Artigo Fes

9 Jun 2005 21:10 AR AR248-BE07-11.
tex XMLPublishSM (2004/02/24) P1: KUV

10.1146/annurev.bioeng.6.040803.140103
Annu. Rev. Biomed. Eng. 2005. 7:327–60

doi: 10.1146/annurev.bioeng.6.040803.140103
Copyright c 2005 by Annual Reviews. All rights reserved
First published online as a Review in Advance on March 23, 2005
FUNCTIONAL ELECTRICAL STIMULATION

∗
FOR NEUROMUSCULAR APPLICATIONS
P. Hunter Peckham
Department of Biomedical Engineering, Case Western Reserve University, Cleveland,
Ohio 44106; Department of Veterans Affairs, Cleveland, Ohio 44106; MetroHealth
Annu. Rev. Biomed. Eng. 2005.7:327-360. Downloaded from www.annualreviews.org
Medical Center, Cleveland, Ohio 44109; email: pxp2@case.edu
Jayme S. Knutson
by University of Edinburgh on 06/28/12. For personal use only.
Department of Veterans Affairs, Cleveland, Ohio 44106; email: jsk12@case.edu
Key Words motor neuroprosthesis, spinal cord injury, stroke rehabilitation,

bladder stimulator, phrenic nerve pacing
■ Abstract Paralyzed or paretic muscles can be made to contract by applying elec-
trical currents to the intact peripheral motor nerves innervating them. When electrically
elicited muscle contractions are coordinated in a manner that provides function, the
technique is termed functional electrical stimulation (FES). In more than 40 years
of FES research, principles for safe stimulation of neuromuscular tissue have been
established, and methods for modulating the strength of electrically induced muscle
contractions have been discovered. FES systems have been developed for restoring
function in the upper extremity, lower extremity, bladder and bowel, and respiratory
system. Some of these neuroprostheses have become commercialized products, and
others are available in clinical research settings. Technological developments are ex-
pected to produce new systems that have no external components, are expandable to
multiple applications, are upgradable to new advances, and are controlled by a combi-
nation of signals, including biopotential signals from nerve, muscle, and the brain.
CONTENTS
INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
PHYSIOLOGICAL AND TECHNOLOGICAL PRINCIPLES OF FES . . . . . . . . . . 328
Electrical Activation of the Neuromuscular System . . . . . . . . . . . . . . . . . . . . . . . . 328
Configurations of Neuroprosthetic Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
CLINICAL APPLICATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333
Upper Extremity Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333
Lower Extremity Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
Bladder and Bowel Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 341
Respiratory Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 344
∗
The U.S. Government has the right to retain a nonexclusive, royalty-free license in and to
any copyright covering this paper.
1523-9829/05/0815-0327$20.00 327
9 Jun 2005 21:10 AR AR248-BE07-11.tex XMLPublishSM (2004/02/24) P1: KUV
328 PECKHAM KNUTSON
FUTURE DIRECTIONS AND DEVELOPMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . 347

Emerging Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 348
Emerging Control Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 350
CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352
INTRODUCTION
Functional electrical stimulation (FES) is the application of electrical current to
excitable tissue to supplement or replace function that is lost in neurologically
impaired individuals. In addition to the chronic applications for restoration of

function described in this paper, electrical stimulation has also been used for many
therapeutic applications. The purpose of therapeutic electrical stimulation is to
improve tissue health or voluntary function by inducing physiological changes that
remain after the stimulation is used. In contrast, the purpose of an FES intervention
is to enable function by replacing or assisting a person’s voluntary ability. In FES
applications, stimulation is required to achieve a desired function; therefore, FES
systems are usually designed to be worn by the user. FES devices or systems that are
used as a substitute for lost neurological function are often called neuroprostheses.
Both sensory and motor function can be restored with FES. Auditory and visual
neuroprostheses are examples of FES used to restore sensory system functions.
This review is limited to applications of FES to the neuromuscular system. The
concept is to provide functional restoration through electrical activation of intact
lower motor neurons using electrodes placed on or near the innervating nerve fibers.
Appropriate electrical stimuli can elicit action potentials in the innervating axons,
and the strength of the resultant muscle contraction can be regulated by modulating
the stimulus parameters. A functional limb movement can be produced by properly
coordinating several such electrically activated muscles.
This article begins with a review of the physiological and technological prin-
ciples of electrical stimulation applied to the neuromuscular system. Next, the
development and status of neuroprostheses for restoring upper extremity, lower
extremity, bladder and bowel, and respiratory functions are highlighted. Finally,
future directions and developments of emerging FES technology and control meth-
ods are presented.
PHYSIOLOGICAL AND TECHNOLOGICAL

