UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Cervical intervertebral kinematics and neck muscle responses during simulated vehicle rollovers Schoenfeld, M'Beth 2018

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

Item Metadata

Download

Media
24-ubc_2018_november_schoenfeld_mbeth.pdf [ 4.99MB ]
Metadata
JSON: 24-1.0371612.json
JSON-LD: 24-1.0371612-ld.json
RDF/XML (Pretty): 24-1.0371612-rdf.xml
RDF/JSON: 24-1.0371612-rdf.json
Turtle: 24-1.0371612-turtle.txt
N-Triples: 24-1.0371612-rdf-ntriples.txt
Original Record: 24-1.0371612-source.json
Full Text
24-1.0371612-fulltext.txt
Citation
24-1.0371612.ris

Full Text

CERVICAL INTERVERTEBRAL KINEMATICS AND NECK MUSCLE RESPONSES DURING SIMULATED VEHICLE ROLLOVERS by  M’Beth Schoenfeld  B.A.Sc, The University of British Columbia, 2015  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF APPLIED SCIENCE in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Biomedical Engineering)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)   August 2018  © M’Beth Schoenfeld, 2018  ii  The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, a thesis/dissertation entitled:  Cervical intervertebral kinematics and neck muscle responses during simulated vehicle rollovers  submitted by M’Beth Schoenfeld  in partial fulfillment of the requirements for the degree of Master of Applied Science in Biomedical Engineering  Examining Committee: Dr. Peter Cripton Co-supervisor Dr. Gunter Siegmund Co-supervisor  Dr. Mark Carpenter Supervisory Committee Member Dr. Antony Hodgson Additional Examiner    Additional Supervisory Committee Members:  Supervisory Committee Member  Supervisory Committee Member iii  Abstract Vehicle rollovers account for 3% of motor vehicle crashes yet cause one-third of all crash-related fatalities. Despite advanced cervical spine injury models, a discrepancy exists between clinically reported injuries and cadaver test pathologies. One possible explanation for this discrepancy is that the intervertebral posture and simulate muscle tone used in cadaver models (and computer models) typically mimic an upright and relaxed condition that may not exist during a rollover. The aim of this work was to characterize vertebral alignment and neck muscle responses in the cervical spine by studying a human subject in a simulated impending headfirst impact, in an upside-down configuration. A custom inversion device was built to expose human subjects to a 321 ms inverted free fall drop. An onboard fluoroscopic C-arm captured cervical vertebral motion while indwelling electromyography captured the response of 8 superficial and deep neck muscles. The subject shoed consistent muscular responses in 4 repetitions of the free-fall exposure. Moreover, the muscle response pattern was different from the scheme used in existing cervical spine injury models and observed in previous quasi-static tests conducted in our lab. The general trends in muscle-induced changes to vertebral alignment were consistent with our previous work. C3-C6 translated anteriorly and inferiorly in response to the inverted free fall stimulus, and the head moved into flexion. These observations suggest that, at the time of impact, the in vivo state of the neck may differ considerably from its initial alignment prior to the forewarned impact. The in vivo data set acquired from this experiment of vertebral and muscular responses could be used to improve and validate current injury models and advance injury prevention strategies in rollover crashes.  iv  Lay Summary Vehicle rollovers are a significant source of major injuries. Specifically, rollovers can cause neck injury, through headfirst impact when an occupant “dives” into the roof during roof-to-ground impact. Rollover neck injury mechanisms are poorly understood and although cadaver studies have been used to study rollovers, they lack muscles and because rollovers take several seconds to develop, muscles are important to the spine posture during rollovers. Neck injury is sensitive to neck muscle activity and posture, yet the pre-impact state of the neck is unknown. To fill this gap, we exposed a human volunteer to a simulated vehicle rollover and recorded their neck muscle activity and posture. We found that neck muscle activity and posture differ from the manner in which they’ve been represented in previous research. This information can be used to improve existing methods of studying neck injury in rollover crashes.            v  Preface This thesis represents my own work under the guidance of Dr. Peter Cripton and Dr. Gunter Siegmund. The original project idea was developed by Dr. Robyn Newell, Dr. Gunter Siegmund, and Dr. Peter Cripton. Mircea Oala-Florescu completed the mechanical design, fabrication, and assembly of the inversion apparatus. I completed the commissioning of the inversion apparatus under the help and guidance of Jeff Nickel. The verification of the inversion apparatus (Chapter 3) was completed by myself with help from Jeff Nickel. The experimental protocol (Chapter 4) was developed by myself with input from Jason Fice, Daniel Mang, Dr. Gunter Siegmund, and Dr. Peter Cripton. Data analysis was completed by me, under the guidance of Dr. Jean Sebastien Blouin and Dr. Gunter Siegmund, with input from Dr. Peter Cripton.  A version of Chapter 4 has been published as part of the 14th Annual Injury Biomechanics Symposium conference proceedings: Schoenfeld MS, Fice J, Blouin JS,  Siegmund GP, Cripton P. Cervical vertebral kinematics and neck muscle responses during an inverted free fall simulating a vehicle rollover: pilot data from an in vivo human subject experiment.  The research conducted in Chapter 4 of this thesis was approved by the University of British Columbia’s Clinical Research Ethics Board (certificate # H07-01445) and the subject gave their written informed consent.   vi  Table of Contents  Abstract .......................................................................................................................................... ii Lay Summary ............................................................................................................................... iv Preface .............................................................................................................................................v Table of Contents ......................................................................................................................... vi List of Tables ................................................................................................................................ xi List of Figures .............................................................................................................................. xii List of Abbreviations ................................................................................................................. xiv Glossary ........................................................................................................................................xv Acknowledgements .................................................................................................................... xvi Dedication ................................................................................................................................. xviii Chapter 1: Introduction ................................................................................................................1 1.1 Overview and motivation ................................................................................................ 1 1.2 Axial impact injury to the cervical spine ........................................................................ 2 1.2.1 Scenarios ..................................................................................................................... 2 1.2.2 Cervical spine injury mechanisms .............................................................................. 4 1.2.3 Cervical spine injury in vehicle rollovers ................................................................... 6 1.2.4 Influence of neck musculature .................................................................................... 8 1.2.5 Influence of cervical spine posture ........................................................................... 11 1.3 Existing methods of studying axial impact cervical spine injury ................................. 13 1.3.1 Cadaver studies ......................................................................................................... 13 1.3.2 Computer models ...................................................................................................... 15 vii  1.3.3 Anthropomorphic test devices (ATDs) ..................................................................... 18 1.3.4 Human volunteers ..................................................................................................... 19 1.3.4.1 In vivo neck kinematics ..................................................................................... 21 1.3.4.2 Electromyography ............................................................................................. 22 1.4 Thesis objectives and hypotheses ................................................................................. 23 Chapter 2: Commissioning of an inversion apparatus to simulate impending headfirst impact ............................................................................................................................................26 2.1 Design Goals ................................................................................................................. 26 2.1.1 Rationale for minimum free fall duration ................................................................. 28 2.1.2 Rationale for limiting deceleration to 1g .................................................................. 30 2.2 Mechanical description of final design ......................................................................... 31 2.3 Design commissioning .................................................................................................. 34 2.3.1 Derivation of linear motor motion profile ................................................................ 34 2.3.1.1 Linear motor control program ........................................................................... 38 2.3.2 Retro-fit of an existing fluoroscopic C-arm .............................................................. 40 2.3.3 Image analysis method .............................................................................................. 43 2.3.4 EMG accommodation ............................................................................................... 48 2.3.5 Risk mitigation .......................................................................................................... 50 2.3.5.1 Hardware ........................................................................................................... 50 2.3.5.2 Software ............................................................................................................ 51 Chapter 3: Verification of an inversion apparatus to simulate impending headfirst impact55 3.1 Introduction ................................................................................................................... 55 3.2 Verification methods ..................................................................................................... 56 viii  3.2.1 Apparatus .................................................................................................................. 56 3.2.2 Instrumentation ......................................................................................................... 58 3.2.3 Experimental verification testing .............................................................................. 59 3.2.4 Analysis..................................................................................................................... 60 3.2.4.1 Linear encoder data ........................................................................................... 60 3.2.4.2 Accelerometer data ........................................................................................... 60 3.2.4.3 Free fall duration ............................................................................................... 62 3.3 Verification results ........................................................................................................ 63 3.3.1 Kinematics ................................................................................................................ 63 3.4 Discussion ..................................................................................................................... 67 3.5 Conclusion .................................................................................................................... 70 Chapter 4: Cervical vertebral kinematics and neck muscle responses during an inverted free fall simulating a vehicle rollover: pilot data from an in vivo human subject experiment72 4.1 Introduction ................................................................................................................... 72 4.2 Methods......................................................................................................................... 73 4.2.1 Human subject .......................................................................................................... 73 4.2.2 Custom-built inversion device .................................................................................. 74 4.2.3 Conditions ................................................................................................................. 74 4.2.4 Electromyography ..................................................................................................... 75 4.2.4.1 Maximum voluntary contractions (MVC’s) ..................................................... 76 4.2.5 Cervical spine posture ............................................................................................... 77 4.2.6 Head orientation ........................................................................................................ 79 4.2.7 Data processing notes ............................................................................................... 80 ix  4.2.8 Error estimates of head and neck posture ................................................................. 80 4.3 Results ........................................................................................................................... 81 4.3.1 Muscle activity .......................................................................................................... 81 4.3.2 Neck and head posture .............................................................................................. 86 4.3.2.1 Error estimates of head and neck posture ......................................................... 87 4.4 Discussion ..................................................................................................................... 88 4.5 Conclusion .................................................................................................................... 93 Chapter 5: Discussion ..................................................................................................................95 5.1 Overview ....................................................................................................................... 95 5.2 Strengths and limitations............................................................................................... 98 5.2.1 Strengths ................................................................................................................... 98 5.2.2 Limitations ................................................................................................................ 99 5.3 Potential applications .................................................................................................. 103 5.4 Future work & recommendations ............................................................................... 105 5.5 Conclusions ................................................................................................................. 108 Bibliography ...............................................................................................................................109 Appendices ..................................................................................................................................119 Appendix A Code used to generate the linear motor’s position profile ................................. 119 A.1 Matlab code used to generate the linear  motor’s position profile .......................... 119 A.2 Matlab code used to convert position increments to DMC commands .................. 122 Appendix B Linear motor control program flowchart and DMC code .................................. 123 B.1 Program Flowchart.................................................................................................. 123 B.2 DMC Code .............................................................................................................. 124 x  Appendix C Risk mitigation documents ................................................................................. 137 Appendix D Experimental Protocol ........................................................................................ 150  xi  List of Tables Table 1-1 Cervical spine injury classification ................................................................................ 6 Table 3-1 Tabulated maximum and minimum accelerations experienced by the subject carriage and surrogate ................................................................................................................................. 66 Table 4-1 Summary of normalized EMG rms average ................................................................. 84 Table 4-2 Summary of vertebral and Frankfort plane angles. ...................................................... 87 Table 4-3 Error estimates of vertebral translations ....................................................................... 87 Table 4-4 Error estimates of vertebral and Frankfort plane angles .............................................. 88 Table 5-1 Criteria for risk (severity) level .................................................................................. 137 Table 5-2 Semi-quantitative probability levels ........................................................................... 138 Table 5-3 Hazard analysis risk index table ................................................................................. 138 Table 5-4 Risk action key ........................................................................................................... 138 Table 5-5 Hazard analysis table .................................................................................................. 140  xii  List of Figures Figure 1-1 The effect of head pre-flexion in cervical spine buckling ........................................... 13 Figure 2-1 Photo of the final inversion rig design ........................................................................ 32 Figure 2-2 Steel bracing securing x-ray source and image intensifier assembly to the subject carriage .......................................................................................................................................... 34 Figure 2-3 Depiction of motion between the subject carriage (S) and the linear motor (M) ....... 35 Figure 2-4 Diagram of the problem set-up ................................................................................... 36 Figure 2-5 Calculated motion profiles .......................................................................................... 38 Figure 2-6 Right angle optics ........................................................................................................ 41 Figure 2-7 Fluoroscopic image of two human vertebra captured at 200 Hz and 1280x800 resolution....................................................................................................................................... 43 Figure 2-8 Image distortion due to misalignment between a source and detector ........................ 44 Figure 2-9 A scaled scan of the steel bead array used to rectify images ...................................... 45 Figure 2-10 The bead array plate is mounted in front of the x-ray source and is visible in the fluoroscope’s field of view. .......................................................................................................... 45 Figure 2-11 Spherically distorted (left) and undistorted (right) images collected from the fluoroscopic C-arm ....................................................................................................................... 46 Figure 2-12 Visual summary of the photo rectification process applied to an image of the bead array plate ...................................................................................................................................... 47 Figure 2-13 T-shaped mounting shelf for EMG pre-amplifiers .................................................... 48 Figure 2-14 The pre-amplifiers mounted relative to an example subject ..................................... 49 Figure 2-15 Upper limit switch mounted to positive stop ............................................................ 52 Figure 2-16 L-brackets and optical sensors .................................................................................. 54 xiii  Figure 3-1 The custom-built inversion apparatus raised to the fixed drop height with an ATD subject ........................................................................................................................................... 57 Figure 3-2 Fluoroscopic C-arm centered on ATD's neck ............................................................. 58 Figure 3-3 ASTM F2291 – 17 coordinate systems ....................................................................... 61 Figure 3-4 An example of the free fall duration measurement ..................................................... 62 Figure 3-5 Kinematic profiles of the inversion apparatus and surrogate ...................................... 65 Figure 3-6 Measured accelerations compared to ASTM F2291 – 17 limits ................................. 67 Figure 4-1 Four video frames from the inverted drop condition .................................................. 75 Figure 4-2 Sample of the image analysis process ......................................................................... 78 Figure 4-3 Method by which vertebral angles were determined .................................................. 79 Figure 4-4 Method by which Frankfort plane angle was measured .............................................. 79 Figure 4-5 Unprocessed EMG signal from Trial 1 ....................................................................... 82 Figure 4-6 Onset of normalized rms EMG for each trial .............................................................. 83 Figure 4-7 Normalized rms EMG for each muscle across all four trials. ..................................... 85 Figure 4-8 Summary of vertebral translations throughout all trials .............................................. 86 Figure 4-9 Summary of maximum muscle activity between free fall and a voluntary bracing posture ........................................................................................................................................... 90 Figure 4-10 Comparison of cervical spine posture during free fall and a voluntary bracing task 92  xiv  List of Abbreviations  ATD  Anthropomorphic test device CoG  Center of gravity CSI   Cervical spine injury DMC  Dynamic motion control EMG  Electromyography G  Gravity I-F  Inverted-forward I-R  Inverted-relaxed LS  Levator scapulae MultC4 Multifidus MVC  Maximum voluntary contraction RMS  Root mean square SCM  Sternocleidomastoid SsCap  Splenius capitus SsCerv  Semispinalis cervacis STH  Sternohyoid Trap  Trapezius U-R  Upright-relaxed  xv  Glossary  Biofidelity  the ability of a given surrogate to approximate the behavior of a living human under comparable loading conditions  Optocollic reflex  reflexive movements of the head (by neck muscles) that stabilize the visual image on the retina; it is triggered by neural signals that measure retinal image slip  Somatosensory system  a multifaceted system of neurons and sensory celss that provide an organism with information about the physical state of its body including temperature, limb position, and pressure on the skin  Vestibular system  a collection of several cell groups that receive sensory input from the vestibular apparatus (the semicircular canals, utricle, and sacculus) and convey this information to a variety of other brain regions  Vestibulocollic reflex  a reflexive counterrotation of the head when the body turns (using neck muscles); it stabilizes the visual image on the retina and is triggered by head-rotation signals from the semicircular canals    xvi  Acknowledgements I would like to thank my co-supervisors, Dr. Peter Cripton and Dr. Gunter Siegmund, for their support and guidance throughout this project. Peter, thank-you for the many hours you spent securing the grants and ethics approvals to complete this project. I also appreciate the time you spent helping me prepare for presentations, talking-through my concerns, and contributing insightful edits to manuscripts and this thesis. I owe particular thanks to Dr. Gunter Siegmund, who’s feedback and insights guided my learning experience. Thank-you for carving out the time in your schedule to discuss my questions at length, share your knowledge, and promote high-quality research. I feel lucky to have had access to someone like you.  Thank-you to Jeff Nickel for your providing your expertise in comedy and electrical instrumentation, in that order. Without your help this project would have felt insurmountable.   Thank-you to Mircea Oala-Florescu for your hard work in designing, building, re-designing, re-building, and re- re- designing and building the inversion rig.   Thank-you to my fellow students and the OIBG lab group for your collaboration and support. Special thanks are owed to Jason Fice for enduring 18 hours of testing, and many more of trouble-shooting Windows Vista (on a weekend). I’m eternally grateful for your selfless attitude and willingness to help.    Thank-you to my parents, sisters, and friends, for all the smiling and nodding you did while I went on-and-on-and-on about my project.   xvii   Lastly, thank-you to Jacob, for all the hot nights.  xviii  Dedication  To knowing all the things I never knew I didn’t know.   1  Chapter 1: Introduction  1.1 Overview and motivation Injury to the cervical spine may result in upper spinal cord injury which is associated with devastating neurological consequences such as tetraplegia and death. Cervical spine injury may be the result  of headfirst axial impact where the head is suddenly arrested, and the neck is loaded by the incoming momentum of the torso. This loading scenario is common in equestrian sports, mountain biking, diving and vehicle rollover crashes. Vehicle rollover crashes have been reported as the most frequent trauma mechanism leading to cervical spine injury [1], and as such this thesis is presents and discusses cervical spine injury in the context of vehicle rollover crashes primarily.   Clinically-reported cervical spine injuries, as a result of rollover crashes, differ from laboratory-induced cervical spine injuries (section 1.2.3) . It has been suggested that the absence of musculature, and muscle-induced changes in cervical spine posture, in lab-induced injuries are responsible for this disparity[2]. Cadaver studies and computational models have indicated that pre-impact spinal alignment [3–5] and neck musculature [6–8] significantly affect the injury mechanism (sections 1.2.4 & 1.2.5). However, current methods of studying axial impact cervical spine injury do not use cervical spine postures or muscle activations relevant to real-world axial impact cervical spine injury [6,7,9–13] (section 1.3). Furthermore, these existing methods are validated using data from human surrogates or human volunteers in unrepresentative loading scenarios (ex. frontal impacts) [6,7,12,13]. Little information is known about the exact in vivo state of the neck prior to an impact due to the challenges involved with human subject testing [14]. Previous work from our lab has investigated in vivo neck kinematics and muscle activity 2  under static conditions [15–17], but the in vivo state of the neck under impending headfirst impact is still unknown.   The ideas introduced above are discussed in more detail and depth in the forthcoming sections of this introductory chapter. The first section provides an overview of axial impact injury to the cervical spine, discussing the injury patterns observed in clinical cases versus laboratory-induced injuries. The influence of musculature and cervical spine posture is explored by discussing the results of previous cadaveric studies and computational models.  In the second section, existing methods of studying axial impact injury to the cervical spine are defined with a discussion of their limitations. Special attention is given to the previous human subject work conducted in our lab, which laid the foundation for this thesis. Lastly, this introductory chapter concludes with an explicit statement on the objectives any hypothesis of this thesis.  