the presumed cervical spine posture at the time of failure and the predominant mode of failure.  These are: compressive flexion (CF), vertical compression (VC), distractive flexion (DF), compressive extension (CE), distractive extension (DE), and lateral flexion (LF).  Their system is thus planar with the first five groups occurring in the mid-sagittal plane and the last in the coronal plane.  They did not find axial rotation to be a predominant mode of failure and deemed it to be an associated force which acts to “lateralize” the other groups. They chose to specify the location of injury in terms of a functional spinal unit, for example instead of specifying that only C4 was fractured, they would say that the injury occurred at C4,5  with the underlined “4” indicating the fracture.  For each group, three to five sub-levels, with a higher number being more severe, were identified.  For the purposes of this thesis it is not necessary to detail the entire classification method.  The central idea behind the classification is that at any given level, the combined loading consisting of axial force, shear force(s), and moment(s) can be represented as single force acting at some eccentricity to produce the given injury. For the loadings presented in this thesis only compressive-extension, vertical compression, and compressive-flexion are required.  These classifications can be thought of as having a vertical component of force but acting at differing eccentricities relative to the column anatomy and thus giving rise to a range of different injuries as shown in Figure 1-14.  As described by Myers et al. [62], when the resolved compressive force acts posteriorly to the  24 vertebral bodies, posterior element fractures of the spinous process and anterior soft injuries such as anterior longitudinal ligament rupture and vertebral disc ruptures are produced.  Compression fractures occur when the force acts through the vertebral bodies. Compression-flexion injuries such as burst fractures, wedge fractures, and facet dislocations occur with increasing anterior eccentricity.  These are only a small subset of the types of fractures that can be produced but demonstrate the classification method.  An important implication of this classification is that a given external loading can produce different classifications at different spinal levels.  This is not a limitation of the method or paradoxical but is rather explained by the curvature in the cervical spine.  Depending on the location of the force and posture of the spine, different levels of the spine experience different magnitudes of bending moments and in some cases in opposing directions.  It has been suggested that buckling in the cervical column is responsible for compression-flexion injuries at some levels with compression-extension at others [63].  1.6.2 Spinal Cord Injuries (SCI)  Injuries to the spinal cord occur in some but not all spinal column injuries.  Due to proximity of the spinal cord within the spinal canal, virtually all insults to the spinal cord are caused by some non-physiological interaction between the spinal column and the cord in response to an applied external force.  An exception to this would be SCI occurring through violence such as gunshots or stabbing.  However, spinal cord injury can occur without  Figure 1-14:  Compressive cervical spine injury classification based on eccentricity  A range of compression-extension, compression, and compression-flexion injuries with increasing anterior eccentricity.  Image obtained from Winkelstein and Myers, 1997 [62] and used with permission from Wolter Kluwer Health.   25 radiological abnormalities (SCIWORA) to the spinal column [64].  Approximately 45% of SCI occur with accompanying vertebral dislocation, with and without fracture, burst fractures account for 30%, other compressive fractures 10%, and SCIWORA account for 15% [1].  In addition, vertebral dislocations are most correlated with complete SCI causing full loss of sensation below the lesion [1].  Four characteristic mechanisms of SCI are: impact with persistent compression, impact alone with transient compression, distraction, and laceration/transection [65].  In head-first axial impacts, which are the focus of this thesis, the first two of these are predominant.  These four mechanisms use the term compression to mean any non-tensile strain but it has since been shown that the spinal cord develops distinct lesions from tension, shear, and contusion [66].  Shear injuries are most often caused by facet dislocations or other types of fracture-dislocations where the displaced vertebra, or portion of a fractured vertebra, “pinches” the spinal cord.  Contusion injuries can occur from burst fractures of the vertebral bodies which can drive bone fragments posteriorly into the spinal cord.  Tensile injuries typically occur when the cord is stretched due to a localized flexion-distraction or flexion- extension column injury although tethering of the spinal cord along the column is thought to affect tensile cord strains as well [66].  1.6.3 Factors Affecting Injury and SCI in Head­First Impacts  Much research has been done to study cervical spine injuries that occur as a result of head-first impact [34, 35, 39, 67-75].  Due to the complex nature of the cervical spine structure, the array of injuries that can develop from a head-first impact is large.  It has been shown that there are many factors that influence the development and degree of injuries. Impact Velocity  McElhaney et al. [69] showed that axial compressive injuries can occur at surprisingly low impact velocities of 3.1 m/s by reconstructing several diving injuries.  This is equivalent to a free fall from a height of 0.5 meters.  A review study that combined the results of several full cadaver impact tests used the inferior head velocity relative to the torso as an indicator of injury and found that the average head velocity (after incorporating restitution) was 6.32 ± 1.29 m/s for serious neck injuries and 3.75 ± 2.17 m/s for non-injurious impacts [76].  The difference between  26 head velocity and impact velocity in drop tests is due to the head rebounding such that peak head velocity was 22% higher than impact velocity in the drop tests against barriers of infinite mass. It is clear from the observed sports and activities where these injuries occur that the potential for much larger impact speeds does exist.  Although focused on concussion, 182 helmet to helmet impacts occurring in professional football were reconstructed and it was found that that the average closing speeds for concussion-producing tackles was 9.3 ± 1.9 m/s [77].  The closing speed is not indicative of the change in speed (a measure of impact severity) which depends on the mass ratio of the striking athlete vs. the struck athlete (or inanimate object) and the coefficient of restitution. Posture at Impact – Observational  The posture that the cervical spine assumes during impact will strongly affect the way it responds to a compressive load.  In some of the earliest full cadaveric head-first drop tests it was observed that if the head and cervical spine was in the mid-sagittal plane that the response of the neck will largely stay in this plane.  However, the presence of lateral bending and axial rotation increases the risk of fracture and out of plane loading [78]. A study that reviewed 60 football impacts that produced neurological injuries and were captured by television cameras was able to reconstruct the injury mechanism in 51 of the impacts.  They found that axial loading with the head partially flexed was responsible for every single instance [52] and is shown in Figure 1-15.  Throughout this thesis, the lordosis-removed posture in the right hand image of Figure 1-15 will be referred to as an “aligned” posture or an “aligned column”. This posture in many researchers’ opinions represents a worst case scenario for the development of compressive injuries in axial compressive injuries [79].    27  Figure 1-15: Neutral and aligned head and neck postures The neutral posture (left) in most individuals displays a lordotic curvature.  When the head is flexed approximately 30 degrees (right) the neck lordosis is removed and the cervical spine posture is aligned.  Images adapted from Torg et. al, 1990 [52] and used with permission from Sage Publications.   It is also noteworthy that while this posture is the most dangerous one for neck injury to the striking player if contacting another player or inanimate object with large mass (or momentum), it is also the most dangerous in terms of concussion to the struck player in a head-on-head collision [38].  When the striking player assumes the aligned head-neck-torso posture the neck is in its stiffest possible posture which maximizes the engagement between the striking player’s torso and head through the neck.  This effectively increases the momentum of the striking player causing a more severe head impact to the struck player [38]. The aligned posture has also been the focus of many laboratory investigations.  McElhaney performed constant velocity load to failure tests with cadaveric cervical spine specimens [80]. His specimens were ‘straightened’ to remove the natural lordosis.  He noted that with zero eccentricity (i.e. the centre of C1 aligned over T1) compression injuries were produced but that by moving the base of the specimen either 1 cm anteriorly or posteriorly the classic flexion or extension type injuries were produced.  Alternatively stated; a bending pattern response would occur in either the anterior (flexion) or posterior (extension) directions as opposed to an axial buckling response with only 1 cm of eccentricity. Very similar findings were also reported by Pintar and Maiman [79, 81]. In a manuscript describing the outcome of 30 cadaveric head and neck specimens where eccentricity was varied prior to compressive loads being applied, the  28 mean eccentricities for compression-extension, compression, compression-flexion, and hyperflexion type injuries were -0.5, 0.1, 2.3, and 5.3 cm respectively.  When they categorized the injuries into fracture and non-fracture groups the mean eccentricities were 0.6 and 5.2 cm respectively.  In this study positive eccentricity was defined such that the center of the occipital condyles was anterior to the center of the vertebral body of C7. Alignment of the Load with the Crown of the Helmet and Axis of the Spine  In one of the earliest studies that re-created known injury scenarios for axial compressive injuries, 3 reported situations where football players suffered paralysis following a head-first impact with a football tackling block were reconstructed [82].  The tackling block was a spring- loaded device commonly used for linesmen to practice on.  In their tests, they used a helmeted GM Hybrid III dummy and subjected it to a range of impacts thought to have caused injuries in these young football players.  Among their findings was that it was absolutely necessary for the tackling block to hit the very crown of the helmet for significant loads to result.  They reported that in one impact where the tackling block made contact in a location slightly off the crown of the helmet, that the peak axial neck loads were reduced by 76%, the superior-inferior head accelerations reduced by 92%, and the peak moment measured at the atlanto-occipital joint decreased by 48%.  The same effect was observed some 2 years later in a study using helmeted full cadavers [72]. End Conditions Matter  In one of the earliest studies looking at neck injuries from head-first impacts, cadavers were fitted with football helmets and axial impacts delivered to the top of the head with a pendulum.  The authors were unable to experimentally produce axial compressive injuries without the addition of a full constraint at the atlanto-occipital joint as without this the heads would simply deflect, the neck bending out of the way without significant neck load developing [72].  This study, like all cadaveric studies, lacked the stabilization offered by active musculature but it highlighted the flexibility of the  cervical column and the difficulty of experimentally loading it in axial compression. Another early study that highlighted the effect of constraint, although using a very artificial method inserted a metal rod into the vertebral canal and then loaded cadaveric cervical spine specimens in compression with a pneumatic mechanical press.  This study produced bi-lateral  29 facet dislocations at the level where the rod terminated and created a stress riser due to the abrupt change in stiffness [83].  McElhaney loaded full cadaveric cervical spines eccentrically to produce a combined compression bending loading [84].  His work showed that the bending stiffness of the cervical spine was also sensitive to the constraint.  In particular, when moving from a pinned-pinned to a fixed-pinned set of end conditions, the bending stiffness increased by more than a factor of 8 whereas classical beam theory (using a non-segmented beam-column) predicts that it should double.  This was attributed to the fact that the pinned end of the specimens were constrained to move along the vertical direction and this combined with the fixed end to create a shear force which acted over a long moment arm and stiffened the bending response [84]. A compelling study performed by Nightingale et al. specifically looked at the effects of end conditions on the cervical spine and how they affected its response to axial forces [71]. Cadaveric cervical spine specimens were mounted on three specially built test apparatuses to test three different end conditions: unconstrained, rotational-constrained, and fully-constrained. A linear actuator applied an axial displacement quasi-statically to the spines while a load cell measured the forces and moments that developed due to the applied displacement. The rotational and full constraint spine specimens had average axial stiffnesses that were 8.5 and 12.2 times as stiff as the unconstrained spines. The increased stiffness prevented the spines from escaping the applied load and injuries developed. For each constraint pattern 6 spine specimens were tested. All 6 unconstrained spines escaped injury while all 6 rotation-constrained specimens sustained bilateral facet dislocations and all 6 fully-constrained specimens developed compression type injuries [71]. The importance of the inertial end-conditions on the neck imposed by the head has also been shown in drop test experiments with cadaveric head and cervical spine specimens [34].  The specimens were impacted onto various angled and padded surfaces and all specimens that were subjected to rigid posterior-to-vertex impacts escaped injury due to rapid forward translation of the head upon impact. Another important finding was that the human head when stopped by a perpendicular, rigid, and frictionless impact surface, has sufficient inertia to provide a constraining end condition sufficient for neck injury development as opposed to the neck doing work on the head to move it out of the way. These studies indicated that increased constraint on the head prevented energy dissipation through displacement and rotation of the spine.   30 Surface Conditions, Friction and Compliance (padding)   The effect of compliant padding on the impact surface was also studied using the above mentioned Duke University model of cadaveric head-first impact [34] which found that the padding restricted the head from escaping and provided a more constraining end condition which lead to neck injury.  It was concluded that this “pocketing” of the head in the padding was the risk factor.  Later, Camacho [85, 86] performed finite element simulations validated against these experimental drops and showed that increasing friction in a near vertex head-first impact does increase the risk of neck injury development by dissuading the head from moving along the impact surface during impact.  The compliance of the padding did not in fact promote injury, thus invalidating the “pocketing theory” as long as the padded surface had low friction [86].  This was a numerical simulation and it is unclear if there exists such a padding that can maintain low tangential friction while under high compression.  Even with a low friction, once the head compresses a volume of padding, there is an elastic resistance to tangential motion as additional padding must be compressed to allow tangential motion. Cervical Muscular Contraction  The literature is unclear with respect to the contribution of the effect of muscles on neck response in head-first impacts.  On the one hand, even some of the earliest full cadaver [75] and cadaveric cervical spine specimen tests [74] have attempted to include the stabilizing effect of musculature in a rudimentary way.  However, other researchers have concluded that because the timeframe (after first head contact) in which injuries occur (2-19 ms) [34] is far less than that required to actively recruit cervical muscle forces (54-92 ms) [87], that the effect of muscles are minimized in near-vertex impacts.  As it was discussed in section 1.5.3 regarding buckling, muscle contraction could be present prior to impact.  If the difference between the isometric strength of the neck at full contraction [24] versus its passive resistance [88] can be interpreted as an increased bending stiffness, then under muscular contraction this increased bending stiffness should also act to increase the cervical spine’s critical load and buckling response.  Although it is unproven at this point, it seems that full muscular contraction may be a risk factor for injury development in head-first impacts as it could act similarly to a constraint.  Both in the cervical spine [44] and the lumbar spine [89, 90], it has been shown that muscle contractions and co- contractions increase the axial load through the spine in addition to resisting applied bending  31 loads.  It seems logical that in a head-first impact, the muscles could increase the risk for injury development by increasing the internal load even prior to impact, but also by the stiffening argument affecting buckling above.  A recent cervical spine finite-element simulation that included muscles looked at both active and passive muscle contractions and concluded that full muscular contraction doubled the risk of fracture development over passive contraction as indicated by maximum principal strain in the vertebrae [91].  Passive muscular contraction increased fracture risk only slightly over no contraction. It seems logical that the same stability offered by the cervical musculature that acts to increase the static load carrying capacity could hold the spine in a more stabilized posture when subject to a dynamic loading from the massive torso and that this could aggravate the risk for neck injury. This topic clearly demands further study. 1.7 Helmets and Head Injury  Helmets have historically been designed to protect the head against two types of head injury that occur in impacts. The first is soft tissue (brain) injury caused by high linear accelerations and the second is localized skull fractures from contacting pointed objects. These two injury modes require different approaches and demand that the helmet designer determine a suitable compromise between energy absorption and penetration tolerance. To protect against object penetration, a lightweight shell of high stiffness is used. To protect against brain injury, high density padding of much lower stiffness is used. In theory, if one were not concerned with penetration, there would be no outer shell (as in some recreational bicycle helmets) as only the softer foam is necessary to protect against linear brain injuries.  There is a relatively new type of helmet technology called the MIPS (Multi Impact Protection System, MIPS AB, Stockholm Sweden) which adds a level of protection against angular acceleration injuries which are known to cause traumatic and mild traumatic brain injury [92].  This technology is being offered in ski and snowboard helmets among other sports and involves a breakaway liner between the head and outer shell to dissipate any angular impulse occurring as a result of oblique helmet impact. 1.8 Helmets and Neck Injuries  Contemporary helmets have not been designed to prevent or protect against neck injuries.  Manufacturers make this claim on every helmet presumably because they have been litigated against continually for at least the last two decades.  The football helmet market has far fewer manufacturers than it did at one point due to litigation-related bankruptcies [8].  It has been  32 reported based on a mathematical model that crash helmets can aggravate neck injuries due to the extra weight on the head which then adds to the loading the neck experiences from indirect head loading situations such as whiplash [93].  A similar finding was reported based upon a retrospective study looking at basal skull fractures in motorcycle collisions where it was concluded that a helmet weight beyond 1500 grams was associated with increased basal ring fractures [94].  However, a study looking at pediatric helmet use in snow-sports where helmets are much lighter found no significant difference in neck injuries between helmeted and unhelmeted participants [95].  1.9 Prior Preventative Strategies   Contemporary approaches to neck injury protection in sports where helmets are worn do not adequately protect the user against axial loading vectors. The devices on the market, and described in patents, can be subdivided into collars that do and do not attempt to couple the helmet to the shoulders. The central idea behind coupling the helmet to the shoulders is to create an alternate load path between the helmet and torso in parallel with the neck to attempt to shield force away from the neck onto the shoulders.  Some of the simpler non-coupling collars are made of foam and aim to protect the user against injuries associated with extreme neck motion and make no claims to protect against axial loads.  The more advanced non-coupling collars for the motorsports disciplines are made of materials such as carbon fiber and polycarbonates that are also designed to limit the maximum head motion and decelerate the head motion in a controlled manner and even to be frangible in severe impacts and thus absorb energy that would otherwise be transferred into the neck.  The most commercial of these in the so-called action sports is called the Leatt Brace™ which is marketed towards motocross and mountain biking disciplines (Figure 1-16).  33  Figure 1-16:  The Leatt Brace™ neck brace for mountain biking and motorcycle markets The Leatt Brace rides on the user’s shoulder girdle and interfaces with the helmet to prevent extreme flexion, extension, and lateral bending motions.  Images obtained from www.leatt-  There is no peer-reviewed research on this device that we are aware of although self- published results [96] show a decreased range of motion compared to tests without the brace using an inverted Hybrid III dummy attached to a pendulum subjected to flexion and extension loading.  The Leatt Brace™ claims to protect against combined axial loading with bending once the helmet contacts the brace by coupling the helmet directly to the shoulders.  However, it seems clear that their device could not offer protection to a purely compressive loading vector as the brace sits approximately 50 mm below the underside of the helmet to allow for mobility (Figure 1-16).  This is more than double the average failure compression of the cervical spine (18 mm) when in an aligned posture [33]. Similar collars have been used in football for some time and there have been published studies on their effectiveness at limiting hyperextension and lateral flexion of the helmeted head in conjunction with football shoulder pads [97-100].  The earliest of these [98] did not report any loads transmitted (although they were measured to ensure consistent trials in the repeated measure design) but rather compared the range of motion of a simple foam collar to the Cowboy Collar™ and also to a custom fit orthotic.  Five football players applied either extension or lateral flexion moments to their own helmets via an adjustable pulley mechanism with: helmet only, helmet and shoulder pads, and helmet, shoulder pads and one of three respective collars.  Their results showed that none of the collars produced a significant reduction in lateral flexion motion but that all three collars reduced maximum hyperextension by 32% to 48% where the  34 foam collar and Cowboy Collar were not significantly different from each other and the custom orthotic provided the maximum reduction [98].  A second study produced almost identical results but applied the loads by the hands of a researcher using a pressure transducer to quantify the forces [97]. Only one study of football collars that we are aware of applied axial loading in addition to extension and lateral bending [99].  Axial forces were applied to the top of a football helmet clad Hybrid III dummy with a materials testing machine at 5 and 7 m/s while wearing three different football collars (Figure 1-17) with shoulder pads in a normal and “raised” position.  Figure 1-17:  Football neck collars Three popular football neck collars.  The Cowboy Collar (left), Bullock Collar (middle), and Kerr Collar (right).  Image adapted from Rowson et al., 2008 [99] and used with permission from Wolters Kluwer Health.  Of the three collars tested, two of them (the Kerr collar and the Bullock collar) provided force reductions that increased with increasing loading rate and which were also increased with the shoulder pads in the “raised” position.  There were some aspects about this study which were unclear.  In particular, the “control” group, which was the dummy wearing the shoulder pads but without a neck collar, also showed force reductions at the lower-neck compared to the input force from the materials testing machine.  In some cases the control group showed slightly larger force reductions than the least-performing collars, but in all cases the collar which was in contact with the helmet prior to load application produced the largest force reductions (the Kerr Collar); from a 15% reduction with normal shoulder pads and a 5 m/s impact up to 43% reduction with raised shoulder pads and the 7 m/s impact speed.  It is unclear whether or not a similar force reduction would occur with a more compliant human neck and torso.  As far as we are aware,  35 neither the human or the Hybrid III shoulder girdle stiffness has been characterized in the superior-inferior direction but we assume that the Hybrid III torso is stiffer than the human. Helmet-shoulder coupling neck braces are very similar to the collars described above.  In our opinion the defining difference is whether or not the collar or brace is in contact (or close proximity) with the underside of the helmet prior to impact.  A prototype model of one of these helmet shoulder coupling devices was recently published [101] and represents an embodiment of a patented device by one of the authors [102]; both are shown in Figure 1-18.   Figure 1-18:  Schematic of helmet-shoulder coupling device (left) and prototype version (right). Helmet-shoulder coupling device which has been patented[102] and a recent prototype version which demonstrated significant axial neck force reductions.  Image adapted from Engsberg et al., 2009 [101] and used with permission from Nature Publishing Group.  The prototype is quite crude in that could not be worn by an actual human.  The vest portion was made of fiberglass molded to the Hybrid III dummy and was reinforced with a steel member overtop of the head and onto the shoulders.  Unfortunately the authors did not state whether the axial neck loads were measured at the upper or lower-neck but the maximum axial neck force produced with the prototype device with an impact speed of 6 m/s was 275 N compared to neck forces near 11,000 N (data not tabulated in paper) with a 5.5 m/s impact speed and wearing a conventional motorcycle helmet without the device [101].  There are several limitations with this work, namely that the device was essentially rigid and also that it was rigidly mounted to the Hybrid III torso which we would expect is far less compliant than the human shoulder girdle.   36 Perhaps in some activities such as race car driving it would be possible for participants to wear a device already in contact with the shoulders, but in many of the sports the constricting nature would make it impractical.  Considering the human cervical spine fails at only 18 mm of compression when in an aligned posture [79], and also that the human shoulder girdle is far more compliant, it remains to be determined whether or not this strategy could be effective.  The only other study we are aware of that considered this helmet-shoulder coupling strategy was based upon a lumped parameter mathematical model and studied the characteristics required for a helmet with an air-bag in its base that deploys onto the shoulders.  In their simulations, they concluded that in order for such an airbag to reduce the force in the neck to below 3000 N, for a 4.6 m/s impact into a fixed barrier, assuming that the neck and device would have to fully arrest a 36 kg torso, that the airbag would have to develop 45 kN of force and deploy in 4 ms [103].  Their conclusion was that no such feasible device could be developed and thus that there is no engineered solution to preventing catastrophic spinal injuries from head-first impacts in collision sports.  This deployment time is on the same order as the fastest possible deployment for modern automotive airbags in the most severe collisions [104] and certainly would pose challenges.  It would seem that if the helmet-shoulder coupling strategy is to work that the helmet rim would need to be in contact with the device prior to impact. 1.10 Induced Head Motion in Axial Impacts  We are only aware of one published study that investigated the effect of purposely induced head motion during impact on neck loading [105].  This was a finite element study using a model of the cervical spine and skull validated against the Duke University drop testing [34] that evaluated an experimental automobile roof which would cause anterior translation upon impact by way of an asymmetric spring assembly.  It was unclear how this roof would have been constructed physically but in their computer simulation the asymmetric deformation of the springs were caused simply by decreasing the anterior spring stiffness relative to the posterior springs such that when the roof incurred a vertical load, the headliner would move anteriorly.  The simulation showed that lower axial forces resulted in the neck when anterior translation causing head flexion was induced upon contact.  Reductions in axial load at T1 of 44% were realized for a 15 degree oblique impact delivered posterior to the crown of the head when the head was deflected anteriorly upon impact. Similarly, a 27% reduction was realized for perpendicular vertex impacts [105].  37 Another study that looked at inducing head motion to protect against neck injuries in automotive rollovers studied a seat-mounted airbag that deployed and pushed the head forward into flexed posture [106].  The important distinction with this device was that the induced head motion was intended to occur well before vertical impact as in rollovers the collision occurs over a much longer duration than the actual axial loading of the neck.  This device was intended to move the head forward to increase the survival space for the occupant and also the airbag would act to soften the blow to the head. 1.11 Conclusions from Literature about Cervical Spine Injuries in Head­First Impact  1. The cervical spine cannot absorb the momentum of the torso without injury in an inverted fall except at very low impact speeds. 2. The cervical spine is difficult to load axially unless certain conditions of head-neck-torso alignment combined with head constraint exist.  Most impacts to the head cause the flexible cervical spine to bend without developing neck injury. 3. If the spine is in a lordosis-removed straightened posture which occurs when the head is flexed forward by approximately 15-30 degrees, the risk of compressive injuries is highest because the momentum of the torso must be managed predominantly by the bony vertebrae along the spine’s stiffest axis rather than dissipating energy into the musculature over larger rotations and displacements. 4. A very small amount of eccentricity, which is about half of the average depth of a typical vertebral body, changes the reaction pattern of the cervical bony structure from an axial response to a combined compressive-bending response. 5. The cervical spine has a low initial axial stiffness that rapidly increases.  In a head-first impact this has the effect of creating a bi-modal loading for the head where there is a lag between head and neck axial load development.  38 6. The cervical spine stiffens axially with increasing constraint upon the head.  The increased constraint dissuades the dissipation of energy into the paravertebral musculature through rotation and displacement of the head and neck and promotes instead the development of high strain energy in the bony vertebrae beyond their limits causing injury. 7. The head has sufficient inertia, once stopped in an axial head-first impact, to provide a constraining end condition on the cervical spine in a frictionless, perpendicular impact onto a rigid surface.  High friction and unfortunate impact angles and head-neck-torso alignments increase this constraining end condition further. 8. When the alignment and surface conditions align favourably to ensure a component of head impact velocity along the impact surface, injuries have been avoided completely at impact velocities known to cause injuries with other more head-constraining impact combinations.  In these cases the head avoided stopping immediately and this allowed the cervical spine to “escape” the incoming momentum of the torso. 9. Engineered environments such as automobile roofs seem to have the potential to provide neck injury protection 39 1.12 Research Questions and Objectives  The overarching objective of this doctoral dissertation is to address the question:  Can a helmet be used to induce head motion at impact in an aligned column head-first impact to prevent or mitigate the catastrophic cervical spine injuries typical of head-first impact without exacerbating or causing head or additional cervical spine injuries?  The approach taken required the development of mechanical head and neck surrogates and helmet prototypes and to conduct the first proof-of-concept testing of the helmet using the mechanical surrogates.  In addition, a test protocol using cadaver specimens and state of the art instrumentation was developed and characterized.  This work was encapsulated in the four specific objectives below that were addressed in four respective research chapters.  1. Design and construct a robust mechanical neck model that captures the time lag between head and neck axial load development as observed in the human head and neck in aligned column head-first impact and provide a realistic range of motion and bending stiffness in the sagittal plane. 2. Characterize the mechanical head and neck model response to changes in pre- impact posture (away from purely aligned), impact surface friction, impact surface stiffness, and changes in the incident angle of the impact surface in order to determine the worst-case axial impact scenarios for evaluation of the helmet. 3. Design, construct, and evaluate a mechanical helmet prototype of realistic size, inertia, and mass that induces head-motion upon impact for use with the mechanical head and neck to determine whether or not inducing head motion in head-first impacts is a useful strategy for preventing neck injuries and also whether or not this strategy can be realized without increasing alternate head or neck injuries.  40 4. Develop a cadaveric cervical spine model for head-first impact suitable for testing with improved prototype helmets, that is improved over our lab’s existing model, by utilizing a 3D Hybrid III surrogate head and a physiologically-based muscle force replication system to create a stabilized and aligned neck posture.  1.13 Scope  Chapter 2 provides the design rationale for a new sagittal plane mechanical neck specifically designed for simulating aligned column head-first impacts, which is the worst case scenario for catastrophic neck injuries in sports and transportation.  Chapter 2 also presents the results of two different characterization tests, specifically drop testing onto a rigid perpendicular impact platform to assess the time lag between the head and neck force development and also bending flexibility tests.  Chapter 3 describes the results of further drop-testing characterization experiments to determine the new neck model’s sensitivity to a wider range of impact variables and which of these pose the greatest risk for neck injury; specifically these variables are the impact platform padding stiffness, pre-impact neck postural adjustments away from purely aligned, the effect of varying platform friction, as well as the effect of varying the impact platform angle.  Chapter 3 also contains the results of a 2nd drop-testing experiment utilizing both the newly developed head and neck model, but also the industry standard Hybrid III head and neck to provide a comparison between the two models in a full-factorial experiment. Chapter 4 provides the design rationale for the mechanical helmet prototype which is the fundamental motivation for this dissertation along with the results of a drop testing experiment to assess the performance of the helmet prototype.  Chapter 5 presents a new cadaveric cervical spine model of head-first impact which offers several improvements over an existing model in our laboratory that allows it to be used with contemporary helmets as well as future 3D helmet prototypes which are improved versions of the helmet presented in Chapter 4.  These improvements include the use of the 3D biofidelic Hybrid III dummy head and a method of simulating neck muscle forces to create an aligned pre-impact posture by applying forces along physiologic directions.  41 Chapter 2: A New Biofidelic Sagittal Plane Neck Model for Head­First  Impacts  2.1 Introduction  Cervical spine injuries from head-first impacts during sports and transportation accidents occur worldwide and they often have serious consequences including complete or incomplete spinal cord injury (SCI) and paralysis. These injuries typically affect a young demographic [1] and roughly half of all SCI occur in the cervical spine. Billions of dollars of annual health care costs occur worldwide for acute cervical SCI. Automotive collisions are the most common source of cervical SCI followed by violence, falls and then sports injuries [2].  It is difficult to correlate the fraction of cervical SCI that occur from head-first impacts or in people wearing helmets because detailed mechanism information is not collected in most epidemiological studies of SCI. Studies looking at smaller populations of high school level athletes have estimated the incidence rate of catastrophic neck injuries in hockey and football at 2.56 and 0.68, respectively, per 100,000 participants [9]. A review of video taped catastrophic neck injuries in US football between 1971 to 1987 concluded that the majority occurred from head-first impacts [52]. The typical injury involved a player dropping and flexing his head slightly prior to impact to use it as the first point of contact in a tackle. This is referred to as spearing or spear-tackling in football and it results in an impulse applied to the top of the head with the neck’s natural lordosis (curvature) largely removed due to the flexed head posture [52]. This aligned cervical posture is associated with grave risk for catastrophic injury since there is nowhere for the momentum of the following torso to be arrested other than the bony vertebrae [3]. The mechanisms of neck injury in head-first impacts have been studied by many researchers with cadaveric head and neck specimens [63, 107], full cadavers [78, 108, 109], computer simulations [85, 105] and anthropometric test devices. In particular, the Hybrid III dummy has been used to study head-first impacts in sports [82, 110] and automotive rollovers [111-113].  The standard Hybrid III neck was designed to provide a biofidelic flexion/extension response under high-speed rear-end and frontal collisions [114]. However, the Hybrid III neck’s stiffness in axial compression is higher than in humans [74] and its response to  42 changes in loading constraint is less sensitive than the much more flexible human cervical spine [71]. It has been reported that the Hybrid III head and neck temporal force development between head and lower-neck, even onto a rigid impact surface, had no apparent time lag between first head contact and lower-neck load development [115] and we have also observed this in our lab. This is significantly different than the so-called bimodal response in axial force observed at the impact platform in cadaveric head-first impact experiments [34, 63, 78]. In the cadaver experiments a first mode arises from deceleration of the head with no concurrent loading at the lower-neck while the second mode is largely due to torso deceleration and results in approximately equal load at the head and lower-neck. For these reasons, investigators at the Medical College of Wisconsin have criticized the use of the Hybrid III neck to make conclusions on neck injury outcomes in head-first impacts and this remains an important issue in the automotive safety engineering community especially in the context of rollover injuries [74, 75, 107, 109, 116]. There are specialized anthropometric dummies for side, frontal and rear-end automotive impacts and dummy necks have been developed for low speed rear impacts [27, 28, 117] or for omnidirectional low speed impacts [118]. In the automotive arena, specialized dummies have resulted in effective airbags, active head rests, and side impact airbags. However, to the authors’ knowledge, no anthropometric dummy neck has been designed for use in head-first impacts. Investigations of injury mechanisms and prevention approaches in automotive rollovers and head-first sports impacts do require a biofidelic surrogate neck. In head-first impact research there has been some evidence that the degree of constraint on the head affects neck injury risk [34, 63, 71, 72, 78, 85] and further that lowering the degree of head constraint [62] through an engineered device [105] could be a preventative strategy. To appropriately investigate and develop such engineering approaches to lower head constraint, a surrogate neck requires a biofidelic bimodal temporal response to impact. Therefore, our objective was to design and build a mechanical surrogate neck with repeatable and biofidelic temporal kinetic and kinematic responses and to subject it to a series of drop test experiments against a rigid surface perpendicular to the direction of incoming velocity to determine the temporal pattern of head and lower-neck axial load development.  In addition, instrumented bending experiments were conducted to determine the neck’s range of motion (ROM) and flexibility characteristics in the sagittal plane. 43 2.2 Materials and Methods  Our overall approach was to design and build a head-neck model that modeled the neck’s physiologic sagittal plane bending and axial impact behavior.  The model was designed for use on our drop tower carriage and with the lower-neck load cell described here had a mass of 15.2 kg, to approximate the upper torso of the 50th percentile male [34]. 2.2.1 Surrogate Neck Model Design  The articulating aluminum mechanical neck is shown in Figure 2-1.   Figure 2-1:  New head and neck model on drop tower Sagittal compressive 7-segment (SC7) articulated neck in an upright posture (left) and shown inverted with the surrogate head on the drop tower with human overlay for orientation (right).  Both views show the left side. Right hand image from Nelson and Cripton, 2010 [119] and used with permission from Taylor and Francis.  The neck design incorporates 8 vertebrae, T1 to C1, that connect to a surrogate head. The head was designed to match human sagittal inertia, mass, CoG location, size, and shape parameters [120, 121]. The neck’s geometry was designed to match the average sagittal centers of rotation (COR) for each vertebra [42, 43], sagittal vertebral body size [122-124],  44 axial displacement [33], sagittal bending range of motion (ROM) [125] and the ratio of flexion to extension bending stiffness at each intervertebral level [50]. A comparison to human mass properties for the head and neck is presented in Table 2-1.  