MAGNETIC RESONANCE IMAGING CHANGES FOLLOWING A NON-INVASIVE MODEL OF MILD TRAUMATIC BRAIN INJURY IN RATS

Transcription

1 MAGNETIC RESONANCE IMAGING CHANGES FOLLOWING A NON-INVASIVE MODEL OF MILD TRAUMATIC BRAIN INJURY IN RATS By FRITHA CLAIRE SAUNDERS A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2017

2 2017 Fritha Claire Saunders

3 To my fiancé, Christopher, my parents, Sue and Alan, and my sister, Leah. To my mentors in radiology who have inspired, enthused and guided me from the beginning. And to the victims of traumatic brain injury, for whom we strive to help. 3

4 ACKNOWLEDGMENTS I would like express my sincere appreciation to my committee; Drs. Davenport, Berry, Winter, Giglio and Porter for their hard work, advice, support and encouragement. I would also like to thank Drs. Spoldi, Abbott, and Dan Mungas, as well as Sally O Connell, Mary Wilson, Bobbie Davis, and Matt Bolin, without whom this project would not have been possible. Additionally, my thanks to Drs. Sherry Adams, Fernando Garcia-Pereira, Sheila Carrera-Justiz, Jill Condrey, Mrs Brandy Winter and the team at ACS and the Metabolic Building, for their help and advice throughout the course of the project. Last but never least, I would like to thank Christopher and my family for their love, support, encouragement and help throughout the data collection process, and the writing of this thesis. 4

5 TABLE OF CONTENTS ACKNOWLEDGMENTS... 4 LIST OF TABLES... 7 LIST OF FIGURES... 8 LIST OF SYMBOLS AND ABBREVIATIONS... 9 ABSTRACT CHAPTER 1 INTRODUCTION Incidence and Costs of Traumatic Brain Injury Incidence of Traumatic Brain Injury in the United States Traumatic Brain Injury in Sports Traumatic Brain Injury in the Military Economic Costs of Traumatic Brain Injury Social Costs of Traumatic Brain Injury Mild Traumatic Brain Injury and Concussion Patient Recognition of Concussion Physician Diagnosis and Management of Mild Traumatic Brain Injury Classification of Brain Injury Focal Brain Injury Diffuse Brain Injury Experimental Traumatic Brain Injury Models Direct Brain Deformation (Percussion Concussion) Models Inertial Acceleration Models Impact Acceleration Models Experimental Species Imaging Techniques in Human Clinical Traumatic Brain Injury Magnetic Resonance Imaging Sequences in Experimental Brain Injury Magnetic Resonance Imaging in the Acute Stage of Experimental Brain Injury Traumatic Brain Injury Model Characterization by Magnetic Resonance Imaging Timing of MRI Scans Quantitative and Qualitative Variables Conclusion and Identified Gap in the Literature MATERIALS AND METHODS Aims, Research Questions and Hypotheses Animals Time-Course Anesthesia

6 Magnetic Resonance Imaging Traumatic Brain Injury Euthanasia Intensity Measurements Statistical Analysis Normalized Data One-Way Analysis of Variance RESULTS Study Design Rats Weight-Drop and Mortality Anesthesia and Mortality Scan Protocol Brain Intensities Within-Group Variability J Injury J Injury J Injury Between-Group Variability Control group (0 J) Experimental Groups Day DISCUSSION Changes in Intensity J Injury J Injury J versus 1 J Injury Injury Severity Statistical Discussion Rat Mortality Weight-Drop Anesthesia Limitations Future Research Conclusion APPENDIX TABLES LIST OF REFERENCES BIOGRAPHICAL SKETCH

7 LIST OF TABLES Table page 1-1 Comparison of the impact energies used during weight-drop traumatic brain injury investigations in various closed skull rodent models, with either open or closed cranial Percentage of patients diagnosed with various lesions using CT and MRI, following TBI, and the level of significance for the difference in detection rates between modalities Magnetic Resonance Imaging parameters for the study protocol including the fast spin echo T2 weighted sequences used in data analysis Four sets of ANOVA tests were performed using standard and normalized data, the first and third comparing within The breakdown products of hemoglobin, with their associated phase of hemorrhage and their appearance on T1-weighted and T2-weighted MRI sequences A-1 Comparison of studies of experimental TBI that have assessed MRI findings in the acute stage following injury. Only conventional A-2 Summary of p values obtained from One Way ANOVA tests and Kruskal- Wallis tests A-3 P values from ANOVA and Paired t-tests comparing MRI intensities at 0, 3 and 7 days following a 0.75 J injury, for brain regions with significant, or nearsignificant results using

8 LIST OF FIGURES Figure page 1-1 Rendition of the murine impact acceleration experimental setup used in the publication Schematic of the time-course of the study methodology. A pre-injury MRI scan was performed, followed by the 0.75 J or 1 J weight-drop Dorsal (left) and transverse (right) MRI images showing slice location and ROI placement. Dorsal plane Average MRI intensities for the combined left and right piriform lobe data, at pre-injury, and days 3 and 7, with a 0.75 J injury Average MRI intensities for the combined left and right occipital lobe data, at pre-injury, and days 3 and 7, with a 0.75 J injury Average MRI intensities for the combined left and right temporal lobe data, at pre-injury, and days 3 and 7, with a 0.75 J injury Average MRI intensities for the combined left and right thalamic data, at preinjury, and days 3 and 7, with a 0.75 J injury Average MRI intensities for the cerebellar data, at pre-injury, and days 3 and 7, with a 1 J injury Average MRI intensities for the brainstem data, at pre-injury, and days 3 and 7, with a 1 J injury Average MRI intensities for the frontal lobe data, at day 7, with 0 J, 0.75 J and 1 J injuries Average MRI intensities for the cerebellar data, at day 7, with 0 J, 0.75 J and 1 J injuries Average MRI intensities for the piriform lobe data, at day 7, with 0 J, 0.75 J and 1 J injuries Average MRI intensities for the normalized piriform lobe data, at day 7, with 0 J, 0.75 J and 1 J injuries

9 Degrees % Percent LIST OF SYMBOLS AND ABBREVIATIONS ANOVA cm CT DICOM Analysis of variance Centimeters Computed tomography Digital imaging and communications in medicine F Degrees Fahrenheit g J KW m MRI PACS ROI T TBI Grams Joules Kruskal-Wallis Meters Magnetic resonance imaging Picture archive and communication system Region of interest Tesla Traumatic brain injury 9

10 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science MAGNETIC RESONANCE IMAGING CHANGES FOLLOWING A NON-INVASIVE MODEL OF MILD TRAUMATIC BRAIN INJURY IN RATS Chair: Clifford R. Berry Major: Veterinary Medical Sciences By Fritha Claire Saunders May 2017 Traumatic brain injury (TBI) is a common injury, which has high social and economic costs. Mild TBI is the most common form of TBI, and has been called a silent epidemic. There is an especially high incidence of TBI amongst athletes and military personnel. Traumatic brain injury can cause focal or diffuse brain lesions, and these can be modeled using many experimental methods. The experimental techniques used to model TBI each produce a homogenous injury and are used to model an aspect of human disease. Weight-drop models have been shown able to create diffuse brain injury in rodents and have been adapted to produce non-invasive models. The aims of this study were 1) to investigate the acute changes in MRI brain intensity in rats over the first 7 days following a non-invasive, experimental weight-drop method of diffuse, mild traumatic brain injury, and 2) to determine whether there were differences in brain intensity using various impact injuries. Three groups of Sprague Dawley rats were used, with 5 animals in each group. The 3 groups received either 0 J (sham injury; control group animals), 0.75 J or 1 J 10

11 weight-drop injuries. The weight-drop injury was performed by dropping a 276 g weight from a height of m (0.75 J injury) or m (1 J injury), onto the skin on the dorsal surface of the anesthetized rat s head. A closed-cell foam pad was adhered to the tip of the weight to disperse the impact energy over the surface of the head. Magnetic resonance imaging scans were performed before injury and at 3 and 7 days post-injury. T2 weighted fast spin echo sequences were obtained in transverse and dorsal planes. Average T2 intensity values were measured by placing regions of interest (ROIs) on the olfactory bulbs, frontal, temporal, piriform, and occipital lobes, thalami, brainstem and cerebellum, using the transverse T2 weighted images. Differences between injury level and between day were assessed using analysis of variance (ANOVA) tests, paired t-tests, and as needed, Kruskal-Wallis tests. No differences were found in intensity between injury level or day for the control group rats. When a 0.75 J injury was used, the intensities within the piriform and occipital lobes was significantly lower at day 7 compared to the pre-injury scans. When a 1 J injury was used, the intensity within the cerebellum was significantly higher at day 7 compared to the pre-injury scans. The T2 hypointensity found using the 0.75 J injury is most likely due to the presence of hemorrhage. The T2 hyperintensity found using the 1 J injury is most likely due to the presence of cytotoxic edema, possibly with concurrent hemorrhage. The changes identified in this project are important in their mild severity and their inability to be detected upon subjective examination; objective measurement of MRI scans of TBI patients may be needed to detect post-injury changes. 11

12 INTRODUCTION Traumatic brain injury (TBI) is a remarkably common injury, with an especially high incidence amongst athletes and military personnel, and which has high economic and social costs. The American Congress of Rehabilitation Medicine describes TBI as an alteration in brain function, or other evidence of brain injury, caused by an external force. 1 Mild TBI is the most common form of TBI, and has been called a silent epidemic due to under-reporting and under-diagnosis. Concussion is a common sequela of TBI. Concussion is defined as loss of consciousness followed by posttraumatic amnesia. Traumatic brain injury can be caused by sports injuries, combat injuries, assault, falls, vehicle accidents and shaken-baby syndrome in children. Traumatic brain injury can cause focal or diffuse brain lesions, and these can be modeled using many different experimental methodologies. Computed tomography (CT) can be used to identify surgical lesions following TBI in the emergency department setting, however magnetic resonance imaging (MRI) is the most common technique used to image the brain, both clinically and experimentally. Although advanced imaging techniques provide detailed anatomic and functional information on the brain, conventional MRI modalities remain the predominant clinical imaging technique in hospitals. Incidence and Costs of Traumatic Brain Injury Incidence of Traumatic Brain Injury in the United States The incidence of traumatic brain injuries in the United States is estimated at 1.7 million cases annually, with 275, 000 hospitalizations and 52, 000 deaths. 2 Around 75% of the TBIs sustained in the United States annually are classified as mild TBI. 3 It is likely 12

