UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Intradermal injections through hollow microneedles Shrestha, Pranav 2018

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

Item Metadata

Download

Media
24-ubc_2018_may_shrestha_pranav.pdf [ 194.98MB ]
Metadata
JSON: 24-1.0365782.json
JSON-LD: 24-1.0365782-ld.json
RDF/XML (Pretty): 24-1.0365782-rdf.xml
RDF/JSON: 24-1.0365782-rdf.json
Turtle: 24-1.0365782-turtle.txt
N-Triples: 24-1.0365782-rdf-ntriples.txt
Original Record: 24-1.0365782-source.json
Full Text
24-1.0365782-fulltext.txt
Citation
24-1.0365782.ris

Full Text

INTRADERMAL INJECTIONS THROUGH HOLLOW MICRONEEDLES by  Pranav Shrestha  B.Eng., University of Victoria, 2015  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF APPLIED SCIENCE in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES  (Mechanical Engineering)   THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  April 2018  © Pranav Shrestha, 2018   ii  The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, a thesis/dissertation entitled:  Intradermal injections through hollow microneedles  submitted by Pranav Shrestha in partial fulfillment of the requirements for the degree of Master of Applied Science in Mechanical Engineering  Examining Committee: Boris Stoeber, Mechanical Engineering Supervisor  Dana Grecov, Mechanical Engineering Supervisory Committee Member  Gary Schajer, Mechanical Engineering Supervisory Committee Member  Additional Examiner   Additional Supervisory Committee Members:  Supervisory Committee Member  Supervisory Committee Member iii  Abstract  Hollow microneedles are a promising alternative to conventional drug delivery techniques such as oral drug administration and hypodermic injections, and are used for delivering drugs and therapeutics into the skin. Although the benefits of intradermal drug delivery have been known for decades, our understanding of fluid absorption by skin tissue has been limited due to the difficulties in imaging a highly scattering biological material such as skin. In this thesis, we report the results from ex-vivo injection experiments into excised porcine skin tissue using hollow microneedles. We introduce the use of optical coherence tomography (OCT) for real-time imaging of skin tissue at the micro-scale during intradermal injections through hollow microneedles. We identify two modes of flow into the skin – microinjection, a region of high transient flow-rate, and microinfusion, a region of lower steady-state flow-rate. We relate the two modes of flow to tissue deformation. Using images from the OCT, we find that the skin tissue behaves like a deformable porous medium and absorbs fluid by locally expanding rather than rupturing to form a fluid filled cavity. We measure the strain distribution in a cross section of the tissue to quantify local tissue deformation using digital image correlation (DIC), and find that the amount of volumetric expansion of the tissue corresponds closely to the volume of fluid injected. Mechanically restricting the tissue expansion limits fluid absorption into the tissue, and allowing the tissue to expand leads to increased fluid absorption. Our experimental findings can provide physical insights for optimizing the delivery of drugs into the skin for different therapeutic applications, and for better modelling fluid flow into biological tissue.   iv  Lay Summary  Drug delivery is essential for improving human health: by preventing life-threatening diseases (as with vaccine delivery), or by treating abnormal conditions (as with insulin delivery for diabetic patients). The skin has been considered as a potential site for drug delivery, partly due to its strong immune response. Although useful, injecting into the skin using the conventional hypodermic needles is challenging and requires skilled personnel. Hollow microneedles, a potential alternative, are small needle-like devices that can access the skin more reliably for effective drug delivery. They also improve patient compliance and are potentially self-administrable. In this thesis, we study injections into the skin using microneedles and observe how the injected fluid interacts with the skin tissue. We find that the amount of fluid injected into the skin relates closely to the amount of expansion of the skin tissue. This new observation can help optimize drug delivery into the skin using microneedles.   v  Preface  A majority of the work mentioned in this thesis has been submitted for publication: P. Shrestha, and B. Stoeber, “Fluid absorption by skin tissue during intradermal injections through hollow microneedles,” Apr. 2018 (submitted). I performed all the experiments and analyses, and wrote most of the manuscript.   A portion of the results from Chapter 3 and Chapter 4 were presented at a conference: P. Shrestha, B. Stoeber, “Tissue expansion and fluid absorption by skin tissue following intradermal injections through hollow microneedles”, presented at 70th Annual Meeting of the APS Division of Fluid Dynamics, Denver, Colorado, Vol. 62, No. 14, 21 Nov 2017.  The single hollow microneedles used for the experiments were fabricated by Microdermics Inc. The PDMS moulding described in section 2.2 was performed by Hongbae Sam Park.   vi  Table of Contents  Abstract ....................................................................................................................................iii Lay Summary ........................................................................................................................... iv Preface ....................................................................................................................................... v Table of Contents ..................................................................................................................... vi List of Tables ............................................................................................................................. x List of Figures .......................................................................................................................... xi List of Symbols ....................................................................................................................... xix List of Abbreviations ............................................................................................................. xxi Acknowledgements ............................................................................................................... xxii Dedication............................................................................................................................. xxiii Chapter 1: Introduction............................................................................................................ 1 1.1 Skin and intradermal injections .................................................................................... 1 1.1.1 Primary layers of the skin..................................................................................... 1 1.1.2 Immune response of the skin ................................................................................ 3 1.1.3 Intradermal injections .......................................................................................... 4 1.2 Microneedles ............................................................................................................... 7 1.3 Previous injection experiments using hollow microneedles .......................................... 9 1.4 Visualization of injections ......................................................................................... 10 1.5 Biological tissue as a deformable porous medium ...................................................... 12 1.6 Objectives ................................................................................................................. 16 Chapter 2: Materials and methods ......................................................................................... 17 vii  2.1 Intradermal injections using hollow microneedles ...................................................... 17 2.2 Experimental setup .................................................................................................... 19 2.3 Skin sample preparation............................................................................................. 24 2.4 Injection procedure .................................................................................................... 25 2.5 OCT image acquisition .............................................................................................. 26 2.5.1 Skin refractive index estimation ......................................................................... 28 2.5.2 Additional capabilities of OCT .......................................................................... 29 2.5.3 Other imaging modalities ................................................................................... 29 2.6 Speckle reduction in OCT images .............................................................................. 30 2.7 Sample OCT image processing .................................................................................. 33 2.8 Digital image correlation ........................................................................................... 34 2.8.1 Time difference between DIC results ................................................................. 35 2.8.2 Calculating 2D strain ......................................................................................... 38 2.9 Volume of tissue expansion ....................................................................................... 38 2.9.1 Surface deformation technique ........................................................................... 40 2.9.2 3D Volumetric strain technique .......................................................................... 41 2.9.3 Technique validation – DIC and volume of expansion ....................................... 43 2.10 Correlation maps ....................................................................................................... 46 Chapter 3: Experimental Results ........................................................................................... 48 3.1 Microneedle insertion ................................................................................................ 48 3.2 Flow-rate and pressure during injections .................................................................... 50 3.2.1.1 Flow-rate offset correction ............................................................................. 51 3.3 Microinjection and microinfusion .............................................................................. 53 viii  3.4 OCT imaging of tissue microstructure ....................................................................... 56 3.4.1 OCT images pre-retraction ................................................................................. 56 3.4.2 OCT images post-retraction ............................................................................... 57 3.4.3 3D OCT images ................................................................................................. 61 Chapter 4: Skin tissue deformation ........................................................................................ 63 4.1 2D strain fields .......................................................................................................... 63 4.1.1 Sample DIC data ................................................................................................ 63 4.1.2 2D strain fields during intradermal injections ..................................................... 67 4.2 Volume of tissue expansion ....................................................................................... 69 4.3 Correlation maps ....................................................................................................... 70 4.4 Multiple/continuous retractions.................................................................................. 74 4.4.1 Multiple retractions ............................................................................................ 74 4.4.2 Continuous retraction ......................................................................................... 78 Chapter 5: Conclusions and future work ............................................................................... 83 5.1 Summary of observations .......................................................................................... 83 5.2 Conclusions ............................................................................................................... 85 5.3 Applications of research findings ............................................................................... 86 5.4 Limitations of experimental techniques ...................................................................... 86 5.5 Suggested future work ............................................................................................... 87 References ............................................................................................................................... 88 Appendices .............................................................................................................................. 93 Appendix A OCT scan parameters ........................................................................................ 93 Appendix B Imaging techniques and experimental setup ....................................................... 95 ix  B.1 Micro-CT .............................................................................................................. 95 B.2 Histology ............................................................................................................... 96 B.3 Injection system details .......................................................................................... 96 B.4 Sensors and tubing ................................................................................................. 97 Appendix C LABVIEW ........................................................................................................ 98 Appendix D Wavelet transform based filtering .................................................................... 100 D.1 Wavelet-2D decomposition .................................................................................. 100 D.2 Thresholding detail coefficients ........................................................................... 101 Appendix E ......................................................................................................................... 102 E.1 Example of Ncorr window ................................................................................... 102 E.2 Ncorr strain .......................................................................................................... 103  x  List of Tables  Table 2.1: Sensors and actuators used in the experimental setup and the corresponding parameters controlled or measured. ........................................................................................... 24 Table 2.2: Comparison of theoretical and calculated changes in area and volume of test image. 45  xi  List of Figures  Figure 1.1: Different layers of the skin - epidermis, dermis and hypodermis [derivative of “Skin layers” (c) Madhero88 (http://commons.wikimedia.org/wiki/File:Skin_layers.png); used under a Creative Commons Attribution-Share Alike 3.0 Unported License] ............................................ 2 Figure 1.2: Needle angles for different types of injections, highlighting the difficulty in performing a near parallel intradermal injection (10˚ - 15˚ insertion angle). [“needle-insertion-angles-1” (c) British Columbia Institute of Technology (http://opentextbc.ca/clinicalskills/wp-content/uploads/sites/82/2015/09/needle-insertion-angles-1.png); used under a Creative Commons Attribution 4.0 International License] ......................................................................... 6 Figure 1.3: Representative elementary volume for a porous medium such as a biological tissue 12 Figure 2.1: Cross-sectional schematic of skin and microneedle showing 5 steps of a successful intradermal injection: i) microneedle impacting the stretched skin (with velocity vim) to penetrate stratum corneum; ii) microneedle insertion; iii) application of input pressure (Pin) with no initial flow due to high dermal resistance; iv) retraction of microneedle (by dr) ; v) successful injection of fluid with the formation of a raised papule (characteristic of intradermal injections). ............ 18 Figure 2.2: Simplified schematic of the experimental setup (top; not to scale) showing the major components and their interactions, with variables indicating injection parameters controlled and/or measured; and picture (bottom) showing the different components of the experimental setup. ........................................................................................................................................ 20 Figure 2.3: Picture of the injection system (dashed grey box in Figure 2.2 schematic), showing the impact system, retraction system, tissue holder and OCT. Detail (blue box) showing xii  microneedle and excised porcine skin sample (stretched) in the tissue holder. Scale bars (white and black) are 10 mm. ............................................................................................................... 21 Figure 2.4: Steps for stretching the excised skin sample: First the skin sample is clamped firmly, and then stretched uniformly in radial directions by the PDMS backing. ................................... 22 Figure 2.5: Cross-sectional schematic (mid-section) showing the microneedle inserted into the stretched skin sample (top) with OCT imaging the skin and the microneedle through the transparent PDMS backing. OCT image (bottom) showing a cross-section of skin with the microneedle inserted. The scale bar is 0.5 mm........................................................................... 27 Figure 2.6: Setup for measuring the refractive index of skin (left); Corresponding OCT image (right) around the edge of the skin,  showing an apparent increase in height of skin in the image. ................................................................................................................................................. 28 Figure 2.7: Additional capabilities of the OCT, a: 3D image of the skin surface after injection, showing hole created by the microneedle and the swelling of the skin around the microneedle imprint. b: one cross-section of the 3D image, showing the microneedle imprint and the swelling of skin around the MN. c: OCT image of a single microneedle, which can be used to determine the height of the MN. The scale bars are 0.5 mm. ...................................................................... 30 Figure 2.8: Image processing of a sample OCT image, showing the different types of techniques used. In the thesis, OCT images are presented in the forms IIb, IIIb, or IV. The scale bars are 0.5 mm. .......................................................................................................................................... 34 Figure 2.9: Schematic for calculating the cumulative displacement and strain fields for InjA using DIC results from 9 pairs of wavelet denoised OCT images (Δt = 58 ms), with a time interval of 0.52 s between the first reference  (WD1) and last denoised OCT image (WD10). ... 36 xiii  Figure 2.10: Schematic for calculating the cumulative displacement and strain fields for InjB using DIC results from 8 pairs of averaged OCT images (Δt = 13.5 ms), with a time interval of 108 ms between the first reference, (a+b)/2, and last averaged OCT image, (i+j)/2. ................... 37 Figure 2.11: Two cases - before (A) and during (B) injection of water into skin – with different porosity and permeability. ......................................................................................................... 39 Figure 2.12: Left: OCT image before injection, showing the un-deformed state of the skin that is used as a reference. The surface of the skin is shown in white dashed lines and the microneedle is shown in yellow dotted lines. Right: OCT image during injection, showing the new position of the skin surface (blue dashed line), and the original un-deformed position (white dashed line). The area between the blue and white dashed lines is the added area, which is used for calculating the volume assuming axisymmetric conditions around the central axis of the microneedle (red dashed line). The scale bar (black) is 0.5 mm. ........................................................................... 40 Figure 2.13: Images used for validating volume expansion calculations, showing original patterned circle (left), expanded to increase radius from 300 pixels to 325 pixels (middle), and rotated by 10 degrees (right)...................................................................................................... 