1 EVALUATION OF THE EFFECT OF NEEDLE INSERTION SPEED IN CONVECTION ENHANCED DELIVERY INTO THE RAT BRAIN By GONZALO FERNANDO CASANOVA GARCIA A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PAR TIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2013
2 2013 Gonzalo Fernando Casanova Garcia
3 To my daughter Alexandra
4 ACKNOWLEDGEMENTS I acknowledge to people who support ed this work and made it possible. First, I thank my advisor, Dr. Malisa Sarntinoranont, for her support, teaching and supervision throughout my entire Ph.D. study and also for her generous financial support. I also thank all other membe rs of my dissertation committee, Dr. Paul Richar d Carney, Dr. Thom as Angelini, and Dr. Ashok V. Kumar for their valuable suggestions and advice for my research. Especially, it is a pleasure to mention Dr. Paul Richar d Carney who provided the laboratory for the in vivo experiments and allowed me to joint his excellent research team. Also, it is a pleasure to mention Dr. T h om as Angelini who has provided gr e at comments and insightful thought. I express my gratitude for my lab members: Dr. Jung Hwan Kim, Dr. S ung Jin Lee, Tatiana Nobrega, Wei Dai, Magdoom Mohamed, Yu Hong, and Saranya Canchi who provided teaching, encouragement and lots of great ideas. I specially thank Dr. Svetlana Kantorovich and Francisco Delgado for their training and support during my enti re research. I would like to thank my family for love and support.
5 TABLE OF CONTENTS Page ACKNOWLEDGEMENTS ................................ ................................ ............................... 4 LIST OF FIGURES ................................ ................................ ................................ .......... 9 LIST OF ABBREVIATIONS ................................ ................................ ........................... 11 ABSTRACT ................................ ................................ ................................ ................... 13 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 15 1.1 Motivation ................................ ................................ ................................ ......... 15 1.2 Specific Aims ................................ ................................ ................................ .... 21 1.2.1 Specific Aim 1: Evaluate the Influence of Needle Insertion Speed on The Backflow Phenomenon During CED ................................ ............... 21 1.2.2 Specific Aim 2: Determine the Tissue Damage and Pre Stress Between the Needle and the Tissue and Its Influence on Backflow for Different Needle Insertion Speeds ................................ ......................... 22 1.2.3 Specific Aim 3: Evaluate the Effect of Insertion Speed on Force, Tissue Deformation and Damage During Insertion and Pre Stress During Retraction in Real Time in Live Brain Tissue .............................. 22 2 BACKGROUND AND PREVIOUS STUDIES ................................ .......................... 24 2.1 Brain Tissue ................................ ................................ ................................ ...... 24 2.2 Convection Enhanced Delivery ................................ ................................ ......... 25 2.3 Backflow ................................ ................................ ................................ ........... 25 2.4 CED Models ................................ ................................ ................................ ...... 28 2.5 Backflow Modeling ................................ ................................ ............................ 30 2.6 Needle Insertion Mechanics ................................ ................................ .............. 31 2.7 Brain Tissue Mechanical Properties ................................ ................................ .. 33 2.8 Possi ble Biological Processes Affecting Tissue Disruption ............................... 36 2.8.1 Residual Stress ................................ ................................ ...................... 36 2.8.2 Edema ................................ ................................ ................................ ... 36 2.8.3 Hemorrhage ................................ ................................ ........................... 37 2.8.4 Gliosis ................................ ................................ ................................ .... 37 2.8.5 Blood Brain Barrier Disruption ................................ ............................... 38 2.9 Figures ................................ ................................ ................................ .............. 38 3 INFLUENCE OF NEEDLE INSERTION SPEED ON BACKFLOW IN 0.6% AGAROSE HYDROGEL ................................ ................................ ......................... 40 3.1 Overview ................................ ................................ ................................ ........... 40
6 3.2 Infusion Experiments ................................ ................................ ........................ 40 3.3 Damage and Plastic Deformation ................................ ................................ ..... 41 3.4 Pre stress Calculation ................................ ................................ ....................... 42 3.5 Backflow Model ................................ ................................ ................................ 43 3.6 Results ................................ ................................ ................................ .............. 45 3.7 Discussion ................................ ................................ ................................ ........ 47 3.8 Summary ................................ ................................ ................................ .......... 51 3.9 Figures and Tables ................................ ................................ ........................... 52 4 EVALUATON OF THE INFLUENCE OF NEEDLE INSERTION SPEED ON BACKFLOW DURING CED INTO THE RAT BRAIN ................................ .............. 56 4.1 Overview ................................ ................................ ................................ ........... 56 4.2 Methodology ................................ ................................ ................................ ..... 56 4.2.1 Animal Preparation and Surgical Procedures ................................ ........ 56 4.2.2 Needle Insertion and CED System ................................ ........................ 57 4.2.3 Backflow Experiments ................................ ................................ ........... 58 4.2.4 Backflow as Percentage of Tracer Infusate Outside the Target ............. 59 4.2.5 Backflow Distances ................................ ................................ ................ 62 4.3 Results ................................ ................................ ................................ .............. 62 4.4 Discussion ................................ ................................ ................................ ........ 64 4.5 Summary ................................ ................................ ................................ .......... 66 4.6 Figures ................................ ................................ ................................ .............. 67 5 EVALUATON OF TISSUE DAMAGE AND PRE STRESS BETWEEN THE NEEDLE AND THE TISSUE AND ITS INFLUENCE ON BACKFLOW FOR DIFFERENT NEEDLE INSERTION SPEEDS ................................ ........................ 71 5.1 Overview ................................ ................................ ................................ ........... 71 5.2 Methodology ................................ ................................ ................................ ..... 72 5.2.1 Animal Preparation and Surgical Procedures ................................ ........ 72 5.2.2 Needle Insertion and Pressure Measurement System ........................... 72 5.2.3 Needle Insertion Experiments ................................ ................................ 73 5.2.4 Infusion Pressure Measurements ................................ .......................... 73 5.2.5 Hole Meas urement Along the Needle Track ................................ .......... 73 5.2.6 Histological Assessments ................................ ................................ ...... 74 5.2.7 Pre Stress Estimation ................................ ................................ ............ 75 5.2.8 Statistical Analysis ................................ ................................ ................. 76 5.3 Results ................................ ................................ ................................ .............. 77 5.3.1 Infusion Pressure ................................ ................................ ................... 77 5.3.2 Hole Measurements ................................ ................................ ............... 78 5.3.3 Histology ................................ ................................ ................................ 81 5.3.4 Pre Stress ................................ ................................ .............................. 82 5.3.5 Pre Stress Change and Error Estimation ................................ ............... 82 5.4 Discussion ................................ ................................ ................................ ........ 84 5.5 Summary ................................ ................................ ................................ .......... 90 5.6 Figures ................................ ................................ ................................ .............. 91
7 6 EVALUATON OF THE EFFECT OF INSERTION SPEED ON FORCE, DEFORMATION, TISSUE DAMAGE, AND PRE STRESS ................................ ..... 99 6.1 Overview ................................ ................................ ................................ ........... 99 6.2 Methodology ................................ ................................ ................................ ... 100 6.2.1 Animal Preparation and Surgical Proc edures ................................ ...... 100 6.2.2 Needle Insertion and Force Measurement System .............................. 100 6.2.3 Force Measurement on Brain Tissue ................................ ................... 101 6.2.4 Force Measurement on 0.6% Agarose Hydrogel ................................ 102 6.2.5 Pre Stress Model ................................ ................................ ................. 102 6 .2.6 Statistical Analysis ................................ ................................ ............... 104 6.2.7 Phenomenological Models ................................ ................................ ... 104 6.3 Results ................................ ................................ ................................ ............ 105 6.3.1 Needle Insertion ................................ ................................ ................... 105 6.3.2 Needle Retraction ................................ ................................ ................ 107 6.3.3 Pre Stress ................................ ................................ ............................ 109 6.3.4 Phenomenological Models ................................ ................................ ... 110 6.4 Discussion ................................ ................................ ................................ ...... 112 6.5 Summary ................................ ................................ ................................ ........ 119 6.6 Figures and Tables ................................ ................................ ......................... 120 7 CONCLUSIONS AND FUTURE WORK ................................ ............................... 126 7.1 Summary and Conclusions ................................ ................................ ............. 126 7.2 Future Work ................................ ................................ ................................ .... 129 7.3 Figures ................................ ................................ ................................ ............ 134 LIST OF REFERENCES ................................ ................................ ............................. 136 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 144
8 LIST OF TABLES Table Page 3 1 Hole diameter and radial stretch ratio s for varying gauge needles inserted at a fast insertion speed ................................ ................................ ......................... 52 6 1 Average penetration force, force at the end of insertion, and dimpling for brain tissue and hydrogel ................................ ................................ ................. 1 20 6 2 Average resting force, and dragging force for brain tissue and hydrogel ........ 120 6 3 Friction stress for varying needle insertion speed averaging a long the entire needle track, and for each of the three regions in the brain .............................. 120 6 4 Model constants for friction stress ................................ ................................ ... 121 6 5 Constants for phenomenological model of backflow as function of friction stress. ................................ ................................ ................................ ............... 121
9 LIST OF FIGURES Figure Page 2 1 Magnetic resonance image of dru g infused in dog brain. ................................ ... 38 2 2 Insertion force on agarose hydrogel and porcine brain. ................................ ...... 39 3 1 Material displaced due to cannula ret raction. ................................ ..................... 52 3 2 Backflow distances measured at varying flow rates, needle diameters, and insertion speeds. ................................ ................................ ................................ 53 3 3 Tracer distribut ions in hydrogel. ................................ ................................ .......... 53 3 4 Accumulation of hydrogel at the needle tip. ................................ ........................ 54 3 5 Ins ertion of the needle showing no accumulation of mat erial. ............................ 54 3 6 Needle tip after passing through a 20 mm thick slice of hydrogel. ...................... 54 3 7 Hole in the hydrogel left by the needle aft er retraction. ................................ ...... 55 3 8 Hydrogel pre stress for varying needle gauges. ................................ ................. 55 3 9 Predicted and experimental backflow distances. ................................ ................ 55 4 1 Coronal section of the rat brain showing a schematic of the needle track. ......... 67 4 2 Brain tissue slice with EBA distribution and corr esponding concentration map. ................................ ................................ ................................ ................... 68 4 3 EBA backflow distribution along the needle track. ................................ .............. 69 4 4 EBA distributions close to the needle track. ................................ ....................... 69 4 5 Evans blue concentration close to the needle track. ................................ ........... 70 4 6 Backflow as a function of insertion speed and flow rate. ................................ .... 70 5 1 Hole of expanded to the diameter of the needle by the action of compressive pre stress ................................ ................................ ................................ ........... 91 5 2 Infusion pressure results du ring CED ................................ ................................ 91 5 3 Different types of tissue damage left after needle retraction ............................... 92 5 4 Area of the hole after needle retraction ................................ .............................. 93
10 5 5 Variation of hole measurements ................................ ................................ ......... 94 5 6 Equivalent hole diameters as function of depth along the needle track .............. 95 5 7 Hole perimeters as function of depth along the needle track .............................. 96 5 8 Needle tissue damage observed without staining ................................ ............... 96 5 9 Needle tissue damage observed with H&E staining ................................ ........... 97 5 10 Pre stress along the needle track ................................ ................................ ....... 98 6 1 Schematic representation of forces acting on the needle during the retraction 121 6 2 Typical force during insertion as a function of time in a rat brain in vivo ........... 122 6 3 Average force as function of insertion depth ................................ .................... 122 6 4 Typical curve of retraction force as function of time for brain tissue ................. 123 6 5 Curves of average force during needle retraction ................................ ............. 123 6 6 Friction stress as a function of insertion depth ................................ .................. 124 6 7 Comparison of experimental average friction stress values with the phenomenological models ................................ ................................ ................ 125 6 8 Average backflow percentage for varying friction stress ................................ ... 125 7 1 EBA distribution in tissue slices. ................................ ................................ ....... 134 7 2 Backflow for varying insertion speed for rats infused with blebbistatin ............. 134 7 3 Tissue damage after infusions ................................ ................................ .......... 135 7 4 Areas of the hole left by the needle after retraction on normal rats and pre infused with bleb bistatin ................................ ................................ ................... 135
11 L IST O F ABBREVIATIONS T ERM : Definition Radius of the needle BF Backflow percentage BBB Blood brain barrier Concentration of EBA cc corpus callosum CED Convection enhanced delivery CPu Caudate putamen DTI D iffusion tensor imaging Dimpling EBA Evans blue albumin ec External capsule F Needle retraction f orce measured by the load cell FEM Finite element method FBM Finite volume method Dragging force Gradient deformation tensor Penetration force Resti ng force G Green index of a color RGB figure GM Gray matter h Thickness of the brain slice H&E Hematoxylin and eosin
12 Light i ntensity Incident light energy Path length of the light inside the specimen Lss Lateral stripe of striatum M Factor that acco unts the increased path length of detected photons Mass of Evans blue MRI Magnetic resonance imaging N Factor that accounts for scattering of light PEEK Polyether ether ketone Material constant Energy function WM White matter Light absorption coef ficient Kronecker delta tensor Stretch ratio Friction coefficient Shear modulus Diameter of the hole left by the needle Cauchy stress tensor Radial stress Shear stress on the needle surfa ce
13 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy EVALUATION OF THE EFFECT OF NEEDLE INSERTION SPEED IN CONVECTION EN HANCED DELIVERY INTO THE RAT BRAIN By Gonzalo Fernando Casanova Garcia August 2013 Chair: Malisa Sarnti noranont Major: Mechanical Engineering Flow back along the needle track (backflow) is an important problem during direct infusion of drugs into the brain tissue, e.g. convection enhanced delivery (CED). In this study, the effect of needle insertion speed on backflow and tissue injury was evaluated. Needle insertion and convection enhanced delivery experiments were performed in vivo in the rat brain. Needle s were introduced at three insertion speeds (0.2, 2, and 10 mm/s) followed by Evans blue a lbumin (EBA) infusion at three flow rates Backflow was quantified as the percentage of tracer outside the target T he hole left in tissue along the needle track was measured to evaluate tissue disruption and injury at two time points after needle retraction. Radial stress at the needle tissue interface produced by tissue compaction during needle insertion (pre stress) was considered to influence backflow since pressur e in the infusate has to overcome pre stress to separate the tissue from the needle and produce backflow. Therefore, hole measurements were used to estimate pre stress using a hyperelastic model for the tissue. The insertion force and the displacement of t he material before puncture (dimpling), which is proportional to the surface strain, were measured to evaluate tissue damage. The retraction force was used to calculate friction stress which is proportional to pre stress Backflow increased
14 with increasing insertion speed as much as 2.46 fold for 10 mm/s compared to 0.2 mm/s Fast insertion speeds produced more injury including tissue bleeding and disruption Significant changes in hole areas were observed between time points after needle retraction indicat ing some tissue swelling before fixation. P re stresses were compressive ranging between 0 and 485 Pa. Average pre stress at 10 mm/s was 1.46 fold the average value at 0.2 mm/s. P re stresses were found to vary along the needle track and were smaller within white matter T he insertion force and the dimpling were increased for increasing insertion speed which explains the higher tissue damage for this insertion speed. Average friction stress was decreased for increasing insertion speed. For this test system, b ackflow was increased with higher insertion speed probably due to the lower pre stress produced which was due to greater local injury at the tissue needle interface.
15 CHAPTER 1 INTRODUCTION 1.1 Motivation Successful treatment of neurological disorders requires effect ive techniques to deliver drugs to the central nervous system (CNS). Systemic delivery can be problematic because of the blood brain barrier (BBB)  which is a layer of cells that cover the inside surface of cerebral blood vessels and prevents the passiv e passage of large molecules from the bloodstream into the extracellular space, the interconnected fluid filled space in which cells and blood vessels are embedded . Convection enhanced delivery (CED) is technique to deliver macromolecules directly to t he CNS and bypass the BBB. In CED, a cannula (needle) is inserted directly into the parenchyma and infusate is delivered at controlled flow rates into the extracellular space. With convection (advection) as the dominant transport mechanism, this technique can provide local delivery and larger distribution volumes than diffusion dependent methods . Since CED requires the surgical insertion of a needle into brain tissue, this raises specific concerns about minimizing injury and proper targeting. Even with stereotactic positioning, backflow (leak back) along the needle track continues to be a major source of improper targeting, especially at high flow rates [4, 5]. With backflow, infusate flows back along the outer cannula wall instead of penetrating into th e tissue at the needle tip. This backflow phenomenon is normally undesirable since large distribution volumes in tissue are not achieved, and drugs can reach regions of the brain where they are not effective, toxic or result in unintended side effects. Pre vious backflow studies have focused on developing new cannula designs to minimize backflow [6 8], imaging to
16 identify backflow in real time [9, 10], and modeling studies to predict the extent of backflow [11 13]. Backflow can be produced by injury or flui d tissue interactions or by a combination of these effects. Tissue disruption and injury associated with needle insertion may create a fluid filled gap between the needle and the surrounding tissue through which infusate can easily flow with little resista nce. Edema may also be produced in response to local tissue injury by the needle insertion, e.g. vascular damage and bleeding. Edema is an abnormal accumulation of fluid in the tissue which can result in swelling. Such tissue edema around the cannula may a lso introduce a high conductivity layer of traumatized tissue around the cannula through which infusate can easily flow . Alternatively, swelling may expand cells and tissue and help to close gaps at the needle tissue interface. Even when no fluid fill ed gap is present at the needle interface, fluid pressure at the needle tip is increased with infusion and this pressure can push tissue away from the cannula surface to create a fluid filled gap (intrinsic backflow) [11, 12]. Related to this mechanism, ti ssue coring, defined as tissue going inside the cannula during the insertion process, has being implicated as another cause of increased b ackflow [15 17]. Peaks in pressure which are required to clear the obstruction at the needle tip may create a bolus in fusion for backflow. Tissue stresses which are introduced or are pre existing may also influence the mechanics of backflow. Pre stress which we define as the radially directed tissue stress at the needle tissue interface is created by tissue displacement and subsequent compaction around the needle with insertion or tissue swelling due to injury. The introduction of this stress may reduce backflow since fluid pressure would have to
17 overcome it to separate the tissue from the needle surface . In our prev ious studies, we have evaluated the effect of pre stress on CED backfl ow in hydrogel tissue phantoms. The presence of pre stress resulted in reduced backflow . However, experimental measures or mechanical analysis of this stress is cur rently missing fo r CNS tissues. In living tissues, tissue swelling can be produced due to vascular disruption (vasogenic edema) or osmotic imbalances due to cell injury (cytotoxic edema) during needle insertion and can affect backflow if it produces changes in the stress d istribution before or during the injection process. Therefore it is also possible that swelling continues to occur after needle insertion resulting in time dependent pre stress and backflow effects. In addition, it should be noted that residual stresses al ready exist in brain tissues even before needle insertion. Pre existing tensile residual stresses in white matter and compressive residual stress in gray matter have been measured in murine brain slice studies . Such residual stresses may also influenc e backflow by altering the mechanics of tissue disruption, changing fluid gap dimensions, or changing the magnitude of pre stress. Studies on local CNS tissue injury have been previously conducted for insertion of electrodes and needles and for ballistic t ests. Blood vessel rupture and hemorrhage have been previously reported as evidence of tissue damage during electrode insertion in cortical tissue [19, 20]. In previous CED studies by White et al. , regions of increased tissue injury were also observed to be coincident with backflow. They reported hemorrhage mainly within white matter regions, and pressure induced tissue fracture near to the needle tip. In ballistic impact tests, Williams et al.  detected tissue fracturing and hemorrhage as the only acute (5 min after injury) reactions after
18 injury. This study also provided some information on the time evolution of tissue reaction, and over longer time periods (6 hrs), inflammatory cells, and cell and fiber tracts degeneration were observed. However, it should be noted ballistic tests are at much higher strain rates than needle insertion tests which can result in differences in tissue damage. Even though there are no previous studies that measure acute tissue swelling after needle insertion, there are some studies that have detected early tissue swelling due to other kinds of trauma. Byard et al.  detected brain tissue swelling 30 min after traumatic brain injury by measuring increases in intracranial pressure. Brain tissue swelling due to low oxyg en supply (hypoxia) has also been detected using magnetic resonance imaging as early as 20 min after traumatic insult , or even during insult injury . Few previous studies have focused on evaluating insertion speed on tissue damage [19, 20] and to the best of our knowledge none have looked at the effect of insertion speed on both: brain tissue damage and backflow. Previous studies have investigated the effe ct of insertion speed on electrode insertion. Fast insertions of sharp electrodes and electrod e arrays into cortical tissue have been previously shown to cause less vascular rupture and bleeding [19, 20]. Also, excessive tissue deformation prevented complete insertion of the electrodes arrays at lower speeds . Axial force measurements have been a useful tool to evaluate tissue deformation and damage as a function of insertion depth and speed [25 29 ]. The same trend for insertion force and for tissue damage has been observed across tissue types even though the effect of insertion speed on tissue damage has not bee n the same across tissue types. For insertions through human skin, high insertion force has been associated with greater
19 tissue disruption as evidenced by more intensiv e bleeding and reported pain [25 ]. In heart and liver tissue, it has b een found that greater insertion force and greater damage are produced inserting the nee dle with low insertion speed [26, 27 ]. In brain tissue, increased insertion force and increasing tissue deformation have been produced w ith high insertion speed [28, 29 ]. Insertion force is proportional to the energy delivered by the needle to the tissue and the energy should be related to tissue damage. Therefore, monitoring insertion force may be a simple way to evaluate the influence of insertion parameters on damage in vivo and in real time. Before surface puncture, surface tissue is primarily displaced in the axial direction of the needle. This axial displacement increas es the deformation of tissue [30 ] and with puncture may produce tissue damage far away from the n eedle. Dimpling is defined as the displacement of the tissue surface in the axial direction before puncture and has been previously used to evaluate the effect of needle insertion speed on brain tissue deformation [28, 29 ]. These studies found that dimplin g was increased for brain tissues for increasing insertion speeds. Therefore, tissue dimpling may also be used to predict the extent of tissue damage in real time during needle implantation. Because brain tissue is a heterogeneous material, acute brain tis sue damage due to needle insertion may be affected by the differences in tissue composition within different regions along the needle pathway. Force during insertion is expected to be sensitive to regions in the brain because each region is expected to hav e diff erent composition, stiffness [ 31 ], residual stres s [ 18 ], and failure strength [ 3 2 ]. Additionally, insertion factors like insertion speed [25 29], needle and tip geometry [29 ] affects tissue damage and insertion force.