13 that the incidence of TBI (especially mild TBI) in the general population is higher than commonly reported, due to exclusion of patients who were evaluated in primary care settings, or who did not seek medical care. This has been suggested in a study that investigated incidence of TBI by examining records from hospitals, outpatient clinics, general practitioners and death certificates, and estimated an incidence of TBI that was higher than the previously reported values from high-income, developed countries. 4 Traumatic Brain Injury in Sports Athletes not only have a high incidence of concussion, but they have an increased risk of experiencing repeat concussions. It has been estimated that in the United States, between 2001 and 2009, there were 173,000 emergency department visits for sports and recreation-related TBI in people under 20 years of age. 5 Concussion has been reported to have an incidence of 6.3% in collegiate football players, with 6.5% of these concussed players undergoing a repeat concussion within the same season. 6 Additionally, it was found that football athletes with a history of 3 or more previous concussions were 3.0 times more likely to obtain a concussion, and had longer recovery times, than players with no concussion history. Concussion rates are also high in professional football players, with a reported average incidence of 131 concussions per National Football League season, or an average of 0.4 concussions per game. 7 Football is not the only contact sport facing concerns of TBI in its players; 9.8% of Australian non-professional rugby players, and 1% of high-school soccer players were found to have suffered a concussion over the previous year, and concussions have been found to be the most common injury in female ice hockey players. 8, 9, 10 One retrospective study investigating televised professional rugby league games identified a probable concussive event in 31% of games, with an average of

14 concussions per game; this is similar to that noted in National Football League games. 11 Additionally, concussion has been reported as a concern in non-contact sports including baseball and volleyball. 8 The high incidence of TBI in both amateur and professional athletes, especially in contact sports, is concerning, both on a social level, as well as on an economic level. Traumatic Brain Injury in the Military Traumatic brain injury is a major cause of morbidity and mortality in military personnel. According to a 2013 report by the Centers for Disease Control and Prevention, 4.2% of all military personnel serving between 2000 and 2011 were diagnosed with TBI, with 77% of these injuries classified as mild TBI. 12 It has been reported that between 12-23% of military personnel who served in Iraq and Afghanistan experienced a TBI while deployed. 13 Traumatic brain injury is so common in military personnel that it has been labelled the signature injury of the Afghanistan and Iraq wars. Economic Costs of Traumatic Brain Injury Traumatic brain injury is associated with high direct and indirect costs. Estimating the economic costs associated with TBI is difficult due to uncertainty associated with which outcomes can be attributed to TBI, and exclusion of patients who did not seek medical treatment or present to a hospital. 12 One study estimated the total economic costs of TBI in the United States in 2000 at $60 billion; including direct medical and rehabilitation costs, as well as societal costs. 14 Data provided by the United States Military Health Service in 1992, at the end of the Iraq war, produced an estimate of total associated costs of deployment-associated traumatic brain injury at approximately $42 million. 15 These estimates give an indication of the scale of the economic impact of TBI. 14

15 The economic impact of sports-related TBI is a concern for many contact and non-contact sports across the world. A review of claims made to the New Zealand Accident Compensation Corporation system found that of the total sports-related claims made, 6.4% of the claims were for moderate-to-severe sports-related concussions (concussions that required medical care, as well as rehabilitation and income replacement from time off-work related to the injury). However, these claims accounted for 79% of the sports-related payouts, costing a total of $16,500,000 over the 10 year period ( ). The majority of the sports-related claims were for rugby union players. 16 Although the total economic costs associated with TBI are hard to assess, the economic impact of these injuries is high. Social Costs of Traumatic Brain Injury It is not just the high economic impact of TBI that is a source of concern; the social costs associated with TBI are substantial. As well as mortality, TBI can cause significant morbidity. In one report, only 15% of the total costs associated with TBI were direct medical costs, with the remaining 85% of the costs incurred from productivity losses. 14 In the acute phase, TBI can cause loss of consciousness and amnesia. Following TBI, patients can experience decreased attention span, problem-solving abilities, language skills and ability to learn new information, as well as sleep disturbances, irritability, hyperactivity, mania, anxiety, personality changes and aggression. 17 These changes can affect the ability to integrate into society, form relationships, and maintain employment. Additionally, TBI has been linked to increased incidence of alcoholism, epilepsy, depression, suicide and Alzheimer s disease. 18, 19, 20, 21 It has been estimated that people in the United States suffer permanent disability related to TBI. 22 The societal costs and productivity losses 15

16 associated with TBI are high, and estimates suggest that many Americans are living with the after-effects of TBI. Mild Traumatic Brain Injury and Concussion Mild TBI is the most common form of TBI, however TBI-associated concussions may not be recognized, diagnosed or treated. As many as 75% of the TBIs sustained annually in the United States are classified as mild TBI, causing mild TBI to be called a silent epidemic. 3 While most people with mild TBI recover, mild TBI can lead to longlasting effects such as lack of concentration, tiredness, vertigo and personality changes. 23, 22 Individuals affected by mild TBI can have difficulty returning to normal activities and work. 3 Additionally, repeated mild TBIs can have a cumulative effect, with delayed recovery from subsequent brain injuries. 6 Concussion is a common sequela of TBI. Concussion is defined as loss of consciousness followed by post-traumatic amnesia, however in mild concussions patients can retain consciousness and memory, but be confused, disoriented, and nauseous, or have hearing, vision and equilibrium disturbances. 24, 23, 25 When undiagnosed, mild TBI-associated concussion leads to insufficient monitoring and medical care. The lack of diagnosis can stem from affected individuals failing to recognize that they have experienced a concussion, or to recognize the need for medical examination and treatment, or alternatively, from medical practitioners failing to diagnose concussion, or to implement appropriate monitoring or therapeutics. 25, 26, 27, 28 Patient Recognition of Concussion Recognition of the symptoms and understanding of the importance of concussion, especially mild concussion amongst members of the public is low. In a 16

17 survey of college football and soccer players, only 23.4% of concussed football players and 19.8% of concussed soccer players realized that they have suffered a concussion. 25 These players reported experiencing symptoms of concussion following a blow to the head (loss of consciousness, confusion, disorientation, headaches, dizziness, nausea, memory difficulties, blurred or impaired vision, hearing problems or light sensitivity), however these players did not recognize that these were symptoms of a concussion. In a study of Italian soccer players, 62% of concussed players did not report their injury because they did not think the injury was severe enough to warrant reporting. 26 Another reason for the under-reporting of concussive symptoms in sports athletes stems from the disinclination to stop playing following a concussive event. 29 This could be extrapolated to a lack of understanding of the gravity of concussive injuries and their potential long-term effects. In a study of televised professional rugby league games, 60% of players who were retrospectively identified as displaying concussive behaviors continued to play, or returned to play within the same game; 3 of these players had experienced loss of consciousness. 11 Additionally, the televised commentary provided for viewers, suggested that 8 out of the 12 concussed players who returned to play had experience concussive symptoms. This suggests that rugby league viewers are exposed to practices involving concussed players returning immediately to play. This may have an impact on the way amateur players react to their own concussive events, and may down-play the importance of seeking medical treatment and withdrawing from play. 11 Although mild TBI can have acute and longlasting neurological effects, the recognition of concussive symptoms, and the importance of concussion is underappreciated amongst members of the public. 17

18 Physician Diagnosis and Management of Mild Traumatic Brain Injury The recognition of concussion is not only difficult for members of the public; making a diagnosis of concussion is challenging for medical practitioners. Accurate diagnosis is vital for appropriate treatment, long-term monitoring, and the determination of return to play for athletes. 30 Concussion can be difficult to diagnose due to variable and subtle clinical signs, rapid resolution of clinical signs following rest, variability in the literature of the exact definition and clinical grading of concussions, and limitations of detecting acute, mild TBI using conventional imaging modalities. 27 Because the brain often appears normal on imaging studies of mild TBI patients, diagnoses are frequently made based on clinical symptoms; a process complicated by unreliability in the selfreporting of symptoms, the vague nature of the clinical signs, and the overlap of clinical signs with other diagnoses. 31, 32 In one study, 56% of patients identified by study personnel as having a mild TBI (using the Centers for Disease Control and Prevention criteria for diagnosing mild TBI) were not documented as having mild TBI by the attending emergency department physician. 28 The greatest discrepancy between physicians and the study personnel was for the finding of confusion, whereas the greatest agreement was for the finding of loss of consciousness. 28 However, the diagnosis of concussion is being made with increasing frequency in emergency department settings; a review of admissions at pediatric hospitals found that the number of patients presenting to emergency departments with concussion from 2001 to 2010 increased, while the overall number of admissions remained stable. 33 This statistic may represent increasing awareness of mild TBI, its symptoms and diagnostic indices, or may represent a true increase in the incidence of mild TBI. 18

