Restoring Sensation in Human Upper Extremity Amputees Using Chronic Peripheral Nerve Interfaces

RESTORING SENSATION IN HUMAN UPPER EXTREMITY

AMPUTEES USING CHRONIC PERIPHERAL NERVE

INTERFACES

DANIEL WANEI TAN

Submitted in partial fulfillment of the requirements

for the degree of Doctor of Philosophy

Dissertation Advisor: Dr. Dustin J. Tyler

Department of Biomedical Engineering

CASE WESTERN RESERVE UNIVERSITY

August, 2014

CASE WESTERN RESERVE UNIVERSITY

SCHOOL OF GRADUATE STUDIES

We hereby approve the thesis/dissertation of

DANIEL WANWEI TAN

candidate for the degree of Doctor of Philosophy*.

Committee Chair

Dustin J. Tyler, Ph.D. Biomedical Engineering, Associate Professor

Committee Member

Kenneth J. Gustafson, Ph.D. Biomedical Engineering, Associate Professor

Committee Member

Robert Kirsch, Ph.D. Biomedical Engineering, Department Chair

Committee Member

Grover C. Gilmore, Ph.D. Dean of Mandel School of Applied Social Sciences

Date of Defense

7/1/2014

*We also certify that written approval has been obtained for any proprietary material contained therein.

Table of Contents

Contents Table of Contents ...... 4 Table of Figures ...... 7 Table of Equations ...... 9 Table of Tables ...... 10 Acknowledgements ...... 11 Abstract ...... 12 Chapter 1 : INTRODUCTION...... 14 Significance ...... 14 Past Research in Sensory Feedback for amputees via nerve stimulation ...... 17 Intraneural microstimulation in man of 1980’s ...... 18 Current Research in Sensory Feedback via Direct Afferent Nerve Stimulation ...... 19 Alternate methods of sensory feedback: ...... 21 Sensory substitution ...... 21 Targeted Sensory Reinnervation ...... 22 Sensory feedback via CNS stimulation...... 23 Chapter 2 : SPECIFIC AIMS ...... 25 Innovation ...... 25 Specific Approach...... 26 Specific Aims ...... 28 Aim I. Characterize sensory perception from chronically‐implanted, peripheral nerve cuff electrode stimulation...... 28 Aim II: Demonstrate functional improvement of a sensory feedback‐enabled prosthetic ...... 31 Risks ...... 33 Paper Chapter Description ...... 35 Chapter 3 : Neural interface provides stable, natural, touch perception to prosthetic hand users for more than one year ...... 37 Abstract ...... 37 One Sentence Summary: ...... 38 Introduction ...... 39 Results ...... 41 Cuff Electrode Peripheral Interface is Selective and Stable ...... 41

Stimulating with Constant Stimulation Intensity Produces Paresthesia ...... 43 Stimulation Intensity Modulation with a Time‐Variant Pulse Width Results in “Natural” Pressure Perception ...... 45 Sensory Feedback Improves Functional Performance and User Confidence ...... 53 Subjects Report That Sensory Feedback Eliminates Pain in the Phantom Hand ...... 55 Discussion ...... 56 Conclusion: ...... 60 Materials and Methods: ...... 60 Study Design ...... 60 Methods ...... 61 Experimental Setup ...... 62 Generic Framework of Electrical Stimulation ...... 63 Stimulating With Time Invariant Parameters ...... 64

Stimulating With a Time‐Variant Pulse Width, PW(i,t) ...... 64 Threshold Detection Method ...... 65 Contralateral Pressure Matching ...... 65 Functional Testing ...... 65 Statistical Analysis ...... 66 Supplementary Materials ...... 67 Chapter 4 Stability and selectivity of a chronic, multi‐contact cuff electrode for sensory stimulation in a human amputee ...... 71 INTRODUCTION ...... 71 METHODS ...... 73 A. Surgical Implantation ...... 73 B. Nerve Stimulation ...... 75 RESULTS ...... 77 A. Sensory Locations & Modalities ...... 77 B. Sensory Thresholds ...... 82 C. Super threshold recruitment of area ...... 85 DISCUSSION ...... 85 CONCLUSION ...... 90 Chapter 5 Toward an artificial hand: natural sensory feedback improves task performance ...... 92 Abstract: ...... 92 Introduction: ...... 93

Methods: ...... 97 Results ...... 104 Discussion: ...... 109 Conclusion: ...... 116 Chapter 6 Summary ‐ Conclusion ...... 117 Specific Aim I: Characterize sensory perception from chronically‐implanted, peripheral nerve cuff electrode stimulation...... 117 Specific Aims II: Demonstrate functional improvement of a sensory feedback‐enabled prosthetic. 118 Future Investigative Priorities ...... 119 Applications ...... 120 Appendix A: Additional Observations ...... 123 Aim 1: Characterization of Sensation ...... 123 Proprioception ...... 125 Attenuation of Sensation ...... 130 Multiple Location Detection ...... 132 Time Delay of Response ...... 132 Field‐steering with multi‐channel stimulation ...... 133 Difficulties with Clinical work ...... 134 Improving the stimulus waveform ...... 134 The Magic of 1 Hz ‐ Patterned Stimulation Intensity ...... 135 Aim 2: Functional Tests ...... 136 Optimizing the Feedback scheme ...... 136 Functional tasks ...... 138 Phantom Pain Reduction ...... 142 Embodiment ...... 142 Functional use of Pain ...... 144 Subject Preference ...... 145 References ...... 146

Table of Figures

FIGURE 2.1 GENERAL SURGICAL IMPLANT DIAGRAM ON THE MEDIAN, ULNAR AND RADIAL NERVES ...... 27 FIGURE 3.1 STABILITY AND SELECTIVITY OF IMPLANTED CUFF ELECTRODE SYSTEM ...... 42 FIGURE 3.2 WAVEFORM PATTERNS ...... 44 FIGURE 3.3 FULL‐SCALE MODULATION, SINUSOIDAL PW ENVELOPE ...... 47 FIGURE 3.4 SMALL‐SCALE, OFFSET (SSO) MODULATION ...... 51 FIGURE 3.5 FUNCTIONAL TASKS WITH SENSORY FEEDBACK ...... 54 FIGURE 3.6 SUBJECT S2 PERCEPTUAL LOCATIONS AT NEAR-THRESHOLD STIMULATION LEVELS...... 68 FIGURE 3.7 STABLE ELECTRODE IMPEDANCES ACROSS STUDY DURATION...... 68 FIGURE 3.8 SUBJECT S2 CONTRALATERAL PRESSURE MATCHING...... 69 FIGURE 4.1 A. NERVE CUFF ELECTRODES IMPLANTED IN THE FOREARM OF THE AMPUTEE SUBJECT 1. OPEN‐HELIX PERCUTANEOUS LEADS ARE PASSED INDIVIDUALLY THROUGH THE SKIN SO THAT THE SKIN WILL GROW INTO THE OPEN‐HELIX, PREVENTING PISTONING OF THE LEAD AND REDUCING THE RISK OF BACTERIA ENTRY (INSET). B. IN SUBJECT 2, THE ELECTRODES ARE IMPLANTED IN THE MID‐FOREARM...... 74 FIGURE 4.2 IN SUBJECT 1, STIMULATION PROVIDES SENSORY RESPONSE IN 19 LOCATIONS ON THE PHANTOM HAND WHICH COVER CLASSIC INNERVATION PATTERNS FOR THE MEDIAN (M1‐8), ULNAR (U1‐8), AND RADIAL (R1‐4) NERVES. (A) SHOWN IS THE PERCEIVED LOCATIONS AT STIMULATION THRESHOLD. (B) SUPRA‐THRESHOLD STIMULATION LEADS TO UNIQUE PERCEPT AREA RECRUITMENT. SHOWN IS EARLY MEASUREMENTS, WEEK 2, BEFORE RESPONSES “SETTLED”. (C) IN 104, TYPICAL PERCEPT AREAS COVERED APPROXIMATE MEDIAN AND RADIAL INNERVATION PATTERNS OF THE HAND AND ON THE ARM. SOME RADIAL CHANNELS ALTERNATE BETWEEN TWO DISTINCT LOCATIONS, THE HAND AND THE ARM (R3 AND R4)...... 80 FIGURE 4.3 THE CROSS SECTION OF THE FINE IMPLANTED ON THE MEDIAN NERVE OF SUBJECT 2 AND THE CORRESPONDING, CHANNEL‐SPECIFIC PERCEPT AREAS ARE SHOWN...... 81 FIGURE 4.4 PATTERNS OF PERCEPT AREAS OVER TIME IN SUBJECT 104...... 82 FIGURE 4.5 THRESHOLD AND IMPEDANCE MEASUREMENTS OVER TIME INDICATE STABLE NEURAL INTERFACE...... 84 FIGURE 4.6 ALL PERCEPT AREAS, INCLUDING SUPER‐THRESHOLD RESPONSE, ELICITED FROM WEEK 2 TO 56 FOR CONTACTS M2, M6, M4, M7 IN SUBJECT 2 IS SHOWN WITH AN OVERLAY OF THE PROPER PALMAR DIGITAL NERVES OF MEDIAN NERVE ADAPTED FROM TEXTBOOK NEUROANATOMY...... 85 FIGURE 5.1(A) S1 WAS IMPLANTED WITH 2 FINES AND 1 SPIRAL NERVE CUFF. THE FINES WERE IMPLANTED ON THE MEDIAN AND ULNAR NERVES. THE SPIRAL WAS IMPLANTED ON THE RADIAL NERVE. LEADS WERE TUNNELED TO THE LATERAL UPPER ARM, WHERE THEY EXIT AS 20 HELICAL WIRES. S2’S AMPUTATION IS IN THE PROXIMAL FOREARM AND CUFFS WERE IMPLANTED IN THE DISTAL ARM. (B) THE SUBJECT’S PROSTHETIC HAND WAS MOUNTED WITH LOW‐PROFILE PRESSURE SENSORS ON THE PADS OF D1‐D3 AS WELL AS A BEND SENSOR MEASURING THE D1‐D2 ANGLE (NOT SHOWN). BOTH SUBJECTS USED THEIR OWN PROSTHETIC HAND FOR THE TESTS: THE OTTO BOCK SENSORHAND SPEED WITH 1 DEGREE OF FREEDOM AND 1 GRIP PATTERN. THE INTERNAL SLIP SENSOR WAS DISABLED. THE PRESSURE AND BEND SENSORS REGULATED THE STIMULUS APPLIED TO THE NERVES. (C) S2 USING HIS INSTRUMENTED PROSTHETIC TO LOCATE AND REMOVE MAGNETIC BLOCKS FROM A METAL PLATFORM WHILE BLINDFOLDED...... 98 FIGURE 5.2: ODT‐2 CORRECT RESPONSES FOR S1 (LEFT) AND S2 (RIGHT) ON TWO DIFFERENT DAYS OF EXPERIMENTATION. BOTH SUBJECTS WERE ALWAYS ABLE TO IDENTIFY THE PRESENCE OF A WOODEN BLOCK WITH THEIR INTACT HAND (NOT SHOWN). WITHOUT SENSORY FEEDBACK, ONLY S1 ON DAY 2 PERFORMED DIFFERENTLY THAN CHANCE. PROVIDING FEEDBACK ABOUT THE PRESSURE EXPERIENCED AT THE FINGERTIPS WAS FOUND TO SIGNIFICANTLY IMPROVE ACCURACY. FURTHER ADDITION OF APERTURE FEEDBACK RESULTED IN IMPROVED PERFORMANCE APPROACHING THAT OF THEIR INTACT HAND...... 105

FIGURE 5.3: TOTAL NUMBER OF FAILURES ON THE MBB TEST VERSUS PERCENTAGE OF MAGNETIC BLOCKS SUCCESSFULLY REMOVED FROM THE METAL TABLE FOR S1 (TOP ROW) AND S2 (BOTTOM ROW) ON DAY 1 OF TESTING (LEFT COLUMN) AND DAY 2 OF TESTING (RIGHT COLUMN) WHEN USING THE PROSTHETIC HAND. FAILURE WAS DEFINED AS THE SUM OF 1) ATTEMPTS TO MOVE A BLOCK WHEN ONE WASN’T IN THE HAND (EMPTY PINCH); 2) DROPPED BLOCKS WHILE IN THE PROCESSES OF A MOVEMENT; 3) BLOCKS PUSHED OFF THE TABLE; AND 4) BLOCKS REMAINING ON THE TABLE AFTER TWO MINUTES. SUCCESS RATE INCREASED AND FAILURES WERE REDUCED WHEN THE SUBJECT WAS BLINDED AND SUPPLEMENTED WITH PRESSURE AND HAND APERTURE FEEDBACK, TRENDING TOWARD SIGHTED, PERFORMANCE WITHOUT SUPPLEMENTAL FEEDBACK. IN ONE CASE (S1, DAY 2), THESE SCENARIOS WERE NOT STATISTICALLY DIFFERENT...... 106 FIGURE 5.4 EMBODIMENT IMPROVES WITH SENSORY FEEDBACK ENABLED DURING FUNCTIONAL TASKS...... 109

Table of Equations EQUATION 3.1 ...... 63 EQUATION 3.2 ...... 64 EQUATION 5.1 ...... 100

Table of Tables TABLE 3.1 SENSATION QUALITIES REPORTED DURING A SINGLE EXPERIMENTAL SESSION. M6 TRANSITIONED TO A SENSATION OF A NEEDLE WITHIN A VEIN AT HIGHER STIMULATION...... 48 TABLE 3.2 AVERAGE CHANNEL RESPONSE FOR EACH CUFF WITH FULL‐SCALE MODULATION, WHERE N IS THE NUMBER OF UNIQUE, NATURAL RESPONSES PER NERVE, PER EXPERIMENTAL VISIT. NATURAL, NON‐TINGLING SENSATION WAS ACHIEVED ON EVERY CHANNEL WITH SINUSOIDAL VARYING PW STIMULATION. COLUMNS MAY NOT SUM TO 100% SINCE SOME CHANNELS HAVE TO MULTIPLE SENSATIONS DEPENDING ON STIMULATION...... 48 TABLE 3.3 PERCEIVED SENSATIONS ELICITED FROM SMALL‐SCALE PULSE WIDTH MODULATION ON ALL AVAILABLE CHANNELS. ... 52 TABLE 3.4 TAPES PAIN SURVEY DATA ...... 69 TABLE 5.1 SUBJECTS ENROLLED IN THE SENSORY RESTORATION STUDY ...... 98 TABLE 5.2 CONDITIONS UNDER WHICH ODTS AND MBBS WERE PERFORMED ...... 101

Acknowledgements

Research supported by  VA Merit Review #A6156R  NIH TATRC #W81XWH-07-2-0044

Louis Stokes Cleveland Department of Veterans Affairs Medical Center, Cleveland, OH 44106.

Restoring Sensation in Human Upper Extremity Amputees Using Chronic Peripheral Nerve Interfaces

Abstract

DANIEL WANWEI TAN

We restored normal sensation for up to two years in the perceived hands of amputees.

The sense of touch is essential to experience and manipulate the world around us.

Despite increasingly sophisticated mechatronics, prosthetics still do not convey sensation

back to amputees. Sensory perception is important for control of the prosthetic limb, a

sense of embodiment, and in the reduction of phantom pain. Those with limb loss rely

primarily on visual feedback for control. Chronically-implanted, single-channel nerve

cuff electrodes have produced sensation in the perceived hand with perceptions such as fist clenching, vibration, and paresthesia, and sensations was over large regions of the hand. Intrafascicular electrodes have produced discrete, tactile sensation but were only implanted for four weeks or less and paresthesia was associated with 30% - 50% of the

stimulating channels. A need exists to provide natural, discrete, tactile sensation in a

stable, long-term neural interface.

We implanted multi-channel cuff electrodes, the spiral and FINE, on the peripheral

nerves of two upper-limb amputees. Discrete, tactile sensory perception was recruited

from 97% of the electrode channels. Long-term stability was demonstrated with up to

two years of stable stimulation threshold and impedance measures. Stimulation pulse

charge and frequency modulated the size of the percept area and intensity, respectively.

Moreover, we discovered patterned stimulation intensity elicited natural tactile modalities

of pressure, light moving touch, and vibration, while avoiding paresthesia. The

functional benefit of sensory feedback was evaluated through functional and activities of

daily living (ADL) tests. Sensors mounted on the prosthetic hand provided grip pressure and opening span feedback through sensory stimulation to the subjects. We found

sensory feedback improved success rates in blinded object detection, object localization,

and controlled pressure manipulation tasks. Performance with sensory feedback while

blindfolded was similar to performance while sighted. Results from a standardized ADL

test, the Southampton Hand Assessment Procedure (SHAP), show that the sensory

feedback does not degrade myoelectric control. Sensory feedback reduces reliance on

visual feedback when using prosthesis.

Chapter 1 : INTRODUCTION

Our hands are the primary means by which we explore and manipulate our environment. They perform hundreds of intricate movements every minute without conscious thought or even visual attention.

The overall goal of the presented work is to improve the amputee’s quality of life by enabling sensory feedback through their prosthetic device. Sensory feedback for touch and proprioception are vital components of normal hand function. Prosthetic limb users must rely on visual and auditory cues to provide the necessary feedback. This limited feedback results in an additional cognitive load, which detracts from functional tasks and social interactions. Usable sensory feedback in a prosthetic limb should improve prosthetic control, leading to improvement in functional activities of daily living tasks.

Sensory feedback that provides natural modalities of sensation may also improve embodiment of the prosthetic hand and reduce uncomfortable phantom limb sensation or pain. The results of this work may also be relevant to other applications of afferent nerve stimulation, such as bladder control, obesity control, and pain management.

Significance

As of 2013, an estimated 650,000 individuals in the United States live with an upper limb amputation and the incidence rate is 46,000 new cases per year (Ziegler-Graham et al. 2008). Of this population, approximately 65,000 individuals are categorized as major upper-limb amputees, meaning disarticulation at the wrist or a more proximal level, and

are suitable for upper-limb prostheses. The most common cause of upper limb

amputation is trauma (83%) from workplace accidents and motor vehicular accidents . A

large portion of the amputee population consists of young, active individuals;

approximately 56% are within working age (Ziegler-Graham et al. 2008). From 2000 to

2011, recent U.S. military activity (OND, OIF, OEF) has resulted in an additional 500

veterans with major upper limb amputations (AFHSC 2012).

For people with upper limb amputations, a variety of functional prosthetic options are available. The most basic devices consist of a simple mechanical split-hook design that is attached to a shoulder-controlled cable system (i.e. Bowden). Lowering or raising the

contralateral shoulder pulls the cable, opening and closing the split-hook gripper. The

cable’s resistance provides an indirect sense of grip force and grasp-opening feedback.

However, the shoulder-harness cable system limits the range of motion to the anterior

plane, preventing use when the arm is extended laterally and superiorly at the shoulder

joint. Modern approaches to prosthetic control involve battery-powered, motorized split-

hook grippers (i.e. Otto Bock Electric Greifer), which are triggered by electromyography

(EMG) signals sensed from the residual limb. For a realistic aesthetic, the Greifer can be

replaced with the myoelectric hand, a three-prong metal gripper with a hand-shaped PVC

and silicon cover. Users reject myoelectric hands less than non-myoelectric hands

(Biddiss, Beaton, and Chau 2007) and most users (>83%) wear their myoelectric hands

more than 8 hours a day(Pylatiuk, Schulz, and Döderlein 2007). Recently, advanced

myoelectric hands featuring multiple degrees of freedom and positionable thumbs for

varying grasp patterns, such as the iLimb (TouchBionics) and the Michelangelo (Otto

Bock), have become commercially available. Other highly dexterous hands have been

developed under DARPA’s Revolutionary Prosthetics program and are currently seeking

commercialization, such as the Johns Hopkins University Applied Physics Laboratory

(APL) Modular Prosthetic Limb (MPL) and the DEKA Luke arm.

Despite advanced developments, modern hands and advanced myoelectrics lack a

way to provide natural sensory feedback to users. Users of standard myoelectric systems

typically rely on vision and incidental sound and motor vibration in the prosthetic socket to provide control feedback (Childress 1980). Some modern myoelectric systems utilize feedback internal to the prosthetic device but outside of the user’s control. A thumb slip sensor was incorporated into Otto Bock electric hands, which automatically activated hand closure when slip was detected. Unfortunately, this also prevented intended release and uncontrolled grip force, making it difficult to perform actions such as letting go of another person’s hand. The iLimb features individual motors in each finger digit.

Current feedback from each motor limits each digit’s flexion force, allowing digits to wrap and conform around objects of various shapes. However, since the force limits are not under the control of the user, the hand has difficulty adapting from grasping delicate objects, such as a human hand, to heavy force tasks, such as a pulling a heavy door open.

Other researchers are developing advanced sensors for prosthetics specifically to mimic the human hand’s touch receptors (Edin et al. 2008; Wettels et al. 2008).

Sensory feedback is a critical element of complex activities. Lack of sensory feedback in myoelectric upper limb prostheses makes complex motions difficult with

16 artificial limbs and requires the user to maintain visual attention on their prosthetic limbs to perform even the simplest tasks. The significant cognitive load required from the user for limited function results in up to 41% of users ultimately abandoning their prostheses(Biddiss and Chau 2007, 2007; McFarland et al. 2010). Sensory feedback is reported to be a commonly requested feature by current users (Biddiss et al. 2007;

Pylatiuk et al. 2007).

The addition of sensory feedback may also improve control of the prosthetic hand

(Riso 1999; Witteveen et al. 2012), help secure light objects in hand grasp (Scott et al.

1980), and improve overall prosthetic use confidence (Pylatiuk, Kargov, and Schulz 2006;

Shannon 1979). Sensory feedback may lead to improved embodiment of the prosthetic limb (Ehrsson et al. 2008; Marasco et al. 2011; Tsakiris and Haggard 2005), which may reduce uncomfortable phantom sensation or pain(Chan et al. 2007; Hunter, Katz, and

Davis 2003; Ramachandran and Rogers-Ramachandran 1996). Consequently, significant research investment in developing mechanically advanced prostheses(Resnik et al. 2012) will not likely achieve its full potential to improve prosthetic function and amputee quality of life without the addition of natural, stable, and long-term sensory feedback.

Past Research in Sensory Feedback for amputees via nerve stimulation

Researchers have long been investigating neural interfaces to provide natural sensory feedback to amputees. These early works are notable for their achievements; however the electrode technology was immature needed to be developed further in selectivity and

17 stability. In the 1970’s, Clippinger et al. implanted single-contact, nerve cuff electrodes on the median nerve in 10 amputees (Clippinger, Avery, and Titus 1974). Using a voltage-controlled stimulator, subjects felt tingling/vibration/paresthesia on the perceived hand from 0-35 Hz, above which transitioned to fist-clenching sensation or an increased area of paresthesia. Above 100 Hz, some subjects experienced a reduction of sensation.

Anani et al. stimulated the radial and median nerves of healthy subjects and amputees using intraneural fine wire electrodes, eliciting perceptions of paresthesia. To compensate for electrode positional instability, AM and FM current-controlled stimulation was evaluated for transmitting information within the paresethesia(Anani,

Ikeda, and Körner 1977; Anani and Körner 1979). It is suggested that these early efforts resulted in foreign sensations of tingling, vibration, tapping and flutter because

“stimulation activated many different types of cutaneous afferents all at once and with unnatural synchronicity”(Riso 1999).

Intraneural microstimulation in man of 1980’s

Much of our current understanding of the four tactile receptors found in glabrous skin was developed with microneurographic stimulation and recording techniques in the

1980’s(Vallbo and Johansson 1984). Understanding the nature of tactile receptors may lead to insight on perceptual responses from nerve stimulation. Microstimulation of specific peripheral afferent axons gave rise to specific perceptual responses according to the receptor type(Ochoa and Torebjörk 1983; Vallbo et al. 1984). Activation of single

Meissner (FAI) or Pacinian (FAII) afferents with a single stimulation pulse gives rise to a single tap and flutter or vibration with a continuous train of pulses. Activation of single

Merkle (SAII) afferents with continuous train of pulses lead to sustained pressure, with

the intensity related to the frequency of the stimulus train. Activation of single stretch-

receptor Ruffinian (SAII) afferents unreliably elicits perceptions of hand proprioception.

Later work would show that proprioception is a complex sensation arising from a

population of Ruffinian (SAII) afferents, muscle-spindle primary endings, and motor

command signals (Collins and Prochazka 1996; Proske and Gandevia 2009). Indeed,

intraneural recordings using concentric needles indicate perceptions may arise from

population encoding of SAII and PC afferents activation(Hallin, Carlstedt, and Wu 2002).

Current Research in Sensory Feedback via Direct Afferent Nerve

Stimulation

Recent work in sensory stimulation for amputees has focused on short-term

intraneural interface approaches. In 2003, a microelectrode array (MEA) was implanted

in the median nerve of a normal human for enhancing control of a prosthetic arm and

wheelchair with sensory feedback(Warwick et al. 2003). However, stimulation did not provide natural, tactile sensory feedback. Rather, it produced muscle twitch, which the subject used as sensory feedback. The MEA was removed after 3 months due to a loss of function in 17 of 20 electrode channels. In 2004 and 2005, Dhillon and Horsch implanted Longitudinal Intrafasicular Electrodes (LIFE) in the median nerve of 8

amputee subjects for two weeks (Dhillon and Horch 2005; Dhillon et al. 2004). The

LIFE is a single-contact electrode that is surgically implanted using partial epineurial

dissection in order to visualize fascicles and is placed longitudinally within the nerve.

Paresthesia, pressure and proprioception were the primary responses from nerve

stimulation. The modality and area of perceived sensation were modulated by pulse

amplitude and pulse width while stimulation frequency modulated sensation intensity.

70% of the contacts remained active but thresholds were noted to increase throughout the

14 day trial. An interesting finding of this work is that sensory pathways are readily

triggered from neural stimulation even in a chronic limb-loss of 30 years, suggesting

sensory research can benefit a wide limb-loss population of any age. A follow-up study

investigated functional use of sensory feedback for object size and compliance

recognition(Horch et al. 2011). Three sizes of 25, 50, and 75 mm and three stiffnesses of

soft foam rubber ( 6.8 N mm), medium foam rubber ( 11.5 N mm), and a hard piece of

wood (effectively inﬁnite stiffness) were presented. Of the two subjects, the subject that

did not have proprioceptive feedback had difficulty discriminating size and compliance.

The other subject who had some sense of proprioception from stimulation was able to significantly discriminate compliance, but not size. The results suggest proprioception

and touch feedback are necessary for object discrimination.

In 2010, Rossini and Micera implanted in an amputee with a related interface technology called the Transverse Intrafascicular Multichannel Electrode (TIME), which appears to only differ from LIFE in how it is implanted (Rossini et al. 2010).

Stimulation on 9 of 32 contacts produced touch and tingling sensations and frequency was found to modulate intensity. Improvements with phantom limb syndrome were noted with stimulation and for up to 1 week post-removal. The electrodes were implanted for 4 weeks, but the sensory stimulation failed at 10 days due to increasing thresholds; encapsulation was observed at explant. In a separate study, the sensory feedback was used in a functional task of object shape recognition to distinguish between a cylinder, cone, or sphere. However, the sensory feedback used was only relying on two

percept areas - broad median and ulnar areas - and the quality of the sensation was not

described (Raspopovic et al. 2014).