20 In this study, the effect of th e needle insertion speed on tissue damage and backflow was evaluated by performing needle insertion followed by CED in the caudate putamen of the rat brain in vivo CED experiments infused visible Evans Blue Albumin (EBA) tracer at three flow rates and at three needle insertion speeds. Backflow was quantified as the percentage of tracer infusate outside the target Tissue damage and r egional variation of the hole left in tissue along the needle track was also quantified. Hole diameters were used as an input into a mechanics model to estimate pre stresses assuming brain tissue behaved as a neo Hookean material. N eedle insertion force measurements were performed in vivo in rat brain with different insertion speeds. Insertion force was used as a parameter to ev aluate tissue damage during insertion. In addition, dimpling allowed evaluation of the effect of insertion speed on surface tissue deformation which may also be linked to the extent of eventual tissue damage with penetration. Force measurements during retr action were used to evaluate the influence of the needle insertion speed on friction stress which is proportional to pre stress This approach allowed the evaluation of pre stress in real time, at an early time point. T his is the first study to provide a visual way to evaluate mechanical damage along the needle track within white and gray matter regions. Also, it is the first to quantify pre stress in CNS tissues as function of insertion speed and study its relation with backflow. The results of this study may be used in the future to improve the insertion processes of needles, electrodes, probes or other devices in the brain to reduce tissue injury
21 1.2 Specific Aims The objective of this research was to improve the reliability of CED as a drug delivery techn ique by reducing backflow. This w as accomplished by evaluating the effect of the needle insertion speed on local tissue damage and backflow. In vivo infusion experiments performed at different needle insertion speeds w ere used to evaluate the distribution of a tracer in rat brain. Pre stress and tissue damage (mechanical disruption) were evaluated by needle insertion experiments. Also, local displacement produced in the tissue by the needle insertion, w as evaluated by histology using hematoxylin and eosin s taining (H&E) on tissues surrounding the needle track. Additionally, in vivo need le insertion experiments were performed to measure the force during insertion with different insertion speed s These experiments allowed for the correlation between tissue dam age, pre stress and backflow as a function of needle insertion speed and provide d understanding of the mechanics of needle insertion and backflow. 1.2.1 Specific Aim 1: Evalua te the Influence o f Needle Insertion S peed o n The Backflow Phenomenon During CED A prel iminary study performed on agarose hydrogel showed that backflow was affected b y the needle insertion speed [13 ]. Moreover, damage in the tissue due to the insertion of electrodes was dependent on insertion speed [19, 20, 28, 29 ]. Therefore, it is importan t to evaluate whether backflow is dependent on needle insertion speed in live brain tissue. To evaluate the effect of needle insertion speed, a visible tracer, Evans Blue Albumin (EBA), w as infused in the caudate putamen of rat brain in vivo with different needle insertion speed s and different flow rate s The regions of the brain reached by the infusate were detected by slicing the tissue and imaging the infused
22 region. The volume of infusate inside and outside a specific brain r egion, the caudate putamen ( CPu) was calculated from the images. The percentage of the total injected volume that leaked outside the CPu was used as indi cator of backflow. Tracer distributions were analyzed to characterize CED backflow patterns in different brain tissue regions. 1.2.2 Spec ific Aim 2: Determine t he Tissue Damage and Pre Stress Between t he Needle a nd the Tissue and I ts Influence on Backflow f or Different Needle Insertion Speed s During insertion, the tissue is displaced and conformed around the needle. As a consequence of this deformation, a pre stress exists between the tissue and the outer needle wall. This pre stress may reduce the backflow because the infusate has to overcome this pre stress to separate the tissue from the needle and produce backflow. However, during the in sertion process, tissue is displaced also in axial direction which may produce higher damage in a wider volume of tissue Because of this damage, pre stress may be reduced or even gaps can exist between the tissue and the needle as a consequence of tissue disrup tion. The hypothesis was that this damage is dependent on needle insertion speed. T herefore, the size of the holes left by the needle were evaluated postmortem on tissue slices. T he pre stress was estimated by using the contraction of the hole in a h yperelastic model. Histological analysis w as used to further evaluate the damage and the displacement patterns of tissue produced by the needle insertion process. 1.2.3 Specific Aim 3: Evaluate the Effect of Insertion Speed on Force, Tissue D eformation and D amag e During Insertion and P re S tress During Retraction in Real T ime in L ive Brain T issue Axial force measurements during electrode implantations have been a useful tool to evaluate tissue deformation and damage as a function of insertion depth and
23 insertion s peed. Greater insertion force has been associated with greater tissue damage in several biological tissues [ 25 29 ]. Therefore, insertion force was used in this study to evaluate the influence of insertion speed on tissue deformation and dam age. Dimpling is the displacement of the tissue surface in axial direction before puncture and has been used to evaluate the effect of needle insertion spee d on brain tissue deformation [1, 28, 29 ] Therefore, in this study, dimpling was used to evaluate axial deformation of tissue before failure Additionally force measurement during retraction was used to calculate the friction stress at the needle tissue interface which is proportional to pre stress. In this way the influence of needle insertion speed on pre stress was evaluated in real time and at an early time point where the influence of swelling or other biological process is negligible. Specific Aim 2 allow ed direct observ ation of the disruption in tissue produced by the needle. However because it was a postmortem technique the calculated pre stress correspond ed to a time point later than when the infusion started during the CED backflow experiments of Specific aim 1 This disadvantage was overcome in Specific aim 3 by evaluating the pre stress using the force retr action measurement because with this approach the retraction was performed in vivo at the same time point when the infusion started.
24 CHAPTER 2 BACKGROUND AND PREVIOUS STUDIES 2.1 Brain T issue Brain tissue is composed of neurons, glia cells, blood vessels and extracel lular matrix Neurons are formed by one cell body, which contains the nucleus and surrounding cytoplasm, one axon and one or more dendrites. Although processing information and control of the body functions depend on neurons, they are only approximately 25 % of the cells in brain  Glia cells are specialized cells in the brain which provides protection, insulation and physical support to neurons. Glia consists of astrocytes, o ligodendrocytes, and microglia. Astrocytes are star like cells that make up 30 6 0% of the glia cells  Oligodendrocytes form the myelin: an electrically insulating layer that sheaths the axons of dendrites. Astrocytes and microglia are the cells that mainly respond to tissue injury. The inside surface of the brain blood vessels are covered by a layer of endothelial cells known as blood brain barrier (BBB) which limit the diffusion of large molecules and prevent in this way that bacteria or toxic substances reach the brain tissue  The cells and blood vessels are embedded in the e xtracellular space which is formed by a liquid phase and ions and molecules floating or attached to the cells. Extracellular space is important for diffusion and transport of molecules between blood stream and cells; therefore it is important for drug deli very  In general, brain tissue can be classified as Gray Matter (GM) and White Matter (WM). GM consists of neuronal cell bodies, dendrites, astrocytes, microglia and axons. WM consists mostly of glia cells and myelinated axons which produces the white appearance. White matter has anisotropic behavior with respect to elastic properties
25 and hydraulic conductivity defined as the ability of a porous media to allow a fluid to pass through it. 2.2 Convection Enhanced Delivery Because of the presence of the BBB poor penetration of some drugs into brain tissue after intravenous injec tion has been known to occur [33 ]. To overcome this obstacle, CED has emerged as a technique to deliver drugs directly into the parenchyma . The technique basically consists of th e insertion of a needle or catheter into the target tissue An infusion pump is connected to the catheter to drive the flow. Once in the brain, fluid flow is pressure driven through the extracellular space and spread throughout clinically significant volum es of brain parenchyma. In homogeneous tissues, drugs would ideally spread uniformly throughout the infusion volume. However, tissue tortuosity and heterogeneity cause a non ideal behavior and variable distribution patterns. Efficient distribution of the c onvected drug depends on many parameters, such as infusate characteristics, catheter type, flow rate and tissue characteristics [34 ] 2.3 Backflow An optimal infusion in CED has been frequently difficult to obtain because of the reflux of the infusate along the catheter track especially at high flow rates. With backflow, infusate flows back along the outer cannula wall instead of penetrating into the tissue at the needle tip. This backflow phenomenon is normally undesirable since large distribution volumes in tissue are not achieved, and drugs can reach regions of the brain where they are not effective, toxic or result in unintended side effects. Figure 2 1 shows a Magnetic Resonance Image (MRI) of the distribution of a MR visible tracer in a dog brain Relati vely good distribution of infusate is observed in Figure 2 1a however
26 considerabl e backflow was observed when a high er flow rate was used ( Figure 2 1b ) [9 ] Raghavan et al. [35 ] explained two causes of backflow. The first was the existence of a gap betwee n the needle and the tissue because of tissue disruption due to needle insertion. In this case the infusate can easily flow back through the gap with little resistance. The second is pressure associated with the infusion process which pushes the tissue cau sing its separation from the cannula wall creating a gap where the infusate can flow back (Int r insic backflow). Chen et al. [34 ], performing infusions in the rat brain, found that backflow increases with higher flow rate and cannula diameter. The higher ba ckflow with the greater cannula diameter can be due to the larger diameter having a greater probability of producing tissue disruption as m entioned by Raghavan et al. [35 ]. A new mechanism of backflow w as suggested by White et al. [14 ]. They performed in v ivo CED experiments in rat and pig using silica cannulas with diameters between 0.2 and 0.7 mm and two tip geometries (rounded and blunt tip). The tissue damage was evaluated using histology and the infusate distribution was measured using MRI. They claim that the backflow is produced by a high conductivity layer of traumatized tissue around the cannula where the infusate can flow easily. Therefore, to produce backflow, the tissue does not necessarily need to be damaged to the point of producing a gap betwe en the cannula and the tissue. In their study, smaller backflow was achieved with the smaller rounded tip needle Due to the heterogeneity of the brain and the higher conductivity of WM with respect to GM, backflow has been found to be dependent o n the nee dle pathway  Therefore, the location of the catheter can be optimized to reduce backflow [35 ].
27 Another important factor in backflow is the cannula design. Neeves et al. [8 ] reported the fabrication and evaluation of microprobes made of silicon with sq uare cross sections of 0.1x0.1 mm. The probes were tested in vivo in rat caudate putamen and considerably lower backflow was obtained in comparison with a 30 gauge blunt tip needle. The reason for the backflow reduction beside s the small size, may be the sharp tip geometry of the microprobes which may reduce tissue damage and increase the tissue compaction and pre stress around the microprobes. A cannula with a small diameter at the tip, a slightly larger diameter shaft and a sharp transition between the t wo sections (step cannula) was evaluated by infusing int o the caudate putamen of rats [6 ]. With the step cannula, infusions of up to 3 L /min were achieved without significant backflow. I n contrast, with a normal cannula, backflow was observed at flow rate s as low as 0.5 L /min The reduction in backflow with the step cannula is likely due to a seal effect produced by tissue compa c tion mainly in axial direction at the step of the cannula. An in dwelling step cannula was evaluated infusing in the putamen of a non human primate. MRI showed that free backflow infusions were achieved at a flow rate of 3 L /min [7 ]. A more sophisticated cannula which controls backflow by applying a suction pressure at the outer cannula wall some distance in the axial cannula dire ction has also been reported [3 6 ]. The suction not only absorbs any infusate that can be flowing back along the needle track, but also avoids the separation of the tissue from the cannula wall. This cannula was evaluated in a tissue phantom and with a comp utational m odel showing reduced backflow. However because of the complexity and probably the expensive cost of this cannula, it is not an optimal solution for clinical reduction of backflow.
28 2.4 CED M odels M odeling is an important tool for the analysis and im provement of CED. One effort in modeling infusion into the brain was the work of Basser [3 7 ] where the tissue was considered a biphasic medium composed of a solid matrix and an interstitial fluid. Analytical solutions for interstitial pressure, fluid veloc ity, and extracellular fraction were obtained. Solid matrix was considered isotropic homogeneous linear elastic and both solid and liquid were considered incompressible. To simplify the analytical solution, an infinite spherical geometry was assumed with a spherical infusion cavity; in this way, all the variables are dependent only on the radial coordinate. At the infusion cavity, constant pressure or constant flow rate were used as boundary conditions. With this model, dilatation of the material and increa se in the porosity were found near the infusion site. Another analytical solution for the linear poro elastic model was developed by Chen et al. [3 8 ]. They compared the analytical solution for infusion pressure and volume distribution with experimental val ues obtained on a tissue phantom gel. The porosity value needed for the model was obtained by adjusting the analytical solution to the experimental values. The value obtained was 0.25 which is close to the rep orted for brain tissue (0.26) [ 39 ] A model con sidering linear poroelastic material with spherical symmetry was pr esented by Smith and Humphrey [ 40 ]. They obtained the transient response of pressure and fluid velocity in brain and tumor tissue. They included transvascular fluid exchange in this model b ut they found no important infused fluid loss to the vasculature in the case of brain tissue. A poro elastic model considering an isotropic media and spherical sy mmetry was solved numerically [4 1 ]. In this study, a sensitivity analysis of
29 pore pressure, fl ow velocity, dilatation and solute concentration with respect to the mechanical properti es of the tissue was performed. The main limitation of these previously developed CED models is the assumption of isotropy and spherical symmetry which produces a solut ion depending only on the radial coordinate. This spherical model also assumes no backflow Moreover, because of the heterogeneity and anisotropy of the tissue, the pressure, velocity, dilatation, and drug distribution are not symmetric. A model including more realistic tissue geometry was developed to simulate the drug infusion in spinal cord using th e Finite Element Method (FEM) [4 2 ]. The use of MR images of the spinal cord allowed the model to consider the heterogeneity of the tissue and the anisotropy o f white matter. Although that model was three dimensional it was simpl ified by projecting a single MR image because of the small variation in white mater in the cranio caudal direction. The model considered the infusion site as a spherical cavity in a rigi d porous media. In other studies, instead of FEM, Finite Volume Method (FVM) has been use d to model CED in spi nal cord [4 3 ] or human brain [4 4 ] using Diffusion Tensor Imaging (DTI). The use of DTI in those models allowed consideration of not only the heter ogeneity but also the anisotropy of the tissue. By using FVM and DTI, Kim et al. [4 5 ] developed a voxelized model of infusion into the rat brain. The use of a voxelized model allowed them to build a structured mesh. The final drug distribution was reasonab ly predicted when compared with experimental resul t s presented by Astary et al. [4 6 ]. T hese studies [4 1 4 5 ] considered the tissue as a rigid porous media. Therefore, it is not possible to model tissue swelling. Furthermore, the relative movement of the tis sue near the needle produces separation of the tissue from the
30 needle. Also, pressure in the fluid furthers the separation of tissue from the needle and produces backflow which ultimately affects the drug distribution. Predicted displacements of up to 8 mm of the tissue at the infusion site, were reported by Smith and Garcia [4 7 ] using a numerical simulation of CED considering the brain as a poroelastic media with spherical symmetry, infusion in a spherical cavity, and considering the solid matrix as a non linear elastic material. They also predicted considerable changes in porosity due to deformation of the tissue. However the radial displacement repo rted [4 7 ] may be overestimated because they considered infusion in a spherical cavity at the needle tip and did not consider that under relatively high pressure, tissue separates from the needle and produces backflow. This backflow releases the pressure at the needle tip, increases the infusion area, and may prevent further expansion of the spherical cavity at t he needle tip. Therefore, it is important to consider the tissue needle interface in modeling tissue deformation. 2.5 Backflow M odeling Morrison et al.  and Raghavan et al. [12 ] have proposed models to calculate infusion pressure and backflow dis tances as function of flow rate Backflow distances were defined as the distances in axial direction from the needle tip to the point of maximum tracer penetration back along the needle track. Both models considered an axisymmetric geometry with the needle at the c enter surrounded by a linear poro elastic material and a narrow region between the tissue and the needle where the backflow takes place. These models account for the separation of the tissue from the needle and calculate the backflow distance. They found t hat backflow distances increased with high needle diameter. These models did not deal with the drug distribution or the infusion parameters inside the tissue but only at the tissue needle interface. These studies also
31 did not consider the effect of pre str ess produced by the deformation of the tissue during the needle insertion but assume zero stress at the tissue interface. 2.6 Needle Insertion M echanics Chen et al. [4 8 ] performed force measurement during the insertion of a 3 mm diameter catheter into 0.6% a garose hydrogel and in vivo porcine brain. Forces of the same order were obtained with both materials. As the catheter was moved against the hydrogel or the brain tissue, the force increased followed by an abrupt decrease when the catheter penetrated the s ample (Fig ure 2 2). After that, the force increased again until the catheter was stopped. Forces have also been measured during insertion in rat brain of cylindrical stainless steel probes with different tips and at d ifferent insertion speeds [ 49 ]. As expe cted, sharp probes resulted in smaller penetration forces than flat punch probes, and in general, higher forces were detected during faster insertions. The stiffness of the tissue was also rate dependent based on the viscoelastic nature of brain tissue [ 50 ]. Welkenhuysen et al. [2 8] and Andrei et al.  doing force measurements during insertion of probes at several speeds, found that besides the force, the dimpling was also rate dependent. However, they found no difference in chronic response of the tiss ue due to the implanted device for the speeds studied. This chronic response consisted of the activation of astrocytes which formed a compact sheath surrounding the probe to isolate it from the rest of the tissue. There was also microglia activation which can destroy the extraneous body of damaged cells. These pre vious studies suggested that, not only are the elastic properties of the brain rate dependent (viscoelastic material) but the rupture strength is also rate dependent However, the insertion speeds evaluated were in a range smaller than those evaluated in this study.
32 With respect to needle insertion modeling, some models have been developed to simulate the needle insertion process [26 30 ] However, they do not provide information about the stress fi eld around the needle due to tissue displacement or damage; therefore, they can not be used to estimate the pre stress between the needle and the tissue. Subbaroyan et al. [ 51 ] developed a finite element simulation to calculate the stresses in brain tissue around an inserted electrode; however that model did not consider the stresses due to the insertion process, because it was assumed that the electrode was already implanted and only stresses due to movement of the electrode were evaluated. A more complete simulation was developed to evaluate the penetration of a s olid rod on myocardial tissue [5 2 ] considering non linear viscoelastic material and failure of the tissue. This study was focused on penetration force which was compared numerically and experiment ally with good agreement; however, less focus was given to pre stress. The presence of pre stress between tissue and a needle used for CED was first mentioned by Raghavan et al. [12 ]. However, this pre stress was not quantified or considere d in their model Metz et al. [5 3 ] measured the elastic modulus of brain tissue inserting a balloon catheter inside the brain and pumping water and measuring pressure. Although those measurements could be used to quantify pre stress, the time point where those measurement s were taken was not clear. Therefore the relaxation of the brain may have affect ed the measurement Moreover, due to the large diameter (1.83 mm), this catheter would not be ideal for pre stress testin g in rodent brains. Tissue strain due to electrode ins ertion has been e valuated by Bjornsson et al. [19 ] by injecting vascular fluorescent markers into the brain before electrode insertion. This method was
33 useful to evaluate strain and vasculature damage due to needle insertion for their specific tip geometry However, it did not provide local information about the stress field around the electrode beca use of the limited resolution of vascular density and because the relation between strains and stresses was not provided Although Raghavan et al. [12 ] recogniz ed tissue disruption around the needle as a main cause of backflow, no study has been done so far to quantify this phenomenon. In the study performed by Bjornsson et al. [1 9 ], only the vasculature disruption was evaluated and in the stu dy performed by Whi te et al. [14 ], the tissue with BBB disruption was evaluated but gaps in the tissue due to tissue disruption were no t considered important for backflow Other stu dies like those by Welkenhuysen et al. [2 8] studied the chronic reaction of tissue due to elec trode implantation but not the acute response or the presence of tissue disruption along the needle track 2.7 Brain Tissue Mechanical P roperties A wide range of mechanical properties have been measured experimentally for brain tissue  which can result in a wide range of predicted mechanical behavior. This in the modeling. In vitro studies  have shown that GM presents an approximately isotropic behavior and WM beh aves as transver se ly isotropic during oscillatory shear tests. T he hydraulic conductivity of the tissue, which governs the conductance of fluid through porous media, is still not clearly determined for brain tissues Frequently, a simplified approach is ta ken to estimate the hydraulic conductivity assuming that the infusion takes place in a spherical cavity whose radius is equal to the needle radius. However, the infusion site is not a spherical cavity because of backflow [12 ]. Backflow
34 distances along the needle track are potentially smaller if the needle is inserted through regions with greater hydraulic conductivity  since with high conductivity fluid can penetrate into the tissue rather than flows back along the needle track T herefore, conductivity is important in the infusate distribution and the magnitude of backflow. DTI studies have shown isotropic diffusivity of GM and transversely anisotropic diffusivity of WM , and based on that, anisotropic conductivity has also been assumed for WM  For excised brain tissu e samples, Fransechini et al. [32 ] showed that brain tissue behaves like a solid non linear almost incompressible material with hardening and softening at high deformations and fibrous like fractures. In this study it was also shown that the mechanical properties change with the direction of applied force and the region of the brain. I ncompressibility of excised brain tissue was also found by Libertiaux et al. [5 5 ] by doing uniaxial compression tests and measuring the strain by using image analysis. The incompressibility of excised tissue may be because of cytotoxic edema which moves fluid from the extracellular space to the inside of the cells. However, live tissue may be compressible because the presence of fluid in the extracellular space which may be squeezed due to compression producing global tissue shrinkage. Some hyper elastic models have been proposed for brain tissue. For example, Mi ller and Chinzei [5 7 ], propos ed an Ogden hyper elastic model Lee et al. [5 8 ] used a modified n eo Hookean model to obtain the mechanical prop erties brain tissue by fitting the results of a finite element model to experimental values of displacement of tissue slices during indentation.