The center of gravity (CoG) locations for each vertebra, relative to the C7 vertebra CoG are, are presented in Table 2-2  and are shown in the (left) lateral view of Figure 2-2.  Machine drawings of the mechanical head and neck can also be seen in Appendix A. 45 Table 2-1:  Surrogate head and neck mass and inertia comparisons  a: HIII head mass from (King Foster, J.O. et al. 1977)[126] b: Human head & head and neck mass and inertia from (Walker, Harris et al. 1973)[120] c: Human vertebral mass and inertia from (Camacho, Nightingale et al. 1997)[50] d: HIII inertia from (Hubbard and McLeod 1974)[121] e: Sum of vertebral masses of (Camacho, Nightingale et al. 1997)[50] and head mass of (Walker, Harris et al. 1973)[120]  Table 2-2:  Surrogate head and neck center of gravity location comparisons  a: The lordotic posture as shown in Figure 2-2:  T1 flexed 35° wrt horizontal, all other levels 5° extension b: Human CoG from (Camacho, Nightingale et al. 1997)[50] c: Hybrid III dimensions interpreted from Denton 1716A upper neck load cell drawing (Robert A. Denton Inc., Michigan, USA) d: Human dimensions interpreted from (Walker, Harris et al. 1973)[120]   HIIIa Humanb Humanc Present Study Humanc Present Study Humanb HIIId kg kg kg kg kg*cm 2 kg*cm 2 kg*cm 2 kg*cm 2 Head 4.38 4.38 ± 0.59 4.38b 5.46 - 209 233 ± 58 238 Head and Neck 5.44 6 ± 0.79 4.63e 7.25 - 496 446 ± 37 na grams grams kg*cm2 kg*cm2 C1 - - 40.4 203.4 0.063 0.956 - - C2 - - 50.8 442.5 0.110 1.958 - - C3 - - 36.3 235.3 0.045 0.643 - - C4 - - 36.6 200.5 0.047 0.546 - - C5 - - 37.1 166.3 0.049 0.467 - - C6 - - 43.9 115.7 0.069 0.279 - - C7 - - 50.5 50.8 0.119 0.085 - - Mass Inertia Iyy Level Vertebra Anterior X Cephalad Z Anterior X Cephalad Z Anterior X Cephalad Z Present Study HIII c C1 -1.50 12.57 1.42 12.19 2.32 10.52 X (anterior) 1.5 1.8 C2 -1.63 10.06 1.09 9.70 2.10 8.73 Z (superior) 5.2 4.8 C3 -0.74 7.77 1.75 7.42 1.95 6.73 C4 -0.64 5.79 1.47 5.49 1.85 4.97 C5 -0.66 3.84 0.97 3.68 1.54 3.29 C6 -0.33 1.98 0.58 1.88 1.00 1.63 Present Study Humand C7 0 0 0 0 0 0 X (posterior) 0.4 1.4 T1 0.10 -2.03 -0.84 -1.85 na na Z (inferior) 3.2 2.3 Head CoG wrt C0/C1 Joint (cm) Head and Neck CoG wrt Head CoG (cm) Vertebral CoG locations relative to C7 CoG (centimeters) Present Study Aligned Present Study Lordotic a Human Lordotic b 46 Table 2-3:  Surrogate neck model vertebral dimensions and human comparison  Human Vertebral Dimensions (mm) Human COR locations† (mm) Present Study COR locations† (mm) Level Body Depth DVB a   Body Height HVB a   Level Human CoRX b Human CoRZ b Level Model CoRX Model CoRZ C1 - 14.1c C0/C1 1.5e 14.1f C0/C1 -0.6 14.1 C2 17d 18.5d C1/C2 -4.1 30 C1/C2 -6.2 30 C3 17.7 15.4 C2/C3 4 9.4 C2/C3 6 9.2 C4 18.3 14.7 C3/C4 4.3 9.7 C3/C4 6.5 9.9 C5 18.5 14.5 C4/C5 6 10.4 C4/C5 9 10.8 C6 19.0 14.4 C5/C6 6.4 12.9 C5/C6 9.6 13.5 C7 19.6 15.2 C6/C7 6.4 17.2 C6/C7 9.8 17 Present Study Vertebral Dimensions (mm) Stiffness Ratiog Level  Body Depth DVB Body Height HVB Body Width WVB RA RP Overall Depth Level Ext/Flex C1 25.5 15 63.5 31.7 37.3 69 C0/C2 1.39 C2 25.5 18.5 113 19.5 32.6 52.1 C3 26.5 15 99.3 20 26.7 46.7 C2/C3 2.81 C4 27.4 15 85.6 18.4 33.7 52.1 C3/C4 1.77 C5 27.8 15 70.9 18.1 29.1 47.2 C4/C5 3.37 C6 28.5 15 59.2 18.7 25.8 44.5 C5/C6 2.57 C7 30 15 35.1 20.2 22.3 42.5 C6/C7 1.9 a. Vertebral body height and depth from (Kandziora, Pflugmacher et al. 2001)[124] b. Human COR data from (Dvorak, Panjabi et al. 1991)[43] c. C1 dimenions interpreted from (Dong, Hong et al. 2003)[122] d. C2 dimensions from (Wilke, Kettler et al. 1997)[123] e. C0 CORx on C1 determined relative to C1/C2 CORx from (Dong, Hong et al. 2003)[122] f. C0 CORy on C1 assumed from height of lateral mass of atlas from (Dong, Hong et al. 2003)[122] g. Stiffness ratios interpreted from (Camacho, Nightingale et al. 1997)[50] †: The COR listed with a given vertebra is for the superior vertebra Dimensions as shown in Figure 2-4  47 The vertebral dimensions and COR locations are presented in Table 2-3 and the frontal view shown in Figure 2-2 respectively.  The surrogate head had a cast aluminum upper cap with a shape to represent the human head vertex. The upper cap was covered with 3 layers of serrated rubber sheeting (McMaster part #856956K, New Jersey USA) underneath 2 mm thick leather to represent the scalp. The neck connected to the head through a pivot joint representing the atlanto-occipittal joint (C0/C1) that allowed only sagittal rotation. At all other levels, both sagittal rotation and pure compression between adjacent vertebrae were allowed. Stacked serrated rubber sheets (McMaster part #856956K, New Jersey USA) were placed at all joints inferior to C0/C1 to create a non linear resistance to axial deformation with a low initial stiffness (i.e. toe region) that increased with further displacement. This is representative of the axial behavior of the cervical discs in vitro [31, 127].  Figure 2-2:  Schematics of as-built model showing lateral and frontal views Left: The vertebral Center of Gravity (CG) locations for the C7 to T1 vertebrae in a representative lordotic posture (fully distracted) for comparison to human vertebral CG locations (relative to the T1 CG) as published in Camacho et al., 1999 [85]and presented in Table 2-2 according to the T1 coordinate system shown.  Right: Front view of the as-built surrogate spine in an aligned and fully distracted posture.  Both views show that the articulation between adjacent vertebrae is through a slotted connection allowing relative sagittal rotation and axial compression. Right hand image from Nelson and Cripton, 2010 [119] and used with permission from Taylor and Francis.   At all intervertebral levels inferior to C0/C1, bolts passed through slots to locate the average COR while allowing relative compression and sagittal bending articulation. In an aligned  48 posture, the slots allowed 3.75 mm of pure compression from the fully distracted to the fully compressed states and served as the endpoints to axial displacement in an aligned posture at each intervertebral level resulting in a maximum overall neck (T1 to C0) compression of 26.3 mm. The neck was intended to be preloaded by a follower load mechanism to offer a target range of impact compression closer to 18 mm which was the average value for aligned cadaveric cervical columns [33]. In sagittal bending, the geometry of the vertebrae (and the stiffness of the intervertebral discs) determine the ROM. Figure 2-3 shows schematics of the surrogate spine’s articulating joints (without discs) at ROM endpoints in flexion and extension for both the fully compressed and distracted states. In an aligned and fully distracted posture, each disk was prescribed a thickness of 5 mm to match the average human cervical disc height for C2/C3 to C6/C7 [124].  Although the human C1/C2 does not have a disc, in our model the C1/C2 joint had the same axial displacement offered at other levels. This compromise was made to ensure a wide ROM in both C0/C1 and C1/C2 as has been consistently observed in the radiographic ROM studies [128]. The C1 vertebra was hollow and had a transverse internal bolt that passed through a slotted “odontoid-like” cephalad protrusion on the superior surface of C2 (Figure 2-3). At all intervertebral levels caudal to C1/C2, bi-lateral bolts protruded from an inferior vertebra through bi-lateral slots on the adjacent superior vertebra (Figure 2-1). The protruding bolts located the segmental COR [42, 43] at these levels and also served to guide bi-lateral follower load cables attached to a spring-loaded mechanism that served to apply an adjustable axial compressive force. This simulated the effects of stabilizing musculature. In order to locate a given vertebra’s COR on an inferior vertebra, the lateral dimension of our model became progressively larger moving cranially along the spine Figure 2-2.   49  Figure 2-3:  Range of motion for fully compressed and fully distracted conditions The maximum allowable rotations for extension and flexion bending under full compression and full distraction.  Note that these images are from a conceptual model.  The as-built model shown in figures 2-1and 2-2 has identical sagittal geometry.  Top two images adapted from Nelson and Cripton, 2010 [119] and used with permission from Taylor and Francis.  The neck was designed around an aligned posture to represent the worst case posture for compressive neck injury in head-first impacts [3]. The simplified vertebral shape was representative of the vertebral bodies which are thought to take the majority of load during compression in a slightly or moderately flexed posture [129]. While this shape did not  50 represent the human posterior vertebral elements anatomically, the relatively stiffer response to sagittal extension than flexion [50] of the cadaveric osseoligamentous cervical spine provided by the vertebral posterior elements was simulated through geometry.  Figure 2-4 shows a sagittal comparison between a human cervical vertebra and the simplified block representation used in this model. Referring to Figure 2-4, the dimensions Dvb and Hvb were determined from anatomical measurements of the human cervical vertebral bodies [124]. The intervertebral COR for a functional spinal unit determined by Dvorak and Panjabi (1991) were reported relative to coordinate systems located at the inferior and posterior margin of the inferior cervical vertebral body. In the as-built model of the surrogate spine, the vertebral body heights (Hvb) were anatomical [122-124]. The depths (Dvb) were determined as follows: Dvb for C7 was set arbitrarily but then the relative size between adjacent superior vertebral bodies (i.e. adjacent Dvb) along the cervical spine was preserved according to the mean vertebral body depth data of Kandziora et al. [124]. The dimensions labeled CoRx and CoRz in the model were determined by combining the vertebral body data of Kandziora with the COR data of Dvorak [43] to determine the COR locations relative to the depth and height of a given arbitrarily sized vertebral body. The dimension labeled Rp was determined by considering the relative bending stiffness ratio between extension and flexion at each intervertebral level. A graphical digitization of the flexibility curves for the cadaveric cervical spine [50] was used to estimate the slopes of the linear portions and the ratio of extension to flexion bending stiffness (Table 2-3) at each intervertebral level.  The specific design of the model dictated that the stiffness in bending was proportional to the volume of rubber disc being compressed for a given rotation.  Figure 2-4 shows a comparison of the volume of rubber compressed in extension to flexion bending for a given rotation. The ratio of the two areas shown, was used to determine the desired dimension labeled Rp, the dimension from the COR to the posterior margin of the vertebra after the dimension labeled Ra, the equivalent anterior dimension, had already been determined as described above.  The posterior dimension was calculated as: Rp = Ra * (SR)^0.5 where SR was the ratio of extension to flexion bending stiffness described above for each intervertebral level. The lateral dimension of C7 was based on anatomy [124] with the lateral dimension of all vertebrae superior to C7 increasing as described above. For the C7-T1 joint, the COR data from C6-C7 was used as data for C7-T1 was unavailable.  51   Figure 2-4:  Surrogate neck vertebral dimensions related to human anatomy and segmental stiffness Surrogate neck vertebral geometry relate to anatomic vertebral body dimension[122-124], average center of rotation locations, and cadaveric osseoligamentous cervical spine segmental stiffness[50].  The left image shows specified dimensions.  Dvb is determined from anatomical references.  CoRx and CoRz are located relative to Dvb along the posterior margin of the vertebral body.  Ra is a consequence of Dvb and CoRx. The right image shows two vertebrae and a disc.  Rp was calculated relative to Ra to produce a stiffer extension bending than flexion in proportion to the area of ‘disc’ compressed in extension to flexion.  Images adapted from Nelson and Cripton, 2010 [119] and used with permission from Taylor and Francis.   2.2.2 Methods: Flexibility Testing  Full surrogate spine (C0-T1) flexion-extension flexibility tests were performed on our custom spine motion simulator [130]. The simulator can apply pure non-constraining dynamic moments at controlled loading rates. The surrogate spine was fixed to the table at the cranial end and moments were applied with an articulated shaft to T1. A torque load cell (TRT-200, Transducer Techniques,Temecula, CA) measured the applied moment and was sampled at 20 Hz. A high speed camera (Phantom V9, Vision Research, Wayne NJ, USA) aligned perpendicular and lateral to the surrogate spine recorded kinematics at a resolution of 1440 x 1080 pixels and at 20 fps. A protocol consisting of 6 degrees of applied rotation per second with a torque limit of 6 Nm was applied to ensure the neck moved through the majority of its ROM. This protocol was applied with 0, 78, and 104 N follower loads. The 78  52 and 104 N magnitudes were an attempt to recreate 75 and 100 N of in vivo static compression reported in neutral and slightly flexed postures [45]. To determine vertebral rotations, markers on the vertebrae were tracked with TEMA tracking software (Image Systems, Linkoping, Sweden). Through comparisons with known motion we have established that the error from the marker tracking results in a mean accuracy for intervertebral angle measurement of 0.5º or less. The ROM was defined as maximum flexion angle plus maximum extension angle and the neutral zone (NZ) was defined as the difference between the flexion and extension angles at zero torque on the unloading cycles. ROM and NZ were calculated and compared to published studies.  2.2.3 Methods: Drop Testing  Twelve drop tests were performed from a drop height of 0.5m onto a rigid perpendicular surface, six with a 104 N follower load and six without a follower load. This number of tests was performed to establish the standard deviation of impact metrics as a measure of model repeatability.  The test apparatus is shown schematically in Figure 2-5. Without the follower load, the posture of the head would hang in extension due to gravity so thin fishing line was used to hold the head for the 0 N drops to create similar pre-impact postures to those with a 104 N follower load (Figure 2-6). The fishing line consistently went slack at impact and did not affect the test response. The head velocity at impact was consistently approximately 3 m/s which matches the velocity of cervical spine injuries from diving injury reconstructions [69]. This velocity has also been shown to cause cervical spine injuries in cadaveric laboratory studies [34]. Strain-gauge based uniaxial load cells (Omega LC 402-5K, accuracy = 0.1% of full scale output which corresponds to an accuracy of 22.2 N in this case) were used to measure axial force under the impact platform and at the lower- neck connection to the drop tower carriage. The load cells were collected at a frequency of 255 kHz and low pass filtered through a bi-directional 4th order Butterworth filter with 1500 Hz low pass cut off frequency. The high speed camera (Phantom V9) captured video of 1632 x 1200 pixel resolution at 1000 fps and was located perpendicular to the impact to enable kinematic calculations using the same techniques and with the same accuracy as in the sagittal bending tests (displacement accuracy = 0.5 mm). The peak head and neck forces, impulses, and the temporal lag between head and neck loading were calculated. Descriptive  53 statistics and a matched pairs t-test analysis comparing these quantities with no follower load to 104 N follower load was performed using SPSS, (SPSS Inc, Chicago, USA).  We anticipated that the time lag would be affected by follower load but not the peak head forces or neck forces.  Figure 2-5:  Schematic of surrogate head and neck on drop tower Drop testing experimental apparatus and instrumentation consisting of two uniaxial load cells, one underneath the impact platform (Fz Head) and one between the T1 vertebrae and the impact carriage(Fz Neck) representing the torso.  Image adapted from Nelson and Cripton, 2010 [119] and used with permission from Taylor and Francis.     54 2.3 Results  2.3.1 Flexibility Testing – Effects of Follower Load  The spine motion simulator moved the entire surrogate neck through a smooth controlled motion up until the 6 Nm load endpoints shown in Figure 2-6. Without a follower load the bending response was highly nonlinear with a characteristic NZ and logarithmic shaped flexibility curve as shown in Figure 2-7.  The addition of 78 and 104 N of follower load reduced the NZ, reduced ROM, and increased hysteresis upon unloading. The subaxial (C2 – C7) and upper neck (C0-C2) ROM and NZ in flexibility testing, at all moment levels, was found to be lower with each incremental follower load (Figure 2-7). A similar trend was observed for the segmental ROM at most levels although the addition of the 78 N follower load accounted for the majority of the difference (Figure 2-8).  The ROM and NZ data shown in Figure 2-7, Figure 2-8, and Table 2-4 generally indicate that the bending stiffness increased with incremental increases in follower load and that the extension response was stiffer than the flexion response.  Figure 2-6:  Drop testing pre-impact postures and flexibility testing motion endpoints Top: Pre-impact postures for zero follower load showing the fishing line required to provide the desired aligned pre-impact posture (left) and 104 N follower load (right). Bottom: The end points of a 6 Nm flexion/extension bending test protocol with zero follower load.  The extension end point at just over 68º of rotation (left) and the flexion endpoint at just over 80º of rotation (right).  55  Figure 2-7:  Upper and lower surrogate neck flexibility testing curves Upper spine (C0-C2) and lower spine (C2-C7) flexion-extension moment-rotation response at 0, 78, and 104 N follower loads.  The ROM was determined on the loading cycles for 1, 2, and 6 Nm torques (solid markers) and the NZ was determined on the unloading cycle (hollow markers).  In all tests a positive angle indicates flexion.  The tests started at 0 deg, or in an aligned posture, and the neck was continuously loaded and unloaded in both flexion and extension starting first with flexion as shown.  The addition of the 78 and 104 N follower loads changed the shape of the flexibility curves by creating more resistance to bending which resulted in decreased ROM and NZ.  The follower load also increased the hysteresis observed upon unloading.   56  Figure 2-8:  Segmental surrogate neck flexibility curves Intervertebral flexibility curves for 0, 78, and 104 N follower loads showing only the loading portions at all segmental levels along with the mean curve-fit equations from Camacho et al., 1997[50].  57  Table 2-4:  Neutral zone and range of motion results from flexibility testing   Neutral Zone (degrees) Range Of Motion (degrees)   Present Study, 6 Nm  In Vitro 4.5 Nmb Present Study 0 N Follower Load Present Study 78 N Follower Load Present Study 104 N Follower Load In Vitroa,b,c In Vivoa,d,e Spinal Level 0 N 78 N 104 N 1 Nm 2 Nm 6 Nm 1 Nm 2 Nm 6 Nm 1 Nm 2 Nm 6 Nm 1 Nma 2 Nma 4.5 Nmb   C0/C1 8 8 7.5 - 10.5 17.5 22.5 8.5 16.5 22 8 16.5 22.5 - - - 30d C1/C2 6 2 0.5 14.7 (9.7) 9.5 11 12.5 1 2 7.5 0.5 1.5 6.5 - - 23.8 (4.9) 12e C2/C3 11 2.5 2 6.8 (2.7) 12 14.5 16.5 2 8 16 1.5 5.5 15.5 9 10.5 11.1 (3) 11 C3/C4 14.5 3 2 6.7 (3.2) 16.5 18 20 3 6.5 19.5 2 4 19.5 11 15 12 (4) 16 C4/C5 13 5 3.5 7.6 (2.8) 15 17 18.5 4.5 8.5 16.5 3 6.5 15 12 13.5 13.3 (3.5) 19 C5/C6 10.5 6.5 4.5 7.6 (3.1) 12.5 12.5 16.5 6.5 8.5 14 4.5 7 11.5 11.5 14 11.9 (4.4) 19 C6/C7 21 15 11 6.6 (4.9) 22.5 24.5 25 10.5 13 22.5 9.5 11 18 10.5 14 11.6 (5) 16 C7/T1 8.5 4 3 - 6.5 11 18.5 6 6.5 6.5 4.5 4.5 5.5 4 5 - - C2/C7 70 31 3 - 79 88 96 26.5 45 88 20.5 33.5 79 54.3 66.8 59.9† 80 C0/C2 14 8.5 8.5 - 20 28.5 35 9.5 18.5 29.5 8.5 17.5 29 32c - - a. 1 Nm and 2 Nm In Vitro ROM from (Miura, Panjabi et al. 2002)[125] b. 4.5 Nm In Vitro ROM and NZ from (Wen, Lavaste et al. 1993)[131] c. 1.5 Nm In Vitro ROM interpreted from (Camacho, Nightingale et al. 1997)[50] d. C0/C1 In Vivo ROM from (Penning 1978)[132] e. C1/C2 In Vivo ROM from (Dvorak, Froehlich et al. 1988)[133]    58 2.3.2 Drop Testing – Effects of Follower Load  The head and neck axial loading histories were initially out-of-phase as shown in Figure 2-9. For all drops, the impulse measured under the head impact platform was bimodal. Loading at the lower-neck load cell began after completion of the mode 1 head impulse and was in-phase with mode 2 head loading.  Figure 2-9:  Drop testing temporal head-neck forces Head (solid lines) and Lower-neck (dashed lines) force with and without a 104 N follower load.  Time zero is the start of the 1st mode of the head impulse.   The presence of a 104 N follower load significantly decreased the time lag between head and neck loading onset and significantly decreased the neck impulse duration as shown in Table 2-5. The 104 N follower load had a larger mode 1 mean peak head force and mean peak neck force. The ratio of mode 2 to mode 1 mean peak head force was 67 and 57% for the 104 N follower load and no follower load respectively. Similarly, the mean peak neck force to mode 1 peak head force ratios were 42 and 45% for 104 N and zero follower loads. The peak forces  59 measured were repeatable having a coefficient of variation of 2% or less for peak neck force and 12% or less for the peak impact platen force (Table 2-5).  The impulse shapes for both head and neck force were also repeatable for both series of drops with and without a 104 N follower load (Figure 2-10) and the impulse magnitudes (area under the curve) for head and neck forces showed coefficients of variation less than 3% (Table 2-5).  60 Table 2-5:  Drop testing impact parameters   Mode 1 Mode 2     Follower Load Peak Head Force (N)* Head Impulse (Ns)* Impulse duration (ms) Peak Head Force (N) Head Impulse (Ns)* Head Impulse Duration (ms) Peak Neck Force (N)* Neck Impulse (Ns)* Neck Impluse duration (ms)* Time Lag (ms)* 0 N Mean 17113 15.6 1.6 11455 85.5 17.7 7094 74.4 18.68 6.15 S.D. 529 0.5 0.0 803 0.5 2.2 91 0.6 0.61 0.09 C.O.V. 3% 3% 1% 7% 1% 12% 1% 1% 3% 1% 104 N Mean 19105 16.3 1.5 10863 88.5 17.3 8542 74.8 15.82 2.23 S.D. 1134 0.2 0.1 1318 1.4 0.9 155 0.3 0.21 0.16 C.O.V. 6% 1% 5% 12% 2% 5% 2% 0% 1% 7% P (two tailed matched pairs t-test) 0.02 0.02 0.05 0.37 0.01 0.67 0.00 0.04 0.00 0.00 Nightingale 96a Mean 8205 13.2 na 3912 na na 3281 45 na 2 S.D. 324 0.8 na 1089.8 na na 807.8 3.8 na 0.3 C.O.V. 4% 6% na 28% na na 25% 8% na 19% a. Data is from 3 drops onto a rigid perpendicular impact platform from (Nightingale, McElhaney et al. 1996)[34] *indicates a significant difference (p<0.05) between 0N and 104N follower load groups from matched pairs t-test    61  Figure 2-10:  Drop testing head and neck force temporal repeatability Repeatability of head and neck loading for 0 N and 104 N follower loads. Each impact was naturally slightly offset in time from the other impacts and this was not corrected to allow for visibility of the data.  The head and neck kinematics were also highly repeatable. The neck response was always column-like and it did not buckle as it compressed.  The amount of compression during impact was affected by the follower load as shown in Figure 2-11.  Figure 2-11:  Drop testing neck compression C1-T1 compression for typical drops with and without a 104 N follower load.  62 Rather than buckling, the neck had a self-aligning tendency where the intervertebral angles between C1 and C6 would tend to decrease throughout the impact.  The intervertebral angles for two typical drops, one with and one without a 104 N follower load, are shown in Figure 2-12.  Throughout impact at levels C7/T1 through to C1/C2 no angles larger than 5 degrees were recorded. The largest angular motions after impact occurred at the most cranial (Head/C1) and caudal (C6/C7) levels.   Figure 2-12:  Drop testing intervertebral angles Intervertebral rotations for a typical drop with and without a 104 N follower load.  A positive angle denotes flexion.  2.4 Discussion  We have developed an aluminum sagittal plane surrogate neck for head-first impacts and report here on a series of tests to characterize its response with different follower loads in flexion-extension bending and in head-first impacts. An objective for the neck was to incorporate a repeatable and biofidelic head and neck axial force response, including a time lag between head and neck axial force development during a head-first impact. This time lag  63 has been observed consistently in cadaver tests. It was also desired for the neck to have a biofidelic ROM and bending stiffness in the sagittal plane. Detailed design and alteration of the articulations between vertebrae was performed to meet these design goals. We designed bi-lateral bolts that protruded from the inferior vertebrae through bilateral inferiorly projecting components of the adjacent superior vertebrae to achieve the desired bending and axial impact performance. This is conceptually similar to the design of the Bio RID [28] and BioRID2 [27] necks but an important difference was that the bolts in our model protrude through vertical slots instead of holes on their respective articulating vertebrae. Thus the vertebrae are unconstrained in sagittal rotation and are also allowed a range of axial compression that is important for head-first impact biofidelity and is not offered by other dummy necks. Also, unlike other dummy necks, this design has incorporated motion at an anatomically-based upper cervical spine. The bimodal nature of the head impulse in the present study was in good agreement with many previous cadaveric studies. The cadaveric head-first impacts performed by Nightingale (1996) showed a mean time lag between head contact and load development at T1 of 1.7 ±0.3 ms for the impacts onto a rigid surface and with no follower load. Our model with a 104 N follower load demonstrated a mean time lag of 2.2 ±0.2 ms. Our average value is within 30% of the average cadaver value from the Nightingale study and the standard deviations of our model measurements and those cadaver measurements overlap. This is similar agreement to some measures of biofidelity that have been applied to other ATD components (i.e. Hybrid III neck flexion, Hybrid III chest compression etc) where the dummy is required to fit a corridor of cadaver responses that represents the range of cadaver responses [114]. Furthermore, the model developed for studying aligned cadaveric cervical columns at the Medical College of Wisconsin displayed approximately a 1ms lag time between head and the onset of lower-neck loading [33] and recent but as yet unpublished cadaveric testing in our lab (presented in Chapter 5) showed close to a 3 ms lag time for a specimen in the aligned stiffest axis configuration under a simulated muscle force replication. As our surrogate neck’s lag time currently lies in between these published values and is within one standard deviation of Nightingale’s results, we judge it to be biofidelic with respect to head-neck force time lag when 104 N of follower load is applied.  64 In the tests conducted using our model without the follower load, the head-neck force time lag was unacceptably longer. Our model with a follower load might then better represent the cadaveric human neck and in fact a follower load slightly larger than 104 N may shorten the time lag between head and neck load development to more closely approximate Nightingale’s test series if that was a priority. The use of the follower load in our model appears essential for a biofidelic temporal kinetic response. The follower load induced pre-impact compression and thus there was less compression displacement available, and a shorter time period, before the rubber discs “bottomed out” causing the neck force to develop. The metallic nature of the head and neck resulted in peak forces for both the head and neck that were much larger than would have been present with biologic tissue [33, 34]. This was in part due to the inherently stiffer material involved with the aluminum surrogate neck compared to bones and discs in the cadaveric neck. It was also in part due to the aligned geometry of the surrogate neck when it was dropped and likely also to the fact that it did not exhibit buckling. Using our model with a 104 N follower load, the mean peak forces at the head for modes 1 and 2 were 133 and 178% of the average value measured by Nightingale (1996) with human cadaveric head and neck specimens in lordotic postures dropped from 0.5 m onto a perpendicular impact surface. Similarly, the peak neck forces with our model were approximately 160% larger than those measured by Nightingale (1996) or Pintar (1995).  Although the magnitudes of the forces were larger than those measured with cadaver specimens the ratios of the peak values were in better agreement. With the 104 N follower load, the ratio of mean peak head force in mode 2 to mode 1 was 57% in our model compared to 48% in the cadaveric work of Nightingale (1996). Similarly, the ratio for mean peak neck force to mean peak head force (mode 1) was 45% for our model compared to 40% for Nightingale’s. Future iterations of our design will aim to reduce the peak forces but maintain a similar temporal and kinetic force magnitude pattern to that observed here. This may be achieved by using thicker rubber elements as well as a more compliant material for vertebrae.  Although this model has been designed solely for head-first impacts it was thought important to incorporate a realistic ROM and stiffness in flexion-extension bending since axial impacts can induce various degrees of sagittal plane rotation in cadavers and this  65 occurred in our model as well. Without a follower load, the overall moment-rotation response was non linear and exhibited a large NZ similar to in vitro spine kinematics [37, 125].  The ROM data from flexibility testing without a follower load was within an acceptable tolerance of published in vivo values given the large variability reported [125]. However, incrementally increasing the follower load increased the bending stiffness and lowered the ROM to a given moment input, thus producing motion that was not in agreement with published in vitro or in vivo results. This is because the simplified geometry in our model creates a coupling between axial compression and sagittal bending where sagittal rotation is reduced as axial compression increases. Our model’s geometry and follower load mechanism combined to simulate an antagonistic muscular resistance to rotation which is not found with in vivo or in vitro ROM testing.  Hence the bending stiffness increased and ROM decreased when a compressive follower load was applied with our model.  Comparisons of ROM values from our model to those in the literature must be interpreted in light of this consideration. It is clear that the design requirements for a biofidelic sagittal bending response and axial impact response are at odds with each other. In the current iteration the temporal kinetic pattern and time lag representing the degree of phase-shift between head and neck appear biofidelic at 104 N follower load. However, in bending, the shape of the flexibility curves were best without a follower load (Figure 2-7 and Figure 2-8) but adding the 78 or 104 N follower load reduced the ROM well below those that have been published for cadavers or human subjects. In future we will work to achieve optimum biofidelity in bending and in axial impact with a single level of follower load.   In our model, the simplified vertebral geometry ensures that the response from the neck is column-like without buckling. This is an effect of the coupling between axial compression and sagittal bending described above. As the compressive load developed and the neck compressed, the vertebrae experienced moments that caused their intervertebral angles to approach zero. This is in contrast to the natural flexibility, postural curvature, and slender nature of the human cervical spine. The natural cervical spine exhibits bending (or even buckling) in response to most head-first impacts [33, 34]. While we recognize the self- aligning nature of our model as a lack of biofidelity, we hypothesize that in an aligned posture the human spine must be attempting to respond axially before failure and our model’s  66 propensity for axial response is desirable. We base this hypothesis on the injury patterns observed in some of the more catastrophic compressive injuries including vertebral body burst fractures that occur in collision sports where one would expect the stabilizing muscles to be tensed at impact [3, 33]. We hypothesize that the static structural stabilizing effect provided by muscles would also apply over the initial milliseconds of a dynamic impact much shorter in duration that the muscle reaction time [34] and further, that the force magnitudes required to create these catastrophic compressive injuries could only be generated along the “stiffest axis” of an aligned or nearly-aligned column [33]. Our interpretation is that if a device is under development to prevent neck injury by reducing head constraint in head-first impacts in this most vulnerable posture, that it should be developed with a neck model that readily assumes axial load in a similar fashion. This makes our model suited more to device development and phenomenological observations of impact where the worst case spinal alignment always occurs rather than a predictor of specific human injury in a given impact. Since our model is not frangible, it can only be biofidelic up until the development of a large axial neck load. Compressive spine injuries resulting from cadaveric head-first impact experiments against similar rigid surfaces developed in as little as 3 ms after head contact at the first local maximum of the lower-neck load [34]. It is likely this same temporal point in our model that defines the endpoint of our model’s biofidelic range. In terms of device development, it is necessary for the surrogate model to simulate the injurious environment as accurately as possible over the duration that the device’s deployment occurs. What happens after that is not as much of a concern, at least initially. A published contemporary example of this approach is the MIPS (Multi-directional Impact Protection System, MIPS AB, Stockholm Sweden) helmet testing program [134]. In this program only a helmeted HIII head (without neck or torso) is dropped onto a rapidly moving impact platform [92] and the rotational head kinematics are studied.  Obviously the response after initial impact cannot be biofidelic without a neck or torso but their application presumably only requires a model with biofidelic response for the mode 1 head impact force and for a period of time where the head motion will not be influenced by the neck or torso. In order to evaluate the effect of changing head constraint upon neck reaction forces, we feel that an accurate temporal pattern for both  67 the mode 1 and mode 2 head impact forces and for the development of neck loads (such as we present here) is necessary.  A disadvantage of our current iteration is that we are unable to measure the reaction forces and moments at the upper cervical spine. This makes our results more difficult to compare to other dummy studies since the forces are likely to be different at the upper and lower end of the neck in some phases of the head-first impact.  Future iterations of our current neck design may perhaps incorporate an upper neck transducer instead of the current anatomically based upper neck for this reason. While we are encouraged by the results obtained so far, there are many other impact variables that should be considered, such as pre-impact head/neck posture and oblique angled impact surfaces with varying friction and padding. We will present results of experiments investigating these parameters in the future.  The current iteration of our design constrains against biofidelic axial rotation and lateral bending ranges of motion. The basis for limiting the neck’s motion to the sagittal plane included observations of injury patterns resulting from real-world head-first impacts and published neck kinematics during head-first impact. Anecdotally we know that athletes in many of the sports where these injuries occur (hockey, football, motocross) are most frequently in sagittal postures. Predominant injury patterns from head-first impacts include bilateral dislocations, bilateral fracture dislocations, and burst fractures [3] and all suggest that the injury occurred with a posture in the sagittal plane. Flexion-extension bending in the sagittal plane has the highest ROM and lowest bending stiffness [135, 136] for the cervical spine. While we are aware that non-planar impacts must also be considered, our interpretation of the epidemiology associated with head-first impacts suggests that much can be gained from studying these impacts in the sagittal plane.  Within the sagittal plane, this model currently constrains against anterior-posterior shear translation. This was a compromise made for ease of design in manufacturing the slots in the caudal protrusions on each vertebra to locate the COR. This constraint is known to have a stiffening effect upon the axial response in both the cadaveric thoracic [137] and cervical [71] spines and should be rectified.   Design engineering often requires compromise as certain aspects of a design are in competition with others. In this case, a major compromise was to limit the model to the  68 sagittal plane. We prioritized locating the average COR which resulted in the caudal projections from each vertebra and the increasing lateral dimensions moving superiorly along the neck. These features resulted in the mass and inertia for each vertebra being larger than the human case and anterior-posterior shear deformations being constrained against. In addition, this design of articulations resulted in a similar range of available compression for C1/C2 as other inferior spinal levels, unlike the human case. Despite these compromises that we were required to make in this project, this neck appears to have utility for our purposes. We feel that the results illustrate that the model is biofidelic in many of the key ways that we require for our cervical spine injury prevention research. Future changes and improvements utilizing more advanced geometry, materials, and manufacturing techniques are suggested by our current results and we will aim to use this information to build improved versions of this neck in the future.   This test series was a first characterization of a newly developed surrogate neck for head-first impact simulation. Our model exhibited a biofidelic temporal response between head and neck development with 104 N of follower load. The flexion-extension flexibility testing showed a rotation vs. moment response and ROM that was generally similar to cadaver and human volunteer data when no follower load was applied. Future research should focus on developing a neck that exhibits biofidelic bending and head-first impact performance with a single follower load level. The intended function for this head and neck model was to conduct preliminary testing of an injury prevention helmet concept that changes the constraint on the head at impact through inducing controlled and guided motion of the head and an inner shell inside the outer shell of the helmet [138]. Although our neck model’s geometry and kinematic response is greatly simplified compared to the human cervical spine, its aligned posture and tendency to respond axially provide a worst case scenario for axial neck loading with accurate and repeatable head and neck temporal kinetics. This worst-case spinal alignment situation is an advantage in the evaluation of neck injury prevention strategies and devices such as the helmet under development in our lab.  69 Chapter 3: Characterization and Comparison of a New Biofidelic  Sagittal Plane Surrogate Neck to the Hybrid III Head and Neck for  Head­First Impacts  3.1 Introduction  Cervical spine injuries from head-first impacts during sporting collisions and in motorcycle and bicycle transportation contexts are relatively rare but devastating injuries. For example, the incident rates for catastrophic cervical spine injury with neurological consequences in football and hockey are estimated to be 0.52 and 2.56 per 100,000 participants among high school level athletes in the United States [139, 140]. The incidence of cervical spine fractures with or without SCI has been recorded as 0.79 and 0.21 per 100,000 days of snowboarding and skiing, respectively, at the Whistler/Blackcomb ski area in Canada [10]. A study analyzing European motocross racers over a 12 year period, where each race lasts approximately 20-30 minutes, found 3 cervical spine fractures with accompanying SCI out of 1500 recorded accidents that occurred over an estimated 66,000 hours of riding time [11].  Although these incidence rates are low, the severity of the injuries are catastrophic (partial or complete quadriplegia) and permanent. Preventative measures such as rule changes and education can reduce injury incidence rates [8, 141] but the risks associated with collision sports and motorcycle and bicycle transportation are inherent to the dynamics involved and as such, will always be present.  Helmets are effective at protecting against brain injuries [142, 143], however there is a need for neck injury prevention devices to prevent broken necks; we need tools to develop these. In the automotive context, there has been a great push to develop and incorporate passive safety systems such as airbags or active head restraints into vehicles to provide injury protection in the motor vehicle collisions that inevitably do occur. These passive safety systems have been developed in large part with biofidelic application-specific mechanical crash test manikins. Despite the fact that automotive rollovers are over-represented in serious (AIS 3 and higher) head and neck injuries [144] there has not been a surrogate neck available that is biofidelic in head-first impacts to test injury prevention approaches for these important automotive, transport, and collision-sport impact situations. We have recently developed and reported [119] on a mechanical surrogate head and sagittal compressive seven vertebra neck for conducting head-first impact tests that was  70 presented in Chapter 2. Throughout this chapter and the rest of the thesis it will be referred to as the SC7 neck.  We designed our model to simulate the straightened cervical spine posture which has been shown to be a worst case scenario for catastrophic cervical injuries [3], and to provide a biofidelic range of motion in flexion-extension bending and axial compression.  The SC7-model’s response to impact in inverted drop testing has thus far been characterized with limited instrumentation in one impact scenario: an aligned neck posture dropped onto a low friction and rigid impact surface normal to the incoming velocity. However, in reality, head-first impacts occur over a wide range of sporting and transportation contexts with much variability in impact conditions.   Many cadaveric experiments have shown that the cervical spine’s axial response is sensitive to conditions such as pre-impact posture [33, 78], the obliqueness of the surface impacted [34], and surface padding [33, 34, 39] . In addition, finite element models of head- first impact validated against cadaveric impacts have also shown that high friction between the head and impact surface is a risk factor for neck injury [85]. This study also showed that the neck injury risk exacerbation from padding was due to increased surface friction and not from pocketing as was previously reported [34]. A common conclusion regarding head-first impacts is that the risk of neck injury increases with increasing constraint between the head and impact surface at impact that could arise due to friction, pocketing, or due to surface geometry [34, 62, 72, 78, 110].   The Hybrid III dummy neck was developed for frontal and rear-end automobile collisions and head-first impact biofidelity was not considered in its development [126]. It has been reported that the Hybrid III dummy neck was not as sensitive to changes in end conditions as the human cadaveric neck during quasi-static axial loading and its stiffness is reported to be over 50 times that of the human cadaveric cervical spine [71, 74]. Full manikin inverted head-first impacts with the Hybrid III dummy showed that the lower-neck loading magnitudes varied with the incident angle of the dummy neck relative to the impact surface normal [115]. At present, except for the few situations reviewed above, the sensitivity to changes in constraint, impact orientation, and surface friction for both our SC7 neck, and the Hybrid III neck in these impact conditions is unknown.  