A common result among implanted intraneural interface reports is the observation of

increasing stimulation thresholds throughout the study trial, possibly due to encapsulation

increasing the impedance to stimulation. In addition, both Horsch and Micera’s reports do not discuss in detail the quality of the perceived tactile sensation and the level of

paresthesia.

Alternate methods of sensory feedback:

Sensory substitution

Sensory substitution is an alternative approach to provide feedback via non-invasive

transcutaneous-electrical or vibro-tactile stimulation of the skin. Fingertip force sensors

transmit feedback information by proportionally activating either vibrational or electrical

stimulation on the skin. Subjects are trained to interpret the sensation as grip force or position. Early work has shown that sensory substitution increases confidence(Shannon

1979) and helps in securing light objects during hand grasp(Scott et al. 1980). Recent approaches have been shown to reduce the force applied to objects via grasping pressure feedback(Panarese et al. 2009; Pylatiuk et al. 2006) and to provide positional feedback about hand opening(Witteveen et al. 2012). Subjects are also trained to interpret various amplitude or frequency modulation schemes to increase discrimination(Cipriani,

D’Alonzo, and Carrozza 2012; Stepp and Matsuoka 2012). These approaches to sensory feedback have been commercially unsuccessful, possibly because sensory substitution suffers from cross-modality and cross-spatial translation, i.e. the subject needs to interpret

coded vibration or electrical tingle sensation as tactile input and in a referred location

which is unnatural. Cross-modality and cross-spatial interpretation leads to additional

cognitive load and may be distracting to subjects (Pylatiuk et al. 2006).

A similar approach that shows promise is utilizing air-bladder pressure for referred

sensation on the residual limb(Antfolk et al. 2012). Pressure bulbs mounted to the

prosthetic hand push pressure toward air bladders against the residual limb. If the subject

has referred sensation on the residual limb, i.e. touch on the residual limb elicits touch perception on the phantom hand, which matches the locations of the sensor mounted on the prosthetic hand, both matched-modality and matched-spatial characteristics are achieved. The approach is mechanically simple; it is cost-effective to produce and maintain with minimal training required. However, it is limited to subjects with easily distinguished referred sensation on the residual limb, which may represent 62% of the traumatic-injury amputee population.

Targeted Sensory Reinnervation

Targeted Muscle Reinnervation (TMR) may be an option for providing sensory

feedback in the future. TMR is a recent technique (2004) where peripheral nerves are

routed to pectoral muscles to improve EMG recording for controlling a prosthetic

arm(Kuiken et al. 2007). In some subjects, the afferent nerves also reinnervated with the

pectoral skin, leading to referred sensation on the perceived hand, i.e. touch on the pectoral skin elicited a perception of touch on the perceived (phantom) hand(Marasco,

Schultz, and Kuiken 2009). This technique is called Targeted Sensory Reinnervation

(TSR). Most of the reinnverated skin lead to tingling sensations however a small portion

discretely elicited natural touch modalities. Ongoing research seeks to take advantage of

this aspect with a controlled tactor unit mounted over the reinnvation site to provide natural sensory feedback and improve embodiment(Marasco et al. 2011).

Sensory feedback via CNS stimulation.

An alternative approach to providing sensory feedback is by stimulating the central

nervous system rather than the peripheral nervous system. In animal studies of non- human primates (NHP), cortical stimulation with microelectrodes produced flutter touch

(possibly light vibration) discrimination matching natural mechanical stimulation(Romo et al. 2000). Graded force from a prosthetic finger was detected by a NHP using microelectrode arrays(MEA) implanted on the somatosensory cortex (Berg et al. 2013).

Recording electrodes using MEAs in humans have allowed control of 7-DOF prosthetic arm(Collinger et al. 2013), but have yet to be used for stimulating sensory perception. In addition, long-term reliability is an issue for cortically-implanted MEAs as studies have shown neural cell loss, glial scarring, and a drop in the number of functioning electrodes

(Biran, Martin, and Tresco 2005; Polikov, Tresco, and Reichert 2005). The biological response appears to be highly variable (Polikov et al. 2005) and may be caused by local, late onset neurodegenerative disease-like states in the tissue surrounding the chronic electrodes(McConnell et al. 2009). In human studies, deep brain stimulation of the thalamus using cycling stimulation patterns leads to sensation modalities of movement

(24%), mechanical (12%), temperature (2%) and pain (7%) on the hand (Heming et al.

2011). However, the majority of the response was tingling sensation (55%). Depending on the subject, the portion of stimulation that was rated as "natural" varied from 0-30%, and naturalness was easier to achieve with lower stimulation values of <127% of threshold current (Heming, Sanden, and Kiss 2010). CNS interfacing has led to several

23 notable, long-term neuroprosthesis cases but all involve subjects with tetraplegia and limited mobility(Collinger et al. 2013; Simeral et al. 2011) . Current CNS interfacing techniques are not appropriate for the generally active amputee population and it is unknown how many amputees would be willing to undergo brain surgery, which is more invasive than a peripheral neural interface approach.

Chapter 2 : SPECIFIC AIMS

The central hypothesis is that extraneural, multi-contact cuff electrodes chronically

implanted on the median, ulnar, & radial nerves will provide functionally beneficial

sensory and proprioception feedback to the user of upper-arm prosthetic devices.

Specific Aim I: Characterize sensory perception from chronically-implanted, peripheral

nerve cuff electrode stimulation.

Specific Aims II: Demonstrate functional improvement of a sensory feedback-enabled

prosthetic.

Innovation

The primary innovation of this proposal is the application of multi-contact cuff

electrodes for afferent nerve stimulation in amputees. Cuff electrodes, such as the self-

sizing spiral electrode(Naples et al. 1988), have been successfully implemented in

chronic, human trials for motor control applications in spinal cord injury(Fisher et al.

2009; Polasek et al. 2009). Recent advances in cuff electrode design have enabled a

selective response to stimulation. We will primarily use the Flat Interface Nerve

Electrode (FINE), which reshapes the nerves such that fascicles are closer to the surface

of the nerve, allowing a multi-contact cuff electrode to selectively activate individual

fascicles of a nerve bundle(Leventhal and Durand 2003; Tyler and Durand 2003). The

FINE has been tested interoperatively in the femoral nerve for motor recruitment(Schiefer et al. 2010). This project is one of the first chronic studies using

FINE electrodes in human subjects. Fascicles are thought to be somatopically organized

in the peripheral nerve(Hallin and Wu 2002). In unpublished histological studies of the

upper-extremity peripheral nerves, the elongated cross-section and multi-fascicular

structure make upper-extremity peripheral nerves an excellent candidate for the

application of the FINE interface (Brill, unpublished). Selective activation of fascicles

makes the FINE an ideal technology to provide selective sensory response.

It is unknown how well sensory perception can be elicited though a multi-contact,

cuff electrode. Researchers have long been investigating neural interfaces to provide

natural sensory feedback to amputees. Early neural stimulation research with single-

contact nerve cuff electrodes on the median nerve in amputees provided a sense of

paraesthesia or a proprioception of fist clenching(Clippinger et al. 1974). Early intraneural electrode stimulation of the median, ulnar and radial nerves for sensory feedback struggled with stability of the electrode position(Anani and Körner 1979).

Recently, distinct and graded pressure sensation on the phantom hand was demonstrated using intrafascicular electrodes implanted in median and ulnar nerves(Dhillon and Horch

2005). Touch and tingling sensation were elicited from ~50% of multi-contact intrafascicular electrodes in another study(P. M. P. Rossini et al. 2010). However, with only 2-4 week acute clinical trials, and rising stimulation threshold trends, the long term safety and stability of intrafascicular electrodes are unknown(Dhillon and Horch 2005; P.

M. P. Rossini et al. 2010). The stability of the cuff electrodes makes it ideal for application in non-paralyzed, active individuals.

Specific Approach

As a first-in-field study, up to 5 amputee subjects may be recruited for the project.

Subjects will be unilateral, upper-limb amputees who will not be at high risk for infection

26 and who will be without significant residual limb or phantom pain. Baseline outcome measures will evaluate function using activities of daily living (ADL) and non-ADL tasks as well as assess overall quality of life. Subjects will be implanted with nerve cuff electrodes on the median, ulnar and radial nerves of the upper arm which innervate sensory receptors of the hand

(Error! Not a valid bookmark self-reference.).

The size of the cuff electrodes will be selected Figure 2.11 General surgical implant diagram on the median, ulnar and radial nerves based on nerve dimensions from histological studies of the human cadavers and recommendations of surgical team during implantation.

The 8-channel FINE cuff design will be the preferred electrodde although the 4-channel spiral cuff may be used if the nerve diameter is small or at the surgical team’s recommendation. Depending on the amputation llevel of the subject, the electrodes will be ideally implanted distal to motor-branching points to maximize afferent nerve response and minimize interference with the electromyography (EMG) control of myoelectric hands. Additionally, electrodes will be implanted so as to not interfere physically with the wearing of the myoelectric socket. The electrode leads will be connected to open-helix, percutaneous leads which will be tunneled to an exit site in the lateral deltoid of the upper arm (Error! Not a valid bookmark self-reference.). After 2-

4 weeks of recovery, the subject will begin weekly experimental session where stimulation is provided through the percutaneous leads through a computer-controlled

nerve stimulator. The study duration for each subject is 6 months minimum with option

to continue for up to 2 years.

Specific Aims

Aim I. Characterize sensory perception from chronically‐implanted, peripheral nerve

cuff electrode stimulation.

Rationale and Hypothesis. It is unknown if a multi-contact, extraneural cuff interface

will result in a selective sensory response during stimulation of afferent fibers. We

hypothesize that nerve stimulation will elicit sensory perception characterized with at

least 3 unique percept areas, 2 unique modalities, and 3 levels of intensity through at least

50% of the contacts of all implanted cuff electrodes.

Objective 1.1 Determine the percept area, modality and intensity of perceived sensation and their relationship to stimulation parameters.

Experimental Approach

From weeks 2-8 post-op, experimental sessions will focus on mapping stimulation

parameters to perceptual response. Monopolar, single-contact stimulation will be provided while systematically varying stimulation parameters. If perceptual response is

elicited, the subject will indicate the percept area by sketching the location on a hand diagram. The subject will be asked to describe the modality of the sensation and make a comparison to a normal sensation felt by the intact hand. The subject will be asked to rate the intensity of the sensation on an open-ended scale. If pressure sensation is perceived, subject will match the intensity of the sensation by pressing on a pressure sensor with the contralateral, intact hand. Levels of intensity will be measured in a

constant parameter, repeated magnitude production experiment (Gescheider 1997). If

proprioception is perceived, the subject will match the position of the phantom hand with

the contralateral, intact hand. The parameters of the stimulation will be varied to map the

relationship between the contact channel, stimulation pulse width, pulse amplitude, and

frequency, and the resulting percept area, modality, and intensity of sensation.

Expected Results, Interpretation, Possible Pitfalls.

We expect at least 3 unique percept areas, 2 unique modalities, and 3 levels of intensity

will be elicited from nerve stimulation with multi-contact, cuff electrodes. Nerve cuffs

implanted on medial, ulnar, and radial nerves are expected to provide sensation in the

three corresponding sensory innervation regions of the hand. Multi-contact cuffs are expected to provide multiple, distinct percept areas within each innervation region. The receptors for pressure and proprioception have the largest diameter axons of all receptor types. Extraneural stimulation tends to recruit large diameter fibers with the lowest stimulation energy; therefore, pressure and proprioception are the primary expected sensory modalities. Other researchers in the field have found pressure and proprioception

from direct nerve stimulation in amputees(Dhillon et al. 2004; P. M. P. Rossini et al.

2010). Varying stimulation parameters may control the intensity of sensation(Dhillon and Horch 2005; P. M. P. Rossini et al. 2010). We expect at least 3 levels of intensity

will be controllable through the stimulation parameters and would indicate a relationship

which may be used for pressure or proprioceptive feedback. Due to subject fatigue, it

may not be possible to thoroughly map all contact stimulation parameters to sensory response during a single session. Priority will be given to sensory contacts which produce sensory responses useful in pinch and power grip hand positions. Addition,

channels or stimulation parameters which result in painful or uncomfortable sensations

will be avoided.

Objective 1.2 Evaluate the stability of the neural interface over study duration.

Experimental Approach

Perceptual threshold will be determined using an unbiased, stair-case search

method(Kaernbach 1990). Thresholds will be determined for each contact. Stability of

the interface will be determined through repeated, weekly measures of threshold and

contact-to-contact electrode impedance over time. Stability of the percept area of

sensation will also be noted.

Expected Results, Interpretation, Possible Pitfalls.

We expect the thresholds and impedances to remain statistically-unchanged or decreasing

throughout the 6 month study period, which would indicate a stable neural interface. The

FINE has been shown to be stable in chronic animal studies(Leventhal, Cohen, and

Durand 2006; Leventhal and Durand 2004; Tyler and Durand 2003). The multi-contact spiral electrode, predecessor to the FINE, was found to be stable in the upper extremity of two human subjects for up to 3 years(Polasek et al. 2009) and in the lower extremity of one human subject up to 1 year(Fisher et al. 2009). However, the cuff electrodes in those trials were implanted in tetraplegic or hemiplegic individuals with limited mobility.

Amputees generally lead active lives(Ziegler-Graham et al. 2008). A chronic implant in an active individual will demonstrate the clinical viability of the neural interface platform to a wider patient population.

Aim II: Demonstrate functional improvement of a sensory feedback‐enabled prosthetic

Rationale and Hypothesis. The purpose of this aim is to determine if sensory feedback can improve the functional capability of a prosthetic user. Pressure sensors and bend sensors will be mounted on the subject’s myoelectric hand. The sensor readings will be used to control the stimulation system to feedback hand grip force and position information to the subject. Pressure and position will be mapped to stimulation parameters using the perceptual mapping results from Aim 1. Subject will be given an opportunity to train with the closed-loop, sensory-feedback enabled prosthetic and suggest modification to the control scheme per the subject’s preference. We hypothesize that performance with sensory feedback will be significantly better than without sensory feedback.

Objective 2.1 Evaluate functional performance with and without sensory feedback in validated, standard activities of daily living tests.

Experimental Approach

The subject will participate in standardized measures of function tests with and without the sensory feedback system enabled. The Box and Blocks test is a timed, functional test where the number of blocks the subject can move from one tray to another within one minute is recorded. The subject will also participate in standardized activities of daily living tasks (ADL). The ADLs chosen for this proposal is the Southampton Hand

Assessment Procedure (SHAP). The SHAP is a clinically-validated, hand-function test, originally developed to assess the effectiveness of upper limb prostheses(Light, Chappell,

and Kyberd 2002). The SHAP is a timed test, where the subject will manipulate a variety

of everyday objects using various grip patterns.

Expected Results, Interpretation, Possible Pitfalls.

We expect that the subject will show significant improvement in the scores of Box and

Blocks, and SHAP when using sensory feedback. This result would imply greater

myoelectric function in terms of control or attention. In normal humans, sensory

feedback enables a secure grip and allows adaption to changing angles or loads of the

manipulated object(Augurelle et al. 2003; Jenmalm and Johansson 1997; Johansson,

Häger, and Riso 1992; Monzée, Lamarre, and Smith 2003) .Previous work has shown

feedback is useful to improve myoelectric control(Pylatiuk et al. 2006; Witteveen et al.

2012). However, these tests may not show an improvement because they were not

designed to account for sensory feedback. Box and Blocks and SHAP benefit from

advanced, dexterous, myoelectric hands, but if a subject only had a basic myoelectric, the

hand mechanics may be the overriding limiting factor.

Objective 2.2 Evaluate functional performance with and without sensory feedback

in novel tasks to illustrate sensory capabilities.

Experimental Approach

The unique nature of sensory feedback may require novel tests to show functional

improvement with sensory feedback. Standard tests which evaluate sensory aspects of prosthetic function do not exist. We will develop functional tests which demonstrate functional enablement due to sensation. Prosthesis users primarily rely on vision for

feedback (Childress 1980). Functional tests will be designed to explore performance in

sighted and blinded tasks with and without sensory feedback. Subject will be blinded to

32 visual and auditory feedback(Horch et al. 2011; Raspopovic et al. 2014). We will examine the subject’s capability for object detection.

Expected Results, Interpretation, Possible Pitfalls.

We expect subjects to show significantly improved object detection and object manipulation while blinded with sensory feedback over no sensory feedback. This represents a significant advancement in functional capability since amputees would be disinclined to attempt usage of a prosthetic in visually-occluded situations or environments. Applications include low-light environments such as evenings and theatres and social interactions such as handshakes. It is possible that subject may have to engage in significant re-training of their prosthetic to incorporate or optimize the use of sensory feedback with motor control.

Risks

Major safety risks to the project include surgical risk during implant, pain from nerve stimulation, and infection of the percutaneous risk. Surgical risk will be mitigated with utilizing an Operating Room team consisting of both experienced orthopeadic surgeons and research personnel. Nerve stimulation risk will be mitigated with providing stimulation in gradually increasing charge and safety-limited equipment. The largest risk to the project is infection of the percutaneous leads during the study trial. If an infection occurs which cannot be resolved with topical or oral antibiotics, the implanted components must be removed, thus halting the study for the affected subject. However, the occurrence is low. An intramuscular electrode failure rate study of 858 electrodes has shown percutaneous lead removal due to infection at 0.5%(Knutson et al. 2002). We will

mitigate this risk with instructions to the subject on daily care and cleansing of the site and visual inspection by a clinician on each visit (every 1-2 weeks).

Paper Chapter Description

Chapter 3: Restoring and controlling a sense of touch in the phantom hands of two human upper extremity amputees using long-lived peripheral nerve interfaces and patterned stimulation intensity for control of perceptual quality (accepted to Science

Translational Medicine, to be published in future issue)

This paper provides a general overview of the initial results with two subjects. Both

Aims 1 and Aim 2 are described including percept areas, modality, and intensity characteristics of percepts and a delicate object manipulation functional task. The stimulation method which promotes natural sensory percepts without paresthesia is described.

Chapter 4: Stability and selectivity of a chronic, multi-contact cuff electrode for sensory stimulation in a human amputee (paper not submitted)

This paper discusses results of Aim 1 detailing the stability of percept area, stimulation threshold, and impedance in both subjects up to 12 and 24 months in subject 1 and subject 2, respectively. The stability and selectivity of multi-contact, cuff electrodes is discussed as an ideal platform for providing long-term, nerve interfacing in clinical settings.

Chapter 5: Direct sensory feedback in humans improves task performance (paper not submitted)

This paper details the results of Aim 2 Functional Testing with both subjects, focusing on two functional tasks: blinded object detection and blinded object localization. A

35 distinction is made between the roles of contact pressure and hand position feedback in the functional tests.



Chapter 3 : Neural interface provides stable, natural, touch perception to prosthetic hand users for more than one year

(This paper has been accepted to be published in a future issue of Science Translational Medicine)

Authors: Daniel Tan1,2†, Matthew Schiefer1,2†, Michael Keith1,2,3, J. Robert Anderson1,4,

Joyce Tyler3, Dustin J. Tyler1,2,3*

Affiliations:

1Louis Stokes Veterans Affairs Medical Center, Cleveland, OH.

2Case Western Reserve University, Cleveland, OH.

3MetroHealth Medical Center, Cleveland, OH.

4University Hospitals Rainbow Babies & Children's Hospital, Cleveland, OH

*Corresponding author: [email protected]

†authors contributed equally to the work.

Abstract

Implanted peripheral nerve interfaces in two human subjects with upper limb amputation provide stable, natural touch sensation in their hands for more than one year. Touch perception on the fingers and hand is essential for fine motor control, contributes to our sense of self, allows for affective communication, and aids in our fundamental perception of the world. Despite increasingly sophisticated mechatronics, prosthetics still do not

convey sensation back to their wearers. Electrical stimulation through implanted peripheral nerve cuff electrodes that do not penetrate the nerve produces touch perceptions at many locations on the phantom hand with repeatable, stable responses in two subjects for 16 and 24 months. Patterned stimulation intensity (PSI) produces sensation that the subjects describe as natural and without “tingling,” or paresthesia.

Different patterns produce different types of sensory perception at the same location on the phantom hand. Our two subjects report tactile perceptions they describe as natural tapping, constant pressure, light moving touch, and vibration. Changing average stimulation intensity controls the size of percept area. Complex stimulation of peripheral

nerves can affect higher order, upstream cognitive processing with broader application.

Changing stimulation frequency controls sensation strength. Artificial touch sensation

improves the subjects’ ability to control grasping strength of the prosthesis and better

manipulate delicate objects. Electrical stimulation through peripheral nerve electrodes

produces long-term sensory restoration in limb loss.

One Sentence Summary: Electrical stimulation using a patterned intensity paradigm

applied through cuff-type peripheral nerve interfaces provides stable, multiple, and

repeatable touch perceptions in the fingers, thumb and hand for 16 and 24 months in two

subjects with upper limb loss.

Introduction

The sense of touch is essential to experience and manipulate the world around us

(Augurelle et al. 2003; Robles-De-La-Torre 2006). In addition to loss of function, loss of sensation is a devastating consequence of upper limb amputation. Sensory perception is important for control of the prosthetic limb, a sense of embodiment (Marasco et al. 2011), and in the reduction of phantom pain (Ramachandran and Rogers-Ramachandran 1996).

Those with limb loss rely primarily on visual but also on auditory feedback from the device motors for control (Childress 1980). The prosthesis is perceived by the user as a foreign tool extending beyond, but not as part of, their body (Murray 2008). Sensory substitution, such as vibration on a residual limb with intensity proportional to pressure at the prosthetic fingertip, improves prosthetic control in limited situations (Pylatiuk et al.

2006; Witteveen et al. 2012), but has not been widely adopted as the vibration is often described as distracting (Pylatiuk et al. 2006). Single-channel nerve cuff electrodes produced sensation in the perceived hand nearly four decades ago (Clippinger et al.

1974). Each subject, however, reported different sensory perceptions such as general fist clenching, vibration, and paresthesia. The sensation was over large regions of the hand.

More recently, intrafascicular electrodes produced tactile sensation but were only implanted for four weeks or less and some exhibited a continuous threshold increase and full loss of sensory stimulation capability as early as 10 days (Dhillon et al. 2005;

Raspopovic et al. 2014; Rossini, Micera, and Benvenuto 2010). Paresthesia was associated with 30% - 50% of the stimulating channels. However, the restored sensation allowed a subject to correctly identify three different objects, illustrating the value of sensory feedback on functional control (Raspopovic et al. 2014). Electrodes inserted into

the sensory cortex of non-human primates (Berg et al. 2013) demonstrate localized

sensory feedback.

In this study, non-penetrating peripheral nerve cuff electrodes (Naples et al. 1988;

Tyler and Durand 2003) provide multiple tactile perceptions at multiple locations on the

phantom hand with stable responses for 21 months. Cuffs contain either four or eight

independent stimulus channels. Each cuff is an electrically insulating silicone sheath

with multiple electrical contacts spaced evenly around the outside of the peripheral nerve.

The first subject, S1, is a 49-year-old male who suffered a wrist disarticulation in a 2010

industrial accident. We implanted cuffs on his median, radial, and ulnar nerves in his

mid-forearm in May 2012, providing a total of 20 stimulation channels; eight each on the

median and ulnar nerve and four on the radial nerve (Figure 1.A). The second subject,

S2, is a 47-year-old male with a below elbow amputation resulting from a 2004 industrial

accident. We implanted cuffs on his median and radial nerves in his mid-upper arm in

January 2013, providing eight stimulation channels on each nerve. After surgery, leads

from the cuffs protruded from the subjects’ upper arm for connection to stimulation

equipment during lab visits. Both subjects report stable sensory perception from stimulation over the entire testing period. A stimulation paradigm developed during the

study produces tactile sensation that both subjects describe as natural and without

“tingling,” or paresthesia. This sensory feedback has improved the subjects’ performance

on functional tasks. Subjects report cessation of pain they had perceived in the phantom

limb prior to the study.

Results

Cuff Electrode Peripheral Interface is Selective and Stable

During monthly visits by the subjects to the Cleveland Veterans Affairs Medical

Center, lasting between 4 and 6 hours, the subjects described perceived sensations in

response to trains of electrical pulses on each channel (Figure 3.1.A). We identified the

threshold for sensory perception by slowly increasing the intensity of the stimulation

pulse until the subject indicated feeling a sensation. The subject was blind to pulse

intensity. Random insertion of null pulses ensured the subject was not anticipating

stimulation. For each change in stimulus intensity, the subjects verbally described the

perceived sensation and traced its location on a schematic of a hand. Of the 20 available

channels in S1, stimulation produced sensation from 19 channels at 15 unique locations on the perceived hand (Figure 3.1.B). Of the 16 available channels in S2, stimulation

produced sensation from 14 channels at 9 unique locations (Figure S1). Over the duration of the study, the locations of percepts were repeatable and stable (Figure 3.1.C). In both subjects, percepts were produced at multiple, independent, small, and well-defined locations on the hand, including the thumb and finger tips. All perceived locations were consistent with innervation patterns of the median, ulnar, and radial nerves on which the electrodes were implanted. The peripheral sensory pathways and the subject’s sensory perception do not appear to have reorganized following injury, which confirms earlier studies (Dhillon et al. 2004). The sensory stimulation threshold remained stable the first

8 weeks after implant (Figure 3.1.D), ranging from 40.7 to 95.5 nC for S1 and 95 to

141 nC for S2. The thresholds were stable up to 68 weeks (Figure 3.1.E). Electrode

impedances remain stable around 3 k (Figure S2).

Figure 3.1 Stability and selectivity of implanted cuff electrode syystem

A. We implanted three cuffs with 20 channels total in the forearm of subject S1: 4‐contact spiral cuff on the radial nerve of the forearm and 8‐contact FINEs on median and ulnar nerves. The electrode leads run subcutaneously to the upper arm and connected to open‐helix percutaneous leads via Letechepia connectors (Knutson et al. 2002;

Letechipia et al. 1991; Polasek et al. 2009). A Universal External Control Unit (UECU, Ardiem Medical, Indiana, PA) supplied single‐channel, charge‐balanced, mono‐polar nerve stimulation.

B. Sensation locations at threshold stimulation levels week 3 post‐op. Cuff electrodes were highly selective, with each contact (M1‐8, U1‐8, R1‐4) producing either a unique location or unique sensation. Ulnar locations presented the most overlap at threshold, but differentiated in area expansion at suprathreshold responses. The subject drew the boarders. Areas outside the hand represents a small wrap‐around of sensation on the digit.