19 Even following diagnosis of concussion, medical treatment and monitoring may be brief; many patients are discharged within hours of admission and receive no followup plan, or education of potential on-going symptoms or complications. 34 The majority of patients presenting to emergency departments with mild TBI are discharged within 24 hours. 34 A challenging aspect of the care of mild TBI patients is deciding which patients can safely be sent home, and at which stage discharge should be performed. 35 The current recommendations are that patients with mild TBI and negative CT scans are at low risk for developing an intracranial lesion, and thus may be safely discharged from the emergency department, with information on post-concussive symptoms. 35 However, follow up is vital for that minority of patients who will experience on-going, chronic symptoms. Diagnosis of mild TBI and associated concussion is important for monitoring and treatment, however concussion may be missed due to lack of recognition of the symptoms and importance by the public, and difficulties associated with reaching a diagnosis by medical practitioners. Classification of Brain Injury There are several systems used to characterize brain injury, including classification based on spatial factors (focal or diffuse injury) or temporal factors (primary or secondary injury). 36 Using the spatial classification, focal injuries are injuries that cause tissue damage localized at the site of injury, while diffuse brain injury includes injury occurring around the periphery of impact, as well as in remote brain locations. Focal injuries are comprised of contusions, lacerations and hematomas, while diffuse injuries involve brain swelling, ischemic injury and diffuse axonal injury. 36 Temporal classification of brain injury divides brain injury into primary injuries that occur at the moment of injury, and secondary (delayed) injuries that produce effects 19

20 hours to days after impact, and that progress over time. 37, 36 Primary injuries result from mechanical damage to the brain during the initial insult, and include contusions, lacerations, intracranial hemorrhage, and diffuse axonal injury. Because primary injuries are usually irreversible, they are often untreatable. 37 As an example, severe, diffuse, primary mechanical axonal injury can produce immediate and prolonged coma. 36 Secondary injuries occur as a result of a cascade of cellular and molecular events initiated by the primary insult, leading to ischemia, breakdown of the blood-brain barrier, edema, altered cerebral perfusion or intracranial pressure, delayed diffuse axonal injury, and impaired cellular metabolism. 37 These changes can result in delayed neuronal death or dysfunction, which can cause systemic effects such as delayed loss of consciousness, apnea and hypoxia. 36 The nature of these secondary changes, and the delay in their occurrence, means that there is the potential for the prevention, or treatment of secondary injuries. For example, hemorrhage and hematomas associated with focal brain injury can induce secondary brain herniation and brainstem compression, leading to delayed-onset coma. It is recognized however that patients often present with a combination of focal and diffuse, and primary and secondary brain injury. 38 Focal Brain Injury Focal brain injury involves localized tissue damage, and is comprised of contusions (bruising occurring with intact pia-arachnoid membranes), lacerations (bruising occurring with torn pia-arachnoid membranes), and hematomas. Contusional hemorrhages can be characterized as cortical (within the cerebral cortex) or subcortical. Additionally, contusions are described as coup ; contusions below the site of impact, or contrecoup ; contusions on the contralateral side of the brain to the impact 20

21 caused by oscillation of the brain within the cranial vault. 36 Depending on the severity of injury, contusions manifest as microhemorrhages, or as larger hemorrhages affecting gyri and causing a wedge-shaped area of tissue necrosis. 23 Hematomas are caused by the tearing of blood vessels, leading to hemorrhage. The vascular damage occurs at the moment of injury, however clinical signs may be delayed due to increased intracranial pressure and cerebellar herniation as the hemorrhage expands into a space-occupying lesion. 23 Hematomas can be caused by a penetrating wound that slices vessels, or by rapid acceleration or deceleration of the head (i.e. whiplash injury) or brain (following impact) leading to the tearing of vessels. 36 Intracranial hematomas result from hemorrhage within the brain and meninges, and can be intracerebral hematomas, subarachnoid hematomas, and subdural hematomas. Subarachnoid hemorrhage can not only occur as a focal injury, but also as a result of widespread vascular damage. 39 Epidural hematomas result from hemorrhage between the dura mater and the skull, leading to separation of the dura mater from the skull. 23, 36 Diffuse Brain Injury Diffuse brain injury is caused by ischemic-hypoxic damage from loss of brain perfusion, brain swelling secondary to edema and/or congestion, diffuse vascular injury and diffuse axonal injury. 23, 39 The most common form of diffuse brain injury is diffuse axonal injury. Diffuse axonal injury is characterized by multifocal axonal damage with axonal swelling within the deep and subcortical white matter. 40, 23 The predominant causes of diffuse axonal injury are rotational accelerationdeceleration forces, such as those occurring during automobile accidents. 41, 36 Acceleration and deceleration produce strain and shear forces. Brain is an incompressible tissue giving it low susceptibility to compression injuries, however 21

22 because brain has low innate rigidity, it is very sensitive to strain (tissue deformation as a result of mechanical force). 42 Shearing forces have the highest magnitude in areas of differential tissue rigidity, such as at the interfaces between gray and white matter, brain and cerebrospinal fluid, and brain and dura mater. 43 Rapid brain deformation (dynamic loading) is more likely to produce brain injury than slow deformation (static loading). 23 Axons are a viscoelastic tissue, whereby when a slow stretching force is applied, the axon will elongate. However, if a stretching force is rapidly applied, the axon acts in a brittle manner and can break (axotomy). Axotomy may occur immediately at the time of injury due to severe stretching forces or transection (primary axotomy), or can occur several hours later (secondary axotomy). Secondary axotomy is a result of cytoskeleton disruption. This causes impairment of axoplasmic transport, with resultant swelling of the axon. Axonal swelling occurs over 3-6 hours post-injury in people. Over the next hours, the proximal segment of the axon detaches from the distal segment, and the distal segment undergoes Wallerian degeneration. 23, 36 Thus, diffuse axonal injury can manifest either immediately at the time of injury, or can produce delayed-onset clinical signs. 23, 36 Brain injury can be classified in several ways, including by spatial factors (focal or diffuse injury) or by temporal factors (primary or secondary injury). Brain injury in people is a heterogenous disease, and focal and diffuse, and primary and secondary injuries may all occur in one patient. 38 Experimental Traumatic Brain Injury Models Human traumatic brain injury is a complex, heterogenous disease comprised of a variety of lesions with varying pathogeneses. Experimental animal models of TBI produce homogenous, repeatable injury with control of variables (such as gender and 22

23 age), and as such, are individually incapable of modelling the entire spectrum of human lesions. 44, 36, 45 Each model allows researchers to investigate an aspect of the disease with the assumption that it can be extrapolated to human TBI. 41 Experimental brain injury can be divided into two injury mechanisms; percussion concussion (direct brain deformation) and acceleration concussion injuries. 46 Percussion concussion models involve direct impact injury onto the dura, from an object, gas or fluid striking the dura mater via a craniotomy. 41, 36 This creates direct brain deformation (see below). Acceleration concussion models produce brain injury due to acceleration or deceleration motion causing deformation of the brain within the cranial vault. 36 They can be classified as inertial acceleration models and impact acceleration models. Inertial acceleration models are where the head is accelerated without providing any impact, or by providing an indirect impact to a part of the body other than the head (see below). Impact acceleration models are where the brain is accelerated by impacting the exterior of the head or the skull. 41, 36 Percussion concussion models most commonly produce focal brain injury, while acceleration concussion models are more likely to produce diffuse brain injury, although there is likely to be a degree of overlap in many of the models. 36 Direct Brain Deformation (Percussion Concussion) Models Direct brain deformation injury can be induced via fluid percussion, controlled cortical impact, and weight-drop methods. Fluid percussion injury models are the most widely characterized and utilized models of experimental TBI. 47, 36 Fluid percussion injury is induced following trephination of the skull, by rapidly impacting a fluid bolus against the intact dura mater. 47 The bolus expands concentrically outwards from the 23

24 trephination site through the epidural space, leading to diffuse loading of the brain. 36 Injury severity can be adjusted by altering the force of the fluid pressure pulse. The pulse is produced by swinging a pendulum from a known height, to contact the piston of a saline fluid reservoir. The impact produces a pressure pulse in the saline that is forced into the cranial vault. Alterations in the force of the pressure pulse are achieved by changing the height of the pendulum. 47, 36 Fluid percussion models have been utilized in many species including rabbits, cats, rats, dogs and pigs. 36 The trademark injury of the fluid percussion injury model is the initial production of subarachnoid hematomas and intraparenchymal contusions, later leading to cortical necrosis and cavitation. 48 With increasing injury severity, fluid percussion models produce diffuse axonal injury in the brainstem. 48, 38 An advantage of these models is the ability to produce varying and controlled severities of injury. A disadvantage is that biomechanical analysis of the injuries is difficult because fluid flow characteristics (and therefore characteristics of the mechanical loading from the fluid pulse on the brain) are dependent on brain geometry and species. 44 Controlled cortical impact models (rigid percussion models) utilize a rigid impactor, or pressurized air to strike the intact dura mater through a craniotomy hole in the restrained head. This method was developed in ferrets in 1988, 49 and has since been used in other species including rats, mice and pigs. 50, 51, 52 Controlled cortical impact models produce focal injuries including contusions, intraparenchymal petechial hemorrhages, and epidural and subdural hematomas. 49, 50 With increasing severity of injury, axonal injury is produced within the brainstem. 50 An advantage of controlled 24

25 cortical impact models is the ability to control the velocity and depth of impaction of the brain, allowing the severity of injury to be controlled. 49, 50 Direct cortical compression can also be produced by dropping a weight onto the intact dura mater via a craniotomy hole. Weight-drop TBI methods have been designed to produce focal contusion injury. 53, 54 A disadvantage of the direct brain deformation models is their invasive nature. The need for surgery increases the complexity of the model, the time, equipment and skills required for preparation, the risk of intra-operative complications and post-surgical infection, and the addition of an extra variable into the model. Inertial Acceleration Models Inertial acceleration models have been designed to produce translational and rotational forces on the head, creating diffuse brain injury. 36 This technique was developed in non-human primates. 55, 56 The sedated animal s body is restrained in a mobile chair leaving the head unrestrained, and an impact is applied to the base of the chair inducing acceleration of the head through an arc, and inertial acceleration brain injury. A disadvantage of unrestrained head models is that variability in head motion makes biomechanical analyses difficult, and introduces a source of variability into the results. 44 In response to this limitation, restrained-head inertial acceleration models have been developed in the pig by securing the head to the apparatus, then rapidly rotating the head through a º arc from left to right. 57 This model produces diffuse axonal injury in both deep white matter and at the junction of white and gray matter. Because the acceleration and head motion are controlled, comparison between experiments is possible. These models avoid production of skull fractures and local skull deformation. However, biomechanical interpretation is complicated due to the 25

