Citation
An optical stimulation cuff for use in rodent peripheral nerves toward control of an advanced prosthesis

Material Information

Title:
An optical stimulation cuff for use in rodent peripheral nerves toward control of an advanced prosthesis
Creator:
Hogan, Laura Kathryn
Place of Publication:
Denver, CO
Publisher:
University of Colorado Denver
Publication Date:
Language:
English

Thesis/Dissertation Information

Degree:
Master's ( Master of science)
Degree Grantor:
University of Colorado Denver
Degree Divisions:
Department of Bioengineering, CU Denver
Degree Disciplines:
Bioengineering
Committee Chair:
Bodine, Cathy
Committee Members:
Weir, Richard
Caldwell, John
Benninger, Richard

Notes

Abstract:
The purpose of this research was to create an LED-based stimulation cuff that could be implanted around a mouse sciatic nerve. Currently available prosthetic devices lack sensory feedback, which is necessary for embodiment of the limb; this causes a high rejection rate, especially among upper-limb unilateral distal prosthesis users. Sensory feedback has been shown in several studies to increase ownership and usefulness of the limb. Electrical stimulation has largely been used in humans, with some success. However, electrical nerve stimulation implants have several drawbacks, chief among them non-specificity. Thus, the field of optogenetics is useful for this application, since it allows for subsets of axons to be targeted and preferentially activated using light and light-sensitive proteins. In this study, an implantable nerve cuff was designed and iterated. The final design, including three LEDs, was implanted around transgenic mouse sciatic nerves, and stimulation was applied. Compound action potentials were recorded and compared to recordings from electrical stimulation. These experiments found that this cuff could be a viable alternative to electrical stimulation, especially should it be improved, due to optogenetics’ advantage of selective stimulation. However, further experiments and validation must be performed.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
Copyright Laura Kathryn Hogan. Permission granted to University of Colorado Denver to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

An Optical Stimulation Cuff for Use i n Rodent Pe ripheral Nerves Toward Control of a n Advanced Prosthesis by LAURA KATHRYN HOGAN B.S., Wichita State University, 2014 A thesis submitted to the Faculty of the Graduate School of the University of Colorad o in fulfillment of the requirements for the degree of Master of Science Bioengineering Program 2018

PAGE 2

ii This thesis for the Master of Science degree by Laura Kathryn Hogan has been approved for the Bioengineering Program by Cathy Bodine, Chair Ri chard Weir, Advisor John Caldwell Richard Benninger Date: July 20, 2018

PAGE 3

iii Hogan, Laura Kathryn (M.S., Bioengineering) An Optical Stimulation Cuff for Use in Rodent Peripheral Nerves Toward Control of an Advanced Prosthesis Thesis directed by Research Assoc iate Professor Richard Weir ABSTRACT The purpose of this research was to create an LED based stimulation cuff that could be implanted around a mouse sciatic nerve. Currently available prosthetic devices lack sensory feedback, which is necessary for embodim ent of the limb; this causes a high rejection rate, especially among upper limb unilateral distal prosthesis users. Sensory feedback has been shown in several studies to increase ownership and usefulness of the limb. Electrical stimulation has largely been used in humans, with some success. However, electrical nerve stimulation implants have several drawbacks, chief among them non specificity. Thus, the field of optogenetics is useful for this application, since it allows for subsets of axons to be targeted and preferentially activated using light and light sensit ive proteins. In this study, an implantable nerve cuff was designed and iterated. The final design, including three LEDs , was implanted around transgenic mouse sciatic nerves, and stimulation was a pplied. Compound action potentials were recorded and compared to recordings from electrical stimulation. These experiments found that this cuff could be a viable alternative to electrical stimulation, especially should it be improved, due to advantage of selective stimulation. However, further experiments and validation must be performed. The form and content of this abstract are approved. I recommend its publication. Approved: Richard Weir.

PAGE 4

iv ACKNOWLEDGEMENTS The author would like to thank se veral people for their assistance in the completion of this study. Drs. Richard Weir, John Caldwell, Cathy Bodine, Richard Benninger, Emily Gibson, and Diego Restrepo offered expertise as committee members or advisors in various fields related to the proje ct. Stephen Huddle, Dr. Arjun Fontaine, Dr. Hans Anderson, Dr. Matthew Davi d son , and other current and former members of the Weir lab offered support and help on technical aspects of the study. Dr. Cara Mitchell and the rest of the RC1 vivarium staff assi sted in animal handling and surgery training. Andrew Scallon from the Optogenetics and Neural Engineering Core offered expertise in optical fiber fabrication as well as equipment rental . Drs. Baris Ozbay and Greg Futia assisted in extraneous work related t o optical systems and imaging. Nick George built and maintained the CAP recording system used in later experiments. Karen Purba assisted in preparation and execution of the last several months of experiments. Kelly Waugh, Becky Breaux, and Brian Burne supp orted the author in her clinical internship at Assistive Technology Partners. One year of this project was funded by a pre doctoral CCTSI grant, TL1 TR001081, from July 2016 to June 2017. Finally, the author would like to thank her friends and family, in p articular her husband James Hogan and her sister and parents, Alex, Georgie, and Bruce Elson, for moral support during the process of graduate school.

PAGE 5

v TABLE OF CONTENTS I. INTRODUCTION ................................ ................................ ................................ ....................... 1 Literature Review ................................ ................................ ................................ ............ 1 Part 1: The State of Current Prostheses and Sensory Feedback ................................ . 1 Amputations and Prosthetics. ................................ ................................ .................. 1 Current Feedback Technologies in Use. ................................ ................................ . 3 Part 2: Optogenetics as a Means of Communicati ng with the Nervous System ......... 4 Methods and Advantages. ................................ ................................ ....................... 4 Current Optogenetic Peripheral Nerve Interfaces. ................................ .................. 5 Specific Aims ................................ ................................ ................................ .................. 6 II. EXPERIMENTAL DESIGN ................................ ................................ ................................ ...... 7 Cuff Design ................................ ................................ ................................ ..................... 7 Initial Design Choices ................................ ................................ ................................ . 7 Optical Fibers vs. Light Emitting Diodes (LEDs). ................................ ................. 7 Iterating LEDs. ................................ ................................ ................................ ........ 9 Shape of the cuff. ................................ ................................ ................................ .. 10 Cuff Fabrication ................................ ................................ ................................ ........ 10 Nerve Recording ................................ ................................ ................................ ........... 14 Mouse Strains Used ................................ ................................ ................................ .. 14 Compound Action Potential (CAP) Recording ................................ ......................... 15

PAGE 6

vi Construction of Suction Electrodes. ................................ ................................ ..... 15 Experimental Setup. ................................ ................................ .............................. 18 Electrical Stimulation and Recording. ................................ ................................ .. 20 Implantation of Optical Cuff and Optical Stimulation. ................................ ........ 20 Data Analysis. ................................ ................................ ................................ ....... 22 III. RESULTS A ND DISCUSSION ................................ ................................ ............................. 23 Results ................................ ................................ ................................ ........................... 23 Completed Cuffs ................................ ................................ ................................ ....... 23 3D Printed Cuffs. ................................ ................................ ................................ .. 23 Molded Cuffs. ................................ ................................ ................................ ....... 24 Microprobes Cuff. ................................ ................................ ................................ . 26 In House Cuffs. ................................ ................................ ................................ ..... 29 Strain Relief Strategies. ................................ ................................ ........................ 29 Electrical Stimulation Results ................................ ................................ ................... 35 Cuff Stimulation Results ................................ ................................ ........................... 39 Laser Stimulation Results ................................ ................................ ......................... 37 Discussion ................................ ................................ ................................ ..................... 49 Cuff Design ................................ ................................ ................................ ............... 49 Things to Consider When Comparing Signals ................................ .......................... 50 The Shape of the Signal ................................ ................................ ............................ 53

PAGE 7

vii Limitations of Study and Future Work ................................ ................................ ..... 55 Conclusions ................................ ................................ ................................ ................... 58 REFERENCES ................................ ................................ ................................ ............................. 60 APPENDICES ................................ ................................ ................................ .............................. 65 A. Extraneous Work ................................ ................................ ................................ . 65 GRIN Lens Cuff for Optical Recording ................................ ................................ ... 65 Dorsal Root Ganglion Exposure Surgery ................................ ................................ . 68 B. CCTSI TL1 Predoctoral Fellowship ................................ ................................ ... 70

PAGE 8

viii LIST OF FIGURES Figure 1: 60um optical fiber. ................................ ................................ ................................ .......... 7 Figure 2: First LED from Cree. ................................ ................................ ................................ ....... 8 Figure 3: Surface Mount 0402 LEDs from Bivar. ................................ ................................ .......... 9 Figure 4: Cuff fabrication process. ................................ ................................ ............................... 13 Figure 5: Pre soldered LED installed on 1 mm ID cuff. ................................ ............................... 14 Figure 6: Fully assembled suction electrode. ................................ ................................ ................ 15 Figure 7: Experimental setup for nerve stimulation and recording. ................................ ............. 18 Figure 8: CAP recording setup. ................................ ................................ ................................ .... 19 Figure 9: Sciatic and tibial nerve in optical cuff. ................................ ................................ .......... 21 Figure 10: First iteration: 3D printed cuff. ................................ ................................ ................... 23 Figure 11: A) 3D printed cuff in euthanized mouse. ................................ ................................ .... 24 Figure 12: Molded cuff. ................................ ................................ ................................ ................ 25 Fi gure 13: Implanting the stimulation cuff into a mouse. ................................ ............................. 26 Figure 14: Stimulation cuffs received from Microprobes, with a pen for scale. .......................... 26 Figure 15: Shining laser light (473nm) from connected fiber (on right) to nerve cuff. ................ 27 Figure 16: Implanting the Microprobes cuff in a euthanized mouse. ................................ ........... 28 Figure 17: The first iteration of my in house cuffs. ................................ ................................ ...... 29 Figure 18: Conceptual designs of strain relief strategies. ................................ ............................. 30 Figure 19: First strain relief strategy of gluing everything into flexible tubing. .......................... 30 Figure 20: Strain relief using strain gauges glued directly to the cuff. ................................ ......... 32 Figure 21: Pre soldered LED glued onto a 1 mm ID cuff. ................................ ........................... 33 Figure 22: Three LED cuff illuminated and encapsulating the nerve. ................................ ......... 33

PAGE 9

ix Figure 23: Pre soldered LEDs on a narrower, thicker walled cuff. ................................ .............. 34 Figure 24: Representativ e CAPs following electrical stimulation. ................................ ............... 35 Figure 25: Signal peaks following electrical stimulation (n=3). ................................ ................... 36 Figure 26: Recording area under the signal while increasing stimulation voltage. (n=3) ............ 37 Figure 28: A typical signal following optical stimulation. ................................ ........................... 40 Figure 27: Varying frequency for prolonged cuff stimulation of ChR2+ axons. ......................... 40 Figure 29: 10 Hz signals from three different nerves. ................................ ................................ .. 41 Figure 30: 20 Hz stimulation over 30 seconds. ................................ ................................ ............. 43 Figure 31: Area under the signal following 10 Hz stimulation. ................................ ................... 44 Figure 32: Area under the signal following 20 Hz stimulation. ................................ ................... 45 Figure 33: Area under signal following 30 and 40 Hz stimulation. ................................ .............. 46 Figure 34: Area under the signal following 60 Hz stimulation. ................................ ................... 47 Figure 35: Double peak signals during optical stimulation. ................................ ......................... 48 Figure 36: Varying frequency of laser stimulation to see degradation rate of the signal. ............ 38 Figure 37: Varying stimulation pulse length at 10 Hz to determine the optimal value. ............... 39 Figure 38: Solidworks model of GRIN lens cuff. ................................ ................................ ......... 65 Figure 39: GRI N lens cuffs from Microprobes that could not be used. ................................ ....... 66 Figure 40: Third iteration of the Microprobes cuff. ................................ ................................ ...... 67 Figure 41: Imaging through GRIN lens cuff. ................................ ................................ ................ 68 Figure 42: Dorsal root ganglion exposure surgery. ................................ ................................ ...... 70