C. Repeated, weekly overlapping threshold locations of channels M2, M3, M4, M5, and M8 over weeks 3 through

10 Post‐op indicated consistent location perception. Locations remained stable for all stimulation waveform used.

D. Mean, normalized threshold charge density for all channels on the median (blue), ulnar (green), and radial (red) cuffs of S1 shown as the solid line. Shaded areas indicate the 95% confidence interval. An unbiased, step‐wisse

search determined the threshold. Frequency was a constant 20 Hz. During weeks 2‐8, percept thresholds for S1

were 95.5 ± 42.5 nC (n = 59), 70.7 ± 59.2 nC (n=50), and 40.7 ± 12.4 nC (n = 24) for the median, ulnar, and radial

nerves, respectively. Linear regression of the threshold stimulation intensity for perception over 8 weeks for every

channel was unchanging (18/19, ANOVA test, p ≥ 0.067) or decreasing (1/19, ANOVA, p = 0.044). S2 was also stable

(p ≥ 0.087) with thresholds of 141 ± 46 nC and 95 ± 47 nC for the median and radial nerves, respectively.

E. Threshold tracking of median channels M3, M4, and M5 to 68 weeks and ongoing show no significant change in

threshold over time (p=0.053, 0.587, 0.773 respectively).

Stimulation produced sensory perception before muscle activity that would interfere

with myoelectric control. In subject 1, the cuffs were implanted distal to the motor

branches of residual muscles. In subject 2, the electrodes were located on mixed motor

and sensory nerves. Stimulation did not interfere with standard myoelectric prosthesis

control, even when the motor and sensory fibers are in the same nerve.

Both subjects participating in this study are active individuals. Their normal, daily

activity does not affect the peripheral nerve electrode performance or affect the sensory

restoration provided by the electrode nor have the implants affected their daily activities.

Stimulating with Constant Stimulation Intensity Produces Paresthesia

The standard nerve stimulation paradigm is a train of identical, charge-balanced,

square electrical pulses characterized by pulse amplitude (PA), pulse width (PW), and

either pulse repetition frequency (f) or interpulse interval (IPI = 1/f) (Figure 3.2.A).

Traditionally, these three parameters are time invariant and fixed in value: PA0, PW0, and

IPI0 = 1/f0. With constant stimulation intensity pulse trains with stimulation frequency

ranging from 1 to 1,000 Hz, the subjects reported an unnatural sensation of paresthesia,

described as “electrical,” in 96% of 151 trials over a 10-month period. We systematically

43 mapped the sensory perceptions produced for many different constant values of PW0,

PA0, and f0. Increasing PA0 or PW0 increased the intensity and the size of the percept area

(Figure 3.2.B). The increasing area of sensation was somatotopically organized, suggesting a somatotopic organization of sensory fibers withiin the peripheral nerve.

During stimulation trains up to 60 sec, paresthesiia did not resolve into a natural sensation, as was previously reported with intrafascicular electrodes (Dhillon et al. 2005).

Figure 3.2 Waveform patterns

A. Square, charge‐balanced, cathodic‐first stimulation pulsing pattern. Prior neural stimulation paradigms

maintain parameters, such as pulse amplitude (PA), pulse width (PW) and interpulse interval (IPI) or frequency (f),

constant.

B. In general, constant PA and PW modulate the area of the perception. M5 shows a channel‐specific recruitment

pattern as PW was increased from 24 to 60 µs. M3 showed that percept area increases as PA increases from 1.1 to

2.0 mA. These recruitment patterns match sensory nerve innervation patterns of the digital nerve.

C. An example of full‐scale modulation, using a sinusoidal (1 Hz) PW envelope that produces a natural sensation of

pulsing pressure (top plot). The schematic resulting stimulation waveform where the IPI is 0.1 sec (10Hz) is shown

in bottom plot. Our stimulation trials typically used an IPI of 0.01 sec (100Hz).

Stimulation Intensity Modulation with a Time‐Variant Pulse Width Results in

“Natural” Pressure Perception

Full‐scale modulation feels like a pressure pulse. Modulation of stimulation intensity

resulted in S1’s description of the perception changing from “tingly” to “as natural as can

be.” He variously described the sensation as pulsing pressure, constant pressure,

vibration, tapping, and rubbing on a texture. In the full-scale modulation pattern, the

width of the pulses in the pulse train (f0 = 100 Hz) followed a slow (fmod = 1 Hz) sinusoidal envelope. The pulse width varied between 0 µs and a maximum of B µs

(Figure 3.2.C, see methods). In response to this stimulation, S1 reported a 1 Hz pulsing pressure sensation and described it, “as if I was feeling my own pulse or heartbeat - just like putting my fingers here,” as he demonstrated his fingers against his carotid artery pulse in his neck. When the peak pulse width was set to a first noticeable level, B = Bth, the sensation was described as the back of a pen repeatedly pushing “very lightly” on a small, localized area of the skin (Figure 3.3.A). The sensory transformation occurred in

both subjects. Both subjects could tap synchronously with the perceived pressure pulse

using the intact limb for visual confirmation of the pulse. The tapping matched the

frequency of the intensity modulation, fmod, as fmod varied between 0.5 and 4 Hz. As the

maximum pulse width, B, increased, the intensity of the pressure pulse increased and the

pulsing frequency remained matched to fmod.

There was psychometric correlation between perceived intensity of the pressure pulse

and the maximum pulse width. The subjects rated their perception of 5 different values of

peak pulse width, B. Each stimulation level was repeated multiple times and all presented

in random order. The subject was blind to the stimulation level. The verbal rating of

perceived intensity correlated significantly with B (p < 0.05, R2 = 0.85, n=22) (Figure

3.3.B).

At the finger tips, the subjects described the pulsing sensation to be like pressing on

the tip of a ball-point pen. The perceived sensory modalities across all 19 active channels

and all trials in subject 1 included pulsing pressure (86%), light moving touch (7%), or

tapping (7%) (Tables 3.1 & 3.2). S2 reported perceiving pulsing pressure during

stimulation with contacts on the median nerve (75%) and vibration during stimulation

with contacts on the radial nerve (75%). The range of modulation producing natural

sensation is defined by the threshold pulse width, Bth, and maximum pulse width, Btingle, with Bth < B < Btingle. When B increased above Btingle, the subject reported a light tingle

sensation in addition to and surrounding the region of natural perception (Figure 3.3.A).

When B was increased past an upper limit, BMask, where B > BMask > Btingle, paresthesia

dominated and overwhelmed the natural sensory perception (Figure 3.3.C).

Figure 3.3 Full‐scale modulation, sinusoidal PW envelope

A. At threshold (Bth), a pulsing pressure was felt at the bold circle area (M3, M4, M8, blue). Increasing the PW to a secondary threshold (Btingle) introduced an additional pulsing paresthesia, which typically coovered a larger area that overlapped the pressure location. Increasing the PW further caused the area of paresthesia to increase but

47 the area of constant pressure did not increase. Light moving touch was described as someone lightly brushing the skin with a finger. It consistently moved in the same directional for a given stimulus (R1, R4, red) .

B. Psychometric rating of sensation intensity as a function of PWmax shows a relationship between PW and tthe strength of the perceived intensity. The subject was provided 5 PWmax levels (100, 114, 131, 150, 167 µs) and each level was presented 3‐6 times in random order.

C. Threshold windows for natural sensation were measured on every channel of median cuff. Pressure occurred at

Bth (green), was accompanied by paresthesia at Btingle (black line, yellow), and was overwhelmed by paresthesia at BMask (red). The largest PW windows for a particular channel were found when PA was lowest. Higher levels of stimulation were avoided for M6 due to pain response.

Table 3.1 Sensation qualities reported during a single experimeental session. M6 transitioned to a sensation of a needle within a vein at higher stimulation.

Table 3.2 Average channel response for each cuff with full‐scale modulation, where n is the number of unique, natural responses per nerve, per experimental visit. Natural, noon‐tingling sensation was achieved on every chhannel with sinusoidal varying PW stimulation. Columns may not sum to 100% since some channels have to multiple sensations depending on stimulation.

Small‐scale, offset (SSO) modulation feels like constant pressure. Increasing the

minimum pulse width reduced the pulsing quality of the sensation. The frequency of the train of stimulation pulses, f0, remained 100 Hz and the modulation envelop frequency, fmod, remained 1 Hz. As the minimum pulse width was raised above 0 sec, the subject felt smaller variation in the pulsing and described a, “lingering pressure,” during the sinusoidal pulse width modulation. When the modulation was sufficiently small, the subject reported a continuous pressure described as though, “someone just laid a finger on my hand.” The typical size of pulse width modulation resulting in constant sensation,

PWpk-pk, was small, on the order of 5 µs. The best results were achieved when the average pulse width, PWoffset, was approximately 90% of the Bth required to produce the natural

pulsing sensation with a full-scale modulation (Figure 3.4.A, see methods).

The quality of the sensation varied depending on the type of skin in which the sensation was perceived. Small-scale pulse width modulation elicited constant pressure sensation when the perception corresponded to glabrous skin only (6 of 6 sites in S1 and

4 of 5 sites in S2). On sites with mixed glabrous and hairy skin or hairy skin only, 5 of 11

sites in S1, small-scale modulation resulted in a sensation of constant pressure. Of the

other 6 of 11 mixed or hairy skin sites, 4 remain as paresthesia, 1 was constant vibration,

and 1 was described as a “cotton ball” lightly rubbed on skin (Table 3.3). Interestingly,

the sensations in hairy skin were described as pulsing, “natural” pressure during full-scale

modulation, but as the modulation envelope was decreased both subjects report that these

resolve to constant vibration, itch, or tingling sensation.

To confirm that the natural sensation is a consequence of the modulation of the simulation and not of other effects, such as sensory accommodation, we repeatedly decreased PWpk-pk modulation to zero and back to a value that produced constant pressure.

The percept always transformed to paresthesia when PWpk-pk was zero and returned to

continuous pressure when modulated. As PWoffset increased, the subject reported an

increase in sensation intensity. When the maximum pulse width exceeded Btingle, pressure

was accompanied by paresthesia. The size and range of the PWpk-pk window needed to produce constant pressure depended on the stimulation channel; pulse amplitude, PA0; offset of the modulation, PWoffset; and pulse train frequency, IPI0 (Figure 3.4.C).

Quality of a sensory perception is a consequence of higher-order processing of

multiple sensory inputs. Time-varying or patterned intensity of the stimulation changed

the quality of sensory perception related to higher-order processing. Variation in

stimulation intensity varies the population of activated sensory fibers, which we define as

population coding. Population coding is different than the traditional temporal coding

resulting from patterned changes of pulse frequency only and was more effective at

controlling sensory perception.

Stimulation Frequency Controls Intensity of Sensation. Microneurography studies of

sensation (Knibestol and Vallbo 1980) show that axonal firing frequency encodes

intensity of pressure sensation. The perception of pressure intensity during small-scale

modulation was a function of the pulse repetition frequency (f = 1/IPI). The subject was

presented with an initial stimulation pulse rate with the IPI = 0.02 sec and instructed that

the perceived intensity was defined as “5.” He then scored subsequent sensations relative

to the initial sensation and the upper end of the scale was unbound. The lightest intensity

(score 1) occuru red with the longest IPPI = 0.2 sec and the subject described the sensation as if, “a finger was just resting on the surface of the skin.” Thhe greatest intensity (score

13) was at the shortest IPPI = 0.002 sec and was reported as “white knuckle” forceful pressure. To relate the perception of pressure to a physical value, the subject pressed their intact hand on a force sensor having a shape and position that matched the perceived shape and position of the phantom sensation. There was a dirrect relationship between the log of frequency and matched pressure sensation (in S1, ANOVA, p < 0.05, R2 = 0.70,

N = 25) (Figuure 3.4.B). Data from subject S2 is shown in Figure S3.

Figure 3.4 Small‐scale, offset (SSO) modulation

A. Typical example of a SSO modulation using sinusoidal (1Hz) PW with offset stimulation on M4 (solid, red line).

PWpk‐pk = 90‐95 µs was the lowest stimulation level that produced constant pressure sensation. For comparison, the threshold for pulsing pressure from full‐scale modulation is shown (dottedd, blue line).

B. Contralateral pressure matching indicated frequency can control intensity of constant pressure sensation. The subject was provided SSO modulation with IPI set to 50, 20, 10,, 5 or 2 ms (20, 50, 100, 200 or 500 Hz) on channel

M4 and asked to match the perceived pressure with his contralateral hand. Perceived constant pressure intensity was on matched to 0‐500 grams (<1 lb).

C. The PWmin‐max window that produced a sensation of constaant pressure was influenced by the PA, which altered both the size and the location of the window. We found frequency has a weaker effect on the window but found it affected the intensity. At PA of 0.5 mA there was no response. For PA 0.8 mA and above, the data suggests that the window for continuous pressure sensation decreases.

Table 3.3 Perceived sensations elicited from small‐scale pulse width modulation on all available channels.

Constant pressure sensation is commonly associated with glabrous skin location on the perceived hand.

*Not Tested: usually stimulation on M6 results in uncomfortable sensation

**Not Tested: connector needed replacement. External connecctor lasted 1.6 years before requiring simple replacement.

Sensory Feedback Improves Functional Performance and User Confidence

Pulling the stem from a cherry requires control of the grasping pressure on the fruit.

If the pressure is too light, the cherry slips; if it is too heavy, the cherry is damaged or crushed (Figure 3.5.A-B). We performed this test under four conditions combining sensory (S) and audiovisual (AV) feedback: S-/AV+, S-/AV-, S+/AV-, and S+/AV+.

Sensory feedback provided to the thumb and index finger matched the physical pressure measured at matched positions on the prosthesis. Stimulation frequency to each location was proportional to force measured by force-sensitive-resistor sensors mounted on the corresponding fingertips of the prosthetic hand. Stimulation frequency to the contact that produced pressure sensation on the thenar eminence of the phantom was proportional to the prosthesis opening span.

When sensory feedback was not provided, the subject successfully plucked 43% and

77% of the cherries without (S-/AV-) or with audiovisual feedback (S-/AV+), respectively. When sensory feedback was provided, the subject successfully plucked

93% and 100% of the cherries without (S+/AV-) or with audiovisual feedback (S+/AV+), respectively (Figure 3.5.E). Sensory feedback significantly improves performance (43% to 93% success) without audiovisual feedback (test of two proportions, p < 0.001, N = 15 per condition). With audiovisual feedback, sensation still significantly improves performance from 77% to 100% success (p < 0.005). Fingertip grip forces are significantly reduced with sensory feedback (Figure 3.5.C-D, F). The subject’s self- reported confidence in performing the functional trials is significantly higher with sensory feedback compared to without sensory feedback (one-tailed t-test, p = 0.0305).

Without sensation, prosthesis users typically use the prosthesis for gross tasks such as brracing and holding. The improved control and confidence resulting from sensory feedback may lead to greater use of the prosthesis for fine activity and improve balanced bilateral activities, and hence, a more normal appearance and integration of the prosthesis into daily activity.

Figure 3.5 Functional tasks with sensory feedback

A. Without the sensory feedback system enabled, the subject was often unable to adequately control the grip force in a delicate task of holding a cherry while removing the stem.

B. With the sensory feedback enabled, the subject felt contact with the cherry and the force applied. He successfully gripped the cherry and removed the stem without damaging the fruit.

C. Total force from thumb and index tip sensors in sighted case without sensory feedback. (* Peak force during the trial.)

D. With sensory feedback enabled, peak forces (*) are greatly reduced.

E. Sighted and blinded performance with the sensory feedback On or Off during the cherry task show an improvement in success rate (test of proportions, p < 0.005, n = 15 per condition).

F. Peak forces are significantly lower in the feedback enabled condition in both blinded and sighted cases (Welch’s t‐test, p < 0.001, n = 15 per condition).

Subjects Report That Sensory Feedback Eliminates Pain in the Phantom Hand

Subject S1 reported that prior to the study, his phantom hand felt like it was always clenched in a fist. Furthermore, multiple times per week he would experience pain that he described as his fist being squeezed in a vice. After sensory stimulation began, he reported that it felt like his phantom hand was opening again, and that over time, the phantom pain episodes diminished and eventually disappeared. Subject S2 reported that prior to sensory stimulation he experienced pain described as, “a nail being driven through [his] thumb” approximately twice per month. Since beginning sensory stimulation, he has not had another episode of phantom pain. The results of the pain survey from the Trinity Amputation and Prosthesis Experience Scales (TAPES) given throughout the study are shown in Table 3.4. These are similar to findings from other single-subject case studies (Horch et al. 2011; P. Rossini et al. 2010) and warrant more

rigorous investigation into the benefits of sensory feedback and neural interfaces in the

management of phantom pain.

Discussion

Peripheral nerve cuff electrode interfaces provided more than a year of multi-

location, multi-perception sensory feedback in two human subjects. Patterned

stimulation intensity controls the quality of sensory perception. Our hypothesis is that the

patterned intensity introduces information in the peripheral nerve by population coding

that has influence on higher-order processes to produce complex sensory perceptions.

Independent control of pulse width, pulse amplitude, stimulation frequency, and the patterns by which these parameters were varied controlled the spatial extent, intensity,

and quality of perception. This level of control was possible at multiple locations

innervated by a single peripheral nerve. There was independent control of 19 different

locations on the hand with only three implanted cuffs having a total of 20 electrical

contacts in S1 on mostly sensory nerves and 9 different locations in the hand and 3 in the

arm from two cuffs having a total of 16 contacts on mixed motor and sensory nerves in

S2. Higher density contact arrays may further improve coverage on the hand. The

implanted electrodes were stable for up to two years in individuals who have active

lifestyles that include chopping wood, home renovation, and camping.

Tasks requiring fine grasp control were difficult without sensation even when looking

at the prosthesis. With sensory feedback functional performance improved, especially in

the audiovisual-blinded case. The addition of sensory feedback alleviates the visual and

attentional demands typically required to use a myoelectric prosthesis and enhances the

user’s confidence in task performance. The subjects report feeling like they are grabbing the object, not just using a tool to grab the object.

The complex perceptions described in response to patterned stimulation intensity paradigms suggest sensory perception is a pattern recognition activity (Makin, Fellows, and Sabes 2013). Paresthesia results from ectopic or unnatural patterns of fiber activation

(Mogyoros, Bostock, and Burke 2000; Ochoa and Torebjork 1980). While perception can arise from activation of a single fiber, it is more natural for many fibers to be active during a sensory task (Dimitriou and Edin 2008). This activity is then integrated centrally to produce meaningful perception. Nerve electrodes activate populations of

axons based on size and location relative to the stimulating contact (Schiefer, Tyler, and

Triolo 2012). Because the SA1, SA2, RA1, and PC fibers all have similar diameters and

stimulation threshold characteristics, it is not possible to selectively activate a population

consisting of a single fiber type. With a constant pulse intensity, there is synchronous activity of a mixed population of axons. This is the abnormal firing pattern observed when paresthesia is induced from ischemic block (Mogyoros et al. 2000) and is typical in traditional electrical stimulation.

The patterned stimulus intensity recruits varying populations of axons at each pulse, thereby creating a pattern in the population activation. At the lowest pulse width of the

SSO modulation paradigm, a small population of axons are supra-threshold and actively firing. Because the pulse width never decreases below this level, this population of the neurons will be active at the frequency of the pulse train. This is similar to a constant,

SA-type pattern of activity in response to a constant pressure. At higher pulse widths the stimulation is sufficient to activate other, slightly more distant or smaller diameter

57 neurons. That population will have a transient or bursting activation pattern that mimics the pattern more typical of RA/PC fibers. The higher-order processing of this population pattern results in the perceived quality of the sensation. Sinusoidal pulse amplitude modulation would likely produce similar population recruitment patterns and sensation.

With modulation patterns more complex than a sinusoid reported here, it is possible to introduce more complex population codes.

The sensation is described by the subjects as “natural.” It is unlikely that the population activation is a perfect mimic of the patterns of natural SA-type and RA/PC- type fibers in response to touch. Thus, the resulting activation pattern is not strictly natural at the axon level. However, processing in the thalamic relays and columns of the primary sensory cortex appear robust to errors in the pattern, resulting in the reported natural sensory perception. Normal processing in the brain is highly tolerant of abnormal patterns and classifies patterns according to best matching prior sensory experiences

(Makin et al. 2013).

The subjects’ responses to restored sensation demonstrate its value to quality of life.

The subjects strongly preferred having sensory feedback. They described the sensation as natural and not requiring additional interpretation, as is required by sensory substitution techniques. When asked about performing object grasping tasks with sensory feedback enabled, S1 stated that, “I knew that I had it,” referring to whatever object was part of the test. Both subjects desire for a fully-implanted system that would provide them with permanent sensation. Whether sensation was natural or paresthetic, S1 stated that “I’d rather have it in a heartbeat,” and “I miss it when I leave.”

When sensation was active, both subjects’ perceived hand and prosthetic hand nearly

perfectly aligned. When sensation was not active, the prosthesis was viewed by the

subjects as a tool that extended beyond their hands. S1 described his normal perception of

the prosthesis as though, “[he] is holding the prosthetic hand [at the base of the

prosthesis] with [his] phantom hand.”

The extraneural electrode selectivity was effective. Every stimulating contact

provided sensation at well-defined and unique locations on a predominately sensory nerve (subject S1). The subject could identify sensation at multiple sites independently.

There is mounting evidence that peripheral nerves are somatotopically organized (Badia et al. 2010; Prodanov, Nagelkerke, and Marani 2007; Watchmaker et al. 1991), and hence, multi-channel cuff electrodes are able to produce somatotopically selective results similar to those reported in acute trials with intrafascicular electrodes (Dhillon et al.

2004; Raspopovic et al. 2014; P. Rossini et al. 2010). The peripheral interface in this trial produced punctate sensations in the hand with 19 of 20 channels (95%) or 9 of 16 (56%) from a proximal implant location. Natural tactile sensation has been reported in prior work, but has not explicitly been quantified or otherwise detailed, making direct comparison difficult with the current study (Raspopovic et al. 2014; P. Rossini et al.

2010). Prior work only examined the response to a pulse train stimulation paradigm. By

introducing patterned stimulation intensity, we are able to control the quality of the

sensation. More than a year after implant and 24 months in the longest subject, these

results have remained stable. On mixed motor and sensory nerves (subject S2), the

results are similar in that isolated and distinct locations were perceived. Even more

59 encouraging, sensation could be produced without motor activation or interference with myoelectric control.

Conclusion:

Peripheral nerve cuff electrodes are stable and produce sensory feedback in the human with multiple modes of sensation at multiple points on the hand. Sensory feedback significantly improves grasping performance using a prosthesis. Patterned stimulation intensity controls perceived sensory quality and has potential application in other areas of somatosensory neuromodulation, such as pain, autonomic function, and DBS. Frequency of stimulation produces a natural, graded intensity of sensation. The human experiments provide data not readily available in non-human studies, specifically in a verbal description of perception. Closing the loop by providing natural sensory feedback significantly improves success rate on a functional task requiring variable grip strength.

Materials and Methods:

Study Design

The central hypothesis is that direct nerve stimulation with selective, non- penetrating peripheral nerve cuff electrodes on the residual upper limb nerves can elicit graded sensation at multiple locations perceived in the missing hand of human subjects with limb loss. The inclusion criteria included unilateral, upper limb loss amputees, age

21 or older, and who are current users of myoelectric prosthesis or prescribed to use one.

Potential subjects were excluded for poor health (uncontrolled diabetes, chronic skin

ulceration, history of uncontrolled infection, active infection) and if significant,

uncontrolled, persistent pain existed in the residual or phantom limb.

This is a first-in-man, self-controlled, non-randomized case study of two subjects

to demonstrate feasibility of cuff electrodes to generate sensory perception. Subject S1

has a wrist disarticulation and was 19 months post-loss at the time of the implant in May

2012. Subject S2 has a below-elbow amputation and was 93 months (7.75 years) post-

amputation at the time of implant in Jan 2013.

Methods

In S1, surgeons implanted three electrodes in the residual limb of a then 46 year-

old male who has a unilateral wrist disarticulation from work-related trauma in 2010. The

surgery is an outpatient procedure. At the time of implant, the subject had been a regular

user of a myoelectric prosthesis for 7 months. Eight-contact Flat Interface Nerve

Electrodes (FINEs) (Tyler and Durand 2003) were implanted on the median and ulnar nerves and a 4-contact CWRU spiral electrode (Naples et al. 1988) was implanted on the

radial nerve. FINE opening size for the nerve was 10 mm wide by 1.5 mm tall for both

the median and ulnar nerves. The internal diameter of the spiral electrode was 4 mm for

the radial nerve. Peripheral nerve histology from human cadavers guided specifications of

electrode sizes and surgeons confirmed proper electrode fit during the procedure. All

electrodes were implanted in the mid-forearm (Figure 3.1.A) and connected to

percutaneous leads (Knutson et al. 2002; Letechipia et al. 1991; Polasek et al. 2009) that

exited through the upper arm.

In S2, surgeons implanted two multi-contact nerve cuff electrodes of a then 46

year-old male who has a below elbow amputation from work-related trauma in 2004. At

61 the time of implant, the subject had been a regular user of a myoelectric prosthesis for 7 years. Eight-contact FINEs were implanted on the median and radial nerves. FINE opening size for the nerve was 10 mm wide by 1.5 mm tall for both the median and radial nerves. All electrodes were implanted in the mid-upper arm. The subject was discharged from the hospital the day after surgery. All other surgical details were identical to S1.

Ardiem Medical (Indiana, PA) supplied implanted components (cuff electrodes, percutaneous leads, connectors) and MOOG (Buffalo, NY) sterilized the components with Ethylene Oxide.

Stimulation experiments began after three weeks to allow the electrodes and tissue response to stabilize. Subjects did not report any adverse sensation in the implanted locations or changes in their phantom sensations. In weekly sessions beginning after the stabilization period, we applied stimulation through each contact for up to 10 sec. For all trials, the subject was blind to the stimulation strength. Following stimulation, the subject would describe any perceived sensation and sketch its location on a blank hand diagram. We randomly intermixed null trials with no stimulation to assure the subject was not anticipating sensation. The Cleveland Department of Veterans Affairs

Medical Center IRB approved all procedures and the study is conducted under an FDA

Investigational Device Exemption.

Experimental Setup

The stimulation system consists of a computer that controls stimulation parameters and sends the commands to a single board computer running xPC Target

(Mathworks, Inc., Natick, MA). Ardiem Medical (Indiana, PA) fabricated a custom- designed stimulator (Cleveland FES Center). An isolator provides optical isolation

between devices plugged into the wall and the subject. To prevent overstimulation we limit charge density to less than 50 µC/cm2 and stimulation to less than a 50% duty cycle

during all stimulation protocols.

The stimulator has 24 channels of controlled-current stimulation outputs, with a

maximum stimulation amplitude of 5.6 mA, a maximum stimulation pulse width of 255

µs and a compliance voltage of 50 V. The stimuli are monopolar, biphasic, charge-

balanced, cathodic-first pulses with return to a common anode. There is a minimum

delay of 0.6 ms between each channel of stimulus, thus the stimulator does not output

truly simultaneous stimulation when multiple channels are active.

Generic Framework of Electrical Stimulation

The generic stimulation waveform, , is a train of pulses, , separated by an

interpulse interval, IPI. For each pulse shape, the pulse parameters, , are selected to

activate a population of neurons. To elicit a sensory perception from touch, the pulse

parameters, , and IPI are a function of measured external inputs over time, (t), and

time, t. Patterns in the pulse parameters will vary the population of axons excited and

affect qualities of sensory perception. Hence,  and IPI are defined as a function of the

desired tactile perception and time, i.e.  ((t), t) and IPI((t),t) (Eq. 3.1).