44 Figure 2.14: Strain fields Exx, Exz and Ezz for the expanded and rotated test sample. The scale bar is 600 pixels, which is the size of the diameter of the circle. ...................................................... 44 Figure 2.15: Example of a correlation map for InjA (5 s after MN retraction), showing the deforming tissue in dark pixels and the stationary tissue in bright pixels; Pin = 100 kPa, the injected fluid is water. The scale bar is 0.5 mm. ........................................................................ 46 Figure 3.1: Setup before insertion (left) with the spring compressed; and after insertion (right) with the spring relaxed and the retraction arms support the microneedle sub-assembly. The scale bars are 10 mm.......................................................................................................................... 48 xiv  Figure 3.2: OCT images before (top) and after (bottom) insertion of the microneedle into porcine skin sample. The scale bars are 0.5 mm. .................................................................................... 49 Figure 3.3: Time evolution of input pressure (top – black curve) and measured flow-rate (bottom – blue curve) for a typical injection; the microneedle was retracted by 0.3 mm at 0 s indicated by the dashed red line. ................................................................................................................... 51 Figure 3.4: MN injection into PDMS. a: Setup showing MN inserted into PDMS (scale bar 10 mm). b, Typical test cycle with pressures alternating between 0 kPa and 100 kPa, and the corresponding flow-rates. c, Plots showing step increase and decrease in pressure, and the corresponding offsets Q0 (0 kPa) and Q0 (100 kPa). .................................................................. 52 Figure 3.5: Two modes of flow - microinjection and microinfusion. Top: Flow-rate of water into excised porcine skin sample, showing the mean (solid blue line) and the standard deviation (blue shading) for five experiments. Water is initially not injected into the skin, even with an input pressure of 100 kPa (light green bar); the microneedle is retracted by 0.3 mm (magenta bar); and water flows into the skin (dark green bar; t = 0 s marking the onset of fluid flow during retraction). When the MN is retracted, the flow-rate increases rapidly (microinjection; inset after 0 s) and decays to a non-zero steady state value (micro-infusion; after approximately 10 s). Bottom: First 20 s of the 5 individual flow-rate measurements. ................................................. 54 Figure 3.6: Volume of water injected into the skin, calculated by integrating the flow-rate over time, shows rapid fluid absorption by tissue during microinjection (inset) followed by slower fluid absorption during microinfusion. Light green bar represents application of 100 kPa input pressure with no flow, magenta bar represents microneedle retraction by 0.3 mm, and dark green bar represents absorption of fluid by the skin. ........................................................................... 55 xv  Figure 3.7: Picture and cross-sectional schematic of the microneedle inserted into the skin (top). OCT images of the skin for InjA (middle; using image processing IIIb – wavelet transform based filter and brightness adjustment) and InjB (bottom; using image processing IV - averaging), showing no movement of skin with Pin = 100 kPa. The time t=0 s is when the microneedle is retracted by 0.3 mm. Input pressure was set to 100 kPa a few seconds before retraction. In both sets of images, the input pressure is set to 100 kPa, but the skin tissue does not deform in response to the applied input pressure before retraction. The scale bars are 0.5 mm. .................. 57 Figure 3.8: Sequence of OCT images (InjA, IIIb – top 2 rows; captured at 17.2 frames/s) showing the microneedle (yellow dotted lines) retracting by 0.3 mm at 0 s, and showing the deformation of skin tissue after 0 s. The surface of the skin, the stratum corneum (white dashed lines), moves upwards towards the base plate of the microneedle and then moves laterally outwards. OCT images (InjB, IV – bottom) captured at a higher frame rate (73.9 frames/s) than InjA, showing skin tissue expansion during initial injection after retracting the microneedle by 0.3 mm at 0 s. Scale bars (black) in a, b, and c are 0.5 mm. ....................................................... 58 Figure 3.9: Sequence of OCT images (InjC, IIb – top 2 rows) showing initial injection of water accompanied by rapid expansion of tissue (microneedle retraction of 0.3 mm at 0 s), followed by gradual expansion. OCT images (InjD, IIb – bottom) showing a gap created when the microneedle is retracted by 0.3 mm at 0 ms. The gap does not form a fluid-filled cavity that grows during the injection. The scale bars (white) are 0.5 mm................................................... 59 Figure 3.10: Injection of viscous fluid (65 ˚Bx) into the skin, with 0.2 mm retractions (each) at 0 s, 6.85 s, and 21.0 s. The scale bar (white) is 0.5 mm. ............................................................... 60 Figure 3.11: 3D imaging of the skin after injection using OCT. The top panels show XZ and YZ planes, showing the hole created by the microneedle (along the vertical yellow line). Horizontal xvi  lines (1 red, 2 yellow and 3 blue) indicate locations for the XY plane sections, shown in the bottom panels. Scale bars: 0.5 mm. ........................................................................................... 62 Figure 4.1: Displacement and strain fields for InjB (cumulative results for a time interval of 108 ms, considering 8 pairs of averaged OCT images). OCT image (top) shows ROI in orange dashed lines, MN in yellow dotted lines, and SC in white dashed lines. The displacement fields ux and uz, and strain fields Exx, Exz, and Ezz are calculated using DIC in Ncorr. The 2D strain fields ɛ2D are calculated in MATLAB using equation (13).  The scale bars are 0.5 mm. ............. 66 Figure 4.2: 2D maps of strain (ɛ2D) for the region bounded by orange dashed lines in the OCT image (left). Regions of local tissue expansion (positive strain) are shown in red, while regions of local tissue compression (negative strain) are shown in blue. OCT images captured at 73.9 frames/s; microneedle retracted at 0 s by 0.3 mm; Pin= 100 kPa; injected fluid is water. The scale bars (black) are 0.5 mm. ................................................................................................... 67 Figure 4.3: Strain maps (ɛ2D) overlaid (50% transparency) on OCT images, considering a region of interest for DIC as the region of the skin before retraction. Each strain map shows cumulative strain using DIC results on 9 pairs of consecutive OCT images, with a total time difference between the reference OCT frame and the last OCT frame of 0.52 s. The ‘V’ shaped region of high tissue expansion grows laterally outwards. The rate of tissue expansion decreases over time (for instance, limits of strain at 0.6 s are an order of magnitude larger than those at 10.6 s). OCT images captured at 17.2 frames/s; Pin = 100 kPa and dr = 0.3 mm. The scale bar (black) is 0.5 mm. .......................................................................................................................................... 68 Figure 4.4: The time evolution of total volume of tissue expansion derived from 3D volumetric strain (orange curve; OCT-DIC) and from surface deformation technique (red curve; OCT-SD) have similar profiles to that of total injected volume (purple curve; Q Sensor). ......................... 70 xvii  Figure 4.5: Correlation maps for InjA; Pin = 100 kPa; dr = 0.3 mm. Bottom right: Evolution of the lower boundary between expanded and stationary tissue tracked over time, showing growth of the expanded region. Successive dashed (colored) lines are 2.9 s apart, with the first (red) dashed line starting 2s after microneedle retraction; the microneedle position is shown by dotted gray lines. The black scale bars are 0.5 mm. .............................................................................. 71 Figure 4.6: Correlation maps for 5 sets of injection experiments (at 1 s and 5 s), showing ‘V’ shaped bands of low correlation (dark pixels) indicating movement of tissue. Bright pixels indicate stationary tissue. Pin = 100 kPa; dr = 0.3 mm (InjA – InjD) and 0.2 mm (InjE); Injected fluid = water (Inj A – InjD), and 65 ˚Bx sugar solution (InjE). The black scale bars are 0.5 mm. ................................................................................................................................................. 73 Figure 4.7: Top: Flow-rate after retraction of 0.3 mm at 0 s, followed by 8 retractions of 0.1 mm (spaced 20 s apart) starting at 300 s. After each retraction, the flow-rate increases rapidly (microinjection) and then decays to a value greater than that of the previous retraction; Pin = 100 kPa. Bottom: The volume of fluid injected is higher for 8 successive retractions of 0.1 mm than for a single retraction of 0.3 mm (calculated by integrating the curve in b from 300 s to 600 s, and from 0 s to 300 s, respectively). .......................................................................................... 75 Figure 4.8: Flow-rate after multiple retractions of 0.1 mm each for 3 different sets of injections. ................................................................................................................................................. 76 Figure 4.9: OCT images for InjG, InjH and InjI considered for the multiple retraction plot. InjG has a deeper insertion depth than InjH and InjI; InjH is around 1.2 mm thick, while InjG and InjI are around 1.1 mm thick; the dermis of InjI was not completely bonded, allowing it to move during injection. The black scale bars are 0.5 mm. .................................................................... 77 xviii  Figure 4.10: Correlation maps and pictures of the microneedle and tissue holder for InjF. The microneedle was retracted slowly by 1 mm; Pin = 100 kPa; Injected fluid: water + green dye. The correlation maps show regions of deforming tissue (dark pixels) and stationary tissue (bright pixels). The stratum corneum gradually swells as the microneedle is retracted back continuously, while allowing tissue expansion. The scale bars in the correlation maps are 0.5 mm.................. 80 Figure 4.11: Flow-rate and retraction distance for InjF, showing a continuous retraction of up to 1 mm at a rate of 0.1 mm/s and an increase in flow-rate during the retraction. The spike in flowrate at around 1.5 s is likely due to the injection of an air bubble at the tip of the microneedle, as shown later. Pin  = 100 kPa; Injected fluid = water + green dye. ....................... 81 Figure 4.12: OCT images of InjG and InjJ showing the rapid injection of air trapped in the microneedle at the beginning of the injection for InjG and in the middle of an injection for InjJ. The input pressure is 100 kPa for both injections, the injection fluid is water for InjG and 65 ˚Bx sugar solution for InjJ. For InjJ, the injection of air results from an air bubble trapped in the system. Note: the actual orientation of InjJ was upside down, with the microneedle and injection system upright – which could have resulted in the air bubble moving up during the injection. ... 82  xix  List of Symbols  Symbol Description Units 𝜙%  Porosity (relative porosity of fluid phase) Dimensionless 𝜙& Relative porosity of solid phase Dimensionless 𝜇 Dynamic viscosity of the fluid 𝑁. 𝑠/𝑚- 𝜆 Stretch ratio or stretch Dimensionless ∆𝜃 Increment of angle Radians 𝑑2 Depth of microneedle insertion 𝑚 𝑑3 Microneedle retraction distance 𝑚 𝐸55 Normal strain in x direction Dimensionless 𝐸56 Shear strain  Dimensionless 𝐸66 Normal strain in z direction Dimensionless 𝜀-8  2D strain Dimensionless 𝜀9 3D volumetric strain Dimensionless 𝐾 Permeability 𝑚- 𝜅 Permeability (for biological applications, including µ) 𝑚</𝑁/𝑠 𝑚 Material constant (permeability-porosity equation) Dimensionless 𝑛&>2? Refractive index of skin Dimensionless 𝑝 Fluid pressure 𝑃𝑎 𝑃2? Fluid input pressure 𝑃𝑎 𝑃C Recorded fluid pressure 𝑃𝑎 xx  𝑄 Recorded fluid flowrate 𝑚E/𝑠 𝑡3 Time when retraction (by 𝑑3) completes  𝑠 ∆𝑡 Time interval between OCT images 𝑠 𝑢 Displacement of solid 𝑚 𝑢5 Displacement field in the x- direction 𝑚 𝑢6 Displacement field in the z- direction 𝑚 𝑣 Velocity of fluid 𝑚/𝑠 𝑣2C Microneedle impact velocity 𝑚/𝑠 𝑣5 Velocity of fluid in the x-direction 𝑚/𝑠 𝑉 Total volume of porous medium 𝑚E 𝑉% Volume occupied by fluid phase (void volume) 𝑚E 𝑉& Volume occupied by solid phase  𝑚E  xxi  List of Abbreviations  2D Two-dimensional 3D Three-dimensional DIC Digital image correlation MN Microneedle OCT Optical coherence tomography PDMS Polydimethylsiloxane ROI Region of interest SC Stratum corneum  xxii  Acknowledgements Firstly, I would like to thank Prof. Boris Stoeber for introducing me to this meaningful project and giving me the wonderful opportunity and creative freedom to explore the topic deeply. You have been the guiding light in this challenging, yet rewarding, journey. I am really grateful for your support, patience, lessons and sense of humor! You have truly made my graduate experience enjoyable and enlightening.  I would like to express my gratitude towards Prof. Dana Grecov and Prof. Gary Schajer for taking out their time to be a part of my examining committee. I appreciate your feedback, attention and cooperation.  I had the immense joy of meeting so many fantastic people while at UBC, and I thank you all for such amazing memories. I thoroughly enjoyed my time with the folks from the lab including Sam, Farzad, Hatef, Bryan, Crystal, Jorge, Hamed, Anika, Mazi, Sahan, Arian, Ramin, Mani, and Christoph. I met some amazing leaders at MEGA including Amin, Hoda, Ali, Pooyan and Parisa. It was great to share the lab space in PPC with Masoud, Ali, Justin, with George always there to help. I am also grateful for the wonderful times at Microdermics with Iman, Mehrsa, Sahan and Kimberley.  Nothing would have been the same without the constant love and support from my family. Daddy, you have been the greatest inspiration and role model; Mummy, you have so meticulously constructed the intellectual foundation on which we stand and grow; Apu dijju, you give us hope in our darkest times to rise above our troubles; Aki dijju, you always give unparalleled and much-needed advice; Bhai, you are the source of my curiosity and are my greatest teacher; Baba, your lessons are immensely valuable; Mama, your composure is contagious; and most importantly dear Kripa, I couln’t have done this without you.    xxiii  Dedication  To my family.  1  Chapter 1: Introduction Microneedles provide a promising alternative to conventional drug delivery techniques such as oral drug administration and hypodermic injections. Microneedles deliver drugs into the skin, which can be modelled as a deformable porous medium. This chapter provides the background required for the following chapters of the thesis, a literature review on previous experiments and theoretical modelling on injections into biological tissue, and a brief overview of the objectives for the thesis.   1.1 Skin and intradermal injections The skin is a potential site for effective drug delivery using microneedles. This section briefly explains the anatomy and functions of the skin, the advantages of injecting drugs into the skin, and the challenges associated with the conventional injection technique.   1.1.1 Primary layers of the skin Skin is the largest organ of the human body. It protects the body from mechanical stresses and external agents, it regulates body temperature, and detects/relays information from the environment about pain, touch and temperature [1].  The three primary layers of the skin [2], from the outside to the inside are: epidermis, dermis and hypodermis (Figure 1.1).   The outermost layer of the skin, the epidermis, is a physical barrier preventing chemicals and micro-organisms from entering the body, and preventing excess water loss from the body. The epidermis is 50 µm – 200 µm thick, depending on the region of the body. Most of the cells in the epidermis (90%) are keratinocytes, while the rest are melanocytes and Langerhans 2  dendritic cells. The outermost layer of the epidermis, the stratum corneum (10 µm – 40 µm thick), contains layers of dead keratinised cells and plays the most important role in the skin’s barrier function [2]. The stratum corneum provides a great physical barrier not just to harmful external agents, but also to useful drugs that need to be delivered into the skin [3]. Thus, effective transdermal drug delivery systems need to first disrupt, physically or chemically, the stratum corneum.   Figure 1.1: Different layers of the skin - epidermis, dermis and hypodermis [derivative of “Skin layers” (c) Madhero88 (http://commons.wikimedia.org/wiki/File:Skin_layers.png); used under a Creative Commons Attribution-Share Alike 3.0 Unported License]  The dermis, situated below the epidermis, is a tough, flexible and elastic layer and is 1 mm to 3 mm thick. This layer is made up of collagen, elastin and reticular fibres, and can be divided into two sublayers: the papillary dermis and the reticular dermis. The papillary dermis (upper sublayer) is a thin layer, consisting of loosely arranged connective tissue. The reticular 3  dermis is the lower sublayer, consisting of a network of connective tissue and collagen fibres that run horizontally. The reticular dermis also contains a dense network of capillary blood vessels and lymphatic vessels that circulate a variety of immune cells including dermal dendritic cells, monocytes, polymorphonuclear lymphocytes, and mast cells [2]. Due to its strong immune response and access to capillary and lymphatic systems, the dermis is an attractive site to deliver drugs and therapeutics. The delivery of drugs into the dermis is called intradermal drug delivery.     The hypodermis or subcutaneous tissue is situated below the dermis and is 1 mm to 2 mm thick. It consists of loosely arranged connective tissue and elastin fibres, and is primarily used for fat storage. Currently, most vaccine delivery is subcutaneous (in the hypodermis) or intramuscular (below the hypodermis), using a hypodermic needle and syringe. Intradermal delivery of vaccines is only widely used for Bacille Calmette-Guérin and rabies vaccines, which are delivered through a hypodermic needle using a technique called the Mantoux technique. There is a renewed interest in intradermal vaccine delivery due to skin’s strong immune response and the development of new and reliable delivery systems such as microneedles that eliminate the difficulties associated with hypodermic needle injections [4].    1.1.2 Immune response of the skin The skin is responsible for generating innate and adaptive immune responses. Innate immune responses are not specific to antigens (foreign substances that induce an immune response) and do not have immunological memory. Adaptive immune responses are specific to antigens and have immunological memory. Skin’s immune response is caused mainly due to the immune cells derived from the bone-marrow that specialize in antigen-presenting properties: the 4  epidermal Langerhans cells and the dermal dendritic cells. Together with macrophages from circulating blood and infiltrating dermal tissue, these cells are vital for the immune system [2].   The immune response of the skin makes delivery of vaccines particularly advantageous into the skin. The most successful vaccine to date, for smallpox (which got eradicated in 1980), was administered into the dermis. A simple bifurcated needle, developed by Benjamin A. Rubin, was dipped into a vial containing the potent vaccine to pick up about 2 µl of it between the two prongs of the needle. The needle, containing the vaccine, was jabbed several times, perpendicular to the skin, to deliver the vaccine into the papillary dermis [2]. Although the technique was effective due to delivery into the dermis, the amount of vaccine delivered could not be controlled precisely, which motivated new delivery systems [5].   1.1.3 Intradermal injections The advantages of intradermal injection of drugs have been known for decades, some of which are as follows:  • Dose sparing effect for vaccines: Some vaccines generate equivalent immune response with a significantly reduced dose for intradermal injections than for deeper injections, due to epidermal Langerhans and dermal dendritic cells. The dose sparing effect has been shown for vaccines including influenza, rabies, inactivated poliovirus, yellow fever, and hepatitis A vaccines, and can have extended benefits such as [4]:  o Potential reduction of cost per injection, including transport and storage of drugs, which could benefit resource-poor settings 5  o Stretch the availability of vaccines with limited supply due to manufacturing capabilities (e.g. in 2009 when the H1N1 influenza vaccine was not available to most low-income countries for 8 months) • Elimination of first-pass effect: The first-pass effect is a phenomenon where drugs are prematurely metabolized by the gastro-intestinal tract and the liver, which greatly reduces the concentration of drugs for systemic circulation [3]. The first-pass effect occurs for oral drug administration, but not for intradermal drug administration. Hence, intradermal drug delivery avoids the side-effects of drugs to the stomach or the liver.   • Consistency in skin thickness: While performing a classic intramuscular vaccination, the appropriate needle length needs to be selected based on the muscle mass of the injection site, thickness of the subcutaneous fat layer, and the patient’s weight. However, such adjustments in needle length are not required for intradermal injections due to the consistency in skin thickness across people with varying demographic profiles [2].   Despite the advantages, widespread use of intradermal drug delivery has been historically limited due to the drawbacks of the standard intradermal injection technique, the Mantoux technique (shown in the far right in Figure 1.2). Developed by Charles Mantoux in 1910 for the intradermal injection of tuberculin used to diagnose tuberculosis, this technique forms the basis of intradermal injections using hypodermic needles, which is still in use today. The Mantoux technique is performed by stretching the skin surface and inserting a hypodermic needle almost parallel to the skin surface. Some of the inherent challenges associated with this technique are as follows [2], [4]:  • Difficulty in performing the technique requiring highly trained medical personnel  6  • Poor consistency of the injected volume  • Invasive nature of hypodermic needle injections  • Potential for needle-stick injuries  • Risk of transmitting blood-borne pathogens through inappropriate needle re-use  • Risk of accidental subcutaneous injections (if a patient receives a reduced dose of a vaccine, to take advantage of the dose sparing effect in intradermal delivery, and the injection ends up deeper in the skin, the patient will not be immunized)   Figure 1.2: Needle angles for different types of injections, highlighting the difficulty in performing a near parallel intradermal injection (10˚ - 15˚ insertion angle). [“needle-insertion-angles-1” (c) British Columbia Institute of Technology (http://opentextbc.ca/clinicalskills/wp-content/uploads/sites/82/2015/09/needle-insertion-angles-1.png); used under a Creative Commons Attribution 4.0 International License]  7  Recent developments of novel devices such as microneedles, which overcome the challenges of the standard intradermal injection technique while preserving the advantages of intradermal injections, have renewed interest in intradermal drug delivery [2], [6], [7].   1.2 Microneedles Microneedles increase the permeability of drugs through the skin by breaching the barrier of the stratum corneum. Although conceptualized decades ago, microneedles gained significant research interest in the late-1990s because microfabrication technology made it possible to fabricate these micron-scale needles. Microneedles have been made with different materials (including silicon, metal, polymer and glass) and varying geometries, but can be broadly classified into the following categories [8]–[10]:  • Solid microneedles: Solid microneedles are usually sharp devices that penetrate or scrape the stratum corneum to create pores through which drugs can permeate. Thus, these microneedles are used as a pretreatment for forming holes in the skin, after which drugs are applied to the skin for a local effect or for systemic circulation following uptake by dermal capillaries. The drugs are either applied using a patch, as is used in transdermal drug delivery, or using a topical formulation such as gels, creams and ointments, as is used for skin treatment.  • Hollow microneedles: Hollow microneedles are like a miniaturization of the conventional hypodermic needles, through which liquid formulations of drugs are injected into the body. These microneedles usually pierce the stratum corneum to inject drugs into the dermis. The fluid flow is driven by a pressure source, which like in the case of hypodermic needles could be provided by a syringe. Since the injection of drugs is 8  pressure-driven, the amount and speed of injected fluid can be controlled. Hollow microneedles can also make use of existing liquid drug formulations, unlike the other types of microneedles that require the development of new drug formulations suited for the particular application.  • Coated microneedles: Solid microneedles can be coated with suitable drug formulations, for not only piercing the skin, but also depositing the drugs after insertion. The drugs coated onto the microneedles subsequently dissolve into the skin, for local or systemic delivery. However, the dose of drug deposited into the skin depends on the amount that can be coated onto the surface of the microneedle, which is usually less than 1 mg for small microneedle arrays, and could be insufficient for many applications.  • Dissolving microneedles: These types of microneedles are made entirely of water-soluble materials, such as polymers and sugars, that can completely dissolve in the skin after insertion. The microneedles usually have drugs encapsulated in them such that after the microneedles dissolve in the skin, the drugs are released, leaving no biohazardous sharps waste. However, the drugs delivered are limited by the amount that can be encapsulated.   Microneedles have been used to deliver a wide range of drugs and therapeutics. In addition to eliminating most of the drawbacks of the standard Mantoux technique, microneedles can improve patient compliance, and potentially permit self-administration. Before the use of microneedles, transdermal drug delivery was limited to low molecular weight and relatively lipophilic drugs that could permeate through the stratum corneum. However, microneedles have enabled the delivery of biotherapeutics (e.g. insulin) and vaccines (e.g. influenza vaccine) into 9  the skin, and bioactives into the eye and into cells [9], [10]. Hollow microneedles have also been investigated for biosensing and extraction of blood and interstitial fluid [11].   1.3 Previous injection experiments using hollow microneedles Studies have shown that the primary resistance to fluid flow during intradermal injections is caused by the skin, rather than the microneedle. The dense dermal tissue was found to cause the largest resistance to fluid flow, and the microneedle was not the rate-limiting barrier. Varying the effective radius of the microneedle tip opening from 22 µm to 48 µm (with areas from 1462 µm2 to 7400 µm2) had no significant effect on fluid injection rate, suggesting that the primary resistance is caused by the skin and not the microneedle. Furthermore, the fluid flow-rate through the microneedle into air was a few orders of magnitude higher than that into the skin. Studies also show that some of the dermal resistance to flow is reduced by partially retracting the microneedle to increase flow-rate. Researchers suggest that either the compression of dermal tissue around the microneedle or the deformation of tissue into the microneedle lumen prevents flow, and retracting the microneedle partially either relieves the dermal compression or removes the deformed tissue, allowing fluid to flow more easily [12].    Previous experiments [12]–[15] have been limited to studying the effect of injection parameters or microneedle geometries on fluid flow-rate, injected volume or perceived pain. The effect of the following parameters on the fluid flow-rate are provided below: • Injection fluid pressure: As expected, an increase in input fluid pressure increased the flow-rate, volume and pain associated. 10  • Microneedle retraction distance: Retracting the microneedle by a larger distance resulted in greater injection volumes, fluid flow-rates, and pain scores. An 11.6-fold increase in flow-rate was achieved by partial retraction of microneedles.  • Microneedle insertion depth: Greater insertion depth of the microneedle, followed by retraction of the microneedle, resulted in larger volumes delivered.  • Use of beveled tip: Flow-rate through a microneedle with a 35˚ beveled-tip was approximately 3-fold higher than that through a blunt-tip microneedle.  • Use of hyaluronidase: Hyaluronidase is an enzyme that breaks down hyaluronan, a glycosaminoglycan within skin collagen fibers, and reduces the flow resistance in the skin. Co-administration of hyaluronidase with the injected fluid resulted in a 7-fold increase in flow-rate.    Although previous studies have considered the effect of injection parameters on fluid flow-rate, the mechanism of fluid flow through skin tissue is not explored adequately and our current understanding of how fluid is absorbed by the skin during intradermal injections is limited. This lack of understanding, which is essential for controlling effective delivery of fluid into the skin, could be attributed to challenges associated with visualizing intradermal injections because skin tissue is a highly optically scattering biological material [16].   1.4 Visualization of injections To visualize the dynamics of intradermal injections, an imaging system requires micron-level spatial resolution, an imaging depth greater than 1 mm in a highly scattering biological medium, and a temporal resolution high enough to capture tissue or fluid movement without 11  significant motion artifacts. One such non-invasive biomedical imaging modality is optical coherence tomography (OCT), which produces two-dimensional (2D) cross-sectional images of internal tissue microstructures in real-time [17]. OCT has been widely used in ophthalmology [18] and dermatology [19], [20], due to its ability to perform non-invasive “optical biopsy” on tissue with micron-level resolution up to depths of 1-2 mm. Previously, the use of OCT had been limited to imaging microneedle penetration into skin tissue [21], [22], or dissolution of polymeric microneedles [23].   Other reported imaging of tissue were mostly carried out post-injection or post-insertion using confocal microscopy [6], [24]–[26], ultrasound echography [14], fluorescence microscopy [13], [27], histology [12], [13], [25], [28], x-ray computed tomography (micro-CT) [29], [30], or two-photon microscopy [31]. In an imaging system, there usually exist trade-offs between imaging depth and spatial resolution, and between spatial resolution and temporal resolution. For instance, fluorescence and confocal microscopy have adequate spatial and temporal resolutions to image injections, but have low imaging depths in skin (up to few hundred micrometers) that only allow visualizing the upper layers of skin after injections. On the other hand, micro-CT and two-photon microscopy have better imaging depths, but are not suitable for real-time imaging of skin cross-sections. The only reported real-time visualization of injections into skin tissue was for deeper subcutaneous injections using high-speed X-ray imaging [32], [33]. However, high-speed imaging techniques using X-ray or ultrasound do not have micron-level spatial resolution required for imaging intradermal injections.    12  1.5 Biological tissue as a deformable porous medium Biological tissue can be considered a porous medium for modelling fluid flow [34]. A porous material consists of a solid matrix with interconnected voids, which are usually assumed to be filled with fluid (in the case of tissue). Utilizing the continuum approach, the porous medium is characterized by macroscopic properties such as porosity and permeability, which are averaged over a sufficiently large representative elementary volume (r.e.v.). The length scale of the r.e.v., 𝑙, is much larger than the pore scale (given by the average size of the pores d), but is smaller than the length scale 𝐿 over which macroscopic properties have to be considered (Figure 1.3), i.e. 𝛿 ≪ 𝑙 ≪ 𝐿. Generally in biological tissues, 𝛿 < 0.1	𝜇𝑚, 𝑙~1	𝜇𝑚, and 𝐿~10	𝜇𝑚 to 1	𝑐𝑚 [35].   Figure 1.3: Representative elementary volume for a porous medium such as a biological tissue  The macroscopic variables averaged over the r.e.v. neglect the details of the pore structure, and are properties defined for the entire porous medium. The porosity of the medium 𝜙% = 𝑉%𝑉 	, (1) 13  which is dimensionless, is the ratio between the void volume occupied by the fluid, 𝑉%, to the total volume of the porous medium, 𝑉. Porosity only depends on the geometry of the medium, and does not provide any information about the interconnectedness of the voids.   The relative porosity of the solid phase 𝜙& = 𝑉&𝑉  (2) can also be described using the volume occupied by the solid, 𝑉&.The relationship between the volumes and porosities of the two phases are given as follows: 𝑉 = 𝑉% + 𝑉%  (3) 1 = 𝜙& + 𝜙%  (4)  The flow conductivity in a porous medium is defined by its permeability, 𝐾. For a porous medium, where viscous effects dominate inertial effects (low Reynold’s number), the relationship between the pressure gradient and fluid flow-rate is given by Darcy’s Law, which states that the fluid flow-rate through a porous medium is linearly proportional to the pressure gradient. The permeability, 𝐾, is the constant of proportionality in Darcy’s Law, which in simplified one-dimensional form (x-direction) can be written as follows [35] defines the average flow velocity in the x-direction 𝑣5 = −𝐾𝜇 𝜕𝑝𝜕𝑥 (5) as a function of the dynamic viscosity of the fluid 𝜇, the pressure 𝑝 and the permeability K. The permeability 𝐾 has units of 𝑚-, and the dynamic viscosity 𝜇 has units of 𝑁. 𝑠/𝑚-. However, in 14  the literature for biological porous media, the fluid viscosity is often incorporated into the permeability, 𝜅, such that 𝜅 = 𝐾/𝜇, with units of 𝑚</𝑁/𝑠 [36]–[38]. An approximate range of permeability in biological tissue is from 𝜅 ≈ 0.5 × 10]^_	𝑚</𝑁/𝑠 for cartilage to 𝜅 ≈3.2 × 10]^-	𝑚</𝑁/𝑠 for fat [36]. Darcy’s law can be also written as ∇𝑝 = −𝑣𝜅	. (6)  Additionally, soft tissue, such as skin tissue, can be considered as a deformable porous medium: the pressure associated with the injected fluid can deform the soft porous matrix as fluid flows through its pores. The individual solid and fluid phases, however, are assumed to be intrinsically incompressible. Any bulk compression results from the reduction in the volume fraction of the fluid (porosity, 𝜙%), rather than the compression of the individual fluid or solid components. Likewise, expansion of the bulk porous medium results from an increase in volume fraction, rather than expansion of individual components.   A compression of the deformable porous medium results in an decrease in porosity, 𝜙% , and an expansion results in an increase in porosity. The relationship between the porosity and deformation of the solid matrix is given by [36]: 𝜙% = 𝜙%c + 𝜙%c∇. u (7)  Here, u is the displacement of the solid. The permeability of the medium depends on the porosity, and the most commonly used form of this dependence  𝜅 = κc exp(𝑚𝜙%), (8) 15  uses a material constant 𝑚. As assumed by some researchers [36], the fluid flow and the deformation of the solid matrix are coupled because the fluid flow deforms the tissue, which subsequently affects the passage of fluid by altering the local permeability. For instance, if fluid flow into tissue expands the tissue, it increases the porosity, 𝜙% , and subsequently increases the local permeability, 𝜅. According to Darcy’s law, an increase in permeability 𝜅 increases flow 𝑣, for the same pressure gradient ∇𝑝.  The current literature on simulations and mathematical models of injections into biological tissue has varying results due to differences in model assumptions. Tissue has been modelled as a mechanically non-linear deformable porous medium [36], where injection into tissue formed a fluid-filled spherical cavity and high cavity pressure caused fluid to flow into the neighbouring tissue. Fluid flow and tissue deformation were found to be coupled and any flow induced deformation of the material increased local tissue permeability, aiding fluid transport. Another model considered the injection of fluid into a layer of deformable porous medium from a point source [37], with applications to subcutaneous injections, assuming linear poro-elasticity and a constant permeability of the medium. Simulations using these assumptions indicated a swelling of the porous medium (with no cavity formation) and a subsequent deformation of the free surface, resembling the raised papule (wheal or bleb) observed after successful intradermal injections. The upper surface of the porous medium was considered to be impermeable, as is the case in the skin with the stratum corneum acting as the impermeable boundary. Since the free surface was impermeable, the entire medium inflated, while lifting the free surface and moving the position of the source (of fluid injection) upwards.  16  Other models have considered flow through skin as porous medium without deformations [39]–[41], flow into a growing cavity in non-porous medium [42], or a combination of spherical expansion in non-porous skin followed by spherical diffusion in porous skin [43]. For creating more accurate models, we require experimental observations of flow through tissue to validate model assumptions and to provide physical insights into the dynamic behavior of tissue during injections. Visualizing intradermal injections in real-time with micron-scale resolution will allow backing the assumptions used in models with physical observations from experiments.   1.6 Objectives The objectives of the thesis are briefly described below:  1. Design an experimental setup to perform repeatable injections into excised skin samples, while controlling and recording injection parameters, and visualizing the skin tissue in real-time. 2. Record the fluid flow-rate (and injected volume) and the deformation of skin tissue at the micron-scale over time during different parts of the injection. 3. Quantify the local deformation of the skin tissue during intradermal injections.  4. Calculate the total volume of skin tissue deformation, and compare it with the volume of fluid injected into the skin.  5. Suggest injection protocols based on observations and results.   Chapter 2 describes the experimental setup and the methods of analyses used in the thesis. Chapter 3 describes the experimental results, while Chapter 4 includes further analyses to extract additional results. Chapter 5 provides the conclusions and outlines potential future work.  17  Chapter 2: Materials and methods This chapter describes the experimental setup designed and built for performing controlled injections into skin samples, and the techniques used for processing the data collected during the injection experiments.   2.1 Intradermal injections using hollow microneedles Before building the final experimental setup to control and observe intradermal injections, we first tested injections using simpler prototypes to find out the injection parameters that repeatedly resulted in successful intradermal injections. We performed all the experiments on excised porcine skin tissue using single hollow microneedles (length: 550 µm - 700 µm; lumen diameter: 80 µm - 100 µm). The general steps required for successfully injecting fluid into the skin sample are shown in Figure 2.1 and are outlined below:  i. Microneedle (MN) impact: To deliver fluids into the dermis, the microneedle needs to first pierce the stratum corneum. It has been reported that impacting the microneedle with a velocity 𝑣2C of 1 m/s to 3 m/s pierces human skin reproducibly [44]. ii. MN insertion: After impact, the microneedle is inserted into the stretched skin sample. After excising a skin sample, it shrinks laterally and therefore should be stretched back to its original dimension to more closely mimic in-vivo skin. In addition, we have observed that un-stretched and stretched skin behave differently – the microneedles penetrate more easily into stretched samples than un-stretched ones, supporting the need to stretch the skin samples.  iii. Application of input pressure: When we applied an input pressure 𝑃2? = 100 kPa with the microneedle inserted, water was not injected into the skin. Some researchers have 18  attributed this inability to inject fluid initially to either the compression of dermal tissue around the microneedle or plugging of the microneedle lumen by dermal tissue during microneedle insertion [13], which had motivated retracting the microneedles partially to relieve the tissue compression and/or remove cored tissue. iv. MN retraction: For our experiments, we found that retracting the MN by 𝑑3 ≈  0.3 mm consistently reduced the dermal resistance and triggered the onset of fluid flow. v. Successful injection: Injection of fluid into the skin formed a raised papule (wheal or bleb), which is characteristic of intradermal injections even with hypodermic needles. One of the main objectives of the research was to observe the tissue microstructure during this step.    