Thus the objectives of this study were to further characterize the SC7 neck’s response to head-first impacts while subject to defined variations in the neck’s pre-impact posture and the stiffness, friction, and alignment  71 (incident angle) of the impact surface with respect to the direction of incoming velocity. We aimed to determine the worst case scenarios for neck injury and also to compare the response of the SC7 head and neck to the Hybrid III head and neck using identical test apparatus and impact conditions. 3.2 Materials and Methods  3.2.1 Materials  Our SC7 model was designed for use on drop towers and allows simulation of a head- first impact with a following torso mass. In the present experiments, impacts were conducted against an impact platen that could be adjusted in the sagittal plane to allow various angled impact surfaces to be tested.  Both our SC7 and the Hybrid III head and neck models were tested with identical apparatus and instrumentation as shown in Figure 3-1.  Figure 3-1:  Schematic of head and SC7 neck model on drop tower Schematic of experimental apparatus shown with our surrogate head and SC7 neck model.  The Hybrid III head and neck was tested with identical instrumentation and apparatus. 72 3.2.2 Experimental Design  The four factors we studied in the preliminary experiment were: friction between the impact surface and the head, compliance of the impact surface, the head-neck posture at impact, and the angle of impact surface with respect to horizontal.  The peak axial forces and moments at the lower-neck [33, 34, 135] were chosen as outcome variables representative of neck injury risk.  A fractional factorial experimental design referred to as the L9 [91, 145] was used to simultaneously vary these four independent variables in 9 unique runs to estimate the active main effects that may exist among the variables. The L9 is an orthogonal array (balanced) design for studying 4 factors, each at 3 levels.  The characteristic of an orthogonal array is that in each column of the design, each factor level is matched an equal number of times with each factor level in the other columns. Referring to Figure 3-2, note that in each column, each posture, platform angle, and platform compliance occur once while each column contains a single friction level.  Across each row, the three friction conditions are represented such that the number of occurrences of all four factors’ levels is balanced. This offers tremendous efficiency over a full factorial experiment (81 runs) at the expense that the main effect estimates, while unbiased by other main effects due to the balanced nature of the design, are all confounded with 2-factor interactions which may or may not be active and remain unknown without additional testing. Table 3-1:  Impact variables studied in L9 fractional factorial  Each run was replicated and the resulting 18 tests were performed in a completely randomized (i.e. no blocking) fashion in one testing session. The active main effects that were uncovered in the initial study were included in the design of the subsequent full factorial study (12 duplicated runs, 24 tests) to compare the response of the SC7 head and  73 neck model to the 50th percentile Hybrid III head and neck. The second full-factorial study also served as a check on the confounded main-effects that were measured in the first study.  Figure 3-2:  Impact variables studied in the L9 fractional factorial The 9 runs making up the L9 orthogonal array.  Note the balanced nature of the array, in each column each posture and platform angle occurs once while along the rows, each level of friction and platform stiffness occurs once. The left column has a friction condition of Teflon- Teflon (low) for the head-surface respectively, while the middle column shows leather-Teflon (medium), and the right column shows leather-rubber (high). Runs 1,6, and 8 show the thickest padding for the “low” stiffness, runs 2,4, and 9 show the “medium” stiffness padding, and runs 3,5, and 7 show the “high” stiffness padding condition. Runs 1,3,4,5,8,and 9 show either flexion or extension postures imparted by a light rope tied to the impact carriage.  74 3.2.3 Methods  All experiments were conducted on our drop tower with a carriage mass of 15.2 kg and dropped from a constant height of 60 cm to create an impact velocity just over 3 m/s. This speed has been shown to be the human tolerance to sub-axial burst fractures from head- first diving impacts [69] and is the same impact speed we used in earlier impact characterization with this model [119]. All tests with our model used 78 N of follower load applied through an adjustable cable mechanism.  This was based on results of earlier testing which showed that 104 N best recreated the head-neck time lag but that range of motion in bending was severely reduced.  Because of the use of angled platforms in this study which we expected to produce larger head and neck rotations, 78 N was used. The four variables at their respective 3-levels comprising the 9 unique runs are shown in Table 3-1 and Figure 3-2.  The angle of the impact platform was set at either -15º, 0º, or 15º. The -15º platform angle was such that the point of head contact was posterior to the head vertex at impact. Three head postures of 10º flexion, neutral, and 10º extension were also tested. The angle used was formed between the vertical and a superior-inferior straight edge on the surrogate head using a level and protractor. This provided a convenient measure of postural change as it was observed that changing the head posture also changed the neck posture. The neck was designed around an aligned posture [119] and our intent here was to study its sensitivity to small postural changes away from fully aligned. In order to impose an initial flexion or extension posture, the surrogate head was tied to the carriage with a light string as shown in Figure 3-2.  Preliminary work showed that the line became slack at impact and did not regain tension through the impact. For each test, the impact platform was moved along the anterior- posterior direction prior to impact to best align the point of initial head contact directly over the center of the anvil and the uniaxial load cell. Three different levels of surface friction and stiffness were also under study and the following methods were used to avoid confounding the two variables: The uppermost layer of the impact surface was always either a 1.5 mm sheet of durometer 30 neoprene rubber either exposed or covered with a 0.5 mm Teflon sheet. Similarly, the leather skull cap of the surrogate head was either left exposed or covered with an exemplar Teflon sheet resulting in three ordinal head-surface friction pairs: Teflon-Teflon (low, µs=0.27), leather-Teflon (medium, µs=0.41), and leather-rubber (high, µs=0.97). For surface stiffness, the neoprene  75 rubber sheet on top of the steel impact surface, with or without Teflon sheet, was the stiffest (high) value and was essentially rigid (Young’s Modulus 200 GPa). The medium stiffness was achieved by adding a 14 mm thick piece of foam (density 24.7 kg/m3, effective Young’s Modulus 90.1 kPa) and the low stiffness used this same 14 mm thick piece under another 25 mm layer of softer foam (density 31.8 kg/m3, effective Young’s Modulus 22.2 kPa).  Because the two foams were stacked, or in series, the overall effective modulus for the softest padding condition was 17.8 kPA. After the L9 experiments, it was observed that the 14 mm thick  § § § § § § § § § § § § § § § § § § § § § § § § § § § § §  § § § § § § § § § § § § § § § § §§                   from permanent deformation for the full factorial.  This gave the low stiffness padding in the full factorial study an overall effective modulus of 15.5 kPa.  Aside from these padding differences, all tests within the fractional and the full factorial experiments used identical instrumentation and apparatus. The stiffness of the paddings were estimated by performing static indentation tests to 15% strain and using the method of Zhang et al., 1997 [146] to determined the Effective Young’s Modulus.  This method is inadequate for full characterization of the foam over the range of compressions experienced in these drop tests but was used here to report the relative difference in stiffness between the “low” and “medium” padding conditions.     Each impact was captured by a high speed camera (Phantom V9, Vision Research, Wayne NJ) set up perpendicularly to the impact at 1000 frames per second, with 10 image- synchronized data channels sampled 78 times per frame.  The  Phantom camera software allows limited control over data acquisition rate; either one sample per frame or the fastest possible which is a function of the number of channels being collected, hence why 78 kHz was used. The data channels are schematically shown in Figure 3-1 and correspond to the 6 axis lower-neck load cell (Denton 4366J) between the lower-neck and the drop tower carriage, a uniaxial impact platform load cell (Omega LC 402-5K), and 3 linear accelerometers (Endevco 7264C) oriented in the three anatomic planes (sagittal, coronal, transverse) at the head center of gravity. For the purposes of characterization, several impact parameters were calculated. These were the peak resultant head accelerations, Head Injury Criteria scores (HIC15), peak lower-neck forces, moments and impulses as well as the peak force and impulse under the impact platform.  The HIC15 is a measure of head injury that  76 takes into account the time integral of head accelerations rather than just peak values.  It has a long history in automotive safety that stems originally from the Wayne State Tolerance curve of head injury which was comprised from a wide variety of results with volunteer testing, cadaveric testing, animal testing, clinical research and considered multiple injury mechanisms [147].   ??? ? ??? ? ? ??????? ? ?????? ?? ?? ? ?.? ??? ? ??? where t2 and t1 represent any two arbitrary 15 ms timeframes during the acceleration pulse.  For the Hybrid III dummy, HIC calculated over 15 ms has an Injury Assessment Reference Value of 700 [30] but here it is used strictly to compare multiple impacts.  Anti-aliasing hardware filters were applied in the amplification board (Analog Devices 3B) to specify a nominal low-pass filter of 1000 Hz which spectral plots showed to be closer to 1100Hz.  Prior to data manipulation, the entire raw data set was digitally filtered with a 4th order low pass Butterworth filter with cutoff frequency of 1000Hz (as is consistent with that specificed by the SAE J211 specification[148]) applied through the filtfilt routine in Matlab (Mathworks, Natick MA, USA) thus making it a bi-directional filter that does not cause a time distortion.  For the L9 study, a “main effects only” ANOVA model was applied using SAS Learning Edition 4.1 (SAS Institute Inc.,Cary, NC, USA) to the lower-neck axial force and sagittal moment as is standard practice according to the Taguchi method [145]. For the second study using both our SC7 head-neck and the Hybrid III head and neck model, a full ANOVA model was applied using SPSS 13.0 Student Edition (SPSS Inc., Chicago, IL). Since both of the physical neck models are inert and highly repeatable, a confidence level of 99% was used for significance in all statistical considerations reported here. For the L9 study, in addition to significance, the relative effect sizes (η2) among independent variables were considered to assess the importance of including them in subsequent testing.    77 3.3 Results  3.3.1 Preliminary L9 Fractional Factorial Experiment  The main effects ANOVA analyses showed that the platform angle accounted for the greatest observed variance for both the peak lower-neck force and sagittal moment; nearly 6 and 5 times the next largest factor which in both cases was platform surface stiffness. Only platform angle and stiffness effects met both significance and magnitude thresholds as posture and friction made much smaller contributions to the observed variability. The main effects plots with 1% Least Significant Difference error bars from this experiment for axial force and sagittal moment are shown in Figure 3-3 and Figure 3-4 respectively.   Figure 3-3:  L9 experiment lower-neck axial force main effects Axial force main effects plots with 99% confidence least significant difference error bars.  Platform angle and platform stiffness were the two dominant variables.   78  Figure 3-4:  L9 experiment sagittal lower-neck sagittal moment main effects Sagittal moment main effects plots with 99% confidence least significant difference error bars.  Platform angle and platform stiffness were the two dominant variables  The largest axial forces occurred in normal impacts and were significantly lower at both oblique angled surfaces with the lowest against the +15º deg platform where the initial point of contact at impact was anterior to the head vertex causing a head flexion upon impact. The lowest sagittal moments were observed in normal impacts. The -15º angle produced a flexion peak moment while the +15º angle resulted in an extension peak moment. The mean peak-values for both injury metrics and other impact parameters in the 9 runs are presented in Table 3-2.  79 Table 3-2:  L9 experiment impact parameters    80 3.3.2 Hybrid III Full Factorial Comparison Experiment  Based on the L9 ANOVA results, only the platform angle and surface stiffness variables were included in the design of a 2nd test series to compare the SC7 head and neck model to the 50th percentile Hybrid III head and neck. The experimental design for the comparison study to the Hybrid III head and neck was a 3 x 2 x 2 factorial consisting of the same 3 platform angles, 2 modified platform stiffnesses conditions (as described above), and the two head and neck models.  The low stiffness now utilized the same 1” thick foam as in the L9 tests covering 2 layers of 5mm PVC exercise (yoga) mat and the medium stiffness was achieved by removing the top 1” low density foam.  The 6 unique runs, shown in Figure 3-5, were each replicated with both models for a total of 24 tests.  Figure 3-5:  Impact variables studied in the Hybrid III full factorial Visual presentation of the 6 runs studied to compare our head and SC7 neck model to the Hybrid III head and neck model.  81 Both models exhibited the same trend of highest axial forces in normal impacts and highest sagittal moments against -15 degree oblique platforms (head contact posterior to vertex) as shown in Figure 3-6 and Figure 3-7. The peak axial forces were also higher at -15 degrees than +15 degrees.  Both models were repeatable in these metrics and it should be mentioned that the error bars show the range for the duplicated runs. For peak axial force the largest coefficients of variation for the duplicated unique runs with our model and the Hybrid III respectively were 1.64% and 0.77% and similarly for peak sagittal moment they were 7.1% and 3.4%.  Both injury metrics, as well as the impact force, head accelerations, and HIC values were significantly higher with the SC7 head-neck model than the Hybrid III over all platform angles and padding conditions (Figure 3-6, Figure 3-7, Table 3-3, Table 3-4). For the axial force, the head-neck model and then platform angle main effects were the 1st and 2nd most dominant effects and together accounted for 88.8% of the observed variance while no large interactions were observed.  For sagittal moment, again the head- neck model and platform angle, along with their interaction were the only large effects observed.  The other main difference between the two models was the temporal relationship between the head and lower-neck force development.  Against the stiffer surface padding, at all platform angles, the SC7 model demonstrated a bi-modal (head) impact force where the 1st mode of head loading was out-of-phase with the lower-neck axial force that became in phase for the 2nd mode of head loading.  The Hybrid III head and neck showed a different relationship.  While the head force was bi-modal, the lower-neck force developed during the 1st mode of the head force and the peak lower-neck force was not in-phase with the 2nd mode of the head force.  With the Hybrid III neck, the peak lower-neck force occurred between the peak head forces in the 1st and 2nd modes.  This temporal trend for the Hybrid III model was observed against both padding stiffnesses.  With the SC7 model, the temporal trend did change with the soft padding; the head force became much more unimodal and was in-phase with the lower-neck force as shown in Figure 3-8.  82  Figure 3-6:  Peak lower-neck axial force – Hybrid III full factorial experiment Peak axial force ANOVA results, the error bars represent the range for each duplicated run.   Figure 3-7:  Peak lower-neck sagittal moment – Hybrid III full factorial experiment Peak sagittal moment ANOVA results, the error bars represent the range for each duplicated run.  83  Figure 3-8:  Full factorial experiment temporal head-neck forces and head accelerations Comparison of the temporal axial force development at the impact platform and lower-neck and head accelerations for the SC7 head-neck and the Hybrid III head-neck corresponding to runs 3 and 6 in the full factorial experiment (Figure 3-5).    84 Table 3-3:  Impact parameters for full factorial experiment   Fx (N) Fy (N) Fz (N) Mode 1 Mode 2 Time lag (Mx (Nm) My (Nm) Mz (Nm) Fz(Ns) Fhead (Ns) Ax (G's) Ay (G's) Az (G's) Ares (G's) HIC15max Mean -1362 122 6608 7868 6340 0.9 11.9 106.7 -6.5 70.4 93.9 -12 5 107 107 69 SD 30 15 9 165 30 0.1 0.4 3.8 0.3 0.2 0.5 41 12 8 8 2 CoV 2% 12% 0% 2% 0% 11% 4% 4% 5% 0% 1% 325% 232% 7% 7% 3% Mean -849 135 7642 9123 7243 1.5 14.2 1.1 -0.1 79.1 103.3 18 6 128 128 87 SD 25 6 13 109 67 0.1 0.6 50.0 1.9 0.3 0.3 1 2 2 2 7 CoV 3% 5% 0% 1% 1% 7% 4% 4638% 1595% 0% 0% 7% 38% 2% 2% 8% Mean 1089 77 6345 8203 6202 1.7 10.0 -104.1 1.9 67.8 90.3 -2 1 110 110 72 SD 8 5 45 76 43 0.3 0.4 3.7 0.1 0.4 0.7 23 4 5 5 0 CoV 1% 7% 1% 1% 1% 18% 4% 4% 4% 1% 1% 1255% 413% 5% 5% 0% Mean -1331 132 7193 9474 6715 0.9 13.5 107.7 -7.2 70.8 94.2 -29 7 157 157 86 SD 30 10 30 64 18 0.1 0.4 4.4 0.1 0.3 0.1 4 2 3 3 2 CoV 2% 7% 0% 1% 0% 11% 3% 4% 1% 0% 0% 15% 28% 2% 2% 2% Mean -919 119 8024 10090 7604 1.0 14.8 2.7 -1.6 79.0 102.7 -17 51 166 166 103 SD 48 18 62 108 237 0.1 1.1 50.2 0.1 0.5 0.5 50 62 2 2 21 CoV 5% 16% 1% 1% 3% 10% 7% 1880% 4% 1% 0% 297% 121% 1% 1% 21% Mean 1024 77 6367 9036 6059 1.1 9.6 -106.8 2.1 67.3 89.5 -17 27 134 134 62 SD 22 3 33 66 8 0.0 0.0 3.9 0.1 0.3 0.2 3 35 2 2 8 CoV 2% 3% 1% 1% 0% 0% 0% 4% 4% 0% 0% 19% 130% 2% 2% 12% Mean -2482 282 8524 10955 n/a 10.0 -25.1 235.5 -4.2 76.2 109.2 44 -23 -73 75 100 SD 242 72 8 352 n/a 0.5 7.6 6.1 20.6 0.3 0.4 11 1 9 12 29 CoV 10% 26% 0% 3% n/a 5% 30% 3% 490% 0% 0% 26% 4% 12% 15% 29% Mean -1244 389 9891 11393 n/a 9.6 -32.1 63.4 -5.8 84.1 117.9 58 -27 -78 90 156 SD 73 85 48 53 n/a 0.1 8.7 4.9 0.1 0.2 0.1 12 9 1 9 18 CoV 6% 22% 0% 0% n/a 1% 27% 8% 2% 0% 0% 22% 32% 1% 10% 11% Mean 29 -30 7624 8756 n/a 10.3 -32.6 -206.1 -8.5 73.1 102.1 -3 -25 -66 71 123 SD 1435 370 27 262 n/a 0.3 8.4 1.0 3.5 0.8 1.0 52 3 0 1 16 CoV 4968% 1230% 0% 3% n/a 3% 26% 0% 41% 1% 1% 2029% 12% 0% 1% 13% Mean -2940 306 9002 16642 11769 5.4 -24.3 270.8 -12.2 75.3 79.3 45 -26 -333 333 417 SD 71 23 73 760 412 0.0 0.3 0.9 1.1 0.5 0.8 3 4 8 8 9 CoV 2% 7% 1% 5% 3% 0% 1% 0% 9% 1% 1% 7% 17% 2% 2% 2% Mean -988 514 10724 18542 12866 5.4 -53.3 65.0 -8.1 82.2 81.9 75 -46 -373 374 593 SD 25 36 167 738 96 0.1 4.4 2.2 2.0 0.1 0.2 5 8 14 14 10 CoV 2% 7% 2% 4% 1% 2% 8% 3% 25% 0% 0% 7% 18% 4% 4% 2% Mean 1250 367 8003 13562 9848 6.0 -36.1 -189.9 -10.6 71.2 69.9 53 -27 -250 251 326 SD 25 11 27 914 701 0.1 4.3 1.5 3.2 1.0 1.4 1 8 18 18 8 CoV 2% 3% 0% 7% 7% 2% 12% 1% 30% 1% 2% 3% 30% 7% 7% 2% SC7 Head- Neck 1 2 3 4 5 6 Fhead (N) Run # Hybrid III Head- Neck 1 2 3 4 5 6 85 Table 3-4:  Impact parameter comparison between SC7 and Hybrid III head-neck models   Equiv. Run FS to HIII Fz (N) My (Nm) Mode 1 Mode 2 Time lag (ms) Impulse Fz(Ns) Impulse Fhead - Mode 1 (Ns) Az (G's) HIC15max 1 29% 121% 39% na 1011% 8% 16% -168% 45% 2 29% 79% 25% na 540% 6% 14% -161% 78% 3 20% 98% 7% na 506% 8% 13% -160% 70% 4 25% 151% 76% 75% 500% 6% -16% -312% 384% 5 34% 83% 84% 69% 440% 4% -20% -326% 473% 6 26% 78% 50% 63% 445% 6% -22% -286% 425% Percent Differences (SC7 greater than Hybrid III) Fhead (N) 86 3.4 Discussion  We undertook two characterization experiments; the first an L9 fractional factorial with our SC7 head and neck model to determine if the lower-neck loading was sensitive to pre-impact posture, platform stiffness, friction, and platform angle, and also what set of these impact conditions produced the highest lower-neck loading. The L9 study could only provide estimates of the main effects that were confounded with possible active (and unknown) interactions and therefore the findings for each variable were to be considered carefully in light of the mechanics.  The main-effects estimates made in the first study influenced the design of the second full factorial experiment to compare the response of our SC7 head-neck model to the industry standard Hybrid III 50th percentile head and neck. The L9 study showed that surface friction had a negligible effect upon both lower- neck injury metrics.  This is explained by a review of the test videos which showed that no observable head translations (until after the peak lower-neck loading) were detected in any of the L9 or other impacts.  The range of platform angles studied here were sufficiently normal to the incident velocity that all of the friction conditions discouraged head motion.  We expect that if larger platform angles were imposed (or possibly lower frictions as will be discussed) that surface friction may have played a larger role in either allowing or preventing head motion which would have then had a stronger effect on the lower-neck loads. The lower-neck injury metrics were also not sensitive to the small deviations in pre- impact posture under study in the L9 experiment. This small range was chosen as the SC7 neck model was designed around an aligned posture [119] and we wanted to test its sensitivity to small postural changes. We expect that the posture would affect these lower- neck injury metrics if larger pre-impact head postural angles were imposed. The small effect of posture on lower-neck loading is explained by the stability of the column. In the SC7 neck model there is a coupling between sagittal bending and axial compression that manifests as a self-aligning of vertebrae throughout the column during impact. As the spine is compressed axially, intervertebral moments are transferred along the column which at all except for the most cranial and caudal levels, align the adjacent vertebrae. Thus small deviations away from an aligned pre-impact posture were not largely influential upon the lower-neck kinetics. Platform angle had the largest effect on the head kinematics and lower-neck loading. The general trend observed in both studies with the SC7 model and the Hybrid III head-neck,  87 was that with oblique impact surfaces (at all platform stiffnesses) the initial point of head contact was eccentric with respect to the head vertex, head CoG, and also the C0/C1 joint.   This moment about C0/C1 caused head flexion or extension. The effect was more pronounced at positive platform angles (+15º) where the geometric effect of moving the point of contact anteriorly added to the posterior location of C0/C1 relative to the head vertex to increase eccentricity and create larger head flexions. Against a normal impact surface, the head simply stopped and was effectively “gripped” such that no further head rotation could be induced.  Thus the lower-neck absorbed a sharper pulse of torso (carriage) deceleration and consequently experienced higher peak lower-neck forces. The oblique platform angles reduced axial loading at the lower-neck by allowing the head to stay in motion during the torso deceleration. The larger head flexion rotations against +15º platforms reduced the axial loading significantly more than extension rotations against -15º impacts. However, these oblique impacts resulted in larger moments at the lower-neck than the normal impacts in approximate proportion to the axial force reduction.  The angle of the platform affected the head’s kinematics upon impact. The neck kinematics were somewhat dependent upon the head kinematics in as much as they determined whether C0/C1 would go into flexion or extension, however due to the self-aligning nature of our SC7 neck model, the sub-axial intervertebral angles were similar for all impacts.  The temporal relationship between head and neck axial force development was not affected by the different platform angles. In this study, the platform padding condition strongly affected the peak head accelerations and impact platen force but played a much smaller role with the peak lower- neck axial force and sagittal moment. This effect is similar to what was observed with cadaveric head-first impacts [34] and also in finite element models based upon the same cadaveric impacts [86] where they found that the lower-neck mechanical loads were only slightly sensitive to changes in padding stiffness. It has been observed in cadaveric experiments [34-36], that the head and neck are initially out-of-phase in head-first impacts and thus what is beneficial for reducing head injury metrics, i.e. deep soft padding, is not necessarily beneficial for the neck and in fact can actually increase the risk of neck injury by adding constraint upon the head. Likely the only reason that the padding slightly reduced lower-neck axial loading here, was because in this experimental model, the mechanically robust necks were forced to completely stop the torso in each impact.  In contrast, the human  88 cervical spine typically cannot solely stop the momentum of the torso in the majority of head-first impacts above 3 m/s impact velocity without undergoing a failure at which point the column can no longer support axial load.  In our study it was observed that the surface stiffness had no effect upon the peak moments that developed with the Hybrid III head-neck but did play a small role with the SC7 neck model. For 0 and +15 degrees, the lower padding stiffness resulted in slightly higher moments that were significantly lower at -15 degrees with low stiffness. The padding’s effect upon peak moments with the SC7 neck was likely due to the essentially rigid endpoints to sagittal rotation at each intervertebral level. The Hybrid III neck’s range of motion does not incorporate a hard endpoint as the rubber elements in the neck govern the range of bending observed. The platform padding played a major role upon the temporal force development pattern between the head and lower-neck and the shape of the (Head) impact impulse for the SC7 model that was not observed with the Hybrid III head and neck.  The SC7 impact force- history (impulse) showed a bi-modal response with two clear impulses against the stiffer platform surface that became less distinct against the softer platform.  The peak lower-neck force was in-phase with the 2nd mode of the (head) impact force against the stiffer padding or the unimodal (head) impact impulse for the softer padding.  The Hybrid III head and neck also showed a bi-modal (head) impact force but the two modes, while identifiable, were two peaks of the same impulse and this response was not affected by the surface padding.  The peak lower-neck force occurred out-of-phase with, and in between, the two peaks of the bimodal impact force for both surface padding conditions.  In addition, with the Hybrid III neck, with both surface paddings, there was less lag time between the axial force development at the impact platform compared to the lower-neck.   All of these observations are explained by the higher initial axial compliance offered by the SC7 neck compared to the Hybrid III.  The SC7 neck incorporates vertical slots at each vertebra to simulate intervertebral compression and this results in a lower initial stiffness in compression.  With both models, the vertical (z) head accelerations were in-phase with the initial head impact and both models exhibited head bounce that was in phase with the lower-neck axial loading as shown in Figure 3-8. The bouncing effect was correctly much less pronounced with the Hybrid III head where the rebound accelerations were closer to 20% to those in the initial  89 impact the impact whereas with our head model, the accelerations were in some cases of similar magnitude although in opposing directions. Our head, while inertially correct, was not adequately damped compared to the Hybrid III head.  However, the compliance of the SC7 neck delayed the load propagation from the head rebound such that the peak lower-neck force was in-phase with the 2nd mode of the head-impact force and thus it did not confound the key injury metric measurement.  With the much stiffer HIII neck, the head rebound loading was transmitted faster which resulted in the peak lower-neck force occurring prior to the 2nd mode of the (head) impact force. In these two experiments, the 1st study was an L9 fractional factorial experiment with only the resolution to provide estimates of main effects. It was used as a “screening” experiment to find the dominant variables in a far lower number of tests than a full factorial experiment. However, the L9 design in particular has some inherent limitations that come at the expense this efficiency and this influenced why it was used as a preliminary and interim study to direct a further full factorial experiment using the dominant variables and not as the study endpoint. The 2nd study served as a “check” on the findings from the first study for the two variables, platform angle and platform stiffness, that were investigated in both experiments and served to allow an objective standalone full factorial investigation of these factors. The full factorial study clearly showed statistically lower peak moments at 15 degrees than -15 degrees, for both padding conditions. The peak moment magnitudes with the SC7 neck differed between the two test series due to the different paddings used.  While the medium padding in the L9 was stiffer (Eeff=90.1 kPa vs. 49.3 kPa) than in the fractional factorial, it was thicker and absorbed more energy before bottoming out.  The findings of the two studies with respect to platform stiffness showed some major differences. In the L9 study, the confounded estimate for the 3-level variable showed the highest lower-neck axial forces with the softest impact platform whereas in the full factorial study with both neck models, the lower-neck axial forces were clearly higher with increasing stiffness. It is an unfortunate limitation that the padding conditions between the two studies were not constant, but the difference between the two studies actually allows an important finding. Plots of the temporal head and neck axial loading with head accelerations for runs 2 and 6 of the L9 compared to runs 4 and 2 respectively in the comparison study are shown in  90 Figure 3-9. These impacts allow a direct comparison of the slightly different “low” and “medium” padding conditions used in the two studies with otherwise identical inputs.    Figure 3-9:  Padding comparison L9 fractional factorial vs. full factorial Temporal head-neck axial forces at the impact platform and lower-neck and head accelerations  with the SC7 head-neck to compare the “soft” and “medium” paddings used in the L9 fractional factorial and the full factorial experiments.  The soft padding condition was the medium padding, plus another thicker low-density layer.  The medium padding in the full factorial study absorbed less energy and gave rise to higher impact forces and head accelerations.  We see that for the low stiffness padding conditions, the much thicker and softer padding dominated the response such that there was no apparent difference in peak impact force, peak head Az acceleration, or lower-neck axial force between the two different “soft” conditions.  The effective Young’s moduli (Eeff) for the two soft paddings was 17.8 kPa and 15.5 kPa for the L9 and the full factorial respectively.  However, when comparing the two “medium” conditions, the peak head forces with the L9 medium padding (Eeff=90.1 kPa) were much lower than in the full factorial study (Eeff=49.3 kPa).  The “medium” padding difference can also be seen in the temporal nature of the two impact forces.  In the L9, while the head force was clearly bi-modal, it was in fact one single impulse whereas in the full factorial study, the head experienced more severe “bouncing” such that a temporal plot of the  91 impact force clearly showed multiple separate impulses.  The medium padding in the L9 study offered more energy absorption than the “medium” padding in the full factorial study where the head bounced more, reached higher peak accelerations, and produced higher forces at the impact platen.  Since the experimental design here dictated that the “soft” padding condition was essentially the “medium” plus another layer of softer padding, this difference seems to explain the differing findings between the two studies for the peak lower-neck axial force. There is likely an optimum amount of energy absorption that padding should have and once there is sufficient absorption to prevent any rebound of the head, further head-energy absorption may place the neck at a greater risk of neck injury if the head is more constrained because of it.  This observation is consistent with the findings about “pocketing” observed by other researchers [34, 86]. In previous head-first impact studies with cadaveric cervical spines it was shown that the degree of head constraint was influential upon neck injury development [34].  Finite element simulations validated against these same cadaveric head-first impacts later showed that increased surface friction was influential upon neck response in a head-first impact. However, friction’s effect on neck injury metrics mainly occurred in moving from a coefficient of friction of 0 to 0.2 and further increases above 0.2 had a much smaller effect [85]. This is consistent with results obtained for a study evaluating an experimental low- friction modification to the exterior of helmets [149] and explains the results observed here in our study. We estimated the static coefficients of friction used in our low, medium, and high conditions (for rigid surfaces) to be 0.27, 0.41, and 0.97 respectively, using a simple inclination experiment. For our lowest friction condition, a weight wrapped in Teflon sheet at rest upon the Teflon-coated impact platform repeatedly began to slide at 22.5º inclination. Given that a maximum impact platform inclination of 15 degrees was used here, the low effect of friction observed in the L9 experiment appears valid. From a constraint perspective, the insufficiently low friction and low platform inclinations were a limitation of our study.  However, since we are interested in neck injury prevention, we desire to study the worst case scenario type impacts consisting of a pre-impact aligned posture with environmental conditions that constrain the head.  Only a very small fraction of head-first impacts that occur give rise to the catastrophic neck injuries this model is being designed to address [8].  This  92 suggests that our worst case aligned spine condition is rare in head-first impacts in these sports.  The effect of oblique platform angles on neck injury metrics was smaller here than reported for the Hybrid III [115] by Frechede at al. In our study the effective mass of the torso did not vary with incident angle but in the Frechede study, the angle of the incoming dummy was changed with respect to a static impact platform and therefore the changes they observed were due to changes in effective mass trajectory as well as any conditions that affected head motion upon impact. In our study, the difference in lower-neck loading was only the effect of changing the head motion with constant torso effective mass and engagement.  The full factorial study provided a fair, statistically powerful, objective, and standalone comparison of the Hybrid III and SC7 head and neck in that it utilized the same experimental apparatus with the same transducers in the same locations. It has been difficult for us to compare our model to published studies with the HIII neck because of the differing experimental conditions and because often loads are reported at the upper-neck for the Hybrid III neck while our SC7 model currently incorporates only a lower-neck transducer.  While the peak lower-neck forces are much higher with both neck models than would be present with biological tissue, they occur at the same temporal point in the impact's neck trace where injuries have occurred in experimental injury biomechanics experiments with full cadavers [75, 78, 108, 150], cadaveric head-neck specimens [34-36, 74], and more recently in our lab, a combination of a Hybrid III head with a cadaveric cervical spine specimen (Chapter 5).  It is common for ATD injury metrics to be far higher due to their infrangible nature than in biological specimens and has resulted in the development of injury assessment reference values (IARV) [82, 151].  Furthermore, the SC7 head and neck model, and in particular using it with peak lower-neck axial force as an injury metric, does appear to be sensitive to changes in end conditions that restrict motion in the sagittal plane during simulated head-first impacts. However, the kinematics of the SC7 neck are much simpler than the human case as this model is not frangible, moves only in the sagittal plane, and furthermore each intervertebral joint is restrained against anteroposterior shear within the sagittal plane. The main biofidelity of this model from a neck injury mitigation standpoint, is capturing the sensitivity of the neck to changes in head constraint. The head constraint is  93 effectively an inertial constraint on the cervical spine as a column. The degree of constraint on the cervical spine has been shown to be influential upon the axial response of human cadaveric cervical and thoracic spine specimens [71, 137] as well as full human cadavers in vertical impact [75] and influenced the injury patterns. Our SC7 model, as a robust ATD neck, is not biofidelic as a column throughout an entire event; rather only in the initial portion of a head-first impact up until the development of a large lower-neck axial load which is commensurate with the temporal location where an injury would have occurred.  The SC7 head and neck model’s temporal axial force response and temporal peak head to neck force ratio against both padded and rigid impacts was in excellent agreement with that observed with human cadaveric cervical spines [34] as discussed in Chapter 1 and a published version of Chapter 1 [119].  The lower-neck loading was also sensitive to changing the incident angle with the impact platform. A perpendicular impact surface prevented any head rotation compared to either oblique surface and this resulted in significantly higher axial forces. In this respect the lower-neck axial force appears to correlate well with the incident angle variable in the sense that more inertially constraining head conditions produced higher loading at the lower-neck. The experimental model considered here was in many ways a worst-case scenario because the neck posture was always nearly aligned and collinear to the torso momentum which was perpendicular (or nearly perpendicular) to the impact surface. The injurious scenarios in sporting and transportation incidents often result in the torso having both translational and rotational components of velocity prior to impact however here the torso (drop tower carriage) had only linear momentum and was further constrained to remain along the incident direction throughout the test. In reality, we speculate based on real-world injury rates [8] that few head-first impacts will have this high of a degree of coalignment between the torso momentum, the aligned cervical spine’s axis, and the normal to the impact surface. Here the carriage representing the torso rides on linear bearings which do allow for up to 3º shaft misalignment such that the carriage is somewhat damped in rotational resistance. However, it would be better to have the head-neck be in free-fall and unconstrained at impact as has been conducted in some early experiments with full cadavers [78] and more recently recreated with full Hybrid III ATD dummys [115]. This is more suited to full manikin drop  94 tests whereas our head-first impact model only represents the torso, head, and neck and is inspired from the cadaveric model developed at Duke University [34].  All of our tests were conducted with an impact speed just over 3.1 m/s which is the minimum speed required to sustain a mid-cervical compressive burst fracture [69]. This is equivalent to only a 50 cm freefall and thus represents the very lower-end of the spectrum for injurious impact velocities. However, the full mass of the torso was arrested by the neck in each drop whereas in real impacts with tangential and rotational components of torso velocity.  The torso momentum is only directed normally to the surface for brief touchdowns such that the neck is rarely tasked with a complete torso arrest. Our limited experimental loading scenarios are likely most representative of head-first impacts occurring in American Football and in Hockey compared to head-first impacts in other sports such as mountain- biking and off-road motorcycling that usually involve significant tangential velocities as well as torso angular velocities. Compromises made in order to design the axial response for the SC7 neck model limit the scenarios over which it can be used. In order to design a high initial compliance for the neck, the articulations incorporated an axial compression at the expense of limiting the model to the sagittal plane. Furthermore, within the sagittal plane, its response is much different than the human cadaveric cervical spine mainly in that our model does not exhibit buckling, even when pre-impact postures and oblique impact surfaces would likely have induced a 1st order buckling, i.e. combined axial-bending response in a human neck. In our model, the pin-constraints and rigid materials preclude anteroposterior shear translations and the hard geometry of each vertebrae provide endpoints to rotation at each level. We recognize this as a lack of biofidelity compared to the human case and argue that this model was designed for only one worst case scenario posture to evaluate experimental devices that change the constraint on the head in the interest of lowering the neck loading as a neck injury prevention strategy.  The model cannot be used to predict specific injuries unless and until we conduct experiments to establish IARVs for it [82, 151]. Its primary purpose is to provide phenomenological insight to compare the severity of different impacts with and without injury prevention devices. Its biofidelic temporal response of axial head and neck loading that is sensitive to impact conditions that affect the degree of head constraints such as incident angle and surface padding support its use in this capacity.   95 The findings in this study are consistent with and complement those of our earlier characterization experiments against a rigid and perpendicular impact platform. While our SC7 head and neck model is still a specialized self-built prototype, it was repeatable and its high initial compliance and axial range of motion offered signifcant biofidelity improvements in temporal force development over the Hybrid III neck for sagittal head-first impacts.  The SC7 lower-neck loading was more sensitive to changes in incident angle that either encouraged or discouraged head rotation.   We would argue that for the purposes of evaluating compressive neck injury prevention strategies and devices, this biofidelic temporal force pattern between the head and lower-neck, improved sagittal plane rotational kinematics, as well as increased sensitivity to impact padding conditions is an advantage over other ATD necks.  This study has provided insight that can be used in future mechanical manikin necks, with the goal of incorporating the high initial compliance characteristics observed here in an omnidirectional model necessary for simulating automotive rollovers. 