PRINCIPLES OF FES
Electrical Activation of the Neuromuscular System

Electrical pulses applied to nerves can elicit action potentials. The stimulating elec-
trode creates a localized electric field that depolarizes the cell membranes of nearby
neurons. If the depolarization reaches a critical threshold, an influx of sodium
ions from the extracellular space to the intracellular space produces an action
FUNCTIONAL ELECTRICAL STIMULATION 329
potential that propagates in both directions away from the site of stimulation.
Action potentials that propagate proximally in the peripheral nerve axons will ulti-
mately be annihilated at the cell body, and action potentials that propagate distally
will be transmitted across the neuromuscular junction and cause muscle fibers to
contract. In general, large-diameter axons (which innervate the larger motor units)
are activated with less current than small axons because the wider spacing between
nodes of Ranvier in large axons produces larger induced transmembrane voltage
changes (1). The activation of large motor units before small motor units is known
as reverse recruitment order and is the opposite of the physiological size principle,
where small motor units are initially recruited, followed by larger motor units.
FES applications for motor function operate under the fundamental principle
that electrical stimulation generally activates nerve rather than muscle. This is be-
cause the threshold charge for directly eliciting muscle fiber action potentials is
much greater than the threshold for producing action potentials in neurons (2), as
shown in Figure 1. Thus, for FES to be effective, the lower motor neurons must be
intact from the anterior horns of the spinal cord to the neuromuscular junctions in
the muscles that are to be activated. Lower motor neuron damage prevents the appli-
cation of FES in cases of polio, amyotrophic lateral sclerosis (ALS), and peripheral
nerve injuries (e.g., brachial plexus). Additionally, the neuromuscular junction and
muscle tissue must be healthy for electrical stimulation to be effective. This ham-
pers the application of electrical stimulation in muscular dystrophies. Neuromuscu-
lar electrical stimulation may be used when the lower motor neurons are excitable
and the neuromuscular junction and muscle are healthy, as is usually the case with
spinal cord injury (SCI), stroke, head injuries, cerebral palsy, and multiple sclerosis.
To date, most motor neuroprostheses have been targeted toward the SCI population.
Figure 1 Strength-duration curve for nerve and muscle tissue. Stimulus magnitude
required to produce a constant muscle response in normal and pharmacologically
denervated cat tibialis anterior. Modified after figure 20 in Reference 2.
330 PECKHAM KNUTSON
Electrical activation of neuromuscular tissue requires at least two electrodes to

produce a current flow. The electrodes are typically arranged in a monopolar or
bipolar configuration. With both configurations, one electrode, generally referred
to as the active electrode, is placed near the peripheral nerve to be stimulated. In
monopolar stimulation, the other electrode, known as the indifferent or return elec-
trode, is placed in a remote area near less excitable tissue, such as tendon or fascia.
Often the reference electrode has a larger surface area than the active electrode.
In bipolar stimulation, the reference electrode is placed near the active electrode.
Multichannel monopolar systems reduce the number of electrodes and leads re-
quired by using only one remote return electrode with several active electrodes
placed near motor points or nerves targeted for excitation. In multichannel bipolar
systems, each active electrode has its own return electrode, requiring more leads;
however, bipolar stimulation may allow greater selectivity of activation because
each electrode pair creates a more localized electric field (3).
Stimulation is delivered as a waveform of electrical current pulses, which are

characterized by three parameters: pulse frequency, amplitude, and duration. The
strength of muscle contraction is controlled by manipulating those parameters.
If the pulse frequency is too low, the muscle responds with a series of twitches.
Above a certain stimulation frequency, known as the fusion frequency, the re-
sponse becomes a smooth contraction. The cumulative effect of repeated stimuli
within a brief period of time is known as temporal summation. Higher stimulus
frequencies produce stronger muscle contractions up to a maximum, but also in-
crease the rate of muscle fatigue. Thus, high stimulus frequencies are generally
avoided. Typically, stimulus frequency rates of 12 to 15 Hz are the minimum re-
quired for summation of muscle twitches if the muscles have been conditioned
to have relatively long-duration twitches. An exercise regimen of low-frequency
muscle stimulation increases the contraction time and fatigue-resistance of muscle
fibers (4, 5). The strength of a muscle contraction may also be increased by increas-
ing the number of motor units activated, an effect known as spatial summation.
This is achieved by increasing the stimulus pulse amplitude and/or pulse dura-
tion, which effectively increases the electric charge injected, producing a larger
electric field and broader region of activation so that more axons and motor units
are activated (6). In most neuroprostheses, the strength of muscle contraction is
controlled by modulating the pulse amplitude or pulse duration, and the stimulus
frequency is set constant and as low as possible to avoid fatiguing the muscle
prematurely.
Stimulus waveforms are generally either monophasic or biphasic in shape. A
monophasic waveform consists of a repeating unidirectional (usually cathodic)
pulse. Biphasic waveforms consist of a repeating current pulse that has a cathodic
(negative) phase followed by an anodic (positive) phase. The first, or primary,
phase elicits an action potential in nearby axons, and the secondary positive pulse
balances the charge injection of the primary pulse. The purpose of the secondary
pulse is to reverse the potentially damaging electrochemical processes that can
occur at the electrode-tissue interface during the primary pulse, allowing neural
stimulation without causing tissue damage (2). The use of charge-balanced wave-
forms is especially important when the stimulating electrode is implanted rather
than placed on the surface of the skin. It is also important to use stimulus parameters
that are appropriate for the dimensions and material composition of the implanted
electrode being used so that the injected charge density per phase remains within
established safe limits, thereby preventing electrode corrosion (2).
Stimulators are designed to regulate either current or voltage. With voltage-
regulated stimulation, the stimulator output is a voltage, and therefore the mag-
nitude of current delivered to the tissue is dependent on the impedance at the
electrode interface (Ohm’s Law). With the use of surface electrodes, the impedance
at the electrode-skin interface increases as the electrode dries or loses contact

with the skin. As electrode impedance increases, the current delivered with a
voltage-regulated stimulator decreases, minimizing the possibility of skin burns
owing to high current densities. Thus, voltage-regulated stimulation is often used
for surface stimulation applications. The motor response, however, is more vari-
able with voltage-regulated stimulation because of impedance-dependent currents.
With current-regulated stimulation, the current is directly controlled and is not af-
fected by changes in the tissue load. Therefore, the quantity of charge delivered per
stimulus pulse can be guaranteed. To ensure that the stimulus charge is maintained
within safe levels, a current-regulated waveform is often used with implanted elec-
trodes. The use of a current-regulated stimulator also increases the likelihood of
obtaining repeatable muscle responses to stimulation.
Configurations of Neuroprosthetic Systems