Equation 3.1

,, ∑ , ∀ , , , ≜ Individual stimulus pulse waveform, such as charge‐balanced, bi‐phasic square Figure 3.2.A , ≜ Stimulation waveform parameters ≜ The sensory data measurements to be feedback to the user

The psychometric relationships between modulation of stimulation parameters

and sensory perception were assessed for  as a charge-balanced, biphasic, square pulse,

having parameters ={PA(,t), PW(,t)} with IPI(t), where PA is pulse amplitude and

PW is pulse width. Perception is controlled by the time-dependent change of the parameters, referred to as patterned stimulation intensity. We systematically examined

each parameters of stimulation according to each of the following conditions.

Stimulating With Time Invariant Parameters

In a typical stimulation paradigm, the pulse intensity parameters are set to a fixed

values, i.e. PA(i) = PA0 and PW(i) = PW0, and the pulse train frequency is constant,

IPI(i) = IPI0 = 1/f0.

Stimulating With a Time‐Variant Pulse Width, PW(i,t)

Full‐scale modulation.

For this set of trials, the following equation (Eq. 2) defines the specific version of

Eq. 1 for the full-scale modulation pattern of stimulation intensity:

Equation 3.2

ttIPIii 10

PAi  PA0

PWaftbimod sin 

a and b are parameters that control the size of the pulse width modulation. In these trials,

the interpulse interval was held constant at IPI0 = 0.01 s. The pulse width was modulated

in a slow (fmod = 1 Hz) sinusoidal envelope with b=a=B/2, where B is the peak of the

varying pulse width. The pulse width, therefore, ranged from 0 µs to B µs (Figure

3.2.D)).

Small‐scale, offset (SSO) modulation. Small modulation is defined by Eq. 2 with a =

(PWmax - PWmin)/2 = PWpk-pk/2 and b = PWoffset. The typical value of PWpk-pk was 5 µs,

IPI0 was 0.01 sec, fmod = 1 Hz, and PWoffset was set to approximately 90% of the Bth

required to produce the natural pulsing sensation (Figure 3.4.A).

Threshold Detection Method

See Supplemental Material.

Contralateral Pressure Matching

We tested pressure matching with SSO modulation on channel M4 (corresponding

to the thenar emminence) of subject S1. For each trial, the subject received 5 seconds of

stimulation and matched the pressure sensation in the contralateral hand by pressing on a

manipulator for 5 sec. The last two seconds of the matched pressure data were averaged

per trial. The manipulator shape resembled the perceived sensation and the user pressed

with the same palmar location on the contralateral hand as the perceived sensation. The

manipulator shape iwa an approximately 1/2” diameter circle flat tip with rounded edges and made of balsa wood. We placed this manipulator on top of a FlexiForce sensors model A201 (Tekscan, Inc., South Boston, MA) with 0-1 lbf range. The sensor was calibrated before each trial and sampled at 10 Hz.

Functional Testing

The subject used his intact hand to pluck the stem off of a cherry with the fruit

grasped by the prosthetic hand. The subject used his standard prosthetic hand, which was

65 the velocity-controlled, SensorHand Speed® (Otto Bock HealthCare, Germany) with automatic slip detection/correction turned off. Control of the hand was driven by subjects using standard surface EMG signals. Thin FlexiForce sensors (Tekscan model A201) mounted on the end pads of the thumb and index finger measured the pressure applied to the cherry by the prosthetic. A bend sensor mounted between index finger and thumb provided measurement of the opening span. The instantaneous stimulation frequency applied to the nerve was linearly proportional to the real-time force sensor output. The frequency range was 10-125 Hz. The sensory feedback stimulation system has a measured mean delay of 112 +/- 35 Ms between measurement and applied sensation.

The subject reports no noticeable delay in sensation during sighted tasks. We administered the test 60 times with sensory feedback off and 30 times with sensory feedback on. For half the trials in each set, designated AV-, the subject wore a sleeping mask to remove visual feedback and noise-protection ear muffs over ear buds playing white noise to remove audio feedback. Failures were defined as any production of juice or visible cracks in the fruit skin during the task, and determined by a single evaluator for all trials. After each trial with the subject blinded, we asked the subject to rate his confidence in his overall performance before the mask and headphones were removed.

Statistical Analysis

Unless otherwise stated, we used ANOVA statistical testing and set significance as α = 0.05.

Supplementary Materials

Threshold Detection Method

We used the Single-Interval Adjustment Matrix, which is an unbiased, adaptive

staircase method (Kaernbach 1990) to determine the sensory perception threshold. We set parameters for a target performance of 50% with true stimulation provided 50% of the

time. The threshold search stopped after 12-16 reversals of perceived and not perceived

sensation. Stimulation was a 1 sec train and repeated if requested by the subject.

Stimulation frequency was constant at 20 Hz. To prevent overstimulation, we increment

stimulation pulse amplitude (PA) and pulse width (PW) by 0.1 mA and 10 µs steps,

respectively, until the rough threshold was determined. Then, we hold PA at constant

0.1 mA below the rough threshold and apply the adaptive staircase method to determine a

precise PW threshold with 1 µs resolution.

Figure 3.6 Subject S2 perceptual locations at near-threshold stimulation levels.

Cuff electrodes were highly selective, even though Subject S2 was implanted in the upper arm at a much more proximal location than Subject S1. This section of the nerve has a combination of both motor and sensory fibers and includes sensory fibers innervating the forearm and upper arm. Sensory response was actually elicited from 14 of 16 available contacts, but some of these were on the residual limb which we excluded from the report. We focused on the 9 of 16 contacts which produces sensation on the perceiveed hand. This is quite remarkable given the proximal location of the electrodes which demonstrates the generalizability of the multi-contact, extraneural interface approach.

Figure 3.7 Stable electrode impedances across study duration.

Impedances were measured across electrode channel pairs by providing a 0.3 mA and 50 µs stimulus pulse train at 20 Hz. Impedance was calculated by measuring the peak voltage drop. In both subjects, linear regression suggests impedances did not change significantly (subject 2, p = 0.271) or exhibited a decrease (subject 1, p = 0.009) throughout the study duration.

Figure 3.8 Subject S2 contralateral pressure matching.

In Subject S2, contralateral pressure matching also indicated frequency can modulate intensity of constant pressure sensation (N = 25, R2 = 0.663, linear regression p < 0.001). The subject was provided SSO modulation with IPI set to 50, 20, 14.3, 10 or 8 ms (20, 50, 70, 100 or 125 Hz) on channel MM6 and asked to match the perceived pressure (blue) with his contralateral hand. Perceived constant pressure intensity was on the order of 0-500 grams (<1 lb) similar to Subject S1. We also requested the subject to rate the intensity (red) on an oppen- scale, which showed surprising similar trends with measured force data.

Table 3.4 TAPES Pain Survey Data

The pain survey from the Trinity Amputation and Prosthesis Experience Scales (TAPES) was administered throughout the study, demonstrating a reduction of phantom pain in both subjects. There may be an assocciation with phantom pain reduction and the provision of natural sensory feedback without paresthesia.

Acknowledgments: We thank Melissa Schmitt for clinical support and management of the IRB compliance. We thank Jennifer Wall for regulatory support and managing FDA compliance. We thank Lee Miller and Sliman Bensmaia for review and editing early drafts of the manuscript.

Funding: This material is based upon work supported by the Department of Veterans

Affairs, Veterans Health Administration, Office of Research Development, Rehabilitation

Research and Development under Merit Review #A6156R and Career Development

Award 1IK1RX000724-01A1.

Author contributions: D.T. and M.S. contributed equally to experiment design, data collection, analysis and documentation. D.T. is a graduate student in the Rehabilitation

Research and Development (RR&D) Service at the Louis Stokes Cleveland Department of Veterans Affairs Medical Center (LSCDVAMC). M.S. and D.J.T. are investigators in the RR&D Service at the LSCDVAMC. M.K. and J.R.A. are surgeons in the

Orthopaedics Service at the LSCDVAMC and conducted the implantation surgeries. J.T. provided occupational therapy support. D.J.T. was the responsible Primary Investigator and contributed significantly to the experimental work.

Competing interests: There are no equity interests amongst the investigators in any entity related to this work. D.J.T. in an inventor of the patents for the FINE electrodes used in the study, but Case Western Reserve University holds the patents on this technology. D.T., M.S., and D.J.T. are named inventors on a provisional patent on the stimulation paradigm described in the paper, which is jointly held by Case Western

Reserve University and the Louis-Stokes Cleveland VA Medical Center.

Chapter 4 Stability and selectivity of a chronic, multi‐contact cuff electrode for sensory stimulation in a human amputee

Daniel Tan1,2, Matthew Schiefer1,2, Michael W. Keith1,3, Robert Anderson1,4, Dustin J.

Tyler1,2

Abstract— Stability and selectivity are important when restoring long-term, functional sensory feedback in individuals with limb-loss. We demonstrate a chronic, clinical neural stimulating system for providing selective sensory response in two upper-limb loss individuals for 1 and 2 years, the longest sensory feedback system to date. Multi-contact cuff electrodes were implanted in the median, ulnar, and radial nerves of the upper-limb. Nerve stimulation produced a selective sensory response on 19 of 20 contacts and 16 of 16 contacts in subjects 1 and 2 respectively.

Multiple distinct percept areas were elicited on the phantom and residual limb.

Consistent threshold, impedance, and percept areas have demonstrated that the neural interface is stable for the duration of this on-going, chronic study.

INTRODUCTION

Natural sensory feedback is an improvement in prostheses highly desired by individuals with limb-loss (Biddiss et al. 2007; Pylatiuk et al. 2007). A permanent sensory feedback system may lead to improved control of the prosthetic hand (Riso 1999; Witteveen et al.

2012), increased embodiment(Marasco et al. 2011), and elimination of uncomfortable phantom limb sensation (Flor et al. 2001; P. M. P. Rossini et al. 2010). Although recent advances have demonstrated the feasibility of sensory feedback (Horch et al. 2011;

Raspopovic et al. 2014), these studies have not demonstrated stability using a long-term neural interface in humans. Choosing an appropriate nerve interface for translation requires a consideration for stability of the nerve interface as well as selective, multi- point sensory response. Stability and selectivity are important when restoring long-term and functional sensory feedback in physically active individuals with limb-loss.

Researchers have long been investigating neural interfaces to provide natural sensory feedback to individuals with limb-loss. Early extra-neural stimulation research with single-contact nerve cuff electrodes on the median nerve in individuals with limb-loss provided a sense of paresthesia or a proprioceptive sensation of fist clenching(Clippinger et al. 1974). Intraneural stimulation of the median, ulnar, and radial nerves for sensory feedback also primarily recruited paresthesia but struggled with stability of the electrode position (Anani and Körner 1979). In 2003, a microelectrode array was implanted in a median nerve of a normal human but stimulation only recruited motor response and did not provide tactile sensory feedback (Warwick et al. 2003). Recently, distinct and graded pressure sensation on the phantom hand was demonstrated using intrafascicular electrodes implanted in median and ulnar nerves of a human amputee(Dhillon and Horch 2005).

Touch and tingling sensations were elicited from ~50% of multi-contact intrafascicular electrodes in another study (P. Rossini et al. 2010). However, with only 2-4 week acute clinical trials that also demonstrated upward trends in stimulation thresholds over time, the long term safety and stability of intrafascicular electrodes are unknown (Dhillon and

Horch 2005; Raspopovic et al. 2014; P. Rossini et al. 2010).

We demonstrate a chronic, clinical neural stimulating system for providing selective, sensory response in two subjects with upper-limb loss. Our approach uses multi-contact

72 cuff electrodes implanted on the peripheral nerve. At 12 and 24 months in two subjects, the system is also the longest sensory stimulation system in humans to date. In addition, it is the first successful, chronic human implant of the Flat Interface Nerve Electrode

(FINE)(Tyler and Durand 2003). The FINE takes advantage of the natural structure of peripheral nerves so that fascicle-selective stimulation is achieved (Schiefer et al. 2013;

Schiefer, Triolo, and Tyler 2008). The resizing Case Western Reserve University spiral electrode (Naples et al. 1988), which has been used extensively in restoring motor control in SCI patients (Fisher et al. 2009; Polasek et al. 2009), is also used in this study to provide sensory feedback.

METHODS

A. Surgical Implantation

Multi-contact nerve cuff electrodes were implanted in an outpatient surgical procedure on the peripheral nerves of two amputee subjects. Subject 1 is a 46 year old male who has a unilateral wrist disarticulation from work-related trauma. At the time of implant, the subject was 19 months (1.5 years) post-amputation. Eight-contact FINEs were implanted on the median and ulnar nerves and a 4-contact CWRU spiral electrode was implanted on the radial nerve. All electrodes were implanted in the mid-forearm and routed beneath the skin to percutaneous leads (Knutson et al. 2002; Letechipia et al. 1991; Polasek et al.

2009) that exited through the upper arm (Figure 4.1). The FINEs were sized 10 mm x 1.5 mm and the spiral cuff was sized 4 mm in diameter. Selection of electrode sizes were based on peripheral nerve histology studies in human cadavers (Brill and Tyler 2014) and with recommendation by surgeons during the implant operation. A spiral was selected for the radial nerve due to the small nerve size and superficial nature of the implant location.

Figure 4.1 A. Nerve cuff electrodes implanted in the forearm of the amputeee subject 1. Open‐helix percutaneous leads are passed individually through the skin so that the skin will grow into the open‐helix, preventing pistoning of the lead and reducing the risk of bacteria entry (inset). B. In subject 2, the electrodes are implanted in the mid‐ forearm.

Subject 2 is a 47 year old male who has a unilateral mid-forearm amputation from work-related trauma. At the time of implant, the subject was 93 months (7.75 years) post- amputation. Eight-contact FINES were implanted on the median, ulnar and radial nerves, however post-surgical imaging showed that the ulnar nerve cuff was not closed properly during surgery. All electrodes were implanted in the mid-upper arm and connected to percutaneous leads that exited through the upper arm (Figure 4.1.B). The size of the

FINES were 10 mm x 1.0 mm on the median and ulnar nerves, and 10 mm x 1.5 mm on the radial nerve.

The implanted electrodes were allowed to stabilize for two weeks before stimulation was applied. Swelling of the implanted limb was controlled by a coompressive sock.

Subjects were provided percutaneous site maintenance instructions and a daily supply of

3MTM TegadermTM or alternative waterproof dressing. Over the course of the study,

subjects did not report adverse pain or unusual phantom sensation outside of experimental

sessions. No infection of the percutaneous sites occurred to date.

Study protocol has been approved by the local Internal Review Board (Louis Stokes

Veterans Affairs Medical Center) and under an FDA Investigational Device Exemption.

Implanted components (cuff electrodes, percutaneous leads, connectors) were

manufactured by Ardiem Medical (Indiana, PA) and sterilized with Ethylene Oxide by

Ethox International (Buffalo, NY).

B. Nerve Stimulation

Subjects participated in nerve stimulation sessions once every 1 to 2 weeks for subject

1 and once every month for subject 2, based on subject availability. In each session,

stimulation was applied through each contact for up to 10 s. We provided a stimulus train

of monopolar, bi-phasic, charge-balanced, cathodic-first square pulses. Anodic return was

through a surface electrode on the dorsal surface of the upper arm immediately proximal

to the elbow. For all trials, the subject was blinded to the stimulation parameters. To

prevent dissolution of the stimulating contact, stimulation was limited to less than the

recommended charge injection limit of 0.5 µC/mm2. To prevent overstimulation, the

stimulation was limited to less than 50% of the duty cycle(Agnew et al. 1989).

Sensory perception thresholds were determined on every contact during weeks 2

through 8 post-op in subject 1. Threshold was determined using the Single-Interval

Adjustment Matrix, an unbiased, adaptive staircase method (Kaernbach 1990). The

parameters were set for a target performance of 50% (t = 0.5), which is the maximal difference between hit rate and false alarm rate, with true stimulation provided 50% of the

time. 50% of the trials were catch trials with no stimulation and were randomly

intermixed with stimulation trials to prevent subject bias. The threshold search was

defined as complete at 12-16 reversals. Stimulation was applied for 1 second and repeated

upon subject request. Stimulation frequency was held constant at 20 Hz. To prevent

uncomfortable sensation from overstimulation, stimulation pulse amplitude (PA) and

pulse width (PW) were incremented by 0.1 mA and 10 µs steps, respectively, until the

rough threshold was determined. Then PA was held constant at one step (0.1 mA) below

the rough threshold while the adaptive staircase method was used to determine a precise

PW threshold with 1 µs resolution.

For all subsequent sessions with subject 1 and all sessions with subject 2, threshold was determined using patterned stimulation intensity. In the full-scale modulation pattern, the

width of the pulses in the pulse train (f0 = 100 Hz) followed a slow (fmod = 1 Hz) sinusoidal envelope, which typically gives rise to a natural, non-tingly pulsing perception

(Tan et al. 2014). The pulse width varied between 0 µs and a maximum of B µs, where B is the measured threshold. Stimulation was applied in 5 second trials and repeated on subject request. The threshold PA level was determined by setting the peak of the sinusoidal PW to the maximum stimulation range (255 µs) and incrementing the PA stepwise by 0.1 mA until sensation was perceived. Then PA was held constant, while a binary search method was used to determine a precise PW threshold to 5 µs precision.

Typically the threshold was determined within 3-4 reversals. In addition, no catch trials were included because neither subject ever reported false positives in response to a catch trial at initial threshold evaluation.

Following stimulation, the subject was asked to describe any perceived sensation and

sketch the percept area on a hand diagram. The percept area drawings were scanned into a

computer for analysis.

Impedance measures were taken between contacts pairs of 1 & 2, 3 & 4, 5 & 6, and 7

& 8 on each electrode cuff using non-perceptible 0.3 mA and 50 µs stimulation pulses at

constant trains of 20, 50 and 100 Hz. Frequencies were selected based on typical

parameters used during functional testing with sensory feedback. The mean of 8 measures

of the resulting peak voltage drop between each pair of contacts was measured to calculate

the impedance.

Unless otherwise noted, ANOVA was used to test significance with α = 0.05.

RESULTS

A. Sensory Locations & Modalities

At two weeks post-op, stimulation provided sensation response on the phantom limb

from 19 of the 20 available contacts in subject 1. The non-responding contact was Ulnar-

4 (U4), and is thought to be due to an oversized cuff electrode. Electrode contacts are

each selective to a specific percept area, as shown at threshold stimulation levels indicated

in Figure 4.2.A. Patterns of sensory perception match the classical touch receptor

innervation areas for the median, ulnar and radial nerves. Although some threshold

percept areas overlap, each contact elicited differing super-threshold recruitment (Figure

4.2.B). Contacts were also selective for unique modalities of sensation. For example,

although three contacts on the median nerve cuff (M1, M6, M7) had overlapping percept

area at the wrist, M6 had a characteristic "deep, uncomfortable, sharp needle" sensation

whereas M7 produced a "superficial, tingly" sensation. Initial testing using constant

parameter stimulation evoked "tingly" or paresthesia on all contacts.

Various stimulation waveforms were explored over the chronic study period until we

were able to repeatedly produce natural, non-tingling tactile sensation with a sinusoidal (1

Hz) PW modulated waveform. Channels were selective to specific tactile sensations

including pressure (86%), vibration or tapping (7%), and light moving touch (7%). No

graded proprioception was observed in subject 1. However, when eliciting a pressure

perception on the tips of digit 1 and 2 (M3, M5), the subject sometimes voluntarily reports

that his perceived hand was in the shape of the “okay” position with the thumb and index

finger pinched together. At threshold stimulation, no motor recruitment was observed in subject 1. This is expected, however, as the implant location was distal to motor axon branching.

Initially, 10 of the 16 available contacts produced sensation in subject 2. However, the number of active channels increased to 16 of 16 by week 27 (Figure 4.2.C). Each contact was selective to characteristic percept areas on the hand. Stimulation through the median nerve cuff produced perceptive fields matching classical touch receptor innervations for the median nerve (Figure 4.3). During stimulation through the radial cuff, one contact

(R8) repeatedly and two contacts (R3 and R4 ) occasionally produced response on the radial innervation of the perceived hand, while the rest of the contacts produced sensations on the residual limb, mostly on the skin of the upper arm (Figure 4.2.C). Stimulation with

the sinusoidal (1 Hz) PW modulated waveform initially produced perceptions described as

tactile pressure, “a cold edge”, or “a jet of water moving across the hand.” Over multiple

visits, these sensations have resolved into simply tactile pressure (75% of channels) on the

median nerve cuff, similar to S102. However the radial produced primarily vibration

(75% of channels). One channel (M4) led to both tactile pressure and a proprioception of middle finger flexion at threshold stimulation levels. EMG recordings and visual confirmation indicates muscle twitch of a residual forearm muscle during perception of proprioception.

The perceptive areas of the stimulating channels resolved into relatively stable and repeatable locations over time in both subjects. Figure 3 shows monthly drawings from

week 2 to week 56 in subject 2. Interestingly, in subject 2, some median nerve cuff (M1,

M8) channels initially elicited percept areas on the dorsum and ulnar/radial innervation

areas. Over time, these shifted and have since settled to characteristic median nerve

innervation areas (Figure 4.4, top row). Most channels (M2, M4, M7) remain unchanged

in position even from the first experimental session (Figure 4.4, middle row). Some channels produce perceptive fields which alternate between two distinct locations, suggesting that the contact borders between two groupings of axons (Fig 3. M3, M8).

Figure 4.2 In subject 1, stimulation provides sensory response in 19 locationns on the phantom hand which cover classic innervation patterns for the median (M1‐8), ulnar (U1‐8), and radial (R1‐4) nerves. (A) Shown is the perceived locations at stimulation threshold. (B) Supra‐threshold stimulation leads to unique percept area recruitment. Shown is early measurements, week 2, before responses “settled”. (C) In 104, typical percept areas covered approximate median and radial innervation patterns of the hand and on the arm. Some radial channels alternate between two distinct locations, the hand and the arm (R3 and R4).

Figure 4.3 The cross section of the FINE implanted on the median nerve of subject 2 and the corresponding, channel‐specific percept areas are shown.

The example nerve cross section shown is from a comparable location taken from human histology studies (Brill and Tyler 2014) and is NOT from subject 2. Multiple measures, including suprathreshold response, up to week 56 is shown for each channel, showing most have a characteristic peercept area, impplies somatotopically organized fascicular structure. A similar drawing is not available for subject 1 since it is unknown the exact order of channels from his cuff electrodes.

Figure 4.4 Patterns of percept areas over time in subject 104.

M1 showed a shift over time toward proper median innervation, settling on the thumb. Most channels remained in the same percept area since the first experimental session (for example, M7). Initially, 5 of 8 channels elicited sensation (week 2). However all channels elicited sensation by week 27 (example M5).

B. Sensory Thresholds

In subject 1, the threshold for sensation perception was determined for all 20 contacts from week 2 through 8 post-op (Figure 4.3.A-C)). Mean peercept thresholds were 95.5 ±

42.5, 70.7 ± 59.2, and 40.7 ± 12.4 nC for the median, ulnar and radial nerves, respectively.

Linear regression on each contact over time produced a slope that was either not significantly different than 0 (n=18, p ≥ 0.103) or significantly decreasing (n = 1, p =

0.044) suggesting no significant increase in threshold over the first 8 weeks. From week 8 to 105, threshold measurements were prioritized for M3, M4, and M5, which produced the most functionally-relevant perceptive fields on fingertip and palmar areas. Mean

82 perceptive thresholds for the median nerve was 96.7 ± 36.8 nC. The threshold continued the trends of either decreasing threshold (M3, p < 0.001) or no change with time (M4, M5, p ≥ 0.09). In subject 2, thresholds on all channels were tracked for week 4 to 74 (Figure

4.5). The mean cuff thresholds were 125.91 ± 41.51 and 120.445 ± 32.47 nC for the median and radial nerves respectively. Again, similar trends were observed with most thresholds either not significantly changing (n = 12, p ≥ 0.20) or significantly decreasing

(n = 3, p ≤ 0.015). A single radial channel was significantly increasing (R8, p = 0.016) over time.

Impedance measures up to 94 weeks in subject 1 and up to 52 weeks in subject 2 are shown in figure 5. In subject 1, the mean channel impedances were 3.12 ± 0.15, 2.66 ±

0.15, and 2.91 ± 0.22 kOhm for the median, ulnar, and radial nerve cuffs respectively. In subject 2, the mean channel impedances were 2.92 ± 0.21 and 3.09 ± 0.19 kOhm for the median and radial nerve cuffs, respectively. Decreasing impedances were strongly correlated with time (p < 0.001) for the median nerve cuff of subject 1 and 2 and ulnar nerve cuff in subject 1 electrodes with Pearson’s correlation coefficient of -0.51, -0.73, and -0.49, respectively. There was no significant change over time for the radial electrodes in both subject (p > 0.42).

Figure 4.5 Threshold and Impedance measurements over time indicate stable neural interface.

Perceptual thresholds were measured for each contact for the first 8 weeks of the median (A), ulnar (B), and radial (C) nerve cuffs of subject 1. Threshold measures using a different stimulation waveform shown up to 105 weeks (E). In subject 2, threshold measures were recorded up to 74 weeks on every contact in the median (F) and radial (G) nerve cuffs. Impedance measures also suggest stable neural interfaces in subject 1(D) and subject 2 (H).

C. Super threshold recruitment of area

Figure 4.6 All percept areas, including super‐threshold response, elicited from wweek 2 to 56 foor contacts M2, M6, M4, M7 in subject 2 is shown with an overlay of the proper palmar digital nerves of median nerve adapted from textbook neuroanatomy (Jenkins 2001). Selective activation can be inferred by the pattern of recruitment and the relationship to underlying neuroanatomy.

DISCUSSION

An ongoing challenge in the field of neural interfaces is to produce selective neural stimulation with an interface that is stable for chronic, long-term clinical application.

Typically, these two goals are conflicting, as more selective interfaces usually require a more invasive approach to integrate with the nerve. Although recent work in intraneeural interfaces, such as intrafascicular electrodes (LIFE/TIME), microelectrode arrays, and regeneration electrodes, have advanced the field of neurmodulation and sensory stimulation, they have not demonstrated reliable, long-term performance in multiple

human subjects. We have demonstrated a peripheral neural interface that can be both

selective and stable in two human subjects, ongoing for more than 1 and 2 years. Such a

system is important for translating neuromodulation to the clinical setting.

Both of our subjects exhibit selective stimulation with approximately 15 unique percept

areas from 19 active contacts in subject 1 and approximately 10 unique percept areas from 16 active contacts in subject 2. Knowing the percept areas and the corresponding

stimulating contact on the cuff electrode may indicate fascicular organization of the nerve

bundle. For example, contacts M1, M5, and M6 all give rise to percept area of the thumb

tip and may be activating the same fascicle (Figure 4.3). Some contacts elicit a sensation

in the same locations but with different perceptual qualities (superficial versus deep

sensation on M6 and M7, subject 1). The high selectivity of sensory recruitment is

perhaps surprising, but gratifying for an external cuff implanted in the upper arm such as

in subject 2. Figure 6 shows an overlay of super-threshold sensory recruitment locations

and the textbook neural anatomy of the proper palmar digital nerve branches of the

median nerve (Jenkins 2001), suggesting these specific branches can be selectively

stimulated from the upper-arm. The results build on the emerging understanding of

neural anatomy that the nerve bundle retains a high level of somatic organization along

the entirety of the peripheral nerve (Hallin and Wu 2002; Prodanov et al. 2007), in

contrast to the plexiform branching and joining described by Sunderland (Sunderland

1945).