96 Chapter 4: An Experimental Neck Injury Prevention Helmet: Inducing  Head Motion to Mitigate Neck Loading in Head­First Impacts  4.1 Introduction  Head-first impacts can cause permanent debilitating cervical spine, and spinal cord injuries that devastate the lives of those afflicted and their families. These impacts occur in many different sports where helmets are worn such as football, hockey, equestrian, and the various bicycle and motorcycle disciplines.  It is difficult to estimate the incidence rate of spinal cord injuries from head-first impacts as most epidemiological studies lack sufficient information describing the injury mechanism.  However, some studies have estimated the rates in high school football and hockey to be 0.68 and 2.56 per 100,000 participants respectively [9]. The incidence of cervical spine fractures with or without SCI has been recorded as 0.79 and 0.21 per 100,000 days of snowboarding and skiing, respectively, at the Whistler/Blackcomb ski area in Canada [10]. A study analyzing European motocross racers over a 12 year period, where each race lasts approximately 20-30 minutes, found only 3 cervical spine fractures with accompanying SCI out of 1500 recorded accidents that occurred over an estimated 66,000 hours of riding time [11].  It seems clear that while the incidence rate for these injuries is low, they are still occurring and there is presently no cure for paralysis from spinal cord injuries which occur predominantly in young healthy males [1]. Many researchers studying cervical spine injuries from head-first impacts have found them to be influenced by conditions between the head and impact surface [34, 78, 85, 86, 105].  In an early investigation using football helmet-clad cadavers, the researchers had to artificially create a fusion at the atlantooccipital joint in order for any significant force to be transmitted through the cervical spine [72].   The spherical shape and low friction of the helmet combined with the inherent flexibility of the cervical spines to simply deflect the head out of the path of incoming torso momentum. Even head-first impact studies with dummies wearing helmets [110, 150, 152] have demonstrated that it can be difficult to significantly load the neck axially as many impacts would deflect the head out of the path of the incoming torso momentum unless the head was somehow constrained.  Nightingale and Myers showed that the cadaveric osseoligamentous cervical spine’s axial stiffness in quasi-static loading increased with increasing constraint at the cranial end of the specimens  and that the injury patterns changed from compressive, to compressive- 97 flexion, to non-injurious as they implied full-, rotational-, and no-constraint loading conditions [71].  They later showed with cadaveric head and neck specimens that in a head- first impact near the tolerance impact speed of 3.1 m/s, that the head once stopped, had sufficient inertia to provide a constraining end condition for neck injury development. Some specimens avoided spine injury when the head translated and rotated along a frictionless and inclined surface while other more constrained specimens developed a wide array of unstable fractures [34].   A similar finding was noted in one of the earliest full cadaver drop-test experiments studying cervical spine injuries in vertical impact [75] although using impact speeds of 4.2 to 4.9 m/s.  In this study the cadavers’ head neck complexes were either restrained or unrestrained.  The restrained specimens had a halo ring mounted to the head with spring- mounted steel cables running from the anterior and posterior portions of the halo ring over turnbuckles fixed to the pelvis to stabilize the head as a crude simulation of muscular forces.  It was noted that the heads of unrestrained cadavers moved anteriorly upon impact while the restrained heads did not.  The restrained cadavers developed higher impact forces and more fractures in both the cervical and thoracic spine than the unrestrained cadavers [75]. The relationship between head constraint and spinal injuries in head-first impacts remains undetermined and is a complex dynamical and structural problem.  However, even while this relationship remains unknown, it has been widely observed that non-injurious impacts can occur when the cervical spine as a structure simply fails to assume load thus “escaping” the incoming momentum of the torso [34, 70, 72] even at impact speeds above that required for catastrophic cervical injuries to develop.  Our interpretation of the literature regarding head-first impact suggests that a prevention strategy for compressive neck injury might therefore involve keeping the head moving along the impact surface through an engineered interface to lower constraint on the cervical spine.  We are in the process of designing a helmet to accomplish this task. We are aware of two related studies that induced head motion with an experimental automobile roof during rollover as a neck injury prevention strategy.  The first was performed with a MADYMO multi-body dynamics model [153] and the second utilized a finite element model of a human head and cervical spine in a head-first impact [105]. The roof structure had an asymmetrical spring that deflected towards the front of the vehicle  98 during impact causing an occupant’s head to translate anteriorly. Reductions in cervical spine axial force of 27% and 44% were reported with this roof for perpendicular and -15º oblique (point of contact posterior to head vertex) impacts respectively.  Knowing that this injury prevention strategy is effective and that these injuries often occur in environments where helmets are worn led us to question whether this strategy could be incorporated into sporting helmets.  We are aware of no other attempts to induce head motion in a head-first impact through the use of a helmet. A distinction should be made that we are aware of multiple helmet designs which aim to allow a rotational degree of freedom between the inner liner and the outer shell of the helmet to reduce the rotational acceleration experienced by the brain due to the friction caused during tangential impacts [92].  While this may appear similar, it is a fundamentally different concept.  An inner and outer shell that have nearly identical radii  can only rotate with respect to each other, but the proposed method in this study uses a larger outer shell such that a relative translation and rotation can occur between shells.  The important consequence of this is that in a purely axial crown impact with no tangential velocity relative to the impact surface, the design under consideration here could induce a general spatial motion of the head consisting of both translation and rotation leading whereas a slip-plane helmet could not. The objective of this work was to assess the efficacy of a novel helmet prototype of our design that induces head motion upon impact as a neck injury prevention strategy in head-first impacts over a range of impact conditions.  4.2 Materials and Methods  A custom mechanical neck, head, and ‘helmet’ were used with a drop tower with impact platform capable of adjusting the angle relative to the direction of incident velocity of the drop tower carriage. The sagittal compressive seven segment (SC7) aluminum and rubber neck shown in Figure 4-1, has exhibited biofidelic mechanics under sagittal plane bending and axial impact. We have previously reported on the development of this head and neck [119] (Chapter 2) and also compared its response to the Hybrid III head and neck over a range of platform angles and padding stiffnesses (in Chapter 3).  The lower-neck axial force and sagittal moment were slightly more sensitive to changes in the head constraint with our model than the Hybrid III.  Both models produced reaction forces and moments far higher  99 than has been observed with human cadaveric specimens.  Against stiffer impact surfaces, our model, unlike the Hybrid IIII head and neck, displayed the characteristic “bi-modal” [34, 78, 119, 154] axial temporal force development at the head and lower-neck which became unimodal against softer impact surfaces as has been observed in cadaveric impact experiments. The prototype ‘helmet’ shown in Figure 4-1 was machined of aluminum and Delrin.  As it contained no substantial padding, it was simply an experimental means of inducing head motion at impact that did not provide head protection like contemporary helmets.  Machine drawings of the prototype helmet can be seen in Appendix B.  The prototype helmet mass and inertia were comparable to a conventional football helmet shown in Table 4-1.  100  Figure 4-1:  Mechanical head, neck, and helmet Mechanical head, neck, and helmet (top left inset).  The surrogate head had bi-lateral arc-shaped protrusions (left) that interfaced with conformal slots in the mechanical helmet (lower right).  The slots were made of Delrin and contained a deployment tab (top right – section view) to prevent relative motion in the absence of a vertical force. 101 Table 4-1:  Mechanical helmet mass and inertia comparison  a: Human head mass and inertia from (Walker, Harris et al. 1973) [120] b:  NFL helmet mass from (Viano and Pellman 2005) [38] c:  Football & Aviation helmet properties from (Njus, Liu et al. 1984) [155]  Conceptually, our head-helmet model represents a helmet with two shells, an inner and outer connected through a passive mechanical guide mechanism that consists of bi-lateral arc- shaped protrusions (pins) from the inner shell that interface with slots located on the internal lateral borders of the outer shell. An important difference between the conceptual model described and the prototype helmet under consideration here is that the passive mechanical guide system was directly affixed to the surrogate head precluding the need for an inner shell.  At impact, the head was guided along one of two possible paths relative to the helmet although, in this prototype design, only one escape path was possible in a given impact.  The helmet could either induce a head flexion with anterior translation (FAT) escape, or a head extension with Head Mass (kg) Inertia (kg-cm^2) Humana 4.38 ± 0.59 233 ± 58 SC7 Head (present study) 5.46 209 Helmet Present study prototype 1.38 84 Modern NFL Football helmetb 1.92 n/a Football /Aviator no guardc Riddel Kralite 0.83 69 Army A-1 0.94 64 Bike IV 1.00 87 Rawlings HC 1.00 88 Riddel PAC-44 1.01 88 Mac 100 MH 1.05 97 Rawlings HND-P 1.07 99 Riddel Ace-1 1.10 97 Wil. F-2101 1.10 95 Army SPH9 1.21 89 Rawlings HND-9 1.22 117 Mean +/- S.D.  1.05 ± 0.11  90 ± 14 Football/Aviator with guardc Riddel PAC-44 1.32 149 Rawlings HC 1.35 154 Riddel Ace-1 1.05 159 Rawlings HND-P 1.45 180 Wil. F-2101 1.49 175 Mean +/- S.D.  1.33 ± 0.17  163± 14 102 posterior translation (EPT) escape.  The motion path was the portion of a 121 mm constant radius arc starting 45º from horizontal such that there was 25 mm of horizontal and 19 mm of vertical relative displacement between the head and helmet along with 15º of relative rotation due to the conformal nature of the “pin” and “slot” mechanism.  These values were assigned based on a review of biomechanical literature which is briefly described here.  The average cervical vertebral body depth is 20 mm [124] and it is thought that, when in a flexed posture, the cervical vertebral bodies carry the majority of an axial load [129, 156] and thus 25 mm of imposed horizontal translation (anterior or posterior) nears the dimension of the load-bearing portion of the column itself.  It has also been shown that the human cadaveric cervical spine is very sensitive to postural changes in axial loading.  Postural changes measured as the anterior- posterior eccentricity away from the aligned  “stiffest axis” posture of -5 and 22 mm changed the injury patterns of the cervical spine from vertical compression to compression-extension, and compression-flexion respectively [81].  Also, from a helmet design perspective, it has been widely observed that the mass and inertia of helmets should be as low as possible while providing the necessary head protection.  Thus we felt that these motion path dimensions were likely large enough to have an effect and also as large as possible for incorporation into a pragmatic helmet in order to keep the research relevant. A replicated full factorial experiment (N=36) was performed consisting of 3 platform angles (-15º, 0º, +15º), 2 padding stiffness conditions (soft - Eeff = 15.5 kPA, medium - Eeff = 49.3 kPA), and 3 escapes (FAT, EPT, NH).  The escapes and platform angles are shown in Figure 4-2.  The two platform stiffnesses were described in chapter 3 alongside a rigid impact platen and are named low (or soft) and medium here for consistency.  The lower padding stiffness condition consisted of a 25 mm layer of soft foam (density 31.8 kg/m3, effective Young’s Modulus 22.2 kPa) overtop of two layers of 5 mm PVC yoga mat (density 193.7 kg/m3, effective Young’s Modulus 49.3 kPa) resulting in a series stiffness of 15.5 kPA. The medium stiffness used only the two layers of the same PVC yoga mat.  The paddings were affixed to the steel impact platform with duct tape.  103  Figure 4-2:  Schematic of full factorial experiment The 18 unique runs making up the 3 x 3 x 2 factorial experiment.  The rows represent the three different escapes, namely Flexion-Anterior-Translation (FAT), Extension-Posterior-Translation (EPT), and No-Helmet (NH) and the columns show the three platform angles, -15°, 0°, and 15°.  The run numbers correspond to low and medium platform padding stiffnesses respectively.  Anterior is to the right in all the schematics such that a positive angle moves the point of contact anterior to the head/helmet vertex.  Note the slot direction in runs 1-12.  Even numbered runs (2,4,6,8,10,12) show the EPT escape, odd numbered runs (1,3,5,7,9,11) show the FAT escape. 104 Each drop was conducted from 60 cm to produce pre-impact velocities near 3.2 m/s which is thought to be near the tolerance speed to cervical spine injuries based upon reconstruction of several shallow water diving incidents [69] and on cadaveric experiments [34, 70].  In each impact, a Teflon sheet covered both the head (or helmet) and the impact surface to create a low-friction environment similar to contemporary helmet design.  We measured the static coefficient of friction between two sheets of exemplar Teflon sheet covering a wooden block on top of the rigid impact platform to be 0.27 using a simple inclination experiment.  Each impact was imaged (Phantom V9, Vision Research, Wayne, NJ) at 1000 fps and synchronized with instrumentation consisting of: 6-axis load cell (Denton 4366J, Humanetics, Plymouth, MI, accuracy 1% of FS = 133 N for force and 4.5 Nm for moment) at the lower-neck, a uniaxial load cell (Omega LC 402, Omega Engineering, Stamford, CT, accuracy = 0.1% of full scale output which corresponds to an accuracy of 22.2 N) under the impact surface, a triaxial gyro sensor (IES 3103, accuracy 0.5% of FS = 0.3 rad/s) and 3 uniaxial accelerometers (Endevco 7264C, Endevco, San Juan Capistrano, CA, accuracy 1.2% of measured output ) at the head CoG.  All signals were sampled at 78 kHz with hardware anti-alias filters set nominally at 1000 Hz which spectral analysis showed to be closer to 1100 Hz.  These values were specific to our equipment and were explained in section 3.2.3.  Fiducial markers on each vertebra, head, and helmet were tracked for calculating 2D kinematics. A schematic of the test apparatus and the positive coordinate systems used for the main sensors are shown in Figure 4-3 and Figure 4-4 respectively. Three-way factorial ANOVA (single dependent outcome variable with factorial independent variables) was performed separately on five injury metrics under consideration.  For all injury metrics, all statistical operations were performed on the average value of the replicated runs in the factorial experiment.  Peak lower-neck axial reaction force and sagittal moment were used as neck injury metrics while peak head accelerations, Head Injury Criteria (HIC 15), and peak head angular accelerations were also considered for comparing head injury potential. In order to visualize the combined effect on both neck injury metrics, a modified variant of the N_IJ and was used where N_IJ =  Fz/Fc + My/Mflex.  The method for calculation at the upper-neck with the Hybrid III calls for using different intercepts for flexion and extension responses.  For our purposes of comparing the three helmet conditions, we instead used the absolute value for both force and moment and used Fc = 8500 N and Mflex = 310 Nm.  Fc was the average lower- 105 neck load measured against a rigid surface with a neck follower load [119] and Mflex was the flexion reference value specified for use with the Hybrid III dummy.  If this model were to be used to predict injury rather than to compare severity between multiple impacts, it would be important to establish Injury Assessement Reference Values (IARVs) [151].  While we do not have an estimate for the Mflex IARV with this model, the Fc of 8500 N was determined from drop tests conducted at or near the established tolerance impact speed for axial compressive cervical spine impacts determined from diving injury reconstructions [69] and cadaveric drop testing using a model similar to this one [34] and is the closet estimate of a compressive IARV for our model currently available.  However, it is important to note that these intercepts apply only to this model and therefore these N_IJ values cannot be compared to those with the Hybrid III where it is normally calculated at the atlanto occipital joint and which are based upon IARVs determined for the Hybrid III head and neck.  The N_IJ here provides a normalized linear combination of the two neck injury metrics calculated at each temporal point.  It is important to remember that this experimental ‘helmet’ did not contain any significant energy-absorbing padding and hence explains why comparisons between the two helmet escapes (FAT and EPT) and the no-helmet (NH) condition were deemed appropriate instead of comparing to a helmet that did not induce head motion for both head and neck injury metrics.  If the helmeted escapes also contained energy absorbing padding these comparisons would be confounded. The null hypothesis was that there would be no difference in peak lower-neck loading or head injury parameters between the two helmet escapes or without the helmet.  Post-hoc multiple comparisons between the three head escapes were made for each unique combination of platform angle and padding, and for each injury metric, using paired t-tests and using the method of False Discovery Rate (FDR) to adjust the significance values [157].  This “modified Bonferroni” method was used instead of the more traditional Bonferroni correction that controls the family- wise error rate (FWER) by dividing alpha by the number of comparisons.  The FDR method orders the raw p-values (Pi ) (from t-test comparisons) from lowest to highest and then moving sequentially from i = 1 to m, for m-comparisons significance is met under the condition that ?? ? ??? ?  . Moving through the sequential p-values, the last comparison to satisfy the condition is the adjusted critical p-value for significance in the post hoc testing.  This method is less  106 conservative and a more powerful alternative to multiple comparisons testing than the Bonferroni method when a large number of comparisons is made [157].   107  Figure 4-3:  Drop testing schematic Mechanical surrogate head, neck, and helmet with a flexion anterior translation (FAT) escape on free standing drop tower.  108  Figure 4-4:  Coordinate systems for head and neck instrumentation Positive coordinate systems for mechanical parameters for the lower-neck load cell and head linear accelerometers and the triaxial angular velocity gyroscrope.  Note that the lower-neck load cell shows the direction of positive reaction forces/moments.  4.3 Results  In general, one or both of the FAT or EPT helmet escapes changed the kinematics of the head in all of the drops compared to the no helmet (NH) tests.  Snapshots from the high speed video for all the runs can be seen in Appendix C.  Figure 4-5 shows the head rotation and head CoG horizontal translation plotted against time for 18 tests (1 per run) comprising the 6 unique combinations of platform padding and angle.  The temporal kinematic patterns observed in Figure 4-5 displayed good repeatability for all the replicated runs.  The variability of the kinematics for the duplicated runs can be seen alongside the average values of induced head translation, head rotation, and neck eccentricity at the time of peak neck force in Table 4-2.  The changing head kinematics also affected the kinematic response of the surrogate neck.  The induced head motion increased the eccentricity of the neck at the time of peak load.  This was defined by the horizontal locations of the C7/T1 and C0/C1 articulation points where a positive  109 value indicated that C0/C1 was anterior to C7/T1.  This definition was adopted to this model and based upon that used by researchers at the Medical College of Wisconsin [81].  In general, larger head horizontal translations increased neck eccentricities.  Table 4-2:  Head and neck kinematic measurements (and range) for duplicated experimental runs  110   Figure 4-5:  Head rotation and horizontal translation Solid lines indicate the head angles and dashed indicate the head CoG horizontal translation.  Positive quantities are flexion and anterior translation.  Black lines show Extension Posterior Translation (EPT), blue lines show Flexion Anterior Translation (FAT), and red lines show No Helmet (NH).  Time zero is first head or helmet contact with the impact surface and each test was plotted just up until any carriage (torso) rebound.  The stars on each curve indicate the point in time where the peak lower-neck axial force occurred.  111 Table 4-3:  Mean peak kinetic impact parameters (and range) by run  Peak Values  Mean +/‐ SD Run 1 ‐571  ± 19 6  ± 163 5726  ± 131 8626  ± 268 ‐1.6  ± 23.8 80.9  ± 1.6 ‐8.6  ± 0.7 1  ± 24 ‐6  ± 58 ‐47  ± 4 54  ± 2 53  ± 5 0.93  ± 0.02 Run 2 ‐833  ± 31 ‐130  ± 22 8349  ± 114 9015  ± 38 ‐12.3  ± 0.4 47.4  ± 3.3 ‐1.5  ± 10.0 ‐24  ± 2 ‐21  ± 1 89  ± 255 180  ± 126 99  ± 47 1.10  ± 0.00 Run 3 ‐1145  ± 224 269  ± 100 7067  ± 40 11257  ± 71 ‐30.3  ± 0.6 107.5  ± 0.7 ‐18.5  ± 0.2 39  ± 14 62  ± 4 ‐95  ± 4 99  ± 6 150  ± 10 1.15  ± 0.01 Run 4 ‐1135  ± 29 ‐4  ± 231 9967  ± 94 11604  ± 31 ‐2.0  ± 14.6 82.2  ± 0.5 ‐18.2  ± 1.4 ‐43  ± 13 ‐47  ± 24 ‐134  ± 8 137  ± 6 280  ± 37 1.31  ± 0.01 Run 5 ‐531  ± 12 ‐24  ± 155 5913  ± 109 10582  ± 126 ‐9.9  ± 5.5 ‐0.1  ± 30.4 ‐5.8  ± 2.8 ‐54  ± 7 69  ± 3 ‐89  ± 2 100  ± 2 147  ± 32 0.75  ± 0.00 Run 6 ‐852  ± 44 82  ± 290 7341  ± 596 8788  ± 380 ‐15.8  ± 11.8 ‐5.6  ± 53.6 9.7  ± 2.5 11  ± 43 51  ± 45 ‐84  ± 30 101  ± 6 115  ± 31 0.92  ± 0.06 Run 7 ‐671  ± 50 ‐22  ± 177 6389  ± 203 12033  ± 421 ‐13.1  ± 0.1 23.6  ± 0.1 0.2  ± 6.6 ‐41  ± 17 3  ± 153 245  ± 7 250  ± 0 561  ± 39 0.78  ± 0.02 Run 8 ‐1044  ± 40 248  ± 22 9260  ± 37 10418  ± 110 ‐11.2  ± 0.3 2.3  ± 74.1 7.9  ± 4.1 35  ± 4 ‐84  ± 38 51  ± 278 198  ± 71 451  ± 186 1.12  ± 0.01 Run 9 952  ± 45 37  ± 301 5823  ± 24 6338  ± 228 ‐1.5  ± 12.3 ‐123.8  ± 3.6 0.4  ± 11.6 4  ± 45 5  ± 45 ‐65  ± 5 67  ± 5 91  ± 7 0.80  ± 0.00 Run 10 1545  ± 26 ‐217  ± 13 3378  ± 207 7247  ± 542 ‐25.1  ± 9.0 ‐186.8  ± 1.8 9.2  ± 3.1 ‐20  ± 2 1  ± 39 ‐73  ± 13 77  ± 13 50  ± 5 0.69  ± 0.00 Run 11 94  ± 1514 ‐309  ± 130 7822  ± 1348 7015  ± 762 ‐28.2  ± 19.0 ‐123.6  ± 3.2 1.1  ± 18.6 ‐53  ± 11 22  ± 150 75  ± 219 162  ± 104 394  ± 288 1.06  ± 0.15 Run 12 1852  ± 10 ‐210  ± 2 3419  ± 36 9079  ± 26 ‐14.4  ± 0.7 ‐233.7  ± 2.7 ‐10.5  ± 0.5 27  ± 1 46  ± 162 184  ± 96 204  ± 123 210  ± 75 0.85  ± 0.01 Run 13 ‐2482  ± 242 282  ± 72 8524  ± 8 10955  ± 352 ‐25.1  ± 7.6 235.5  ± 6.1 ‐4.2  ± 20.6 44  ± 11 ‐23  ± 1 ‐73  ± 9 75  ± 12 100  ± 29 1.75  ± 0.00 Run 14 ‐2940  ± 71 306  ± 23 9002  ± 73 16642  ± 760 ‐24.3  ± 0.3 270.8  ± 0.9 ‐12.2  ± 1.1 45  ± 3 ‐26  ± 4 ‐333  ± 8 333  ± 8 417  ± 9 1.65  ± 0.01 Run 15 ‐1244  ± 73 389  ± 85 9891  ± 48 11393  ± 53 ‐32.1  ± 8.7 63.4  ± 4.9 ‐5.8  ± 0.1 58  ± 12 ‐27  ± 9 ‐78  ± 1 90  ± 9 156  ± 18 1.29  ± 0.02 Run 16 ‐988  ± 25 514  ± 36 10724  ± 167 18542  ± 738 ‐53.3  ± 4.4 65.0  ± 2.2 ‐8.1  ± 2.0 75  ± 5 ‐46  ± 8 ‐373  ± 14 374  ± 14 593  ± 10 1.35  ± 0.02 Run 17 29  ± 1435 ‐30  ± 370 7624  ± 27 8756  ± 262 ‐32.6  ± 8.4 ‐206.1  ± 1.0 ‐8.5  ± 3.5 ‐3  ± 52 ‐25  ± 3 ‐66  ± 0 71  ± 1 123  ± 16 1.41  ± 0.00 Run 18 1250  ± 25 367  ± 11 8003  ± 27 13562  ± 914 ‐36.1  ± 4.3 ‐189.9  ± 1.5 ‐10.6  ± 3.2 53  ± 1 ‐27  ± 8 ‐250  ± 18 251  ± 18 326  ± 8 1.31  ± 0.01 Mz (Nm) Ax (G's) Ay (G's) Az (G's) Ares (G's) HIC15max NIJ maxFx (N) Fy (N) Fz (N) Fhead (N) Mx (Nm) My (Nm) 112 Table 4-4:  ANOVA tables for all injury metrics  113 All of the average impact parameters for the 18 runs (based on 36 tests) are presented in Table 4-3.  The full ANOVA tables for all injury metrics including p values, and eta squared (percent variation) are shown in Table 4-4.  The mean peak lower-neck axial forces ± 1 standard deviation (based on two measurements thus equivalent to the range) are shown in Figure 4-6.  For the axial force injury metric, the ANOVA showed that the helmet escape main effect along with the interaction between platform angle and escape were significant   Against 15 degree platforms which moved the contact point anterior to the vertex of the head, the EPT escape showed significantly smaller lower-neck peak axial forces than a FAT escape, or with no helmet (NH).  Against -15 degree platforms that produced contact points posterior to the head vertex, only the FAT escape produced significantly smaller peak lower-neck axial forces than NH.  Against 0 degree (perpendicular) impact platforms, both escapes reduced peak lower-neck axial forces over NH with the FAT escape being slightly more beneficial than the EPT escape at lowering peak neck axial forces.  Figure 4-6:  Factorial results – lower-neck axial force Solid lines indicate the lower stiffness padding and dashed indicated the higher stiffness.  Red  lines show No Helmet (NH), black lines show Extension Posterior Translation (EPT), and blue lines show Flexion Anterior Translation (FAT). Error bars indicate the range for replicated runs.  114 The ANOVA for peak moment returned significant F tests for all main effects and interactions.  The peak lower-neck sagittal moments without directionality are shown in Figure 4-7.  Against perpendicular impact platforms, the FAT escape produced significantly lower peak moments than NH or with an EPT escape at both padding stiffnesses.  The EPT escape was also beneficial over NH with soft perpendicular impact platforms, but with the stiffer padding it provided no significant difference from NH.  Against -15 degree platforms, both the FAT and EPT escapes were significantly beneficial over NH using both padding stiffnesses although there was no difference between the FAT and EPT escapes.  However, at 15 degrees, only the FAT escape provided a reduction in peak moment compared to NH.  The EPT escape was either not significantly different from NH, or in one case against the stiffer padding, the peak moment was significantly higher than NH.  Figure 4-7:  Factorial results – lower-neck sagittal moment Solid lines indicate the lower stiffness padding and dashed indicated the higher stiffness.  Red lines show No Helmet (NH), black lines show Extension Posterior Translation (EPT), and blue lines show Flexion Anterior Translation (FAT). Error bars indicate the range for replicated runs.     115 The ANOVA for NIJ returned significant results for all effects except the platform angle and stiffness interaction.  The N_IJ shows that mainly due to higher moments, especially without the helmet, the impacts against -15 degree impacts were more severe than 0 degrees or 15 degrees and that, at all platform angles, one or both escapes reduced lower-neck injury metrics compared to tests without the helmet (NH).  At -15 and 0 degrees and with both paddings, the FAT escapes significantly lowered NIJ over both NH and the EPT escape.  Against 15 degree platforms and both paddings, both the EPT and FATescapes lowered NIJ over NH although there was no significant difference between the FAT and EPT.    Figure 4-8:  Factorial results – lower-neck N_IJ Solid lines indicate the lower stiffness padding and dashed indicated the higher stiffness.  Red lines show No Helmet (NH), black lines show Extension Posterior Translation (EPT), and blue lines show Flexion Anterior Translation (FAT). *This metric was adapted from the N_IJ used with the Hybrid III dummy however was calculated with different intercepts and cannot be compared to values reported for the Hybrid III dummy. Error bars indicate the range for replicated runs.   116 The peak resultant head accelerations were strongly dependent upon the surface stiffness and are shown in Figure 4-9. The ANOVA returned significant F tests only for the surface stiffness and helmet escape variables and their interaction.  Against the softer platform at all angles, there were no significant differences among any of the FAT, EPT or NH escape conditions for either linear head injury metric. However, against the stiffer impact platform, at -15 and 0 degree angles both FAT and EPT escapes significantly lowered peak resultant head accelerations over NH; at 15 degrees, only the FAT escape was significantly lower than NH.  Figure 4-9:  Factorial results – peak resultant head CoG acceleration Red lines show No Helmet (NH), black lines show Extension Posterior Translation (EPT), and blue lines show Flexion Anterior Translation (FAT). Error bars indicate the range for replicated runs.  Similar to the head peak accelerations, the HIC15 linear head injury metric was strongly dependent upon surface stiffness and is shown in Figure 4-10.  The ANOVA returned significant  117 F tests for platform angle and stiffness, and their interaction but not for helmet escape or any interactions involving helmet escape.  The only significant corrected matched pairs t-test was that the FAT escape produced lower HIC15 scores than NH at -15 degrees with the stiffer impact surface.  Figure 4-10:  Factorial results – Head Injury Criterion Red  lines show No Helmet (NH), black lines show Extension Posterior Translation (EPT), and blue lines show Flexion Anterior Translation (FAT). Error bars indicate the range for replicated runs.  For the five injury metrics discussed, the ANOVA analyses indicated where the helmet escape factor and/or any 2-factor interactions with the helmet escape were significant and matched pairs t-tests were used to uncover which helmet escapes significantly affected the  118 respective injury metrics.  The raw probabilities from the matched pairs t-tests along with the adjusted p-critical based on the method of False Discovery Ratio [157] and percent differences between the three helmet escape conditions are shown in Table 4-5. We unfortunately had problems with our angular velocity channels.  Our particular sensors do not appear to be well-suited for closed-head impact applications.  While it did not occur in all tests, in the majority of impacts with the helmet, the angular velocity channels became saturated at impact (implying 3600 rad/s) through a time period before observable head rotation.  This precluded their inclusion in the study.  119 Table 4-5:  Factorial ANOVA, post hoc multiple comparisons for head and neck injury metrics  *Significant difference using method of False Discovery Rate (Benjamini, Y. and Y. Hochberg 1995) [157] with an alpha of 0.05.  The adjusted P critical is the cutoff used to indicate significance.   % difference comparisions:  Run X vs Run Y, a negative percentage indicates X lower than Y †Modified for lower-neck loading with model.  Calculated with normalizing intercepts of Fc = 8500N and My = 310 Nm for both flexion and extension.  Not comparable to NIJ for Hybrid III upper neck loading or indicative of whether injury would have developed. Angle, Padding Paired T‐Test Description % diff raw p % diff raw p % diff raw p % diff raw p % diff raw p Run 2 EPT vs Run 13 NH ‐2% 0.1639  ‐80%*  0.0007*  ‐37%*  0.0000* 140% 0.3605 ‐1% 0.9793 Run 1 FAT vs Run 13 NH  ‐33%*  0.0011*  ‐66%*  0.0008*  ‐47%*  0.0003* ‐29% 0.1223 ‐47% 0.1490 Run 1 FAT vs Run 2 EPT  ‐31%* 0.0022*  71%*  0.0060*  ‐16%*  0.0071* ‐70% 0.2912 ‐47% 0.3020 Run 6 EPT vs Run 15 NH  ‐26%*  0.0264* ‐40% 0.0601  ‐29%*  0.0134* 12% 0.2915 ‐26% 0.2449 Run 5 FAT vs Run 15 NH  ‐40%*  0.0005*  ‐66%*  0.0069*  ‐42%*  0.0005* 11% 0.2736 ‐6% 0.7579 Run 5 FAT vs Run 6 EPT ‐19% 0.0794 ‐43% 0.0983 ‐18% 0.0565 ‐1% 0.8473 28% 0.4143 Run 10 EPT vs Run 17 NH  ‐56%*  0.0012*  ‐9%*  0.0056*  ‐51%*  0.0000* 8% 0.5991 ‐59% 0.0265 Run 9 FAT vs Run 17 NH  ‐24%*  0.0002*  ‐40%*  0.0010*  ‐43%*  0.0000* ‐6% 0.3326 ‐26% 0.1267 Run 9 FAT vs Run 10 EPT  72%*  0.0036*  ‐34%*  0.0020*  17%*  0.0005* ‐13% 0.4115 81% 0.0205 Run 4 EPT vs Run 14 NH  11%*  0.0075*  ‐70%*  0.0000*  ‐21%*  0.0003*  ‐59%*  0.0012* ‐33% 0.0368 Run 3 FAT vs Run 14 NH  ‐21%*  0.0009*  ‐60%*  0.0000*  ‐30%*  0.0002*  ‐70%*  0.0008*  ‐64%*  0.0013* Run 3 FAT vs Run 4 EPT  ‐29%*  0.0006*  31%*  0.0006*  ‐12%*  0.0023* ‐28% 0.0235 ‐46% 0.0414 Run 8 EPT vs Run 16 NH  ‐14%*  0.0068* ‐19% 0.0452  ‐17%*  0.0032* ‐47% 0.0757 ‐24% 0.3958 Run 7 FAT vs Run 16 NH  ‐40%*  0.0018*  ‐64%*  0.0015*  ‐42%*  0.0011*  ‐33%*  0.0060* ‐5% 0.3753 Run 7 FAT vs Run 8 EPT  ‐31%*  0.0026*  ‐55%*  0.0064*  ‐30%*  0.0022* 26% 0.4143 24% 0.5019 Run 12 EPT vs Run 18 NH  ‐57%*  0.0000*  23%*  0.0025*  ‐35%*  0.0003* ‐19% 0.6468 ‐36% 0.1611 Run 11 FAT vs Run 18 NH ‐2% 0.8674  ‐35%*  0.0015* ‐19% 0.1413 ‐35% 0.3555 21% 0.7721 Run 11 FAT vs Run 12 EPT 129% 0.0438  ‐47%*  0.0007* 24% 0.1846 ‐21% 0.7491 87% 0.4752 Adjusted P critical 0.0264 0.0069 0.0134 0.0060 0.0013  -15°, Soft  0°, Soft  15°, Soft  -15°, Med  0°, Med  15°, Med Post Hoc Multiple Comparisons Fz My NIJ† Ares HIC15 120 4.4 Discussion  The objective of this work was to assess the efficacy of inducing head motion upon impact as a neck injury mitigation strategy for head-first impacts over a range of simulated impact conditions using mechanical surrogates for the head, neck, and an experimental helmet which did not contain padding.  The helmet induced either a head flexion with anterior translation (FAT) or a head extension with posterior translation (EPT) upon impact and experiments were also performed without the helmet (NH) over 3 platform angles and 2 platform stiffnesses. We observed a strong interaction between the chosen escape direction and platform angle for all three neck injury metrics.  For the perpendicular impacts, both escapes reduced neck injury metrics compared to NH with the FAT escape being slightly preferential to the EPT escape.  For the angled platform impacts, the largest reductions in axial force were achieved when the induced head translation motion was in the same direction that the angled platform would have caused naturally.  The FAT response was preferable for perpendicular and posterior- to-vertex impacts which in this experiment were the negative platform angles and the EPT escape was desirable for anterior-to-vertex impacts.  Some of the runs with the helmet in the factorial experiment were thus “preferred” escapes while others were “unpreferred” and we will refer to them as such herein.  In this study runs # 1 through 12 were helmeted and of those runs (1,3,5,7,10,12) were the preferred escapes (Figure 4-2). The preferred escapes were more successful at inducing head motion as the geometry acted to increase the obliqueness of the impacts which deflected the head and changed the head rotation pattern compared to tests without the helmet. For the oblique platform impacts, the induced head rotation was initially in the opposite direction as compared to NH.  For example, referring to Figure 4-5, for -15° impacts the pre-impact head posture was approximately 3°- 4°of head flexion for FAT and NH drops respectively.  In the NH drops, the head rotated into approximately 5° extension post-impact whereas the FAT escapes rotated into a maximum angle of approximately 12° flexion before undergoing extension rotation back to approximately 5° of flexion. In most cases the preferred escapes were able to prolong the impacts where the time frame was from first head or helmet contact to first visible rebound motion of the carriage (torso) as can be seen in Figure 4-5.  The slower deceleration of the torso resulted in reduced neck injury  121 metrics.  This was most evident for the preferred EPT escape for anterior-to-vertex impacts (positive angles in this study) where the lower-neck axial impulse had a much flatter shape reaching lower peak values over a longer time duration and resulted in a maximum axial force reduction of 56% over NH (Figure 4-11).  In these same particular impacts the preferred EPT escape had a positive effect on the combined loading NIJ metric from lowering both Fz and My moments during the axial force impulse and also by slightly separating their peak values in time (Figure 4-12) compared to the same conditions with NH.  However, these impacts also highlight the one case where we significantly increased the peak sagittal moment, after the axial force impulse, by 23%.   Figure 4-11:  Temporal head and neck axial force development comparison run 12 EPT vs run 18 NH The black lines are the EPT escape and the grey lines are the NH test.  Dashed lines are the impact force, solid lines are lower-neck axial force.    122  Figure 4-12:  Temporal lower-neck axial force and sagittal moment comparsion run 12 EPT vs run 18 NH The black lines are the EPT escape and the grey lines are the NH test.  Dashed lines are lower- neck axial force(Fz) and solid lines are lower-neck sagittal moment (My) where flexion is positive.   Caution must be used in interpreting the head injury metrics in this study in terms of preferred impacts.  The numeric HIC values or peak accelerations do not have established IARVs similar to the N_IJ metric as mentioned previously.  They cannot be used to predict whether a brain injury would or would not have occurred but instead only to make a relative comparison between impacts.  The strong interaction observed between platform angle and helmet escape for the neck injury criteria was not observed for either head injury metric and ANOVA analyses for both peak resultant head acceleration and HIC15 were non-significant.  Despite this, we still conducted the multiple comparisons t-tests as there was high variability with a select few tests which might have been hiding significant differences between helmet escapes in other impact scenarios.  In general, for preferred-escape impacts where significant reductions in neck injury criteria were uncovered, the corresponding linear head injury metrics were smaller or not significantly different from NH.  From a neck injury prevention standpoint, the lack of difference in linear head injury metrics from NH, preferred, or unpreferred impacts is promising.   123 While this study was focused upon the effect of inducing head motion upon neck injury metrics, in general head injuries are a larger problem for society than neck injury and thus it is paramount to ensure that any potential neck injury mitigation strategy, especially one like this for an injury type that is in fact quite rare, does not reduce the protection against head injury which is a helmet’s primary role. A recent study of head impacts incurred in high school football conducted with the HITS (Simbex, Lebanon NH, USA) telemetric helmet showed that the lowest angular accelerations occurred for impacts to the crown of the helmet [158].  Since inducing head motion may raise angular accelerations, they must be kept within injury tolerance levels.  It is clear that while the neck injury mitigation strategy seems promising, this metric must be carefully evaluated in the future.  It has recently been demonstrated through similar drop testing experiments that reduced friction interfaces encouraging head motion in oblique impacts can mitigate peak head angular accelerations in some impacts, but also exacerbate them in others [149].  Based on this and what we have observed here, we believe there exists some optimal path for moving the head to minimize neck reaction forces within the constraints of angular brain injury tolerance. We are not aware of any other helmets that have been proposed to mitigate neck injuries in head-first impacts.  In fact, the literature contains many studies showing that in fact helmets do not appear beneficial for neck injury mitigation.  Some of the earliest studies using cadavers, were studied with 7 different football helmets and found them to have little if any effect the severity of the neck injuries produced [82].  Similar findings have been documented using ATDs [159].  There has in fact been a series of two papers based upon a lumped parameter model showing that the helmet cannot possibly absorb the incoming energy of the torso and thus have concluded that no realistic helmet, i.e. one small enough to actually be worn, could influence neck injuries in a head-first impact [103, 160].  We certainly agree that if the neck must fully arrest the torso, especially in high-speed action sports, that it is unlikely to be able to prevent neck injury beyond the tolerance speeds.  However, a careful examination of some of these real- life impacts in football shows that the complicated dynamics involved expose the cervical spine to torso momentum only in brief chaotic head to ground impacts or tackles.  Often the direction of velocity of the athlete is not aligned axially with the neck and force is only directed along the axis of the neck briefly before the rotating and translating torso begins to contact the ground.  It has been observed that the human cervical spine reacts in a much stiffer manner when loaded  124 axially, developing high forces and injuries over small displacements, when subject to full constraint but when unconstrained can undergo much larger deflections through a combination of compression and bending without injury [71].  We believe that in such brief head to ground impacts or head to object impacts as occur in many of the sports where head-first impact spinal injuries occur, it could be of great benefit to the cervical spine to reduce head constraint over these short durations.  The Duke University cadaveric head-neck drop tower model testing confirmed that the neck can be injured at a point in time where the neck is forced to manage the rebounding inertial load of the head in addition to that of the incoming torso [154].  Another benefit of inducing head motion as we have done with this helmet is to control the deceleration of the head to prevent head rebound. There has been a history of devices, mostly only seen in patents [102, 161-163] that act to connect the helmet to the shoulders such that in a crown impact the device provides an alternate load path to the relatively more fragile cervical spine such that load is deflected onto the torso.  More recently, a prototype of one of these devices has been reported in the literature and shown to greatly reduce the reaction forces at the upper-neck compared to a Hybrid III dummy either without a helmet or wearing a contemporary helmet without any neck protection [101].  However, in their experiments their rigid device was then rigidly affixed to a rigid mechanical         Ā2   Ũ Ũ  é           é    éĀ     | concept the real human fleshy shoulder girdle is far more compliant.  In the most dangerous aligned posture [3], the human cervical spine was shown to have a highly nonlinear stiffness characteristic of an initial soft response attributed to compression of intervertebral discs that becomes much stiffer when the bony components of the spine dominate the response.  A study of 20 cadaveric specimens showed that up until a mean compression of approximately 12 mm, the mean stiffness was approximately 33 N/mm at which point it significantly increased to 555 N/mm and the spines failed at a mean compression of 18 ± 3 mm [33].  A simplified conceptual schematic of the alternate load path provided by a linkage between a helmet and shoulders is presented in Figure 4-13.  Analyzing this mechanical system shows that the series combination of the bony clavicle, shoulder soft tissue, garments, and the prevention device which is in parallel with the neck, must be comparably as stiff, or stiffer than the neck over the range of physiological neck compression in order to shield load away from the neck. While we are not aware of the stiffness of the in-vivo shoulder girdle, even the stiffest component of the series  125 chain, the clavicle, has a stiffness in 3-point bending along the anterior-posterior direction of 95 N/mm [164] which is 5.8 times less than that reported for the aligned cervical spine after it has been compressed by 12 mm.  