26 combination of both acceleration and deceleration phases, with potential for brain injury during either, or both, of these phases of motion. 44 Another limitation is that these models are not feasible in small animals (such as rats). 44, 57, 37 This is because as brain mass decreases, there is an exponential increase in the magnitude of the rotational acceleration forces needed to produce brain injury, leading to the need for extremely high acceleration forces. 58, 37 Impact Acceleration Models Impact acceleration models have developed from focal to diffuse injury models, and have been variably modified to produce non-invasive injury, to alter the surface that the rodent lies on, the site and force of injury, and the species/strain of animal used. Impact acceleration models were initially developed as an open-skull injury model, whereby a craniotomy was performed, and a weight was dropped vertically onto a stainless steel plate lying on top of the dura mater. 59 This model produced focal cortical contusions. Subsequently, a closed-skull weight-drop model was developed that involved dropping a silicon-coated metal weight onto the exposed and unprotected left hemisphere of the skull. 60 This model produced ipsilateral, focal, cortical contusions. Various similar closed-skull models have been used in different rat strains, and in mice, with the weight being dropped onto the exposed and unprotected skull. 61, 62, 63 In 1994, a closed-skull, impact acceleration model was developed in Sprague- Dawley rats that produced diffuse brain injury. 58, 64 This model involved incising the skin over the skull, reflecting the periosteum, and adhering a stainless steel disc to the skull. Brass weights ( g) were dropped from various heights (1 m and 2 m), to determine the force required to produce a severe injury (defined by a 50% mortality rate). In pilot studies, a 100 g weight was dropped from 50 cm onto heads without any 26

27 form of skull protection. In these studies, skull fractures were common. Thus, a steel sheet was glued onto the exposed skull in order to disperse the impact forces over a large area of the skull. This skull cap helped to prevent skull fractures, and allowed higher impact-acceleration levels, thus producing increased severity of brain injury. The authors determined that a severe head injury could be produced using a 450 g weight dropped from 2 m, producing an impact energy of 8.83 J. This technique produced a 44% mortality rate with 12.5% of animals suffering skull fractures. When the 450 g weight was dropped from a height of 1 m, a moderate injury severity was induced (4.41 J), with a mortality rate of 0% and no observed skull fractures. This model provides predominantly shear forces due to acceleration of the head, producing diffuse injuries, with decreased importance of focal impact forces. This is in contrast to the focal injuries produced by impact acceleration models that do not provide skull protection. 37 The main advantages of this method include the production of diffuse injury, and the lack of predominant brainstem lesions seen with dural impact models (such as high severity lateral fluid percussion models). 58 This method (hereby known as the Marmarou method ) has since been widely utilized. 65, 66, 67, 68, 69 Minor adaptations to the Marmarou method have been made by adjusting the size of the weight used and the height from which the weight is dropped, 70, 71, 72, 73, 74 or the animal used; including Wistar strain rats, 75 immature rats, 76 and mice. 77 Several authors have produced non-invasive forms of the weight-drop method. A modification of the Marmarou method to produce a non-invasive technique has been achieved by applying a deformable substance to the impact-surface of the weight. 78, 79 In these studies, no skin incision or other surgical intervention is made. The weight is 27

28 dropped onto the unprotected rat skull, and skull fractures are prevented by a deformable silicon disc or soft conical tip. The soft substance deforms according to the convex shape of the rat skull, allowing dispersion of the kinetic energy over a diffuse area of the skull. The advantages of this non-invasive method are that it does not involve a craniotomy or use a metal skull disc, and so does not require surgery, does not have post-operative complications (such as infection and pain management), and avoids the confounding factors of surgery and open-skull injury. The Marmarou method involves restraint of the head against a foam pad, producing dorsoventral linear acceleration of the brain, with no rotational acceleration component. 80 It has been stated that rotational acceleration is an important component of human traumatic brain injury. 81 This limitation was addressed in a non-invasive, modified model of mild repetitive brain injury in mice. 81 In this model, a perforated foil sheet replaces the deformable foam bed that was used in the Marmarou method. When the weight is dropped and impacts the mouse s head, the foil splits without applying any resistance to the head. The foil is positioned a distance above a foam pad, so that following impact, the mouse rotates 180º around a horizontal plane directed bilaterally through the mouse (Figure 1-1). This allows rapid and free rotational acceleration of the head and torso. This murine method has been adopted by several authors. 82, 83, 84 28

29 Figure 1-1. Rendition of the murine impact acceleration experimental setup used in the publication by Kane et al. (2012) showing rotational motion of the mouse following impact. 81 Diagram by Christopher Mungas, Another modification that has been made is the replacement of the Marmarou foam pad with a spring-loaded platform. 78, 79 The spring-loaded platform allows adjustments to be made to the relative contributions of impact and acceleration forces. 79 The spring-loaded platform alters the resistance to head movement, whereby the amount of movement of the table determines acceleration of the head. 78 The site of brain injury from weight-drop models has also varied. Weight-drop techniques have been used in mice, whereby a mild traumatic brain injury is induced by dropping a weight onto the right 85 or left 61 hemisphere of the skull. The Marmarou method was modified to produce an invasive (surgical) method of frontal traumatic brain injury in Long-Evans rats. 80. This method was designed to produce frontal injury and rostral-caudal (anterior-posterior) and sagittal rotational acceleration of the brain, 29

30 compared to the dorsoventral linear acceleration produced by the Marmarou method. 58, 80 The impact energies used in weight-drop methods to produce TBI have varied widely (see Table 1-1). Marmarou and colleagues 58 induced TBI by dropping a weight onto a steel plate adhered to the exposed skull, using an impact energy of 4.41 J to create a moderate injury, and 8.83 J to create a severe injury. Other investigators utilizing the Marmarou method have used impact energies ranging from 2.94 J to 9.98 J. 65, 73 In other exposed-skull studies, the authors replaced the steel skull plate with a silicon pad adhered to the end of the weight. In these studies, much lower impact energies were used ( J), and focal injuries were produced. 60, 61 Non-invasive, closed-head weight-drop models without any form of skull protection have utilized adult-rat equivalent 1 impact energies ranging from 0.15 J to 5.05 J. 77, 84 In one of these studies, a spherical weight was dropped onto the unprotected skull of Wistar rats. 62 The weight of the spheres was adjusted to produce different impact energies (from J to 1 J). In this study, animals undergoing impacts with energies of J and 0.5 J expressed no change in behavior or physiological parameter, however animals undergoing impacts of J and 1 J showed abnormal postures and convulsions. 62 Two investigative teams have used a non-invasive, closed-head weight-drop model, with protective padding attached to the end of the weight. In these studies, 1 Where impact energies in mice are expressed as the equivalent energy for an adult rat. The impact energies used in mouse studies were converted to the equivalent impact energy that would have been applied to an adult rat using a conversion factor determined by the ratio of the mass of a mouse brain to that of a rat brain (0.21); the mouse impact energy was calculated from the published weights and weightdrop heights, and this value was divided by

31 impact energies ranged from J in Wistar rats, 79 and J in Sprague-Dawley and Wistar rats. 78 These impact energies are lower than the energies used by Marmarou et al. 58 to produce moderate and severe injury, but higher than the impact energies used in the closed-head, exposed skull methods that produce focal injury. In Blaha et al. 79, a 400 g soft-tipped weight was dropped from various heights ( cm) onto the closed head of Wistar rats, to determine gradation of injury severity. Heights of cm ( J) produced 100% mortality rates. Using a dropheight of 40 cm (1.57 J) produced a severe head injury with 50% mortality. Heights of 35 cm and 30 cm produced moderate (1.37 J; 20 40% mortality) and mild (1.18 J; 0% mortality) injuries, respectively. 79 The lower mass-height combinations producing a severe (50% mortality) injury in this study (a 400 g weight dropped from a height of 0.4 m) as compared to those used by Marmarou et al. 58 (a 450 g weight dropped from a height of 2 m) may be related to the difference in weight-drop technique (lack of protective steel skull plate), or may be related to the use of Wistar rats versus Sprague- Dawley rats. It has been shown that under identical injury conditions, Wistar rats had significantly higher mortality rates (50% mortality) than Sprague-Dawley rats (0% mortality). 78 However, in that study, there were similar rates of skull fracture between the rat strains. 78 The impact energy used by various investigators with weight-drop injury techniques has been widely variable, and differs between models utilizing various skull-protection mechanisms. In summary, the techniques and equipment utilized in weight-drop models, as well as the forces used have widely varied, with models evolving from focal to diffuse, and invasive to non-invasive models. 31