Stimulation may be delivered using surface, percutaneous, or implanted systems
(Figure 2). Surface systems, sometimes referred to as transcutaneous systems, uti-
lize electrodes that are placed on the skin and are connected with flexible leads
to a stimulator that may be worn around the waist, the arm, or the leg. Usually, a
sensor or switch that controls the stimulation is also connected to the stimulator.
Surface electrodes are readily available in a variety of sizes from many manufac-
turers. The electrodes are placed on the skin over the nerves or over the “motor
Figure 2 Neuroprosthetic system configurations. S = stimulator, A = anode (ref-

erence electrode), C = cathode (active electrode), ECU = external control unit.
Single-channel monopolar stimulation of one muscle near its motor point is shown for
a surface, percutaneous, and implanted system.
332 PECKHAM KNUTSON
points” of muscles to be activated. The motor point is the site of stimulation that
produces the strongest and most isolated contraction at the lowest level of stimula-
tion. The advantages of surface systems are that they are noninvasive and relatively
technologically simple. Therefore, these systems are easily applied in the clinic,
easily reversible, and relatively inexpensive, making them especially well utilized
in therapeutic applications. However, the repeated placement of the electrodes in
the appropriate locations to get the desired response requires skill and patience.
Also, it can be difficult to achieve isolated contractions or activate deep muscles. In
sensate skin, the stimulation may generate painful sensations because of cutaneous
pain receptor activation. Furthermore, the appearance of the system may draw un-
wanted attention, especially if there are many stimulus channels and leads, and the
donning and doffing and managing of the external leads and components can be-
come unacceptable. These disadvantages have motivated the design of implantable
systems.
Percutaneous systems make use of intramuscular electrodes that pass through

the skin and are implanted into the muscles to be activated. Percutaneous electrodes
can activate deep muscles, can provide isolated and repeatable muscle contractions,
and are less likely to produce pain during stimulation because they bypass the
sensory afferents in the skin. Percutaneous electrodes are generally formed from
a multifilament lead within a single insulator that is wound into a helical configu-
ration (7). An electrode is inserted through the skin and implanted in the muscle
using a hypodermic needle. The electrode leads exit the skin and are connected
to external stimulation equipment. A large surface electrode is used as the return
electrode. The percutaneous interface on the skin is protected by placing a junction
connector over the skin surface where the electrodes exit. The skin at the electrode
site must be cleaned, dressed, properly inspected, and maintained to reduce the
risk of complications (8). Percutaneous systems provide a minimally invasive tech-
nique for investigating the feasibility of restoring functional muscle contractions
without having to prematurely subject research participants to implantable sys-
tem surgery. Percutaneous systems have served as precursors to fully implanted
systems (9) and have provided function in some individuals for periods of years
(10). Various reports have indicated the longevity of percutaneous electrodes (8,
11, 12), and they are now being investigated in a life-sustaining application for
respiration (13).
Implanted neuroprosthetic systems are designed for long-term use. Unlike sur-
face and percutaneous systems, the stimulator is implanted, eliminating the need
for wiring outside of the body to an external stimulator. The implanted electrodes
are connected by leads under the skin to the implanted stimulator. Thus, the elec-
trodes can be made with larger and more durable leads because they do not pass
through the skin. They are often connected to the stimulator using in-line connec-
tors, which permit the surgical removal and replacement of individual electrodes
if necessary. The circuitry of the stimulator is generally sealed in a titanium en-
closure, which serves as the indifferent electrode. The stimulator is implanted in
the chest or abdomen and receives power and command instructions through a
radio-frequency (RF) telemetry link to an external control unit (ECU). The power
demands of multichannel neuromuscular stimulation applications have made the
use of implanted batteries impractical, as they would require replacement on the
order of weeks, as compared with years for cardiac pacing. Advances in battery
technology have made rechargeable implanted batteries with service lives greater
than five years a feasible option for future neuroprosthetic systems.
The RF link allows the device to be fully passive with no active battery, thereby
eliminating the need for replacement of the implanted stimulator because of battery
failure. The telemetry link requires no wires through the skin; rather, a circular coil
(antenna) connected to the ECU is taped to the skin over the implanted stimulator.
The ECU may be worn on the body or carried on a user’s wheelchair. If required,
external transducers or switches for control may be connected to the ECU. The
ECU supplies power to the entire system through its internal rechargeable batteries,
receives signals from the transducers recording control information, and generates
control signals that are transmitted to the implanted stimulator. The system is
tailored to an individual by programming the ECU with stimulus and control
parameters that are required for each individual patient.
Implantable stimulators have the capability of using a variety of stimulation
electrodes. These electrodes may be implanted on the muscle surface (epimysial
electrode) (14), within the muscle (intramuscular electrode) (15), adjacent to a
nerve (epineural electrode) (16), or around a nerve (helix or cuff electrode) (17,
18). Epimysial electrodes have proved to be durable in upper (19) and lower (20)
extremity applications, and are especially useful for activating broad, superficial, or
thin muscles. Intramuscular electrodes allow activation of deeper or smaller mus-
cles, such as the intrinsic muscles of the hand. Nerve-based electrodes are used
when it is difficult to access the target muscle directly or when more complete mus-
cle recruitment can be obtained by stimulating the nerve. These technologies and
principles of electrical activation of the neuromuscular system form the foundation
for FES use in clinical applications.
CLINICAL APPLICATIONS
Numerous neuroprostheses for a variety of applications have reached the clinical
testing stage of development. Some have progressed to commercialization. Today,
there are FDA-approved neuroprostheses for restoring hand function, ambulation,
bladder and bowel control, and respiration. Some of these experimental and com-
mercial neuroprostheses are described below.
Upper Extremity Function