Figure 2.1: Cross-sectional schematic of skin and microneedle showing 5 steps of a successful intradermal injection: i) microneedle impacting the stretched skin (with velocity vim) to penetrate stratum corneum; ii) microneedle insertion; iii) application of input pressure (Pin) with no initial flow due to high dermal resistance; iv) retraction of microneedle (by dr) ; v) successful injection of fluid with the formation of a raised papule (characteristic of intradermal injections).   Based on the steps for intradermal injections, an experimental setup needed to: control injection parameters such as fluid input pressure 𝑃2?, microneedle impact velocity 𝑣2C, insertion depth 𝑑2, and retraction distance 𝑑3; record fluid parameters such as pressure 𝑃C, and flow-rate 19  𝑄; and visualize the tissue microstructure. We used optical coherence tomography (OCT) for real-time visualization of skin tissue deformation during intradermal injections through hollow microneedles.  2.2  Experimental setup We designed the experimental setup in Figure 2.2 to observe the dynamics of intradermal injections. This setup allows impacting the microneedle onto the skin sample, retracting the microneedle, and controlling input fluid pressure. During each experiment, we visualized the deformation of skin tissue due to fluid flow, while continuously recording the flow conditions such as the pressure of the fluid as it enters the needle 𝑃C, and the flow-rate 𝑄. To ensure repeatability between experiments, we controlled the following injection parameters: fluid input pressure 𝑃2?, microneedle impact velocity 𝑣2C, insertion depth 𝑑2, and retraction distance 𝑑3.   A pressure controller supplied compressed air at 𝑃2? to a fluid reservoir containing the injection fluid, which flowed, through a flow-rate sensor and a pressure sensor, to a single hollow microneedle. An impact system and a retraction system controlled the vertical position of the microneedle during insertion and injection, respectively. A tissue holder secured a sample of excised porcine skin tissue, which was visualized, along with the microneedle, using optical coherence tomography. 20    Figure 2.2: Simplified schematic of the experimental setup (top; not to scale) showing the major components and their interactions, with variables indicating injection parameters controlled and/or measured; and picture (bottom) showing the different components of the experimental setup.  The single hollow microneedles were fabricated by electroplating nickel onto polymeric (SU-8) pillars, using the fabrication process detailed in [25]. For inserting the microneedle into the skin, as described earlier, the microneedle first had to impact the skin with an impact velocity of between 1 m/s to 3 m/s [44]. Our impact system uses the energy of a compressed spring with a spring constant k = 343 N/m to insert the microneedle at 𝑣2C of approximately 2 m/s into a sample of excised porcine skin tissue.   21   Figure 2.3: Picture of the injection system (dashed grey box in Figure 2.2 schematic), showing the impact system, retraction system, tissue holder and OCT. Detail (blue box) showing microneedle and excised porcine skin sample (stretched) in the tissue holder. Scale bars (white and black) are 10 mm.  The skin sample was stretched uniformly in radial directions in the tissue holder by a transparent polydimethylsiloxane (PDMS) backing. The skin was stretched to its approximate original dimensions (pre-excision), using steps shown in Figure 2.4. The stretched skin sample is used for the injection experiments (picture of tissue holder and microneedle shown previously in Figure 2.3).   The PDMS (SYLGARD 182 Silicone Elastomer) was prepared by thoroughly mixing the base and curing agents at a 10:1 weight ratio, vacuum de-airing to remove bubbles, and pouring into a hollow cavity of an aged 3-D printed part made of Verowhite (Stratasys Direct, Inc.); the 3-D printed part was left at room temperature and pressure for 1 week, after which the material did not interfere with the curing process of the PDMS. The PDMS backing provided three main 22  functions: stretching the skin sample back to its original dimensions; maintaining a fixed and impermeable boundary condition at the bottom surface of the dermis; and providing optical access for visualizing tissue deformation.   Figure 2.4: Steps for stretching the excised skin sample: First the skin sample is clamped firmly, and then stretched uniformly in radial directions by the PDMS backing.   23  To capture tissue deformation during injections at sub-second temporal resolution and micron-level spatial resolution, we used the Thorlabs TELESTO-II Spectral Domain OCT Imaging System. The OCT uses near-infrared light, with a central wavelength of 1300 nm, to provide 2D cross-sectional images of the skin tissue. Water and PDMS are transparent to the OCT. The OCT has an axial resolution of 5.5 µm in air (4.2 µm in water), a lateral resolution of 13 µm and an imaging depth of 3.5 mm in air. In a highly scattering material such as skin tissue, the imaging depth is reduced to around 1-2.5 mm. The OCT images were recorded using a computer, and were processed and analyzed using MATLAB and ImageJ, as described in sections 2.6, 2.7 and 2.9.  Apart from the OCT image acquisition, other injection parameters were also measured or controlled using a computer. A LABVIEW program (National Instruments, Austin, TX, USA) synchronized the control and measurement of the injection parameters: controlling/measuring the input pressure 𝑃2?; measuring the fluid flow-rate 𝑄; measuring the fluid pressure 𝑃C; and controlling/measuring the retraction distance 𝑑3 (Figure 2.2). An input pressure 𝑃2? between 0-200 kPa was applied to the fluid reservoir by the pressure controller, Elveflow OB1. We used Milli-Q water (successively filtered and de-ionized), by EMD Millipore Corporation, as the injection fluid, which was degassed for 3 minutes before each experiment. The flow-rate 𝑄 and pressure 𝑃C of the injection fluid were measured at 100 Hz using an Elveflow Microfluidic Flow Sensor and an Elveflow Pressure Sensor, respectively.   The retraction distance 𝑑3 was set using a linear motorized stage, Zaber T-LSM050A, with a microstep size (default resolution) of 0.048 µm. Some of the injection parameters, such as 24  the microneedle impact velocity 𝑣2C and the insertion depth 𝑑2, were manually controlled. The impact velocity 𝑣2C was set using the compression of the spring, and our impact system gave a choice of three impact velocities 𝑣2C of 1.3 m/s, 2.0 m/s, and 2.6 m/s based on the three levels of spring compression 10 mm, 15 mm, and 20 mm. The insertion depth of the microneedle was controlled by vertically moving the tissue holder, which was mounted on a single-axis translation stage (z-axis) with a standard micrometer. A list of the sensors and actuators used in the experimental setup are provided in Table 2.1. Appendix B.3 and B.4 show more details of the design and tubing connections of the experimental setup.   Table 2.1: Sensors and actuators used in the experimental setup and the corresponding parameters controlled or measured. Controlled Quantity  Symbol Actuator Range Input Pressure 𝑃2?,k Elveflow OB1 Pressure Controller 0 – 2 bar (0 – 200 kPa) Retraction distance 𝑑3 Zaber T-LSM050A Motorized Stage 50.8 mm, 0.00022 mm/s – 7 mm/s  Measured Quantity Symbol Sensor Range Input Pressure  𝑃2?,C Elveflow OB1 Pressure Controller 0 – 2 bar (0 – 200 kPa) Flowrate  𝑄 Elveflow Microfluidic Flow Sensor 0 – 1 ml/min Pressure 𝑃C Elveflow Pressure Sensor 0 – 30 psi (0 - 206.843 kPa)   2.3 Skin sample preparation  Porcine skin is often used as a model for human skin due to similarities in anatomy and physiology, and it has been found to have similar dermal/transdermal absorption as that of 25  human skin [45]. For our experiments, porcine skin tissue was excised from the abdomen of a female Yorkshire pig at the Center of Comparative Medicine (CCM) at the University of British Columbia. Freshly excised abdominal skin samples were cut into 30	𝑚𝑚 × 25	𝑚𝑚 rectangular pieces and frozen at -80 °C. Before each experiment, the frozen skin samples were thawed and sliced to a thickness of around 1-2 mm by removing subcutaneous fat using surgical scalpels and scissors. Removing the subcutaneous layer was necessary for visualizing the dermis and epidermis using OCT, because of its imaging depth of approximately 1-2.5 mm in biological media. Finally, to account for the contraction of the skin samples post excision to approximately 50% of their in-vivo area [46], the skin was stretched uniformly in radial direction in the tissue holder to its approximate original size, as shown earlier in Figure 2.4.   2.4 Injection procedure After preparing the porcine skin sample, it was mounted onto the tissue holder. The position of the tissue holder was adjusted such that the microneedle base aligned with the initial position of the stratum corneum, after microneedle impact when the spring was relaxed. This adjustment led to consistent insertion depths between experiments, and prevented the microneedle base from applying additional compressive stresses to the skin at impact. At the pre-impact position, with the spring compressed and the microneedle raised, pressure was applied to the fluid to fill the sensors, microneedle and tubing. After expelling water out of the microneedle for a few seconds, the microneedle impacted the skin sample to penetrate the stratum corneum. A constant input pressure, 𝑃2? = 100𝑘𝑃𝑎, was then applied to the fluid using the OB1 pressure controller. The application of input pressure was followed by a rapid increase in flow-rate due to the compliance of the tubing in the system followed by a rapid decay to zero flow-rate. A delay 26  of 10 seconds before beginning the retraction protocol ensured that the tubing compliance did not affect flow-rate readings during the experiment. The OCT captured cross-sectional images at the approximate mid-section of the microneedle. The movement of the skin tissue and the microneedle were observed during fluid injection.  2.5 OCT image acquisition The OCT creates a 1D depth profile of the sample using the interference pattern from the backscattered light (from the sample) and a reference light beam. The system combines multiple 1D scans to form a 2D cross-sectional image and successive 2D tomographic sections can be stacked to provide a 3D image of the sample [17]. To capture the injections at a high frame rate, only one 2D cross-section of the sample (passing through the approximate center of the microneedle) was acquired over time.   The acquisition time of an OCT frame, which corresponded to a single cross-section of the skin, depended on the selected axial scan rate, axial scan averaging, field of view (FOV) and pixel resolution. Appendix A  provides the scan parameters for all injections mentioned in the thesis. A typical OCT image showing a cross-section of the skin at the approximate mid-section of the microneedle is shown in Figure 2.5. As shown in cross-sectional schematic (top) in the figure, the OCT probe is situated below the tissue holder, imaging the skin sample from the bottom (dermal side) through the transparent PDMS backing.  27   Figure 2.5: Cross-sectional schematic (mid-section) showing the microneedle inserted into the stretched skin sample (top) with OCT imaging the skin and the microneedle through the transparent PDMS backing. OCT image (bottom) showing a cross-section of skin with the microneedle inserted. The scale bar is 0.5 mm.   Due to the difference in the refractive indices of air and skin, the vertical dimensions of the skin and its features in an OCT image differ from their actual dimensions. Near infrared light, used by OCT, travels faster in air than in skin because the refractive index of air is lower than that of skin. For the same time, the distance travelled by light in the reference medium (air) is larger than the distance travelled in the sample (skin). Therefore, the samples appear to be elongated vertically in the OCT images by a factor equal to the refractive index of the material (skin), assuming the refractive index of air is unity. The OCT images were adjusted by dividing the vertical dimension of the image by the refractive index of the skin (𝑛&>2?), estimated to be 1.375 ± 0.006. The OCT image shown in Figure 2.5 has been adjusted for the refractive index of the skin. Sample OCT images, including a raw OCT image and the corresponding refractive 28  index corrected OCT image, are shown in Figure 2.8 of section 2.7. Since the entire image was scaled based on the refractive index of skin, the distances in the skin in the OCT images represented actual physical distances in skin and the distances in air in the same OCT images were reduced by a factor equal to 𝑛&>2?. This reduction of distance in air did not affect any results, since all the analysis included distances in the skin.   2.5.1 Skin refractive index estimation The refractive index of the skin was estimated by considering an edge of a skin sample between two parallel glass slides (Figure 2.6). In the figure, the distance of the skin appears larger than the distance of the air gap. The air gap seen in the OCT image is not distorted because the reference medium is air, while the skin is stretched due to its refractive index.    Figure 2.6: Setup for measuring the refractive index of skin (left); Corresponding OCT image (right) around the edge of the skin,  showing an apparent increase in height of skin in the image.  The refractive index of the skin (𝑛&>2?) can be calculated using the heights of skin and air in the OCT image as  𝑛&>2? = ℎq>2?ℎr23 	. (9) 29  We determined the refractive index of skin to be 1.375 ± 0.006, considering 23 OCT images. This estimate of refractive index was used to adjust the sizes of OCT images displayed in the thesis, and to correct the vertical distances used for calculations. The refractive index of skin and is close to that of water. As a result, we expect a small change in refractive index as water is injected into tissue and the error associated with this is negligible. Assuming the refractive index of water to be 1.333, the error is approximately 3%: for a cavity of water with a height of 200 µm, the error is around 6 µm, which is negligible considering that the axial resolution of the OCT is 4.2 µm in water and that the fluid filled cavities observed had heights much lesser than 200 µm.   2.5.2 Additional capabilities of OCT In addition to visualizing the dynamics of intradermal injections in real-time at the micron-scale, the OCT can also be used for visualizing the shape of the skin surface after injection and measuring the height of the microneedle, as shown in Figure 2.7. The 3D image of the skin surface can be used to observe the shape of the raised papule formed after injection. The two images on the left of Figure 2.7 show that the skin surface is raised around the microneedle and the base of the MN prevents the upward movement of the skin directly beneath it.    2.5.3 Other imaging modalities Apart from OCT, we also tested other imaging modalities including x-ray computed tomography (micro-CT), histology and confocal microscopy (Appendix B.1 and B.2). However, the other imaging modalities had shortcomings that prevented them from being suitable for visualizing intradermal injections in real-time. An imaging depth of around 300 µm using the 30  confocal microscope was insufficient for our application. Histology could only be performed after the injection, and micro-CT did not have adequate temporal resolution for real-time imaging. OCT provided the right range of spatial resolution, temporal resolution and imaging depth for imaging the tissue during intradermal injections.     Figure 2.7: Additional capabilities of the OCT, a: 3D image of the skin surface after injection, showing hole created by the microneedle and the swelling of the skin around the microneedle imprint. b: one cross-section of the 3D image, showing the microneedle imprint and the swelling of skin around the MN. c: OCT image of a single microneedle, which can be used to determine the height of the MN. The scale bars are 0.5 mm.   2.6 Speckle reduction in OCT images OCT uses interferometry as its measurement technique and relies on the coherence (spatial and temporal) of optical waves backscattered from the sample. However, in a highly scattering material such as skin tissue, multiple back scattering and forward scattering events of the beam combine randomly to create speckle noise in the interference signal [47]. Speckle noise reduces the signal-to-noise ratio and the contrast of OCT images [48], degrades image quality by producing grainy appearances [49], and limits the performance of subsequent quantitative image analysis [50]. We used two different methods for reducing speckle noise in our OCT images: averaging consecutive OCT images, and wavelet-transform based filtering. The post-processing of OCT images was carried out for injections captured at either of the two frame rates: 17.2 31  frames/s (InjA) or 73.9 frames/s (InjB), corresponding to acquisition times (for each OCT frame) of either 58 ms or 13.5 ms, respectively. Image processing using the two denoising techniques on a sample OCT image of InjA are shown in Figure 2.8 of section 2.7.  Some researchers [51] have averaged axial scans of OCT for real-time reduction of speckle noise, but here we averaged OCT images post-acquisition.  The same speckle noise reduction could have been achieved by longer averaging during OCT recording, but that would have reduced the temporal resolution. Here, we maintain the temporal resolution (13.5 ms) while reducing speckle noise: we average images n-1 and n to image A and average images n and n+1 to image B while maintaining the inter-image period. Averaging was only used for OCT images captured at high frame rates (InjB - 73.9 frames/s), since at lower frame rates (InjA - 17.2 frames/s) averaging produced significant motion artifacts. The greyscale intensity at each pixel location in an image was averaged with that of the corresponding pixel location in the consecutive image. The averaged images were used to quantify tissue deformation using digital image correlation. For the high frame rate images, tissue deformations were captured at sufficiently high temporal resolution such that features of the tissue were not blurred by averaging, after completion of microneedle retraction.   For OCT images captured at 17.2 frames/s (InjA), we used wavelet-transform based filtering to reduce speckle noise. Wavelet filtering has been found to significantly reduce speckle noise and increase the signal-to-noise ratio in OCT images, while preserving strong edges and maintaining image sharpness [49]. Transform domain filtering, such as wavelet and curvelet [50] filtering, relies on the sparse representation of signal and widespread distribution of noise in the 32  decomposition levels of the transform domain, and appropriate thresholds that attenuate noise. The wavelet denoising procedure, like those described in [48], [49], [52], included five main steps: logarithmic conversion, forward wavelet transform, thresholding, inverse wavelet transform, and exponential conversion. Logarithmic transformation (base 10) of the OCT image converted speckle noise, modelled usually as multiplicative noise, into additive noise. Then, a multi-level wavelet decomposition of the logarithmic image produced wavelet coefficients at different decomposition levels (up to level N).   The signal was first decomposed into a low pass sub-band (approximation level) and a high-pass sub-band (detail-level), and the approximation level was further decomposed for multi-level analysis. The 2D maps of the detail coefficients at horizontal, diagonal and vertical subbands were visually analyzed to adjust thresholds at different decomposition levels. A higher threshold was applied to 2D maps that had detail coefficients predominantly represented in regions of high noise outside the skin, where most of the OCT signal was attenuated. Inverse wavelet transform reconstructed the denoised image, and an exponential conversion (base 10) changed the image from the logarithmic scale to the original linear scale. We used the 2D Stationary Wavelet Transform function from the Wavelet Toolbox in MATLAB for denoising images. We applied a near symmetric, eight-tap symmlet wavelet (sym-8) with a four-level decomposition (N=4) and soft thresholding with a greater threshold applied to vertical subbands to preserve horizontal edge information common in OCT images of tissue, as described in [49]. An example of decomposing an image into approximation and detail coefficients, and thresholding the detail coefficients is provided in Appendix D.1 and D.2.  33  2.7 Sample OCT image processing The raw OCT images are processed as described in sections 2.5 and 2.6, and as shown in Figure 2.8. Due to the difference in refractive index of the skin and the reference medium (air), the skin appears elongated (as in ‘I’ of Figure 2.8). The following image processing techniques were utilized: • Refractive index correction: The size of the raw OCT image (I) was adjusted by dividing the height of the image with the refractive index of skin, estimated to be 1.375, as described in section 2.5. The resulting image (IIa) represented the actual physical dimensions of the skin.  o Brightness adjustment: The brightness was adjusted in ImageJ for displaying the image more clearly.  • Wavelet transform based filtering: This filtering technique, as described in section 2.6, was used for denoising images captured at lower frame rates.  o Brightness adjustment: Brightness was adjusted for displaying the images.  • Averaging: Two consecutive OCT images were averaged, as described in section 2.6, only for the images captured at high frame rates. For lower frame rates, averaging produced motion artifacts.  The images displayed in the thesis are in the final forms of IIb, IIIb, or IV. Digital image correlation, for quantifying tissue expansion, was done on images that were denoised either using wavelet transform based filtering (IIIa) for low frame-rate images or averaging (IV) for high frame-rate images.   34   Figure 2.8: Image processing of a sample OCT image, showing the different types of techniques used. In the thesis, OCT images are presented in the forms IIb, IIIb, or IV. The scale bars are 0.5 mm.  2.8 Digital image correlation Digital image correlation (DIC) is a non-contact technique for measuring material deformation, with a wide variety of applications including multiscale biomechanics [53]. For measuring tissue deformation in our OCT images, we used NCorr, an open-source 2D subset-35  based DIC software package implemented in MATLAB [54]. In subset-based algorithms, a reference image is partitioned into subsets or subwindows, smaller regions that are initially contiguous groups of points. Deformations inside each subset are assumed to be homogenous, and deformed subsets are tracked in a current image. NCorr calculates Green-Lagrangian strains 𝐸55, 𝐸56, and 𝐸66, based on four displacement gradients, stus5 , stus6 , stvs5 , and stvs6 , as shown below: 𝐸55 = 12w2𝜕𝑢5𝜕𝑥 + x𝜕𝑢5𝜕𝑥 y- + x𝜕𝑢6𝜕𝑥 y-z (10) 𝐸56 = 𝐸65 = 12 x𝜕𝑢5𝜕𝑧 + 𝜕𝑢6𝜕𝑥 + 𝜕𝑢5𝜕𝑥 𝜕𝑢5𝜕𝑧 + 𝜕𝑢6𝜕𝑥 𝜕𝑢6𝜕𝑧 y (11) 𝐸66 = 12w2𝜕𝑢6𝜕𝑧 + x𝜕𝑢5𝜕𝑧 y- + x𝜕𝑢6𝜕𝑧 y-z (12) NCorr uses its strain window algorithm to compute the displacement gradients and Green-Lagrangian strains, and obtains the entire strain field of the selected ROI [54]. Since the computation of strain involves differentiation, which is susceptible to noise, the NCorr algorithm a least squares plane fit on a subset of displacement data to get the values of the displacement gradients for equations (10) – (12), an example of which is shown in Appendix E.1 and E.2. We used NCorr to calculate strain fields for two cases: averaged (InjB – 73.9 frames/s), and wavelet filtered (InjA – 17.2 frames/s).   2.8.1 Time difference between DIC results The displacement fields and strain fields presented in this thesis are not simply based on the DIC results on a pair of OCT images. The results displayed are cumulative results from 9 pairs and 8 pairs of images for InjA and InjB, respectively. For InjA, the procedure for performing DIC, as shown in Figure 2.9, was as follows: 36  • Wavelet transform based filtering: Each raw OCT image (R1 – R10) was denoised using the technique described in section 2.6 (image processing IIIa in Figure 2.8). The time difference between consecutive denoised images (WD1 – WD10) was the same as that between the raw images (Δt = 58 ms).  • DIC on consecutive images: DIC was performed on consecutive denoised images (WD1 and WD2, WD2 and WD3, and so on), with the reference image being updated after each iteration. The time difference (Δt) between consecutive results was 58 ms.  • Cumulative DIC results: The DIC results from 9 pairs of images (a total of 10 images) were combined in the NCorr software to provide 𝑢5, 𝑢6, 𝐸55, 𝐸56 and 𝐸66 with a total time difference of 0.52 s.   Figure 2.9: Schematic for calculating the cumulative displacement and strain fields for InjA using DIC results from 9 pairs of wavelet denoised OCT images (Δt = 58 ms), with a time interval of 0.52 s between the first reference  (WD1) and last denoised OCT image (WD10).  37  The procedure for performing DIC on OCT images from InjB, as shown in Figure 2.10, was as follows:  • Averaging: Two consecutive raw OCT images (e.g. a and b) were averaged, (a+b)/2, using the technique described in section 2.6 (image processing IV in Figure 2.8). The time difference between consecutive averaged images, e.g. (a+b)/2 and (b+c)/2, was the same as that between the raw images (Δt = 13.5 ms).  • DIC on consecutive images: DIC was performed on consecutive averaged OCT images, e.g. (a+b)/2 and (b+c)/2, with the reference image being updated after each iteration. The time difference (Δt) between consecutive results was also 13.5 ms.  • Cumulative DIC results: The DIC results from 8 pairs of averaged images were combined in the NCorr software to provide 𝑢5, 𝑢6, 𝐸55, 𝐸56 and 𝐸66, with a total time difference of 108 ms between the first reference and last averaged OCT image.   Figure 2.10: Schematic for calculating the cumulative displacement and strain fields for InjB using DIC results from 8 pairs of averaged OCT images (Δt = 13.5 ms), with a time interval of 108 ms between the first reference, (a+b)/2, and last averaged OCT image, (i+j)/2. 38  2.8.2 Calculating 2D strain The DIC algorithm calculates the values of displacement fields 𝑢5 and 𝑢6, and strain fields 𝐸55, 𝐸56, and 𝐸66. After obtaining the strain fields 𝐸55, 𝐸56, and 𝐸66 in Cartesian coordinates from the DIC algorithm, we calculated 2D strain fields 𝜀-8 = 𝐸55 + 𝐸66 + 𝐸55𝐸66 − 𝐸56- 	. (13) In Chapter 4, the 2D strain fields are used for displaying the distribution of strain in the tissue. For calculating the volume of tissue expansion, 3D volumetric strain is calculated, as described later in section 2.9.2. For comparing the results of tissue deformation in 2D, the values of 2D strain are used in the thesis instead of the 3D volumetric strain due to the intuitive physical interpretation of the 2D strain maps. The magnitude of 2D strain provides a measure of tissue expansion and a relative change in area with respect to a reference area element – positive and negative values of 2D strain indicate an increase or decrease in area, respectively.    2.9 Volume of tissue expansion To see how tissue expansion relates to fluid absorption, we estimated the total volume of tissue expansion using OCT images and compared it with the total injected volume from the sensor measurements. We calculated the volume of tissue expansion from OCT images using two methods: surface deformation and volumetric strain.  Firstly, consider injection of water, as is the case in our experiments, into skin. The following assumptions are used: 1. Skin is a deformable porous medium containing fluid filled pores (same as the injected fluid, water). Thus, the porous medium has only two components – fluid and solid. 2. The individual components (solid and fluid) are incompressible. 39  3. During injection, fluid is added at the source position (tip of the microneedle), with the total volume of solid remaining the same (no solid components added).  Consider two cases as shown in Figure 2.11, at different times during the injection, having different macroscopic properties such as porosity and permeability:  • A: Before injection (original un-deformed state of the porous medium) • B: During injection and addition of fluid components   Figure 2.11: Two cases - before (A) and during (B) injection of water into skin – with different porosity and permeability.   The volume of the porous medium is given by the sum of volumes of its fluid and solid components:  𝑉r = 𝑉r% + 𝑉r&	; 		𝑉} = 𝑉}% + 𝑉}& (14)  Taking the difference of the volumes and using assumption 3, we find  𝑉} − 𝑉r = 𝑉}% − 𝑉r%	. (15)  For  𝑉}% > 𝑉r%  (as in the case of injection in our experiments), the left-hand term is the overall expansion/swelling of the porous medium (change in total volume), which is equal to the 40  injected volume of fluid (right hand term). From the OCT images, we calculate the overall expansion (𝑉} − 𝑉r) using the surface deformation (SD) technique, and we calculate the local expansion using 3D volumetric strain.   2.9.1 Surface deformation technique The surface deformation technique considers the swelling of the entire porous medium – a macroscopic view of expansion of the whole skin tissue. Assuming incompressibility of individual (solid and fluid) components of tissue, any fluid added by injecting through the microneedle inflates the porous medium. Thus, the volume of the injected fluid at any instant is equal to the increased bulk volume of the porous medium with respect to its initial un-deformed bulk volume. The overall expansion of the medium is calculated using the un-deformed (before injection) and deformed positions of the skin surface in the OCT images, as shown in Figure 2.12. We used ImageJ to outline the top surface of the skin and MATLAB to calculate the volume.    Figure 2.12: Left: OCT image before injection, showing the un-deformed state of the skin that is used as a reference. The surface of the skin is shown in white dashed lines and the microneedle is shown in yellow dotted lines. Right: OCT image during injection, showing the new position of the skin surface (blue dashed line), and the original un-deformed position (white dashed line). The area between the blue and white dashed lines is the added area, which is used for calculating the volume assuming axisymmetric conditions around the central axis of the microneedle (red dashed line). The scale bar (black) is 0.5 mm. 41  The change in area (using the 2D OCT images) is assumed to be axisymmetric around the central axis of the microneedle. The volume of tissue expansion using the surface deformation technique 𝑉q8 = 𝜋𝑅2‚𝛿𝐴2‚‚2  (16)  is calculated with 𝑖 and 𝑗 spanning the entire domain between the blue and white curves (in Figure 2.12). The radius (in mm) is denoted by 𝑅2‚, the area (in mm2) is denoted by 𝛿𝐴2‚  corresponding to an area element between the two dashed lines and outside the microneedle equivalent to 1 pixel × 1 pixel, and 𝑉q8  is the volume of tissue expansion (in mm3 or µl), which is equivalent to 𝑉} − 𝑉r in equation (13).    2.9.2 3D Volumetric strain technique On the contrary, the 3D volumetric strain technique considered local tissue expansion quantified using strain fields – a microscopic view of the expansion. The Green-Lagrangian strain fields calculated using DIC in 2D Cartesian coordinates were translated into 3D volumetric strain fields in cylindrical coordinates, which were used to calculate the total volume of tissue expansion. The volumetric strain 𝜀9 = ∆𝑉𝑉 , (17) is expressed as a ratio of the change in volume to the original volume. The 2D strains and deformations in Cartesian coordinates are used to calculate the 3D volumetric strain in cylindrical coordinates. The 3D volumetric strain is then used to calculate the volume of tissue 42  expansion. The x- and z- directions of the OCT cross-section in Cartesian coordinates are taken to be aligned with the r- and z- directions in the cylindrical coordinates (with the z- axis aligned with the central axis of the microneedle). The change in volume (or expansion of tissue)  ∆𝑉 = 𝜋𝜀9,2‚𝑟2‚𝛿𝑟𝛿𝑧‚2  (18) is calculated using volumetric strain where 𝑖 and 𝑗 span the region of interest chosen for the DIC algorithm. The change in volume, ∆𝑉, is calculated for each subset and the values of 𝛿𝑟 and 𝛿𝑧 depend on the subset spacing chosen for DIC. The distance between the central axis of the microneedle and the center of the subset is denoted by 𝑟2‚. The 3D volumetric strain for each subset is calculated using the sum of strain invariants (𝐼 , 𝐼-, 𝐼E): 𝜀9 = 𝐼 + 𝐼- + 𝐼E (19) 𝐼 = 𝜀33 + 𝜀66 + 𝜀ˆˆ (20) 𝐼- = 𝜀33𝜀66 + 𝜀66𝜀ˆˆ + 𝜀ˆˆ𝜀33 − 𝜀36- − 𝜀6ˆ- − 𝜀ˆ3- (21) 𝐼E = 𝜀33𝜀66𝜀ˆˆ − 𝜀33𝜀6ˆ- − 𝜀66𝜀ˆ3- − 𝜀ˆˆ𝜀36- + 2𝜀36𝜀6ˆ𝜀ˆ3 (22)  Using the axisymmetric assumption, where 𝑢ˆ = 0 and ssˆ [	] = 0, the 3D volumetric strain in cylindrical coordinates is given by: 𝜀9 = 𝜀33 + 𝜀66 + 𝜀ˆˆ + 𝜀33𝜀66 + 𝜀66𝜀ˆˆ + 𝜀ˆˆ𝜀33 − 𝜀36- + 𝜀33𝜀66𝜀ˆˆ − 𝜀ˆˆ𝜀36- (23)  The components 𝜀33, 𝜀66 and 𝜀36 in cylindrical coordinates correspond to 𝐸55, 𝐸66 and 𝐸56 in Cartesian coordinates derived from DIC. The value of 𝜀ˆˆ  is given by t‹3  for an axisymmetric case, and 𝑢3 can be related to the displacement 𝑢5 derived from DIC by changing 43  the signs of 𝑢5 in the left half of the image: the Cartesian coordinates in the DIC algorithm had the origin at the top left corner with the x-axis pointing to the right, while for the cylindrical coordinates the line r = 0 passed through the central axis of the microneedle (center of the image) with r-axis pointing to the right in the right half of the image and pointing to the left in the left half of the image. For small strains 𝜀2 ≪ 1, the 3D volumetric strain reduces to the first strain invariant, which is equal to the trace of the infinitesimal strain tensor [55]. However, during the initial part of the injection we observed strains greater than 0.1. Therefore, the small strain assumption was not used and the complete equation (23), including higher order terms, was used.  We determined the volume of tissue expansion, using both the techniques (surface deformation and volumetric strain), for each OCT image of InjA, which provided a volume estimate for the two techniques every 58 ms. Since each OCT image was a 2D cross-sectional image at the approximate mid-section of the microneedle, we estimated the volume by assuming the tissue structure and its deformations were axisymmetric around the central axis of the microneedle. This axisymmetric assumption allowed us to estimate the total volume of tissue expansion at a rate equal to the frame rate of OCT acquisition.  2.9.3 Technique validation – DIC and volume of expansion To validate the technique used for calculating the volume of expansion and to check for any errors in the custom MATLAB codes, we considered a simple test case. A 2D image of a pattered circle with known radius (𝑟2 = 300	𝑝𝑖𝑥𝑒𝑙𝑠), was stretched in ImageJ with a final radius (𝑟% = 325	𝑝𝑖𝑥𝑒𝑙𝑠) and rotated by 10 degrees, as shown in Figure 2.13.  44   Figure 2.13: Images used for validating volume expansion calculations, showing original patterned circle (left), expanded to increase radius from 300 pixels to 325 pixels (middle), and rotated by 10 degrees (right).  Digital image correlation was performed on the expanded and rotated images, taking the original image as the reference. The displacement and strain fields were used to calculate the 2D strain 𝜀-8  and 3D volumetric strain 𝜀9, using equations (13) and (23) respectively.   Figure 2.14: Strain fields Exx, Exz and Ezz for the expanded and rotated test sample. The scale bar is 600 pixels, which is the size of the diameter of the circle.   45  The strain fields 𝐸55, 𝐸56 and 𝐸66, as a result of performing DIC on the test images, are shown in Figure 2.15. Since the expansion in the image was prescribed (from a radius of 300 pixels to 325 pixels), the theoretical change in area ∆𝐴k and theoretical change in volume ∆𝑉k  were calculated as ∆𝐴k = 𝜋Ž𝑟%- − 𝑟2- = 4.91 × 10<	𝑝𝑖𝑥𝑒𝑙- (24) ∆𝑉k = 43𝜋Ž𝑟%E − 𝑟2E = 3.07 × 10’	𝑝𝑖𝑥𝑒𝑙E (25)  The 2D strain 𝜀-8  and 3D volumetric strain 𝜀9 were used to estimate the change in area and volume. The volume expansion was assumed to be axisymmetric, similar to the injection experiments. The results for the technique validation are summarized in Table 2.2. The percentage error for the estimated volume change (or volume of expansion) using the DIC results from the expanded image (with the original image as reference) was 2.28 %, while the percentage error for the estimated volume change using the DIC results from the expanded and rotated image (with the original image as reference) was 5.53 %. The percentage errors for the change in area calculated for the expanded image and the expanded + rotated image were 3.67 % and 3.05 % respectively.   Table 2.2: Comparison of theoretical and calculated changes in area and volume of test image.   Expansion only Expansion + rotation  Theoretical Estimated % error Estimated % error ∆𝐴 4.91 x 104 pixel2 5.09 x 104 pixel2 3.67 % 5.06 x 104 pixel2 3.05 % ∆𝑉 3.07 x 107 pixel3 3.14 x 107 pixel3 2.28 % 2.90 x 107 pixel3 5.53 % 46  2.10 Correlation maps To qualitatively evaluate the regions of motion in the tissue, we created correlation maps of the OCT images, which showed how different segments of subsequent OCT images (interrogation windows) were correlated with one another. Each OCT image was divided into square interrogation windows of edge length 3 pixels, and 2D correlation coefficients between interrogation windows in consecutive images were calculated (using the corr2 function in MATLAB). To reduce the effect of noise, a 3-point running average was taken of 3 consecutive correlation maps. A correlation coefficient of 1 indicated that the image segments were identical (or tissue was stationary), while a value of 0 or -1 indicated that the image segments were completely uncorrelated or completely anti-correlated, respectively.    Figure 2.15: Example of a correlation map for InjA (5 s after MN retraction), showing the deforming tissue in dark pixels and the stationary tissue in bright pixels; Pin = 100 kPa, the injected fluid is water. The scale bar is 0.5 mm.   The distribution of correlation coefficients was scaled between 0 and 1 to a 16-bit grayscale image, with bright pixels representing stationary tissue and dark pixels representing deforming tissue or noise. While noise was more prominent in the upper dermis, where the OCT 47  signal was attenuated due to optical scattering, the correlation maps in the lower dermis distinguished stationary tissue from deforming tissue. The time evolution of the regions of deforming and stationary tissue qualitatively showed the growth of the region of injection.   48  Chapter 3: Experimental Results This section includes the experimental results of intradermal injections into porcine skin tissue (ex-vivo) and the observations of fluid flow-rate and internal tissue structure of the skin during the injections.   3.1 Microneedle insertion As described in section 2.4, after mounting the skin sample in the tissue holder, the position of the tissue holder is adjusted using a linear stage with a standard micrometer (Figure 3.1), to ensure similar insertion depths between different samples. After filling the tubing, sensors and microneedle with fluid, the microneedle is impacted onto the skin sample to penetrate the stratum corneum.    Figure 3.1: Setup before insertion (left) with the spring compressed; and after insertion (right) with the spring relaxed and the retraction arms support the microneedle sub-assembly. The scale bars are 10 mm.   49  As shown in Figure 3.1, the spring, which provides the impact energy for the microneedle insertion, is compressed before insertion. At this stage, the microneedle is filled with the injection fluid. To insert the microneedle into the skin, the spring release switch is pushed such that the moving sub-assembly including the microneedle (white 3-D printed parts) is free to move inside the fixed cylindrical housing (blue 3-D printed parts). The stored energy in the spring moves the sub-assembly towards the tissue holder, with a final impact velocity 𝑣2C of approximately 2 m/s. The corresponding OCT images before and after inserting the microneedle into the skin sample are shown in Figure 3.2. The thin skin sample is fixed onto the PDMS backing using double-sided tape, which can be seen between the PDMS and the dermis layer of the skin. The PDMS backing provides support to the skin during insertion to enable microneedle penetration. This support provides stability to the soft skin sample, preventing the skin from displacing vertically during microneedle insertion and retraction.      Figure 3.2: OCT images before (top) and after (bottom) insertion of the microneedle into porcine skin sample. The scale bars are 0.5 mm.  50  As shown in Figure 3.1, after microneedle insertion, the spring is relaxed and the moving sub-assembly (white 3-D printed parts inside the fixed blue housing) is held by the retraction arms. The retraction arms are connected to the linear motorized stage and control the movement of the sub-assembly (with the microneedle) in the vertical direction. Since, the linear motorized stage is connected to the computer and controlled using the LABVIEW interface (Appendix C  shows the LABVIEW interface), the retraction protocol can be adjusted based on the experiment: the retraction speed, the number of successive retractions and the interval between them can be controlled within the operating range of the actuator that has a maximum speed of 7 mm/s as provided in Table 2.1.   