1.2 Axial impact injury to the cervical spine 1.2.1 Scenarios Serious cervical spine injury and spinal cord injuries can result from axial impacts to the head, which occur in a variety of scenarios. Equestrian sports, mountain biking, diving, and vehicle rollover crashes are examples of scenarios associated with axial impact that leads to cervical spine injury.  In British Columbia (BC), Canada, the rate of hospital admissions for horseback riding injuries is approximately 1 admission per 2000 hours of riding [18]. In Alberta, Canada, trauma injuries to the head represented 48% of horseback injury cases reviewed over a 10-year period [19]. In equestrian sports, cervical spine fracture is the 8th most common injury when falling from a horse [20] (according to a review of participating European trauma centers). In some 3  cases, a rider may be ejected forward from the horse and free fall headfirst toward the ground [21]. Christopher Reeves, best known for his role in Superman, sustained complex fractures to this first and second vertebrae after being thrown from his horse and landing on his head [22].  Mountain biking is a sport gaining increased popularity throughout North America, especially in British Columbia. Between 1995 and 2007, the mean risk of spinal injury from mountain biking was 0.20 per 100,000 BC residents. Over the same 13-year period, 3.7% of all provincial spinal cord injuries were sustained while mountain biking [23]. In mountain biking, the distribution of spinal injuries is overwhelming localized to the cervical spine, representing 73.8% of spinal injuries. Of clinically admitted cases, being propelled over the handlebars is the most common injury mechanism, after which there is a 91% chance of direct impact to the head and resultant cervical spine injury [23].  Recreational swimming and diving is thought to be the third most common recreational activity in the United States (US) [24]. Among swimming injuries treated in US hospital Emergency Departments, the most commonly injured body region is the head & neck; of total head & neck injuries, neck injuries account for 11% [25]. In diving, cervical spine injuries occur in cases where the diver’s head strikes the bottom of the pool (or lakebed, etc.). After the diver’s head strikes the bottom of the pool, the rest of the diver’s body is driven into the neck, causing catastrophic neck injury [26].  Vehicle rollovers are a chaotic and complex type of motor vehicle crash which account for one-third of crash-related fatalities [27] and 40% of serious cervical spine injuries [28]. The annual incidence of rollover crash related fatalities is 3.4 per 100,000 people. Neck injury can occur when the vehicle roof hits the ground and the inverted occupant strikes the interior of the 4  vehicle with their head [29,30]. Similarly to diving injuries, the neck is subsequently loaded by the incoming momentum of the torso [31].  Vehicle rollover crashes have been reported as the most frequent trauma mechanism leading to cervical spine injury [1]. Given their higher incidence rate, the rest of this thesis focuses on cervical spine injury in the context of vehicle rollover crashes.   1.2.2 Cervical spine injury mechanisms Cervical spine injury is influenced by the rate, magnitude, and direction of loading. As with most structures that are composed of viscoelastic, composite components, the mechanical properties of the cervical spine and it’s components vary with varying load conditions. Kazarian and Graves evaluated the compressive strength properties of the vertebral centrum (anterior vertebral body) from loading rates between 0.53 and 5334 cm/min. They found that the ultimate failure load and stiffness values approximately doubled from the lowest loading rate to the highest loading rate [32]. Yoganandan studied the dynamic axial tensile loading characteristics of cervical spine ligaments at loading rates from 9 mm/s to 2500 mm/sec [33]. A nonlinear increase in biomechanical properties of force to failure between the low and high loading rates was observed. Again, force to failure and stiffness showed approximately a two- to threefold increase in values from the low to high loading rates [33]. These rate-dependent responses of the cervical spine alter its mechanical properties, which may in turn influence injury mechanisms. Magnitude of loading understandably influences cervical spine injury, as once the magnitude exceeds the injury tolerance, injury occurs. However, critical thresholds for injury are difficult to estimate due to the large number of conditions that have been varied, or uncontrolled for, in past experiments. For example, Nusholtz et al. [3] were able to produce fractures to the 5  C6-C7 vertebra of a specimen with only 1,800 N of impact force, whereas Alem et al. [34] impacted a specimen with 16,000 N in a similar manner with no injury. It is difficult to use force magnitude as a predictor of injury since critical force levels are very sensitive to other mechanical factors such as end conditions, C1-C7 spinal eccentricity, and spinal curvature [9,11,34]. The aforementioned mechanical factors are provided as examples of some of the aspects of cervical spine injury that have been studied and are not an exhaustive list.   Loading direction is an important factor in cervical spine injury. Tensile and compressive loading directions, among others, each have different associated injury patterns. The classification of neck injury is complicated and has changed as more is learned about injury mechanisms. Myers et al. [35] describe one classification scheme in which the majority of cervical injuries are classified into nine categories: compression, compression-flexion, compression-extension, tension, tension-extension, tension-flexion, torsion, shear, and lateral bending (Table 1-1). This classification scheme, and the observed injuries, is based on experimental observations between direction of loading and localized bony injury (using cadaveric cervical specimens). During axial impact, the cervical spine may experience flexion, extension, anterior/posterior translation and compressive loading conditions alone, or in combination. Most commonly, rotation is coupled with a compressive load during an injury event and results in a combination of injuries [36]. Generally, pure compressive injuries occur under higher axial loads than compression-flexion or compression-extension injuries. Compression fractures have been produced at an average loading of 4810 N, while bilateral facet dislocations have been produced at an average loading of 1720 N, purely by adjusting rotational and translational constraints on the spine specimen [37]. 6  Table 1-1 Cervical spine injury classification The classification of cervical spine injuries based on applied forces with experimental validation [35]. The loading conditions are bolded, and their associated cervical spine injuries are listed underneath.   Compression  Tension-extension Jefferson Fracture  Hangman's fracture Multiple atlas fracture  Anterior longitudinal ligamentous damage Vertebral body compression fracture  Disc rupture Teardrop fracture  Horizontal fracture of vertebral body   Teardrop fracture Compression-flexion    Teardrop fracture  Tension-flexion Burst fracture  Bilateral facet dislocation  Wedge compression fracture  Unilateral facet dislocation Hyperflexion sprain   Bilateral facet dislocation  Torsion Unilateral facet dislocation  Atlantoaxial rotary dislocation    Unilateral atlantoaxial facet dislocation  Compression-extension   Hangman's fracture  Shear Clay-shoveler's fracture  Odontoid fracture Posterior element fracture  Transverse ligament rupture Anterior longitudinal ligamentous rupture   Anterior disc rupture  Lateral bending (in combined loading) Horizontal vertebral body fracture  Asymmetric injury Teardrop fracture  Nerve root avulsion    Peripheral nerve injury Tension   Occipitoatlantal disolcation     1.2.3 Cervical spine injury in vehicle rollovers Vehicle rollovers are chaotic and dynamic events in which a vehicle undergoes at least a 90 degree rotation about either its longitudinal or lateral axis [38]. This thesis is concerned with rollovers undergoing a rotation about the longitudinal axis. When lateral forces create a large enough moment about a vehicle’s center of gravity for a sufficient amount of time, the vehicle will rollover. Rollover crashes may occur in one of two ways: tripped or un-tripped. Tripped 7  rollovers occur as a result of collision with objects or sliding sideways into soft soil. Curbs, soft soil/road shoulders, guardrails, pavement surface discontinuities, snow banks, or other physical objects can cause high tripping forces to the vehicle tires and instigate a rollover [39]. Un-tripped rollovers are less common, and usually occur during high-speed avoidance maneuvers on the roadway, such as swerving; either the tires themselves create enough lateral forces to tip the vehicle or the rims dig into the road surface. Un-tripped rollovers may also occur while travelling at a high speed along a curved road, with centrifugal forces generating a roll moment about the vehicle’s longitudinal axis.  In many rollovers, the head interacts with the vehicle’s roof structures, leaving the head and cervical spine vulnerable to compressive forces [40]. This is possible partly because 3-point safety belts were designed for frontal impacts and aren’t designed to limit upwards movement of an occupant. Of clinically reported rollover cases involving cervical spine injury (CSI), the predominant loading mechanism in rollover occupants is an axial load applied to the cervical spine. The neck injury mechanism is described as initial head impact against the roof, followed by subsequent loading by the incoming torso [29,30]. Laterally eccentric load vectors are also frequently associated with injury, creating lateral neck flexion and asymmetric fracture distributions [2]. A large proportion of serious neck injuries in rollover crashes are represented by compression-flexion injury mechanisms [41]. Fracture of the articular facets is recorded as the most common CSI sustained by rollover occupants, comprising 34.7% of the total fracture distribution [2]. Rollover occupants with CSI typically have concurrent injuries. As the head interacts with the interior of a vehicle throughout a rollover, occupants often have evidence of impact to the forehead and face [2,42,43]. Additionally, approximately one-quarter of rollover occupants who sustain neck injuries by roof impact also sustain serious injuries to the thoracic 8  spine [43]. Due to the complex and chaotic nature of a rollover, it is usually impossible to deduce the exact loading mechanism in real-life rollovers.   Ideally, the injury distribution between clinically reported and lab-induced injuries should be fairly similar. However, major differences exist in the type of injury and fracture location between clinical cases and lab-created injuries. Clinically, the majority of rollover occupants experience facet dislocations, often asymmetrical, at the lower end of the cervical spine while lab-induced injuries typically occur as vertebral body fractures, often symmetric. Facet dislocations are the second-least common injury induced in a laboratory setting [2]. These lab-induced injuries are performed by subjecting cadaveric specimens to representative loading conditions. One suggested reason for the injury disparity is that cadaveric specimens lack musculature. Passive musculature pre-loads the cervical column, and determines the spine posture at impact, which could have an effect on injury outcome. Active musculature may have the ability to alter the clinical presentation of injury [44]. Generally, in laboratory tests an upright and neutral cervical spine posture is often simulated, which is not representative of an inverted orientation [15]. Thus, both active musculature and cervical spine posture may be partly responsible for differences in injury types, location of fracture, and fracture symmetry. However, simulation of a relevant cervical spine posture and muscle activation scheme is difficult, as the in vivo state of the neck prior to an impact is still unknown.   1.2.4 Influence of neck musculature In the past, the influence of neck musculature has been neglected in axial impact injury. There are two primary reasons why neck musculature has been overlooked. Firstly, neck injury resulting from axial impact occurs before muscles have sufficient time to reflexively respond to 9  impact itself. Neck injury typically occurs within 20 ms of impact [45], while reflexive muscle activation requires at least 54 ms and an additional 60 ms to fully contract [46]. However, common vehicle roll rates of 200-300 deg/s leave occupants up to 900 ms to react, which is ample time for neck muscles to develop a response between the onset of roll and a headfirst impact after one-half roll [47]. Secondly, it has been assumed that muscles do not significantly affect the cervical spine response in compression because they do not bear compressive loads. Muscles create tensile forces between fixed origin and insertion sites by contracting; thus, muscles only exert ‘pull’ forces and cannot exert ‘push’ forces, similar to a rope, and consequently do not act in compression. Although neck muscles cannot support compressive loading (beyond passive effects), they can alter the biomechanics of the cervical spine and influence injury outcome.  Some computer models investigating axial impact cervical spine injury have concluded that active neck musculature likely increases the chances of neck injury [6,7]. When simulating select neck muscles in their maximally active state, Hu et al. observed that including active neck musculature nearly doubled the risk of fracture [6]. When simulating a muscle activation scheme optimized for an upright and relaxed posture, Nightingale et al. observed, on average, 32% larger peak compressive forces in the cervical spine compared to simulations without musculature [7]. The effects of passive musculature have also been shown to increase the risk of neck injury. A passive muscle model by Hu et al. showed that passive musculature increased injury risks, albeit only slightly higher than those predicted by a no-muscle model [6]. Also, cadaveric specimens constrained by simulated muscle forces have higher rates and severities of injury [8]. Thus, neck injury risk seems to increase with the inclusion of neck musculature, whether active or passive.  10   In the in vivo environment, muscle forces contribute to stabilizing joints, aligning the spine, and thereby indirectly helping to support the weight of the head. Living, healthy humans are able to support more than the weight of their head, whereas the osteoligamentous spine, i.e., a spine with no muscles, has been shown to buckle under compressive loads of approximately 10 N (about ¼ the weight of a human head) [48]. In cadaveric specimens, simulating only a few muscles (using systems of cables and pulleys) stabilizes the  cervical spine by reducing its range of motion [49]. These examples demonstrate musculature’s primary role in regulating the stability of the spinal column. Active, or passive, musculature may stiffen the vertebral column and influence the development of forces throughout the spinal column during an impact. By increasing joint stiffness, active musculature increases the critical buckling load of the cervical spine in axial impact injuries [7]. Additionally, the added mass of neck musculature may provide an inertial resistance to buckling [50]. By adding resistance to buckling, neck musculature may shift the mode of injury [7].  As neck muscles contract, the cervical vertebrae are subjected to a compressive pre-load. Computer models of pre-impact muscle tensing predicted that neck muscles may generate   compressive pre-loads reaching 800-1400 N in the spine [51]. Other numerical simulations suggest these pre-loads may reach up to 40% of the neck’s injury tolerance in compressive loading [52,53]. This compressive pre-load puts the state of the neck closer to injury thresholds prior to an impact, thereby increasing the risk of neck injury. It has been shown that the compressive pre-loads increase caudally, again suggesting that the presence of active musculature is capable of shifting injury location. Therefore, active musculature affects the biomechanics of the spine and the risk of cervical spine fracture prior to an impact [53].   11  1.2.5 Influence of cervical spine posture The initial posture of the cervical spine is one of the most important factors in determining injury type and risk [3,4]. Active musculature generates muscle-induced changes in cervical spine posture, and spinal eccentricity, curvature, and alignment of the head with respect to the spine are important postural factors that influence cervical spine injury in axial headfirst impact.   Neck eccentricity is defined as the anterior or posterior offset between the superior (C1) and inferior (C7) cervical vertebra. Injuries to the spinal column are sensitive to small changes in neck eccentricity. Maiman et. al investigated the effects of pre-injury cervical alignment on axial-impact injury in cadaveric head-neck complexes [5]. The extent of which the occipital condyles were offset from T1 (the first thoracic vertebra) significantly influenced the mechanism of injury. Eccentricities of -0.5 (posterior), 2.3, 5.3, and 0.1 cm (anterior) resulted in compression-extension, compression-flexion, hyperflexion, and vertical compression injury mechanisms, respectively [5]. These results show that even small changes in the eccentricity of the cervical spine can change the mechanism of axial impact injury.   Adjacent vertebrae are coupled via ligaments and musculature, and thus spinal eccentricity and curvature are linked. Spinal curvature relates to localized eccentricities of adjacent cervical vertebra, which affects the path of load transmission throughout the cervical spine. The aligned spine, with no lordosis or kyphosis, is generally accepted as the most vulnerable posture of the osteoligamentous spine [16]. In the aligned configuration, the spinal column cannot bend out of the way  from the incoming load of the torso, and thus the vertebrae often fracture in this scenario. Neck flexion and neck extension allow the neck to bend out of the 12  way of the load path of the incoming torso. Generally, slight lordosis results in extension injuries and slight kyphosis result in flexion injuries [9].  Head alignment is also linked to neck curvature; as the head flexion angle is adjusted, the natural curvature of the spine changes. A computational study by Nightingale et al. demonstrated that pre-impact head flexion angles tend to straighten the cervical spine, and lead to higher-order buckling modes (Figure 1-1)  [53]. Failure loads associated with higher-order buckling are typically larger because the critical load increases with the degree of buckling. Therefore, in cases with a straightened spine, the cervical vertebra may fracture in compression before the spinal column’s critical buckling load is reached. Higher order buckling, which is observed when the spine is pre-flexed by angles near 32 degrees [7], does not necessarily translate to an increased injury risk. The critical buckling load in first-order buckling is lower, and the failure mode is a combination of compression, shear, and bending in a more compliant configuration [7]. Therefore, head alignment may change the type of injury, but not necessarily correspond to a lower injury risk.  13   Figure 1-1 The effect of head pre-flexion in cervical spine buckling A neutral cervical spine posture results in first-order bucking modes (left), whereas 32 degrees of neck pre-flexion results in the expression of higher-order buckling modes (right) [7].   1.3 Existing methods of studying axial impact cervical spine injury 1.3.1 Cadaver studies Post mortem human surrogates (PMHS) provide an exact anatomical representation of human subjects (complete with inter-subject variation). Fresh cadavers are a more representative surrogate, as opposed to embalmed or fresh-frozen specimens, as the material properties of some soft tissues change rapidly post mortem [54]  (muscles, spinal cord, brain, but ligaments and vertebral discs may retain their mechanical properties after death if they’re frozen). Whole-body cadavers have been used in a drop configuration to investigate the interaction of the head, neck, and torso, while including loading effects from the incoming torso momentum [4,8]. Cadavers may also be segmented to study isolated body regions. Yoganandan et al. used cadaver head-14  neck complexes, isolated by transection at T2-T3 to investigate geometric effects of the cervical spine on injury mechanism [10]. When manipulation of the spine end conditions is desired, other researchers have tested PMHS neck segments without a head [37].  When more precise control over loading parameters is desired, other studies have isolated smaller segments of the cervical spine [55,56]. Materials testing machines, pendulum and linear impactors, or dropping specimens from a known height are used to control the loading conditions applied to cadaveric specimens. In order to visualize the spinal column soft tissues, primarily musculature, are typically removed. Since cadavers are essentially atonal, external hardware (e.g., systems of cables, pulleys, and springs) or tailored embalming techniques are used to approximate muscle tone [54]. Additionally, masking tape and other temporary supports (designed to break or detach during impact) are sometimes used to maintain specimens in a pre-impact posture [14]. Generally, these previous cadaver tests have been aimed at studying axial headfirst impact injuries and have not been specific to rollover applications. Two studies have used whole-body cadavers to investigate PMHS response in vehicle rollovers. Lessley et al. characterized the whole-body kinematic response of restrained PMHS in a variety of controlled laboratory rollover conditions [57]. Moffat et al. investigated head excursion during the airborne phase of a rollover by subjecting seat-belted cadavers to rollover simulations under static and dynamic conditions, with dynamic roll rates reaching 280 degrees/s [58]. The results showed cadavers experienced more head excursion than human volunteers under static conditions [58]. Due to the potential for injurious loading in Moffat’s experimental environment, human subjects were not subjected to the dynamic roll conditions in this experiment.  15  Although PMHSs permit testing under injurious conditions, they’re subject to numerous limitations. PMHS testing is generally time-intensive, time-sensitive, expensive, and specimen availability is limited [14]. Biomechanically, segmented neck models lack interaction with the rest of the spine and head. A major limitation of all cadaveric studies is that PMHS lack physiologic functions. Following rigor mortis, cadavers are essentially atonal and unable to simulate the resting muscle tone or contraction dynamics of living humans. Despite methods to simulate muscle tone, reflexive muscular responses prior to, and during, impact cannot be replicated. Reflexive tensing may alter the load path between an occupant and vehicle interior, influencing the dynamic loading of an occupant during impact [14,59]. Even when musculature is simulated, it is often not based on in vivo data relevant to a rollover scenario. Some cadaver studies are conducted in an upright orientation [9–11], although recent studies have shown that inversion alone has a significant effect on in vivo cervical spine posture. Consequently, the results of testing in an upright and neutral posture are somewhat limited when applied to inverted headfirst cervical spine injury, such as in a vehicle rollover. Lack of active musculature and representative cervical spine posture may be partly responsible for the disparity between lab-induced and clinically reported injury distributions [2].  1.3.2 Computer models Computer models are useful tools for evaluating the effects of parameters that may be important to neck injury during an impact. Computer models are well suited to studying problems which are too complicated to tackle with conventional experimental approaches [14]. Unlike cadaveric studies, computer models allow simultaneous simulation of musculature and visualization of the cervical spine during impact injury. However, the validity of the results 16  depends on the quality of the computer model and its input parameters, and in many cases in vivo data is not available. Due to a paucity of data, cervical spine axial impact models are often not validated against relevant in vivo responses. Instead, many models have been validated with in vivo data from human volunteers in frontal impacts, ex vivo cadaver drop tests, and/or cadaveric specimen experiments [6,7,12,51,60]. Furthermore, axial neck injury models are typically based on a 50th percentile male or a single cadaveric specimen [6,7,12,51,51]. Thus, they are limited in describing the general population.  Computer models are able to simulate the effects of passive and active musculature, and thus can potentially overcome a major limitation associated with cadaver studies. Previously there’s been uncertainty in which muscles to include and which muscle activation scheme to employ. Several different muscle activation schemes have been adopted in cervical spine axial impact injury. Commonly, a relaxed activation scheme optimized to reduce muscle fatigue in an upright orientation is simulated [7,51]. This upright relaxed state represents an upright human holding their head upright against gravity with no pre-impact contraction. Chancey et al. simulated muscles in a sub-maximal tensed activation scheme, in which muscle forces were maximized while maintaining head position [51]. The tensed activation scheme represents a human holding their head upright with some level of pre-impact awareness. Other models simulate neck musculature at 100% of their maximum voluntary contraction (MVC) [6,12,13,60]. When simulating muscles at 100% MVC, neck extensor muscles overpower flexor muscles, forcing the head into extension [61]. In contrast, in vivo data from human subjects show the head moving into flexion when adopting a voluntary, quasi-static, bracing posture [17]. To remedy this difference, computational models may apply head restraints (ex. “to keep the head in an upright position a head restraint was modeled when activating all the neck muscles to 100%” 17  [12]) although subsequent effects on the simulation are unclear. Among the muscle activation schemes considered, the relaxed activation scheme by Chancey et al. is closest to the in vivo response of human subjects voluntarily bracing for impact (sub-maximal activity and varied between muscles [62]). However, these in vivo responses were measured under quasi-static and voluntary conditions and may not represent the reflex-mediated responses which develop in a dynamic scenario. The neck muscle activity of human subjects anticipating an inverted headfirst impact is still unknown.   Due to a paucity of data, axial impact neck injury models assume an upright and neutral cervical spine posture. This assumption is likely not representative of a rollover environment since cervical spine posture is altered due to the effects of inversion alone [15], and furthermore when adopting a voluntary bracing posture [62]. When inverted and instructed to maintain a forward gaze (representing an upside-down vehicle occupant maintaining their gaze on the road), human subjects show an average 14.1 mm anterior shift in spinal eccentricity than when compared to their upright-neutral posture [15]. When inverted and braced for impact, the C1 was shifted forward an average of 26 mm relative to the line of action of the torso’s inertia [17]. These changes in spinal eccentricities are larger than the small displacements and changes in eccentricity between the upper and lower cervical spine necessary to significantly influence injury, as discussed in section 1.2.5. Thus, the outcomes of existing computational models of axial impact injury to the cervical spine may not represent the intervertebral geometry and loading characteristics of an inverted occupant anticipating a headfirst impact.  18  1.3.3 Anthropomorphic test devices (ATDs) Anthropomorphic test devices (ATDs), also colloquially called crash test dummies, are mechanical analogs that aim to match the anthropometry, articulations, and structural response of a living human [14]. ATDs are instrumented with sensors to record the accelerations, forces, and displacements which would be experienced by a living human in a crash. Given their intensive instrumentation and associated high cost, ATDs are subject to high durability requirements. These requirements for durability and robustness may be contradictory to those necessary for biofidelity [14]. For example, the thoracic spine of the Hybrid III ATD is fused as a single steel component whereas the human spine is segmented into 12 thoracic vertebrae.  The Hybrid III ATD is commonly used in vehicle rollover research. Originally developed for use in frontal impact testing, the Hybrid III’s body stiffness is intended to be biofidelic under high acceleration events [63].  However, the airborne phase of a rollover, in which occupants may interact with the vehicle roof, involves relatively lower accelerations (approximately 3-4 g of centripetal acceleration [64]).  