If it were assumed that the soft tissue of the shoulder girdle and shoulder padding (as worn in football and hockey) bottomed-out over the same 12 mm as the low-stiffness region of the neck, then the next stiffest element in the series chain alternate load path, the clavicle, would dominate its response.  Now the effective stiffness of the neck and clavicle in parallel would be 651 N/mm and the neck would still be subjected to 85% of the applied load.  While this is a very simplified argument, we feel it is conservative and demonstrates the significant challenges that this alternate load path approach faces.  These challenges are further magnified if the alternate load path device is not in contact with the helmet prior to the impact which is the case with several neck collars designed to limit maximum head rotation such as the Leatt Brace™ mentioned in Chapter 1 which is more advanced, but fundamentally similar to football collars.   Figure 4-13:  Helmet-shoulder coupling load path conceptual schematic Simplified schematic representation of parallel spring elements representing the neck (left) and a series combination of shoulder pads and anatomic elements (right) between the head and shoulders.  Football collars are neck injury prevention devices that act to connect the helmet to the torso through an interfacing between the underside of the helmet and the top of an energy  126 absorbing neck collar that sometimes attaches to shoulder padding.  These collars are intended to limit the range of motion of the head to within physiological limits.  One of three football neck collars tested was shown to reduce the axial force transmitted through the neck by 34% in axial crown loading when the football shoulder pads were raised to be in closer proximity to the base of the helmet prior to impact [99].  However, this study utilized the Hybrid III dummy and the compliance of the shoulder girdle is not known, although we believe it is much stiffer than a human.  A helmet to neck collar load-path faces the same load-shielding challenges described above.  We find it hard to imagine that any coupling device which would allow for sufficient mobility to allow for unimpeded play could have an effective overall stiffness high enough to shield the neck from axial loading.  The reduction of range of head motion offered by these devices may help in some impacts that would cause hyperextension or hyperflexion but they could also add constraint to the head and dissuade it from bending out of the way to help the neck “escape” the momentum of the torso.  In addition, the reduced range of head motion may impede the athlete’s ability to see which could have other dangerous consequences.  The counter-intuitive strategy embodied in our experimental helmet is to promote a more compliant neck response by removing inertial constraint imposed upon the neck by the relatively massive head if suddenly stopped.  The results here suggest that if the head can be arrested in a more controlled manner with only slightly more ride-down, that neck loading parameters could be significantly reduced even while not exacerbating or in some cases further mitigating linear head acceleration parameters as well.   Perhaps a head-motion-inducing helmet combined with a neck collar or other helmet- torso connecting device, could provide further reduction in neck loading by allowing a greater vertical head displacement (i.e ride-down) to account for the relative compliance ratio of the shoulder girdle to the cervical spine over the small displacements that compressive neck injuries occur.  The extra vertical displacement offered by a 2-shelled helmet could allow the torso soft tissue the deflection to effectively “bottom-out” and allow the next lowest stiffness element in the series connection to dominate the load-shielding.  We hypothesize that this would be the clavicle.   In this study it was shown that the preferred deployments resulted in higher neck eccentricity at the point of peak loading than in the equivalent no-helmet tests.  In perhaps the most direct study of eccentricity on axial loading of the human cadaveric head and neck [81], it  127 was shown that the injury type changed from compressive to flexion type injuries when the eccentricity was 1.5 cm (anterior) or larger.  Eccentricities less than 1.5 cm produced compressive fractures. In the same study when they grouped their results by fracture vs no fracture, the eccentricities for each group were 0.6 and 5.2 cm respectively. Using this mechanical neck model, we can only speculate about the effect that this added eccentricity induced by the helmet escapes might have on human necks, but we hypothesize that it may be beneficial by helping to promote a combined bending-compression response instead of purely compressive.  The axial force reductions with the FAT escape at 0º and -15º were similar to those reported for an experimental deformable automobile roof structure [105]. Halldin reported a 27% reduction in axial load at T1 against a 0° roof and a 44% reduction against a -15° roof (posterior to vertex contact) whereas we found a 40% reduction against a perpendicular surfaces and a 29% and 33% reduction for the medium and soft paddings respectively.  A difference was that in the Halldin study the entire head and neck were rotated 15 º relative to a fixed roof orientation such that the effective mass of the torso was more eccentric relative to the cervical spine, whereas in our model, the platform angle was adjusted while the incident velocity stayed constrained to the vertical such that the effective mass was constant.  Halldin did not present moment magnitudes in his paper but did suggest that at one point in the simulation, that an atlantooccipital dislocation was evident, although it was not stated how this conclusion was reached.  Although our non- frangible metallic spine may suggest higher load reductions than would be present with biological tissue, the mechanics of reducing human spine loading would be similar. Our SC7 mechanical neck provided an accurate temporal head-neck force development pattern over the time period where neck injuries have been shown to occur. The axial loading of any beam-column trapped between two large masses (the head and torso), will depend upon the dynamics of those masses in an impact.  The head and helmet models used in this study had realistic mass and inertial properties although the surrogate head mass was nearly the largest value ever documented that we are aware of [165].  Based on another study that published the mean and standard deviation of the human head (4.38 ± 0.59 kg) [120] our head (5.46 kg) would be at the 94th percentile.  The surrogate head mass and inertia were originally matched to average human values [120] however modifications to the head to allow for interaction with the SC7 neck and the prototype helmet increased its mass and inertia.  From an impulse-momentum  128 perspective, the added head mass adds to the already biased mass ratio between the human head and a helmet making these tests somewhat conservative from the standpoint that a less massive head would have also undergone induced head motion from the helmet. Our drop tower carriage weight of 14.2 kg was modeled after the Duke University model [34] and represents only the upper portion of the mass of the 50th percentile torso [34, 154].  Our helmet prototype had sagittal mass properties that were consistent with football helmets [155] and within operating limits deemed suitable for high G environments experienced by helicopter pilots [166].  Our SC7 head and neck model displayed a head and neck temporal force development pattern at the head and lower-neck that is very similar to that observed with human cadaveric testing [119] up until the peak lower-neck axial load. Beyond this point the neck’s kinematic response was likely not biofidelic as it is not frangible and does not buckle the same way a human neck would.  This model does not help uncover the true buckling response of the human cervical spine, but for the timeframe of first head contact up until the point of peak lower-neck axial load development, its response appears accurate enough to be used to investigate, and design injury-prevention mechanisms that deal with the inertial effect of altering head motion at impact during a head-first impact. Our line of thought is that the development of the helmet itself can be advanced with this mechanical model, but that any such design would then have to be further tested with a cadaveric model of head-first impact alongside finite element and multi- body dynamics models for efficient optimization and miniaturization. Although this model demonstrated positive results, they must be interpreted with caution in light of this study’s limitations.  This conceptual model is different from an actual helmet in several important ways, the most obvious being its lack of any padding for head protection.  While the helmet prototype is inertially correct and of similar mass to contemporary helmets, it was machined from aluminum and used large lateral gussets, as shown in Figure 4-1, to provide exceptional rigidity to resist the bending moment created by the pin/slot interface which would appear necessary to ensure that head motion is induced.  Consequently, this helmet prototype was over-designed and much stiffer than even the stiffest contemporary helmet shells.  This has important considerations given the relative mass ratio between the helmet and head. From an impulse-momentum standpoint, using a helmet to alter the path of the much more massive head is certainly a challenge.  In this study the mass ratio between head and helmet was nearly 4 to 1, and thus the helmet’s rigidity certainly was important in deflecting the path of the head upon  129 impact.  Considering that a constraint in helmet design is keeping overall mass and inertia as low as possible, it could prove challenging to design a sufficiently stiff outer shell.  This helmet characteristic, i.e. a stiffer outer shell, could be dangerous to other athletes in contact sports.  A recent study analyzing concussions in American professional football found that the crown of the helmet was used to deliver the blows that caused 61% of the subset of video recorded helmet to helmet concussions [38].  In their study they found that the same injury mechanism responsible for the majority of catastrophic neck injuries for the striking player [3], i.e. a lowered-head spearing type tackle with the crown of the helmet, was also responsible for the majority of concussions from helmet to helmet hits in the struck player.  The difference of course is the mass ratios between the collisions.  When the striking player in the aligned posture contacts a larger mass, i.e. the torso of another player, the head is stopped and rebound head loading combines with torso loading to compress the cervical spinal column beyond its capability in the striking player.  The greater engagement of torso mass on behalf of the striking player, when impacting another player’s helmet, delivers a larger acceleration to the struck player.  This iteration of our head-helmet model did not use a headform with biofidelic overall shape.  Thus the motion-mechanism which would be connected to a user’s head via a chinstrap in an actual 2-shelled helmet was rigidly attached to the head here.  It seems likely that there would be some rotation between the head and inner shell as head-helmet movement has been observed and chinstrap roll-off is an element of helmet design [167].   Additionally, without padding the overall size seems here seems reasonable even with the 25 mm space between the head and helmet prior to impact.  An actual 2-shelled helmet would presumably need to be 25mm taller than contemporary helmets in order to accommodate the same parameters of induced head motion although perhaps specialized padding could be designed to reduce this somewhat.  Furthermore, our head-neck model is limited to the sagittal plane and as such, the helmet shape presented here was cylindrical rather than spherical.  This geometry ensured that the impacts remained planar which in turn, ensured correct operation of the sliding mechanism in all tests reported.  All of these issues are being addressed in our laboratory through the design of subsequent 3D prototypes designed to work with the 50th percentile Hybrid III head. The testing thus far has been over a very limited loading scenario where the incoming velocity remained vertical with respect to a stationary impact platform.  Many of the cycling sports and transportation contexts where these injuries occur involve significant tangential  130 velocities, and we are aware of methods to incorporate tangential velocities [92]. There are important variables still to be considered such as impacts in the presence of lateral bending or axial rotation, realistic helmet retention, and the effect of tangential impact speed among the different sports where head-first impacts occur.  It may be that this strategy is more effective in impacts where the main component of velocity is axial as in hockey or football rather than motor sports although as mentioned above, it could pose potential dangers in contact with sports with head-to-head impacts.  It was an unfortunate but major limitation that our angular velocity sensors proved un- suited to direct head impact loading (as opposed to inertial head loading without a direct contact) with this model in all but a handful of tests.  We acknowledge the importance of closely monitoring the angular accelerations experienced by the head as an indication of potential mild traumatic brain injury or concussion.  We have since upgraded to using a 9-accelerometer array in the Hybrid III head in our newer models to accurately monitor the effect of induced head motion on head angular accelerations. We are unaware of other attempts to induce head motion in a head-first impact through the use of a helmet for neck injury mitigation.  This work, while conceptual, demonstrated that a helmet with realistic mass and inertia can alter head kinematics in head-first impacts and reduce the axial force passing through the neck if a preferred deployment occurs.  The results here suggested that a selector mechanism capable of deploying in the proper direction would be required as non-preferred deployments were not mitigating and in some cases aggravating to neck injury metrics.  The results here, while limited, are encouraging and warrant further exploration. We have since developed more realistic 2-shelled, three dimensional spherical helmet models that work with the Hybrid III 50th percentile head and incorporate realistic head- retention through a chinstrap.   We are actively developing these models using our mechanical surrogate neck and working towards testing them with cadaveric cervical spines and a Hybrid III head in the presence of muscle force replication and a biofidelic surrogate spinal cord. 131 Chapter 5: A New Model of Head­First Impact using Aligned Cervical  Spine Specimens, the Hybrid III ATD Head, and Neck Muscle Force  Simulation  5.1 Introduction  A core challenge in the field of Injury Biomechanics stems from the inability to study real-life injurious scenarios in human subjects for the very obvious ethical reasons.  Therefore, researchers must develop models to attempt to first understand injury mechanisms and then design interventions for prevention.  Neck injuries from head-first impacts have been studied using full cadavers [72, 75, 78, 108], cadaveric head and neck specimens [34, 74], anthropomorphic test devices (ATDs) [82, 110-113, 115], and computational models [85, 86, 105, 168].  All of these types of models have their benefits and their weaknesses.  Computational models, although promising and although they have made important contributions in the Injury Biomechanics field previously, require extensive validation and are unlikely to ever produce the range of injuries that can develop in response to a single injury mechanism.  ATDs provide useful feedback on passive safety device performance but whether or not they have sufficient biofidelity for meaningful injury outcome conclusion in head-first impacts is a contentious issue among researchers [119, 169, 170].  Full cadavers or cadaveric head-and-neck specimens can be difficult and expensive to obtain and their use in impact biomechanics experiments is forbidden by some tissue banks and anatomy departments on ethical grounds. One advantage that cadaveric models have over other types is that the material properties and osseoligamentous anatomy are human. This is why they are used to validate computational models under development.  Until the Injury Biomechanics community has developed and validated a predictive digital model of a full human being that is biofidelic in all potential loading scenarios, it seems that cadaveric models will have utility in this field.  Thus it is vital to continually improve the biofidelity of the cadaveric models alongside the digital models especially if they are going to be used for injury prevention research. A central weakness with all cadaveric models is the lack of both passive and active musculature.  This is especially critical for studying dynamic compressive injuries to the cervical spine.  The cadaveric cervical spine, devoid of musculature, is a very slender and unstable column that buckles under approximately 10 N, which is 4 times less compressive load than the average sized human head normally applies just due to gravity [40].  It is readily observable  132 anecdotally that the in vivo cervical spine can carry vertical loads far in excess of this by considering head stands performed by dancers, acrobats, yogis, and also from published literature.  African porters carried up to 70% of their body weight (while on a treadmill) on their head in one study [48], and in another Nepalese porters carried up to 183% of their body mass using namlo head straps [47].  This would seem to establish musculature plays an instrumental role in stabilizing the cervical spine. The effects of muscular stabilization on neck response in head-first impact are currently unknown and the literature on this is contradictory and at best inconclusive.  Not simulating the effect of cervical musculature in cadaveric models of head-first impact has been justified by the observation that the cervical spine is injured in 2-20 ms after impact, well below the activation time required for cervical musculature and before the head undergoes any significant motion [34].  However, there is no reason, especially in sporting collisions, that the cervical musculature could not be activated prior to impact and furthermore some studies have suggested that the influence of muscle tone may increase the risk of neck injury development.  An early full- cadaver drop test experiment that used a halo ring attached to the head to simulate a “constrained” condition representing muscle tone was found to greatly increase impact forces and to produce more serious spinal injuries [75].  More recently a finite element computational model of the head and cervical spine that included muscular contraction showed that full muscle activation doubled the risk of vertebral fracture in a head-first impact compared to passive contraction based on principle vertebral strain [91].  This suggests that not only does tensed musculature increase the compressive load through the column [44], the muscles “hold” the neck in a more stabilized posture due to an increased bending stiffness.  This increased stability could contribute to the column responding axially developing neck forces sufficient for fracture as opposed to a combined bending-compression response. The cervical spine carries approximately 75 to 150 N of compression in static postures, as calculated through a combination of experimental disc pressure measurements and disc cross sectional area [16, 45, 46]. Compressions as high as 1164 N (at C4/C5) have been predicted by muscle force models under a maximal extension contraction [44].  All of these observations suggest that it is vital that cadaveric models simulate the effect of muscle forces in some fashion if true kinetic and kinematic biofidelity is desired.  133 One experimental method of providing some stabilization of the cadaveric cervical spine is the compressive follower load concept [41].  The central idea behind this strategy is to apply a compressive load that is guided near the approximate center of rotation [42, 43] at each vertebral level such that the compression is applied in a manner that minimizes any shear forces or moments which would act to bend or buckle the cervical spine. This method has been shown to achieve gains in stability as evidenced by preventing buckling under 250 N of applied compression, which is almost 6 times the weight of the average human head, and 25 times the critical load observed without a follower load [41]. While a significant improvement, this load magnitude was modest compared to loads predicted to exist in vivo [44].  Since then, other researchers have developed more advanced methods for simulating muscle forces with the cadaveric cervical spine that more closely recreated an in vivo response to flexibility testing and thus displayed improved biofidelity [136, 171].  These methods were later applied to study dynamic whiplash scenarios where the response of the cadaveric specimens was similar to the motion response displayed by in vivo experiments with volunteers [172, 173].  These last 2 studies by Ivancic et al. highlight a progression in their model of cadaveric whiplash with muscle force replication that was necessary to study the injury mechanism in more realistic surroundings.  The first study [172] developed a benchtop model of cadaveric cervical spine whiplash that created realistic kinematics but which could not be used in an actual rear-end automobile simulation due to the way the muscle forces were applied and also because the mass representing the head did not have a biofidelic shape.  The second study [173] expanded upon these methods to work with the full body and head of a Hybrid III dummy to make the model suitable for use on a larger acceleration sled in order to evaluate active seatbacks as a whiplash injury prevention device.    The motivation behind this study is similar.  We are developing a neck injury prevention helmet for head-first impacts [138], the proof of concept prototype was presented in Chapter 4, and our lab’s existing cadaveric cervical spine model, hereinafter referred to as the Saari model [35, 36, 174], for head-first impact was not suitable for evaluating helmet prototypes for various reasons (to be discussed) without modifications.   The Saari cadaveric cervical spine model for studying head-first impact was the first instance that we are aware of where a follower load was used to stabilize the cervical spine in dynamic compressive impact.  This model utilized a mechanical surrogate head with human  134 cadaveric spines (C0-T2) and a radioopaque surrogate spinal cord [35, 36] and is shown in Figure 5-1.  The surrogate head was the same one used and described in Chapters 2-4 which had the same mass and sagittal inertial properties as the human head [120].   Figure 5-1:  The Saari et al. model of head-first impact Our lab group’s previous model of cadaveric head-first impact utilizing a cadaveric cervical spine from C0-T2, a radioopaque spinal cord, and a mechanical surrogate head.  The spines were stablized by bi-lateral follower load cables where the guides were provided by small eye hooks screwed into the vertebral bodies.  The pre-impact posture was neutral-lordotic and the pre- impact head position was held in place using fishing line that went taut at impact.  (Saari et al, 2006, 2007)[35, 36, 174]   The Saari et al. (2006, 2007, 2011) model of cadaveric head-first-impact with a mechanical surrogate head produced extension injury patterns consistent with a 1st order extension buckling response and captured the interaction between bony fracture and spinal cord compression.  While the model produced clinically observed injuries, it had some shortcomings.  In particular, the model did not produce any mid-cervical (C4 to C6) burst or compressive  135 fractures which are the most common injury in young collision-sport athletes or diving athletes after sustaining a head-first impact [3, 69].  We hypothesized that if the pre-impact posture would have been aligned instead of lordotic, and the spines stabilized using a muscle replication system, that perhaps compressive injuries might have developed.  It has been noted by other researchers that this aligned posture was a requisite for experimentally producing axial compressive burst fractures [33]. Another shortcoming of the previous model related to the surrogate head used.  In particular, only the crown region of the head had a biofidelic shape and thus our previous model could only be used against perpendicular impact platforms with sagittal spinal postures.  Even within the sagittal plane, a significant pre-flexed or pre-extended head posture, either with or without an inclined impact platform, would cause an inaccurate response due to unrealistic head geometry beyond the crown region.  Furthermore, the crude shape of the surrogate head precluded its use with a conventional spherical helmet (although it was used with the 2- dimensional helmet prototype described in Chapter 4).  Another limitation precluding us from using our previous cadaveric model with a helmet (either spherical or our 2D prototype) was the means of positioning the head and neck into the desired pre-impact posture.  This was accomplished by applying forces directly to the head via fishing line along non-physiologic directions which is similar to methods that have been commonly used by other researchers with cadaveric models of axial compression [33, 34, 78].  While it was not necessarily a limitation of their models considering their objectives, it was for us considering our overarching goal of testing a neck injury prevention helmet with cadaveric cervical spines. To overcome these issues, the objective of this study was to create a new model of head- first impact with aligned full cadaveric cervical spines, a 50th percentile Hybrid III biofidelic spherical head, and an advanced neck muscle force replication system that applied stabilizing neck forces along physiologic directions, for the purpose of creating clinically relevant experimental compressive cervical spine injuries and which would be suitable for use in the presence of three-dimensional neck injury prevention helmets under development in our lab.  The goal for these tests was simply to assess the biofidelity of this new model, and in particular the kinetics of head and lower-neck loading that relate best to the rest of this thesis, by comparing its response to other published models.   136 5.2 Materials and Methods  Five fresh-frozen human cervical spines (T2-C0, where C0 represented an approximately 8 x 10 cm portion of the occipital bone) were used for this study. DEXA scans in the lower thoracic and upper lumbar spinal region were performed to assess bone mineral density on all specimens.  In addition, pre-test lateral and coronal x-rays were taken before specimen preparation to screen for any deterioration or abnormalities beyond that expected due to age- related degeneration alone.  The mean age of the donors was 79 ± 12 years and 3 of the 5 specimens were male.  Specimen age, gender, DEXA score and cause of death are shown in Table 5-1. Table 5-1:  Cervical spine specimen information   The occipital bones of each specimen were attached to a 50th percentile Hybrid III (HIII) dummy head by way of two custom-designed mounting plates that replaced the Denton 1716 HIII upper-neck load cell and sandwiched a Delrin cylinder with two O-rings to seal the head. The head needed to be sealed against saline water that we applied to the specimen to keep it hydrated and other miscellaneous biological fluids. The superior plate was mounted on the inside of the HIII head and was used to mount the instrumentation (to be discussed) placed at the head CoG while the exterior plate on the inferior surface of the HIII head provided a flat surface for mounting the occiput.  This external plate had 4 protruding clevis pins that served as anchoring points for the follower load and muscle force strings. Machine drawings locating the anatomical occipital condyle on the HIII ATD head/neck interface (Denton 1716A) were used as reference to accurately position the center of the C0/C1 joint of our specimen.  We used the center of the transverse processes on C1 to estimate the anterior-posterior position of the center of the C0/C1 joint capsule.  In order to maintain approximately physiological coronal and sagittal alignment of the foramen magnum while securely mounting the specimen in such a way to avoid artifactual fractures from stress risers Specimen Age Sex Cause of death Thoracic BMD (g/cm2) Bone Quality Classification H1220 60 m Rt lung cancer 0.668 Osteopenic (0.648 ‐ 0.833 g/cm2) H1221 76 m Brain cancer  0.714 Osteopenic (0.648 ‐ 0.833 g/cm2) H1222 87 m Hrt failure 0.690 Osteopenic (0.648 ‐ 0.833 g/cm2) H1223 87 f Metatastic cancer  0.602 Osteoporotic (< 0.648 g/cm2) H1224 85 f Pneumonia 0.613 Osteoporotic (< 0.648 g/cm2) 137 associated with fixation screws through the occiput, the center of the C0/C1 joint capsule was placed approximately 1 cm more inferior relative to the head CoG than it would be in situ.  The portion of the occiput was trimmed to maximize the contact area within the fixed surface area available on the inferior surface of the HIII head and to maintain optimal alignment.  The occiput was then temporarily mounted to the HIII head using 3 screws passed through the occiput.  One was placed anterior to the foramen magnum near the sagittal plane and was used to wrap a copper wire around to fix to the anterior portion of the outer HIII outer mount plate.  The other two were machine screws placed bi-laterally to secure the occiput to the threaded posterior portion of the exterior adapter plate.  Four additional screws were also passed through each occiput as anchor points for the posterior muscle forces (to be discussed).  Once it was confirmed that these fasteners were aligned with the threaded holes in the exterior HIII mounting plate to provide optimal head-spine alignment, the spine specimens were then removed.  At this point, just prior to potting the inferior end of the specimens, a previously validated elastomeric radioopaque biofidelic surrogate spinal cord [175] with elliptical shape (major and minor diameters of 11.7 and 6.4 mm respectively) was inserted into the spinal canal.  The cord was tethered at the inferior border of T2 and at the occiput via a wire passed transversely through the cord with approximately 5% tensile strain. At this point, the inferior end of the specimens (T2 and the inferior half of the T1 vertebra) were potted in the casting cup such that the anterior margin of the T1 vertebral body was vertically oriented in the casting cup using a laser alignment system to ensure proper alignment in the coronal and sagittal planes.  After the casting had cured, but before the head was mounted ‘permanently’, a lateral x-ray was taken with the specimen in the casting cup.   The HIII head was then mounted using the aforementioned fastening methods through the pre-drilled and pre-aligned holes in the occiput.  The trimmed and aligned occiput created a large teardrop- like cavity against the head which was then tightly filled with epoxy putty (Magic Bond) around all the fasteners to distribute compressive load.  This completed anchoring of the superior end of the specimens to the head. At the inferior end of the specimens, the base of the casting cup used in casting was replaced with an 11” x 7” x 0.5” thick aluminum plate.  This plate located the inferior end of all the inferiorly terminating muscle forces (excludes splenius capitis) and the follower load strings.  A schematic showing the muscle forces and the testing apparatus is shown in Figure 5-2.  138  Figure 5-2:  Testing schematic Schematic of test environment including apparatus as well as applied muscle forces.  The plate also served to connect to the drop tower carriage via a six-axis load cell.  Each string was passed through this plate and tied to a compression spring-screw assembly such that an easily measurable and known tension force could be applied independently to each string to stabilize, apply compressive follower load, and further straighten the cervical spine specimens.  In total there were 15 strings terminating in spring-screw assemblies at this inferior plate; 12 represented muscle forces and the other 3 were the bilateral and anterior follower load cables.  The three follower load cables represented deep cervical muscles such as the longus colli muscles and each required being run over two pulleys in order to clear the casting cup.  These pulleys were mounted on slotted angle brackets providing fine-tuning adjustability along 2  139 directions.  The remaining 12 cables represented the inferiorly terminating muscle forces which included the trapezius, sternocleidomastoid, and the infrahyoid muscle group all of which were applied bilaterally.  In addition, the semispinalis capitus muscles were modeled with four extension springs that anchored superiorly on the aforementioned four additional screws that passed through the occiput and inferiorly attached to the spine at levels C3 through C6.  The method of anchoring the muscle forces and follower load guides to the vertebrae was improved over our lab’s previous model [35, 36, 174] by avoiding the use of any screws into bone which could act as stress risers and create fractures not representative of those observed clinically.  Each muscle force attachment and follower load guide was applied using the same braided fishing line that applied the muscle forces. The fishing line in this context was fashioned into a harness around the vertebral body and posterior elements as shown in Figure 5-3.  Figure 5-3:  Schematic of muscle force vertebral anchoring Method used to avoid stress risers when attaching or guiding muscle force or follower load strings to a typical cervical vertebra.  Fishing line, shown in red, was tied anteriorly through the transversium foramen and posteriorly around the vertebral arches.  Clamshell clamps, shown in blue and split rings shown in green were used to guide the lateral follower load (LFL) and anterior follower load (AFL) strings and to attach the Trapezius (TRAP) and Semispinalis Capititus (SemCap) muscles.  (Image from Van Toen et al, 2009)[176]   To anchor the follower load guides, a string was passed bi-laterally through each foramen transversarium and tied such that it was tight to the anterior margin of the vertebral bodies.  This allowed for mounting clam-shell rings which guided the follower load strings.  The semispinalis capitis and trapezius muscles were connected posteriorly to the vertebrae in a similar fashion by  140 passing bi-lateral strings around the lamina on the vertebral arches.  The muscle force replication system also incorporated a flexion limiter system between C0 and C1 to prevent hyper- physiologic flexion rotations at the upper cervical spine that have been observed in other cadaveric models and would be prevented from occurring in vivo from the chin contacting the chest [172]. The muscle forces were not constant among specimens but instead adjusted as necessary to create a lordosis-removed, axially-aligned spinal posture which we hypothesized to be a requisite for incurring mid-column compressive burst fractures [33].  We attempted to place the anterior-posterior center of the T1 vertebral body over the anterior-posterior center of the occipital condyles to achieve zero eccentricity [81].  After achieving an aligned posture on the bench, the entire specimen and casting cup plate was mounted to a 14.2 kg carriage via a 6-axis load cell on our custom built four-rail drop tower.  The final postural adjustments were made by adjusting the muscle forces with the inverted specimen hanging on the drop tower carriage.  Lateral and anterior pre-test photographs of the completed model hanging on the drop tower are shown in Figure 5-4. The mass of the casting cup, load cell, and impact carriage was approximately 17 kg to represent the mass of the upper human torso that is thought to act on the neck during a head-first impact [34].  The impact platform was aligned perpendicularly to the incoming velocity and covered with 2 layers of 5 mm PVC yoga mat (density 193.7 kg/m3, effective Young’s Modulus 49.3 kPa) and then a 1/32” sheet of gummy neoprene rubber (30A) to create a high friction, high- constraint surface.  The static coefficient of friction between a leather cap on the surrogate head and this gummy neoprene rubber was measured to be 0.97 and the friction between the Hybrid III head wrapped in nylon stocking was likely near to this or in fact higher.  Before each drop, the (surrogate) head and neck specimens were lowered until the head just made contact with the impact surface and then adjustable mechanical stops were set 4 cm below the carriage based on the protocol from our previous model [35, 36]. This distance ensured that enough displacement was allowed to injure the spine but that the torso carriage was halted before the spine was subjected to unrealistic displacements or injuries.  Each drop was conducted from a height of 60 cm to achieve an impact speed near 3.1 m/s which is the estimated threshold for compressive burst fractures [69].  141 Each impact was captured with two high speed video cameras (Phantom V9, Vision Research, Wayne NJ, USA) at a resolution of 1632 x 1200 pixels and at 1000 frames per second.  Instrumentation included a 6 axis lower-neck load cell (Denton 4366J), a uniaxial load cell underneath the impact platform (Omega LC 402-5K), and 3 uniaxial accelerometers (Endevco 7264C) at the HIII head CoG.  These 10 image-synchronized channels were sampled at 78 kHz and filtered with pre A/D anti-aliasing filters to provide a low-pass cutoff frequency of approximately 1100 Hz.  The sign conventions for the instrumentation are shown in Figure 5-5. A single trigger was used to synchronize both video cameras, the DAQ system, and a high speed x-ray system.  Kinetic calculations were performed with a custom Matlab  (Mathworks, Natick, MA) program that included digital filtration using a bi-directional 4th order Butterworth filter with 1000 Hz low-pass cut off frequency.  The reaction moments at the lower-neck load cell, along with appropriate reaction forces, were resolved to present the moments at the approximate centroid location of the C7-T1 intervertebral disc according to the free body diagram shown in Figure 5-6.  Although not shown in Figure 5-6, the lateral bending moment was also resolved to this location using the lateral (Fy) shear force.  The disc centroid locations were estimated using pre-test lateral x-rays taken with the specimen already in the casting cup having known dimensions (in 4 of 5 specimens, estimated from photographs in H1224).  The directionality of the reported Mx and My moments is consistent with that shown occurring at the load cell as the convention shown in Figure 5-5 and should be interpreted to mean the moment exerted by C7 onto T1.  The high speed x-ray images were analyzed frame by frame to determine the maximum spinal cord deformation.  Post-impact, each specimen was imaged with both x-ray and CT (Xtreme CT) and injuries were diagnosed by a spine surgeon (JS) using both the CT and dissection.   142   Figure 5-4:  Photographs of model pre-impact Anterior (left) and left side (right) lateral views of the new HIII-AMFR cadaveric model of head-first impact prior to an impact.   143  Figure 5-5:  Coordinate systems for instrumentation and data analysis  The head and neck loading as well as torso (drop tower carriage) motions were analyzed for each drop by viewing the high-speed video footage.  The Phantom V9 cameras were controlled by manufacturer-supplied software (Phantom 640, Vision Research, Wayne, NJ) which synchronized the analog signals captured with the video and allowed each (or all) signal(s) to be viewed with the video images.  The onset of load at the impact platen load cell was used to mark the beginning of the event and the maximum torso displacement and the point at which it began to rebound were determined from the high-speed images.  The carriage impact velocity was estimated from the high-speed video by tracking a point on the casting cup throughout the last 2, 3, and 4 frames of video prior to impact.  This initial velocity estimate was used alongside the head Az acceleration trace to integrate the head Az signal to estimate the head velocity trace throughout the impact.  144  Figure 5-6:  Free body diagram to resolve moments at C7/T1 Free body diagram for resolving sagittal moment from load cell centroid to the centroid of the C7/T1 disc.  The load cell, casting cup, and aluminum plate have known dimensions.  The sagittal moment at C7/T1 (My_T1) is affected by both the axial force (Fz) and the anterior-posterior shear force (Fx).   5.3 Results  All 5 specimens sustained serious injuries from these impacts.  In four of five specimens (H1221-1224) the spine rapidly began a first-order extension buckling response while the fifth specimen (H1220) underwent a purely axial response.  The spinal cord was successfully imaged in all impacts and the maximum cord compression observed was near 20% compression.  The detailed results of the cord strains and how they relate to column injuries will be presented elsewhere in a first-author publication by co-author Claire Jones.  In all specimens, the spine was successfully aligned using the adjustable muscle forces with peak segmental compressions of approximately 180 N.  Also, the method of attaching the muscle forces and follower loads  145 prevented any fractures associated with stress risers.  A more detailed account of the development of the advanced muscle force system, and a quantitative analysis of the muscle forces that were applied across the segmental levels will be presented elsewhere in a first-author publication by co-author Carolyn Van Toen. For all of the impacts, the neck response remained predominantly in the sagittal plane as evidenced by the high-speed video and this was also confirmed by the lower-neck load cell reactions.  The peak out-of-plane shear forces (Fy) were much smaller than the in-plane shear forces (Fx) and axial forces (Fz).  Similarly, the lateral bending moments (Mx) and axial rotation moments (Mz) were much smaller than the sagittal moments (My).  The head accelerations also showed this trend with the vertical accelerations (Az) being much larger than the anterior- posterior (Ax) which were also much higher than the out of plane accelerations (Ay).  The peak values for these channels along with other impact parameters are presented in Table 5-2.  A representative temporal plot showing all of the lower-neck reaction forces and moments (at C7/T1) as well as head accelerations for specimen H1220 is shown in Figure 5-7.  Equivalent plots for the other specimens are presented in Appendix D.   146  Figure 5-7:  Representative temporal plot of lower-neck forces, moments, and head accelerations Lower-neck forces and impact platen force (top), head CoG Accelerations (middle), and lower- neck moments (bottom) for specimen H1220.  Mx and My moments are those calculated at the approximate C7/T1 disc centroid location.   147 Table 5-2:  Impact parameters for the five drop tests Test # Impact Speed (m/s) Time Lag (ms) Fx (N) Fy (N) Fz (N) Fhead (N) Mx (Nm) My (Nm) Mz (Nm) Ax (G's) Ay (G's) Az (G's) Ares (G's) H1220 3.0 2.5 ‐571 ‐234 3838 7925 17.0 ‐81.8 ‐6.8 10 ‐6 140 141 85 H1221 3.1 2.4 473 119 2302 7439 19.8 84.0 ‐4.6 ‐22 ‐6 123 125 112 H1222 3.2 1.5 ‐893 176 3756 6525 18.2 58.6 7.6 23 7 144 144 144 H1223 3.1 2.3 ‐473 ‐132 1446 6651 ‐16.7 89.5 7.5 ‐20 2 117 117 113 H1224 3.2 2.3 757 ‐162 1411 5924 15.1 101.1 ‐7.5 ‐21 ‐4 98 98 102 Mean 3.1 2.2 ‐141 ‐47 2551 6893 10.7 50.3 ‐0.8 ‐6 ‐1 124 125 111 SD 0.1 0.4 715 182 1193 790 15.4 75.4 7.7 21 6 19 19 22 Peak Forces and Moments Peak Head Accels (G's) HIC_15 max 148 All of the specimens displayed an initial lag time [34-36] between head and lower-neck axial force development and the average lag time was 2.2 ± 0.4 ms.  The peak axial neck forces in all specimens occurred just after the peak impact platen force associated with the stopping of the head and occurred while the neck was being loaded inertially by both the torso and head masses.  