3.2 Flow-rate and pressure during injections After piercing the stratum corneum of the skin during impact, the microneedle tip is positioned in the dermis layer of the skin for injecting fluid, as shown in Figure 3.2. A typical injection, with the applied input pressure 𝑃2? and measured flow-rate 𝑄, is shown in Figure 3.3. An input pressure 𝑃2? of 100 kPa is first applied, which results in a large transient increase in flow-rate. This initial spike in flow-rate occurs due to the compliance of the tubing in the system, and the microneedle is retracted only after the tubing has fully expanded as a result of the pressure increase. The microneedle is retracted by a distance 𝑑3 of 0.3 mm, 10 seconds after applying the input pressure, marked by the red dashed line in Figure 3.3 at 0 s. The retraction of microneedle is followed by a transient increase in flow-rate, indicating the onset of fluid flow into the dermis. The flow-rate decays to a non-zero steady state value, as the injection continues. When we stop applying the input pressure (at approximately 810 s), the tubing in the system, 51  ahead of the flow-rate sensor, contracts. As the tubing contracts, fluid is quickly pushed back in the opposite flow direction, resulting in a sudden spike in negative flow-rate.   Figure 3.3: Time evolution of input pressure (top – black curve) and measured flow-rate (bottom – blue curve) for a typical injection; the microneedle was retracted by 0.3 mm at 0 s indicated by the dashed red line.   3.2.1.1 Flow-rate offset correction The resting flow-rate measurement, when the input pressure was 0 kPa and there was no flow of fluid, was non-zero due to a zero error. Additionally, the flow sensor seemed to have a slight positive cross-sensitivity to pressure. To account for the offset in the flow-rate measurements and to correct the readings, we recorded the flow-rate when the fluid inside the tubes was stationary. The microneedle was inserted into an impermeable PDMS layer to block the tip of the microneedle, as shown in Figure 3.4. The pressure was cycled between 0 kPa and 100 kPa, and the corresponding flow-rates were recorded.  52   Figure 3.4: MN injection into PDMS. a: Setup showing MN inserted into PDMS (scale bar 10 mm). b, Typical test cycle with pressures alternating between 0 kPa and 100 kPa, and the corresponding flow-rates. c, Plots showing step increase and decrease in pressure, and the corresponding offsets Q0 (0 kPa) and Q0 (100 kPa). 53  Figure 3.4c shows the three controlled/measured pressures: 𝑃2?,k is the target input pressure of the pressure controller (Elveflow OB1), 𝑃2?,C is the value of input pressure measured by the Elveflow OB1 controller, and 𝑃C is the fluid pressure measured before the microneedle (after the flow sensor). The steady-state flow-rates 𝑄c(0 kPa) and 𝑄c(100 kPa) at input pressures, 𝑃2?,k of 0 kPa and 100 kPa, respectively, were recorded to correct for the offset error in the flow sensor. The value 𝑄c(𝑃2?,k) at the input pressure 𝑃2?,k was calculated by averaging the steady-state flow-rates, after the decay of flowrate due to the compliance of the tubing (Figure 3.4c). After each experiment of injections into the skin, the MN was inserted into the PDMS holder, shown in Figure 3.4a, to determine the values of 𝑄c(0 kPa) and 𝑄c(100 kPa). The values of  𝑄c(𝑃2?) were used for each injection experiment to correct the raw flow-rate recordings, 𝑄3“”, as follows: 𝑄 = 𝑄3“” − 𝑄c(𝑃2?) (26)  3.3 Microinjection and microinfusion After the onset of fluid flow due to microneedle (MN) retraction, we identified two modes of flow into tissue, defined here as: microinjection (region of high transient fluid flow-rate; Figure 3.5 inset after 0 s), and microinfusion (region of steady-state flow-rate, lower than that in the transient case; Figure 3.5 after ~10𝑠). The sampling rate of the flow sensor was high enough (100 Hz) to capture the transient changes in fluid flow to enable identification of the two modes. The details of the transient flow-rate immediately following retraction had not been reported earlier either due to lower sampling rates [12] or the use of a syringe pump to maintain a constant flow-rate [14], instead of a constant pressure as in our experiments.  54   Figure 3.5: Two modes of flow - microinjection and microinfusion. Top: Flow-rate of water into excised porcine skin sample, showing the mean (solid blue line) and the standard deviation (blue shading) for five experiments. Water is initially not injected into the skin, even with an input pressure of 100 kPa (light green bar); the microneedle is retracted by 0.3 mm (magenta bar); and water flows into the skin (dark green bar; t = 0 s marking the onset of fluid flow during retraction). When the MN is retracted, the flow-rate increases rapidly (microinjection; inset after 0 s) and decays to a non-zero steady state value (micro-infusion; after approximately 10 s). Bottom: First 20 s of the 5 individual flow-rate measurements.  55  By integrating the flow-rate with respect to time, after the microneedle retraction (at t =  0 s), we calculated the total volume of water injected into the tissue as a function of time, shown in Figure 3.6. The injected volume over time provided a measure of fluid absorption by the skin tissue at different stages of the injection. The tissue absorbed water more rapidly initially during microinjection than during microinfusion, as indicated by the steep slope of the volume-vs.-time curve initially followed by a much more gentle slope (Figure 3.6). As elaborated later in the following sections and in Chapter 4, the different rates of fluid absorption, during microinjection and microinfusion, can be related to variations in the rates of tissue expansion derived from OCT images.    Figure 3.6: Volume of water injected into the skin, calculated by integrating the flow-rate over time, shows rapid fluid absorption by tissue during microinjection (inset) followed by slower fluid absorption during microinfusion. Light green bar represents application of 100 kPa input pressure with no flow, magenta bar represents microneedle retraction by 0.3 mm, and dark green bar represents absorption of fluid by the skin.  56  3.4 OCT imaging of tissue microstructure The OCT measures the intensity of backscattered light from the sample; since water is transparent to the OCT, it only records the scattered light from skin tissue and not from the injected fluid. The greyscale intensity of skin tissue in an OCT image has sufficient contrast to allow tracking the expansion of tissue.    3.4.1 OCT images pre-retraction Before retracting the microneedle, the tissue structure in successive OCT images remained identical, even with the application of 100 kPa input pressure, as shown in Figure 3.7. The corresponding flow sensor measurements indicated that no fluid was injected into the tissue. The OCT images of microneedle insertion for most experiments showed that tissue deformed into the lumen of the microneedle, possibly plugging the flow and forming a mode II ring crack. Upon insertion, the base of the microneedle also represented a mechanically fixed boundary condition preventing the upward movement of the stratum corneum and the swelling of the skin tissue. Figure 3.7 shows two OCT images each for InjA and InjB when input pressure was set to 100 kPa. The two images for each injection set are nearly identical showing that the skin tissue does not deform in response to applied input pressure before microneedle retraction (at t = 0 s).  57   Figure 3.7: Picture and cross-sectional schematic of the microneedle inserted into the skin (top). OCT images of the skin for InjA (middle; using image processing IIIb – wavelet transform based filter and brightness adjustment) and InjB (bottom; using image processing IV - averaging), showing no movement of skin with Pin = 100 kPa. The time t=0 s is when the microneedle is retracted by 0.3 mm. Input pressure was set to 100 kPa a few seconds before retraction. In both sets of images, the input pressure is set to 100 kPa, but the skin tissue does not deform in response to the applied input pressure before retraction. The scale bars are 0.5 mm.    3.4.2 OCT images post-retraction As opposed to pre-retraction, the tissue structure in corresponding OCT images post-retraction changed (Figure 3.8): the injected fluid deformed the tissue. In an OCT image, regions of only water appear dark due to the lack of scatter sources, allowing us to identify the formation 58  of fluid-filled cavities. The movement of the MN base removed the fixed boundary condition at the top and created an air gap, allowing the stratum corneum to move freely upwards and the tissue to swell. The skin tissue behaved like a deformable porous medium, as modelled in [36], [37]. However, the fluid did not form a large cavity, which grew over the course of the injection, as suggested in [36]. Instead, the skin tissue expanded locally to absorb the injected fluid (water), as shown in Figure 3.8.  Figure 3.8: Sequence of OCT images (InjA, IIIb – top 2 rows; captured at 17.2 frames/s) showing the microneedle (yellow dotted lines) retracting by 0.3 mm at 0 s, and showing the deformation of skin tissue after 0 s. The surface of the skin, the stratum corneum (white dashed lines), moves upwards towards the base plate of the microneedle and then moves laterally outwards. OCT images (InjB, IV – bottom) captured at a higher frame rate (73.9 frames/s) than InjA, showing skin tissue expansion during initial injection after retracting the microneedle by 0.3 mm at 0 s. Scale bars (black) in a, b, and c are 0.5 mm.  The OCT images showed the evolution of the tissue expansion and the location of the skin surface over time. Initially, the region of locally expanded tissue was limited to a small area near the MN tip, with the surrounding tissue structure in the lower dermis un-expanded. As the 59  injection progressed, the initial region of expanded tissue stayed stationary while the adjacent tissue expanded; this region of expanding tissue moved away from the microneedle tip increasing the region of expanded tissue. The rate of tissue expansion differed between different parts of the injection. Initially, rapid tissue expansion occurred when the flow-rate was high (microinjection). Then as the free surface of the skin bulged upwards and reached the microneedle base, where it flattened and then grew laterally, the flow-rate decayed to the much lower steady-state value, which was associated with slow tissue expansion. Some anatomical features such as hair shafts and vessels can also be seen in the OCT images, as shown in Figure 3.9.    Figure 3.9: Sequence of OCT images (InjC, IIb – top 2 rows) showing initial injection of water accompanied by rapid expansion of tissue (microneedle retraction of 0.3 mm at 0 s), followed by gradual expansion. OCT images (InjD, IIb – bottom) showing a gap created when the microneedle is retracted by 0.3 mm at 0 ms. The gap does not form a fluid-filled cavity that grows during the injection. The scale bars (white) are 0.5 mm.  60  All the injections, whose OCT images are displayed thus far (InjA – InjD), used water as the injection fluid and retracted the microneedle by 0.3 mm at 0 s. However, Figure 3.10 shows OCT images for an injection of a viscous fluid into the skin sample, with multiple retractions (retractions of 0.2 mm at 0 s, 6.85 s, and 21.0 s). The fluid used for the injection was a sugar solution, with 65 degrees Brix (65 ˚Bx). Degrees Brix is a measure of concentration of sucrose in an aqueous solution, where 1 ˚Bx is 1 gram of sucrose in 100 grams of solution. The fluid had a viscosity of approximately 100 mPa.s (since 65 ˚Bx sugar solution at 25 ˚C and 85% purity of sugar has a viscosity of 97.9 mPa.s). Even with a more viscous fluid, the tissue expands to absorb the fluid, as in the case of water. After each retraction, the tissue expansion is rapid, similar to the initial retraction at 0 s. The injection shown in Figure 3.10 was done on a thicker piece of skin than the ones described earlier, and the stratum corneum is not visible in the OCT images.     Figure 3.10: Injection of viscous fluid (65 ˚Bx) into the skin, with 0.2 mm retractions (each) at 0 s, 6.85 s, and 21.0 s. The scale bar (white) is 0.5 mm.   61  3.4.3 3D OCT images OCT can also be used for 3D imaging of the skin tissue at the micron-level. Figure 3.11 shows an example of the sections from a 3D OCT image taken after injecting water (InjC). To capture the 3D images in Figure 3.11, the OCT probe was pointed towards the stratum corneum of the skin, unlike in the real-time imaging of injections shown earlier, where the OCT probe was pointed towards the PDMS and dermis. During the injections, imaging through the stratum corneum side was not possible because the near-infrared light from the OCT cannot pass through the metal microneedle and its metal base. Hence, real-time imaging was only possible through the dermis (and PDMS), and imaging the skin through the stratum corneum was possible after removing the MN.    The top left image in Figure 3.11 shows the XZ plane of the 3D image. This is the plane used for imaging the injections in real time. Note that the location of microneedle insertion is towards the left end of the image in the XZ plane (indicated by the hole created by the microneedle), unlike the other OCT images of injections presented in the thesis where the location of the microneedle was mostly centered in the image. The Y axis on the XZ plane image is positive out of the page. The vertical yellow line in the first image indicates the location of the YZ plane (shown on the right). The three horizontal lines (1 – red, 2 – yellow, and 3 – blue) in the XZ and YZ plane represent the locations for the three images in the bottom (in the XY plane). The hole created by the microneedle can be seen in image 2 (yellow) of the XY plane (located by the intersection of the vertical and horizontal yellow lines). In image 2 and 3 (XY plane), the dark pixels inside the bright region indicate fluid filled cavities inside tissue, respectively. The increasing size of the bright region (tissue cross-section) from 1 to 3 in the XY 62  plane, corresponding to the horizontal lines (1 to 3; top to bottom) in the XZ and YZ planes, indicate the swelling of the skin after injection.      Figure 3.11: 3D imaging of the skin after injection using OCT. The top panels show XZ and YZ planes, showing the hole created by the microneedle (along the vertical yellow line). Horizontal lines (1 red, 2 yellow and 3 blue) indicate locations for the XY plane sections, shown in the bottom panels. Scale bars: 0.5 mm.   63  Chapter 4: Skin tissue deformation The data for intradermal injections from the OCT images and the sensor measurements are analyzed, in this chapter, to relate the skin tissue deformation with the injected fluid flow. Firstly, the strain in skin tissue is calculated to quantify tissue deformation using digital image correlation on OCT images. Secondly, the strain values and OCT images are used to estimate the total volume of tissue expansion, and relate it to the volume of injected fluid. Next, the growth of the expanded region of skin tissue is evaluated using correlation maps. Lastly, the effect of multiple and continuous retractions on tissue expansion and fluid flow is described.   4.1 2D strain fields To quantify the tissue deformation and to relate it to the injected volume, we calculated the Green-Lagrangian strain in the tissue using digital image correlation of the OCT images. We considered two settings: one with a higher frame rate (InjB - 73.9 frames/s) and smaller region of interest (ROI) to obtain strain fields at high temporal resolution during initial injection, and another with a lower frame rate (InjA - 17.2 frames/s) and larger ROI to compare results from flow sensor and OCT over a longer period (50 seconds). In both cases, the ROI excluded some parts of the upper dermis/epidermis where the contribution of noise was significantly higher than for the lower dermis because the OCT signal was attenuated by optical scattering.   4.1.1 Sample DIC data  Figure 4.1 shows two sets of displacement and strain results, at 1 s and 3.5 s, for InjB. Appendix A  lists the scan parameters for the OCT acquisition, and the OCT images for InjB were acquired at high frame rates (73.9 frames/s). The top panel of the figure shows a sample 64  OCT image for InjB, and the region of interest for DIC. The first two panels of the results show the displacement fields, 𝑢5 and 𝑢6, in the horizontal (x-) and vertical (z-) directions, respectively. Positive values (red) indicate displacement in the direction of the (x- or z-) axis, while negative values (blue) indicate displacement in the opposite direction. Since the x- axis points to the right and has its origin at the top left corner, the blue regions in the left half of the 𝑢5 displacement fields are moving to the left and the red regions in the right half is moving to the right – away from the injection site, as expected.   The Green-Lagrangian strain components 𝐸55 and 𝐸66 are longitudinal strains and inform about the increase or decrease of length segments originally oriented (in the reference configuration) along the x- and y- directions, respectively. 𝐸55 and 𝐸66 have positive values (red) if the interrogation window elongates in the corresponding (horizontal or vertical) direction, and negative (blue) values for compression in that direction. The stretch ratio or stretch l is defined, along a particular direction, as the ratio of the length of a deformed line segment to its original length, and can be related to the longitudinal strain [55] 𝜆5 = •1 + 2𝐸55, (27) 𝜆6 = •1 + 2𝐸66, (28)  with 0 < 𝜆 < ∞. The unit elongation along x- and z- directions, ex and ez, can also be related to the longitudinal strains from the Green-Lagrangian or material strain tensor  𝜀5 = 𝜆5 − 1 = •1 + 2𝐸55 − 1, (29) 𝜀6 = 𝜆6 − 1 = •1 + 2𝐸66 − 1. (30) 65   The strain component 𝐸56 is called angular strain and it informs about the variation of angles between orthogonal segments oriented originally along x- and z- directions. If 𝐸56 is positive, the angle between the originally orthogonal line segments, along x- and z- directions, decreases. On the other hand, if 𝐸56 is negative, the angle between the orthogonal line segments increases. The increment of angle ∆θ˜™ = 𝜃% − 𝜃2 = −𝑎𝑟𝑐𝑠𝑖𝑛 w 2𝐸56•1 + 2𝐸55•1 + 2𝐸66z, (31) or the difference between the final angle (𝜃%) and the original angle in the reference configuration (𝜃2 = 𝜋/2), can be related to the strain values [55]. The negative sign results in a decrease of angle for a positive value of 𝐸56 and an increase of angle for a negative value of 𝐸56. In Figure 4.1 the red values indicate positive values of 𝐸56 or a decrease in angle, and the blue values indicate negative values of angular strain or an increase in angle.   The bottom panel of Figure 4.1 shows the 2D strain fields, 𝜀-8 , calculated in MATLAB using equation (13). The magnitude of 2D strain, 𝜀-8 , represents a relative increase/decrease in area of the corresponding interrogation window considered for the DIC algorithm, and provides a measure for local tissue expansion/compression in 2D. The strain magnitude is proportional to the rate of tissue expansion/compression because the time interval between OCT frames used for DIC is constant for each injection. The strain fields provide a distribution of tissue that is either locally expanding, compressing or stationary, with the magnitude representing the extent of tissue deformation – also proportional to the rate of deformation. We use these 2D strain fields to compare the local tissue deformation at different times during the injection, as described in the next section.  66   Figure 4.1: Displacement and strain fields for InjB (cumulative results for a time interval of 108 ms, considering 8 pairs of averaged OCT images). OCT image (top) shows ROI in orange dashed lines, MN in yellow dotted lines, and SC in white dashed lines. The displacement fields ux and uz, and strain fields Exx, Exz, and Ezz are calculated using DIC in Ncorr. The 2D strain fields ɛ2D are calculated in MATLAB using equation (13).  The scale bars are 0.5 mm.   67  4.1.2 2D strain fields during intradermal injections Figure 4.2 shows the 2D strain in skin tissue during the initial phase of the injection, calculated for InjB (captured at a high frame rate of 73.9 frames/s). The magnitude of strain was higher for the first few seconds than later during the injections, indicating a high initial rate of tissue deformation (during microinjection) followed by a lower rate (during microinfusion); positive values of strain represent regions of tissue expansion, while negative values represent regions of tissue compression. Since we assumed incompressibility of individual solid and fluid components of the porous medium, here, tissue compression implies a decrease in void fraction of the fluid in that region or a decrease in local porosity in that region of the tissue. Likewise, tissue expansion indicates an increase in the void volume or local porosity of that particular region in the tissue. Each strain map in Figure 4.2 shows cumulative strain using DIC results on 8 pairs of averaged OCT images, with a total time difference between the reference OCT frame and the last OCT frame of 108 ms, as described earlier in section 2.8.1.   