UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Measuring in vivo internal spinal cord deformations during experimental spinal cord injury using a rat… Lucas, Erin 2010

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

Item Metadata

Download

Media
24-ubc_2010_fall_lucas_erin.pdf [ 2.89MB ]
Metadata
JSON: 24-1.0071208.json
JSON-LD: 24-1.0071208-ld.json
RDF/XML (Pretty): 24-1.0071208-rdf.xml
RDF/JSON: 24-1.0071208-rdf.json
Turtle: 24-1.0071208-turtle.txt
N-Triples: 24-1.0071208-rdf-ntriples.txt
Original Record: 24-1.0071208-source.json
Full Text
24-1.0071208-fulltext.txt
Citation
24-1.0071208.ris

Full Text

Measuring In Vivo Internal Spinal Cord Deformations During Experimental Spinal Cord Injury Using a Rat Model, Radiography, and Fiducial Markers  by  Erin Lucas B.S, Virginia Polytechnic State Institute and State University, 2006  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE in The Faculty of Graduate Studies (Biomedical Engineering)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) August 2010 © Erin Lucas, 2010  Abstract Spinal cord injuries (SCIs) are commonly studied experimentally by causing injury to rodent spinal cords in vivo and analyzing behavioral and histological results post injury. Few researchers have directly investigated the deformation of the in vivo spinal cord during impact, which is thought to be a predictor of injury. This knowledge would help to establish correlations among impact parameters, internal structure deformation, and histological and functional outcomes. The objective of this thesis was to develop a radiographic method of tracking the realtime internal deformations of an anesthetized rat‘s spinal cord during a typical experimental SCI. A technique was developed for injecting fiducial markers into the dorsal and ventral white and grey matter of in vivo rat spinal cords. Two radio-opaque beads were injected into C5/6 in the approximate location of the dorsal and ventral white matter. Four additional beads were glued to the surface of the cord caudal and cranial to the injection site (one dorsal, one ventral). Overall bead displacement was measured during quasi-static compression using standard medical x-ray equipment. Dynamic bead displacement was tracked during a dorsal impact (130mm/s, 1mm depth) by imaging laterally at 3,000 fps using a custom high-speed x-ray system. The internal spinal cord beads displaced 1.02-1.7 times more than the surface beads in the cranial direction and 2.5-11 times more in the ventral direction for the dynamic impact and maximum quasi-static compressions. The dorsal spinal cord beads (internal and surface) displaced more than the ventral spinal cord beads during all compressions. Finite element modeling and experimental measurements suggested that bead migration with respect to the spinal cord tissue was small and mostly insignificant. These results support the merit of this technique for measuring in vivo spinal cord deformation. The differences in bead displacements imply that the spinal cord undergoes complex internal and surface deformations during impact. Many applications of this technique are conceivable including validating finite element and surrogate models of the spinal cord, comparing localized grey and white matter motion during impact to histological findings, and improving SCI preventative and treatment measures.  ii  Preface Dr. Jie Liu is a microsurgery specialist at ICORD (International Collaboration on Repair Discoveries) and performed all of the surgeries for this work. The development of the injection and impact methods was a collaboration between him and me. Drs. Peter Cripton, Tom Oxland, and Wolf Tetzlaff also gave guidance during the development of the methodologies used for this work and provided revisions for the writing of this thesis. Colin Russell is a M.A.Sc. candidate with the Orthopaedic Injury Biomechanics Group (OIBG) at UBC and wrote the denoising algorithm used for the high-speed x-ray videos (Chapter 3). Colin wrote the majority of the description of the denoising algorithm in the Appendix. He also developed the finite element model (for his thesis work) which was used for assessing bead migration and he ran the simulations for this analysis (Chapter 3). Colin wrote the majority of the limitations associated with the finite element model in the discussion of Chapter 3. Robyn Newell is a Ph.D. candidate with OIBG and wrote the algorithm used to quantify the remaining distortion in the x-ray images in the Appendix. Jason Chak is a research engineer with OIBG and wrote the LabView program used for data acquisition during the contusion injuries (Chapter 3). This work was approved by the UBC Animal Care Centre and Ethics Committee. The ethics certificate (number A08-0518) is included in the Appendix.  iii  Table of Contents Abstract .............................................................................................................................. ii Preface ............................................................................................................................... iii Table of Contents .............................................................................................................. iv List of Tables ..................................................................................................................... ix List of Figures .................................................................................................................... x Acknowledgements ......................................................................................................... xiii Dedication ........................................................................................................................ xiv Chapter 1: Introduction .................................................................................................... 1 1.1  Motivation ........................................................................................................................... 1  1.2  Spinal cord anatomy .......................................................................................................... 2  1.3  Spinal cord injury .............................................................................................................. 6  1.4  Animal models of SCI ........................................................................................................ 9  1.4.1  SCI mechanisms ........................................................................................................... 9  1.4.2  Limitations of SCI models .......................................................................................... 11  1.4.3  SCI injury severity ...................................................................................................... 12  1.5  Relationship between SCI mechanics and injury ......................................................... 14  1.6  Biomechanics of the spinal cord during SCI ................................................................. 15  1.6.1  Theoretical models ..................................................................................................... 16  1.6.2  Experimental tests....................................................................................................... 18 iv  1.6.2.1  Tensile tests ....................................................................................................................................... 18  1.6.2.2  Compression tests .............................................................................................................................. 20  1.6.3  Surrogate models ........................................................................................................ 21  1.6.4  Finite element models ................................................................................................. 22  1.7  Imaging of neural tissue deformation ............................................................................ 23  1.7.1  Magnetic resonance imaging ...................................................................................... 24  1.7.2  Radiography................................................................................................................ 25  1.8  Summary ........................................................................................................................... 27  1.9  Thesis objectives and scope ............................................................................................. 28  1.10 Thesis manuscript overview ............................................................................................ 28 1.11 References ......................................................................................................................... 30  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model....................................................................................................... 39 2.1  Introduction ...................................................................................................................... 39  2.2  Methods ............................................................................................................................. 40  2.2.1  Surgical preparation and bead injection ..................................................................... 40  2.2.2  Injection accuracy ....................................................................................................... 42  2.2.2.1  Surgical procedure ............................................................................................................................. 42  2.2.2.2  Bead location verification .................................................................................................................. 43  2.2.3  2.3  Quasi-static compression ............................................................................................ 43  2.2.3.1  Surgical procedure ............................................................................................................................. 43  2.2.3.2  Spinal cord compression procedure ................................................................................................... 45  2.2.3.3  Bead displacement measurement....................................................................................................... 45  2.2.3.4  Bead migration assessment ................................................................................................................ 47  2.2.3.5  Measurement accuracy and repeatability........................................................................................... 48  2.2.3.6  Statistical analysis ............................................................................................................................. 48  Results ............................................................................................................................... 49 v  2.3.1  Injection accuracy ....................................................................................................... 49  2.3.2  Quasi-static compression ............................................................................................ 50  2.3.2.1  Bead displacements ........................................................................................................................... 50  2.3.2.2  Bead migration .................................................................................................................................. 53  2.3.2.3  Measurement accuracy and repeatability........................................................................................... 54  2.4  Discussion.......................................................................................................................... 54  2.5  Conclusions ....................................................................................................................... 57  2.6  References ......................................................................................................................... 59  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model ........................................................................... 61 3.1  Introduction ...................................................................................................................... 61  3.2  Methods ............................................................................................................................. 63  3.2.1  Surgical procedure ...................................................................................................... 63  3.2.2  Impact procedure ........................................................................................................ 64  3.2.3  High-speed x-ray imaging .......................................................................................... 66  3.2.4  Bead displacement and velocity measurement ........................................................... 66  3.2.5  Bead migration assessment ......................................................................................... 67  3.2.6  Bead tracking accuracy and precision ........................................................................ 67  3.2.7  Statistical analysis....................................................................................................... 67  3.2.8  Finite element analysis ............................................................................................... 68  3.3  Results ............................................................................................................................... 69  3.3.1  Bead location, impact parameters, and high-speed x-ray imaging ............................. 69  3.3.2  Bead displacement and velocity ................................................................................. 70  3.3.3  Bead migration ........................................................................................................... 73  3.3.4  Bead tracking accuracy and precision ........................................................................ 73  3.3.5  Finite element analysis ............................................................................................... 73  3.4  Discussion.......................................................................................................................... 74 vi  3.5  Conclusions ....................................................................................................................... 77  3.6  References ......................................................................................................................... 79  Chapter 4: Discussion and Conclusions ........................................................................ 83 4.1  Comparison between current methodology and relevant literature ........................... 83  4.1.1  Neural tissue deformation measurements ................................................................... 83  4.1.2  Fiducial marker types ................................................................................................. 84  4.1.3  High-speed x-ray image quality ................................................................................. 84  4.1.4  Bead tracking accuracy and precision ........................................................................ 86  4.1.5  Marker migration ........................................................................................................ 87  4.2  Spinal cord mechanics during injury ............................................................................. 88  4.2.1  Internal deformation ................................................................................................... 88  4.2.2  Overall force versus displacement .............................................................................. 89  4.3  Impact parameters and injury severity ......................................................................... 91  4.4  Quasi-static compression compared to dynamic contusion ......................................... 93  4.5  Finite element simulation compared to experimental dynamic contusion ................. 95  4.6  Strengths and limitations ................................................................................................ 96  4.6.1  Strengths ..................................................................................................................... 96  4.6.2  Limitations .................................................................................................................. 98  4.7  Future work .................................................................................................................... 103  4.8  Conclusions ..................................................................................................................... 106  4.9  References ....................................................................................................................... 107  Appendix A: Fiducial Marker Selection...................................................................... 111 A1  Marker type evaluation ................................................................................................. 111 vii  A2  Marker selection............................................................................................................. 112  Appendix B: Quasi-Static Bead Displacements .......................................................... 114 Appendix C: Dynamic Bead Displacements and Velocities....................................... 117 Appendix D: High-Speed X-ray Image Processing .................................................... 119 D1  Image distortion correction ........................................................................................... 119  D2  Denoising......................................................................................................................... 121  Appendix E: Sensor and TEMA Data Processing ...................................................... 123 E1  Sensor data processing .................................................................................................. 123  E1.1 Filtering ........................................................................................................................ 123 E1.2 Inertial compensation ................................................................................................... 125 E1.3 Time syncing ................................................................................................................ 128 E2.  TEMA data processing .................................................................................................. 129  E2.1 Filtering ........................................................................................................................ 129 E2.2 Pixel to millimeter conversion factor ........................................................................... 130 E2.3 Magnification error ...................................................................................................... 131 E2.4 TEMA data syncing with sensor data .......................................................................... 132  Appendix F: Sensor Specifications and Accuracy ...................................................... 133 F1  Load cell calibration ...................................................................................................... 133  F2  LVDT accuracy .............................................................................................................. 133  F3  Actuator repeatability and PID settings ...................................................................... 134  Appendix G: Ethics Board Certificate ........................................................................ 135 viii  List of Tables Table 1-1: ASIA Impairment Scale (Maynard et al., 1997) ............................................................7 Table 3-1: Average spinal cord bead displacements at maximum impactor depth, maximum velocities during impact, and post-impact displacements (avg±STD). ............71 Table 3-2: Maximum differences in displacements of nodes with beads compared to nodes without beads (control). ...........................................................................................74 Table A-1: Powder type fiducial markers evaluated....................................................................111 Table A-2: Bead type fiducial markers evaluated........................................................................111 Table B-1: Spinal cord bead displacements during compression. ...............................................114 Table B-2: Spinal cord bead post-compression displacements....................................................115 Table C-1: Average spinal cord bead displacements at maximum impactor depth, maximum velocities during impact, and post-impact displacement ................................117  ix  List of Figures Figure 1-1. Divisions of the spinal column and spinal cord ©Elsevier Ltd., 2005 (Drake et al., 2005).. ........................................................................................................................3 Figure 1-2: A typical vertebrae ©Elsevier Ltd., 2005 (Drake et al., 2005).. ...................................4 Figure 1-3: Anatomy of the spinal cord ©Elsevier Ltd., 2005 (Drake et al., 2005). .......................5 Figure 1-4: Anatomy of a Neuron (http://commons.wikimedia.org/wiki/File:Neuron.svg). ..........5 Figure 1-5: Types of vertebral column fractures which cause spinal cord injury: burst fracture (A), fracture dislocation (B), and distraction (C) (Vaccaro et al., 2005). ..............7 Figure 1-6: MRI of a cervical SCI caused by cord compression due to a burst fracture (indicated by arrow).. ...........................................................................................................8 Figure 1-7: Schematic showing contusion (A), dislocation (B), and distraction (C) type SCI mechanisms in rats created with a machine developed at the University of British Columbia (Choo et al., 2008).................................................................................11 Figure 1-8: Predicted stress within a spinal cord during spinal cord compression (Panjabi and White, 1988).. ..............................................................................................................17 Figure 1-9: Predicted strains within the spinal cord during spinal cord compression (Blight, 1988). ....................................................................................................................18 Figure 1-10: Lateral radiograph of contrast agent fiducial markers injected inside of a feline‘s spinal cord (Maiman et al., 1989). ........................................................................26 Figure 1-11: High-speed x-ray image of fiducial markers inside of a human cadaveric brain (Al-Bsharat et al., 1999). ..........................................................................................26 Figure 2-1: Bead injection procedure. ...........................................................................................41 Figure 2-2: Transverse cross-section of rat spinal cord segment C5 from a rat spinal cord atlas (Left) (Paxinos and Watson, 1986) and bead injection location, depth, and angle determined using the atlas (Right). ..........................................................................43 Figure 2-3: Lateral radiograph showing bead positioning in the cervical spinal cord of an anesthetized rat (A) and associated diagram (B) ...............................................................44 Figure 2-4: Bead centroids selected using MATLAB. The frame and bead centroid selections from one rat are shown (A) ...............................................................................46 Figure 2-5: Approximate location of the beads for the injection accuracy study. .........................49 x  Figure 2-6: Microscope image of a 260µm tantalum bead injected into dorsal grey matter of the spinal cord. ...............................................................................................................50 Figure 2-7: Spinal cord bead displacements in the D/V direction for each compression group (avg ± std). ...............................................................................................................51 Figure 2-8: Spinal cord bead displacements in the Cr/Cd direction for each compression group (avg ± std). ...............................................................................................................52 Figure 2-9: Overlay of spinal cord bead displacements before (yellow), during (red), and after (blue) a 3.0mm compression. ....................................................................................53 Figure 3-1: Lateral high-speed x-ray image showing the internal and surface bead positioning in the cervical spinal cord of an anesthetized rat ............................................64 Figure 3-2: Experimental test set-up. .............................................................................................65 Figure 3-3: Finite element model of the rat cervical spinal cord with bead inclusions developed by C. Russell (Russell et al., 2008). .................................................................69 Figure 3-4: X-ray images from high-speed x-ray video before, during, and after impact. ............70 Figure 3-5: Force versus impactor displacement during impact of the spinal cord .......................70 Figure 3-6: Average spinal cord bead displacements in the D/V direction during impact. ...........71 Figure 3-7: Average spinal cord bead velocities in the D/V direction during impact. ..................72 Figure 3-8: Spinal cord bead displacements from the FEM simulation of the 130mm/s impact.. ...............................................................................................................................74 Figure 4-1: Gel model of spinal cord contusion demonstrating Cr/Cd tissue displacement (Blight, 1988).. ...................................................................................................................89 Figure 4-2: Force displacement data extracted from contusion SCI animal studies (Bresnahan et al., 1987; Choo et al., 2007; Sparrey et al., 2008).. ....................................91 Figure 4-3: Comparison between the dorsal-ventral (A) and cranial-caudal (B) displacements of spinal cord beads during the dynamic contusion and the quasistatic 2mm compression group. .........................................................................................95 Figure 4-4: Example of bead patterns used to obtain an overall map of localized spinal cord deformations. ...........................................................................................................105 Figure A-1: Powder injection needle shown withdrawing powder in preparation for injection............................................................................................................................112 Figure D-1: Distortion grid before (A) and after distortion correction (B). ................................120 xi  Figure D-2: Distortion grid after distortion correction showing how remaining distortion was quantified. .................................................................................................................121 Figure D-3: Images from high-speed x-ray before (A) and after (B) denoising. .........................122 Figure E-1: FFT of raw acceleration data. ...................................................................................123 Figure E-2: FFT of raw load cell data. .........................................................................................123 Figure E-3: FFT of LVDT data. ...................................................................................................124 Figure E-4: Filtered versus raw acceleration data using a 200Hz cut-off frequency...................124 Figure E-5: Filtered versus raw load cell data using a 200Hz cut-off frequency. .......................125 Figure E-6: Filtered versus raw LVDT data using a 200Hz cut-off frequency. ..........................125 Figure E-7: Inertially compensated force data of a blank impact. ...............................................126 Figure E-8: Inertially compensated force data of a blank impact showing remaining offset. ...............................................................................................................................127 Figure E-9: Inertially compensated force data from one experimental test. ................................127 Figure E-10: Inertially compensated force data with offset mask from one experimental test. ...................................................................................................................................128 Figure E-11: Force and acceleration data shifted to LVDT data. ................................................129 Figure E-1: FFT of Internal dorsal bead from one experimental test. ........................................130 Figure E-2: Filtered TEMA data. ................................................................................................130 Figure E-3: Magnification error measurement. ...........................................................................131 Figure F-1: Load cell calibration .................................................................................................133 Figure G-1: Animal care certificate representing ethics approval for the studies presented in this thesis......................................................................................................................135  xii  Acknowledgements I would like to first acknowledge Dr. Peter Cripton for giving me the opportunity to work in the lab, and for his encouragement, optimism, and ingenuity. I would also like to thank Drs. Tom Oxland and Wolf Tetzlaff for their guidance and expertise. I am indebted Dr. Jie Liu for his surgical expertise, patience, and endless support during my experimental testing. I am very thankful for the help and suggestions from all of the students and staff in our lab and at the International Collaboration on Repair Discoveries (ICORD), especially Colin Russell, Tim Bhatnagar, Claire Jones, Robyn Newell, Jason Chak, Maryam Shahrokni, Carolyn Van Toen, Katharine Wilson, Hannah Gustafson, Angela Melnyk, Heather Murray, Clarrie Lam, Femke Streijger, Peggy Assinck, and Yuan Jiang. I would like to thank my funding support including ICORD, Centre for Hip Health and Mobility, Canadian Institutes of Health Research, Vancouver Costal Health Research Institute, and the University of British Columbia. I am grateful for my family and friends for their never-ending support and encouragement. Thanks to Heather Murray and Katharine Wilson for reminding me to have fun, and a special thanks to John Aherne for his love, patience, and positive outlook for the future.  xiii  Dedication  To my family for their endless support, guidance, and love  xiv  Chapter 1: Introduction  Chapter 1: Introduction 1.1  Motivation Spinal cord injury (SCI) is one of the most devastating and traumatic injuries for  which there is no known cure (Kwon et al., 2004). SCIs can result in partial to complete paralysis and can affect just about every aspect of day-to-day living including motor, sensory, respiratory, and cardiac function. Understandably these types of injuries can cause severe psychological distress such as feelings of depression and anxiety (Kennedy and Rogers, 2000). It is estimated that there are approximately 12,000 new SCIs each year in the United States (National Spinal Cord Injury Statistical Center, 2010) and 1,700 per year in Canada (Dryden et al., 2003). Currently, it is estimated that at least 280,000 people are living with SCI in the US and Canada (Rick Hansen Spinal Cord Injury Registry, 2006). The annual economic cost of SCIs is estimated to be $9.7 billion in the US (Berkowitz et al., 1998) and $1.5 billion in Canada (Rick Hansen Spinal Cord Injury Registry, 2006). Globally, the main cause for SCIs are motor vehicle related accidents (40-50%) followed by injury due to falls, sports and recreation, work, and violence (Sekhon and Fehlings, 2001). In Canada, falls represent 31% of the SCI population and are the second leading cause of the injury, following motor vehicle accidents (Pickett et al., 2006). The majority of SCI research focuses on understanding the pathophysiology of the injury and developing treatments. There is still relatively little known about the mechanical response of the spinal cord during injury, which is a fundamental step in understanding the resulting severity of injury. Existing methods of studying the biomechanics of the spinal cord have involved mechanical testing of ex vivo human spinal cords and ex vivo and in vivo animal spinal cords. Neural tissue has been shown to degrade quickly after death so the results from ex vivo studies are most likely not representative of the in vivo response (Hung and Chang, 1981; Oakland et al., 2006). In vivo mechanical testing has involved extensive laminectomies and sensor attachments to the cord which would change the natural environment of the spinal cord and possibly alter its natural physiologic response (Chang et al., 1981; Hung et al., 1975). Only the overall cord deformation has been measured with these studies, leaving the internal localized tissue mechanics unknown. The anatomical 1  Chapter 1: Introduction  structure of the spinal cord is nonhomogeous leaving many to believe that it may have different localized mechanical responses to injury. In particular the white matter of the cord has been thought to be mechanically different than the grey matter (Blight, 1988; Ichihara et al., 2003; Ichihara et al., 2001). Histological findings such as larger amounts of damage in the centralized spinal cord tissue could possibly be attributed to these mechanical differences in the cord, although to the author‘s knowledge this has not been directly studied experimentally (Blight, 1988; Blight and Decrescito, 1986). The ability to measure internal deformation during injury would help define relationships between SCI impact parameters, internal structure deformation, and biological and functional outcomes. The overall objective of this thesis was to develop a method of measuring the internal deformations of an in vivo rat spinal cord during an experimental SCI. This research provided a foundation for many potential applications in the areas of basic science, prevention, and repair of SCI. A direct application of this research is to improve existing models of the spinal cord, such as finite element models (FEMs) and surrogate models, which currently have only been validated using overall spinal cord biomechanical properties. Specifically, in vivo deformations of various regions in the white and grey matter could be used to validate or adjust mechanical properties of these models resulting in a more physiologic representation of the spinal cord. These models can be used to assess the efficacy of injury mitigating devices such as helmets or automotive safety features. In the long term, the added knowledge of the internal deformation of the spinal cord during SCIs could help target treatments to areas suffering particularly large amounts of mechanical damage.  1.2  Spinal cord anatomy The spinal cord is protected by the spinal column which consists of approximately 33  vertebrae (Figure 1-1): seven cervical vertebrae (C1-C7), twelve thoracic (T1-T12), five lumbar (L1-L5), five sacral (S1-S5), and three to four fused vertebrae which make up the Coccyx (Co1-Co4). A typical vertebra is shown in Figure 1-2, though there are size and shape variations depending on the location in the spinal column. In general, a vertebra consists of an anteriorly located vertebral body which supports loads on the spinal column, and a posteriorly located vertebral arch which surrounds the spinal cord and provides 2  Chapter 1: Introduction  muscular and ligamentous attachment points. The vertebral arch is formed by the pedicles, laminae, and transverse and spinous processes. The vertebrae are connected to each other through synovial joints formed by the articular processes on the vertebral arch and intervertebral discs located between the vertebral bodies. The intervertebral discs evenly distribute loads across the vertebral bodies and allow for motion between vertebrae. They consist of an outer fibrocartilage structure called the annulus fibrosus which surrounds a gelatinous center called the nucleus pulposus. The vertebrae are also connected with various ligaments and musculature which will not be discussed in detail.  C1  Cervical enlargement (of spinal cord)  7 cervical vertebrae (CI-CVII)  Pedicles of vertebrae  C2 C3 C4 C5 C6 C7 C8 T1 T2 T3  Spinalganglion  T4 T5 T6  12 thoracic vertebrae (TI-TXII)  T7 T8 T9 T10  Lumbosacral enlargement (of spinal cord)  T11 T12 L1  5 lumbar vertebrae (LI-LV)  L2 Cauda equina  L3 L4 L5  Sacrum (5 fused sacral vertebra I-V) Coccyx (3-4 fused coccygeal vertebra I-IV)  S1 S2 S3 S4 S5 Co  Figure 1-1. Divisions of the spinal column and spinal cord ©Elsevier Ltd., 2005 (Drake et al., 2005). Reprinted from Drake et al., 2004 with permission from Elsevier.  3  Chapter 1: Introduction  Figure 1-2: A typical vertebrae ©Elsevier Ltd., 2005 (Drake et al., 2005). Reprinted from Drake et al., 2004 with permission from Elsevier.  The spinal cord is located in the superior two-thirds of the spinal canal, typically descending until L1 and L2 (Figure 1-1). The spinal cord is surrounded by three meninges which are thought to protect the cord and maintain its shape. The dura mater is the outer most membrane, the arachnoid mater is closely coupled to the dura mater, and the pia mater is closely coupled to the spinal cord. Between the arachnoid mater and pia mater is the subarachnoid space which consists of cerebrospinal fluid (CSF) and denticulate ligaments extending from the pia mater to suspend the spinal cord in the dural sac. The central portion of the spinal cord consist of the characteristic ―H‖ shaped grey matter which contains neuronal cell bodies and other supporting cells (Figure 1-3). The neuronal cell bodies send and receive signals from the brain, body, or other regions in the spinal cord. These signals are called action potentials and are carried through cable like projections from the neural cell bodies called axons (Figure 1-4). The axons travelling through the spinal cord form long tracts which surround the grey matter and form the white matter. Axons exit the cord ventrally (from the front) carrying motor signals and enter the cord dorsally (from the back) carrying sensory information. The axons entering and exiting the spinal cord form rootlets which join together to create the nerve roots. The nerve roots exit between foramina created by the joining of two vertebrae, and the segments of the spinal cord are named based on the level that the nerve roots exit the spinal column (Figure 1-1). There are 31 pairs of nerve roots: eight cervical, twelve thoracic, five lumbar, five sacral, and one coccygeal. Since the  4  Chapter 1: Introduction  spinal cord is shorter than the spinal column, there is a long ―tail‖ of nerve roots called the cauda equina which contains nerve roots exiting the lower levels of the spinal canal. Grey matter White matter  Figure 1-3: Anatomy of the spinal cord ©Elsevier Ltd., 2005 (Drake et al., 2005). Reprinted from Drake et al., 2004 with permission from Elsevier.  