PAGE 10

1 I. INTRODUCTION Literature Review Part 1: The State of Current Prostheses and Sensory Feedback Amputations and Pro sthetics. Limb loss is a traumatic experience that can be caused by a variety of factors. Dysvascular disease is the most common cause (especially when comorbid with diabetes mellitus), while trauma, cancer, and congenital deficiencies also contribute. In 2005, it is estimated that there were 1.6 million Americans living with some level of amputation; by 2050, this number is expected to more than double to 3.6 million. Of this number, 35% are estimated to have lost an upper extremity, largely due to trauma (Ziegler Graham, MacKenzie, Ephraim, Travison, & Brookmeyer, 2008) . Despite this increasing prevalence, current prosthesis design does not meet the needs of users. P rosthesis rejection rate is relatively high, especially among unilateral upper limb users who are able to compensate with their remaining hand. Up to 45% rejection rates have been reported among pediatric amputee populations, while adult populations reject prostheses at a lower (though still significant) rate (26%) (E. A. Biddiss & Chau, 2007) . Rejection is reported to be direct same population of prostheses rejecters reported that, should prosthesis function improve, they would be more likely to use a device in the future (E. Biddiss & Chau, 2007) . Two of the most important factors in prosthesis use are adept control and sensory feedback. One study reports that 88% of those surveyed considered sensory feedback to be at least moderately important to a prosthesis, in particular with regard to grip force and proprioception (Lewis, Russold, Dietl, & Kaniusas, 2012) . Motor control is considered equally imp ortant, especially as it relates to performing activities of daily living (ADLs). This includes

PAGE 11

2 the ability to modulate grip strength as well as perform more than one action at a time for example, rotating the wrist and opening the hand (Cordella et al., 2016) . Sensory feedback and motor control are closely intertwined when considering function and ownership of prosthetic limbs, especially for the upper body. Generally, prostheses are considered more of a tool than a part of the body; this perception may be changed by adding sensory feedback to prosth etic limbs. This feedback will make motor control more natural to the user, particularly during gripping tasks (Wijk & Carlsson, 2015) . Feed back on proprioception and grip strength inform the user on the progress of tasks considered vital (such as holding an object without breaking it) without the user giving visual attention to the prosthetic. Such simple tasks are generally considered import ant by people with amputations more so than complex tasks (Engdahl et al., 2015; Lewis et al., 2012) . There are currently three main classifications of prostheses, and two main methods of control: cosmetic devices, which are passive and are ge nerally only used to simulate the appearance of a natural limb; body powered devices, which use the remaining limb to power the prosthesis; and myoelectric devices, which use surface electromyography (EMG) to read muscle activity in the residual limb. Body powered devices are considered more useful by those with more active workplaces, such as construction workers; myoelectric devices are favored by those with desk jobs (Carey, Lura, & Highsmith, 2014) . Myoelectric prostheses are considered state of the art, but users experience several problems with the device that limit their us efulness and may increase the abandonment rate. These include slippage due to sweating or poor socket fit, weight, grip speed, price, and durability concerns (Carey et al., 2014; Pylatiuk, Schulz, & Döderlein, 20 07) . It is clear that the current state of prosthesis design is not acceptable to many users, leading to abandonment.

PAGE 12

3 Current Feedback Technologies in Use. Sensory feedback is of particular interest to users, with several studies labeling it among the mo st important features for future designs (E. Biddiss & Chau, 2007; Cordella et al., 2016; Lewis et al., 2012; Peerdeman et al., 2011; Wijk & Carlsson, 2015) . Several studies have attempted to close this gap through varying methods, ranging from non invasive, surface level stimulation to complex neural interfaces. The most common non invasive sensory feedback systems include electrotactile, vibrotactile, and mechanotactile stimulation, which aim to provide feedback through small electric current, vibration, and pressure, respectively. Elec trotactile feedback is seen as the most encouraging of the three due to its smaller size and weight, and because it provides a way to directly stimulate the peripheral nervous system without entering the body. However, it is limited in people who use myoel ectric prostheses, as the signal generated by the feedback system interferes with the muscle signals picked up by the prosthesis. In addition, the feedback provided by this electrical stimulation is not intuitive and does not relate to the sensory stimulat ion that (Li, Fang, Zhou, & LIU, 2017) . Other systems include using transcutaneous electrical nerve stimulation (TENS) to stimulate the nerve through the skin; however, this comes with its own set of challenges, including elect rode placement and non intuitive feedback modalities . Systems that use mechanotactile or vibrotactile stimulation can provide homologous feedback appropriate to the stimulation of the hand, but in order to achieve somatotopic feedback, a phantom hand map is required. This map only exists in about half of amputees (and is often incomplete), making such a system impractical for many users a et al., 2017) .

PAGE 13

4 Targeted muscle reinnervation (TMR) is also a promising system to provide somatotopic and homologous feedback as well as more sites for EMG controlled movement; however, it involves a highly invasive surgery to re assign the nerves. Beca use of this, it is typically only performed in those with shoulder disarticulation or other high level amputations for whom standard prostheses are not an option. However, it does allow the user more sites for muscle control, and has been shown to provide intuitive sensory feedback (Cheesborough, Smith, Kuiken, & Dumanian, 2015; Lipschutz, 2017; Scho field, Evans, Carey, & Hebert, 2014) . A strong alternative to these systems is more invasive neural interfaces. These require surgery for implantation, but they interface directly with the peripheral nervous system, allowing for direct contact and more sp ecific activation of particular nerves. However, there is a sharp trade off in these systems between specificity and tissue preservation. Interfaces that do not pierce the nerve must stimulate most or all of the axons within the nerve; those that do pene trate the nerve to activate specific fascicles or axons risk causing tissue damage (S ahyouni et al., 2017) . Less invasive cuffs have been designed to attempt to flatten the nerve (Dweiri, Stone, Tyler, McCallum, & Durand, 2016; Tyler & Durand, 2002) , and to improve electrode nerve contact via coatings or tissue engineering principles to promote nerve growth (Heo et al., 2016; Thompson, Zoratti, Langhals, & Purcell, 2016) . However, a more precise method of stimulation is needed to gain the specificity and safety necessary for long term implantation. Part 2: Optogen etics as a Means of Communicating with the Nervous System Methods and Advantages. The field of optogenetics is a promising option for this. Optogenetics uses light sensitive proteins expressed in peripheral axons, either via transgenic animal lines or ade no associated virus (AAV) injection, to drive or inhibit action potentials

PAGE 14

5 (Deisseroth, 2011) . Channelrhodopsin 2 (ChR2) is the most commonly used activation opsin; it is a non selective cation channel that depolarizes the cell when illuminated with blue light (473nm) (Nagel et al., 2003) . Opto genetics provides an extremely time sensitive method of axon activation, as well as providing precise specificity in a way that even the most invasive electrical systems cannot. Opsins can be expressed in specific subsets of neurons (for example, motor neu rons) via viral injection or transgenic animals , so that when the entire nerve is stimulated with light, only those axons are activated (Deisseroth, 2011) . Optogenetics has also been shown to recruit motor units in a physiological order, rather than a reversed order as seen with electrical stimulation; t his has exciting prospects for therapy and other muscle based work, as muscles fatigue slower when they are recruited properly (Llewellyn, Thompson, Deisseroth, & Delp, 2010) . Current Optogenetic Peripheral Nerve Interfaces. Currently, optogenetics is largely used in the central nervous system to study brain functi ons , novel pathways, and injury ( Klapoetke et al., 2014) . However, there is growing interest in for applications such as chronic pain inhibition, motor control for those who are paralyzed, and sensory stimulation (Ahmad, Ashraf, & Komai, 2015; Bryson, Machado, Lieberam, & Greensmith, 2016; Shrivats M Iyer et al., 2016; Mallory, Grahn, Hachmann, Lujan, & Lee, 2015) . Several methods of communicating with the peripheral nervous system have been tested. One of the simpler methods includes transdermal illumination, in which there is no implant at all; instead, light is sho activated that way (Shrivats Mohan Iyer et al., 2014; Maimon et al., 2017) . However, this method can be less effective if one is trying t o stimulate a deep nerve, or wants to stimulate as

PAGE 15

6 much of the nerve as possible, due to the scattering properties of tissue . Thus, it is most useful for stimulating superficial pain fibers (Towne, Montgomery, Iyer, Deisseroth, & Delp, 2013) . Several stud ies have shown the effectiveness and safety of implantable optical peripheral nerve cuffs (Michoud et al., 2017; Montgomery et al., 2015; Park et al., 2015; Towne et al., 2013) . However, to my knowledge there has been no o ptical nerve cuff fabricated with multiple light sources. Having the nerve illuminated from different angles would ensure that many more expressing axons are activated, and would nearly eliminate the issue of tissue scattering. This would enable the cuff t o gain up to a maximal response from the nerve, driving either stronger muscle contractions or stronger sensations. Specific Aims Currently, there are no optically based cuffs for the rodent peripheral nerve with multiple light sources . These cuffs are an important step in advancing the field of optogenetics and eventually moving into larger animals , and multiple light sources ensure that as many axons are activated as possible . The goal of this study was to design such a nerve cuff and to test it in the mo use sciatic nerve. The cuff needed to be small; flexible so that it would not break during implantation; easy to implant; easy to make; and illuminate as much of the nerve as possible. Thus, the goals of this thesis were: Specific Aim 1: Design an optical neural interface for use in mice. Specific Aim 2: Test this interface in ex vivo nerves and compare the results to electrical stimulation.

PAGE 16

7 II. EXPERIMENTAL DESIGN Cuff Design Initial Design Choices Several cuff designs were considered for this project. Th ere were several criteria that were discussed when determining the best design. These included 1) ease of implantation; 2) robustness of the cuff; 3) cuff size; 4) percent of nerve illuminated; 5) ease of fabrication. Optical Fibers vs. Light Emitting Di odes ( LEDs ) . The first main choice to make was whether the cuff would include LEDs or optical fibers as its light source. Each has its own advantages. LEDs do not require a laser to operate, so are more portable; more of them can easily be put in a single cuff. On the other hand, optical fibers have a more focused cone of light, such that deeper axons might be illuminated and activated; they also are focused at a single wavelength (in our case, 473nm), which is close to the peak activation of channelrhodop sin 2. Neither LEDs nor fibers were easy to install in a cuff. The optical fibers we had readily available were 60 µ m in diameter, which were difficult to see let alone handle and install Figure 1 : 60um optical fiber. A) Unjacketed fiber in mailing envelope. B) Jacketed fiber with loose end for cuff implantation. A B

PAGE 17

8 without jacketing to protect it. Even with 80% of the fiber jackete d, the free end m eant to be installed in the cuff was difficult to handle. Such fabrication was beyond my abilities or those of anyone involved with the project; thus, only one fiber optic cuff was manufactured, by the company Microprobes. Both the jackete d and unjacketed fibers can be seen in Figure 1 . In addition, the initial goal of installing multiple light sources in the cuff was to place them at different angles, such that axons on every side of the nerve would be illuminate d. Because the optical fibers were not very flexible, it would be impossible to have fibers surrounding the whole nerve. At best, two fibers at approximately 90 º to each other could face Figure 2 : First LED from Cree. A) Schematic of LEDs. Bottom view shows the pads where I expected to solder wires TM LEDs Data Sheet CxxxTR2432 . B) Pack of over 200 LEDs next to a 1mm cuff (black part). A B

PAGE 18

9 upward from the surgical opening; even this would be difficult to ach ieve. Because of these difficulties, I decided to use LEDs moving forward . Iterating LEDs. At first, I ordered the smallest LEDs I could find; these were Cree DA243 2 LED chips, with dimensions of 240x320 µ m on their largest side. These can be seen in Figure 2 . These proved far too small to work with ; even under a microscope, it was impossible to solder two separate contacts on the LED. Thus , I instead pursued slightly larger LEDs (standard Figure 3 : Surface Mount 0402 LEDs from Bivar. A) Schematic showing dimensions and s oldering sites (Bivar, 2017) . B) LED on fingertip. A B

PAGE 19

10 size 0402) that would still fit on a cu ff. The LEDs I ordered from Mouser can be seen in Figure 3 . After working with these for several months, I found pre soldered LEDs of the same design on eBay . I decided to order and work with these for later iterations , so I would be able to make cuffs more quickly and with much less effort . Shape of the cuff. The design of the cuff itself was also under consideration. There were two main choices: the first was an oval shape to slightly flatten the nerve, which Tyler and Durand re ported in 2002 (Tyler & Durand, 2002) . This would increase the surface area of the nerve and reduce the average distance between the LED and the axons it is illuminating; overall, a greater percentage of axons within the nerve would be activated. Further studies on the flat in terface nerve electrode (FINE) showed no long term damage to the nerve (>60 mm Hg) pressures were not applied (Freeberg, Stone, Tyler, & Triolo, 2013; Leventhal, Cohen, & Durand, 2006; Tyler & Durand, 2003) . Circular cuffs, on the whole, are easier to produce, becau se round tubing is easily available and virtually unlimited. In addition, the benefits of the FINE would be mitigated somewhat by installing multiple LEDs around the cuff. These reasons, combined with time constraints on the project, re sulted in all via ble cuffs being made circular . Cuff Fabrication Several early cuffs were designed and produced with various techniques before the final design was determined. The first involved simply 3D printing the cuff; the next involved 3D printing a mold and then pouring silicone into it such that the cuff would be made of softer material than our printer could produce. The third was fabricated by Microprobes and incorporated an optical fiber rather than LEDs.