Figure 4.2: 2D maps of strain (ɛ2D) for the region bounded by orange dashed lines in the OCT image (left). Regions of local tissue expansion (positive strain) are shown in red, while regions of local tissue compression (negative strain) are shown in blue. OCT images captured at 73.9 frames/s; microneedle retracted at 0 s by 0.3 mm; 𝑷𝒊𝒏= 100 kPa; injected fluid is water. The scale bars (black) are 0.5 mm.  68  Initially, the deformation of tissue was limited to a region near the MN tip. A ‘V’ shaped region of high positive 2D strain near the MN indicated the site where tissue absorbed fluid, by locally expanding (Figure 4.2 and Figure 4.3). The ‘V’ shaped region of high positive strain, corresponding to high tissue expansion, increased in size with time, moving downwards and sideways (Figure 4.2). The initial strain fields during an injection indicated that initially tissue near the MN tip expanded locally, and then flow through this previously expanded tissue further expanded the surrounding tissue. Thus, the region of high tissue expansion, representing the boundary of fluid saturation, moved away from the MN tip as an injection continued. We observed a similar decrease of strain magnitude and growth of the expansion region over a longer period, as shown in Figure 4.3.    Figure 4.3: Strain maps (ɛ2D) overlaid (50% transparency) on OCT images, considering a region of interest for DIC as the region of the skin before retraction. Each strain map shows cumulative strain using DIC results on 9 pairs of consecutive OCT images, with a total time difference between the reference OCT frame and the last OCT frame of 0.52 s. The ‘V’ shaped region of high tissue expansion grows laterally outwards. The rate of tissue expansion decreases over time (for instance, limits of strain at 0.6 s are an order of magnitude larger than those at 10.6 s). OCT images captured at 17.2 frames/s; Pin = 100 kPa and dr = 0.3 mm. The scale bar (black) is 0.5 mm. 69  Here, the fixed lower boundary prevented the downward movement of the ‘V’ shaped boundary, such that the ‘V’ shape widened and moved mostly laterally outwards. Figure 4.3 also shows that as the ‘V’ shaped region of tissue expansion widens, the stratum corneum (white dashed lines) swells and similarly moves laterally outwards.  4.2 Volume of tissue expansion The total volume of tissue expansion was calculated from the OCT images using 3D volumetric strain and surface (SC) deformation, as described in section 2.9. The tissue deformations recorded in 2D were assumed to be axisymmetric around the central axis of the MN. The total volume of tissue expansion, calculated using these two methods, in Figure 4.4 shows a similar trend as compared to that of the injected volume. A high rate of total tissue expansion and surface deformation indicated by a rapid increase in volume, corresponds to high rates of injected volume. Thus, high fluid absorption is associated with high local tissue expansion, as seen initially during microinjection. A similar link between fluid absorption and tissue expansion was previously observed by modelling skin tissue as deformable porous medium [36], where fluid absorption by the tissue was associated with an overall expansion of the medium and an increase in the local permeability.   The volume of injected water calculated from the flow sensor was higher than the volume of tissue expansion calculated from strain or surface deformation. The volume of tissue expansion from strain was lower than the injected volume because the ROI for DIC was limited, not accounting for local tissue deformation outside the ROI. Both estimates, from strain and 70  surface deformation, also ignored any flow through the un-deformed porous network, since the techniques only account for expanded tissue and cannot measure flow directly.     Figure 4.4: The time evolution of total volume of tissue expansion derived from 3D volumetric strain (orange curve; OCT-DIC) and from surface deformation technique (red curve; OCT-SD) have similar profiles to that of total injected volume (purple curve; Q Sensor).   4.3 Correlation maps To qualitatively observe the evolution of the boundaries between stationary tissue and deforming tissue, we computed correlation maps of the OCT images that provide 2D representations of how similar consecutive images are, as described in section 2.10. In each correlation map, a high correlation coefficient, represented by bright pixels, corresponds to stationary tissue, while a low correlation coefficient, represented by dark pixels, corresponds to deforming tissue or noise. Figure 4.5 shows an example of the correlation maps for InjA at 0.5 s, 5.0 s and 14.0 s. Unlike DIC, the correlation maps calculate only correlation coefficients (not strain) and are computationally faster to achieve. The correlation maps show the regions of 71  motion more clearly than the strain maps, especially when the gradients of strain between expanding and stationary tissue are low, such as during microinfusion.   The thin region of low correlation coefficient between regions of high correlation coefficient (i.e., dark band between bright pixels of tissue at 5.0 s and 14.0 s in Figure 4.5) resembles the ‘V’ shaped region of high strain obtained from DIC, and the movement of this region over time qualitatively shows the growth of the expanded tissue region (bottom right in Figure 4.5). The region of deforming tissue highlighted in the first image of Figure 4.5 successively turns stationary (i.e. the correlation coefficient changes from a low to a high value) as the region of expanded tissue grows, indicating that fluid expands the tissue up to a maximum limit that is most likely given by the injection pressure.    Figure 4.5: Correlation maps for InjA; Pin = 100 kPa; dr = 0.3 mm. Bottom right: Evolution of the lower boundary between expanded and stationary tissue tracked over time, showing growth of the expanded region. Successive dashed (colored) lines are 2.9 s apart, with the first (red) dashed line starting 2s after microneedle retraction; the microneedle position is shown by dotted gray lines. The black scale bars are 0.5 mm. 72  We observed similar results for the growth of the expanded region for multiple injection experiments, as shown in Figure 4.6 for 5 datasets. The shape of deforming tissue, indicated by dark pixels, was similar for all the injection tests. The first four sets (InjA – InjD) are for injection of water into tissue and a retraction distance of 0.3 mm, the last set (InjE) is for injection of viscous fluid (65 ˚Bx sugar solution) and a retraction distance of 0.2 mm. The amount of deforming tissue reduces from 1 s to 5 s for all injections.   The ‘V’ shape of the expansion region could be a result of many factors including the boundary conditions (fixed boundary condition at the PDMS-dermis interface and impermeable boundary condition at the surface), the compression of the skin by the microneedle and the anisotropy [56] of the skin tissue. Ideally, for an isotropic, homogeneous, and infinite porous medium, one would expect the region of expansion from a point source of fluid to be spherical and to grow radially outwards [36].   For an isotropic finite porous medium (or semi-infinite – bounded by the top and bottom), that is homogeneous and deformable, as in the case of [37], an injection of fluid from a point source below the surface results in an expansion of the region near the point source and a consequent lifting of the surface. As reported in [37], the boundary conditions of the finite porous medium affect the shape of expansion and the contours of pressure in the tissue. Our experiments include injections into the skin, which is an anisotropic, non-homogenous and finite porous medium. We expect the region of expansion to be affected by the boundary conditions and the anisotropy of the skin tissue.   73   Figure 4.6: Correlation maps for 5 sets of injection experiments (at 1 s and 5 s), showing ‘V’ shaped bands of low correlation (dark pixels) indicating movement of tissue. Bright pixels indicate stationary tissue. Pin = 100 kPa; dr = 0.3 mm (InjA – InjD) and 0.2 mm (InjE); Injected fluid = water (Inj A – InjD), and 65 ˚Bx sugar solution (InjE). The black scale bars are 0.5 mm.  74  4.4 Multiple/continuous retractions After a single step retraction of the microneedle, the skin tissue absorbs fluid by expanding, and we observed that the expansion (and consequently the fluid flow-rate) decreases during microinfusion, as the stratum corneum reaches the base of the microneedle. To further test the link between microneedle retraction, tissue expansion and fluid flow-rate, we evaluated the effect of multiple retractions and continuous microneedle retraction.   4.4.1 Multiple retractions We compared the injection following a single retraction of 0.3 mm to that associated with multiple successive retractions (8 retractions of 0.1 mm each, in 20 second steps). The behavior of fluid flow and tissue deformation after each successive retraction resembled that after a single retraction, as shown in Figure 4.7. After each successive retraction of 0.1 mm, the flow-rate increased significantly and decayed towards a steady-state value higher than that after previous retractions.   At each retraction, the fixed boundary condition at the stratum corneum was removed, which was followed by the swelling of the tissue due to fluid absorption and the upward movement of the stratum corneum to the MN base. Each time when the free surface reached the MN base, the MN base provided a mechanically fixed boundary, which limited tissue expansion in that direction. Thus, limiting/restricting tissue expansion reduces the fluid flow-rate or fluid absorption, and removing the fixed boundary restricting expansion – allowing tissue to expand – increases fluid flow successively.  75   Figure 4.7: Top: Flow-rate after retraction of 0.3 mm at 0 s, followed by 8 retractions of 0.1 mm (spaced 20 s apart) starting at 300 s. After each retraction, the flow-rate increases rapidly (microinjection) and then decays to a value greater than that of the previous retraction; Pin = 100 kPa. Bottom: The volume of fluid injected is higher for 8 successive retractions of 0.1 mm than for a single retraction of 0.3 mm (calculated by integrating the curve in b from 300 s to 600 s, and from 0 s to 300 s, respectively).  The final steady-state flow-rate and injection volume following 8 successive retractions of 0.1 mm each was higher than for a single retraction of 0.3 mm, as shown in the bottom graph of Figure 4.7. Previous studies [12] have shown that, for a single retraction, a larger retraction distance resulted in a higher injection flow-rate (and volume). Thus, we expected a cumulative retraction of 1.1 mm (0.3 mm for single retraction + 0.8 mm for multiple retractions) to result in a higher flow-rate than a 0.3 mm retraction. However, a single (step) retraction of 1.1 mm was 76  not possible because the height of the microneedle used in our experiments was 0.7 mm, and the microneedle would have retracted completely out of the skin.   By retracting the microneedle successively, and allowing tissue to swell/expand after each retraction, we achieved a cumulative retraction higher than that possible by a single step retraction, resulting in a greater volume of injection. We observed similar behavior of increased flow-rate after each successive retraction for other injections. Figure 4.8 shows the time evolution of flow-rate after 8 successive retractions of 0.1 mm each, showing a sudden increase in flow-rate after each retraction. For all three datasets, the successive retractions (of 0.1 mm each) started around 300 s after a single step retraction of 0.3 mm.    Figure 4.8: Flow-rate after multiple retractions of 0.1 mm each for 3 different sets of injections.   The flow-rates of InjG, InjH and InjI are slightly different from one another possibly due to differences in the skin thickness, insertion depth and/or boundary conditions between the skin samples, as shown in Figure 4.9. For instance, the insertion depth for InjG was the largest among the three. For InjG, the microneedle was inserted into the skin around 0.2 mm deeper than the 77  free surface of the skin, unlike for InjH and InjI, where the microneedle base was almost leveled with the free surface of the skin. This could explain why InjG had the lowest curve among the three injections. The thickness of the skin samples were similar, with InjH being thicker (around 1.2 mm) than InjG and InjI (around 1.1 mm).   Additionally, the lower part of the dermis for InjI was not completely bonded to the PDMS like the other two samples, as seen in Figure 4.9. Due to a lack of complete bonding of the dermis, the PDMS did not form a fixed mechanical boundary at every location of the lower dermis. During injection the lower left surface moved in response to the injection, and allowed a small space between the dermis and the PDMS for the tissue to expand. This additional space for expansion could likely be the reason why InjI has higher flow-rate than InjG and InjH. Note that the microneedles in the OCT images of InjH and InjI cannot be seen clearly since the OCT image does not pass through the central axis of the microneedle for the two cases.   Figure 4.9: OCT images for InjG, InjH and InjI considered for the multiple retraction plot. InjG has a deeper insertion depth than InjH and InjI; InjH is around 1.2 mm thick, while InjG and InjI are around 1.1 mm thick; the dermis of InjI was not completely bonded, allowing it to move during injection. The black scale bars are 0.5 mm.  78  4.4.2 Continuous retraction Similar to multiple retractions, slow and continuous retraction of the microneedle resulted in higher injection volume as compared to a single step retraction. Figure 4.10 shows the correlation maps and pictures of the tissue holder for InjF. The microneedle is slowly retracted at a speed of 0.1 mm/s up to a distance of 1 mm. The first two time frames shown are during retraction, while the last two are after the completion of retraction (1 mm). The injected fluid, water, is mixed with green dye for demonstrating the uptake of fluid from the outside. Green dye was not used for the other experiments described in the paper. The injection pressure was set to 100 kPa.   The continuous retraction of 1 mm would not have been possible with a single step retraction, since the height of the microneedle used was 0.7 mm. Like the successive retractions described earlier, the continuous retraction allows the tissue to swell as the microneedle is retracted, prolonging the microinjection mode of flow and resulting in a greater flowrate (and volume) than a single retraction case.  The swelling of the stratum corneum and uptake of water, as seen in Figure 4.10, results in the formation of a raised papule.   The correlation maps for InjF show the regions of deforming and stationary tissue. The first two correlation maps for 2 s and 6 s can be compared with those for the other injections (InjA – InjE). As shown earlier in Figure 4.6, for the other injections (InjA – InjE) with step retraction of the microneedle (0.2 mm – 0.3 mm), the region of deforming tissue decreased from 1 s to 5 s, forming a thin band of deforming tissue (‘V’ shaped) at 5 s. However, for the continuous retraction case (InjF), the region of deforming tissue increases from 2 s to 6 s, rather than 79  decreasing, because at 6 s the microneedle is still retracting – the retracting microneedle allows the skin to continuously expand and absorb fluid, rather than acting as a barrier for tissue expansion (like in the step retraction case).  The narrow band of deforming tissue, separating the expanded tissue and stationary tissue, can be seen at 12 s and 16 s in Figure 4.10. The top part of the correlation maps showing low correlation coefficients can be attributed to noise because in the top half of the image, the OCT signal was almost completely attenuated. The region of expanded tissue is much larger in this case, covering the skin sample almost completely, unlike for the step retraction case for InjA (Figure 4.5), where at 14 s the region of expanded tissue was bounded by the ‘V’ shape.   The corresponding flow-rate and retraction distance measurements for the continuous retraction case (InjF) are shown in Figure 4.11. The microneedle was retracted slowly at 0.1 mm/s to a total distance of about 1 mm. The flow-rate kept increasing as the microneedle continued to retract and created more space for the skin surface to swell. A spike in flow-rate was observed at around 1.5 s. Looking at the corresponding OCT images, the spike corresponded to an injection of trapped air, which rapidly increased flow-rate at 1.5 s.   80   Figure 4.10: Correlation maps and pictures of the microneedle and tissue holder for InjF. The microneedle was retracted slowly by 1 mm; Pin = 100 kPa; Injected fluid: water + green dye. The correlation maps show regions of deforming tissue (dark pixels) and stationary tissue (bright pixels). The stratum corneum gradually swells as the microneedle is retracted back continuously, while allowing tissue expansion. The scale bars in the correlation maps are 0.5 mm.  81    Figure 4.11: Flow-rate and retraction distance for InjF, showing a continuous retraction of up to 1 mm at a rate of 0.1 mm/s and an increase in flow-rate during the retraction. The spike in flowrate at around 1.5 s is likely due to the injection of an air bubble at the tip of the microneedle, as shown later. Pin  = 100 kPa; Injected fluid = water + green dye.    The injection of air can be seen in OCT images of the injection, as shown in Figure 4.12. The injection of air into skin is usually very rapid, and the corresponding OCT images have significant motion artifacts at the regions of air injection. The injection of air bubbles can also cause rapid rupturing of skin tissue, as shown in the OCT images for InjJ – where the air bubble appeared mid-injection. It is important to note that InjJ was performed with the microneedle and injection system upright, unlike other injections shown where the microneedle was inverted as shown earlier in Figure 2.3. Throughout the thesis, only InjJ and InjE had the setup where the microneedle was upright (and both the injections used 65 ˚Bx sugar solution). We had originally started with an upright system, like in InjJ and InjE, but had encountered the problem of trapped air bubbles moving upwards towards the microneedle and disrupting the flow of fluid mid-injection. We inverted the injection system to eliminate the problem in most cases. InjF was one 82  of the few cases (and the only one presented here) using the inverted system where a trapped air bubble was injected into the skin. The OCT was a useful tool in identifying whether or not any air was injected into the skin based on the motion artifacts produced during the injection of air.     Figure 4.12: OCT images of InjG and InjJ showing the rapid injection of air trapped in the microneedle at the beginning of the injection for InjG and in the middle of an injection for InjJ. The input pressure is 100 kPa for both injections, the injection fluid is water for InjG and 65 ˚Bx sugar solution for InjJ. For InjJ, the injection of air results from an air bubble trapped in the system. Note: the actual orientation of InjJ was upside down, with the microneedle and injection system upright – which could have resulted in the air bubble moving up during the injection.        83  Chapter 5: Conclusions and future work This chapter summarizes the results presented in the thesis, lists the conclusions that can be drawn from the results, describes the applications of the research findings, provides limitations of the experimental work, and suggests possible future research directions.   5.1 Summary of observations We designed and built an experimental setup to observe tissue deformations during intradermal injections, and record flow properties. A summary of the main results and observations from the experiments and their analyses is provided below: 1. Flow sensor recordings:  a. Fluid is not injected into the tissue before MN retraction. b. We identified two modes of flow during intradermal injections -  microinjection at a high transient flow-rate, and microinfusion at a low steady state flow-rate. 2. OCT images:  a. Skin tissue does not deform before MN retraction, when no fluid is injected. b. Initially, during microinjection, skin tissue expands rapidly near the microneedle tip and the stratum corneum moves upwards.  c. Later, during microinfusion, skin tissue expands slowly, the region of expansion grows away from the microneedle, and the stratum corneum moves laterally outwards after reaching the MN base. 3. Strain:  a. A ‘V’ shaped region of high tissue expansion exists during the injection.  84  b. The rate of tissue deformation is higher initially, during microinjection, than later, during microinfusion. c. As the injection continues, the ‘V’ shaped region of high tissue expansion moves away from the MN, increasing the region of expanded tissue 4. Volume of tissue expansion: a. The volume of tissue expansion from strain and surface deformation techniques closely relates to the volume of injected fluid from flow sensor recordings.  