Figure 1-4: Anatomy of a Neuron (http://commons.wikimedia.org/wiki/File:Neuron.svg). Reproduced under the GNU Free Documentation and the Creative Commons Attribution-Share Alike 3.0 Unported Licenses.  5  Chapter 1: Introduction  1.3  Spinal cord injury The majority of acute SCIs are caused by apparent injury to the spinal column  including burst fractures, fracture dislocation, and distraction injuries, as depicted in Figure 1-5 (Sekhon and Fehlings, 2001; Vaccaro et al., 2005). Burst fractures and fracture dislocations are the most prevalent types of spinal column fractures associated with SCI (Pickett et al., 2006; Sekhon and Fehlings, 2001). Burst fractures result from crushing of the vertebral bodies due to axial loading to the vertebral column. Boney projections from burst fractures can protrude into the canal and cause compression of the spinal cord in the anteriorposterior direction. Fracture dislocation is caused by the vertebral column dislocating between two vertebrae in the anterior-posterior or lateral direction resulting in shearing of the spinal cord. Distraction injuries are caused by the spinal column being pulled apart axially, causing tension in the spinal cord (Vaccaro et al., 2005). Spinal cord injuries can also result from congenital abnormalities and degeneration in the spinal column including boney growths in the spinal canal (osteophytes), hypertrophy of ligaments inside the canal, and bulging discs, all of which can encroach upon the spinal cord (Panjabi and White, 1988; Sekhon and Fehlings, 2001). SCIs occur most often at the cervical level which is thought to be due to the smaller vertebrae and greater flexibility of the spinal column in this region (Pickett et al., 2006; Sekhon and Fehlings, 2001). The most prevalent clinical evaluation of SCI severity involves rating the remaining motor and sensory functionality on the American Spinal Injury Association (ASIA) impairment scale shown in Table 1-1 (Maynard et al., 1997). SCI patients are also imaged with MRI post-injury and throughout their recovery to examine the integrity of the spinal cord (Flanders et al., 1990; Ramon et al., 1997). The three main features assessed in MRI are hemorrhage (bleeding), edema (swelling), and cord compression (all depicted in Figure 1-6) (Flanders et al., 1990; Kulkarni et al., 1987; Marciello et al., 1993).  6  Chapter 1: Introduction  A  B  C  Figure 1-5: Types of vertebral column fractures which cause spinal cord injury: burst fracture (A), fracture dislocation (B), and distraction (C) (Vaccaro et al., 2005). Reprinted from Vaccaro et al., 2005 with permission from Wolters Kluwer/Lippincott, Williams & Wilkins.  Table 1-1: ASIA Impairment Scale (Maynard et al., 1997). Reprinted by permission from Macmillan Publishers Ltd: Maynard et al., ©1997  A B  C  D E  Complete. No sensory or motor function is preserved in the sacral segments S4-S5. Incomplete. Sensory but not motor function is preserved below the neurological level and includes the sacral segments S4-S5. Incomplete. Motor function is preserved below the neurological level, and more than half of key muscles below the neurological level have a muscle grade less than 3. Incomplete. Motor function is preserved below the neurological level, and at least half of key muscles below the neurological level have a muscle grade greater than or equal to 3. Normal. Sensory and motor function is normal.  7  Chapter 1: Introduction  Figure 1-6: MRI of a cervical SCI caused by cord compression due to a burst fracture (indicated by arrow). The darker signal in the spinal cord likely represents hemorrhage (horizontal arrow head) and the lighter signal likely represents edema (vertical arrowhead) (Flanders et al., 1990).  The immediate effects of a SCI, called the primary injury, are due to the mechanical insult to the cord and result in destruction of cells and vascular damage (Choo et al., 2007). In the days and weeks after the initial injury a biological sequence of events occur which worsen the injury causing further cell death and axonal damage. This is referred to as secondary injury (Kwon et al., 2004; Sekhon and Fehlings, 2001). The resulting tissue damage typically appears as a centralized lesion around the injury epicenter coupled with cysts, scarring, and a periphery of spared uninjured axons (Flanders et al., 1990; Hayes and Kakulas, 1997; Kwon et al., 2004; Norenberg et al., 2004). Primary damage due to the mechanical insult is thought of as irreversible and treatments are typically focused on reducing the amount of secondary damage due to swelling, hemorrhage and complex cellular responses (Kwon et al., 2004). Clinical trials have been conducted to assess the efficacy of pharmacological therapies, such as methylprednisolone, in the recovery and repair of SCI (Kwon et al., 2004; Tator, 2006). There has been very little clinical success in this area and there currently remains no known cure for SCI (Kwon et al., 2004). 8  Chapter 1: Introduction  A limitation of analyzing causes and treatments of SCI in humans is that it is impossible to control for all of the variables leading up to injury including age, health, integrity of the spinal column, type of injury, and severity of injury. Additionally. detailed analysis of cellular damage can only be done at the time of death. Highly controlled experiments can be conducted using animal models, which have shown to correlate well with the injuries seen clinically (Metz et al., 1998; Metz et al., 2000a; Stokes and Jakeman, 2002).  1.4  Animal models of SCI  1.4.1 SCI mechanisms Animal models can be used to understand the basic science of SCIs including the biomechanics of injury, the tissue and cellular response following injury, and the response to therapeutic strategies. The rat and mouse are the most popular animal models due to the lower cost and easier accessibility compared to larger animals (Kwon et al., 2002). The most common method of causing a SCI to an animal model is simulating contusion due to burst fracture by impacting the dorsal surface of the spinal cord, causing transverse compression to the cord. This has been done by dropping a weight on the cord (Allen, 1911; Basso et al., 1996; Blight and Decrescito, 1986; Dohrmann and Panjabi, 1976; Hung et al., 1975; Noble and Wrathall, 1987; Wrathall et al., 1985; Young, 2002; Young, 2009) or impacting the cord with an actuated device (Anderson, 1985; Bresnahan et al., 1987; Choo et al., 2007; Jakeman et al., 2009; Kearney et al., 1988; Onifer et al., 2007; Scheff et al., 2003). A highly controlled weight drop type impactor is the Multicenter Animal Spinal Cord Injury Study (MASCIS) impactor, more commonly known as the New York University (NYU) impactor (Young, 2002; Young, 2009). Mild, moderate, or severe injuries to rat spinal cords are produced with this impactor by dropping a 10g weight on the dorsal surface of the spinal cord from a height of 12.5, 25.0, or 50.0mm resulting in impact velocities of approximately 490, 690, 970 mm/s and cord compressions (displacement of the dorsal cord surface in the ventral direction) of 2-3mm (Basso et al., 1996; Maikos and Shreiber, 2007; Metz et al., 2000a; Metz et al., 2000b; Young, 2002; Young, 2009). The use of feedback controlled mechanical actuators allows for more control over the impact parameters than drop weight type impactors. These impactors are typically displacement controlled or force controlled as 9  Chapter 1: Introduction  both of these parameters have been shown to be related to injury severity (Anderson, 1985; Basso et al., 1996; Dohrmann et al., 1978; Kearney et al., 1988; Maikos and Shreiber, 2007; Noble and Wrathall, 1987). The Ohio State University (OSU) impactor is the most commonly used displacement controlled impactor (Bresnahan et al., 1987; Jakeman et al., 2009; Noyes, 1987b; Stokes et al., 1992). This model uses an electromagnetic actuator to impact the dorsal surface of rodent spinal cords to user defined depths of around 0.3-1.3mm at a maximum velocity of about 300mm/s (Bresnahan et al., 1987). The OSU impactor has been shown to produce repeatable injuries at graded injury severities (mild, moderate, or severe injury) based on impact depth (Kloos et al., 2005; Ma et al., 2001; Pearse et al., 2005). The Infinite Horizons (IH) impactor is a force controlled device that was recently developed for rodent models to allow for user defined force using a force feedback system (Cao et al., 2005; Ghasemlou et al., 2005; Kim et al., 2009; Scheff and Roberts, 2009; Scheff et al., 2003). The impactor is controlled with a stepping motor which delivers impacts at approximately 130mm/s with forces of 1.0, 1.5, and 2.0N and resulting displacements of 0.8, 1, and 1.2mm for reproducible mild, moderate, and severe injuries to the thoracic spinal cord of a rat (Scheff et al., 2003). There are a limited amount of impactors designed to simulate SCI due to dislocation or distraction. There are only two known dislocation models (Choo et al., 2007; Clarke and Bilston, 2008; Clarke et al., 2008; Fiford et al., 2004) which create either lateral or posterior dislocations of the vertebral bodies causing shearing in the spinal cord. There are a small number distraction models (Cusick et al., 1982; Maiman et al., 1989; Myklebust et al., 1988) which typically restrain the pelvis and apply an axial excursion to the head. Researchers at the University of British Columbia developed the first known SCI machine capable of creating contusions, dislocation, and distraction type injuries (Figure 1-7) (Choo et al., 2007).  10  Chapter 1: Introduction  Figure 1-7: Schematic showing contusion (A), dislocation (B), and distraction (C) type SCI mechanisms in rats created with a machine developed at the University of British Columbia (Choo et al., 2008). Reprinted from Choo et al., 2008 with permission from Elsevier.  SCI due to chronic cord compression is also simulated using animal models. This is typically done by compressing the spinal cord at slow speeds (around 0.02-3mm/s (Blight, 1991; Huang et al., 2007; Hung et al., 1982; Sparrey et al., 2008)) and thus this is commonly referred to as quasi-static compression. Clip compressions are performed by physically clipping the entire spinal cord and then releasing the clips after an extended period of time (Fehlings and Tator, 1995; Guha et al., 1987; Joshi and Fehlings, 2002; Nashmi and Fehlings, 2001). Hydraulic actuators have been used to slowly compress the spinal cord which allows for velocity and displacement control and force measurements (Carlson et al., 1997; Hung et al., 1982). Other methods of quasi-static compression include forceps compression (Blight, 1991), lowering weights onto the cord (Huang et al., 2007), balloon inflation in the spinal canal (Fukuda et al., 2005), and insertion of spacers in the canal (Dimar et al., 1999). Full or partial spinal cord transections are also a commonly performed to ensure completeness of the injury to specific areas of the cord and assess the effectiveness of therapeutic treatments (Kwon et al., 2002). 1.4.2 Limitations of SCI models There are many sources of variability with SCI models which have caused problems in the repeatability and comparison of results between research groups (Kwon et al., 2002). Limitations with weight-drop models include variations in impactor tips, alignment of the impactor, height and weight variations, bouncing of the weight after impact, and lack of 11  Chapter 1: Introduction  direct control over the impact depth, velocity, and force (Blight, 1988; Dohrmann and Panjabi, 1976; Kwon et al., 2002). Some weight drop impactors have incorporated load cells, strain gauges, or accelerometers for force measurements (Dohrmann and Panjabi, 1976; Pintar et al., 1996; Wrathall et al., 1985), but most, including the NYU impactor, do not have the capability to record force (Young, 2009). While actuator controlled impactors allow for control of velocity, depth, and/or force, variation between research groups can still occur due to methodological differences such as the type of impactor tip, alignment of the impactor, and methods of stabilizing the spinal column. There is also variation in determining when cord contact occurs, either by assigning a force threshold (Choo et al., 2007; Pearse et al., 2005; Scheff and Roberts, 2009; Sparrey et al., 2008) or using sensors on the bottom of the impactor to determine dura touch (Young, 2009), which affects measurement of overall cord compression. Limitations with quasi-static compression devices include size variations of the instruments used and lack of direct control over force, rate of compression, and/or cord deformation (Kwon et al., 2002). An inherent limitation of all of these SCI injury devices is that they can only measure the global mechanical response of the spinal cord to impact, leaving the localized internal mechanics unknown. The ability to measure the internal mechanics of the spinal cord could help compare impact methods and standardize experimental protocols. 1.4.3 SCI injury severity Injury severity resulting from experimental SCI can be determined using functional, electrophysiological or histological methods. Basso and colleagues created a 21 point locomotor rating scale called the Basso, Beatie, and Bresnahan (BBB) scale which rates injury severity based on observed walking in an open field for a four minute period of time (Basso et al., 1995). A score of 0 means there is no observable hindlimb movement and 21 represents successful gait coordination, trunk stability, and appropriate paw and tail positioning. Other types of SCI injury severity scoring include further analysis of gait, grid walking, narrow beam crossing, and various reaching tasks (Behrmann et al., 1992; Metz et al., 2000b).  12  Chapter 1: Introduction  Damage, compression, or excess pressure on the spinal cord can inhibit action potentials, so injury severity can be assessed by testing the ability of the axons to transmit signals through the site of injury (Carlson et al., 2003; Carlson et al., 1997; Shi and Blight, 1996). The has been done by stimulating action potentials in sensory (sensory evoked potentials (SEP)) or motor (motor evoked potentials (MEP)) tracts on one side of the injury and recording the amplitude and timing of the response on the other side with decreased amplitude of response or latency in response indicating injury (Metz et al., 2000a). Histology is performed to assess tissue level and cellular level damage and is used to analyze the progression of injury, compare different types of injury, and assess treatment strategies. Most of what is known about the primary and secondary injury has come from histological analysis of SCI in animal models (Choo et al., 2008; Choo et al., 2007; Kwon et al., 2004). To analyze the amount of spared tissue, hemorrhage, and the presence of cysts and lesions in the cord, sections of the spinal cord are commonly stained with substances that bind to components of the white and grey matter causing them to show up as different colors under a light microscope (Basso et al., 1996; Choo et al., 2007; Clarke et al., 2008; Scheff et al., 2003). Neuronal membrane integrity can be assessed by exposing the spinal cord to fluorescently labeled substances that normally would not penetrate the cell membrane. The cells can then be viewed with florescent microscopy and those which contain the fluorescent substance can be assumed to have a compromised cell membrane (Choo et al., 2008; Choo et al., 2007; Stone et al., 2002). MRI has been used to study spinal cord tissue level injury after experimental SCI in rodents (Bilgen et al., 2001; Gareau et al., 2001; Kozlowski et al., 2008). Diffusion tensor MR imaging has been used to track diffusion of molecules through the spinal cord, where any disruption in this diffusion may be due to structural damage (Ellingson et al., 2008; Kim et al., 2009; Kozlowski et al., 2008; Schwartz and Hackney, 2003). These approaches are still relatively new and, to the author‘s knowledge, have yet to be widely adopted by the neuroscience community.  13  Chapter 1: Introduction  1.5  Relationship between SCI mechanics and injury It has been shown by many studies that the severity of injury is correlated with the  mechanical parameters of impact specifically compression depth, velocity, duration, and force (Anderson, 1985; Basso et al., 1996; Bresnahan et al., 1987; Carlson et al., 2003; Carlson et al., 1997; Fehlings and Tator, 1995; Scheff et al., 2003; Sparrey et al., 2008). While many researchers have analyzed these separately, it is thought that injury severity is a function of a combination of these parameters (Dohrmann and Panjabi, 1976; Kearney et al., 1988). At very low velocities, compression depth, duration, and force have been shown to be good predictors of injury (Carlson et al., 2003; Carlson et al., 1997; Dimar et al., 1999; Fehlings and Tator, 1995; Hung et al., 1982; Joshi and Fehlings, 2002). Fehlings et al. investigated the effect of clip compression force on the amount of surviving axons and found that the number of axons decreased with increasing force (Fehlings and Tator, 1995). To determine whether duration of compression affected injury severity, Carlson et al. compressed the spinal cord of canines for either thirty minutes or three hours and found that the thirty minute group had a rapid recovery while the longer compression resulted in larger lesion sizes, no recovery of somatosensory evoked potentials, and substantial functional deficits (Carlson et al., 2003). The spinal cord has been shown by other studies to be tolerant to compression up to 40-50% if done at slow speeds (Gruner et al., 1996; Hung et al., 1982). At higher velocities, both velocity and compression depth have been shown to be correlated with injury (Anderson, 1985; Basso et al., 1996; Bresnahan et al., 1987; Kearney et al., 1988; Maikos and Shreiber, 2007; Sparrey et al., 2008). Weight drop models like the NYU impactor have been shown to produce injuries with decreasing white matter sparing and animal functionality with increasing drop weight height which correspond with greater levels of compression and velocity (Basso et al., 1996; Blight and Decrescito, 1986; Maikos and Shreiber, 2007). Maikos et al. used the NYU impactor with a rat model and reported that damage to the blood spinal cord barrier correlated better with rate of compression than depth of compression with more injury being seen in the grey matter than the white matter (Maikos and Shreiber, 2007). It is not possible to fully separate the effects of velocity and 14  Chapter 1: Introduction  compression in drop weight type models since increasing the drop weight height increases both the amount of compression and velocity of impact. The OSU impactor has been shown to created injuries of increasing severity with increasing impact depth at a constant velocity (Behrmann et al., 1992; Bresnahan et al., 1987). Anderson tested the effects of increasing compression and velocity separately using a pneumatic impactor and found that impact velocity was correlated with hemmorhagic necrosis (cell death), while compression correlated better with latency in SEPs (Anderson, 1985). Kearney et al. conducted a similar study and found at low velocities compression was a better predictor of injury (SEP and histological) and at higher velocities injury severity correlated better with the product of velocity and compression (Kearney et al., 1988). Sparrey et al. impacted in vivo rat spinal cords at slow and fast velocities and found that axon damage increased and the level of hemorrhage extended from the grey matter into the white matter during the faster velocity (Sparrey et al., 2008). Other studies have contradicted these findings showing weak correlations between impact velocities (ranging between 0.05-0.5m/s) and injury severity so this remains a debated subject (Kim et al., 2009; Noyes, 1987a). Impact force has also been shown to be correlated with high velocity injuries. Using the IH impactor, Scheff et al. showed a significant decrease in spared tissue and BBB locomotor outcome with increasing impact force (Scheff et al., 2003). Ghasemlou confirmed these findings using a mouse model with the IH impactor (Ghasemlou et al., 2005). The integral of impact force over time (impulse) and the integral of impact force and cord displacement (energy transferred to the cord) have also been shown to correlate with lesion volume using a feline model and weight drop contusions of varying height and weight (Dohrmann and Panjabi, 1976).  1.6  Biomechanics of the spinal cord during SCI Deformations reported from previous experiments (discussed above) are typically  scalar values (i.e. maximum compression depth), which limits biomechanical analysis of these studies. Direct measurement of dynamic spinal cord deformations during experimental SCIs and comparison to histological and functional deficit would help relate spinal cord mechanics to impact parameters and injury outcome to determine tissue tolerances and to 15  Chapter 1: Introduction  better understand the resulting injury. Biomechanical properties of the spinal cord can also be used to validate and develop surrogate spinal cords (Hall et al., 2006; Jones et al., 2008; Kroeker et al., 2009; Persson et al., 2009; Pintar et al., 1996) and FEMs (Czyz et al., 2008; Greaves et al., 2008; Ichihara et al., 2003; Maikos et al., 2008). Surrogate spinal cords can be inserted into cadaveric spinal canals to investigate the interactions between spinal column fractures and cord compression. FEMs can be used for detailed analysis of localized stresses and strains inside of the cord which could potentially help to develop targeted treatment strategies. Both surrogate and FEMs can also be used to assess the efficacy of preventative devices (Nelson and Cripton, 2008). The following section will discuss past models used for investigating spinal cord biomechanics, with a focus on methodology and general findings. 1.6.1 Theoretical models Several researchers have developed theories about the biomechanical behavior of the spinal cord during a compressive type SCI in attempts to understand histological and functional injuries such as preferential damage to the center of the cord and the resulting ―central cord‖ syndrome seen clinically (Blight, 1988; Blight and Decrescito, 1986; Harrison, 1999; Panjabi and White, 1988; Raynor and Koplik, 1985). Panjabi and White predicted stress distributions in the spinal cord during a SCI based on standard mechanical three-point bending analysis as shown in Figure 1-8 (Panjabi and White, 1988). They suggested that a complex stress profile may be present inside of the cord during compression due to a combination of compressive, tensile, and shear stresses. In this basic theory, compressive stress would be highest directly under the impactor and shearing stress would be highest in the middle of the spinal cord. Tensile stress in the cranial-caudal (Cr/Cd) direction would be uniform throughout the cord due to stretching of the cord from its superior connections with the brainstem and inferior tethering to the spinal column. The bending load would cause compressive stress at the site of impact and tensile stresses on the opposing side. The authors predicted that the internal strain would follow a similar pattern. Blight‘s prediction of the internal strain pattern during compression is shown in Figure 1-9 and included compression under the impactor, centralized Cr/Cd displacement of the internal tissue (which he called centralized ―extrusion‖) due to in the fluid like tendencies of the internal tissue, and shearing from the interface between the cord and the edges of the impactor (Blight, 1988). Blight 16  Chapter 1: Introduction  hypothesized that the central displacement of the spinal cord may be the cause for the centralized damage observed both clinically and with animal models. Blight analyzed this theory further with a simple surrogate spinal cord model. A plastic tube was filled with a gelatinous material and ink lines were injected through the cord in the dorsal-ventral (D/V) direction. The cord was compressed to different levels and the movement of the ink tracks was recorded. The gelatinous material deformed in a parabolic manner, with the maximum Cr/Cd displacement at the center and linear translation at the edges of the tube both of which increased with increasing displacement. Blight related this finding to what has been seen in histology after experimental SCIs where centralized damage increases radially with increasing impact severities (Blight, 1988; Blight and Decrescito, 1986).  A  B  C  F  D  E  G  Figure 1-8: Predicted stress within a spinal cord during spinal cord compression (Panjabi and White, 1988). Reprinted from Panjabi and White, 1988 with permission from Wolters Kluwer Health/ Lippincott, Williams & Wilkins.  17  Chapter 1: Introduction  Figure 1-9: Predicted strains within the spinal cord during spinal cord compression (Blight, 1988). Reproduced with permission from the author. While theoretical models based on engineering principles are useful in understanding the biomechanics of the spinal cord during injury, the complex structural arrangement of the spinal cord, including axially oriented bundles of axons in the white matter and neuronal cell bodies in the grey matter (discussed in Section 1.2), would likely cause localized differences in the patterns of stress and strain within the spinal cord. Experimental models are needed to fully characterize the biomechanical response of the spinal cord to impact. 1.6.2 Experimental tests 1.6.2.1 Tensile tests The most common biomechanical test of the spinal cord is tensile testing in the Cr/Cd direction using standard material testing equipment (Bassi et al., 2009; Bilston and Thibault, 1995; Chang et al., 1981; Chang et al., 1988; Fiford and Bilston, 2005; Hung and Chang, 1981; Hung et al., 1981; Ichihara et al., 2003; Ichihara et al., 2001; Oakland et al., 2006; Tunturi, 1978). The stress-strain response under constant strain rates and stress relaxation results are typically reported with these studies (Bilston and Thibault, 1995; Chang et al., 1981; Chang et al., 1988; Fiford and Bilston, 2005; Hung and Chang, 1981; Hung et al., 1981; Ichihara et al., 2003; Ichihara et al., 2001; Oakland et al., 2006). This has been done ex vivo with human cadaver spinal cords and both ex vivo and in vivo with animal spinal cords. The spinal cord has been shown to change material properties quickly after death so ex vivo material properties are most likely not representative of the in vivo response (Hung and Chang, 1981; Oakland et al., 2006). Hung and Chang‘s research group took a unique 18  Chapter 1: Introduction  approach to measuring in vivo material properties of feline and canine spinal cords in tension (Chang et al., 1981; Hung and Chang, 1981; Hung et al., 1981). Three rings were glued onto the exposed spinal cord and the animal was positioned vertically in an Instron machine. The top ring was loaded in tension while the bottom ring was fixed and the middle ring was connected to the Instron load cell to measure force. This approach was only used to obtain overall mechanical properties of the spinal cord, since internal deformation could not be seen, and its high level of invasiveness brings into question whether the cord response was representative of a spinal cord in its natural physiologic environment. In general, at lower strain rates these studies have shown that the spinal cords exhibited a relatively linear stressstrain response that is not dependent on strain rate. At higher strain rates there has been shown to be an initial linear stress-strain response at low strains followed by a non-linear response at higher strains with the slope of the stress-strain curve (representing stiffness of the spinal cord) increasing with increasing strain (Cheng et al., 2008). These tensile testing studies have normally only reported overall strain and do not typically measure localized strains on the surface of the cord or within the cord. Fiford and Bilston painted markers on the surface of an ex vivo rat spinal cord which were used to calculate localize surface strains in different regions of the spinal cord, however internal deformations were not measured (Fiford and Bilston, 2005). The only study known which has looked at internal deformations during tensile tests was done by Maiman et al. (Maiman et al., 1989). Contrast agent was injected into the center of in vivo feline spinal cords at C4C7 and T6-L1 and the pelvis of the feline was distracted to 5kg, 10kg, and 15kg. The displacement of the markers and vertebrae were compared to determine spine-spinal cord coupling. Ichihara et al. investigated the differences in grey and white matter material properties by excising grey and white matter samples from a bovine spinal cord and testing each section in tension (Ichihara et al., 2001). The results revealed that the grey matter was more rigid (higher young‘s modulus) and fragile (ruptured at lower strains) than the white matter. While this was done ex vivo, it supports the theory that there may be localized differences in internal mechanics.  19  Chapter 1: Introduction  1.6.2.2 Compression tests Many studies collect force and displacement data during contusion SCIs, but these data are typically used for assessing injury severity and relatively few are used to determine biomechanical properties of the spinal cord. Hung et al. imaged the dynamic deformation of an in vivo feline spinal cord dura a drop weight contusion with a high-speed camera (Hung et al., 1975). This study required an extensive laminectomy to fully expose the spinal cord, which changed the physiologic environment and could have potentially affected the spinal cord motion. Additionally, no data were collected on internal cord deformation since it could not be seen without imaging equipment that allowed visualization of internal structures. Pintar et al. obtained stress-strain measurements during dynamic contusion by impacting in vivo feline spinal cords using a drop weight with an accelerometer and linear variable differential transformer (LVDT) attached (Pintar et al., 1996). These sensors were used to generate stress and strain curves, though no information was given about the drop weight height, impact velocity, or number of trials so these data cannot be readily used by other researchers. Sparrey et al. used a modified version of the OSU impactor to impact in vivo rat spinal cord at slow and fast speeds (Sparrey et al., 2008). The main objective of this study was to investigate damage in the white and grey matter, though force displacement curves were reported showing the spinal cord to be much stiffer at the higher velocity. Hung et al. investigated the mechanical response of in vivo feline spinal cords during quasi-static loading by compressing the spinal cord using a displacement controlled Instron machine and measuring the resulting force (Hung et al., 1982). This study had similar results as the tensile tests with an initial linear elastic region at low deformations followed by a nonlinear region at higher deformations, although there were very few animals used for this study. Carlson et al. measured the viscoelastic relaxation of in vivo canine spinal cords by slowly compressing the spinal cord with a hydraulic piston in the spinal canal and measuring the pressure at the impactor/spinal cord interface during sustained compressions (Carlson et al., 1997). They reported a quick relaxation of the tissue during compression, with 51% dissipation of pressure after five minutes. The experimental set up did not allow for direct measurement of the corresponding spinal cord deformation which would give valuable information about what is happening inside the spinal cord during relaxation. 20  Chapter 1: Introduction  Ex vivo compression tests have been performed to examine the effect of the surrounding membranes, CSF, and ligaments on the spinal cord biomechanics (Hall et al., 2006; Jones et al., 2008; Ozawa et al., 2004; Persson et al., 2009). The effects of dura matter and CSF have been tested by impacting ex vivo bovine cords with and without dura and/or CSF and comparing the trajectories of the impactor in high-speed videos of the impact (Hall et al., 2006; Jones et al., 2008). The results have shown that CSF protects the cord by reducing spinal cord occlusion but the dura does not reduce occlusion. Ichihara et al. took MRI images of an uncompressed and compressed ex vivo bovine spinal cord and compared the percent area change in the grey and white matter during compression (Ichihara et al., 2001). This was used to validate a FEM of a bovine spinal cord. This study did not directly measure tissue deformation, however MRI could have potential for measuring internal spinal cord strains if fiducial markers were added to the spinal cord or through advanced image analysis techniques that incorporate deformable image registration (Ming, 2008). While testing ex vivo spinal cords has the benefit of a controlled testing environment with less biological variables, the results from these studies need in vivo validation. It is difficult to obtain strain data from in vivo studies because the entire spinal cord cannot be seen without complete removal of the vertebral arch. Determining when the impactor initially touches the spinal cord is also a challenge when converting displacement data to strain. Even if the dura touch point is known with the use of sensors (i.e. with the NYU impactor), without seeing the spinal cord it is unknown when the spinal cord itself begins to deform. These studies also do not have a method of measuring internal strains, so only the overall strain can be estimated. 1.6.3 Surrogate models Surrogate spinal cords have been developed based on experimental biomechanical data and used for estimating spinal cord compression due to structural failures in the spinal canal (Hall et al., 2006; Jones et al., 2008; Kroeker et al., 2009; Persson et al., 2009; Pintar et al., 1996). Pintar developed a gelatin surrogate spinal cord using the properties obtained from the previously mentioned in vivo feline drop weight contusions (Pintar et al., 1996). The surrogate cord was instrumented with pressure sensors and inserted into cadaver spinal columns. The spinal columns were axially loaded to produce a burst fractures and the 21  Chapter 1: Introduction  pressures in the spinal cord were measured. As previously mentioned, details of the feline contusions used for validation were not supplied, making it difficult to assess the validity of this model. Kroeker et al. developed a surrogate cord using in vivo quasi-static tensile and compressive material properties from Chang and Hung‘s research group (Chang et al., 1981; Hung et al., 1982; Kroeker et al., 2009). This surrogate cord exhibited a similarly shaped force-displacement curve to in vivo spinal cords under tension, however the surrogate cord had a higher stiffness than the in vivo spinal cords. Additional limitations in all known surrogate spinal cords include the lack of dura, CSF, and separate white and grey matter which all could affect the displacement of the cord during impact. In general, in vivo biomechanical data during different mechanisms of injury and loading rates are needed to validate surrogate spinal cords for use under a variety of different impact conditions. 1.6.4 Finite element models FEMs have an advantage over surrogate spinal cords in that localized stresses and strains can be analyzed during a wide variety of simulated injuries. Czyz et al. developed a FEM of the cervical spinal cord based on the geometries of a 21 year old male SCI patient in attempts to recreate the injury and compare the FEM results to the actual deficit seen in MRI (Czyz et al., 2008). A unique aspect of the model was that it incorporated different material properties for the grey and white matter and showed that high strains and strain rates correlated well with the injury that was seen clinically. The material properties used for this model were all from ex vivo tests and given that there are no known studies that have measured the internal strains in the transverse direction, validation of this method is difficult. The author did mention that these types of models are more advanced than our existing knowledge of spinal cord biomechanics but they have potential for determining strain thresholds in neural tissue and for improving prognosis and treatments of SCI. Greaves et al. developed a FEM model of a human cervical spinal column and spinal cord which was used to compare strain patterns of contusion, dislocation, and distraction type injury mechanisms (Greaves et al., 2008). In vivo tensile material properties from Chang and Hung‘s quasi-static tests were used to model the spinal cord. The FEM was validated for contusion using Hung et al.‘s quasi-static compression study and for distraction using Maiman et al.