The objective of upper extremity neuroprostheses is to enable individuals with
upper extremity paralysis to use their hands in activities of daily living (ADL).
The first upper extremity neuroprostheses were developed in the 1960s (21, 22)
334 PECKHAM KNUTSON
using surface electrodes to open and close the hand. This pioneering work has led
to the development and clinical testing of surface (23–27), percutaneous (28, 29),
and implanted (30, 31) hand grasp systems. The primary target population for most
of these neuroprosthetic systems has been individuals with SCI, although some
systems provide therapeutic and functional benefits to additional populations, such
as individuals with stroke and traumatic brain injury.
Handmaster (NESS Ltd., Ra’anana, Israel) is the only surface system that is
commercially available today. The system consists of an adjustable wrist-hand
orthosis with five built-in stimulation electrodes for activating the finger and thumb
muscles (23). The user initiates a preprogrammed opening/closing stimulation
sequence by pushing a button on the system’s control unit. In a study of seven

subjects with C5 or C6 tetraplegia, the system was used at home to practice three
ADL for three weeks. At the end of the three weeks, all seven subjects used the
system successfully to perform ADL that they were unable to perform without
the system. Also, all had increased grip strength, finger motion, and Fugl-Meyer
scores when using the system (32). Handmaster is FDA-approved for therapy and
for providing hand function to individuals with C5 tetraplegia or hemiplegia caused
by stroke. The system has recently become available in the United States through
BioNESS Inc., Valencia, CA, a new company that brings together NESS, Ltd.
with the BION technology of the A.E. Mann Foundation (see Future Directions
and Developments, below).
Bionic Glove is a surface system that has also undergone significant clinical
testing. Developed at the University of Alberta, the system consists of a fingerless
glove with a forearm sleeve that is worn over three or four self-adhesive electrodes
previously placed on the hand and forearm (24). The stimulation is controlled with
wrist movements, which are sensed by a displacement transducer that spans the
wrist joint. Wrist extension beyond a certain angle triggers stimulation of grasp,
and wrist flexion triggers hand opening, thus enhancing the tenodesis grasp used by
individuals with lower cervical SCI. The stimulation is essentially on or off, with
an automatic ramping up and down. Because voluntary wrist extension is required,
the system is applicable to patients with C6 level tetraplegia and lower. An initial
report of nine subjects and a clinical trial of 12 subjects both demonstrated an
increase in the grasp force and a reduction in the time or difficulty in the performing
standardized hand function tests with the Bionic Glove (24, 33). Half of the subjects
in the second study continued to use the system at home after the study. The main
reason subjects discontinued use was insufficient benefit. Reported difficulties with
the system included donning and doffing the glove, achieving selective stimulation,
and lack of sufficient wrist control for heavy objects. The Canadian company that
was marketing the Bionic Glove is no longer active.
FESMate (NEC Medical Systems, Tokyo, Japan) is a percutaneous system that
is commercially available in Japan (34). It uses up to 30 percutaneous electrodes.
Stimulus patterns for several hand grasps and upper extremity motions are based on
templates of natural muscle activation previously recorded from able-bodied sub-
jects, which are customized with user-specific stimulus thresholds and maximums
(28). Several different command sources have been implemented, including head
switches, voice, sip and puff, and shoulder motion. This system has been applied to
individuals with cervical SCI (C4 to C6) and hemiplegia to produce hand, forearm,
elbow, and shoulder function, but no formal assessment of outcomes is available.
Freehand is an implanted system, developed at Case Western Reserve Univer-
sity (CWRU) and the Cleveland VA Medical Center (14, 35), to provide lateral
and palmar grasp to persons with C5 or C6 tetraplegia. It consists of a stimula-
tor/receiver implanted in the chest and eight epimysial or intramuscular electrodes
implanted at the motor points of hand and forearm muscles. The external com-
ponents include a radio-frequency-transmitting coil taped to the chest over the
implant, a programmable external control unit, and a transducer for detecting con-
tralateral shoulder motions. The movement of the shoulder proportionally controls
the degree of hand opening and closing. The Freehand system received FDA ap-
proval in 1997, and has been implanted in more than 250 people at several centers
worldwide (30, 36–39).