A study investigating the dynamic response of the Hybrid III neck to accelerations less than 15 g concluded that the Hybrid III neck is too stiff to respond in a human-like manner in the axial and fore-aft directions [65]. Similarly, the neck and shoulder complex of the Hybrid III has been reported as too stiff to allow for a biofidelic kinematic response during the roll-onset of a vehicle rollover [66]. In studies comparing the response of Hybrid III ATDs, cadavers, and human volunteers in vehicle rollover simulations, Hybrid III ATDs exhibited less head excursion than cadavers or human volunteers, particularly in the vertical and lateral directions [58,63]. Thus, using Hybrid III ATDs to study neck injury risk in rollovers may be misleading since the ATD head and neck may in a different location and orientation than a human at the time of an injurious impact [67].  19   1.3.4 Human volunteers The human response may be studied through clinical data from people involved in actual crashes (useful for injury trends) or laboratory tests [14]. Video footage from dashboard and highway cameras also offer insight on the human response in actual crashes. As clinical data is frequently gathered from medical records and police reports, a full mechanical description of the injury-producing events is often lacking. Some databases (e.g., CIREN) conduct field studies and offer a professional analysis of medical and engineering evidence to determine injury causation, however the number of investigations performed are limited. These investigations offer data-rich insight into specific rollover cases; however, the small sample size precludes them from representing the full spectrum of vehicle rollovers which occur. Laboratory tests are helpful at understanding human response, but testing is limited to sub-injurious exposure levels. These laboratory tests are also resource intensive, requiring the approval of a review board to assure human welfare, scientific rationale, and all legal and ethical standards are upheld [14]. Due to these restrictions, only a small number of human volunteer tests relevant to cervical spine injury in vehicle rollovers have been conducted.  Rollover laboratory tests involving human subjects have typically been limited to external factors such as gross occupant motion. Human subjects have been used to study head excursion, a measure of head-to-roof clearance in an inverted posture, under static conditions [58,63,68–72]. These excursion studies have primarily been concerned with the effectiveness of various seat-belt configurations rather than the injury mechanism of axial impact injury. Some dynamic studies, investigating head excursion or occupant response, have been done under dynamic conditions [63,72,73], with constant roll rates ranging between 100 – 230 deg/s. Other 20  dynamic studies have investigated occupant kinematics and muscle activity in pre-rollover scenarios [74,75], and the first phase of a rollover in which the vehicle’s wheels lose contact with the road surface [66]. Gross occupant motion has been tracked using either passive or active markers, while muscle activity has been recorded using surface electromyography (EMG) on superficial neck muscles (e.g., sternocleidomastoid and trapezius) [66,74,75]. While valuable in investigating global body motion and the response of superficial muscles to vehicle dynamics, none of these studies have paid special attention to cervical spine alignment, activity of deeper cervical muscles, or their effect on axial impact injury to the cervical spine. Our lab has previously conducted in vivo human subject studies which quantify intervertebral kinematics and neck muscle activity in response to tasks relevant to headfirst cervical spine injury. Newell et al. investigated how the cervical spine re-aligns (via fluoroscopy) and neck muscles respond (via electromyography) under various conditions, particularly in upside-down configurations [15,16,62]. In one study exploring the role of gravity on cervical alignment, seated subjects were statically inverted as their cervical spine posture and muscle activity were recorded in both a relaxed position, which results in a head that is extended relative to the inverted torso, and with the head flexed to achieve a level, albeit upside down, gaze. Both tasks revealed a cervical spine posture and muscle activation scheme that differed from an upright-relaxed posture, although the observed inter-subject variation was large [15,16].  In another study, Newell et al. investigated the effect of actively tensing neck muscles on cervical spine realignment [17]. Subjects were instructed to mimic a protective response in preparation for a headfirst impact. The consequent cervical spine posture and muscle activation scheme differed from the aforementioned inverted-forward condition, and furthermore from the upright-relaxed condition [62]. Although the cervical spine posture and muscle response in each study 21  differed (and showed large inter-subject variation), the muscle activation scheme remained sub-maximal and muscle-dependent. While these studies have been invaluable in understanding headfirst cervical spine injury, they were completed under static conditions. Since rollovers are dynamic events, it is not clear whether humans would react to a headfirst axial impact by adopting any of the muscle activation patterns or the head/neck postures studied [16].   1.3.4.1 In vivo neck kinematics The gold standard for measuring and tracking motion of in vivo spine biomechanics is radiographic image analysis (RIA), an imaging technique using x-rays.  In an effort to circumvent the radiation exposure associated with RIA, some researchers have investigated the comparative accuracy of external motion tracking [76–79]. However, external motion tracking of the cervical vertebra is generally not sufficient due to the complexity of the spinal column and overlying soft tissues [77]. In the cervical spine, skin movement with respect to bone has been reported to be as much as 3 mm for slow dynamic movements (not representative of those present during a rollover), leading to discrepancies between vertebral angles (of 4.8 degrees) measured using external skin markers versus radiographic landmarks [78]. Thus, when studying intervertebral kinematics of the cervical spine, RIA is necessary to accurately determine vertebral displacements and angles.  Within RIA, X-ray fluoroscopy provides moving projection radiographs and is used as a method for studying complex relative motion of bony structures. Portable fluoroscopy machines, called fluoroscopic C-arms, contain both an x-ray generator and an image intensifier. Via an x-ray tube, the x-ray generator produces an emergent x-ray beam that is conically shaped. The x-ray generator and image intensifier are aligned such that the central x-ray of the cone-beam path 22  strikes the image intensifier at a right angle. The image intensifier converts the fluoroscopic x-rays, incident on its input face, into a visible light image on an output screen. An optical coupling between the image intensifier output screen and a camera allows images from the intensifier to be converted into a video signal [80].  The maximum speed of a fluoroscopy system is limited by the refresh rate of the scintillator, a material that fluoresces when struck by a charged particle or high-energy photon (such as those in the incident x-ray beam), within the image intensifier and the frame rate of the optically-coupled camera. The fluoroscopic C-arm in this thesis uses a sodium-doped cesium iodide scintillator, with a decay time of 5 𝜇s [80], making the camera frame rate (30 Hz) the limiting factor. Camera sensitivity and shutter speed further limit frame rate, since low sensitivity and high shutter speed produce dim, low quality images. To record images at higher frame rates, common fluoroscopic equipment may be enhanced with an improved image capture system [81].   1.3.4.2 Electromyography Electromyography (EMG) is a valuable tool for studying relative activations and recruitment patterns of muscles. EMG is a technique used to record the electrical potential produced by muscle cells when these cells are electrically or neurologically activated.  There are two categories of EMG: surface EMG and indwelling EMG. Surface EMG is a noninvasive approach in which a pair of surface electrodes are adhered to prepared skin above a muscle of interest. The potential difference between the two electrodes is measured and corresponds to measured activity of any muscles within the sensing volume. Surface EMG cannot reliably discriminate between the activity of adjacent muscles and is therefore generally restricted to 23  measuring superficial muscle activity (since potentials from deep muscles would be confounded by superficial activity, and vice versa). Additionally, surface EMG is influenced by the thickness of subcutaneous tissue which is dependent on a subject’s body composition. Indwelling, or intramuscular, EMG is an invasive approach in which fine wire electrodes are inserted into muscle via a hypodermic needle. In the case of this thesis, two fine wire electrodes are inserted and referenced to each other electrically. Both wires are electrically insulated, with only the tips exposed to measure the potential within the muscle of interest.  Direct measurement of muscles forces in live humans is not typical, due to the invasive and injurious nature, thus EMG is used as a proxy of muscle force. The relationship between the electrical signal that a  muscle emits and the force it applies is complex, and is confounded by various factors such as muscle orientation, cross-sectional area, length, and contraction velocity [16]. There is no direct relationship between muscle activity and the force generated. Nonetheless, EMG is useful for studying the relative activation levels and recruitment of muscles. EMG recordings are typically normalized to muscle activity recorded during a 100% maximum voluntary contraction (MVC) to assess the relative activity level between muscles, although achieving MVCs in some muscles, particularly individual neck muscles, can be difficult.   1.4 Thesis objectives and hypotheses There is a need to better understand axial impact cervical spine injury to re-create clinically relevant injuries in laboratory settings. The epidemiology of axial impact cervical spine injury reveals vehicle rollover crashes to be a major contributor, and thus this thesis is presented in the context of vehicle rollover crashes. Previous work in our lab has investigated the effects of 24  inversion and voluntary bracing tasks on cervical spine posture and neck muscle activity under static conditions. However, there is a gap in existing literature regarding the reflex-mediated neck muscle response and muscle-induced changes in cervical spine realignment in anticipation of an inverted headfirst impact. In an effort to fill this gap, this thesis focuses on extending our previous work to a dynamic scenario which better simulates an inverted impending headfirst impact. Identifying how the cervical spine realigns and neck muscles respond when anticipating a headfirst impact is important if we ultimately want to better understand and ultimately prevent axial impact neck injury. A better understanding of the relationship between muscle action and spinal alignment will aid in elucidating the mechanics of the cervical spine and spinal cord injuries. This work may also help improve and validate existing cervical spine injury models.  The specific objectives for this thesis were: 1) To develop an apparatus which simulates impending headfirst impact with sufficient time to observe a muscular and postural response; 2) To analyze the in vivo cervical spine kinematics in response to an impending headfirst impact; 3) To examine the in vivo neck muscle activity in response to an impending headfirst impact;  I hypothesize that during simulated headfirst impacts: 1. Muscle activity will increase in response to a simulated impending headfirst impact compared to upright-neutral and inverted-forward conditions; 25  2. Cervical spine posture will change dynamically to characterize a more protective bracing posture; The dynamically adopted posture will differ from a quasi-statically adopted posture after several hundred milliseconds;    26  Chapter 2: Commissioning of an inversion apparatus to simulate impending headfirst impact This chapter addresses the preliminary stages of the first goal:  to develop an apparatus that simulates impending headfirst impact. In this chapter, I outline the design goals set for the apparatus at the project onset, describe the final apparatus design, and detail the commissioning process.   2.1  Design Goals The overall goal for designing an inversion apparatus was to create a means for studying the in vivo muscle responses and cervical spine posture that develop in anticipation of an inverted headfirst impact. While seated in an inverted orientation, the apparatus should displace a human subject a short distance to simulate a free fall drop before decelerating the subject to rest in a noninjurious manner. The apparatus should also accommodate the recording of indwelling EMG electrodes in the neck muscles and fluoroscopy imaging of the cervical spine. A more detailed sequence of events for the proposed human subject tests is as follows: 1. Subject securely seated in an upright orientation 2. Subject is inverted to an upside-down orientation 3. Subject undergoes a vertical free fall translation throughout which fluoroscopy and EMG data are recorded 4. Subject undergoes a noninjurious deceleration throughout which fluoroscopy and EMG data may be recorded 5. Subject reaches a full stop  27  6. Subject is returned to the upright position  7. Subject may be released from the apparatus  Based on this sequence of events, a list of need statements was generated to outline the goals and scope of the project. These need statements are as follows: 1. Restrains human subjects using a 5-point safety restraint harness 2. Allows human subject to be securely seated in an upright orientation 3. Allows human subject to be securely seated in an inverted orientation 4. Allows inverted vertical translation of a human subject equivalent to free fall 5. The free fall duration is at least 426 ms, subject to physical constraints 6. Deceleration rate is limited to 1g 7. Accommodates fluoroscopic C-arm imaging of the subject’s neck throughout the free fall phase 8. Accommodates on-board EMG collection 9. The human subject’s head and neck motions are visible to a motion and/or video capture system  10. Mitigate relative motion between the fluoroscopic C-arm and the human subject 11. Vertical height is limited to 12’ (ceiling height of the test facilities) 12. Restrains human subject’s feet and arms and while inverted and throughout the fall  These need statements were then submitted to Mircea Oala-Florescu for the subsequent mechanical design and fabrication of the inversion apparatus. No need statements were formed around peak velocity or impact velocity, an important metric in studies on impact injury to the 28  head and neck[8,26,40,43,45,53,82–85], since the emphasis was on the pre-impact response. An important consideration was free fall duration, due to the time-sensitivity of muscle responses and postural lag, thus special attention was given to need statement 5. Need statement 6 also received special consideration due to safety and comfort concerns associated with being deceleration of the volunteer subjects while inverted.   2.1.1 Rationale for minimum free fall duration Falling is a powerful vestibular stimulus for the postural control system. At the onset of free fall there is deformation of the body’s soft tissues and a sudden change to cutaneous pressure and forces acting in skeletal joints all of which could signal falling. Following an abrupt stimulus, muscle activity may show a short latency, generalized,  muscle activation classified as a “startle response”. In young populations the startle response diminishes within 70 ms [86], after which the remaining underlying muscle activity is classified as a postural response.  Sudden free fall can be a potent startling stimulus [87] and, in this project, it is important to consider how this relates to an automotive rollover. Startle is easily inhibited by small warning signs that precede a startling stimulus and vehicle rollovers are often preceded by strong visual and somatosensory cues. Additionally, startle responses have a relatively short duration while vehicle rollovers are relatively long duration events, with common roll rates of 200-300 deg/s [12] leaving up to 900 ms before vehicle inversion. Since the time for the startle response to diminish is considerably less than the time taken for a common vehicle inversion, the underlying postural response is likely more appropriate at a time relevant to headfirst impact during a rollover crash.  29  Neck muscle onset times and subsequent postural lag govern the minimum free fall duration. Previous studies have looked at neck muscle activity in response to abrupt vertical movements. In a study by Aoki et al., human subjects were seated upright and exposed to an abrupt vertical movement via a hydraulic servo system. Vertical ascent and descent produced different muscle onset latencies. In vertical ascent, the sternocleidomastoid and trapezius muscles showed latencies  of 24±1.7 ms and 24.8±3.9 ms, respectively, from the onset of acceleration. Vertical descent yielded latencies of 44.1±3.1 ms and 48.6±4.1 ms in sternocleidomastoid and trapezius, respectively [88]. Bisdorff et al. exposed human subjects to a free fall stimulus from the supine position via an electromagnetic-release tilting couch. Healthy subjects showed onset latencies of approximately 56±6 ms in the sternocleidomastoid and 57±7 ms in the trapezius muscles [87] from the onset of acceleration. Ito et al. placed human subjects in a supine position and exposed the head to free fall by abruptly removing the head-support. Onset latencies of the sternocleidomastoid were approximately 20-25 ms from the onset of acceleration[89]. Although none of these studies are representative of our perturbation, i.e., an abrupt free fall while inverted, they suggest the expected neck muscle latencies to be near 20 – 64 ms.  Muscle-induced changes in cervical spine posture lag behind muscle onset. In other words, muscle forces take time to develop and cause postural displacements because the inertial loads resistance of the stationary segments of the spine must first be overcome. Nijhuis et al. studied reaction times in human subjects who voluntarily rotated their head in response to an acoustic startle stimulus. The onset of head movement following the stimulus was 313±114 ms [90] measured from the time at which head velocity exceeded 15mm/s as determined by a position-based motion tracking system. Given that Bisdorff et al. showed neck muscle onset 30  times to be similar between acoustic and free fall startle stimuli [87], our postural lag is expected to be similar. Thus, the hypothesized minimum free fall duration for capturing muscle activity and muscle-induced changes in cervical spine posture is 427 ms (313 ms + 114 ms).  However, physical constraints of the test facility (ceiling height, ceiling beams, pipes, etc.) in combination with deceleration limits, described below, limit the free fall duration. Thus, the free fall duration should be maximized as much as practically possible.   2.1.2 Rationale for limiting deceleration to 1g Human tolerance to acceleration depends on several factors: magnitude, duration, direction, location of application, and body posture. In this experiment, human subjects will be decelerated from a non-zero velocity while upside-down. Consequently, blood flow is forced in a feet-to-head direction causing increased arterial and venous pressures cranial to the heart. Increased pressure in the aortic arch and carotid arteries results in pronounced bradycardia  [91]. Increased venous pressure may result in facial swelling, broken capillary vessels, sinus pain, and headache. Potential neurological effects include sensory disturbances, confusion, loss of consciousness [91], blurred vision, and red-colored visual fog [92]. Research on the tolerance to acceleration in this configuration has been limited due to volunteer subject discomfort and researchers’ fears of untoward side effects [91]. An estimate of reasonable tolerances to negative-g acceleration, based on fighter pilots and human centrifuge studies, are -4.5 g for 15 s, -3 g for 30 s [93], and -2 g for 5 min [92]. Even more conservative tolerances are given by investigations into eye injury: -4.5 g for 1 s, -3 g for 10 s, and -2 g for 100 s. General fatigue, sleep deprivation, anxiety, and mental stress are among factors which may decrease acceleration tolerance [92].  31  Recreational activities and amusement park rides abide by guidelines that limit the magnitude of acceleration. The Canadian Bungee: Code of Safe Practice limits the g-force imposed on a jumper where the force is taken on the torso to -4.5 g [94]. To keep the experience comfortable, some bungee cord manufacturers limit g-forces between -2.4 – -2.8 g [95,96]. Amusement park rides abide by industry standards which set limits based on magnitude, direction, and duration of acceleration exposure [97]. The ‘Hellevator’ at the Pacific National Exhibition in Vancouver, Canada, exposes ride-goers to -2.5 – -3 g. Similarly, the ‘Dropzone’ at the Santa Cruz Boardwalk in Santa Cruz, USA, exposes ride-goers to -3 g.  Based on the potential untoward effects, incomplete data on tolerance to acceleration in this orientation, and likelihood of experimenting on students (who may be prone to factors such as psychological stress) that decrease acceleration tolerance, we have chosen to conservatively limit our acceleration magnitude to -1 g. Another contributing reason is that the total inversion time is unknown. The subject will likely be inverted for some time prior to the onset of free fall. Pre-fall inversion time may include safety checks, likely being raised to a fixed drop height, and any operator delay in issuing a trigger command. The subject will also likely be inverted for some time after being decelerated, as the process by which the subject is returned to upright may involve un-doing numerous safety mechanisms. The additive effects of exposure to various acceleration magnitudes and durations in an upside-down orientation is unclear. Thus, deceleration is conservatively limited to -1 g.   2.2  Mechanical description of final design   The mechanical design and fabrication of the inversion apparatus was completed by Mircea Oala-Florescu of MEA Forensic Engineers & Scientists. The final design consisted of a 32  rail-and-carriage-like apparatus coupled to a pre-existing feedback-controlled linear motor (Figure 2-1).   Figure 2-1 Photo of the final inversion rig design  The steel rail-and-carriage-like apparatus is coupled to a pre-existing feedback-controlled linear motor. Horizontal displacement of the linear motor controls vertical displacement of the subject carriage.   Two steel posts were bolted to the concrete floor and attached to a steel ceiling beam via tubular extensions. A horizontal spreader bar separating the top ends of the two posts ensures the posts remain parallel. Perpendicular to the spreader bar, two angled struts bolted between the concrete floor and steel posts offer additional stability. A linear rail-and-carriage slide mounted to the vertical posts allow vertical translation of an attached internal subject carriage. The 33  internal subject carriage is mounted to the linear slide via spherical bearings which allow it to be rotated relative to the vertical posts. Travel of the linear slide is limited by positive stops at the top and bottom end of the rails. Removable pins can be fastened through the subject carriage and into the linear slide to constrain rotation to the 0° (upright) and 180° (inverted) configurations. Two links attached to the linear slides couple the subject carriage to cross-bracing fastened to the linear motor. Thus, horizontal motion of the linear motor results in vertical motion of the subject carriage. The linear motor is feedback controlled, allowing execution of pre-programmed motion profiles. In case of power loss to the linear motor (the subject carriage would move downwards under the influence of gravity), two shock absorbers bolted below the subject carriage are intended as a redundant means of decelerating the subject carriage at -1g. The subject is seated in an adjustable-height bucket seat (36 series – Intermediate 20-degree Layback, Kirkey Racing Fabrication INC., St. Andrew’s West, ON) and secured with a 75-mm-wide 5-point harness (RCI Racers Choice, INC., Tyler, TX) with their feet restrained using snowboard bindings. A fluoroscopic C-arm (OEC 9400, GE) is secured to the top-end of the subject carriage. The fluoroscopic C-arm is mounted on a rail-carriage system allowing fore-aft adjustment. Fore-aft adjustment of the C-arm and seat-height adjustment allow the C-arm to be centered on the subject’s cervical spine.  During vertical motions, inertial loading on the x-ray source and image detector cause mechanical flexing of the C-arm. This flexing results in misalignment between the x-ray source and image detector, leading to distorted images. To mitigate relative motion, additional steel bracing was used to secure the x-ray source and image intensifier to the subject carriage (Figure 2-2). The final mechanical design implicitly satisfied need statements 1 – 4, 7, 9 – 12. However, need statements 4 – 6, and 8 needed to be addressed through further work.  34             Figure 2-2 Steel bracing securing x-ray source and image intensifier assembly to the subject carriage The steel bracing is fastened to the x-ray source and follows the contour before fastening to the subject carriage (left). The Image Intensifier assembly is braced to the subject carriage just above the camera (right).   2.3 Design commissioning The design commissioning process addressed outstanding need statements 4 – 6, and 8.   2.3.1 Derivation of linear motor motion profile Deriving the linear motor’s motion profile throughout the vertical translation addresses need statements 4 – 6: 4. Allows inverted vertical translation of a human subject equivalent to free fall 5. The free fall duration is at least 426 ms, subject to physical constraints 35  6. Deceleration rate is limited to 1g The desired subject acceleration profile consists of free fall for 426 ms (or as close to 426 ms as practically possible), followed by  deceleration at -1 g. Vertical motion of the subject carriage is controlled by horizontal motion of the linear motor, which must be programmed. Since the subject carriage and linear motor move on perpendicular axes, and are coupled by two links, motion of the subject carriage depends on the instantaneous angle of the links (Figure 2-3). In other words, if the linear motor was travelling at a constant rate, the rate at which the subject carriage travelled would depend on the instantaneous angle of the coupling links. The grey lines represent a the coupling links at different times, while the red lines denote an instantaneous position. On the left, large motions of the motor result in small motions of the subject carriage. On the right, small motions of the motor result in large motions of the subject carriage.    S M M S Figure 2-3 Depiction of motion between the subject carriage (S) and the linear motor (M) 36  The relationship between motion of the subject carriage and motion of the linear motor is a classic related-rates problem. Given the ideal acceleration profile of the subject carriage, the corresponding acceleration profile of the motor can be calculated as follows: (1)  We let the subject carriage, linear motor, and intersection of their axes represent the vertices of a right-angled triangle. (2) We let 𝑎, 𝑏, and 𝑐 represent the following (Figure 2-4): - 𝑎 represents the instantaneous distance between the subject carriage and axes intersection, I - 𝑏 represents the instantaneous distance between the linear motor and the axes intersection, I -  𝑐 represents the length of the coupling link between the subject carriage and linear motor Representation of the problem set-up in calculating the linear motor’s (M) acceleration profile from the subject carriage’s (S) ideal motion profile. The subject carriage (S), linear motor (M), and the intersection of their movement axes (I) form the vertices of a right-angled triangle. The letters a and 𝒃 represent the instantaneous distance of the subject carriage from I, and c represents the length of the coupling link.   S a I b M c Figure 2-4 Diagram of the problem set-up 37  Then the following analysis applies: 𝑙𝑒𝑡 𝑎 = 𝑎(𝑡), 𝑏 = 𝑏(𝑡), 𝑎𝑛𝑑 𝑐 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 From Pythagoras: 𝑎2 + 𝑏2 = 𝑐2 Taking the first derivative with respect to time, and dividing by 2, gives: 𝑎𝑎′ + 𝑏𝑏′ = 0 Taking the second derivative with respect to time gives: 2𝑎′ + 𝑎𝑎′′ + 2𝑏′ + 𝑏𝑏′′ = 0 Solving for 𝑏′′, acceleration of the linear motor, gives: 𝑏′′ = −2𝑎′ + 𝑎𝑎′′ + 2𝑏′𝑏 Where 𝑎′ is determined using the backward difference method and 𝑎 is determined using information about the initial and final angles of the coupling links, and applying the kinematic formula (where t=time): 𝑎 = 𝑎′𝑡 + (12) 𝑎′′𝑡2 𝑏 is determined using instantaneous values for 𝑎, 𝑐, and Pythagoras theorem. The position, velocity, and acceleration curves of both the subject and linear motor are shown in Figure 2-5. These curves generate both a free fall and 1g deceleration phase over the available distance of 1m. Further details can be found in Appendix A  , where the Matlab script used to generate the curves is included. 