35 It has been found that mechanical properties of brain tissue are strongly dependent on the region a nd the direction in the brain. v an Dommelen et al. [ 59 ] reported, in general, a higher value in the shear modulus of WM with respect to GM measured by indentation in vitro of swine brain. Besides direction and region of th e brain, differences in mechanical properties of human and swine brain tissue were found in shear tests by Prange and Margulies [ 60 ]. The shear values reported in this study were considerably s maller than those reported in [ 59 ]. Although the hyperelastic m odel is suitable for brain tissue under quasi static conditions, the viscoelastic model is more appropriate under dyna mic conditions. Some studies [ 50 6 0 62 ] have determined the viscoelastic mechanical properties of the brain by indentation. High variabil ity in these mechanical properties was also found. For example, in indentation tests in vivo [ 50 ], it was found that the viscoelastic properties depend on the indentation speed and whether the indented tissue was preconditioned or not. The rate dependent m aterial properties of biological tissues can be explained by the mechanical interaction of the different components like fibers, cells, or bundles of cel l s with the surrounding media [6 3 ]. Also the water content which dissipates energy because of relative motion between the solid and liquid phases produces a viscoelastic behavior. In the specific case of brain tissue, the interaction between bundles of cells, cells, and the materials within the cell, the interaction between different branches or components like microtubules and neurofilaments, may dissipate energy. This energy dissipation would be the main reason for the highly viscoelastic mechanical properties of brain tissue
36 2.8 Possible Biological Processes Affecting Tissue D isruption 2.8.1 Residual S tress Residu al stress is an important factor affecting hole measurement s in brain slices pre stress and t herefore backflow. Xu et al. [18 ] found evidence of residual stress in the mouse brain. They analyzed 500 m thick slices in coronal plane containing sections of gray matter from the cortex and the thalamus and sections of white matter from the corpus callosum, cingulum, and external capsule. They found that the white matter was under tension residual stress ranging between 0 and 1.2 kPa, and the gray matter under compression residual stress ranging between 0 and 0.2 kPa The backflow experime nts in the present study were performed in a region with an architecture that is similar to the region studied by Xu et al. [18 ] and the GM of the caudate putamen will be consi dered instead of the thalamus. Therefore, it is expected that the compressive residual stress in the putamen may help in reducing backflow. However, the tension residual stress in the corpus callosum may open the hole, separate the tissue from the needle a nd increase the backflow to that region 2.8.2 Edema Edema is an abnormal accumulation of fluid in tissues that produces swelling. It can be caused by injury infection or disease. Cervos Navarro and Lafuente [6 4 ] reported that edema is one of the earliest events after brain injury and is produced mainly in WM Uterberg et al. [6 5 ] describe four types of brain edema: vasogenic edema which is caused by BBB barrier disruption and produces accumulation of extracellular water; cytotoxic edema characterized by accumula tion of intracellular water; interstitial edema observed in hydrocephalus; osmotic edema caused by imbalances of substances that produce flow of water into cells. The first two types of edema are
37 reported to be important in brain injury [6 5 ] but the influ ence of edema on CED has not been studied. By ard et al. [22 ] studied the evolution of br ain tissue swelling with time. They impacted the left temporal lobe region of a sheep and monitored the intracranial pressure after the injury by inserting a probe 1.5 cm dep th into the left parietal lobe. Since the probes were implanted after the impact, the pressure was not monitored immediately after impact; so that the first pressure measurement was taken 30 min. after impact. They performed measurements in an injure d and in a non injured animal. Significant increase in pressure was observed in the injured animal even at the first time point (30 min) and the pressure continuously increased during 4 hours of the monitoring period. Interestingly, a slightly higher value was detected initially for the non injured animal but this pressure decreased to a steady state value; probably this initial high value was due to swelling produced by the insertion o f the probe. Williams et al. [6 6 ] evaluated the response of the brain ti ssue with time after ballistic penetration. They reported evidence of inflammation 6 hours after injury. In the specif ic case of CED, White et al. [14 ] reported evidence of edema around the cannula. 2.8.3 Hemorrhage White et al. [14 ] found evidence of hemorrhag e around a catheter used for CED especially within WM regions. They hypothesized that this hemorrhage increased the spread of infusate in WM region H emorrhage is likely evidence of tissue fracturing which would be a low resistant path for backflow during CED. 2.8.4 Gliosis Astrocytes and microglia cells react to injury or infections to protect the brain tissue. It has been reported that this reactive process starts around 24 hours p ost injury
38 [6 6 ]. If that is the case it should not affect backflo w during CED pro cesses shorter than 24 hours Welkenhuysen et al. [2 8] evaluated the influence of electrode insertion speed on the astrocytes reaction by performing histology 6 weeks post implantation. They did not find any influence of the insertion speed on the reaction 2.8.5 Blood Brain Barrier D isruption Groothuis et al. [6 7 ] evaluated the change in permeability of blood vessels in brain tissue around inserted cannulas in the rat. After the cannula implantation, they injected sucrose intravenously at different time points. They reported an increase of 19 fold in vessel permeability of brain tissue for the time point immediately after inserti on of cannulas. White et al. [14 ] reported BBB disruption in brain tissue around a cannula used for CED and they claimed that the incre ase in permeability of the tissue produces backflow. 2.9 Figures Figure 2 1 Magnetic resonance image of drug infused in dog brain: a) relatively good distribution of infusate, b) considerable backflow .
39 Figure 2 2. Ins ertion force on agarose hydrogel (left) and porcine brain (right) .
40 CHAPTER 3 INFLUENCE OF NEEDLE INSERTION SPEED ON BACKFLOW IN 0.6% AGAROSE HYDROGEL 3.1 Overview A study to evaluate the effect of needle insertion speed (0.2 and 1.8 mm/s) as well as needle dia meter and flow rate on the extent of backflow and local damage to surrounding tissues was conducted on a transparent tissue phantom 0.6% (w/v) agarose hydrogel Needle insertion experiments were also performed to evaluate local damage at the needle tip a nd to back out the pre stress in the surrounding media for speed conditions where local ized damage was not excessive. Pre stress values were then used in an analytical model of backflow. 3.2 Infusion Experiments Infusion tests were performed on 0.6% (w/v) aga rose based hydrogel (TreviGel 5000, Trevigen Inc., Gaithersburg, MD). The infusion system consisted of a syringe pump driving a 100 l gas tight syringe (Hamilton, Reno, NV) coupled to 40 cm of minimally compliant polyaryletheretherketone (PEEK) tubing (1 mm inner diameter and 1.58 mm outer diameter). This infusion line was coupled to a stainless steel blunt tip needle which was inserted into the hydrogel. Two different needle diameters were tested, 22 and 32 gauge (0.7176 mm and 0.235 mm outside diameter, 50.8 mm length, Hamilton, Reno, NV). The needle was driven by hand with a micrometer attached to a stereotaxic frame (Kopf, Tujunga, CA) to a 2 cm depth at two speed levels, 0.2 mm/s ( mm/s) (low) and 1.8 mm/s ( mm/s) (high) These speeds were in the range of those evaluated by Bjornsson et al.  Multiple infusion rates were used to Reprinted with permission from C Insertion Speed on Backflow f or Convection 134, 041006 1.
41 L of bovine serum albumin labeled with Evans b lue (1:2 molar ratio) solution was i nfused at 0.5, 1, 2, 5, /min. Experiments were repeated six times for each test group. Backflow distances were measured immediately after the end of infusion using the micrometer of the stereotactic frame. Backflow distances were measured as the length fro m the cannula tip to the point of maximum dye penetration back along the needle track. 3.3 Damage and Plastic Deformation D amage was investigated by imaging the needle tip and surrounding media during insertion through hydrogel slices. Experiments were perfor med on 2 cm thick hydrogel slices using 22 gauge stainless needles. Needles were advanced completely through the specimen. Tests were conducted using the same set up and insertion speeds tested in backflow experiments. Images were recorded with a digital c amera located directly over the samples. This experiment was repeated 10 times for each insertion speed. In another series of experiments, diameters of the open holes left after needle retraction were measured to quantify plastic deformation and provide a basis for estimating residual stress We employed 28, 22, and 20 gauge blunt tip needles with 0.362, 0.7176, and 0.9081 mm outside diameters, respectively. Needles were inserted through a 2 cm hydrogel slice and left in place for five minutes before retrac tion. These experiments were also performed at the same two insertion speeds. Measurements were repeated 10 times at different sites for each needle.
42 3.4 Pre stress Calculation A Neo Hookean pre stress model was developed to model the stress state at a time po int after the needle was fully inserted. An idealized insertion was assumed where tissue was radially displaced and conformed around the inserted needle. It should be noted that this model does not account for the no pre stress case where tissue damage re sults in a gap between the tissue and the needle. Also, the pre stress condition was determined at equilibrium some time after needle insertion; therefore viscoelastic effects were not considered. Plastic deformation within the hydrogel was estimated from hole measurements in hydrogel (see previous Section 3 .2). The stress in the radial direction (pre stress) was calculated as the compression load on the surface of the hole necessary to re expand the opening radius ( ) after the retraction of the needle to the radius ( a ) of the needle. Our group has previously reported mechanical properties using a Neo Hookean model for 0.6% hydrogel [58 ]. A shear he current study, the hydrogel was considered approximately incompressible and stresses were calculated using  ( 3 1) where s is a material constant, is the gradient deformation tenso r, and is the Kronecker delta tensor. The simplified energy function was ( 3 2) where is the shear modulus. Axial symmetry and a semi infinite medium were considered because of the relatively s mall diameter of the needle compared with the tissue sample. Also, a plane strain condition was assumed because of the relatively
43 large length of the needle with respect to the diameter. This was assumed to be valid along the axial length of the needle whe re strains are basically planar due to the radial displacement of material. At the needle tip, stress and strain distributions are more complex and plane strain conditions no longer apply. Stretch ratios were calculated by using a volume conservation appr oach. The material in the annular zone between + and was displaced to the annular zone between a+dR and a (Fig ure 3 1 ) where R and r are radial coordinates. Expressing this relation in equation form ( 3 3) Therefore, the stretch ratio in the radial direction was ( 3 4) The constant s was calculated far away from the cannula where the stress and displacements were zero, and the stretch ratio was equal to unity. From Equation ( 3 1 ) s was equal to After calculating the derivative of W the stress (Equation ( 3 1 ) ) becomes ( 3 5) For the pre stress, only the normal stress in the radial direction was considered (3 6 ) 3.5 Backflow Model A simplified backflow model was implemented in which calculated pre stress was used to estimate backflow distance. The model was based on the model of Morrison et al. [11 ], in which total flow ( ) is infused through a hemisphere at the needle tip and a cylindrical surface surrounding the needle When pre stress was present, backflow
44 distances were assumed to be small. Therefore in this study, the pressure loss within the annular backflow region was neg lected. Backflow distances were calculated by doing a mass balance of infusate in the tissue surrounding the needle law The flow through the hemisphere at the tip of the cannula was taken to be half that for spherical cavity whose radius was assumed equal to the radius of the needle On the cylindrical surface, only uni directional flow ( in the radial direction was considered The backflo w distance (Z) was found to be (3 7 ) were is the needle diameter, is the hydraulic conductivity of the tissue phantom, is the infusion pressure, and is a constant. In this study, a value of was used [11 ]. (Varying this parameter between 2 and 6 cm resulted in only small changes i n predicted backflow distance.) For the permeability, a value of was used as reported by Raghavan et al. [12 ]. Unlike previo us studies [11, 12 ], backflow distances did not depend on the porosity but only on the permeability of the porous media. Influence of porosity is implicit in the permeability. At the porous media interface, the infusion pressure, Po, was set equal to the total stress in the porous media ( 3 8 )
45 where is the pre stress, is the interstitial fluid pressure and is the porosity assumed equal to 0.6 During inf usion, the infusion pressure and the interstitial pressure at the needle tissue interface was assumed to be equal. 3.6 Results Figure 3 2 shows measured backflow distances for the two needle diameters, the two insertion speeds, and the flow rates evaluated. When the needle was inserted at high speed, backflow was considerably smaller than at the low i nsertion speed For the 1.8 mm/s insertion speed, backflow distances smaller than 1 cm were obta ined for both needle diameters. Infusions using the larger 22 gauge needle resulted in a slightly lower backflow distances Extensive backflow was obtained in e xperiments performed using the 0.2 mm/s insertion speed see Figure 3 3 In the majority of cases, backflow reached the surface of the hydrogel; therefore, the maximum reported backflow distance was 2 cm. Typical distribution patterns obtained inserting ne edles at high and low insertion speeds are shown in Fig ure 3 3 a and 3 3 b respectively. For those experiments performed inserting the needle slowly, dye patterns were not as axisymmetric as those obtained when the needle was inserted more rapidly. Also, th e tracer penetration distance ( e ) was smaller and not clearly defined when the needle was inserted slowly. When needles were inserted at the lower speed (0.2 mm/s), accumulation of hydrogel and a large zone of damage was observed at the needle tip. Materia l was observed to build up and advance at the front of the needle tip during forward insertion, see Fig ure 3 4 During retraction the accumulated material initially followed backwards movement of the needle tip (Fig ure 3 4 c ), but at a certain retraction di stance (~0.3 mm) this material detach ed. The material build up was not transitory and was maintained
46 after the needle stopped and following needle retraction. Small amounts of accrued material were also observed to go inside the hollow needle tip during in sertion (arrow in Fig ure 3 4 d ) In addition, the zone of accumulated material was wider than the outer cannula diameter contributing to a region of damage w ider than the inserted cannula. Build up of hydrogel at the needle tip was not observed when the nee dle was inserted at higher speed (1.8 mm/s) indicating less material damage during insertion, see Fig ure 3 5 The difference in hydrogel damage modes between insertions at high and low speeds was more evident when the needle was allowed to exit the opposi te side of the hydrogel sample. Fig ure 3 6 a and 3 6 b show two representative cases of needles inserted at low speed In Fig ure 3 6 a there was an approximately symmetric accumulation of hydrogel material at the tip. The more common case, where the material was accumulated on one side of the needle, is shown in Fig ure 3 6 b. Average thickness of accumulated material at the tip in the radial direction was 0.18 mm ( mm). When the needle was inserted at high speed, no measurable accumulation was observed o n the needle tip Figure 3 7 shows surface hole s left in the hydrogel after needle retraction For high speed insertion, open diameters were measured to be smaller than the needle outer diameter (Figure 3 7 a) Table 3 1 summarizes average diameter measurem ents, and calculated radial stretch ratios for different diameter needles inserted at high speed. When the needle was inserted at the low speed, holes were not clearly smaller than the inserted needle, see Fig ure 3 7 b. In the majority of the cases, the hol e was not circular
47 but irregular in shape. These larger diameter holes reflected greater local damage, and no pre stress or backflow was predicted for these low speed cases. Pre stress was calculated for high insertion speed tests. Fig ure 3 8 shows the ca lculated residual stress corresponding to measured radial stretch ratios for varying needle gauge. There was no significant difference in the average stress values, even though needle diameters were considerably different. The average value of pre stress f or all three gauges was calculated to be The calculated infusion pressure was Backflow distance as a function of flow rate was predicted for high insertion speed tests using the simplified model Fig ure 3 9 compares experimental and calculated backflow d istances for 32 and 22 gauge needles. Calculated backflow values were determined to be greater than experimentally measured values at high flow rates, but predicted values were smaller than experimental values at low flow rates. It should be noted that the model did not apply when the flow rate was lower than 0.3 for the 32 gauge needle and 2 for the 22 gauge needle in which case all the infusate was infused at the cannula tip (hemisphere at end of needle) and no backflow along the cylindrical surface was predicted 3.7 Discussion In this study, influence of insertion speed on backflow was investigated using an agarose hydrogel which has been previously used as a facsimile of brain tissue for exploratory experiments [ 12, 48 ]. The advantage of using this tissue phantom is that it is transparent and allows fo r direct measurement of backflow distances.
48 Some variation in the needle insertion speed was expected since mi crometers were driven by hand. However, backflow and damage findings were found to be significantly different for the two speed levels tested, a nd an important reduction in backflow was obtained by inserting the needle at a faster speed (P value < 0.01) This improvement may be explained by lower damage observed at this insertion speed. D amage was assessed by observing accumulation of hydrogel at the needle tip during needle insertion and by the extent of plastic deformation in the hydrogel after needle removal. For the faster insertion speed, there was minimal observed accumulation of hydrogel at the needle tip which corresponded to smaller backfl ow distances. On the other hand when the needle was inserted more slowly, extensive accumulation of material at the needle tip was observed, and the holes left by the needles were irregularly shaped. It is likely that the accumulated material at the tip cr eated a channel in the hydrogel that was larger than the outer diameter of the needle, forming a gap between the tissue and cannula that promoted extensive backflow. In most test cases, backflow was observed to reach the surface of the sample. Comparable backflow distances were obtained at the faster insertion speed and these backflow distances were smaller than those measured by Raghavan et al. [12 ] in hydrogel infusion experiments. This was probably due to differences in local hydrogel conditions near th e needle. In their experiments the needle was put in place before the hydrogel solidified; therefore, no pre stress was present due to needle insertion. These results suggest that the presence of pre stress may reduce the extent of backflow. In hydrogel, e vidence of pre stress was the reduction of the insertion hole diameter after needle retraction. In brain tissues, White et al. [1 4 ] have conducted similar needle
49 insertion experiments to investigate local tissue damage. They also saw that holes left in tis sue after needle retraction were always smaller than the needle suggesting that pre stress may also exist in brain tissues. Other evidence for pre stress in tissues is indirect and based on studies measuring needle penetration and retraction forces in hydr ogels and brain tissue [ 28, 29, 49 ]. These forces may result from radial pre stress and friction, as well as, adhesion between the tissue and the needle surface. In general, pre stress and backflow will depend not only on elastic material properties but a lso on the plastic properties like yield strength. In this study, a simple approach based on an experimental measure of hydrogel deformation, i.e., changes in open hole diameter after needle retraction, was us ed to calculate the pre stress. In this way, pr e stress was estimated directly without using plastic properties of the material which are poorly known. This approach makes several assumptions including that the material was incompressible, all displaced material conformed around the needle with no accu mulation at the needle tip, there were no hardening effects, and plain strain conditions existed. Our current model also does not account for consolidation effects due to compaction of tissue at the needle tissue interface. Rather, neoHookean mechanical pr operties were used based on a previous brain tissue slice indentation s tudy by our group (Lee et al [58 ]) in which mechanical properties were calculated for varying Poisson ratio. Any consolidation of tissues would also have occurred during compressive ind entation testing. In the presence of pre stress, backflow distances were measured to be slightly shorter for larger diameter needles While it might be expected that larger diameter needles have larger contact stress with the hydrogel, the pre stress estim ated using
50 hole contraction measurements predicted simi lar values for varying gauges. Larger diameter needles also have a greater surface area at the tip of the needle and along the cylindrical surface around the needle. Because of this greater surface are a, less backflow distance is required for larger needles for the same flow rate and pre stress since they have greater surface area for infusate penetration into hydrogel compared to smaller needles When considering only the pressure necessary to open a g ap between the needle and an elastic material (without pre stress), Morrison et al.  and Raghavan et al. [11 ] found that the backflow was higher for larger needle diameters. This is because only an elastic deformation around the needle was considered. When compared to experimentally measured backflow distances, the simplified backflow model assumed a linear increase in backflow dista nce with increasing flow rate. This model tended to underpredict backflow distances at lower flow rates and overpredict ba ckflow at higher flow rates. This discrepancy may be partially due to the tracer penetration distance which was maximum at low flow rates, in addition to certain model assumptions. For example, the idealized hemisphere a nd cylindrical geometry; also, the m odel assumed uni directional flow with a constant infusion pressure al ong the fluid tissue interface. However, some loss of pressure likely occurs within the annular gap between the needle and tissue, and in this case, the fluid pressure within the tip reg ion should be higher than the pre st ress value used in this study. With higher infusion tip pressure, more infusate is advected into tissue at the needle tip, and less infusate flows back to create backflow. This effect becomes larger at higher flow rates and explains backflow overprediction as flow rate increased.
51 Accumulation of material at the needle tip and the corresponding damage in the surrounding media was a main cause of backflow; therefore, the needle tip geometry may be an important factor since different tips may result in different patterns of material accumulation. In the present study, only blunt tip needles were tested but it is possible that sharp tip needles produce less backflow if they move more material away from the needle or if less t issue sticks to the needle. 3.8 Summary For hydrogel infusions, the extent of backflow was found to vary with needle insertion speed. When the needle was inserted slowly, an accumulation of deformed material at the tip of the needle produce a region of damage that resulted in a gap between the needle and tissue that promoted backflow. On the other hand, when the needle was inserted faster, contact between the surrounding media and the cannula was maintained T he resulting pre stress considerably reduced backflo w. For these faster insertions, no significant difference in the pre stress as a function of the needle diameter was found. This finding was reflected in small differences in measured backflow for varying needle gauge. For varying flow rates, the simplifie d backflow model tended to underpredict backflow distances at lower flow rates and overpredicted backflow at higher flow rates
52 3.9 Figures and Tables Table 3 1. Hole diameter and radial stretch ratios for varying gauge needles inserted at a fast insertion speed (1.8 mm/s) Needle gauge Outer diameter of needle (mm) Hole diameter (mm) Radial stretch ratio 28 0.365 0.257 0.023 0.716 0.064 22 0.717 0.548 0.040 0.763 0.055 20 0.908 0.643 0.047 0.708 0.052 Figure 3 1. Material displaced due to cannula retraction. ( needle diameter; position of the free surface after needle retraction).