The torso loaded the neck by continuing its cephalad momentum and the head loading was evidenced by a concurrent negative (cephalad)  Z acceleration (Az).  The major difference in the temporal patterns between the two distinct column responses was that in the axial response (H1220), the lower-neck axial force was also bimodal and developed a second compressive peak.  In contrast, with the extension specimens (H1221-1224), after the peak compressive neck force occurred, it rapidly dropped below zero and became tensile as the impact carriage contacted and eventually rebounded off the mechanical stops and began to move upwards.  A qualitative description of the impacts and the temporal nature of their kinetics that is consistent with the rest of the thesis, alongside the resulting neck injuries will be presented here:  H1220:  A sequence of images from the high-speed video is shown in Figure 5-8.  Upon impact the head stopped immediately as evidenced by high-speed video, the impact platen load cell, as well as the head z acceleration.  After a lag of approximately 2.5 ms, axial compression began to develop at the lower-neck.  The impact force showed a bimodal shape where the 2nd mode had two local maximums as shown in Figure 5-9.  The first mode was primarily due to head loading as evidenced by the head z acceleration with no concurrent axial force at the lower-neck.  The 1st and 2nd local maximums on the 2nd mode of the impact platen force corresponded with the maximum torso compression (as evidenced from the high speed video) and a 2nd mode of head loading respectively as evidenced by the head z acceleration trace.  From analyzing the video images, the 2nd  local impact platen force maximum was also the point at which the torso began to rebound after impacting the mechanical stops and thus definitively marked the end of the useful portion of the test.  The axial force history at T1 also showed an overall bi-modal shape where the 1st mode had 3 localized peaks.  The 2nd localized peak of the 1st mode corresponded with the peak axial neck force of 3838 N for this drop.  The head z acceleration trace confirmed that the peak neck force was due to concurrent loading from head rebound (negative or cephalad  149  Figure 5-8:  High speed video frames for specimen H1220 direction), in addition to torso momentum, at this point.  Based on the high speed video, the first localized peak of the 1st mode of neck force appeared to correspond to a fracture of C1 which subsequently initiated and propagated further at the 2nd and 3rd local maximums.  After the 3rd localized peak, the T1 axial force dropped rapidly.  The carriage reached its maximum downward compression at the T1 force local minimum (as evidenced from high-speed video) signifying the start of the 2nd mode of T1 axial force.  This 2nd mode of T1 neck loading produced no further fractures.  Throughout the impact duration the head rotated very little.   150  Figure 5-9:  H1220 kinetic analysis plot Axial force, impact platen force, and C7/T1 sagittal moment (top) and head z acceleration and head z velocity (bottom) for specimen H1220.   The C7/T1 sagittal moment was primarily an extension reaction moment that developed before any lower-neck axial load development.  A review of the raw load cell outputs showed that this initial extension moment was due to the contribution of an anteriorly directed Fx shear reaction force with its large lever arm (marked “B” in Figure 5-6) which developed with the impact platen force.  After this point, the My extension peaks were in-phase with the lower-neck axial force peaks and highlight the axial force contribution to the C7/T1 extension moment.  This was due primarily to resolving moments about the C7/T1 disc centroid which was anterior to the load cell centroid such that a purely axial compressive force would also produce an extension moment about this location. Upper cervical spine injuries included a 3-part burst fracture of C1 to the posterior and anterior ring, a transverse ligament rupture without any odontoid process involvement, as well as a basilar skull fracture of the left occipital condyle. Mid-column injuries included a bi-lateral  151 incomplete facet capsule injury, ligamentum flavum tear, as well as anterior disc injuries all at the C5/C6 level. H1221:  Immediately upon impact, the cervical spine began to assume an extension-compression bending response as shown in Figure 5-10.  Figure 5-10:  High speed video frames for specimen H1221 As T1 was fixed, this required a large flexion rotation at C7/T1 with extension bending at all superior levels that created a head extension rotation.  The axial loading developed at T1 after approximately a 2.4 ms lag.  The impact platform displayed a bi-modal force pattern while the T1 neck force history was unimodal and developed a peak force of 2302 N and can be seen in Figure 5-11.  152  Figure 5-11:  H1221 kinetic analysis plot Axial force, impact platen force, and C7/T1 sagittal moment (top) and head z acceleration and head z velocity (bottom) for specimen H1221.  This peak neck force again occurred at a time when the head was rebounding from, but still in contact with, the padded surface as evidenced by the negative head z acceleration and positive impact platen force traces.  The high-speed video and integrated z direction head velocity show that the head was moving caudally at the time of peak lower-neck force.  It was not obvious from the high-speed video where the injuries developed temporally in this test.  The impact carriage continued its downward compression well past the peak T1 neck force approximately until the peak of the 2nd mode of the impact platen force as evidenced from high-speed video with the synchronized analog force traces.  This test showed that just after impact, an initial extension reaction moment developed at C7/T1, which peaked at the onset of lower-neck axial loading.  This was again due to an anteriorly directed Fx shear reaction force.  From this point onwards, the sagittal reaction moment was a flexion moment in good agreement with the flexion rotation observed at C7/T1 from the high-speed video.  153 Post-test dissections revealed a C7 compressive fracture to the inferior endplate of the vertebral body and also at C7/T1 there was ligamentous damage associated with hyper- physiologic flexion rotation.  This consisted of a complete rupture of the interspinous ligaments and ligamentum flavum as well as a bi-lateral facet dislocation.  Other extension injuries in the column included spinous process fractures at C3 and C4. H1222:  A sequence of images from this impact is shown in Figure 5-12.  Figure 5-12:  High speed video frames for specimen H1222 Immediately upon impact the spine began an extension bending response similar to H1221 with head extension rotation as well as anterior translation which occurred as the head rebounded (as observed from high speed video) from the impact surface.  The impact platen force in this test  154 was unfortunately confounded with the inertial force from a much more massive impact platen.  The impact platen force in this impact did not display a bimodal shape as in H1220 and H1221 although this is largely due to the inertial effect of the heavier impact surface and is shown in Figure 5-13.  Figure 5-13:  H1222 kinetic analysis plot Axial force, impact platen force, and C7/T1 sagittal moment (top) and head z acceleration and head z velocity (bottom) for specimen H1222.  The T1 loading developed after a lag of 1.5 ms from head contact.  Initially, the extension bending appeared to be distributed along the column up until the T1 neck force reached its peak value of 3756 N at which point a hyper-physiologic extension rotation occurred in the upper cervical spine as the torso continued compressing the spine causing a C1 burst (Jefferson) fracture with an injury to the atlantooccipital joint as well as a C2 odontoid fracture.  The C7/T1 sagittal moment started out as an extension moment (this was confirmed to be due to an anteriorly directed Fx shear reaction force) which peaked when the lower-neck axial force developed.  This overall column extension required extreme flexion rotation at C7/T1 causing a  155 large C7/T1 flexion moment and produced a complete interspinous ligament and ligamentum flavum rupture as well as a facet joint capsule rupture at this level.  The carriage stopped its motion (as evidenced from high speed video) at the same time the force at T1 became tensile.  The C5/C6 level suffered an incomplete interspinous ligament rupture, ligamentum flavum rupture, incomplete facet capsule rupture and an incomplete anterior longitudinal ligament rupture. H1223: A sequence of images for this specimen is shown in Figure 5-14.  Figure 5-14:  High speed video frames for specimen H1223  156 This specimen also developed an extension-bending response with local flexion at C7/T1 after head contact as the carriage continued its downward momentum.  The impact force did not display a bimodal pattern in this drop as evidenced in Figure 5-15.  Figure 5-15:  H1223 kinetic analysis plot Axial force, impact platen force, and C7/T1 sagittal moment (top) and head z acceleration and head z velocity (bottom) for specimen H1223.  The lower-neck force developed in the second half of the impact platen impulse after a 2.3 ms lag.  As the T1 neck force was developing, its rate of development flattened approximately 7 ms post-contact as significant extension began.  Further carriage compression increased the T1 neck force up until its peak value of 1446 N was reached at approximately 12 ms post-impact.  The head z acceleration (negative or cephalad direction) shows that head rebound contributed to the peak T1 neck force.  Downward carriage motion continued past this point and at 15 ms post- impact, an anterior column injury that included a visible tearing of the anterior longitudinal ligament (ALL) occurred at C3.  Beyond this point, the T1 neck force dropped rapidly and became tensile (negative) at the same time that the impact platen force returned to zero  157 indicating that the head rebounded to a degree that it was no longer in contact with the impact platen.  Although the impact platform and the most superior portion of the head were not visible in the high-speed video, markers on the head showed head motion in the caudal direction and the integrated head z velocity (Figure 5-15) also suggested that the head was not in contact with the impact surface.  The spine’s ability to carry compressive load was diminished such that this continued head rebound contributed to further spinal extension rotation while the T1 neck force remained tensile, as opposed to loading the spine in compression as it had prior to the ALL injury.  At approximately 20 ms post impact, the carriage reached its maximum cephalad displacement (as evidenced from the video and the image synchronized axial neck force trace) just prior to the maximum tensile force at T1and subsequently carriage rebound.  Post-test dissection revealed compression-flexion injuries at C7/T1 consisting of an anterior vertebral body endplate compression fracture with a ligamentum flavum rupture and bi-lateral facet capsule ruptures. Other injuries diagnosed at the C3/C4 level included bi-lateral facet capsule ruptures, an incomplete ligamentum flavum tear, an inferior vertebral body fracture (attributed to osteoporotic bone), a fracture to the left uncinate process on C3, as well as an inferior endplate compression fracture to C4.  The C2/C3 level also incurred bilateral facet capsule injuries and a ligamentum flavum rupture. H1224: This specimen also responded with a compression-extension response creating a local flexion at C7/T1 which can be seen in Figure 5-16.  The kinetic response was almost identical to H1223 in that the head force also did not display a bimodal pattern (Figure 5-17).  The T1 neck force developed after a 2.3 ms lag time and reached a peak compressive value of 1411 N while it was being loaded due to the carriage moving downward (high speed video) and the head simultaneously moving upward (high speed video, integrated head z velocity, and cephalad head z acceleration).  This peak neck force created extension injuries at C3 that compromised the spines compressive load carrying capacity.  The T1 neck force became tensile at 13 ms post- impact while the head was still rebounding and peaked as the carriage hit the mechanical stops and reached the end of its compression.  This specimen also showed an initial extension moment at C7/T1 attributed to an anterior Fx shear reaction force that became a flexion moment as C7/T1 experienced a hyper-physiologic flexion rotation.   158  Figure 5-16:  High speed video frames for specimen H1224  159  Figure 5-17:  H1224 kinetic analysis plot Axial force, impact platen force, and C7/T1 sagittal moment (top) and head z acceleration and head z velocity (bottom) for specimen H1224.  Post injury dissections showed compressive-flexion injuries at C7/T1 consisting of: a T1 compression fracture of the vertebral body, a C7/T1 disc rupture, bilateral facet capsule ruptures, and a ligamentum flavum tear.  Extension injuries at C3/C4 included an anterior longitudinal ligament and disc annulus rupture as well as an incomplete facet capsule rupture.  In addition injuries were observed at C6/C7 consisting of an inferior endplate fracture, anterior longitudinal ligament rupture and an anterior annulus rupture of the C6/C7 disc.    160 5.4 Discussion  The motivation for this work was to extend and build upon our previous cadaveric model of head-first impact where the cervical lordotic posture was preserved and all specimens displayed a 1st order extension buckling response.  Our goals were to produce common clinically observed mid cervical column compressive fractures which are thought to occur in a lordosis- removed aligned spinal posture, and to extend the model to allow it to be used to evaluate three dimensional prototypes of the neck injury prevention helmet discussed in Chapter 4.  In order to achieve these aims, we improved our model’s biofidelity in 2 main ways: improved surrogate head anatomy by using the industry standard Hybrid III head, and improved column stability by use of an advanced muscle force replication system applied along physiologic lines of action. Despite successfully aligning the specimens and applying up to 177 N of compressive muscle force across the column, 4 of the 5 specimens responded with a 1st order extension buckling response.  The remaining specimen displayed an aligned column response and developed a burst fracture of C1 in the upper cervical spine.  Both of the column responses in these 5 impacts produced injuries that are observed clinically and none of the observed fractures or ligamentous failures appeared to have developed as a result of stress raisers associated with mounting to the surrogate head or application of muscle forces or the follower loads.  The modified follower load and muscle force system was very successful in both aligning and stabilizing the cervical spine and holding this pre-impact posture by applying forces along physiologic directions.  These methods were successful in that this model would be suitable for use with either standard 3D helmets, or with future iterations of the neck injury prevention helmet which was the subject of this thesis. As far as we are aware these were the first tests combining a cadaveric cervical spine with a Hybrid III 50th percentile head in a head-first impact.  Other groups have tested cadaveric cervical spines with a mass representing a surrogate head on a whiplash sled [136, 172], one that used a Hybrid III head on a whiplash sled [177] and another which used cadaveric cervical spines with an otherwise full Hybrid III dummy, also for studying whiplash [173].  We are unaware of other studies that have used a muscle force replication system to stabilize the cervical spine in a study of head-first impact.  In addition, this was the first cadaveric cervical spine inverted head-first drop test to be conducted with an aligned pre-impact posture that we are aware of although the aligned posture has been studied in dynamic compression in an upright  161 posture [33].  The peak axial neck forces were consistent with those measured by other researchers [33, 34, 81, 109] although the literature shows wide variability in this metric. Both the means of attaching the muscles to the bones, as well as attaching the cervical spine to the HIII head, appear to have preserved the initial temporal relationship between head and lower-neck loading observed in cadaveric head and cervical spine specimens [34, 78]. In all five tests, the lag times between head and neck development were in good agreement with those reported or observed elsewhere [33, 34].  This lag time is important for studying the concept of induced head motion on neck injury prevention in head-first impacts as the torso momentum does not immediately act upon the spine due to initial laxity in the spine. Specimens H1220, H1221, and H1222 showed a bimodal impact force, also similar to that reported by other researchers [34, 78] but the peak T1 compressive neck force occurred between the two modes of impact platen force.  This is unlike the results for cadaveric head and neck specimens not under a follower load or muscle force replication system [34], and the presence of the follower load likely explains the difference.  A larger follower load should theoretically consume most of the available low-stiffness laxity and thus these results may be more representative of those experienced in vivo.  This was observed in the study presented in Chapter 2 [119] where an increasing follower load shortened the lag-time between head and lower-neck force development in a surrogate neck model.  Taking this to the extreme, if the spine were perfectly rigid, there would be no lag time between the head and lower-neck axial force development and the full momentum of the torso would immediately act on the head, through the neck, at impact.  The results for the HIII neck in Chapter 3 are a closer approximation to this situation.  It should also be pointed out here that in specimen H1222, which was, chronologically, the first drop we conducted, we used a much heavier impact platen over the impact platen load cell.  The static effect of this additional mass was corrected for but the inertial effect confounded the impact platen force trace.  Therefore the impact platen force was actually much more bimodal in this impact than the plot shown in Figure 5-13 would suggest.  A close inspection of the kinetic traces for the other 4 drops shows that at the time point where the lower- neck and impact forces were equal, the head z acceleration was zero or nearly zero.  But in H1222, there is a large discrepancy in the impact and lower-neck forces at the time of zero head z acceleration and this inertial loading from the impact platform explains it.  162 While the degree of initial time lag between head and neck axial load development appeared realistic in all specimens, unfortunately the shape of the impact platen trace was not biofidelic in two of the specimens that reacted with a compression-extension response (H1223 and H1224).  In particular, it did not display the characteristic bimodal shape.  In these cases, a combination of head bounce, spine injury, and likely the torso contacting the mechanical stops precluded the coupling (phase alignment) between head and lower-neck axial loading as has been reported [34] to occur in the 2nd mode of impact platen loading and discussed throughout this thesis.  In fact, in all of the extension responses, the lower-neck actually developed tensile loading at some point but it was more apparent in specimens H1223 and H1224 where the peak tensile loads were in fact larger than the peak compressive loads.  This is clearly an artifact, and some of the soft tissue injuries were likely due to this phenomenon.  Other researchers have shown that the peak lower-neck force occurs when the neck is simultaneously loaded by the torso and head rebound but that after a short period, the becomes in-phase with with the head and neck to load the impact platen [34] which did occur here as well.  In this model, the degree of head bounce was too large such that in H1223 and H1224 when the torso reached its mechanical stops to prevent over-driving the specimens, the head was still rebounding.  The large head bounce could be attributed to the padding on the impact surface as well as the vinyl skin on the HIII head, and deformation from the aluminum headform.  In the one specimen that reacted axially, when the head rebounded, the spine resisted this rebound and was thus loaded compressively by it.  In the H1223 and 1224 extension cases, as the head rebounded, it created further extension bending to a much more compliant spine (incapable of supporting further compressive force from injury) that did not stop the head rebound.  Thus the head was still moving upwards, not in contact with the impact surface, when the carriage hit the stops.  Because of this, the impact platen force did not develop a 2nd defined mode as has been observed in multiple cadaveric studies [34, 78].  The T1 carriage then in fact rebounded off the mechanical stops and started an upward velocity which then added to the observed tensile axial neck forces. Perhaps the biggest strength of this model that relates to the helmet development described throughout this thesis, is the improved biofidelity offered by the HIII head that allows testing with a single size of a 3D helmet prototype.  Considering the high costs associate with creating prototypes, this is critical to the concept development, as developing multiple sizes would initially be cost-prohibitive.  In addition, the HIII head provides access to standardized  163 and state of the art instrumentation options for measuring head impact parameters.  Although it was challenging to ensure proper alignment when mounting the occiputs to the HIII head (see below), the use of a surrogate head allowed for another important strength of the model over intact head and neck specimens.  This was the method of tying the follower load and muscle force guides to the vertebrae which avoided creating stress risers by otherwise using screws or eyelets in the bone.  These benefits would have been much more challenging, if not impossible, without access to the spinal canal at the superior end of the specimens through the foramen magnum.  This is in addition to the fact that the use of intact cadaveric heads in impact biomechanics experiments is a sensitive issue and the multiple freeze-thaw cycles of the brain present additional challenges, not to mention the added psychological difficulties from working with intact heads.  The method of mounting the occiputs to the HIII head did not produce any artifactual fractures at the occiput and preserved the initial time lag between head and neck loading observed with intact head and neck specimens [34] and full cadavers [78].   The major advance made with this model was the development of the advanced muscle force replication system that stabilized the cervical spines and imposed an aligned pre-impact column posture by applying forces along physiologic directions without creating stress risers.  This muscle force system, while a step forward in its own right, will allow this model to be used in the future with helmets by allowing aligned pre-impact postures to be created without requiring some method of positioning the head with forces applied along non-physiologic lines of action.  The model also allowed for 2D x-ray images of the surrogate spinal cord that were synchronized with the high speed video cameras to correlate with spinal column injuries.  This will be a step forward in understanding how spinal cord injuries actually occur over the milliseconds of their initial mechanical insult.  It has been a long term aim of our lab to better understand the relationship between spinal cord and spinal column injuries with the hopes of eventually influencing the standardized tests being performed on mainly rodent models around the world to advance SCI regeneration treatments.  While these experimental mechanical insults to rodents are highly standardized, the relationship between the severity of the SCI inflicted to those occurring in actual human injuries remains unclear. While this model incorporated many improvements over Saari et al.[35, 36], it was not without limitations.  It was our hope that we would be able to produce mid-cervical column burst fractures which are commonly observed in young athletes in collision sports but none were  164 produced.  There are several possible explanations such as head-column eccentricity that was affected by head alignment, the magnitude of applied muscle forces, and also the means in which the muscle forces were applied.  Although we were able to straighten the cervical spines with the muscle forces, in some specimens the crown of the head at impact was anterior to the C7/T1disc centroid and this eccentricity may have influenced the column response to promote bending.  This was most likely due to the alignment achieved when mounting the cervical spines to the HIII head.  We noted considerable variability in the postures and the stiffnesses of the cervical spine specimens such that some specimens required less moment to align.  This finding is consistent with that noted by McElhaney where it took between 5 and 30 Nm of moment to align specimens that were initially cast in their natural lordotic postures [80].  Another possible reason is that we were only able to apply a maximum total compressive force across the column nearing 180 N.  Presumably, larger forces, which are known to exist in vivo with the same moment arms, would hold the aligned posture more securely and thus provide more resistance against the applied moments that occurred in impact.  A finite element study of head-first impact that included cervical musculature in automotive rollovers noted only a moderate increase in fracture risk moving from zero muscle contraction to passive muscle contraction, but the fracture risk doubled in simulations with maximal contraction [91]. Our means of applying the muscle forces via strings and elastic springs meant that the strings acted along straight lines of action although in vivo muscles in the cervical spine act along lines of action that are curved and wrap around bony anatomy and other muscles and they are affected by other muscles [178].  As the spines were compressed, the majority of the strings went loose at impact such that the muscle forces were rapidly reduced effectively to zero at impact.  It remains unknown if a similar effect would occur with in vivo muscles, which are known to have a viscoelastic rate dependency while our springs were theoretically free of viscous effects.  Another limitation of the muscle forces, are that they terminated at single points whereas tendons terminate over a distributed area.  The effects of these muscle simplifications remain unknown.  As mentioned above, all of the specimens that displayed a bending response developed tensile forces at the lower-neck at some point in the impact after being initially compressed.  This is grossly inaccurate and can be attributed to the impact carriage being arrested by the mechanical stops on the drop tower.  A similar effect was observed with our group’s previous model where a 3 cm stop gap was used to avoid over-driving the spinal specimens.  In these  165 experiments, we increased that gap to 4 cm based on the fact that the average failure compression in the cervical spine has been reported at 18 mm [81] and we thought this increased compression would remove this artifactual tensile force development at the lower-neck.  However, the 18 mm of compression at failure was reported for aligned specimens that reacted axially in an experiment using a mechanical ram.  As evidenced by H1220, when the column responds axially, a 4 cm stop gap was sufficient to provide the biofidelic bimodal response.  Similarly, this stop gap was appropriate in H1221 and H1222 which responded with compression-extension but where the neck retained some ability to carry compression after injury.  Unfortunately, in specimens H1223-H1224, the extension bending response created injuries that decimated the axial load carrying capacity.  This when combined with the significant upward bouncing of the HIII head, meant that the upward bouncing of the head contributed to a further extension bending which did not “stop” the upward motion of the head and consequently, the head was not in contact with the impact surface when the impact carriage (representing the torso) hit the mechanical stops.  Not only did the carriage hit the mechanical stops, but it bounced off them which contributed to the tensile forces experience at T1.  In fact, specimens H1223 and H1224 experienced a larger peak tensile force than peak compression force.  In the future, some means of arresting the impact carriage to avoid this bounce would be desirable.  While a drop tower model is not without limitations, it offers benefits such that the amount of energy that is input to the specimens is easily specified and is finite.  This makes it well suited to studying an injury prevention device as opposed to pure characterization of mechanical properties.  In the latter case, materials testing machines offer better control as well as the ability to monitor both force, velocity, and position to fully characterize a tissue or structure [33].  However, when using these materials testing machines, there is essentially infinite energy that ensures an injury will be produced unless some imposed limit of maximum force or displacement is enforced by the use of a control loop.  As it relates to this study, the drop tower model allows the spine the opportunity to be injured, or not, with the given energy input. When performing these drop tests, we lowered the specimens until the head made contact with the impact surface in order to determine where to set the mechanical stops.  In addition, we would apply the weight of the impact carriage quasi-statically by gradually lowering the ratchet- pulley to observe the quasi-static response of the column.  In all cases, the quasi-static response  166 of the cervical column was the same one as when it was impacted near 3 m/s, i.e. all the specimens that dynamically buckled in extension also did so quasi-statically.  We used this method as a last check and opportunity to increase muscle forces to try and further straighten the spine.  In the future, this protocol should be continued, and where it appears the column will respond with a bending response rather than an axial one, we should set the mechanical stops at a greater distance than 4 cm, although at some point the specimens will begin to be over-driven.  At extreme extensions, if the carriage continues downwards, the pulley mechanism on the anterior face of the casting cup will eventually come into contact with the vertebral bodies.  While this did not occur until well after the useful part of any of these experiments, our post-test analyses of injuries were not necessarily able to distinguish between injuries that occurred during the biofidelic “compressive” portion of the tests and any injuries that occurred later due to either a tensile force caused by carriage bounce or interacting with the pulley mechanism. There were also no accommodations made to correct for specimen differences.  As described, the spines were loaded and injured due to inertial loading from both the incoming torso and the simultaneous head rebound.  However, the impact carriage and head mass were the same for all specimens and the impact speeds were also nearly the same in all the drops.  Specimens H1223 and H1224 were from female donors and it has been documented that female necks have significant size, muscular strength, and bone strength differences compared to males [33, 70, 179].  Presumably these female specimens were subject to relatively more severe impacts although the stature of any of the donors was unknown.  These specimens also had the lowest (lumbar) bone mineral density scores which would actually be classified as osteoporotic.  This gender and bone quality difference certainly contributed to the response difference observed in these two drops, although there was only one fracture that could definitely be attributed to poor bone quality.  It should be noted that all of the specimens were either osteoporotic or osteopenic and this may have affected the observed fracture patterns. While it is a limitation seldom mentioned by researchers, this model was extremely labour-intensive.  By our estimates, each test lasting approximately 20 ms came about after approximately 100 to 110 person-hours of preparation, and this is not considering the post-test time to analyze the large amount of data produced.  We could likely devise ways to decrease preparation time slightly, but the labour-intensive nature of this model would make it challenging to use with large numbers of specimens that would be required to achieve sufficient statistical  167 power given the wide variability exhibited by biological cadaveric specimens.  In particular, considering that this model was developed to work towards testing a neck injury prevention helmet and that there are a large number of experimental conditions known to affect neck injury outcome in head-first impacts, this model might be best suited to use as validation for a computational finite element model. This model produced a high proportion of upper cervical spine injuries.  Four of five specimens had bony injuries of C1 or C2.  Two of these were C1 Jefferson fractures, and two were hangman fractures of C2 (traumatic spondylolisthesis).  There are several possible reasons for this.  One may be simply that due to the wide variability of spinal injuries that can result from a head-first impact, a sample size of only 5 specimens is not large enough to conclude that there was some systematic reason, as opposed to just a random sample.  Another could be due to the age of the cadaveric specimens.  An epidemiologic account of 657 patients suffering 717 cervical spine fractures and fracture-dislocations (not all from head-first impact) found that the percentage of upper cervical spine injuries increased gradually with age [180].  Another reason may be that without musculature, there is no avenue for dissipation of torso energy except directly into the spinal column.  Whereas, in vivo, an extension bending motion as displayed by 4 of 5 specimens would also cause lengthening of flexor muscles in the neck as well as paravertebral musculature which would dissipate some energy.  Thus, even though the impact velocities here represented those associated with the lower end of the spectrum where catastrophic cervical column and cord injuries occur in vivo, the injuries may instead represent injuries that are more severe and would occur with higher incoming velocities to younger athletes with better bone quality in the real-world.     There are some other limitations associated with the drop tower model.  The constraint that the carriage imposes against rotation on T1 creates a non-physiologic “stress riser” (abrupt boundary in rotational stiffness) that surely contributed to the flexion injuries observed at C7/T1 in all the extension specimens.  In reality, the moment would be transferred through into the thoracic column.  The thoracic region of the spine has much less mobility than the cervical region so the relatively fixed nature of T1 would provide some constraint against rotation but not nearly as rigidly as here.  The drop tower model also ensured that the momentum of the torso stayed collinear with the direction of impact velocity.  In reality the torso would begin to rotate and eventually make contact with the impact surface.  Ivancic et al. have demonstrated the  168 progression of a benchtop model of cadaveric cervical spine whiplash with muscle force replication [172] into a full-scale automobile seat model [173].  Our cadaveric model of cervical spine head-first impact too could likely be modified to work with the torso of a Hybrid III dummy such that a free-drop could be performed that is less contrained than the drop tower carriage was here due to being guided by rails.  Overall, while these methods were labour intensive and not without problems, they produced clinically observed injury patterns that occurred due to column buckling as reported by other researchers.  The methods were novel in several different ways including the Hybrid III head being used with cadaveric cervical spines in a drop tower model and the development of an advanced muscle force replication system along physiologic lines of action in a drop tower model.  These methods, with some improvements, will be of significant benefit for quantifying and advancing the basic understanding of dynamic spinal column and cord deformations in head- first impacts.  This model will also be significantly beneficial for testing injury prevention helmet prototypes under development in our lab.  169 Chapter 6: Discussion and Conclusions  6.1 Overview of Findings  The degree to which the head is constrained in a head-first impact has clear implications on the risk for spinal injury development [34, 71, 72, 75, 84-86, 91, 150].  The more the head is constrained at impact, from mechanisms such as pocketing into padding [34, 86], high surface friction [85, 150], unfavourable incident angles [34], and potentially from muscular contraction [75, 91], the greater the likelihood that spinal injury and spinal cord injury will develop.  We are aware of three studies that have investigated inducing head motion through an engineered interface, either a sliding roof-liner [105, 153] or a head-rest mounted airbag [106] as a neck injury prevention strategy against automotive rollovers and all concluded that this strategy had great potential.  However, we are not aware of any studies investigating this strategy through the use of a helmet despite the fact that the majority of sporting catastrophic spinal injuries occur in the presence of helmets and from a blow to the head/helmet.  Because of the catastrophic financial and quality of life effects of spinal cord injuries on individuals and societies that were documented in the previous chapters, it is clear that there is a need for neck injury prevention devices from head-first impact in sports and transportation where helmets are worn. This thesis was focused on developing and testing a neck injury prevention helmet from the outset, although other work had to be undertaken first.  The SC7 neck described in this thesis was originally designed and built out of necessity; the two main reasons being that: 1.) at the time our lab did not have (nor have access to) a mechanical surrogate neck and 2.) the main purpose for the SC7 neck was to interface with an existing mechanical surrogate head and a self- designed mechanical helmet prototype.  The first study with the SC7 neck presented in Chapter 2 was the characterization of its bending flexibility and range of motion, as well as its response to impact against a rigid and perpendicular impact platen.  These tests helped determine the magnitude of follower load that best represented the temporal nature of the cadaveric  head cervical spine axial load development presented by the group from Duke University [34] as well as in our own group [35, 36] in drop testing.  However, this follower load magnitude greatly restricted the range of motion of the SC7 neck in flexibility testing.  It was noted in Chapter 2 that while the SC7 neck model had uncovered some key attributes that could and should be incorporated into a future revised surrogate neck model dedicated to performing axial compressive impacts, that it should ultimately be redesigned to achieve both the desired head- 170 neck time lag in axial loading with a biofidelic flexion-extension bending stiffness and range of motion with a single level of follower load.  The fact that the neck was only intended to be used in near-vertex, aligned-column impacts that would not result in significant flexion or extension bending justified its continued use in this limited capacity without redesign.  Since the drop tests presented in Chapter 2 were conducted only against a rigid and perpendicular impact platform, and there are many factors known to affect neck injury outcome in head-first impacts, further characterization testing including additional factors had to be performed on the SC7 neck to determine if it could be used to assess a prototype neck injury prevention helmet.  After the characterization testing presented in Chapter 2, our lab acquired a Hybrid III head, neck, as well as lower-neck, upper-neck, and head CoG instrumentation.  The SC7 head and neck model was adapted to work with the HIII head instrumentation and lower-neck instrumentation.  In addition, an adjustable-angle impact platen was designed and built to proceed with further characterization.  The next testing performed was the L9 fractional factorial experiment presented in Chapter 3.  This experiment uncovered the dominant factors affecting the lower- neck injury metrics and showed that the SC7 neck was sensitive to changes in head constraint.  These results influenced the design of a two additional full factorial experiments, the first to compare the SC7 head and neck to the Hybrid III head and neck against the dominant impact parameters affecting lower-neck injury metrics that was also presented in Chapter 3; the second full factorial experiment studied the two dimensional prototype helmet and was presented in Chapter 4.  The cadaveric cervical spine model of head-first impact presented in Chapter 5 was developed in order to test future three-dimensional helmet prototypes.  6.2 Comparisons to Existing Findings  This thesis was exploratory in nature thus it is difficult to find directly equivalent studies for comparison.  In each dedicated manuscript chapter the research was compared to published literature.  In this section the dissertation research as a whole will be placed in context with the wider body of knowledge in the realm of head-first impact and spinal column and cord injury.  The SC7 ATD neck model described, and tested in Chapters 2-4 is the only ATD neck we are aware of that is specifically designed for and intended for axial compressive loading.  It was designed and built as part of this thesis research which makes it difficult to compare to other ATD studies.  It can be difficult to make comparisons between studies even using standardized  171 ATD models because differences in the experimental method of load application, loading rates, and boundary conditions strongly affect the results reported.  Recognizing the difficulty of making comparisons among testing with different methods, several studies have characterized the isolated Hybrid III neck alongside identical tests of the cadaveric neck; in quasi-static compression under the effects of varying end-conditions [71], combined bending and axial loading [84], and dynamic axial compression [74, 181, 182] in order to make comparisons using identical loading conditions, loading rates, and apparatus.  Similarly, in order to make comparisons between ATDs or ATD necks, they must be tested in identical loading scenarios.  This was the motivation for including a direct comparison between the SC7 head and neck and the Hybrid III head and neck in Chapter 3.  Briefly this testing showed that the SC7 neck developed higher forces and moments in identical loading environments compared to the Hybrid III (summarized in Table 3-4) due to a combination of the SC7 head having higher stiffness and less damping compared to the Hybrid III head and also that the SC7 neck became essentially rigid after the available range of compression offered by the rubber discs was consumed.  The major difference between the two head-neck models that has been discussed throughout the thesis was that the SC7 head and neck displayed a delay between platen force and lower-neck axial force development.  This is due to the high initial axial compliance in the SC7 neck which is not present in the Hybrid III neck.  With the Hybrid III head and neck, the temporal head and neck development pattern was virtually unaffected by the change in surface padding while it changed significantly with the SC7 neck.  The temporal change in head-neck loading due to padding for the SC7 is consistent with that reported for cadaveric cervical head-first impacts in a drop tower model [34].  