32 An alternative approach to impact-acceleration injury is to use a pneumatic acceleration device. 86 Here, a pressure wave accelerates a rod at the rat s head, impacting either the unprotected skin, or a metal plate glued to the exposed skull. 86, 87 Table 1-1. Comparison of the impact energies used during weight-drop traumatic brain injury investigations in various closed skull rodent models, with either open or closed cranial skin, and with various techniques used to protect the skull from skull fractures Investigators Species/Strain Open/Closed Skin Kane et al., 2012, 81 Silva et al., 2011, 62 Mychasiuk et al., 2014, 83 Tang et al., 1997, 77 Zohar et al., 2003, 85 Yang et al., 2013, 88 Yoon et al., 2012, 63 Yu et al., 2014, 84 Blaha et al., 2010, 79 Englebourghs et al., 1998, 78 Skull Protection Lowest Impact Energy Level (J) 2 Mice Closed None 4.44 Highest Impact Energy (J) 2 Wistar rats Closed None Mice Closed None Mice Closed None Wistar rats Closed Padded tipped weight Juvenile Sprague- Dawley rats Closed None 0.75 Mice Closed None Sprague- Closed None Dawley rats Mice Closed None 5.05 Sprague- Dawley rats Closed Silicon tipped weight Where impact energies in mice are expressed as the equivalent energy for an adult rat. The impact energies used in mouse studies were converted to the equivalent impact energy that would have been applied to an adult rat using a conversion factor determined by the ratio of the mass of a mouse brain to that of a rat brain (0.21); the mouse impact energy was calculated from the published weights and weightdrop heights, and this value was divided by

33 Table 1-1. Continued Investigators Species/Strain Open/Closed Skin Marmarou et al., 1994, 58 Adelson et al., 1996, 76 Barzo et al., 1996, 89 Gabrielian et al., 2011, 67 Geeaerts et al., 2006, 72 Hallam et al., 2004, 90 Ito et al., 1996, 91 Kallakuri et al., 2003, 71 Lammie et al., 1999, 70 Li et al., 2013, 69 Maughan et al., 2000, 65 Lu et al., 2001, 75 Shafieian et al., 2009, 66 Turner et al., 2014, 92 Ucar et al., 2006, 73 Skull Protection Sprague- Dawley rats Open Skull plate Juvenile Open Skull Sprague- plate Dawley rats Sprague- Open Skull Dawley rats plate Sprague- Open Skull Dawley rats plate Sprague- Open Skull Dawley rats plate Sprague- Open Skull Dawley rats plate Sprague- Open Skull Dawley rats plate Sprague- Open Skull Dawley rats plate Sprague- Open Skull Dawley rats plate Sprague- Open Skull Dawley rats plate Sprague- Open Skull Dawley rats plate Wistar rats Open Skull plate Sprague- Open Skull Dawley rats plate Sprague- Open Skull Dawley rats plate Sprague- Open Skull Dawley rats plate Lowest Impact Energy Level (J) 3 Highest Impact Energy Level (J) Where impact energies in mice are expressed as the equivalent energy for an adult rat. The impact energies used in mouse studies were converted to the equivalent impact energy that would have been applied to an adult rat using a conversion factor determined by the ratio of the mass of a mouse brain to that of a rat brain (0.21); the mouse impact energy was calculated from the published weights and weightdrop heights, and this value was divided by

34 Table 1-1. Continued Investigators Species/Strain Open/Closed Skin Yan et al., 2013, 74 Zakira et al., 2012, 68 Ghadiri et al., 2014, 93 Kilbourne et al., 2009, 80 Chen et al., 1996, 61 Shapira et al., 1988, 60 Skull Protection Sprague- Dawley rats Open Skull plate Sprague- Open Skull Dawley rats plate Wistar rats Open None, with sharp, pointed rod Long-Evans Open Ball rats impacted rods attached to malar processes Mice Open Silicon tip Lowest Impact Energy Level (J) Highest Impact Energy Level (J) Sabra rats Open Silicon tip 1.10 An alternative approach to impact-acceleration injury is the use of a pneumatic acceleration device. 86 In these models, a pressure wave is used to accelerate a rod at the head of rats, impacting either the unprotected skin overlying the skull, or a metal plate glued to the exposed skull. 86, 87 4 Where impact energies in mice are expressed as the equivalent energy for an adult rat. The impact energies used in mouse studies were converted to the equivalent impact energy that would have been applied to an adult rat using a conversion factor determined by the ratio of the mass of a mouse brain to that of a rat brain (0.21); the mouse impact energy was calculated from the published weights and weightdrop heights, and this value was divided by

35 Experimental Species A wide variety of animal species have been used to model human TBI, including ferrets, rats, mice, pigs, and non-human primates. 49, 50, 51, 52, 88 The most frequently chosen animal model used in acute TBI studies is the Sprague-Dawley rat strain. 94, 95, 96, 86, 97, 98, 99, 100, 101, 102 The Wistar rat strain is less-commonly chosen, but still frequently appears in the literature, 103, 104, 105 as do various mouse strains 106 and rabbits. 107, 54 The ease of comparing study results is increased when variability due to the use of different species or strains is removed. However, it should be noted that there is a great deal of heterogeneity in clinical human TBI that is not represented in the highly controlled situation of experimental TBI. This has been suggested as a factor for the low success rates of clinical trials for drugs developed from experimental TBI models. 45 Imaging Techniques in Human Clinical Traumatic Brain Injury In human emergency settings, CT is the initial imaging modality used to assess patients with traumatic head injury. 108, 109, 110 Computed tomography is effective at detecting skull fractures, hemorrhage and mass-effects in the brain. The goal of CT in the emergency setting is to diagnose intracranial lesions that may require surgery. 35 It is a useful modality in the acute phase because of wide availability, and fast scan time. It is therefore a good imaging modality for use in critical patients, and provides improved patient compliance and decreased motion artifacts. 108 However, MRI is more sensitive than CT for the detection of parenchymal brain lesions; in a study of patients with mild TBI, who were imaged using MRI and CT, parenchymal lesions were detected in 50% of patients with CT, and 75% of patients with MRI. 108 The detection rates of various intraaxial and extra-axial lesions in patients during the first 5 days following TBI, were assessed using CT and MRI. 111 The lesions for which there was a difference in 35

36 detection rate (percentage of patients in which the lesion was detected) between CT and MRI are shown in Table 1-2. For these lesions, MRI was superior for the detection of all intra-axial and extra-axial lesions, except for skull fractures, in which CT had a higher detection rate. Table 1-2. Percentage of patients diagnosed with various lesions using CT and MRI, following TBI, and the level of significance for the difference in detection rates between modalities using a McNemar test 111 Lesion Diagnosis by CT (%) Diagnosis by MRI (%) p value Subdural hematoma Diffuse axonal injury < Cortical contusions < Subarachnoid hemorrhage Skull fractures Unfortunately, many patients with mild TBI do not have apparent changes on conventional MRI. The development of advanced MRI techniques (such as diffusion tensor imaging) is proving promising for the imaging of mild TBI. 35 However, advanced MRI techniques are largely limited to research and academic institutions. 35 MRI is currently not routinely used for TBI in the emergency setting due to cost-prohibitions and limited availability of MRI. 35, 34 Magnetic Resonance Imaging Sequences in Experimental Brain Injury There are many MRI sequences that can be used to characterize TBI. The most commonly used techniques include T1 weighted spin-echo, T2 weighted spin-echo, susceptibility weighted, diffusion weighted and diffusion tensor imaging. T1 and T2 weighted spin-echo sequences are useful for detecting fluid (such as edema, inflammation and hemorrhage). Susceptibility weighted images are used to detect early hemorrhages. Diffusion tensor imaging detects abnormalities in the diffusion of water 36

37 molecules in the white matter tracts of the brain, indicating axonal disruption. 110, 98 T1 and T2 weighted spin-echo sequences are commonly-available MRI sequences, and therefore the remainder of this thesis will focus on studies that have utilized T1 and T2 weighted spin-echo MRI sequences, with decreased emphasis on advanced MRI sequences, such as diffusion tensor imaging and susceptibility weighted imaging. Magnetic Resonance Imaging in the Acute Stage of Experimental Brain Injury Several studies have been performed that have assessed MRI changes within the brain during the acute stage following experimental TBI (Table A-1). The TBI model, scan timing, experimental species and the variable measured on the MRI scans has varied between the studies. Traumatic Brain Injury Model Characterization by Magnetic Resonance Imaging The degree of characterization of the acute changes seen on MRI following TBI has varied between the various models of experimental TBI (Table A-1). These changes have been well categorized for the lateral fluid percussion, 103, 94, 95, 96, 112, 105, 101 and controlled cortical impact 106, 97, 104, 99, 102 models of TBI, however they have not been well described for the Marmarou weight-drop model. 98 Additionally, to the author s knowledge, acute MRI changes have not been assessed in non-invasive weight-drop models. Timing of MRI Scans The timing of MRI scans to document changes in the acute stage of injury has been variable between studies. Various studies have investigated the peracute stage of injury (<24 hours), with additional scans implemented at 1, 2, 3, 4, 7, 8 and 9 days postinjury (Table A-1). The most common timing of scans to document chronology and progression of lesions over the acute period is 1, 3 and 7 days post injury. 54, 107, 101,

38 Quantitative and Qualitative Variables The most common way to analyze lesion severity using T2 weighted spin-echo images is by measuring lesion volume. This is typically done by outlining the hyperintense lesion and, either counting the number of slices the lesion appears on to determine the area the lesion spans, 94, 95, 96, 97, 104, 107, 105, 102 or calculating the lesion volume as a percentage of total brain volume. 99 An alternative analysis technique is to calculate and compare the T2 value or T2 relaxation time of the lesion, or of brain regions. 106, 94, 96, 112, 105, 100 Other authors have measured ventricular volume, 104 quantified midline-shift of the brain due to brain swelling, 101 or have qualified the presence of lesions (contusions, edema, hemorrhage). 103, 86 Although MRI findings in the acute stage of TBI have been well documented for some experimental models of TBI, they have not been well described for weight-drop models. Additionally, the scan timing and analysis methods that have been used are widely variable between studies. Conclusion and Identified Gap in the Literature Traumatic brain injury, especially mild TBI is a common injury with a high incidence amongst amateur and professional athletes, as well as military personnel. The economic and social costs associated with TBI can be substantial, and people affected by TBI may experience long-lasting effects, which can affect their ability to function effectively in society and in employment. The symptoms of TBI, especially mild TBI are not well recognized amongst members of the public, with under-presentation of affected individuals to medical centers. Additionally, reaching a diagnosis of TBI is difficult for medical practitioners, and decision-making regarding treatment and monitoring is complicated. 38