38  The calculated acceleration, velocity, and position curves for both the subject carriage and linear motor. The maximum travel of the subject carriage is limited to 0.95 m. The free fall duration is 312.5 ms, and the total drop time is 632.8 ms.    To execute a pre-determined motion trajectory, the linear motor requires position-based data; the linear motor executes position-increments over a specified time interval. The control card generates motion steps at 256 Hz, thus, the linear motor’s position profile was fragmented into time intervals of 1/256. The Matlab code used to convert the position profile to position increments at 256 Hz is also included in Appendix A  .  2.3.1.1 Linear motor control program A program to control the linear motor was written in Dynamic Motion Control (DMC) code and run using the Galil DMC Smart-Terminal. Upon executing, the program performs numerous safety checks (discussed in section 2.3.5) prior to accepting any user input. Figure 2-5 Calculated motion profiles 39  Subsequently, the user can control motion of the linear motor using six pre-determined motion profiles which are listed below: 1. Cancel (CNL): the linear motor holds its current position while the user is given options of exiting the program or re-selecting a different motion profile.  2. Steady-state raise (SSR): the linear motor moves at a steady pre-programmed velocity to raise the subject carriage from the current position to the top-most (maximum) position.  3. Incremental raise (INCR): the linear motor moves at a steady pre-programmed velocity to raise the subject carriage from the current position by a user-defined distance.  4. Steady-state lower (SSL): the linear motor moves at a steady pre-programmed velocity to lower the subject carriage from the current position to the initial (minimum) position.   5. Incremental lower (INCL): the linear motor moves at a steady pre-programmed velocity to lower the subject carriage from the current position by a user-defined distance. 6. Drop (DROP): the linear motor executes the calculated motion profile (section 2.3.1) from the top-most (maximum) position to the initial (minimum) position.  The Cancel option may be commanded from any position; however, a prompt is displayed if it’s issued above the initial (minimum position). Motions two through five are commanded in a Position Relative mode, while motion six is commanded in a Contour Mode. Position Relative mode uses pre-defined kinematic parameters to command distance-based movements whereas Contour Mode uses arrays of position-based data to execute motion trajectories.  Motions two and three can be issued from any position below the top-most (maximum) position. Likewise, motions four and five can be issued from any position above the initial (minimum) position. Motion six can only be executed from the top-most (maximum) 40  position since the calculated trajectory requires the full travel of the subject carriage. Regardless of mode, all motions are executed within a 2 mm position error via feedback control. If the current position deviates from the intended position by more than 2mm, the program jumps to a position error handling routine while the linear motor holds the current position. More detail regarding the control program can be found in the program flowchart and DMC code included in Appendix B  .   2.3.2 Retro-fit of an existing fluoroscopic C-arm Retro-fitting the pre-existing fluoroscopic C-arm is related to need statement 7: accommodates fluoroscopic imaging throughout the free fall phase. The pre-existing fluoroscopic C-arm was intended for static applications. As such, the image collection rate was limited to 30 Hz by a Vidicon camera. While suitable for a static application, the low collection rate would result in motion blur if applied to dynamic imaging tasks. In our case, such blur would preclude accurate analysis of vertebral body motion. Increasing the frame rate required removing the existing Vidicon camera, and replacing it with a high-speed video camera. The Vidicon camera was optically coupled to the image intensifier via two lenses and a right-angle mirror. The two lenses collimate and focus the image, while the right-angle mirror reflects the image at 90° to the Vidicon camera [80] (Figure 2-6). 41   Figure 2-6 Right angle optics The optical coupling between the image intensifier (left) and Vidicon camera (bottom). The first lens (left) converts the image to a collimated (infinity focus) image, and the second lens (bottom) is focused to form the image on the input plane of the camera. The angled mirror reflects the collimated image by 90° to the camera. Reproduced from OEC 9400 C-arm Service Manual. OEC-Diasonics; 1992.  The Vidicon camera was removed, and a high-speed camera (Phantom V12.1M, Vision Research Inc., Wayne, NJ) was mounted in its place. The mirror and first lens were left untouched, while the second lens was replaced with a lens compatible with the high-speed camera (NIKKOR 50mm f/1.4, Nikon, Melville, NY). This lens was manually focused to form the image on the input plane of the camera.  The fluoroscopic C-arm was equipped with two fluoroscopic modes: Fluoro Auto and Fluoro Manual. Fluoro Auto automatically controls the x-ray tube’s current and voltage, and the camera gain to obtain the proper video exposure. The relationship between the parameters is 42  established by software algorithms. Fluoro Manual requires the operator to manually set the x-ray tube’s current and voltage and offers no video gain adjustments. Removing the Vidicon camera precluded the use of Fluoro Auto, since the software algorithms relied on feedback from the camera to adjust the parameters. Consequently, installing the high-speed camera forced the operator to use the system in manual mode.  The complete absence of the Vidicon camera results in an error code precluding operation of the C-arm. To bypass this error, the Vidicon camera remains electrically connected to the system via a 25 pin D-Sub cable (although it provides no function). This configuration allows the C-arm to operate with the high-speed camera installed.  Recording images at an increased frame rate results in darker images. The increased frame rate means the camera shutter is open for a shorter amount of time and the image sensor is exposed to less light. This imposes limits on the frame rate used by the high-speed camera. Through trial and error, a frame rate of 200 Hz at a resolution of 1280x800 was adequate to distinguish vertebral landmarks (Figure 2-7). 43   Figure 2-7 Fluoroscopic image of two human vertebra captured at 200 Hz and 1280x800 resolution Two dissected human vertebrae separated by a Styrofoam plate are statically held within the fluoroscope’s field of view. Key vertebral landmarks such as vertebral corners are discernable.   2.3.3 Image analysis method Development of an image analysis method is related to need statement 10: no relative motion exists between the fluoroscopic C-arm and the human subject. As a consequence of imaging in a dynamic environment, inertial loading on the x-ray source and image detector causes mechanical flexing of the C-arm. This flexing results in misalignment between the x-ray source and image detector, leading to distorted images (Figure 2-8). Although additional steel bracing securing the x-ray source and image intensifier to the subject carriage mitigates relative motion (Figure 2-2), it is not eliminated. Some movement and vibration are still evident when analyzing the fluoroscopic images. This may be due to motion or vibration of the C-arm’s 44  internal components (ex. phosphor screen, right-angle mirror) . To remedy this issue, an image analysis method was developed that is independent of relative motion or vibration of components. The image analysis method uses an array of 1mm steel beads (Figure 2-9) and photo rectification software to rectify the images. The array of steel beads is mounted within the fluoroscope’s field of view (Figure 2-10), such that the beads are visible for the full duration of any recording. Prior to rectifying the images, additional distortions must be removed.  In this example a light source is analogous to the x-ray source while a piece of paper is analogous to the image detector. If the source and detector are properly aligned (left) so that the central beam is perpendicular to the detector, then projected images are undistorted. If the source and detector are misaligned (right), then dimensions of projected images are stretched and distorted.  Figure 2-8 Image distortion due to misalignment between a source and detector 45   Figure 2-9 A scaled scan of the steel bead array used to rectify images 1mm steel beads are spaced 0.5” apart and embedded within an acrylic sheet mounted within the fluoroscope’s field of view. Known spacing between the beads allows the photo rectification software to rectify image distortion.   Figure 2-10 The bead array plate is mounted in front of the x-ray source and is visible in the fluoroscope’s field of view. Aluminum extrusions fastened to the subject carriage are used to hold the bead array plate in front of the x-ray source (left). The bead array plate remains in the fluoroscope’s field of view, such that the bead array is included in any fluoroscopic image (right). The fluoroscopic image in this figure has been undistorted and rectified.  46  Fluoroscopes produce images with severe nonlinear pincushion and S-shaped distortion [98]. This spherical distortion affects spatial relationships between the beads in the bead array and must be removed before the images can be rectified. Spherical distortion can be removed using XMALab, which requires an image of a standardized sheet of perforated metal to correct the fluoroscopic images. A distortion correction algorithm compares the spacing between holes in the fluoroscope image with the idealized spacing and calculates a transformation matrix for correcting the images [81] (Figure 2-11). The transformation matrix is then applied to all images of interest and may be exported as a set of undistorted images.  Figure 2-11 Spherically distorted (left) and undistorted (right) images collected from the fluoroscopic C-arm  The undistorted images may then be imported into photo rectification software (PC-Rect, MEA Forensic, Richmond, BC). This software converts oblique images into scaled plan views. Since the misalignments and vibrations are dynamic, each image requires a different rectification transformation. Although the actual spacing between the steel beads remains constant, the apparent spacing appears different depending on the degree of misalignment. Once imported into 47  the photo rectification software a square formed by four of the steel beads is identified. Then the known spacing between the beads is enforced to rectify the image (Figure 2-12). The image may then be exported for further analysis.   Figure 2-12 Visual summary of the photo rectification process applied to an image of the bead array plate An undistorted image (left) is imported into photo-rectification software and four steel beads are selected to form a square (middle). Known spacing between the steel beads are enforced by the software to rectify the image (right).   Further analysis may include measuring distances between points in the images (such as the distances between vertebral landmarks). Special care must be taken if making comparisons between different images. Although the images have been rectified, misalignment may shift the location of points on the screen. In other words, the on-screen x-y location of a point fixed in space may differ between frames due to changes in projection angle. Thus, there is no common reference point from which to make measurements. One solution is to superimpose frames such that the beads from the bead array coincide. Distances may then be measured relative to a common bead.   48  2.3.4 EMG accommodation A small shelf mounted on the rear of the seat addresses need statement 8: accommodates on-board EMG collection. The small T-shaped shelf is dimensioned to accommodate a set of pre-amplifiers (NL844, Digitimer Ltd., Hertfordshire, UK). The pre-amplifiers are fastened to the T-shaped shelf using Velcro (Figure 2-13). Velcro is placed at the interface between the T-shaped shelf and first amplifier, and between the first and second amplifier. A Velcro strap is then threaded through the T-shaped shelf and fastened around the pre-amplifiers for additional fixation.   Figure 2-13 T-shaped mounting shelf for EMG pre-amplifiers The T-shaped mounting shelf accommodates the pre-amplifiers. The pre-amplifiers have mating Velcro adhered to their surfaces. Velcro allows the position of the pre-amplifiers to be easily adjusted and accommodate for various subject torso lengths. A Velcro strap may be threaded through slots on either side of the ‘T’ and around the pre-amplifiers for additional security.  49   The pre-amplifiers are fastened inferior to the insertion site of the indwelling EMG electrodes. The EMG electrodes run between the muscle bellies, where they are inserted, and the pre-amplifiers. The pre-amplifiers are connected to an off-board amplifier which is then connected to a DAQ.   Figure 2-14 The pre-amplifiers mounted relative to an example subject The pre-amplifiers are mounted inferior to the insertion sites (neck area). The Velcro connection allows them to be mounted at various angles to the subject, as well as each other.  50    2.3.5 Risk mitigation A Hazard Analysis was conducted to mitigate risk associated with the inversion apparatus. Risk severity and semi-quantitative probability levels were determined and incorporated into a Risk Index Table. A Risk Action key was then used to specify an appropriate action based on the risk index. The full Hazard Analysis along with the severity and semi-quantitative probability levels, Risk Index Table, and Risk Action key are included in Appendix C  . In the following sections, key outcomes of the Hazard Analysis are discussed in terms of hardware or software design features or actions.   2.3.5.1 Hardware Most hardware concerns were associated with the failure of fasteners or mechanical components. In mitigating mechanical failure, it was ensured all fasteners and critical components had a minimum safety factor of 6 [99] under the most adverse expected loading conditions. To avoid failure associated with loose connections, a pre-test checklist was developed in which operators check select fasteners prior to beginning an experiment. The pre-test inspection checklist can be found in Appendix D  as part of the experimental protocol. The pre-test checklist also includes two-person verification checks on critical components. The subject’s safety harness connection is verified by two separate members of the experimental team before inversion takes place. Similarly, the anti-rotation pins between the subject carriage and vertical supports are checked by two parties before the apparatus may be raised. In raising 51  the apparatus, two positive stops are implemented at the top end of the rails to prevent the carriage from being raised beyond available travel.   2.3.5.2 Software The linear motor control program (Appendix B  ) implemented several safety features. Character strings of four letters were used to input commands while numbers were used to enter distances. While numbers may have been used for both, the separation of inputs reduces the opportunity for unintended commands. The program employs two automatic subroutines: a position error subroutine and a limit switch subroutine. The position error subroutine monitors the error between actual and commanded motor positions. If this error exceeds 2mm, the program enters the position error subroutine, in which the motor holds its current position and prompts the user for input. Physical obstacles causing interference to the motor, such as debris in the track, or over-current to the motor would result in calling the position error subroutine.  The limit switch routine is called if a limit switch is activated. Four limit switches are installed on the inversion apparatus: two at the top-end of the subject carriage (Figure 2-15), and two along the linear motor track. The limit switches form boundaries at positions of maximum and minimum intended travel of the subject carriage and linear motor. The limit switches are activated through physical contact with either the subject carriage or the linear motor. Once a limit switch is activated, the limit switch subroutine is called and the linear motor holds its current position and the user is prompted for input. The limit switches mitigate opportunity for the subject carriage and linear motor to travel outside of their intended range.  52   Figure 2-15 Upper limit switch mounted to positive stop The upper limit switch is mounted to the positive stop using a piece of an aluminum L-bracket. The limit switch’s metal lever contacts the carriage-slide and the program instructs the linear motor to crease motion.  A guarded ‘kill switch’ was installed and could be activated in an emergency. The ‘kill switch’ disables communication between the servos and the linear motor. Pressing the ‘kill switch’ has the same effect as cutting power to the motor. If the subject carriage is suspended by the motor, pressing the ‘kill switch’ will cause it to drop under the influence of gravity that is mediated by the dynamics of the connected system. If a subject was seated in the subject carriage, they would experience the passive free fall of the subject carriage into the shock absorbers. The rate at which the subject carriage falls would be determined by gravity and the resistance of mating components (ex. friction between the carriage-and-rail slide and friction between the linear motor and its track). Note that the shock absorbers were specified to 53  decelerate both the subject and subject carriage (and linear motor since via the coupling links) at 1 g. Thus, activating the ‘kill switch’ (or power loss) results in a drop and deceleration which would be no worse than the intended experiment. Once the ‘kill switch’ is pressed, the operator cannot issue any commands to the motor. The motor has effectively been ‘turned off’. Additional instrumentation was added to allow conditional safety checks to occur in the control program. Two interrupt switches were installed, which act as emergency stop switches. The emergency stop switches differ from the ‘kill switch’. Unlike the ‘kill switch’, the operator can continue to issue commands to the motor after pressing the emergency stop switches. If the emergency stop switches are pressed the linear motor will decelerate abruptly and hold its current position, suspending the subject carriage. The operator is then prompted for input on how to proceed. An inversion switch was mounted to the pivot pin of the subject carriage. Upon executing the program, the inversion switch is queried to ensure the experiment is not performed right-side-up. This check is important because the subject carriage is mounted asymmetrically: the distance between the pivot pin and top-most edge is longer when right-side-up than when inverted. If the experiment were performed right-side-up, the subject carriage would interfere with the ceiling at the maximum position. If the subject carriage were to drop while right-side-up, due to this asymmetry, the subject carriage would not adequately compress the shock absorbers. If the shock absorbers were not compressed, the linear motor would bear the full deceleration load. Relying solely on the motor is dangerous because if power to the motor was lost, and a subject was seated in the subject carriage, the subject carriage would collide with positive stops mounted at the end of travel. In the worst case, this scenario is equivalent to an un-decelerated impact into the positive stops from a height of 1 m. Such an impact would likely cause harm to both the subject and the inversion apparatus. The inversion switches function to 54  prevent this scenario. If the subject carriage is not inverted, the control program informs the user and does not execute.  Prior to inversion, the shock absorbers are retracted and secured using small L-shaped brackets. Four optical sensors were implemented to detect the presence of these L-brackets (Figure 2-16). After inversion, these L-brackets are removed to allow the shock absorbers to extend and compress. If the L-brackets had not been removed, and the subject carriage was raised, a power loss would result in an impact between the subject carriage and positive stops. If an L-bracket is not removed, the control program displays an error message until the L-bracket is removed. Additional details regarding the control program’s behavior after a switch or sensor has been activated can be found in Appendix B  .    Figure 2-16 L-brackets and optical sensors The L-brackets ensure the rod of the shock absorbers remain retracted (left). The optical sensors detect the removal of the L-brackets (right).    55  Chapter 3: Verification of an inversion apparatus to simulate impending headfirst impact 3.1 Introduction Vehicle rollovers are chaotic and dynamics events which are responsible for almost a third of all crash-related fatalities [100]. Phases of a vehicle rollover can be divided into the pre-trip, trip, airborne, and impact phases. During the air-borne phase, the vehicle is rotating through the air and free-falling towards the earth. In some vehicle rollovers, at the end of the airborne phase, the vehicle roof impacts the ground and occupants dive headfirst into the vehicle roof. This diving motion and subsequent loading of the neck by the incoming momentum of the torso [30,63,70] may lead to serious or fatal cervical spine injuries.  Efforts to reduce cervical spine injuries in vehicle rollovers have included investigating occupant kinematics and dynamics during simulated vehicle rollovers. Human surrogates, such as anthropomorphic test devices (ATDs) and post mortem human surrogates (PMHS), are commonly used in these dynamic tests. However, few studies have been performed with human subjects [63,75,101,102]. Most of the studies performed on human volunteers are external and quantify gross occupant motion [63,75,101,102]. Few human volunteer studies are available which examine parameters important to cervical spine injury, such as neck muscle activity and cervical spine posture [15,17]. The in vivo neck muscle response and cervical spine posture prior to impact is still unknown.  Chapter 2 of this thesis outlined the design and commissioning of an inversion apparatus that is intended to simulate the free fall portion of a rollover as an occupant dives toward the roof prior to impact. The apparatus is a means for studying the in vivo human muscle responses and 56  cervical spine postures that develop in anticipation of an inverted headfirst impact. The objective of this chapter is to verify key performance parameters of the apparatus. Free fall duration is quantified to determine if adequate time is available to observe neck muscle responses and postural-induced changes (as described in 2.1.1). The acceleration of the subject carriage and a surrogate are measured to confirm they remain within safe limits (as described in 2.1.2). ASTM F2291 – 17 Standard Practice for Design of Amusement Rides and Devices prescribes safe acceleration limits for amusement park rides. Compliance with an industry standard constitutes an acceptable level of risk, and therefore this standard is used to assess the safety of the measured accelerations.  3.2 Verification methods 3.2.1 Apparatus A custom inversion apparatus (Figure 3-1) was designed and built to expose a human ATD surrogate (and eventually a volunteer human subject) to an inverted free fall drop that simulates the free fall portion of a rollover crash. For the purposes of this chapter, the human subject will be replaced by a crash test dummy surrogate. The surrogate is seated in a bucket seat (36 series – Intermediate 20-degree Layback, Kirkey Racing Fabrication, INC., St. Andrew’s West, ON) and secured with a 75-mm wide 5-point harness (RCI Racers Choice INC., Tyler, TX) with the feet restrained in snowboard bindings. The surrogate is then inverted and elevated to a fixed drop height by the feedback-controlled linear motor. The motor is programmed to execute a motion profile that exposes the surrogate to a controlled 312 ms inverted free fall drop, before decelerating to rest. An onboard fluoroscopic C-arm (OEC 9400, GE) retrofitted with a high-speed camera (Phantom V12.1M, Vision Research Inc., Wayne, NJ) captures sagittal plane 57  images of the human surrogate’s neck (Figure 3-2). Further detail on the apparatus can be found in Chapter 2, where the design and commissioning process is described in detail.   Figure 3-1 The custom-built inversion apparatus raised to the fixed drop height with an ATD subject The custom-built inversion apparatus with the seated human surrogate inverted and raised to the fixed drop height. The human surrogate is seated in the subject carriage, which can be rotated to an inverted position (shown) and raised or lowered by commanding the coupled linear motor. 58   Figure 3-2 Fluoroscopic C-arm centered on ATD's neck The onboard fluoroscopic C-arm is used to capture sagittal plane images of the surrogate’s neck area. The C-arm can be centered on the surrogate’s neck in the fore-aft direction by shifting along a rail system. The seat height can be adjusted to raise or lower the surrogate into the fluoroscope’s field of view.  3.2.2 Instrumentation A BioRID II anthropomorphic test device (ATD) was used as a human surrogate. To fit within the bucket seat, the BioRID II’s arms were removed at the shoulder joint. Less the mass of both arms, the BioRID II weight was 69.26±0.36 kg. Two accelerometers were used to record the accelerations experienced by the ATD and subject carriage. A 7.5 g, 3-axis accelerometer (34103A, Summit Instruments, Akron, OH) was mounted at the center of gravity (CoG) of the ATD head, such that the original seismic center was preserved. Another 3-axis accelerometer 59  mounted on an ¼” aluminum place was adhered (Scotch Extreme Mounting Tape, 3M, London, ON) to the subject carriage to record the accelerations of the frame near the seat mounting point. The accelerometer was mounted along the vertical center-line of the subject carriage, below the bucket seat, on the outside edge of the subject carriage frame. This mounting location is different from the location specified in ASTM F2137 – 16. The standard specifies the accelerometer should be placed in a plane reasonably approximating the location of the posterior aspect of a patron’s upper torso (31-41 cm above seat level and 6-13 cm fore of the upper torso contact surface). However, since the motion profile does not include any rotational movements (in which centripetal and centrifugal accelerations may be experienced), and the seat is rigidly coupled to the subject carriage (ideally), the accelerometer should measure the same accelerations on both the seat and subject carriage. In reality, the seat and subject carriage are not rigidly coupled, and the joint between the two dampens the perturbations from the subject carriage to the seat. Thus, the accelerations measured on the subject carriage are larger and therefore more conservative than those which would be measured from the seat. Both accelerometers were sampled at 4000 Hz.   An incremental linear encoder, which provides feedback to the liner motor’s controller, was used to record the actual position of the linear motor.  3.2.3 Experimental verification testing The human surrogate (BioRID II ATD) was inverted and subjected to the controlled free fall drop and deceleration profile for 10 trials (calculated profile: Figure 2-5; programmed profile: Figure 3-5).  60   After raising the subject carriage to the fixed drop height, the operator issued the drop command. The drop command initiated an outgoing trigger pulse used to synchronize data collection between the accelerometers and encoder data. A 0.5 s delay followed the trigger pulse prior to the linear motor executing any position commands. After the 0.5 s delay, the linear motor executed the programmed motion profile. Data was recorded for the 0.5 s delay until 1-2 seconds after the subject carriage came to rest.   3.2.4 Analysis All data analysis was done using Matlab (R2017a, MathWorks Inc., Natick, MA).  3.2.4.1 Linear encoder data The actual position profile was collected from the motion control card at 256 Hz using a storage scope (Servo Design Kit, Galil Motion Control, Rocklin, CA) compatible with the motor control program software (Galil DMC Smart-Terminal, Galil Motion Control, Rocklin, CA). The actual position array was recorded for 6 out of 10 drop trials. This data was converted from pulses-per-mm to meters using a known conversion factor of 500 pulses-per-mm.  3.2.4.2 Accelerometer data Accelerometer data was acquired in accordance with ASTM F2137 – 17: Standard Practice for Measuring the Dynamic Characteristics of Amusement Rides and Devices [103] (triaxial accelerations measured). However, the data channel range was ±7.5 g rather than the prescribed ±10 g. This was deemed acceptable since the expected accelerations were ~1.5 g and exploratory tests showed no data-clipping when using the 7.5 g accelerometers. Note that ASTM 61  F2291 – 17 and ASTM F2137 – 16 only apply to the accelerometer mounted to the inversion apparatus. However, these standards were also applied to the surrogate’s accelerometer for consistency. Accelerometer voltages were then converted to g’s using each sensor’s 3x3 calibration matrix. The offset was removed by averaging the first 250 ms (1000 samples) of the signal and subtracting it from the data. Data was then filtered with a 4-pole, single pass, Butterworth low pass filter using a corner frequency of 5 Hz as per ASTM F2291 – 17 [104]. The data set axes were transformed to a coordinate system consistent with ASTM F2291 – 17 (Figure 3-3). Maximum and minimum values were found in each direction for the accelerometer adhered to the subject carriage. The analysis was completed for 9 out of 10 trials. Trial 1 was omitted due to a corrupt data file (contained no data upon opening). The standard deviation was calculated assuming a randomly distributed sample. Lastly, the filtered acceleration signals (with the offset due to gravity re-instated) along the X-Y, Y-Z, and X-Z axes were compared with the allowable base limits outlined in ASTM F2291 – 17.  