PAGE 20

11 The next major iteration of cuffs would b e the last, with only changes to that design from that point forward. Polyamide tubing was used as the base, with a slit cut lengthwise through which to slip the nerve. Small holes were cut with an X ACTO knife through which the LED would project, shining light directly onto the nerve. From this base, wires needed to be soldered to the loose LEDs in order to power them. Several versions of this were iterated due to poor strain relief within the LEDs themselves. A working prototype was eventually created wit h two LEDs soldered to its surface at approximately 90 ° to each other. 1 mm inner diameter (ID) tubing was initially used due to its availability in the lab. However, 0.7 mm ID tubing with thic ker walls was later considered because the adult mouse sciatic ly ranges from 0.6 to 0.8 mm. 0.7 mm ID for the cuff would be an ideal fit. Ultimately, 1 mm was chosen due to its flexibility and ease of handling ; its slightly increased size made it significantly easier to install LEDs on its out er surface . At this point, creating a functioni ng cuff required several steps: 1. Cut a 5 7 mm length of tubing. 2. D epending on how many LEDs are wanted on the cuff, cut the corresponding number of holes in the cuff at various angles (seen in Figure 4 A) . T ake care to leave enough room lengthwise at one angle to later cut the slit into the cuff. 3. P ut solder on each LED pad. 4. Cut the strain gauges (see Figure 4 B) in half using the X ACTO knife, and put solder o n each half. 5. Put the LED into the hole so it is partially impressed into the cuff, and then glue it in . 6. Glue one half of the strain gauge on each side of the LED lengthwise along the cuff. 7. Wait at least an hour for the glue to dry. (Ideally, wait 24 hours .)

PAGE 21

12 8. Cut two very short lengths of thin wire, approximately the distance between the LED soldering pad and its corresponding strain gauge. Use the soldering iron to attach this piece of wire to these two points, forming a connection. 9. Cut two equal lengths of wire at least a foot long. Strip both ends and put solder on one end. Solder this end to the strain gauge on either side of the LED. The finished product can be seen in Figure 4 C. 10. Cover the entire LED assembly with a liberal amo unt of glue and let dry. 11. Repeat steps 3 10 for each LED you wish to put on the cuff. 12. Using an X ACTO knife, carefully cut a slit down the length of the cuff without compromising any of the LED connections. For ease of nerve implantation, widen this slit by making a second one parallel to it, 0.5 1 mm away.

PAGE 22

13 This process was extremely time consuming, with many possibilities for error; the initial solder joints were 200 x 500 µ m in size, and could not touch each other at any point along the circuit . This mad e it technically difficult to fabricate, as well as increased the possibility of creating a poor solder joint. Thus, when I found the pre soldered LEDs, I immediately ordered them and started using them instead. This significantly reduced the fabrication time required for each cuff , as well as reduced because the strain gauges were no longer necessary; all I had to do was cut a hole in the tubing and glue the LED in. T he results of these Figure 4 : Cuff fabr ication process. A) LED next to 1 mm tubing with hole cut out. B) 1 mm ID cuff with LED and full sized strain gauge . C) Wires soldered to each LED pad using half strain gauges . A C B

PAGE 23

14 streamlined LEDs can be seen in Figure 5 . Due to this, I was able to install three LEDs on a single cuff; this became the final cuff design. Nerve Recording Mouse Strains Used Transgenic m ice expressing ChR2 in cholinergic axons via the ChAT promoter (ChAT Cre ChR2 mice) were advantageous to this study and used for all optical stimulation experiments. These were over as needed had access to their entire colony. Wild type C57BL/6 mice were also used for initial experiments and troubleshooting due to their easier availability and greater numbers. All animals, experiments, and handling were approved by the Institut ional Animal Care and Use Committee (IACUC) at the University of Colorado Health Sciences Center, with accreditation by the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC) . Figure 5 : Pre soldered LED installed on 1 mm ID cuff.

PAGE 24

15 Compound Action Potential (CAP) Recording I record ed electrical CAPs following several types of stimulation. These included electrical stimulation; optical stimulation with a laser and optical fiber; and optical stimulation with my LED based cuff. Construction of Suction Electrodes. Suction electrodes (seen in Figure 6 ) were an integral part of every experiment I carried out. They were used both for electrical stimulation and electrical recording. The construction of these electrodes consisted of several steps: 1. To make the gl ass pipette, take a standard glass pipette with a 1 mm ID, score it with a diamond tip at the center, and break it in half to make it the correct length. 2. Hold one end of the pipette up to a Bunsen burner and rotate it so all sides get equally heated. The s ize of the finished pipette is largely estimated based on experience; the pipette meant to hold the sciatic nerve should be approximately 600 700 µ m in diameter, while the pipette for the tibial nerve should be roughly Figure 6 : Fully assembled suction electrode.

PAGE 25

16 half of that. Using the Bunsen burner , bend the pipette slightly 1/3 the length away from the tapered end in order to help it fit the chamber. a. Fitting the electrode more tightly to the recording end of the nerve ensures that as much signal as possible is collected without the rest being lost within the saline. A truly perfect pipette would result in a suction fit of the nerve within the electrode. 3. To make the base of the electrode, you will need a strong round magnet approximately 1 inch in diameter, small male/female connectors (indicated by the white arrow in Figure 6 ), and a narrow, hollow metal rod 1.5 inches in length. Glue these together as shown in Figure 6 with epoxy, ensuring both ends of the metal rod hang over the ends of the magnet . Also ensure that the contacts on the connectors are accessible on both ends. Let dry for at least 15 minutes. 4. Cut two lengths of thin Teflon coated wire slightly longer than the pipette. Under a microscope , using forceps, carefully strip the Teflon from both ends of the wires for 3 4 mm. 5. Solder one of these wires into each of the female connections, ensuring that solder gets far enough into the connector that a continuous circuit is created, and that the connection is solid. These joints are liable to br eak, either due to a poor solder joint or the fragility of the wires; handle with care. 6. Soak the free ends of both silver wires in bleach for 30 60 minutes to chlorinate them. 7. Affix a short, properly sized piece of rubber tubing to this end of the metal ro d. Thread one of the silver wires through the glass pipette, then carefully push the

PAGE 26

17 pipette into the rubber tubing until secure . Wrap the second wire around the pipette. 8. On the other end of the metal rod, attach a 1 2 foot length of rubber tubing. On the loose end of this, install a properly sized (10 cc) blunt syringe. 9. Obtain dual shielded wire as seen in Figure 6 . Pull the shielding back from each end and strip 3 4 mm from both wires at one e nd, and 1 c m from the other end . Sol der the first ends into the free connector ports, ensuring the connection is solid with sufficient solder to keep it in place. Place a small amount of solder on the other ends of the wires to ensure the multiple strands stay together. 10. Loosen the set screws from a dual banana plug, push the free ends of the wire into the space, and then tighten them so that the connection is secure. 11. Use a multimeter to ensure there is conductivity from the banana plugs to the ends of the silver wires; if not, one of your joi nts is not well made.

PAGE 27

18 E xperimental Setup. The mouse was euthanized in the lab. After secondary euthanasia was performed, the sciatic and tibial nerve were carefully excised from one hind leg and stored in a glucose based mouse saline solution as describe d by Fontaine, et al (Fontaine, Gibson, Caldwell, & Weir, 2017) . Care was taken during extraction to ensure the nerve was not stretched or handled directly with sharp forceps; whe n possible, only connective tissue was held. This saline and the nerve were stored in a custom made chamber similar to the one shown in Figure 7 A. Saline was drawn up into each suction electrode until the interior wire was subme rged. Care was taken to ensure that the exterior wire was connected as well. Then, t he nerve was pulled into each suction electro de on either end for recording. For initial optical fiber stimulation, Figure 7 : Experimental setup for nerve stimulation and recording. A) Empty chamber with sciatic and tibial electrodes. B) Nerve in chamber with h arp and suction electrodes. (Image in B courtesy of Dr. Arjun Fontaine.) A B

PAGE 28

19 a harp was placed across the exposed nerve to keep it st ationary at the bottom of the chamber ( Figure 7 B). The tibial electrode, considered the recording electrode, was attached to the CAP c via insulated alligator clips ; for electrical st imulation, the sciatic e lectrode (the stimulating electrode) was connected to the stimulator , pictured in Figure 8 A. The overall CAP recording system is pictured in Figure 8 B. A Mol ecular Devices Digi data 1550A digitizer was used to communicate information between the software and external hardware. An A M Systems Microelectrode AC amplifier was used to improve the Figure 8 : CAP recording setup. A) Stimulator used for all experiments, giving a wide range of variables. B) The overall CAP recording computer, digiti zer, amplifier, and Faraday cage, which held the stimulating electrodes, the stage for the chamber, recording electrodes, dissecting scope, and lamp. A B

PAGE 29

20 recording signal. The pClamp suite of software was used for data collection (Clampex) an d analysis (Clampfit). Electrical Stimulation and Recording. For electrical stimulation, both suction electrodes are required. 50 µ s pulses ranging from 0.1 V to 10 V were supplied via the sciatic electrode, at frequencies ranging from 10 Hz to 60 Hz. The tibial electrode recorded the subsequent action potentials and portrayed them in Clampex as graphs of response in mV vs. time in ms. Implantation of Optical Cuff and Optical Stimulation. I performed two types of optical stimulation, involving an optical fiber powered by a laser, and an LED cuff powered by an Arduino. The fiber stimulation was performed first ; the nerve was gently held down to the bottom of the bath by a harp. The 120 µ m fiber was attached to the 473 nm laser, and positioned so it was dire ctly touching the nerve within the bath via a micromanipulator. Powers ranging from an estimated < 1 mW to 25 mW were applied. Frequencies between 10 Hz and 40 Hz were successfully recorded from. Pulse lengths between 1 ms and 10 ms were applied. Attachmen t of the optical cuff occurred at the end of the experiment in case the nerve was damaged during insertion. The cuff was placed in an optimal position within the bath and then held down along the edge of the chamber by modeling clay. The nerve was carefull y moved into the slit along the cuff without shoving, tearing, or stretching. This process could often take several minutes, due to difficulties with connective tissue on the nerve and air bubbles getting trapped within the cuff, preventing the nerve from staying in place. Then, the nerve was pulled back into the recording electrode. The stimulation electrode was removed and set aside. This setup can be seen in Figure 9 .

PAGE 30

21 Once the nerve was securely within both the cuff and the re cording electrode , the cuff was gently manipulated so that the LEDs and their exposed wires were not within the bath. This was necessary to prevent electrical interference from drowning out the CAPs within the recorded signal. For short periods of time, th e chamber could be mostly drained of saline , with the nerve exposed to the air ; I put a piece of hydrophobic Parafilm underneath the cuff, and drained excess saline so only the liquid submerging the recording electrode wires remained. Carefully drying off the cuff at this point and putting Dow Corning vacuum silicone grease along the exposed wires helped to further minimize the electrical artifact. However, experiments in this state must be done quickly or else the nerve will dry out and cease to respond to stimulation. An Arduino Uno was coded to turn on the LED at various frequencies. Initially, it accomplished this by joining to an nScope breadboard and creating a complete circuit with a 100 resistor. However, it was discovered that the resistor was no t necessary to protect the LED, Figure 9 : Sciatic and tibial nerve in optical cuff . The tibial end is dr awn up into the recording electrode.

PAGE 31

22 and that when the nScope was removed from the system, the electrical noise in the recorded signal dropped significantly. Thus, I began plugging the LED ends directly into the Arduino power and ground pins. To turn on multipl e LEDs simultaneously, I ran them in parallel to these pins using extra pieces of wire and alligator clips. Signals resulting from 10 60 Hz light stimulation were obtained, using a 5 ms pulse width and maximum LED output. It was difficult to determine this output with a power meter due to the spectrum of wavelengths output by LEDs, but it was much smaller than the power output of the laser (potentially on the order of 100s of µ W) . However, I stimulated the nerve with up to 3 LEDs at once, and the cone of l ight from an LED is broader than that of an optical fiber, potentially illuminating a larger section of axons. Data Analysis. Because this is a proof of concept design project, no true statistical analysis was necessary for this data . However, I worked in Clampfit, Excel, and Matlab to compare traces, average them across multiple runs, and determine signal size between different trials.