Against a rigid impact surface there were two very well defined modes of impact platform loading, the first from stopping the head, the second attributed to the combined head, neck, and torso loading.  Against a soft padded surface, the impact platen force trace became unimodal such that it was not possible to distinguish separate impulses. In Nightingale’s PhD thesis [181] (section 5.4) he conducted drop tower tests using the Hybrid III head and neck against a rigid and one padded surface described as a 3.8 cm thick open polyurethane padding.  He reported that the Hybrid III head and neck displayed a unimodal impact force against the rigid surface, and a bimodal force pattern was observed with the padded surface.  All of the drops we conducted with the Hybrid III head and neck (Chapter 3) were against padded impact surfaces and produced bimodal impact platen force traces with an  172 observable time lag between head and lower-neck loading which moved the peak neck force to occur between the two modes of head force.  Our initial impression was that the SC7 neck model was a better representation of the cadaveric cervical spine than the Hybrid III as the peak lower- neck forces occurred in-phase with the second mode of head loading which is what was reported for the Duke University cadaveric cervical spine drop tower model (which did not simulate neck muscle-related pre-impact compression of the spine) [34].  These tests are widely cited and several numerical models have been validated against them.  However, after analyzing the cadaveric cervical spine model of impact described in Chapter 5, where muscle force simulation was included, we found that the pattern of head-neck force development with muscle force simulation was in better agreement with the Hybrid III neck than the SC7 neck in that the peak neck forces occurred between the two modes of head loading.  The difference is that the increased follower load by the muscle force application removed the initial compliance of the cervical spine and increased the coupling between the head and neck.  It is interesting that for the two “best” cadaveric drop tests in Chapter 5, (H1220 and H1221) both displayed a pattern of head-neck force development similar to the Hybrid III even though H1220 experienced a pure compression response and H1221 displayed a 1st order extension buckling response from initial impact.  Prior to conducting the cadaveric impacts in Chapter 5 it was unknown whether the lag time reported by Nightingale was due to the non linear material response in the cervical spine that results in the widely reported “neutral zone” [37] or if it was attributed to the structural buckling of the column, or some combination of these mechanics.  This concept of bimodal loading and head-neck time lag was first described by Nusholtz [73, 78] in some of the earliest inquiries into compressive neck injuries from head-first impact.  However, it was not possible to measure the lag time between head and lower-neck loading with full cadaver models unless some means of measuring internal force measurement was inserted into the column and these would have affected the column response.  The Duke University drop tower model was the first we are aware of to report this lag between head and lower-neck load development and they did this using isolated osseoligamentous head and neck specimens and did not simulate the compressive effect of musculature on the column prior to impact.  It is difficult to compare the lag times with our cadaveric model to the Duke model because this lag time is strongly influenced by surface padding.  Nightingale reported an average lag time of 1.7 ± 0.3 ms against rigid impact surfaces and 6.9 ± 1.7 ms against their 3.8 cm open polyurethane foam padding [34].  Our cadaveric  173 model, under approximately 180 N of overall compression throughout the cervical spine, and dropped onto a padded platform where the padding consisted of two layers of a 5 mm thick PVC yoga mat and an additional layer of a durometer 30 neoprene rubber, resulted in a time lag of 2.2 ± 0.4 ms.  Based on this it appears that if we had dropped onto a rigid platform our lag times would have been shorter.  Furthermore, we postulate that under full isometric muscular contraction producing loads as high as 1164 N as predicted by a numeric model [44], that this time lag between head and lower-neck loading would be largely eliminated.  The helmet prototype presented and evaluated in Chapter 4 is unlike any other research we are aware of in the sense that it is the only example of using a passive helmet-mounted mechanism to induce head motion in a head-first impact as a neck injury prevention strategy. Despite this, the results that we obtained are consistent with other researchers’ findings about the effect of head constraint on neck injury development.  Hodgson et al. conducted a study on bicycle helmet efficacy conducted head-first impacts with the Hybrid III dummy against 30° (and other) angled platens where the surface was a concrete slab and found that considerably higher neck loading and sagittal moments were produced when soft shelled helmets or the unhelmeted dummy “hung up” on the impact surface compared to hard-shelled helmets that slid along the surface at impact [150].  Deflecting the head along the impact surface in a head-first impact to create neck lateral bending was suggested as a possible neck injury prevention strategy by Bishop et al. [183].  It was shown that against oblique impact platforms which would be equivalent to a 20° angle in this thesis, the neck was still fully contained between the stopped head (at impact) and the torso such that the neck was subjected to an axial compressive load greater than 4500 N but they observed a 10% reduction in peak load compared to a perpendicular impact.  When they increased the incident angle to what would be equivalent to 40° in this thesis, they observed a 45% reduction in peak axial forces.  No mention was made of the friction conditions between the head and barrier.  Of interest is that in both the work of Hodgson and Bishop [150, 183], an angled platform alone was not able to allow for relative head (or helmet) sliding against the impact surface until angles greater than 40° or 30° respectively.  This is of interest to this thesis research where oblique angles only up to 15° away from perpendicular were studied.  First it confirms our finding in the no-helmet testing that a 15° angle is not sufficiently oblique to expect that the head could continue to stay in motion without some intervention and secondly, it highlights the success of this helmet strategy in that we were able to encourage head  174 motion that produced considerable axial force reductions against perpendicular and 15° angled impact surfaces with an appropriately chosen head motion escape, even over a range of padding conditions.  In the Duke University drop testing model using cadaveric head and neck specimens, soft padding was shown to dissuade head motion along -15° oblique impact platforms that caused injury where head motion did occur against a rigid and lubricated -15° surface where no injuries were produced [34].  This again highlights the success of this helmet concept for encouraging head motion even against a platform with soft padding which did contain the helmet, but since the head motion occurred relative to the helmet (or the outer shell) still kept the head in motion relative to the impact surface for the duration of the torso deceleration and resulted in axial neck force reductions. The force reductions, while encouraging, must be interpreted cautiously in light of the high stiffness of the SC7 neck model once the rubber vertebral discs have “bottomed-out” in compression.  These force reductions may be higher than what would occur with a more compliant neck.  Because ATD’s are designed to be robust and not to be frangible in high load impact environments, they develop forces much higher than what would be present with human tissue.  This is the reason for determining Injury Assessment Reference Values (IARVs) [151].  IARVs are established to be lower bounds of injury threshold such that a recorded force below the IARV would be very unlikely to produce injury but that forces exceeding an IARV would not necessarily predict injury as there is a wide range of variability in human tolerance to injury in any given loading modality.  The force reductions measured with the helmet using the SC7 neck suggest that the helmet was effective at reducing the impact severity and that this should result in a less severe impact with cervical spines.  Of course this needs to be studied further to be confirmed, but the force reductions measured here suggest that the helmet should prevent cervical spine fractures that would occur with biological specimens, where an ATD model would simply develop a high reaction force. The only other studies that directly induced head motion as a neck injury prevention strategy in compressive impacts were the seat mounted airbag [106] and the sliding automobile roof liner [105, 153] that were mentioned earlier.  The airbag, while similar enough to mention is actually different than the helmet in that the head motion is induced well before the head impact.  Its strategy involves actively forcing the head forward at a controlled rate into a “rollover- protected” posture well prior to impact to increase the available head space between the head and  175 roof liner before the roof-to-ground impact causes roof crush.  It also acts to increase ride-down for the occupant in the eventual contact between head and roof by absorbing energy [106].  For this reason it is not considered to be the same strategy as the helmet described herein which acts passively during the impact to create a more oblique head contact.  This motion-inducing helmet is most similar to the asymmetrical automobile roof [105, 153] which was compared to in Chapter 4 and is most similar to our flexion-anterior-translation (FAT) escape.  The roof has been evaluated using a finite element model [105] and this model was based on an earlier unpublished MADYMO model that is documented in a KTH (Kungliga Tekniska Högskolan translation Royal Institute of Technology), Stockholm Sweden, master’s thesis [153] of the same roof structure.  As mentioned in Chapter 4, the loading reductions of axial force for our helmet were similar to those observed in the finite element study although the loading scenarios were slightly different. An important difference was that in the finite element model, the orientation of the roof structure and direction of incoming velocity stayed constant while the head and neck model were rotated by 15 degrees such that the effective mass of the torso was more eccentric in their model relative to the neck.  This would explain why they observed a larger reduction at -15 degrees than we did (44% compared to 27% in our study).  Unfortunately no moment data was reported in their finite element study.  In addition, although not described thoroughly, an imposed motion path was applied to the T1 vertebrae in the FEA model which had been determined from the prior MADYMO study utilizing an ellipsoid model of a complete Hybrid III dummy against a similar roof structure that was dropped from a free-fall height of 1 m which we would interpret to mean the T1 data was generated from an impact with 4.43 m/s (compared to our 3.1 m/s impacts) although it is unclear.  The earlier MADYMO study provided many details about the roof system not presented in the published finite element study including moments and some details of the motion parameters.  However, the impact speed is not stated.  Instead, an acceleration pulse was applied to the dummy.  The roof model most similar to our helmet (used a 3 cm vertical displacement) produced a 29% reduction in axial force and an accompanying 17% reduction in lower-neck sagittal moment compared to tests against a conventional automobile roof.  This compares to our 40% reduction in axial force against perpendicular impact surfaces with accompanying 66% and 64% reduction in moment against soft and medium paddings respectively.  It should also be noted that they observed an 83% increase in the upper neck moment (atlantooccipital joint), however since our model was not capable of measuring an upper  176 neck moment, we cannot compare this.  It should be stressed that this MADYMO study includes sparse detail although due to the paucity of available data to compare to was included here for thoroughness.  In fact, we were unable to find a finalized and published version of this master’s thesis through the KTH University and after a direct request to the principal investigator, we obtained a draft copy of the thesis, the only version that they were able to locate. The helmet design presented in Chapter 4 is only one embodiment and perhaps there are other ways of accomplishing induced head motion than the design presented here.  The preliminary design used bi-lateral conformal pin and slot mechanisms in order to induce the head motion passively upon impact.  This design was used so that either the flexion-anterior- translation or the extension-posterior-translation escape could be deployed within a single helmet model.  Having the mechanism centered about the coronal plane of the helmet allowed this.  Perhaps, if only one escape was deemed necessary, a single rail along the sagittal plane could be used.   In this investigation which only considered impact velocities that were nearly perpendicular to the impact platen without any tangential incoming velocity, it was concluded that a selector mechanism would be necessary as the deployment direction had to increase the obliqueness of the impact in order to provide reductions in neck injury metrics.  The testing considered in Chapter 4 was likely most applicable to football or hockey impacts where the impacting surface (either another player’s torso or the boards after a check from behind respectively) is nearly perpendicular to the direction of incoming velocity.  In other sports such as motorcycling and mountain biking, the tangential velocity relative to the impact surface is usually much higher than the normal velocity.  In addition, the typical mechanism for a head-first impact involves the rider being pitched over the handle bars [12] such that the torso has both a rotational velocity and a linear velocity.  While it is undetermined at this point, it may be that only a single escape is necessary which might allow for a different design than the bi-lateral slot mechanisms used here. It has been noted in the literature that due to the phase shift of head and neck axial load development in head-first impacts, preventing head and neck injury can sometimes be at odds with each other [62].  An example of this is deep soft padding which is very beneficial for head injuries as it lowers head accelerations but as discussed throughout this thesis, when the head is stopped in a head-first impact, load is not yet developing through the neck so the padding does  177 not in fact help with neck loading.  In fact, the padding which reduces the head impact severity can aggravate the risk of neck injury development in a head-first impact by further constraining the head by providing a resistance to head motion along the impact surface.  This can result in the neck being exposed to the momentum of the incoming torso long enough for injury to occur.  Brain injuries and in particular mild traumatic brain injuries (concussion) occur much more frequently than cervical spine injuries, and even given the catastrophic nature of some cervical spine injuries, brain injuries are a much larger problem to society as a whole [62].  Thus it is imperative that a neck injury prevention helmet which is aiming to prevent catastrophic but rare injuries does not aggravate head injuries to the wearer such as concussions.  The helmet must also not create alternate neck injuries to the wearer. It was our intention to monitor for the possibility of alternate neck injury in the testing in Chapter 4 and this is why the neck axial loading, sagittal moment, and a combined neck injury metric were used alongside measures of the head acceleration to evaluate the prototype helmet efficacy in Chapter 4.  We also intended to monitor the possibility of alternate head injury which is why head accelerometers were included.  In addition, we attempted to measure head rotational acceleration but were unsuccessful and this will be discussed further as a limitation of the study.  Moving forward, it will be important to show that such a neck injury prevention helmet does not create alternate head or neck injuries to players wearing the helmet other athletes in collision sports.  It has been noted in American college football by way of comparing data from 1959- 1963 to 1971-1975 that the adoption of improved protective helmets decreased the incidence of serious hemorrhagic brain injuries but that this improved protection resulted in players initiating tackles with the crown of the helmet (spear-tackling) which resulted in a significant increase in the number of cervical spine injuries for the striking player [52, 184].  It has more recently been shown that spear-tackling is an injury mechanism that can create severe concussions to the struck player in helmet-helmet contact [38].  Based just on these findings from football, it appears that in contact sports any modifications to safety equipment must also coincide with behavioral modifications such as educational awareness campaigns and rule changes such as penalties that are enforced.  Rule changes and behavioural modifications are truly injury prevention strategies in the sense that if the head-first impact never occurs, then there is no chance of injury occurring.  Rule changes such as the penalty in college football for spear-tackling [8], Head’s up hitting in football and Head’s up hockey are examples of these types of campaigns and there is  178 epidemiological evidence to suggest that they are effective [52, 185].  However there is also evidence to suggest that catastrophic spinal cord injuries from head-first impacts are still occurring regularly [2], and also observational studies in football showing that non-injurious spear-tackling head-first impacts are frequently occurring [8, 186, 187], and that is why a neck injury prevention helmet such as this one is necessary.  The challenge will lie in ensuring that adoption of such a neck injury prevention helmet does not result in increased use of the head as a point of contact and thus the behavioral modification campaigns must continue alongside the adoption of new safety gear such as this helmet. Estimating the effect that an injury prevention helmet like this might have in the real world is certainly not a trivial task.  There are many reasons for this.  These catastrophic cervical spine injuries, while occurring regularly, are extremely rare compared to the number of estimated head-first impacts, most of which are non-injurious [8, 188].  Thus in order to retrospectively observe a statistical difference, it would require very widespread adoption of the helmet which is paradoxical in that widespread adoption would be more likely to occur after such a difference was observed.  In addition, as is the case with all safety equipment in impacts, just because an impact occurred in the presence of a piece of safety equipment and was non-injurious, that does not necessarily indicate that an injury would have been produced without the device, thus anecdotal accounts of non-injury must be evaluated carefully.  Despite the difficulties, an approach has been taken to attempt to estimate the possible effect of a successful neck injury prevention helmet.  This approach was developed by Dr. Cripton, thesis advisor for this body of work and CEO of Cripton Technologies, the newly formed spinoff company to continue this line of research and commercialize the helmet.  The data for cadaveric head-first impacts in Chapter 5 as well as the Duke University model discussed throughout this thesis [70] show that the average injurious neck force for a 3.1 m/s impact was 2066 N.  In addition it also showed that the average peak head force in these same impacts was very close to two times the neck force, or 4132 N.  A new tool developed to determine the head impact severity of football tackles producing concussions in American football consists of an array of accelerometers inside a standard football helmet.  Several research studies have been conducted using this helmet system (HITS, Simbex, Lebanon, NH) which have provided the first ever recorded data about the severity of head impacts in real football scenarios [158, 188-191].  One such study followed a group of ten football players over the course of the 2007 football season (practices and games)  179 and kept track of the number of head impacts, location of impacts on the helmet, as well as impact severity measured as linear and angular accelerations [158].  By making an assumption using the average human head mass of 4.36 kg [120] the average head force to accompany the average injurious neck force from a head-first impact corresponds to an acceleration of 948 m/s2 or approximately 97 g’s.  The study by Rowson [158] determined that of the 1712 impacts recorded, four of these were above 100 g’s, one was above 120 g’s, and none were higher than 140 g’s.  Based on the simplified example presented above, and assuming that the highest acceleration was perhaps just below 140 g’s, then a prevention helmet that lowered the neck force by 40% would have lowered all of the impact severities below the average neck failure force of 2066N.  The preferred deployments in Chapter 4 when averaged across both padding stiffnesses showed a 41% average reduction that ranged from a 21% reduction up to a 57% reduction, although these force reductions depended upon the properties of the SC7 neck.  Therefore it is reasonable to assume that they would be lower with a human neck.  However, with the Rowson paper, a 20% force reduction would have removed 4 of the 5 potentially injurious scenarios.  Of the several published studies using the HITS helmet only one other provides data in a fashion which allows this type of estimate [188].  A study by Greenwald determined that of 289,916 impacts imparted to 449 football players from 13 different collegiate teams over 3 full seasons, the top 1% (2,899) were above 103.4 g’s although the highest acceleration magnitudes were not presented.  The 98th percentile cutoff level (top 2%) was 86.4 g’s.  Again, considering if such a prevention helmet could reduce neck forces by 20%, the 1% of the impacts with the highest severity could be lowered to below the top 2% level which is below that which would cause the average failure neck axial force of 2066N.  It should be pointed out that even though the top 1% of these impacts had the potential to cause cervical spine injury, no such injuries were observed.  In addition, only 17 concussions occurred, 3 of which were confirmed to be to impacts to the crown of the head, although 13% (37,689) of all impacts were to the crown region and the highest severity impacts were associated with impacts to the crown region of the helmet [188].  We are very aware of the great number of simplifying assumptions such as head mass and direction of acceleration required to make this estimate but felt it was important nonetheless.  It would be highly beneficial to expand upon this analysis as there is much data available for football but due to the fact that we are considering neck injuries and it was generated for studying concussions, the published papers often present the data in ways that  180 do not lend themselves well for our purposes.  We will obtain the raw dataset to expand upon this analysis for football.  It would also be beneficial to consider a similar analysis for other sports, but unfortunately no equivalent dataset exists for other sports where hits to the crown of the helmet do not occur as frequently as in football.  Even within the sport of American football, it is estimated that there are 1 million participants at the high school level, seventy five thousand at the collegiate level, and two thousand at the professional level [192] and thus a successful prevention helmet which could reduce the impact severity of the more severe head-first impacts could prevent the catastrophic spinal cord injuries which unfortunately still regularly occur.  6.3 Strengths and Limitations  A key strength of the SC7 neck lies in the design of its articulations which allows for a simultaneous biofidelic range of motion in pure axial compression and flexion-extension bending.  The articulation points and range of motion were designed to match in vivo human center of rotation and range of motion data synthesized from the published literature.  This articulation design allows for axial compression with an initial high compliance which is novel to ATD necks and allows for an otherwise very robust (non frangible) ATD neck to display a time lag between head and neck development which has been noted in cadaveric studies of human head-first impact.  This characteristic has been reported to be missing in the most widely used Hybrid III head and neck [181].  The SC7 neck design includes an adjustable spring-loaded follower load type mechanism to “tune” the time lag between head and neck axial force development.  In addition, the SC7 neck incorporated the anatomically unique upper cervical spine with odontoid-like articulation between the C1 and C2 vertebrae.  The SC7 neck model, for both of the neck injury metrics used throughout the study, showed sensitivity to incident angle and platform padding.  The Hybrid III neck displayed similar sensitivity against incident angle but showed no sensitivity to the two platform paddings used, thus precluding its use from studying padding (of the type we tested) as a neck injury prevention strategy.  The SC7 neck was also highly repeatable for both of the sagittal plane injury metrics without the helmet.  Against perpendicular impacts, which produced much smaller moments, the coefficient of variation was 8% or lower.  With the helmet, the model was also generally very repeatable with the exception of two “non-preferred” runs where the helmet induced the opposite head motion to the one which would increase obliqueness which is at the heart of the injury prevention strategy.  For the  181 preferred helmet escapes, the maximum coefficient of variation for peak axial force and sagittal moment were 6% and 3% respectively.  A key strength of the helmet model presented in Chapter 4 is that it was of realistic mass, inertia, and size compared to contemporary helmets.  The results are especially encouraging in light of the fact that this helmet was able to successfully alter the path of the much more massive head which was 25% heavier than the 50th percentile male head [120].  In addition, due to the realistic helmet size, the magnitudes of the translations applied to the head are feasible for incorporation into helmets that could be worn by users.  This helps ensure the relevance of the research as it has been established before that if a helmet was made abnormally large, it could act to prevent neck injuries but that such a helmet would impede user mobility to the degree that it could not actually be worn.  The helmet testing also used a Teflon- Teflon surface pair, which we measured to have a static coefficient of friction of 0.27, between helmet and impact surface such that we were not biasing the helmets ability to induce motion by using an artificially high friction to “grip” the helmet upon impact. There were several strengths associated with the new cadaveric model of head-first impact presented that will help with the continued development of the neck injury prevention helmet.  The key strength of this model was the development of a light weight muscle force replication system suitable for use on our drop tower where the muscle forces that stabilized the specimens and held the initial posture were applied along physiologic directions thus allowing a helmet to be worn.  It has long been suspected that cervical musculature played a role in the neck response in head-first axial impacts [75, 91, 154].  In addition, observational accounts of real- world head-first impact injuries producing paralysis [3, 52] as well as laboratory studies have suggested that the lordosis-removed posture is required to produce axial compressive cervical spine injuries [74].  The methods used by other researchers to produce these injuries in the lab would not be suited to testing a helmet that induced head motion as a neck injury prevention strategy as the head was stabilized by dead weights, springs, and a halo ring [33, 74, 107].  Another key strength of these methods was the use of the Hybrid III head which will allow for a single size of future helmet prototypes to be used.  This is important due to the high cost of developing helmet prototypes but also because helmet fit is an important variable for performance and thus having a consistent fit prevents confounding results which might occur otherwise if using post mortem human heads and neck as there is wide variability in human head sizes and shapes.  182 There were some limitations associated with the experimental SC7 head and neck model, the drop tower, and the instrumentation used.  While the SC7 neck incorporated some features that increased its temporal head and neck force development biofidelity, these were achieved through the design of the vertebral articulation which, in order to simplify the design, limited the model to the sagittal plane.  Many catastrophic injuries do occur in the sagittal plane, although it has been suggested that the presence of lateral bending and axial rotation prior to impact affect injury outcomes and in fact increase the likelihood of spinal injury in a head-first impact [78].  In both pure axial compression and flexion-extension bending, the SC7 neck had high compliance throughout the available range of motion, but unfortunately reached hard endpoints to that motion.  Thus the benefits provided came at the cost of high forces and moments developing when the discs “bottomed out” in either bending or compression and this explains why the SC7 neck developed higher peak axial forces and sagittal moments than the Hybrid III neck.  Although the Hybrid III neck’s bending stiffness is likely too high for accurate use in head-first impacts, its range of motion in both bending and axial compression is ultimately dictated by the elastic response of the butyl rubber elements as opposed to the relatively rigid metallic response of the SC7 neck.  The articulations in the SC7 model also do not allow any anterior-posterior shear displacement to occur between adjacent vertebrae.  This surely adds to its highly stabilized axial response to compressive impacts.  This axial response was also guaranteed by the self- aligning nature of the SC7 neck under axial compression.  While we argue it is a benefit for studying the very limited worst case loading scenario of aligned column sagittal near vertex impacts, the limitation is that this feature also contributes to the limited number of loading scenarios that can be studied.  In particular, the SC7 neck model cannot buckle and its bending stiffness is proportional to the compressive load.  As presented in Chapter 2, even the presence of a 78 N or 104 N follower load affected the bending stiffness and range of motion.  When much higher compressions develop, such as those occurring during impact loading, the neck can only bend (from C1 to C7) in the presence of a moment sufficiently high enough to overcome the compression.  This is because compression results in engagement of the adjacent vertebral surfaces (equivalent to adjacent vertebral body endplates) which causes a “self-aligning” of the vertebrae and in essence acts as a further constraint against bending.  Our cadaveric testing presented in Chapter 5 showed that four of five specimens responded with a first order extension buckling response while developing compressive load even though the columns were aligned  183 with great care and stabilized in that posture by the muscle force replication system.  When compared to the cadaveric testing, the SC7 model, although not frangible, best represented the one specimen that did not buckle and this is also the worst-case response that we feel would have caused the most severe spinal column injury. While the SC7 neck was limited to studying aligned column sagittal plane impacts, for the initial exploration of this injury prevention concept we would argue this was a sufficient range of impacts to consider.  Several points in the literature justify this.  First, as has been discussed, it is only a very small subset of head-first impacts that produce catastrophic cervical spine injuries [8, 188] and the majority of this subset has been shown to be aligned column impacts to the crown of the head [3, 52, 184, 193].  In impacts where the initial posture of the head and neck is not aligned, an impact to the crown of the head will promote bending of the spine which can dissipate energy over a larger displacement (or rotation) into the spinal musculature as opposed to the aligned posture in which the energy is absorbed primarily by the bony vertebrae of the aligned segmented column.  This is not to say that injuries will always be avoided in these cases, in fact full cadaveric drop tests showed that initial postures with lateral bending and head axial rotation increased fracture risk [78], but rather that inducing either a lateral bending or axial rotation would not be beneficial.  It would also be more difficult to design into the helmet, which will be addressed below.  While the SC7 neck did not have a lateral bending or axial rotation range of motion built into it, tolerances in the design did in fact allow some degree of both of these motions, although none were observed in the testing due to the fact that the platform angles studied were all in the sagittal plane.  The design used here, could be extended to include both anterior-posterior shear displacement, lateral bending, and some degree of axial rotation.  As the majority of the head’s axial rotation range of motion occurs at the C1/C2 joint [14], this joint would require substantial redesign.  However, for the design used here, where articulations between adjacent vertebrae consisted of bolts extending bi- laterally through slots on the adjacent superior vertebra, instead of having the slots being essentially rigid in the anterior-posterior direction, a more compliant slot could be designed which would allow both anterior-posterior shear displacement and an increased range of motion in axial rotation.  Another improvement to the design used here would be to make the intervertebral disc height larger and decrease the height of the vertebral bodies.  There is a limit to this approach which would still allow for the articulation design here to work but this could  184 help alleviate the hard endpoint to both pure axial compression and flexion-extension bending when the adjacent vertebral bodies engage with each other. The surrogate head model used with the SC7 neck also had some limitations.  The crown region of the head was made of solid cast aluminum.  It was covered with two soft sheets of serrated rubber (the same material used for intervertebral discs) and then a layer of leather to simulate the scalp but it produced extremely high impact forces against rigid or nearly rigid surfaces.  In the L9 study in Chapter 3, we had installed a newly designed adjustable angle impact platen, which added considerable static weight over the platen used in Chapter 2 and the combination caused the impact platen load cell to saturate.  Thus no further testing was conducted against nearly rigid impact surfaces.  The factorial comparison study in Chapter 3 also showed that the impact platen forces with the SC7 neck and surrogate head were considerably higher than in tests with the validated Hybrid III head which lead us to conclude that the SC7 surrogate head is too stiff and requires further damping.   The surrogate head was originally designed to be used in the cadaveric drop tests by Saari that have been referenced throughout the thesis [35, 36, 174].  It was designed to match published mass and inertia of the human head [120].  Modifications were made to it to interface with both the SC7 neck (addition of an ‘occiput’ or C0 vertebra) as well as the prototype helmet.  These necessary modifications unfortunately increased the mass of the head but as mentioned above, the fact that the helmet prototype in Chapter 4 could induce motion to a more massive head was not necessarily a limitation.  These limitations of the initial surrogate head were part of the motivation behind the development of the new cadaveric model for head-first impact presented in Chapter 5 and were all overcome with the use of the Hybrid III head in that chapter and this will be the case in all future testing in our lab.  The surrogate head used with the SC7 neck in Chapters 2-4 has thus since been replaced with the Hybrid III head due to these limitations. The helmet model used in Chapter 4 was of realistic mass and overall size however its shape was cylindrical instead of spherical like helmets that fit on anatomically correct heads.  This compromise was made to ease construction and in light of the fact that the SC7 neck model was a sagittal plane model only.  This limitation helped to ensure that the head motion stayed in the sagittal plane.  In addition, the helmet was constructed of aluminum and as such was much stiffer than contemporary helmets.  The helmet was essentially rigid.  As mentioned in Chapter 4,  185 this prototype helmet was essentially an embodiment, or conceptual model of a dual-shell helmet where the two shells are connected through a guide mechanism that induces motion to the inner shell and the users head.  The helmet prototype only had one shell and the interface between the head and helmet was accomplished through the bi-lateral protrusions rigidly attached to the head.  This simplification was justified only as an initial ‘proof-of-concept’ evaluation but it precluded the normal means of securing a helmet to a users head through the friction of a good helmet fit and also the use of a chin strap.  One other notable limitation of the helmet, is that it was more of an experimental device to induce head motion that an actual “helmet” per se as it did not contain energy absorbing padding.  This is, after all, a helmet’s primary role but for the purposes of the preliminary investigation of the effect on neck injury metrics was not deemed necessary.  It is noteworthy however that the linear accelerations were reduced in preferred deployment impacts or not significantly different from no-helmet tests even without head padding due to the more controlled and prolonged manner in which the head was arrested.  All of these limitations with the helmet prototype presented in Chapter 4 have been rectified in subsequent 3D helmet prototypes under development which now incorporate an inner and outer shell with padding and a chinstrap that are being tested with the Hybrid III headform and the SC7 neck. As discussed above, we are well aware that a successful injury prevention helmet must “do no harm” and not create alternate injuries.  As such, it was our intention to closely monitor head injury metrics throughout the research in this thesis.  Brain injuries are known to be proportional to both the linear and the rotational accelerations experienced [194].  Although not reported, all of the impacts conducted in Chapters 3-5 attempted to measure the angular accelerations experienced by the head.  Unfortunately, at the time of acquiring our Hybrid III head, neck, and instrumentation, we were ill-advised to use a triaxial angular gyroscope to measure angular velocity instead of using an array of accelerometers to measure angular acceleration.  These angular gyroscopes have been shown to perform acceptably well and even agree with angular accelerometer arrays in inertial loading scenarios where there is no direct contact with the head.  Poor agreement was found for direct blows to the head of a Hybrid III dummy head containing both a triaxial gyroscope and 9-accelerometer array using a novel method of filtering the angular velocity output prior to differentiating it to angular acceleration [195].  We too discovered that our gyroscope sensor was not well suited to impacts with direct head contact and in particular that it was not adequately isolated from shock loading.  Although  186 it did not happen on every single impact, on the majority of impacts one or more of the three angular velocity channels would saturate upon impact implying that an angular velocity of 3600 rad/s was present even before the 1 ms duration point and well before any observable head rotation from the high-speed video taken at 1000 frames per second.  Multiple attempts at filtering this data prior to differentiating, as well as curve-fitting attempts were made, but ultimately we did not have enough confidence in the output from the sensor to include the data.  This issue has also been rectified with the purchase of an additional Hybrid III head designed to accommodate a 9 accelerometer array which was shown to be the most reliable method for measuring angular accelerations in short duration impact loading[195]. While our drop tower model provides an accurate depiction of what is occurring over approximately the first 25 milliseconds of a head-first impact, it has some limitations compared to impacts that occur in the real world.  In particular, as currently designed, the incoming velocity of the head and torso is always collinear and the torso is constrained (against rotation and translation) along the vertical direction of impact.  This is an excellent starting point but does not allow for the study of tangential velocities nor allow for an investigation of impacts where the head and torso are both rotating and translating prior to impact.  The drop tower scenario puts a large constraint on the lower cervical spine at the level of the carriage which surely influences the response.  It has also been noted in full cadaver drop tests, particularly those that remained in the sagittal plane, that the thoracic vertebrae were frequently injured [73].  In our current drop tower model of head-first impact we would be unable to conclude whether thoracic injuries would have been created.  Moving forward with the helmet development, these issues will eventually need to be overcome in order to make the experimental drop tests more like real-life conditions.   187 6.4 Conclusions  1. Incorporating a high compliance degree of freedom along the axial direction of a mechanical ATD neck simulated the time lag that has been observed to exist in the human head and neck in axial compressive impact loading.  The design of biofidelically- placed articulation points allowed for this degree of freedom and provided a non-linear resistance with low initial stiffness to sagittal bending similar to that observed with human cadaveric necks.  The temporal pattern of head and neck axial load development in impact, and the ROM in bending, were in good agreement with appropriate published research studies. 2. The neck injury metrics were most affected by the impact platform angle and the surface padding.  The highest axial forces occurred against perpendicular impacts with larger neck moments against oblique surfaces.  Platform angles that moved the contact point posterior to the head vertex created the largest peak moments which were extension moments.  Deep padding changed the impact platen impulse from distinctly bimodal to unimodal and reduced the phase-shift between axial head and lower-neck force. 3. A helmet that induces head motion in head-first impact can reduce the axial load experienced by the neck when it increases the obliqueness of an impact such that the head continues rotating and translating throughout the torso deceleration.  The effectiveness of this strategy was highly dependent upon incident angle such that an optimum helmet would be capable of moving the head anteriorly with head flexion in some impacts and posteriorly with head extension in others.  Inducing head motion in a head-first impact either lowered the linear head accelerations or at least did not significantly increase them compared to identical impacts without induced head motion. 4. Our newly developed model for cadaveric cervical spine head-first impact successfully stabilized the cervical column along physiologic directions while moving the head-neck into the most dangerous aligned posture without the need for any non-physiologic head forces which would preclude the use of a helmet.  