5. Correlation maps: a. Initially, a large portion of the tissue deforms, soon after retraction b. Later, tissue deformation is limited to a narrow band – similar to the ‘V’ shaped region of high tissue expansion from strain fields. 6. Multiple or continuous retractions:  a. Increased volume of fluid injected and rapid tissue deformations are achieved during successive and continuous retractions of the microneedle b. The microinjection and microinfusion modes of flow occur after each successive retraction of the microneedle.  c. In the correlation maps for continuous retraction, the large initial region of deforming tissue takes longer to turn stationary as compared to the single retraction case.     85  5.2 Conclusions  Using the observations and results presented in the thesis, the following conclusions can be drawn:  • OCT as a visualization tool [using observation 2]: We demonstrated OCT as a suitable tool for visualizing tissue deformation during intradermal injections and as a potential imaging modality for observing the dynamic behavior of biological tissue with micron-level resolution in real time. • Fluid absorption by skin tissue [using observations 1 and 2]: Skin tissue was found to absorb the injected fluid by locally expanding, without forming a single large fluid-filled cavity. • Skin tissue as a deformable porous medium [using observations 3 and 4]: We found that more fluid was injected when the overall tissue expansion was high, indicating that skin tissue behaved like a deformable porous medium with variable permeability: local tissue expansion due to fluid absorption increased the porosity and permeability of tissue locally, thus increasing fluid flow through that region and aiding further fluid absorption in adjacent regions. • Growth of tissue expansion [using observations 3 and 5]:  Fluid flow through previously expanded tissue further expands surrounding tissue, increasing the region of fluid saturation. • Preventing/allowing tissue expansion [using observation 6]:  We found that limiting the tissue expansion by the MN base, that formed a fixed boundary, reduced the fluid flow-rate into the tissue; and removing this fixed boundary increased the flow-rate. Retracting the microneedle successively or continuously, while allowing the tissue to swell, resulted 86  in a larger cumulative retraction distance (and consequently a larger amount of injected fluid) than was possible for a single step retraction.   5.3 Applications of research findings By taking a closer look at the dynamics of tissue deformation during intradermal injections and understanding the interactions between fluid and tissue, we can better design injection systems to deliver drugs at therapeutically relevant time scales and volumes. The injection protocol of drugs can be controlled based on the therapeutic requirement: a rapid bolus injection of drugs can be accomplished by allowing tissue to freely expand (e.g. by retracting MN successively or continuously, to prolong the microinjection mode); or a slow and controlled infusion of drugs can be accomplished by restricting tissue expansion (e.g. by retracting the MN in a single short step, to maintain the microinfusion mode). Our experimental results can also provide physical insights for mathematical modelling or numerical simulations of flow through biological tissue, especially for intradermal delivery of drugs into skin tissue.  5.4 Limitations of experimental techniques Visualizing the tissue deformations using OCT was only possible for thin skin samples (up to around 1.5 mm), due to the imaging depth of OCT in an optically scattering biological tissue. Additionally, real-time imaging using OCT was limited to one cross-section of the skin tissue, and 3D images of tissue structure could only be taken before or after injections. The movement of fluid (transparent to OCT) in skin tissue cannot be tracked, and the OCT only records the movement of the tissue. Using particles to track the movement of fluid is possible outside the skin tissue, since the particles in the fluid produce speckles in an OCT image. 87  However, the internal structure of the tissue also produces similar speckles, and the particles in the fluid cannot be distinguished easily from the tissue.   5.5 Suggested future work Although OCT has been used to image the internal structure of biological tissue for dermatology and ophthalmology, it has not been widely used for imaging the dynamic processes of skin tissue, such as during intradermal injections. Our work introduced the use of OCT for real-time imaging of skin tissue during intradermal injections, and can motivate similar investigations using OCT on the interaction of biological tissue with its external physical environment. An extensive study observing the effects of many injection parameters on the dynamics of injections, using tools such as OCT, can be performed to advance this work. The effect of injection parameters, such as fluid viscosity, injection pressure, microneedle penetration depth, and microneedle geometry, on tissue deformations and flow-rate can be further studied.   Our work mostly considered the injection of water into skin tissue at 100 kPa, so it would be interesting to see how this differs for injections at higher pressures and with different fluid viscosities. Since the study was limited to thin skin samples (epidermis+dermis), the effect of skin thickness on fluid flow-rate can be studied to compare our results with expected in-vivo injections. Additionally, observing the fluid drainage over time, as opposed to the fluid injection, would also provide meaningful insights. A possible study could look at the effect of injection parameters, such as injection volume, pressure and fluid viscosity, on the fluid drainage rate.   88  References  [1]  G. J. Tortora and B. Derrickson, Principles of Anatomy and Physiology, 7th ed. Harper Collins College Publishers, 1993. [2]  P. H. Lambert and P. E. Laurent, “Intradermal vaccine delivery: Will new delivery systems transform vaccine administration?,” Vaccine, vol. 26, no. 26, pp. 3197–3208, Jun. 2008. [3]  R. Langer, “New Methods of Drug Delivery,” Science, vol. 249, no. 4976, pp. 1527–1533, Sep. 1990. [4]  J. Hickling, K. Jones, M. Friede, D. Zehrung, D. Chen, and D. Kristensen, “Intradermal delivery of vaccines: potential benefits and current challenges,” Bull. World Health Organ., vol. 89, no. 3, pp. 221–226, Mar. 2011. [5]  M. R. Prausnitz and R. Langer, “Transdermal drug delivery,” Nat. Biotechnol., vol. 26, no. 11, pp. 1261–1268, Nov. 2008. [6]  B. Stoeber and D. Liepmann, “Arrays of hollow out-of-plane microneedles for drug delivery,” J. Microelectromechanical Syst., vol. 14, no. 3, pp. 472–479, Jun. 2005. [7]  P. E. Laurent et al., “Evaluation of the clinical performance of a new intradermal vaccine administration technique and associated delivery system,” Vaccine, vol. 25, no. 52, pp. 8833–8842, Dec. 2007. [8]  H. S. Gill and M. R. Prausnitz, “Coated microneedles for transdermal delivery,” J. Controlled Release, vol. 117, no. 2, pp. 227–237, Feb. 2007. [9]  Y.-C. Kim, J.-H. Park, and M. R. Prausnitz, “Microneedles for drug and vaccine delivery,” Adv. Drug Deliv. Rev., vol. 64, no. 14, pp. 1547–1568, Nov. 2012. [10] T.-M. Tuan-Mahmood et al., “Microneedles for intradermal and transdermal drug delivery,” Eur. J. Pharm. Sci., vol. 50, no. 5, pp. 623–637, Dec. 2013. [11] S. A. Ranamukhaarachchi, C. Padeste, M. Dübner, U. O. Häfeli, B. Stoeber, and V. J. Cadarso, “Integrated hollow microneedle-optofluidic biosensor for therapeutic drug monitoring in sub-nanoliter volumes,” Sci. Rep., vol. 6, no. 1, Sep. 2016. [12] W. Martanto et al., “Microinfusion Using Hollow Microneedles,” Pharm. Res., vol. 23, no. 1, pp. 104–113, Jan. 2006. 89  [13] P. M. Wang, M. Cornwell, J. Hill, and M. R. Prausnitz, “Precise microinjection into skin using hollow microneedles,” J. Invest. Dermatol., vol. 126, no. 5, pp. 1080–1087, 2006. [14] J. Gupta, S. S. Park, B. Bondy, E. I. Felner, and M. R. Prausnitz, “Infusion pressure and pain during microneedle injection into skin of human subjects,” Biomaterials, vol. 32, no. 28, pp. 6823–6831, Oct. 2011. [15] J. Gupta, D. D. Denson, E. I. Felner, and M. R. Prausnitz, “Rapid local anesthesia in human subjects using minimally invasive microneedles,” Clin. J. Pain, vol. 28, no. 2, p. 129, 2012. [16] T. Lister, P. A. Wright, and P. H. Chappell, “Optical properties of human skin,” J. Biomed. Opt., vol. 17, no. 9, p. 090901, Sep. 2012. [17] D. Huang et al., “Optical Coherence Tomography,” Science, vol. 254, no. 5035, Nov. 1991. [18] W. Drexler and J. Fujimoto, “State-of-the-art retinal optical coherence tomography,” Prog. Retin. Eye Res., vol. 27, no. 1, pp. 45–88, Jan. 2008. [19] A. Alex et al., “Multispectral in vivo three-dimensional optical coherence tomography of human skin,” J. Biomed. Opt., vol. 15, no. 2, p. 026025, 2010. [20] E. Sattler, R. Kästle, and J. Welzel, “Optical coherence tomography in dermatology,” J. Biomed. Opt., vol. 18, no. 6, pp. 061224–061224, 2013. [21] S. A. Coulman et al., “In Vivo, In Situ Imaging of Microneedle Insertion into the Skin of Human Volunteers Using Optical Coherence Tomography,” Pharm. Res., vol. 28, no. 1, pp. 66–81, Jan. 2011. [22] J. Enfield, M.-L. O’Connell, K. Lawlor, E. Jonathan, C. O’Mahony, and M. Leahy, “In-vivo dynamic characterization of microneedle skin penetration using optical coherence tomography,” J. Biomed. Opt., vol. 15, no. 4, p. 046001, 2010. [23] R. F. Donnelly et al., “Optical coherence tomography is a valuable tool in the study of the effects of microneedle geometry on skin penetration characteristics and in-skin dissolution,” J. Controlled Release, vol. 147, no. 3, pp. 333–341, Nov. 2010. [24] I. Mansoor et al., “A microneedle-based method for the characterization of diffusion in skin tissue using doxorubicin as a model drug,” Biomed. Microdevices, vol. 17, no. 3, Jun. 2015. 90  [25] I. Mansoor, Y. Liu, U. O. Häfeli, and B. Stoeber, “Arrays of hollow out-of-plane microneedles made by metal electrodeposition onto solvent cast conductive polymer structures,” J. Micromechanics Microengineering, vol. 23, no. 8, p. 085011, Aug. 2013. [26] I. Mansoor, U. O. Hafeli, and B. Stoeber, “Hollow Out-of-Plane Polymer Microneedles Made by Solvent Casting for Transdermal Drug Delivery,” J. Microelectromechanical Syst., vol. 21, no. 1, pp. 44–52, Feb. 2012. [27] J. J. Norman et al., “Hollow microneedles for intradermal injection fabricated by sacrificial micromolding and selective electrodeposition,” Biomed. Microdevices, vol. 15, no. 2, pp. 203–210, Apr. 2013. [28] W. Martanto, J. S. Moore, T. Couse, and M. R. Prausnitz, “Mechanism of fluid infusion during microneedle insertion and retraction,” J. Controlled Release, vol. 112, no. 3, pp. 357–361, May 2006. [29] M. Thomsen et al., “Visualization of subcutaneous insulin injections by x-ray computed tomography,” Phys. Med. Biol., vol. 57, no. 21, pp. 7191–7203, Nov. 2012. [30] K. Comley and N. Fleck, “Deep penetration and liquid injection into adipose tissue,” J. Mech. Mater. Struct., vol. 6, no. 1, pp. 127–140, 2011. [31] T. Rattanapak et al., “Transcutaneous immunization using microneedles and cubosomes: Mechanistic investigations using Optical Coherence Tomography and Two-Photon Microscopy,” J. Controlled Release, vol. 172, no. 3, pp. 894–903, Dec. 2013. [32] J. H. Chang, N. C. Hogan, and I. W. Hunter, “High-speed X-ray imaging of needle-free jet injections,” in Biomedical Imaging (ISBI), 2014 IEEE 11th International Symposium on, 2014, pp. 473–476. [33] J. W. Mckeage et al., “High-speed X-ray analysis of liquid delivery during jet injection,” in Engineering in Medicine and Biology Society (EMBC), 2017 39th Annual International Conference of the IEEE, 2017, pp. 296–299. [34] A.-R. A. Khaled and K. Vafai, “The role of porous media in modeling flow and heat transfer in biological tissues,” Int. J. Heat Mass Transf., vol. 46, no. 26, pp. 4989–5003, Dec. 2003. [35] S. M. Becker and A. V. Kuznetsov, Eds., Transport in Biological Media. Saint Louis, MO, USA: Elsevier Science & Technology, 2013. 91  [36] S. I. Barry and G. K. Aldis, “Flow-induced deformation from pressurized cavities in absorbing porous tissues,” Bull. Math. Biol., vol. 54, no. 6, pp. 977–997, 1992. [37] S. Barry, G. Aldis, and G. Mercer, “Injection of fluid into a layer of deformable porous medium,” Appl. Mech. Rev., vol. 48, no. 10, pp. 722–726, 1995. [38] S. I. Barry, G. N. Mercer, and C. Zoppou, “Deformation and fluid flow due to a source in a poro-elastic layer,” Appl. Math. Model., vol. 21, no. 11, pp. 681–689, 1997. [39] Y.-G. Lv, J. Liu, Y.-H. Gao, and B. Xu, “Modeling of transdermal drug delivery with a microneedle array,” J. Micromechanics Microengineering, vol. 16, no. 11, pp. 2492–2501, Nov. 2006. [40] R. Zhang, P. Zhang, C. Dalton, and G. A. Jullien, “Modeling of drug delivery into tissues with a microneedle array using mixture theory,” Biomech. Model. Mechanobiol., vol. 9, no. 1, pp. 77–86, Feb. 2010. [41] G. Pontrelli and F. de Monte, “A two-phase two-layer model for transdermal drug delivery and percutaneous absorption,” Math. Biosci., vol. 257, pp. 96–103, Nov. 2014. [42] F. Yang, K. Chen, and Z.-G. Feng, “Analytical model of initial fluid infusion by a microneedle drug delivery system,” in Biomedical Engineering and Informatics (BMEI), 2011 4th International Conference on, 2011, vol. 2, pp. 913–917. [43] K. Chen, M. Pan, and Z.-G. Feng, “Modeling of Drug Delivery by A Pump Driven Micro-Needle Array System,” Open Biomed. Eng. J., vol. 10, no. 1, pp. 19–33, Mar. 2016. [44] F. J. Verbaan et al., “Improved piercing of microneedle arrays in dermatomed human skin by an impact insertion method,” J. Controlled Release, vol. 128, no. 1, pp. 80–88, May 2008. [45] B. Godin and E. Touitou, “Transdermal skin delivery: Predictions for humans from in vivo, ex vivo and animal models,” Adv. Drug Deliv. Rev., vol. 59, no. 11, pp. 1152–1161, Sep. 2007. [46] S. A. Ranamukhaarachchi et al., “A micromechanical comparison of human and porcine skin before and after preservation by freezing for medical device development,” Sci. Rep., vol. 6, no. 1, Oct. 2016. [47] J. M. Schmitt, S. H. Xiang, and K. M. Yung, “Speckle in optical coherence tomography,” J. Biomed. Opt., vol. 4, no. 1, pp. 95–106, 1999. 92  [48] A. Ozcan, A. Bilenca, A. E. Desjardins, B. E. Bouma, and G. J. Tearney, “Speckle reduction in optical coherence tomography images using digital filtering,” J. Opt. Soc. Am., vol. 24, no. 7, pp. 1901–1910, 2007. [49] D. C. Adler, T. H. Ko, and J. G. Fujimoto, “Speckle reduction in optical coherence tomography images by use of a spatially adaptive wavelet filter,” Opt. Lett., vol. 29, no. 24, pp. 2878–2880, 2004. [50] Z. Jian, Z. Yu, L. Yu, B. Rao, Z. Chen, and B. J. Tromberg, “Speckle attenuation in optical coherence tomography by curvelet shrinkage,” Opt. Lett., vol. 34, no. 10, pp. 1516–1518, 2009. [51] M. Szkulmowski, I. Gorczynska, D. Szlag, M. Sylwestrzak, A. Kowalczyk, and M. Wojtkowski, “Efficient reduction of speckle noise in Optical Coherence Tomography,” Opt. Express, vol. 20, no. 2, Jan. 2012. [52] M. Gargesha, M. W. Jenkins, A. M. Rollins, and D. L. Wilson, “Denoising and 4D visualization of OCT images,” Opt. Express, vol. 16, no. 16, pp. 12313–12333, 2008. [53] H. Wang, W. Lai, A. Antoniou, and A. Bastawros, “Application of Digital Image Correlation for Multiscale Biomechanics,” in Handbook of Imaging in Biological Mechanics, 1st ed., CRC Press, 2014, pp. 141–151. [54] J. Blaber, B. Adair, and A. Antoniou, “Ncorr: Open-Source 2D Digital Image Correlation Matlab Software,” Exp. Mech., vol. 55, no. 6, pp. 1105–1122, Jul. 2015. [55] X. Oliver and C. Agelet de Saracibar, Continuum Mechanics for Engineers: Theory and Problems, 2nd ed. 2017. [56] S. H. Hussain, B. Limthongkul, and T. R. Humphreys, “The Biomechanical Properties of the Skin,” Dermatol. Surg., vol. 39, no. 2, pp. 193–203, Feb. 2013.    93  Appendices Appendix A  OCT scan parameters The acquisition parameters of the OCT for all the injections discussed in the thesis are provided in this section. The following table provides the parameters for InjA to InjE.   Parameter InjA InjB InjC InjD InjE Injection fluid water water water water sugar soln. First retraction (mm) 0.3 0.3 0.3 0.3 0.2 Frame rate (frames/s) 17.2 73.9 17.2 71.4 6.86 Acquisition time (s) 0.0581 0.0135 0.0581 0.014 0.146 Size Z (pixels) 1024 1022 1024 1022 1024 Size Z (mm)  3.61 3.60 3.61 3.60 3.61 Size X (pixels) 1274 700 1274 700 2243 Size X (mm) 4.5 4 4.5 4 6.46 Pixel spacing in Z (mm/pixel) 0.0035 0.0035 0.0035 0.0035 0.0035 Pixel spacing in X (mm/pixel) 0.0035 0.0057 0.0035 0.0057 0.0029 A scan averaging 2 1 2 1 3 B scan averaging 1 1 1 1 1    94  The following table provides the parameters for InjF to InjJ.   InjF InjG InjH InjI InjJ Injected fluid water water water water sugar soln. First retraction (mm) 1 0.3 0.3 0.3 0.2 Frame rate (frames/s) 14.3 14.3 14.3 14.3 6.86 Acquisition time (s) 0.070 0.070 0.070 0.070 0.146 Size Z (pixels) 1024 1024 1024 1024 1024 Size Z (mm)  3.61 3.61 3.61 3.61 3.61 Size X (pixels) 1558 1558 1558 1558 2243 Size X (mm) 5.5 5.5 5.5 5.5 6.46 Pixel spacing in Z (mm/pixel) 0.0035 0.0035 0.0035 0.0035 0.0035 Pixel spacing in X (mm/pixel) 0.0035 0.0035 0.0035 0.0035 0.0029 A scan averaging 2 2  2  2  3 B scan averaging 1 1 1 1 1       95  Appendix B  Imaging techniques and experimental setup The other imaging modalities investigated include X-ray microcomputed tomography (micro-CT), histology and confocal microscopy. Some sample images from micro-CT and histology are shown in this section. More details of the experimental setup are also provided.   B.1 Micro-CT Micro-CT uses X-rays to detect different objects with different attenuation coefficients of X-rays. An iodine based contrasting agent (350Iohexol, Omnipaque) was injected into the skin tissue to get images of different cross-sections after the injection, as shown in the following figure. The bright regions represent injected fluid. The scale bar is 5 mm.   96  B.2 Histology Histology of the skin samples is carried out by fixation of tissue sample, tissue sectioning using a microtome, and staining. Tattoo ink was mixed with water, and injected into the skin sample using hollow microneedles. Some sample images of the histology sections are shown below.    B.3 Injection system details Some of the design details of the injection system are shown in section and detail views below.   97  As shown in the figure for the injection system, the impact mechanism has a simple ‘spring release switch’ that allows loading and releasing the plunger leg for impact. The microneedle attaches to a male luer lock adapter, so the microneedle can be removed after each experiment. Detail C in the figure also shows a section of the hole for the tubing, and the 3D printed screw (bottom) that attaches the plunger leg to the microneedle housing. The skin holder is removable, and can be slid into place before injection and removed for cleaning after experiments.   B.4 Sensors and tubing Some of the details of the connections between the pressure controller, fluid reservoir, flow sensor, pressure sensor and tubing in the system are shown below.    98  Appendix C  LABVIEW The LABVIEW interface used to control and measure injection parameters is shown below. Some of the benefits of using the LABVIEW program are: • Synchronization of data from Zabertech linear motorized stage, OB1 pressure controller, Microfluidic sensor reader (pressure and flow-rate sensors)  • Systematic flow of commands (commands cannot be called without completing pre-requisite commands preceding them) • Automatic data recording and saving measured/controlled data (Q, Pin, Pm, dr)   99   100  Appendix D  Wavelet transform based filtering The decomposition of an image into its approximation and decomposition levels, and the subsequent thresholding of the detail components are shown in this section.   D.1 Wavelet-2D decomposition The decomposition of an image into its approximation and detail coefficients is shown in the figure below. The figure shows the decomposition of the 2D detail coefficients into horizontal, diagonal and vertical components.       101  D.2 Thresholding detail coefficients The threshold values for the vertical detail coefficients are listed on left panel. The images show the raw and denoised images.            102  Appendix E   E.1 Example of Ncorr window This example shows the NCorr window with the reference and current images for InjB.   The Ncorr window for selecting the subset radius and subset spacing are shown below. The region of interest is the shaded rectangular region.   103  The displacement fields are first calculated, as shown below.     E.2 Ncorr strain  Instead of directly using the values for displacement determined from DIC, Ncorr uses a strain window algorithm. The size of the subset window can be independently controlled for determining strain.  The plane parameters for equations (10)-(13) are determined using a least squares plane fit on a subset of displacement data 𝑢ž“?Ÿ(𝑥, 𝑦) = 𝑎t,ž“?Ÿ + w 𝜕𝑢𝜕𝑥ž“?Ÿz 𝑥 + w 𝜕𝑢𝜕𝑦ž“?Ÿz 𝑦 𝑣ž“?Ÿ(𝑥, 𝑦) = 𝑎¡,ž“?Ÿ + w 𝜕𝑣𝜕𝑥ž“?Ÿz 𝑥 + w 𝜕𝑣𝜕𝑦ž“?Ÿz𝑦    104      

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0365782/manifest

Comment

Related Items