39 Traumatic brain injury can be classified by spatial or temporal factors, and patients are likely to present with a combination of focal and diffuse, and primary and secondary injuries. Although human TBI is a heterogenous disease, most experimental models of TBI produce a specific component of the disease. There are a wide variety of experimental TBI models in the literature, each with their own advantages and disadvantages. The lateral fluid percussion and controlled cortical impact models have been well-characterized, but both models primarily produce focal injury. A weight-drop model has been developed that produces diffuse brain injury. This technique has been adapted to create non-invasive models, and to vary the type and magnitude of the impact forces the head experiences. Although CT is the most commonly used imaging modality in the acute stage following human TBI, MRI is better able to characterize brain parenchymal lesions. Various studies have used MRI to characterize brain changes in the acute stage of injury following experimental TBI, however this has not been thoroughly performed in diffuse-injury weight-drop models. More investigations are needed to characterize the brain changes seen on MRI during the acute stage following experimental TBI using the weight-drop model as a model of diffuse brain injury, especially following mild TBI. 39

40 MATERIALS AND METHODS Aims, Research Questions and Hypotheses The aims of this study were 1) to investigate the acute changes in MRI brain intensity in rats over the first 7 days, following a non-invasive, experimental weight-drop method of diffuse, mild traumatic brain injury, and 2) to determine whether there were differences in brain intensity using various impact injuries. The following research questions were put forward. Is there any difference in MRI intensity for any brain region: 1. between 0 J, 0.75 J and 1 J injuries during the study time period? 2. before injury and at 3 and 7 days post-injury for each injury level? The null hypotheses are as follows: 1. That there are no significant differences between the MRI intensities for any brain region, between 0 J, 0.75 J and 1 J injuries. 2. That there are no significant differences between the MRI intensities for any brain region, before injury or at 3 and 7 days post-injury. The alternative hypotheses are two-tailed hypotheses i.e. that there is a nondirectional difference between the groups. Animals All experimental procedures were approved by the University of Florida Institutional Animal Care and Use Committee. Male, 250 g Sprague Dawley strain rats were ordered 1. Upon arrival, the rats were housed in groups of 2 under barrier isolation conditions, in a quiet, temperature and humidity-controlled (73-77º F, 32-64% humidity) 1 Harlan Laboratories, Inc., Indianapolis, IN 40

41 room, maintained on a 12-hour light-to-dark cycle. The rats had ad libitum access to food and water. The rats were randomly divided into 3 groups according to the maximum potential cranial impact energy applied. Experimental injury group animals were subjected to either a 0.75 J injury or a 1 J injury. Control group rats were subjected to the same anesthetic and MRI protocols as the experimental injury group rats, however underwent no cranial injury. Each group consisted of 5 rats. Time-Course A pre-injury MRI scan was performed under general anesthesia. This scan provided an internal (within-animal) control, in addition to the external control provided by the control group rats. The traumatic brain injury was induced at time point = day 0. The rats were subsequently re-scanned at days 3 and 7 following injury, and then euthanized at day 7. In order to reduce the number of rats used in this study, the MRI images from the control rats were obtained from another study. MRI TBI MRI MRI Pre-injury Day 0 Day 3 Day 7 Figure 2-1. Schematic of the time-course of the study methodology. A pre-injury MRI scan was performed, followed by the 0.75 J or 1 J weight-drop injury at day 0 time point. Repeat MRI scans were performed at 3 and 7 days post-injury 41

42 Anesthesia All MRI scans and weight-drop procedures were performed under general anesthesia. General anesthesia was induced by flowing isoflurane 2 vaporized in 100% oxygen through a rodent induction box. For the weight-drop procedure, the rats were induced to a depth of anesthesia where there was absence of the peripheral nociceptive response. For the MRI scans, following the loss of the righting reflex, the rats were maintained on vaporized isoflurane on a non-rebreathing anesthesia circuit to a depth where there was absence of the peripheral nociceptive response (toe pinch response). A peripheral pulse oximetry sensor was used to monitor oxygen saturation and heart rate during anesthesia. Subsequent to unstable anesthesia of the rats during initial MRI scans with multiple mortalities (see results section), warmed heat pads were placed under the rats during the MRI scans, resulting in more-stable planes of anesthesia. Following MRI, the rats were returned to the induction box and maintained on 100% oxygen until righting reflexes had returned. During recovery, the animals were monitored until normal awareness and movement had returned, before the rats were returned to their cage and had access to food and water. 2 Isoflurane, USP; Piramal Healthcare Ltd, Andhra Pradesh, India 42

43 Magnetic Resonance Imaging Scans were performed using a 1.5 T MRI scanner 3 and a knee coil 4. Lubricant 5 was applied to the rats eyes. Rats were positioned in dorsal recumbency (supine position) on warmed heat foam pads used to raise the rat to the level of the center of coil. The pads cushioned the bodies of the rat and prevented any compression that could restrict respiratory motion. The nose of the rat was inserted into the end of the anesthesia circuit system. A polystyrene cushion was placed under the head, and tape was placed around the maxilla and palate securing the head to the cushion, preventing head movement during respiration. A second foam pad was positioned between the ventral surface of the rat and the coil. The MRI protocol consisted of fast spin echo T2-weighted images in transverse and dorsal planes. Initially, gradient echo, T2* transverse images were also obtained, however, this sequence was removed from the protocol early in the study due to patient mortality and motion artifact, and was not used in data analysis (see results section). Magnetic resonance imaging parameters are outlined in Table 2-1. Table 2-1. Magnetic Resonance Imaging parameters for the study protocol including the fast spin echo T2 weighted sequences used in data analysis Sequence Transverse T2 Dorsal T2 Echo time (TE) (ms) Repetition time (TR) (ms) Slice thickness (mm) Interslice space (mm) Number of averages 7 7 Flip angle ( ) Sequence length (minutes) 21:32 21:32 3 Vantage Titan MRI System, Toshiba America Medical Systems, Inc, Duluth, GA. 4 QD Knee coil, Toshiba America Medical Systems, Inc, Duluth, GA. 5 LubriFresh P.M., Major Pharmaceuticals, Livonia, MI. 43

44 Images were stored in a Picture Archive and Communication System (PACS) in Digital Imaging and Communications in Medicine (i.e. DICOM) format. Traumatic Brain Injury Traumatic brain injury was induced using a closed head weight-drop injury technique, which was modified from previously described weight-drop techniques. 58, 78, 79 It was designed to induce a diffuse impact injury using non-invasive (i.e. non-surgical) techniques. A 276 g weight was constructed out of a copper pipe filled with lead. The non-magnetic weight was designed to be able to be used within the MRI suite. A piece of firm closed-cell foam was attached to the end of the weight to dampen resonance from the impact of the weight with the head, and to diffuse the impact over a wide cranial area. General anesthesia was induced in the induction box as described above, then the rats were rapidly transferred onto a firm closed-cell foam pad within a second induction box. The rats were positioned in ventral recumbency (prone position) with the head aligned to a predetermined mark. The thoracic limbs were flexed at the shoulder to position the thoracic limb and manus in a caudal location parallel to the thorax. Foam pads were placed on either side of the head to provide lateral stabilization. A wooden platform with a 1.5 hole drilled partially through the top surface was positioned over the head and aligned such that the hole overlay the central aspect of the dorsal surface of the head, at a location caudal to the eyes and rostral to the ears. The box lid was closed and a 1.5 PVC pipe was passed through a pre-drilled 1.5 hole in the lid to rest within the hole in the wooden platform. This arrangement stabilized the pipe and prevented movement of the pipe during weight-drop. The pipe had pre-drilled holes along its length, allowing the weight to be dropped from a variety of heights to produce the 44

45 desired impact energy. The weight was lowered to the pre-determined height within the PVC pipe, onto a metal rod placed through the appropriate hole. The metal rod was removed and the weight was dropped onto the anesthetized rat s head. The 0.75 J impact injury was produced by dropping the 276 g weight from a distance of m. The 1 J impact injury was produced by dropping the 276 g weight from a distance of m. The control rats were placed under anesthesia but did not receive a weight-drop injury. Euthanasia The rats were euthanized using inhalation isoflurane followed by surgical transection of the diaphragm and aorta. Intensity Measurements Brain intensity measurements were performed on DICOM images reviewed on a PACS workstation 6, with the rats presented in a randomized order, and the measurer blinded to rat identification and injury level. Measurements were performed in triplicate. The dorsal and transverse plane images were first subjectively reviewed for abnormalities (such as mass effect, subjective changes in intensity and the presence of lesions). All measurements were then obtained from the transverse (axial) plane images, while the dorsal (coronal) plane images were used to assist in region localization. Regions of interest (ROIs) were placed on the olfactory bulbs, frontal lobe, temporal lobe, piriform lobe, thalamus, occipital lobe, brainstem and cerebellum (see Figure 2-2). A separate ROI was used for the left versus right sides, except for the olfactory bulbs, brainstem and cerebellum, where the left and right sides were combined 6 MergePACS workstation, Merge Healthcare Inc, Chicago, IL 45

46 in a single ROI that spanned midline. The slice number that each ROI was placed on was recorded, and the same slice was used for subsequent repetitions. For each ROI, the average T2 intensity value of the included voxels was recorded. Each ROI was placed 3 times for each scan in a fashion such that the measurer was blinded to the intensity values obtained in the prior measurement(s). Figure 2-2. Dorsal (left) and transverse (right) MRI images showing slice location and ROI placement. Dorsal plane images were used to assist in region localization: (A) olfactory bulbs, (B) frontal lobes, (C) parietal, temporal and piriform lobes and thalami, (D) occipital lobes, (E) cerebellum and brainstem. Statistical Analysis The mean and standard deviation were calculated for the 3 repeated measurements of each variable. The mean values were used for all further analysis. For the variables measured on each the left and right side of the brain (frontal, parietal, 46