PAGE 32

23 III. RESULTS AND DISCUSSION Results Completed Cuffs 3D Printed Cuffs. Several cuff designs were fabricated and tested over the course of this project. The first was designed in SolidWorks and printed on our la with V eroblack plastic material. This initial design ( Figure 10 A) used the FINE cuff as a basis, and contained an opening on one side through which the nerve could slip through. Unfortunately, the printer was not capable of printing such fine features, and most of the cuffs did not print correctly . In a few, after over an hour of careful cleaning and del icate manipulation, the opening on the side was visi ble (center cuff in Figure 10 B); however, the cuffs were extremely brittle and broke easily. Several broke before I was able to clean one off well enough to try and implant it into a mouse ( Figure 11 ). Figure 10 : First iteration: 3D printed cuff. A) Solidworks model of cuff. B) Printed results. Ruler marks are mm increments. The translucent material surrounding the cuffs on the right is supp ort material from the printer. A B

PAGE 33

24 While I was holding the cuff open to implant around the nerve, it tore along its short axis. This happened to several cuffs I attempted, which led me to believe that the material we were using for this cuff was too brittle for this use. In addition, as can be seen in Figure 11 A, the cuff is very bulky compared to the sciatic nerve and the body cavity in which it resides. Because o f these major design flaws, I decided to attempt other designs that had greater chances of success. Molded Cuffs. The previous cuff had no points of attachment for either LEDs or optical fibers; it likely would not have been possible, due to the poor spa tial resolution of the final product. However, in working toward my next iteration, I decided to add support material for optical fibers, to be added into th e cuff after fabrication. For these, I create d a 3D printed mold in V eroclear, and then planne d to mold the cuff out of silicone. Silicone is much more flexible than any of our available printing materials, and has a similar durometer to th e nerve and surrounding tissue. This is advantageous because it would have minimized damage to the nerve from the cuff while implanted . However, my first attempt was unsuccessful, as seen in Figure 12 . I did not have a clear understanding of how molding should work, and consequently made the overall size of the mold Figure 11 : A) 3D printed cuff in euthanized mouse . B) 3D printed cuff after breaking, in petri dish with excised sciatic nerve. A B

PAGE 34

25 far t oo small, even for such a small part. I filled both halves of the mold with S ilicone s Inc. XP 592 and left it in a pressure pot overnight to cure. Even though the silicone solidified, part of the cuff fell apart when separating the two halves, and the cuff would not release from the mold. I was unable to test this cuff in a mouse. The fine details of the cuff appeared better in the mold than they did in the 3D printed cuff; the opening to the side is visible on the right half of the mold, and the extrusio ns for the fibers are also clearly visible. However, there would have been enormous modifications to make to my cuff and mold design before I found a workable solution. In addition, a nother lab member had struggled with molding for several months with litt le success. Because of these reasons , and because I would only be performing acute experiments rather than long term implantations, I decided that matching the durometer of the cuff to the surrounding tissue was not necessary for the scope of this project , and decided to take another approach to creating my cuff . Figure 12 : Molded cuff. A) Soli dworks model of one half of the mold. B) The mold in practice, with the failed cuff still visible on the right half. A B

PAGE 35

26 Microprobes Cuff . The third iteration I attempted was a fiber based cuff sent by Microprobes, made completely out of silicone. This cuff ha d a several positives, but one critical drawba ck . The fact that the cuff came with an optical fiber was extremely helpful, because it eliminated my need to stably install it myself. It also had very good flexibility, and never broke or tore when I was handling it, even while in a mouse ( Figure 14 ) . Between the two variants, the Figure 13 : Stimulation cuffs received from Microprobes, with a pen for scale. They are identical except for the extra support around the optical fiber on the left. The horizontal bars are metal placeholders for the nerve while outside the body. A B Figure 14 : Implanting the stimulation cuff into a mouse. A) Cuff next to the nerve, showing the sli t allowing the cuff to be opened (indicated by white arrow). B) Cuff successfully implanted around the nerve.

PAGE 36

27 cuff with the extra support material around the optical fiber (on the left in Figure 13 ) was easier to work with, because I could grip that silicone wit h forceps without fear of damaging the fiber. However, the cuff had a drawback that made it nearly unusable: the length of fiber installed was far too short for any functional use. In order to connect the cuff to a laser, an FC/PC connector on the free en d of the fiber is required. Due to the short length of fiber attached to the cuff itself, I was unable to attach it directly to the cuff. Instead, I attached the connector to a long length of independent fiber and then lined the two fibers up in a micro manipulator to try and pass as much light through as possible. The results of this can be seen in Figure 15 ; some light was able to pass through to the cuff, and it is illuminated. However, a power measurement at t he cuff was not possible due to its design the slit along the cuff was at 90° to the fiber, and I could not reliably open the cuff far enough to get an accurate measurement . However, it was clear that the power loss at the fiber junction was very large. Th is situation was far from ideal; working with bare 60 µm diameter fibers requires extreme care, and even with that, the fiber is extremely brittle and prone to break. This can be Figure 15 : Shining laser light (473nm) from connected fiber (on right) to nerve cuff . I used a micromanipulator groove to line up the fibers. The blue light on the micromanipulator plate shows where the two fibers were pushed against each other.

PAGE 37

28 seen in Figure 15 ; the FC/PC connec ted fiber was originally 1 2 feet long, but it broke several times along its length until it was reduced to a few inches. This problem could have been easily remedied by a second iteration of the cuff from Microprobes. I sent them 60 µm fibers with protect ive jackets around them ( Figure 1 B), with the request that they simply install the fiber as is in the existing cuff. However, they never followed up on this despite repeated prompting. Thus, I decided to test the existing fiber with the laser and micromanipulator once in a euthanized animal, to see whether this avenue was worth pursuing. This setup can be seen in Figure 16 . Though this system mechanically worked, it was not easy to build . In addition, I had no way of recording CAPs from the nerve while it was in the mouse, so I had to visually watch for a muscle twitch due to activation of motor axons. I did not see anything; this could have been for a variety of reasons (including nerve degeneration post mortem, co contraction of different muscle groups, or a signal too small to generate a noticeable contraction). However, these discouraging LEDs ins tead, meant we decided against pursuing this line of inquiry. Figure 16 : Implanting the Microprobes cuff in a euthanized mouse.

PAGE 38

29 In House Cuffs . I decided at this point to make my own cuffs in the lab, so that I had total control over the design and fabrication process. It would also increase turn around time, becau se The first design involved simply an LED, two wires, and 1 mm ID polyamide tubing with a hole for the LED as in Figure 4 A . This design can be seen in Figure 17 . It was relatively crude, with a lot of excess solder spilling over the sides of the LED. In addition, even though I was able to make two discrete solder joints, the LED it self was not able to withstand strain on the tabs; the with the slightest strain , rendering it useless. Thus, I was unable to test this cuff functionally in mic e, though I was able to ascertain that the tubing itself was an appropriate ID and length for the sciatic nerve . Strain Relief Strategies . It was clear that I needed to do something to relieve the strain on so the cuff could be reliably han dled and implanted. Working with Steve Figure 17 : The first iteration of my in house cuffs. A) Close image of LED and wiring on cuff. B) Cuff implanted around sciatic nerve. A B

PAGE 39

30 Huddle, I developed and tested two different methods of strain relief. Early concept sketches are in Figure 18 . The first (on the right in Figure 18 ) involved a hollow , 3D printed tube with a window in the middle, out of which the LED faced. First, two wires would be pushed into the tubing one from each end until they w ere about a mm apart (the length of the LED). They would be gl ued in at the point of entry into the tube; this would relieve strain on the LED, since the tube would Figure 18 : Conceptual designs of strain r elief strategies. The first I attempted was the right side; this shows the hollow tubing, the hole cut in the middle for the LED, and the wires traveling through the tubing. On the left is the design I eventually decided on. Strain gauges (wire pads) to p revent anything from pulling on the LEDs, as part of a more compact strain relief system. (Drawn by Steve Huddle) Figure 19 : First strain relief strategy of gluing everything into flexible tubing.

PAGE 40

31 absorb all of it. Then, the LED pads were soldered to each wire. More glue was applied to the entire assembly, and it was allowed to dry. This system wor ked better than no strain relief at all; however, it was not sufficient to be handled during implantation for two reasons. I was able to carry it in a bag to lab meeting without it breaking, but a soldering pad on the LED broke off on the walk back to the lab (visible on the solder in Figure 19 , indicated by red arrow) . to fully protect the cuff from damage . In addi tion, due to the bulk of the 3D printed tubing, it was unclear how I wou ld attach this to the cuff without greatly increasing its volume. Thus, I decided against this design, and worked on creating a prototype using the second strain relief method. This involved using strain gauges (as seen in Figure 4 B) and a generous amount of epoxy to relieve strain on the LED. By not attaching the long wires to the LED itself, instead putting in the strain gauges as a middle man, the LED should never break. This is indeed what I found; the solder p ads on the LEDs did not break from the LEDs in this design. Images of this cuff can be seen in Figure 20 . The drawbacks to this design are the large amount of exposed solder and the increased surface area it takes up on the cuff. Both of these problems are due to the strain gauges; however, it was the best design I had so far, and at the time it did not seem possi ble to hand solder these LEDs satisfactorily and prevent them from breaking. Thus, I built several of th em with one and two LEDs on the surface.

PAGE 41

32 Pre Soldered LEDs . Soon after the strain gauge cuffs were completed, I found pre soldered LEDs on eBay ( Figure 21 ). Because they were the exact same type of LED (0402, a standard 1x0. 5x0.5 mm size) as the ones I had previously been using, I immediately ordered them, hoping that they would streamline the fabrication process significantly. They worked perfectly; the solder joints and LED solder pads were solid, and I had to apply a signi ficant amount of force, pulling the two wires apart, to break the LED. These eliminated the strain gauges from the design, and removed the need for soldering altogether. In addition, due to the smaller footprint, I was able to make a cuff with three LEDs o n it ( Figure 22 ). Four LEDs would have also been possible, if I had spaced them closer together. Figure 20 : Strain relief using strain gauges glued directly to the cuff. A) Close image of LED, strain gauges on either side, and short pieces of wire connecting them. B) The full setup, with the LED illuminated. A B

PAGE 42

33 The only issue with the se LEDs was that the wires attached to them were only a foot long; this was not long enough t o comfortably reach from the chamber to the Arduino I was using to power the circuit. However, this was easily remedied by soldering additional wire to the nois e. At the same time, because I had LEDs that took up less space, I decided to order some 0.7 mm ID tubing from Microlumen to better fit the adult mouse sciatic nerve. Because the pre Figure 21 : Pre so ldered LED glued onto a 1 mm ID cuff. Figure 22 : Three LED cuff illuminated and encapsulating the nerve. Also pictured is the sciatic electrode.

PAGE 43

34 soldered LEDs were slightly taller than I was expecting, I ordered a sli ghtly thicker wall (0.3 mm) than the 1 mm tubing. I planned to inset the LED within the tubing for better stability and a more streamlined design. However, while the tubing was the correct size for the nerve and fit the LEDs very well, it was significantl y more stiff due to its thicker walls. Because of this, it was much harder to cut out openings for the LEDs without deforming the material, and it was also impossible to cut a properly sized slit along the length of the cuff without damaging the integrity of the tubing. This can be seen clearly in Figure 23 ; I was unable to cut a parallel line to the slit, preventing the opening from being large enough. This, in addition to the material being more difficult to handl e, meant it would have been impossible to get the nerve inside the cuff without damaging it. In addition, the smaller tubing size made it more difficult to fit extra LEDs on the cuff; on the cuff that I made, I was only able to put two. These two reasons c onvinced me to stay with the 1 mm ID cuff, even though it is slightly large for the sciatic nerve, because the ease of use and safety of the nerve supersede a perfect fit. The final cuff design was the cuff in Figure 22 . Figure 23 : Pre soldered LEDs on a narrower, thicker walled cuff. Note the tubing deformation along the longitudinal cut.