The mean threshold for eliciting perceptual sensation in our amputee subjects is 69.1 ±

36.0 nC and 109.7 ± 43.2 in subject 1 and 2 respectively. This is on average greater than the 25±17 nC found in previous chronic motor recruitment studies with a spiral nerve cuff implanted on upper extremity nerves in humans (Polasek et al. 2007) and the 21±18 nC found during interoperative studies with the FINE cuff implanted on lower extremity nerves in humans (Schiefer et al. 2010). Our chronic implant of the FINE should exhibit higher thresholds than interoperative FINE studies as encapsulation is an additional factor in chronic studies. The FINEs in our study also exhibit a much wider variance of thresholds than the spiral, as contacts near the center of the rectangular cuff will be closer to the nerve than contacts near the edge of the cuff. During the implant procedure of subject 1, the surgeons commented that the ulnar nerve was slightly smaller than the chosen cuff size. This may explain contact U2’s high threshold and the inactive U4 contact as they are likely to be on the edge of the nerve. The difference from previous studies may also be attributed to motor recruitment threshold being determined by EMG recordings whereas sensation threshold are being determined by conscious subject perception, a completely different neural pathway involving higher-order central processing. Additionally, axonal retrograde degeneration may have occured after amputation (Aitken and Thomas 1962; Chu 1996; McComas, Sica, and Banerjee 1978).

In subject 2, who had been an 8 year chronic amputee, during implantation surgery it was noted that the median and ulnar peripheral nerves of subject 2, who was 8 years post amputation, were smaller than expected from histological studies [unpublished].

We have demonstrated that multi-contact, cuff electrodes are ideal for stable, chronic nerve interfaces. Of all the active channels (n = 35) from both subjects, 97%

have no significant change (n = 30) or trend toward a decrease (n = 4) in thresholds over time with impedances showing similar trends of stability. Taken together, the threshold and impedance measures suggest a stable electrode interface (Figure 4.5). An increase in threshold and/or impedance would have suggested continued tissue encapsulation, nerve impairment, and/or electrode degradation. In addition, there was recruitment of additional channels over time to nearly 100% of available channels in subject two by week 27 (Figure 4.4, bottom row). The mechanism for the recruitment of additional channels is unclear, which may be peripherally or centrally-mediated. Axonal retrograde degeneration occurs after amputation (Aitken and Thomas 1962; Chu 1996; McComas et al. 1978). Tissue activity may lead to regeneration via functional hyperaemia, but it is unknown if occasional electrical stimulation leads to sufficient activity to encourage nerve regeneration. Increased encapsulation around the outside of the nerve cuff may

create a high impedance shell that focuses stimulation charge to the target area. In

contrast to our results, intraneural interfaces have not demonstrated long-term stability in

human trials. A peripherally-implanted MEA in a human exhibited a reduction of 85% of

functional channels (3 of 20) after 3 months(Warwick et al. 2003). Several studies of

intrafascicular electrodes implanted for up to 1 month exhibited rising

thresholds(Raspopovic et al. 2014; P. Rossini et al. 2010).

The percept areas of both subject reflected classical innervation patterns of median, ulnar,

and radial nerves. This suggests cortical reorganization in long-term individuals with

limb-loss does not interfere with sensory restoration from nerve stimulation, which is

consistent with previous reports by Dhillon et al. (Dhillon et al. 2004). Unexpectedly, early stimulation sessions of the median nerve in subject 2 resulted in sensations from

88 classically ulnar and radial innervated locations on the perceived hand. In contrast, subject 1 has always reported locations consistent with innervation patterns of the stimulated median, ulnar or radial nerve. Interestingly, those non-median nerve locations

(M1, M8) in subject 2 shifted during the first few months of stimulation and ultimately stabilized in median nerve locations. The expanded cortical representation, which included ulnar and radial percept areas, may have reduced to the original cortical representation of the median nerve during stimulation of the previously “quiet” deafferented pathways.

The large number of percept areas on the hands of both subjects suggest that our approach is a viable method for a sensory-enhanced prosthetic hand to provide multi-point sensory restoration. A sensory feedback system using our approach can provide matched- location feedback, capable of restoring sensation across the whole hand. This approach is more natural than providing feedback through sensory substitution methods, which usually require training to associate cross-locale sensation. In addition, both subjects reported fingertip percept areas, which are functionally important in grasp patterns. Fingertip recruitment was also reported in other studies (ref) and is likely due to a higher innervation density than other areas of the hand. Using all available contacts in a wide variety of locations, is it easy to imagine implementing a sensor-enhanced prosthetic hand and feedback system capable of improving activities of daily living, providing natural touch, and enhancing embodiment of prosthetic limbs.

CONCLUSION

Multi-contact cuff electrodes that elicit selective nerve responses have been achieved

by demonstrating characteristic percept areas and stimulating thresholds for each contact.

Stability has been demonstrated in percept area, threshold, and impedance measures from

5 electrode cuffs chronically implanted in two human amputee subjects for more than 1

and 2 years. This study is an excellent indicator of a clinically-ready, chronic system for

selective stimulation of the peripheral nerves with nearly 100% of the available contacts

providing sensory perception and stable threshold stimulation levels and locations. The

technology is not only applicable for amputees, but may also be applied in other clinical,

peripheral, neural prosthesis applications as well, including neural recording and pain

therapy.

Acknowledgment Special thanks to Joyce Tyler for provided occupational therapy support and Melissa Schmitt for clinical and study coordinator support.

Research supported by VA Merit Review #A6156R. 1 Louis Stokes Cleveland Department of Veterans Affairs Medical Center, Cleveland, OH 44106. 2 Case Western Reserve University, Cleveland, OH 44106. 3 MetroHealth Medical Center, Cleveland, OH 44109. 4 UH Rainbow Babies & Children's Hospital, Cleveland, OH 44106.

Chapter 5 Toward an artificial hand: natural sensory feedback improves task performance

Authors: Matthew Schiefer1,2†, Daniel Tan1,2†, Michael Keith1,2,3, J. Robert Anderson2,4, Joyce Tyler3, Dustin J. Tyler1,2,3*

Affiliations: 1Case Western Reserve University, Cleveland, OH.

2 Louis Stokes Veterans Affairs Medical Center, Cleveland, OH.

3MetroHealth Medical Center, Cleveland, OH.

4University Hospitals Rainbow Babies & Children's Hospital, Cleveland, OH

*Corresponding author: [email protected]

†authors contributed equally to the work.

Abstract:

Restoration of tactile feedback through direct nerve stimulation improved gripping and

task performance with a prosthesis for limb loss subjects. Tactile feedback is critical to

grip and manipulation of objects. Its absence results in subject reliance on visual and

auditory cues. A chronically implanted, stable nerve interface that provides natural

sensory feedback improved functional task performance in two subjects. The prosthetic

hand was instrumented with low-profile force sensitive resistors to record normal forces on the thumb, index, and middle fingers of the prosthetic hand during object

manipulation. Natural sensory feedback with intensity proportional to the measured force

was provided to the subject through the peripheral nerve interfaces. The subjects

perceived the sensation directly on the perceptual hand at locations corresponding to the

force locations on the prosthetic hand. A bend sensor measured hand opening span. The

intensity of pressure on the thenar eminence or the third digit was modulated by the

opening span. Three functional tests were performed with the subjects blindfolded. In

the first test, the subject was asked to determine whether or not a wooden block had been

placed in his prosthetic hand. In the second test, the subject had to locate and remove

magnetic blocks from a metal table. The third test was the Southampton Hand

Assessment Procedure (SHAP). Blindfolded performance with sensory feedback was

found to be similar to sighted performance in the wooden block and magnetic block

tasks. . Results from the SHAP show that sensory feedback does not degrade

myoelectric control. An embodiment questtionaire given after conducting functional

tasks indicated an improved sense of embodiment with sensory feedback. Sensory

feedback reduces reliance on visual feedback when using a prosthesis.

Introduction:

The sense of touch is essential to the human experience. Touch is fundamental to

adroit manipulation of objects and social interactions (Augurelle et al. 2003; Robles-De-

La-Torre 2006). Beyond the loss of function, a devastating consequence of upper extremity (UE) amputation is the loss of sensation. Without tactile sensation, individuals with limb-loss are forced to rely on visual and auditory cues for task performance.

Without sensory feedback, the prosthesis is disassociated from the sense of self,

perceived as an extension beyond the body and accompanied by a decrease in confidence

during prosthetic use (Tan et al. 2014). Sensation is important for improved control of

the prosthetic limb (Riso 1999; Witteveen et al. 2012), prosthesis embodiment (Marasco

et al. 2011; Tan et al. 2014), and for the reduction of phantom pain (Ramachandran and

Rogers-Ramachandran 1996; Tan et al. 2014).

There are approximately 41,000 individuals in the United Stated living with major

UE loss (Ziegler-Graham et al. 2008). Additionally, 500 military personnel experienced

major UE amputations during combat operations between 2000 and 2011 (AFHSC 2012).

The standard in UE prosthetics has long been a cable-based, body-powered system which

is used in about 31% of the limb-loss population (Biddiss et al. 2007). Compared to other

UE prostheses, body-powered systems provide the most usable sensory feedback.

However, this feedback is in the form of the tension developed on a cable and not a direct

sensation in response to environmental interactions.

An alternative option is the myoelectric prosthesis, which relies on voluntary

contraction of the muscles in the residual limb to control the position of the hand, and is

used in about 43% of the limb-loss population (Biddiss et al. 2007). While myoelectric

prostheses have greatly improved cosmesis and functional grip patterns over recent years,

they lack sensory feedback, forcing users to rely entirely on visual and auditory cues, which has proven to be one of the primary complaints associated with these devices

(Atkins, Heard, and Donovan 1996; Light et al. 2002). Tasks requiring split attention become difficult.

Multiple strategies exist to supplement feedback. The method of sensory substation provides a surrogate sensation, typically vibration, somewhere on the body. The

surrogate sensation’s intensity is scaled to reflect a change in the system, such as the amount of pressure applied at the prosthetic fingertips or the position of the prosthetic hand. Sensory substitution requires both modal interpretation, i.e. vibration encodes

pressure, and spatial interpretation, i.e. sensation at a distant location on the body encodes

an event occurring at the end of the prosthetic. Despite the required interpretation,

sensory substitution has improved prosthetic control in laboratory settings (Pylatiuk et al.

2006; Witteveen et al. 2012), but has not been widely adopted, possibly because the sensations are not natural or due to the additional effort required to interpret the feedback.

Even with sensory substitution, visual feedback was required to accurately comprehend

hand aperture (Witteveen, Rietman, and Veltink 2014).

An alternative strategy for sensory restoration is electrical stimulation of residual

sensory pathway. Peripheral cutaneous (Scott et al. 1980; Szeto and Saunders 1982),

sensory cortex (Berg et al. 2013), and deep brain thalamic (Heming et al. 2011)

stimulation have been explored. While cutaneous stimulation is attractive because it is not invasive, it also lacks spatial and modal specificity. The primary benefit to intracranial stimulation would be perceived sensation in its intended location: the phantom hand. However, in the case of deep brain stimulation, the percept was often distributed across a large area, unnatural, and paresthetic (tingling).

The ideal sensory feedback mechanism would be one that produces the same perception as the natural limb. In the 1970s and 1980s, researchers attempted to produce various sensations with extraneural cuff or intranerual fine wire electrodes (Anani et al.

1977; Anani and Körner 1979; Clippinger et al. 1974; Ochoa and Torebjork 1980; Ochoa and Torebjörk 1983). Typically, subjects reported paresthesia, vibration, or pulsing

95 spread across the phantom hand. Intrafascicular electrodes have produced additional sensations of touch, pressure, and movement at more punctate locations (Dhillon and

Horch 2005; Dhillon et al. 2004, 2005; Horch et al. 2011; Raspopovic et al. 2014; P. M. P.

Rossini et al. 2010) but the stimulation threshold increased over the 28-day study and complete loss of ability to produce sensation was reported as early as 10 days (P. M. P.

Rossini et al. 2010). Paresthesia was associated with 30%-50% of the stimulating channels. Nonetheless, the restored sensation allowed a subject to correctly identify three different objects using sensory onset timing, illustrating the value of sensory feedback on functional control (Raspopovic et al. 2014).

Until recently, a chronically implanted, stable system which provides natural sensory feedback to an individual with limb-loss did not exist. Here, we, used extraneural cuff technology previously found effective for motor recruitment in humans (Fisher et al.

2009; Grinberg et al. 2008; Polasek and Schiefer 2009; Schiefer et al. 2010; Tarver et al.

1992) and animals (Naples et al. 1988; Tarler and Mortimer 2004; Tyler and Durand

2003) combined with novel neural stimulation paradigms (Chapter 3). The stable and selective nature of our system has allowed us to conduct experiments for up to 24 months, providing sufficient time to determine how to produce stable sensations of natural pressure without the ubiquitously reported paresthesia (Chapter 3). This study investigates the effects of sensory restoration on an limb-loss individual’s performance on functional tasks and embodiment of the prosthesis.

Methods:

Two subjects with a UE amputation were implanted with chronic, direct sensory feedback systems (Table 5.1, A). Cuffs were implanted around three nerves in the forearm of subject S1 and two nerve in the upper arm of subject S2. Cuffs were either the 8-channel flat interface nerve electrode (FINE) (Tyler and Durand 2002) or the 4- channel self-sizing spiral nerve cuff electrode (Naples et al. 1988) (Ardiem Medical,

Indiana, PA). In both subjects, percutaneous lead wires were routed from the implanted cuffs to an exit site in the lateral arm. Each implant was an outpatient surgery, performed at the Louis Stokes Cleveland Department of Veterans Affairs Medical Center

(LSCDVAMC) and were approved by the LSCDVAMC Institutional Review Board (IRB) and the FDA under an Investigational Device Exemption (IDE). Surgery required 4-6 hours for nerve dissection, cuff implants, lead tunneling, percutaneous lead positioning, suture, and follow-up X-ray. Subjects healed and the cuffs were allowed to stabilize for 3 weeks before experiments began.

S1 S2 Amputation History Location Wrist Prox. Forearm Year 2010 2004 Cause Forging Industrial Hammer Shredder Amputation Outcomes Neuroma No No Phantom Pain Mild Excruciating Level 2/Week 2/Month Pain Frequency Phantom No Yes Paresthesia Implant Location Mid-Forearm Upper Arm Date May 2012 January 2013 Nerve Cuffs Median FINE FINE (10.0 x 1.5 (10.0 x 1.5 mm) mm)

Ulnar FINE -- (10.0 x 1.5 mm)

Radial Spiral FINE (4.0 mm) (10.0 x 1.0 mm) Total Electrodes 20 16 Table 5.1 Subjects enrolled in the sensory restoration study

Figure 5.1(A) S1 was implanted with 2 FINEs and 1 spiral nerve cuff. The FINEs were implanted on the median and ulnar nerves. The spiral was implanted on the radial nerve. Leads were tunneled to the lateral upper arm, where they exit as 20 helical wires. S2’s amputation is in the proximal forearm and cuffs were implanted in the distal arm. (B) The subject’s prosthetic hand was mounted with low‐profile pressure sensors on the pads of D1‐D3 as well as a bend sensor measuring the D1‐D2 angle (not shown). Both subjects used their own prosthetic hand for the tests: the Otto Bock SensorHand Speed with 1 degree of freedom and 1 grip pattern. The internal slip sensor was disabled. The pressure and bend sensors regulated the stimulus applied to the nerves. (C) S2 using his instrumented prosthetic to locate and remove magnetic blocks from a metal platform while blindfolded.

Stimulus was delivered to one or more contacts inside the cuffs implanted around the subjects’ nerves via the percutaneous leads attached to a custom-built, computer- controlled stimulator, the universal external control unit (UECU). The programmable stimulator delivered waveforms with customizedd pulse width (PW), pulse amplitude (PA),

and frequency (fstim). Details of the stimulus have been previously published (Tan et al.

2014). Each subject participated in 2-4 experimental sessions per month.

To assess functional outcomes, the prosthetic hand was instrumented with low-profile

pressure sensors (Tekscan FlexiForce, Boston, MA) mounted over the pads of the thumb

(D1), index (D2), and middle (D3) fingers (Error! Reference source not found.B).

Additionally, a bend sensor (A.G.E., New York, NY) measured the opening aperture, i.e.

the angle between the thumb and index finger. The pressure sensors on D1, D2, and D3 regulated the stimulus delivered to specific contacts in the FINE on the median nerve. As the pressure, P, on the sensor increased, the frequency of stimulation increased in a linear fashion according to

Equation 5.1

∗ ∗ (1) 125

where Pmax was the maximum pressure expected to be measured during task performance.

Typically, m was set to 1.1, b was set to 14.5, and α was set to 0.05, meaning that any

pressure less than 5% of the maximum pressure resulted in no stimulation. However, the

equation was fine-tuned to compensate for baseline pressure readings. Sensor sampling

rate was 20 Hz. The subject did not perceive a delay between seeing the physical sensor

contact and the perceived sensation. The subjects’ standard myoelectric control of the

prosthetic was use for all tests. Therefore functional tests only evaluated the effect of

sensory feedback.

100

Functional Tests

Three tests were administered to assess the effect of sensory feedback. The first test was a forced-choice object detection test (ODT) that quantified the subject’s ability to determine if a wooden block had been placed between D1 and D2 during a pinch grip.

The subject was presented with a choice 20 times, in which half of the presentations included a block and the other half did not. Block presentation was randomized. The test was scored based on the percentage of correct responses. The ODT was conducted with the prosthetic hand, while occluding vision and hearing, under 3 conditions (Error!

Reference source not found.): 1) without sensory feedback; 2) with pressure feedback only; and 3) with pressure and opening aperture feedback. Vision was occluded with a blindfold. Hearing was occluded by playing white noise through ear-bud headphones covered by noise-canceling ear muffs. Before each condition, subjects were allowed to practice until they were satisfied with their performance (about 5 to 10 minutes). Each condition was tested three times. The test was also performed with the intact hand while blinded and with the prosthetic under typical unblinded conditions. These latter scenarios were later abandoned because subjects performed these tests perfectly, which was not surprising because neither subject had tactile or visual deficits.

Feedback Hand Vision & Hearing Pressure Aperture Code

Intact (I) No Yes Yes , & Yes No No , Prosthetic No No No , (P) No Yes No , No Yes Yes , Table 5.2 Conditions under which ODTs and mBBs were performed

101

Highlighted row reflects normal operating condition. Prior to each set of trials, subjects were

allowed to practice for a self‐selected amount of time to become familiar with the specified

conditions.

The second test was a modification of the standardized box and blocks test (mBB).

This test required the subject to locate, pick up, and remove as many magnetic blocks as

possible (up to 10) from a metal platform in a 2 minute time span. Each side of the

wooden block was 2.54 cm. A small neodymium magnet (0.32x0.64x0.16 cm3) was

attached to one side to prevent the subject from knocking the block off the test platform,

provide a resistance to removal and require a minimum grasp pressure. The metal

platform was 34.3 x 40.6 cm2. The stance assumed and the technique employed to complete the task was self-selected by the subject. The test was scored based on the number of blocks successfully removed and the number of failures. Failures were defined as 1) an attempt to move a block when one was not in the hand, 2) dropping a block, or 3) pushing the block off the platform. Blocks left on the table after 2 minutes were also counted as failures. The amount of time required to remove each block was calculated. Finally, efficiency was quantified as the number of successes normalized by the sum of the number of successes and failures. The mBB was conducted three times under all five conditions.

The Southampton Hand Assessment Procedure (SHAP) is a standardized and validated assessment of upper limb function (C. M. Light et al. 2002) that has been applied to myoelectric prosthetics (Kyberd 2011). The SHAP measures manipulation of household objects to assess a person’s ability to perform Activities of Daily Living

(ADLs) involving different grips. The duration of time to complete each task is recorded.

102

This test was performed prior to surgical implant and post-implant under two conditions:

& & , and , . . This test was not specifically designed to assess sensory

feedback but it does serve as a standardized assessment of function within prosthetics.

As such, this test was conducted to assess if the sensory feedback system reduced or otherwise interfered with myoelectric control.

Proportional Analysis, Chi-Square, ANOVA, and Tukey’s test for multiple comparisons were used to assess if sensory feedback had a statistically significant affect.

The measure of time as a significant factor was evaluated for tests repeated on multiple days. If time was not found to be significant, those data were pooled. In all cases, p <=

0.05 was considered significant.

To evaluate the subject's level of embodiment during functional tasks with

sensory feedback, we administered an adapted embodiment questionnaire (Ehrsson et al.

2008; Marasco et al. 2011) immediately following functional testing. The questionnaire

has 9 statements which the subjects rate between -3 and +3 corresponding to strongly

disagree and strongly agree, respectively. Three statements correlate with embodiment,

while 6 evaluate task compliance and suggestibility. We presented the statements in

random order. We adapted the survey by changing the phrase “touch of the investigator”

to “touch of the objects [manipulated during functional tasks]”. We pooled data from

functional tests on both subjects and compared responses to the questionnaire when

sensory feedback was off to responses when the sensory feedback was on using the two-

sided, paired t-test.

103

Results

Neither subject experienced complications due to the surgery, implant, or

percutaneous leads. There have been no complications due to infection over the 888 percutanous-lead-months. S1 has exhibited no evidence of motor recruitment during stimulation on any contact. S2 has exhibited motor recruitment when the stimulus levels were too high on a small number of contacts. These levels were always greater than the sensory thresold. Lack of motor recruitment from sensory stimulation is desirable for an individual using a myoelectric prosthesis else it may interfere with myoelectric control and adultrate intended movements or initiate unintended movements.

ODT

When presented with a forced-choice test to determine if a wooden block had been

placed between D1 and D2 during a pinch grip, both subjects performed similarly under

the varied conditions (Figure 5.2). An ANOVA revealed that there was a weakly

significant difference between subjects (p = 0.053), but the results were not significant

across time (p = 0.435). As expected, both subjects always identified the presence of a

block with their intact hand during initial tests. Without any visual, auditory, tactile, or

hand aperture feedback, the subjects did not perform statistically different than chance

(0.245 ≤ p ≤ 0.897) with a single exception of S1 on Day 2 (p = 0.001). Addition of force

perception significantly improved performance in S1 (p = 0.018 ), but not in S2 (p =

0.209). Further inclusion of hand aperture resulted in an accuracy of 86% ± 11% and

94% ± 4% for S1 and S2, respecitvely, which was significantly greater than chance (p <

104

0.001) and under the , condition ( p < 0.002) for both subjects. This analysis was also confirmed using a 2 test on tables of true/false positives/negatives.

Figure 5.2: ODT‐2 correct responses for S1 (left) and S2 (right) on two different days of experimentation. Both subjects were always able to identify the presence of a wooden block with their intact hand (not shown). Without sensory feedback, only S1 on Day 2 performed differently than chance. Providing feedback about the pressure experienced at the fingertips was found to significantly improve accuracy. Furtther addition of aperture feedback resulted in improved performance approaching that of their intact hand.

mBB

When blindfolded and audition occluded but using their intact hand or when using their prosthetic hand with visual and auditory feeedback, both subjects were able to remove all blocks from the magnetic table in less than the alllotted two minutes. There was no statistical difference between these two conditions. Conversely, under the

, condition neither subject was able to remove all blocks from the table within the allotted two minutes (Figure 5.3). There was, howwever, a significant difference overall between the performances of the two subjects on the mBB: S1 had a higher success rate than S2 (p = 0.001). Additionally, both subjects eexhibited a significant increase in the

105 number of blocks removed during their second experiment when compared with their first experiment (p <= 0.022), even when separated by more than 31 days.

Figure 5.3: Total number of failures on the mBB test versus percentage of magnnetic blocks successfully removed from the metal table for S1 (top row) and S2 (bottom row) on day 1 of testing (left column) and day 2 of testing

(right column) when using the prosthetic hand. Failure was defined as the sum of 1) attempts to move a block when one wasn’t in the hand (empty pinch); 2) dropped blocks while in the processes of a movement; 3) blocks pushed off the table; and 4) blocks remaining on the table after two minutes. Success rate increased and failures

were reduced when the subject was blinded and supplemented with pressure and hand aperture feedback, trending toward sighted, performance without supplemental feedback. In one case (S1, Day 2), these scenarios were not statistically different.

106

Without any feedback, S1 removed 30% ± 10% of the blocks on Day 1, while S2,

although able to locate blocks on the table, was unable to remove them. This led to the significant difference (p = 0.031) between the two subjects for the first experimental sessions. However, on the second experimental day S1 removed 33% ± 12% and S2 removed 27% ± 15% of the blocks, which was not significantly different.

In 50% of trials, more blocks were removed under , than under , ,. In

trials where more blocks were removed, the increase was significant (p <= 0.012), but in

trials where fewer blocks were removed, the decrease was not significant (p >= 0.794).

In all trials, both subjects removed more blocks under , than under , .

As with the success rate, the two subjects exhibited a similar number of failures.

Unlike the number of successes, the number of failures were not significantly different between the two subjects (p = 0.064). The greatest number of failures occurred without any sensory feedback, regardless of subject and day and was always significantly greater

& than , and , (p <= .003). While the success rate was perfect under these

latter two conditions, S1 did have failures when using his prosthetic hand a under normal

operating situation. However, the difference in the number of failures under these two

conditions was not significant. The number of failures exhibited by both subjects was

found to significantly decrease with time (p <= 0.049). Addition of pressure sensation

alone always reduced the number of failures, though only once by a significant amount.

Addition of hand aperture feedback further reduced the number of failures to a level that

was always significantly less than under , (p <= 0.012). In fact, the number of

& failures under , was not significantly different than under , in three of the

four experiments.

107

Based on the success and failure observed for each subject, their efficiency was

& calculated and normalized by the efficiency found under the , normal operating

condition. While the efficiency for both subjects was always less than 1.0, it was as high

as 0.89 for S1 on Day 2 when pressure and apeture feedback was provided. Typically,

however, the efficiency remained below 0.5. On the whole, S1 was found to be more

efficient that S2 (p < 0.001) and was able to achieve an efficiency during , that

& was not statistically different than that achieved during , .

SHAP

Both subjects performed the SHAP prior to surgery and then again 358 (S1) and 124

(S2) days post-implant without sensory feedback. A paired t-test of the 26 SHAP tasks showed no significant difference between pre and post-implant assessments (0.182 ≤ p ≤

0.937). For post-implant tests, supplementing the subject with sensory feedback did

improve performance. S1 had a 22.4% improvement from an Index of Function (IoF)

score of 58 to 71, which was significant (p = 0.026). S2 had a 7.8% improvement in IoF

score from 51 to 55.