This model preserved the widely observed time lag between head and lower-neck axial force development even while under the presence of up to 180 N of pre-impact compressive muscle force.  No bony or soft-tissue injuries along the cervical spinal column were deemed to be artifacts from the muscle force application system.  Multiple cervical spine specimens produced clinically  188 relevant spinal injuries with a single ATD head thus allowing for a single size of prototype helmet to be developed and tested. These testing methods are appropriate and necessary for the pre-commercial testing of future three dimensional helmet prototypes.   189 6.5 Contributions  This dissertation research has offered contributions to the literature and provides significant motivation for further development of biofidelic ATD necks and also this concept of a neck injury prevention helmet in head-first impacts.  These contributions are listed below. 1. First anthropometric test device neck specifically designed to incorporate the phase-shift between head and lower-neck axial load development observed in the human head and neck by offering a range of pure compression in head-first impacts while providing a realistic bending stiffness and bending range of motion. 2. First study to experimentally show that inducing head motion during a head-first impact can mitigate neck injury metrics and that this can be done while simultaneously lowering the linear head accelerations. 3. First study to induce head motion in a head-first impact with the use of a helmet model of realistic mass and inertia. 4. First study to develop a drop tower model of head-first impact using the Hybrid III head with cadaveric cervical spines. 5. First cadaveric cervical spine model of head-first impact to stabilize the cervical spine and thus assume an aligned pre-impact posture by the use of an advanced muscle force replication system applied along physiologic lines of action.  190 6.6 Applications of Research Findings  In this thesis the applications of the work are perhaps very obvious.  From the outset the work has been towards the development of a new type of neck injury prevention helmet for head-first impacts.  Towards this end two different head and neck models were also developed.  The performance of the SC7 head and neck has provided suggestions for design changes to be made to future versions of a dedicated axial compression ATD neck.    The helmet model developed and presented in Chapter 4 is essentially the first prototype in a series of prototypes.  This work formed the basis for securing intellectual property for this strategy and has led to the creation of UBC spinoff startup company, Cripton Technologies Inc., to develop the helmet further.  Although applicable to many different sports and transportation activities, the testing done thus far is most applicable to American football and hockey and those will be the initial targets.  The methods and cadaveric model for head-first impact developed in Chapter 5 represent a breakthrough in cadaveric cervical spine head-first impact models for their ability to simulate musculature and to relate spinal cord strains to spinal column kinematics.  Currently, every computational model of head-first impact that we are aware of is validated against the Duke University model of head-first impact that did not simulate musculature nor monitor the spinal cord throughout impact.  In addition to helping further the helmet development, the results of these and future drop tests using these methods, will perhaps be used as validation to advance the state of the art in computationally modeling the cervical spine and spinal cord in head-first impacts.    191 6.7 Future Research  This thesis research was the first step towards studying the effects of inducing head- motion during head-first impacts through the use of a helmet as a neck injury prevention strategy.  This body of work suggests that this concept has merit and as such, it has provided motivation for a great deal of future work to further study the efficacy of this strategy.  This thesis research is already directly contributing to, or will contribute to, some of this future work.  Additional studies that build upon this thesis work are suggested. 1. The SC7 neck has been modified to interface with the Hybrid III ATD head and is being used to develop three-dimensional spherical helmet prototypes.  In a recursive fashion, new prototypes are being tested and refined based on observations.  Once a functioning and sufficiently robust three-dimensional helmet has been developed, a study will be undertaken similar to the full factorial experiment in Chapter 4 that used the two- dimensional helmet prototype.  This study will use the same impact parameters as in Chapter 4 and will be aiming to replicate the force reductions observed with the two- dimensional prototype but with several improvements, namely: a helmet with two shells that uses realistic helmet retention (a chin strap), a spherical outer shell as opposed to cylindrical, and a nine accelerometer array in the Hybrid III head to measure rotational head accelerations to accurately assess the potential for alternate brain injuries as a result of induced head motion. 2. The newly developed model of cadaveric cervical spine head-first impact (HIII- AMFR) presented in Chapter 5 will be used to conduct an incomplete factorial study similar to that presented in Chapter 4 to assess the efficacy of three-dimensional neck injury prevention helmets but initially focusing only on the “preferred deployments” versus “no deployments” where the latter represents a helmeted impact without head- motion-inducing technology.  As a starting point, 24 cadaveric cervical spines will be used in the HIII-AMFR model to conduct the following experiment: three platform angles (-15°,0°, 15°), two platform stiffnesses (low and medium density padding used in Chapter 4), two replications, and two escapes.  At the -15° (posterior to vertex contact) and 0°(perpendicular) platform angles, the “preferred” escape of flexion-anterior- translation (FAT) as well as no-escape should be performed while at 15° (anterior to vertex contact) the preferred escape of extension-posterior-translation (EPT) and no- 192 escape should be performed.  This study will also use a newly upgraded Hybrid III ATD head with a nine accelerometer array to directly measure head angular accelerations as well as linear accelerations.  Lower-neck forces and moments will be measured (as in Chapter 5) and spinal kinematics as well as spinal cord compressions measured with high speed x-ray and making use of a biofidelic surrogate spinal cord.  3. A rigid body multi-body dynamics model of the cervical spine, Hybrid III ATD head, and neck injury prevention helmet is currently under development using ADAMS software (MSC software, Santa Ana, CA).  This model has produced similar results to those obtained experimentally in Chapter 4 and will be used to study the effects of different head-motion paths on lower-neck forces and moments.  This model will be used to optimize the path of induced head motion also with the intent of determining how much (or little) displacement of the head is required while still providing force reductions in order to miniaturize this concept as much as possible.  This model can also be used to address impacts involving tangential impact velocities and determine how they affect this mitigation strategy. 4. The HIII-AMFR drop testing and the ADAMS model will eventually be extended to include pre-impact postures that include lateral bending with some axial rotation to study the effect of FAT and EPT induced head motion escapes on injury. 5. After the first study listed above using the SC7 neck and Hybrid III head, as well as the ADAMS multibody dynamics model to investigate inducing head motion in the presence of high tangential velocities, a more advanced drop testing method will be developed.  While the drop tower model used thus far has helped immensely to develop the concept, eventually a less constrained physical drop testing experiment consisting of a Hybrid III dummy torso will be used with the SC7 neck and Hybrid III head.  This study will drop the helmeted dummy in unrestrained free fall onto both stationary and moving impact platforms [92] to study the effect of tangential velocities. 6. Building upon the study listed directly above, the HIII-AMFR model will be improved upon to work with a Hybrid III dummy torso and head to conduct unrestrained free-fall cadaveric cervical spine drop test experiments onto angled stationary and moving impact platforms.  193 7. Aside from the technical and scientific studies listed above, an economic investigation is being undertaken to conduct market research and develop a business plan to assess the profitability of manufacturing and selling such a neck injury prevention helmet.  Due to the rarity of the catastrophic injuries which this helmet aims to prevent, in order for this helmet to have an impact on injury prevention it will require widespread adoption.  In addition to first establishing that the helmet “does no harm” in terms of causing alternate injuries or hindering mobility of players in regular usage, it will also have to generate sufficient revenue for manufacturers to proceed given the known litigious nature of protective safety sporting equipment.  194 References   1. Sekhon, L.H. and M.G. Fehlings, Epidemiology, demographics, and pathophysiology of acute spinal cord injury. Spine, 2001. 26(24 Suppl): p. S2-12. 2. NSCISC, N.S.C.I.S.C., Spinal Cord Information Network—Facts and Figures at a Glance, in National Spinal Cord Injury Statistical Center University of Alabama at Birmingham. 2010. 3. Torg, J.S., J.T. Guille, and S. Jaffe, Injuries to the cervical spine in American football players. J Bone Joint Surg Am, 2002. 84-A(1): p. 112-22. 4. Tator, C.H. and V.E. Edmonds, National survey of spinal injuries in hockey players. Canadian Medical Association Journal, 1984. 130(7): p. 875. 5. Wyndaele, M. and J.J. Wyndaele, Incidence, prevalence and epidemiology of spinal cord injury: what learns a worldwide literature survey? Spinal Cord, 2006. 44(9): p. 523-9. 6. Brydon, V., Rick Hansen Foundation Backgrounder. 2010. 7. Cantu, R.C. and F.O. Mueller, Catastrophic spine injuries in American football, 1977-2001. Neurosurgery, 2003. 53(2): p. 358-62; discussion 362-3. 8. Heck, J.F., et al., National Athletic Trainers' Association Position Statement: Head- Down Contact and Spearing in Tackle Football. J Athl Train, 2004. 39(1): p. 101- 111. 9. Langer, P.R., P.D. Fadale, and M.A. Palumbo, Catastrophic Neck Injuries in the Collision Sport Athlete. Sports Medicine and Arthroscopy Review, 2008. 16(1): p. 7. 10. Tarazi, F., M.F.S. Dvorak, and P.C. Wing, Spinal injuries in skiers and snowboarders. The American journal of sports medicine, 1999. 27(2): p. 177. 11. Gobbi, A., B. Tuy, and I. Panuncialman, The incidence of motocross injuries: a 12- year investigation. Knee Surgery, Sports Traumatology, Arthroscopy, 2004. 12(6): p. 574-580. 12. Dodwell, E.R., et al., Spinal Column and Spinal Cord Injuries in Mountain Bikers. The American journal of sports medicine, 2010. 38(8): p. 1647. 13. Ferrario, V.F., et al., Active range of motion of the head and cervical spine: a three- dimensional investigation in healthy young adults. J Orthop Res, 2002. 20(1): p. 122- 9. 14. Bogduk, N. and S. Mercer, Biomechanics of the cervical spine. I: Normal kinematics. Clinical Biomechanics, 2000. 15(9): p. 633-648. 15. Huelke, D.F. and G.S. Nusholtz, Cervical spine biomechanics:  a review of the literature. Journal of Orthopaedic Research, 1986. 4(2): p. 232-245. 16. Nachemson, A.L., Disc pressure measurements. Spine, 1981. 6(1): p. 93-97. 17. Cripton, P.A., G.A. Dumas, and L. Nolte, A minimally disruptive technique for measuring intervertebral disc pressure in vitro: application to the cervical spine. J.Biomech., 2001. 34(4): p. 545-549. 18. Drake, R., W. Vogl, and A. Mitchell, Gray’s anatomy for students. Churchill Livingstone, Edinburgh, 2004. 19. Milne, N., The role of zygapophysial joint orientation and uncinate processes in controlling motion in the cervical spine. Journal of anatomy, 1991. 178: p. 189.  195 20. Moore, K.L. and A.M.R. Agur, Essential clinical anatomy. 2nd ed. 2002, Philadelphia: Lippincott Williams & Wilkins. xii, 691. 21. Panjabi, M.M., A.A. White, and R.M. Johnson, Cervical spine mechanics as a function of transection of components. Journal of Biomechanics, 1975. 8(327): p. 336. 22. Vasavada, A.N., S. Li, and S.L. Delp, Influence of muscle morphometry and moment arms on the moment-generating capacity of human neck muscles. Spine, 1998. 23(4): p. 412-422. 23. Blouin, J.S., et al., Neural control of superficial and deep neck muscles in humans. Journal of neurophysiology, 2007. 98(2): p. 920. 24. Siegmund, G.P., et al., Electromyography of superficial and deep neck muscles during isometric, voluntary, and reflex contractions. Journal of biomechanical engineering, 2007. 129: p. 66. 25. King Foster, J., K. J.O., and M.J. Wolanin. Hybrid III - A Biomechanically-Based Crash Test Dummy. in 21st Stapp Car Crash Conference. 1977. New Orleans, Louisiana: SAE. 26. Mertz, H.J. and L.M. Patrick. Strength and response of the human neck. 1971. UCLA: 15th STAPP car crash conference. 27. Davidsson, J., BioRID II Final Report. Report at Chalmers University of Technology, Department of Machine and Vehicle Systems, Göteborg, Sweden, 1999. 28. Davidsson, J., et al. BioRID I–A New Biofidelic Rear Impact Dummy. in 1998 International IRCOBI Conference on the Biomechanics of Impacts. 1998. Goteborg, Sweden. 29. Eppinger, R.H., J.H. Marcus, and R.M. Morgan, Development of dummy and injury index for NHTSA's thoracic side impact protection research program. SAE Technical Paper, 1984. 840885. 30. Eppinger, R., et al., Development of improved injury criteria for the assessment of advanced automotive restraint systems–II. National Highway Traffic Safety Administration, 1999. 31. Panjabi, M.M., et al., Three-dimensional load-displacement curves due to forces on the cervical spine. Journal of Orthopaedic Research, 1986. 4: p. 152-161. 32. Panjabi, M.M., V.K. Goel, and K. Takata, Physiologic strains in the lumbar spinal ligaments. An in vitro biomechanical study 1981 Volvo Award in Biomechanics. Spine, 1982. 7(3): p. 192. 33. Pintar, F.A., et al. Dynamic characteristics of the human cervical spine. in 39th Stapp Car Crash Conference. 1995. San Diego, CA: Society of Automotive Engineers. 34. Nightingale, R.W., et al., Dynamic response of the head and cervical spine to axial impact loading. Journal of Biomechanics, 1996. 29(3): p. 307-318. 35. Saari, A., Itshayek, E, Nelson, T.S., Morley, P.L.,Cripton, P.A. Spinal Cord Deformation during injury of the Cervical Spine in Head-First Impact. in International Research Conference on the Biomechanics of Impact (IRCOBI). 2006. Madrid, Spain. 36. Saari, A., E. Itshayek, and P.A. Cripton, Cervical spinal cord deformation during simulated head-first impact injuries. Journal of Biomechanics, 2011. 44(14): p. 2565- 2571.  196 37. Oxland, T.R. and M.M. Panjabi, The onset and progression of spinal injury:  a demonstration of neutral zone sensitivity. Journal of Biomechanics, 1992. 25(10): p. 1165-1172. 38. Viano, D.C. and E.J. Pellman, Concussion in professional football: Biomechanics of the striking player-Part 8. Neurosurgery, 2005. 56(2): p. 266. 39. Nightingale, R.W., W.J. Richardson, and B.S. Myers, The effects of padded surfaces on the risk for cervical spine injury. Spine, 1997. 22(20): p. 2380-7. 40. Panjabi, M.M., et al., Critical load of the human cervical spine: an in vitro experimental study. Clinical Biomechanics, 1998. 13(1): p. 11-17. 41. Patwardhan, A.G., et al., Load-carrying capacity of the human cervical spine in compression is increased under a follower load. Spine, 2000. 25(12): p. 1548-54. 42. Amevo, B., D. Worth, and N. Bogduk, Instantaneous axes of rotation of the typical cervical motion segments: a study in normal volunteers. Clinical Biomechanics, 1991. 6: p. 111-117. 43. Dvorak, J., et al., In vivo flexion/extension of the normal cervical spine. J.Orthop.Res., 1991. 9(6): p. 828-834. 44. Moroney, S.P., A.B. Schultz, and J.A.A. Miller, Analysis and measurement of neck loads. Journal of Orthopaedic Research, 1988. 6: p. 713-720. 45. Hattori, S., et al., Cervical intradiscal pressure in movements and traction of the cervical spine. Zeitschrift fur Orthopadie, 1981. 119: p. 568-569. 46. Pooni, J.S., et al., Comparison of the structure of human intervertebral discs in the cervical, thoracic and lumbar regions of the spine. Surgical and Radiologic Anatomy, 1986. 8: p. 175-182. 47. Bastien, G.J., et al., Energetics of load carrying in Nepalese porters. Science, 2005. 308(5729): p. 1755. 48. Lloyd, R., et al., No'free ride'for African women: a comparison of head-loading versus back-loading among Xhosa women. South African Journal of Science, 2010. 106(3-4): p. 01-05. 49. Nightingale, R.W., et al., Inertial properties and loading rates affect buckling modes and injury mechanisms in the cervical spine. J.Biomech., 2000. 33(2): p. 191-197. 50. Camacho, D.L.A., et al. Experimental flexibility measurements for the developement of a computational head-neck model validated for near-vertex head impact. 1997. Lake Buena Vista, Florida: 41st STAPP Car Crash Conference. 51. Blackburn, J.T., et al., The relationships between active extensibility, and passive and active stiffness of the knee flexors. Journal of Electromyography and Kinesiology, 2004. 14(6): p. 683-691. 52. Torg, J.S., et al., The epidemiologic, pathologic, biomechanical, and cinematographic analysis of football-induced cervical spine trauma. The American journal of sports medicine, 1990. 18(1): p. 50. 53. Kazarian, L., Classification of Simple Spinal Column Injuries, in Impact Injury of the Head and Spine, C.L. Ewing, Thomas, Daniel J., Sances Anthony, Larson, Sanford J., Editor. 1983, Charles C Thomas: Springfield, Illinois. p. 72-93. 54. Landells, C.D. and P.K. Van Peteghem, Fractures of the atlas: classification, treatment and morbidity. Spine, 1988. 13(5): p. 450-2. 55. Grauer, J.N., et al., Proposal of a modified, treatment-oriented classification of odontoid fractures. The Spine Journal, 2005. 5(2): p. 123-129.  197 56. Anderson, L.D. and T. D'alonzo, Fractures of the odontoid process of the axis. Journal of Bone and Joint Surgery.American Volume, 1974. 56-A(8): p. 1663-1674. 57. Benzel, E.C., et al., Fractures of the C-2 vertebral body. Journal of Neurosurgery, 1994. 81: p. 206-212. 58. Effendi, B., et al., Fractures of the ring of the axis. A classification based on the analysis of 131 cases. The Journal of bone and joint surgery. British volume, 1981. 63(3): p. 319. 59. Francis, W.R., et al., Traumatic spondylolisthesis of the axis. Journal of Bone and Joint Surgery-British Volume, 1981. 63(3): p. 313. 60. Hadley, M.N., et al., Axis fractures resulting from motor vehicle accidents. Spine, 1986. 11(9): p. 861-864. 61. Allen, B.L., Jr., et al., A mechanistic classification of closed, indirect fractures and dislocations of the lower cervical spine. Spine, 1982. 7(1): p. 1-27. 62. Winkelstein, B.A. and B.S. Myers, The biomechanics of cervical spine injury and implications for injury prevention. Med Sci Sports Exerc, 1997. 29(7 Suppl): p. S246- 55. 63. Nightingale, R.W., et al., Experimental impact injury to the cervical spine:  relating motion of the head and the mechanism of injury. Journal of Bone and Joint Surgery.American Volume, 1996. 78-A(3): p. 412-421. 64. Hendey, G.W., et al., Spinal cord injury without radiographic abnormality: results of the National Emergency X-Radiography Utilization Study in blunt cervical trauma. J Trauma, 2002. 53(1): p. 1-4. 65. Dumont, R.J., et al., Acute spinal cord injury, part I: pathophysiologic mechanisms. Clin Neuropharmacol, 2001. 24(5): p. 254-64. 66. Choo, A.M., et al., Contusion, dislocation, and distraction: primary hemorrhage and membrane permeability in distinct mechanisms of spinal cord injury. J Neurosurg Spine, 2007. 6(3): p. 255-66. 67. Bates-Carter, J., R.P. Ching, and A.F. Tencer. Transient changes in canal geometry during axial burst fracture of the cervical spine. 1995. Detroit: 5th Injury prevention through biomechanics symposium. 68. Chang, D.G., et al., Geometric changes in the cervical spinal canal during impact. Spine, 1994. 19(8): p. 973-980. 69. McElhaney, J., et al., Biomechanical Analysis of Swimming Pool Neck Injuries. Society of Automotive Engineers SP-79, 1979: p. 47-53. 70. Nightingale, R.W., et al. The dynamic responses of the cervical spine: buckling, end conditions and tolerance in compressive impacts. 1997. Lake Buena Vista, Florida: 41st STAPP Car Crash Conference. 71. Nightingale, R.W., et al., The influence of end condition on human cervical spine injury mechanisms. Society of Automotive Engineers Transactions, Paper 912915, 1991: p. 391-399. 72. Hodgson, V.R. and L.M. Thomas. Mechanisms of cervical spine injury during impact to the protected head. 1980. Warrendale,PA: 24th STAPP car crash conference. 73. Nusholtz, G.S., et al., Response of the cervical spine to superior-inferior head impact. 1981: Society of Automotive Engineers, 400 Commonwealth Dr, Warrendale, PA, 15096, USA.  198 74. Yoganandan, N., A. Sances, Jr., and F. Pintar, Biomechanical Evaluation of the Axial Compressive Responses of the Human Cadaveric and Manikin Necks. J Biomech Eng, 1989. 111(3): p. 250-5. 75. Yoganandan, N., et al., Experimental spinal injuries with vertical impact. Spine, 1986. 11(9): p. 855-860. 76. Viano, D.C. and C.S. Parenteau, Analysis of head impacts causing neck compression injury. Traffic Injury Prevention, 2008. 9(2): p. 144-152. 77. Pellman, E.J., et al., Concussion in professional football: Reconstruction of game impacts and injuries. Neurosurgery, 2003. 53(4): p. 799. 78. Nusholtz, G.S., et al. Cervical spine injury mechanisms. 1983: 27tth Stapp car crash conference. 79. Pintar, F.A., et al. Dynamic characteristics of the human cervical spine. 1995. Warrendale,PA: 39th STAPP car crash conference. 80. McElhaney, J.H., et al. Cervical spine compression responses. 1983. Warrendale,PA: 27th Stapp car crash conference. 81. Maiman, D.J., N. Yoganandan, and F.A. Pintar, Preinjury cervical alignment affecting spinal trauma. J Neurosurg Spine, 2002. 97(1): p. 57-62. 82. Mertz, H.J., et al., An Assessment of Compressive Neck Loads Under Injury- producing Conditions. The Physician and Sports Medicine, 1978. 83. Bauze, R.J. and G.M. Ardran, Experimental production of forward dislocation in the human cervical spine. J Bone Joint Surg Br, 1978. 60-B(2): p. 239-45. 84. McElhaney, J.H., et al., Combined bending and axial loading responses of the human cervical spine. SAE Paper #881709, 1988: p. 21-28. 85. Camacho, D.L.A., R.W. Nightingale, and B.S. Myers, Surface friction in near-vertex head and neck impact increases risk of injury. Journal of Biomechanics, 1999. 32(3): p. 293-301. 86. Camacho, D.L.A., R.W. Nightingale, and B.S. Myers, The influence of surface padding properties on head and neck injury risk. Journal of biomechanical engineering, 2001. 123: p. 432. 87. Foust, D.R., et al. Cervical range of motion and dynamic response and strength of cervical muscles. 1973. Warrendale,PA: 17th STAPP car crash conference. 88. McGill, S.M., et al., Passive stiffness of the human neck in flexion, extension, and lateral bending. Clinical Biomechanics, 1994. 9: p. 193-198. 89. Granata, K.P. and W.S. Marras, The influence of trunk muscle coactivity on dynamic spinal loads. Spine, 1995. 20(8): p. 913-919. 90. Granata, K.P. and W.S. Marras, Cost-benefit of muscle cocontraction in protecting against spinal instability. Spine, 2000. 25(11): p. 1398. 91. Hu, J., et al., A Numerical Investigation of Factors Affecting Cervical Spine Injuries During Rollover Crashes. Spine, 2008. 33(23): p. 2529. 92. Aare, M. and P. Halldin, A New Laboratory Rig for Evaluating Helmets Subject to Oblique Impacts. Traffic Injury Prevention, 2003. 4(3): p. 240-248. 93. Huston, R.L. and J. Sears, Effect of protective helmet mass on head/neck dynamics. Journal of Biomechanical Engineering, 1981. 103: p. 18-23. 94. Konrad, C.J., et al., Are fractures of the base of the skull influenced by the mass of the protective helmet? A retrospective study in fatally injured motorcyclists. The Journal of Trauma, 1996. 41(5): p. 854.  199 95. Macnab, A.J., et al., Effect of helmet wear on the incidence of head/face and cervical spine injuries in young skiers and snowboarders. Injury Prevention, 2002. 8(4): p. 324. 96. Geisinger, A., et al., Innovations at BMW Motorrad in the Development of Equipment for Motorcycle Riders to reduce the Risk of Injuries based on the Example of the Neck Brace System. Development trends of motorcycles: With 10 tables, 2007: p. 199. 97. Gorden, J.A., et al., Effects of football collars on cervical hyperextension and lateral flexion. Journal of Athletic Training, 2003. 38(3): p. 209. 98. Hovis, W.D. and T.J. Limbird, An evaluation of cervical orthoses in limiting hyperextension and lateral flexion in football. Medicine & Science in Sports & Exercise, 1994. 26(7): p. 872. 99. Rowson, S., et al., Biomechanical analysis of football neck collars. Clinical Journal of Sport Medicine, 2008. 18(4): p. 316. 100. Stuber, K., Cervical collars and braces in athletic brachial plexus injury and excessive cervical motion prevention: a review of the literature. The Journal of the Canadian Chiropractic Association, 2005. 49(3): p. 216. 101. Engsberg, J.R., et al., Spinal cord and brain injury protection: testing concept for a protective device. Spinal Cord, 2009. 47(8): p. 634-639. 102. Landau, W.M., Brain and Spinal Cord Protector. 1998: United States. 103. Burstein, A.H. and J.C. Otis, The response of the cervical spine to axial loading: feasibility for intervention. Head and neck injuries in sports, 1994: p. 142. 104. Theisen, M., S. Eisele, and M. Rolleke, The computation of airbag deployment times with the help of precrash information. SAE Technical Paper, 2002: p. 01-0192. 105. Halldin, P. Investigation of Conditions that Affect Neck Compression-Flexion Injuries using Numerical Techniques. in 44th Stapp Car Crash Conference. 2000: SAE. 106. Heudorfer, B., et al. Roofbag-a concept study to provide enhanced protection for head and neck in case of rollover. in 16th International Technical Conference on the Enhanced Safety of Vehicles. 1998. Windsor, Ontario, Canada: NHTSA. 107. Pintar, F.A., et al. Kinematic and anatomical analysis of the human cervical spinal column under axial loading. 1989. Warrendale,PA: 33rd STAPP car crash conference. 108. Nusholtz, G.S., et al. Response of the cervical spine to superior-inferior head impact. 1981: 25th Stapp car crash conference. 109. Sances, A.J., et al., Compression studies of the human and hybrid III cervical spine, in 1986 Advances in Bioengineering, S.A. Lantz and A.I. King, Editors. 1986, ASME: New York. p. 46-47. 110. Bishop, P.J., Factors related to quadriplegia in football and the implications for intervention strategies. Am J Sports Med, 1996. 24(2): p. 235-9. 111. Orlowski, K.F., R.T. Bundorf, and E.A. Moffatt. Rollover crash tests—the influence of roof strength on injury mechanics. in 29th Stapp Car Crash Conference. 1985. Washington, DC: SAE. 112. Moffatt, E.A., et al., Matched-pair rollover impacts of rollcaged and production roof cars using the controlled rollover impact system (CRIS). 2003, SAE. p. 1-10. 113. Bahling, G.S., et al. Rollover and drop tests - the influence of roof strength on injury mechanics using belted dummies. in 34th Stapp Car Crash Conference. 1990. Orlando, Florida: SAE.  200 114. King Foster, J., Kortge J.O., Wolanin, M.J. Hybrid III - A Biomechanically-Based Crash Test Dummy. in 21st Stapp Car Crash Conference. 1977. New Orleans, Louisiana: SAE. 115. Fréchède, B., et al., Hybrid III ATD in Inverted Impacts: Influence of Impact Angle on Neck Injury Risk Assessment. Annals of Biomedical Engineering, 2009. 37(7): p. 1403-1414. 116. Sances, A., Jr. and S. Kumaresan, Comparison of biomechanical head-neck responses of hybrid III dummy and whole body cadaver during inverted drops. Biomed Sci Instrum, 2001. 37: p. 423-7. 117. Yoshida, H. and S. Tsutsumi, Experimental analysis of a new flexible neck model for low-speed rear-end collisions. Accid Anal Prev, 2001. 33(3): p. 305-12. 118. Powell, A.M. and L.E. Bilston, Design and validation of a new omnidirectional dummy neck For low speed impacts. International Journal of Crashworthiness, 2002. 7(4): p. 463-474. 119. Nelson, T.S. and P.A. Cripton, A New Biofidelic Sagittal Plane Surrogate Neck for Head-First Impacts. Traffic Injury Prevention, 2010. 11(3): p. 309-319. 120. Walker, L.B., E.H. Harris, and U.R. Pontius. Mass, volume, center of mass, and mass moment of inertia of head and head and neck of human body. 1973. Warrendale,PA: 17th STAPP car crash conference. 121. Hubbard, R.P., McLeod, D.G., Definition and Development of a Crash Dummy Head, in 18th Stapp Car Crash Conference. 1974, Society of Automotive Engineers, Inc., Warrendale, Pennsylvania, USA: Ann Arbor, Michigan, USA. 122. Dong, Y., et al., Quantitative anatomy of the lateral mass of the atlas. Spine, 2003. 28(9): p. 860-3. 123. Wilke, H.J., et al., Anatomy of the sheep spine and its comparison to the human spine. Anat Rec, 1997. 247(4): p. 542-55. 124. Kandziora, F., et al., Comparison between sheep and human cervical spines: an anatomic, radiographic, bone mineral density, and biomechanical study. Spine, 2001. 26(9): p. 1028-37. 125. Miura, T., M.M. Panjabi, and P.A. Cripton, A method to simulate in vivo cervical spine kinematics using in vitro compressive preload. Spine, 2002. 27(1): p. 43-8. 126. Foster, K.J., J.O. Kortge, and M.J. Wolanin, Hybrid III~a biomechanically-based crash test dummy. SAE International, 1977. 127. Cripton, P.A., et al., Force Transmission in the Human Cervical Spine During Destructive Axial Compression:  Effect of Flexion and Extension Postures, in 28th Annual Meeting of the Cervical Spine Research Society. 2000: Charleston, SC, USA. 128. Ordway, N.R., et al., Cervical flexion, extension, protrusion, and retraction. A radiographic segmental analysis. Spine., 1999. 24(3): p. 240-247. 129. Goel, V.K. and J.D. Clausen, Prediction of load sharing among spinal components of a C5-C6 motion segment using the finite element approach. Spine, 1998. 23(6): p. 684-691. 130. Goertzen, D.J., C. Lane, and T.R. Oxland, Neutral zone and range of motion in the spine are greater with stepwise loading than with a continuous loading protocol. An in vitro porcine investigation. J Biomech, 2004. 37(2): p. 257-61. 131. Wen, N., et al., Three-dimensional biomechanical properties of the human cervical spine in vitro, I. Analysis of normal motion. European Spine Journal, 1993. 2: p. 2-11.  201 132. Penning, L., Normal movements of the cervical spine. AJR.Am.J.Roentgenol., 1978. 130(2): p. 317-326. 133. Dvorak, J., et al., Functional radiographic diagnosis of the cervical spine:  flexion/extension. Spine, 1988. 13(7): p. 748-755. 134. Halldin, P., A. Gilchrist, and N.J. Mills, A new oblique impact test for motorcycle helmets. International Journal of Crashworthiness, 2001. 6(1): p. 53-64. 135. Panjabi, M.M., et al., Mechanical properties of the human cervical spine as shown by three-dimensional load-displacement curves. Spine, 2001. 26(24): p. 2692-700. 136. Panjabi, M.M., et al., Development of a system for in vitro neck muscle force replication in whole cervical spine experiments. Spine, 2001. 26(20): p. 2214-9. 137. Zhu, Q., et al., Translational constraint influences dynamic spinal canal occlusion of the thoracic spine: An in vitro experimental study. Journal of Biomechanics, 2008. 41(1): p. 171-179. 138. Nelson, T.S. and P.A. Cripton. Inducing Head Motion with a Novel Helmet during Head-First Impact Can Mitigate Neck Injury Metrics: An Experimental Proof-of- Concept Investigation using Mechanical Surrogates. in International Research Council on the Biomechanics of Impacts. 2008. Bern, Switzerland. 139. Banerjee, R., M.A. Palumbo, and P.D. Fadale, Catastrophic cervical spine injuries in the collision sport athlete, part 1: epidemiology, functional anatomy, and diagnosis. Am J Sports Med, 2004. 32(4): p. 1077-87. 140. Reynen, P.D. and W.G. Clancy, Jr., Cervical spine injury, hockey helmets, and face masks. Am J Sports Med, 1994. 22(2): p. 167-70. 141. Watson, R.C., C.D. Singer, and J.R. Sproule, Checking From Behind in Ice Hockey: A Study of Injury and Penalty Data in the Ontario University Athletic Association Hockey League. Clinical Journal of Sport Medicine, 1996. 6(2): p. 108. 142. Attewell, R.G., K. Glase, and M. McFadden, Bicycle helmet efficacy: a meta- analysis. Accident Analysis & Prevention, 2001. 33(3): p. 345-352. 143. Liu, B., et al., Helmets for preventing injury in motorcycle riders. Cochrane database of systematic reviews (Online), 2004(2): p. CD004333. 144. Yoganandan, N., et al., Epidemiology and Injury Biomechanis of Motor Vehicle Related Trauma to the Human Spine. 1989: Society of Automotive Engineers, 400 Commonwealth Dr, Warrendale, PA, 15096, USA. 145. Taguchi, G., Y. Yokoyama, and Y. Wu, Taguchi methods: design of experiments. 1993: ASI press Dearborn, MI. 146. Zhang, M., Y.P. Zheng, and A.F.T. Mak, Estimating the effective Young's modulus of soft tissues from indentation tests—nonlinear finite element analysis of effects of friction and large deformation. Medical Engineering and Physics, 1997. 19(6): p. 512-517. 147. Kleinberger, M., et al., Development of Improved Injury Criteria for the Assessment of Advanced Automotive Restraint Systems. 1998. 148. SAE J211 - Instrumentation for Impact Test—Part 1—Electronic Instrumentation. 2003, SAE. 149. Finan, J., R. Nightingale, and B. Myers, The influence of reduced friction on head injury metrics in helmeted head impacts. Traffic Injury Prevention, 2008. 9(5): p. 483-488.  202 150. Hodgson, V.R., Impact, Skid, and Retention Tests on a Representative Group of Bicycle Helmets to Determine their Head-Neck Protective Charactertistics. 1990, Wayne State University: Detroit, MI. 151. Mertz, H.J., A.L. Irwin, and P. Prasad, Biomechanical and scaling bases for frontal and side impact injury assessment reference values. Stapp Car Crash J, 2003. 47: p. 155-88. 152. Mertz, H.J., et al., An Assessment of Compressive Neck Loads Under Injury- producing Conditions. The Physician and Sports Medicine, 1978. 6(11): p. 95-106. 153. Freiholtz, E., Numerical Analysis of an Impact Protection System with Focus on the Head, in Department of Aeronautics. 2000, KTH - Royal Institute of Technology: Stockholm, Sweden. 154. Myers, B.S. and R.W. Nightingale, Review: The Dynamics of Near Vertex Head Impact and its Role in Injury Prevention and the Complex Clinical Presentation of Basicranial and Cervical Spine Injury. Traffic Injury Prevention, 1999. 1(1): p. 67- 82. 155. Njus, G.O., Y. Liu, and T.A. Nye, The inertial and geometrical properties of helmets. Medicine & Science in Sports & Exercise, 1984. 16(5): p. 498. 156. Pal, G.P. and H.H. Sherk, The vertical stability of the cervical spine. Spine, 1988. 13(5): p. 447-449. 157. Benjamini, Y. and Y. Hochberg, Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society. Series B (Methodological), 1995. 57(1): p. 289-300. 158. Rowson, S., et al., Linear and angular head acceleration measurements in collegiate football. Journal of biomechanical engineering, 2009. 131: p. 061016. 159. Bishop, P.J. and R.P. Wells, The inappropriateness of helmet drop tests in assessing neck protection in head-first impacts. The American Journal of Sports Medicine, 1990. 18(2): p. 201. 160. Otis, J.C., A.H. Burstein, and J.S. Torg, Mechanisms and pathomechanics of athletic injuries to the cervical spine. Athletic Injuries to the Head, Neck, and Face. St Louis, Mo: Mosby--Year Book Inc, 1991: p. 438-456. 161. Carveth, W.H., Protective hockey helmet. 1996: United States. 162. McNeil, J.D., K. McNeil, and J. McNeil, Helmet providing cervical spine protection. 2005: United States. 163. Rush Iii, G.A., Helmet. 1996: United States. 164. Proubasta, I.R., et al., Biomechanical evaluation of fixation of clavicular fractures. Journal of the Southern Orthopaedic Association, 2002. 11(3): p. 148. 165. Yoganandan, N., et al., Physical properties of the human head: Mass, center of gravity and moment of inertia. Journal of Biomechanics, 2009. 42(9): p. 1177-1192. 166. Laporte, S., et al. Influence of Helmet Inertial Properties on Neck Injury Risk During Ejection: A Numerical Human Model Approach. 2005: Springer. 167. Thom, D.R., et al. Evaluating the positional stability of motorcycle helmets. 2000: Human Factors and Ergonomics Society. 168. Yang, K.H., et al. Developement of a finite element model of the human neck. 1998. Tempe, Arizona: 42nd STAPP Car Crash Conference.  203 169. Herbst, B., et al. Fidelity of Anthropometric Test Dummy Necks in Rollover Accidents. in 16th Annual International Technical Conference on the Enhanced Safety of Vehicles. 1998. Windsor, Ontario. 170. Piziali, R., et al., Injury causation in rollover accidents and the biofidelity of Hybrid III data in rollover tests, in SAE International Congress and Exposition. 1998, SAE: Detroit, Michigan. p. 105-118. 171. Kettler, A., et al., Mechanically simulated muscle forces strongly stabilize intact and injured upper cervical spine specimens. Journal of Biomechanics, 2002. 35(3): p. 339-346. 172. Ivancic, P.C., et al., Biofidelic whole cervical spine model with muscle force replication for whiplash simulation. European Spine Journal, 2005. 14(4): p. 346-355. 173. Ivancic, P.C., D. Sha, and M.M. Panjabi, Whiplash injury prevention with active head restraint. Clinical Biomechanics, 2009. 24(9): p. 699-707. 174. Saari, A., Spinal Cord Deformation during Axial Impact Injury of the Cervical Spine, in Mechanical Engineering. 2007, The University of British Columbia: Vancouver, BC. p. 126. 175. Kroeker, S.G., et al., The development of an improved physical surrogate model of the human spinal cord--Tension and transverse compression. Journal of Biomechanics, 2009. 42(7): p. 878-883. 176. Van Toen, C., et al. Simulation of head-first impact using cervical spine specimens, simulated neck muscles, and a Hybrid III ATD head. in Ohio State University's 5th Annual Injury Biomechanics Symposium. 2009. Columbus, Ohio. 177. Nuckley, D.J., et al., Examining the relationship between whiplash kinematics and a direct neurologic injury mechanism. International Journal of Vehicle Design, 2003. 32(1): p. 68-83. 178. Vasavada, A.N., et al., Defining and evaluating wrapping surfaces for MRI-derived spinal muscle paths. Journal of Biomechanics, 2008. 41(7): p. 1450-1457. 179. Vasavada, A.N., J. Danaraj, and G.P. Siegmund, Head and neck anthropometry, vertebral geometry and neck strength in height-matched men and women. Journal of Biomechanics, 2008. 41(1): p. 114-121. 180. Ryan, M.D. and J.J. Henderson, The epidemiology of fractures and fracture- dislocations of the cervical spine. Injury, 1992. 23(1): p. 38-40. 181. Nightingale, R.W., The Dynamics of Head and Cervical Spine Impact, in Biomedical Engineering. 1993, Duke University: Durham, NC. p. 167. 182. Nusholtz, G.S., Kinematics of the human cadaver spine in response to superior inferior loading of the head. GM crown head impact-neck. Final report. 1986. 183. Bishop, P.J. and W. Lu, Neck Segment Loading in a Dummy Model: Effect of Compression and Lateral Bending. Clinical Journal of Sport Medicine, 1994. 4(2): p. 113. 184. Torg, J.S., et al., National football head and neck injury registry: report on cervical quadriplegia, 1971 to 1975. Am J Sports Med, 1979. 7(2): p. 127-32. 185. Cook, D.J., et al., Evaluation of the ThinkFirst Canada, Smart Hockey, brain and spinal cord injury prevention video. Injury Prevention, 2003. 9: p. 361-366. 186. Heck, J.F., The incidence of spearing by high school football ball carriers and their tacklers. Journal of athletic training, 1992. 27(2): p. 120.  204 187. Heck, J.F., The incidence of spearing during a high school's 1975 and 1990 football seasons. Journal of athletic training, 1996. 31(1): p. 31. 188. Greenwald, R.M., et al., Head impact severity measures for evaluating mild traumatic brain injury risk exposure. Neurosurgery, 2008. 62(4): p. 789. 189. Broglio, S.P., et al., Head impacts during high school football: a biomechanical assessment. Journal of athletic training, 2009. 44(4): p. 342-349. 190. Guskiewicz, K.M., et al., Measurement of head impacts in collegiate football players: relationship between head impact biomechanics and acute clinical outcome after concussion. Neurosurgery, 2007. 61(6): p. 1244. 191. Mihalik, J.P., et al., Measurement of head impacts in collegiate football players: an investigation of positional and event-type differences. Neurosurgery, 2007. 61(6): p. 1229. 192. Cantu, R.C. and F.O. Mueller, Brain injury-related fatalities in American football, 1945-1999. Neurosurgery, 2003. 52(4): p. 846. 193. Torg, J.S., B. Sennett, and J.J. Vegso, Spinal injury at the level of the third and fourth cervical vertebrae resulting from the axial loading mechanism: an analysis and classification. Clinics in sports medicine, 1987. 6(1): p. 159. 194. Zhang, L., K.H. Yang, and A.I. King, A proposed injury threshold for mild traumatic brain injury. J Biomech Eng, 2004. 126(2): p. 226-36. 195. Bussone, W., et al., Six-Degree-of-Freedom Accelerations: Linear Arrays Compared with Angular Rate Sensors. SAE Technical Paper, 2010: p. 01-1017.   205 Appendix A  Machine Drawings of SC7 Head and Neck   Figure A-1:  SC7 head, neck, and helmet prototype assembly  206  Figure A-2:  SC7 head assembly    207  Figure A-3:  Skull cap      208  Figure A-4:  Head tube    209  Figure A-5: Motion guide inner   210  Figure A-6:  Head steel mass    211  Figure A-7:  C0 occiput – as built   212  Figure A-8:  SC7 neck assembly    213   Figure A-9:  C1 vertebra – as built   214  Figure A-10:  C2 vertebra – as built    215   Figure A-11;  C3 vertebra – as built   216  Figure A-12:  C4 vertebra – as built    217  Figure A-13:  C5 vertebra – as built    218  Figure A-14:  C6 vertebra – as built    219   Figure A-15:  C7 vertebra – as built   220  Figure A-16:  T1 vertebra – as built 221 Appendix B  Machine Drawings of Prototype Helmet   Figure B-1:  Prototype helmet full assembly  222   Figure B-2:  Prototype helmet top cap sheet 1 of 2 – preformed flat state   223   Figure B-3:  Prototype helmet top cap sheet 2 of 2 – formed state   224   Figure B-4:  Prototype helmet side support   225  Figure B-5:  Prototype helmet side support gusset    226  Figure B-6:  Prototype helmet motion deployment guide - right    227  Figure B-7:  Prototype helmet motion deployment guide - left  228 Appendix C  High Speed Video Sequences for Chapter 4 Experiments   Figure C-1:  High speed video frames with post-head-contact time for run 1  229   Figure C-2:  High speed video frames with post-head-contact time for run 2   230   Figure C-3:  High speed video frames with post-head-contact time for run 3   231   Figure C-4:  High speed video frames with post-head-contact time for run 4   232   Figure C-5:  High speed video frames with post-head-contact time for run 5   233   Figure C-6:  High speed video frames with post-head-contact time for run 6   234   Figure C-7:  High speed video frames with post-head-contact time for run 7   235   Figure C-8:  High speed video frames with post-head-contact time for run 8    236   Figure C-9:  High speed video frames with post-head-contact time for run 9   237   Figure C-10:  High speed video frames with post-head-contact time for run 10   238   Figure C-11:  High speed video frames with post-head-contact time for run 11   239   Figure C-12:  High speed video frames with post-head-contact time for run 12   240   Figure C-13:  High speed video frames with post-head-contact time for run 13   241   Figure C-14:  High speed video frames with post-head-contact time for run 14   242   Figure C-15:  High speed video frames with post-head-contact time for run 15   243   Figure C-16:  High speed video frames with post-head-contact time for run 16   244   Figure C-17:  High speed video frames with post-head-contact time for run 17   245   Figure C-18:  High speed video frames with post-head-contact time for run 18  246 Appendix D  Full Instrumentation Traces for Chapter 5 Experiments    Figure D-1: Full instrumentation output for specimen H1220    247  Figure D-2: Full instrumentation output for specimen H1221    248  Figure D-3: Full instrumentation output for specimen H1222    249  Figure D-4: Full instrumentation output for specimen H1223    250  Figure D-5: Full instrumentation output for specimen H1224    


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items