Figure 3-3 ASTM F2291 – 17 coordinate systems Front Side 62  The coordinate systems used for the surrogate (yellow) and subject carriage (red) are as shown. Note that in this figure the surrogate and subject carriage are pictured in an inverted orientation. When re-righted, the coordinate systems are fixed to both the surrogate and the subject carriage. The z-axis of the surrogate is aligned along the longitudinal axis of the head, passing through the head’s CoG. The coordinate axes of the two accelerometers are not aligned, and due to relative motion between the ATD and subject carriage, the spatial relationship between the accelerometers changes dynamically.   3.2.4.3 Free fall duration The free fall duration was determined using the unfiltered z-axis accelerometer data from the sensor adhered to the subject carriage. Free fall duration was calculated as the time between the onset of free fall and the end of the free fall phase. The onset of free fall was defined as the sample where the rms average of the signal for a 25 ms window centered around the sample was two standard deviations above the average signal before the trigger fired in each trial (250 ms immediately preceding the trigger).  At the end of the free fall phase the linear motor begins to decelerate and the subject carriage contacts two shock absorbers. This deceleration and contact results in a relatively large sharp spike in the acceleration data. Thus, the end of free fall was defined as the sample for which successive differences in filtered accelerometer data were greater than 1 g. An example of the free fall duration is shown in Figure 3-4. Figure 3-4 An example of the free fall duration measurement 63    Free fall duration was defined as the time between the onset of free fall and the end of the free fall phase. The blue line is the unfiltered accelerometer data for the sensor adhered to the subject carriage (from Trial 2). The red line is the rms average calculated using a moving 25 ms window, included because it was a step in determining the free fall onset. The first black asterisk marks the onset of free fall, and the second black asterisk marks the end of the free fall phase immediately prior to deceleration.   3.3 Verification results 3.3.1 Kinematics Motion of the linear motor (position and acceleration) was consistent across all trials (Figure 3-5). . The linear motor stayed within the programmed 2 mm tolerance of the commanded motion profile in all 6 trials for which it was recorded (confirmed through the absence of a position error). Data from the accelerometer adhered to the subject carriage showed minimal acceleration in the x- and y-direction, but significant activity in the z-direction (Figure 3-5). The subject carriage’s acceleration in the z-direction hovered around -1 g and 1 g with two clear local minima and maxima. The accelerometer mounted at the BioRID II head’s CoG showed minimal acceleration in the y-direction, and significant activity in the x- and z-direction (Figure 3-5). The shape of the head’s accelerometer trace followed the shape of the subject carriage’s accelerometer trace in each direction. Accelerations for both the subject carriage and dummy were largest in the z-directions. The maximum and minimum accelerations for both the subject carriage and dummy, in each direction, are included in Table 3-1. 64  The average free fall duration using 9 out of 10 trials was 321.3 msec with a 26 msec standard deviation.  Total net accelerations (inclusive of earth’s gravity) measured on the subject carriage satisfied the requirements specified in ASTM F2291 – 17. All combined x-y, x-z, y-z accelerations were within the most conservative limits as outlined by ASTM F2291 – 17. A visual summary of the measured accelerations (inclusive of Earth’s gravity) compared to the specified limits is shown in (Figure 3-6).   65   Figure 3-5 Kinematic profiles of the inversion apparatus and surrogate The executed linear motor profile was within the programmed 2mm tolerance of the commanded profile for all six trials. The precision makes it look as though only one trial of data was plotted although all 6 trials are presented. The measured accelerations from the accelerometer adhered to the center line of the subject carriage showed significant activity in the z-direction (left). The measured accelerations from the accelerometer mounted at the BioRID II’s head CoG generally showed an amplified version of the accelerations measured on the subject carriage. One clear difference is the large fore-act accelerations experienced by the head’s CoG in the x-direction.   66     Table 3-1 Tabulated maximum and minimum accelerations experienced by the subject carriage and surrogate Mean (and standard deviation) values for maximum and minimum accelerations as measured at the centerline of the subject carriage and BioRID II head’s CoG. Both accelerometers experienced the largest accelerations in the z-direction. The mean and standard deviations were calculated using nine out of ten trials. The offset due to Earth’s gravity (1g) has been subtracted.  Subject carriage 0.050 0.008 0.034 0.015 1.247 0.009 -0.040 0.010 -0.010 0.002 -1.144 0.002BioRID II head CoG 0.497 0.040 0.075 0.023 1.874 0.076 -0.938 0.054 -0.075 0.021 -1.346 0.025Maximums MinimumsX [g] Y [g] Z [g] X [g] Y [g] Z [g]67   Figure 3-6 Measured accelerations compared to ASTM F2291 – 17 limits The accelerations measured by the accelerometer adhered to the centerline of the subject carriage are plotted for 9 out of 10 trials. The ASTM F2291 – 17 base limits are the outermost series of elliptical curves centered at (0,0). The most conservative limits (50% of the base limit) are the innermost series of elliptical curves centered at (0,0). The conservative limits are prescribed for changes in direction with peak-to-peak transition times <200 msec (x and y directions) or exceeding -2 g (z direction).  ASTM F2291 – 17 does not address sustained exposure to acceleration in excess of 90 s.   3.4 Discussion This chapter verified key performance parameters of the inversion apparatus designed and commissioned in Chapter 2. These parameters included the programmed linear motor trajectory, free fall duration, and the accelerations experienced by both the subject carriage and a human surrogate.  The executed linear motor trajectory remained within the programmed 2 mm tolerance of the commanded trajectory for all trials. This is confirmed by the absence of the position error subroutine being called. The reliable trajectory creates a repeatable acceleration profile of the 68  subject carriage, which leads to consistent experiences for the surrogate (Figure 3-5). Thus, when testing human subjects, this motion profile provides a consistent and reliable stimulus. Different subject masses should not affect the observed oscillations since the weight of the subject carriage is kept constant at 319 kg to avoid re-tuning the linear motor’s PID controls between subjects. In practice, the subject’s (or surrogate’s) weight is recorded and the difference is added via supplemental weights.  Any motions exhibited by the subject could be attributed to their behavior rather than varying perturbations.  The accelerations measured at the CoG of the BioRID II head follow the same general shape as those measured on the subject carriage, except for the x-direction. In the x-direction the accelerations experienced at the BioRID head CoG are larger, reaching a peak of -0.939±0.054 g. This can be explained by the offset CoG within the head (with respect to the head-neck joint), and the BioRID II’s flexible neck. In principle, the head and neck are a mass (head and neck) and spring (neck) system. The free fall stimulus, or any motion of the subject carriage, acts as a base excitation to the head-neck system. The offset CoG within the head creates a moment while the flexible neck permits relative motion between the head and torso (more rigidly coupled to the subject carriage). Thus, the head may have larger displacement amplitudes and experience accelerations different from the subject carriage.   The measured accelerations comply with the most conservative limits set in ASTM F2291 – 17. We interpret this compliance as an acceptable level of risk. Amusement park ride-goers voluntarily go on rides that expose them to directional changes in accelerations. This recreational participation can be interpreted as ride-goers exposing themselves to an acceptable level of risk. Thus, we think conformance with standards set for amusement park rides may be used to guide safe experimental limits. Note that according to ASTM F2291 – 17, there is room 69  to increase the magnitude of the subject carriage deceleration. Increasing the deceleration magnitude would bring the subject to rest within a shorter distance. As a result, more of the available travel could be spent in free fall to extend the free fall duration. However, we decided against increasing the deceleration magnitude because the time for which a subject may be inverted during the experiment is unknown. Factors such as operator delay, securing safety mechanisms, or performing cable management duties may add to inversion time. Prolonged inversion time is undesirable since it creates subject discomfort and may lead to injury. The measured free fall duration is sufficient to capture both the startle and postural response from neck muscles. In response to abrupt vertical descent, humans have shown onset latencies of 24 ms (SD: 1.7 ms) and 24.8 ms (SD:3.9 ms) in the sternocleidomastoid and trapezius muscles, respectively [88]. A free fall stimulus from the supine position has showed muscle latencies of  56±6 ms in the sternocleidomastoid and 57±7 ms in the trapezius muscles [87]. Also, the startle response is has been shown to diminish after approximately 70 ms in young populations [86]. Thus, if a startle response is elicited, it will likely occupy approximately 130 ms of the measured 321±26 ms free fall duration. This estimate leaves approximately 190 ms to record the underlying postural response of interest.   The free fall duration may not be sufficient to capture all of the muscle-induced changes in cervical spine posture. The reaction time for human subjects to begin rotating their heads in response to an acoustic stimulus has been shown to be 313±114 ms [90] based on the onset of head motion. Although the measured free fall duration is within this range (321±25 ms), it would not be sufficient to capture the peak response of subjects with longer latencies. The free fall duration may only be sufficient to capture changes in cervical spine posture from a subset of subjects with faster reaction times.  70  This inversion apparatus represents rollovers to a limited extent. The sensations occupants experience during a rollover are different from those they experience during an isolated free fall. Rollovers often involve 3 – 4 g of centripetal acceleration [64]., As a vehicle trips, prior to rolling, there is a sharp increase in lateral acceleration and an upward acceleration on the far-side occupant. During the initial half roll, the occupant moves upwards (while being accelerated downward by gravity) and rolls into an inverted posture. Technically the occupants are free-falling while they are still moving with an upwards velocity [105]. The vestibular and somatosensory input may be the same, since each system’s receptors detect changes in acceleration and forces, respectively [106]. However, the visual input of moving upwards is different than moving downwards. In other words, the input to the vestibulocollic reflex may be similar but the input to the optocollic reflex is different. To stimulate the optocollic reflex in a more realistic manner during the simulated rollover, a pair of virtual reality goggles could be included with the visual input of ‘falling upwards’.    The inversion apparatus outlined in Chapter 2 was verified in this chapter. The free fall duration is sufficient to capture the neck muscle response of human subjects. Although, the duration may only be sufficient to capture initial changes in cervical spine posture for a subset of human subjects. The accelerations measured on the subject carriage conform to the most conservative limits set in ASTM F2291 – 17, and are interpreted as acceptable. The results of the inversion apparatus verification confirm its suitability for human subject testing.  3.5 Conclusion The custom inversion apparatus discussed in Chapter 2 was evaluated for suitability in human subject testing. The motion profile of the linear motor was consistent, and resulted in 71  consistent accelerations of the subject carriage and a human surrogate. The acceleration levels conformed to ASTM F2291 – 17, and thus we interpret them to be sub-injurious. The free fall duration was measured and is appropriate for studying neck muscle responses and early changes in cervical spine posture. These results indicate the measured accelerations of the inversion apparatus are reasonable for use with human subjects.  72  Chapter 4: Cervical vertebral kinematics and neck muscle responses during an inverted free fall simulating a vehicle rollover: pilot data from an in vivo human subject experiment  4.1 Introduction Although rollovers account for only 3% of motor vehicle crashes [27], they are responsible for 33% of all motor vehicle fatalities [100] and 40% of serious cervical spine injuries [28]. Vehicle rollovers are a chaotic and complex type of motor vehicle collisions in which the exact injury mechanism is poorly understood. Neck injury can occur when the vehicle roof hits the ground and the inverted occupant strikes the interior of the vehicle with their head [29,107]. It’s been suggested that the cervical spine is subsequently loaded by the incoming momentum of the torso [31]. Currently there is a disparity between lab-induced injuries and those reported from rollovers clinically. In cadaver studies, almost half of lab-induced injuries occur as vertebral body fractures while, in the rollover cases, unilateral or bilateral facet dislocations were common (39.1%) as were facet joint fractures (34.7%) but vertebral body fractures occurred to less than 15% of the case occupants  [2]. The inability to replicate real-world rollover injuries may be due to assumptions used to study cervical spine injury. In cadaveric tests, musculature is removed to aid visualization of the cervical spine during an impact test, although muscle tone may be simulated using systems of cables and pulleys. It’s been suggested that the absence of neck musculature is likely responsible for the disparity between lab-induced and real-world injuries [2]. Computational modelling has suggested that active neck musculature may shift the mode of injury [7], and more than double the risk of neck 73  fracture [6]. The validity of these findings depends on the simulation of realistic muscle activation schemes. Current activation schemes aren’t based on, or validated to, measured in vivo responses. Existing models assume 100% maximum voluntary contraction (MVC) [6] or a scheme optimized for upright, neutral posture [51,108], neither of which represent a vehicle rollover. The accurate simulation of muscle forces and spine postures from an impending headfirst impact, in future cadaver tests, may drive the injuries from cadaveric tests towards concordance with clinically observed injuries. Previous work in our lab has shown that muscle activity and cervical spine posture are altered due to inversion alone [15] and differ yet again when voluntarily bracing after being instructed to brace for an impact under quasi-static conditions [62]. However, the cervical spine posture and muscle activity of an occupant immediately before a headfirst impact is unknown. Therefore, the objective of the current study was to capture the in vivo dynamic cervical spine re-alignment and muscle activity of a human subject in response to a simulated impending headfirst impact.   4.2 Methods 4.2.1 Human subject To date, one asymptomatic, 32-year-old, male human subject participated in this study. The male subject met the following exclusion criteria: no prior whiplash injury, history of neck or back pain, known disease affecting muscles or nerves, history of balance problems, known heart condition, skin disease, history of cancer, or involvement in a study involving radiation in the last year. In the case of a female subject, pregnancy would be an additional exclusion criterion. The subject’s height was 1.78 m, his weight was 80.9 kg, and his head and neck circumference were 0.585 m and 0.385 m, respectively. This study was approved by the 74  University of British Columbia’s Clinical Research Ethics Board, and the subject gave his written informed consent. The subject gave additional permissions to present photographs and video recordings of his participation in the experiment.   4.2.2 Custom-built inversion device A custom device was designed and built to expose human subjects to an inverted free fall drop that simulates a short phase of a rollover crash. Subjects are first seated in an upright posture and then inverted and elevated to a fixed drop height by a feedback controlled linear motor. The linear motor is programmed to expose subjects to a controlled 312 ms inverted free fall drop, before decelerating to rest (peak deceleration of 1.34g). An onboard fluoroscopic C-arm (OEC 9400, GE) retrofitted with a high-speed camera (Phantom V12.1M, Vision Research Inc., Wayne, NJ), was used to capture sagittal plane images of the cervical vertebra. A shelf mounted to the seat back accommodated pre-amplifiers used to collect Electromyography (EMG) data.   4.2.3 Conditions The subject was seated in a bucket seat (36 series – Intermediate 20-degree Layback, Kirkey Racing Fabrication INC., St. Andrew’s West, ON) secured with a 75-mm wide 5-point harness (RCI Racers Choice Inc., Tyler, TX) with his arms strapped to his thighs and feet restrained using snowboard bindings. The subject adopted two static postures and one dynamic posture: upright with a neutral posture and relaxed muscle activity (U-R), inverted while maintaining a forward gaze (I-F), and inverted while subjected to an inverted free fall drop (D) (Figure 4-1). The upright-relaxed condition is akin to initial conditions currently used in 75  cadaveric and computational models while the inverted-forward condition represents inverted occupants maintaining their gaze on the road. The drop condition is intended to represent inverted occupants with pre-impact awareness. Prior to the onset of the free fall drop, the subject was instructed to adopt a forward gaze. One trial was performed for each static condition, whereas four trials were performed for the drop condition with approximately 30 minutes rest between trials. Neck muscle activity and fluoroscopy were recorded throughout all trials.   Figure 4-1 Four video frames from the inverted drop condition The first three frames show the subject being raised to the fixed drop height. The last frame shows the subject during the free fall drop condition.  4.2.4 Electromyography Muscle activity was measured using unilateral indwelling electrodes for 8 neck muscles: sternocleidomastoid (SCM), trapezius (Trap), levator scapulae (LS), splenius capitis (SPL), semispinalis capitis (SsCap), semispinalis cervacis (SsCerv), and multifidus (MultC4) [109] on the left side only. Ultrasound guidance was used to insert the indwelling electrodes into the muscle bellies at the C5-C6 level (Trap) and C4-C5 level (remaining muscles). EMG signals 76  were amplified at 100x gain and sampled at a frequency of 4000Hz. Two hardware filters band-passed frequencies between 50-2000Hz (NL844 & NL144, Digitimer Ltd., Hertfordshire, UK), after which a digital 4th-order high-pass Butterworth filter with a cutoff frequency of 50 Hz was applied. The onset of muscle activity was defined as the sample where the average value of rectified EMG in a 25 ms window around the sample was two standard deviations above the average resting EMG before the stimulus in each trial [110]. To gauge muscle activity relative to maximal muscle activity, the root mean square (RMS) of each signal was calculated for a moving 25 ms window centered around each point for every trial. For the static conditions, windows were chosen in which fluctuations in muscle contractions and neck posture were minimal. For the drop condition, windows were chosen to capture the maximal RMS activity of a muscle during the free fall drop (0 to 312 ms). The RMS average using a 100 ms time window was also calculated for the 100 ms prior to the onset of free fall in each trial to indicate pre-drop quasi-static muscle activity. Each muscle’s RMS activity was normalized to the maximum RMS activity recorded for that muscle in a maximum voluntary contraction (MVC).   4.2.4.1 Maximum voluntary contractions (MVC’s) The seated subject was secured with a chest strap to a rigid backboard while wearing a snug skateboard helmet that was attached to a 6-axis load cell (45E15A-UU760, JR3, Woodland, CA) above the subject’s head [15]. Two repetitions of 10 isometric contractions were performed with verbal encouragement [111] and real-time visual feedback representing force/moment magnitude and direction. The MVCs were performed for 3-5 seconds in a randomized order in 10 directions: flexion, extension, left lateral bending, right lateral bending, four 45° oblique combinations (flexion/left lateral bending, flexion/right lateral bending, extension/left lateral 77  bending, extension/right lateral bending), and left and right axial rotations. The EMG signals were amplified, filtered, and collected as described above. A moving 25 ms window was used to calculate RMS activity at each point. The maximum RMS value for each muscle, regardless of direction, was used for normalization.   4.2.5 Cervical spine posture To determine cervical spine posture in the sagittal plane, the fluoroscopic C-arm recorded images of the cervical vertebra at 200Hz. Fluoroscopic images were corrected for spherical distortion using XMA Lab [81]. Inertial loads due to the dynamic environment resulted in mechanical flexing of the C-arm, and consequent misalignment of the x-ray source and image intensifier. To correct for image distortion due to misalignment, a 10x10 array of 1mm steel beads spaced uniformly apart was mounted in the field of view. Images were imported into photo rectification software (PC-Rect, MEA Forensic, Richmond, BC) and rectified separately using a square formed by four of the steel beads. Photoshop (Photoshop CS4, Adobe Systems, San Jose, CA) was used to manually outline each vertebral body along high-contrast boundaries. To be able to make comparisons between images, two individual images were superimposed such that the beads from the bead array coincided, providing a common reference. Superimposed images where imported into video tracking software (TEMA Automotive, Image Systems AB, Linköping, Sweden) where vertebral corners, midpoints, and reference points were manually identified. A sample of the image analysis process is shown below in Figure 4-2. Four image comparisons were performed: upright-relaxed (U-R) vs. inverted-forward (I-F), and initial vs. final frames for three of the four drop trials. Initial frames were the last frame prior to free fall onset, and final frames were the last frame prior to the onset of deceleration. 78   Figure 4-2 Sample of the image analysis process  The images in this sample have been corrected for spherical distortion, vertebral bodies outlined (faint red), images rectified, and superimposed to allow comparison between the identified vertebral corners. The beads from the bead array of the initial image coincide with the beads from the bead array in the final image. The 2mm steel bead array related to the Frankfort plane angle is visible for both frames. The common reference line connects points 37 and 38. Changes in cervical spine posture were quantified with two measures: the displacement of vertebral bodies and change in vertebral angle (𝜃𝑉) between images. The displacement was measured between the inferior midpoints of each vertebra. Distances were normalized to a reference line common to all frames and reported as a fraction thereof. Vertebral angles were determined by a line drawn through the two inferior corners of each vertebra and measured relative to the common reference line, as shown in Figure 4-3. 79   Figure 4-3 Method by which vertebral angles were determined Vertebral angles were determined by drawing a line through the two inferior corners of the vertebra and measured relative to the common reference line. The dotted horizontal line is parallel to the common reference line described in Figure 4-2.  4.2.6 Head orientation To determine head orientation, a bead array (six-2mm beads) was fastened to the head and projected into the field of view of the fluoroscope. The Frankfort plane angle (𝜃𝐹) was defined as a line from the mid-point of the left tragus and inferior midpoint of the left orbit and was measured relative to the common reference line (positive=extension) (Figure 4-4).  Prior to the experiment, the beads and Frankfort plane landmarks were digitized using a 3D motion capture system (Optotrak Certus, Northern Digital Inc., Ontario, Canada) to establish the initial offset. Thus, angular motion of the beads could be used to determine angular motion of the Frankfort plane.  Figure 4-4 Method by which Frankfort plane angle was measured Frankfort plane angle was measured relative to the common horizontal (the dotted line parallel to the common horizontal reference line), with extension in the positive direction. 80  4.2.7 Data processing notes The upright-relaxed, inverted-forward, and drop conditions were performed before the MVCs. During the MVC portion of the experiment, the indwelling electrode wires for STH and SsCap may have been damaged as the subject transferred from the inversion device to the MVC set-up. Cervical vertebra C1-C6 were visible in the 23cm field of view for all trials, except C6 in the upright-relaxed condition. Prior to the experiment, the mirror which optically couples the camera and image intensifier had unknowingly come loose. As a result, images were dark and out-of-focus. Due to image quality, the boundaries of C1-C2 and C7 were not discernable and could not be analyzed. Additionally, fluoroscopic images were not recorded for the free fall duration of Trial 2 due to human error.  All head/neck posture variables and EMG signals were analyzed using Matlab (R2017a, MathWorks Inc., Natick, MA).   4.2.8 Error estimates of head and neck posture Error was estimated for vertebral translations and angles and Frankfort plane angle. The initial and final frames from Trial 1 were undistorted, outlined with vertebral bodies, and rectified 10 times separately. These 20 initial and final frames were then randomly matched to form 10 pairs of initial and final frames. Each pair of images was then superimposed such that the beads from the bead array coincided. Vertebral displacement, angle and Frankfort plane angle was then determined for each pair of images. The average displacement and angle per vertebra was calculated, as well as the average initial and final Frankfort plane angle. The standard deviation for each of these parameters was calculated assuming a normally distributed sample.  81   4.3 Results 4.3.1 Muscle activity In each trial, unprocessed EMG signals showed a large initial burst of activity followed by several moderately sized bursts. Muscle activity approached minimal levels prior to the onset of deceleration (Figure 4-5). The onset of muscle activity occurred within the first 100 ms of each trial. There appeared to be a consistent onset of muscle activity after the free fall stimulus over the four trials (Figure 4-6).  82   Figure 4-5 Unprocessed EMG signal from Trial 1  This unprocessed EMG signal has been high-pass filtered at 50Hz. Rms EMG is included and has been overlaid. Vertical scale bars represent 4 mV. The onset of deceleration is 312 ms. 83   Figure 4-6 Onset of normalized rms EMG for each trial The onset of muscle activity was consistent across the four trials.  While upright and relaxed, the subject had minimal muscle activity with SsCap most active at 11.59% MVC (Table 4-1). When inverted and maintaining a forward gaze, muscle activity increased in five of the eight muscles. Activity in the extensor muscles MultC4, SsCap, and 84  SsCerv decreased compared to the upright and relaxed condition. In the drop trials, all muscles were minimally active in the 100 ms prior to the onset of free fall. During free fall, maximum activity levels of all muscles were significantly higher than compared to either static posture (Table 4-1). Most muscles increased activity by an average of more than 40%, excluding LS and SPL, which increased by less than 20%. The muscle activation pattern was generally consistent across all four trials, apart from SsCap and SPL in Trial 2 (Figure 4-7).  