PAGE 44

35 Electrical Stimulation Results To gain a baseline to which to compare my cuff, I recorded CAPs from ChAT Cre ChR2 nerves following electrical stimulation. Due to trouble with the CAP recording system, of the 6 nerves I recorded from, only 3 responde d to electrical stimulation. Despite attempts to minimize electrical artifact, there was still a consistent stimulus artifact present in all signal s, visible as the sharp positive slope in Figure 24 . This artifact ranged from monophasic to triphasic, depending on the trace ; it possibly varied based on the amplitude of the stimulation voltage. Figure 24 : Representative CAPs following electrical stimulation. The signal is represented by the red box; the rest of the curve is stimulus artifact. -5 -3 -1 1 3 5 7 9 15 17 19 21 23 25 27 Voltage (mV) Time (ms) Typical Signal After Electrical Stimulation

PAGE 45

36 Figure 25 and Figure 26 . Two types of signal data were considered: the peak signal (in mV) and the area under the signal (in mV · ms). I desired to know the most consistent way to measure CAPs following stimulation. Based on this small sample size, the area under the signal was a much more consistent measurement of nerve response; the peak response values in Figure 25 varied widely depending on which nerve the recording came from. For example, stimulation of nerve 6 resulted in much higher peak signals than nerves 1 and 2, which resulted in the nearly 4 mV response at only 1 V of stimulation. The large error bars in Figure 25 are due to this variation , as measuring signals from the same nerve even an hour separated in time gave consistent responses . Figure 25 : Signal peaks following electrical stimulation (n=3). 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 2 4 6 8 10 12 Peak signal (mV) Stimulation voltage (V) Peak Signal vs. Stimulation Voltage

PAGE 46

37 Measuring the area under the positive half of the signal proved to be a more consistent way of measuring CAPs between nerves, as seen in Figure 26 . There was still some variation among different nerves (nerve 6, again, recorded uncommonly large signals after 0.1 and 1 V of stimulation) but it appears that measuring the area under the signal rather than the peak has normalized the response somewhat due to the width of the signal . This is shown by the sma l ler error bars and more consistent rise in response when increasing the stimulation voltage. Laser Stimulation Results My very early experiments included stimulating ChAT Cre ChR2 nerves with an optical fiber connected to a 473 nm laser, as a precursor to the cuff and to establish a n optical baseline for future experiments . Optical fiber stimulation has the advantage of completely lacking electrical artifact in the signal (best seen in Figure 27 ), even when the fiber is submerged and Figure 26 : Recording area under the signal while increasing stimulation voltage. (n=3) 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 2 4 6 8 10 12 Area Under Signal (mV*ms) Stimulation Voltage (V) Area Under Signal Vs. Stimulation Voltage

PAGE 47

38 butted directly against th e nerve. It also requires less handling of the nerve to position it correctly. However, its drawbacks (design of th e cuff and portability) meant I decided against using laser stimulation. As these results were preliminary, exact details of the experiments were not properly notated. In particular, power measurements were not taken for the laser to determine the power being shone on the nerve in watts. However, other information obtained from these tests was valuable in informing later cuff experiments. Figure 27 shows the recorded signa l over time at different stimulation frequencies. Unlike electrical stimulation, optical stimulation cannot be driven at high frequencies indefinitely without a severe drop off in signal. Anything above a 20 Hz stimulation p attern will likely drop to zero in less than one second. Figure 28 summarizes experiments performed to determine the optimal laser pulse length during stimulation. Based on these res ults, it was determined that a 3 5 ms pulse len gth was Figure 27 : Varying frequency of laser st imulation to see degradation rate of the signal. A D: 10 40 Hz, respectively. A B C D

PAGE 48

39 appropriate for ChR2 activation; 5 ms was used in all cuff experiments. However, in future, a 3 ms pulse length will likely be sufficient for maximum activation. Cuff Stimulation Results I obtained signals from four nerves using my recording suction electrode following optical stimulation from my 3 LED cuff ( Figure 22 ). The initial barrier to recording these signals was overwhelming electrical noise; I eventually determined that this was due to the exposed wires of the LED sitting submerged in the bath. Other contri buting factors include more distal exposed solder joints, and the breadboard I initially used to connect the LED to the Arduino. The Figure 28 : Varying stimulation pulse length at 10 Hz to determine the optimal value.

PAGE 49

40 initial noise had peaks in excess of 600 mV; the signal after significantly dampening this noise is in Figure 29 . Figure 30 : Varying frequency for prolonged cuff stimulation of ChR2+ axons. -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 1000 2000 3000 4000 5000 6000 Area under signal (mV*ms) Time (ms) 10 Hz, Area Under the Signal Over Time [Nerve 5] 10 Hz 20 Hz 30 Hz 40 Hz -3 -2 -1 0 1 2 3 167 169 171 173 175 177 179 Voltae (mV) Time (ms) Typical Signal After Optical Stimulation -3 -2 -1 0 1 2 3 159 161 163 165 167 169 171 Voltage (mV) Time (ms) Measured Noise After Optical Stimulation Figure 29 : A typical signal following op tica l stimulation. On the left, t he nerve is in the recording electrode; on the right, it is ejected. .

PAGE 50

41 Just as with laser stimulation, I saw a drop in signal over time dependent on the frequency of stimulation. This is due to the limitations of ChR2 ; newer, more efficient versions of chann elrhodopsin can be stimulated at higher frequencies (for example, Chronos) (Hight et al., 2015) . This phenomenon can be seen in Figure 30 . It was difficult to compare different nerv very different artifacts and signal sizes. For example, Nerve 5 gave extremely consistent signals throughout the experiment. Nerve 6, tested immediately after on the same day, performed poorly a nd resulted in very small signals in the same setup. This dilemma is best visualized in Figure 31 , which compares 10 Hz signals from three different nerves. Therefore, I have compared s ignals within the same nerve rather than between different samples, in an effort to visualize trends and gain initial data. Figure 31 : 10 Hz signal s from three different nerves. Nerve s 3 and 5 stay relatively constant, but nerve 6 drops off quickly. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0 1000 2000 3000 4000 5000 6000 Area under the signal (mV*ms) Time (ms) 10 Hz, Area Under the Signal vs. Time [All Nerves] NERVE 3 LED 3 NERVE 5 LED 1 NERVE 5 LED 2 NERVE 5 LED 3 NERVE 6 LED 1 NERVE 6 LED 123

PAGE 51

42 Frequencies of up 20 Hz in healthy nerves were able to continue for 30 seconds or more with a 50% drop in signal ( Figure 32 ) . 10 Hz signals , in mo st cases ( Nerve 5 in Figure 31 ) degraded slightly but maintained over 50% of its peak. Frequencies of 30 Hz or more, however, r esulted in degradation of the signal at varying speeds, depending on the frequency. In addition to analyzing signal size, I also wanted to know whether different LEDs, or multiple LEDs, would affect the CAP. I tested each LED individually, as well as combinations of two and all three. These results are all summarized in Figures 31 34 . At thi s time, it does not appear that different LEDs stimulate larger groups of axons, nor does increasing the number of LEDs. However, it appears that LED 3 may give a slightly smaller signal than LEDs 1 and 2 in a few graphs ( Figure 33 A a nd Figure 34 A ). This is of interest because LED 3 was the closest to the recording electrode , and illuminated the proximal end of the tibial nerve rather than the sciatic. A smaller evoked signal may be due to less axons being pr esent in that part of the nerve, or a larger distance from the LED to the nerve due to the over sized cuff, decreasing the power of the LED by the time it reaches the nerve. However, m ore experiments must be performed to analyze this possibility; it appear ed in several LED 3 s timulations, but not all.

PAGE 52

43 Figure 32 : 20 Hz stimulation over 30 seconds. The signal dropped by approximately 50% but continued for the length of stimula tion.

PAGE 53

44 Figure 33 : Area under the signal following 10 Hz stimulation. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0 1000 2000 3000 4000 5000 6000 Area under signal (mV*ms) Time (ms) Comparing Different LEDs at 10 Hz [Nerve 5] LED 1 LED 2 LED 3 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0 1000 2000 3000 4000 5000 6000 Area Under Signal (mV*ms) Time (ms) Comparing Different Numbers of LEDs at 10 Hz [Nerve 6] LED 1 LEDs 123

PAGE 54

45 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0 500 1000 1500 2000 2500 3000 3500 Area under signal (mV*ms) Time (ms) Comparing Different LEDs at 20 Hz [Nerve 3] LED 1 LED 2 LED 3 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0 500 1000 1500 2000 2500 3000 3500 Area under signal (mV*ms) Time (ms) Comparing Different LEDs at 20 Hz [Nerve 4] LED 1 LED 2 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0 500 1000 1500 2000 2500 3000 3500 Area under signal (mV*ms) Time (ms) Comparing Different LEDs at 20 Hz [Nerve 5] LED 1 LED 2 LED 3 LEDs 1&2 LEDs 123 Figure 34 : Area under the signal following 20 Hz stimulation.

PAGE 55

46 Figure 35 : Area under signal following 30 and 40 Hz stimulation. -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 500 1000 1500 2000 2500 Area under signal (mV*ms) Time (ms) Comparing Different Numbers of LEDs at 40 Hz [Nerve 6] LED 1 LEDs 123 0 0.1 0.2 0.3 0.4 0.5 0.6 0 500 1000 1500 2000 2500 Area under signal (mV*ms) Time (ms) Comparing Different LEDs at 30 Hz [Nerve 5] LED 1 LED 2 LED 3

PAGE 56

47 Figure 36 : Area under the signal following 60 Hz stimulation. -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0 100 200 300 400 500 600 700 Area under signal (mV*ms) Time (ms) Comparing Different LEDs at 60 Hz [Nerve 4] LED 1 LED 2 LED 3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0 100 200 300 400 500 600 700 Area uinder signal (mV*ms) Time (ms) Comparing Different Numbers of LEDs at 60 Hz [Nerve 6] LED 1 LEDs 123

PAGE 57

48 In summary, the typical initial voltage of a cuff induce d signal is between 0.5 and 0.7 mV · ms . This decreases over time at a rate dependent on the stimulation frequency of the cuff. Any stimu lation frequency above 20 Hz will likely not be viable for long term stimulation; 10 Hz stimulation causes the signal to stay relatively constant, due to the 95 ms off time between stimulations that allow ChR2 to recover. An interesting observation seen i n several optical (and electrical) stimulations was a double peak, as seen in Figure 37 . These typically appeared clearly at the beginning of a CAP train, but degraded or disappeared entirely several hundred milliseconds later. Th ough thi s will need further testing, it may be two distinct groups of CAPs (for example, from larger diameter and then smaller diameter axons) traveling at different speeds down th e nerve. As the signal d ecreases over time, the secondary, smaller group may also see attrition until it is undetectable. -3 -2 -1 0 1 2 3 4120 4125 4130 Voltage (mV) Time (ms) Two Peak Optical Signal Over Time -3 -2 -1 0 1 2 3 214 219 224 Voltage (mV) Time (ms) Two Peak Optical Signal Figure 37 : Double peak signals during optical stimulation.

PAGE 58

49 Discussion Cuff Design Designing and iterating the nerve cuff was the bulk of this stu dy , and required the most trial and error . At first, my only goal was to successfully solder wires to LEDs, and glue these to tubing to make a cuff. However, as the problems with my design s continued to grow, I began worrying about things such as the size of the tubing; the width of the opening; how much un insulated solder was exposed to the air; the angle of the LEDs relative to each other and the saline bath; and the ultimate usability of the cuff whether the nerve could be successfully installed inside of it , and whether it could stimulate the nerve and allow me to record the signals . Many of these things overlapped, and all of them were vital to the success of later CAP recordings . My main trouble with the cuff, ultimately, was the noise it introduced i nto the recorded signal due to its electrical connections to the Arduino. I spent several full days of work determining what exactly was causing the noise and how I could best minimize it to Figure 29 . The first solution that help ed was simply putting electrical tape around all larger exposed solder joints. This included all solder joints distal to the cuff which I used to elongate the wires from the Arduino to the cuff. Sylastic insulation was a promising suggestion; I just needed to paint it over the small solder joints directly on the cuff and let it dry over the weekend. In theory, this should have totally insulated all of the exposed solder and prevented it from interfering with the signal. However, because the Sylastic we had in the lab was several years old, it did not harden properly over the weekend, rendering it useless.