Fifty-one individuals with C5 or C6 tetraplegia were enrolled in a multicen-
ter clinical study of the safety and effectiveness of Freehand (30). The results
showed that the neuroprosthesis increased the pinch force of every subject, and it
enabled 98% of the participants to grasp and move more objects in a standardized
grasp-release test. The system particularly helped with the heavier objects, but
improvement in moving the lighter objects was also seen in weaker subjects. Free-
hand enabled every participant to perform at least one of the tested ADL tasks with
greater independence, and 78% of the participants could perform three tasks with
greater independence. Ninety-seven percent of 35 respondents said they would
recommend the neuroprosthesis to others, and 91% stated that the neuroprosthesis
improved their quality of life. Eighty-four percent of the participants reported reg-
ular device usage for functional activities, and an additional 8% used the device
for exercise. The implanted electrodes and leads have proven to be safe, reliable,
and durable. In an analysis of 238 electrodes, only three electrode-lead failures
and one electrode infection occurred, and the stimulus levels required to initiate
muscle contractions were stable over time (19). Despite the clinical successes and
high patient acceptance, the manufacturer of Freehand recently withdrew from the
SCI market and another manufacturer is currently being sought.
A second-generation Freehand system has been developed by the same investi-
gators that provides greater upper-limb function and incorporates implanted control
methods, thereby eliminating the need for the external shoulder position sensor.
Enhanced function is provided through additional stimulation channels, which are
used to activate hand intrinsic muscles, triceps, or pronator quadratus. Two new
approaches to implanted control have been developed. The first approach employs
an implantable joint angle transducer (IJAT) designed to detect wrist movements
(40). It consists of a magnet and an array of Hall effect sensors, which are implanted
in the lunate carpal bone and radius. Stimulation is proportionally controlled by
wrist position. The second approach is to extract control signals from myoelectric
(or electromyographic, EMG) signals recorded from muscles that remain under
336 PECKHAM KNUTSON
voluntary control (41). Potential control muscles include those synergistic to the
grasp movement, such as wrist extensors and, to a lesser extent, brachioradialis, as
well as nonsynergists such as sternocleidomastoid or trapezius. Because the IJAT
and myoelectric signal (MES) control strategies make use of signals derived from
the ipsilateral side, they allow the neuroprosthesis to be implemented bilaterally,
which is expected to provide significant additional function.
New technology was developed to make this implantable control and enhanced
function possible. A new implantable stimulator/telemeter (42) was developed
with the capacity to activate 10 or 12 muscles and to telemeter control signals to
an external control unit. Also, intramuscular electrodes were designed to stimulate
small muscles in the hand (15). Finally, the IJAT and recording electrodes were
made for detecting control signals. Two configurations of the stimulator/telemeter
exist. The first version has 10 stimulation channels and is compatible with the
IJAT. The second version has 12 stimulation channels and 2 channels for myo-
electric recording (Figure 3). The 12-channel implant also contains circuitry for
suppressing stimulus artifact in the recorded myoelectric signals.
Eight tetraplegic subjects have received the advanced system. Four subjects have
a 10-channel implant with an IJAT at the wrist, one has the 10-channel system but
uses an external joystick for control (31), and three subjects have the 12-channel
implant with MES control. One of the subjects with the 12-channel system has
recently received an additional 12-channel implant for bilateral upper extremity
Figure 3 Schematic representation of the CWRU/VA 12-channel im-

planted upper extremity neuroprosthesis with myoelectric control capability.
function. The follow-up period has been as long as 7 years for the 10-channel/IJAT
configuration and 1 year for the 12-channel/MES configuration. Each subject has
demonstrated the ability to grasp and release objects and use the system for func-
tional purposes. The clinical results thus far indicate a high level of satisfaction
with the advanced system, which provides greater functionality through the in-
creased workspace and improved grasp posture. Follow-up assessments and more
extended clinical trials are underway.
Lower Extremity Function

Much research has been conducted on the use of electrical stimulation in the lower
extremity, with various objectives, approaches, and populations being studied. One
of the earliest objectives of FES was to prevent the foot from dragging (footdrop)
during the swing phase of gait (43), an application that was mainly targeted to
individuals with poststroke hemiplegia. Also in the early 1960s, the first report of
the use of FES to enable a paraplegic to stand appeared (44). The first FES systems
for restoration of walking in hemiplegia and paraplegia were developed in the late
1970s (45). Today, these three objectives—prevention of footdrop, restoration of
standing and transfer, and restoration of walking—continue to be the focus of
several FES research programs.
FOOTDROP SYSTEMS Following Liberson’s first demonstration of an FES sys-

tem for hemiplegic footdrop (43), several researchers produced and tested similar
single- and multichannel footdrop systems in the 1960s and 1970s (see 46 and
47 for reviews). These systems used surface electrodes positioned on the tibialis
anterior and on the common peroneal nerve where it crosses the head of the fibula.
A heel switch, worn in the shoe of the paretic side, turned the stimulation on
when the foot was lifted off the ground and off at heel strike. Implanted systems
were also developed in the United States (48) and in Ljubljana, Slovenia (49, 50).
These consisted of a stimulator/receiver implanted in the medial thigh region or
calf with a cuff or epineural electrode that stimulated the peroneal nerve directly
and an external heel switch for control. Some challenges encountered in these
early surface and implanted systems included difficulties in properly placing the
surface electrodes, false triggering of the stimulation, inadvertent elicitation of re-
flex spasms in the plantarflexor muscles, pain or discomfort from the stimulation,
mechanical failure of the switch and other components, and difficulty in achieving
balanced dorsiflexion with a single electrode. However, these early trials demon-
strated the efficacy of FES for footdrop and provided incentive for the development
of improved systems.
Today, one FES footdrop system is commercially available and four are pro-
gressing toward commercialization. The Footlifter (Elmetec A/S, Arhus, Denmark)
(51) is a single-channel surface system with a heel switch. It is commercially
available in Europe and is used by more than 3500 Danish patients, according
to the company Web site. Walkaide, developed at the University of Alberta, is a
338 PECKHAM KNUTSON
self-contained surface system worn as a cuff below the knee, with a built-in tilt
sensor that detects step intention (52). It is the only footdrop system with FDA
approval, and has been recently licensed to Innovative Neurotronics, Inc., a new
subsidiary of Hanger Orthopedic Group, Inc. The Odstock Footdrop Stimulator
(ODFS), developed in Salisbury, U.K., is a single-channel footswitch-triggered
surface stimulator (53). It is the only footdrop system that has been tested in a
randomized controlled study, which demonstrated that the system increased walk-
ing speed and decreased walking effort. The effectiveness of the ODFS has also
been studied with positive results in subjects with multiple sclerosis (54). Clinical
studies of ODFS in the United States are being initiated. ActiGait (Neurodan A/S,
Aalborg, Denmark) is an implanted system that uses a four-channel cuff around the
common peroneal nerve (55). The stimulator is implanted on the lateral aspect of
the upper thigh. The signal from a wireless external footswitch is RF-telemetered
to an external controller worn at the waist. This system has recently received the
CE mark, a symbol indicating that it complies with the regulatory requirements