Table 4-1 Summary of normalized EMG rms average Normalized EMG rms average based on a 100 ms window, applied to the 100 ms prior to the drop onset. Maximum %MVC values recorded for each muscle across all 4 trials and both static postures (U-R = upright-relaxed, I-F = inverted-forward gaze).  %MVC 100ms prior to drop onset Maximum %MVC %MVC Muscle Trial 1 [%] Trial 2 [%] Trial 3 [%] Trial 4 [%] Trial 1 [%] Trial 2 [%] Trial 3 [%] Trial 4 [%] U-R [%] I-F [%] Trap 2.72 3.96 4.50 4.57 31.67 57.83 66.02 42.06 3.15 9.29 LS 3.51 2.59 2.18 5.44 13.69 24.47 19.03 22.87 6.96 13.20 SCM 5.35 6.22 5.05 7.12 50.36 65.83 53.79 43.41 1.32 12.66 STH 3.45 5.62 4.83 3.68 98.18 127.96 95.16 72.24 4.52 12.52 MultC4 1.09 1.12 1.19 1.09 42.49 68.10 50.90 39.20 5.11 1.29 SsCerv 1.87 1.26 1.32 1.22 48.42 52.40 56.53 66.23 1.88 1.41 SsCap 2.09 2.07 2.16 2.13 102.17 304.23 87.99 110.92 11.59 2.81 SPL 2.00 1.79 1.92 1.79 17.52 52.37 12.80 15.12 2.64 3.33  85   Figure 4-7 Normalized rms EMG for each muscle across all four trials. 86  Muscle scale bars indicate 100% of maximal voluntary contractions (MVCs). 4.3.2 Neck and head posture Neck vertebral postures were different in all three conditions. In the inverted-forward condition, vertebral bodies moved posteriorly and superiorly, when compared in the same orientation, to the upright-relaxed condition. Prior to the onset of free fall, the initial inverted-forward vertebral postures were consistent with the static inverted-forward condition. After experiencing free fall and prior to the onset of deceleration, vertebral bodies moved anteriorly and inferiorly – indicating neck flexion (Figure 4-8). Vertebral angles increased with inversion, and more so with exposure to free fall. Frankfort plane moved in concert with the vertebrae, indicating head flexion with exposure to free fall (Table 4-2).  Figure 4-8 Summary of vertebral translations throughout all trials Vertebral translations between the initial posture prior to the onset of free fall, and the final posture prior to the onset of deceleration. Distances are expressed as a fraction of the common horizontal reference line. The static postures are upright-relaxed (U-R) and inverted-forward (I-F). Vertebra C3-C6 were identified for all conditions except for the upright-relaxed condition, of which C6 was missing. The conditions have been plotted in the same orientation for comparison; C3 is plotted as the top-most vertebra with C4 immediately below, and so on. 87  Table 4-2 Summary of vertebral and Frankfort plane angles.  Vertebral and Frankfort plane angles were measured relative to the horizontal reference line (bottom-most row of bead array plate). Decreasing Frankfort plane angles indicate flexion. The Frankfort plane assumes that the Image Intensifier and Optotrak’s Y-Z plane are parallel.      Trial 1 Trial 2 Trial 3 Trial 4   U-R I-F Initial Final Initial Final Initial Final Initial Final  Vertebral angles [deg] C1 - - - - - - - - - - C2 - - - - - - - - - - C3 19.75 22.30 22.85 29.90 - - 24.22 42.29 27.98 40.78 C4 22.58 30.00 17.27 29.14 - - 24.24 40.38 24.07 40.75 C5 20.48 23.81 24.00 31.58 - - 31.20 38.45 21.46 35.88 C6 - 22.71 26.28 34.93 - - 24.74 37.14 30.42 45.70 C7 - - - - - - - - - -  Frankfort plane angles [deg]  1.62 13.40 12.22 7.22 - - 6.30 -4.54 8.80 1.62  4.3.2.1 Error estimates of head and neck posture The calculated mean and standard deviation of vertebral translations for each pair of images is included in Table 4-3. The calculated mean and standard deviation of vertebral angles and Frankfort plane angles is included in Table 4-4.  Table 4-3 Error estimates of vertebral translations Means (and standard deviations) of vertebral translations calculated for 10 pairs of initial and final frames from Trial 1. Pixel distances have been normalized to the pixel length of the common reference line.  x y  Vertebral translations [px/px] C1 - - - - C2 - - - - C3 0.1091 (0.0048) -0.1569 (0.0065) C4 0.1399 (0.0030) -0.1655 (0.0141) C5 0.1679 (0.0032) -0.1779 (0.0079) C6 0.1889 (0.0049) -0.1740 (0.0074) C7 - - - - 88   Table 4-4 Error estimates of vertebral and Frankfort plane angles Means (and standard deviations) of vertebral angles for 10 pairs of initial and final frames from Trial 1.   Initial  Final  Vertebral angles [deg] C1 - -  - - C2 - -  - - C3 25.90 (4.08)  28.76 (3.94) C4 20.34 (2.91)  27.56 (4.33) C5 17.73 (1.46)  34.02 (2.06) C6 26.03 (3.12)  33.34 (2.96) C7 - -  - -  Frankfort plane angles [deg]  8.29 (0.62)  6.57 (1.52)  4.4 Discussion The objective of this study was to measure in vivo muscle activation patterns and the realignment of human cervical vertebrae in response to an inverted, free fall, impending headfirst impact. Overall, we observed sub-maximal increases in muscle activity followed by muscle-induced anterior motion of the cervical spine and combined flexion-retraction of the head. These observations support our previous findings that the in vivo state of the neck, at a time relevant to a headfirst impact during a rollover crash, may differ considerably from its initial alignment prior to a forewarned impact [62]. In previous work done in our lab, inverted subjects were instructed to ‘brace for impact’ under quasi-static conditions [62]. Their responses were performed without any actual threat and depended solely on their interpretation of the instructions. The current study applies a free fall stimulus under the threat of impending headfirst impact, generating a more realistic reflexive neck muscle response. These dynamic conditions are more likely to reflect the posture and 89  muscle state immediately before an inverted headfirst impact. Applied to vehicle rollovers, the end of free fall approximates the instant of head contact with the vehicle roof, and thus the state of the spine at this instant is potentially relevant to catastrophic neck injuries in rollover crashes.    Our results align with previous work done in our lab that the in vivo muscle activation levels of inverted subjects differ considerably from those of upright subjects. All muscle activation levels for the static conditions are within two standard deviations of previous studies [15] except for STH, SsCap, and SPL in the upright-relaxed condition. Activity levels associated with STH and SsCap may be artificially high as the indwelling wires may have been damaged prior to the MVC procedure. Consequently, the absolute activity levels of STH and SsCap are not reliable, although the shape of the muscle activation pattern and relative magnitudes of muscle activity between trials are still valid. Under dynamic conditions, we observed a different muscle activation scheme than the one seen in inverted subjects who voluntarily adopted a quasi-static bracing posture (Figure 4-9) [62]. In the quasi-static bracing task, Trap and SPL increased the most (mean increase of 36.2% and 22.7%, respectively), whereas we observed six muscles with mean increases greater than 40%, excluding SPL. Another of these muscles was MultC4, which had previously increased the least (mean increase of 4.3%) in the quasi-static task. The quasi-static bracing task reported high levels of between-subject variability, with confidence intervals as high as 68%. Nevertheless, in our dynamic muscle activation scheme, 5 muscles exceeded the level of muscle activity from the quasi-static bracing task (Figure 4-9). 90   Figure 4-9 Summary of maximum muscle activity between free fall and a voluntary bracing posture Reproduced here are neck muscle activations from an inverted, quasi-static, voluntary bracing task (grey). Mean muscle activity is identified with a point, while the horizontal lines indicate a 95% confidence interval. Overlaid in red is the maximum muscle activity recorded during each free fall trial from this work. The red arrows on the left identify five muscles that consistently exceeded the 95% confidence intervals of the quasi-static bracing task in each trial.   The observed muscle activation pattern was consistent across all four free fall trials, except for SPL and SsCap in Trial 2. After consulting the raw data, the peaks in these two muscles did not resemble EMG activity and should not be interpreted as such. Damage to the 91  wires was not suspected since the subsequent responses in trials 3 and 4 were consistent with trial 1. Since the artefact occurs after a large initial burst had already taken place, onset latency is not affected by the artefact. The consistency of the reflexive muscle response across all four trials suggests that between-subjects variability may be reduced in a reflexive response.  It should be noted that the EMG onset of Trap, LS, SCM, STH is not a true onset, as the muscles were already active while the subject tried to maintain an inverted-forward gaze (Figure 4-7). Cervical vertebral translations followed the same trend as the quasi-static bracing task [62]. Unfortunately, a comparison of distances is precluded by the missing vertebra (C1-C2, C7) and the yet undetermined scaling factor between the image intensifier and the subject plane. The decrease in Frankfort plane angle was not observed in previous work, suggesting a reflexive response generates a different head posture than a quasi-static tensing.         Our observed cervical spine posture follows the same general trend as in our previous quasi-static studies [17]. In both this thesis and our previous work, the cervical vertebra moves anteriorly, and inferiorly (Figure 4-10). The distance between initial and final vertebra positions is larger for the superior vertebra (C3 in Figure 4-10) than the inferior vertebra (C6 in Figure 4-10), indicating flexion rather than a whole-body anterior and inferior translation occurred between the initial and final postures. However, due to lack of instrumentation on the subject’s torso, the relative contributions of neck or thoracic flexion are unclear.   92   Figure 4-10 Comparison of cervical spine posture during free fall and a voluntary bracing task Outlined in red is exemplar data of vertebral motions from an inverted, quasi-static, voluntary bracing task [62]. Outlined in dark blue is data of vertebral motions from each free fall trial in this experiment. Both data sets show anterior and inferior translations of the vertebra. A full comparison is precluded by missing C1-C2, and C7 vertebral bodies in this experiment.  Most existing cervical spine injury models assume an upright posture with muscle activity simulated at 100% MVC [6] or to represent an upright-relaxed posture [51,108]. These assumptions give little consideration to the effect of pre-impact awareness. Our results show the in vivo muscle activity (Figure 4-7) and cervical spine posture (Figure 4-8) during free fall differ from the upright-relaxed condition. Aside from artificially high activations in SPL and SsCap, none of our muscles exceeded 70% MVC, yet all exceeded upright-relaxed activation levels. This difference between inverted occupants maintaining a forward gaze and those preparing for impact (Figure 4-7) illustrates the effect of pre-impact awareness. While different from one another, both conditions showed increased muscle activity and a shift in cervical spine posture when compared with the upright-relaxed condition. Thus, the initial conditions used in current cervical spine injury models may not be representative of a headfirst impact in a rollover.  93  Cadaver studies have shown that injuries to the spinal column are sensitive to the overall spinal eccentricity. Average eccentricities of -5mm, 1mm, 23mm, and 53mm are reported to result in compression-extension, vertical compression, compression-flexion, and hyperflexion injuries, respectively [5]. Thus, the cervical spine posture represented in the initial conditions should be carefully considered as they are likely to change the injury outcome. Further work is needed to evaluate how much the eccentricity of subjects exposed to our free fall stimulus changes and how relevant these changes will be to the risk of different neck injuries. For this study, a single human subject was exposed to an inverted free fall intended to simulate a short phase of a vehicle rollover. These results should be interpreted carefully as more subjects are needed before reaching definitive conclusions. Additionally, rollovers are dynamic and complex events, and pre-rollover dynamics are not captured in this study. Centripetal accelerations in a rollover environment could influence both muscular and postural responses and future experiments are necessary to understand these effects. The aim of this study was to capture muscle activity and posture in a condition which may exist immediately before an inverted headfirst impact. Our findings are relevant to other circumstances of inverted, impending headfirst impact as it provides evidence that a reflexive response to a free fall stimulus can generate significant change in muscle activity and cervical spine posture.   4.5 Conclusion A custom inversion device was built to simulate impending headfirst impacts and to capture cervical spine posture and muscle activity. An in vivo data set of vertebral and muscular responses, in the context of pre-impact in a rollover environment, was collected. Sub-maximal increases in muscle activity were observed, followed by muscle-induced anterior motion of the 94  cervical spine and flexion of the head. These results indicate the initial conditions used in current cervical spine injury models may not reflect those present during an inverted headfirst impact. An in vivo data set of vertebral and muscular responses could be used to improve and validate current injury models and advance injury prevention strategies.  95  Chapter 5: Discussion 5.1 Overview The overarching goal of this work was to advance understanding of cervical spine injury in headfirst axial impacts, such as those seen in vehicle rollovers. Cadaver studies, computer models, and ATDs have been useful tools in studying axial impact cervical spine injuries. However, each of these tools is limited in their biofidelity. Cadaver studies suffer primarily from their inability to simulate dynamically active musculature. Computer models and ATDs are validated to cadaveric data or are developed using human data from unrepresentative scenarios. Furthermore, for lack of better data, most of these tools assume an upright and relaxed neck posture. However, previous work in our lab has shown that neck posture and muscle activity are orientation and task dependent, indicating that an upright and relaxed posture is likely not relevant to a rollover environment. The aim of this thesis was to investigate neck posture and muscle activity in an environment relevant to cervical spine injury in vehicle rollovers. To accomplish this, an inversion apparatus capable of safely exposing subjects in an inverted free fall was first developed. Thereafter, a pilot human subject experiment was conducted in which muscle activity and cervical spine posture were recorded.   The first objective of this thesis was to develop an apparatus which simulates impending headfirst impact with sufficient time to observe a muscular response and postural re-alignment of the cervical spine. This was accomplished through the design, commissioning (Chapter 2:), and verification (Chapter 3:) of an inversion apparatus built by an MEA Employee, Mircea Oala-Florescu. The measured free fall duration was 321.3 ms with a standard deviation of 26 ms, which is sufficient to capture both the startle and postural response from neck muscles [86–88]. However, the measured free fall duration may be insufficient to capture all of the muscle-96  induced changes in cervical spine posture. In some subjects, head motion lags a stimulus by up to 427 ms [90], which exceeds the measured free fall duration. The free fall distance, and thus duration (given fixed acceleration limits), could not be increased due to dimensional constraints of the test facility. The acceleration levels recorded on the apparatus complied with ASTM F2291 – 17, which specifies acceleration levels for amusement rides, and thus we interpret the inversion apparatus is suitable for use with human subjects  Previous dynamic rollover studies involving human subjects have been ex vivo and limited to gross occupant motion [66,73–75,102]. In some studies, surface electrodes have been placed on the skin to record superficial neck muscle activity[74,75] and external marker tracking has described occupant motion [73]. While valuable, these studies cannot accurately describe deep neck muscle activity or changes in vertebral alignment under rollover conditions. External surface markers are not sufficient to track motion of the underlying bony structures [77] and surface EMG is limited to superficial neck muscles. The inversion apparatus developed in this thesis accommodates sagittal plane fluoroscopy, which is the gold standard for tracking bony landmarks, and indwelling EMG, which allow access to deep cervical muscles. To our knowledge, this is the first apparatus which may expose a human subject to inverted free fall while recording the in vivo state of the neck.   The second and third objectives of this thesis were to analyze the in vivo cervical spine kinematics and neck muscle activity in response to an impending headfirst impact. To accomplish these objectives, the inversion apparatus was used to expose a human subject to a controlled free fall drop while recording cervical vertebral motion (via fluoroscopy) and neck muscle activity (via indwelling EMG). The muscle response pattern was different from both the upright-relaxed scheme used in existing cervical spine injury models and patterns observed in previous quasi-static tests conducted in our lab [62], confirming the first hypothesis of this thesis. 97  Moreover, the subject showed consistent muscular responses in all 4 repetitions of the free fall exposure. Cervical spine posture changed dynamically and characterized a more protective bracing posture, confirming the second hypothesis of this thesis. C3-C6 translated anteriorly and inferiorly in response to the inverted free fall stimulus, and the head moved into flexion.  In our previous work, the effects of inversion and adopting a quasi-static bracing posture on neck muscle response and cervical spine posture was studied in eleven human subjects [112,113].  Large inter-subject variations were observed in both tasks, with some subjects adopting different strategies. In the quasi-static bracing task, some subjects straightened their spine while others increased their spinal curvature with confidence intervals on muscle activity reaching as high as 68% (i.e. the 95% confidence interval spanned muscle activation levels of 20% to 88%) [112]. In a study by Bisdorff et al., reflexive and voluntary neck muscle activity was compared. In the reflexive task, supine subjects were exposed to a free fall stimulus via a tilting couch. In the voluntary task, subjects were asked to contract their neck muscles as quickly as possible following an auditory signal. The reflexive response showed shorter latencies with reduced inter-subject and inter-muscle variability [87]. These findings suggest that, as we expand our subject sample, we may see less inter-subject and inter-muscle variation than in previous quasi-static bracing tasks. This corresponds to a more predictable state of the neck prior to a headfirst impact. Consequently, this would provide CSI models and injury prevention strategies with more reliable universal initial conditions.  On the other hand, if a more cohesive response is not observed, large inter-subject variation may suggest one-size-fits-all solutions are not an appropriate prevention strategy for cervical spine injuries caused by headfirst impacts.   98  5.2 Strengths and limitations 5.2.1 Strengths As current cervical spine injury (CSI) models employ muscle activation schemes and postures typical of an upright and relaxed posture, a data set representative of a person’s active response in a dynamic rollover environment is necessary. This thesis work addresses both by simulating the free-fall phase of a dynamic rollover environment and capturing the resulting in vivo response of a human subject. A major strength of this study is the synchronized in vivo data set of cervical spine kinematics and neck muscle activity, for both deep and superficial muscles. The use of a fluoroscopic C-arm in this work allows a description of exact cervical spine postures via tracking vertebral landmarks. The use of indwelling EMG electrodes allows access to deep neck muscle activity, enabling a more comprehensive data set compared to studies using surface electrodes. This work uses tools and techniques capable of accurately describing the in vivo state of the neck prior to a headfirst impact. Thus, the resulting data set is appropriate for informing the initial conditions (i.e. neck muscle activity and cervical spine posture prior to headfirst impact) of CSI models.   Previous studies in our lab used the same in vivo techniques to investigate the effects of inversion [15] and a voluntary bracing posture [62] on neck muscle activity and cervical spine posture. These previous studies were conducted under quasi-static conditions. To achieve their braced posture, these prior participants were verbally instructed to brace in preparation for a hypothetical headfirst impact. In the absence of a fear or startle stimulus, the participant’s voluntary response depended on their interpretation of the experimenter’s instructions. However, rollovers are dynamic events in which a reflexive muscle response is likely, and this reflexive response may differ from the voluntary response [87]. The biggest strength of this thesis over our 99  previous in vivo work is the dynamic nature of the tests. In this work, the free fall stimulus provides an element of startle, which generates a reflexive response. Reflexive responses to free fall stimuli produce less variable neck muscle activations (onset latencies, inter-subject and inter-muscle variability) than voluntary responses [87]. This decreased variability suggests the subject’s responses in this work may be more representative of a rollover occupant than our previous in vivo, quasi-static, work.  This work makes several contributions to furthering understanding CSI in axial impact injury scenarios. This is the first in vivo human subject study to quantify neck realignment and muscle patterns in response to an inverted free fall. The results demonstrate that an upright and relaxed posture is likely not representative of the state of the neck in a rollover scenario. This pilot work provides a basis for future human subject testing. Based on the success of the pilot tests, additional human subjects may be recruited for participation. The image analysis method also offers a new approach for correcting fluoroscopic images collected in a dynamic environment.  5.2.2 Limitations The most important limitation of this work is the simplification of vehicle rollover dynamics. Firstly, the inversion apparatus was not a realistic simulation of the movements leading up to a rollover. Vehicle rollovers, whether tripped or un-tripped, are subject to pre-rollover dynamics in which large lateral loads create lateral acceleration for the occupants and tipping moments for the vehicle.  A subset of tripped rollovers may be the result of an avoidance maneuver into soft soil, guard rails, or a steep slope. Un-tripped rollovers involve high speed collision avoidance maneuvers such as swerving [39]. During all of these maneuvers, whether 100  tripped or un-tripped, the pre-roll dynamics of a vehicle may have an appreciable effect on an occupant’s position, posture, and muscle activity prior to the inverted free fall that develops for a far-side occupant near the end of the first half roll. Secondly, the inversion apparatus did not incorporate any of the centripetal effects present in a rollover. Rollovers are complex events which often involve 3-4 g of centripetal acceleration [64]. As a consequence of these centripetal accelerations, belted occupants are forced upward and outwards (relative to the occupant compartment) as a vehicle rotates during a rollover [114]. Upward and outward excursions often result in head contact with the vehicle’s upper interior prior to the vehicle roof impacting the ground.  This head-to-roof interaction influences head-to-neck geometry. As centripetal effects force the body outwards, and the roof provides an obstacle to the head, the neck shows an increase in lateral bending (both inboard and outboard occupants). In cases where head excursion is sufficient for head-to-roof contact, the neck can experience compressive loading as a function of the roll rate, prior to roof-to-ground impact [115]. These changes in head-neck geometry and compressive pre-loading of the neck influence cervical spine posture, and perhaps muscle activity, prior to roof-ground impact. The inversion apparatus used in this work does not incorporate any centripetal effects, meaning the influence of roof-head geometry and compressive pre-loading on cervical spine posture and muscle activity is not accounted for. Thus, this work may only be directly relevant to a small subset of vehicle rollovers with low roll rates, in which the centripetal effects cause minimal head excursion. Nevertheless, the general findings of altered neck posture and muscle activity suggest that the current assumptions of existing cervical spine injury models need to be re-examined. Lastly, the inversion apparatus does not simulate a vehicle roof, and some occupants have shown an active response to lack of head room [73].  101  Another important limitation of this work is the restrictions placed on occupant motion. A 5-point harness was used to restrain the subject, as opposed to the 3-point safety belt used by many rollover occupants. Three-point safety belts are most useful in frontal collisions as they prevent the body from moving forward into the vehicle interior [116,117]. However, they do not effectively prevent upwards motions and permit substantial head excursion during rollover tests [114,118]. The 5-point safety harness used in this work limits the amount of head excursion via a lap belt with crotch strap, and two shoulder straps mounted below the top of the shoulders. The difference in shoulder contact points between the 5-point and 3-point safety belts create different belt-shoulder interactions. The difference in shoulder interactions likely influences both the postural and muscular outcomes of this work. For safety reasons, a 5-point harness was necessary for this study.  Restraining the subject’s hands to their thighs also imposes limits on occupant behavior. Reviewing dashcam videos of vehicle rollovers shows subjects have a variety of arm responses. Drivers may keep their hands on the wheel [119,120], raise both arms towards the roof [121], or attempt a T-shaped posture, placing one arm over the passenger seat and the other on the window [122]. Other occupants may keep their hands on the wheel initially, eventually raising one arm toward the vehicle roof before roof-ground impact [123]. These responses affect torso motion and thus the torso-neck interface, which is an important factor in cervical spine injury [4,8].   The sensations that occupants experience during a rollover are different from those the subject experienced during the isolated free fall. As the subject is suddenly released into free fall from a stationary posture, cutaneous receptors sense a rapid unloading of the 5-point harness and otoliths (or gravireceptors) sense a downward acceleration. This stimulus has a clean, sudden onset which evokes a startle response followed by a voluntary bracing response. During a 102  rollover, the sensory stimuli are considerably more complicated: not only are they multidirectional, they also develop over time. The free fall phase of a rollover does not begin until well into the event (and true free fall may not occur at all in many rollovers), and it can even begin while the occupant still has an upward velocity. Under these circumstances, the onset of free fall may not evoke a startle response. Startle is easily inhibited by small warning cues that precede a stimulus, and in rollovers there are large multi-directional accelerations preceding the free fall phase [105]. In the free fall experiment, the startle response takes time to develop into head and neck displacements. By the time meaningful head and neck displacements have developed, muscle activity has transitioned into the voluntary portion of the response. Thus, the protective muscle response reported in this work may not be present during the free fall phase of a rollover. Additionally, the reported cervical spine posture may be unrelated to the voluntary portion of the postural response. Head and neck motion lags muscle response, and the voluntary muscle response may not have sufficient time, in this experiment, to generate muscle-induced changes to cervical spine posture. Due to the startle response from an artificial initial posture in our experiment, the cervical spine posture recorded may not be relevant to the circumstance of catastrophic neck injuries in rollovers [105]. The data recorded prior to the free fall onset may already include a preparatory response. The fluoroscopic C-arm requires an operator to turn on the x-ray beam prior to the free fall onset. Once the x-ray beam is on, the C-arm emits a quiet high frequency humming noise. To reduce radiation exposure, the subject was exposed to free fall within the next 1-2 seconds. Thus, the subject may have begun anticipating the drop at the onset of the sound, rather than the onset of free fall. Also, the wait time before the free fall onset was not randomized, the subject was dropped as soon as reasonably possible (3-5 second wait time at the fixed height) to minimize 103  subject exposure to inversion. This consistent wait time may also contribute to a preparatory response before free fall onset. A preparatory response prior to free fall onset creates a false initial condition, which is subsequently used for data comparison. The neck has over 20 pairs of muscles, and the activity of only 8 neck muscles is reported in this work. The recorded muscle activity may suffer from several assumptions made when using indwelling EMG. Although ultrasound guidance was used to place the electrode tip near the middle of the desired muscle belly, it was not confirmed that the tip remained in place throughout the experiment. The recorded activity may only represent the activity at the wire tip. Thus, if the tip migrated, the measured activity may be recorded from a location superficial (and along the insertion track) to the desired muscle belly. Secondly, the EMG signals were normalized to activity levels from MVCs. However, it’s unclear whether subjects were able to exert themselves maximally since performing maximal exertions with one’s neck is unfamiliar and potentially not possible in a complex multiply redundant muscle system like the neck. Lastly, it’s possible that the discomfort and awareness of indwelling EMG electrodes in the subject’s neck may have influenced his responses.  