PAGE 59

50 Dow Corning silicone grease worked marginally well; I painted this over the exposed proximal solder joints, and the size of the artifact went down slightly . It was not a perfect solution (it made the cuff slick and sticky to handle, and was not designed to dry), but for acute experiments, it worked reasonably well at reducing the noise. The final solution we found was simply reducing the saline levels around the cuff so that hydrophobic Parafilm, and placing the electrodes carefully on the edges of it such that both leads were submerged in small puddles of saline, but the LEDs were not. This is best seen in Figure 9 . This solution, combined with the electrical tape and silicone grease, resulted in signals similar to Figure 29 . This process of noise dampening was laborious; even after I realized the solution, properly placing the grease and sucking up the saline without damaging the nerve or the cuff was difficult. In future cuff iterations, decreasing exposed solder at all joints sho uld be the priority. Things to Consider When Comparing Signals There are several points to consider when comparing CAP recordings, especially between nerves and between electrical and optical stimulation. Importantly, each nerve can be vastly different fr om the last in nearly every aspect. The diameter of both the sciatic and tibial nerves can vary depending on the age and size of the mouse; a young adult mouse will be smaller than a year old animal. Such a change will make a noticeable difference in suc tion electrode fit, and thus how much of the signal is recorded. The dissection of the nerve is vitally important to its later performance during recording. A nerve with excess connective tissue, muscle, or fat still attached to it will perform poorly dur ing stimulation, especially optical stimulation. This is because the excess tissue diffuses the

PAGE 60

51 light before it can reach the nerve, meaning that fewer axons will be stimulated. In addition, if too much connective tissue is present around the ends of the n erve, it may not easily fit into the suction electrodes. Handling of the nerve during and after dissection is equally important. The nerve must be kept moist with a saline solution so that it does not dry out or become damaged due to the degrading muscle t issue around it. Dissecting both the left and right nerves as soon as possible after euthanasia is optimal; keeping them in a saline bath and testing one after the other is much better than dissecting and testing one, and then dissecting and testing th e other. Regarding handling of the nerve, it is important that the nerve itself be handled as little as possible. In addition, care must be taken not to stretch the nerve. During extraction, the sciatic end of the nerve may be gripped with forceps and gen tly pulled; however, a fresh cut should be made at that end of the nerve with a razor blade such that the end of the nerve that will go into the electrode is fresh and intact. During manipulation, especially into the cuff, the nerve should not be pinched, stretched, or jammed; grip connective tissue whenever possible. At least one of the nerves (nerve 6) used in this study may have been impacted by this, resulting in its lower optical signals than other nerves. Its electrical recordings were as high as othe rs; thus, it is likely that the nerve was damaged during implantation into the cuff, after electrical recording was completed. Nerve health is not the only thing one must consider when comparing CAP recordings. Simple variations in nerve anatomy, how the n erve is fitted into the electrode and cuff, and how long the nerve has been stimulated may also impact both the artifact and the signal. Thus, when possible, it is always best to compare signals within the same nerve.

PAGE 61

52 In addition, during data analysis, due to the irregular artifact seen in nearly all optical signals, I used software. Though care was taken to place the cursors as accurately as possible, introducing any human intera ction into such data collection could introduce significant error. Finally, w hen comparing electrical and optical cuffs, one must keep in mind how many axons will be stimulated by the light. Electrical stimulation, in theory, activates 100% of axons within the nerve, while optical stimulation, in t his case, only activated motor axons, ignoring sensory and sympathetic neurons. In the sciatic nerve, an estimated 6 % of axons are motor neurons (Schmalbruch, 1986) . However, I cannot simply com pare 6% to 100% of axons within the nerve. This is due to the myelination of some axons within the nerve. According to that same study, approximately 29% of axons within the sciatic nerve were myelinated (including motor axons and myelinated sensory axons) . Unmyelinated axons propagate sig nals significantly slower than their myelinated counterparts; thus, the signal I saw immediately after both electrical and optical stimulation axons are motor neuron s . When comparing the obtained electrical and optical stimulation data, the signals from both cuff and laser stimulation began at 0.6 mV · ms, while the electrical signals were about 2.5 mV · ms. These numbers align closely with the make up of the sciatic nerv e. Future experiments will need to be performed to confirm this, but it is possible that my cuff is indeed stimulating 100% of motor axons within the nerve. In this case, further major iterations of the cuff may not be necessary; designs that minimize nois e and streamline the system as much as possible are all

PAGE 62

53 that would be needed. Fabricating a cuff that contains several LEDs placed at angles to each other will ascertain that a larger signal cannot be obtained; however, such a design may not be necessary. In fact, a cuff with only one LED on it may be sufficient to activate all of the mouse Should future experiments move to larger animals, however, it would be worthwhile to test whether more LEDs at angles are needed due to the larger diameter of the sciatic nerve. The Shape of the Signal A biphasic signal in the response after both electrical and optical stimulation was noted. This is due to the nature of the recording electrode, as described by the McGill Physiology Virtual Lab . In brief, the shape of the CAP is determin ed by the current measured in the saline between the two leads of the recording electrode. The wire external to the glass pipette is the ground or reference electrode, while the one inside the pipette is the recording electrode. In essence, the two leads of the electrode measure the current flow in and out of the section of the nerve that they are closest to. Thus, if the nerve is stimulated extremely distal to the recording electrode (for example, on the sciatic end by the stimulating suction electrode), it will take some time for the signal to propagate down the nerve to the recording electrode (action potentials travel at at least 10 m/s, or 10 mm/ms, in myelinated axons) (Purves et al., 2001) . Because the sciatic and tibial nerve ar e 2 3 cm in length in mice, it will take several milliseconds for the signal to propagate to the recording electrode. The downward slope seen in all sample action potentials is due to depolarization of the nerve before the action potentials arrive at the e lectrode. When an action potential begins, there

PAGE 63

54 is a large influx of Na+ ions at the initiation of the signal. This influx then diffuses down its voltage gradient both ways down the nerve in order to depolarize the membrane further. The negative signal (o r outward current of positive ions) seen in the signals is these Na+ ions traveling to the distal end of the nerve and then diffusing back across the membrane to return to the saline. This is seen as an outward current, even though there is no true action potential at that time proximal to the recording electrode. Eventually, the signal reaches the recording electrode, and there is a large influx of Na+ ions at the electrode due to the CAP; this is the large peak seen in all signals. An important question t o consider is why the signal tapers off during optical stimulation with both laser and cuff stimulation , and not with electrical stimulation . The answer lies in the way ChR2 functions. go es through its conformational changes (Lorenz Fonfria et al., 2013) . It was determined that there are five different states ChR2 exists in; these are labelled the resting ChR2 state, and P 1 through P 4 . Most of these states have a half life of 10 m s or less; however, P 4 , the state of inactivation immediately following channel closure, has a half life of 20 seconds. P 4 is only entered by about 25% of ChR2 within a nerve at a given time ; the other 75% of ChR2 goes directly from P 3 to the ChR2 resting state. It is not yet known what causes this split pathway; it has been hypothesized that it is related to the re protonation of ChR2 over time (Lórenz Fonfría & Heberle, 2014) . 4 , lasted less than 15 ms, it should be able to activate at much higher frequencies than seen in these experiments. However, this P 4 frequencies, P 4 would be utilized more often than at lower frequencies. This would explain the

PAGE 64

55 rapid drop in signal at higher frequencies; as more ChR2 goes into this ina ctivation state, there is less and less of it to allow ions to pass in and out of the cell. Each of these studies mentioned that such a long inactivation state is highly unusual for a rhodopsin, so it is likely that newer rhodopsins such as Chronos have a much shorter P 4 stage. With this, the rhodopsin could be stimulated at higher frequencies without as much loss. However, to my knowledge this has not been confirmed with scientific study. Limitations of Study and Future Work Because this study was explorat ory and largely based around cuff design, there are several components of the design and experiments that can be improved. Also of note are expectations for future experiments, including the immediate future and the lo ng term goal of this project. The cuff, while functional, had a relatively crude design, with everything cut and glued by hand. If this could somehow be manufactured instead, it would likely decrease the exposed solder joints and create a tighter fit for the LEDs to shine through the tubing. Bo th of these would benefit stimulation and recording by minimizing room for error within the cuff itself. Creating a cuff with at least four LEDs on it would be a good test to ensure that a larger signal cannot be obtained . With a 7 mm cuff, using the pre s oldered LEDs, it would certainly be possible to fit four or possibly five LEDs along its length. In addition, staggering the LEDs at different angles to the cuff opening would be advantageous to ensure that as many axons are illuminated as possible. Howeve r, using this design would necessitate an extremely tight insulation around all LED solder joints, because one or two would likely be required to be submerged in saline.

PAGE 65

56 Using tubing that is properly sized for the sciatic nerve may increase nerve illuminat ion, decrease overall size, and improve cuff fit by decreasing the distance from the LEDs to the nerve. The sciatic nerve ranges from 600 800 µm in diameter, depending on the size of the mouse; using a 700 µm ID cuff would be optimal. Time staggered light stimulation would also be something to explore, especially in conjunction with the angled LEDs. For example, take four LEDs at 60° to each other on the cuff, all flashing at 20 Hz. However, instead of all four flashing in sync, they will be staggered from each other by 10 ms. This may make it difficult to measure signals because of the near constant illumination of different parts of the nerve; however, it may result in more axons being activated because the nerve is not being overstimulated. In conjunctio n with the angled LEDs, it should be tested again whether stimulating different areas of the nerve or more of the nerve creates results, I saw that the average optical signal was 20% of the average electrical signal very si milar to expected results in an ideal experiment. However, it would be imprudent not to complete further experiments ensuring that all possible axons are being activated. Eventually, in vivo studies could be performed during a terminal or survival surgery. This would complicate recordings, and it is likely that an impaling recording electrode would need to be used to measure signals. Alternatively, if the signal is strong enough, a muscle twitch could be visualized, or muscle contraction could be directly m easured. The long term goals for this project are complex and potentially decades away; however, once they are realized, this system has the potential to change the way prostheses are used. Eventually, this system should be integrated into both upper and lower limb prostheses in order to provide sensory feedback to the user.

PAGE 66

57 sensory axons. To do this, an adeno associated virus injection targeted toward such axons would need to be approved by the FDA, and such injections would need to gain traction with the public before they could be widely used in medical settings. The person would then express channelrhodopsin exclusively in sensory axons ideally, exclusively in sensory axo ns of a single nerve of the arm or leg. A nerve cuff similar to mine would be implanted surgically around the nerve. Using wireless LEDs would be ideal, should These LEDs would be connected to pressure sensors on the prosthetic limb, and would light up when pressure is applied. It is possible that multiple cuffs could be implanted around several distal nerves in order to give even more specific feedback. For exam ple, a cuff attached to the median nerve would receive input from the prosthetic thumb, while one implanted around the ulnar nerve would be stimulated by the fourth and fifth fingers. Initial set up for such a system would likely be lengthy in order to es tablish a baseline for was great variation between mouse nerves, there will also be for humans. Thus, a baseline power output for the LEDs would be set onc e the cuff is implanted, potentially scaling up with greater pressure on the prosthesis. Different LEDs at angles to the nerve and potentially at different points along its length would be illuminated, and the user would report what the sensation felt like . In this way, clinicians could complete a map connecting parts of the nerve to brain activity, and wire the prosthetic accordingly to give intuitive sensory feedback. I performed no chronic studies on my cuff; however, other cuffs mostly electrically sti mulating have been tested over several years and have been found to be safe, so long as they

PAGE 67

58 do not impale the nerve. Thus, I would not expect major problems with this system causing tissue damage long term. Scar tissue and an immune response could cause p roblems for any implantable system, and must be monitored during early experiments to ensure it does not progress. In short, the cuff I have built lays a groundwork for potentially life altering prosthesis designs in the future. Giving prosthesis users a s ense of touch is of the utmost importance when considering usability and acceptance rate; this system promises to do exactly that, giving specific and intuitive feedback without damaging the fragile nervous system. Conclusion s In this study, I designed and iterated an optical stimulation cuff for use in the mouse sciatic nerve , based on LED light . I tested it on excised nerves and compared it to both electrical stimulation and stimulation with an optical fiber, and recorded signals from all three. Optical s timulation of both kinds was found to activate about 20% of the axons that electrical stimulation does; Optogenetic stimulation is advantageous due to its specificity. While electr ically stimulated signals also include myelinated sensory axons, optically stimulated signals include only axons that contain channelrhodopsin. Such a system can be implemented in a variety of ways due to viral injections, and is ideal for neural stimulati on. With this, one can have extremely specific axonal activation without using an extremely invasive, penetrating electrode that will cause tissue damage. Within optical stimulation, LEDs are preferred over optical fibers due to their portability and flexi bility within the mouse, minimizing damage to the surrounding tissue. The LEDs used

PAGE 68

59 in this study were wired; however, wireless LEDs exist, and can be implemented in nerve cuffs in order to create a fully implanted system. Overall, more experiments in this vein must be performed in order to statistically validate the study. Different cuff designs must be considered as well in order to determine that all of the target axons are being activated. The presented cuff design shows promising preliminary results, a nd is worthy of further study to validate this early data. Such a system, in the future, could be used as a viable method of delivering specific, time sensitive, intuitive sensory feedback to users of prosthetic limbs, revolutionizing the way prostheses ar e used around the world.