53 Figure 3 2 Backflow distances measured at varying flow rates, needle diameters, and insertion speeds. 22G = 22 gauge needle; 32G = 32 gauge needle; Slow = insertion speed of 0.2 mm/s; Fast = insertion speed of 1.8 mm/s; Each point correspond s to the av erage of n = 6; bars indicate standard deviation. Figure 3 3 Tracer distributions in hydrogel for a needle inserted at: a) 1.8 mm/s and b) 0.2 mm/s. Both Figures correspond to infusions with a flow rate of 0.5 Z is the reported backflow dista nce; is the perpendicular tracer penetration distance in hydrogels (0.7176 mm needle diameter)
54 Figure 3 4 Accumulation of hydrogel at the needle tip during: a) insertion at 0.2 mm/s, b) stop, c) retraction, and c) detachment of the accumulated ma terial at the needle tip (needle diameter = 0.7176 mm). Figure 3 5 Insertion of the needle at 1.8 mm/s showing no accumulation of material at the needle tip: a) insertion and b) retraction (diameter of the needle = 0.7176 mm). Figure 3 6 Needle tip after passing through a 20 mm thick slice of hydrogel: a) and b) accumulation of hydrogel at a tip after insertion at 0.2 mm/s, c) no accumulation of material after insertion at 1.8 mm/s (diameter of the needle = 0.7176 mm).
55 Figure 3 7 Hole in t he hydrogel left by the needle after retraction: a) needle inserted at 1.8 mm/s (the circle outlines the hole), b) needle inserted at 0.2 mm/s. The dark region surrounding the needle is EBA tracer. The needle was placed on the hydrogel surface next to the hole to provide a length scale (needle diameter = 0.7176 mm) Figure 3 8 Hydrogel pre stress for varying needle gauges. Pre stress was calculated from hole diameters after fast insertion see Table 2 1. Figure 3 9 Predicted and experimental back flow distances for a) 32 gauge and b) 22 gauge needles.
56 CHAPTER 4 EVALUATON OF THE INFLUENCE OF NEEDLE INSERTION SPEED ON BACKFLOW DURING CED INTO THE RAT BRAIN 4.1 Overview Chapter 4 presents the evaluation of backflow as function of needle insertion speed as men tioned in Specific A im 1. CED experiments infused visible tracer EBA at three flow rates and three insertion speed s in to the CPu of the rat brain in vivo The distribution of the tracer was evaluated by imaging slices of the brain Backflow was quantified as the percentage of the total infused volume which leaked outside the bounds of the target region through the needl e track. 4.2 Methodology 4.2.1 Animal Preparation and Surgical P rocedures Experiments were performed on male Sprague Dawley rats (300 350 g) using p rotocols and procedures approved by the University of Florida Institutional Animal Care and Use Committee. Anesthesia was initiated with xylazine (10 mg/kg, SQ) and isoflurane ( 4%) in oxygen delivered at 1.0 L /min. The head was shaven and disinfected with iodine/alcohol. Then animals were placed in a stereotaxic frame (model 900, David Kopf Instruments, Tujunga, CA ), and inha lation anesthesia (1.5% in 0.5 L /min of oxygen) was delivered via a nose mask. Body temperature was maintained (~37 o C) with a heating pad during the entire procedure. The skull was exposed by a mid sagittal incision that began between the eyes and extended caudally to the level of the ears to expose the bregma and lambda. Two holes of 2 mm diameter were drilled by hand into the skull abo ve the caudate putamen (CPu). Because dura mater is a high resistant membrane that covers the brain, inserting the needle into the tissue with intact dura mater can produce excessive needle resistance that results in greater tissue damage.
57 Therefore, dura mater was carefully taken off and any residual blood was cleaned using 1X phosphate buffered saline (1X PBS) and cotton swabs. The exposed surface of the brain was kept wet with PBS during the course of the experiment. The CPu was chosen as the target beca use it is a relatively homogeneous region composed mainly of gray matter and has been previously used for CED backflow studies [8 3 4 ]. To reach the CPu, the needle must pass through the cort ex, which is mainly gray matter and the external capsule (ec) whi ch is a white matter region (Figure 4 1 [69 ] ) The needle was stereotactically inserted at a 5 mm depth within the CPu. Bilateral infusions were conducted on the right and the left sides (AP = 0.5, ML = 3, DV = 5 ) 4.2.2 Needle Insertion and CED S ystem For al l experiments, a 32 gauge (0.235 mm diameter) blunt tip stainless steel needle ( Hamilton Reno, NV ) was used. Blunt tip needles were used in this research because needles with this kind of tip have been used in clinical trials [4 5 ], are readily commercia lly available, and have been used by our lab in previous CED studies. Moreover, even though new cannulas to minimize backflow have being designed [6, 7], the tip geometry of many of these new designs is a basic blunt tip Needle insertion (z direction) was controlled using a linear stage system (model LP28T, Applied Motion Products, Watsonville, CA ) mounted on a milling machine frame (Micro Mill DSLS3000, Micro Proto Systems, Chandler, AZ ). The stereotaxic frame and rat were placed on the x y table of the m achine frame, and the x y table was used to position the needle over the drilled holes in the rat skull at target AP and ML coordinates The needle was coupled to the linear stage using an electrode holder (model 900, David Kopf Instruments, Tujunga, CA ) w hich also provided alignment. The
58 infusion system consisted of a syringe pump driving a 100 L gas tight syringe ( Hamilton, Reno, NV ) connected to 40 cm of minimally compliant polyetheretherketone (PEEK) tubing (1 mm inner diameter and 1.58 mm outer diame ter). This infusion line was coupled to the needle via a reducing union ( Valco Instruments, Houston, TX ). 4.2.3 Backflow E xperiments E xperiments were conducted to measure backflow distributions at three insertion speeds : 0.2, 2, and 10 mm/s and three flow rates ( 0.5, 1, and 2 L /min ) Experiments were repeated 5 time for each experimental condition (45 experiments, 23 rats because bilateral injection). The first two speed levels were chosen because are in the range e valuated by Bjornsson et al. [19 ] and are clo se to the s p eed levels of previous study [13 ]. The highest insertion speed was chosen ac cording to Rousche & Normann [20 ]. They perform pneumatic implantation of electrode arrays into the cortices of cat brains with different speeds and found that an inser tion speed higher than 8.3 mm/s was necessary to minimize bleeding from pial vessels and perform a complete insertion. At each insertion speed 4 of Evans Blue albumin ( EBA ) was injected at flow rates 0.5, 1, and 2 L /min (5 experiments for each condition). EBA was prepared with 20 mg/m L of bovine serum albumin and 1.25 mg/m L of Evans b lue in 1X PBS. I njections were started 2 min after needle i nsertion to allow tissue relaxation. Also needles were retracted from the brain 3 min after infusion to allow pressure relief. All the retractions were performed at 0.2 mm/s, and it was cleaned with 3% hydrogen peroxide followed by 1XPBS after each retract ion Following CED experiments, rats were immediately euthanized by decapitation and freshly excised brains were sectioned into 100 m slices on a cryostat (~ 20 o C) in
59 the coronal plane. Slices within infused brain regions were mounted on microscope slides and placed on a white surface. White light was projected on each slice, then slices were imaged using a CMOS camera and saved as a RGB file. Backflow was quantified by using two different measures: (1) the percentage of tracer infusate outside of the targeted CPu region and (2) the distance from the needle tip to the point of maximum tracer penetration back along the needle track 4.2.4 Ba ckflow as Percentage of Tracer Infusate Outside the Target Previous studies [8, 46 ] have calculated the infused volume using a threshold intensity value and assuming constant concentration in the pixels with intensity above t his threshold. However, this ap proach introduces some error because the concentration of infusate is not necessarily constant through out infused brain tissues In this study, slice images were converted into concentration maps of Evans blue, which was the only visible component of the t racer. The d istributed tracer region inside the CPu was then outlined integrated and used to calculate the mass of tracer outside the target region Images of the slices were analyzed using a MATLAB subroutine and p ixel intensity was converted to dye con 70 ] ( 4 1 ) where is the pixel intensity, is the path length of light inside the specimen, is the Evans blue concentration, is the absorption or extinction coefficie nt, and is the incident light energy. When light travels inside a non transparent biological tissue, some photons are absorbed like in a transparent medium, some others take a non straight path and reach the detector (camera), and others take a non straight path, do not reach
60 the detector, and result in loss of light by scattering [ 70 ]. Therefore, in the modified M is an introduced factor that accounts for the increase in path length of photons that are detected but take a non straight pa th, and N is a factor which accounts for the scatte ring of light. Equation (4 1 ) can be written as (4 2 ) where and is the thickn ess of the slice. In Equation (4 2 ), was used because the light was assumed to cross the specimen then reflect back to the camera. Since is associated with light absorption, the absorption of the red component of the white light ( R ) was used to calculate concentration. A gray scale picture was obtained by normalizing at every pixel the red value by the total pixel intensity , where G is the green component and B is the blue component of the white light. The constant is thus the intensity of the background and was obtained by applying Equation (4 2 ) to a reference tissue slice without any tracer. A known mass of E vans blue ( ) was introduced into the brain based on the known concentration of Evans blue in EBA (1.25 mg/mm 3 ). To obtain the constant we applied mass conservation of Evans blue in tissue assuming no capillary clearance or metabolism of EBA over the course of the experiment. (4 3 ) where and are volume, differential of volume, area, differential of area, and number of slices respectively. Solving for (4 4 )
61 Once the constants and were obtained, Equation (4 3 ) was used to estimate the mass of Evans blue inside the target CPu ( ). The mass of Evans blue outside the CPu ( ) was not calculated directly from the images because blood was frequently observed outside the CPu, especially in white matter regions, which may int roduce error in the Evans blue calculation. Instead, was calculated as (4 5 ) were is the total mass of Evans blue infused. The backflow percentage ( ) was calculated as (4 6 ) An example of a tissue slice image and its corresponding Evans blue concentration map are shown in Fig ure 4 2 within the infused region. CPu and white matter regions where easily delineated on each slice as shown in Figure 4 2 a The r egion with EBA inside the CPu was isolated on each slice by circling and removing the region outside CPu using ImageJ software. F igure 4 2 c shows the magnitude of the gradient of Evans blue concentration. It can be seen that the gradient is higher inside t he CPu. Therefore, if constant concentration within infused regions is assumed some error is introduced. For the brain corresponding to the slice shown in Figure 4 2 backflow percentage was calculated with both approaches: constant and spatially varying c oncentrations. The backflow percentage calculated considering the concentration map was 21.9%. For the constant concentration assumption, pixels with intensity higher than a threshold of 0.317 mg/mm 3 deter [7 1 ] were used to calc ulate the backflow volume. The backflow in that case was determined as
62 (4 7 ) where and are the total number of pixels and the number of pixels inside CPu respectively. The obtained backflow was 19.17%. In that case, the backflow was underestimated by considering constan t concentration. 4.2.5 Backflow Distances Backflow distances from the needle tip to the point of maximum tracer penetration back along the needle track were also determined For each infusion site, a slice containing the needle track or as close as was possible was used. B ecause the position of the needle tip was difficult to identify in the brain slices, backflow distance s were calculated as the difference between the total insertion depth (5 mm) and the distance (d) from the point of maximum dye penetration alo ng the needle track to the surface of the brain, see Figure 4 3. 4.3 Results Figure 4 4 s hows tracer CED distributions in brain slices for variable insertion speeds. The backflow can be observed as leakage of infusate into the white matter region of the extern al capsule (ec) and the corpus callosum (cc) outside of the targeted CPu. Infusate was not observed to spread into the cortex due to backflow. In general, backflow increased with increasing insertion speed. E vidence of bleeding was frequently found around the external capsule especially for insertions at 10 mm/s Leakage of infusate into the lateral stripe of the striatum (Lss), as shown at the bottom of Figure 4 4 a and 4 4 c, was frequently found (~ 65% of cases) independent of flow rate or insertion speed Leakage to the lateral ventricles was observed in two cases, both of them performed with 10 mm/s insertion speed. No evidence of tracer
63 transport between the infused region in the Lss and external capsule was observed in any of the slices. For that reaso n, tracer distributions in the Lss were not considered in backflow calculations. Inside the target CPu, concentration was greater close to the needle track and decreased with the di stance from the needle (Figure 4 5). In the external capsule the concentra tion was more uniform (smaller gradient of concentration) along the infused region but was considerably low for r egions with hemorrhage (Figure 4 5c). For 10 mm/s insertion speed, the concentration and the infused volume inside the target was smaller than in those experiments performed with slow insertion. Average backflow percentages for varying insertion speed and infusion rate are shown in Fig ure 4 6 a Significant influence of the flow rate and the insertion speed was found (p value < 0.05 in both case s ) Average backflow percentages were 16.7, 30.8, and 36.6% for flow rates of 0.5, 1, and 2 L /min, respectively. Average backflow at 0.5 L /min was significantly smaller than the average at the other two flow rates (p value < 0.05), but there was no significant difference between average backflow at 1 and 2 L /min (p value = 0.22). Average ba ckflow percentage was also found to increase for increasing insertion speed. The average values for insertion speeds of 0.2, 2, and 10 mm/s were 17.0, 27.7, and 39.4%, respectively. The averages at the three insertion speed were all significantly different (p value < 0.05). Comparing backflow percentages obtained at 10 mm/s with respect to 0.2 mm/s insertion speeds, it was 2.46, 2.27, and 2.26 fold for 0.5, 1, and 2 l/min flow rates, respectively. Since backflow was diverted away from the needle track into white matter, backflow distances along the needle track was relatively constant over
64 the insertion speed and flow rates tested ( Fig ure 4 6 b ) and approximately cor respond to the distance between the needle tip and the beginning of the cortex No significant differences between flow rates or insertion speeds on backflow distance was found (p value = 0.52 for insertion speeds, and p value = 0.65 for flow rates). 4.4 Discu ssion In all CED experiments, backflow was observed even though it was smaller for the low flow rate and low insertion speed. Backflow was significantly increased for the higher flow rate and insertion speed. To induce backflow, tracer infusate likely had to overcome pre stresses which were predicted to exist in CPu to separate tissue and form gap at the needle interface. Once backflow reached the white matter of the external capsule, it was found to accumulate into th is region. Similar distribution pattern s have been previously observed in rat backflow studies [8, 11, 34] Interestingly in those previous studies as in the present study, there was no significant presence of the infusate in the corpus callosum even though this region is formed by white matte r as is the external capsule. This is p robably because the hydraulic conductivity may be higher in the external capsule than in the corpus callosum. The combination of low pre stresses and the higher hydraulic conductivity of the white matter with respect to gray matter [4 3 ] contributed to the diversion of infusate into white matter. There did not seem to be sufficient pressure to open up the tissue gap in the cortex region since no tracers were observed to reach the cortex or the surface of the brain durin g infusion. With increasing insertion speed, there was greater accumulation of EBA in the external capsule and corpus callosum. This was reflected in greater backflow measured as percentage of the total tracer infused. While backflow distances were not sig nificantly different this was to be expected since tracer was diverted from the needle track to the
65 white matter region. According to Raghavan et al.  backflow distance was higher when the needle was inserted only through gray matter than when it was i nserted only through white matter due to the higher hydraulic conductivity of white matter. Therefore, in the present work, the high conductivity region of the external capsule, prevented infusate from trave ling back longer distances along the needle track and reaching the cortex or even the surface of the brain. This does not mean that in the present study, backflow was decreased due to the presence of the whit e matter region, but the distribution of the leaked infusate changed ; infusate was diverted in to the white matter region instead of travel ing longer distances along the needle track. Moreover, backflow is not only determin ed by hydraulic conductivity but additional factors such as residual stress and tissue damage in the different regions of the brain can have important influence on backflow. As observed in this study, backflow distances were dependent on the needle pathway and the geometry of white matter intersection. With this white matter diversion of infusate, backflow percentage is a better measu re of the extent of backflow. The approximately uniform concentration at the white matter may be because the high conductivity of this region where infusate can flow easily with relatively low pressure. With respect to flow rate, backflow increase with inc reasing flow rate which was in agreement with numer ous previous studies [11, 12, 34 ]. However, with respect to insertion speed, the results of the present study are opposite to that of our previous hydrogel study (Chapter 3) where backflow was increased fo r low insertion speed. Since similar experimental parameters (test system, insertion rates, needle tip) were used, the different outcome is likely due to the difference in the mechanical behavior of
66 the materials. At low insertion speed in hydrogels, an ac cumulation of material was formed at the needle tip which created a larger zone of damage. This eliminated pre stress, leaving a gap between the needle and the tissue, which increased backflow. In contrast, in the present study, low insertion speeds produc ed smaller damage and higher pre stress in brain tissue as is shown in Chapter 5 With high insertion speed, hemorrhage was more frequent and severe which may be associated to tissue damage. The hemorrhage was mainly observed at the white matter region. In the hemorrhage region s, the Evans blue concentration was f ound to be considerably smaller. Therefore, bleeding itself probably did not increased backflow. The higher damage at the white matter may be because higher rupture stress in this region as suggest ed by the results of probe insertion force measurement rep orted by Welkenhuysen et al. [2 8 ]. They found and important increasing in insertion force when the probe tip met the white matter tracks. Wi th higher strength, white matter could be deformed greater and displace a wide volume of material before being broken producing more damage in the surrounding tissue. 4.5 Summary Backfl ow was greater for higher needle insertion speed in the CPu of the rat brain using a blunt tip stainless steel needle. These results are opposite to the backflow results obtained in hydrogel ( Chapter 3 ). Increased insertion speed not only produced greater backflow but also it frequently produced hemorrhage especially at the white matter region o f the external capsule and the corpus call osum.
67 4.6 Figures Figure 4 1. Coronal section of the rat brain showing a schematic of the needle track (in red) targeting the caudate putamen (CPu). Slice picture taken from [ 6 9 ]
68 Figure 4 2 Brain tissue slice with EBA distribution and corresponding concentration map: a) slice with EBA distribution with insertion speed 0.2 mm/s, b) concentration map of Evans Blue ( mg /mm 3 ) and c) magnitude of the gradient of concentration (mg/mm 3 mm)
69 Figure 4 3. EBA backflow distribution along the needle track (freshly excised 100 m brain slice). For backflow calculations, distance along the needle track from the brain surface to the top most backflow redin (d) was measured. Figure 4 4 EBA distributions close to the needle track for infusions performed at 0.5 /min with insertion speed: a) 0.2 mm/s, b) 2 mm/s, and c) 10 mm/s (ec: external capsule, CPu: caudate putamen, Lss: lateral strip of the striatum, cc: corpus callosum)
70 Fi gure 4 5. Evan s blue concentration close to the needle track (mg/mm 3 ) for infusions /min with insertion speed: a) 0.2 mm/s, b) 2 mm/s, and c) 10 mm/s. Concentrations corresponding to brain slices shown in Figure 3 4. Figure 4 6 Backflow as a function of insertion speed and flow rate following 4 L CED of EBA infusate : a) percentage of the EBA infusate outside the target CPu, and b) backflow distances. Bars indicate 1 standard deviation; n = 5 for each test group
71 CHAPTER 5 EVALUATO N OF TISSUE DAMAGE AND PRE STRESS BETWEEN THE NEEDLE AND THE TISSUE AND ITS INFLUENCE ON BACKFLOW FOR DIFFERENT NEEDLE INSERTION SPEED S 5.1 Overview Backflow can be produced by injury or fluid tissue interactions or by a combination of these effects. Tissue di sruption and injury associated with needle insertion may create a fluid filled gap between the needle and the surrounding tissue through which infusate can easily flow with little fluid resistance. T issue edema around the cannula may also introduce a high conductivity layer of traumatized tissue around the cannula through which infusate can easily flow [14 ]. With infusion, fluid pressure at the needle tip is increased and this high pressure can push tissue away from the cannula surface to also create a flui d filled gap [ 11, 12 ]. Tissue stresses which are introduced or are pre existing may also influence the mechanics of backflow. Pre stress is created by tissue displacement and subsequent compaction around the needle with insertion The introduction of this stress may reduce backflow since fluid pressure would have to overcome it to separate the tissue from the needle surface [13 ]. C hapter 5 presents the evaluation of the effect of needle insertion speed on tissue damage and pre stress N eedle insertion foll owed by CED experiments where performed in the caudate putamen of the rat brain in vivo CED infusion pressure was monitored to evaluate the extent of tissue coring. Local tissue damage in the vicinity of the needle was evaluated at two time points after n eedle retraction Excised tissue slices were evaluated histologically. The hole left in tissue along the length of the needle track was used as a damage parameter Hole diameters were also used as an input into
72 a mechanics model to estimate pre stress assu ming brain tissue behaved as a neo Hookean material Finally, damage assessment was used to understand the influence of injury and pre stress on backflow. 5.2 Methodology Two series of in vivo rat brain experiments were conducted: (1) needle insertion experime nts to measure local tissue damage along the needle track (infusion pressure measurements and hole measurements), n= 12 rats, (2) additional histological assessment, n = 3 rats 5.2.1 Animal P repara tion and Surgical P rocedures Experiments were performed on male Sprague Dawley rats (300 350 g) using protocols and procedures approved by the University of Florida Institutional Animal Care and Use Committee. For both series of experiments, t he surgical procedure was th e same as described in Section 4 .2.1 for backfl ow experiments 5.2.2 Needle Insertion and Pressure Measurement S ystem The needle insertion was performed using the same system described in Section 4 .2.2. Infusion pressures were measured using a custom designed fiber optic pressure transducer (model FOP MIV NS 663, FISO Technologies, Quebec, CA ) that was connected in line to the infusion system (descr ibed in Section 4 .2.2) between the syringe and the needle. Pressure was monitored at 1 Hz during the insertion, infusion, and retraction process. Prior to each brai n infusion, a reference pressure was recorded outside the brain with the infusion pump running at the test infusion rate, and with the needle at the same stereotaxic level as within the rat brain during CED. This value was a measure of the pressure drop wi thin the infusion line and was subtracted from the in vivo pressure measured during infusion in brain to determine pressure at the needle tip.