As this was a pilot experiment, only one male human subject was tested. The results cannot be generalized and more subjects are needed to also represent both sexes and other age groups.    5.3 Potential applications Injury prevention strategies in vehicle rollovers may benefit from this work. Novel roof designs [124] and seat mounted airbags [125] have been proposed as methods of preventing cervical spine injury. The aim of these devices is to actively force an occupant’s neck into 104  flexion, thereby allowing the head and neck to escape the incoming loading of the torso. Since the performance of these devices relies on changing posture, knowledge of the initial posture of the neck is imperative. The initial cervical spine posture may influence the location of the major loading vector, alter the buckling mechanics of the spine [108], and influence the injury mechanism [5]. The level of muscle activity may also stiffen the vertebral column and change the forces that develop through the column during impact. Thus, designers of injury prevention devices may use this work to address the initial in vivo state of the neck, as its dynamic response is affected by cervical spine posture and muscle activity.  Hybrid III ATDs, designed for frontal impact scenarios, are typically used in vehicle rollover testing [58,126]. The Hybrid III neck is more stiff than a human’s neck [127], and thus is limited in its biofidelity. This work provides an in vivo data set of muscular responses which may be useful in developing a neck that better represents the neck of a live human. A more biofidelic neck would help prevent cervical spine injury by improving the assessment of potential neck injuries in vehicle rollover studies.   Some existing CSI models assume an upright-relaxed pre-impact posture (see section 1.3). However, this work has demonstrated that the upright and relaxed state of the neck does not represent the neck in an inverted, impending headfirst impact. These results provide an appropriate data set for improving and validating current and future cadaveric and computational cervical spine models, under physiologic conditions and in the context of preparing for inverted, headfirst impact.  Cadaveric CSI models may implement a more representative cervical spine posture, as spinal alignment is an influential parameter in axial impact injury mechanism [5,9,10]. Typically computational models simulate muscle activations based on results of muscle optimization strategies (i.e. minimize energy to hold head upright under the effects of gravity 105  [108]). This work provides EMG data of dynamic, active muscular response for 8 neck muscles. This active muscular response may be coupled with the observed cervical spine posture to simulate more realistic pre-impact postures and neck muscle loads. Simulating the in vivo state of the neck may expose some of the causes of the disparity between laboratory-induced and clinically reported injuries.  Although this work focuses on cervical spine injury in vehicle rollovers, it may be extended to other scenarios of axial impact injuries. In equestrian sports, cervical spine fracture is the 8th most common injury when falling from a horse [20]. In some cases, a rider may be ejected forward from the horse and free fall headfirst toward the ground [21]. In mountain biking, the distribution of serious injuries is overwhelmingly localized to the cervical spine. In clinically admitted cases, being propelled over the handlebars is the most common injury mechanism. Of those admitted to hospital for an over the handle bars injury, 91% suffered a direct impact to the head and resultant cervical spine injury [23]. In diving, cervical spine injuries occur in cases where the diver’s head strikes the bottom of the pool (or ground in natural settings) [26]. Although the dynamics of diving are more complicated due to hydrodynamic effects, the injury mechanism, in which the head strikes a stationary surface and the neck is loaded by the torso, is very similar. As all these scenarios involve axial impact injury to the cervical spine, this thesis may be relevant to their CSI models and injury prevention strategies.   5.4 Future work & recommendations This thesis work served as a pilot test to confirm the feasibility of the inversion apparatus and experimental protocol. Future work should include the recruitment of additional participants subject to exclusion criteria disclosed in the ethics application. Additional participants will 106  provide a more representative overview of the population and allow statistically significant trends to be observed.   Future work may also consider increasing the deceleration magnitude. Abiding by the most conservative limits, ASTM F2291 – 17 prescribes a deceleration limit of -2 g. Other human subjects in rollover testing have been exposed to -3 g for 20 seconds [118]. Given that the inversion apparatus has a fixed travel, increasing deceleration magnitude (used while stopping the subject’s free fall) would allow an increase in free fall duration (2 g would add approximately 90 ms and 3 g would add approximately 130 ms). This increase would address aforementioned limitations regarding the evolution of cervical spine posture. Increasing free fall duration would allow additional time for the postural muscle activity to evolve into muscle-induced changes of cervical spine posture. Since rollovers are relatively long duration events, the postural response (of both muscles and cervical spine realignment), which is likely relevant to neck injury.   Changes in the image analysis method for future human subject testing is recommended. Several steps in the image analysis method used in this thesis were largely manual, meaning accuracy is dependent on the analyzer. Due to poor image quality, certain manual steps such as tracing vertebral outlines introduced subjectivity. Variations may be large in both intra- and inter-analyzer results. This variability makes meaningful comparisons to published literature difficult. In future work, steps in which the analyzer subjectively selects points should be automated.  An automatic image rectification program which recognizes beads from the bead array and enforces known geometrical relationships would also be valuable. Firstly, it would eliminate variability associated with manually selecting the beads and, secondly, it would allow more frames from the free fall sequence to be analyzed. Previously only two frames from each 107  sequence were rectified; because the time required to manually rectify more than sixty frames was prohibitive. An automated rectification program would reduce the necessary rectification time, overcoming the labor-intensive deterrents. Including image stabilization subsequent to rectification would also be beneficial. The image stabilization program would stabilize the position of the bead array between frames. This would remove the need for an operator to manually superimpose beads from the bead array to allow inter-frame comparisons. Automatic rectification and image stabilization would allow a previously developed vertebral tracking algorithm to be used [16]. The vertebral tracking algorithm automatically identifies and tracks vertebral landmarks between frames, eliminating subjective judgements from the analyzer. Poor image quality may preclude the tracking algorithm from identifying vertebral landmarks. Thus, an investigation into methods for improving image quality should be performed. Lastly, scale objects should be included in the subject plane and image intensifier plane, allowing intervertebral translations to be dimensionalized into length units (i.e. transform from units of px/px to cm or mm).    Future extensions of this work should include an improved representation of pre-rollover and rollover dynamics. The most limiting feature of this work, in the context of rollovers, was the lack of pre-rollover and rollover dynamics. The effect of centripetal accelerations on in vivo muscle activity and cervical spine posture should be investigated. Two major data quality concerns should be considered. First, using indwelling EMG in dynamic scenarios can create motion artefact in the acquired signal. Second, lateral bending of the neck (due to centripetal accelerations) may interfere with sagittal plane fluoroscopy. Lateral bending would create out-of-plane motion (interfering with magnification scaling factors) and may cause the participant’s mandible to occlude the cervical spine. Bi-planar fluoroscopy may be used to create frame-by-108  frame 3D reconstructions of the cervical spine [81], overcoming these concerns associated with 2D fluoroscopy.   5.5 Conclusions An inversion apparatus was developed as a means for simulating the free fall phase of a rollover. Neck injury often occurs at the end of the free fall phase of a rollover when the head contacts the stationary vehicle roof and the neck is loaded by the incoming momentum of the torso. Thus, the apparatus simulates an environment relevant to cervical spine injury in vehicle rollovers (and other activities involving headfirst impact while upside down).  A pilot experiment was conducted in which a human subject was placed in the inversion apparatus and subjected to an inverted free fall drop. In general, the shape of the cervical spine was consistent with previous work in our lab, showing anterior and inferior translation of the vertebrae (relative to an upright-relaxed condition). Muscle activity increased substantially during free fall, exceeding levels recorded in upright-relaxed, quasi-static, configurations. The measured muscle activity also exceeded levels observed during previously quasi-static bracing tasks. These changes in neck muscle activity and cervical spine posture suggest the in vivo state of the neck differs from the state assumed in existing cervical spine injury models. This pilot experiment demonstrates that the inversion apparatus and experimental protocol are feasible for additional human subject testing.      109  Bibliography 1.  Yadollahi M, Paydar S, Ghaem H, Ghorbani M, Mousavi SM, Taheri Akerdi A, et al. Epidemiology of Cervical Spine Fractures. Trauma Mon [Internet]. 2016 Mar 16;21(3). Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5124335/ 2.  Foster JB, Kerrigan JR, Nightingale RW, Funk JR, Cormier JM, Bose D, et al. Analysis of cervical spine injuries and mechanisms for CIREN rollover crashes. In: Proceedings of the International Research Council on the Biomechanics of Injury conference [Internet]. 2012 [cited 2013 Sep 20]. p. 61–75. Available from: http://www.ircobi.org/downloads/irc12/pdf_files/15.pdf 3.  Nusholtz GS, Melvin JW, Huelke DF, Alem NM, Blank JG. Response of the Cervical Spine to Superior-Inferior Head Impact. In 1981 [cited 2016 May 28]. Available from: http://papers.sae.org/811005/ 4.  Nusholtz GS, Huelke DE, Lux P, Alem NM, Montalvo F. Cervical Spine Injury Mechanisms. In 1983 [cited 2016 May 28]. Available from: http://papers.sae.org/831616/ 5.  Maiman DJ, Yoganandan N, Pintar FA. Preinjury cervical alignment affecting spinal trauma. J Neurosurg Spine. 2002;97(1):57–62.  6.  Hu J, Yang KH, Chou CC, King AI. A Numerical Investigation of Factors Affecting Cervical Spine Injuries During Rollover Crashes: Spine. 2008 Nov;33(23):2529–35.  7.  Nightingale RW, Sganga J, Cutcliffe H, Bass CR “Dale.” Impact responses of the cervical spine: A computational study of the effects of muscle activity, torso constraint, and pre-flexion. J Biomech. 2016 Feb 29;49(4):558–64.  8.  Yoganandan N, Sances AJ, Maiman DJ, Myklebust JB, Pech P, Larson SJ. Experimental Spinal Injuries with Vertical Impact. Spine. 1986 Nov;11(9):855–60.  9.  Pintar FA, Yoganandan N, Voo L, Cusick JF, Maiman DJ, Sances A. Dynamic Characteristics of the Human Cervical Spine. In 1995 [cited 2016 Nov 27]. Available from: http://papers.sae.org/952722/ 10.  Yoganandan N., Pintar F., Gennarelli T., Eppinger R., Voo L. Geometric effects on the mechanism of cervical spine injury due to head impact. In 1999.  11.  Maiman DJ, Sances A, Myklebust JB, Larson SJ, Houterman C, Chilbert M, et al. Compression injuries of the cervical spine: A biomechanical analysis. Neurosurgery. 1983;13(3):254–60.  12.  Halldin P, Hedenstierna S, Brolin K, Holst H von. Finite Element Analysis of the Effects of Head-Supported Mass on Neck Responses [Internet]. ROYAL INST OF TECH 110  STOCKHOLM (SWEDEN), ROYAL INST OF TECH STOCKHOLM (SWEDEN); 2006 Sep [cited 2016 May 28]. Available from: http://www.dtic.mil/docs/citations/ADA456789 13.  Brolin K, Hedenstierna S, Halldin P, Bass C, Alem N. The importance of muscle tension on the outcome of impacts with a major vertical component. Int J Crashworthiness. 2008 Sep 11;13(5):487–98.  14.  Crandall JR, Bose D, Forman J, Untaroiu CD, Arregui-Dalmases C, Shaw CG, et al. Human surrogates for injury biomechanics research. Clin Anat N Y N. 2011 Apr;24(3):362–71.  15.  Newell RS, Blouin J-S, Street J, Cripton PA, Siegmund GP. Neck posture and muscle activity are different when upside down: A human volunteer study. J Biomech. 2013 Nov 15;46(16):2837–43.  16.  Newell, Robyn. Are inversion, posture, motion and muscle effects important to spinal alignment? [Internet]. University of British Columbia; 2014 [cited 2016 May 26]. Available from: http://dx.doi.org/10.14288/1.0167424 17.  Newell RS, Siegmund GP, Blouin J-S, Street J, Cripton PA. Cervical Vertebral Realignment When Voluntarily Adopting a Protective Neck Posture.  [Miscellaneous Article]. Spine. 2014 Jul;39(15).  18.  Sorli JM. Equestrian injuries: a five year review of hospital admissions in British Columbia, Canada. Inj Prev. 2000 Mar 1;6(1):59–61.  19.  Ball CG, Ball JE, Kirkpatrick AW, Mulloy RH. Equestrian injuries: incidence, injury patterns, and risk factors for 10 years of major traumatic injuries. Am J Surg. 2007 May 1;193(5):636–40.  20.  Weber CD, Nguyen AR, Lefering R, Hofman M, Hildebrand F, Pape H-C. Blunt injuries related to equestrian sports: results from an international prospective trauma database analysis. Int Orthop. 2017 Oct 1;41(10):2105–12.  21.  Bourdet N, Willinger R. Head impact conditions in case of equestrian accident. IRCOBI Biomech Impacts Conf. 2015;  22.  Riding Accident Paralyzes Actor Christopher Reeve (washingtonpost.com) [Internet]. [cited 2018 Jul 20]. Available from: http://www.washingtonpost.com/wp-dyn/articles/A99660-1995Jun1.html 23.  Dodwell ER, Kwon BK, Hughes B, Koo D, Townson A, Aludino A, et al. Spinal Column and Spinal Cord Injuries in Mountain Bikers: A 13-Year Review. Am J Sports Med. 2010 May;38(8):1647–52.  111  24.  Statistical abstract of the united states-participation in selected sports activities [Internet]. US Census Bureau; 1991 [cited 2012 Jan 24]. Available from: http://www.census.gov/compendia/statab/past_years.html 25.  Pollard KA, Gottesman BL, Rochette LM, Smith GA. Swimming injuries treated in US EDs: 1990 to 2008. Am J Emerg Med. 2013 May 1;31(5):803–9.  26.  Mcelhaney J, Snyder RG, States JD, Gabrielsen MA. Biomechanical Analysis of Swimming Pool Neck Injuries. In 1979 [cited 2016 Jun 21]. Available from: http://papers.sae.org/790137/ 27.  Safercar.gov [Internet]. [cited 2016 Dec 1]. Available from: http://www.safercar.gov/Vehicle-Shoppers/Rollover/Fatalities 28.  Yoganandan N, Haffner M, Maiman DJ, Nichols H, Pintar FA, Jentzen J, et al. Epidemiology and Injury Biomechanics of Motor Vehicle Related Trauma to the Human Spine. In 1989 [cited 2016 Oct 29]. Available from: http://papers.sae.org/892438/ 29.  Moffatt EA, Cooper ER, Croteau JJ, Orlowski KF, Marth DR, Carter JW. Matched-Pair Rollover Impacts of Rollcaged and Production Roof Cars Using the Controlled Rollover Impact System (CRIS). Soc Automot Eng. 2003 Mar;2003-1–172.  30.  James Raddin, Joseph Cormier, Brian Smyth, Jeffery Croteau, Eddie Cooper. Compressive Neck Injury and its Relationship to Head Contact and Torso Motion during Vehicle Rollovers. SAE Int J Passeng Cars-Mech Syst. 2009 Apr 9;2(1):914–41.  31.  Bahling GS., Bundorf RT., Moffatt EA., Orlowski KF. The influence of increased roof strength on belted and unbelted dummies in rollover and drop tests. J Trauma. 1995 Apr;38(4):557–63.  32.  KAZARIAN L, GRAVES GA. Compressive Strength Characteristics of the Human Vertebral Centrum. Spine. 1977;2(1):1–14.  33.  Yoganandan N, Pintar FA, Butler J, Reinartz J, Sances AJ, Larson S. Dynamic response of human cervical spine ligaments. Spine. 1989;14(10):1102–10.  34.  Alem NM, Nusholtz GS, Melvin JW. Head and Neck Response to Axial Impacts. In 1984 [cited 2016 May 28]. Available from: http://papers.sae.org/841667/ 35.  Myers BS., Winkelstein BA. Epidemiology, classification, mechanism, and tolerance of human cervical spine injuries. Crit Rev Biomed Eng. 1995;23:307–409.  36.  Harris JH Jr, Mirvis SE. The radiology of acute cervical spine trauma [Internet]. 3rd ed. Baltimore: Williams & Wilkins; 1996.  37.  Nightingale RW, Doherty BJ, Myers BS, McElhaney JH, Richardson WJ. The influence of end condition on human cervical spine injury mechanisms. SAE Technical Paper; 1991.  112  38.  National Highway Traffic Safety Administration. The National Highway Traffic Safety Administration’s Rating System for Rollover Resistance: An Assessment. Washington: Transportation Research Board; 2002.  39.  NHTSA. Types of Rollovers [Internet]. safercar.gov. [cited 2018 Jul 17]. Available from: https://www.safercar.gov/Vehicle-Shoppers/Rollover/Types-of-Rollovers 40.  Viano DC, Parenteau CS. Analysis of Head Impacts Causing Neck Compression Injury. Traffic Inj Prev. 2008 Apr 4;9(2):144–52.  41.  Herbst BF. Fidelity of anthropometric test dummy necks in rollover accidents. Int Tech Conf Enhanc Saf Veh. 1998;1998:2093–7.  42.  Ridella S., Eigen AM. Biomechanical investigation of injury mechanisms in rollover crashes from the CIREN database. In 2008.  43.  Bambach MR, Grzebieta RH, McIntosh AS, Mattos GA. Cervical and thoracic spine injury from interactions with vehicle roofs in pure rollover crashes. Accid Anal Prev. 2013 Jan;50:34–43.  44.  Pintar F., Voo LM., Yoganandan N., Cho TH., Maiman DJ. Mechanisms of hyperflexion cervical spine injury. In 1998.  45.  Nightingale RW, McElhaney JH, Richardson WJ, Myers BS. Dynamic responses of the head and cervical spine to axial impact loading. J Biomech. 1996 Mar 1;29(3):307–18.  46.  Foust DR, Chaffin DB, Snyder RG, Baum JK. Cervical range of motion and dynamic response and strength of cervical muscles. SAE Tech Pap 730975. 1973;285–308.  47.  Gopal M., Baron K., Shah M. Simulation and testing of a suite of field relevant rollovers. SAE Tech Pap No 2004-01-0335. 2004;  48.  Panjabi MM, Cholewicki J, Nibu K, Grauer J, Babat LB, Dvorak J. Critical load of the human cervical spine: an in vitro experimental study. Clin Biomech. 1998 Jan;13(1):11–7.  49.  Kettler A, Hartwig E, Schultheiss M, Claes L, Wilke HJ. Mechanically simulated muscle forces strongly stabilize intact and injured upper cervical spine specimens. J Biomech. 2002 Mar;35(3):339–46.  50.  Nightingale RW, Camacho DL, Armstrong AJ, Robinette JJ, Myers BS. Inertial properties and loading rates affect buckling modes and injury mechanisms in the cervical spine. J Biomech. 2000 Feb;33(2):191–7.  51.  Chancey VC., Nightingale RW., Van Ee CA., Knaub KE., Myers BS. Improved estimation of human neck tensile tolerance: Reducing the range of reported tolerance using anthropometrically correct muscles and optimized physiologic initial conditions. Stapp Car Crash J. 2003;47:135–53.  113  52.  Dibb AT, Cox CA, Nightingale RW, Luck JF, Cutcliffe HC, Myers BS, et al. Importance of muscle activations for biofidelic pediatric neck response in computational models. Traffic Inj Prev. 2013;14 Suppl:S116-27.  53.  Nightingale RW, McElhaney JH, Camacho DL, Kleinberger M, Winkelstein BA, Myers BS. The Dynamic Responses of the Cervical Spine: Buckling, End Conditions, and Tolerance in Compressive Impacts. In 1997 [cited 2016 Jun 21]. Available from: http://papers.sae.org/973344/ 54.  Crandall J. Preservation of Human Surrogates for Impact Studies [PhD Dissertation]. University of Virginia; 1994.  55.  Carter JW, Ku GS, Nuckley DJ, Ching RP. Tolerance of the cervical spine to eccentric axial compression. Stapp Car Crash J. 2002 Nov;46:441–59.  56.  Pintar FA., Yoganandan N., Pesigan M., Reinartz J., Sances A Jr ., Cusick JF. Cervical vertebral strain measurements under axial and eccentric loading. J Biomech Eng. 1995 Nov;117(4):474–8.  57.  Lessley DJ, Riley P, Zhang Q, Foltz P, Overby B, Heltzel S, et al. Occupant Kinematics in Laboratory Rollover Tests: PMHS Response. Stapp Car Crash J Ann Arbor. 2014 Nov;58:251–316.  58.  Moffatt EA, Cooper ER, Croteau JJ, Parenteau C, Toglia A. Head Excursion of Seat Belted Cadaver, Volunteers and Hybrid III ATD in a Dynamic/Static Rollover Fixture. In 1997. Available from: http://papers.sae.org/973347/ 59.  Begeman PC, King AI, Levine RS, Viano DC. Biodynamic Response of the Musculoskeletal System to Impact Acceleration. SAE Trans. 1980;89:3992–4007.  60.  Vasavada AN, Danaraj J, Siegmund GP. Head and neck anthropometry, vertebral geometry and neck strength in height-matched men and women. J Biomech. 2008;41(1):114–21.  61.  Oi N, Pandy MG, Myers BS, Nightingale RW, Chancey VC. Variation of neck muscle strength along the human cervical spine. Stapp Car Crash J. 2004 Nov;48:397–417.  62.  Newell RS, Siegmund GP, Blouin J-S, Street J, Cripton PA. Cervical Vertebral Realignment When Voluntarily Adopting a Protective Neck Posture.  [Miscellaneous Article]. Spine. 2014 Jul;39(15).  63.  Moffatt E, Hare B, Hughes R, Lewis L, Iiyama H, Curzon A, et al. Head excursion of restrained human volunteers and hybrid III dummies in steady state rollover tests. Annu Proc Assoc Adv Automot Med. 2003;47:445–65.  64.  Orlowski KF, Bundorf RT, Moffatt EA. Rollover crash tests—the influence of roof strength on injury mechanics. Soc Automot Eng. 1985;851734.  114  65.  Seeman MR., Muzzy III WH., Lustick LS. Comparison of human and Hybrid III head and neck dynamic response. SAE Tech Pap No 861892. 1986;  66.  Adamec J., Praxl N., Miehling T., Muggenthaler H., Schonpflug M. The occupant kinematics in the first phase of a rollover accident - Experiment and simulation. In 2005. p. 145–56.  67.  Zhang Q, Lessley DL, Riley P, Toczyski J, Lockerby J, Foltz P, et al. Occupant Kinematics in Laboratory Rollover Tests: ATD Response and Biofidelity. Stapp Car Crash J. 2014;58(101133951):317–60.  68.  Arndt MW, Mowry GA, Dickerson CP, Arndt SM. Evaluation of experimental restraints in rollover conditions. In: 39th Stapp Car Crash Conference. San Diego, California, USA: Society of Automotive Engineers, Inc., Warrendale, Pennsylvania, USA; 1995. p. 101–10.  69.  Arndt MW, Mowry GA, Baray PE, Clark DA. The Development of a Method for Determining Effective Slack in Motor Vehicle Restraint Systems for Rollover Protection. In 1997. Available from: http://papers.sae.org/970781/ 70.  Bahling GS, Bundorf RT, Kaspzyk GS, Moffatt EA, Orlowski KF, Stocke JE. Rollover and drop tests - the influence of roof strength on injury mechanics using belted dummies. Soc Automot Eng. 1990;902314.  71.  James MB, Allsop DL, Nordhagen RP, Decker RL. Injury mechanisms and field accident data analysis in rollover accidents. SAE; 1997.  72.  herbst  brian, forrest  stephen, wang  phillip, chng  david, friedman  donald, friedman  keith. the ability of 3 point safety belts to restrain occupants in rollover crashes. IRCOBI Conf Proc. 1996;  73.  Friedman D., Chng D. Human subject experiments in occupant response to rollover with reduced headroom. SAE Tech Pap No 980212 [Internet]. 1998; Available from: http://dx.doi.org/10.4271/980212 74.  Muggenthaler H., Adamec J., Praxl N., Schonpflug M. The influence of muscle activity on occupant kinematics. In 2005. p. 313–23.  75.  Yamaguchi G. Electromyographic Activity and Posturing of the Human Neck During Rollover Tests. J Biomech. 2005;  76.  Ono K, Kaneoka K, Wittek A, Kajzer J. Cervical Injury Mechanism Based on the Analysis of Human Cervical Vertebral Motion and Head-Neck-Torso Kinematics During Low Speed Rear Impacts. In 1997 [cited 2016 Nov 28]. Available from: http://papers.sae.org/973340/ 77.  Johnson GM. The correlation between surface measurement of head and neck posture and the anatomic position of the upper cervical vertebrae. Spine. 1998 Apr 15;23(8):921–7.  115  78.  Wu SK, Lan HH, Kuo LC, Tsai SW, Chen CL, Su FC. The feasibility of a video-based motion analysis system in measuring the segmental movements between upper and lower cervical spine. Gait Posture. 2007 Jun;26(1):161–6.  79.  Descarreaux M, Blouin JS, Teasdale N. A non-invasive technique for measurement of cervical vertebral angle: report of a preliminary study. Eur Spine J. 2003 Jun;12(3):314–9.  80.  OEC 9400 C-arm Service Manual. OEC-Diasonics; 1992.  81.  Brainerd EL, Baier DB, Gatesy SM, Hedrick TL, Metzger KA, Gilbert SL, et al. X-ray reconstruction of moving morphology (XROMM): precision, accuracy and applications in comparative biomechanics research. J Exp Zool Part Ecol Genet Physiol. 2010 Jun 1;313A(5):262–79.  82.  Bidez MW, Cochran JE, King D, Burke DS. Occupant Dynamics in Rollover Crashes: Influence of Roof Deformation and Seat Belt Performance on Probable Spinal Column Injury. Ann Biomed Eng. 2007 Jul 20;35(11):1973–88.  83.  Fréchède B, McIntosh A, Grzebieta R, Bambach M. Hybrid III ATD in Inverted Impacts: Influence of Impact Angle on Neck Injury Risk Assessment. Ann Biomed Eng. 2009 May 8;37(7):1403–14.  84.  Myers BS, Nightingale RW. 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. J Crash Prev Inj Control. 1999;1(1):67–82.  85.  Roberts C, Kerrigan JR. Injuries and Kinematics: Response of the Cervical Spine in Inverted Impacts. In 2015.  86.  Bisdorff AR, Bronstein AM, Gresty MA, Wolsley CJ, Davies A, Young A. EMG-responses to sudden onset free fall. Acta Oto-Laryngol Suppl. 1995;520 Pt 2:347–9.  87.  Bisdorff AR, Bronstein AM, Gresty MA. Responses in neck and facial muscles to sudden free fall and a startling auditory stimulus. Electroencephalogr Clin Neurophysiol. 1994 Dec;93(6):409–16.  88.  Aoki M, Matsunami K ’ichi, Han X y, Yamada H, Muto T, Ito Y. Neck muscle responses to abrupt vertical acceleration in the seated human. Exp Brain Res. 140(1):20–4.  89.  Ito Y, Corna S, von Brevern M, Bronstein A, Rothwell J, Gresty M. Neck muscle responses to abrupt free fall of the head: comparison of normal with labyrinthine-defective human subjects. J Physiol. 1995 Dec 15;489(Pt 3):911–6.  90.  Nijhuis LBO, Janssen L, Bloem BR, van Dijk JG, Gielen SC, Borm GF, et al. Choice reaction times for human head rotations are shortened by startling acoustic stimuli, irrespective of stimulus direction. J Physiol. 2007 Oct 1;584(Pt 1):97–109.  116  91.  Flight Surgeon’s Manual. Naval Airspace Medical Institute; 1991. Report No.: Third Edition.  92.  Pandolf KB, Burr RE, Office of the Surgeon General, Dept. of the Army, United States. Medical aspects of harsh environments [Internet]. Falls Church, Va;Bethesda, Md;Fort Sam Houston, Tex;Washington, D.C; Office of the Surgeon General, United States Army; 2001.  93.  Christy RL. Effects of radial, angular, and transverse acceleration. Aerosp Med. 1971;(2nd ed.):187.  94.  Ontario Government. Canadian Bungee Code of Safe Practice [Internet]. 2000. Available from: https://www.tssa.org/corplibrary/ArticleFile.asp?Instance=136&ID=B5BB95176D3243FE94FF1809BBC596DA 95.  Cords and Equipment [Internet]. Bungee Consultants International. [cited 2017 Jan 10]. Available from: http://www.bungeeconsultants.com/new-page/ 96.  Lawrence M. Personal Communication. 2016.  97.  F2783 Standard Practice for Design, Manufacture, Operation, Maintenance, and Inspection of Amusement Rides and Devices, in Canada. 2014.  98.  Wang J, Blackburn TJ. The AAPM/RSNA physics tutorial for residents: X-ray image intensifiers for fluoroscopy. Radiographics. 2000;20(5):1471–7.  99.  WCB Standards [Internet]. WorkSafe BC; 2004 [cited 2018 Jun 26]. Available from: https://www.worksafebc.com/en/law-policy/occupational-health-safety/searchable-ohs-regulation/wcb-standards/wcb-standards#SectionNumber:StandardWPL3 100.  