‘s spine/spinal cord coupling data (Hung et al., 1982; Maiman et al., 1989). There are no known studies 22  Chapter 1: Introduction  which track the cord deformation during a dislocation type injury so in vivo validation for this injury mechanism was not possible. The assumption that the spinal cord material properties are isotropic, linearly elastic, and homogenous and the lack of dynamic validation were limitations with this model. Ichihara developed a FEM of one bovine cervical spinal cord segment using separate white and grey material properties which incorporated viscoelasticity found in their ex vivo quasi-static tensile tests (Ichihara et al., 2003). This model is limited in that it was only developed for quasi-static compression, uses ex vivo material properties, and only models one segment of the spinal cord. Additionally, tensile properties were used for compression and validation was only done with a simplistic comparison between the volume change in MRI image of an ex vivo compressed bovine cord compared to the FEM. Maikos et al. developed a FEM of a rat spinal cord to simulate a drop weight impact using the NYU impactor and compare internal strain distribution to disruption of the blood brain barrier seen in previous experimental tests (Maikos et al., 2008; Maikos and Shreiber, 2007). This model assumed isotropy and used material properties from ex vivo tensile tests at low strain rates as a baseline and altered these properties until simulated drop weight trajectory matched what was seen experimentally (Fiford and Bilston, 2005; Maikos et al., 2008; Maikos and Shreiber, 2007). Viscoelastic components were added from a human brain study conducted at high strain rates since no such study exists in the spinal cord (Mendis et al., 1995). The white and grey matter properties were adjusted until the strain patterns matched locations of hemorrhage seen experimentally. This FEM model of the spinal cord most likely a close representation of what happens during an acute SCI, but the author does note that different combinations of parameters reproduced similar impactor trajectories so further experimental validation is needed. FEMs of the spinal cord have great potential, but they are more advanced than our current knowledge of in vivo spinal cord biomechanics, and further experimental tests which measure the localized internal spinal cord response to various loading rates and conditions are needed for thorough validation.  1.7  Imaging of neural tissue deformation The lack of research on internal spinal cord biomechanics is due in part to  complications of implementing appropriate imaging techniques to measure internal spinal cord deformations during injury. Medical imaging equipment capable of imaging the internal 23  Chapter 1: Introduction  structures of tissue is needed to visualize such deformation during injury. MRI and radiography are both promising methods of imaging spinal cord deformation. 1.7.1 Magnetic resonance imaging MRI is commonly used for imaging neural tissue and creates images with great contrast of the white and grey matter. The deformation of the human spinal cord has been measured during various degrees of non-injurious flexion and extension using tagged MRI (Margulies et al., 1992; Yuan et al., 1998). Tagged MRI uses an MRI sequence to create parallel lines in the tissue that remain with the tissue for a period of time. To image spinal cord deformation the spinal cord was first tagged with lines, and the initial position of the lines was compared to the position after flexion or extension of the neck. The deformation of the lines was then measured to determine strain within the cord. The limitation with this approach is that it is limited to quasi-static deformations since the image acquisition speed is not fast enough to capture deformation during a single dynamic impact such as an acute SCI. One method that was developed to capture dynamic deformations of the in vivo rat brain during a dynamic impact was to collect only a fraction of the image to reduce the acquisition time and then repeat the impact at least 64 times until the full image was acquired (Bayly et al., 2006; Bayly et al., 2005). Repeated impacts to the spinal cord at injurious levels would cause damage to the neural tissue which would more than likely change the biomechanics of the spinal cord. Ming developed an image registration technique in MRI to track the deformation of an in vivo rat spinal cord during flexion and extension (Ming, 2008). This technique is again limited to quasi-static deformation because of the speed of image acquisition. This methodology was validated against the FEM of the spinal cord developed by Greaves et al. (Greaves et al., 2008). Given the previously discussed limitations of the current state of FEMs, further validation with a physical model is needed such as tracking the deformation of fiducial markers in the spinal cord. A potential technique for imaging dynamic deformation with MRI is magnetic particle imaging (MPI) which tracks displacement of MRI contrast agent at temporal resolutions of up to 21.5ms (Weizenecker et al., 2009). This approach has been used for tracking the movement of blood through the heart of an in vivo mouse but has yet to be used in the central nervous system.  24  Chapter 1: Introduction  1.7.2 Radiography Despite the fact that neural tissue is nearly translucent to x-rays, radiography is a promising method for measuring spinal cord motion during impact due to the ability to collect images at high-speeds. Plain film x-ray and high-speed x-ray have been used to image animals, cadaveric human specimens, and human volunteers during biomechanical tests (AlBsharat et al., 1999; Bass et al., 2006; Bauman and Chang, 2010; Brainerd et al., 2010; Hardy et al., 2001; Hardy et al., 2007; Hardy et al., 2008; Kaneoka et al., 1999; Luan et al., 2000; Martin et al., 2010; Ono et al., 1997; Panjabi et al., 1995; Panjabi et al., 2000). Due to the radio-translucency of neural tissue, using this imaging modality requires the inclusion of radio-opaque fiducial markers. Smith injected iron wire pins into the spinal cord segments of ex vivo monkeys and took radiographs to measure the overall displacement of each segment during various movements such as lifting the arms and the legs (Smith, 1956). This study was limited to understanding the normal physiologic motion of the spinal cord rather than investigation of spinal cord biomechanics during injury. The only known study to image in vivo spinal cord deformation using radiography was done by Maiman et al. (Maiman et al., 1989). As previously mentioned, contrast agent was injected into the spinal cords of in vivo felines to measure Cr/Cd marker displacement during distraction and contusion injuries (Figure 1-10). Dynamic displacement of the markers was not reported with this study, and most likely was not measurable since standard fluoroscopy equipment was used which has relatively slow image acquisition speeds. Additionally, localized deformation in the D/V direction could not be measured since only one marker was injected per spinal cord segment. There have been studies which have imaged the dynamic deformation of the brain during impact using high-speed x-ray techniques (Al-Bsharat et al., 1999; Hardy et al., 2001; Hardy et al., 2007; Nusholtz et al., 1984; Zou et al., 2007). Fiducial markers have been injected into a cadaveric brain to track dynamic brain deformation during impact using high-speed x-ray (Figure 1-11) (Al-Bsharat et al., 1999; Hardy et al., 2001; Hardy et al., 2007; Zou et al., 2007). The markers were 5mm long and 2.5mm diameter, which would be too large for inclusion in the spinal cord.  25  Chapter 1: Introduction  Figure 1-10: Lateral radiograph of contrast agent fiducial markers injected inside of a feline‘s spinal cord (Maiman et al., 1989). The arrow heads are pointing to fiducial markers inside of the cervical spinal cord. Reprinted from Maiman et al., 1989 with permission from Wolters Kluwer/Lippincott, Williams & Wilkins.  Figure 1-11: High-speed x-ray image of fiducial markers inside of a human cadaveric brain (AlBsharat et al., 1999). Reproduced from Al-Bsharat et al., 1999 by permission of The Stapp Association.  The use of fiducial markers to track tissue deformation during high-speed impacts requires testing to assure the beads are not migrating with respect to the tissue. Typically if beads are glued directly to the tissue it is assumed that they will not migrate (Hardy et al., 2008; Panjabi et al., 1995; Panjabi et al., 2000). While this may be a valid assumption for glued markers, migration of markers not rigidly affixed should be evaluated. Maiman et al. tested marker migration by stretching ex vivo canine spinal cords various distances and comparing the intermarker displacements to the absolute cord displacements which occurred in a 1:1 ratio (Maiman et al., 1989). This approach is limited since internal versus absolute spinal cord deformation is unknown. They also reported that the force-displacement curves were the same with and without beads. Additionally, they dissected the spinal cord and noted that there was no disruption of surrounding structures. The methods were not fully disclosed 26  Chapter 1: Introduction  for this validation and no data were given on the results including accuracy of the bead measurement approach. Hardy et al. tracked the motion of brain tissue using neutral density targets (NDTs) of a similar density to neural tissue to reduce the chance of migration (Hardy et al., 2001). Migration of the NDTs was assessed by comparing the marker locations before and after impact which was shown to be minimal. Dissection of the brain was also performed revealing very little disruption of the surrounding tissue, and limited signs of the injection track. The migration of NDTs was further assessed using a FEM of the brain, which showed that the markers had negligible motion with respect to the surrounding tissue during impact (Hardy et al., 2007). Ex vivo methods of investigating marker migration in a tendon has been investigated by cyclic loading of a tendon with markers and comparing the position before and after loading (Roos et al., 2004). As previously discussed, the spinal cord is known to change material properties quickly after death, so ex vivo validation would not be appropriate.  1.8  Summary There is very little known about how the spinal cord deforms during an injury.  Existing experimental research in this area typically focuses on the overall spinal cord biomechanics, and very few studies have investigated internal deformations. Anatomical differences in the structure of the white and grey matter, differences in white and grey matter responses to mechanical tests, and localized tissue damage seen in histology give a strong indication that damage to the spinal cord is to some extent a result of nonhomogenous deformations inside of the cord. The ability to track the internal deformation of the spinal cord would help relate spinal cord tissue mechanics with functional and histological damage which can be used for determining injury thresholds, tying together existing SCI experimental methods, and associating localized areas of injury with different injury mechanisms. In vivo spinal cord deformation measurement during a variety of injuries can be used to develop more physiologic surrogate and FEMs of the spinal cord as well as to develop treatments targeted towards specific areas suffering from large amounts of mechanical damage.  27  Chapter 1: Introduction  1.9  Thesis objectives and scope The overall objective of this thesis was to develop a method of tracking internal  deformations of an in vivo rat spinal cord during a typical contusion type SCI using radiography. The specific objectives of this thesis were to use anesthetized rats to: 1. Develop an in vivo method of injecting multiple fiducial markers into the cervical spinal cord which (a) are tolerated by the rat with no noticeable physiologic effects, (b) can be injected into specific locations of the white and grey matter, and (c) are visible with x-ray. 2. Develop a method of tracking internal spinal cord displacement during quasi-static compression using standard x-ray equipment 3. Develop a method of tracking internal spinal cord displacement during dynamic contusion using high-speed x-ray equipment 4. Assess the likelihood of fiducial marker migration with respect to the spinal cord tissue during impact 5. Collect initial experimental data of internal spinal cord deformation during quasistatic and dynamic SCIs The scope of this thesis included imaging the bead displacements laterally in twodimensions with the intention of using bi-planar imaging in the future. The bead displacements were tracked in vivo in order to maintain the physiologic nature of the spinal cord. Only two beads were injected per spinal cord segment because of the bead size relative to the spinal cord diameter. Histology was not performed for the studies included in this thesis, although this could be incorporated in future studies to determine the SCI severity and potential damage due to the bead injections. This thesis was limited to non-survival studies and all animals were euthanized directly after impact. Future work would involve survival studies and testing the functional outcome, but this was out of the scope of this thesis.  1.10  Thesis manuscript overview The objectives of this thesis were accomplished with two main studies: a quasi-static  compression study (Chapter 2), and a dynamic contusion study (Chapter 3). The quasi-static 28  Chapter 1: Introduction  compression study (Chapter 2) involved development of the fiducial marker injection technique and assessment of marker placement accuracy as well as development of the methodology of tracking the fiducial marker displacement during quasi-static compression. Initial experimental data were collected on internal spinal cord deformation during multiple levels of compression and bead migration was assessed for each of these cases. The dynamic contusion study (Chapter 3) involved developing the methodology to track the dynamic displacement of the spinal cord fiducial markers using high-speed x-ray. Initial internal deformations were collected during a typical contusion type SCI, and migration was assessed in these tests. Finite element analysis (FEA) was also used in this study to further assess fiducial marker migration. The overall conclusions of these studies and their relation to relevant literature are discussed in Chapter 4.  29  Chapter 1: Introduction  1.11  References  Al-Bsharat, A. S., W. N. Hardy, K. H. Yang, T. B. Khalil, S. Tashman and A. I. King (1999). Brain/Skull relative displacement magnitude due to blunt head Impact: New experimental data and model. Proc. 43rd Stapp Car Crash Conference, pp. 321-332, Society of Automotive Engineers, Warrendale, PA. Allen, A. R. (1911). "Surgery of experimental lesions of spinal cord equivalent to crush injury of fracture dislocation of spinal column." J.A.M.A 57: 878-880. Anderson, T. E. (1985). "Spinal cord contusion injury: experimental dissociation of hemorrhagic necrosis and subacute loss of axonal conduction." J Neurosurg 62(1): 115-119. Bass, C. R., R. S. Salzar, S. R. Lucas, M. Davis, L. Donnellan, B. Folk, E. Sanderson and S. Waclawik (2006). "Injury risk in behind armor blunt thoracic trauma." Int J Occup Saf Ergon 12(4): 429-442. Bassi, M., P. F. Jarzem, M. Steibel, P. Barriga, J. Ouellet and R. Reindl (2009). "An in vitro study on the effects of freezing, spine segment, repeat measurement, and individual cord properties on cord interstitial pressure." Spine (Phila Pa 1976) 34(4): 351-355. Basso, D. M., M. S. Beattie and J. C. Bresnahan (1995). "A sensitive and reliable locomotor rating scale for open field testing in rats." J Neurotrauma 12(1): 1-21. Basso, D. M., M. S. Beattie and J. C. Bresnahan (1996). "Graded histological and locomotor outcomes after spinal cord contusion using the NYU weight-drop device versus transection." Exp Neurol 139(2): 244-256. Bauman, J. M. and Y.-H. Chang (2010). "High-speed X-ray video demonstrates significant skin movement errors with standard optical kinematics during rat locomotion." Journal of Neuroscience Methods 186(1): 18-24. Bayly, P. V., E. E. Black, R. C. Pedersen, E. P. Leister and G. M. Genin (2006). "In vivo imaging of rapid deformation and strain in an animal model of traumatic brain injury." Journal of Biomechanics 39(6): 1086-1095. Bayly, P. V., T. S. Cohen, E. P. Leister, D. Ajo, E. C. Leuthardt and G. M. Genin (2005). "Deformation of the human brain induced by mild acceleration." J Neurotrauma 22(8): 845-856. Behrmann, D. L., J. C. Bresnahan, M. S. Beattie and B. R. Shah (1992). "Spinal cord injury produced by consistent mechanical displacement of the cord in rats: behavioral and histologic analysis." J Neurotrauma 9(3): 197-217. Berkowitz, M., P. O‘Leary, D. Kruse and C. Harvey (1998). Spinal cord injury: An analysis of medical and social costs. New York, Demos Medical Publishing Inc. Bilgen, M., R. Abbe and P. A. Narayana (2001). "Dynamic contrast-enhanced MRI of experimental spinal cord injury: in vivo serial studies." Magn Reson Med 45(4): 614622. Bilston, L. E. and L. E. Thibault (1995). "The mechanical properties of the human cervical spinal cord in vitro." Ann Biomed Eng 24(1): 67-74. Blight, A. (1988). "Mechanical factors in experimental spinal cord injury." J Am Paraplegia Soc 11(2): 26-34. Blight, A. R. (1991). "Morphometric analysis of a model of spinal cord injury in guinea pigs, with behavioral evidence of delayed secondary pathology." J Neurol Sci 103(2): 156171. 30  Chapter 1: Introduction  Blight, A. R. and V. Decrescito (1986). "Morphometric analysis of experimental spinal cord injury in the cat: the relation of injury intensity to survival of myelinated axons." Neuroscience 19(1): 321-341. Brainerd, E. L., D. B. Baier, S. M. Gatesy, T. L. Hedrick, K. A. Metzger, S. L. Gilbert and J. J. Crisco (2010). "X-ray reconstruction of moving morphology (XROMM): precision, accuracy and applications in comparative biomechanics research." J Exp Zool A Ecol Genet Physiol. Bresnahan, J. C., M. S. Beattie, F. D. Todd, 3rd and D. H. Noyes (1987). "A behavioral and anatomical analysis of spinal cord injury produced by a feedback-controlled impaction device." Exp Neurol 95(3): 548-570. Cao, Q., Y. P. Zhang, C. Iannotti, W. H. DeVries, X. M. Xu, C. B. Shields and S. R. Whittemore (2005). "Functional and electrophysiological changes after graded traumatic spinal cord injury in adult rat." Exp Neurol 191 Suppl 1: S3-S16. Carlson, G. D., C. D. Gorden, H. S. Oliff, J. J. Pillai and J. C. LaManna (2003). "Sustained spinal cord compression: part I: time-dependent effect on long-term pathophysiology." J Bone Joint Surg Am 85-A(1): 86-94. Carlson, G. D., K. E. Warden, J. M. Barbeau, E. Bahniuk, K. L. Kutina-Nelson, C. L. Biro, H. H. Bohlman and J. C. LaManna (1997). "Viscoelastic relaxation and regional blood flow response to spinal cord compression and decompression." Spine 22(12): 1285-1291. Chang, G. L., T. K. Hung, A. Bleyaert and P. J. Jannetta (1981). "Stress-strain measurement of the spinal cord of puppies and their neurological evaluation." J Trauma 21(9): 807810. Chang, G. L., T. K. Hung and W. W. Feng (1988). "An in-vivo measurement and analysis of viscoelastic properties of the spinal cord of cats." J Biomech Eng 110(2): 115-122. Cheng, S., E. C. Clarke and L. E. Bilston (2008). "Rheological properties of the tissues of the central nervous system: a review." Med Eng Phys 30(10): 1318-1337. Choo, A. M., J. Liu, M. Dvorak, W. Tetzlaff and T. R. Oxland (2008). "Secondary pathology following contusion, dislocation, and distraction spinal cord injuries." Exp Neurol 212(2): 490-506. Choo, A. M., J. Liu, C. K. Lam, M. Dvorak, W. Tetzlaff and T. R. Oxland (2007). "Contusion, dislocation, and distraction: primary hemorrhage and membrane permeability in distinct mechanisms of spinal cord injury." J Neurosurg Spine 6(3): 255-266. Clarke, E. C. and L. E. Bilston (2008). "Contrasting Biomechanics and Neuropathology of Spinal Cord Injury in Neonatal and Adult Rats following Vertebral Dislocation." J Neurotrauma 25(7): 817-832. Clarke, E. C., A. M. Choo, J. Liu, C. K. Lam, L. E. Bilston, W. Tetzlaff and T. R. Oxland (2008). "Anterior fracture-dislocation is more severe than lateral: a biomechanical and neuropathological comparison in rat thoracolumbar spine." J Neurotrauma 25(4): 371-383. Cusick, J. F., J. Myklebust, M. Zyvoloski, A. Sances, Jr., C. Houterman and S. J. Larson (1982). "Effects of vertebral column distraction in the monkey." J Neurosurg 57(5): 651-659.  31  Chapter 1: Introduction  Czyz, M., K. Scigala, W. Jarmundowicz and R. Beidzinski (2008). "The biomechanical analysis of the traumatic cervical spinal cord injury using finite element approach." Acta Bioeng Biomech 10(1): 43-54. Dimar, J. R., 2nd, S. D. Glassman, G. H. Raque, Y. P. Zhang and C. B. Shields (1999). "The influence of spinal canal narrowing and timing of decompression on neurologic recovery after spinal cord contusion in a rat model." Spine (Phila Pa 1976) 24(16): 1623-1633. Dohrmann, G. J. and M. M. Panjabi (1976). ""Standardized" spinal cord trauma: biomechanical parameters and lesion volume." Surg Neurol 6(5): 263-267. Dohrmann, G. J., M. M. Panjabi and D. Banks (1978). "Biomechanics of experimental spinal cord trauma." J Neurosurg 48(6): 993-1001. Drake, R., W. Vogl and A. Mitchell (2005). Gray‘s Anatomy for Students, Churchill Livingstone. Dryden, D. M., L. D. Saunders, B. H. Rowe, L. A. May, N. Yiannakoulias, L. W. Svenson, D. P. Schopflocher and D. C. Voaklander (2003). "The epidemiology of traumatic spinal cord injury in Alberta, Canada." Can J Neurol Sci 30(2): 113-121. Ellingson, B. M., J. L. Ulmer, S. N. Kurpad and B. D. Schmit (2008). "Diffusion tensor MR imaging in chronic spinal cord injury." AJNR Am J Neuroradiol 29(10): 1976-1982. Fehlings, M. G. and C. H. Tator (1995). "The relationships among the severity of spinal cord injury, residual neurological function, axon counts, and counts of retrogradely labeled neurons after experimental spinal cord injury." Experimental Neurology 132(2): 220228. Fiford, R. J. and L. E. Bilston (2005). "The mechanical properties of rat spinal cord in vitro." J Biomech 38(7): 1509-1515. Fiford, R. J., L. E. Bilston, P. Waite and J. Lu (2004). "A vertebral dislocation model of spinal cord injury in rats." J Neurotrauma 21(4): 451-458. Flanders, A. E., D. M. Schaefer, H. T. Doan, M. M. Mishkin, C. F. Gonzalez and B. E. Northrup (1990). "Acute cervical spine trauma: correlation of MR imaging findings with degree of neurologic deficit." Radiology 177(1): 25-33. Fukuda, S., T. Nakamura, Y. Kishigami, K. Endo, T. Azuma, T. Fujikawa, S. Tsutsumi and Y. Shimizu (2005). "New canine spinal cord injury model free from laminectomy." Brain Research. Brain Research Protocols 14(3): 171-180. Gareau, P. J., L. C. Weaver and G. A. Dekaban (2001). "In vivo magnetization transfer measurements of experimental spinal cord injury in the rat." Magn Reson Med 45(1): 159-163. Ghasemlou, N., B. J. Kerr and S. David (2005). "Tissue displacement and impact force are important contributors to outcome after spinal cord contusion injury." Exp Neurol 196(1): 9-17. Greaves, C. Y., M. S. Gadala and T. R. Oxland (2008). "A three-dimensional finite element model of the cervical spine with spinal cord: an investigation of three injury mechanisms." Ann Biomed Eng 36(3): 396-405. Gruner, J. A., A. K. Yee and A. R. Blight (1996). "Histological and functional evaluation of experimental spinal cord injury: evidence of a stepwise response to graded compression." Brain Res 729(1): 90-101.  32  Chapter 1: Introduction  Guha, A., C. H. Tator, L. Endrenyi and I. Piper (1987). "Decompression of the spinal cord improves recovery after acute experimental spinal cord compression injury." Paraplegia 25(4): 324-339. Hall, R. M., R. J. Oakland, R. K. Wilcox and D. C. Barton (2006). "Spinal cord-fragment interactions following burst fracture: an in vitro model." J Neurosurg Spine 5(3): 243250. Hardy, W. N., C. D. Foster, M. J. Mason, K. H. Yang, A. I. King and S. Tashman (2001). "Investigation of Head Injury Mechanisms Using Neutral Density Technology and High-Speed Biplanar X-ray." Stapp Car Crash J 45: 337-368. Hardy, W. N., M. J. Mason, C. D. Foster, C. S. Shah, J. M. Kopacz, K. H. Yang, A. I. King, J. Bishop, M. Bey, W. Anderst and S. Tashman (2007). "A study of the response of the human cadaver head to impact." Stapp Car Crash J 51: 17-80. Hardy, W. N., C. S. Shah, M. J. Mason, J. M. Kopacz, K. H. Yang, A. I. King, C. A. Van Ee, J. L. Bishop, R. F. Banglmaier, M. J. Bey, R. M. Morgan and K. H. Digges (2008). "Mechanisms of traumatic rupture of the aorta and associated peri-isthmic motion and deformation." Stapp Car Crash J 52: 233-265. Harrison, D. E. (1999). "A review of biomechanics of the central nervous system - Part 2: Spinal cord strains from postural loads." Journal of Manipulative and Physiological Therapeutics 22(5). Hayes, K. C. and B. A. Kakulas (1997). "Neuropathology of human spinal cord injury sustained in sports-related activities." J Neurotrauma 14(4): 235-248. Huang, W. L., K. J. George, V. Ibba, M. C. Liu, S. Averill, M. Quartu, P. J. Hamlyn and J. V. Priestley (2007). "The characteristics of neuronal injury in a static compression model of spinal cord injury in adult rats." Eur J Neurosci 25(2): 362-372. Hung, T. K., M. S. Albin, T. D. Brown, L. Bunegin, R. Albin and P. J. Jannetta (1975). "Biomechanical responses to open experimental spinal cord injury." Surg Neurol 4(2): 271-276. Hung, T. K. and G. L. Chang (1981). "Biomechanical and neurological response of the spinal cord of a puppy to uniaxial tension." J Biomech Eng 103(1): 43-47. Hung, T. K., G. L. Chang, J. L. Chang and M. S. Albin (1981). "Stress-strain relationship and neurological sequelae of uniaxial elongation of the spinal cord of cats." Surg Neurol 15(6): 471-476. Hung, T. K., H. S. Lin, L. Bunegin and M. S. Albin (1982). "Mechanical and neurological response of cat spinal cord under static loading." Surg Neurol 17(3): 213-217. Ichihara, K., T. Taguchi, I. Sakuramoto, S. Kawano and S. Kawai (2003). "Mechanism of the spinal cord injury and the cervical spondylotic myelopathy: new approach based on the mechanical features of the spinal cord white and gray matter." J Neurosurg 99(3 Suppl): 278-285. Ichihara, K., T. Taguchi, Y. Shimada, I. Sakuramoto, S. Kawano and S. Kawai (2001). "Gray matter of the bovine cervical spinal cord is mechanically more rigid and fragile than the white matter." J Neurotrauma 18(3): 361-367. Jakeman, L. B., D. M. McTigue, P. Walters and B. T. Stokes (2009). The Ohio State University ESCID Spinal Cord Contusion Model. Animal Models of Acute Neurological Injuries: 433-447.  33  Chapter 1: Introduction  Jones, C. F., S. G. Kroeker, P. A. Cripton and R. M. Hall (2008). "The effect of cerebrospinal fluid on the biomechanics of spinal cord: an ex vivo bovine model using bovine and physical surrogate spinal cord." Spine 33(17): E580-588. Joshi, M. and M. G. Fehlings (2002). "Development and characterization of a novel, graded model of clip compressive spinal cord injury in the mouse: Part 1. Clip design, behavioral outcomes, and histopathology." J Neurotrauma 19(2): 175-190. Kaneoka, K., K. Ono, S. Inami and K. Hayashi (1999). "Motion analysis of cervical vertebrae during whiplash loading." Spine (Phila Pa 1976) 24(8): 763-769; discussion 770. Kearney, P. A., S. A. Ridella, D. C. Viano and T. E. Anderson (1988). "Interaction of contact velocity and cord compression in determining the severity of spinal cord injury." J Neurotrauma 5(3): 187-208. Kennedy, P. and B. A. Rogers (2000). "Anxiety and depression after spinal cord injury: A longitudinal analysis." Archives of Physical Medicine and Rehabilitation 81(7): 932937. Kim, J. H., T. W. Tu, P. V. Bayly and S. K. Song (2009). "Impact speed does not determine severity of spinal cord injury in mice with fixed impact displacement." J Neurotrauma 26(8): 1395-1404. Kloos, A. D., L. C. Fisher, M. R. Detloff, D. L. Hassenzahl and D. M. Basso (2005). "Stepwise motor and all-or-none sensory recovery is associated with nonlinear sparing after incremental spinal cord injury in rats." Exp Neurol 191(2): 251-265. Kozlowski, P., D. Raj, J. Liu, C. Lam, A. C. Yung and W. Tetzlaff (2008). "Characterizing white matter damage in rat spinal cord with quantitative MRI and histology." J Neurotrauma 25(6): 653-676. Kroeker, S. G., P. L. Morley, C. F. Jones, L. E. Bilston and P. A. Cripton (2009). "The development of an improved physical surrogate model of the human spinal cord-Tension and transverse compression." Journal of Biomechanics 42(7): 878-883. Kulkarni, M. V., C. B. McArdle, D. Kopanicky, M. Miner, H. B. Cotler, K. F. Lee and J. H. Harris (1987). "Acute spinal cord injury: MR imaging at 1.5 T." Radiology 164(3): 837-843. Kwon, B. K., T. R. Oxland and W. Tetzlaff (2002). "Animal models used in spinal cord regeneration research." Spine 27(14): 1504-1510. Kwon, B. K., W. Tetzlaff, J. N. Grauer, J. Beiner and A. R. Vaccaro (2004). "Pathophysiology and pharmacologic treatment of acute spinal cord injury." Spine J 4(4): 451-464. Luan, F., K. H. Yang, B. Deng, P. C. Begeman, S. Tashman and A. I. King (2000). "Qualitative analysis of neck kinematics during low-speed rear-end impact." Clin Biomech (Bristol, Avon) 15(9): 649-657. Ma, M., D. M. Basso, P. Walters, B. T. Stokes and L. B. Jakeman (2001). "Behavioral and histological outcomes following graded spinal cord contusion injury in the C57Bl/6 mouse." Exp Neurol 169(2): 239-254. Maikos, J. T., Z. Qian, D. Metaxas and D. I. Shreiber (2008). "Finite element analysis of spinal cord injury in the rat." J Neurotrauma 25(7): 795-816. Maikos, J. T. and D. I. Shreiber (2007). "Immediate damage to the blood-spinal cord barrier due to mechanical trauma." J Neurotrauma 24(3): 492-507.  34  Chapter 1: Introduction  Maiman, D. J., J. Coats and J. B. Myklebust (1989). "Cord/spine motion in experimental spinal cord injury." J Spinal Disord 2(1): 14-19. Marciello, M. A., A. E. Flanders, G. J. Herbison, D. M. Schaefer, D. P. Friedman and J. I. Lane (1993). "Magnetic resonance imaging related to neurologic outcome in cervical spinal cord injury." Arch Phys Med Rehabil 74(9): 940-946. Margulies, S. S., D. F. Meaney, L. B. Bilston, L. Thibault, N. G. Campeau and S. J. Riederer (1992). In Vivo Motion of the Human Cervical Spinal Cord in Extension and Flexion. IRCOBI Conference, Verona, Italy. Martin, D. E., N. J. Greco, B. A. Klatt, V. J. Wright, W. J. Anderst and S. Tashman (2010). "Model-Based Tracking of the Hip: Implications for Novel Analyses of Hip Pathology." J Arthroplasty. Maynard, F. M., M. B. Bracken, C. Graham, J. F. Ditunno, W. H. Donovan, T. B. Ducker, S. L. Garber, R. J. Marino, S. L. Stover, C. H. Tator, R. L. Waters, J. E. Wilberger and W. Young (1997). "International Standards for Neurological and Functional Classification of Spinal Cord Injury." Spinal Cord 35(5): 266-274. Mendis, K. K., R. L. Stalnaker and S. H. Advani (1995). "A constitutive relationship for large deformation finite element modeling of brain tissue." J Biomech Eng 117(3): 279-285. Metz, G., I. Klusman, H. van de Meent, A. Curt, M. E. Schwab and V. Dietz (1998). Spinal Cord Injury in Rats and Humans. Metz, G. A., A. Curt, H. van de Meent, I. Klusman, M. E. Schwab and V. Dietz (2000a). "Validation of the weight-drop contusion model in rats: a comparative study of human spinal cord injury." J Neurotrauma 17(1): 1-17. Metz, G. A., D. Merkler, V. Dietz, M. E. Schwab and K. Fouad (2000b). "Efficient testing of motor function in spinal cord injured rats." Brain Res 883(2): 165-177. Ming, K. (2008). Towards In Vivo MRI Based Analysis of Spinal Cord Injury. Department of Electrical and Computer Engineering. Vancouver, The University of British Columbia: 101. Myklebust, J. B., D. J. Maiman and J. F. Cusick (1988). "Axial tension model of spinal cord injury." J Am Paraplegia Soc 11(2): 50-55. Nashmi, R. and M. G. Fehlings (2001). "Changes in axonal physiology and morphology after chronic compressive injury of the rat thoracic spinal cord." Neuroscience 104(1): 235-251. National Spinal Cord Injury Statistical Center, (2010). "Spinal Cord Injury Facts and Figures at a Glance." Retrieved February, 2010, from www.nscisc.uab.edu. Nelson, T. and P. A. Cripton (2008). "Inducing head motion with a novel helmet during head-first impact can mitigate neck injury metrices: an experimental proof-of-concept investigation using mechanical surrogates." Proceedings of the International Research Council on the Biomechanics of Impacts. Bern, Switzerland. Noble, L. J. and J. R. Wrathall (1987). "An inexpensive apparatus for producing graded spinal cord contusive injury in the rat." Exp Neurol 95(2): 530-533. Norenberg, M. D., J. Smith and A. Marcillo (2004). "The pathology of human spinal cord injury: defining the problems." J Neurotrauma 21(4): 429-440. Noyes, D. H. (1987a). "Correlation between parameters of spinal cord impact and resultant injury." Experimental Neurology 95(3): 535-547.  35  Chapter 1: Introduction  Noyes, D. H. (1987b). "Electromechanical impactor for producing experimental spinal cord injury in animals." Med Biol Eng Comput 25(3): 335-340. Nusholtz, G. S., P. Lux, P. Kaiker and M. A. Janicki (1984). Head impact response—skull deformation and angular accelerations. 28th Stapp Car Crash Conference, Chicago, Illinois, USA, Society of Automotive Engineers, Inc., Warrendale, Pennsylvania, USA. Oakland, R. J., R. M. Hall, R. K. Wilcox and D. C. Barton (2006). "The biomechanical response of spinal cord tissue to uniaxial loading." Proc Inst Mech Eng [H] 220(4): 489-492. Onifer, S. M., A. G. Rabchevsky and S. W. Scheff (2007). "Rat models of traumatic spinal cord injury to assess motor recovery." ILAR J 48(4): 385-395. Ono, K., K. Kaneoka, A. Wittek and J. Kajzer (1997). Cervical injury mechanism based on the analysis of human cervical vertebral motion and head-neck-torso kinematics during low speed rear impacts, Lake Buena Vista, Florida, 41st STAPP Car Crash Conference. Ozawa, H., T. Matsumoto, T. Ohashi, M. Sato and S. Kokubun (2004). "Mechanical properties and function of the spinal pia mater." J Neurosurg Spine 1(1): 122-127. Panjabi, M. and A. White, 3rd (1988). "Biomechanics of nonacute cervical spinal cord trauma." Spine 13(7): 838-842. Panjabi, M. M., M. Kifune, L. Wen, M. Arand, T. R. Oxland, R. M. Lin, W. S. Yoon and A. Vasavada (1995). "Dynamic canal encroachment during thoracolumbar burst fractures." J Spinal Disord 8(1): 39-48. Panjabi, M. M., T. Oda and J. L. Wang (2000). "The effects of pedicle screw adjustments on neural spaces in burst fracture surgery." Spine (Phila Pa 1976) 25(13): 1637-1643. Pearse, D. D., T. P. Lo, Jr., K. S. Cho, M. P. Lynch, M. S. Garg, A. E. Marcillo, A. R. Sanchez, Y. Cruz and W. D. Dietrich (2005). "Histopathological and behavioral characterization of a novel cervical spinal cord displacement contusion injury in the rat." J Neurotrauma 22(6): 680-702. Persson, C., S. W. McLure, J. Summers and R. M. Hall (2009). "The effect of bone fragment size and cerebrospinal fluid on spinal cord deformation during trauma: an ex vivo study." J Neurosurg Spine 10(4): 315-323. Pickett, G. E., M. Campos-Benitez, J. L. Keller and N. Duggal (2006). "Epidemiology of traumatic spinal cord injury in Canada." Spine (Phila Pa 1976) 31(7): 799-805. Pintar, F. A., M. B. Schlick, N. Yoganandan and D. J. Maiman (1996). "Instrumented artificial spinal cord for human cervical pressure measurement." Bio-Medical Materials and Engineering 6(3): 219-229. Ramon, S., R. Dominguez, L. Ramirez, M. Paraira, M. Olona, T. Castello and L. Garcia Fernandez (1997). "Clinical and magnetic resonance imaging correlation in acute spinal cord injury." Spinal Cord 35(10): 664-673. Raynor, R. B. and B. Koplik (1985). "Cervical cord trauma. The relationship between clinical syndromes and force of injury." Spine 10(3): 193-197. Rick Hansen Spinal Cord Injury Registry, (2006). "Spinal Cord Injury Facts and Statistics." Retrieved October, 2009, from http://rickhansenregistry.org/sites/rickhansen/files/Facts_and_Stats_V2006.pdf. Roos, P. J., M. L. Hull and S. M. Howell (2004). "How Cyclic Loading Affects the Migration of Radio-Opaque Markers Attached to Tendon Grafts Using a New Method: A Study 36  Chapter 1: Introduction  Using Roentgen Stereophotogrammetric Analysis (RSA)." Journal of Biomechanical Engineering 126(1): 62-69. Scheff, S. and K. N. Roberts (2009). Infinite Horizon Spinal Cord Contusion Model. Animal Models of Acute Neurological Injuries: 423-432. Scheff, S. W., A. G. Rabchevsky, I. Fugaccia, J. A. Main and J. E. Lumpp, Jr. (2003). "Experimental modeling of spinal cord injury: characterization of a force-defined injury device." J Neurotrauma 20(2): 179-193. Schwartz, E. D. and D. B. Hackney (2003). "Diffusion-weighted MRI and the evaluation of spinal cord axonal integrity following injury and treatment." Exp Neurol 184(2): 570589. Sekhon, L. H. and M. G. Fehlings (2001). "Epidemiology, demographics, and pathophysiology of acute spinal cord injury." Spine (Phila Pa 1976) 26(24 Suppl): S212. Shi, R. and A. R. Blight (1996). "Compression injury of mammalian spinal cord in vitro and the dynamics of action potential conduction failure." J Neurophysiol 76(3): 15721580. Smith, C. G. (1956). "Changes in length and position of the segments of the spinal cord with changes in posture in the monkey." Radiology 66(2): 259-266. Sparrey, C. J., A. M. Choo, J. Liu, W. Tetzlaff and T. R. Oxland (2008). "The distribution of tissue damage in the spinal cord is influenced by the contusion velocity." Spine (Phila Pa 1976) 33(22): E812-819. Stokes, B. T. and L. B. Jakeman (2002). "Experimental modelling of human spinal cord injury: a model that crosses the species barrier and mimics the spectrum of human cytopathology." Spinal Cord 40(3): 101-109. Stokes, B. T., D. H. Noyes and D. L. Behrmann (1992). "An electromechanical spinal injury technique with dynamic sensitivity." J Neurotrauma 9(3): 187-195. Stone, J. R., D. G. Rubin, A. O. Dialo, D. O. Okonkwo and G. A. Helm (2002). "Relationship of 40 kD, 10 kD, and 3 kD fluorescent indicators of alterted axolemmal permeability to impaired axoplasmic transport in traumatic axonal injury." J. Neurotrauma 19: 1278. Tator, C. H. (2006). "Review of treatment trials in human spinal cord injury: issues, difficulties, and recommendations." Neurosurgery 59(5): 957-982; discussion 982957. Tunturi, A. R. (1978). "Elasticity of the spinal cord, pia, and denticulate ligament in the dog." J Neurosurg 48(6): 975-979. Vaccaro, A. R., R. A. Lehman, Jr., R. J. Hurlbert, P. A. Anderson, M. Harris, R. Hedlund, J. Harrop, M. Dvorak, K. Wood, M. G. Fehlings, C. Fisher, S. C. Zeiller, D. G. Anderson, C. M. Bono, G. H. Stock, A. K. Brown, T. Kuklo and F. C. Oner (2005). "A new classification of thoracolumbar injuries: the importance of injury morphology, the integrity of the posterior ligamentous complex, and neurologic status." Spine (Phila Pa 1976) 30(20): 2325-2333. Weizenecker, J., B. Gleich, J. Rahmer, H. Dahnke and J. Borgert (2009). "Three-dimensional real-time in vivo magnetic particle imaging." Phys Med Biol 54(5): L1-L10. Wrathall, J. R., R. K. Pettegrew and F. Harvey (1985). "Spinal cord contusion in the rat: production of graded, reproducible, injury groups." Exp Neurol 88(1): 108-122. Young, W. (2002). "Spinal cord contusion models." Prog Brain Res 137: 231-255. 