108

Embodiment

Both subjjects reported higher embodiment with sensory feedback enabled. Both subjects responded similarly and their results were pooled. Without sensory feedback, agreement with embodiment statements and contrrol statement did not differ significantly

(p = 0.348). With sensory feedback, embodimennt statementss were significantly different and greater than control statements (p < 0.001). Control statements with feedback were significantly better than the control statement without feedback (p = 0.012) (Figure 5.4).

Figure 5.4 Embodiment improves with sensory feedback enabled during functional tasks.

Discussion:

This study marks multiple milestones. S1 is the longest and S2 is the most proximal recipient of nerve cuff electrodes to restore sensation. There were many differences between the two subjects. S1 received implants in the middle forearm, distal to major motor branches, while S2 received implants in thhe middle arm and motor recruitment was observed, though at levels much higher than that required to restore sensation. Another primary difference between S1 and S2 was the time that lapsed between the amputation and the implant. S1 was implanted 1 year and 6 months after his amputation, whereas S2

109

was implanted nearly 7 years and 8 months after his amputation. Additionally, S1 had

been a regular user of his myoelectric prosthesis for 7 months prior to implant, whereas

S2 infrequently (“about once per month”) used his myoelectric prosthesis prior to implant.

Post-implant, both subjects use their myoelectric prosthesis as their primary prosthesis.

Despite these differences, both subjects perceived natural sensations of pressure on

two important locations for pinch grip – the thumb (D1) and index finger (D2) – among

other punctate locations (for complete description, see Chapter 3). Despite any cortical

reorganization that may have taken place, both subjects immediately felt the effects of

stimulation in these localized areas, where sensations have remained stable to date.

Perhaps this was serendipity. Alternatively, this may point to a preservation of

underlying axonal organization within nerves contrary to the plexiform disorganization

proposed by Sunderland (Sunderland 1978) as well as preserved cortical neural networks.

Others have proposed using intrafascicular electrodes to facilitate stimulating only

one or a very small population of axons. While there may be advantages of achieving

such a resolution, the results in [Chapter 3] and here suggest that it is not necessary to

interface with axons on a near 1:1 basis. Instead, we believe the key to producing

punctate sensations is an underlying axonal organization along the nerve. Axons from

adjacent regions of the hand maintain close proximity, at least as proximal as the mid-

arm.

With the major differences between S1 and S2, one might have expected to see

dramatic differences in performance on the functional tests. Interestingly, S1 and S2

showed similar performance on all of the tests. Enough so, that under many conditions

the ANOVA revealed that there was no statistical difference between the subjects. This

110

suggests that the nerve cuffs used in the study are robust enough to produce the necessary

localized natural sensations needed for pinch grip.

Time was often found to have a significant effect on the results and the effect of time

was greater for S1 than for S2. Although the time in the study was reported as Days 1

and 2, this was not to suggest that the experiments occurred on consecutive days. Each

subject’s first experiment occurred approximately 5 months post-implant, while 260 days

and 32 days lapsed before the second experiment for S1 and S2, respectively. During the

interim, both subjects returned for other experiments. This may suggest that learning

occurred between the first and second experiments, which would lend further support for

expected improvement over time. Alternatively, this may reflect the natural variance in

human performance with sensory feedback.

While training with sensation is not expected to alter the outcomes of the ,

tests that lack all sensory feedback, it is likely there would be an improvement in the

, and possibly the , tests. Even with the 5-10 minutes of training that the

subjects received prior to the ODT-2 and mBB tests, the outcomes were trending toward

intact hand performance. This suggests two things. First, the sensations must be close to

natural because minimal training was needed to understand the information that was being conveyed to the subject. Second, further training would likely result in outcomes that mimicked those obtained with the intact hand, with the exception that time to locate objects would likely remain unchanged.

During the ODT-2 tests without supplemental sensory feedback, both subjects’

accuracy was slightly better than chance. When asked what information they were using

to determine if there was block in their prosthetic hand, both subjects indicated they were

111 relying solely on the time required to open and close the hand as determined by the small amount of vibration they could feel through the socket as the motor operated the hand.

Performance improved on the ODT-2 when pressure feedback was provided. If pressure was, in fact, the only feedback the subjects received, then theoretically the subjects should have performed at chance, unable to distinguish the difference between squeezing the fingers against a hard block and squeezing the fingers against each other. However, pressure feedback alone allowed the subjects typically to perform at a level significantly greater than chance. As was the case when there was no sensory feedback, the subjects relied on timing to determine if there was or was not a block in their hand. This time, however, instead of relying solely on the transmitted vibration of the motor, they also relied on the onset of the sensation of pressure. If the sensation of pressure occurred sooner, there was a block in the hand. Although pressure significantly improved accuracy, performance was still not as accurate as with their intact hand in both subjects.

While proprioceptive feedback alone (, ) was never assessed on the ODT-2, it is likely that the increase in performance found with pressure and proprioceptive

feedback (, ) was due predominately to the proprioceptive feedback. Following the test, both subjects immediately indicated that they knew if there was an object in their hand not because of the pressure on D1 and D2, but because they knew if their hand was fully closed or partly closed. Additionally, one subject explicitly stated that he ignored the sensation of pressure that was provided. These results suggest that, for hard, immutable objects, a sensation of the hand’s position is sufficient to distinguish sizes and, therefore, the presence or absence of an object, assuming a sufficiently tight grip.

112

In addition to the ODT-2, the mBB test illustrated the importance of receiving proprioceptive sensation. As with the ODT-2, pinch force alone typically improved successes and decreased failures in the mBB. Having only a sensation of force frequently confounded the subject: it was not clear if the subject was squeezing an object or squeezing the fingers of the prosthetic together. Proprioceptive feedback allowed the subjects to better gauge the degree to which the hand was opened or closed. We found that continuous levels of proprioception were too difficult for the subject to discern, in

part because sensations attenuate over time, making the precise sensation challenging to gauge. Instead, we used a feedback strategy in which the D1-D2 angle was discretized.

Subjects were able to reliably distinguish three levels. Just noticeable difference tests should be conducted to determine the finest useable resolution. It may be possible to

move back to a graded feedback given sufficient training.

On the mBB, it is most likely the case that the differences observed in success rate

under , were due mostly to chance and whether or not the block fell between D1

and D2 during pinch grasp. It was found that both subjects assumed similar stances and

used similar techniques to locate blocks with their prosthetic hand while blinded.

Specifically, subjects stood, provided additional balance support by pushing the

contralateral hand against the table, and, with the fingers of the hand pointing downward

toward the magnetic table, prodded the table in a descending motion. Objects were

located by determining when the prosthetic finger tips were stopped at a higher elevation.

Upon gross localization, the subjects then attempted to resolve the nearest part of the

block with D1, though occasionally a block was encountered with one of the other digits.

A video illustrating mBB performance has been made available online(Tyler 2013).

113

With regard to the normalized efficiency, performance under , wasn’t

& significantly different from the positive prosthetic control , and was

significantly greater than the negative control , for S1. Importantly, this illustrates that with pressure and proprioception, the subject achieved an efficiency that was not significantly different than that when he used his prosthetic hand with vision. In many

ways, the , , , , and , tests are overly challenging. It’s unlikely that an

individual with limb-loss will conduct a task with the prosthetic hand without devoting

any visual attention to it. However, what this study suggests is that it should be possible

for an individual with limb-loss to devote less visual attention to the prosthesis provided

there is natural sensory feedback about the prosthetic’s interaction with the environment.

This may then allow the user to devote visual attention to a simultaneous task. Although

both subjects had similar efficiencies in their first experiment and both subjects

experienced an improved efficiency over time, this degree of efficiency was not achieved

by S2 during his second experiment.

In addition to the overall improvement on the ODT-2 and mBB tests, the subjects

were also found to have an increased confidence in their abilities when receiving sensory

feedback. On day 2, S1 reported that he believed he had correctly identified a block in

40% of the ODT-2 trials under , . His confidence increased to 75% when he was

supplemented with pressure feedback and further still to 80% when he also received

proprioceptive feedback. Similarly, S2 reported a confidence of 30% on , and

, trials. However, when he was additionally supplemented with proprioceptive

feedback, his confidence level also increased to 80%. While both outperformed their

114

confidence under , , it’s important to note that neither was strongly confident in his

abilities until both pressure and proprioceptive feedback were supplemented.

Results from pre-implant and post-implant SHAP tests without sensory feedback suggests that functional performance using the myoelectric prosthesis was not affected by

the surgery, surgical implants, electrical nerve stimulation, nor percutaneous leads. In

addition to having no negative impact on performance, both subjects’ SHAP score improved when provided with sensory feedback compared to the same-day evaluation without feedback. These tests were only administered once under both conditions, so it is possible the improved scores fall within the normal range of variance. Assumed

intrasubject variance was based on published SHAP reliability measures in able-bodied

subjects and those using a 1 DoF myoelectric prosthesis (Kyberd 2011). Alternatively,

& & because in both cases , was performed after , , the improvement may

have been due to learning. The improvement is, in fact, surprising because both subjects

reported that they ignored sensory feedback and relied entirely on vision, which is not

surprising during a task requiring speed. Regardless of the improvement, the important

point is that performance on the SHAP was not adversely affected by the presencence or

functioning of the sensory feedback system, which is essential to move forward to the

next stage: a take-home system.

This study contributes to a growing body of knowledge that suggests that

percutaneous leads, when maintained and cared for properly, do not lead to infection. A survey of 858 percutaneous electrodes with the same or similar lead design and implanted in the upper extremity found that only 0.6% were removed due to a suspected infection (Knutson et al. 2002). With regard to the percutaneous sites, both subjects

115

reported that there was no discomfort, irritation, or itching. Neither subject has had a

problem at their percutaneous sites except for 4 occurrences of a clogged sebaceous pore,

which was treated by expunging the clog and providing topical and oral antibiotics to

mitigate a potential infection.

Agreement ratings for Embodiment statements were significantly greater than that of

the control statement, suggesting greater embodiment of the prosthetic hand when

sensory feedback was provided. Embodiment will likely to be increased with agency

(Dolezal 2009; Murray 2008; de Vignemont 2011) giving reason to examine both

embodiment and functional tasks together. Our subject has stated that when he uses the

Greifer, he feels as if his hand is holding a split-hook tool. Whereas using the visually-

realistic myoelectric hand with sensation, he feels as if that is his hand as the prosthesis.

Even if a subject cannot attain improved control, a perceptually intact hand is a highly

beneficial aspect of sensory feedback.

Conclusion:

Direct sensory feedback improves functional capability by 1) improving the

individual’s ability to manipulate objects; 2) increasing the individual’s sense that the

prosthesis is, in fact, a part of his body; and 3) improving confidence. Our multi-channel

nerve cuff electrodes provide a platform solution that has produced chronic, stable, natural, and selectively restored sensations in humans. In general, restoring a sensation of pressure and proprioception was found to increase the number of successes while decreasing the number of failures on two functional tasks in both subjects. Both subjects also showed improvement over time and tended toward the results obtained with positive

116

controls. Continued use and learning is expected to further improve performance.

Perhaps as important, both subjects reported increased confidence in their actions and a clear preference for their restored sensation. These data suggest that continued long term studies should be conducted, take-home systems should be developed, and additional subjects should be enrolled.

Chapter 6 Summary ‐ Conclusion

The overall goal of the research was to provide sensory feedback to amputees. Our

hypothesis: Extraneural, multi-contact cuff electrodes chronically implanted on the

median, ulnar, & radial nerves will provide functionally beneficial sensory and

proprioception feedback to the user of upper-limb prosthetic devices.

Specific Aim I: Characterize sensory perception from chronically‐implanted,

peripheral nerve cuff electrode stimulation.

In a first-in-field trial, we implanted a total of 5 multi-contact cuff electrodes with 36

stimulating channels in two human subjects for the purpose of exploring sensory response from nerve stimulation. Sensory perception was achieved on 97% (35 of 36) of the

stimulating channels which is beyond our expected result of 50% response (Chapter 3 and 4). We mapped the relationship from parameters of the stimulation to the resulting percept area, modality, and intensity of sensation. Multi-contact cuff electrodes selectively recruit more than 3 percept areas, approximately 15 unique percept areas in subject and 10 in subject 2, notably including distinct finger areas in both subjects

(Chapter 3 and 4). We discovered a unique stimulation waveform called patterned stimulation intensity which transitioned tingling sensation into natural tactile modalities

117

of sensation including pulsing pressure, constant pressure, light moving touch, vibration,

sharpness (Chapter 3 and 4), and finger flexion (Chapter 4 and Appendix A). In doing so,

we achieved our aim by demonstrating more than 2 modalities of sensory perception

through a multi-contact cuff nerve interface. The intensity of the perceived sensation was

mapped through frequency or PW modulation, with intensity indicated by contralateral

pressure matching, demonstrating more than 3 levels of intensity (Chapter 3 and

Appendix A). We characterized threshold and impedance response and found that the

chronic implants were stable for up to two years in human subjects. Accomplishing this

specific aim demonstrates that natural sensory perception is achievable though a multi-

contact, extra-neural interface, which has not been shown before.

Specific Aims II: Demonstrate functional improvement of a sensory feedback‐enabled prosthetic.

Using a simple, closed-loop sensory feedback system with pressure and hand-opening

sensors mounted on the subjects’ own myoelectric prosthetic, we conducted a range of

functional tasks. In a standardized, activities of daily living task, we found sensory

feedback only improved performance in 1 of 2 subjects (Chapter 5) and is insufficient for

evaluating the effects of sensory feedback (Chapter 5 and Appendix A). We developed

several tests to evaluate the effect of sensory feedback in blinded tasks (Chapter 3, 5 and

Appendix A). In doing so, we found both subject demonstrated improved outcomes in

blinded object detection, object localization, reduction of failures (Chapter 5), and

delicate object manipulation (Chapter 3). While performing functional tasks with sensory

feedback, we found both subjects reported significantly improved embodiment of the

prosthetic hand (Chapter 5). Both subjects reported a drop in phantom limb pain

118 throughout the course of the study (Chapter 3). Accomplishing this specific aim demonstrates that sensory feedback, through our multi-contact, extra-neural interface, can improve function in terms of object manipulation as well as in terms of psychosocial benefit.

Future Investigative Priorities

Biomedical Engineering research presents a unique challenge: to investigate questions of basic understanding while applying the results to clinical-relevant applications.

Specifically, research presented in this report represents a balance between investigating the basic science of stimulating perception and its application of restoring sensation in amputees. Future work should continue to pursue and develop both paths of investigation.

The following list is the top three topics are highlighted here with additional detail in the

Appendix A.

Basic science:

 What is the underlying mechanism for natural perception from nerve stimulation?

Why does the patterned intensity waveform produce natural sensation?

 How can we better control the proprioceptive sense?

 How can we take full advantage of multi-channel, field-steering effects?

Application to sensory restoration:

 Develop a fully-implantable, take-home system.

 Optimize the prosthetic sensor inputs to sensory feedback stimulation control

scheme.

119

 Investigate the effect of embodiment and the reduction of phantom pain.

Applications

Level 1: Lower‐Limb Prosthetics

The benefit of sensory feedback stimulation is not limited to prosthetic users of the

upper-limb. Sensory feedback stimulation is readily applicable to lower-limb amputees.

In fact, the prevalence of major lower-limb amputations is approximately three times the

prevalence of upper-limb amputations (AFHSC 2012; Ziegler-Graham et al. 2008).

Many advanced micro-controlled prosthetics legs exist on the market, providing a range

of mobility options including running and stair-climbing. Still, users must rely on visual

feedback or directly feel the ground-impact of the prosthetic through the socket-stump

interface. The sensation is cross-locale and must be interpreted, which is unnatural,

similar to sensory substitution (Chapter 1). By providing sensory stimulation in the form

of a natural heel-strike pressure, the amputee may improve weight distribution symmetry

(Sabolich and Ortega 1994) , improve balance confidence (Miller, Speechley, and Deathe

2002), allowing adaption to uneven grounding (Fan et al. 2008) and reduce falls

(Kulkarni et al. 1996). Similar to what has been shown for the upper-limb, sensory feedback will likely increase embodiment and reduce phantom pain in the lower limb

(Ehrsson et al. 2008; Foell and Flor 2013; Marasco et al. 2011; P. Rossini et al. 2010).

Level 2: Alternate nerve stimulation medical therapies

The sinusoidal PW stimulation waveform represents a major advancement in in any

therapeutic electrical stimulation which results in tingling sensation. The most notable

120

application of this research is in stimulation for pain management. Current pain-relief

stimulation systems using Transcutaneous Electrical Stimulation (TENS) have an

unfortunate side-effect of tingling sensation as the electricity must pass through skin

receptors to reach target nerves or muscles. The stimulation waveform may provide the

same pain-relief while changing the unnatural tingling sensation until a natural, massage-

like pulsing sensation, allowing better rest for the patient. Other afferent stimulation

applications may also benefit from the unique waveform. Afferent stimulation is

currently being investigated as a method to improve bladder control in SCI(Bruns,

Bhadra, and Gustafson 2008, 2009). In deep brain stimulation (DBS), stimulation of the subthalamic nucleus can reduce tremor in Parkinson’s disease or inhibit epilepsy, however a side effect may include tingling sensation during stimulation. Patterned stimulation may target desirable fiber tracts while reducing unwanted recruitment.

Level 3: Commercial and non‐treatment uses in normal humans.

The method of sensory stimulation with nerve cuff electrodes may be used for non-

medical applications in normal, healthy humans. Given that the cuff electrode approach

has not caused any pain or perceived sensation outside of stimulation in both subjects of

our study, one could envision implanting these cuffs on normal people to give them

enhanced sensory feedback. A natural sense of touch feedback may improve the remote

operation of robotics, such as controlling the NASA-DARPA humanoid-limbed

Robonaut in space walks or provide surgeons with a delicate touch in telemedicine and/or minimally-invasive, DaVinci robotic surgery system. Military application may include controlling drone aircraft, allowing pilots to “feel” air currents over the wings. Besides

121

industrial application, consumers could have natural sensory feedback to enhance their

experience of virtual worlds, or simply allow the first sense of human touch across the internet. Video telephony applications, such as FaceTime, currently allow vision and sound communication. What if a third dimension of touch communication was possible?

A distance relationship between a deployed soldier and his family could be made more bearable with vision, sound and touch communication.

Beyond normal tactile feedback, it is simple to consider using feedback to expand the

limits of human sensory modalities and detection. Pioneering individuals have already

implanted magnets in their fingertips sub-dermally to detect magnetic field or metallic

objects. Consider sensors that detect magnetic fields, the location of Wi-Fi networks, or

radiation, which is translated into a natural touch sensation. A sense of synesthesia

could be experienced, allowing normal people to detect color, sound, smell, or light

translated as a touch sense.

122

Appendix A: Additional Observations

This chapter reports on data which has yet to be published, additional detail on data already presented in Chapters 3, 4, and 5, and observed cases which were not collected in detail but deserve mention for future work in the field. Data will be sectioned by Aims, with topic sections self-contained including a brief description of methods used and implications for future work.

In this section, subject 1 is referred to as S102 and subject 2 is referred to as S104.

Aim 1: Characterization of Sensation

Intensity modulation considerations

In Chapter 3, we demonstrated that frequency modulates the intensity of perceived sensation in S102. However, the frequency modulation’s relationship with intensity is also dependent on which specific contact that was tested. In S104, we conducted the contralateral pressure matching protocol (see Chapter 3) on contacts M1, M4, M6 and R8.

We placed shaped manipulators on the pressure sensor designed to replicate the shape of the perceived sensation when the subject presses on the sensor. We also requested the subject to provide an open-scale intensity rating for stimulus presentation. The results are shown in figure A1

123

Figure A1 Pressure matching traces from 4 channels of subject S104.

The following statistics were calculated using pressure-match data. Using a linear fit model, frequency was found to have a significant effect on intensity in M1, M6, and R8

(all p < 0.005) but a weak or poor effect on intensity in M4 (p = 0.071, R = 0.153).

Across all contacts, 20 Hz was significantly weaker than all oother frequencies (p < 0.008).

125 Hz was significantly stronger than the mid frequencies 50-70-100 Hz in only contact

M1 (p < 0.001). Interestingly, there appears to be a peak at 70 Hz on R8. Although it is not significantly different than 50Hz or 100 Hz, it is interesting that the psychometric curve follows the same trend. Across all contacts, the psychometric curve appears to follow the same trends as the pressure match curve.

124

The weak frequency relationship of M4 may be related to its unique perceptual response.

Only M4 produces a sense of Digit 3 flexion proprioception along with pressure along the

ulnar edge of Digit 3. If the proprioception sensation is related to motor activation,

which EMG recordings have indicated, then perhaps the frequency relationship levels out

at 30 Hz due to tetanic contraction.

The peak response at 70 Hz for R8 is unique, but not completely unexpected. We have observed that sometimes frequencies above 100 Hz lead to a drop in intensity, possibly

due to the refractory period of nerve firing. Future work may involve a collecting intensity scores across more points of frequency. The resulting frequency relationship

may also elucidate the underlying tactile afferents of the perceived sensation, by

comparison to frequency response characteristics of FA, SA and PC afferents (Bensmaia

2008; Johansson and Flanagan 2009; Johnson 2001; Vallbo and Johansson 1984).

Proprioception

Proprioception or kinesthetic sense is the sense of limb position and movement.

Proprioception is a complex sensation arising from both Ruffinian (SAII) stretch-receptor afferents and muscle (spindle primary endings) receptors. During muscle contraction, an additional cue is provided by centrally generated motor command signals(Proske and

Gandevia 2009). The nature of proprioception as a sensation elicited by a population of afferents may explain why proprioception is generally difficult to achieve solely through afferent nerve stimulation(Collins and Prochazka 1996; Hallin et al. 2002).

Proprioception has been reported by other sensory research teams; however, it is unclear if any were solely from afferent nerve stimulation without muscle contraction.

125

Proprioception of elbow flexion was described in 3 subject with LIFEs implanted in the upper-arm, median nerve specifically proximal to motor branching points (Dhillon and

Horch 2005). Digit tip flexion proprioception was also reported with LIFEs implanted in the upper-arm median nerve (Dhillon et al. 2004, 2005) with stimulation frequency correlated with joint flexion rate and static joint position. It is unclear how many of the 8 subjects reported proprioception. In a more recent study, LIFEs were implanted in the upper arm of 2 subjects, but only one reported proprioception from stimulation (Horch et al. 2011). In all reported studies resulting in proprioception, it may have been from activating muscle contraction, although it is unclear as they do not present EMG data from sensory stimulation.

In two studies, the TIME were implanted in the upper-arm of an amputee for 4 weeks, but no proprioception was reported (Raspopovic et al. 2014; P. M. P. Rossini et al. 2010).

Throughout this study, we have observed instances of proprioception which we have categorized as controlled, instantaneous, and volitional proprioception:

1. Controlled Proprioception

In S104, stimulation of contact M4 directly controls a perception of digit 3 flexion. The subject describes the sensation as a finger flexion and/or “tensing” of the palm to wrist area below digit 3. The subject was asked to contralaterally match the sensation while wearing the Cyberglove on the intact hand. The Cyberglove measures every joint angle of the fingers. A stimulus with the full-wave sinusoidal (1 Hz) PW waveform produced flexion of exactly 1 Hz. Two trials of 20 sec stimulation was applied and repeated for two PW levels (frequency did not appear to modulate the intensity of proprioception).

126

The mean and standard deviation was calculated for the 40 samples of peak flexion and peak-to-peak flexion for each PW level. The angle of the mid joint away from the palm is shown in figure A2 (left). Both the peak-to-peak flexion and peak flexion of the two

PW levels were significantly different (two-tailed t-test. p < 0.001), suggesting that PW can modulate the intensity of perceived oscillatory flexion. Additionally, we attempted a small-signal sinusoidal (1 Hz) stimulation, which the subject described as a constant flexion, also shown in figure A2 (right). This was explored at two PW levels demonstrating the PW can control a constant flexion perception as well (significant difference, two-tailed t-test. p < 0.001). Additional trials would strengthen this conclusion.

Figure A2 Finger flexion proprioception can be modulated with intensity. (Left) Full- wave patterned intensity stimulation produced modulated depth of percieved flexion.

(Right) Small-scale patterned intensity produced constant flexxion.

In S104, proprioception is elicited only from the cchannel which also elicits EMG response. Muscle twitch was observed at the instant proprioception was perceived and was verified through EMG recordings. It is unknown if proprioception is due to stimulating stretch or primary spindle fiber endings or from muscle recruitment. Our

127

current hypothesis is that all observed proprioception has been a result from motor activation. A future study with motor-block would definitively establish if the proprioception comes from muscle activation.

In S102, no controlled proprioception has been reported. No motor activation is expected since the implants are distal to motor branching points of the nerve, and stimulation of stretch receptors alone may not be enough to produce proprioceptive sensation (Collins

1996). Stretch receptors may also be stimulated in an uncoordinated population compared to what is require for controlled proprioception (Hallin and Wu 2002).

2. Instantaneous Proprioception

During tactile stimulation studies (Chapter 3), occasionally our subjects will volunteer

that they noticed a change in their hand or finger position. This change is instantaneous, without any sort of transitory movement, which the subject indicates with his contralateral hand. A common case we’ve observed is a change to the “OK” hand position with digit 1 and digit 2 touching and the rest of the digits more outstretched. This is usually associated with stimulation that results in a perceived field of the tips of digit 1

and/or digit 2 (for example, M3 and M5 in S102). Our hypothesis is that the

instantaneous change is caused by unconscious cortical processing: if no object is being

manipulated by the hand, then the sensation appearing at the fingertips must be from the

“OK” position. This hand position generally does not relax or change after stimulation is

off, so it is not quite controlled.

Another example of this effect is a proprioception of digit 2 and digit 3 straightened together when tactile sensation was elicited between them. Altering the stimulation

128 parameters, we were able to repeatedly elicit a slight separation of the two digits either up or down in reference to each other and return to a tight side-by-side position. We suspect recruitment of skin stretch receptor afferents may have played a role in this proprioception.

3. Volitional Proprioception

When stimulation is applied, volitional proprioception “unlocks” the frozen phantom hand so that the subject could volitionally attempt to move and feel his perceptual hand moving. In one case, stimulation on a specific channel allowed the subject to rotate his perceived thumb. When the stimulation was off, the perceived thumb was “frozen” in place, and the subject could not volitionally change its position. This was repeatable on the experimental day which this was tested. A quick test was conducted on all other channels of the median nerve, but no other channel presented this volitional proprioception effect. Our hypothesis is that either the stretch receptor afferents play a role in this perception, or tactile stimulation is providing a signal to the cortex that the amputated hand is intact and thus should be allowed to be considered movable.

Proprioception is difficult detect when experiments are focused on tactile perceptions.

Tactile sensations are much easier to detect as they represent new sensations whereas with proprioception, the subject always has a background sense of his phantom hand. All three proprioception types were found when the subject volunteered the perception.

Many times, this was noticed after reporting on elicited tactile sensation. It is impossible to prevent bias in this case as, in general, the subject needs to be directed to pay attention to any proprioception in order to detect it. In the case of volitional proprioception, the

129

subject needs to be directed to attempt moving the perceived fingers in order to see if the

effect exists.