that allow it to be legally marketed in the countries within the European Free Trade
Association and European Union. The Finetech Dropped Foot System (Finetech
Medical Ltd., United Kingdom) is a relatively new implanted system that was de-
veloped at the University of Twente, Netherlands (56). It consists of a dual-channel
stimulator implanted below the knee, two bipolar epineural electrodes implanted
on the two branches of the common peroneal nerve to allow control of inversion
and eversion, and an external footswitch. Initial reports of clinical testing indicated
that the amount of eversion accompanying dorsiflexion could be adjusted (57) and
that the system increased walking speed by 24% (58). A larger trial is underway
to fully demonstrate the safety and efficacy of the Finetech Dropped Foot System,
and plans are being made to test the system in SCI.
STANDING/TRANSFER SYSTEMS Enabling persons with paraplegia to stand from

a seated position and transfer to another surface is the objective of some lower
extremity FES systems. The functional goals associated with standing include
reaching for high objects, having face-to-face interactions with other people, and
transferring between surfaces independently or with minimal assistance. The earli-
est demonstration of standing with FES was achieved by stimulating the quadriceps
and glutei muscles for knee and hip extension (44). Later studies showed that stand-
ing could be achieved in some subjects with stimulation of the quadriceps alone
(59, 60), but a more recent study suggests that adding stimulation of the hip exten-
sors could speed up the rising phase, thereby reducing the overall work effort (61).
A balance aid has been required in all FES standing systems. Presently, there are
no commercial or FDA-approved systems for FES-aided standing; however, one
implantable system has reached the multicenter clinical trial stage of development
(62).
The implantable standing neuroprosthesis (62) developed at the Cleveland VA
Medical Center and CWRU uses the same 8-channel stimulator/receiver as the
Freehand system (see Upper Extremity Function, above). Epimysial electrodes are

Figure 4 Schematic representation of the CWRA/VA 8-channel implanted

standing system.
implanted bilaterally on the vastus lateralis, gluteus maximus, and semimembra-

nosus (or alternatively, the posterior adductor magnus) for knee and hip extension,
and intramuscular electrodes are implanted in the lumbar erector spinae for trunk
support. The stimulator/receiver is implanted in the anterior lower abdomen, as
shown in Figure 4 (63). A command ring worn around the index finger and operated
by the thumb is used to send stand and sit command signals to an external control
unit worn around the waist. Balance and assistance is provided by the upper limbs
and a support device or assistant. In a pilot study of 12 subjects with C6 to T9 SCI,
the strength of knee and hip extension produced with stimulation was adequate to
achieve standing in all the subjects (62). The subjects generally maintained more
than 85% of their body weight on their legs and could stand for periods ranging
from 3 to 40 min. While standing, they were frequently able to release one hand
from the balance aid and manipulate objects in the environment. Some were even
able to use the system for limited swing-through walking with a walker. The system
340 PECKHAM KNUTSON
also reduced the effort and assistance required to transfer, especially from a lower
surface to a higher one. All of the subjects expressed satisfaction with the device
and indicated that they would go through the surgery and rehabilitation again to
achieve the same results (63, 64). The clinical trial is currently ongoing at multiple
centers in the United States.
In a separate study at Shriners Hospital in Philadelphia, the 8-channel
CWRU/VA stimulator was implanted in nine pediatric paraplegic subjects and
was compared with long leg bracing during several functional upright mobility
activities. Gluteus maximus, gluteus medius, posterior adductor magnus, and the
femoral nerve were stimulated. The rectus femoris was released to prevent hip
flexion. Ambulation was possible using a swing through pattern with crutches or
a walker and an ankle foot orthosis. The subjects could don the FES system faster
than the long leg braces, and could stand from a seated position and transfer into
an inaccessible bathroom stall more quickly and independently with the FES sys-
tem. Walking and stair climbing performance was similar with the two assistive
devices. The FES system was preferred for the majority of activities tested (65).
AMBULATION SYSTEMS The ultimate goal of lower extremity FES systems is to