73 5.2.3 Needle I nsertio n E xperiments Experiments were conducted at the three insertion speed used for backflow experimen ts (0.2, 2, and 10 mm/s) and 4 rats were tested at each insertion speed (12 total rats tested). Two min after needle insertions 4 of tracer was infused at 2 L /min to reproduce conditions of backflow and to provide easier identification for hole meas urements in fixed tissue sections (see section 5 .2.5 ). Needles were retracted 3 min after inf usion to allow pressure relief. All retractions were performed at 0.2 mm/s. To evaluate possible changes in our damage parameter, hole diameter, due to acute injur y swelling, tissues were fixed at two different time points after needle retraction, 10 and 25 min Bilateral insertion experiments were conducted on each rat. After the first retraction, needle was cleaned with 3% hydrogen peroxide followed by 1XPBS. A se cond needle insertion was performed in the opposite side of the brain. The second retraction was 15 min after the first one. Then, the rat was immediately euthanized by perfusion fixation using 10% formalin (~10 min). 5.2.4 Infusion Pressure M easurements In line infusion pressure provided a separate measure of needle tissue interaction. In previous studies, spikes in infusion pressures have indicated the occurrence tissue coring within the needle tip [ 15 17 ]. In line pressure measurements were performed in vivo a t the same three insertion speeds (on the same 12 rats used for hole measurement) to monitor changes in in line fluid pressure during insertion and CED at 2 L /min 5.2.5 Hole Measurement Along the Needle Track Fixed brains were sectioned into 100 m slices in the horizontal plane on a cryostat (~ 20 o C). Approximately 45 slices were obtained along each needle track.
74 Slices were put in 1X PBS at room temperature during 30 min and then mounted on microscope slides for imaging. Needle holes on each slice were imag ed on a microscope (model IX 71, Olympus America Inc., Center Valley, PA ) and a digital camera (model SPOT RT3, Diagnostic Instruments, Inc., Sterling Heights, MI ) using bright light. Color RGB images of the hole were used to measure the area and the perim eter using a MATLAB script. Images were converted to gray scale and the hole was detected by considering only pixels with intensity higher than a threshold value determined from surrounding tissue in each slice using Otsu's method [ 7 1 ]. The pixel length (1 .81 m) was used to convert from pixels to perimeter and area measures. Since the holes were not necessarily circular, the measured areas were used to calculate equivalent diameter which was defined as the diameter of a circle with the same area of the hole. Also the aspect ratio, defined as the ratio between the equivalent perimeter calculated using the equivalent diameter and the measured perimeter, was used to evaluate the shape of the hole. The aspect ratio is 1 for a circular hole, and the aspect ratio go es to zero as the shape becomes more irregular. In some slices, especially those close to the brain surface, the hole was difficult to measure because of excessive bleeding produced during the dura mater removal or because the hole was closed. These holes were not included in the data analysis. 5.2.6 Histolog ical Assessments In an additional test group (n=3 rats ), needles were inserted at the three insertion speeds and 4 L of Albumin in solution with 1XPBS ( 20 mg/mL ) was infused at 2 L /min. Evans b lue was not infused to allow Hematoxilin and Eosin ( H&E ) staining.
75 Fixed brain tissue was sliced into 50 m thick slices in horizontal planes and standard procedures for H&E staining were followed [14 ]. 5.2.7 Pre Stress E stimation Hole diameters were input into a tissue m echanics model to estimate pre stresses along the length of the needle track. The model assumed an idealized insertion where tissue is radially displaced and conformed around an inserted needle. Axial symmetry and a semi infinite medium were considered bec ause of the relatively small diameter of the needle compared with the surrounding tissue. Also, a plane strain condition was assumed because of the relatively large length of the needle with respect to the diameter. This was assumed to be valid along the a xial length of the needle where strains are basically planar due to the radial displacement of material. The pre stress was determined at equilibrium after needle insertion; therefore viscoelastic effects were not considered. Pre stress was calculated at t he most immediate tissue fixation time point (10 min) where swelling was smaller In view of the fact that pre existing residual stress change the hole size, the approach to estimating pre stress considers the net effect of pre existent residual stress and pre stress generated during needle insertion. Pre stress was calculated using the model described in Section 3.6, as the compression load on the surface of the hole necessary to re expand the equivalent hole diameter ( ) to the diameter of the inserted n eedle ( a ) ( Figure 5 1 ) assuming the tissue as an incompressible neo Hookean material Equation 3 6 was used to calculate pre st r ess. Pre stress was calculate using equilibrium shear modul i reported by Elkin et al. [ 31 ] measured by micro indentation in cor onal slices for different regions of white matter
76 and gray matter in the adult rat brain. They reported a value of 485 Pa for the cortex and a value of 238 Pa for the corpus callosum. They did not reported values for external capsule and CPu. Therefore in this study 485 Pa was used for the gray matter region of the cortex and the CPu, and a value of 238 Pa was used for the external capsule. Pre stress was calculated with shear modulus of white matter at two points: 2.8 and 2.9 mm depths where according to a rat brain atlas [ 69 ] a ~0.2 mm thick layer of white matter corresponding to the external capsule is located. 5.2.8 Statistical Analysis Average infusion peak pressure and steady state pressure was compared for the three insertion speeds by using one way ANOVA. The total average hole areas, diameters, and pre stresses, were compared for the three insertion speeds and the two time points by using the Kruskal Wallis test because data was not normally distributed. Also, the Kruskal Wallis test was used to compare av erages areas, diameters, and pre stresses among regions in the brain along the needle track (cortex, external capsule, and CPu). Average holes area for each point along the insertion track was compared for the three insertion speeds and for the two time po ints by using two way ANOVA. At some depths along the needle track, the hole was difficult to see in images and was not measurable. Therefore statistical analysis was performed only for those depths where the complete data set which consisted of four repet itions was obtained. Data are presented as mean 1 standard deviation. All p values were considered significant.
77 5.3 Results 5.3.1 Infusion Pressure Small variations in infusion pressure were observed during needle insertion. When the infusion was started, th e pressure increased to a peak value between 14 and 25 kPa and then decreased to an approximately steady state value between 2.7 and 7.4 kPa ( Figure 5 2a ). The peak pressure may be indicative of tissue coring or some other initial tissue resistance to allo w infusate to come out. After infusion was stopped, the pressure decreased to a value above zero and went to zero after the needle was fully retracted. For infusions performed with needle insertion 2 and 10 mm/s, after the peak pressure, the pressure decre ased and reach a local minimum ( in Figure5 2a ) and then increased again between 0.2 and 2.46 kPa with respect to forming a second small peak ( ). Figure 5 2b s hows the averages of peak, steady state and the second peak of pressure ( ) for the different insertion rates. The difference between average peak and steady state values was not significant with respect to insertion speed (p value = 0.62 for peak pressure; p value = 0.29 for steady state pressure). That peak pressure did n ot depend on insertion speed indicated that tissue coring or tissue resistance due to tissue separation was not insertion rate dependent. The difference in the second peak between insertion speed 2 and 10 mm/s was not significant (p value = 0.6). Magnitude of the second peak of pressure is given by the difference The average of that difference was 0.54 and 0.84 kPa for 2 and 10 mm/s insertion speed respectively. The averages of these magnitudes were also not significantly different (p value = 0.39 ).
78 5.3.2 Hole Measurements In the majority of tissue slices the hole left by the needle appeared as empty space ( Figure 5 3a ), but in other slices (~5% of slices), especially those close to the surface on the cortex, the hole appeared as a narrow crack surrounded by EBA ( Figure 5 3b ). In other cases (~20% of slic es), only a region with accumulation of red blood cells was observed (Fig ure 5 3c ) where it was difficult to distinguish the actual track of the needle. Points close to the surface at insertion depths shallower than 1.4 mm were not considered in statistica l analysis since the complete data was no obtained for these positions. Therefore 36 hole measurements per experiment were considered. For two experiments performed at 10 mm/s insertion speed, the area of the hole close to the white matter regions was prac tically the same as the cross area of the needle (0.0434 mm 2 ). For the rest of the slices, the area of the hole was always significantly smaller than the area of the needle. The measured hole area in slices was averaged at each spatial location along the n eedle track for each insertion speed and fixation time point, Figure 5 4 Hole area was not constant with insertion depth. The hole increased in size starting at the cortex, between 2 and 3.5 mm depth reached a maximum, and decrease again for higher depths The region of maximum area corresponded to the white matter region (external capsule). Fig ure 5 4 d shows the aspect ratios of the holes along the needle track for the 0.2 mm/s insertion speed. Holes were more circular (higher average value and lower stan dard deviation) between depths of 2 and 3 mm. Holes were found to be less circular at the tip of the needle and near the surface of the brain. The aspect ratio profiles for other insertion speeds were found to be similar and are not shown.
79 The area of all the holes along the needle track for each insertion speed and time point was averaged and compared by using Kruskal Wallis test because the data was found to be non Gaussian (Fig ure 5 5a ). As shown in Fig ure 5 5 b, the average area increased for increasing insertion speed. Also average area was larger for earlier fixation, 10 min time point. The total area averages were 0.0105 0.007, 0.01492 0.009, and 0.0197 0.010 mm 2 for insertion speed s 0.2, 2, and 10 mm/s, respectively. Each average area was diffe rent from the others (p value < 0.05 in all the cases). The total average area for 10 min fixation time point was 0.01624 0.010 mm 2 and for the 25 min time point was 0.0139 0.009 mm 2 The difference between the two values was significant (p value < 0.0 5 ). Changes in hole area were more significant with insertion spee d than with fixation time point; the average area for 10 mm/s was 1.87 fold and 1.32 fold the average area for 0.2 mm/s and for 2 mm/s respectively, while the average area for 10 min before fixation was up to 1.17 fold the average area for 25 min after fixation. There was no interaction between insertion speed and fixation time point as evidence d by the fact that average hole area s at 10 min were always greater than average areas at 25 min. I n Fig ure 5 5b it can be seen that average hole area has a slowed growth with respect to insertion speed. Fig ure 5 5 c shows the results of an analysis of variance performed to compare the area of the holes at each point along the needle track. With respect to insertion speed, the difference was significant at the majority of the points between 2 and 4 mm depths where increased insertion speed produced increasing area of the holes except within a narrow region around 3 mm depth which probably corresponds to the white matter region. With respect to the fixation time dependent evaluation of tissue damage,
80 the difference was significant at 12.5% of the depth measures, where the p values were smaller than 0.05. Fig ure 5 5 d shows the average hole area for the thr ee main regions along the needle track. The average values were 0.0133 0.011 0.0218 0.009 and 0.016 0.008 mm2 for the cortex, external capsule and CPu, respectively. Each average value was significantly different from the others (p value < 0.05 in all the cases). The maximum difference in average area was between external capsule and cortex where average area of the external capsule was 1.64 fold the average area in the cortex. This ratio is smaller than the ratio between averages areas at 10 mm/s a nd 0.2 mm/s insertion speed. Therefore, the effect of insertion speed seems to be more significant than the variation introduced within different region s in the brain. Equivalent hole diameter (Figure 5 6) presented similar patterns as hole area profiles. The hole diameter was always smaller than the needle diameter in gray matter. Within white matter regions, equivalent hole diameters were practically the same as the needle diameter for two experiments performed at 10 mm/s insertion speed. The average diam eters were 0.107 0.042 0.137 0.047 and 0.152 0.046 mm for insertion speeds 0.2, 2, and 10 mm/s, respectively. Each value was significantly different from the others (p value < 0.05 ). For the different fixation time points, the averages were 0.138 0.049 and 0.125 0.047 mm for 10 and 25 min, respectively. These averages wer e also different (p value < 0.05 ) Figure 5 7 shows the hole perimeters as a function of insertion depth for the three insertion speeds and the two time points. Perimeters prese nted similar patters as diameters and areas. The average perimeters were 0.38 0.14, 0.48 0.15, and 0.53
81 0.15 mm for insertion speed 0.2, 2, and 10 mm/s respectively. All these averages were statistically different (p value < 0.05). The average perime ter for 10 min time point was 0.49 0.16 and for 25 min was 0.44 16 mm with significant difference between them. 5.3.3 Histology Several rupture mechanisms were observed in brain tissue slices. Figure 5 8a shows cracks and tissue stained with blood close to t he needle hole. In this study, tissue fracture is defined as a crack or narrow opening in interior regions of the tissue slice with bleeding. Blood that is present in the extracellular space before the rat death is not cleared with perfusion fixation. Ther efore, the presence of blood indicates that fracturing occurred in vivo during needle insertion or retraction and not during the tissue processing, e.g. slicing and mounting. Bleeding and fracturing was observed mainly in white matter regions of the extern al capsule and in the CPu. Individual cell separation from the tissue was also frequently observed at the surface of the hole ( Figure 5 8 ). These cells were detached from the bulk of tissue. Narrow strips of tissue were also torn from the bulk of tissue ( arrow in Figure 5 8b ). Layers of cells or fibers were observed to be curved at the hole boundary ( Figure 5 9a) and provided evidence of radial tissue compaction which was sometimes observed, especially at slower insertion speeds of 0.2 mm/s. The main diffe rence seen with varying insertion speed was the intensive bleeding and tissue fracturing often observed with higher speeds, 2 and 10 mm/s ( Figure 5 9b ). Fractures between 20 and 150 m long were found. Tissue fracturing was observed in approximately 35% of the needle insertion experiments at 2 or 10 mm/s insertion speeds. At slower insertion speeds, the bleeding was only observed close to the hole and tissue fracturing was not observ ed.
82 5.3.4 Pre Stress Figure 5 10 shows pre stress, calculated from equivalent hole diameters at the earliest fixation time point, for the three insertion speeds. The obtained pre stress values were between 0 and 485 Pa. Pre stress around the white matter region was 0 at five points corresponding to the two experiments performed at 10 mm/s insertion speed. A verage pre stress values along the needle track were 351.7 92, 266.7 109, and 240.6 125 Pa for insertion speeds of 0.2, 2, and 10 mm/s, respectively ( Fig ure 5 10d ). The average value for 0.2 mm/s was found to be significantly higher than the averages obtained for the other two speeds (p value <0.05 ), but there was no significant difference between averages for 2 and 10 mm/s (p value = 0.16). Pre stress wa s not constant along the needle track. It was maximum in the cortex close to the surface, was minimum in the white matter region and increased again with greater depths. The average pre stress values were 301.3 132, 115.7 42, and 296.0 98 Pa for cort ex, external capsule, and CPu respectively. The average value for external capsule was significantly different from the o ther two regions (p value < 0.05 ), but there was no significant difference between cortex and CPu (p value = 0.5) 5.3.5 Pre Stress Change and Error E stimation In S ection 5 .3.2 it was found that the measured hole areas changed depending on the fixation time. Therefore, it is possible that the process that reduced hole size between 10 and 25 min was present in the tissue even before the 10 min ti me point. If that is the case, the hole after retraction should be even bigger and should introduce a change in pre stress with respect to the pre stress at zero time point To estimate the error in pre stress, the error in area and hole diameter should be estimated first. To estimate the error in area, the average area was assumed to change linearly with time
83 between 0 and 25 min. Doing an extrapolation the average value at zero time point ( ) after needle retraction was 0.0178 mm 2 This would be the average area immediately after needle retraction, after viscoelastic relaxation, and before any biological process (as swelling) take place. The ratio between the areas at 0 and 10 min was (5 1 ) The equivalent diameter at zero time point ( ) as function of area was calculated as the diameter of the circle with area equal to the hole area (5 2 ) It is also possible that the fixation with 10% formalin produced tissue shrinkage that reduced the size of the hole and produced an overestimation of pre stress. To estimate the error because of fixation, shrinkage of 8% in volum e reported by Quester et al. [ 7 2 ] was assumed. The percentage of shrinkage can be calculated as (5 3 ) where and are the volumes at time 0 and 10 min of a cylinder of tissue of the same size of the needle hole. The most critical case, where the hole area reduction occurs only in the radial direction was considered. Also, we assumed that the shrinkage was uniform along the needle length ( 5 4 ) where is the length of t he hole. Following the s ame approach as in Equation (5 2 ) the equivalent hole diameter corrected for fixation ( ) was (5 5 ) Combining the effects of time and fixation we have
84 (5 6 ) The diameters were corrected using this factor and with the new diameters, the pre stresses were calculated again. After correction, pre stress decreased and became zero at 8 points around white matter region (~1.8% of the points) for three experiments. T he global pre stress average before correction was 307.25 Pa and after the correction the average was 292.8 Pa. Therefore there was an average error of 4.9% associated with the combined effects of tissue swelling and fixation. 5.4 Discussion In this study, tis sue damage was evaluated for varying needle insertion speeds Inline pressure was monitored during CED to evaluate the influence of insertion speed on tissue coring. Histological assessment and hole measurements were used to characterize tissue damage due to needle insertion. The size of the hole was measured along the needle track for three needle insertion speeds and at two fixation time points to evaluate the influence of insertion speed and tissue swelling on tissue disruption. This is the first study t o evaluate tissue disruption in this way and estimate pre stress along the needle track. Tissue damage, evaluated as the size of the hole left by the needle after retraction, bleeding, and tissue fracturing, was found to increase for increasing insertion speeds and was highe r within white matter regions. A statistically significant difference in hole areas with respect to insertion speed was found. A slowed growth of average hole areas with respect to insertion speed was also found. Therefore for insertion speeds higher than 10 mm/s, areas may reach an asymptotic value. There are not previous needle inserti on studies to directly compare. In previous studies [ 28 29 ]
85 electrodes with 10 50 o tip angles were inserted in vivo into a similar region of the brain They found that the surface displacement of the tissue before puncture (dimpling) was increased for increasing insertion speed. Dimpling is a measure of deformation and it is likely that greater deformation produces greater damage. If that is the case, o ur results are in agreement with those previous studies. Previous brain tissue studies have noted greater dimpling and insertion forces with increasing insertion sp eed [28, 29 ]. These higher deformation and force measu res may indicate greater damage in agr eement with the present study. The hemorrhage results of the present study are in contrast with previous studies on electrode insertion [19, 20] which have found that fast insertion of sharp tip electrodes produced less tissue damage (blood vessels rupture and bleeding). Probably the most important difference with experiments reported in  and  was the geometry because sharp electrodes were used. Moreover, there was a difference in tissue region. In these previous studies [19, 20], the electrodes wer e inserted only though the cortex. In the present study, the needle passed through the cortex, the external capsule, and the CPu and it was found that hemorrhage was observed primarily in the external capsule and the CPu. In the cortex no substantial bleed ing was observed except that it was sometimes produced during the dura mater removal. The hole left in tissue by the needle was used to evaluate tissue dama ge along the needle track. Hole area s were relatively small in the cortex, maximum within the white matter region and decreased again in the CPu. The decrease in hole area in the CPu indicates that damage did not accumulate with depth and there was negligible tissue accumulated at the needle tip. Hole measurements also showed that hole is more
86 circular i n shape at the inner cortex and the white matter region and was more irregular close to the surface and at the CPu. The irregular shape of the hole would result in non uniform pre stresses which may increase the likelihood of backflow in smaller pre stress i.e. less compacted, r egions at the needle periphery. The variation of the hole size along the needle track may be because of pre existing residual stresses in the brain as reported by Xu et al. . By analyzing cuts in brain slices, they reported that the white matter region is under tension and residual stress es are up to 1.2 kPa. Their cuts shared a similar compositional pathway (gray matter close to the surface, white matter, then gray matter) to the needle pathway in this study. Thus, tensile resid ual stress in white matter regions may be relieved when the tissue is cut by the needle resulting in a larger hole than that in a region without tensile residual stress. Conversely, compressive residual stress may tend to close holes. Tissue damage around the hole and hemorrhaging were also used to evaluate tissue damage along the needle track. Cell separation and strip formation were rupture types found at the surface of the holes. This strip formation may correspond to a fracture typical of a fibrous mate rial, which has been previously reported in tensile tests of human brain tissue [ 32 ]. However, this type of damage was observed in all regions along the needle track not only in white matter region. More intensive hemorrhaging was found in white matter reg ions and this is in agreement with hemorrhage patterns seen by White et al. in their CED studies into the rat CPu . Another observation of our study was tissue fracturing found primarily within white matter regions and within the CPu. White et al.  also reported tissue fracturing but only close to the needle tip (CPu) and not within white matter regions. Higher injury in white matter may be
87 produced by higher rupture stress in this region as suggested by previous probe insertion force measurements [ 28, 29 ] where significant increases in insertion force where found for white matter tracks. With higher strength, white matter could be deformed more and displace a large volume of tissue before failure, causing more damage. If tissue coring occurs during needle insertion, the pressure at the beginning of the infusion has to increase to expel tissue out and clear the needle. Tests were conducted to determine if higher insertion speeds produce greater coring as the cause of greater backflow; however, pressu re measurements showed no significant difference in the peak expulsion pressures indicating that tissue coring was not the main cause of insertion rate dependent changes in backflow. The peak and steady state pressure were in general smaller than those rep orted in previous studies [15, 17, 4 8 ] which may be due to the use of larger cannulas (23 25 gauge) or differences in the insertion system which resulted in less tissue coring. Incidence of backflow increased with the appearance of a second small peak in p ressure observed at 2 and 10 mm/s insertion speed. Therefore, the second small peak may be better indicator of backflow than the peak pressure. This small peak may be produced due by dynamic fluid tissue along the wall or with the cored tissue. With respe ct to the two fixation time points evaluated, significant difference in the hole size was found. Swelling, inflammation or other injury processes may be increasing tissue volume, during the 25 min of the experiment, to decrease the measured hole area. Byar d et al. , inserted a pressure sensor in a sheep brain 30 min after traumatic brain injury in vivo and detected significant increase in intracranial pressure
88 (~80%), probably due to brain tissue swelling. That increase in pressure may have started at a time point earlier than 30 min. Also, experiments on hypoxia have detected swelling using magnetic resonance imaging at times less than 20 min after insult [23, 24]. Therefore, hole measurements are a way to measure the effects of needle tissue tearing on ly as long as the time between the retraction and the measurement is minimized. If swelling starts immediately after injury, even the results of the experiments performed at the shortest possible time, 10 min before fixation, are be affected by swelling. T herefore, hole measurements include the combined effects of tissue swelling, tissue tearing due to needle in sertion, and residual stresses. Reported pre stress values therefore account for swelling at these specific fixatio n time points. These volumetric i ncreases in tissue with swelling will i ncrease pre stresses with time. If these changes are large enough, they may reduce backflow. However, previous CED studies have not found important influence of additional wait time before infusion in preventing backf low [34 ]. While fixation stops further swelling, this tissue fixation may also produce v olumetric tissue contraction [72 ] which may be a source of hole measurement error. Moreover, because it is a post mortem technique, these measures may be sensitive to tissue preparation, e.g. slicing and mounting. Despite these sources of error, statistical analysis showed a significant influence of needle insertion speed on tissue disruption, and there was no interaction between insertion speed and fixation time point. Therefore these measurement errors were probably smaller than the differences in damage introduced at the tested insertion speeds. Also, it is expected that these errors are
89 uniformly spread along the needle track and hole measurements still provide impor tant profile information. Hole measurements were used to calculate pre stress which provides important information for understanding mechanics at the tissue needle interface. This is the first study where pre stress is estimated for brain tissue, and this knowledge can be used to improve needle implantation surgical procedures or to improve prediction of backflow. Calculated pre stress was found to be significantly smaller for high insertion speed because of the greater tissue damage produced. Although pre stress was non uniform along the needle track and was smaller within white matter region, it was predicted to be always present within the CPu which is the region where infusate moves back up the needle track to reach the white matter region. Therefore, wi thin CPu there were likely no gaps between the needle and the tissue. The pre stress reported in this study considers the net effect of stress due to tissue deformation during needle insertion, swelling, and residual stress. Therefore, compressive pre stre ss was smaller in white matter region because of the presence of tensile residual stress. Without residual stress, pre stress due to needle insertion would be larger in white matter and smaller in gray matter than currently predicted. The errors in the ho le measurement previously mentioned also affect pre stress estimates. Other source of pre stress error is the idealized model based on a Neo Hookean incompressible material assumed. Reported brain tissue properties vary significantly [ 54 ], and if another c onstitutive model or different material properties are applied, the pre stress values will be different from those reported here. Moreover, at the brain surface and at the needle tip, plane strain conditions no longer apply and
90 stresses and strains are mor e complex than considered by our model. The error in pre stress based on previous studies of fixation shrinkage [72 ] was estimated assuming contraction of a cylinder (diameter equal to the hole diameter) in the radial direction, which is the most critical case. Also average hole areas at 10 and 25 min fixation time points were linearly extrapolated to estimate the hole area at a fixation time of 0 min. The combined effect of volumetric changes due to swelling and fixation resulted in an small net decrease in pre stress of ~5%. There exist some techniques to directly estimate stress at the tissue needle interface using probes or microballoons as reported in previous studies [ 53 73 ]. However those devices are too large to be used in the rat brain. Moreover, with those techniques only one average value for stress is obtained, and information about regional variation along the needle track is missing. Therefore, a new technique to evaluate pre stress in real time (right before infusion), in vivo and along the needle track needs to be developed. Further studies looking at additional factors influencing pre stress such as different brain location or needle tip geometry are also needed 5.5 Summary Tissue damage and pre stress during needle insertion and CED was evalu ated post mortem Tissue damage evaluated as hole area, hemorrhage, and tissue fracturing were found to increase with insertion speed. Tissue damage was found to vary along the needle track and was greater in the white matter region of external cap sule. Ti ssue pre stresses along the needle track are also reported. Increased backflow with high insertion speed was probably produced by increasing tissue damage which decreased or eliminate compressive pre stress.