The results noted here are very promising. Due to proprioception’s high utility in

providing grip opening/closing feedback of a prosthetic hand, it deserves continued

investigation.

Attenuation of Sensation

In normal humans, the sense of touch can habituate or attenuate if constantly or

frequently applied, such as the sensation of clothing on skin. Much like normal humans,

our subjects attenuates to the intensity of sensations from nerve stimulation if they are

constantly or frequently applied. This leads to difficulty in collecting intensity data or

conduct just noticeable difference (JND) experiments. A mitigating solution is to provide

extra “rest” time between stimulation trials to allow the system to “reset” to the sensation.

However, this limits the overall amount of data that can be collected due to time

constraints of clinical testing.

An experiment was conducted to characterize the attenuation rate. Continuous trains of

full-wave sinusoidal (1 Hz) PW stimulus at 100 Hz was provided to subject S102 for up to 120 sec or until stimulation was no longer detectable. Every 10 sec, the subject was

asked to report the current percent intensity given that the initial sensation was 100%.

Each trial was repeated three times with at least 1 minute rest period of no stimulation

between each trial. Stimulus was provided at or near threshold (within +10 us above

threshold). For method details, see Attenuation Protocol in appendix.

130

Results were similar across all trials for a single contact, average attenuation rate shown in figure A3. Using paired t-tests of response ratiing vs. time and a pair-wise comparison, all contacts were found to produce a significantly different attenuation pattern (p<=0.014).

Perceptually, every contact initially began with a sensation of either light tapping or tingling. Interestingly, every contact resolved into a natural sensation of tapping different than the sometimes initial sensation of tingling within 30-40 seconds. At the conclusion of one of the attenuation tests of M4, we increased PW by +5us and found that the perception increased from 10% to 30%. It is noteworthy that the subject reported the sensation was "as normal as can be" and it "felt like a finger pushing on a part of your skin." The data suggests that in the event of an initial tinglinng perception, 30-60 seconds of continuous, low-level stimulation may be used to condition the system into a natural sensation. The data also suggested that if sensation attenuation is inevitable, then perhaps a sensory feedback system should turn off long stimulations to reduce power consumption.

131

Figure A3. Attenuation rates over 120 seconds from S102.

Attenuation needs to be a key factor in future studies of sensory stimulation, especially if

it is to be used in a take-home, daily-use system. It may be possible to develop

stimulation waveforms which “reset” the perceptual system. In normal humans, sensory

system is inherently designed to respond to “changes” in the environment. The ideal

sensory feedback system may need to mimic this natural system to account for

habituation.

Multiple Location Detection

A demonstration was set-up where multiple sensors were mounted on the subject’s

prosthetic hand, each triggering a separate channel of stimulation on the median nerve. It

is known that the subject can detect all 4 channels of stimulation simultaneously. This

may be due to our stimulation system which cannot actually produce stimulation on

multiple channels at the same time; each channel is offset from another by at least 0.6 us.

It is unknown how many locations the subject can detect simultaneously with this setup.

Future work should determine this to maximize the number of sensing areas that can be

mounted on the subjects hand. Additionally, patterned activation of multiple channels

and percept areas may provide motion or slip sense to the subject.

Time Delay of Response

The current feedback system has a sampling delay of 50 us with an unknown processing and stimulation delay time, with a total delay of 50-100 us. Our subject has reported that he does not detect a time discrepancy between seeing/touching the pressure sensors mounted on his prosthetic hand and the stimulated sensation of the current feedback

system. It would be wise to determine the exact latency of our feedback system and the

132 upper bound of latency which the time delay would become noticeable and no-longer nature. This would be useful for the setting design parameters of the take-home sensory feedback system.

Field‐steering with multi‐channel stimulation

The presented work in this dissertation focused on sensory response from single-contact stimulation. Applying stimulation in more than one contacts should create directed

“fields” of current, which may activated areas of the nerve which were not accessible from single-contact stimulation. We typically used internal anodes (and thus field- steering) to ensure the stimulation current does not interfere with the EMG control in functional tasks. Generally, distributing the anodic return on three channels prevents interference. We attempted multi-channel stimulation and anodic return to test field- steering effect. Since the nerve stimulator cannot produce true simultaneous stimulation, we routed the a single stimulation channel into a switch board which allowed routing of the current to any combination of channels set as cathodes and anodes. We succeeded in recruiting additional unique locations in both subjects. This data is to be presented in a future paper.

Field-steering presents too many combinations (8! * 8! For a single channel) to exhaustively test and practically implement in a sensory feedback system, but it is something to consider in case single-contact stimulation does not recruit desired sensory locations.

133

Difficulties with Clinical work

Due to limited time with human subject availability, it is often not practical to conduct

rigorous, repeated trials of all characteristics of sensory stimulation. Although this

dissertation reports on a myriad of sensory stimulation characteristics with repeated

measure in a single experimental session, it is unknown how many of these

characteristics are repeatable over multiple sessions. Repeated measures over multiple

sessions were prioritized for thresholds, impedances, and perceptions, but many of the

detailed characteristics of perception were balanced in favor of continued investigation to produces a functional sensory feedback system.

Improving the stimulus waveform

Bursting is known to occur in various sensory systems and is thought to represent a

distinct mode of neural signaling (Bruns et al. 2008; Krahe and Gabbiani 2004). Based

on this principle, our first breakthrough was the discovery that stimulating with an initial

burst of 2 to 10 pulses at 250-500Hz followed by a slower, time-invariant pulse train,

resulted in a natural tapping sensation on the skin instead of paresthesia (Figure 2.C).

When we applied this “bursting” pattern to the median nerve, the subject felt a tapping sensation at the slower, time-invariant rate, ftap, in 45 of 48 trials (93%). The subject described the sensation as “running a comb” across the small area on the perceived hand.

For ftap between 2 and 20 Hz, each stimulus pulse manifested as a single tap, characteristic of a rapidly adapting (RA) receptor response (16). At 2 Hz, the subject

could synchronously tap a finger of the intact limb to illustrate the perceived tapping

frequency. The tapping sensation remained consistent during stimulation and did not

134 devolve into paresthesia that characterized the identical stimulation train without the initial bursting pattern. Hence, the initial bursts influence the perception of the subsequent stimulus train, suggesting that the nature of the perception is established by the initial pattern of activity and subsequent interpretation is based on that established expectation.

In later experiments, it was found that bursting caused initial taps to be of an increased intensity. In the research field of afferent stimulation for micturition, bursting of 2–10 pulses at 100–200 Hz repeated at continuous stimulation frequencies evoked significantly larger bladder responses than continuous (single pulse) stimulation (Bruns et al. 2008).

This effect may be attributed to the frequency modulation of intensity as described in

Chapter 3. Bursting has not been implemented in our sensory feedback in our reported functional tasks. However, a future application may be to revert the effect of attenuation

(see appendix?); for example, a burst may “wake-up” or “reset” the perception to a habituated sensation.

The Magic of 1 Hz ‐ Patterned Stimulation Intensity

Chapter 3 presented the sinusoidal (1 Hz) PW modulation as pattern stimulation intensity which may recruit axons at a variety of firing rates, simulating normal input from a variety of mechanoreceptor types. Natural, constant pressure sensation was achieved with a small-signal, sinusoidal pulse width. An alternate hypothesis is that natural sensation results from using a 1 Hz intensity modulation, which is similar to normal heart rate. Consider the intense throbbing sensation one might feel from a hammer

135

accidentally hitting one’s thumb and the subject’s response to full-wave 1Hz PW

modulation: “it feels like [measuring] my pulse”. Recent research has indicated that the

vascular sensory system may substitute for conscious tactile perception in individuals

without cutaneous sensory receptors (Bowsher et al. 2009). Unless we focus, we do not detect the underlying pulse of our blood during every day touch sensation. When we can feel our own pulse, it is likely primarily from cutaneous receptors. However receptors in cutaneous blood vessels may also play a role. In order to induce constant pressure, the unperceived small-signal sinusoid stimulation may provide the underlying pulsatile pressure needed to simulate a normal sensory system. It may have been fortunate that we selected an intensity modulation of 1 Hz, which is within the normal human heart rate of

60-100 BPM. To explore this further, future studies will examine the effect of adjusting the 1 Hz intensity modulation to the subjects measured heart rate and to values outside of normal human rates.

Aim 2: Functional Tests

Optimizing the Feedback scheme

A major effort in future work is to focus on optimizing the sensory feedback scheme used

during functional task experiments. No formal method to develop or test the feedback

scheme was used in this initial report since the major aim was to simply provide enough

feedback in order to demonstrate feasibility of impacting functional tasks.

136

The functional testing results presented in Chapter 5 were implemented using a very

simple sensory feedback control scheme. The PA and the PW (set to either small-signal or full wave sinusoidal (1 Hz)) were set while pressure sensor input mounted on the prosthetic hands controlled the frequency. However, we know not all channels are linearly mapped to frequency (see above). We have also observed that PW fairly

consistently produces an increase in intensity (see two figures, rating scale and the all contacts colored one. see STM paper). PW alone or some combination of PW and Freq may be better control inputs for distinguishing intensity. For detection of hand-opening, we have given the subject full-wave pulsing pressure feedback modulating the frequency.

We have observed that sometimes the subject has difficulty in distinguishing even 2 to 3 levels of intensity. It is sometimes difficult to determine even JND and the number of noticeable intensity levels simply because of the attenuation effect. A better scheme we have recently explored is to modulate the sinusoidal PW envelope frequency to scale from 1 Hz to 5 Hz, which the subject has reported is easier to detect and does not change with attenuation of intensity.

Another control scheme is to use a Proportional-Derivative (PD) controller for intensity than simply proportional (pressure-to-frequency) alone as is currently provided. We briefly explored this concept and found that the subject responded favorably to a P/D ratio of 0.2. The natural sensory system has RA and SA type fibers which have firing responses similar to derivative and proportional feedback, respectively (Hallin et al. 2002;

Johansson and Flanagan 2009; Johnson 2001). Peripheral nerve firing models such as those by Bensmaia et al, may provide insight on what firing patterns to attempt(Bensmaia

137

2008; Hollins and Bensmaïa 2007). Our earlier explorations with bursting patterns may

improve sensation and may be similar to a derivative control effect.

Functional tasks

The development and use of modified tests of function was because it was determined

that all standardized tests of function were inadequate to evaluate the effect of sensory

feedback. Standardized tests of function such as box blocks, SHAP, ACMC, are better

suited to measure improvements in prosthetic functional dexterity rather than the effect of

sensory feedback. The standard prosthetic hand used in experiment is bulky, making it impossible to operate in the limited space of the traditional Box & Block test and to pick

up blocks which have settled into a uniform plane. All standardized tests of function are

also sighted and most are timed. In this case the subject would rather rely on vision and

speed, which by probability alone would achieve a greater score than a new tactile

sensation. A common complaint among our amputee is that the bulkiness of their

myoelectric hand interferes visual feedback of object grasping. The griefer is sometimes

preferred because the mechanics restricts less of the object contact vision, even though

it’s not a real hand. In normal hands, our tactile senses compliment and sometimes

replace the need for vision. Therefore, blinded tasks are ideal for tests of sensory

feedback effects.

ACMC was conducted on both subjects, with and without sensation, except never

evaluated due to change in qualified personnel during the study. In future studies, we

would use the AM-ULA instead which can be scored via published guidelines instead of

trained personnel. The AM-ULA was developed by Linda Resnick to increase

quantification in functional performance evaluation (ref) .

138

In developing our own blinded functional tasks (paper 3) as part of Aim 2, we have

attempted a number of tasks which were not as successful and a set of guidelines for

functional tasks. A summary of those tasks is presented here for future reference.

List of activities which the subject has difficulty with (self-reported) or wishes he were capable of.

 dropping delicate dishes/china

 opening water bottles without spilling

 toothpaste, condiment squeeze bottles

 Red/Blue SOLO disposable plastic cups

 Banana

Derived list of functional testing characteristics

 Blinded tasks

 Attention-distraction

 Two-handed tasks, since any one handed tasks that required dexterous movement

or fine motor control could be completed with the intact hand.

 Delicate objects. Pressure is useful in handling delicate objects. With stiff/hard

objects, the prosthetic would be maximally closed for a secure grip without need

for precision control. Precision control is when a minimum force must be used to

secure the object while staying under the maximum force which would destroy

the object.

 Delicate objects which are also deformable. The subject’s control is position-

based grip opening/closing. Therefore, graded sensory feedback is not useful in

139

cases of delicate objects that cannot be deformed. If the subject had an advanced

myoelectic prosthesis with effort-based operation like human muscle, then

pressure-feedback can be applied to non-deformable objects as well.

Summary of attempted functional tasks:

Attention‐Distraction Task Subject has stated that he sometimes drops delicate plates. We hypothesisze that this is due to giving attention to family members while doing the task. The subject was directed to take plates, bowls (break-resistant) and plastic cups off a shelf and set up the dinner table. Then the subject was instructed to move all dishes to a side table (labeled ‘sink’).

The subject was instructed not to drop any dishes. At the same time the subject was to pay attention to a screen which presented the D-FES Color-Word interference test. The

Color-Word interference task shows words “red”, “blue”, and “green” but asks the subject to respond with the actual color of the word. It is scored for number of correct responses. In a second test, the subject was asked to move delicate cups from one end of the table to another without crushing the cups, while continually adding the numbers 3,

2, then 1 in that order to a running total. In both tasks, the subjects unexpectedly performed worse with sensory feedback enabled. We hypothesize that because the subject currently needs to pay attention to the sensation, less attention is spent on other tasks. Without a take-home system and continued sensory feedback, the subject cannot integrate sensation with his unconscious control which we would then expect improve attention-distraction task performance.

Delicate Manipulation – Cup Crush Task Graded pressure feedback should benefit the handling of deformable, delicate objects.

We presented the subject with a blinded task to grasp and raise disposable, plastic cup to

140

should height (to simulate drinking) and replace the cup while minimizing crushing or deformation. Two stiffness of cups were used: soft, opaque plastic cup and stiff-strong, clear plastic cup. In some cups, we added 8oz of weight to simulate a full cup. We found that the subject did not permanently deform any cup which was our definition of crush.

However, using the recorded video we analyzed the amount of deformation during the task and found sensory feedback allowed significantly less deformation during grasp than without sensory feedback in the stiff, clear cup cases or weighted cases (Mann-Whitney test, p ≤ 0.005). Report is in appendix. The results warrant repeating with a higher number of trials before publication.

Water Bottle Opening Task The subject was instructed to open a plastic water bottle without spilling while blinded.

The bottle is to be gripped with the prosthetic hand with enough force to prevent slip but

not so much force that the water will spill when the intact hand twists off the cap.

Weighing the bottles before and after opening allows us a quantitative measure of

performance. Anecdotally, we found sensory feedback reduced the amount of water

spilled, however, this task depends on a number of factors for success. Since this is a

power grip, sensors need to be mounted along the entirety of the finger instead of simply

the fingertip used in our published studies. Sensors (and all nearby stimulation

equipment) should also be water-proofed. Mounting sensors with tape reduces the grip

afforded by the grasp. A layer of duct tape on top of the sensors approximately restores

the grip of the subjects prosthetic. In fact, if high-friction tape such a Nextell was

wrapped around each finger, the subject could probably open bottles easily without

spilling. A protocol (see appendix) has been drafted for future studies.

141

Ketchup bottle squeeze task An exploratory demonstration of control with sensory feedback suggested more precise amounts of condiment was achievable with sensory feedback, even with vision. This suggested a full protocol could be explored as the weight of each squeezed amount could be used for a quantified measure. The subject commented that he felt more control with the feedback.

Phantom Pain Reduction Both Subjects self-reported a reduction in phantom pain of 85-95%. A Pain Survey from

Trinity Amputation and Prosthesis Experience Scales is shown in Chapter 3. It is unknown whether pain reduction was from providing previously missing tactile sensations or from the surgical procedure. However, as both subject have reported a significant reduction in phantom pain, this suggest phantom pain as an application of this work. Phantom pain is a common (51%) problem among amputees with 64% of those suffering experiencing moderate to intense phantom pain(Kooijman et al. 2000). Case studies have demonstrated a reduction in phantom pain through sensory stimulation

(Dietrich et al. 2012; P. M. P. Rossini et al. 2010).

Embodiment An adapted embodiment survey (Ehrsson et al. 2008; Ehrsson, Holmes, and Passingham

2005; Marasco et al. 2011) was given post function tasks of activities of daily living when sensory feedback was on and off. These surveys had 3 statements which signified embodiment and 7 statements which were control statements of no embodiment. The subject rated his agreement with each statement in a continuous scale from -3 of complete disagreement to + 3 of complete agreement.

142

Both subjects responded similarly and their results were pooled. Agreement ratings for

Embodiment statements were significantly greater than that of the control statement, suggesting greater embodiment of the prosthetic hand when sensory feedback was provided (Figure A.4).

Our hypothesis is that embodiment occurs with simultaneous visual and tactile feedback, similar to the effect of the RHI.

In additional to this embodiment survey, we propose that future work should examine embodiment with additional measures. Both subjects have indicated that their perceived

hand lines up or extends with/to the prosthetic hand more when sensory feedback is

enabled. A self-reported diagram would be useful to track positional embodiment. In

addition, Marasco has proposed utilizing temperature monitoring as more unbiased and

quantifiable measures of embodiment (Marasco et al. 2011).

Although researchers prioritize improved functional control which is easier to measure, embodiment is a highly important aspect of sensory feedback to amputees. In fact, embodiment will likely to be increased with agency (Dolezal 2009; Murray 2008; de

Vignemont 2011) giving more reason to examine both embodiment and functional testing.

Our subject has stated that when he uses the Greifer, he feels as if his hand is holding a split-hook tool. Whereas using the visually-realistic myoelectric hand with sensation, he feels as if that is his hand as the prosthesis. Even if a subject cannot attain improved control, a perceptually intact hand is a highly beneficial aspect of sensory feedback.

143

Figure A.4 Embodiment demonstrated with questionnaire taken after functional tasks.

Functional use of Pain

A small percentage of contacts reliably recruit uncomfortable, deep throbbing sensation as low stimulus levels with sinusoidal (1Hz) PW stimulation and uncomfortable pain at higher levels. Although our study purpose was to not recruit painful sensation, there is utility in providing pain feedback to prevent damage to the prosthetic hand. In fact, embodiment of the prosthetic may increase, when the user reacts to noxious stimuli by pulling his prosthetic hand away from the damagiing source. Pain will beecome an important and useful aspect of completely natural sensory restoration in future developments.

144

Subject Preference

It is important to note that both subjects often reported a preference for the electrical nerve stimulation, requesting take-home systems and frequently initiating the scheduling of the next experiment. Indeed, both subjects remain highly engaged and enthusiastic about taking part in the study even after nearly 60 combined visits.

145

References

AFHSC. 2012. “Amputations of Upper and Lower Extremities, Active and Reserve Components, U.S. Armed Forces, 2000-2011.” Medical Surveillance Monthly Report 19(6):2–6.

Agnew, W. F., D. B. McCreery, T. G. Yuen, and L. a Bullara. 1989. “Histologic and Physiologic Evaluation of Electrically Stimulated Peripheral Nerve: Considerations for the Selection of Parameters.” Annals of biomedical engineering 17(1):39–60. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/2537589).

Aitken, J. T., and P. K. Thomas. 1962. “Retrograde Changes in Fibre Size Following Nerve Section.” Journal of anatomy 96:121–29. Retrieved June 28, 2014 (http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1244178&tool=pmcentrez&rend ertype=abstract).

Anani, a B., K. Ikeda, and L. M. Körner. 1977. “Human Ability to Discriminate Various Parameters in Afferent Electrical Nerve Stimulation with Particular Reference to Prostheses Sensory Feedback.” Medical & biological engineering & computing 15(4):363–73.

Anani, A., and L. Körner. 1979. “Discrimination of Phantom Hand Sensations Elicited by Afferent Electrical Nerve Stimulation in below-Elbow Amputees.” Medical Progress Through Technology 6(3):131–35.

Antfolk, Christian et al. 2012. “Sensory Feedback from a Prosthetic Hand Based on Air-Mediated Pressure from the Hand to the Forearm Skin.” Journal of rehabilitation medicine : official journal of the UEMS European Board of Physical and Rehabilitation Medicine 44(8):702–7. Retrieved December 15, 2013 (http://www.ncbi.nlm.nih.gov/pubmed/22729800).

Atkins, D. J., D. C. Y. Heard, and W. H. Donovan. 1996. “Epidemiologic Overview of Individuals with Upper-Limb Loss and Their Reported Research Priorities.” J Prosthetics and Orthotics 8:2–11.

Augurelle, Anne-Sophie, Allan M. Smith, Thierry Lejeune, and Jean-Louis Thonnard. 2003. “Importance of Cutaneous Feedback in Maintaining a Secure Grip during Manipulation of Hand-Held Objects.” Journal of neurophysiology 89(2):665–71. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/12574444).

Badia, Jordi, Arán Pascual-Font, Meritxell Vivó, Esther Udina, and Xavier Navarro. 2010. “Topographical Distribution of Motor Fascicles in the Sciatic-Tibial Nerve of the Rat.” Muscle & nerve 42(2):192–201. Retrieved May 27, 2013 (http://www.ncbi.nlm.nih.gov/pubmed/20544926).

Bensmaia, Sliman J. 2008. “Tactile Intensity and Population Codes.” Behavioural brain research 190(2):165–73.

Berg, J. A. et al. 2013. “Behavioral Demonstration of a Somatosensory Neuroprosthesis.” IEEE transactions on neural systems and rehabilitation engineering : a publication of the IEEE Engineering in Medicine and Biology Society 21(3):500–507. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/23475375).

Biddiss, Elaine a, and Tom T. Chau. 2007. “Upper Limb Prosthesis Use and Abandonment: A Survey of the Last 25 Years.” Prosthetics and orthotics international 31(3):236–57. Retrieved October 5, 2012 (http://www.ncbi.nlm.nih.gov/pubmed/17979010).

146

Biddiss, Elaine, Dorcas Beaton, and Tom Chau. 2007. “Consumer Design Priorities for Upper Limb Prosthetics.” Disability & Rehabilitation: Assistive Technology 2(6):346–57. Retrieved December 13, 2013 (http://informahealthcare.com/doi/abs/10.1080/17483100701714733).

Biddiss, Elaine, and Tom Chau. 2007. “Upper-Limb Prosthetics: Critical Factors in Device Abandonment.” American journal of physical medicine & rehabilitation / Association of Academic Physiatrists 86(12):977–87. Retrieved October 5, 2012 (http://www.ncbi.nlm.nih.gov/pubmed/18090439).

Biran, Roy, David C. Martin, and Patrick a Tresco. 2005. “Neuronal Cell Loss Accompanies the Brain Tissue Response to Chronically Implanted Silicon Microelectrode Arrays.” Experimental neurology 195(1):115–26. Retrieved January 27, 2014 (http://www.ncbi.nlm.nih.gov/pubmed/16045910).

Bowsher, David et al. 2009. “Absence of Pain with Hyperhidrosis: A New Syndrome Where Vascular Afferents May Mediate Cutaneous Sensation.” Pain 147(1-3):287–98. Retrieved May 2, 2013 (http://www.ncbi.nlm.nih.gov/pubmed/19836135).

Brill, Natalie, and Dustin J. Tyler. 2014. “Optimization of High Density Nerve Cuff Stimulation in Upper Extremity Nerves.” Case Western Reserve University.

Bruns, Tim M., Narendra Bhadra, and Kenneth J. Gustafson. 2008. “Variable Patterned Pudendal Nerve Stimuli Improves Reflex Bladder Activation.” IEEE transactions on neural systems and rehabilitation engineering 16(2):140–48. Retrieved (http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3375815&tool=pmcentrez&rend ertype=abstract).

Bruns, Tim M., Narendra Bhadra, and Kenneth J. Gustafson. 2009. “Intraurethral Stimulation for Reflex Bladder Activation Depends on Stimulation Pattern and Location.” Neurourology and Urodynamics 28(March 2009):561–66.

Chan, Brenda L. et al. 2007. “Mirror Therapy for Phantom Limb Pain.” The New England journal of medicine 357(21):2206–7. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/22495728).

Childress, D. S. 1980. “Closed-Loop Control in Prosthetic Systems: Historical Perspective.” Annals of biomedical engineering 8(4-6):293–303. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/7027836).

Chu, Nai-shin. 1996. “Long-Term Effects of Finger Amputation on Stump Skin Sensibility and Digital Nerve Conduction.” Muscle & Nerve 19(8):1049–51.

Cipriani, Christian, Marco D’Alonzo, and Maria Chiara Carrozza. 2012. “A Miniature Vibrotactile Sensory Substitution Device for Multifingered Hand Prosthetics.” IEEE transactions on biomedical engineering 59(2):400–408. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/22042125).

Clippinger, F. W., R. Avery, and B. R. Titus. 1974. “A Sensory Feedback System for an Upper- Limb Amputation Prosthesis.” Bulletin of prosthetics research Fall:247–58. Retrieved March 6, 2012 (http://www.ncbi.nlm.nih.gov/pubmed/4462906).

Collinger, Jennifer L. et al. 2013. “High-Performance Neuroprosthetic Control by an Individual with Tetraplegia.” Lancet 381(9866):557–64. Retrieved January 26, 2014 (http://www.ncbi.nlm.nih.gov/pubmed/23253623).

147

Collins, D. F., and a Prochazka. 1996. “Movement Illusions Evoked by Ensemble Cutaneous Input from the Dorsum of the Human Hand.” The Journal of physiology 496 ( Pt 3:857–71. Retrieved (http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1160870&tool=pmcentrez&rend ertype=abstract).

Dhillon, G. S., and K. W. Horch. 2005. “Direct Neural Sensory Feedback and Control of a Prosthetic Arm.” IEEE Transactions on Neural Systems and Rehabilitation Engineering 13(4):468–72.

Dhillon, G. S., T. B. Krüger, J. S. Sandhu, and K. W. Horch. 2005. “Effects of Short-Term Training on Sensory and Motor Function in Severed Nerves of Long-Term Human Amputees.” Journal of neurophysiology 93(5):2625–33. Retrieved January 4, 2013 (http://www.ncbi.nlm.nih.gov/pubmed/15846000).

Dhillon, Gurpreet S., Stephen M. Lawrence, Douglas T. Hutchinson, and Kenneth W. Horch. 2004. “Residual Function in Peripheral Nerve Stumps of Amputees: Implications for Neural Control of Artificial Limbs.” The Journal of Hand Surgery 29(4):605–15; discussion 616–8. Retrieved November 5, 2012 (http://www.ncbi.nlm.nih.gov/pubmed/15249083).

Dietrich, Caroline et al. 2012. “Sensory Feedback Prosthesis Reduces Phantom Limb Pain: Proof of a Principle.” Neuroscience letters 507(2):97–100. Retrieved May 28, 2014 (http://www.sciencedirect.com/science/article/pii/S0304394011014807).