Conroy C, Hoyt DB, Eastman AB, Erwin S, Pacyna S, Holbrook TL, et al. Rollover crashes: Predicting serious injury based on occupant, vehicle, and crash characteristics. Accid Anal Prev. 2006 Sep;38(5):835–42.  101.  Friedman D, Chng D. Human Subject Experiments In Occupant Response To Rollover With Reduced Headroom. In 1998 [cited 2018 Mar 19]. Available from: http://papers.sae.org/980212/ 102.  Friedman K., Friedman D. Upper interior head, face and neck injury experiments. In 1998.  103.  ASTM F2137 - 16 Standard Practice for Measuring the Dynamic Characteristics of Amusement Rides and Devices. American Society for Testing and Materials;  104.  ASTM F2291 - 17 Standard Practice for Design of Amusement Rides and Devices. American Society for Testing and Materials;  105.  Siegmund G. Free-fall perception. 2016.  117  106.  Peterson BW, Richmond FJ. Control of head movement. New York: Oxford University Press; 1988.  107.  Raddin J, Cormier J, Smyth B, Croteau J, Cooper E. Compressive neck injury and its relationship to head contact and torso motion during vehicle rollovers. SAE Tech Pap No 2009-01-0829 [Internet]. 2009; Available from: http://dx.doi.org/10.4271/2009-01-0829 108.  Nightingale RW, Sganga J, Cutcliffe H, Bass CR “Dale.” Impact responses of the cervical spine: A computational study of the effects of muscle activity, torso constraint, and pre-flexion. J Biomech. 2016 Feb;49(4):558–64.  109.  Blouin J-S, Siegmund GP, Carpenter MG, Inglis JT. Neural Control of Superficial and Deep Neck Muscles in Humans. J Neurophysiol. 2007 Aug 1;98(2):920–8.  110.  Hodges PW, Bui BH. A comparison of computer-based methods for the determination of onset of muscle contraction using electromyography. Electroencephalogr Clin Neurophysiol Mot Control. 1996;101(6):511–9.  111.  Gandevia SC. Spinal and supraspinal factors in human muscle fatigue. Physiol Rev. 2001 Oct;81(4):1725–89.  112.  Newell RS, Siegmund GP, Blouin J-S, Street J, Cripton PA. Cervical Vertebral Realignment When Voluntarily Adopting a Protective Neck Posture: Spine. 2014 May;1.  113.  Newell RS, Blouin J-S, Street J, Cripton PA, Siegmund GP. Neck posture and muscle activity are different when upside down: A human volunteer study. J Biomech. 2013 Nov 15;46(16):2837–43.  114.  Newberry WN, Imler S, Carhart M, Dibb A, Balavich K, Croteau J, et al. Belted Occupant Kinematics and Head Excursion During the Airborne Phase of Vehicle Rollover: Evaluation of the Effects of Rollover-Deployed Curtain Airbags. In 2014 [cited 2016 Aug 2]. Available from: http://papers.sae.org/2014-01-0527/ 115.  Ashby BM, Lai W, Carhart MR, Newberry W, Weaver B, Corrigan CF. Compressive Neck Preloading During the Airborne Phase of Vehicle Rollover. In SAE International; 2007. Available from: http://dx.doi.org/10.4271/2007-01-0377 116.  J.D. States, Benedict JF. Safe and unsafe upper torso restraints for occupant protection in motor vehicles. Proc 7th Stapp Car Crash Conf. 1963;312–23.  117.  Tourin B, Aldman B, Michelson I, Mitchell J. Restraining characteristics of harnesses. Proc 6th Stapp Car Crash Conf. 1962;138–49.  118.  Moffatt E, Hare B, Hughes R, Lewis L, Iiyama H, Curzon A, et al. Head Excursion of Restrained Human Volunteers and Hybrid III Dummies in Steady State Rollover Tests. Annu Proc Assoc Adv Automot Med. 2003;47:445–65.  118  119.  PluPekoInside2. Dual dash cam rollover [Internet]. [cited 2017 Jan 10]. Available from: https://www.youtube.com/watch?v=Jb1ZOS_nNYM 120.  Pete Naz. Argon Racing- Pacific Forest Rally 2014 high speed roll [Internet]. Available from: https://www.youtube.com/watch?v=DJaMaaTzBg0 121.  feeboi. car roll over [Internet]. Available from: https://www.youtube.com/watch?v=2IpCIYF0lVU 122.  James MB, Nordhagen RP, Schneider DC, Koh S-W. Occupant Injury in Rollover Crashes: A Reexamination of Malibu II [Internet]. Warrendale, PA: SAE International; 2007 Apr [cited 2013 Jul 7]. Report No.: 2007-1–369. Available from: http://digitallibrary.sae.org.ezproxy.library.ubc.ca/searchresults/ 123.  Zach Argrave. 700bhp skyline crashed and rolled [Internet]. Available from: https://www.youtube.com/watch?v=9jQ5QcH1WC8 124.  Ejima S., Ono K., Kaneoka K., Fukushima M. Development and validation of the human neck muscle model under impact loading. In 2005. p. 245–55.  125.  Heudorfer B., Breuninger M., Karlbauer U., Kraft M., Maidel J. Roofbag - A concept study to provide enhanced protection for head and neck in case of rollover. In 2005.  126.  Bahling GS, Bundorf RT, Kaspzyk GS, Moffatt EA, Orlowski KF, Stocke JE. Rollover and Drop Tests - The Influence of Roof Strength on Injury Mechanics Using Belted Dummies. In 1990 [cited 2018 Apr 2]. Available from: http://papers.sae.org/902314/ 127.  Yoganandan N., Sances A Jr ., Pintar F. Biomechanical evaluation of the axial compressive responses of the human cadaveric and manikin necks. J Biomech Eng. 1989 Aug;111(3):250–5.  128.  Jaggi B. Lecture 7: Risk Management. 2015; The University of British Columbia.  129.  Bruno J. Question re: device risk management. 2018.  119  Appendices Appendix A  Code used to generate the linear motor’s position profile This appendix contains the commented Matlab code used to generate the position, velocity, and acceleration profiles of the linear motor. These profiles were calculated based on the idealized acceleration profile of the subject carriage. The position and velocity profiles of the subject carriage are also calculated in this code. A.1 Matlab code used to generate the linear  motor’s position profile %AUTHOR: M’BETH SCHOENFELD %CALCULATE SLED PROFILE FOR THE DROP %IDEAL SOLUTION   clc close all clear all   % sled parameters TM = 1024; %TM 1000 DT = 2;  dt = (1/TM)*(2^DT); %[s] time step   % Rig parameters c = 1.71; %[m] link length piston_stroke = 500/1000; %[m] piston stroke length theta_max = 75.0*pi/180; % [rad] initial angle (at top of drop height) theta_min = 22.4*pi/180; %[rad] final angle (bottom of drop height) theta = theta_max; %[rad] initial angle for calculations   %% Calculate throw max_h = c*sin(theta_max); %[m] max subject height min_h = c*sin(theta_min); %[m] min subject height throw = max_h-min_h; %[m] total throw available for subject %throw = 0.95; %[m] 05.07.2017 measurment from limit switch stops %complex geometry--> start decel 2.5 cm early so that travel < throw drop_dist = (1/2)*throw-2.5/100; %[m] want to free fall for 2.5cm short of half the total throw decel_dist = throw-drop_dist;   %% Create arrays a = []; %[m] subject position b = []; %[m] sled position  da = []; %[m/s] subject velocity db = []; %[m/s] sled velocity d2a = []; %[m/s/s] subject acceleration d2b = []; %[m/s/s] sled acceleration   120  % populate initial conditions % a refers to the subject % b refers to the sled a(1) = c*sin(theta); %[m] b(1) = c*cos(theta); %[m] da(1) = 0; %[m/s] db(1) = 0; %[m/s] d2a(1) = -9.81; %[m/s/s] % d2b(1) = (-a(1)/b(1))*((-b(1)/a(1))^2)*(d2a(1)); %[m/s/s] d2b(1) = (-a(1)*d2a(1)-da(1)^2-db(1)^2)/b(1);   %% Loop i = 2; %index to access array elements. Start @ 2 bc the first elements have already been initialized travel = 0; %[m] cumulative distance travelled throughout the fall   %while abs(travel) < throw %run until the subject has moved through the entire throw distance while da(i-1) <= 0 %run until the subject has come to a rest     if abs(travel) < drop_dist          d2a(i)= -9.81; %[m/s/s] free fall through the drop distance     else         d2a(i) = 9.81; %[m/s/s] start deceleration once subject has fallen through the drop distance     end          delta_a = da(i-1)*dt + (1/2)*d2a(i-1)*dt^2; % [m] distance through which subject drops per time step     travel = travel + delta_a; %[m] cummulative distance subject has dropped     a(i) = a(i-1) + delta_a; %[m] current subject height     b(i) = sqrt(c^2 - a(i)^2); %[m] current sled position     da(i) = da(i-1) + d2a(i-1)*dt; %[m/s] current subject velocity %     db(i) = -a(i)*da(i)/b(i); %[m/s] current sled velocity %     d2b(i) = -a(i)*(da(i)^2)*d2a(i)/(b(i)*(b(i)^2)); %[m/s/s] current sled acceleration     db(i)=-a(i)*da(i)/b(i);     d2b(i)=(-a(i)*d2a(i)-da(i)^2-db(i)^2)/b(i);      i=i+1; end   %% PLOT  plot(a); hold on plot(b); hold on title('position') legend('sub','sled')   figure() plot(da); hold on plot(db); hold on title('velocity') legend('sub','sled')   figure() plot((b-b(1))*1000,d2a); hold on plot((b-b(1))*1000,-d2b); hold on 121  title('acceleration') legend('sub','sled') xlabel('sled position [m]')   figure() title('subj. velocity vs. position'); hold on plot(abs(a-a(1)),abs(da)); hold on xlabel('subj. position [m]') ylabel('subj. vel. [m/s]')   figure() title('sled velocity vs. position'); hold on plot(abs(b-b(1)),abs(db)); hold on xlabel('sled position [m]') ylabel('sled vel. [m/s]')   %% Check profiles vel_check=diff_5pt(b, dt); accel_check=diff_5pt(db, dt); vel_check_2=diff(b)/dt; accel_check_2=diff(db)/dt;   figure() title('vel_check'); hold on plot(vel_check); hold on plot(db); hold on plot(vel_check_2); hold on legend('5pt stencil','calc','matlab diff')   figure() title('accel check'); hold on plot(accel_check); hold on plot(d2b); hold on plot(accel_check_2) legend('5pt stencil','calc','matlab diff')   figure() plot((b-b(1))*1000, d2b*1000) xlabel('Sled Position [mm]') ylabel('Sled Acceleration [mm/s/s]') title('sled acceleration as f(position)')   %% Store motion profiles a_da_d2a_b_db_d2b=[-a' da' d2a' b' db' d2b'];   %% Generate DMC Commands galil=b-b(1); galil=galil*(-1); %the sled needs to complete the drop in the -x direction (galil coordinate system) generateDMC( galil );  122  A.2 Matlab code used to convert position increments to DMC commands This section of the appendix contains a the function generateDMC() which is called in the last line of the code above. This section of code takes the position increments and converts them to a format compatible with the linear motor’s control software, Galil DMC.  function generateDMC( sled_pos ) %This function takes in a position array [in m] and generates the DMC %commands in a text file to excecute the sled profile.  %Make sure the time-base is 1/512 s   fileID = fopen('sledprofile_june13.txt','wt'); [R,C]=size(sled_pos);   if C>R %convert to column vector if necessary     sled_pos=sled_pos';     [R,C]=size(sled_pos); end   sled_pos=sled_pos*1000; %convert from [m] to [mm] Ppmm=500; %500 quadrature counts per mm counts=sled_pos*Ppmm; %sled position in quadrature counts   pos_index=0:1:(R-1);   pos_index=pos_index';   for i=1:length(pos_index)     %fprintf(fileID,'Pos[%i]=@INT[%.0f]\n', pos_index(i),counts(i));     %%prints to text file [COUNTS]     %fprintf('Pos[%i]=@INT[%.0f]\n', pos_index(i),counts(i)) %prints to     %command window [COUNTS]     %fprintf('%.0f\n', counts(i))%testline          fprintf(fileID,'Pos[%i]=@INT[Ppmm*%.4f]\n', pos_index(i),sled_pos(i)); %prints to text file [MILLIMETERS]     fprintf('Pos[%i]=@INT[Ppmm*%.4f]\n', pos_index(i),sled_pos(i)) %prints to command window [MILLIMETERS] end sled_pos fclose(fileID);   end 123  Appendix B  Linear motor control program flowchart and DMC code This appendix includes a program flowchart of the program used to operate the feedback-controlled linear motor. Also included is the program’s DMC code, which was run on Galil DMC Smart-Terminal. B.1 Program Flowchart 124  B.2 DMC Code #ROLLOVER NO Confirm timebase is 1024 microseconds TM 1000 NO Set position error limit ER 1000 KP 12 KD 15 KI 0.4 NO Disable Off-ON-Error(ServoHere) OE 0 NO Set interrupts for pin 1 II 1,2,,1 CB 1 NO DP 0 NO Correct any typos in the PID controls. IF (_KPx<>12) KP 12 MG "KP reset." MG "KP: ",_KPx ENDIF IF (_KDx<>15) KD 15 MG "KD reset." MG "KD: ". _KDx  ENDIF IF (_KIx<>0.4) KI 0.4 MG "KI reset, KI: ",_KIx ENDIF NO Test the assumption that sled pos starts at 0 IF (_TPx<>0) MG " " MG "Out of position by [counts]: ",_TPx MG " "  MG "Exiting Program. Reset and Re-download." JP #END ENDIF NO Check status of Inversion Switch IF (@IN[4]=1) MG " " MG "ERROR: Subject NOT inverted!" MG " " MG "Exiting Program... " MG " " JP #END ENDIF  NO Check status of piston L-bracket switches IF (@IN[3]=0) MG " " MG "ERROR: L-Brackets are NOT removed!" MG " " MG "Exiting Program... " MG " " 125  JP #END ENDIF NO Initialize Constant Variables NO My variables MaxPos=565428 MinPos=0 Stop=1 Ppmm=500 Shape=1 Pos=0 JP #MAIN NO ------------------------------------------------- NO Define sled behavior when LIMIT SWITCH IS TRIPPED NO Decelerate at 62.5 m/s/s to stop in 20cm #LIMSWI DT 0 CD 0 AB 1 SH MG " " IF (_LFX=0) MG "**************************************" MG " ERROR! UPPER LIMIT SWITCH ACTIVATED. " MG "**************************************" ELSE MG "**************************************" MG " ERROR! LOWER LIMIT SWITCH ACTIVATED. " MG "**************************************" ENDIF MG " " RE 1 NO ---------------------------------------------- NO Define sled`X behaviour on INTERRUPT ROUTINE NO Pause & hold the sled motion wherever its at #ININT NO ABort without leaving program NO Disable contour mode if active DT 0 CD 0 AB 1 SH IF (@IN[1]=1) MG " " MG "************************************** " MG " ERROR! OPERATOR INTERRUPTED MOTION." MG " " NO WANT PROGRAM TO RETURN TO MAIN OPTIONS MENU MG " Release E-STOP to continue..." MG "************************************** " MG " " #CLEAR JP #CLEAR, @IN[1] = 1 ENDIF IF (@IN[2]=0) NO THE MOTOR WOULD TURN OFF 126  MG "************************************** " MG " ERROR! SERVO AMP INTERRUPTED MOTION." MG " " MG " Waiting for SERVO to be cleared..." MG "************************************** " #CLEAR1 JP #CLEAR1, @IN[2]=0 ENDIF MG " " MG " E-STOP released... Continue selection > " RI 0 NO ------------------------------------------------- NO Define sled behavior on POSITION ERROR #POSERR MG "********************************************* " MG " ERROR! POSITION ERROR [COUNTS] = ", _TEX MG "********************************************* " DT 0 CD 0 AB 1 SH RE  NO ------------------------------------------------- NO ------------------------------------------------- NO ---------- * * * MAIN PROGRAM * * * ------------- NO ------------------------------------------------- NO ------------------------------------------------- #MAIN NO Declare arrays DM Pos[164] DM Diff[163] NO ------------------------------------------------- NO Initialize variables subject to change NO My variables  NO MaxPos=565428 NO MinPos=0 NO Stop=1 NO Ppmm=500 NO Shape=1 NO Pos=0 Speed=0 Accel=0 Decel=0 Dist=0 Trav=0 mTrav=0 J=0 N=163 NO Initialize arrays J=0 #INIT1 Diff[J]=0  J=J+1 JP #INIT1, J<163 J=0 127  #INIT2 Pos[J]=0 J=J+1 JP #INIT2, J<164 NO ---------------------------------------------- NO Get user input for sled motion SH MG " " IN "ACTION (CNL,SSR,INCR,SSL,INCL,DROP)> ",Shape{S4} IF ((Shape<>"CNL") & (Shape<>"SSR") & (Shape<>"INCR")) IF ((Shape<>"SSL") & (Shape<>"INCL") & (Shape<>"DROP")) JP #ERR001 ENDIF ENDIF NO -------------SHAPE 1------ #SHAPE1 IF (Shape="CNL") MG "CANCEL" MG " " #REPEAT IN " Stop (SH, MO)> ",Stop{S2} JP #RIGHT, ((Stop="SH") | (Stop="MO"))  MG "ERROR! INVALID ENTRY." JP #REPEAT #RIGHT IF (Stop="SH") MG "SERVO-ING" SH ENDIF IF (Stop="MO") MG "MO/EXIT" MG " " Pos=_TPX IF (Pos<>0) MG "   Position [mm]: ",Pos/Ppmm MG " " MG "  ***SUBJECT WILL DROP***" MG " "  ENDIF IN "  EXIT? (MO, MAIN)>",Exit{S4} MG " " IF (Exit="MO") MO JP #END ELSE MG " " MG "  Back to MAIN..."  ENDIF ENDIF JP #MAIN ENDIF NO -----------------------------SHAPE 2-------- #SHAPE2 IF (Shape="SSR") IF (_LFX=0) 128  MG " " MG "********************************************** " MG "ERROR: LIMIT SWITCH ACTIVE! LOWER CARRIAGE... " MG "********************************************** " MG " " JP #MAIN ENDIF MG " " MG "SS RAISE" Pos=_TPX Trav=MaxPos-Pos IF (Trav<=0) MG " ERROR! MAX HEIGHT REACHED/EXCEEDED. " NO SH JP #MAIN ELSE mmTrav=(Trav/Ppmm) MG " "  MG " Travel [mm]= ",mmTrav Dist=mmTrav Accel=50 Decel=50 Speed=100 MG " "  MG " System ARMED..." MG " " IN " Press L to Launch (C-Cancel, L-Launch)> ",Launch{S1} IF (Launch="L") MG " *LAUNCHED* " JP #RUN1 ELSE MG " *CANCELLED* " NO SH JP #MAIN ENDIF ENDIF ENDIF NO -----------------------------SHAPE 3----- #SHAPE3 IF (Shape="INCR") IF (_LFX=0) MG " " MG "********************************************** " MG "ERROR: LIMIT SWITCH ACTIVE! LOWER CARRIAGE... " MG "********************************************** " MG " " JP #MAIN ENDIF MG " " MG "INC RAISE" Pos=_TPX Trav=MaxPos-Pos IF (Trav<=0) MG " " MG "********************************************** " 129  MG " ERROR! MAX HEIGHT REACHED/EXCEEDED. " MG "********************************************** " MG " " NO SH JP #MAIN ELSE mmTrav=(Trav/Ppmm) MG " " MG "  Available travel [mm]= ",mmTrav MG " " IN "  Enter distance [mm]> ",Dist Dist=@ABS[Dist] MG " ",Dist cDist=Dist*Ppmm cDist=@INT[cDist] NO MG "  Distance in Counts = ",cDist NO IF (@ABS[Dist]>@ABS[mTrav]) IF (@ABS[cDist]>@ABS[Trav]) MG " ERROR! DISTANCE > AVAILABLE TRAVEL." NO SH JP #MAIN ELSE NO The commented lines of code below are used only when troubleshooting. NO IN "  Enter speed [mm/s]> ",Speed NO MG " ",Speed NO IN "  Enter acceleration [mm/s/s]> ",Accel NO MG " ",Accel NO IN "  Enter deceleration [mm/s/s]> ",Decel NO MG " ",Decel NO MG " " NO Set motion parameters: Accel=50 Decel=50 Speed=100 MG " "  MG " System ARMED..." MG " " IN " Press L to Launch (C-Cancel, L-Launch)> ",Launch{S1} IF (Launch="L") MG " *LAUNCHED* " JP #RUN1 ELSE MG " *CANCELLED* " NO SH JP #MAIN ENDIF ENDIF ENDIF ENDIF NO -----------------------------SHAPE 4-------- #SHAPE4 IF (Shape="SSL") MG " " MG "SS LOWER" Pos=_TPX 130  Trav=Pos-MinPos IF (Trav<=0) MG "********************************************** " MG " ERROR! MIN HEIGHT REACHED/EXCEEDED. " MG "********************************************** " NO SH JP #MAIN ELSE Trav=-1*@ABS[Trav] mmTrav=(Trav/Ppmm) Dist=mmTrav Accel=100 Decel=100 Speed=200 MG " " MG " Travel [mm]= ",mmTrav MG " " MG " System ARMED..." MG " "  IN " Press L to Launch (C-Cancel, L-Launch)> ",Launch{S1} IF (Launch="L") MG " *LAUNCHED* " JP #RUN1 ELSE MG " *CANCELLED* " NO SH JP #MAIN ENDIF ENDIF ENDIF NO -----------------SHAPE 5--------------- #SHAPE5 IF (Shape="INCL") MG "INC LOWER" Pos=_TPX Trav=Pos-MinPos IF (Trav<=0) MG "********************************************** " MG " ERROR! MIN HEIGHT REACHED/EXCEEDED. " MG "********************************************** " NO SH JP #MAIN ELSE mmTrav=(Trav/Ppmm) MG " " MG "  Available travel [mm]= ",mmTrav MG " " IN "  Enter distance [mm]> ",Dist Dist=-1*@ABS[Dist] MG " ",Dist cDist=Dist*Ppmm cDist=@INT[cDist] NO MG "  Distance in Counts ",cDist IF (@ABS[cDist]>@ABS[Trav]) MG "********************************************** " 131  MG " ERROR! DISTANCE > AVAILABLE TRAVEL." MG "********************************************** " NO SH JP #MAIN ELSE NO The commented code is for troubleshooting only: NO IN "  Enter speed [mm/s]> ",Speed NO MG " ",Speed NO IN "  Enter acceleration [mm/s/s]> ",Accel NO MG " ",Accel NO IN "  Enter deceleration [mm/s/s]> ",Decel NO MG " ",Decel NO MG " " NO Set motion parameters: Accel=100 Decel=100 Speed=200 MG " "  MG " System ARMED..." MG " " IN " Press L to Launch (C-Cancel, L-Launch)> ",Launch{S1} IF (Launch="L") MG "*LAUNCHED* " JP #RUN1 ELSE MG "*CANCELLED* " NO SH JP #MAIN ENDIF ENDIF ENDIF ENDIF NO -----------------SHAPE 6---------------- #SHAPE6 IF (Shape="DROP") MG "DROP" Pos=_TPX IF (@ABS[(MaxPos-Pos)]>100) MG "**************************************************** " MG " ERROR! SUBJECT CARRIAGE NOT RAISED TO DROP HEIGHT." MG "**************************************************** " NO SH JP #MAIN ELSE JP #DROP #WELCOME MG " " MG " System ARMED..." MG " " IN " Press L to Launch (C-Cancel, L-Launch)> ",Launch{S1} IF (Launch="L") MG "*LAUNCHED* " JP #RUN2 ELSE MG "*CANCELLED* " 132  NO SH JP #MAIN ENDIF ENDIF ENDIF #DROP Pos[0]=@INT[Ppmm*-0.0000] Pos[1]=@INT[Ppmm*-0.2792] Pos[2]=@INT[Ppmm*-1.1158] Pos[3]=@INT[Ppmm*-2.5063] Pos[4]=@INT[Ppmm*-4.4452] Pos[5]=@INT[Ppmm*-6.9249] Pos[6]=@INT[Ppmm*-9.9359] Pos[7]=@INT[Ppmm*-13.4667] Pos[8]=@INT[Ppmm*-17.5046] Pos[9]=@INT[Ppmm*-22.0351] Pos[10]=@INT[Ppmm*-27.0428] Pos[11]=@INT[Ppmm*-32.5113] Pos[12]=@INT[Ppmm*-38.4234] Pos[13]=@INT[Ppmm*-44.7613] Pos[14]=@INT[Ppmm*-51.5070] Pos[15]=@INT[Ppmm*-58.6422] Pos[16]=@INT[Ppmm*-66.1486] Pos[17]=@INT[Ppmm*-74.0080] Pos[18]=@INT[Ppmm*-82.2024] Pos[19]=@INT[Ppmm*-90.7142] Pos[20]=@INT[Ppmm*-99.5261] Pos[21]=@INT[Ppmm*-108.6213] Pos[22]=@INT[Ppmm*-117.9837] Pos[23]=@INT[Ppmm*-127.5975] Pos[24]=@INT[Ppmm*-137.4476] Pos[25]=@INT[Ppmm*-147.5195] Pos[26]=@INT[Ppmm*-157.7992] Pos[27]=@INT[Ppmm*-168.2736] Pos[28]=@INT[Ppmm*-178.9298] Pos[29]=@INT[Ppmm*-189.7558] Pos[30]=@INT[Ppmm*-200.7401] Pos[31]=@INT[Ppmm*-211.8715] Pos[32]=@INT[Ppmm*-223.1397] Pos[33]=@INT[Ppmm*-234.5347] Pos[34]=@INT[Ppmm*-246.0471] Pos[35]=@INT[Ppmm*-257.6677] Pos[36]=@INT[Ppmm*-269.3882] Pos[37]=@INT[Ppmm*-281.2002] Pos[38]=@INT[Ppmm*-293.0962] Pos[39]=@INT[Ppmm*-305.0686] Pos[40]=@INT[Ppmm*-317.1105] Pos[41]=@INT[Ppmm*-329.2153] Pos[42]=@INT[Ppmm*-341.3764] Pos[43]=@INT[Ppmm*-353.5878] Pos[44]=@INT[Ppmm*-365.8437] Pos[45]=@INT[Ppmm*-378.1386] Pos[46]=@INT[Ppmm*-390.4671] Pos[47]=@INT[Ppmm*-402.8242] Pos[48]=@INT[Ppmm*-415.2049] 133  Pos[49]=@INT[Ppmm*-427.6046] Pos[50]=@INT[Ppmm*-440.0188] Pos[51]=@INT[Ppmm*-452.4432] Pos[52]=@INT[Ppmm*-464.8735] Pos[53]=@INT[Ppmm*-477.3059] Pos[54]=@INT[Ppmm*-489.7364] Pos[55]=@INT[Ppmm*-502.1614] Pos[56]=@INT[Ppmm*-514.5771] Pos[57]=@INT[Ppmm*-526.9801] Pos[58]=@INT[Ppmm*-539.3671] Pos[59]=@INT[Ppmm*-551.7346] Pos[60]=@INT[Ppmm*-564.0796] Pos[61]=@INT[Ppmm*-576.3990] Pos[62]=@INT[Ppmm*-588.6896] Pos[63]=@INT[Ppmm*-600.9486] Pos[64]=@INT[Ppmm*-613.1731] Pos[65]=@INT[Ppmm*-625.3603] Pos[66]=@INT[Ppmm*-637.5074] Pos[67]=@INT[Ppmm*-649.6117] Pos[68]=@INT[Ppmm*-661.6706] Pos[69]=@INT[Ppmm*-673.6815] Pos[70]=@INT[Ppmm*-685.6418] Pos[71]=@INT[Ppmm*-697.5491] Pos[72]=@INT[Ppmm*-709.4009] Pos[73]=@INT[Ppmm*-721.1946] Pos[74]=@INT[Ppmm*-732.9281] Pos[75]=@INT[Ppmm*-744.5987] Pos[76]=@INT[Ppmm*-756.2043] Pos[77]=@INT[Ppmm*-767.7425] Pos[78]=@INT[Ppmm*-779.2110] Pos[79]=@INT[Ppmm*-790.6075] Pos[80]=@INT[Ppmm*-801.9298] Pos[81]=@INT[Ppmm*-813.1756] Pos[82]=@INT[Ppmm*-824.2069] Pos[83]=@INT[Ppmm*-834.8962] Pos[84]=@INT[Ppmm*-845.2561] Pos[85]=@INT[Ppmm*-855.2985] Pos[86]=@INT[Ppmm*-865.0348] Pos[87]=@INT[Ppmm*-874.4757] Pos[88]=@INT[Ppmm*-883.6312] Pos[89]=@INT[Ppmm*-892.5110] Pos[90]=@INT[Ppmm*-901.1243] Pos[91]=@INT[Ppmm*-909.4798] Pos[92]=@INT[Ppmm*-917.5856] Pos[93]=@INT[Ppmm*-925.4498] Pos[94]=@INT[Ppmm*-933.0799] Pos[95]=@INT[Ppmm*-940.4831] Pos[96]=@INT[Ppmm*-947.6662] Pos[97]=@INT[Ppmm*-954.6359] Pos[98]=@INT[Ppmm*-961.3984] Pos[99]=@INT[Ppmm*-967.9599] Pos[100]=@INT[Ppmm*-974.3262] Pos[101]=@INT[Ppmm*-980.5027] Pos[102]=@INT[Ppmm*-986.4949] Pos[103]=@INT[Ppmm*-992.3079] 134  Pos[104]=@INT[Ppmm*-997.9466] Pos[105]=@INT[Ppmm*-1003.4159] Pos[106]=@INT[Ppmm*-1008.7202] Pos[107]=@INT[Ppmm*-1013.8640] Pos[108]=@INT[Ppmm*-1018.8516] Pos[109]=@INT[Ppmm*-1023.6870] Pos[110]=@INT[Ppmm*-1028.3742] Pos[111]=@INT[Ppmm*-1032.9170] Pos[112]=@INT[Ppmm*-1037.3190] Pos[113]=@INT[Ppmm*-1041.5839] Pos[114]=@INT[Ppmm*-1045.7149] Pos[115]=@INT[Ppmm*-1049.7154] Pos[116]=@INT[Ppmm*-1053.5887] Pos[117]=@INT[Ppmm*-1057.3377] Pos[118]=@INT[Ppmm*-1060.9654] Pos[119]=@INT[Ppmm*-1064.4747] Pos[120]=@INT[Ppmm*-1067.8683] Pos[121]=@INT[Ppmm*-1071.1490] Pos[122]=@INT[Ppmm*-1074.3192] Pos[123]=@INT[Ppmm*-1077.3815] Pos[124]=@INT[Ppmm*-1080.3382] Pos[125]=@INT[Ppmm*-1083.1917] Pos[126]=@INT[Ppmm*-1085.9442] Pos[127]=@INT[Ppmm*-1088.5979] Pos[128]=@INT[Ppmm*-1091.1547] Pos[129]=@INT[Ppmm*-1093.6168] Pos[130]=@INT[Ppmm*-1095.9859] Pos[131]=@INT[Ppmm*-1098.2641] Pos[132]=@INT[Ppmm*-1100.4530] Pos[133]=@INT[Ppmm*-1102.5543] Pos[134]=@INT[Ppmm*-1104.5697] Pos[135]=@INT[Ppmm*-1106.5008] Pos[136]=@INT[Ppmm*-1108.3491] Pos[137]=@INT[Ppmm*-1110.1159] Pos[138]=@INT[Ppmm*-1111.8028] Pos[139]=@INT[Ppmm*-1113.4110] Pos[140]=@INT[Ppmm*-1114.9417] Pos[141]=@INT[Ppmm*-1116.3962] Pos[142]=@INT[Ppmm*-1117.7756] Pos[143]=@INT[Ppmm*-1119.0810] Pos[144]=@INT[Ppmm*-1120.3134] Pos[145]=@INT[Ppmm*-1121.4737] Pos[146]=@INT[Ppmm*-1122.5630] Pos[147]=@INT[Ppmm*-1123.5820] Pos[148]=@INT[Ppmm*-1124.5316] Pos[149]=@INT[Ppmm*-1125.4124] Pos[150]=@INT[Ppmm*-1126.2252] Pos[151]=@INT[Ppmm*-1126.9707] Pos[152]=@INT[Ppmm*-1127.6494] Pos[153]=@INT[Ppmm*-1128.2619] Pos[154]=@INT[Ppmm*-1128.8085] Pos[155]=@INT[Ppmm*-1129.2899] Pos[156]=@INT[Ppmm*-1129.7063] Pos[157]=@INT[Ppmm*-1130.0581] Pos[158]=@INT[Ppmm*-1130.3455] 135  Pos[159]=@INT[Ppmm*-1130.5689] Pos[160]=@INT[Ppmm*-1130.7283] Pos[161]=@INT[Ppmm*-1130.8239] Pos[162]=@INT[Ppmm*-1130.8557] Pos[163]=@INT[Ppmm*-1130.8557] JP #CALC NO ------------------------------------------ #CALC NO calculate the difference array I=0 #LOOP1 J=I+1 Diff[I]=Pos[J]-Pos[I] I=I+1 JP #LOOP1,I<N JP #WELCOME NO ------------------------------------------ #RUN1 NO clear output bits CB 1 NO Check motion parameters are within range IF (@ABS[Speed]>200) MG "Error: Speed > 200 mm/s." JP #MAIN ENDIF IF (@ABS[Accel]>100) MG "Error: Accel > 100 mm/s/s." JP #MAIN ENDIF IF (@ABS[Decel]>100) MG "Error: Decel > 100 mm/s/s." JP #MAIN ENDIF IF (@ABS[Dist]>1131) MG "Error: Dist > 1131 mm." JP #MAIN ENDIF NO Run Position Relative Mode NO convert [mm] units to [counts] units DistC=Dist*Ppmm SpeedC=Speed*Ppmm AccelC=Accel*Ppmm DecelC=Decel*Ppmm NO execute motion profile  AC AccelC DC DecelC SP SpeedC PR DistC MG " " NO MG "     Dist",Dist NO MG "     Speed",Speed NO MG "     Accel ",Accel NO MG "     Decel ",Decel SH NO Set bit 1 high, sled output to DAQ.  136  SB 1 BG X AM X CB 1 MG " "  MG " PR Motion Complete..." SH JP #MAIN NO ------------------------------------------ #RUN2 NO clear output bits CB 1 NO Run Contour Mode CM X DT 2 NO DT 4 C=0 NO Set bit 1 high, sled output to DAQ. SB 1 #MOVING CD Diff[C] WC C=C+1 JP #MOVING, C<I SH CB 1 NO Reset modes DT 0 CD 0 MG " CM Motion Complete..." JP #MAIN NO ------------------------------------------- NO -------- ERROR CONDITIONS ----------------- NO ------------------------------------------- #ERR001 SH MG " " MG " #ERR001-- Invalid Entry" JP #MAIN NO ------------------------------------------- NO ----------------- END --------------------- NO ------------------------------------------- #END MG " " MG " " MG " * * * * * EXITED * * * * * " EN     137  Appendix C  Risk mitigation documents This index contains all the documents used to complete a hazard analysis. Hazards and harms were determined, after which they were assigned levels of severity and probability of occurrence. Hazards were then assigned a risk index score using a Risk Index Table. The appropriate action was then determined using a Risk Action Key. The criteria for risk severity, semi-quantatative probability levels, risk index table, risk action key, and format of the hazard analysis was recommended by a course in Clinical and Industrial Biomedical Engineering given at the University of British Columbia by Professor Bruno Jaggi [128]. These items were used to align with requirements set by ISO 14971. The risk index used here has also been used in industry and reviewed by the FDA [129]. All of these aforementioned items are presented in succession below: Table 5-1 Criteria for risk (severity) level The severity level assigned to each harm was determined using these levels of severity. Severity was separated into three different categories: impact on human health, environment, or property.  Severity Category Definition Human Health Impact Environmental Impact Property Impact 1 Negligible No noticable effect No noticable effect Damage < $1k 2 Marginal Discomfort or minor injury not requiring medical intervention Isolated exposure to harmful substance requiring intervention $1k < Damage < $10k 3 Significant Reversible injury requiring medical intervention Localized chronic exposure to harmful substances $10k < Damage < $100k 4 Catastrophic Death or irreversible injury affecting quality of life Widespread chronic exposure to harmful substances $100k < Damage  138  Table 5-2 Semi-quantitative probability levels Probability of occurrence was determined using the semi-quantitative probability levels in this table. The definitions are based on a per use basis.  Likelihood of Occurrence Definition f Frequent ≥ ≈ 10^−3 p Probable < ≈ 10^−3 and ≥ ≈ 10^−4 o Occasional < ≈ 10^−4 and ≥ ≈ 10^−5 i Improbable < ≈ 10^−5 and ≥ ≈ 10^−6 r Remote < ≈ 10^−6  Table 5-3 Hazard analysis risk index table After being assigned a severity and probability level, the risk index table is used to assigned a risk index score.     Severity (S) Likelihood of Occurance (P) 4 Catastrophic 3 Significant 2 Marginal 1 Negligible f, frequent 20 18 14 9 p, probable 19 16 12 6 o, occasional 17 15 10 4 r, remote 13 11 7 2 I, improbable 8 5 3 1  Table 5-4 Risk action key The risk index score is used to determine the appropriate level of action. The risk action key prescribes the appropriate level of action using a risk index assigned by the risk index table.  Risk Index (RI) Action 139  1-4 Acceptable without further investigation. 5-11 Acceptable, although further mitigation should be considered. 12-15 Undesirable, further mitigation is indicated. May be accepted with adequate cost benefit rationale.  16-20 Unacceptable, further mitigation is required.  140  Table 5-5 Hazard analysis table The hazard analysis table documents the hazards, harms, and their associated risk indices and subsequent design actions.141  142  143  144  145  146  147  148  149   150  Appendix D  Experimental Protocol This appendix contains the intended experimental protocol for a human subject experiment. The pre-test checklist and other precautions born from the risk mitigation process are included here. The full experimental protocol is shown on the following pages.  151  152  153  154  155  156  157  158  159  160   

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0371612/manifest

Comment

Related Items