37  Chapter 1: Introduction  Young, W. (2009). MASCIS Spinal Cord Contusion Model. Animal Models of Acute Neurological Injuries: 411-421. Yuan, Q., L. Dougherty and S. S. Margulies (1998). "In vivo human cervical spinal cord deformation and displacement in flexion." Spine (Phila Pa 1976) 23(15): 1677-1683. Zou, H., J. P. Schmiedeler and W. N. Hardy (2007). "Separating brain motion into rigid body displacement and deformation under low-severity impacts." J Biomech 40(6): 11831191.  38  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model 2.1  Introduction Spinal cord injuries (SCIs) are extremely debilitating injuries and new injuries affect  approximately 12,000 people a year in the United States (National Spinal Cord Injury Statistical Center, 2010) and 1,700 per year in Canada (Dryden et al., 2003). While the majority of SCIs are due to spinal column fracture causing high-speed compression or sheering to the spinal cord, they can also occur from degenerative changes in the spinal column such as a bulging disc, ligamentous hypertrophy, and boney growths which cause chronic cord compression (Panjabi and White, 1988; Pickett et al., 2006). Residual cord compression can also result from a remaining fragment of bone lodged in the spinal canal after an acute SCI (Dimar et al., 1999). At very slow strain rates (i.e. quasi-static compression), injury severity has been shown to be correlated with compression depth and duration (Carlson et al., 2003; Carlson et al., 1997; Gruner et al., 1996; Hung et al., 1982). This is not surprising from an engineering standpoint, since increasing the compression depth would cause higher strains within the cord and increasing the duration of compression would cause relaxation in the tissue (Blight, 1988; Carlson et al., 1997). With an in vivo canine model it has been shown that the pressure within the spinal cord dissipates to 51% while under compression for five minutes and to 13% while under compression for three hours (Carlson et al., 1997). The animals in the five minute group fully recovered despite the rapid relaxation, but the animals in the three hour group did not fully recover. This study suggests that there is relaxation of the tissue inside of the spinal cord, and there may be a threshold of relaxation before persistent injury occurs. The ability to directly measure the tissue deformations during such an injury would help determine injury thresholds. To the author‘s knowledge, only one in vivo experimental study has been conducted which has directly measured internal deformation of the spinal cord during injury. Maiman et al. injected contrast agent into in vivo feline spinal cords, and imaged the cranial-caudal (Cr/Cd) displacement of the markers during various levels of distraction to the spinal column, _________________  A version of this chapter will be submitted for publication. Lucas E., Liu J., Tetzlaff W., and Cripton P.A. (2010) Radiography used to measure quasi-static spinal cord deformation in an in vivo rat model.  39  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model  and after a contusion injury (Maiman et al., 1989). Only one marker was injected per spinal cord segment, so the internal deformations of multiple points within one segment of the spinal cord could not be measured. Additionally, dorsal-ventral (D/V) deformations were not reported in this study. Theoretical, finite element, and experimental models have proposed that the mechanical response of the internal spinal cord tissue varies depending on location within the spinal cord which could result in different patterns of damage (Blight, 1988; Blight and Decrescito, 1986; Greaves et al., 2008; Ichihara et al., 2003; Ichihara et al., 2001; Maikos et al., 2008; Maikos and Shreiber, 2007; Sparrey et al., 2008). The heterogeneous anatomy of the spinal cord, including the long tracks of axons in the outer white matter surrounding neuronal cell bodies in the interior grey matter, also supports the theory that internal deformations during compression may vary depending on location within the cord. The ability to measure the internal deformations of multiple points within the grey and white matter of the spinal cord during experimental SCI would help explain differences between internal tissue deformations (i.e. white versus grey matter), define injury thresholds, and validate the deformations seen in existing models of the spinal cord such as finite element and surrogate cord models (Greaves et al., 2008; Kroeker et al., 2009; Maikos et al., 2008). The overall objective of this study was to develop a method of measuring internal deformations of an in vivo rat spinal cord using radiography and fiducial markers inside the cord. The specific objectives were to (1) determine the success rate of injecting fiducial markers into specific areas of the white and grey matter of the spinal cord, (2) measure the displacement of internal and surface fiducial markers during a quasi-static compression using radiography, and (3) assess marker migration with respect to the spinal cord tissue.  2.2  Methods  2.2.1 Surgical preparation and bead injection Sixteen Sprague Dawley male rats (360-430g) were used for this study, which was approved by the UBC Animal Care Centre and Ethics Committee (Appendix G). The rats were deeply anesthetized with an intramuscular injection of ketamine (72mg/kg; BimedaMTC, Cambridge, Ontario, Canada) and xylazine (10mg/kg; Bayer Inc., Etobicoke, Ontario, Canada). Top-ups of ketamine were administered for maintenance during surgery and x-ray 40  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model  imaging as per clinical signs of nociceptive reflexes. Following anesthetic, the animal was placed in a stereotaxic surgical frame (David Kopf Instruments, Tujunga, California) and a local anesthetic (0.5mL lidocaine with 2% epinephrine) was injected into the cervical musculature. A 200g weight was then hung on the tail of the rat to straighten the spinal column, further separate the vertebrae, create slight tension on the cord, and reduce spinal cord and spinal column movement due to breathing. An incision was then made along the C2-T1 vertebrae and the cervical musculature was split and retracted or cut to expose the cervical spinal column. The dorsal ligaments between the cervical vertebrae at the injection site were removed to expose the spinal cord. Full and/or partial laminectomies (see Sections 2.2.2 and 2.2.3) were performed to expose the dorsal surface of the spinal cord. Radio-opaque tantalum beads (260µm diameter) were used as the fiducial markers after previous evaluation and comparison to other powder and bead type markers (Appendix A). To inject the beads, a channel was first created in the spinal cord to the desired bead location using a 350µm diameter needle and the predetermined injection depth and angle (see Section 2.2.2). The bead was then placed on top of the channel and pushed to the bottom of the channel using a 30 gauge needle (300µm) with a flattened tip (Figure 2-1). The rat‘s physiologic reaction to the bead injections was monitored using a pulseoximeter to observe changes in heart rate and blood oxygenation. All animals were euthanized immediately after testing via intracranial perfusion with phosphate buffered saline (injection accuracy study) or via intraperitoneal overdose with 5% chloral hydrate (5%, 100 mg/kg, i.p.; BDH Chemicals, Toronto, Ontario, Canada) (quasi-static study).  A  B  C  Figure 2-1: Bead injection procedure. A channel was made through the cord to the desired depth (A), the bead was placed on the top of the channel (B), and then pushed to the bottom of the channel (C).  41  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model  2.2.2 Injection accuracy 2.2.2.1 Surgical procedure Seven anesthetized rats were used to assess the accuracy of injecting beads into the dorsal and ventral aspects of the white and grey matter in the spinal cord. The dorsal ligaments between C3-C8 vertebrae were removed to expose the spinal cord and a slight portion of the laminae on each side was shaved down to increase the cord exposure. The required injection depth and angle for these beads were determined using a rat spinal cord atlas (Paxinos and Watson, 1986). Figure 2-2 shows the atlas as well as a simplified diagram with the injection depth, angle, and lateral position used for injecting beads into the dorsal and ventral, white and grey matter. For the white matter injections, the lateral injection distance was measured from a vein that runs along the dorsal aspect of the cord approximately in the mid-sagittal plane. The lateral injection position for the grey matter injections was located at the dorsal horns of the grey matter which were visible under a surgical microscope. Two beads were injected into five spinal cord segments per animal, which resulted in fifteen injections trials per bead location. Two beads per spinal cord segment, five segments per animal, and seven animals resulted in a total of 70 bead injections, but ten of these beads were excluded due to using different injection parameters. The 60 remaining bead trials resulted in fifteen injection trials for each of the four desired bead locations. The rats were euthanized directly after the bead injections.  42  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model  Figure 2-2: Transverse cross-section of rat spinal cord segment C5 from a rat spinal cord atlas (Left) (Paxinos and Watson, 1986) and bead injection location, depth, and angle determined using the atlas (Right). Reprinted from Paxinos and Watson, 1986 with permission from Elsevier.  2.2.2.2 Bead location verification The spinal cords were immediately harvested after death and fixed for 24 hours in paraformadahyde to harden the tissue for easier dissection. After fixation, small transverse slices were cut from the spinal cord until the beads were visible. A small section containing the beads was cut and the bead location was analyzed under a microscope. Some beads were in different cutting planes, requiring a second section to be cut. The bead location was graded qualitatively by the surgeon and a second observer, where the marker was considered to be accurately placed if approximately 80% or more of the marker was in the appropriate tissue in the spinal cord. Otherwise the bead was considered to be misplaced. 2.2.3 Quasi-static compression 2.2.3.1 Surgical procedure Nine anesthetized rats were used for the quasi-static compression study. A full laminectomy of C5 and partial laminectomies of C4 and C6 were performed to expose the dorsal and lateral sides of the spinal cord in this area. Two beads were injected into the dorsal and ventral white matter between spinal cord segments C5 and C6 using the injection parameters shown in Figure 2-2. The internal beads will be referred to as the internal dorsal 43  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model  (ID) and internal ventral (IV) beads. Four additional beads were glued with cyanoacrylate to the pia mater (closely adhered to the surface of the spinal cord) through small holes made in the dura mater. Figure 2-3 shows the bead locations in a lateral radiograph. The surface beads were located cranial and caudal to the injection site, with two dorsal and two ventral beads. The surface beads will be referred to as the cranial dorsal (CrD), cranial ventral (CrV), caudal dorsal (CdD), and caudal ventral (CdV) beads. To avoid applying excess glue to the cord and surrounding tissue, the beads were adhered to the flattened tip of a needle with petroleum jelly, dipped into liquid cyanoacrylate, and then carefully touched to the spinal cord through hole in the dura. The bead adhered directly to the surface of the spinal cord without the need for drying the cord. The transverse process of C5 was filed down to create a flat surface for gluing a plastic ―T‖ shaped piece with three glued reference beads to detect vertebral column motion. The ―T‖ consisted of a 1mm bead on the bottom to measure spinal column translation in the D/V and Cr/Cd directions and two 400µm tantalum beads on the top to detect vertebral column rotation (see Figure 2-3). Before x-ray procedures, the scapulae were removed as this was a non-survival study and they were shown to interfere with the visibility of the cervical spinal cord markers in the x-ray images.  Impactor  Impactor  ―T‖  Caudal Surface Beads  Cranial Surface Beads  Spinal Column Reference Bead  A  Internal Beads  Cr (+)  Cranial Surface Beads  D (+) Cd V  B  Spinal Column Reference Bead  Caudal Surface Beads  Internal Beads  Figure 2-3: Lateral radiograph showing bead positioning in the cervical spinal cord of an anesthetized rat (A) and associated diagram (B). Internal beads were in the dorsal (D) and ventral (V) internal aspects of the cord. Surface beads were on the ventral and dorsal aspects of the surface of the cord both cranial (Cr) and caudal (Cd) to the internal beads. Horizontal arrows point to the impactor beads and vertical arrows point to the T-beads (A). The locations of the impactor and ―T‖ piece are shown in the diagram for clarity (B). Note that the internal beads are shown visible in the lateral view in the diagram (B) but in actuality they were inside of the cord and would not be visible from the surface without x-ray. The locations of the internal and surface beads are demonstrated with transverse cross-sections in the diagram.  44  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model  2.2.3.2 Spinal cord compression procedure An impactor was fabricated by gluing a 12.5cm glass micropipette to a 7.5x5.5x3.3mm piece of polycarbonate with the 7.5x5.5mm surface used for the impacting surface. To ensure the impacting surface was perpendicular to the pipette, the pipette was held in the injection holder of the stereotaxic frame which has precise angular control in the x, y, and z directions. The pipette was then slowly lowered down until coming into contact with the polycarbonate base which was resting on a flat surface of the stereotaxic frame. The two pieces were then glued in place and left to dry. Three 400µm tantalum beads were also glued to the front surface of the impactor to detect impactor displacement in the x-ray images (Figure 2-3). The impactor was held with the injection holder of the stereotaxic frame and was manually lowered (using the attached micrometer (±0.05mm)) to the surface of the spinal cord until visually touching the dura. This initial position was used for each subsequent compression. Three compressions were performed on each rat by manually lowering the impactor to a depth of 0.5mm, 2.0mm, and 3.0mm at a rate of approximately 0.1mm/s. The impactor was returned to its original starting position between each compression. Lateral radiographs were taken before, during, and after each compression. The animal was then euthanized via intraperitoneal overdose with 5% chloral hydrate. The spinal cord was immediately harvested, fixed in 4% Paraformaldehyde, and sectioned in transverse slices to determine the internal bead locations in the dorsal or ventral, white or grey matter (as described with the injection accuracy study). 2.2.3.3 Bead displacement measurement Bead positions were measured before, during, and after compression using MATLAB‘s Image Processing toolbox (The MathWorks, Natick, MA) to analyze the digital x-ray images. A grey threshold was first applied to the image to demarcate the beads from the surrounding structures (Figure 2-4B) and the weighted centroid of each bead was found using the regionprops and weightedcentroid commands in MATLAB. The reference bead is not visible in the threshold image in Figure 2-4B because a different threshold value was needed for the reference bead due to a different level of contrast than the spinal cord beads. If the beads could not be individually separated (with grey thresholding) from other beads or 45  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model  structures due to overlap, the center of the bead was determined visually and manually selected in MATLAB. Beads which were completely obstructed from view were excluded from analysis. After selection of the bead centroids was complete, the centroids were plotted on each radiographic image and were visually verified (Figure 2-4C).  A  Spinal cord beads  Stereotaxic frame  B  Cranium C2  D  Frame reference  Spinal cord beads Spinal column reference bead  C  Figure 2-4: Bead centroids selected using MATLAB. The frame and bead centroid selections from one rat are shown (A). The bead centroids were selected by first applying a grey threshold (B) and then using MATLAB to calculate the weighted centroid of the bead. The bead centroid locations were plotted on the radiograph and verified after each test (C). The frame reference was manually selected as the tip of a thread (D) on the stereotaxic frame.  Bead displacements in the vertical and horizontal directions of the x-ray images were considered to be D/V and Cr/Cd displacements, respectively. The testing surface, x-ray equipment, and x-ray cassettes were all level to the ground, thus the x-ray cassette acted as a consistent coordinate system for all measurements. Reference points on the stereotaxic frame and on the spinal column were used for measuring individual bead displacements. The frame reference point was the tip of a thread on the stereotaxic frame which was manually selected 46  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model  in radiographic images (Figure 2-4D). The spinal column reference point was the 1mm spinal column reference bead and was selected using the same weighted centroid technique described above (Figure 2-4C). The impactor beads and spinal column reference bead locations were measured relative to the frame reference. The impactor beads were used to determine the pixel/mm conversion factor. This was calculated by dividing the average impactor displacement (of all three impactor beads, before/during and during/after compression) in pixels by the actual amount that the impact was displaced using the micrometer (0.5mm, 2.0mm, or 3.0mm) resulting in a separate conversion factor for each level of compression. The spinal column reference bead was used to determine spinal column translation during compression. The spinal cord beads, impactor beads, and top two ‗T‘ bead locations were all measured relative to the spinal column reference bead to account for any spinal column displacement during compression. Since spinal column movement during compression would affect the overall compression depth, the impactor beads were used to determine the ―effective compression‖ to the spinal cord, defined as the intended compression (0.5mm, 2.0mm, or 3.0mm) minus the spinal column displacement. The internal spinal cord strain in the D/V direction was estimated as the change in distance between the two internal beads before and during compression divided by the distance before compression. The overall spinal cord strain was estimated by dividing the effective compression by the D/V diameter of the spinal cord. For this study the diameter of each rat‘s spinal cord was not directly measured, but it was estimated to be 3.0mm using a rat spinal cord atlas (Paxinos and Watson, 1986). 2.2.3.4 Bead migration assessment Since the internal beads were not directly glued to the spinal cord, it was necessary to consider the possibility of bead migration with respect to the spinal cord tissue. If the bead migrated during compression due to rupturing or deforming the tissue, it was assumed that it would not return to its original position after compression (thus it would have a postcompression displacement). Post-compression displacement could also be due to real residual cord deformation from the compression itself, so a method of separating bead migration from real residual cord deformation was needed. It was assumed that residual cord deformation would be apparent in post-compression displacements of the surface beads since the surface 47  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model  beads were directly glued to the spinal cord surface. Thus, post-compression displacements of the internal beads exceeding the surface beads were attributed to bead migration. 2.2.3.5 Measurement accuracy and repeatability To estimate the accuracy of the bead displacement measurements, the displacements of each of the impactor beads during compression were separately compared to the actual displacement measured with the micrometer (0.5, 2.0, or 3.0mm ± 0.05mm) and the absolute values of the differences were averaged. To estimate the accuracy of the post-compression displacement measurements, the absolute value of the post-compression displacements of the impactor beads were averaged. Since the impactor was returned to its original position after each compression, the post-compression position of the impactor beads should be zero. The precision was defined as the standard deviation of these measurements. The repeatability of finding the bead centroid locations was determined by repeating the bead centroid selections of one radiograph five times. A radiograph taken during a 3mm compression which required some manual bead centroid selections was chosen as a ―worstcase‖ scenario. The repeatability was defined as the mean standard deviation of each spinal cord bead position. Variability in the measured bead positions due to removing and replacing the cassette between x-ray images (to read the images in the digital reader) was determined by taking a series of five radiographs of one rat with bead inclusions. The impactor was lowered to touch the dorsal surface of the spinal cord before taking these images, but the spinal cord was not compressed between images. The variability of the spinal cord bead positions was defined as the mean standard deviation of bead positions relative to the spinal column reference bead. The variability of the impactor bead positions relative to the frame reference was calculated in the same manner to assess the integrity of the chosen frame reference point. 2.2.3.6 Statistical analysis A repeated measures analysis of variance was performed on the spinal cord bead displacements during compression and post-compression using bead position (ID, IV, CrD, CrV, CdD, CdV) and compression level (0.5, 2.0, 3.0mm) as two factors. A Student Newman  48  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model  Keuls post-hoc test was performed to identify significant differences between beads. Pvalues less than 0.05 were considered significant.  2.3  Results  2.3.1 Injection accuracy The rats showed no apparent respiratory or cardiac changes during any of the bead injections. Figure 2-5 shows the approximate bead locations for the fifteen injection trials in the dorsal and ventral aspects of the white and grey matter. Overall, 85% of the beads were in the correct location within the dorsal or ventral, white or grey matter. The injections in the dorsal grey matter were the most successful with 93% correct placement. The ventral grey matter injections had an 87% success rate, and the dorsal and ventral white matter injections were 80% successful. Figure 2-6 shows an example of a bead successfully injected into the dorsal grey matter. The dorsal white matter placement errors were typically due to the bead being injected too deep and the ventral white matter errors were due to the beads not being injected deep enough. Dorsal white matter injections Ventral white matter injections Dorsal grey matter injections Ventral grey matter injections Intended bead locations  Figure 2-5: Approximate location of the beads for the injection accuracy study. The shading within each group represents injections made in different spinal cord segments of the same rat.  49  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model  Figure 2-6: Microscope image of a 260µm tantalum bead injected into dorsal grey matter of the spinal cord. The slight hemorrhage next to the bead is due to injection of another bead out of the plane of this image.  2.3.2 Quasi-static compression Harvested spinal cords from the quasi-static compression study showed the ID beads were in the dorsal white matter and the IV beads were in the ventral white matter, as intended. Some data exclusions from the nine rats tested were necessary. In one animal, the internal beads were covered by the spinal column reference bead during the 3.0mm compression so the compression displacement analysis was excluded for these beads. For a second animal the cranial ventral bead came off before the compressions and thus could not be used for analysis. For three out of the 27 total compressions, the centroids of two of the beads were manually selected (as discussed in the methods) due to the inability to separate the beads using automatic thresholding. 2.3.2.1 Bead displacements 2.3.2.1.1  Spinal cord  Figure 2-7 and Figure 2-8 show the D/V and Cr/Cd displacements of the spinal cord beads within the 0.5, 2.0, and 3.0mm compression groups. In the D/V direction, all of the dorsal beads (surface and internal) moved significantly more than all of the ventral beads for all three compression levels (p=0.0001-0.04). On average, the ID bead moved 110%, 98%, and 65% more than the IV bead for the respective 0.5mm, 2.0mm, and 3.0mm compressions (p=0.0001). The ID bead also moved farther than the CdD bead for the 3.0mm compression 50  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model  (p=0.0001) and the CrD bead for the 0.5mm (p=0.0001) and 3.0mm (p=0.01) compressions. In the Cr/Cd direction, the internal beads moved more than all of the surface beads for all compression levels (p=0.0001-0.046). The ID bead moved more than the IV bead for the 2.0mm (p=0.008) and 3.0mm (p=0.0002) compressions. Additional data can be found in Appendix B. 0.4  *  *  0.5mm 2mm  *  Displacement (mm)  -0  3mm  -0.4  -0.8  -1.2  *  *  -1.6  *  * -2  CrD  CrV  ID IV Bead Location  CdD  CdV  Figure 2-7: Spinal cord bead displacements in the D/V direction for each compression group (avg ± std). Negative values represent displacements in the ventral direction. Stars indicate significance in all three compression groups unless otherwise noted with the individual compression group markers (diamond, square, or circle). Significance between the surface beads is not marked, since comparing displacements between surface beads was not an objective of this study.  51  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model  2  *  Displacement (mm)  1.6  0.5mm 2mm 3mm  *  1.2  * 0.8 0.4 0 -0.4  CrD  CrV  ID IV Bead Location  CdD  CdV  Figure 2-8: Spinal cord bead displacements in the Cr/Cd direction for each compression group (avg ± std). Negative values represent displacements in the caudal direction. Stars indicate significance in all three compression groups unless otherwise noted with the individual compression group markers (diamond, square, or circle). Significance between the surface beads is not marked, since comparing displacements between surface beads was not an objective of this study.  The average internal spinal cord strain (defined as the change in distance between the ID and IV beads divided by the original distance between these beads) in the D/V direction was 15% ± 9%, 34% ± 13%, and 41% ± 11% for 0.5mm, 2.0mm, and 3.0mm compressions, respectively. The strains from the 2.0mm and 3.0mm compression groups were significantly larger than the 0.5mm group (p=0.001 and 0.0002 respectively), but they were not significantly different from each other (p=0.09). The overall spinal cord strain, defined as the effective compression divided by the 3.0mm D/V diameter from the rat spinal cord atlas, was estimated to be 17% ± 9%, 32% ± 13%, and 45% ± 9% for 0.5, 2.0, and 3.0mm compressions. These strains are similar to the internal strains, but statistical analysis was not performed due to the limitations in assuming the spinal cord diameter. 2.3.2.1.2  Spinal column  Overlays of radiographs before, during, and after a typical 3.0mm compression are shown in Figure 2-9. Figure 2-9A shows the bead displacements with respect to the spinal column reference bead and Figure 2-9B with respect to the frame reference. The displacement of the spinal column reference bead evident in Figure 2-9 demonstrates that the 52  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model  spinal column moved during compression, reducing the amount of compression seen by the spinal cord (effective compression). The average spinal column bead displacement, reflecting spinal column translation, was 0.04mm ± 0.26mm, 0.94 ± 0.32mm, and 1.56mm ± 0.28mm for compression groups 0.5mm, 2.0mm, and 3.0mm respectively. The effective compressions significantly increased between the 0.5mm and 2.0mm groups (p=0.0005) and the 2.0mm and 3.0mm groups (p=0.009), with averages of 0.56mm ± 0.25mm, 1.03 ± 0.33mm, and 1.35 ± 0.28 for the 0.5mm, 2.0mm, and 3.0mm compression groups respectively. The average displacement of the upper T-beads relative to the spinal column reference bead was 0.08mm ± 0.1mm, which reflects rotation or out of plane movement of the spinal column.  Impactor  Impactor  Spinal Column (Before&After)  C3  A  Spinal Column Reference Bead Spinal Column  C3 Cr (+)  B Spinal Column (During)  D (+) Cd V  Figure 2-9: Overlay of spinal cord bead displacements before (yellow), during (red), and after (blue) a 3.0mm compression. Radiographs are overlaid with respect to the spinal column reference bead (A) and the frame reference (B). Overlay A shows the ―effective compression‖ to the cord and B shows the actual displacement of the impactor and the displacement of the spinal column during compression.  2.3.2.2 Bead migration The post-compression displacements of the ID bead were not significantly different than the surface dorsal beads (p=0.08-0.5 (D/V) and 0.7 (Cr/Cd)) and IV beads were not significantly different than the ventral surface beads (p=0.3 (D/V) and 0.7 (Cr/Cd)). The average absolute value of the differences in post-compression displacements of the internal beads compared to the surface beads were 0.05mm ± 0.05mm (ID) and 0.04mm ± 0.03mm (IV) in the D/V direction and 0.06mm ± 0.04 (ID) and 0.05 ± 0.04 (IV) in the Cr/Cd direction. Additional data is available in Appendix B. 53  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model  2.3.2.3 Measurement accuracy and repeatability The average absolute value of the difference between the measured impactor bead displacements and the actual impactor displacement during compression was 0.024mm (representing accuracy) and the standard deviation was 0.02mm (representing precision). The average absolute value of the measured post-compression displacements of the impactor beads after compression was 0.092mm (representing accuracy) and the standard deviation was 0.05mm (representing precision). The repeatability test, performed by repeating the bead centroid selections of one radiograph five times, showed no difference in the centroid locations for all of the beads and the location of the frame reference. For the variability study, performed by taking five radiographs of one rat with bead inclusions, the mean standard deviation of the spinal cord bead positions relative to the reference bead was 1.01 pixels in the D/V direction and 0.57 in the Cr/Cd direction. Using the average conversion factor from all of the tests (0.09 mm/pixel) these values were 0.09mm and 0.05mm respectively. The variation of the impactor beads relative to the frame reference was 0.67 pixels (0.06mm) (D/V) and 0.49 pixels (0.045mm) (Cr/Cd).  2.4  Discussion To the author‘s knowledge, this was the first time internal deformations within one  spinal cord segment have been measured in vivo during an experimental SCI. The injection accuracy results showed that fiducial markers could successfully be injected into specific locations within the spinal cord. The dorsal grey matter injections were the most successful, most likely because the dorsal horn is visible from the dorsal surface of the spinal cord, enabling controlled injection directly into this region. The variability of the white matter injections possibly resulted from measuring the lateral injection locations from the edge of the dorsal vein which was not always perfectly centralized and was not present in some animals (requiring an estimation of the center of the cord). A more accurate method of measuring the injection location, such as measurement from the visible dorsal horn position, would likely increase the injection accuracy. Another cause of variation could be due to variations in the sizes of the spinal cords. Since the spinal cord diameter increases as the rat 54  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model  matures to adulthood, this variation could be reduced by strictly regulating the age of the animals (Young, 2002) . The bead displacement results indicate that the internal spinal cord moves differently depending on location. The ID bead had larger displacement than the IV bead, which is intuitive since the ID bead was closer to the site of compression. The internal beads also had different displacements than the surface beads, highlighting the importance of tracking internal deformations and not solely relying on surface deformations. The presence of Cr/Cd movement of the internal beads which exceeds the surface beads suggests that the internal spinal cord tissue displaces more than the surface of the cord during compression. This theory of internal tissue displacement has also been proposed by experimental and theoretical studies (Blight, 1988; Blight and Decrescito, 1986; Maiman et al., 1989). Maiman et al. reported Cr/Cd spinal cord displacement with a drop weight resting on the spinal cord, though there were no markers on the surface of the spinal cord making it unclear if the deformation was localized deformation or overall cord translation. Since the spinal cord tissue deformation was being measured with fiducial markers, it was necessary to assess bead migration with respect to the tissue. This is a difficult task since the tissue movement and bead movement cannot be simultaneously measured. It was assumed that bead migration would be due to the bead pushing through the spinal cord tissue during compression and deforming or rupturing through the tissue. Any migration of the bead in this sense would show up as a post-compression displacement. The internal bead postcompression displacements were compared to the glued surface beads to account for real residual deformation of the tissue due to compression. This approach assumes that the surface and interior aspects of the spinal cord would have the same residual deformation after compression. The surface beads were glued to the pia mater which is directly coupled to the spinal cord surface and has been shown to restore the shape of the cord after compression (Ozawa et al., 2004). It is possible that this membrane would cause the surface of the cord to have less residual deformation than the internal spinal cord. If this was true, the values of migration found in this study would be an overestimate. It was also seen in this study that the internal beads moved more than the surface beads in the Cr/Cd direction which may also cause greater internal residual deformation. This migration assessment could possibly be 55  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model  confounded if the beads backed out of the injection track during compression. However, it was assumed that the hydrostatic pressure of the surrounding tissue would cause the injection track to close. A minimal residual injection track was confirmed through spinal cord dissections. Since this was a quasi-static study, inertial effects of the bead during compression were assumed to be minimal. Inertial effects should be addressed in dynamic impact studies. Additional investigation into the possible effects of bead presence in the spinal cord on its motion could be assessed with the use of finite element modeling (Chapter 3). The rotation and translation of the spinal column during compression was a limitation that increased the variability between tests. While it is well known that the spinal column should be clamped for dynamic injuries (Choo et al., 2009; Dohrmann et al., 1978), it was thought that the added weight to the tail and the slow compression rate in this study would limit spinal column motion due to breathing or compression. The prevalence of spinal column motion in this quasi-static study (which was controlled for by using a spinal column reference point) emphasizes the importance of clamping the spinal column even for very slow compression injuries. The spinal column displacement resulted in different absolute compression levels than intended, but the effective compression increased with increasing intended compression so it was still appropriate to compare displacements between these groups. Performing three compressions on the same spinal cord was a potential limitation since the spinal cord may suffer some structural damage after each compression. Hung et al. showed that the spinal cord exhibits the same force-displacement response for repeated quasi-static compressions under 36% strain which is greater than the first two compression groups in the present study (Hung et al., 1982). A limitation of the surface strain calculation was that an atlas was used to estimate the spinal cord D/V diameter since the cord diameters were not directly measured. This calculation also assumed that the impactor, aligned by eye with the spinal cord, was perfectly perpendicular to the spinal cord surface. The internal spinal cord strain measurement assumed that the beads were in the same sagittal and transverse planes and it did not account for any lateral displacement of the beads. While these limitations likely caused error in the strain measurements, this approach for measuring strain suggests one way of analyzing the 56  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model  bead deformations. In the future, lateral bead displacement could be measured using biplanar x-ray, and internal and surface strains could be compared to histological outcome for determining tissue injury thresholds. The variation in bead position noticed between sequential radiographic images could have been due to breathing effects, switching the x-ray cassettes between each image, and variations in the global reference point selection. While the variance was slightly higher than desired, the overall trends in the data were most likely valid due to the statistical significance and repeatability of results between animals. For future more precise studies, it is recommended that the spinal column is clamped to reduce breathing effects. Additionally, a bead should be attached to the frame and used as the global reference for increased repeatability of this location. A precisely positioned cassette holder should also be implemented and the same cassette should be used for each image to avoid any slight difference in geometry between the cassettes. There are many conceivable applications of this methodology including a variety of studies in basic science, engineering, and neuroscience. The most direct application is the improvement of existing finite element and surrogate spinal cord models, since the current methodology allows for measurement of in vivo internal deformations and strains. This approach could also be used to link mechanical variables of experimental SCIs to histological and functional outcomes. While this study was limited to quasi-static compressions, deformations due to high-speed contusions could be imaged by triggering the x-ray source at maximum compression or through the use of high-speed x-ray (Chapter 3) (Hardy et al., 2007; Ono et al., 1997).  