47 temporal, piriform and occipital lobes, and the thalamus), box and whisker plots were used to compare the distribution of the data between the left and right measurements. No differences were found in the distributions between sides, so the left and right measurements from each scan were averaged for each variable, and the mean value was used in further analysis. Statistical analysis was performed using Minitab 7 as described below. Normalized Data Upon subjective and objective analysis of the data, it became apparent that the intensity values for all assessed regions of the brain were higher in the control animals than in either of the experimental groups (0.75 J or 1 J). This difference prevented the experimental group data being directly compared to the control group data. In order to allow comparisons, the data values were normalized to a baseline value (pre-injury); the data for each rat at days 3 and 7, from all three injury levels (0 J, 0.75 J and 1 J) were divided by the corresponding pre-injury data value for that rat. This provided a normalized intensity ratio for the day 3 and 7 data, which allowed comparisons to be made between the control and experimental data. One-Way Analysis of Variance Four sets of one-way analysis of variance (ANOVA) tests were performed, with two sets being performed on the standard data, and another two sets of ANOVA tests performed using normalized data (see below and Table 2-2). Each set of the ANOVA tests were run individually for all the brain ROIs: 1. For the standard data, the first set of ANOVA tests were used to assess within-group variability over time i.e. to compare how the brain MRI intensity values within each 7 Minitab , Minitab Inc. State College, PA 47

48 injury level group (0 J, 0.75 J and 1 J) changed over time (between pre-injury, and days 3 and 7). E.g. Compared each injury level across the three days (i.e J at pre-injury, and days 3 and 7) to assess temporal changes at an injury level, or intraanimal variation. 2. For the standard data, the second set of ANOVA tests were used to assess between-group variability at any single time point i.e. to compare how the brain MRI intensity values varied between the injury level groups (0 J, 0.75 J and 1 J) at each time point (pre-injury, and days 3 and 7). E.g. Compared all the injury levels at each day (i.e. 0 J, 0.75 J and 1 J at day 3) to assess injury intensity differences at a time point. 3. For the normalized data, the first set of ANOVA tests, were used to assess withingroup variability over time relative to a baseline (pre-injury) value. E.g. Compared 0.75 J at day 3 to day For the normalized data, the second set of ANOVA tests, were used to assess between-group variability at a time point. E.g. Compared each of the 3 injury levels between days 3 & 7, using the data that was normalized to pre-injury. The Ryan-Joiner test was used to test normality. If the assumption of normality was violated, the Kruskal-Wallis non-parametric test was used. The assumption of equal variances was tested, and was found to have been upheld in all instances. Tukey s test was used to perform post-hoc analysis for differences between the groups. Variables from the sets of ANOVA tests that assessed within-group variability, and that were found to have significant differences, were retested using paired student t-tests to ensure that the differences obtained were due to within-animal variations. A p-value of <0.05 was considered significant. 48

49 Table 2-2. Four sets of ANOVA tests were performed using standard and normalized data, the first and third comparing within-group variability over time, and the second and fourth comparing between-group variability at any point in time First Set of ANOVA Tests: Measuring how MRI intensity changed within the group (injury level) over time (day), using standard data Injury Level (J) Days 0.75 J Pre-injury, 3, 7 1 J Pre-injury, 3, 7 0 J Pre-injury, 3, 7 Second Set of ANOVA Tests: Comparing MRI intensity between groups (injury level) at each time point (day), using standard data Day Injury Level (J) Pre-injury 0, 0.75, 1 3 0, 0.75, 1 7 0, 0.75, 1 Third Set of ANOVA Tests: Measuring how MRI intensity changed within the group (injury level) over time (day), using data normalized to the pre-injury time point Injury Level (J) Days 0.75 J 3, 7 1 J 3, 7 0 J 3, 7 Fourth Set of ANOVA Tests: Comparing MRI intensity between groups (injury level) at each time point (day), using data normalized to the pre-injury time point Day Injury Level (J) 3 0, 0.75, 1 7 0, 0.75, 1 49

50 RESULTS Study Design Rats A total of 19 rats were included in this study. One rat was euthanized immediately following a 1 J weight-drop injury due to thoracic limb paralysis and 3 three rats died under anesthesia during pre-injury MRI scans (see below). This resulted in a total of 15 rats in the study. Five rats were used in each of the injury level groups (control group [0 J], 0.75 J and 1 J). Weight-Drop and Mortality At the beginning of the study period, the first rat underwent the pre-injury MRI scan. This scan was prolonged, lasting for 4.5 hours due to repeated scan runs because of motion artifact. Following the MRI scan, a 1 J weight-drop injury was performed. Upon recovery from anesthesia, it was identified that this animal suffered from thoracic limb paralysis, with bilateral loss of withdrawal reflexes, and with retention of motor function within the pelvic limbs. This rat was subsequently euthanized and removed from the study and subsequent analysis. No other rats died following traumatic brain injury. Rats recovered consciousness rapidly following anesthesia and rapidly displayed normal grooming behaviors, feeding, and social interactions with their cage mate. There was no evidence of porphyrin staining, bleeding, or behaviors representative of pain or stress following injury. Anesthesia and Mortality An appropriate plane of anesthesia was difficult to maintain on the rats. The rats would alternate between being too light and too deep under anesthesia. When the plane 50

51 of anesthesia was too light, the result was motion artifact in the images. As a result of a plane of anesthesia that was inappropriately deep, 3 rats died during the course of the pre-injury MRI scan. Many of the rats displayed dorsoventral motion of the head similar to a nodding motion throughout their scans, producing motion artifact on the images. Additionally, despite the use of low vaporizer settings and high oxygen flow rates, the animals became unstable under anesthesia and 3 rats died. While under anesthesia, it was noted that the rats were using excessive abdominal motion during respiration, suggestive of respiratory disease. However, no other indicators of respiratory disease were identified upon physical examination; no abnormal respiratory sounds were identified on lung auscultation, discharge was not identified from the nose or eyes, abnormal respiratory rate, sounds or effort were not identified, and abdominal breathing was not identified in awake animals. It is thought that the combination of abdominal breathing, head bobbing, and unexplained anesthetic deaths could be due to occult respiratory disease that was exacerbated by anesthesia. Necropsies were not performed. The following attempts were made to reduce head motion and to reduce anesthetic mortalities: Vaporizer level was adjusted (increases and decreases were both attempted depending on the situation). Oxygen flow rate was increased. Anesthetic tubing length was increased to maximize the distance between the anesthetic machine and the magnet, to reduce any interference of the magnet on the vaporizer. 51

52 Replaced the patient mask surrounding the rat s head with the tip of the nose inserted into the end of the tube, to attempt to prevent any compression of the airways from the mask, or airway kinking from head positioning. Placed a piece of tape over maxillae, through the oral cavity to restrain the jaw, to reduce head mobility. Removed the T2* sequence from the scanning protocol; this sequence is the longest sequence, greatly increasing scan time, and the gradients required for this sequence produce a lot of noise. The noise during this sequence lightened the patient plane of anesthesia and may have contributed to them waking up. Used heat pads around the rats during the MRI scan to prevent hypothermia, and to assist with maintaining a stable plane of anesthesia. Delayed starting the MRI scan for 5 minutes following induction of anesthesia to stabilize plane before providing noise stimulation. Scan Protocol Due to high levels of rat mortality during anesthesia for the MRI scans, attempts were made to reduce scan duration. The longest sequence in the MRI protocol was the T2* sequence, and this sequence additionally provided high levels of noise stimulation due to the intermittent application of magnetic gradients. This was felt to contribute to the difficulty in regulating a light plane of anesthesia without anesthetic mortality. As a result, following the pre-injury scanning of the first 7 rats, of which 3 died under anesthesia, the T2* sequence was removed from the scanning protocol. For the rats on which a T2* sequence had already been obtained on the pre-injury scans, this sequence was excluded from analysis. Brain Intensities No brain lesions or changes in intensity were identified when the MRI studies were subjectively examined. A summary of the results from the ANOVA tests is presented in Table A-2, where the p value of the ANOVA test (or the Kruskal-Wallis test if the assumption of normal 52

53 distribution was violated) is presented for each situation at each location. The p values which reached or neared significance are presented in bold. Within-Group Variability 0 J Injury For the control group (0 J injury level) rats, there was no significant difference in the mean MRI intensity levels between pre-injury, day 3 and day 7, for any brain location (p > 0.05). This finding was true for both the standard and normalized data J Injury For the 0.75 J injury level group, ANOVA tests performed on the standard data showed significant differences in mean MRI intensities between days at the piriform lobes (p = 0.004) and occipital lobes (p = 0.030). The p values for the differences in mean intensity neared significance at the temporal lobes (p = 0.068) and the thalami (p = 0.059). A summary of the brain regions with significant or near significant results using ANOVA tests are presented in Table A-3, along with the results of the associated paired t-tests. Piriform lobes: The mean MRI intensities within the piriform lobes at each time point are as follows: pre-injury = 3931, day 3 = 3343, and day 7 = Using an ANOVA test with Tukey s post-hoc tests on the standard data for the piriform lobes, mean MRI intensity pre-injury was significantly higher than post-injury (days 3 and 7). Although there was a trend for mean intensity to decrease between day 3 and 7, there was no statistical difference in mean intensity between these days. Using paired t-tests, the only statistically significant difference was found between pre-injury and day 7 (p = 0.016), with no significance difference found between pre-injury and day 3 (p = 0.107), and days 3 and 7 (p = 0.237). This information is graphically displayed in Figure 3-1, 53