PAGE 69

60 REFERENCES Ahmad, A., Ashraf, S., & Komai, S. (2015). Optogenetics applications for treating spinal cord injury. Asian Spine Journal , 9 (2), 299 305. http://doi.org/10.4184/asj.2015.9.2.299 Antfolk, C., D Alonzo, M., Controzzi, M., Lundborg, G., Rosen, B., Sebelius, F., & Cipriani, C. (2013). Artificial redirection of sensation from prosthetic fingers to the phantom hand map on transradial amputees: Vibrotactile versus mechanotactile sensory feedback. IEEE Transactions on Neural Systems and Rehabilitation Engineering , 21 (1), 112 120. http://doi.org/10.1109/TNSRE.2012.2217989 Biddiss, E. A., & Chau, T. T. (2007). Upper limb prosthesis use and abandonment: a survey of the last 25 years. Prosthetics and Orthoti cs International , 31 (3), 236 57. http://doi.org/10.1080/03093640600994581 Biddiss, E., & Chau, T. (2007). Upper limb prosthetics: critical factors in device abandonment. American Journal of Physical Medicine & Rehabilitation / Association of Academic Physi atrists , 86 (12), 977 987. http://doi.org/10.1097/PHM.0b013e3181587f6c Bivar. (2017). Surface Mount LED Blue, 0402 Package. Retrieved from https://www.mouser.com/datasheet/2/50/Bivar 3 22 2017 SM0402BWC 1159202.pdf Bryson, J. B., Machado, C. B., Lieberam, I ., & Greensmith, L. (2016). Restoring motor function using optogenetics and neural engraftment. Current Opinion in Biotechnology , 40 , 75 81. http://doi.org/10.1016/j.copbio.2016.02.016 Carey, S. L., Lura, D. J., & Highsmith, M. J. (2014). Differences in my oelectric and body powered upper limb prostheses: Systematic literature review. Journal of Research Rehabilitation and Development , 52 (3), 17 19. http://doi.org/10.1007/s10488 012 0430 z.Public Cheesborough, J. E., Smith, L. H., Kuiken, T. A., & Dumanian, G. A. (2015). Targeted muscle reinnervation and advanced prosthetic arms. Seminars in Plastic Surgery , 29 (1), 62 72. http://doi.org/10.1055/s 0035 1544166 Compound Action Potential. (n.d.). Retrieved July 8, 2018, from http://www.medicine.mcgill.ca/physio/ vlab/CAP/recording.htm Cordella, F., Ciancio, A. L., Sacchetti, R., Davalli, A., Cutti, A. G., Guglielmelli, E., & Zollo, L. (2016). Literature Review on Needs of Upper Limb Prosthesis Users. Frontiers in Neuroscience , 10 , 209. http://doi.org/10.3389/fnins .2016.00209 Cree ® TR2432 TM LEDs Data Sheet CxxxTR2432 Sxx00. (2011). Retrieved from www.marktechopto.com/pdf/products/datasheet/Cree_TR2432 B.pdf D Anna, E., Petrini, F. M., Artoni, F., Popovic, I., Simani , I., Raspopovic, S., & Micera, S. (2017). A somat otopic bidirectional hand prosthesis with transcutaneous electrical nerve stimulation based sensory feedback. Scientific Reports , 7 (1), 10930.

PAGE 70

61 http://doi.org/10.1038/s41598 017 11306 w Deisseroth, K. (2011). Optogenetics. Nature Methods , 8 (1), 26 29. http: //doi.org/10.1038/nmeth.f.324 Dweiri, Y. M., Stone, M. A., Tyler, D. J., McCallum, G. A., & Durand, D. M. (2016). Fabrication of High Contact Density, Flat Interface Nerve Electrodes for Recording and Stimulation Applications. Journal of Visualized Experim ents , (116), e54388 e54388. http://doi.org/10.3791/54388 Engdahl, S. M., Christie, B. P., Kelly, B., Davis, A., Chestek, C. A., & Gates, D. H. (2015). Surveying the interest of individuals with upper limb loss in novel prosthetic control techniques. Journa l of Neuroengineering and Rehabilitation , 12 , 53. http://doi.org/10.1186/s12984 015 0044 2 Fischer, G., Kostic, S., Nakai, H., Park, F., Sapunar, D., Yu, H., & Hogan, Q. (2011). Direct injection into the dorsal root ganglion: Technical, behavioral, and his tological observations. Journal of Neuroscience Methods , 199 (1), 43 55. http://doi.org/10.1016/j.jneumeth.2011.04.021 Fontaine, A. K., Gibson, E. A., Caldwell, J. H., & Weir, R. F. (2017). Optical Read out of Neural Activity in Mammalian Peripheral Axons: Calcium Signaling at Nodes of Ranvier. Scientific Reports , (February), 1 11. http://doi.org/10.1038/s41598 017 03541 y Freeberg, M. J., Stone, M. A., Tyler, D. J., & Triolo, R. J. (2013). Chronic Response of the Cat Sciatic, Median, and Ulnar Nerves to a C ompliant, Composite Flat Interface Nerve Electrode (C FINE). In IEEE Neural Engineering Short Papers (p. 1). IEEE Engineering in Medicine & Biology. Retrieved from https://neuro.embs.org/files/2013/0518_FI.pdf Heo, D. N., Song, S. J., Kim, H. J., Lee, Y. J ., Ko, W. K., Lee, S. J., Kwon, I. K. (2016). Multifunctional hydrogel coatings on the surface of neural cuff electrode for improving electrode nerve tissue interfaces. Acta Biomaterialia , 39 , 25 33. http://doi.org/10.1016/j.actbio.2016.05.009 Hight, A. E., Kozin, E. D., Darrow, K., Lehmann, A., Boyden, E., Brown, M. C., & Lee, D. J. (2015). Superior temporal resolution of Chronos versus channelrhodopsin 2 in an optogenetic model of the auditory brainstem implant. Hearing Research , 322 , 235 41. http://doi .org/10.1016/j.heares.2015.01.004 Iyer, S. M., Montgomery, K. L., Towne, C., Lee, S. Y., Ramakrishnan, C., Deisseroth, K., & Delp, S. L. (2014). Virally mediated optogenetic excitation and inhibition of pain in freely moving nontransgenic mice. Nature Biot echnology , 32 (3), 274 8. http://doi.org/10.1038/nbt.2834 Iyer, S. M., Vesuna, S., Ramakrishnan, C., Huynh, K., Young, S., Berndt, A., Delp, S. L. (2016). Optogenetic and chemogenetic strategies for sustained inhibition of pain. Scientific Reports , 6 , 305 70. http://doi.org/10.1038/srep30570

PAGE 71

62 Jacques, S. J., Ahmed, Z., Forbes, A., Douglas, M. R., Vigenswara, V., Berry, M., & Logan, A. (2012). AAV8 gfp preferentially targets large diameter dorsal root ganglion neurones after both intra dorsal root ganglion an d intrathecal injection. Molecular and Cellular Neuroscience , 49 (4), 464 474. http://doi.org/10.1016/j.mcn.2012.03.002 Klapoetke, N. C., Murata, Y., Kim, S. S., Pulver, S. R., Birdsey Benson, A., Cho, Y. K., Boyden, E. S. (2014). Independent optical exci tation of distinct neural populations. Nature Methods , 11 (3), 338 346. http://doi.org/10.1038/nmeth.2836 Leventhal, D. K., Cohen, M., & Durand, D. M. (2006). Chronic histological effects of the flat interface nerve electrode. Journal of Neural Engineering , 3 (2), 102 113. http://doi.org/10.1088/1741 2560/3/2/004 Lewis, S., Russold, M. F., Dietl, H., & Kaniusas, E. (2012). User demands for sensory feedback in upper extremity prostheses (pp. 1 4). IEEE. Li, K., Fang, Y., Zhou, Y., & LIU, H. (2017). Non invasive Stimulation based Tactile Sensation for Upper extremity Prosthesis: a Review. IEEE Sensors Journal , 17 (9), 2625 2635. http://doi.org/10.1109/JSEN.2017.2674965 Lipschutz, R. D. (2017). Impact of Emerging Technologies on Clinical Considerations: Targeted Muscle Reinnervation Surgeries, Pattern Recognition, Implanted Electrodes, Osseointegration, and Three Dimensional Printed Solutions, (12). Llewellyn, M. E., Thom pson, K. R., Deisseroth, K., & Delp, S. L. (2010). Orderly recruitment of motor units under optical control in vivo. Nature Medicine , 16 (10), 1161 1165. http://doi.org/10.1038/nm.2228 L ó renz Fonfr í a, V. A., & Heberle, J. (2014). Channelrhodopsin unchained: Structure and mechanism of a light gated cation channel. Biochimica et Biophysica Acta Bioenergetics , 1837 (5), 626 642. http://doi.org/10.1016/j.bbabio.2013.10.014 Lorenz Fonfria, V. A., Resler, T., Krause, N., Nack, M., Gossing, M., Fischer von Mollard , G., Heberle, J. (2013). Transient protonation changes in channelrhodopsin 2 and their relevance to channel gating. Proceedings of the National Academy of Sciences , 110 (14), E1273 E1281. http://doi.org/10.1073/pnas.1219502110 Maimon, B. E., Zorzos, A. N ., Bendell, R., Harding, A., Fahmi, M., Srinivasan, S., Herr, H. M. (2017). Transdermal optogenetic peripheral nerve stimulation. Journal of Neural Engineering , 14 (3), 034002. http://doi.org/10.1088/1741 2552/aa5e20 Mallory, G. W., Grahn, P. J., Hachmann , J. T., Lujan, J. L., & Lee, K. H. (2015). Optical Stimulation for Restoration of Motor Function After Spinal Cord Injury. Mayo Clinic Proceedings , 90 (2), 300 307. http://doi.org/10.1016/j.mayocp.2014.12.004 Mason, M. R. J., Ehlert, E. M. E., Eggers, R., Pool, C. W., Hermening, S., Huseinovic, A., Verhaagen, J. (2010). Comparison of AAV serotypes for gene delivery to dorsal root ganglion neurons. Molecular Therapy : The Journal of the American Society of Gene

PAGE 72

63 Therapy , 18 (4), 715 24. http://doi.org/10.103 8/mt.2010.19 Michoud, F., Sottas, L., Browne, L. E., Asboth, L., Latremoliere, A., Sakuma, M., Lacour, S. P. (2017). Optical cuff for optogenetic control of the peripheral nervous system. Journal of Neural Engineering . http://doi.org/10.1088/1741 2552/aa 9126 Montgomery, K. L., Yeh, A. J., Ho, J. S., Tsao, V., Mohan Iyer, S., Grosenick, L., Poon, A. S. Y. (2015). Wirelessly powered, fully internal optogenetics for brain, spinal and peripheral circuits in mice. Nature Methods , 12 (10), 969 974. http://doi. org/10.1038/nmeth.3536 Nagel, G., Szellas, T., Huhn, W., Kateriya, S., Adeishvili, N., Berthold, P., Bamberg, E. (2003). Channelrhodopsin 2, a directly light gated cation selective membrane channel. Proceedings of the National Academy of Sciences of the United States of America , 100 (24), 13940 5. http://doi.org/10.1073/pnas.1936192100 Ozbay, B. N., Futia, G. L., Ma, M., Hughes, E. G., Restrepo, D., & Gibson, E. A. (2018). Three dimensional multiphoton imaging of brain activity in freely moving mice using a miniature microscope with variable focus lens (Conference Presentation). In T. G. Brown, C. J. Cogswell, & T. Wilson (Eds.), Three Dimensional and Multidimensional Microscopy: Image Acquisition and Processing XXV (Vol. 10499, p. 70). SPIE. http://doi.org /10.1117/12.2309388 Ozbay, B. N., Losacco, J. T., Cormack, R., Weir, R., Bright, V. M., Gopinath, J. T., Gibson, E. A. (2015). Miniaturized fiber coupled confocal fluorescence microscope with an electrowetting variable focus lens using no moving parts. O ptics Letters , 40 (11), 2553 6. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/26030555 Park, S. Il, Brenner, D. S., Shin, G., Morgan, C. D., Copits, B. A., Chung, H. U., Rogers, J. A. (2015). Soft, stretchable, fully implantable miniaturized optoelect ronic systems for wireless optogenetics. Nature Biotechnology , 33 (12), 1280 1286. http://doi.org/10.1038/nbt.3415 Peerdeman, B., Boere, D., Witteveen, H., Huis in `tVeld, R., Hermens, H., Stramigioli, S., Misra, S. (2011). Myoelectric forearm prostheses: State of the art from a user centered perspective. The Journal of Rehabilitation Research and Development , 48 (6), 719. http://doi.org/10.1682/JRRD.2010.08.0161 Puljak, L., Kojundzic, S. L., Hogan, Q. H., & Sapunar, D. (2009). Targeted delivery of pharmaco logical agents into rat dorsal root ganglion. Journal of Neuroscience Methods , 177 (2), 397 402. http://doi.org/10.1007/s11103 011 9767 z.Plastid Purves, D., Augustine, G. J., Fitzpatrick, D., Katz, L. C., Lamantia, A. S., McNamara, J. O., & Williams, S. M. (2001). Neuroscience (2nd editio). Sunderland, MA: Sinauer Associates. Retrieved from https://www.ncbi.nlm.nih.gov/books/NBK10921/ Pylatiuk, C., Schulz, S., & D ö derlein, L. (2007). Results of an Internet survey of myoelectric prosthetic hand users. Prosth etics and Orthotics International , 31 (4), 362 370. http://doi.org/10.1080/03093640601061265