91 5.6 Figures Figure 5 1. Hole of diameter expanded to the diameter of the needle ( a ) by the action of compressive pre stress ( ) Figure 5 2. Infusion pressure results during CED at 2 L/min : a) typical pressure as function of time during infusion for the three needle insertio n speeds; b) average peak, second peak ( ) and steady state pressure as a function of insertion speed (Bars indicate standard deviation; n = 8)
92 Figure 5 3 Different types of tissue damage left after needle retraction: a) approximatel y circular hol e surrounded by the infused EBA; b) hole as a slender opening/crack; c) red blood cell accumulation where it was difficult to measure the hole opening All the figures are at the same scale. 100 thick slices obtained from fixed brain tis sue
93 Figure 5 4 Area of the hole after needle retraction as a function of insertion depth for varying fixation time point s (10 and 25 min ites ) an d insertion speeds: a) 0.2 mm/s; b) 2 mm/s; c) 10 mm/s; d) Aspect ratio of holes as a fun ction of depth along the needle track for insertion speed 0.2 mm/s and 10 min before fixation. (Each data point corresponds to the average of n = 4 hole measures. Bars represent standard deviation
94 Figure 5 5 Variation of hole me asurements: a) histogram showing non Gaussian distribution of the hole area s for 10 mm/s needle insertion speed and 10 min fixation time (combined data from 4 needle tracks, 161 total hole measurements); b) average hole area s for the three insertion speed and the two time points evaluated considering depths greater than 1.4 mm. Each point corresponds to the average of n = 144 hole measurements; c) p value as function of insertion depth to evaluate the influence of insertion speeds (0.2, 2, and 10 mm/s) and time points (10 and 25 min before fixation) with depth a l ong the needle track; d) average hole areas for the three main tissue regions along the needle track. Each value corresponds to the average of n = 336, n = 48, n= 504 measurements for cortex, extern al capsule (ec) and caudate putamen (CPu) respectively. For cortex points between 1.4 and 2.7 mm depth were considered. For external capsule, points at depth 2.8 and 2.9 mm w ere considered. For caudate putamen, points at depth higher than 2.9 mm were consi dered. Bars represent standard deviation
95 Figure 5 6 Equivalent hole diameters as function of depth along the needle track for varying time point (10 and 25 min before fixation) with the needle inserted at: a) 0.2 mm/s, b) 2 mm/s, c) 10 mm /s (Each date point corresponds to the average of n = 4 measures. Bars represent standard deviation
96 Figure 5 7. Hole perimeters as function of depth along the needle track for varying time point (10 and 25 min before fixation) with t he needle inserted at: a) 0.2 mm/s, b) 2 mm/s, c) 10 mm/s (Each date point corresponds to the average of n = 4 measures. Bars represent standard deviation Figure 5 8 Needle t issue damage observed without staining: a) cell separation and bleeding nest to a needle hole in the external capsule for insertion at 10 mm/s b) fibrous tissue torn (arrow) next to a hole in the CPu for insertion at 0.2 mm/s. Both images are at the same scale. 50 m tick fixed brain tissue slices
97 Figure 5 9 Needle t issue damage observed with H&E staining: a) hole in external capsule white matter region showing deformation and compaction of the tissue around the needle hole In the enclosed region, layers of cells can be seen to conform around the hole (needle insertion at 0.2 mm/s); b) damage in caudate putamen showing extensive bleeding and tissue f racture at some distance of the hole (needle insertion at 10 mm/s)
98 Figure 5 10 P re stress along the needle track for the needle inserted at: a) 0.2 mm/s, b) 2 mm/s, c) 10 mm/s (Bars represent the calculated standard deviation, n = 4 ) d) Average pre stress for varying insertion speeds. Each average corresponds to the average of n = 14 4 pre stress values
99 CHAPTER 6 EVALUATON OF THE EFFECT OF INSERTION SPEED ON FORCE, DEFORMATION TISSUE DAMAGE, AND PRE STRESS 6.1 Overview Since CED requires the surgical insertion of a needle into brain tissue, this raises s pecific concerns about acute tissue damage which may produce improper drug targeting  i.e. backflow. Therefore, insertion tissue damage may limit the potential of CED treatment and needs to be characterized. Axial force measurements have been a useful tool to evaluate tissue deformation and dam age as a function of insertion depth and insertion speed [25 30 ]. Independently of the influence of insertion speed, insertion force has been associated with greater tissue damage. Dimpling is an important parameter to evaluate tissue deformation, because it is related to displacement of tissue in the axial direction before the needle penetration. Additionally force measurement during needle retraction allows quantifying the friction stress which is proportional to pre stress. The refore force measurement is a simple way to evaluate the influence of insertion process on acute tissue damage and pre stress. In Chapter 6 needle insertion force measurements were performed in vivo in rat brain with different insertion speeds. Mechanical parameters such as penetra tion force ( ) and force at the end of insertion ( ) were used to understand tissue damage during insertion. In addition, dimpling allowed evaluation of the effect of insertion speed on surface tissue deformation which may also be linked to the extent of eventual tissue damage with penetration. Force measurements during retraction were used to evaluate the influence of the needle insertion speed on pre stress at the needle tissue interface. Agarose hydrogel has been used as a phantom material for CED s tudies [ 12, 48 ]. However, under the same insertion conditions, the effect of needle insertion speed on
100 backflow and damage in hydrogel Chapter 3, was opposite to brain tissue as shown in Chapter 4 To further understand the difference in mechanical damage behavior between these materials, needle penetration force and dimpling were also measured on 0.6% agarose hydrogels. Proposed methods are a simple approach to evaluate the mechanics of insertion in tissues, in real time, as function of the insertion dept h. Phenomenological models of friction stress as a function of insertion speed, and correlations between backflow and friction stress were also obtained. By comparing with backflow data collected in Chapter 4 correlation between backflow and friction stre ss was also evaluated. This study provides information on the variation in tissue damage along the needle track and its dependence on insertion speed which can be used to understand insertion mechanics and improve the process to insert needles or electrode s in the brain. 6.2 Methodology 6.2.1 Animal Preparation and Surgical P rocedures Experiments were performed using protocols and procedures approved by the University of Florida Institutional Animal Care and Use Committee. A nimal prepara tion and surgical procedures w ere the same as described in S ection 4 .2.1 for backflow experiments. 6.2.2 Needle Insertion and Force Measurement System A blunt tip stainless steel needle ( Hamilton Company, Reno, NV ) with diameter 0.235 mm (32 gauge) was used. Needles were inserted to a depth of 5 mm into the CPu. Based on backflow experiments (Chapter 4 ) three insertion speeds were used: 0.2, 2, and 10 mm/s. Insertion speeds were controlled using a linear stage system (model LP28T, Applied Motion Products, Watsonville, CA ) mounted on a millin g
101 machine frame (Micro Mill DSLS3000, Micro Proto Systems, Chandler, AZ ). The stereotaxic frame and rat were placed on the x y table of the machine frame and the x y table was used to position the needle over on the drilled holes in the rat skull target co ordinates AP = 0.5 and ML = 3. The insertion and retraction force were measured with an axial tension/compression miniature load cell (model 31, Honeywell Sensotec, Columbus, OH ). The needle was held to the load cell with a custom polyether ether ketone ( PEE K) holder to provide alignment. The load cell signal was recorded by using an interface card (model LCIC WIM BEN, Omega Engineering, Stamford, CT ). Data were recorded at 1 kHz, 5.8 kHz, and 17.5 kHz for needle insertion speeds of 0.2, 2, and 10 mm/s, re spectively. 6.2.3 Force Measurement on Brain T issue The needle tip was positioned on the test tissue surface to set the zero position of the linear actuator. Then the needle was moved up 5 mm to start the needle movement from that point. In this way the accelera tion of the needle occurred outside the tissue and the needle had a constant speed when it reached the tissue. Before needle insertion, a baseline signal was recorded from the load cell over ~5 s to provide a zero force level. The acceleration and decelera tion of the needle was 300 mm/s 2 for all the experiments. During needle insertion, the needle was considered in contact with the tissue once the insertion force increased above the noise level (0.06 mN). After insertion, the needle was left in place for 2 min and then retracted at 0.2 mm/s to provide estimates of friction stress The needle was cleaned with 3% hydrogen peroxide followed by 1XPBS after each retraction. To simulate the condition of no tissue coring (tissue entering through the needle tip), th e needle hole was fill with cyanoacrylate glue.
102 Experiments were repeated 10 times for each insertion speed (15 rats with bilateral injection) From the recorded force profile, the peak force at the point of puncture or penetration (penetration force ( ) ), the force 2 min after the needle insertion (resting force ( ) ), the maximum force during the retraction (dragging force ( ) ), and the dimpling ( ) were used for comparison between varying insertion speeds. Dimpli ng was the distance between the spatial coordinate where the needle made contact with the tissue and the point of tissue puncture. 6.2.4 Force Measurement on 0.6% Ag a rose H ydrogel The methodology for force measurement on hydrogel was the same as in brain tissue Needles were inserted to the same 5 mm depth into samples of 0.6% (w/v) agarose based hydrogel (TreviGel 5000, Trevigen Inc., Gaithersburg, MD). The hydrogel samples were prepared by mixing 0.6 g of agarose powder and 100 ml of deionized water. The mixtu re was covered and heated until boiling and then poured into petri dishes to a 10 mm thickness, and allowed to solidify at room temperature. After solidification, a ~1 mm thick layer of deionized water was maintained on samples during the entire experiment to prevent dehydration. 6.2.5 Pre Stress M odel An approach similar to the used by Sharp et al. [49 ] was used to calculate friction stress at each point along the needle track during retraction. When the needle is being retracted, the contact of the tissue with the needle generates contact stresses in normal ( ) and axial directions ( ) (Fig ure 6 1 ). The radial contact stress is the pre stress, and
103 the axial stress is the friction stress that is overcome by the force of the actuator which is measured by the load cell. Friction stress and pre stress are related by ( 6 1) where is the friction coefficient which can be produced by contact between the needle and the solid matrix of the tissue, viscous forces in a film of fluid located between the needle and the tissue or a combinatio n of the two situations. To avoid variation in the friction coefficient between experiments, retractions were all performed at the same retraction speed (0.2 mm/s) Doing a balance of forces at time t when the needle is being retracted and a length is inside the tissue, neglecting inertial and gravity forces, and considering that the radial stress is not necessarily constant along the needle track ( 6 2) where is a differential of cylindrical surface area of the needle, is a integration variable and D is the needle diameter. Using the fundamental theorem of calculus to take the derivative with respect to of Equation ( 6 2 ), the radial stress at the tip of the needle can be dete rmined ( 6 3) The friction coefficient for in vivo needle insertion under similar conditions has not been previously measured. (It has been measured ex vivo with artificial lubrication [74 ]). Therefore, in this s tudy, the friction stress which is the product of the pre stress and the friction coefficient is reported ( ). To calculate the derivative in Equation ( 6 3 ) the force data was first filtered to eliminate the force variation due to the heart beating. T he derivative was then calculated
104 using the forward difference method dividing the total insertion depth (5 mm) into intervals of ~0.1 mm. 6.2.6 Statistical A nalysis Average insertion and retraction parameters (penetration force, resting force, dragging force, d impling, and pre stress) were compared for the three insertion speeds by using one way analy sis of variance (ANOVA). Since the insertion and retraction force along the needle track presented a non Gaussian distribution, comparison between average forces an d friction stresses for the three insertion speeds and considering all the points along the needle track where performed using a Kruskal Wallis test. This test was also used to compare pre stresses at the different regions in the brain along the needle tra ck. Data are presented as mean 1 standard deviation. All p values were considered significant. 6.2.7 Phenomenological M odels Based on the friction stresses experimental values, exponential models were chosen for friction stress as a function of inserti on speed. Also, as the insertion protocols in the present study were the same as in the CED backflow study (Chapter 4 ) backflow experimental values from that study were fit to a linear model as a function of friction stress obtained in this study. Only th e average friction stress values for each insertion speed were considered. The constants needed in the models were calculated by fitting the values of the model to the experimental values using the least square method. The goodness of the friction stress m odels was evaluated using the coefficient of determination ( ). The correlation between backflow and friction stress was evaluated by using the linear correlation coefficient ( ) defined as
105 ( 6 4) where are the backflow and average backflow values and and are the friction stress and the average friction stress. 6.3 Results 6.3.1 Needle I nsertion A typical force vs. time curve for needle insertion in brain tissue in vivo is shown in Fig ure 6 2 After the needle made contact with tissue, the force increased approximately linearly until the tissue was punctured or penetrated. After surface penetration, the force sometim es decreased slightly and then increased again. Immediately after the needle w as stopped the force decreased. After some additional time, ~ 0.6 s, the force stopped decreasing and the greatest variation in force was due The aver age force from 10 experiments at each insertion depth was calculated and is shown in Fig ure 6 3 For brain tissue (Fig ure 6 3a ), the average force was greater for high insertion speed. The average slope of the curve between the contact point and the punctu re point was 1.45 1.22, 2.32 1.68, 1.95 0.85 mN/mm for insertion speeds of 0.2, 2, and 10 mm/s, respectively. There was no significant difference between these averages (p value = 0.2). The total average force, calculated from the point when the need le made contact with tissue until the needle stopped was 1.344 0.61, 1.975 0.88, and 2.998 1.33 mN for insertion speeds of 0.2, 2, and 10 mm/s. Each average was significantly different from the others (p value <0.05 in all the cases). After puncture, the force increased with insertion depth until approximately 3
106 mm depth, which correspond to the start of the CPu. For depths greater than 3 mm, the force did not show a clear increasing or decreasing trend, but oscillated in value. Fig ure 6 3 b shows aver age force as function of insertion depth on hydrogel. A much higher peak force was observed at surface puncture for all the three insertion speeds. In nine of the ten experiments with a n insertion speed of 0.2 mm/s, a second peak force (38.15 2.2 mN) was also observed at 4.52 mm depth. For the insertion speed of 2 mm/s, the second peak (7.1 0.7 mN) was always observed at 1.83 0.14 mm depth. For insertion speed 10 mm/s no clear second peak was observed. The average slope of the curves before punct ure was 1.92 0.21, 2.51 0.17, 2.52 0.09 mN/mm for insertion speeds of 0.2, 2, and 10 mm/s. There was no difference between the average at 2 and 10 mm/s (p value = 0.8). The total average forces along the entire needle track were 11.0 8.4, 5.1 4.9 and 5.2 4.2 mN for insertion speeds of 0.2, 2, and 10 mm/s respectively. All these average s were significantly different. Average insertion forces were higher in hydrogel in comparison with brain tissue and presented an opposite trend with insertion sp eed. The average penetration force ( ), force at the end of insertion ( ), and dimpling with their corresponding standard deviation are shown in Table 6 1. For brain tissue, the penetration force at 0.2 mm/s was significantly smaller than at the other speeds tested. There was no significant difference between the penetration force at 2 mm/s and 10 mm/s (p value = 0.6). Increased insertion speed produced increased average with significant difference among all of them (p values < 0.05). Dimpling was increased for increasing insertion speed also. However, only the difference between the average dimpling at 0.2 mm/s and 10 mm/s was significant.
107 In the case of hydrogel, the penetration force at 2 mm/s was significantly higher than t he force for 0.2 and 10 mm/s (p value < 0.05 in both cases) and there was no difference between the force at 0.2 and 10 mm/s (p value = 0.058). Even though presented a maximum at 0.2 mm/s and a minimum at 2 mm/s insertion speed, the difference among the three averages values was not significant (p value > 0.5 in all the cases). Insertion speed had opposite effect on dimpling in hydrogel when compar ed to brain tissue. While dimpling increased with increasing insertion speed for brain tissue, it decreased for hydrogel. The averages dimpling were all significantly different each other for hydrogel. For hydrogel, force measures did not show clear sensit ivity over the range of insertion speed tested. However, dimpling was more sensitive to ne edle insertion rate dependence. 6.3.2 Needle R etraction Fig ure 6 4 shows a typical curve of retraction force as function of time for brain tissue. The raw data showed perio dic fluctuation because of the heart beating. Curves of retraction force for hydrogel were similar to brain tissue with the exception of the heart beating. When the needle retraction started, the measured force quickly changed from a compressive (shown wit h + sign) initial value, i.e. the resting force, to a tensile or pulling value (shown with sign). Once the pulling force reached a maximum magnitude value, i.e. dragging force, it decreased until reaching approximately zero when the needle reached the su rface of the sample. Because at the beginning of the retraction process, the force corresponds not only to friction force but also to the relaxation of compressed tissue at the needle tip, only the force after the maximum value was used to compare between insertion speeds and to calculate the friction stress. Thus, data
108 corresponding to depths greater than 4.5 mm were not considered in friction stress calculations. The average retraction forces from 10 experiments at each insertion depth are shown in Fig ure 6 5 For brain tissue (Fig ure 6 5 a), the magnitude of the retraction force was greater for those needles inserted at 0.2 mm/s and was smaller for those needles inserted at 10 mm/s. The average retraction force was 0.832 0.51, 0.609 0.32, and 0.257 0 .15 mN for insertion speeds of 0.2, 2 and 10 mm/s, respectively. The average retraction forces were all different from each other. In comparison with brain tissue, opposite behavior was observed during retraction in hydrogels except at the point of maximum magnitude, i.e. dragging force (Fig ure 6 5 b). The average retraction force was 0.068 0.05, 0.265 0.13, 0.282 0.15 mN for insertion speeds of 0.2, 2, and 10 mm/s, respectively. These values were also all different each other. Table 6 2 shows the ave rage resting forces ( ) and dragging forces ( with their corresponding standard deviations. For brain tissue, the resting force decreased for increasing insertion speed. All the averages were significantly different. Resting force in hydrogel was also decreased with increasing insertion speed and all the values where different. Dragging force was decreased for increasing insertion speed for brain tissue and all the averages were different each other. For hydrogel, average dragging force presented a min imum for 2 mm/s insertion speed which was significantly smaller than the averages at 0.2 and 10 mm/s (p value < 0.05 in both cases). However there was no significant difference between the average dragging force at 0.2 and 10 mm/s insertion speed (p value = 0.4).