enable individuals with complete paraplegia to walk again. Pioneering work by
Kralj, Bajd, and others in Ljubljana, Slovenia, introduced the technique of eliciting
a flexion withdrawal reflex of the hip, knee, and ankle by stimulating the peroneal
nerve (45, 66). This action, elicited by a single electrode, produces lower extremity
motion that can be substituted for the swing phase of gait. Stimulation of the
quadriceps for knee extension during the stance phase of gait completes the gait
cycle. This stimulation technique gave rise to the only FDA-approved FES system
for ambulation available. ParastepTM (Sigmedics, Inc., Fairborn, Ohio) uses four
to six channels of bilateral surface stimulation of the quadriceps, peroneal nerves
(for reflex withdrawal), and, if necessary, the glutei to enable individuals with
T4 to T12 paraplegia to walk with a walker (67). A microprocessor/stimulator
unit is worn at the waist, and using a walker with controls built into the handles,
individuals with paraplegia can stand and walk with reciprocal gait for limited
distances. Use of the system has additional medical benefits, such as increased
blood flow to the lower extremities, lowered heart rate at subpeak work intensities,
increased muscle mass, reduced spasticity, and psychological benefits.
Direct activation and control of individual muscles using either percutaneous
or implantable systems is an alternative to eliciting spinal reflexes for ambulation.
The percutaneous technique was most extensively developed by researchers at
the Cleveland VA Medical Center and CWRU, who synthesized complex lower
extremity muscles by activating up to 48 muscles under the control of a pro-
grammable microprocessor-based external stimulator (68). Stimulation patterns
were selected and initiated by the user through a hand-operated switch. With train-
ing, the subjects progressed to using a rolling walker for support, and some could
use axillary crutches and were able to climb stairs (69). The walking distance in
well-conditioned subjects was consistently 300 m at an average speed of 0.5 m/s
(68). Follow-up has been reported for over 17 years with two subjects who can use
their systems for exercise, standing, and walking with stand-by assistance (10).
One of these individuals performs exercise walking for five minutes five days per
week at a walking rate of approximately 0.6 m/s. The percutaneous system demon-
strated the ambulation that is achievable by activating individual muscles, and has
allowed the investigators to determine the most useful combination of muscles to
be activated in a system with an implanted receiver/stimulator and fully implanted
electrodes (70).
Several groups have developed implanted systems for ambulation, but few sub-
jects have been studied by any of these groups. A 12-channel system for activa-
tion of L2-S2 motor roots was attempted but did not provide adequate selectivity
of activation (71). A 22-channel system developed from cochlear implant tech-
nology (72) provided standing and limited swing-through gait in two subjects.
A 16-channel system consisting of two 8-channel stimulator/receivers implanted
bilaterally has provided one subject the ability to stand and walk for 25 m with
stand-by assistance (73). This approach, developed in Cleveland, provides a clin-
ical pathway for a subject with an 8-channel system for standing to progress to a
walking system by implantation of a second device.
Hybrid systems, which use both electrical stimulation and conventional external
bracing (74–76), take advantage of the added stability provided by bracing and
reduce the energy expenditure required for walking with FES alone. The bracing
supports the user’s body weight, and the stimulation provides propulsion. The most
widely tested hybrid system is the reciprocating gait orthosis (RGO) (77), which
uses surface stimulation of hip extension on one side to cause the RGO mechanism
to move the contralateral limb forward. Thus, walking is achieved by alternating
stimulation of the hip extensors. Although hybrid RGOs are often successful in
allowing users to stand and walk with less energy consumption, they are difficult to
put on and are often cosmetically unacceptable, and therefore have low long-term
usage rates.
In summary, all of the FES systems for standing and walking require the use of a
walker or standing frame for stability. Walking with FES requires considerable en-
ergy. The physical effort involved in standing and ambulation with FES is estimated
to be four to six times greater than normal (67, 68). Few individuals are able to walk
with any system without at least minimal assistance. Long and continued periods
of muscle strength conditioning are required as well as strategies to avoid muscle
fatigue. Because of these limitations, it is not anticipated that FES for walking in
paraplegia will soon replace the wheelchair as the primary mobility aide; however,
FES systems for standing, transferring, and short distance mobility and maneuver-
ing around barriers are expected to be more widely available in the near future.
Bladder and Bowel Function

Injury above the sacral levels of the spinal cord results in loss of bladder and
bowel control. Numerous complications related to this loss of function include
342 PECKHAM KNUTSON
frequent urinary tract infections, renal deterioration, bladder or kidney stones, and
autonomic dysreflexia. These complications threaten to compromise the health and
quality of life of the individual, and can lead to major long-term, and even deadly,
consequences. The functional goal of a bladder neuroprosthesis is to produce
effective micturition (bladder voiding) and continence, and also improve bowel
function.
Micturition and continence are normally performed by the coordinated action
of the bladder (detrusor muscle) and the urethral sphincter, two muscles that are
controlled by neural circuits in the brain and lumbrosacral spinal cord. Suprasacral
lesions generally result in detrusor-sphincter dyssynergia (detrusor and urethral
sphincter contract simultaneously rather than reciprocally) and neurogenic detrusor

overactivity (NDO), which is characterized by involuntary detrusor contractions
during the filling phase. These effects result in elevated bladder pressure, which is
the primary risk factor for renal deterioration. Thus, reduction of bladder pressure
is also an important objective.

The use of electrical stimulation for restoring bladder control has been most
advanced by the work of Giles Brindley (78) in the United Kingdom. The Finetech-
Brindley Bladder System (Finetech Medical Ltd., United Kingdom) is an implanted
FES system that enables micturition by inducing a detrusor contraction through
stimulation of the sacral spinal nerve roots. Two pairs of tripolar electrodes (active
contact flanked on each side with a return contact) are implanted bilaterally on
the sacral spinal nerve roots either intradurally on the anterior (motor) roots in the
cauda equina via a lower lumbar laminectomy (79), or extradurally on the mixed
sacral nerves in the sacral canal via a laminectomy of S1–S3 (80). The advantage of
the intradural approach is that it attempts to only activate efferent (motor) fibers in
order to minimize activation of reflexes through the posterior sacral roots (Figure 5).
Figure 5 Innervation of bladder and external sphincter from sacral