PAGE 73

64 Sahyouni, R., Chang, D. T., Moshtaghi, O., Mahmoodi, A., Djalilian, H. R., & Lin, H. W. (2017). Functional and Histological Effects of Chronic Neural Electrode Impl antation. Laryngoscope Investigative Otolaryngology . http://doi.org/10.1002/lio2.66 Schmalbruch, H. (1986). Fiber composition of the rat sciatic nerve. The Anatomical Record , 215 (1), 71 81. http://doi.org/10.1002/ar.1092150111 Schofield, J. S., Evans, K. R ., Carey, J. P., & Hebert, J. S. (2014). Applications of sensory feedback in motorized upper extremity prosthesis: a review. Expert Rev Med Devices , 11 , 499 511. http://doi.org/10.1586/17434440.2014.929496 Thompson, C. H., Zoratti, M. J., Langhals, N. B., & Purcell, E. K. (2016). Regenerative Electrode Interfaces for Neural Prostheses. Tissue Engineering Part B: Reviews , 22 (2), 125 135. http://doi.org/10.1089/ten.teb.2015.0279 Towne, C., Montgomery, K. L., Iyer, S. M., Deisseroth, K., & Delp, S. L. (2013). Optogenetic control of targeted peripheral axons in freely moving animals. PloS One , 8 (8), e72691. http://doi.org/10.1371/journal.pone.0072691 Tyler, D. J., & Durand, D. M. (2002). Functionally selective peripheral nerve stimulation with a flat interface n erve electrode. IEEE Transactions on Neural Systems and Rehabilitation Engineering , 10 (4), 294 303. http://doi.org/10.1109/TNSRE.2002.806840 Tyler, D. J., & Durand, D. M. (2003). Chronic Response of the Rat Sciatic Nerve to the Flat Interface Nerve Electro de. Annals of Biomedical Engineering , 31 (6), 633 642. http://doi.org/10.1114/1.1569263 Wijk, U., & Carlsson, I. (2015). Forearm amputees views of prosthesis use and sensory feedback. Journal of Hand Therapy , 28 (3), 269 278. http://doi.org/10.1016/j.jht.20 15.01.013 Ziegler Graham, K., MacKenzie, E. J., Ephraim, P. L., Travison, T. G., & Brookmeyer, R. (2008). Estimating the prevalence of limb loss in the United States: 2005 to 2050. Archives of Physical Medicine and Rehabilitation , 89 (3), 422 9. http://doi.org/10.1016/j.apmr.2007.11.005

PAGE 74

65 A PPENDICES A. Extraneous Work GRIN Lens C uff for Optical Recording As part of my planned doctoral work, I was going to record from peripheral nerves optically, in addition to stimulating them. To this end, I helped develop, order, and test several variants of a 1 mm diameter gradient index lens ( GRIN lens) implantable cuff for use in the mouse sciatic nerve. A GRIN lens acts as a miniaturized, elongated objective lens for an optical system. These properties allow the microscope itself to be further away from the sample, allowing a part of the micr oscope to be implanted in an animal for imaging in vivo , as shown by Ozbay et al (Ozbay et al., 2015, 2018) . Our group desired to impl ant a GRIN lens to image from both the sciatic nerve and the vagus nerve. My focus was the sciatic nerve, and I designed and began to implement a peripheral nerve cuff to image using a fiber coupled microscope (FCM). Figure 38 sho ws the initial Figure 38 : Solidwor ks model of GRIN lens cuff.

PAGE 75

66 SolidWorks model of the GRIN lens, a sleeve within the cuff to allow easy removal of the lens, and the lens itself. The initial prototyping was performed in contact with the company Microprobes; however, they proved difficult to work with. The incorrect, unusable cuffs from Microprobes are shown in Figure 39 and Figure 40 . The cuffs from Microprobes were made of two sections of polyamide tubing, glued toget her to make a T shape. Along the length of tubing meant for the nerve was a slit to allow the nerve to be placed inside of the cuff. In the first iteration, the dimensions of both halves of the cuff are incorrect; Microprobes did not have a clear understa nding of what we wanted the cuff to look like. I attempted to clarify by sending the Solidworks model from Figure 38 ; however, the second cuff they sent us was similarly incorrect, as can be seen in B. The two types of tubing we s ent them were switched, leaving a 1 mm inner diameter to house a 700 µ m nerve, and tubing too small to fit the 1 mm GRIN lens. Accordingly, the hole cut into the cuff through which the lens would image was also far too small. The length of the cuff was als o twice as long as it should be, seen on the right Figure 39 : GRIN lens cuffs from Microprobes that could not be used. A) F irst iteration. B) Center and right, second iteration. A B

PAGE 76

67 compared to a previously sent stimulation cuff; a 5 7 mm length is appropriate for cuffing the sciatic nerve, but the cuff Microprobes sent was one centimeter long. The third cuff Microprobes sent ( Figure 40 ) had the correct dimensions for both the nerve and in a euthanized ChAT Cre tdTomato mouse and imaged through it. T hese results can be seen in Figure 41 A. However, also visible in this image is the edge of the polyamide tubing used to house the nerve. Upon focusing the FCM on the tubing surrounding the nerve rather than the ne rve itself, we discovered that the hole in the tubing was incorrectly placed and sized too small, partially obscuring the field of view ( Figure 41 B). The problems caused by this were twofold. First, imaging throu gh the polyamide tubing is not easily achieved, as seen in Figure 41 A . The axons would appear to continue past the obstruction, but no axons can be seen past the tubing. In addition, the GRIN lens works optimally when pressed directly against the tissue it is imaging. With the tubing separating the Figure 40 : Third iteration of the Micropro bes cuff. A) Implanted around the sciatic nerve of an adult ChAT ChR2 Cre mouse, A B

PAGE 77

68 two, there was a small (<1mm) gap between them, which was sufficient to make the images less bright and focusing the FCM much more difficult. We used a tdTomato expressi ng mouse specifically because it has very strong expression and fluoresces strongly under the FCM; had we used a mouse expressing YFP, we may not have seen anything. Attempts to carefully cut out the excess material to solve this problem were unsuccessfu l because of the fragility of the cuff. The glue holding the two portions together was part of the material blocking the GRIN lens, so breaking that away would have caused the cuff to fall apart. Thus, we decided that making GRIN lens cuffs ourselves would be cost and time effective. However, I had minimal involvement with this process. Dorsal Root Ganglion Exposure Surgery As a portion of my planned doctoral work, I was going to inject AAV s containing Chron os, a channelrhodopsin variant, into the dorsal root ganglion (DRG) of mice at spinal levels L3 and L4. This would give exclusive expression of Chronos in sensory axons of the sciatic nerve, allowing me to have two distinct populations of neurons within the nerve when injected into ChAT Cre GCaMP6s mice. This would have allowed me to stimulate sensory Figure 41 : Imaging through GRIN lens cuff. A) ChAT axo ns expressing tdTomato, captured through the GRIN lens and FCM. At the bottom right, indicated by the white arrow, the edge of the polyamide tubing is visible. B) The hole through which the nerve was meant to be imaged, as seen through the GRIN lens. B A

PAGE 78

69 axons via Chronos and record from motor axons using GCaMP6s, which simulates a real life situation in which optogenetics may be used for pro sthetic limbs. To this end, I needed to develop and carry out a DRG injection surgery, which had previously only been reported in rats (Fischer et al., 2011; Jacques et al., 2012; Mason et al., 2010; Puljak, Kojundzic, Hogan, & Sapunar, 2009) . Using these reported procedures as a guide, I worked with Dr. Cara Mitchell, a veterinarian f rom the University vivarium, to develop and practice a surgical procedure to access the DRG. A simplified diagram of the anatomy is seen in Figure 42 , adapted from (Puljak et al., 2009) . An incision was made alo ng the lumbar spine of the mouse (determined from the position of the iliac crest, which can be palpated). The muscle covering the spine was then carefully dissected away to expose the spine. The correct levels of the lumbar spinal cord were determined by examination of the spinous and transverse processes as well as location of the iliac crest (which is level with L6). For these preliminary experiments, only the L3 DRG was of consideration. A laminectomy was then performed in order to expose the DRG. Figure 42 B shows a mouse with much of the spine removed in order to be tter visualize the spinal cord and the left L3 DRG (indicated by white arrow). Not pictured is the right L3 DRG, which was also exposed to determine symmetry. It was determined that the secondary method of euthanasia used on these mice during preliminary e xperiments , cervical dislocation, breaks the cervical spine and may damage or dislocate the lower spinal cord in the process. During an earlier dissection, the cauda equina was visualized at a higher spinal level than anticipated . Thus, in later experiment s, an alternate form of secondary euthanasia was used, typically a bilateral thoracotomy.

PAGE 79

70 When my thesis was changed to a Master s level, this part of the project was deemed unnecessary, so I did not pursue further surgical experience in this area. B. CCTSI TL1 Predoctoral Fellowship From July 2016 to June 2017, this research was funded under a Colorado Clinical and Translational Science Institute (CCTSI) Pre D octoral TL1 Fellowship (TR001081). This funded most of my stipend for that year, my tuition, travel to a TL1 based conference for students, and extraneous expenses, totaling $30,682. This fellowship was created to encourage the translational aspect of bas ic science, allowing pre doctoral students who perform basic research to observe the tangible outcomes of their work. Thus, a condition of receiving this fellowship was to perform shadowing in at least one clinical setting relevant to my research for the y ear that I worked under the grant. B Figure 42 : Dorsal root ganglion exposure surgery. A) Diagram of rat spine, roughly analogous to mice; adapted from Puljak et al., 2009. SP = spinous process; SAP = superior articular process; IAP = inferior articular process; TP = transverse process; IC = iliac crest. B) Mouse spine blunt dissection to expose spinal cord and L3 DRG, indicated by large white organ in the cen ter and white arrow, respectively. B

PAGE 80

71 Several rehabilitation departments from throughout the hospital come together once a month to meet with children with limb amputation s or deformities, to determine their needs from a medical and prosthetic standpoint. I was not allowed into the same room that the appointments were occurring in instead, I was on the other side of an observation window due to the high number of people inv olved in the clinic. However, I still gained insight into the operations of such a large hospital, as well as first hand exposure to different types and levels of amputation and pediatric prosthetic devices. The second clinic I attended quickly turned into a long term internship due to its proximity to my career interests: Assistive Technology Partners (ATP) at the University of Colorado Denver Auraria Campus. Once a week, I attend ed three to four appointments with the therapists who work there, helping wit h wheelchair evaluation, design, and delivery concepts. The clinic has two main client facing aspects: wheelchair evaluations and augmentative and alternative communication (AAC) devices. I interfaced nearly exclusively with the former, though the latter their power wheelchair to their AAC device. I also learned about other accessibility issues disabled people face in everyday life, including education, employment, ADA complianc e, insurance coverage, and extraneous medical issues caused by or compounded by their disability. In this way I gained wide exposure to many struggles the disability community faces, and first hand accounts of how they could best be tackled. Though both sh adowing opportunities afforded me excellent insight into different aspects of the medical device world and alternate suggestions of where my career path could take me, I s largely

PAGE 81

72