2.5  Conclusions Fiducial markers were successfully injected into the dorsal and ventral white and grey  matter of in vivo rat spinal cords. The markers were easily seen with x-ray and displacements during quasi-static compression were measured in radiographic images. Results showed differences in displacements of beads inside of the spinal cord and on the surface of the cord, suggesting that the spinal cord moves in a complex non-uniform manner and that internal deformations should be considered. Migration of the markers with respect to the spinal cord 57  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model  tissue appeared to be minimal, indicating that they are appropriate for measuring spinal cord deformation. Future applications of this methodology include imaging internal spinal cord relaxation during sustained compression, measurement of dynamic spinal cord deformation using high-speed x-ray (Chapter 3), validation of finite element and surrogate cord models, and comparison of localized spinal cord deformations with histological and functional outcomes.  58  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model  2.6  References  Blight, A. (1988). "Mechanical factors in experimental spinal cord injury." J Am Paraplegia Soc 11(2): 26-34. Blight, A. R. and V. Decrescito (1986). "Morphometric analysis of experimental spinal cord injury in the cat: the relation of injury intensity to survival of myelinated axons." Neuroscience 19(1): 321-341. Carlson, G. D., C. D. Gorden, H. S. Oliff, J. J. Pillai and J. C. LaManna (2003). "Sustained spinal cord compression: part I: time-dependent effect on long-term pathophysiology." J Bone Joint Surg Am 85-A(1): 86-94. Carlson, G. D., K. E. Warden, J. M. Barbeau, E. Bahniuk, K. L. Kutina-Nelson, C. L. Biro, H. H. Bohlman and J. C. LaManna (1997). "Viscoelastic relaxation and regional blood flow response to spinal cord compression and decompression." Spine 22(12): 1285-1291. Choo, A. M., J. Liu, Z. Liu, M. Dvorak, W. Tetzlaff and T. R. Oxland (2009). "Modeling spinal cord contusion, dislocation, and distraction: characterization of vertebral clamps, injury severities, and node of Ranvier deformations." J Neurosci Methods 181(1): 6-17. Dimar, J. R., 2nd, S. D. Glassman, G. H. Raque, Y. P. Zhang and C. B. Shields (1999). "The influence of spinal canal narrowing and timing of decompression on neurologic recovery after spinal cord contusion in a rat model." Spine (Phila Pa 1976) 24(16): 1623-1633. Dohrmann, G. J., M. M. Panjabi and D. Banks (1978). "Biomechanics of experimental spinal cord trauma." J Neurosurg 48(6): 993-1001. Dryden, D. M., L. D. Saunders, B. H. Rowe, L. A. May, N. Yiannakoulias, L. W. Svenson, D. P. Schopflocher and D. C. Voaklander (2003). "The epidemiology of traumatic spinal cord injury in Alberta, Canada." Can J Neurol Sci 30(2): 113-121. Greaves, C. Y., M. S. Gadala and T. R. Oxland (2008). "A three-dimensional finite element model of the cervical spine with spinal cord: an investigation of three injury mechanisms." Ann Biomed Eng 36(3): 396-405. Gruner, J. A., A. K. Yee and A. R. Blight (1996). "Histological and functional evaluation of experimental spinal cord injury: evidence of a stepwise response to graded compression." Brain Res 729(1): 90-101. Hardy, W. N., M. J. Mason, C. D. Foster, C. S. Shah, J. M. Kopacz, K. H. Yang, A. I. King, J. Bishop, M. Bey, W. Anderst and S. Tashman (2007). "A study of the response of the human cadaver head to impact." Stapp Car Crash J 51: 17-80. Hung, T. K., H. S. Lin, L. Bunegin and M. S. Albin (1982). "Mechanical and neurological response of cat spinal cord under static loading." Surg Neurol 17(3): 213-217. Ichihara, K., T. Taguchi, I. Sakuramoto, S. Kawano and S. Kawai (2003). "Mechanism of the spinal cord injury and the cervical spondylotic myelopathy: new approach based on the mechanical features of the spinal cord white and gray matter." J Neurosurg 99(3 Suppl): 278-285. Ichihara, K., T. Taguchi, Y. Shimada, I. Sakuramoto, S. Kawano and S. Kawai (2001). "Gray matter of the bovine cervical spinal cord is mechanically more rigid and fragile than the white matter." J Neurotrauma 18(3): 361-367.  59  Chapter 2: Radiography Used to Measure Quasi-Static Spinal Cord Deformation in an In Vivo Rat Model  Kroeker, S. G., P. L. Morley, C. F. Jones, L. E. Bilston and P. A. Cripton (2009). "The development of an improved physical surrogate model of the human spinal cord-Tension and transverse compression." Journal of Biomechanics 42(7): 878-883. Maikos, J. T., Z. Qian, D. Metaxas and D. I. Shreiber (2008). "Finite element analysis of spinal cord injury in the rat." J Neurotrauma 25(7): 795-816. Maikos, J. T. and D. I. Shreiber (2007). "Immediate damage to the blood-spinal cord barrier due to mechanical trauma." J Neurotrauma 24(3): 492-507. Maiman, D. J., J. Coats and J. B. Myklebust (1989). "Cord/spine motion in experimental spinal cord injury." J Spinal Disord 2(1): 14-19. National Spinal Cord Injury Statistical Center, (2010). "Spinal Cord Injury Facts and Figures at a Glance." Retrieved February, 2010, from www.nscisc.uab.edu. Ono, K., K. Kaneoka, A. Wittek and J. Kajzer (1997). Cervical injury mechanism based on the analysis of human cervical vertebral motion and head-neck-torso kinematics during low speed rear impacts, Lake Buena Vista, Florida, 41st STAPP Car Crash Conference. Ozawa, H., T. Matsumoto, T. Ohashi, M. Sato and S. Kokubun (2004). "Mechanical properties and function of the spinal pia mater." J Neurosurg Spine 1(1): 122-127. Panjabi, M. and A. White, 3rd (1988). "Biomechanics of nonacute cervical spinal cord trauma." Spine 13(7): 838-842. Paxinos, G. and C. Watson (1986). The Rat Brain in Stereotaxic Coordinates: Second Edition, Academic Press, Inc. Pickett, G. E., M. Campos-Benitez, J. L. Keller and N. Duggal (2006). "Epidemiology of traumatic spinal cord injury in Canada." Spine (Phila Pa 1976) 31(7): 799-805. Sparrey, C. J., A. M. Choo, J. Liu, W. Tetzlaff and T. R. Oxland (2008). "The distribution of tissue damage in the spinal cord is influenced by the contusion velocity." Spine (Phila Pa 1976) 33(22): E812-819. Young, W. (2002). "Spinal cord contusion models." Prog Brain Res 137: 231-255.  60  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model 3.1  Introduction Spinal cord injuries (SCIs) are extremely debilitating injuries and affect approximately  12,000 people a year in the United States (National Spinal Cord Injury Statistical Center, 2010) and 1,700 per year in Canada (Dryden et al., 2003). It is well known that the severity of SCI is correlated with the biomechanics of impact, specifically velocity and compression (Kearney et al., 1988). Limited data exists however on the biomechanics of the spinal cord during impact, specifically how the spinal cord deforms during impact. Understanding how the cord deforms could help define the mechanical injury tolerances of the spinal cord tissue. Some researchers have studied the deformation of the spinal cord during SCI using surrogate spinal cords (Bilston and Thibault, 1997; Kroeker et al., 2009; Saari et al., 2006; Van Toen (Nee Greaves) et al., 2008) and others have modeled it with finite element models (FEMs) (Czyz et al., 2008; Greaves et al., 2008; Li and Dai, 2009; Maikos et al., 2008). Such models require accurate spinal cord material properties which have often been obtained from ex vivo mechanical testing of human and animal spinal cords (Bilston and Thibault, 1995; Fiford and Bilston, 2005; Oakland et al., 2006). The material properties of the spinal cord have been shown to change quickly after death, necessitating in vivo testing for accurate representation of the in vivo response of the spinal cord during SCI (Hung and Chang, 1981; Oakland et al., 2006). In vivo animal models have been widely used to study SCIs, but research is typically focused on post-injury results such as histology, behavioral assessment, and the effectiveness of treatments. There is a paucity of studies evaluating the in vivo spinal cord biomechanics during injury. The various components of the spinal cord (i.e. the white and grey matter components inside of the cord) may have different material properties and thus they may move differently under impact (Ichihara et al., 2001). Understanding the biomechanics of the structure and subcomponents of the spinal cord can give insight into the histology and behavioral results and help validate existing models of the spinal cord. This enhanced knowledge of the interactions between impact parameters, spinal cord biomechanics, and _________________ A version of this chapter will be submitted for publication. Lucas E., Liu J., Russell C., Tetzlaff W., and Cripton P.A. (2010) High-speed radiography used to measure dynamic spinal cord deformation in an in vivo rat model.  61  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model  injury could also help improve preventative devices such as helmets (Nelson and Cripton, 2008) and could help in the development of targeted treatments. The limited research on spinal cord biomechanics is due in part to the complications of implementing appropriate imaging techniques to record spinal cord deformations during injury. Hung et al. used a high-speed camera to image in vivo feline spinal cord movement during a SCI, but this approach required a large laminectomy and it could not show internal tissue deformation (Hung et al., 1975). MRI has been used to study tissue damage and repair after experimental SCI in rodents (Bilgen et al., 2001; Gareau et al., 2001; Kozlowski et al., 2008), but it currently cannot be used to capture the dynamic nature of a single injurious high-speed impact due to acquisition speed limitations (Bayly et al., 2006). High-speed x-ray has been used to image movement of animal and cadaveric human specimens during biomechanical tests (Al-Bsharat et al., 1999; Bauman and Chang, 2010; Brainerd et al., 2010; Hardy et al., 2001; Hardy et al., 2007; Hardy et al., 2008; Martin et al., 2010; Ono et al., 1997; Sundararajan et al., 2004; Zou et al., 2007). The deformation of the cadaveric brain has been measured during impact using radio-opaque markers and high-speed x-ray (AlBsharat et al., 1999; Hardy et al., 2001; Hardy et al., 2007). This is a promising method for imaging spinal cord motion during a SCI but requires the inclusion of radio-opaque fiducial markers in the cord since neural tissue is radio-translucent. Maiman et al. injected contrast agent into in vivo feline spinal cords and measured the displacement of the markers after distraction and contusion injuries using fluoroscopy. This was not imaged at high-speeds and only a single marker was injected into each spinal cord segment giving limited information on the dynamic deformation of localized areas within the cord (Maiman et al., 1989). There is a need for a technique to image high-speed motion of several fiducials in an in vivo spinal cord as this will allow calculation of localized internal deformations and overall surface deformations during impact. The objectives of this study were to (1) develop a method of using high-speed x-ray to track internal and surface deformations of an in vivo rat spinal cord at high collection rates during experimental SCI, (2) collect initial surface and internal spinal cord deformation data during an experimental contusion SCI, and (3) assess bead migration with respect to the spinal cord tissue during impact using an experimental and FEM approach. 62  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model  3.2  Methods  3.2.1 Surgical procedure Twelve anesthetized Sprague Dawley rats (390-460g) were used for this study, which was approved by the UBC Animal Care Centre and Ethics Committee (Appendix G). Buprenorphine (0.03mg/kg, Temgesic®, Reckitt Benkiser Healthcare Ltd.,UK) was injected subcutaneously and the rat was deeply anesthetized with isoflurane (4% for introduction and 1-2% for maintenance as per clinical signs of nociceptive reflexes). Following anesthetic, the animal was placed in a stereotaxic surgical frame (David Kopf Instruments, Tujunga, California) and a local anesthetic (0.5mL lidocaine with 2% epinephrine) was injected into the cervical musculature. A full laminectomy of C5 was performed to expose the dorsal and lateral sides of the spinal cord in this area. Spherical radio-opaque tantalum beads (260µm diameter, Bal-tec, Los Angeles, CA) were used as fiducial markers to measure movement and deformation of the spinal cord tissue during impact. One bead was injected into the dorsal cord, approximately 0.6mm below the dorsal surface, and another was injected into the ventral cord, approximately 2.1mm below the dorsal surface (Figure 3-1). The required injection depth and angle were determined using a rat spinal cord atlas (Paxinos and Watson, 1986) as discussed in Chapter 2, and these injection locations were intended to be in the approximate location of the dorsal and ventral white matter. The internal beads will be referred to as the internal dorsal (ID) and internal ventral (IV) beads. To inject the beads, a channel was first created in the spinal cord to the desired bead location using a 350µm diameter needle. The bead was then placed on top of the channel and pushed to the bottom using a 30 gauge needle (300µm) with a flattened tip. Four additional beads were glued with cyanoacrylate to the pia mater (closely adhered to the surface of the spinal cord) through small holes made in the dura. These beads were located on the lateral surface of the cord, caudal and cranial to the injection site as shown in Figure 3-1. These surface beads will be referred to as the cranial dorsal (CrD), cranial ventral (CrV), caudal dorsal (CdD), and caudal ventral (CdV) beads. A 1mm diameter bead was glued to the transverse process of C5 to measure spinal column motion and will be referred to as the spinal column reference bead (Figure 3-1). The scapulae were removed since this was a non-survival study and they interfered with bead visualization. 63  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model  Prior to transportation to the high-speed x-ray, the rat was weaned off of the gas anesthesia with an intramuscular injection of ketamine (72mg/kg; Bimeda-MTC, Cambridge, Ontario, Canada) and xylazine (10mg/kg; Bayer Inc., Etobicoke, Ontario, Canada). Top-ups of ketamine were administered for maintenance as per clinical signs of nociceptive reflexes. Vertebral Clamp  Vertebral Clamp Impactor  D (+) Cd  Caudal Surface Beads  C2  Cranial Surface Beads  Cr (+) V  Internal Beads  Spinal Column Reference Bead  Figure 3-1: Lateral high-speed x-ray image showing the internal and surface bead positioning in the cervical spinal cord of an anesthetized rat. Internal beads were in the dorsal (D) and ventral (V) internal aspects of the cord. Surface beads were on the ventral and dorsal aspects of the surface of the cord both cranial (Cr) and caudal (Cd) to the internal beads.  3.2.2 Impact procedure The spinal column was stabilized before impact by clamping C3&C4 and C6&C7 and fixing them to the test frame using articulating arms (Figure 3-2). Using a hydraulic actuator (Instron Dynamight, Instron, Norwood, MA), the dorsal surface of the spinal cord was impacted at the location of the C5 vertebrae at 130mm/s (approximately the same speed as the Infinite Horizons impactor (Scheff et al., 2003)) to a depth of 1mm which is a similar compression depth to other actuator driven impactors (0.8-1.3mm) (Choo et al., 2007; Pearse et al., 2005; Scheff et al., 2003; Sparrey et al., 2008). The impactor had a 4mm x 2.5mm flat rectangular surface. The impactor was first lowered until it visually touched the dura of the spinal cord and was then raised 17mm above the surface of the spinal cord and accelerated downwards to 1mm past the dura contact position. The rat was euthanized immediately after impact via an overdose of chloral hydrate (5%, 100 mg/kg, i.p.; BDH Chemicals, Toronto, 64  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model  Ontario, Canada). The spinal cord was harvested, fixed in 4% Paraformaldehyde, and dissected to determine the internal bead locations (dorsal or ventral, white or grey matter) using the procedure discussed in Chapter 2.  High-speed Camera Image Intensifier Hydraulic actuator LVDT  Accelerometer Load cell Impactor  X-ray Tube  Articulating arms Clamps Exposed spinal cord Figure 3-2: Experimental test set-up. The high-speed x-ray tube and image intensifier were on either side of the Instron Dynamight and impactor. In the inset photo a rat is shown positioned below the impactor with the spinal column clamped and fixed to the test frame with articulating arms.  Force, displacement, and acceleration data were collected at 20kHz from a load cell (5lb (22.2N), model 31, Honeywell-Sensotec, Columbus, OH, ±0.15% full-scale nonlinearity (0.03N)), linear variable differential transformer (LVDT) (±25mm stroke, model 8841, Instron, ±0.10mm accuracy (Appendix F)), and accelerometer (50g , model 355B03, PCB Piezotronics, Depew, NY, ≤1% full-scale non-linearity (0.5g)) which were all attached to the actuator and impactor. Inertial compensation of the force data was applied using the accelerometer data (Appendix E). These data were filtered using a 4th order Butterworth filter with a 200Hz cut off frequency selected based on Fast Fourier Transform (FFT) and raw data analysis (Appendix E). The actual dura touch point was determined to be the displacement corresponding to when the force increased by 0.01N (similar to touch force of 0.015N used by Sparrey et al. (Sparrey et al., 2008)).  65  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model  3.2.3 High-speed x-ray imaging The impact was imaged laterally with high-speed x-ray (40kV and 2mA) at 3,000 frames per second (Figure 3-2). The high-speed x-ray equipment consisted of a 640Watt generator (Gulmay FC-160, TSG X-ray, Atlanta, GA) supplying an x-ray tube (Comet MXR160, TSG X-ray, Atlanta, GA) and an image intensifier (PS93QX-P20, Precise Optics, Bay Shore, NY) set to a four inch field of view. A Phantom V12 high-speed camera (Vision Research, Wayne, NJ) was mounted to the back of the image intensifier and collected x-ray images at 3,000fps with 800x800 resolution. The x-ray video was undistorted using XROMM (X-ray Reconstruction of Moving Morphology) software (Brainerd et al., 2010), and denoised using custom 3D curvelet image denoising software developed with the CurveLab toolkit (www.curvelet.org) (Candes et al., 2006) (Appendix D). 3.2.4 Bead displacement and velocity measurement The motion of the impactor, clamps, and beads were tracked using TEMA Motion software (Image Systems, Wayne, NJ) and filtered with a 100Hz 4th order Butterworth filter (Appendix E). The mm/pixel conversion factor was calculated by dividing the average actuator displacement (from the LVDT) from ten separate impacts by the tracked impactor displacement (from beads glued to the impactor) in pixels. All bead displacements were measured relative to the spinal column reference bead. The dimensions of the x-ray videos represented the coordinate system for all measurements. Bead displacements in the vertical and horizontal directions of the x-ray images were considered to be dorsal-ventral (D/V) and cranial-caudal (Cr/Cd) displacements, respectively. The D/V and Cr/Cd displacement of each spinal cord bead at the time of the maximum impactor depth was recorded for statistical analysis. Bead velocity was calculated by fitting a cubic spline to the displacement versus time data and differentiating the resulting spline. The maximum D/V and Cr/Cd bead velocity during impact was recorded for statistical analysis. The internal spinal cord strain in the D/V direction was estimated as the change in D/V distance between the two internal beads before compression and during maximum compression divided by the distance before compression. The overall spinal cord strain was estimated by dividing the impact depth by the D/V diameter of the spinal cord. For this study  66  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model  the diameter of each rat‘s spinal cord was not directly measured, but the D/V diameter was estimated to be 3mm using a rat spinal cord atlas (Paxinos and Watson, 1986). 3.2.5 Bead migration assessment Since the internal beads were not directly glued to the spinal cord, it was necessary to verify that these beads were not migrating with respect to the spinal cord during impact. It was assumed that if the internal beads migrated during impact, they would not return to their original position before impact and thus would have a post-impact displacement. Residual tissue deformation due to the impact would also contribute to the post-impact bead displacement, so a method of separating bead migration from real residual tissue deformation was needed. It was assumed that the surface beads‘ post-impact displacements would be due to residual tissue deformation rather than migration since the beads were glued directly to the spinal cord. The internal beads‘ post-impact displacements were compared to the glued surface beads‘ post-impact displacements. Internal bead displacements exceeding the surface bead displacements were attributed to bead migration. 3.2.6 Bead tracking accuracy and precision The bead tracking precision was estimated by averaging the standard deviation of each spinal cord bead‘s position (with respect to the spinal column reference bead) before the impactor made contact with the cord. The bead tracking accuracy was determined by imaging ten additional impacts with high-speed x-ray without an animal present. A 260µm tantalum bead was glued to a piece of plastic attached to the tip of the impactor. A segment from a cadaveric rat spinal cord was then glued overtop of the bead to simulate a bead inside of the cord. The bead displacement was tracked using the TEMA motion software and compared to the LVDT data to assess the tracking accuracy. The accuracy was defined as the mean absolute value of the difference between the tracked displacement and the LVDT displacement. 3.2.7 Statistical analysis A repeated measures analysis of variance was performed on the bead displacements at maximum impactor depth, the maximum bead velocities during impact, and the post-impact bead displacements of the experimental impact tests with bead position (i.e. ID, IV, CrD, CrV, CdD, CdV) as the factor. A Student Newman Keuls post-hoc test was performed to 67  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model  identify significance between beads. The bead displacements for the accuracy tests were analyzed in the same manner. A matched pairs t-test was performed for assessing white versus grey matter motion, with the internal bead location (white or grey matter) as a categorical factor. P-values less than 0.05 were considered significant. 3.2.8 Finite element analysis The effect of the bead inclusions on the spinal cord motion was assessed using a finite element simulation of the impact (Figure 3-3). The dynamic 3D FEM used for this simulation was developed by C. Russell at the University of British Columbia (see Preface) and consisted of a rat cervical spinal column and spinal cord model (Russell et al., 2008) with the same dura and spinal cord material properties used by Maikos et al. for their FEM of the thoracic rat spinal cord (Maikos et al., 2008). While Maikos et al. developed models with different white and grey matter material properties, the current model used the properties from their standard homogenous model since the difference in in vivo white and grey matter deformation during impact has yet to be determined. Maikos et al.‘s model had hyperviscoelastic material properties incorporating an Ogden hyperelastic strain energy function and Prony series viscoelasticity. Their model was validated at approximately 490mm/s by matching the trajectory of the impactor during cord compression to experimental drop-weight tests (Maikos et al., 2008). Unlike the model developed by Maikos et al, the current model incorporated a fluid like cerebrospinal fluid (CSF) which was modeled using the Smoothed Particle Hydrodynamics (SPH) method proposed by Monaghan as an efficient means of modeling incompressible flow of water (Monaghan, 2005). The vertebrae and discs were constrained in all directions to simulate clamping of the spinal column, and the impactor was modeled to the same dimensions as the impactor used in the experimental tests. Additionally, the nodes at the cranial and caudal edges of the dura were constrained from movement in the axial direction.  68  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model  * *  Figure 3-3: Finite element model of the rat cervical spinal cord with bead inclusions developed by C. Russell (Russell et al., 2008). A lateral view is shown (Left) as well as a transverse section with black asterisks denoting bead position (Right). The impactor is positioned above the spinal cord before impact.  Bead inclusions were modeled by adding a point mass of a 260 µm spherical tantalum bead (approximately 0.15mg) to a node in the dorsal and ventral cord (Figure 3-3). The impactor was positioned approximately 0.3mm above the cord and was displaced 1.3mm at a constant velocity, for a 1.0mm contusion. The spinal cord with and without bead inclusions was impacted at 130mm/s to simulate the experimental parameters, and the displacement of the nodes with beads were compared to a control without beads. This simulation was repeated at 1m/s to simulate the higher velocities used by some rodent SCI impactors (Choo et al., 2007; Young, 2002). The simulations at each speed were repeated with ten-times the bead mass to investigate an inertial effect due to a higher bead mass.  3.3  Results  3.3.1 Bead location, impact parameters, and high-speed x-ray imaging Harvested spinal cords showed that all of the ID beads were in the dorsal white matter, and the IV beads were in the ventral white matter for eight of the animals and the ventral grey matter for the remaining four. The bead motion was highly visible in the highspeed x-ray video after distortion correction and denoising. The average (±STD) resolution of the high-speed x-ray images was 0.125 ± 0.001 mm/pixel. Figure 3-4 shows images from the x-ray video taken before, during, and after impact. The average maximum force during impact was 1.14N ± 0.3N and the average impact depth and velocity were 1.11mm ± 0.10mm and 131.6mm/s ± 1.3mm/s, respectively (±STD). The average force and 69  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model  displacement during impact is shown in Figure 3-5. The peak in force at maximum displacement was due to a normal twitching response of the rat that occurred during all tests after the impactor reached its maximum displacement.  DURING  BEFORE  AFTER  Figure 3-4: Lateral x-ray images from high-speed x-ray video before, during, and after impact.  1.6 1.4  Force (N)  1.2 1 0.8 0.6 0.4 0.2 0 0  0.2  0.4 0.6 0.8 1 Displacement (mm)  1.2  Figure 3-5: Force versus impactor displacement during impact of the spinal cord. The averages are shown with the solid red line, and standard deviations in displacement and force are shown with the blue shading.  3.3.2 Bead displacement and velocity At maximum compression, the ID bead had approximately 65% larger ventral displacement than the IV bead (p=0.0002) and 14%-94% larger ventral displacement than all of the surface beads (p=0.0001-0.02) (Table 1-1). The IV bead also had 54%-70% larger ventral displacement than the ventral surface beads (p=0.0001). These trends are also evident when looking at the average bead displacements over time during impact (Figure 3-6). In the cranial direction, the ID bead moved more than all of the surface beads (p=0.002-0.03) and the IV bead moved more than the CdD, CdV, and CrV surface beads (p=0.002-0.01). There 70  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model  were no significant differences in the maximum displacements of the IV beads in the grey versus white matter (p= 0.7 (Cr/Cd) and 0.6 (D/V)). Table 3-1: Average spinal cord bead displacements at maximum impactor depth, maximum velocities during impact, and post-impact displacements (avg±STD). Negative values represent motion in the ventral or caudal direction. Additional data can be found in Appendix C. Displacement at Maximum Impactor Depth (mm) Cr/Cd D/V Direction Direction 0.06 ± 0.10 -0.62 ± 0.24  Bead Caudal Dorsal  Maximum Velocity During Impact (mm/s) Cr/Cd D/V Direction Direction 10.33 ± 15.28 -84.45 ± 30.21  Cr/Cd Direction -0.01 ± 0.06  D/V Direction -0.10 ± 0.10  Post-Impact Displacement (mm)  Cranial Dorsal  0.12 ± 0.12  -0.73 ± 0.14  17.00 ± 19.96  -92.79 ± 22.84  0.01 ± 0.08  -0.11 ± 0.09  Internal Dorsal  0.19 ± 0.14  -0.85 ± 0.13  22.79 ± 14.28  -99.89 ± 12.48  0.04 ± 0.07  -0.09 ± 0.10  Caudal Ventral  0.09 ± 0.04  0.05 ± 0.10  11.84 ± 11.71  5.11 ± 18.64  0.02 ± 0.05  -0.02 ± 0.07  Cranial Ventral  0.08 ± 0.08  0.08 ± 0.14  13.43 ± 19.18  12.46 ± 16.15  0.02 ± 0.07  -0.03 ± 0.09  Internal Ventral  0.20 ± 0.18  -0.30 ± 0.11  18.68 ± 20.26  -38.51 ± 14.57  0.05 ± 0.07  -0.01 ± 0.08  0.5  Dura contact  Ventral Beads  Displacement, mm  0  -0.5 Caudal Dorsal Cranial Dorsal Internal Dorsal Caudal Ventral Cranial Ventral Internal Ventral Impactor  -1  -1.5  0  5  10  Dorsal Beads  15 Time, ms  20  25  30  Figure 3-6: Average spinal cord bead displacements in the D/V direction during impact. The end of the data series represents the bead displacement at maximum impact depth as recorded in Table 1-1.  71  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model  The average internal spinal cord strain (change in distance between the internal beads during maximum compression divided by the original distance between the beads) was 49% ± 15%. The overall spinal cord strain (impact depth divided by 3mm approximate D/V diameter (Paxinos and Watson, 1986)) was estimated to be 37% ± 3%. Comparing the strains of each rat individually, the internal strains were an average of 13% ± 15% larger than the estimated overall strain. The statistical significance of these differences was not assessed due to the limitations in assuming the spinal cord diameter. The bead velocities in the D/V direction during impact are shown in Figure 3-7. The ID bead had significantly larger maximum velocities in the ventral direction than the IV bead (p=0.0002) and ventral surface beads (p=0.0001), but had no significant difference with the dorsal surface beads (p=0.1 and 0.3) (Table 1-1). The IV bead had significantly larger maximum velocities than the ventral surface beads in the ventral direction (p=0.0001). There were no significant differences in the maximum bead velocities in the Cr/Cd direction (p=0.2). There were also no significant differences in the maximum velocities of the IV beads in the grey versus white matter in either direction (p= 0.7). 20 0  Velocity, mm/s  -20  Ventral Beads  -40 -60  Caudal Dorsal Cranial Dorsal Internal Dorsal Caudal Ventral Cranial Ventral Internal Ventral Impactor  -80 -100  Dorsal Beads -120 -140  0  5  10  15 Time, ms  20  25  30  Figure 3-7: Average spinal cord bead velocities in the D/V direction during impact. The maximum values represent the data recorded in Table 1-1.  72  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model  3.3.3 Bead migration In the ventral direction, there were no significant differences in the post-impact displacement (assumed in this study to be an indicator of bead migration) of the ID bead compared to the dorsal surface beads (p=0.4 and 0.7), nor with the IV bead compared to the ventral surface beads (p=0.5 and 0.6) (Table 1-1). In the cranial direction, the ID bead had significantly higher post-impact displacements than the dorsal surface beads (p=0.0004 and 0.03), with an average displacement about 0.04mm greater than the surface bead averages. The IV bead also had significantly higher post-impact displacements than the ventral surface beads in the cranial direction (p=0.001 and 0.003), which was also an average of 0.04mm greater than the surface beads. 3.3.4 Bead tracking accuracy and precision The bead tracking precision, defined in this study as the mean standard deviation of each spinal cord bead‘s position before impact, was 0.021mm ± 0.01mm. For the bead tracking accuracy study the average absolute value of the differences between the impactor displacement (measured with the LVDT) and the internal spinal cord bead displacement (measured using TEMA) was 0.019 ± 0.017mm, although this difference was not statistically significant (p=0.9). 3.3.5 Finite element analysis The FEM bead displacements versus the control simulation for 130mm/s impacts are shown in Figure 3-8. The maximum differences in displacement for the 130mm/s and 1m/s impacts are listed in Table 3-2. For the 130mm/s impact, the maximum differences between the beads and the control were lower than the bead tracking precision (0.02mm) and represented approximately 2% of the maximum IV bead displacement. For the 1m/s impact, the differences were within the standard deviation of the bead tracking precision and represented approximately 10% of the maximum IV bead displacement. For the ten-times mass simulation with the 130mm/s impact the differences were also below the level of precision. For the ten-times mass case with the 1m/s impact the differences represented approximately 30% of the IV bead displacement and were above the bead tracking precision.  73  Displacement, mm  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model  -0.3 -0.35 -0.4  Displacement, mm  0  7.2  -0.2  7.4 Time, ms  7.6  -0.4 -0.6  Impactor Internal Ventral Control Internal Dorsal Control Internal Ventral Bead Internal Dorsal Bead  -0.8 -1 0  2  4  6 Time, ms  8  10  Figure 3-8: Spinal cord bead displacements from the FEM simulation of the 130mm/s impact. The inset shows a magnified view of the displacements. Minimal variation was seen between the beads and the control. Table 3-2: Maximum differences in displacements of nodes with beads compared to nodes without beads (control). All values are in millimeters.  