109 In general, the mechanical behavior of hydrogel during insertion was opposite to brain tissue. One exception was the resting force that decreased with insertion speeds for both hydrogel and brain tissues. 6.3.3 Pre S tress Because the beginning of the cu rve of force during retraction does not entirely corresponds to friction but also to relaxation of the compressed tissue at the needle tip, the friction stress was not calculated before the point where maximum force value (dragging force) was reached. This point was not the same for all the curves (Fig ure 6 5 ) and was located at a depth greater than 4.5 mm. Therefore, friction stresses were not calculated for depth between 4.5 and 5 mm. Fig ure 6 6a, 6 6b, and 6 6c show the friction stress for insertion spee ds 0.2, 2, and 10 mm/s for brain tissue. The friction stress was higher for slower insertion speed with a minimum value between 2 and 3 mm depth, and it was close to zero for the entire needle track for insertion at 10 mm/s. The difference in friction stre ss between insertion speeds was significant except around 2.8 mm depth, which corresponds to the white matter region of the external capsule, and close to the surface of the brain (Fig ure 6 6d ). For hydrogel, when the needle was inserted at 0.2 mm/s, the f riction stress was close to zero for the middle part of the needle track and was higher at the needle tip and at the surface of the sample (Fig ure 6 6 e). For higher insertion speed the friction stress was more uniform along the needle track and, in general higher than that obtained at 0.2 mm/s (Fig ure 6 6 f, and 6 6 g). The difference in friction stress between insertion speeds was not significant at the needle tip and close to the surface (Fig ure 6 6 h). The average friction stress values for the three regio ns in the brain along the needle track, considering the three insertion speeds was 0.227 0.27, 0.222 0.19,
110 and 0.383 0.30 kPa for cortex, external capsule and CPu respectively. Average friction stress in CPu was significantly greater than average fri ction stress in cortex and external capsule (p value < 0.05 in both cases). There was no significant difference between average friction stress in cortex and in external capsule (p value = 0.15). Table 6 3 shows the total average (global) friction stress c onsidering all the points along the needle track and the 10 experiments for each insertion speed. These averages were significantly different from each other. Table 6 3 also shows the average friction stress value for each region in the brain along the nee dle track depending on the needle insertion speed. In all three regions the friction stress decreased with increasing insertion speed. The global average was closer to the average of CPu than to the average of the other regions. The average friction stress es for hydrogel are also included in Table 6 3. All the three average s were significantly different. 6.3.4 Phenomenological M odels In Table 6 3 it was observed that friction stress in brain tissue decreased for increasing insertion speed. Data was fit to an expo nential function of the form ( 6 5) where is needle insertion speed in mm/s and and are constants to be determine by fitting the model to the experimental values shown in Table 5 3. The constants and the coe fficient of determination for the global brain and the three regions are shown in Table 6 4. The results of the model and the comparison with the experimental values are shown in Fig ure 6 7 a. This model predicts that as speed goes to zero, pre stress has i ts maximum value equal to and pre stress tends to zero as insertion speed increases. Taking the derivative with respect to S, it was seen that the rate of friction
111 stress decrease at is given by the product of and Bot h constants are greater in CPu. Therefore i n this region the rate of friction stress decreasing with respect to insertion speed is greater (0.082 kPa/(mm/s)) and the minimum rate of change of friction stress with respect to insertion speed at was at the external capsule (0.028 kPa/(mm/s)). The rate of change of friction stress tends to go to zero as the insertion speed increases. The maximum value of was obtained at the external capsule which means a better fit with the experimental values in this region. Because friction stress in hydr ogel was increasing with increasing insertion speed, the proposed model was ( 6 6) The obtained constants are shown in Table 6 4 and the experimental results are compared to the model in Fig ure 6 7 b. As inse rtion speed increases, the friction stress in this model tended to an asymptotic value equal to kPa which is the maximum average friction stress value that can be achieved according to this model. Also, the model predicted that a minim um average friction stress, kPa, would be obtained inserting the needle at approximately zero insertion speed. The rate of change of friction stress with respect to insertion speed was maximum at and was given by the product kPa/(mm/s). Average backflow values for three insertion speeds (0.2, 2, and 10 mm/s) and three flow rates (0.5, 1, and 2 L/min) were taken from Chapter 4 and were plotted with the global average friction stress obtained at each insertion speed (Fig ure 6 8 ). For the approximately linear relation backflow vs. friction stress, the correlation coefficients ( ) considering the nine average backflow values was 0.723. The negative value means
112 that backflow was decreasing for increasing pre stress. The critical value of correlation coefficients for n = 9 samples was 0.666 for confidence level of 95% [75 ] Therefore, there was a linear decreasing trend with 95% confidence level. Table 6 5 shows the constants and obtained by fitting the data shown in Fig ure 6 8 to a linear function of backflow ( ) vs. friction stress ( 6 7) By doing a linear extrapolation to zero average friction stress, the maximum backflow percentage was found to be 30.63, 52.67, and 62.11% for infusion flow rates of 0.5, 1, and 2 L/min, respectively. Equation ( 6 7 ) may also be use d to calculate the required friction stress to have zero backflow at each flow rate. Those values were 0.83, 0.92, and 0.93 kPa for flow rates of 0.5, 1, and 2 L/min, respectively. However, the phenomenolo gical model of friction stress, Equation ( 6 6 ), fo und that the maximum global average friction stress that can be achieved for the experimental condition of the present study was 0.573 kPa. Given this maximum average friction stress and Equation ( 6 7 ) backflow is always predicted to be present and the mi nimum backflow percentage was 9.7, 19.8, and 23.8% for flow rates 0.5, 1, and 2 L/min, respectively. 6.4 Discussion Influence of needle insertion speed on force, deformation, damage and friction stress was evaluated on brain tissue and hydrogel tissue phantom For other biological tissues such as cardiac tissues it has been found that insertion force and tissue damage was decreased with high insertion speed [26, 27]. However, in brain tissue, this study found that increased insertion speed produce greater forc e and damage which is in agreement with previous studies of electrode insertion performed on rat brain tissue in
113 vivo [28, 29]. These results showed that brain tissue behaves in the opposite way to other tissues such as cardiac tissues. This mechanical beh avior during insertion may be because rate dependent failure strength, stiffness, or other mechanical properties of brain tissue which may have a higher effect at high insertion speed producing higher forces and damage on a wider volume of tissue During i nsertion, the stress distribution at the needle tip is complex and tissue may be broken by shear or tensile stress. Therefore both the shear and tensile strength affects the tissue rupture and the force during insertion. However there are not studies abo ut the rate dependency of these mechanical properties. The higher force w ith high insertion speed may also be due to higher friction stresses between the needle and the tissue. As the needle is inserted, fluid from the extracellular space is in contact with surface of the needle. This fluid is more viscous than water because of the presence of organic molecules . In a viscous fluid or for a viscoelastic tissue interface, the shear stresses are proportional to strain rate. Therefore, with increasing inserti on speeds, higher shear stresses may be produced between the needle and the tissue and between cells which are surrounded by fluid. In this way, the energy transfer from the needle may be spread through a wider region in the brain with increasing insertion and deformation rates. This energy transfer may be one reason why damage in tissue was found relatively far away from the needle surface. In contrast, with low insertion speed the fluid component of tissue (at the tissue needle interface and in extracellu lar spaces) has more time to redistribute and does not absorb as much of the energy introduced by the needle because with low insertion speed low strain rates are expected, and shear stresses are proportional to strain rates in a fluid. As a result, tissue deformation and stresses are concentrated
114 closer to the needle tip, producing injury only close to the needle surface and affecting less wide region of tissue. This rate dependent damage was seen in histological sections where, for high insertion speed, h emorrhage and tissue fracturing was observed in a wide region around the needle; whereas for low insertion speed, no significant damage was observed besides the needle hole itself (Chapter 5 ). In brain tissue, the average insertion force, the force at the end of insertion the penetration force, and dimpling were increasing for increased insertion speed. However, only the average insertion force and were significantly different for the three evaluated insertion speeds. Therefore, th ese two more sensitive parameters: the average insertion force and were more closely related and better indicators of tissue damage which was increased for increasing insertion speed. High sensitivity to insertion force was also found for the r esting force which was significantly different for the three insertion speeds. However, the trend was opposite to other parameters measured during insertion (average insertion force and ) because it decreased with increasing insertion speed. The refore, the resting force should be more associated with friction stress after insertion and was a good indicator of pre stress. Resting force is produced by the combination of friction stress along the needle and compressive stress at the tip. As high ins ertion speed s produced lower pre stress, the friction stress was also low which was reflected in the the restin g force. The force during insertion in brain tissue increased with insertion depth for depth smaller than ~3 mm where it reached a maximum. This maximum peak was located approximately at the transition between external capsule and CPu, and is close to the region where tissue damage was greater as observed in histological sections. The
115 increasing of force (the slope of the curve force vs. depth) wa s higher for high insertion speed. Therefore, the increasing of force with insertion depth may be due to friction stress between the tissue and the needle which is proportional to the needle insertion length and probably also proportional to the insertion speed. For depths greater than 3 mm, the force did not show a clear increasing or decreasing trend, but oscillated in value. In this region the friction coefficient may be smaller than in the cortex and in the external capsule. Another possible reason may be a lower rupture strength inside the CPu in the insertion needle direction or a different rupture mechanism in this region with respect to inner cortex and external capsules which may produce a differen t hole geometry as evidenced by the smaller aspect r atio of the holes found in the CPu (Figure 5 4d, Chapter 5 ) In hydrogel, the forces during insertion were not only quantitatively but also qualitatively different from those in brain tissue. In hydrogel, the second peaks of force observed for insertion s peeds 0.2 and 2 mm/s may be due to accumulation of material at the needle tip as shown in the hydrogel study (Chapter 3) The increase of the force to form a second peak and the sudden decrease in the force may be because the accumulated material increased to a critical size and then was separated from the needle tip. When the needle is free of accumulation, there is less damage with insertion and higher pre stresses are present. This can explain why friction stress was higher for low insertion speeds close to the surface of the sample (around 1 mm depth) and close to the needle tip (around 4 mm depth). Moreover, the high friction stress at the needle tip for 0.2 mm/s insertion speed explains the high dragging force obtained at that insertion speed which was comparable to that obtained at 10 mm/s. The difference in
116 tissue damage and pre stress between hydrogel and brain tissue may be because of microstructural and compositional differences in both materials which produce s different stress distribution s and di fferent rupture mechanism s In brain tissue the cells are surrounded by viscous fluid and connected by extracellular matrix consisting of long chain organic molecules . Therefore the stresses in brain tissue may be transferred mainly through the fluid l ayer as the long chain molecules are extended to allow for high deformation. Hydrogel is formed by short polymer chains which may become fully extended at smaller deformation due to extensive cross linking (hydrogen bound connections ) preventing impor tant viscous dissipation Therefore in this material the solid matrix may be connected by weaker bounds and if stresses are transferred mainly along macromolecular chains (not through viscous fluid as in brain tissue) then they can be easily separated to p roduce a more brittle material [76 ]. Other possible reasons for the rate dependent behavior may be higher elasticity of cells and extracellular matrix components, or stresses may be transferred differently through larger cells components (cell membrane and neurofilaments) which can move relative to each other versus through the uniform and homogenous the structure of hydrogel. The mechanical behavior of brain tissue during needle insertion was also opposite to cardiac tissue. The reason may be that, in card iac tissue, the cells are connected directly to each other through intercalated discs  while in brain tissue they are surrounded by fluid. Dimpling is closely related to tissue damage because with dimpling the tissue is displaced in the axial directio n. Greater axial displacement before puncture implies that more tissue is being moved in axial direction following puncture and not only displaced radially as in an ideal insertion. Therefore, dimpling may be directly related to the extent
117 of tissue damage Dimpling measures were found to be sensitive to changes in insertion speed on both materials. The opposite behavior of brain tissue and hydrogel was especially clear in dimpling. Increased dimpling with increased insertion speed was observed in brain tis sue which explains the greater damage in a wide region observed with faster insertion Greater dimpling may produce failure of tissue due to tensile stress relatively far away from the needle and less failure due to shear close to the needle tip. For hydro gels increased insertion speed produced smaller dimpling and smaller tissue damage as evidenced by the higher friction stress. In brain tissue, friction stress was sensitive to changes in insertion speed, and was smaller for those experiments performed at the fastest speed, 10 mm/s. The differences in friction stress between regions in the brain are not necessarily due to differences in pre stress because the friction coefficient may be different for different regions. However, inside a particular region th e friction coefficient is not expected to change if the speed is kept constant. For that reason, the retractions were all performed all at the same speed (even though the insertions were performed with different speeds). Then the friction coefficient durin g retraction can be assumed constant inside a particular region for all the experiments and the changes in friction stress with insertion speed were expected to be only because of change in pre stress. Therefore, the smaller friction stress observed for hi gh insertion speed can be explained by the low pre stress due to higher tissue damage produced during insertion. During insertion with higher insertion speeds, higher forces and dimpling were produced which indicated greater tissue damage and decreased pre stress. Insertion speed did not affect significantly friction stress at external capsule since it was small for all insertion speeds.
118 In brain tissue, calculated friction stress was not constant along the needle track. It was smaller between 2 and 3 mm d epth where the inner cortex and the external capsule are located. Mechanical properties of different regions in the brain (stiffness and failure strength) affect the insertion mechanics and therefore friction stress and tissue damage. The low friction stre ss around the white matter region can be due to the tensile residual stress that exists in this region as reported by Xu et al.  and/or the smaller stiffness of white matter as reported by Elkin et al. [31 ]. The residual tensile stress would reduce the contact compressive stress between the needle and the tissue in the radial direction (pre stress). In any case, the measurement of friction stress detects the net effect of tissue stress produced by needle insertion plus residual stress which pre exists i n the brain. The shape of the friction stress curves were similar to the shape of pre stress calculated in Chapter 5 using the hole diameters. However, in the pre stress calculated from the hole diameters there was no significant difference between pre str ess at 2 and 10 mm/s probably because effects of swelling and tissue disruption on the hole size are opposite Greater tissue swelling may be generated by the greater injury at 10 mm/s that closed the hole more than at 2 mm/s. In the present study there wa s significant difference between friction stresses at the three insertion speeds because the measurement was done at an early time point where the influence of tissue swelling or other biological reaction is smaller. The phenomenological models of friction stress highlighted insertion speed trends for brain tissue and hydrogel. For brain tissue the insertion speed should be kept as low as practically possible to reduce injury and increase the pre stress. For hydrogel a small increase in insertion speed prod uces increase in friction stress. Therefore high
119 insertion speed is recommended in this material. Asymptotic behavior of friction stress for both materials was predicted. In brain tissue, the low pre stress obtained with high insertion speed explains the g reater backflow obtained for that insertion speed in Chapter 4 In hydrogel, pre stress was greater for high insertion speed which explains the low backflow with high insertion speed obtained previously in this material (Chapter 3) Therefore on both mater ials, under the particular experimental conditions, the most relevant parameter for evaluating backflow was pre stress. To induce backflow, tracer likely had to overcome pre stresses to separate tissue and form a gap at the needle interface 6.5 Summary Increa sed insertion force was produced with increasing insertion speed which produced more damage in the tissue. As a consequence of greater tissue damage, smaller pre stress was obtained with higher insertion speed. Therefore, high insertion speed is not recomm ended on brain tissue. Higher tissue damage and smaller friction stress was found around the white matter region of the external capsule. In general, brain tissue presented an opposite mechanical behavior during insertion with respect to insertion speed in comparison with hydrogel and other biological tissues reported in the literature. On brain tissue and on hydrogel, pre stress was found to be important parameter to produce backflow during CED.
120 6.6 Figures and Tables Table 6 1 Average penetration force ( ), force at the end of insertion ( ), and dimpling ( ) for brain tissue and hydrogel at the three evaluated needle insertion speeds. Data are presented as the average of n = 10 measures standard deviation Brain tissue Hydrog el Speed 0.2 mm/s 2 mm/s 10 mm/s 0.2 mm/s 2 mm/s 10 mm/s (mN) 0.923 0.60 1.607 0.92 1.773 1.01 22.11 2.5 26.77 1.8 23.82 1.2 (mN) 2.29 0.80 3.02 0.43 4.42 0.51 4.51 2.0 3.96 0.85 4.32 1.20 (mm) 0 .654 0.17 0.745 0.23 0.904 0.22 1.15 0.02 1.06 0.02 0.94 0.02 Table 6 2. Average resting force ( ), and dragging force ( ) for brain tissue and hydrogel at the evaluated needle insertion speeds. Data are presented a s the average of n = 10 measures standard deviation Brain tissue Hydrogel Speed 0.2 mm/s 2 mm/s 10 mm/s 0.2 mm/s 2 mm/s 10 mm/s (mN) 1.0530.22 0.6230.02 0.2650.15 1.7570.15 0.7980.12 0.5300.19 (mN) 2.4890.52 2.0 050.33 1.1710.46 1.0170.19 0.6480.13 1.1100.30 Table 6 3. Friction stress (kPa) for varying needle insertion speed averaging along the entire needle track (Global), and for each of the three regions in the brain: caudate putamen (CPu), external ca psule (ec), and cortex. Each value corresponds to the average of 500, 170, 20, and 310 measures for global, CPu, ec, and cortex respectively Speed 0.2 mm/s 2 mm/s 10 mm/s Global 0.585 0.27 0.390 0.17 0.167 0.08 CPu 0.584 0.34 0.408 0.22 0.158 0.14 ec 0.291 0.22 0.259 0.16 0.115 0.13 Cortex 0.356 0.33 0.220 0.24 0.103 0.12 Hydrogel 0.070 0.06 0.142 0.06 0.171 0.04
121 Table 6 4. Model constants for friction stress obtained by fitting phenomenological models to the average friction stress experimental values vs. insertion speed (mm/s) Constants (kPa) (kPa) (s/mm) Global 0.5730 0.1385 0.96 Cpu 0.5818 0.1415 0.98 ec 0.3029 0.0933 0.99 Cortex 0.3400 0.1424 0.92 Hydrogel 0.0555 0.1162 0.6968 0.99 Table 6 5. Constants for phenomenological model of backflow as funct ion of friction stress. Values were obtained by fitting the model to experimental values at three flow rates Flow rate (%) (%/kPa) 0.5 L/min 30.637 36.544 0.95 1 L/min 52.676 57.203 0.99 2 L/min 62.112 66.789 98 Figure 6 1. S chematic representation of forces acting on the needle during the retraction ( = len g th of the needle inside the tissue at time t F = retraction force measured for the load cell, = pre stress, = friction stress )
122 Figure 6 2 Typical fo rce during insertion as a function of time in a rat brain in vivo at 2 mm/s insertion speed ( : penetration force, : force at the end of insertion) Figure 6 3. Average force as function of insertion depth for three differe nt insertion speeds: a) in vivo brain tissue, and b) 0.6% agarose hydrogel
123 Figure 6 4. Typical curve of retraction force as function of time for brain tissue obtained after inserting the needle at 0.2 mm/s. Retraction was performed at 0.2 mm/s. Positi ve values correspond to compressive forces and negative values correspond to tensile forces Figure 6 5. Curves of average force during needle retraction as a function of insertion depth for: a) brain tissue, b) 0.6% agarose hydrogel. Retraction wa s performed at 0.2 mm/s in all the experiments. Resting and dragging forces are indicated on the force profile for a needle inserted at 0.2 mm/s in brain tissue
124 Figure 6 6 Friction stress as a function of insertion dept h for varying insertion speed : b rain tissue (a, b, and c), and 0.6% agarose hydrogel (e, f, and g) Each data point corresponds to the average of n = 10 measures ; d) and h) p value s to evaluate the difference in average friction stress between insertion sp eeds at each position along the needle track (depth) for brain tissue and 0.6% agarose hydrogel respectively (T he line in F igures d and h represents 0.05; n = 10 for each depth position)
125 Figure 6 7. Comparison of experimental average friction str ess values with the phenomenological models: a) average friction stress and the respective model considering the entire brain (global), caudate putamen (CPu), external capsule (ec), and cortex. Only the average friction stress values at each region in the brain along the needle track for each insertion speed were considered; b) average friction stress and the corresponding model for hydrogel Figure 6 8 Average backflow percentage for varying friction stress. Backflow was quantified as the percentage o f infused volume (4 L of Evans blue albumin) leaked outside the target region (caudate putamen)
126 CHAPTER 7 CONCLUSIONS AND FUTURE WORK 7.1 Summary and C onclusions The goal of this study was to understand the influence of needle insertion speed on tissue deformation, tissue damage, and backflow during CED and to evaluate the relationship between backflow and pre stress. To achieve this goal, CED experiments were performed in vivo to evaluate backflow with three different insertion speeds and flow rates. Infusion press ure was measured to evaluate the dependence of insertion rate on tissue coring and its effect on backflow. Hole imaging and measurement in excised tissue slices was implemented as a method to visually evaluate tissue disruption and to estimate pre stress f or varying insertion speeds at two time points. Histology was used to determine the extent of injury for the various insertion speeds. Force measurement techniques were applied during needle insertion and retraction. Dimpling was also measured during inser tion to evaluate tissue deformation. R etraction force measurements allowed evaluat ion of friction stress which is function of pre stress. Detailed aspects hig hlighted in each chapter are summarized below. Chapter 3 presented the evaluation of needle insert ion on backflow and damage in agarose hydrogel. In this phantom material, it was found that fast needle insertion considerably decreased the damage, and backflow. In Chapter 4 the influence of needle insertion speed on backflow was evaluated by performing CED experiments in vivo with three insertion speeds: 0.2, 2, and 10 mm/s. It was found that increased insertion speed produced increasing backflow. Hemorrhage was frequently observed, specially in the white matter of the external capsule.