Dimitriou, Michael, and Benoni B. Edin. 2008. “Discharges in Human Muscle Spindle Afferents during a Key-Pressing Task.” The Journal of physiology 586(Pt 22):5455–70. Retrieved May 27, 2013 (http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2655390&tool=pmcentrez&rend ertype=abstract).

Dolezal, Luna. 2009. “The Remote Body: The Phenomenology of Telepresence and Re- Embodiment.” Human Technology 5(November):208–26. Retrieved March 2, 2012 (http://www.humantechnology.jyu.fi/articles/volume5/2009/dolezal.pdf).

Edin, B. B. et al. 2008. “Bio-Inspired Sensorization of a Biomechatronic Robot Hand for the Grasp-and-Lift Task.” Brain research bulletin 75(6):785–95. Retrieved May 28, 2014 (http://www.sciencedirect.com/science/article/pii/S0361923008000166).

Ehrsson, H. Henrik et al. 2008. “Upper Limb Amputees Can Be Induced to Experience a Rubber Hand as Their Own.” Brain : a journal of neurology 131(Pt 12):3443–52. Retrieved June 15, 2011 (http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2639202&tool=pmcentrez&rend ertype=abstract).

Ehrsson, H. Henrik, Nicholas P. Holmes, and Richard E. Passingham. 2005. “Touching a Rubber Hand: Feeling of Body Ownership Is Associated with Activity in Multisensory Brain Areas.” The Journal of neuroscience : the official journal of the Society for Neuroscience 25(45):10564–73. Retrieved July 7, 2011 (http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1395356&tool=pmcentrez&rend ertype=abstract).

Fan, Richard E. et al. 2008. “A Haptic Feedback System for Lower-Limb Prostheses.” IEEE transactions on neural systems and rehabilitation engineering : a publication of the IEEE

148

Engineering in Medicine and Biology Society 16(3):270–77. Retrieved June 20, 2014 (http://ieeexplore.ieee.org/articleDetails.jsp?arnumber=4463638).

Fisher, LE, DJ Tyler, JS Anderson, and RJ Triolo. 2009. “Chronic Stability and Selectivity of Four- Contact Spiral Nerve-Cuff Electrodes in Stimulating the Human Femoral Nerve.” Journal of neural engineering 6(4). Retrieved May 7, 2013 (http://iopscience.iop.org/1741- 2552/6/4/046010).

Flor, Herta, Claudia Denke, Michael Schaefer, and Sabine Grüsser. 2001. “Effect of Sensory Discrimination Training on Cortical Reorganisation and Phantom Limb Pain.” Lancet 357(9270):1763–64.

Foell, Jens, and Herta Flor. 2013. “Phantom Limb Pain.” Pp. 417–30 in Handbook of Pain and Palliative Care, edited by Rhonda J. Moore. New York, NY: Springer New York. Retrieved April 1, 2014 (http://link.springer.com/10.1007/978-1-4419-1651-8).

Gescheider, George A. 1997. Psychophysics : The Fundamentals. Mahwah, N.J.: L. Erlbaum Associates.

Grinberg, Yanina, Matthew a Schiefer, Dustin J. Tyler, and Kenneth J. Gustafson. 2008. “Fascicular Perineurium Thickness, Size, and Position Affect Model Predictions of Neural Excitation.” IEEE transactions on neural systems and rehabilitation engineering 16(6):572– 81. Retrieved (http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2918421&tool=pmcentrez&rend ertype=abstract).

Hallin, R. G., T. Carlstedt, and G. Wu. 2002. “Population Behaviour of Human Cutaneous Mechanoreceptive Units.” Behavioural brain research 135(1-2):19–26.

Hallin, R. G., and G. Wu. 2002. “Novel Information on Peripheral Tactile Mechanisms in Man Acquired with Concentric Needle Electrode Microneurography.” Behavioural brain research 135(1-2):11–18. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/12356428).

Heming, Ethan a, Ryan Choo, Jonathan N. Davies, and Zelma H. T. Kiss. 2011. “Designing a Thalamic Somatosensory Neural Prosthesis: Consistency and Persistence of Percepts Evoked by Electrical Stimulation.” IEEE transactions on neural systems and rehabilitation engineering : a publication of the IEEE Engineering in Medicine and Biology Society 19(5):477–82. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/21622082).

Heming, Ethan, Andrew Sanden, and Zelma H. T. Kiss. 2010. “Designing a Somatosensory Neural Prosthesis: Percepts Evoked by Different Patterns of Thalamic Stimulation.” Journal of neural engineering 7(6):064001. Retrieved February 4, 2014 (http://www.ncbi.nlm.nih.gov/pubmed/21084731).

Hollins, Mark, and Sliman J. Bensmaïa. 2007. “The Coding of Roughness.” Canadian Journal of Experimental Psychology/Revue canadienne de psychologie expérimentale 61(3):184–95.

Horch, Kenneth, Sanford Meek, Tyson G. Taylor, and Douglas T. Hutchinson. 2011. “Object Discrimination with an Artificial Hand Using Electrical Stimulation of Peripheral Tactile and Proprioceptive Pathways with Intrafascicular Electrodes.” IEEE transactions on neural systems and rehabilitation engineering : a publication of the IEEE Engineering in Medicine and Biology Society 19(5):483–89. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/21859607).

149

Hunter, J. P., Joel Katz, and Karen D. Davis. 2003. “The Effect of Tactile and Visual Sensory Inputs on Phantom Limb Awareness.” Brain 126(3):579–89. Retrieved February 21, 2014 (http://www.brain.oupjournals.org/cgi/doi/10.1093/brain/awg054).

Jenkins, David B. 2001. Hollinshead’s Functional Anatomy of the Limbs and Back. 8th ed. Philadelphia: Saunders.

Jenmalm, P., and R. S. Johansson. 1997. “Visual and Somatosensory Information about Object Shape Control Manipulative Fingertip Forces.” The Journal of neuroscience : the official journal of the Society for Neuroscience 17(11):4486–99. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/9151765).

Johansson, R. S., C. Häger, and R. Riso. 1992. “Somatosensory Control of Precision Grip during Unpredictable Pulling Loads. II. Changes in Load Force Rate.” Experimental brain research 89(1):192–203. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/1601097).

Johansson, Roland S., and J. Randall Flanagan. 2009. “Coding and Use of Tactile Signals from the Fingertips in Object Manipulation Tasks.” Nature reviews. Neuroscience 10(5):345–59. Retrieved March 20, 2014 (http://www.ncbi.nlm.nih.gov/pubmed/19352402).

Johnson, K. O. 2001. “The Roles and Functions of Cutaneous Mechanoreceptors.” Current opinion in neurobiology 11(4):455–61. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/11502392).

Kaernbach, Christian. 1990. “A Single Interval Adjustment Matrix (SIAM) Procedure for Unbiased Adaptive Testing.” Journal of the Acoustical Society of America 88(6):2645–55.

Knibestol, B. Y. M., and A. B. Vallbo. 1980. “Intensity of Sensation Related to Activity of Slowly Adapting Mechanoreceptive Units in the Human Hand.” The Journal of physiology (300):251–67.

Knutson, Jayme S., Gregory G. Naples, P. Hunter Peckham, and Michael W. Keith. 2002. “Electrode Fracture Rates and Occurrences of Infection and Granuloma Associated with Percutaneous Intramuscular Electrodes in Upper-Limb Functional Electrical Stimulation Applications.” Journal of rehabilitation research and development 39(6):671–83. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/17943669).

Kooijman, C. M., P. U. Dijkstra, J. H. Geertzen, a Elzinga, and C. P. van der Schans. 2000. “Phantom Pain and Phantom Sensations in Upper Limb Amputees: An Epidemiological Study.” Pain 87(1):33–41. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/10863043).

Krahe, Rüdiger, and Fabrizio Gabbiani. 2004. “Burst Firing in Sensory Systems.” Nature reviews. Neuroscience 5(1):13–23. Retrieved March 5, 2012 (http://www.ncbi.nlm.nih.gov/pubmed/14661065).

Kuiken, Todd a et al. 2007. “Targeted Reinnervation for Enhanced Prosthetic Arm Function in a Woman with a Proximal Amputation: A Case Study.” Lancet 369(9559):371–80. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/17276777).

Kulkarni, J., S. Wright, C. Toole, J. Morris, and R. Hirons. 1996. “Falls in Patients with Lower Limb Amputations: Prevalence and Contributing Factors.” Physiotherapy 82(2):130–36. Retrieved May 23, 2014 (http://www.sciencedirect.com/science/article/pii/S0031940605669684).

150

Kyberd, Peter J. 2011. “The Influence of Control Format and Hand Design in Single Axis Myoelectric Hands: Assessment of Functionality of Prosthetic Hands Using the Southampton Hand Assessment Procedure.” Prosthetics and orthotics international 35(3):285–93. Retrieved August 22, 2013 (http://www.ncbi.nlm.nih.gov/pubmed/21937574).

Letechipia, J. E., P. H. Peckham, M. Gazdik, and B. Smith. 1991. “In-Line Lead Connector for Use with Implanted Neuroprosthesis.” IEEE transactions on bio-medical engineering 38(7):707–9. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/1879864).

Leventhal, Daniel K., Mark Cohen, and Dominique M. Durand. 2006. “Chronic Histological Effects of the Flat Interface Nerve Electrode.” Journal of neural engineering 3(2):102–13. Retrieved June 6, 2014 (http://iopscience.iop.org/1741-2552/3/2/004).

Leventhal, Daniel K., and Dominique M. Durand. 2003. “Subfascicle Stimulation Selectivity with the Flat Interface Nerve Electrode.” Annals of Biomedical Engineering 31(6):643–52. Retrieved May 23, 2014 (http://link.springer.com/10.1114/1.1569266).

Leventhal, Daniel K., and Dominique M. Durand. 2004. “Chronic Measurement of the Stimulation Selectivity of the Flat Interface Nerve Electrode.” IEEE transactions on bio-medical engineering 51(9):1649–58. Retrieved June 6, 2014 (http://www.ncbi.nlm.nih.gov/pubmed/15376513).

Light, CM, PH Chappell, B. Hudgins, and K. Engelhart. 2002. “Intelligent Multifunction Myoelectric Control of Hand Prostheses.” Journal of Medical Engineering & Technology 26(4):139–46.

Light, Colin M., Paul H. Chappell, and Peter J. Kyberd. 2002. “Establishing a Standardized Clinical Assessment Tool of Pathologic and Prosthetic Hand Function: Normative Data, Reliability, and Validity.” Archives of Physical Medicine and Rehabilitation 83(6):776–83. Retrieved August 23, 2013 (http://linkinghub.elsevier.com/retrieve/pii/S0003999302095795).

Makin, Joseph G., Matthew R. Fellows, and Philip N. Sabes. 2013. “Learning Multisensory Integration and Coordinate Transformation via Density Estimation.” PLoS computational biology 9(4):e1003035. Retrieved May 27, 2013 (http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3630212&tool=pmcentrez&rend ertype=abstract).

Marasco, Paul D., Keehoon Kim, James Edward Colgate, Michael A. Peshkin, and Todd A. Kuiken. 2011. “Robotic Touch Shifts Perception of Embodiment to a Prosthesis in Targeted Reinnervation Amputees.” Brain : a journal of neurology 134(Pt 3):747–58. Retrieved March 6, 2012 (http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3044830&tool=pmcentrez&rend ertype=abstract).

Marasco, Paul D., Aimee E. Schultz, and Todd a Kuiken. 2009. “Sensory Capacity of Reinnervated Skin after Redirection of Amputated Upper Limb Nerves to the Chest.” Brain : a journal of neurology 132(Pt 6):1441–48. Retrieved June 10, 2013 (http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2685921&tool=pmcentrez&rend ertype=abstract).

McComas, A. J., R. E. Sica, and S. Banerjee. 1978. “Long-Term Effects of Partial Limb Amputation in Man.” Journal of Neurology, Neurosurgery & Psychiatry 41(5):425–32. Retrieved June 28, 2014 (http://jnnp.bmj.com/content/41/5/425.short).

151

McConnell, George C. et al. 2009. “Implanted Neural Electrodes Cause Chronic, Local Inflammation That Is Correlated with Local Neurodegeneration.” Journal of neural engineering 6(5):056003. Retrieved March 3, 2014 (http://www.ncbi.nlm.nih.gov/pubmed/19700815).

McFarland, Lynne V, Sandra L. Hubbard Winkler, Allen W. Heinemann, Melissa Jones, and Alberto Esquenazi. 2010. “Unilateral Upper-Limb Loss: Satisfaction and Prosthetic-Device Use in Veterans and Servicemembers from Vietnam and OIF/OEF Conflicts.” Journal of rehabilitation research and development 47(4):299–316. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/20803400).

Miller, William C., Mark Speechley, and A. Barry Deathe. 2002. “Balance Confidence Among People With Lower-Limb Amputations.” Physical Therapy 82(9):856–65. Retrieved June 20, 2014 (http://ptjournal.apta.org/content/82/9/856.short).

Mogyoros, I., H. Bostock, and D. Burke. 2000. “Mechanisms of Paresthesias Arising from Healthy Axons.” Muscle & nerve 23(3):310–20. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/10679707).

Monzée, Joël, Yves Lamarre, and Allan M. Smith. 2003. “The Effects of Digital Anesthesia on Force Control Using a Precision Grip.” Journal of neurophysiology 89(2):672–83. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/12574445).

Murray, Craig D. 2008. “Embodiment and Prosthetics.” Pp. 119–29 in Psychoprosthetics, edited by Pamela Gallagher, Deirdre Desmond, and Malcolm MacLachlan. London: Springer.

Naples, G. G., J. T. Mortimer, A. Scheiner, and J. D. Sweeney. 1988. “A Spiral Nerve Cuff Electrode for Peripheral Nerve Stimulation.” IEEE Transactions on Biomedical Engineering 35(11):905–16.

Ochoa, J., and E. Torebjörk. 1983. “Sensations Evoked by Intraneural Microstimulation of Single Mechanoreceptor Units Innervating the Human Hand.” The Journal of physiology 342:633– 54. Retrieved (http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1193981&tool=pmcentrez&rend ertype=abstract).

Ochoa, Jose L., and H. Erik Torebjork. 1980. “Paraesthesiae from Ectopic Impulse Generation in Human Sensory Nerves.” Brain 103(4):835–53.

Panarese, Alessandro, Benoni B. Edin, Fabrizio Vecchi, Maria C. Carrozza, and Roland S. Johansson. 2009. “Humans Can Integrate Force Feedback to Toes in Their Sensorimotor Control of a Robotic Hand.” IEEE transactions on neural systems and rehabilitation engineering 17:560–67.

Polasek, Katharine H., Harry a Hoyen, Michael W. Keith, and Dustin J. Tyler. 2007. “Human Nerve Stimulation Thresholds and Selectivity Using a Multi-Contact Nerve Cuff Electrode.” IEEE transactions on neural systems and rehabilitation engineering 15(1):76–82. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/17436879).

Polasek, Katharine H., Harry A. Hoyen, Michael W. Keith, Robert F. Kirsch, and Dustin J. Tyler. 2009. “Stimulation Stability and Selectivity of Chronically Implanted Multicontact Nerve Cuff Electrodes in the Human Upper Extremity.” IEEE transactions on neural systems and rehabilitation engineering 17(5):428–37. Retrieved

152

(http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2927980&tool=pmcentrez&rend ertype=abstract).

Polasek, KH, and MA Schiefer. 2009. “Intraoperative Evaluation of the Spiral Nerve Cuff Electrode on the Femoral Nerve Trunk.” Journal of neural engineering 6(6):1–12. Retrieved May 7, 2013 (http://iopscience.iop.org/1741-2552/6/6/066005).

Polikov, Vadim S., Patrick a Tresco, and William M. Reichert. 2005. “Response of Brain Tissue to Chronically Implanted Neural Electrodes.” Journal of neuroscience methods 148(1):1–18. Retrieved January 22, 2014 (http://www.ncbi.nlm.nih.gov/pubmed/16198003).

Prodanov, Dimiter, Nico Nagelkerke, and Enrico Marani. 2007. “Spatial Clustering Analysis in Neuroanatomy: Applications of Different Approaches to Motor Nerve Fiber Distribution.” Journal of neuroscience methods 160(1):93–108. Retrieved May 27, 2013 (http://www.ncbi.nlm.nih.gov/pubmed/17049615).

Proske, Uwe, and Simon C. Gandevia. 2009. “The Kinaesthetic Senses.” The Journal of physiology 587(Pt 17):4139–46. Retrieved February 21, 2014 (http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2754351&tool=pmcentrez&rend ertype=abstract).

Pylatiuk, Christian, Artem Kargov, and Stefan Schulz. 2006. “Design and Evaluation of a Low- Cost Force Feedback System for Myoelectric Prosthetic Hands.” Journal of Prosthetics and Orthotics 18(2):57–61. Retrieved (http://content.wkhealth.com/linkback/openurl?sid=WKPTLP:landingpage&an=00008526- 200604000-00007).

Pylatiuk, Christian, Stefan Schulz, and Leonhard Döderlein. 2007. “Results of an Internet Survey of Myoelectric Prosthetic Hand Users.” Prosthetics and orthotics international 31(4):362–70. Retrieved December 20, 2013 (http://www.ncbi.nlm.nih.gov/pubmed/18050007).

Ramachandran, V. S., and D. Rogers-Ramachandran. 1996. “Synaesthesia in Phantom Limbs Induced with Mirrors.” Proceedings of the Royal Society B: Biological Sciences 263(1369):377–86. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/8637922).

Raspopovic, Stanisa et al. 2014. “Restoring Natural Sensory Feedback in Real-Time Bidirectional Hand Prostheses.” Science translational medicine 6(222):222ra19. Retrieved March 20, 2014 (http://www.ncbi.nlm.nih.gov/pubmed/24500407).

Resnik, Linda et al. 2012. “Advanced Upper Limb Prosthetic Devices: Implications for Upper Limb Prosthetic Rehabilitation.” Archives of physical medicine and rehabilitation 93(4):710–17. Retrieved August 19, 2013 (http://www.ncbi.nlm.nih.gov/pubmed/22464092).

Riso, R. R. 1999. “Strategies for Providing Upper Extremity Amputees with Tactile and Hand Position Feedback--Moving Closer to the Bionic Arm.” Technology and Health Care 7(6):401–9. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/10665673).

Robles-De-La-Torre, G. 2006. “The Importance of the Sense of Touch in Virtual and Real Environments.” IEEE Multimedia 13(3):24–30. Retrieved (http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=1667971).

Romo, R., A. Hernández, A. Zainos, C. D. Brody, and L. Lemus. 2000. “Sensing without Touching: Psychophysical Performance Based on Cortical Microstimulation.” Neuron 26:273–78.

153

Rossini, Paolo M. PM et al. 2010. “Double Nerve Intraneural Interface Implant on a Human Amputee for Robotic Hand Control.” Clinical neurophysiology 121(5):777–83. Retrieved January 31, 2013 (http://www.ncbi.nlm.nih.gov/pubmed/20110193).

Rossini, PM, Silvestro Micera, and Antonella Benvenuto. 2010. “Double Nerve Intraneural Interface Implant on a Human Amputee for Robotic Hand Control.” Clinical Neurophysiology 121(5):777–83.

Sabolich, John A., and Giovani M. Ortega. 1994. “Sense of Feel for Lower-Limb Amputees: A Phase-One Study.” Journal of Prosthetics & Orthotics Spring. Retrieved June 20, 2014 (http://journals.lww.com/jpojournal/Abstract/1994/00620/Sense_of_Feel_for_Lower_Limb_A mputees__A_Phase_One.3.aspx).

Schiefer, Matthew A. et al. 2013. “Selective Activation of the Human Tibial and Common Peroneal Nerves with a Flat Interface Nerve Electrode.” Journal of neural engineering 10(5):056006. Retrieved January 9, 2014 (http://www.ncbi.nlm.nih.gov/pubmed/23918148).

Schiefer, Matthew A., Katharine H. Polasek, R. J. Triolo, G. C. J. Pinault, and D. J. Tyler. 2010. “Selective Stimulation of the Human Femoral Nerve with a Flat Interface Nerve Electrode.” Journal of neural engineering 7(2):26006. Retrieved April 2, 2013 (http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2915830&tool=pmcentrez&rend ertype=abstract).

Schiefer, Matthew A., Ronald J. Triolo, and Dustin J. Tyler. 2008. “A Model of Selective Activation of the Femoral Nerve with a Flat Interface Nerve Electrode for a Lower Extremity Neuroprosthesis.” IEEE transactions on neural systems and rehabilitation engineering 16(2):195–204. Retrieved (http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2920206&tool=pmcentrez&rend ertype=abstract).

Schiefer, Matthew A., Dustin J. Tyler, and Ronald J. Triolo. 2012. “Probabilistic Modeling of Selective Stimulation of the Human Sciatic Nerve with a Flat Interface Nerve Electrode.” Journal of computational neuroscience 33(1):179–90. Retrieved January 9, 2014 (http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3357453&tool=pmcentrez&rend ertype=abstract).

Scott, R. N., R. H. Brittain, R. R. Caldwell, A. B. Cameron, and V. A. Dunfield. 1980. “Sensory- Feedback System Compatible with Myoelectric Control.” Medical & biological engineering & computing 18:65–69.

Shannon, G. F. 1979. “A Myoelectrically-Controlled Prosthesis with Sensory Feedback.” Medical & biological engineering & computing 17:73–80.

Simeral, J. D., S. P. Kim, M. J. Black, J. P. Donoghue, and L. R. Hochberg. 2011. “Neural Control of Cursor Trajectory and Click by a Human with Tetraplegia 1000 Days after Implant of an Intracortical Microelectrode Array.” Journal of neural engineering 8:025027.

Stepp, Cara E., and Yoky Matsuoka. 2012. “Vibrotactile Sensory Substitution for Object Manipulation: Amplitude versus Pulse Train Frequency Modulation.” IEEE transactions on neural systems and rehabilitation engineering : a publication of the IEEE Engineering in Medicine and Biology Society 20(1):31–37. Retrieved (http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3395369&tool=pmcentrez&rend ertype=abstract).

154

Sunderland, S. 1978. Nerves and Nerve Injuries. 2nd ed. Churchill Livingstone.

Sunderland, Sydney. 1945. “The Intraneural Topography of the Radial, Median and Ulnar Nerves.” Brain 68(4):243–98. Retrieved June 14, 2014 (http://brain.oxfordjournals.org/content/68/4/243.extract).

Szeto, A. Y., and F. A. Saunders. 1982. “Electrocutaneous Stimulation for Sensory Communication in Rehabilitation Engineering.” IEEE Transactions on Biomedical Engineering 29(4):300–308.

Tan, Daniel et al. 2014. “Neural Interface Provides Stable, Natural, Touch Perception to Prosthetic Hand Users for More than One Year.” Science Translational Medicine accepted.

Tarler, Matthew D., and J. Thomas Mortimer. 2004. “Selective and Independent Activation of Four Motor Fascicles Using a Four Contact Nerve-Cuff Electrode.” IEEE transactions on neural systems and rehabilitation engineering 12(2):251–57. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/15218938).

Tarver, W. B., R. E. George, S. E. Maschino, L. K. Holder, and J. F. Wernicke. 1992. “Clinical Experience with a Helical Bipolar Stimulating Lead.” Pacing and clinical electrophysiology 15(10):1545–56. Retrieved (http://www.scopus.com/inward/record.url?eid=2-s2.0- 0026666252&partnerID=40&md5=6f8a27ace620a4efeb8e2e95e84d8dbe).

Tsakiris, Manos, and Patrick Haggard. 2005. “The Rubber Hand Illusion Revisited: Visuotactile Integration and Self-Attribution.” Journal of experimental psychology. Human perception and performance 31(1):80–91. Retrieved July 24, 2011 (http://www.ncbi.nlm.nih.gov/pubmed/15709864).

Tyler, DJ J., and DM M. Durand. 2002. “Functionally Selective Peripheral Nerve Stimulation with a Flat Interface Nerve Electrode.” Neural Systems and Rehabilitation … 10(4):294–303.

Tyler, Dustin J. 2013. “Restoring Natural Human Sensation in Amputees Video.” YouTube. Retrieved October 30, 2013 (http://www.youtube.com/watch?v=DIodb8qM9N0).

Tyler, Dustin J., and Dominique M. Durand. 2003. “Chronic Response of the Rat Sciatic Nerve to the Flat Interface Nerve Electrode.” Annals of Biomedical Engineering 31(6):633–42. Retrieved April 29, 2013 (http://link.springer.com/10.1114/1.1569263).

Vallbo, A. B., and R. S. Johansson. 1984. “Properties of Cutaneous Mechanoreceptors in the Human Hand Related to Touch Sensation.” Human neurobiology 3(1):3–14. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/6330008).

Vallbo, A. B., K. A. Olsson, K. G. Westberg, and F. J. Clark. 1984. “Microstimulation of Single Tactile Afferents from the Human Hand. Sensory Attributes Related to Unit Type and Properties of Receptive Fields.” Brain : a journal of neurology 107 ( Pt 3:727–49.

De Vignemont, Frédérique. 2011. “Embodiment, Ownership and Disownership.” Consciousness and cognition 20(1):82–93. Retrieved July 27, 2011 (http://www.ncbi.nlm.nih.gov/pubmed/20943417).

Warwick, Kevin et al. 2003. “The Application of Implant Technology for Cybernetic Systems.” Archives of neurology 60(10):1369–73. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/14568806).

155

Watchmaker, Greg P., Cesar a. Gumucio, R. Evan Crandall, Michael a. Vannier, and Paul M. Weeks. 1991. “Fascicular Topography of the Median Nerve: A Computer Based Study to Identify Branching Patterns.” The Journal of Hand Surgery 16(1):53–59. Retrieved May 27, 2013 (http://linkinghub.elsevier.com/retrieve/pii/S0363502310800139).

Wettels, Nicholas, Veronica J. Santos, Roland S. Johansson, and Gerald E. Loeb. 2008. “Biomimetic Tactile Sensor Array.” Advanced Robotics 22(8):829–49. Retrieved June 1, 2014 (http://dx.doi.org/10.1163/156855308X314533).

Witteveen, H., E. Droog, J. Rietman, and P. Veltink. 2012. “Vibro- and Electrotactile User Feedback on Hand Opening for Myoelectric Forearm Prostheses.” IEEE transactions on biomedical engineering 59(8):2219–26. Retrieved (http://www.ncbi.nlm.nih.gov/pubmed/22645262).

Witteveen, Heidi Jb, Hans S. Rietman, and Peter H. Veltink. 2014. “Vibrotactile Grasping Force and Hand Aperture Feedback for Myoelectric Forearm Prosthesis Users.” Prosthetics and orthotics international. Retrieved May 2, 2014 (http://www.ncbi.nlm.nih.gov/pubmed/24567348).

Ziegler-Graham, Kathryn, Ellen J. MacKenzie, Patti L. Ephraim, Thomas G. Travison, and Ron Brookmeyer. 2008. “Estimating the Prevalence of Limb Loss in the United States: 2005 to 2050.” Archives of physical medicine and rehabilitation 89(3):422–29. Retrieved May 21, 2013 (http://www.ncbi.nlm.nih.gov/pubmed/18295618).

156