Bead Internal Dorsal Internal Ventral  3.4  Difference in bead displacement and from control (mm) 130mm/s Impact 1m/s Impact 10X Bead 10X Bead Bead Mass Bead Mass Mass Mass 0.005 0.014 0.029 0.092 0.006 0.009 0.015 0.047  Discussion The results indicate the merit of this technique for measuring in vivo spinal cord  deformation. The differences in bead displacements and velocities imply that the spinal cord undergoes complex internal and surface deformations during impact. The larger displacements and velocities of the dorsal beads versus the ventral beads in the ventral direction were expected since the spinal cord was impacted at the dorsal surface. The greater displacement of the internal beads versus the surface beads in the Cr/Cd direction indicates that the internal tissue displaces more than the surface during impact which could possibly be 74  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model  due to the Poisson effect. This Cr/Cd tissue displacement has been proposed in theoretical and experimental models to be a cause for centralized damage commonly seen in histology after a contusion injury (Blight, 1988; Blight and Decrescito, 1986; Maiman et al., 1989). In the D/V direction, the internal strains were higher than the estimated overall strains which suggests that the central cord could become more damaged due to the higher levels of strain. However, there are limitations with the overall strain measurement including not directly measuring the spinal cord diameter and assuming the impactor was exactly perpendicular to the spinal cord surface. The assumptions that the internal beads were in the same sagittal and transverse plane and that there were no lateral movements of the beads were also limitations of the internal strain measurement. The strain calculations were used to estimate differences between inner and outer spinal cord motion. For more precise strain measurements, the D/V diameter of each rat‘s spinal cord should be measured before impact, and bi-planar x-ray should be used to detect any lateral bead movements. The differences in white matter and grey matter material properties are currently debated in published studies (Ichihara et al., 2001; Ozawa et al., 2001). While there was not a significant difference between the ventral white and ventral grey matter motion in this study, further studies are needed with larger animal numbers and more bead locations to quantitatively examine differences between white and grey matter motion. A larger animal model (such as a porcine model) is recommended for this type of study, as the larger size of the spinal cord would allow for finer detection of differences in the white and grey matter. A crucial step for validating this methodology was verifying that the internal markers did not migrate with respect to the spinal cord tissue during the impact. Many researchers that have used radio-opaque fiducial markers in tissue and bone have assumed that the markers do not move independently of the tissue (Hardy et al., 2008; Panjabi et al., 1995; Panjabi et al., 2000; Snelderwaard et al., 2002). While this is probably a valid assumption for markers glued to tissue or bone (as the markers in the previously mentioned studies were), migration of markers not rigidly affixed should be evaluated. Hardy et al. tracked the motion of brain tissue using fiducial markers of a similar density to neural tissue to reduce the chance of migration (Hardy et al., 2001). Marker migration was assessed in their study by comparing the marker locations before and after impact. The current study also assumes that 75  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model  any bead migration would permanently displace the bead from its original position postimpact. While there were no significant differences between the internal and surface postimpact displacements in the D/V direction, there were significant differences in the Cr/Cd direction indicating a possibility of migration. As previously discussed (Chapter 2), it is possible that the internal spinal cord tissue experiences more residual tissue deformation than the surface of the cord. In this case the detected migration may actually be residual deformation of the internal tissue that has exceeded the surface tissue deformation. It was also assumed that the injection tract was smaller than the bead diameter and the beads would not move within the tract during impact due to the surrounding hydrostatic pressure. It was assumed that if the beads were moving in the injection track this would have caused variability in the post-impact bead displacements which would be detectable with the given method of assessing migration. Additionally, this movement would have likely been visible in the displacement data during impact, since the static friction coefficient would be higher than the dynamic friction coefficient and the change from the static to dynamic state would cause a sudden change in displacement. This was not noticed in the displacement data. The inherent limitation of using radiography is that the spinal cord tissue cannot be seen so it is not known definitively whether post-impact displacement is a good predictor of migration. A limitation of using radiography is the need to inject beads into the spinal cord. While some bleeding was seen along the injection track during post-mortem dissection, it appeared to be minimal and the residual track appeared to be smaller than the bead diameter. In the future, histology should be performed to assess the size of the remaining injection track and damage due to the bead injection. The inclusion of beads inside of the cord and on the surface required creating small holes in the dura. This could have affected the movement of the spinal cord due to loss of CSF which has been shown to reduce spinal cord occlusion during impact (Jones et al., 2008). A full laminectomy was performed with this study because a large area of the spinal cord was being impacted. To avoid laminectomy, the beads can be injected between two laminae by removing the ligaments and hanging a small weight on the tail to increase the intervertebral space before injection (Chapter 2). The scapulae were removed as they were shown to interfere with the visualization of the beads. For survival studies, alternative approaches should be taken such as hooking small weights to the front limbs to pull the scapulae downwards and out of the beam of the x-ray. In the future, the 76  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model  blood pressure should also be regulated during the surgical and impact procedures since a change in perfusion pressure due to the loss of blood from the surgical procedures may alter the cord‘s response to impact. Another potential limitation of using fiducial markers is the possibility that the spinal cord moves differently with bead inclusions. A FEM was used to further investigate the effect of beads on the spinal cord deformation. The results indicated that the difference in the spinal cord displacement with and without bead inclusions was small and below the precision of the high-speed x-ray tracking method. The difference in displacement with ten-times the mass at 1m/s velocity was above the x-ray precision. This emphasizes the need to keep the bead mass as low as possible. FEMs of spinal cords are limited with high velocity impacts due to a limited amount of research on the in vivo material properties of the spinal cord at high strain rates. The current FEM used rat spinal cord material properties from Maikos et al. (Maikos et al., 2008). Maikos et al. did not fit these material properties directly to experimental material testing data, and therefore they may not perfectly describe the stressstrain behavior of the tissue. However, they calibrated the properties to match experimental weight-drop results for an impact velocity of 489mm/s, which is the only known spinal cord FEM validation at high strain rates. The current model is the only known spinal cord FEM to incorporate a fluid like CSF layer, but this method has yet to be validated against appropriate experimental results. Even with these limitations, the current FEM was thought to be a good estimate of spinal cord material properties and was assumed to be adequate for assessing general effects that bead inclusions may have on the spinal cord‘s response to impact. Future FEA simulations could include the full bead geometry, a small injection track, and additional bead locations. The effects of the bead inclusions could also be further investigated with surrogate cord models or with repeated in vivo or ex vivo experimental measurements.  3.5  Conclusions A method was successfully developed to measure in vivo rat spinal cord biomechanics  using high-speed x-ray. Bead migration during impact was shown to be small and mostly insignificant. Results showed differences in displacements and velocities of beads inside of the spinal cord and on the surface of the cord, suggesting that the spinal cord moves differently depending on location within the spinal cord. Future applications of this technique 77  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model  involve investigating white versus grey matter movement, comparing bead movements with histological and behavioral outcomes, and validating spinal cord models. This information would improve our knowledge on the basic science of spinal cord injury mechanisms and could be used to improve existing clinical treatments and preventative measures.  78  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model  3.6  References  Al-Bsharat, A. S., W. N. Hardy, K. H. Yang, T. B. Khalil, S. Tashman and A. I. King (1999). Brain/Skull relative displacement magnitude due to blunt head Impact: New experimental data and model. Proc. 43rd Stapp Car Crash Conference, pp. 321-332, Society of Automotive Engineers, Warrendale, PA. Bauman, J. M. and Y.-H. Chang (2010). "High-speed X-ray video demonstrates significant skin movement errors with standard optical kinematics during rat locomotion." Journal of Neuroscience Methods 186(1): 18-24. Bayly, P. V., E. E. Black, R. C. Pedersen, E. P. Leister and G. M. Genin (2006). "In vivo imaging of rapid deformation and strain in an animal model of traumatic brain injury." Journal of Biomechanics 39(6): 1086-1095. Bilgen, M., R. Abbe and P. A. Narayana (2001). "Dynamic contrast-enhanced MRI of experimental spinal cord injury: in vivo serial studies." Magn Reson Med 45(4): 614622. Bilston, L. E. and L. E. Thibault (1995). "The mechanical properties of the human cervical spinal cord in vitro." Ann Biomed Eng 24(1): 67-74. Bilston, L. E. and L. E. Thibault (1997). "Biomechanics of cervical spinal cord injury in flexion and extension: a physical model to estimate spinal cord deformations." International Journal of Crashworthiness 2(2): 207-218. Blight, A. (1988). "Mechanical factors in experimental spinal cord injury." J Am Paraplegia Soc 11(2): 26-34. Blight, A. R. and V. Decrescito (1986). "Morphometric analysis of experimental spinal cord injury in the cat: the relation of injury intensity to survival of myelinated axons." Neuroscience 19(1): 321-341. Brainerd, E. L., D. B. Baier, S. M. Gatesy, T. L. Hedrick, K. A. Metzger, S. L. Gilbert and J. J. Crisco (2010). "X-ray reconstruction of moving morphology (XROMM): precision, accuracy and applications in comparative biomechanics research." J Exp Zool A Ecol Genet Physiol. Candes, E., L. Demanet, D. Donoho and L. Ying (2006). "Fast Discrete Curvelet Transforms." Multiscale Modeling & Simulation 5(3): 861-899. Choo, A. M., J. Liu, C. K. Lam, M. Dvorak, W. Tetzlaff and T. R. Oxland (2007). "Contusion, dislocation, and distraction: primary hemorrhage and membrane permeability in distinct mechanisms of spinal cord injury." J Neurosurg Spine 6(3): 255-266. Czyz, M., K. Scigala, W. Jarmundowicz and R. Beidzinski (2008). "The biomechanical analysis of the traumatic cervical spinal cord injury using finite element approach." Acta Bioeng Biomech 10(1): 43-54. Dryden, D. M., L. D. Saunders, B. H. Rowe, L. A. May, N. Yiannakoulias, L. W. Svenson, D. P. Schopflocher and D. C. Voaklander (2003). "The epidemiology of traumatic spinal cord injury in Alberta, Canada." Can J Neurol Sci 30(2): 113-121. Fiford, R. J. and L. E. Bilston (2005). "The mechanical properties of rat spinal cord in vitro." J Biomech 38(7): 1509-1515. Gareau, P. J., L. C. Weaver and G. A. Dekaban (2001). "In vivo magnetization transfer measurements of experimental spinal cord injury in the rat." Magn Reson Med 45(1): 159-163. 79  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model  Greaves, C. Y., M. S. Gadala and T. R. Oxland (2008). "A three-dimensional finite element model of the cervical spine with spinal cord: an investigation of three injury mechanisms." Ann Biomed Eng 36(3): 396-405. Hardy, W. N., C. D. Foster, M. J. Mason, K. H. Yang, A. I. King and S. Tashman (2001). "Investigation of Head Injury Mechanisms Using Neutral Density Technology and High-Speed Biplanar X-ray." Stapp Car Crash J 45: 337-368. Hardy, W. N., M. J. Mason, C. D. Foster, C. S. Shah, J. M. Kopacz, K. H. Yang, A. I. King, J. Bishop, M. Bey, W. Anderst and S. Tashman (2007). "A study of the response of the human cadaver head to impact." Stapp Car Crash J 51: 17-80. Hardy, W. N., C. S. Shah, M. J. Mason, J. M. Kopacz, K. H. Yang, A. I. King, C. A. Van Ee, J. L. Bishop, R. F. Banglmaier, M. J. Bey, R. M. Morgan and K. H. Digges (2008). "Mechanisms of traumatic rupture of the aorta and associated peri-isthmic motion and deformation." Stapp Car Crash J 52: 233-265. Hung, T. K., M. S. Albin, T. D. Brown, L. Bunegin, R. Albin and P. J. Jannetta (1975). "Biomechanical responses to open experimental spinal cord injury." Surg Neurol 4(2): 271-276. Hung, T. K. and G. L. Chang (1981). "Biomechanical and neurological response of the spinal cord of a puppy to uniaxial tension." J Biomech Eng 103(1): 43-47. Ichihara, K., T. Taguchi, Y. Shimada, I. Sakuramoto, S. Kawano and S. Kawai (2001). "Gray matter of the bovine cervical spinal cord is mechanically more rigid and fragile than the white matter." J Neurotrauma 18(3): 361-367. Jones, C. F., S. G. Kroeker, P. A. Cripton and R. M. Hall (2008). "The effect of cerebrospinal fluid on the biomechanics of spinal cord: an ex vivo bovine model using bovine and physical surrogate spinal cord." Spine 33(17): E580-588. Kearney, P. A., S. A. Ridella, D. C. Viano and T. E. Anderson (1988). "Interaction of contact velocity and cord compression in determining the severity of spinal cord injury." J Neurotrauma 5(3): 187-208. Kozlowski, P., D. Raj, J. Liu, C. Lam, A. C. Yung and W. Tetzlaff (2008). "Characterizing white matter damage in rat spinal cord with quantitative MRI and histology." J Neurotrauma 25(6): 653-676. Kroeker, S. G., P. L. Morley, C. F. Jones, L. E. Bilston and P. A. Cripton (2009). "The development of an improved physical surrogate model of the human spinal cord-Tension and transverse compression." Journal of Biomechanics 42(7): 878-883. Li, X. F. and L. Y. Dai (2009). "Three-dimensional finite element model of the cervical spinal cord: preliminary results of injury mechanism analysis." Spine (Phila Pa 1976) 34(11): 1140-1147. Maikos, J. T., Z. Qian, D. Metaxas and D. I. Shreiber (2008). "Finite element analysis of spinal cord injury in the rat." J Neurotrauma 25(7): 795-816. Maiman, D. J., J. Coats and J. B. Myklebust (1989). "Cord/spine motion in experimental spinal cord injury." J Spinal Disord 2(1): 14-19. Martin, D. E., N. J. Greco, B. A. Klatt, V. J. Wright, W. J. Anderst and S. Tashman (2010). "Model-Based Tracking of the Hip: Implications for Novel Analyses of Hip Pathology." J Arthroplasty. Monaghan, J. (2005). "Smoothed particle hydrodynamics." Reports on Progress in Physics 68(8).  80  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model  National Spinal Cord Injury Statistical Center. "Spinal Cord Injury Facts and Figures at a Glance." Retrieved February, 2010, from www.nscisc.uab.edu. Nelson, T. and P. A. Cripton (2008). "Inducing head motion with a novel helmet during head-first impact can mitigate neck injury metrices: an experimental proof-of-concept investigation using mechanical surrogates." Proceedings of the International Research Council on the Biomechanics of Impacts. Bern, Switzerland. Oakland, R. J., R. M. Hall, R. K. Wilcox and D. C. Barton (2006). "The biomechanical response of spinal cord tissue to uniaxial loading." Proc Inst Mech Eng [H] 220(4): 489-492. Ono, K., K. Kaneoka, A. Wittek and J. Kajzer (1997). Cervical injury mechanism based on the analysis of human cervical vertebral motion and head-neck-torso kinematics during low speed rear impacts, Lake Buena Vista, Florida, 41st STAPP Car Crash Conference. Ozawa, H., T. Matsumoto, T. Ohashi, M. Sato and S. Kokubun (2001). "Comparison of spinal cord gray matter and white matter softness: measurement by pipette aspiration method." J Neurosurg 95(2 Suppl): 221-224. Panjabi, M. M., M. Kifune, L. Wen, M. Arand, T. R. Oxland, R. M. Lin, W. S. Yoon and A. Vasavada (1995). "Dynamic canal encroachment during thoracolumbar burst fractures." J Spinal Disord 8(1): 39-48. Panjabi, M. M., T. Oda and J. L. Wang (2000). "The effects of pedicle screw adjustments on neural spaces in burst fracture surgery." Spine (Phila Pa 1976) 25(13): 1637-1643. Paxinos, G. and C. Watson (1986). The Rat Brain in Stereotaxic Coordinates: Second Edition, Academic Press, Inc. Pearse, D. D., T. P. Lo, Jr., K. S. Cho, M. P. Lynch, M. S. Garg, A. E. Marcillo, A. R. Sanchez, Y. Cruz and W. D. Dietrich (2005). "Histopathological and behavioral characterization of a novel cervical spinal cord displacement contusion injury in the rat." J Neurotrauma 22(6): 680-702. Russell, C., T. E. Chung and T. Oxland (2008). "Creation of a finite element model of the rat cervical spine from magnetic resonance images." Journal of Biomechanics 41(Supplement 1): S516-S516. Saari, A., E. Itshayek, T. Nelson, P. L. Morley and P. A. Cripton (2006). Spinal Cord Deformation During Injury of the Cervical Spine in Head-First Impact. International Research Council on the Biomechanics of Impacts, Madrid, Spain. Scheff, S. W., A. G. Rabchevsky, I. Fugaccia, J. A. Main and J. E. Lumpp, Jr. (2003). "Experimental modeling of spinal cord injury: characterization of a force-defined injury device." J Neurotrauma 20(2): 179-193. Snelderwaard, P., J. H. De Groot and S. M. Deban (2002). "Digital video combined with conventional radiography creates an excellent high-speed X-ray video system." J Biomech 35(7): 1007-1009. Sparrey, C. J., A. M. Choo, J. Liu, W. Tetzlaff and T. R. Oxland (2008). "The distribution of tissue damage in the spinal cord is influenced by the contusion velocity." Spine (Phila Pa 1976) 33(22): E812-819. Sundararajan, S., P. Prasad, C. K. Demetropoulos, S. Tashman, P. C. Begeman, K. H. Yang and A. I. King (2004). "Effect of Head-Neck Position on Cervical Facet Stretch of Post Mortem Human Subjects during Low Speed Rear End Impacts." Stapp Car Crash J 48: 331-372. 81  Chapter 3: High-Speed Radiography Used to Measure Dynamic Spinal Cord Deformation in an In Vivo Rat Model  Van Toen (Nee Greaves), C., T. S. Nelson, C. F. Jones, J. Street and P. A. Cripton (2008). Development of an in vitro model of head-first impact with a hybrid III head, surrogate spinal cord and simulated neck muscles. NHTSA 36th International Workshop on Human Subjects for Biomechanical Research, San Antonio, TX. Young, W. (2002). "Spinal cord contusion models." Prog Brain Res 137: 231-255. Zou, H., J. P. Schmiedeler and W. N. Hardy (2007). "Separating brain motion into rigid body displacement and deformation under low-severity impacts." J Biomech 40(6): 11831191.  82  Chapter 4: Discussion and Conclusions  Chapter 4: Discussion and Conclusions This thesis presented a novel method for measuring in vivo spinal cord deformations in a rat during quasi-static compression and dynamic contusion type experimental SCIs. This method involved injecting multiple radio-opaque markers into the internal structures of anesthetized rat spinal cords and measuring internal deformation during injury with standard static x-ray (quasi-static compression) and high-speed x-ray (dynamic contusion). While MRI is most commonly used for imaging neural tissue, radiography was used for this study because of its ability to track motion at fast speeds. The spinal cord motion was tracked using fiducial markers since neural tissue is radio-translucent. Spherical tantalum beads were chosen as the fiducial markers because of their high contrast with the surrounding tissue and bone, biocompatibility, consistent size and shape, and ease and repeatability of injections. These beads could be injected into the dorsal and ventral, white and grey matter. The beads showed little tendency for migration with respect to the spinal cord tissue, indicating that they are appropriate for tracking spinal cord motion. Future applications for this methodology include conducting larger studies with a variety of bead patterns and different SCI mechanisms and determining relationships between internal spinal cord deformation, impact parameters, and injury outcome.  4.1  Comparison between current methodology and relevant literature  4.1.1 Neural tissue deformation measurements Tracking spinal cord deformation with fiducial markers and radiography enables measurement of internal and surface deformations during both quasi-static and dynamic type experimental SCIs. The only other known study to track internal spinal cord deformation during impact was done by Maiman et al., who injected contrast agent into various segments of in vivo feline spinal cords and used fluoroscopy to image the overall displacement after distraction and drop weight contusion injuries (Maiman et al., 1989). Dynamic deformation was not reported with their study, and most likely would not have been possible due to acquisition speed limitations of standard fluoroscopy equipment (typically less than 100fps). Additionally, it is possible that liquid or powder based markers would deform during injury, which could reduce the marker tracking precision. High-speed x-ray and fiducial markers 83  Chapter 4: Discussion and Conclusions  have been used to measure dynamic deformation of a human cadaver brain during impact (Al-Bsharat et al., 1999; Hardy et al., 2001; Hardy et al., 2007; Zou et al., 2007). The markers were made by encasing beads inside of plastic tubes in efforts to reduce the overall density of the markers. However, this also increases the marker size making them too large for implementation in the spinal cord. The fiducial markers used in the current study were solid beads which would not change shape during impact and were small enough for injection inside of a rat spinal cord. Other studies have tracked in vivo human brain and spinal cord deformations during normal flexion and extension motion using MRI tagging (Bayly et al., 2005; Margulies et al., 1992; Yuan et al., 1998). Using MRI tagging techniques to image dynamic deformation requires repeated impacts which could irreversibly damage the neural tissue at injurious loads (Bayly et al., 2006). The capture rate of the high-speed xray equipment was fast enough to image a single dynamic injury. 4.1.2 Fiducial marker types Tantalum beads are commonly used as fiducial markers in biomechanical studies (Bauman and Chang, 2010; Brainerd et al., 2010; Tashman and Anderst, 2003) though lead, tin, gold, and stainless steel markers have also been used (Hardy et al., 2001; Hardy et al., 2008; Miller et al., 2006; Panjabi et al., 1995). Typical marker sizes range from 0.8mm 3mm which is much larger than the 260µm beads used in the current study (Brainerd et al., 2010; Craig et al., 2008; Hardy et al., 2007; Hardy et al., 2008; Martin et al., 2010). Smaller beads enable more precise measurements of deformations in tissues such as spinal cords, but the size is typically limited by the resolution of the x-ray system. One study used 0.1mm gold shavings to track orbital eye muscle movement in a monkey using fluoroscopy, but standard fluoroscopy has higher resolution than high-speed x-ray (Brainerd et al., 2010) and these markers could not be easily seen with the current high-speed x-ray system (Miller et al., 2006). To the author‘s knowledge, there are no known studies which have tracked the movement of beads smaller than 0.8mm with high-speed x-ray. 4.1.3 High-speed x-ray image quality The high-speed x-ray image quality was dependent on image distortion, noise, and resolution. Image distortion is apparent in fluoroscopic images due to projecting the image from a curved surface inside of the image intensifier onto a flat surface inside of the camera. 84  Chapter 4: Discussion and Conclusions  The distortion was corrected for by imaging a grid with regularly spaced holes before a testing series and using XROMM software developed at Brown University to determine the distortion correction transform (Brainerd et al., 2010). A similar approach was taken by the previously discussed high-speed x-ray studies which have used mesh grids, perforated sheets, or regularly spaced bead patterns to develop distortion correction algorithms (Brainerd et al., 2010; Hardy et al., 2001; Martin et al., 2010; Tashman and Anderst, 2003). The x-ray images were also denoised using custom denoising software developed by C. Russell at the University of British Columbia (See Preface and Appendix D) which dramatically increased the image quality and bead tracking ability. Details of image denoising typically have not been disclosed in high-speed x-ray studies, though Hardy et al. did mention using a custom noise subtraction algorithm (Hardy et al., 2007). Pixel size is commonly used as a measure of image resolution for medical imaging equipment. The pixel size of the high-speed x-ray system was approximately 125µm which was sufficient for imaging the 260µm tantalum beads. Pixel size has not been reported in most of the high-speed x-ray studies found, but one study did report a 270 µm pixel size (Craig et al., 2008). Pixel size is not necessarily the best comparison between x-ray equipment since it is also dependent on the proximity of the test object to the image intensifier, where objects farther away will be more magnified resulting in a smaller pixel size. Additionally, higher magnification and smaller fields of view in the image intensifier decreases the pixel size. A better method of directly comparing between high-speed x-ray equipment is to image a standard resolution pattern attached directly to the face of the image intensifier to avoid magnification differences. The resolution patterns for fluoroscopy equipment consist of pairs of light and dark lines which are used to determine how close the lines can be while still being able to visually separate them. The visible line pairs per millimeter (lp/mm) are reported. The image resolution of the high-speed x-ray equipment used for the current study was 2.8lp/mm at a four inch field of view. High-speed x-ray equipment at Brown University has been reported to have a resolution of 1.5lp/mm at a twelve inch field of view and 2.2lp/cm at a magnified field of view (assumed to be six inches based on standard image intensifier levels of magnification) (Brainerd et al., 2010).  85  Chapter 4: Discussion and Conclusions  4.1.4 Bead tracking accuracy and precision The bead tracking accuracy and precision needed to be high for this study since small spinal cord deformations were being measured. Previous bead tracking studies using single static x-ray images, have determined measurement error by measuring the position of the beads several times and averaging the difference in the measured versus actual positions (Panjabi et al., 1995; Panjabi et al., 2000). Precision was defined as the standard deviation of these measurements. In high-speed x-ray bead tracking studies, error has been determined with separate tests where the beads were moved a known distance and the tracked displacement was compared to the actual displacement (Brainerd et al., 2010; Tashman and Anderst, 2003). In these tests precision was defined in the actual experimental test as the standard deviation of the distance between two beads rigidly connected to the same structure (i.e. bone) (Brainerd et al., 2010; Tashman and Anderst, 2003). The measurement error of the quasi-static study was defined as the average absolute values of the differences between the measured and actual impactor bead displacement. The error was 0.024mm for the compression measurements and 0.092mm for the post-impact displacement measurements. The precision was defined as the standard deviation of the measurement error and was 0.02mm for the compression measurements and 0.05mm for the post-impact displacement measurements. These values were similar to those in a study done by Panjabi et al. in which standard x-ray was used to measure displacement of beads glued inside of a spinal canal before and after a burst fracture (Panjabi et al., 2000). The precision in this study was found to be 0.028-0.038mm depending on the measurement. The accuracy and precision of the current system could be improved even further with more precise control of the impactor depth (currently ±0.05mm) and a more precise method of registering (aligning) the images before, during, and after compression. The measurement error of the dynamic contusion study was assessed with a separate accuracy study which compared the measured displacement of a bead covered with cadaveric spinal cord to the LVDT reading of displacement. The mean absolute value of the difference between the tracked displacement and the LVDT displacement was 0.019 mm. The bead tracking precision was defined in the actual experimental tests as the mean standard deviation of the spinal cord bead movement before impact which was 0.021mm. Precision and error of 86  Chapter 4: Discussion and Conclusions  other high-speed x-ray studies tracking bead displacements have ranged from 0.02 – 0.7mm (Brainerd et al., 2010; Hardy et al., 2001; Martin et al., 2010; Tashman and Anderst, 2003). 4.1.5 Marker migration Marker migration was assessed in the current studies by measuring the postcompression displacements of the internal spinal cord beads. Post-compression displacements of the internal beads exceeding the glued surface beads were attributed to migration. For the quasi-static compression the average differences in post-compression displacements between the internal beads and the surface beads were 0.06±0.04mm in the D/V direction and 0.04±0.04mm in the Cr/Cd direction. For the high-speed contusions the average differences were 0.04±0.04mm in the D/V direction and 0.04±0.03mm in the Cr/Cd direction. These values were considered negligible since they were close to the precision of the x-ray equipment and the differences were mostly insignificant with a repeated measures analysis of variance. Post-mortem dissection showed that the beads were surrounded by the spinal cord tissue and the injection track appeared to be small. A small remaining injection track would reduce the chance of the bead backing out of the track during impact. Possible inertial effects from the added bead mass were analyzed using a FEM of a rat spinal cord with the bead mass added to dorsal and ventral nodes. The bead displacements were tracked during a simulated impact and compared to a control impact without bead inclusions. At the impact speed for the current study (130mm/s) the maximum difference between nodes with and without the added bead mass was 0.006mm which represented approximately 1-2% of the maximum bead motion and was less than the precision of the x-ray equipment. Hardy et al. used a similar analysis of assessing migration of markers used in the brain by measuring the post-impact displacement of the markers as well as verifying that the beads were surrounded by brain tissue and the injection track was small (Hardy et al., 2001). The average standard deviations of the markers‘ post-impact displacements were 0.36mm and 0.56mm for two different cadavers, which was considered to be insignificant in their study. Their assessment of migration possibly exceeded the true value of migration since real residual tissue deformation due to the impact would also be included in the post-impact displacement. To account for this in the current studies, residual tissue deformation was estimated as the deformation of the surface beads since they were glued to the surface of the 87  Chapter 4: Discussion and Conclusions  spinal cord and the internal bead post-impact displacements were compared to the surface beads. This approach would be more difficult with the brain because of its larger and more complex shape. Inertial effects of the beads were also assumed to be negligible in Hardy et al.‘s study due to the low bead mass. This group later assessed the inertial effects of the beads on the brain tissue movement with the use of a FEM and found that the beads add 1.5% of the maximum motion (Hardy et al., 2007). This is similar to the values found in the current study and was also considered insignificant. Maiman et al. assessed migration of the contrast agent markers in the feline spinal cord by confirming with dissection that the markers were surrounded by spinal cord tissue (Maiman et al., 1989). The effect of beads on the tissue mechanics was analyzed by stretching the cord and comparing the intermarker displacements to the overall cord displacements at various levels of distraction, which confirmed a 1:1 correlation. They also compared the force displacement curves with and without beads to verify there were no differences. There were no reported data on these migration studies, so it cannot be directly compared to the current study.  4.2  Spinal cord mechanics during injury  4.2.1 Internal deformation To the author‘s knowledge, no known experimental studies have measured internal spinal cord deformation in the D/V direction, limiting direct comparison between the deformations found in the current study and existing literature. Displacement of the spinal cord tissue in the Cr/Cd direction has been shown with experimental and theoretical studies and has been hypothesized to be a cause for centralized damage seen after contusion injuries (Blight, 1988; Blight and Decrescito, 1986; Maiman et al., 1989). Blight demonstrated this concept by dropping a weight onto a flexible tube filled with gel and showed that ink lines injected in the D/V direction displaced away from the site of injury with the greatest displacement seen in the central tissue (Figure 4-1). He called this displacement ―extrusion‖ of the spinal cord tissue, but in engineering terms extrusion refers to pushing or pulling material through a die. A more appropriate mechanical term might be that the spinal cord is exhibiting Possion effect where material expands in the direction perpendicular to the loading direction. For both the dynamic and the quasi-static compressions in this thesis, Cr/Cd bead displacement was seen and the internal bead displacements exceeded the surface bead 88  Chapter 4: Discussion and Conclusions  displacements. The displacement of the tissue in the Cr/Cd direction could be due to Poisson expansion from compressing the tissue in the D/V direction.  A  B  Figure 4-1: Gel model of spinal cord contusion demonstrating Cr/Cd tissue displacement (Blight, 1988). Ink lines were injected into the spinal cord (A) and were shown to deform in a parabolic manner during contusion (B). Reproduced with permission from the author.  Maiman et al. measured the Cr/Cd displacement of the contrast agent markers inside of feline spinal cords after a drop weight contusion injury to the thoracic spinal cord (Maiman et al., 1989). Although there were no markers directly under the impact site, a marker was located 2mm from the impact site. The average Cr/Cd displacement of this marker was 1.407 ± 0.089mm which was much higher than the average internal displacements in the current dynamic impact study of 0.19 ± 0.14mm (ID) and 0.2 ± 0.18mm (IV). The higher Cr/Cd displacement could be a result of the relative size differences of the rat and feline spinal cord, the higher impact velocity (approximately 1.7mm/s compared to 0.13mm/s), and the placement of the markers in the center of the cord. If the Cr/Cd displacement does take a parabolic shape, as proposed by Blight in the theory above, larger displacements would be seen in the center of the cord compared to the outer portions of the cord. 4.2.2 Overall force versus displacement The average force-displacement response of the spinal cord during the dynamic injury can be compared to other rat SCI studies which have reported force-displacement curves. Figure 4-2 shows a comparison between the average force-displacement data from the dynamic contusions and data extracted from some of the few in vivo rat contusion studies that have reported this information (Bresnahan et al., 1987; Choo et al., 2007; Sparrey et al., 2008). A contusion injury to the cervical spinal cord of one rat using the IH impactor (Scheff and Roberts, 2009) was performed in our lab and was included in this comparison since no 89  Chapter 4: Discussion and Conclusions  other IH force-displacement graphs were found to be reported in literature. The forcedisplacement data from the current study correlated closely with the IH impactor data, as expected since the impact velocities were approximately the same. A larger impactor was used for the current study which could explain the higher force between 0.5-1mm compared to the IH impactor. Sparrey et al. impacted the thoracic spinal cord of rats at low and high speeds and reported the force-displacement response at both speeds (3mm/s and 300mm/s) (Sparrey et al., 2008). The forces from the current study are initially similar to the slower impact until about 0.5mm. At the higher displacements the maximum force exceeds the force generated by the slower impact and has a similar stiffness (the slope of the forcedisplacement plot) as the faster impact. It was thought that the forces would be more closely related to the higher speed impact, however differences could be attributed to the smaller size of the thoracic cord and potential differences in proportions of grey and white matter compared to the cervical cord. Bresnahan et al. also impacted the thoracic spinal cord of rats at 430mm/s which was a similar velocity to the higher speed impact in Sparrey et al.‘s study, but the forces were much lower (Bresnahan et al., 1987). The forces with this impact were also lower than the current study, which was not expected due to the higher impact velocity. The lower force could be attributed to the smaller sized impactor (2mm diameter compared to 4mm x 2.5mm) or methodological differences such as clamping which could put varying amounts of tension on the cord. Choo et al. impacted the cervical spinal cord of rats at 1m/s which resulted in higher forces than the current study (Choo et al., 2007). The higher force was expected due to the fact that the spinal cord is a viscoelastic material and was shown in the study by Sparrey et al. to be stiffer at higher strain rates (Cheng et al., 2008; Sparrey et al., 2008).  90  Chapter 4: Discussion and Conclusions  3  Sparrey et al., 2008 (3mm/s) Infinite Horizons (130mm/s) Lucas, current (130mm/s) Sparrey et al., 2008 (310mm/s) Bresnahan et al., 1987 (430mm/s) Choo et al., 2007 (1,000mm/s)  2.5  Force (N)  2  1.5  1  0.5  0 0  0.2  0.4  0.6  0.8 1 1.2 Displacement (mm)  1.4  1.6  1.8  Figure 4-2: Force displacement data extracted from contusion SCI animal studies (Bresnahan et al., 1987; Choo et al., 2007; Sparrey et al., 2008). Data were estimated from graphs in Choo et al., 2007 with permission from the American Association of Neurological Surgeons. Data were estimated from graphs in Bresnahan et al., 1987 with permission from Elsevier. Data were estimated from graphs in Sparrey et al., 2008 with permission from Wolters Kluwer/Lippincott, Williams & Wilkins.  A limitation of comparing force-displacement data between research groups is that differences such as impactor geometry, impactor spinal cord interface, spinal cord size, impact location, and clamping methods can affect the resulting force-displacement curves. Comparison between stress and strain would help correct for differences in geometry, but stress-strain graphs are rarely reported with contusion type impacts because of the difficulty in determining the contact area due to the cylindrical shape of the spinal cord. Thus, comparison of force-displacement data was used as a general idea of how the current study relates to other rat SCI studies that have been performed.  