127 Chapter 5 descri bes i nfusion pressure measurement, hole imaging and histology analysis used to evaluate the tissue damage and to estimate the pre stress for three needle insertion speeds and two time points. No significant difference in peak or steady state pressure was d etected. This means that there was not a rate dependent tissue coring effect, and tissue coring was probably not the major reason for differences in backflow. Hole measurements demonstrated that with high insertion speed, a bigger hole is produced by the n eedle, which means there was less pre stress and backflow was more easily produced. The larger hole diameter in the white matter region of the external capsule was probably due to tensile residual stress present in that region. Time had a significant influ ence on hole diameter between the moment of needle retraction and tissue fixation. The time dependent size of the hole may introduce variation of the calculated pre stress with respect to the moment of the actual CED infusion. The imperfect circular shape of the hole, the tissue fixation which may produce shrinkage of the tissue, and assumption s associated with a simplified neo Hookean model for the brain are some sources of possible error in the pre stress estimation that need to be improved. Histology sh owed that greater tissue fracturing and hemorrhage was produced with high insertion speed. Therefore, high insertion speed, beside greater backflow, produced more severe trauma to the tissue which may be harmful for the brain. No evidence of accumulation a nd dragging of material at the needle tip was observed for brain tissue as was seen for hydrogel ( Chapter 3 ) The white matter region of the external capsule presented the smaller pre stress, and the greater tissue trauma. Moreover, backflow occurred as le akage of fluid to this white matter region. Therefore,
128 the anatomy of the target and the surrounding tissue is important for controlling tissue trauma and backflow. Such factor s may be considered, in choosing the appropriate trajectory of the needle Speci fically, infusions should be preferably performed by inserti ng the needle in a trajectory such that the pre stress is high along the needle track except for close to the tip. If the needle has to pass through a region with tensile pre stress and/or high hy draulic conductivity, the needle tip should be as far as possible of that region. The evaluation of needle insertion and retraction force on brain tissue and 0.6% agarose hydrogel was presented in Chapter 6 On brain tissue, increased insertion speed produ ced increasing insertion force and increasing deformation Higher deformation as measured by dimpling, produced gre ater tissue damage This behavior was opposite to other biological tissues reported in literature where smaller insertion force and tissue da mage was produced with high insertion speed. Retraction force permitted the evaluation of the friction stress which is a function of pre stress and the friction coefficient. This approach allowed evaluation of tissue damage and pre stress at an early time point, in vivo and along the needle track. Friction stress was not constant along the needle track and was mi nimal at the inner cortex and the external capsule. Pre stress was considerably smaller for high insertion due to the greater damage produced with this insertion speed. An opposite behavior of deformation and pre stress with respect to insertion speed was observed in agarose hydrogel which explains the opposite trend of backflow with respect to insertion speed in these materials. T he most useful measure for predicting backflow was the friction stress evaluated with the retraction force measurement. This fri ction stress had a better
129 correlation with backflow, was more sensitive, there were less error sources, and is easier and faster to measure compared to histological hole measures and dimpling E xperimental work with brain tissue has some limitations relat ed to the size of the samples, the softness of brain tissue, the lack of transparency of brain tissue, and the necessity of keep the tissue alive during the entire experiment. For these reasons, it was not possible to use techniques to quantify the stress and strain field in the tissue during the insertion and retraction, to visualize the predominant failure mode of the tissue at the needle tip, and to visualize the flow and spreading for infusate at the needle tip in real time. However, the experimental me thods developed in this study allowed us to answer basic questions and find the influence of insertion speed on tissue damage and backflow and the relation between pre stress and backflow. 7.2 Future W ork The developed experimental work provided useful insight into the effect of insertion speed on backflow and tissue mechanics. With further understanding of insertion mechanics of CED gained from this study, additional studies are needed to consider other important factors such as geometry and material surface p roperties of the needle. In this study only blunt tip needles were evaluated. In view of the fact that backflow is dependent on tissue damage, a rounded or a sharp tip needle may produce less damage  and backflow may be reduced. Also the needle diamete r may change the pre stress and affect backflow. The adhesion forces between the needle and tissue can also have a significant effect on tissue damage and backflow. If the adhesion is high, the tissue may tend to bind to the needle causing considerable def ormation and damage is produced during the insertion. On the other hand, high adhesion may prevent the separation of the tissue from the needle during the injection reducing
130 backflow. Therefore, the adhesive properties of the material of the needle need to be optimized. Also the possibility of using biocompatible lubricant should be considered. Consistent with previous studies , the anatomy of the target and the insertion trajectory are shown to be determinant for tissue damage and backflow. Additional experimental work is needed on other regions of the brain and in larger animals with different brain architecture. The technique to estimate the pre stress using the hole measurement may be improved by using Finite Element Method to expand the hole to the needle diameter. In this way the hole does not need to be assumed perfectly circular. Moreover, the irregular shape of the hole may be used with FEM to obtain the variation of pre stress in tangential direction Also with FEM a more advanced, no t necessari ly incompressible model for the tissue can be used. Also, FEM can be used to better account for tissue heterogeneity and anisotropy along the needle track. The tissue contraction and its variation with the brain region due to fixation may be more precisely evaluated and that evaluation may be used to predict better the size of the hole without fixation and account for a more realistic hole geometry Alternatively, a new technique for imaging the hole without fixation may be developed. With respect to the es timation of pre stress, the friction coefficient needs to be measured to determine the pre stress more precisely. Previous studies have meas ured the fiction coefficient [74 ]. However, they used artificial lubrication on their experiments and that may produ ce significantly different results. Therefore, the friction coefficient needs to be measured between brain tissue and the material of the needle (stainless steel) without lubrication. The measurement of the friction coefficient may be useful not
131 only for t he pre stress estimation, but also to calculate the force necessary for the design of robotic surgical systems and in modeling when tissue cutting or penetration is involved. This study found that residual stress was an important factor in tissue damage. H owever, the analysis of residual stress was based on a study where residual stress were evaluated on mouse brain, in coronal planes and in a region (the thalamus) different from the region used in this study [ 18 ]. Even though the architecture of the thalam us and the region of the CPu evaluated in this study are similar, the residual stress may not be the same in magnitude or distribution. Also, horizontal plane is the most relevant for the needle trajectory of this study Therefore, the residual stresses in rat brain, in the CPu, and in horizontal plane need to be measured. Additional work needs to be done to develop computational models to predict the infusate distribution including backflow. A more realistic model should include the heterogeneity and aniso tropy of tissue, the poroelastic behavior of tissue, the tissue disruption or the pre stress at the needle tissue interface obtained in the present study and the separation of tissue from the needle to produce backflow. In this study, it was found that h igher insertion speed produced great er deformation which produced greater damage. But the reason for the larger deformation as evidenced by the higher dimpling is still unclear. Therefore, mechanical studies at the cellular level should be performed to und erst and why in brain tissue, the rate dependent insertion forces and damage behaves in a manner opposed to some other tissues like cardiac and liver tissues.
132 Cellular mechanics may have an important effect on stresses at the needle tissue interface and ma y influence backflow. Internal pre stress has been reported to exist in cells even if there are no external forces applied to them . If a drug that produces changes in those internal pre stresses is infused in the brain before the actual tracer infusat e, the pre stress at the needle tissue interface may also be changed and that would change the backflow. A preliminary experimental study was performed to test for changes in backflow by injecting blebbistatin before infusing the tracer EBA. Blebbistatin i s a molecule that inhibits myosin II and affects cell mobility  and may produce changes in the tissue properties and stress field. Specific al ly, the hypothesis was that blebistatin may relax the tensile residual stress in the white matter region and th e compressive residual stress in the CPu. In this way an increas e of pre stress in the white matter region and a decreas e of pre stress in the CPu would be produced and backflow would be reduced. Blebbistatin was injected by CED in the rat brain in vivo u sing a dose of 20 mg of blebbistatin per kilogram of weight of the animal . The animal surgery was similar to a normal CED expe riment as described in Section 4 .2.1. Backflow was significantly increased when blebbistatin was pre infused in the brain. E BA reached the surface of the brain during one experiment where the needle was inserted at 10 mm/s. Figure 7 1 shows slices of brain after injection of blebbistatin in comparison with a normal brain. It can be observed that on the rats with blebbistatin th e infused region at the target CPu is considerably smaller specially for 10 mm/s insertion speed. Figure 7 2 shows the backflow percentage for rats infused with blebbistatin in comparison with normal rats results tak en from Chapter 4 Significant
133 increase in backflow was found for those rats pre inf used with blebbistatin. Figure 7 3 shows holes left in tissue by the needle on rats pre infused with blebbistatin in comparison with a hole on a normal rat. Considerably bigger holes can be observed on rats pre i nfused with blebbistatin. The measured hole areas are shown in Figure 7 4 and are compared with hole areas of normal rats obtained at the corresponding insertion speed taken from Chapter 5 It can be observed that with blebbistatin, the area was bigger tha n in normal rats and at some regions the hole was bigger than the needle. The high backflow produced with blebbistatin can be explained by the bigger hole which eliminated pre stress and in some regions produced a gap between the needle and the tissue whe re the infusate can easily flow back and even reach the brain surface. It was expected that blebbistatin would relax residual stresses and decrease the backflow. However the result w as the opposite. D espite the negative result, this preliminary study showe d that it is possible to change backflow by affecting cell mechanics However, side effects of any drug used to try to improve infusate distribution sh o uld be evaluated Other studies have been performed to increase the porosity of brain tissue by injectin g mannitol to improve the infusate distribution Mannito l is thought to dilate the extracellular matrix by driving water from the blood vessels and cells to the extracellular matrix . A similar approach may be evaluated as a methodology to decrease bac kflow.
134 7.3 Figures Figure 7 1. EBA distribution in tissue slices : a) normal rat without blebbistain, needle inserted at 0.2 mm/s; b) rat with blebbistatin, needle inserted at 0.2 mm/s; c) rat with blebbistatin, needle inserted at 10 mm/s. 100 m thick f resh brain tissue slices. Initially 5 L of blebbistatin dissolved in DMSO (dimethyl sulfoxide) was injected at 2 L/min. After 20 min of the blebb istatin injection, 5 L EBA was injected. Two animals were used to measure backflow as the percentage of i nfusate leaked outside the target CPu. Bilateral injections were used to insert the needle at 0.2 mm/s on one side of the brain and at 10 mm/s on the other side. Figure 7 2. Backflow for varying insertion speed for rats infused with blebbistatin (n = 2 rats) compared with normal rats (n = 5 rats ).
135 Figure 7 3. Tissue damage after infusions at 2 L/min: a) hole in the cortex of normal rat, needle inserted at 10 mm/s, b) hole in the cortex of a rat with blebbistatin, needle inserted at 0.2 mm/s, c) hole in the CPu of a rat with blebbistatin, needle inserted at 0.2 mm/s. All figures are at the same scale. 100 thick slices obtained from fixed brain tissue Figure 7 4. Areas of the hole left by the needle after retraction on normal rats and pre infused with blebbistatin a) insertion performed at 0.2 mm/s; b) insertion performed at 10 mm/s. O ne animal was used to measure the hole after retraction. Bilateral injection was performed to insert the needle at 0.2 mm/s on one side and at 10 mm/s on the other side
136 LIST OF REFERENCES  of brain tissue to pp. 1 18.  Reports on Progress in Physics, 64, pp. 815 884.  Ro Enhanced Delivery in the Treatment of 351.  Sampson, J.H., Archer, G., Pedain, C., Wembacher Schroder, E., Westphal, M, possible explanation for the 309.  Kunwar, S. Prados, M., Chang, S.M., Berger, M.S., Lang, F.F., Piepmeier, J.M., tox (IL13 PE38QQR) in the recurrent malignant glioma: a report by the cintredekin pp. 837 844.  d placement for convection Neuroscience Methods, 187, pp. 46 51.  Rosenblunth, K.H., Luz, M., Mohr, E., Mittermeyer, S., Bringas, J., Bankiewicz, dwelling cannula for convectio Journal of Neuroscience Methods, 196, pp. 118 123.  Neeves, K.B., Lo, C.T., Foley, C.P., Saltzman, W.M., Olbricht, W.L., 2006, al of Controlled Release, 111, pp. 252 262.  Fiandaca, M.S., Forsayeth, J.R., Dickinson, P.J., Bankiewicz, K.S., 2008, guided convection enhanced delivery platform in the treatment of e American Society for Experimental Neuro Therapeutics, 5, pp. 123 127.  Yin, D., Valles, F.E., Fiandaca, M.S., Bringas, J.,Gimenez, F., Berger, M.S., guided convection en hanced delivery or therapeutics in human and non human S203.  Morrison, P.F., Chen, M.Y., Chadwick, R.S., Lonser, R.R., Oldfield, E.H., 1999, eter diameter,
137 Regulatory, Integrative and Comparative Physiology, 277, pp. 1218 1229.  R catheters into elastic tis sue: I. Azimuthally symmetric backflow in homogeneous 304.  C Insertion Speed on Backflow for Convection Biomec hanical Engineering, 134, 041006 1.  White, E., Bienemann, A., Malone, J., Megraw, L., Bunnum, C., Wyatt, M., Gill, mechanics in achieving high flow convection enhanced Neuroscience Methods, 199, pp. 87 97.  Bobo, R.H., Laske, D.W., Akbasac, A., Morrison, P.F., Dedrick, R.L., Oldfield, Proc. Natl. Acad. Sci, 91, pp. 2076 2080.  Kim, J.H., Astary, G.W., Nobrega, T.L., Kantorovich, S., Carney, P.R., Mareci, enhanced MRI of Gd albumin delivery to the rat hippocampus in vivo by convection Journal of N euroscience Methods, 209, pp. 62 73.  Prabhu, S.S., Broaddus, W.C., Gillies, G.T., Loudon, W.G., Chen, Z.J., Smith, B., infusion: a model for direct controlled delivery of th Neurology, 50, pp. 367 375.  Biomech Model Mechanobiol, 8, pp. 253 262.  Bjornsson, C.S., Oh, S.J., Al Kofahi, Y.A., Lim, Y.J., Smit h, K.L., Turner, J.N., Neural Engineering, 3, pp. 196 207.  Rousche, P.J., Normann, R.A Engineering, 20, pp. 413 422.  Williams, A.J. Hartings, J.A., Lu, X.M., Rolli, M.L., Tortella, F.C., 2006, istic like brain injury in the rat: differential time courses of Neurotrauma, 23 (12), pp. 1828 1846.
138  s cerebral swelling follow trauma? Observations using an animal model and possible s131.  Dubowitz, D.J., Dyer, E.A.W., Theilmann, R.J., Buxton, R.B., Hopkins, S.R., 252.  acute edema following cerebral hypoxia ischemia in neonatal compared with juvenile rats using magnetic resonance imagi 101 106.  International Journal of Mechanical Sciences, 65, pp. 157 167.  Mahvash, M 343.  dle tissue interaction forces Physics, 34, pp. 665 680. [ 28 ] insertion speed on tissue response and insertion mechan ics of a chronically implanted silicon Engineering, 58 (11), pp. 3250 3259. [ 29 ] model prediction the in vivo insertion b ehavior of micromachined neural Journal of Neural Engineering, 9, 016005. [ 30 ] IEEE Transaction on Robotics and Automation, 19 (5), pp. 864 875. [31 ] Elkin, B.S ., Ilankova, A., Mor Properties of the P orcine Brain: I n Journal of Biomechanical Engineering Transactions of ASME, 133 / 0711009 1. [32 ] Franceschini, G., Bigoni, D., Regi tnig, P., Deform S imilarly to Filled Elastomers and Follows Consolidation T o f the Mechanics and Physics of S olids, 54, pp. 2592 2620.
139 [33 ] Wang, P., Olbricht, Convection Enhanced Drug Delivery: a Annals of Biomedical Engineering, 38 (8), pp. 2512 2519. [34 ] Chen, M.Y., Lonser, R.R., Morrison P.F., Governale, L.S., Oldfield, E.H., 1999, A ffecting Convection E nha nced D elivery to the Striatum: a Systemic Examination of Rate of Infusion, C annula Size, I nfusate C oncentration, and T issue C annula S ealing T ime, Journ al of Neurosurgery, 90, pp. 315 320. [35 ] Raghavan, R., Brady, M., Rodriguez Ponce, M.I., Hartlep, A., Pedain, C., S on Enhanced Delivery of Therapeutics for B rain Neurosugery Focus, 20 (3), pp. 1 13. [36 ] Ivanchenko, O., Ivanch Free C 061003 1. [3 7 ] Bas u ring Infusion into Micro vascular R esearch 4, pp. 143 165. [3 8 ] Chen, Z., Broaddus, W.C., Viswanathan, R.R., Raghavan, R., Guillie s, G., 2002, Drug Delivery via Positive P ressure Infusion: Experimental and M odeling S tudies of P oroel IEEE Transactions on Biomedical Engineering, 49 (2), pp. 85 96. [ 39 ] Sarntinoranont, M., Banerjee, R.K., Lonser, R.R., Morrison, P. Computational Model of Direct I nterstitial Infusion of M acro molecules into the Ann Biomed Eng 31, pp. 448 461. [ 40 ] Smith, J.H., Humphre Fluid Exchange During Infusion into B rain Microvascular Research, 73, pp. 58 73. [4 1 ] Chen, X., Sarnt T ransport for D irec Annals of Biomedical Engineering, 35 (12), pp. 214 5 2158. [4 2 ] Sarntinoranont, M., Banergee, R.K., Lonser, Computational Model of Direct Interstitial Infusion of M acro molecules into the Annals of Biomedical Engineering, 31, pp. 448 461. [4 3 ] Sarntinoranont. Model of I nterstitial Transpor t in the Spinal C ord Annals of Biomedical Engineering, 34 (8), pp. 1304 1321. [4 4 ] Linninger, A.A., Somayahi, M.R., Erickson, T., G ou, X., Pe nn, R.D., 2008, ransport in A nisotropic and Journal of Biomechanics, 41, pp. 2176 2187.
140 [4 5 ] Kim, J.H., Mareci, T.H. t Infusion into the Corpus Callosum and Hippocampus of the Rat Brain: Model D ev Med. Biol. Eng. Comput 48, 203 214. [4 6 ] Astary, G.W., Kantorovich, S, Carney, P.R., Mareci, T.H., Sarntinoranont, M., vection Enhanced Delivery of Gadolinium Labeled A lbumin in the Rat Hippocampus in Vivo Journal of Neuroscience Methods, 187, 129 137. [4 7 ] Smith, J.H., Garcia, J.J., 20 Controlled Infusion in Brain: Fluid Transpor t and Tissue Deformation A naly Journal of B iomechanics, 42, 2017 2025. [4 8 ] Chen Z.J., Gillies, G.T., Broaddus, W.C., Prabhu, S.S., Filmore, H., Mithcell, R.M., Corwin, rain Tissue Phantom for Intraparenchym al Infusion S 101, pp. 314 332. [ 49 ] Sharp, A.A., Ortega, A.M., Restrepo, D., Curran Ev In Vivo P enetration Mechanics and Mechanical Properties of Mouse Brain Tissue at Micrometer S ions on Biomedical Engineering, 56, pp. 45 53. [ 50 ] Gefen, A., in Vivo and in Situ B rain Tissues Journal of Biomechanics, 37, pp. 1339 1352. [ 51 ] Subbaroyan, J., Mati ite Element Model of the Mechanical Effects of Implantable M icroele Journal of Neural Engineering, 2, pp. 103 113, [5 2 ] Forsell, C., V entricular Tissue Due to D ee p Penetration: the I mp act of Constitutive Journal of Biomechanics, 44, pp. 45 51. [5 3 ] Metz, H., McElhan omparison of the E lasticity of L ive Journal of Biomechanics, 3, pp. 453 4 58. [5 4 ] Hrapko, M., van Dommelen, J.A.W., Peters, G.M.W., Wismans, J.S.H.M., 2008, echani cal J of Biomechanical Engineering, 130/031003 1.  Prange, M.R., Ma dependent Engineering, 124, pp. 244 252.
141 [5 6 ] Libertiaux, V., Pascon F., Ces erification o f Brain T issue Incompressibility U sing D igital Image C orrelation Mechanical Behavior of Biomedical M aterials, 4, pp. 1177 1185. [5 7 ] M Mechanical P ropert Journal of Biomechani cs, 35, pp. 483 490. [5 8 ] Lee, S.J., Sun, J., Flint, J.J., Gou, S., Xie, H.K., King, M.A., Sarntinoranont, M., B ased Indentation Technique for A cute Rat Brain Tissue Slices and Thin B Part B: Applied Biomaterials, 97B, pp. 84 95. [ 5 9 ] van Dommelen, J.A.W., van der Sande, T.P.J., Hrapko, M., Pet ers, G.W.M., roperties of B rain Tissue by I ndent ation: Interregional Journal of the Mechanical Behavior of Biom edical Materials, 3, pp. 158 166. [ 60 ] Prange, M.T., Marguli D ependent P roperties of the B rai Journal of Biomechanical Engineering Transactions of ASME, 124, pp. 244 252. [6 1 ] Journal of Biomechanics, 40, pp. 117 124. [6 2 ] Prevost, T.P., Balakrishnan, A., Suresh, S., Socra Brain T 95. [6 3 ] Fung, Y.C., Biomechanics. Mechanical Properties of Living T issues. Springer Verlag. 1993. [6 4 ] Cervos Navarro, J. &La C S14. [6 5 ] Unte rberg, A.W., Sover, J., Kress, B., K T 1029. [6 6 ] Williams, A.J., Wei, H.H., Dave, J.R., To Neuroinflammatory Response Following Experimental Pen etration Ballistic Brain Injury in the R Journal of Neuroinflammation, 4 ( 17 ) pp. 2 12. [6 7 ] Groothuis, D.R., Ward, S., Schalageter, K.E., Itskovich, A., Schwerin, S.C., Allen, C.V., Dills, C., Brain Barrier Perme ability Associated with Insertion of B rain C annulas and Microdialysis P Research, 803, pp. 218 230.
142 [68 ] Mas e, G.T., Mase, G.E., Continuum Mechanics for E ngineers S econd E dition, CRC Press, New York, 1999. [69 ] Paxinos, G., Watson, C., 19 Academic Press, San Diego, CA. [ 70 ] M Historical Perspectives and Relevance in Near Infrared M onitoring of O pti International Jour nal of Industrial Ergonomics, 40, pp. 125 134. [7 1 ] Sezigin, 146 165. [7 2 ] Quester R., Schroder, R., rain Stem D uring F ormalin F ixa Journal of Neuroscience Methods, 75, pp. 81 89. [7 3 ] sweelling in cerebral embol Surgical Neurology, 41, pp. 477 481.  Behavior of Biomedical Materials, 14, pp. 163 171.  Prentice hall [76 ] agarose gels. I. Rheo 184 195. [77 ] composite of adhering junctions connecting heart muscle cells of vertegrates. I. Molecualr definition in in tercalated disks of cardiomyocytes by immunoelectron 69 82.  ling in Mechanobiology, 1, pp. 95 108.  Laboratory Investigation, 90, pp. 448 458.
143  Neeves, K.B., Sawyer, J.A., Foley, C.P., Saltzman, W.M., Olbricht, W.L., 132.
144 es-CO BIOGRAPHICAL SKETCH es-CO Gonzalo Fernando Casanova Garcia was born in Pupiales, Colombia. He obtained a Bachelor of Science in mechanical engineering from Unversidad del Valle en 200 3 and a Master of Science in mechanical engineering from the same university in 2006. In 2007 he st art working as auxiliary professor at Universidad del Valle and in 2009 he earned a Fulbright Colciencias DNP grant and joined Dr. Malisa He received his Ph.D. from Universit y of Florida in the sum mer of 2013. His research interests are in drug distribution in brain tissue, as well as, in vivo penetration mechanics.