54 and when subjectively assessed, a progressive trend of decreasing mean intensity is identified from pre-injury to day 3 to day 7. Figure 3-1. Average MRI intensities for the combined left and right piriform lobe data, at pre-injury, and days 3 and 7, with a 0.75 J injury Occipital lobes: The mean MRI intensities within the occipital lobes at each time point are as follows: pre-injury = 2884, day 3 = 2576, and day 7 = Upon ANOVA analysis with Tukey s post-hoc tests using the standard data for the occipital lobes, mean MRI intensity pre-injury was significantly higher than at 7 days post-injury. There was no difference in mean intensity between pre-injury and day 3, or days 3 and 7. Paired t-tests showed no statistically significant differences between days (pre-injury to day 3, p = 0.169; pre-injury to day 7, p = 0.079; day 3 to 7, p = 0.365). However, the test for the difference between pre-injury and day 7 neared significance. This information is graphically displayed in Figure 3-2, and when subjectively assessed, a progressive trend of decreasing mean intensity is identified from pre-injury to day 3 to day 7. 54

55 Figure 3-2. Average MRI intensities for the combined left and right occipital lobe data, at pre-injury, and days 3 and 7, with a 0.75 J injury Temporal lobes: The mean MRI intensities within the temporal lobes at each time point are as follows: pre-injury = 2997, day 3 = 2727, and day 7 = Utilizing an ANOVA test on the standard data for the temporal lobes, the p value for the differences in mean intensity between days neared, but did not reach significance (p = 0.068). When paired t-tests were performed, the mean intensity at pre-injury was significantly higher than at day 7 (p = 0.046), while the differences in mean intensity between preinjury and day 3 (p = 0.204) and days 3 and 7 (p = 0.402) were not significant. This information is graphically displayed in Figure 3-3, and when subjectively assessed, a progressive trend of decreasing mean intensity is identified from pre-injury to day 7. Subjectively a decreasing trend is identified from day 3 to day 7, however this trend is weak with overlap of the distributions, and similar mean values. 55

56 Figure 3-3. Average MRI intensities for the combined left and right temporal lobe data, at pre-injury, and days 3 and 7, with a 0.75 J injury Thalami: The mean MRI intensities within the thalami at each time point are as follows: pre-injury = 2627, day 3 = 2334, and day 7 = The ANOVA test on the standard data for the difference in mean intensity within the thalami between days produced a p value that neared, but did not reach significance (p = 0.059). The differences between pre-injury and day 3 (p = 0.196), pre-injury and day 7 (p = 0.106), and days 3 and 7 (p = 0.566) were not significant using paired t-tests. This information is graphically displayed in Figure 3-4, and when subjectively assessed, a progressive trend of decreasing mean intensity is identified from pre-injury to day 7. Subjectively a mildly decreasing trend is identified from day 3 to day 7, however there is overlap of the distributions, with similar mean values. 56

57 Figure 3-4. Average MRI intensities for the combined left and right thalamic data, at pre-injury, and days 3 and 7, with a 0.75 J injury Regions without significant differences between days: Using the standard data, there were no significant differences between days for the olfactory lobes, frontal lobes, parietal lobes, cerebellum or brainstem when a 0.75 J injury was used (Table A- 2). Normalized data: When assessing the normalized data for the 0.75 J injury level, there were no significant differences between pre-injury, and days 3 or 7 for any brain location (Table A-2). 1 J Injury For the 1 J injury level group, ANOVA tests performed on the standard data produced a significant result for the differences in mean MRI intensities between days at 57

58 the cerebellum (p = 0.008). The p value for the differences in mean intensity between days trended towards significance at the brainstem (p = 0.082). Cerebellum: The mean MRI intensities within the cerebellum at each time point are as follows: pre-injury = 2809, day 3 = 2747, and day 7 = Using ANOVA and Tukey s post-hoc tests on the standard data for the cerebellum, mean MRI intensity at day 7 was significantly higher than at pre-injury and at day 3, with no difference between pre-injury and day 3. Interestingly, using paired t-tests, there was a significant difference only between days 3 and 7 (p = 0.046), with the p value for the difference between pre-injury and day 7 almost reaching significance (p = 0.059). There was no significant difference between pre-injury and day 3 (p = 0.533). This information is graphically displayed in Figure 3-5, with the day 7 data of much higher intensity than the remaining days. When subjectively assessed, the boxplot shows a mild trend towards decreasing intensity from pre-injury to day 3. 58

59 Figure 3-5. Average MRI intensities for the cerebellar data, at pre-injury, and days 3 and 7, with a 1 J injury When the 1 J data were normalized relative to pre-injury, and the ANOVA tests were repeated, the intensity within the cerebellum at day 7 remained significantly higher than at day 3 (p = 0.045). Paired t-tests confirmed a statistically significant difference in the data between these days (p = 0.044). Brainstem: The mean MRI intensities within the brainstem at each time point are as follows: pre-injury = 2532, day 3 = 2524, and day 7 = The p value for the ANOVA test on the standard brainstem data neared statistical significance (p = 0.082). However, when a Ryan-Joiner test for normality was performed, the data were not normally distributed (p < 0.010). Upon the non-parametric Kruskal-Wallis test for differences between the days, a statistically significant result was obtained (p = 0.046). Similar to the cerebellar findings, using paired t-tests, the p value for the difference between mean MRI intensity at day 7 compared to day 3 neared significance (p = 59

60 0.076), while the differences between pre-injury and day 7 (p = 0.176) and pre-injury and day 3 (p = 0.937) did not reach significance. When assessing the boxplots of the brainstem data in Figure 3-6, the day 7 data is of higher intensity than the remaining days. In contrast to the cerebellar data, there is not a subjective decrease in the brainstem data at day 3. Upon normalization of the data, there was no longer a significant difference between days using a 1 J injury at the brainstem using the Kruskal-Wallis nonparametric test (p = 0.251). Figure 3-6. Average MRI intensities for the brainstem data, at pre-injury, and days 3 and 7, with a 1 J injury Regions without significant differences between days: Using both the standard and normalized data, there were no significant differences found between 60

61 days for the olfactory lobes, frontal lobes, parietal lobes, temporal lobes, occipital lobes, piriform lobes and the thalami when a 1 J injury was performed (Table A-2). Between-Group Variability Control group (0 J) The mean MRI intensities of the control group were found to be statistically higher than the intensities for the 0.75 J and 1 J experimental groups across all regions of the brain, and at all time points (pre-injury, and days 3 and 7). This produced p values for the ANOVA tests that were close to 0 for all tests in the between-group variability sections (Table A-2). Upon inspection of the Tukey s post-hoc tests, for all the regions of the brain except the regions discussed in the preceding and following sections (cerebellum at day 7 using the standard data, and the piriform lobes at day 7 using the normalized data), the only differences in intensity between injury levels were between the control group (0 J) and the collective experimental groups (0.75 J and 1 J), with no differences between the 0.75 J and 1 J groups. A boxplot of the data from the frontal lobes at day 7, exemplifying the higher intensity of the control group data compared to the experimental group data is provided in Figure 3-7. In this figure, the control group is significantly different from both the 0.75 J and 1 J groups, with no difference between the 0.75 J and 1 J groups (p = 0.000). 61

62 Figure 3-7. Average MRI intensities for the frontal lobe data, at day 7, with 0 J, 0.75 J and 1 J injuries Experimental Groups When comparing the two experimental injury levels (0.75 J and 1 J) at each time point, using ANOVA tests on the standard data, the only brain location for which there was a statistically significant difference in intensity between the injury levels, was the cerebellum at day 7 (Table A-2). When the data were normalized, the only brain region for which there was a difference in intensity between injury levels was the piriform lobes at day 7. Day 7 Cerebellum: The mean MRI intensities within the cerebellum at day 7 for the three injury levels are as follows: 0 J = 4370, 0.75 J = 2761, 1 J = The ANOVA test produced a statistically significant difference between the three injury levels. Using Tukey s post-hoc tests, differences in intensities were found between all three injury 62

63 levels. As described above, the 0 J intensity was higher than that of the 0.75 J and 1 J injury levels. The 0.75 J injury level had the lowest intensity, while the 1 J injury had an intermediate intensity that was significantly different from the other injury levels (Figure 3-8). This finding is similar to the trends found in the 0.75 J and 1 J within-group variability sections for other regions of the brain, where intensities decreased over time using 0.75 J, and intensities increased over time using 1 J. When the data were normalized, significant differences were not identified between the three injury levels at day 7 in the cerebellum (p = 0.250). Figure 3-8. Average MRI intensities for the cerebellar data, at day 7, with 0 J, 0.75 J and 1 J injuries 63

64 Piriform lobes: The mean MRI intensities within the piriform lobes at day 7 for the three injury levels are as follows: 0 J = 5455, 0.75 J = 3087, 1 J = On the standard data, there was a significant difference between injury levels at the piriform lobes (p = 0.000) as described above. However, on Tukey s post-hoc tests, only the control group and experimental groups were different, with no differences between 0.75 J and 1 J injury levels (Figure 3-9). Figure 3-9. Average MRI intensities for the piriform lobe data, at day 7, with 0 J, 0.75 J and 1 J injuries When the data were normalized to baseline, and comparisons were able to be made not only between the 0.75 J and 1 J groups, but also to the control group, a significant difference was found between the groups (p = 0.023). The 0.75 J group had 64

65 significantly lower intensity at day 7 than both the 0 J and 1 J groups. There was no difference in intensity between the 0 J and 1 J groups (see Figure 3-10). Figure Average MRI intensities for the normalized piriform lobe data, at day 7, with 0 J, 0.75 J and 1 J injuries Regions without significant differences between days: No significant differences were found between injury levels at any day for the olfactory lobes, frontal lobes, parietal lobes, temporal lobes, occipital lobes, thalami or brainstem, using the standard or normalized data. 65