4.3  Impact parameters and injury severity Since injury severity was not assessed with the studies presented in this thesis, the  impact parameters will be compared to other studies for a general prediction of injury severity. The average compressions for the quasi-static study (Chapter 2) were 0.56mm, 1.03mm, and 1.35mm which corresponded to strains of 0.18, 0.34, and 0.45. This was on the lower end of other quasi-static compressions which have compressed the spinal cord 30-80% at rates between 0.02-3mm/s (Blight, 1991; Gruner et al., 1996; Huang et al., 2007; Hung et 91  Chapter 4: Discussion and Conclusions  al., 1982; Sparrey et al., 2008). At low velocities, SCI severity has been shown to correlate better with compression depth rather than rate of compression (Hung et al., 1982; Kearney et al., 1988). During normal flexion of the neck in human volunteers, strains of up to 14% have been reported, and although these were axial strains it suggests that the lowest compression group would have resulted in little to no injury (Yuan et al., 1998). A forceps compression test to a rat thoracic spinal cord showed that rats subjected to an overall compressive strain of 0.36 were indistinguishable from the sham group after 21 days (Gruner et al., 1996). This indicates that the middle compression groups might have resulted in minor injuries. Hung et al. reported mild injury severity (noticeable recovery of motor function in the rear limbs of the cats after ten days) after quasi-static compressions of approximately 50% strain suggesting that the highest compression group would have resulted in a mild injury (Hung et al., 1982). In the dynamic contusion study the spinal cord was impacted to an average depth of 1.1mm (approximately 37% of the cord diameter), an average speed of 132mm/s, and average maximum impact force of 1.14N. This was approximately the same speed as the IH impactor (Scheff et al., 2003) and was a similar compression depth as studies which have used the IH and other actuator driven impactors (0.8-1.3mm) (Choo et al., 2007; Pearse et al., 2005; Scheff et al., 2003; Sparrey et al., 2008). In a recent study the cervical spinal cord of adult rats were impacted using the IH and mild to moderate injuries (based on recovery of forelimb deficits) were reported with forces of approximately 2.0N-2.5N (Anderson et al., 2009). Since the average maximum force for the current dynamic study was 1.14N this would suggest that the resulting injuries would be more minor. While many researchers focus on the effect of one impact variable on injury severity such as impact depth, velocity, and force, it had been suggested that injury severity is dependent on a combination of these parameters (Dohrmann and Panjabi, 1976; Kearney et al., 1988). Kearney et al. impacted the cervical spinal cords of ferrets with a pneumatic impactor at various combinations of compression (25-65%) and velocity (1.5-6m/s) and found that the probability of recovery was more closely correlated with the product of velocity and compression (VC) than with either of these variables alone (Kearney et al., 1988). The data from this study cannot be directly compared to the current study, since the velocities are much higher than what was used for the dynamic study and rodent models of SCI typically use impact speeds 1m/s or 92  Chapter 4: Discussion and Conclusions  less. The VC of the current study can be compared to other rat contusion injury studies as an approach for estimating injury severity that takes into account both the compression depth and the velocity. The VC criteria is not commonly reported with SCI studies (to the author‘s knowledge only Kearney et al. have used this criterion for SCI), but the impact velocity has commonly been reported and the strain can be estimated using the reported compression depth and the approximate size of the spinal cord in the impacted region. The impact velocity of the current dynamic study was approximately 130mm/s with an overall strain of approximately 37% resulting in a VC of 48mm/s (130mm/s*0.37). Mild, moderate, and severe injuries (based on behavioral BBB score and spared white and grey matter) to the thoracic spinal cord of a rat with the IH impactor resulted in VC values of 39, 48, 58mm/s respectively (Scheff et al., 2003). Graded impacts to the rat cervical spinal cord with the OSU had VC values of 80, 95, and 110mm/s for mild, moderate, and severe injuries (based on functional tests such as the BBB score and histological analysis involving white and grey matter volumes), respectively (Pearse et al., 2005). These values suggest that the current dynamic impact might have caused a mild injury. It is important to note that comparisons of impact parameters and resulting injury severity between research groups is very difficult given different experimental protocols including animal type, impactor type, location of SCI, and clamping methods, as well as the actual classification of injury severity whether it be via functional, behavior, or histological measures. Additionally, impact parameters causing a mild injury in the thoracic spinal cord may cause a different severity of injury in the cervical cord (Pearse et al., 2005). In the future, it would be more useful to directly compare the spinal cord deformations during injury to detailed histological and functional results, rather than simply categorizing the injury into generic groupings of severity. For an accurate and more detailed assessment of the injury severity associated with the current studies, future studies would need to be conducted which include functional, behavior, and/or histological analysis.  4.4  Quasi-static compression compared to dynamic contusion The ability to compare internal deformations during different impact scenarios would  give insight into the mechanisms behind the resulting tissue damage. While the quasi-static compression study and the dynamic contusion study were not designed for direct 93  Chapter 4: Discussion and Conclusions  comparisons between the two, the impactors were of a similar size and the depth of compression for the dynamic test (1.1±0.10mm) was similar to the second compression group of the quasi static test (1.03±0.33mm). The spinal cord deformations from these two studies will be compared to investigate if the spinal cord moved differently under these two different injury mechanisms. The internal bead displacements from the dynamic contusions and the second compression group are shown in Figure 4-3. In the D/V direction there do not appear to be differences between the maximum displacements of the internal beads for the quasi-static or dynamic impacts. This could be due to the large standard deviation for the quasi-static displacements, or it is possible that the beads actually do move the same overall distance. The ventral surface beads appear to move differently in the two studies. The ventral surface beads in the dynamic study move in the dorsal direction compared to the ventral surface beads in the quasi-static study which move in the ventral direction. This could imply that the surface of the spinal cord moves differently depending on the rate of compression. In the Cr/Cd direction, it appears that the internal dorsal bead moves farther in the quasi-static compressions than the dynamic compressions. This could possibly be due to relaxation within the spinal cord during quasi-static compression causing more Cr/Cd displacement than with a dynamic impact. The higher displacement would not necessarily result in more damage since injury severity has been shown to be dependent on both rate and depth of compression (Kearney et al., 1988). However Sparrey et al. did report that quasi-static compressions resulted in a more longitudinal (Cr/Cd direction) spread of hemorrhage than the dynamic compressions (Sparrey et al., 2008), which supports the finding in the current study since the quasi-static compressions resulted in more Cr/Cd displacement than the dynamic contusions.  94  Chapter 4: Discussion and Conclusions  0.4  1.2  Dynamic Quasi-Static  0 -0.2 -0.4 -0.6 -0.8  0.8 0.6 0.4 0.2 0  -1 -1.2  Dynamic Quasi-Static  1 Displacement (mm)  Displacement (mm)  0.2  -0.2  A CdD  CrD  ID IV Bead Position  CdV  CrV  -0.4  B CdD  CrD  ID IV Bead Position  CdV  CrV  Figure 4-3: Comparison between the dorsal-ventral (A) and cranial-caudal (B) displacements of spinal cord beads during the dynamic contusion and the quasi-static 2mm compression group.  The average internal strain of the dynamic contusion was 0.49 which exceeded the estimated overall applied strain (0.37) by an average of 13%. The higher internal strains could possibly be due to inertia of the internal dorsal tissue allowing further deformation of the dorsal tissue when the impactor starts to slow down. The average internal strains with the quasi-static compression were 0.17, 0.32, and 0.45 which were similar to the estimated overall applied spinal cord strains of 0.18, 0.34, and 0.45. The inertial effects most likely would not have been seen with the quasi-static compressions since they were applied at slow rates, which could explain why the internal strains were similar to the overall strains. Finite element studies using separate grey and white matter properties have also reported high centralized strains during dynamic compressions (Ichihara et al., 2003; Maikos et al., 2008) and more uniform strains during quasi-static compressions (Ichihara et al., 2003). The finding of the current study would need to be verified with more testing which should include direct measurement of the spinal cord diameter and bi-planar x-ray imaging for more accurate strains measurements.  4.5  Finite element simulation compared to experimental dynamic contusion A FEM with and without bead inclusions was used as a general assessment of  whether the beads would displace more than the spinal cord tissue during impact due to an inertial effect. FEA of the 130mm/s dynamic impact showed that the bead motion increased the cord displacement about 1% of the maximum cord deformation indicating a very minor 95  Chapter 4: Discussion and Conclusions  inertial effect due to the bead inclusions. The maximum displacements of the internal beads during the FEM simulation of the 130mm/s impact were 0.66mm for the ID bead and 0.36mm for the IV bead. These displacement were close to the displacements with the experimental dynamic study (ID: 0.85 ± 0.13mm and IV: 0.30 ± 0.11mm) suggesting that the FEM was a close approximation of in vivo spinal cord movement. The average impact depth in the experimental test (1.1mm) was 0.1mm deeper than the finite element test (1.0mm) which could attribute to the difference in ID displacement. Differences in the ID displacement could also be due to the homogeneity of the material properties in the FEM (no difference between the white and grey matter). Differences in the amount of CSF surrounding the cord for the experimental and finite element tests could also affect cord motion since the CSF has been shown in experimental tests to reduce the amount of cord compression (Jones et al., 2008). The experimental tests would most likely have less CSF since holes were created in the dura during injection.  4.6  Strengths and limitations  4.6.1 Strengths To the author‘s knowledge, the current study represented the first time that internal spinal cord deformations had been tracked dynamically during impact. Overall spinal cord tissue mechanics during experimental SCI have been shown in many studies to be a predictor for injury (discussed in Chapter 1), but until now a method had not been developed for visualizing localized internal deformations during in vivo experimental SCI. Histological findings have shown different injury patterns inside of the cord result from different SCI mechanisms (Choo et al., 2007; Sparrey et al., 2008), but possible relationships between these findings and the internal mechanics of the in vivo spinal cord have not been determined. The methods developed in this thesis can now be used to directly image spinal cord deformations during a variety of impact scenarios. Conducting this study in vivo was a major strength since the spinal cord was kept in its natural physiologic environment during testing, and the spinal cord material properties would not have been affected by changes seen with ex vivo spinal cords (Oakland et al., 2006). A previous in vivo study imaged spinal cord deformations with a high-speed camera 96  Chapter 4: Discussion and Conclusions  during impact, and required extensive laminectomy which still would not have fully exposed the spinal cord due to the structure of the vertebrae (Hung et al., 1975). The use of radiography allows for imaging through the spinal canal enabling measurement of spinal cord deformation without the need for extensive laminectomies. Additionally, the bead injection method was developed so that minimal laminectomy was needed. This would allow for measurement of spinal cord deformation during contusion, distraction, and dislocation type injuries while leaving the spinal column intact (Choo et al., 2007). The visibility of the spinal canal in the x-ray images also enables direct measurement of spinal column movement during impact which would affect the overall compression and displacement measurements. Spinal column displacement during impact has previously either been assumed to be negligible since clamps were used to stabilize the spinal column (Anderson et al., 2009; Jakeman et al., 2009) or has been estimated by measuring the vertebral clamp motion (Young, 2009). The tantalum bead markers were small and compact with high contrast and could not deform during impact making them preferable over powder or liquid type markers for measuring displacement during experimental SCI. Due to the bead size and injection methods, multiple beads could be injected per spinal cord segment which enables measurement of internal D/V and Cr/Cd displacement at various levels in the cord. The beads could be injected into specific locations of the white and grey matter with a high success rate which can be used in the future to characterize deformation in different regions of the spinal cord. Bead migration and inertial effects during impact were assessed with experimental and FEM techniques which both seemed to imply that the beads have minimal movement with respect to the spinal cord tissue. A strength of the experimental assessment of bead migration was that it was conducted during the actual experimental test rather than ex vivo or under different testing conditions which could have altered the environment of the cord and changed the results. The bead tracking precision in the high-speed x-ray video was high, enabling detection of small difference in movement between different areas of the spinal cord. The speed of the x-ray system (3,000fps) also allowed for high temporal resolution of the bead displacement, considering the displacement took place over a very short period of time. The 97  Chapter 4: Discussion and Conclusions  ability to run at high-speeds was due to the quality of the image intensifiers and the light sensitivity of the high-speed cameras used to capture the x-ray images. The high level of precision was due to the contrast of the markers, the resolution of the x-ray system, and the custom denoising software which dramatically improved the image quality. 4.6.2 Limitations There were several limitations of this work, some of which were associated with method development and could be improved in future studies and others which were a necessary part of this technique. While the use of radiography allows for high-speed imaging the associated limitations are exposure to radiation, the size and cost of the high-speed x-ray equipment, and the necessity for fiducial marker inclusions. Since this was a terminal study radiation to the animal was not a factor. In survival studies the radiation can be limited to the duration of impact. For quasi-static type studies where longer exposure times would be needed, static or high-speed x-ray images could be taken in brief intervals during compression. The cost of the high-speed x-ray equipment could make implementation of this approach difficult for other labs. Standard x-ray equipment triggered to take images at specific times during impact would be a lower cost option. The high-speed x-ray equipment also requires a large space for operation and could be difficult to implement without appropriate facilities. The use of radiography required fiducial marker inclusions which could possibly damage the spinal cord tissue, migrate with respect to spinal cord during impact, change the mechanics of the spinal cord, or cause tissue reactions which could inhibit survival studies. Tantalum beads were chosen because of the biocompatibility of the material (Aronson et al., 1985), but histology could not be performed to assess tissue reactions or damage due to injection because the blade used for histology would not be able to cut through the bead. The injection track was visible due to hemorrhage inside of the track and appeared to be minimal, but it was difficult to assess the overall size of the track given the thick slices of the spinal cord being imaged. To the author‘s knowledge there have been no studies that have looked at tissue damage due to spinal cord injection, though treatments have been administered into the thoracic spinal cord of rats using 33 gauge needles (approximately 210um) and the injections were shown to cause no behavioral abnormalities (Rosenberg et al., 1999; Teng and 98  Chapter 4: Discussion and Conclusions  Wrathall, 1997). To further assess damage, histology could be performed around the bead location, or tantalum powder could be injected into the cord so that histology can be performed directly at the injection location. Survival studies could also be conducted to assess functional deficit due to the injections. Since histology cannot be performed with these beads, parallel experiments would be needed for comparison of spinal cord biomechanics to histological results. The presence of an injection track could possibly cause the beads to translate within the track during impact. This would likely be due to inertial forces which have exceeded the static friction of the bead with the surrounding tissue. It was assumed that the change from the static to the dynamic state would be visible in the bead displacement data during the dynamic impact since the dynamic friction forces would be lower than the static friction forces. This change in displacement was not noticed in the displacement data. Additionally, any displacement within the track should be detectable with the bead migration assessment since this would likely cause variability in the post-impact bead displacement. The possibility of bead migration was assessed by comparing the post-compression displacements of the internal beads to the surface beads. It was assumed that surface bead post-compression displacement would be a measure of residual cord deformation due to the impact and that internal bead post-compression displacements exceeding the neighboring surface beads would be an indication for bead migration. The main assumption with this analysis was that the internal spinal cord would have the same residual deformation as the surface of the spinal cord after impact. This was a limitation of this approach since the relationship between the residual deformation of the internal spinal cord and the surface of the spinal cord is not known. It is possible that the surface of the spinal cord would deform less than the internal spinal cord because the surface is directly coupled to the pia mater which has been shown to restore the shape of the spinal cord after compression (Ozawa et al., 2004). If this is true, it would make the current assessment of migration an overestimate. A FEM of the spinal cord was used to analyze the inertial effects of the beads during impact. FEMs of the spinal cord have not been validated against in vivo internal spinal cord strains during impact making this a limitation for use with the current study. Additionally, many FEMs of the spinal cord have only been validated for quasi-static loading since there is very limited biomechanical data currently available during dynamic loading. However, the 99  Chapter 4: Discussion and Conclusions  FEM that was used with this study was similar to the rat spinal cord FEM created by Maikos et al. which has been validated against drop weight type impacts at 0.49m/s (Maikos et al., 2008). While the internal deformations have not been validated in this model, they were similar to the deformations found in the current study suggesting that the model was appropriate for a general assessment of the effects of the beads during impact. An additional limitation of this approach was that the bead mass was only added to one node, and the full geometry was not modeled. This was thought to represent a worst case scenario for inertial effects since all of the mass was concentrated to one point and its motion would be resisted by less spinal cord tissue. In the future, the full geometry could be modeled to see if there are any changes in the cord mechanics. There were some limitations associated with the surgical procedure. For both studies the scapulae were removed as they were shown to interfere with the visualization of the beads. For survival studies, alternative approaches should be taken such as hooking small weights to the front limbs to pull the scapulae downwards and out of the beam of the x-ray. The need to puncture a hole in the dura was also a limitation since it resulted in the loss of CSF which has been shown to protect the spinal cord (Jones et al., 2008). It was noticed that only a small amount of CSF leaked from the holes during injection, and the injection depths and angles were chosen so only one hole was needed for multiple injections. The addition of the surface beads required more holes in the dura which likely caused more CSF leakage. In the future the holes could be covered with Vaseline to help prevent further leaking during impact. While the placement of the bead injections within the white and grey matter was shown to be fairly successful, the bead position needed to be confirmed after impact since on average 15% of the beads were not in the correct location for the injection accuracy study. Four of the twelve ventral beads in the dynamic study were in grey matter instead of white matter. For severe impacts, it may be difficult to discern between white and grey matter and the tissue may have displaced due to impact, so an alternative method of assessing bead position should be considered. A possible approach would be to take pre-surgical MR images and overlay CT or bi-planar high-speed x-ray images after bead injection to verify bead location within the white and grey matter. 100  Chapter 4: Discussion and Conclusions  An additional limitation of the bead inclusions was that only two can be injected per spinal cord segment due to the size of the beads and the precision of the injections. The use of smaller beads or a larger animal model would enable more bead injections per segment and more precise measurements of internal spinal cord deformations. Since the beads are currently close to the high-speed x-ray resolution, using smaller beads would require an image intensifier with a smaller field of view resulting in higher resolution of the x-ray images. Since only a small portion of the four inch field of view was used for the dynamic study, using a smaller field of view would definitely be a feasible option. For both studies, dural-touch was visually confirmed and used as the starting position of the impactor which may have caused variability between animals. Use of a sensor to confirm dura touch could help reduce this variability (Young, 2009). The alignment of the spinal column with the impactor was also visually verified. This was easier with the quasistatic study than with the dynamic study since a stereotaxic frame was used which allowed adjustment of the tooth bar to straighten the spinal column. For the dynamic study the rats were not in a stereotaxic frame because of restrictions in space with the Instron equipment. The rats were partially suspended with articulating arms attached to the spinal column clamps, and the clamps were adjusted so that the spinal cord was parallel to the bottom of the impactor. Slight misalignments between the impactor and the spinal cord may have caused some of the bead displacements in the Cr/Cd direction. Additional testing would need to be conducted with the impactor precisely aligned with the spinal column to verify the presence of Cr/Cd displacement. Clamps were used to stabilize the spinal column during the dynamic impact, but every animal twitched after maximum compression which caused small movement of the clamps and increased compression to the cord. This resulted in variation between animals so the bead displacements could only be analyzed up to the maximum impactor displacement (before the twitch). In the future, the clamps should be more rigidly attached to a frame to limit variability between animals. For the quasi-static study, clamps were not used to stabilize the spinal column, so the compression depth was less than intended for each compression group. Since the spinal column was visible in the x-ray images, the actual compression to the spinal cord could be determined by subtracting the impactor displacement 101  Chapter 4: Discussion and Conclusions  from the spinal column displacement. The compression depth significantly increased for each compression group, but the standard deviations were high causing variability in the spinal cord bead displacements. The lack of spinal column clamps also allowed for movement of the spinal column and spinal cord during respiration. A series of images were taken of a rat before impact and the spinal cord bead positions were shown to vary about 35% of a bead diameter which causes variability in the deformation measurements. The use of clamps and slight suspension of the animal would reduce breathing effects. The repeated compressions to the spinal cord for the quasi-static study may have caused damage to the spinal cord tissue which would affect the displacement of the spinal cord during compression. Hung et al. showed that the spinal cord exhibits the same forcedisplacement response for repeated quasi-static compressions under 36% strain which was greater than the first two compression groups (Hung et al., 1982). The velocity of compression was not directly controlled which could also cause some variation. The same study by Hung et al. showed that with quasi-static compressions, the rate of compression did not change the spinal cord force-displacement. The bead tracking precision for the post-compression displacements in the quasistatic study was higher than the differences found between the internal beads and the surface beads, limiting the bead migration assessment. The level of precision was still about 34% of the bead diameter which should be sufficient for measuring any major amounts of migration. Additionally, since the impacts were quasi-static, inertial effects could be neglected so the tendency for migration would be lower than with a dynamic impact. Precision could be increased for this type of study by having a circular bead reference point on the testing frame to register each image together more accurately. A precise cassette holder could also be used to ensure the same placement with each image. The use of fluoroscopy would also increase precision since there is no need to change cassettes between images. Lateral bead movement was not accounted for in either of these studies since only lateral images were acquired. Since the injection location was completely underneath of the impactor and the beads were located close to the mid-sagittal plane, it was assumed that lateral bead movement would be minimal. Future studies could use bi-planar x-ray to image three dimensional bead motions, which would enable further characterization of spinal cord 102  Chapter 4: Discussion and Conclusions  deformation during impact. The use of bi-planar imaging would also help limit exclusion of beads due to bead overlap. There are also limitations with the measurement and comparison of the overall and internal strains with these studies. The spinal cord diameters were estimated for the strain calculations using an atlas of a rat spinal cord and were not directly measured. This would obviously affect the overall strain measurement and in the future the spinal cord diameter should be directly measured. The measurement of internal strains assumed that the beads were in the same plane and no cord rotation or lateral bead movement occurred. These limitations could affect the strain measurements and the comparisons between the internal and surface strains. The main purpose of these measurements was to estimate the overall and internal strains and demonstrate a potential method of analyzing the spinal cord deformations and comparing them between studies. In the future bi-planar x-ray should be used to more accurately measure internal spinal cord strains. For the dynamic study, the impact speed was on the lower end of the range of impact speeds commonly used with rodent models (0.1m/s-1m/s) (Basso et al., 1996; Choo et al., 2007; Ghasemlou et al., 2005; Maikos and Shreiber, 2007; Scheff et al., 2003; Sparrey et al., 2008). The ability to track higher speed impacts was not directly assessed, but it is thought that the high-speed x-ray images could be obtained at 4,000-5,000fps by increasing the current to the x-ray tube. At an impact speed of 1m/s, a 1mm impact depth would take approximately 1ms and could be captured with four to five frames. This should be sufficient for capturing the dynamic movement of the beads during impact.  4.7  Future work The methodology presented in this thesis has been developed with the intention to be  used with larger animal studies with more bead patterns and various impact scenarios to fully understand the biomechanical response of the spinal cord during impact. It has been postulated through analysis of histological findings and mechanical testing of ex vivo white and grey matter that the internal structures of the spinal cord respond differently to impact (Blight, 1988; Ichihara et al., 2001). Future studies can involve injecting beads into various locations of the white and grey matter to compare differences in deformations in these 103  Chapter 4: Discussion and Conclusions  regions. Bi-planar x-ray is available in the Orthopaedic Injury Biomechanics Lab at the University of British Columbia and it could be used for a three dimensional measurements of deformation. Since only two beads can be injected per segment with a rat model, bead patterns could be strategically chosen so that results can be overlapped. Figure 4-4 demonstrates this idea where three groups of animals have two bead injection locations each, and one injection location overlaps with another group. This type of approach would help characterize strains as if more than two beads were injected per segment. Larger animal models such as pigs have larger spinal cords so more beads could be injected into the spinal cord. Currently, the Orthopaedic Injury Biomechanics Group at UBC has a pig SCI model that has been developed, so the bead tracking could be implemented with this model. In the future, the animal‘s blood pressure should be regulated during the surgical and impact procedures since a change in perfusion pressure may alter the cord‘s response to impact. While the current studies did not involve survival or histological analysis, survivability after bead injections could be assessed in future studies. If the rats are not adversely affected by the bead injections, the spinal cord deformations could be compared to functional deficit through behavioral analysis. Histology can also be performed either with a parallel group of animals without bead injections, or with the inclusion of powder markers. The internal strain and strain rates can be mapped to the histological analysis of primary and secondary injury to determine if strain and/or strain rate are predictors for histological findings. The internal strains and strain rates could also be compared to findings in MRI. If MRI results could be related to strain, it may be possible to know more about how an injury occurred just by analyzing a patient‘s MRI. Spinal cord strains during different types of SCI such as contusion, distraction, and dislocation could be compared in order to understand differences in how the spinal cord becomes injured in these scenarios. The added knowledge of the internal deformation during different injury scenarios could help develop targeted treatments for areas known to suffer high strains/strain rates.  104  Chapter 4: Discussion and Conclusions  Figure 4-4: Example of bead patterns used to obtain an overall map of localized spinal cord deformations.  Comparison of internal strains during contusion, distraction, and dislocation has been done using a FEM of the human spinal cord (Greaves et al., 2008), and internal strains in a FEM model of a rat spinal cord have been shown to correlate with compromise to the blood spinal cord barrier during a drop weight impact (Maikos and Shreiber, 2007). However, these internal strains in the FEMs have not been validated against in vivo spinal cord deformations during a dynamic injury. Using the current bead tracking methodology, internal spinal cord deformations can now be tracked in various impact scenarios and used to validate FEMs. If positive correlations were made between the internal strains and histological and functional outcomes, injury tolerances of the spinal cord tissue could be determined and results from FEMs could be directly linked to the injury severity outcome. The spinal cord deformations can also be used for validating surrogate spinal cords. FEM and surrogate cord models can be used for assessing the efficacy of preventative devices such as helmets and automotive safety features. This approach could also be used as a mechanical measure for determining the repeatability of impact methods, or for comparing different types of impactors to each other like the NYU, OSU, and IH in order understand the similarities and differences of these impactors from a mechanical standpoint. This methodology could be further developed for tracking deformation of brain during traumatic brain injury. This has been done in cadaver human brains with high-speed 105  Chapter 4: Discussion and Conclusions  x-ray and fiducial markers (Hardy et al., 2007) and in rat brains with repeated uninjurious impacts and tagged MRI (Bayly et al., 2006) but has yet to be done with single impacts with in vivo animals. These beads could also be added to other organs in the body to determine biomechanical properties or to image deformation of the organ inside of a cadaver during injurious situations such as automobile accidents (Hardy et al., 2008).  4.8  Conclusions A method was developed for injecting radio-opaque markers into the grey and white  matter of the cervical spinal cord of in vivo rats. The displacement of the spinal cord beads were tracked during quasi-static and dynamic type SCIs using standard medical x-ray equipment and high speed x-ray equipment. Migration of the fiducial markers with respect to the spinal cord tissue was assessed with finite element modeling and experimental measures and appeared to be small and insignificant. The bead displacements during impact varied depending on the location within the cord, with the dorsal beads typically moving farther than the ventral beads. The internal spinal cord beads moved more in the dorsal-ventral and cranial-caudal directions than the neighboring beads glued to the surface of the spinal cord. The internal strains spinal cord strains appeared to exceed to overall applied strain in the dynamic contusions, while the internal and overall strains were similar for the quasi-static compressions. The results indicate that the spinal cord deforms in a complex manner that is dependent on the location within the cord. In the future, the spinal cord strains and strain rates could be compared to histological and functional outcome to determine tissue tolerances and the dependence of injury severity on strain and strain rate. Surrogate and finite element models could also be validated by comparing internal deformations to deformations found using the methodology presented in this thesis.  106  Chapter 4: Discussion and Conclusions  4.9  References  Al-Bsharat, A. S., W. N. Hardy, K. H. Yang, T. B. Khalil, S. Tashman and A. I. King (1999). Brain/Skull relative displacement magnitude due to blunt head Impact: New experimental data and model. Proc. 43rd Stapp Car Crash Conference, pp. 321-332, Society of Automotive Engineers, Warrendale, PA. Anderson, K. D., K. G. Sharp and O. Steward (2009). "Bilateral cervical contusion spinal cord injury in rats." Exp Neurol 220(1): 9-22. Aronson, A. S., N. Jonsson and P. Alberius (1985). "Tantalum markers in radiography. An assessment of tissue reactions." Skeletal Radiol 14(3): 207-211. Basso, D. M., M. S. Beattie and J. C. Bresnahan (1996). "Graded histological and locomotor outcomes after spinal cord contusion using the NYU weight-drop device versus transection." Exp Neurol 139(2): 244-256. Bauman, J. M. and Y.-H. Chang (2010). "High-speed X-ray video demonstrates significant skin movement errors with standard optical kinematics during rat locomotion." Journal of Neuroscience Methods 186(1): 18-24. Bayly, P. V., E. E. Black, R. C. Pedersen, E. P. Leister and G. M. Genin (2006). "In vivo imaging of rapid deformation and strain in an animal model of traumatic brain injury." Journal of Biomechanics 39(6): 1086-1095. Bayly, P. V., T. S. Cohen, E. P. Leister, D. Ajo, E. C. Leuthardt and G. M. Genin (2005). "Deformation of the human brain induced by mild acceleration." J Neurotrauma 22(8): 845-856. Blight, A. (1988). "Mechanical factors in experimental spinal cord injury." J Am Paraplegia Soc 11(2): 26-34. Blight, A. R. (1991). "Morphometric analysis of a model of spinal cord injury in guinea pigs, with behavioral evidence of delayed secondary pathology." J Neurol Sci 103(2): 156171. Blight, A. R. and V. Decrescito (1986). "Morphometric analysis of experimental spinal cord injury in the cat: the relation of injury intensity to survival of myelinated axons." Neuroscience 19(1): 321-341. Brainerd, E. L., D. B. Baier, S. M. Gatesy, T. L. Hedrick, K. A. Metzger, S. L. Gilbert and J. J. Crisco (2010). "X-ray reconstruction of moving morphology (XROMM): precision, accuracy and applications in comparative biomechanics research." J Exp Zool A Ecol Genet Physiol. Bresnahan, J. C., M. S. Beattie, F. D. Todd, 3rd and D. H. Noyes (1987). "A behavioral and anatomical analysis of spinal cord injury produced by a feedback-controlled impaction device." Exp Neurol 95(3): 548-570. Cheng, S., E. C. Clarke and L. E. Bilston (2008). "Rheological properties of the tissues of the central nervous system: a review." Med Eng Phys 30(10): 1318-1337. Choo, A. M., J. Liu, C. K. Lam, M. Dvorak, W. Tetzlaff and T. R. Oxland (2007). "Contusion, dislocation, and distraction: primary hemorrhage and membrane permeability in distinct mechanisms of spinal cord injury." J Neurosurg Spine 6(3): 255-266. Craig, M., C. Bir, D. Viano and S. Tashman (2008). "Biomechanical response of the human mandible to impacts of the chin." J Biomech 41(14): 2972-2980.  107  Chapter 4: Discussion and Conclusions  Dohrmann, G. J. and M. M. Panjabi (1976). ""Standardized" spinal cord trauma: biomechanical parameters and lesion volume." Surg Neurol 6(5): 263-267. Ghasemlou, N., B. J. Kerr and S. David (2005). "Tissue displacement and impact force are important contributors to outcome after spinal cord contusion injury." Exp Neurol 196(1): 9-17. Greaves, C. Y., M. S. Gadala and T. R. Oxland (2008). "A three-dimensional finite element model of the cervical spine with spinal cord: an investigation of three injury mechanisms." Ann Biomed Eng 36(3): 396-405. Gruner, J. A., A. K. Yee and A. R. Blight (1996). "Histological and functional evaluation of experimental spinal cord injury: evidence of a stepwise response to graded compression." Brain Res 729(1): 90-101. Hardy, W. N., C. D. Foster, M. J. Mason, K. H. Yang, A. I. King and S. Tashman (2001). "Investigation of Head Injury Mechanisms Using Neutral Density Technology and High-Speed Biplanar X-ray." Stapp Car Crash J 45: 337-368. Hardy, W. N., M. J. Mason, C. D. Foster, C. S. Shah, J. M. Kopacz, K. H. Yang, A. I. King, J. Bishop, M. Bey, W. Anderst and S. Tashman (2007). "A study of the response of the human cadaver head to impact." Stapp Car Crash J 51: 17-80. Hardy, W. N., C. S. Shah, M. J. Mason, J. M. Kopacz, K. H. Yang, A. I. King, C. A. Van Ee, J. L. Bishop, R. F. Banglmaier, M. J. Bey, R. M. Morgan and K. H. Digges (2008). "Mechanisms of traumatic rupture of the aorta and associated peri-isthmic motion and deformation." Stapp Car Crash J 52: 233-265. Huang, W. L., K. J. George, V. Ibba, M. C. Liu, S. Averill, M. Quartu, P. J. Hamlyn and J. V. Priestley (2007). "The characteristics of neuronal injury in a static compression model of spinal cord injury in adult rats." Eur J Neurosci 25(2): 362-372. Hung, T. K., M. S. Albin, T. D. Brown, L. Bunegin, R. Albin and P. J. Jannetta (1975). "Biomechanical responses to open experimental spinal cord injury." Surg Neurol 4(2): 271-276. Hung, T. K., H. S. Lin, L. Bunegin and M. S. Albin (1982). "Mechanical and neurological response of cat spinal cord under static loading." Surg Neurol 17(3): 213-217. Ichihara, K., T. Taguchi, I. Sakuramoto, S. Kawano and S. Kawai (2003). "Mechanism of the spinal cord injury and the cervical spondylotic myelopathy: new approach based on the mechanical features of the spinal cord white and gray matter." J Neurosurg 99(3 Suppl): 278-285. Ichihara, K., T. Taguchi, Y. Shimada, I. Sakuramoto, S. Kawano and S. Kawai (2001). "Gray matter of the bovine cervical spinal cord is mechanically more rigid and fragile than the white matter." J Neurotrauma 18(3): 361-367. Jakeman, L. B., D. M. McTigue, P. Walters and B. T. Stokes (2009). The Ohio State University ESCID Spinal Cord Contusion Model. Animal Models of Acute Neurological Injuries: 433-447. Jones, C. F., S. G. Kroeker, P. A. Cripton and R. M. Hall (2008). "The effect of cerebrospinal fluid on the biomechanics of spinal cord: an ex vivo bovine model using bovine and physical surrogate spinal cord." Spine 33(17): E580-588. Kearney, P. A., S. A. Ridella, D. C. Viano and T. E. Anderson (1988). "Interaction of contact velocity and cord compression in determining the severity of spinal cord injury." J Neurotrauma 5(3): 187-208.  108  Chapter 4: Discussion and Conclusions  Maikos, J. T., Z. Qian, D. Metaxas and D. I. Shreiber (2008). "Finite element analysis of spinal cord injury in the rat." J Neurotrauma 25(7): 795-816. Maikos, J. T. and D. I. Shreiber (2007). "Immediate damage to the blood-spinal cord barrier due to mechanical trauma." J Neurotrauma 24(3): 492-507. Maiman, D. J., J. Coats and J. B. Myklebust (1989). "Cord/spine motion in experimental spinal cord injury." J Spinal Disord 2(1): 14-19. Margulies, S. S., D. F. Meaney, L. B. Bilston, L. Thibault, N. G. Campeau and S. J. Riederer (1992). In Vivo Motion of the Human Cervical Spinal Cord in Extension and Flexion. IRCOBI Conference, Verona, Italy. Martin, D. E., N. J. Greco, B. A. Klatt, V. J. Wright, W. J. Anderst and S. Tashman (2010). "Model-Based Tracking of the Hip: Implications for Novel Analyses of Hip Pathology." J Arthroplasty. Miller, J. M., E. A. Rossi, M. Wiesmair, D. E. Alexander and O. Gallo (2006). "Stability of gold bead tissue markers." J Vis 6(5): 616-624. Oakland, R. J., R. M. Hall, R. K. Wilcox and D. C. Barton (2006). "The biomechanical response of spinal cord tissue to uniaxial loading." Proc Inst Mech Eng [H] 220(4): 489-492. Ozawa, H., T. Matsumoto, T. Ohashi, M. Sato and S. Kokubun (2004). "Mechanical properties and function of the spinal pia mater." J Neurosurg Spine 1(1): 122-127. Panjabi, M. M., M. Kifune, L. Wen, M. Arand, T. R. Oxland, R. M. Lin, W. S. Yoon and A. Vasavada (1995). "Dynamic canal encroachment during thoracolumbar burst fractures." J Spinal Disord 8(1): 39-48. Panjabi, M. M., T. Oda and J. L. Wang (2000). "The effects of pedicle screw adjustments on neural spaces in burst fracture surgery." Spine (Phila Pa 1976) 25(13): 1637-1643. Pearse, D. D., T. P. Lo, Jr., K. S. Cho, M. P. Lynch, M. S. Garg, A. E. Marcillo, A. R. Sanchez, Y. Cruz and W. D. Dietrich (2005). "Histopathological and behavioral characterization of a novel cervical spinal cord displacement contusion injury in the rat." J Neurotrauma 22(6): 680-702. Rosenberg, L. J., Y. D. Teng and J. R. Wrathall (1999). "Effects of the sodium channel blocker tetrodotoxin on acute white matter pathology after experimental contusive spinal cord injury." J Neurosci 19(14): 6122-6133. Scheff, S. and K. N. Roberts (2009). Infinite Horizon Spinal Cord Contusion Model. Animal Models of Acute Neurological Injuries: 423-432. Scheff, S. W., A. G. Rabchevsky, I. Fugaccia, J. A. Main and J. E. Lumpp, Jr. (2003). "Experimental modeling of spinal cord injury: characterization of a force-defined injury device." J Neurotrauma 20(2): 179-193. Sparrey, C. J., A. M. Choo, J. Liu, W. Tetzlaff and T. R. Oxland (2008). "The distribution of tissue damage in the spinal cord is influenced by the contusion velocity." Spine (Phila Pa 1976) 33(22): E812-819. Tashman, S. and W. Anderst (2003). "In-vivo measurement of dynamic joint motion using high speed biplane radiography and CT: application to canine ACL deficiency." J Biomech Eng 125(2): 238-245. Teng, Y. D. and J. R. Wrathall (1997). "Local blockade of sodium channels by tetrodotoxin ameliorates tissue loss and long-term functional deficits resulting from experimental spinal cord injury." J Neurosci 17(11): 4359-4366.  109  Chapter 4: Discussion and Conclusions  Young, W. (2009). MASCIS Spinal Cord Contusion Model. Animal Models of Acute Neurological Injuries: 411-421. Yuan, Q., L. Dougherty and S. S. Margulies (1998). "In vivo human cervical spinal cord deformation and displacement in flexion." Spine (Phila Pa 1976) 23(15): 1677-1683. Zou, H., J. P. Schmiedeler and W. N. Hardy (2007). "Separating brain motion into rigid body displacement and deformation under low-severity impacts." J Biomech 40(6): 11831191.  110  Appendix A: Fiducial Marker Selection  Appendix A: Fiducial Marker Selection A1  Marker type evaluation The markers evaluated were tantalum, tungsten, barium, lead metallic powder (Table  A-1) and tantalum, aluminum oxide, calcium carbonate miniature beads (Table A-2). Two different injection techniques were used to inject the powder and the beads. For the powder injections, a custom injector was made by attaching a glass needle (180µm diameter) to the cannula of a 2mL Hamilton syringe with melted wax (Figure A-1). The powder was mixed with Ringer‘s solution at the maximum concentration that could be injected with the syringe. The needle was then slowly inserted into the spinal cord to the desire marker depth, and then retracted 200µm to create a pocket for injection. The powder was then injected and the needle was withdrawn one minute after injection to allow the powder to settle to the bottom of the channel. For the bead injections, the methods used were those discussed in Chapters 2 and 3. The rat was euthanized after injections and x-ray images were taken (65kV, 10ma) to assess the visibility of marker types. Table A-1: Powder type fiducial markers evaluated.  Powder Barium Lead Tungsten Tantalum  Concentration (mg/0.1ml dH20) 100 60 500 20  Volume (µL) 0.1,0.2 0.2 0.2 0.2  Supplier Scholar Chemistry, West Henrietta, NY Boreal Northwest, St. Catharines, ON Inframat Advanced Materials, Willington, CT Inframat Advanced Materials, Willington, CT  Table A-2: Bead type fiducial markers evaluated.  Beads Tantalum Calcium carbonate Aluminum oxide  Bead diameter (µm) 260 & 400 200 & 400 200 & 400  Supplier Bal-tec, Los Angeles, CA Harper International, Lancaster, NY Harper International, Lancaster, NY  111  Appendix A: Fiducial Marker Selection  Figure A-1: Powder injection needle shown withdrawing powder in preparation for injection.  A2  Marker selection The tantalum beads had the best visibility in the x-ray images and were much easier  to inject than the powder (Figure A-2). While the 400µm beads had the best contrast, the smaller 260µm beads were selected since they occupy less of the spinal cord. The calcium carbonate and aluminum oxide beads and the tungsten and tantalum powders were not visible in the x-ray images. The barium and lead powder were visible in x-ray though the required marker size was larger than 400µm for adequate contrast. The powder injection method was very difficult and unreliable because the powders frequently clogged the injection needle and the size and shape of the marker was variable. The size of the powder markers typically needed to be larger than the tantalum beads to have the same level of contrast. The barium markers were the most reliable powder markers and had better contrast than the lead powder.  112  Appendix A: Fiducial Marker Selection  400µm Tantalum  Lead powder 260µm Tantalum Barium powder Spinal column  Figure A-2: Lateral x-ray of rat cervical spine showing comparison of 400µm tantalum beads, barium powder, lead powder, and 260µm tantalum beads.  113  Appendix B: Quasi-Static Bead Displacements  Appendix B: Quasi-Static Bead Displacements Table B-1: Spinal cord bead displacements during compression. All values are in millimeters and values in bold are Avg±STD of each bead group. Negative values represent cranial or ventral displacement. Displacement During Compression (mm) Compression Depth Bead  Caudal Dorsal  Cranial Dorsal  Internal Dorsal  Caudal Ventral  0.5mm Cr/Cd Direction 0.008 0.007 0.009 -0.017 -0.128 0.042 0.020 0.069 0.037 0.01 ± 0.06 -0.002 0.007 0.022 0.019 -0.190 0.022 -0.031 0.008 0.003 -0.02 ± 0.07 0.049 0.008 0.022 0.267 0.068 0.189 0.103 0.153 0.169 0.11 ± 0.09 0.022 -0.028 -0.027 -0.032 -0.138 0.023 -0.014 0.003 0.017 -0.02 ± 0.05  D/V Direction -0.259 -0.071 -0.039 -0.404 -0.936 -0.378 -0.334 -0.439 -0.408 -0.36 ± 0.26 -0.118 -0.033 -0.026 -0.184 -0.749 -0.282 -0.028 -0.334 -0.363 -0.24 ± 0.23 -0.292 -0.414 -0.316 -0.413 -0.934 -0.232 -0.308 -0.438 -0.254 -0.40 ± 0.21 0.028 -0.084 -0.039 -0.200 -0.699 -0.086 -0.186 -0.122 -0.106 -0.17 ± 0.21  2mm Cr/Cd Direction 0.040 0.139 0.044 0.181 0.117 0.007 0.167 0.449 0.112 0.14 ± 0.13 0.014 0.186 0.037 0.177 0.163 -0.031 0.127 0.449 0.108 0.14 ± 0.14 0.327 0.220 0.100 0.655 0.765 0.461 0.220 0.669 0.303 0.41 ± 0.24 0.036 0.096 -0.019 0.071 0.098 0.051 0.054 0.337 0.089 0.09 ± 0.10  D/V Direction -0.757 -0.670 -0.281 -0.872 -0.801 -0.973 -0.647 -1.595 -0.489 -0.79 ± 0.37 -0.621 -0.592 -0.730 -0.691 -0.682 -0.911 -0.433 -1.550 -0.444 -0.74 ± 0.34 -0.678 -0.815 -0.903 -0.797 -0.820 -0.817 -0.614 -1.551 -0.325 -0.81 ± 0.33 -0.118 -0.456 -0.171 -0.495 -0.312 -0.425 -0.543 -0.989 -0.133 -0.40 ± 0.27  3mm Cr/Cd Direction 0.068 0.041 0.129 0.278 0.211 0.093 0.326 0.123 0.045 0.15 ± 0.10 0.058 0.054 0.185 0.293 0.298 Covered 0.288 0.137 0.048 0.17 ± 0.11 0.505 0.246 0.221 0.860 0.944 Covered 0.409 0.318 0.311 0.48 ± 0.28 0.079 0.068 0.021 0.189 0.102 0.273 0.115 0.123 0.079 0.12 ± 0.07  D/V Direction -0.769 -0.722 -0.329 -1.062 -0.976 -1.742 -0.894 -1.348 -0.823 -0.96 ± 0.40 -0.693 -0.615 -0.564 -0.956 -0.891 Covered -0.618 -1.362 -0.795 -0.81 ± 0.26 -0.815 -1.028 -0.937 -0.985 -0.903 Covered -0.877 -1.272 -0.682 -0.94 ± 0.17 -0.143 -0.501 -0.262 -0.615 -0.537 -1.051 -0.797 -0.795 -0.310 -0.56 ± 0.29  114  Appendix B: Quasi-Static Bead Displacements  Table B-1 continued: Displacement During Compression (mm) Compression Depth Bead  Cranial Ventral  Internal Ventral  0.5mm Cr/Cd Direction -0.005 0.014 -0.033 -0.028 -0.165 0.030 No Bead -0.011 0.029 -0.02 ± 0.06 0.033 0.044 0.013 0.126 0.042 0.071 -0.009 0.098 0.054 0.05 ± 0.04  2mm  D/V Direction 0.021 -0.031 -0.001 -0.103 -0.744 -0.034 No Bead -0.150 -0.114 -0.14 ± 0.25 -0.135 -0.172 -0.082 -0.080 -0.898 -0.056 -0.127 -0.121 -0.036 -0.19 ± 0.27  Cr/Cd Direction 0.065 0.077 0.016 0.095 0.010 0.016 No Bead 0.297 0.121 0.09 ± 0.09 0.277 0.130 0.106 0.405 0.724 0.130 0.082 0.582 0.150 0.29 ± 0.23  D/V Direction -0.204 -0.430 -0.223 -0.424 -0.481 -0.440 No Bead -1.031 -0.147 -0.42 ± 0.28 -0.314 -0.350 -0.433 -0.314 -0.274 -0.413 -0.357 -1.110 -0.088 -0.41 ± 0.28  3mm Cr/Cd Direction 0.064 0.005 0.012 0.186 0.188 Covered No Bead 0.107 0.108 0.10 ± 0.07 0.391 0.040 0.136 0.577 0.820 0.351 0.177 0.262 0.185 0.33 ± 0.24  D/V Direction -0.243 -0.555 -0.294 -0.675 -0.681 Covered No Bead -0.848 -0.308 -0.51 ± 0.23 -0.372 -0.496 -0.508 -0.511 -0.512 -1.168 -0.553 -0.785 -0.240 -0.57 ± 0.27  Table B-2: Spinal cord bead post-compression displacements. All values are in millimeters and values in bold are Avg±STD of each bead group. Negative values represent cranial or ventral displacement. Post-Compression Displacement (mm) Compression Depth Bead  Caudal Dorsal  Cranial Dorsal  0.5mm Cr/Cd Direction -0.015 -0.037 -0.007 -0.128 -0.109 0.027 0.046 -0.015 0.007 -0.03 ± 0.06 -0.023 0.069 -0.005 0.082 0.120 -0.026 0.036 -0.012 -0.033 -0.02 ± 0.06  D/V Direction -0.066 0.010 -0.030 -0.124 -0.756 -0.060 -0.068 -0.072 -0.194 -0.15 ± 0.23 -0.056 -0.008 -0.021 -0.062 -0.718 -0.043 0.065 -0.011 -0.165 -0.11 ± 0.23  2mm Cr/Cd Direction -0.012 0.04 -0.086 -0.026 -0.001 -0.024 -0.073 0.038 0.067 -0.01 ± 0.05 -0.026 0.065 -0.102 -0.015 -0.03 -0.024 -0.025 -0.021 0.045 -0.01 ± 0.05  D/V Direction -0.151 -0.193 -0.114 -0.127 -0.087 -0.035 -0.006 -0.319 0.213 -0.09 ± 0.15 -0.095 -0.194 -0.19 -0.048 -0.068 -0.035 -0.105 -0.305 0.186 -0.09 ± 0.14  3mm Cr/Cd Direction -0.003 0.026 -0.002 -0.005 0.014 -0.028 -0.062 0.042 0.072 -0.01 ± 0.04 0.003 -0.035 0.006 0.02 0.029 0.028 0.074 -0.012 -0.072 0.00 ± 0.04  D/V Direction -0.072 -0.079 -0.024 -0.036 -0.086 -0.131 0.151 0.086 -0.016 -0.02 ± 0.09 -0.061 -0.067 -0.029 -0.017 -0.075 -0.13 -0.049 0.103 0.011 -0.03 ± 0.07  115  Appendix B: Quasi-Static Bead Displacements  Table B-2 continued: Post-Compression Displacement (mm) Compression Depth Bead  Internal Dorsal  Caudal Ventral  Cranial Ventral  Internal Ventral  0.5mm Cr/Cd Direction -0.050 -0.006 0.055 0.032 0.015 0.018 0.042 0.062 0.081 0.03 ± 0.04 -0.016 -0.022 -0.023 -0.095 -0.149 0.022 -0.002 -0.013 0.034 -0.03 ± 0.06 -0.033 -0.026 -0.055 -0.082 -0.146 0.034 No Bead 0.004 -0.002 -0.02 ± 0.08 -0.033 -0.014 0.049 -0.01 0.003 0.005 0.013 0.047 0.029 0.01 ± 0.03  D/V Direction -0.060 -0.101 -0.010 -0.134 -0.756 0.006 -0.062 -0.086 -0.117 -0.15 ± 0.23 -0.015 -0.027 -0.024 -0.113 -0.719 -0.04 -0.032 -0.004 -0.068 -0.12 ± 0.23 -0.052 -0.009 0.017 -0.089 -0.735 -0.033 No Bead -0.017 -0.031 -0.12 ± 0.25 -0.05 -0.04 -0.023 -0.035 -0.843 -0.024 -0.038 0.000 0.024 -0.11 ± 0.27  2mm Cr/Cd Direction 0.021 0.043 -0.123 0.047 0.045 0.006 -0.085 0.097 0.056 0.01 ± 0.07 -0.004 0.016 -0.067 -0.038 0.016 -0.058 -0.035 -0.012 0.044 -0.02 ± 0.04 0.040 0.034 -0.036 -0.048 -0.022 -0.045 No Bead -0.026 0.02 0.04 ± 0.16 0.032 0.097 -0.098 0.031 0.043 -0.02 -0.071 0.068 0.07 0.02 ± 0.07  D/V Direction -0.072 -0.005 -0.068 -0.079 -0.086 -0.054 0.038 -0.38 0.184 -0.06 ± 0.15 -0.030 -0.087 -0.033 -0.063 -0.038 -0.005 -0.017 -0.24 0.035 -0.05 ± 0.08 -0.062 -0.042 -0.087 -0.032 -0.023 -0.034 No Bead -0.214 0.032 -0.06 ± 0.07 -0.008 -0.02 -0.078 -0.021 0.011 -0.007 -0.027 -0.334 0.072 -0.05 ± 0.12  3mm Cr/Cd Direction 0.048 0.038 -0.086 0.134 0.138 0.155 0.003 -0.044 0.007 0.04 ± 0.08 0.003 0.03 -0.047 0.046 0.002 0.095 0.052 0.021 -0.052 0.02 ± 0.05 0.010 0.002 -0.039 0.042 0.021 0.048 No Bead -0.012 -0.015 0.01 ± 0.03 0.044 0.038 -0.071 0.077 0.061 0.058 0.043 -0.164 -0.033 0.01 ± 0.08  D/V Direction -0.073 -0.079 -0.004 -0.051 -0.057 -0.048 0.033 0.154 0.032 -0.01 ± 0.07 -0.002 0.032 -0.026 -0.018 -0.077 -0.144 -0.013 -0.019 -0.017 -0.03 ± 0.05 -0.014 -0.003 0 -0.005 -0.073 -0.13 No Bead 0.022 0 -0.03 ± 0.05 -0.052 -0.035 0.015 0.03 -0.024 -0.098 0.009 0.091 0.061 0.00 ± 0.06  116  Appendix C: Dynamic Bead Displacements and Velocities  Appendix C: Dynamic Bead Displacements and Velocities Table C-1: Average spinal cord bead displacements at maximum impactor depth, maximum velocities during impact, and post-impact displacement. All values are in millimeters and values in bold are Avg±STD of each bead group. Negative values represent cranial or ventral displacement.  Bead  Caudal Dorsal  Cranial Dorsal  Internal Dorsal  Displacement at Max Impactor Depth (mm) Cr/Cd D/V Direction Direction -0.116 -0.572 -0.015 -0.674 -0.125 -0.714 0.104 -0.496 0.162 -0.641 0.075 -0.150 0.115 -0.765 0.023 -0.843 0.054 -0.980 0.149 -0.237 0.064 -0.725 0.189 -0.625 0.06 ± 0.10 -0.62 ± 0.24 -0.179 -0.812 0.069 -0.707 0.153 -0.794 0.230 -0.640 0.292 -0.776 0.065 -0.406 0.187 -0.737 0.100 -0.982 0.076 -0.773 0.157 -0.719 0.068 -0.823 0.195 -0.584 0.12 ± 0.12 -0.73 ± 0.14 -0.083 -0.632 0.130 -0.848 0.200 -0.997 0.293 -0.822 0.308 -0.985 0.054 -0.879 0.469 -0.869 0.186 -1.051 0.179 -0.913 0.138 -0.761 0.194 -0.844 0.242 -0.609 0.19 ± 0.14 -0.85 ± 0.13  Maximum Velocity During Impact (mm/s) Cr/Cd D/V Direction Direction -26.442 101.287 8.633 118.316 -14.686 112.128 19.983 52.966 23.331 91.363 12.662 21.455 12.450 90.082 22.630 104.835 14.696 104.674 19.400 40.288 17.397 88.253 13.853 87.766 10.33 ± 15.28 84.45 ± 30.21 -40.591 134.982 32.664 91.590 37.900 100.421 27.692 76.602 30.073 105.224 12.691 38.710 25.136 92.863 14.186 109.974 15.683 89.883 15.480 97.427 17.717 97.488 15.427 78.335 17.00 ± 19.96 92.79 ± 22.84 -10.356 82.349 22.825 107.336 35.599 99.504 40.556 95.331 28.865 127.607 16.310 107.388 40.940 108.208 12.542 98.476 25.703 102.474 11.414 87.519 22.536 99.184 26.602 83.351 22.79 ± 14.28 99.89 ± 12.48  Post-Impact Displacement (mm) Cr/Cd D/V Direction Direction -0.052 -0.195 -0.134 -0.074 -0.022 -0.151 0.023 -0.013 0.100 -0.051 -0.008 0.102 -0.059 -0.180 -0.024 -0.087 0.043 -0.299 0.005 -0.054 0.055 -0.078 -0.009 -0.126 -0.01 ± 0.06 -0.10 ± 0.10 -0.007 -0.219 -0.167 -0.107 0.069 -0.152 0.065 -0.076 0.143 -0.061 -0.013 0.055 -0.023 -0.061 -0.027 -0.169 0.036 -0.305 -0.036 -0.020 0.062 -0.121 0.000 -0.135 0.01 ± 0.08 -0.11 ± 0.09 0.020 -0.205 -0.094 -0.104 0.054 -0.095 0.078 -0.098 0.151 -0.022 -0.006 0.073 0.007 -0.023 -0.011 -0.097 0.075 -0.309 -0.024 0.002 0.135 -0.114 0.035 -0.137 0.04 ± 0.07 -0.09 ± 0.10  117  Appendix C: Dynamic Bead Displacements and Velocities  Table C-1 continued:  Bead  Caudal Ventral  Cranial Ventral  Internal Ventral  Displacement at Max Impactor Depth (mm) Cr/Cd D/V Direction Direction 0.042 0.310 0.063 0.013 0.035 0.039 0.114 0.032 0.157 0.034 0.116 0.127 0.100 -0.005 0.064 -0.046 0.112 -0.122 0.121 0.111 0.051 0.028 0.103 0.049 0.09 ± 0.04 0.05 ± 0.10 0.125 0.094 -0.049 0.232 0.018 -0.007 0.087 0.142 0.220 -0.001 0.061 0.300 0.195 0.104 0.010 0.034 0.149 -0.133 0.065 0.192 -0.024 -0.151 0.050 0.066 0.08 ± 0.08 0.08 ± 0.14 -0.129 -0.210 0.104 -0.164 0.268 -0.482 0.257 -0.346 0.510 -0.344 0.085 -0.302 0.455 -0.395 0.132 -0.178 0.256 -0.297 0.087 -0.339 0.262 -0.450 0.082 -0.148 0.20 ± 0.18 -0.30 ± 0.11  Maximum Velocity During Impact (mm/s) Cr/Cd D/V Direction Direction -10.784 -42.087 10.845 -11.657 -9.664 14.495 22.323 -5.669 28.373 -8.396 17.558 -18.605 17.654 -12.826 7.988 -18.581 19.578 22.715 14.480 -9.459 12.662 9.842 11.052 18.897 11.84 ± 11.71 -5.11 ± 18.64 22.405 -25.408 -33.883 -44.776 21.444 -14.572 18.192 -12.918 38.972 -8.007 14.869 -24.845 20.774 -13.663 12.142 -9.932 33.311 17.256 10.167 -13.766 -9.433 11.677 12.152 -10.570 13.43 ± 19.18 -12.46 ± 16.15 -21.316 17.463 11.734 23.036 34.227 53.796 24.667 41.234 49.727 42.041 17.854 36.991 42.706 60.991 14.567 47.995 23.807 24.614 -12.403 53.446 23.730 41.427 14.858 19.076 18.68 ± 20.26 38.51 ± 14.57  Post-Impact Displacement (mm) Cr/Cd D/V Direction Direction 0.007 -0.127 -0.125 0.041 0.034 -0.071 0.062 0.027 0.098 0.045 0.018 0.104 -0.019 -0.091 0.003 -0.062 0.029 -0.131 0.013 0.017 0.058 0.001 0.018 -0.023 0.02 ± 0.05 -0.02 ± 0.07 0.018 -0.146 -0.134 -0.074 0.022 -0.076 0.026 0.085 0.156 0.018 -0.016 0.116 -0.025 -0.043 0.009 -0.024 0.036 -0.178 -0.005 0.090 0.108 -0.055 -0.001 -0.031 0.02 ± 0.07 -0.03 ± 0.09 -0.003 -0.142 -0.082 0.016 0.094 -0.072 0.078 0.031 0.169 0.080 -0.002 0.096 0.043 0.021 0.006 0.033 0.159 -0.170 -0.004 0.022 0.119 -0.014 0.034 -0.041 0.05 ± 0.07 -0.01 ± 0.08  118  Appendix D: High-Speed X-ray Image Processing  Appendix D: High-Speed X-ray Image Processing D1 Image distortion correction A perforated metal sheet (1.6mm hole diameter, 3.18mm center-to-center spacing, 0.9mm thickness, PSC-11618036, Metal Supermarkets, Mississauga, Ontario) was used for correcting the distortion in the high-speed x-ray videos. The variability of the hole spacing on the sheet was measured by taking fifty caliper (±0.005mm) measurements in the vertical, horizontal, and two diagonal directions, resulting in two-hundred measurements. The average standard deviation of these measurements was 0.018mm. The grid was attached to the face of the image intensifier and an x-ray image was taken before the high-speed x-ray testing. The distortion in the grid image was corrected using the open source XROMM (X-ray Reconstruction of Moving Morphology) software developed at Brown University as mentioned in Chapter 3. Figure D-1 shows an image before and after distortion correction. This software is run using MATLAB and creates a distortion correction transformation matrix based on the distortion in the grid image. This is then applied to the x-ray videos. The general steps of the XROMM algorithms are: 1. The centroid is calculated for each circle in the distorted grid image after a user selected grey threshold has been applied. 2. The idealized hole positions in the grid image are found by averaging the distance from the center hole to its six equally spaced neighboring holes to determine the correct hole to hole spacing. The correct alignment of each row is found by averaging the angles between the six holes surrounding the center hole (with the center hole as the reference point). 3. A local weighted means transformation matrix is created using the cp2tform function from the Image Processing Toolbox in MATLAB. The XY coordinates of the measured centroids and the XY coordinates of the idealized locations are the inputs for this function. 4. The transformation matrix is then applied to the grid image and to each frame in the x-ray video for distortion correction.  119  Appendix D: High-Speed X-ray Image Processing  B  A  Figure D-1: Distortion grid before (A) and after distortion correction (B).  To quantify the amount of residual distortion, MATLAB was used to create lines between the centroids of two selected circles, and the distance between the center of the line and the centroid of the center circle was calculated (using an algorithm developed by R. Newell – see Preface). This was done five times in the vertical, horizontal, and diagonal directions and the result was averaged giving an average residual distortion of 0.166 pixels. This value is approximately 0.02mm, using the conversion factor 0.133mm/pixel which was calculated by dividing the actual hole diameter (1.6mm) by the measured hole diameter in pixels (12pixels). Since the precision of bead tracking was 0.02mm, this remaining distortion was considered negligible.  120  Appendix D: High-Speed X-ray Image Processing  Figure D-2: Distortion grid after distortion correction showing how remaining distortion was quantified. Vertical, horizontal, and diagonal lines were created through bead centroids the deviation of the center of the line from the center circle was measured.  D2  Denoising A custom denoising algorithm developed by C. Russell (see Preface) was used to  denoised the high-speed x-ray video. Figure D-3 shows a frame from high-speed x-ray video before and after denoising. This algorithm uses a three dimensional curvelet thresholding technique. In basic terms, three dimensional curvelets are shaped like discs of varying scales and orientations, and are fit to features in sequential frames of the x-ray video, with the time dimension of the video extending the two spatial dimensions to yield 3D ―feature trajectories‖. Discs that align well with a feature, such as that describing a circular crosssection of a bead, would have a good fit and would be given a high coefficient. Discs overlaid on noise would have a very poor fit and would be given a low coefficient. Discs on the edges of features would have a worse fit than those inside of a feature, but would still have a much higher coefficient than noise. The lower coefficients are then filtered away, based on a specified threshold, which reduces the noise while preserving features and edges. For a more detailed mathematical explanation of three dimensional curvelets, refer to Candes et al., 2006 as referenced in Chapter 3. 121  Appendix D: High-Speed X-ray Image Processing  A  B  Figure D-3: Images from high-speed x-ray before (A) and after (B) denoising.  The specific denoising algorithm was a 3D generalization of the 2D image denoising example included with the CurveLab toolbox (www.curvelet.org). Briefly, the noisy image data is first subjected to a forward curvelet transform, yielding a corresponding set of curvelet coefficients encompassing a range of orientations and scales. These coefficients are then subjected to a threshold that is proportional to the estimated variance of the image noise. The thresholding method can be varied, but it was found that hard thresholding (which replaces all coefficients below the given threshold with zero, and leaves the rest unchanged) yielded the best results for the high-speed x-ray bead tracking images. Finally, the inverse curvelet transform is applied to the thresholded coefficients, yielding a denoised image set. Additionally, because the curvelet transform process is memory intensive and a typical high-speed x-ray image set contained up to 1000 frames, the algorithm divided the total image file into smaller sets 128 frames long and processed each set individually. Due to the boundary artifacts introduced at by the curvelet denoising algorithm (manifested as lines at the outer edges of each frame, and as ghosting effects over the first and last several frames of each processed set), twenty frames of overlap was specified between each set and the full image file was thus reassembled with the intermediate ghosting artifacts omitted. Image sets were read and written as multipage TIFF files to avoid introducing compression artifacts.  122  Appendix E: Sensor and TEMA Data Processing  Appendix E: Sensor and TEMA Data Processing E1 Sensor data processing E1.1  Filtering The accelerometer, LVDT, and load cell data were filtered using a 4th order  Butterworth filter with a 200Hz cut-off frequency. The cut-off frequency was chosen by analyzing the FFT diagrams of the raw data and by analyzing the data after various cut-off frequencies to verify peaks in data were maintained. Figures E-1:3 show the FFT of raw sensor data from one test. Figures E-4:6 show the raw and filtered data. 0.015  0.01  Amplitude  Amplitude  0.015  0.005  0 -5000  -3000  -1000 1000 Freq (Hz)  3000  0.01  0.005  0 -500  5000  -300  -100 100 Freq (Hz)  300  -300  -100 100 Freq (Hz)  300  500  Figure E-1: FFT of raw acceleration data.  0.015  Amplitude  Amplitude  0.015  0.01  0.005  0 -5000  0.01  0.005  -3000  -1000 1000 Freq (Hz)  3000  5000  0 -500  500  Figure E-2: FFT of raw load cell data.  123  0.015  0.015  0.01  0.01  Amplitude  Amplitude  Appendix E: Sensor and TEMA Data Processing  0.005  0 -500  0.005  -300  -100 100 Freq (Hz)  300  500  0 -5000  -3000  -1000 1000 Freq (Hz)  3000  5000  Figure E-3: FFT of LVDT data.  2.5 Acceleration, Raw Acceleration, Filtered  2 1.5 Acceleration (G)  1 0.5 0 -0.5 -1 -1.5 -2 -2.5 400  500  600  700  800 900 Time (ms)  1000  1100  1200  Figure E-4: Filtered versus raw acceleration data using a 200Hz cut-off frequency.  124  Appendix E: Sensor and TEMA Data Processing  1.5 Force, Raw Force, Filtered  1 0.5  Force (N)  0 -0.5 -1 -1.5 -2 -2.5 400  500  600  700  800 900 Time (ms)  1000  1100  1200  Figure E-5: Filtered versus raw load cell data using a 200Hz cut-off frequency.  -0.5 Displacement, Raw Displacement, Filtered  Displacement (mm)  -0.6  -0.7  -0.8  -0.9  -1  -1.1 800  850  900 Time (ms)  950  1000  Figure E-6: Filtered versus raw LVDT data using a 200Hz cut-off frequency.  E1.2  Inertial compensation The inertial compensation factor was determined using force and acceleration data  from six hits without the animal present. The inertial compensation factor was chosen so that its product with the acceleration would remove the peaks in force when the actuator comes to a stop (force peaks being present because of inertia of the mass attached to the load cell). Figure E-7 shows the acceleration, force, and compensated force from one of these hits. The 125  Appendix E: Sensor and TEMA Data Processing  inertial compensation factors were averaged, and the average was used for force compensation of the experimental test data. Remaining offset was noticed which could be due to viscoelasticity in the load cell (Figure E-8). The offset between the start of the first acceleration peak and the end of the last acceleration peak was averaged for the six inertial compensation impacts (referred to here as the offset ―mask‖) and was subtracted from the experimental test data after performing inertial compensation. Figure E-9 shows sample test data with force compensation. Figure E-10 shows the compensated force with offset and the force after the offset ―mask‖ is applied.  2 Acceleration Force, Raw Force, Compensated  Force (N) and Acceleration (G)  1.5 1  Stroke stops 0.5  Actuator returns to start  0 -0.5 -1  Stroke starts -1.5 -2 500  750  1000 Time (ms)  1250  1500  Figure E-7: Inertially compensated force data of a blank impact.  126  Appendix E: Sensor and TEMA Data Processing  0.5  Force (N) and Acceleration (G)  0.4 0.3 0.2  Remaining Offset 0.1 0 -0.1 -0.2 -0.3  Acceleration Force, Raw Force, Compensated  -0.4 -0.5 650  700  750 800 Time (ms)  850  900  Figure E-8: Inertially compensated force data of a blank impact showing remaining offset. The force data was compensated to the force offset between the two large acceleration peaks. This resulted in a remaining offset in the inertially compensated force data.  2 Acceleration Force, Raw Force, Compensated  Force (N) and Acceleration (G)  1.5 1 0.5 0 -0.5 -1 -1.5 -2 500  750  1000 Time (ms)  1250  1500  Figure E-9: Inertially compensated force data from one experimental test.  127  Appendix E: Sensor and TEMA Data Processing  2 Force, Compensated Force, Masked  Force (N) and Acceleration (G)  1.5 1 0.5 0 -0.5 -1 -1.5 -2 500  750  1000 Time (ms)  1250  1500  Figure E-10: Inertially compensated force data with offset mask from one experimental test. The offset mask causes a slight reduction in the overall force.  E1.3  Time syncing There was a slight lag between the LVDT data and the force and acceleration data  since the LVDT data went through the Instron software before the DAQ. The LVDT data was differentiated twice to determine the acceleration, and the initial slope of the LVDT acceleration curve was matched to the accelerometer acceleration (Figure E-11).  128  Displacement (mm), Force (N), and Acceleration (G)  Appendix E: Sensor and TEMA Data Processing  Acceleration LVDT Acceleration Force LVDT Displacement  2.5 2 1.5 1 0.5 0 -0.5 -1 -1.5 -2 800  900  1000 Time (ms)  1100  1200  1300  Figure E-11: Force and acceleration data shifted to LVDT data. The acceleration peaks from the accelerometer and the LVDT were overlaid and the time was shifted so the LVDT data were synced with the accelerometer and load cell data.  E2. TEMA data processing E2.1  Filtering The bead displacement data from TEMA were filtered using a 4th order Butterworth  filter with a 100Hz cut-off frequency. The cut-off frequency was chosen by analyzing the FFT diagrams of the raw data (Figures E-12) and by analyzing true peaks in displacement in the x-ray video and tracking data. The cut-off frequency was chosen so that it did not eliminate any true data peaks. Raw and filtered data for the internal spinal cord beads of one test is shown in Figure E-13.  129  Appendix E: Sensor and TEMA Data Processing  0.05  Amplitude  0.04 0.03 0.02 0.01 0 -500  -300  -100 100 Freq (Hz)  300  500  Figure E-12: FFT of Internal dorsal bead from one experimental test. 1 0  Displacement (pixels)  -1 -2 -3 -4 -5 -6 -7 Raw data Filtered data  -8 -9 700  800  900  1000  1100 1200 Time (ms)  1300  1400  1500  Figure E-13: Filtered TEMA data. The ID bead is curve on the bottom and the IV bead is the curve on the top.  E2.2  Pixel to millimeter conversion factor The start of the LVDT data was clipped for seven of the twelve tests, due to an  incorrect setting in the DAQ LabView program. The LVDT calibration factor was accurate for all tests and only the first couple of millimeters in displacement were clipped, so the LVDT displacement during the impact was still valid for all tests. Since the overall LVDT displacement of the impactor was intended to be used for the calibration factor, it was necessary to find the average LVDT displacement for separate impacts and use this for the 130  Appendix E: Sensor and TEMA Data Processing  calibration factor for all of the tests. The average displacement for the six impacts that did not have clipped LVDT data was 18.0888mm ± 0.005mm. The displacement of five additional impacts was 18.08838mm ± 0.005mm. The average of all of these impacts was 18.0886mm ± 0.005mm. Given the small standard deviation, the average LVDT displacement of 18.0886mm was used for the conversion factor. The average displacement of the three beads on the impactor (tracked with TEMA) was divided the LVDT displacement for the pixel/mm conversion factor (as discussed in Chapter 3). E2.3  Magnification error Since the beads in the impactor (used for the conversion factor) were not directly in  line with the beads inside of the spinal cord, there will be error due to magnification. Magnification occurs because the x-ray source is projecting a cone of x-rays from a small point. The farther the object is from the image intensifier screen, the larger the magnification. This magnification can estimated using simple trigonometry, as illustrated in Figure E-14A (Boone, 2000 – see reference below). This approach was used for determining the magnification error resulting from using the beads on the impactor for the conversion factor (Figure E14B).  A  B  Figure E-14: Magnification error measurement. Standard trigonometry can be used for estimating magnification of an object in an x-ray beam (A). This approach was taken for estimating error in the mm/pixel conversion factor due to magnification of the impactor beads which were out of plane from the spinal cord beads.  The impactor was 2.5mm in wide, so it was assumed that the impactor beads were 1.25mm from the center of the spinal cord (Y). The distance between the x-ray source and impactor was approximately 453mm (X). This gave an estimated magnification of 1.0028, which means a displacement of 18mm would be magnified to 18.05mm at the center of the 131  Appendix E: Sensor and TEMA Data Processing  spinal cord. The average conversion factor for the high-speed x-ray tests was 0.125mm/pixel, and with an overall displacement of 18mm, this corresponds to about 144 pixels. With a 18.05mm displacement, the new conversion factor would be 0.1253mm/pixel. If a bead displacement of 10 pixels is being measured, the actual conversion factor would convert this to 1.25mm and the corrected conversion factor would convert this to 1.253mm. Since the precision of the bead tracking was approximately 0.02mm, this slight magnification error of 0.003mm was considered insignificant. II to tube = 59.3cm II to impactor = 14.0cm Tube to impactor = 45.3mm J.M. Boone, Handbook of Medical Imaging, Volume 1. Physics and Psychophysics, Chapter 1. X-ray Production, Interaction, and Detection in Diagnostic Imaging, SPIE Press, Bellingham, USA, 2000.  E2.4  TEMA data syncing with sensor data There was a slight offset between TEMA tracking data from the high-speed x-ray and  the sensor data. The offset was compensated for by matching the slope of the TEMA impactor displacement with the slope of the LVDT displacement.  132  Appendix F: Sensor Specifications and Accuracy  Appendix F: Sensor Specifications and Accuracy F1 Load cell calibration The load cell (5lb, model 31, Honeywell-Sensotec, Columbus, OH) was calibrated in tension by hanging small weights on the end and recording the output voltage. Compression calibration was performed by placing weights on top of the load cell and recording the output voltage. The equation for the best-fit line was found and used to find the Volt to Newton conversion (Figure F-1).  3  Force Readout (V)  2  1  0 -30  -20  -10  0  10  20  30  -1  -2  y = 0.0941x + 0.0762 R² = 0.9998 5V = 53.3252N -5V = -53.9447N  -3 Force Applied (N)  Figure F-1: Load cell calibration  F2 LVDT accuracy The LVDT was standard with the Instron Dynamight model 8841 and had a stroke distance of ±25.4mm and a manufacturer‘s reported error of 0.250mm (0.5% full scale). The error was found to be lower with the current test set-up. During ten impacts, the LVDT reading was compared to a caliper measurement (±0.005mm) of the actuator displacement 133  Appendix F: Sensor Specifications and Accuracy  which resulted in an overshoot in the LVDT reading of 0.10mm ± 0.03mm. This overshoot in the LVDT reading would cause error in the bead measurement since the conversion factor is based on the LVDT displacement. The average conversion factor for the high-speed x-ray tests was 0.125mm/pixel, and with an overall displacement of 18mm this corresponds to about 144 pixels. With 18.10mm of displacement, the new conversion factor would be 0.1257mm/pixel. If a bead displacement of 10 pixels is being measured, the actual conversion factor would convert this to 1.250mm and the corrected conversion factor would convert this to 1.257mm. Since the precision of the bead tracking is approximately 0.02mm, this slight error of 0.007mm is considered insignificant.  F3 Actuator repeatability and PID settings The Instron Dynamight (model 8841) actuator was used for these impacts. The average displacement of ten impacts was 18.0886mm ± 0.005mm. The standard deviation represents the high repeatability of the actuator. The PID settings for the actuator were chosen to minimize rise time and overshoot. The PID settings were varied (P: 9-20; I: 0-1; D: 0-0.1) and P=20 I=0 D=0 was chosen as it had the fastest rise time and minimal overshoot.  134  Appendix G: Ethics Board Certificate  Appendix G: Ethics Board Certificate  Figure G-1: Animal care certificate representing ethics approval for the studies presented in this thesis.  135  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0071208/manifest

Comment

Related Items