spinal nerves. Modified after figure 1 in Reference 90.
However, the extradural approach is reported to carry less risk of trauma to the
nerves or cerebrospinal fluid leakage. The electrode leads are tunneled to a re-
ceiver/stimulator, which is placed in the anterior abdominal wall through a separate
incision. Bilateral posterior rhizotomies of the S2–S4 spinal nerves are usually per-
formed to eliminate involuntary detrusor contractions (associated with NDO) and
thereby provide continence. The benefits of posterior rhizotomy include abolition
of reflex incontinence, increased bladder capacity and compliance, prevention of
autonomic dysreflexia triggered from the bladder or bowel, and protection of the
kidneys from uteric reflux and hydronephrosis (81). However, posterior rhizotomy
also abolishes other potentially useful sacral reflexes, such as reflex erection, ejac-
ulation, defecation, and sacral sensation if present. The implant is powered and
controlled by an external control unit with a transmission antenna. The user selects
the program for bladder or bowel control and then turns the device on to deliver
the programmed pattern of stimulation to the sacral nerves.
Normally, during voiding the sphincter relaxes when the bladder contracts,
but stimulation of the sacral roots results in contraction of both the detrusor and
urethral sphincter. Coactivation occurs because the sacral roots contain both the
small-diameter preganglionic parasympathetic axons innervating the bladder via
the pelvic nerve and the large-diameter somatic motor axons innervating the ex-
ternal urethral sphincter. Because large fibers have a lower threshold for excitation
than smaller fibers, it is difficult to produce contraction of the detrusor without
contraction of the external sphincter. Nevertheless, micturition can be produced
by intermittent stimulation, a technique that takes advantage of the fact that the
relaxation time of the smooth detrusor muscle is longer than that of the striated
external urethral sphincter muscle. Intermittent stimulation (3 to 6 s on, 6 to 9 s
off) leads to sustained contraction of the bladder and relaxation of the sphincter
between stimulus periods. Thus, urine flows during the intervals between stimulus
periods and the bladder is emptied in spurts.
The system has been implanted in more than 2500 patients worldwide (82). The
majority of the first 500 patients had SCI. However, patients with other neurological
disorders, including multiple sclerosis, spinal cord tumors, transverse myelitis,
cerebral palsy, and meningomyelocele have also received the implant (83). In the
United States, the system is known as Vocare and has been approved by the FDA
for individuals with SCI.
A multicenter study has reported that more than 85% of 184 implant recipients
use the system as the primary means of bladder emptying (84). Residual volume
in the bladder following system use was less than 60 mL in 95% of the users
and less than 30 mL in 89%. A substantial decrease in symptomatic urinary tract
infections following use of the implant has been reported in several studies (81,
84–86). Continence is reported in more than 85% of patients (83, 84, 86, 87),
largely because of increased bladder compliance following the posterior rhizotomy
(88, 89). Most users become free of catheters and urine collection bags and can
discontinue anticholinergic medication, which in turn reduces constipation and
other side effects such as dry mouth and drowsiness (81, 90). Satisfaction with the
system has been reported to be high (81), and the hardware has been reliable (91).
344 PECKHAM KNUTSON
Bowel evacuation and penile erection are secondary uses of the Finetech-
Brindley/Vocare system. Regular stimulation of the sacral parasympathetic nerves
contributes to transport of stool through the distal colon into the rectum (92). Most
users report a reduction in constipation and reduced need for laxatives and stool
softeners (92, 93). Some users are able to defecate by a pattern of intermittent
stimulation similar to that used for micturition but with longer intervals between
bursts of stimulation to allow passage of stool (93). This results in a significant
reduction in time that is spent in bowel evacuation (81, 93). Penile erection has
been reported in 60% to 87% of men with intradural electrodes implanted on the
anterior sacral roots (83, 86, 87). In a smaller group of participants with extradu-
ral sacral root stimulation, penile erection by stimulation was achieved in only
approximately 10% (90).
Methods to prevent coactivation of the bladder and external urethral sphincter,
and thereby achieve a more physiological voiding pattern, are being investigated
(94). For example, selective stimulation of the small fibers innervating the bladder
may be possible by arresting action potentials in the large fibers innervating the
sphincter (95) or by elevating the activation threshold of the large fibers above that
of the small fibers (96).
Researchers are also exploring alternatives to posterior rhizotomy for reduc-
ing NDO and providing continence. One approach is to create a neural block
that mimics the function of the rhizotomy (97). Another approach is to induce
inhibitory reflexes through the natural neurological pathways, a technique known
as neuromodulation, which has shown some benefit in able-bodied subjects with
urge incontinence (98). A recent study implemented the Finetech-Brindley/Vocare
system without posterior rhizotomy but with stimulation of the posterior sacral af-
ferents (to suppress bladder activity) in addition to the anterior efferents (to induce
bladder contraction) (99). With this system, neuromodulation increased bladder ca-
pacity, but hyper-reflexia of the external urethral sphincter persisted and prevented
complete emptying in some cases.
Respiratory Function
Individuals with high cervical SCI or central alveolar hypoventilation (CAH) lose
the ability to breathe because central neural connections to the diaphragm have been
disrupted or because central respiratory centers are impaired. These individuals
typically become dependent on a mechanical respirator, which sustains life by con-
tinuously forcing air into and out of the lungs through a tracheostomy at the base of
the neck. Unfortunately, mechanical respiration is associated with substantial mor-
bidity, mortality, inconvenience, physical discomfort, fear of disconnection, diffi-
culty with speech, impaired sense of smell, and encumbered mobility (100). The
purpose of a respiratory neuroprosthesis is to restore breathing by rhythmically ac-
tivating the diaphragm, and thereby eliminate the need for a mechanical respirator.
Electrical stimulation can be used to activate the diaphragm through the phrenic
nerve. This technique, known as phrenic nerve pacing, was introduced in the 1960s
by Glenn and colleagues at Yale University (101). Bilateral synchronous stimula-

tion of the phrenic nerves causes contraction and descent of each hemi-diaphragm,
which subsequently causes a fall in intrathoracic pressure and inspiration. Cessa-
tion of stimulation results in relaxation of the diaphragm, an increase in intratho-
racic pressure, and exhalation. The stimulation cycle is repeated 8 to 14 times per
minute to produce a normal breathing pattern. The user may adjust the number of
breaths per minute and the duration of each breath. Phrenic pacing systems have
allowed users to decrease or even discontinue the use of mechanical respirators
and have enabled more normal breathing. The technique has been applied to more
than 1200 patients worldwide and has become a clinically accepted intervention
in selected individuals (100).

Commercially available phrenic pacing systems consist of electrodes implanted
bilaterally on both phrenic nerves in the cervical or thoracic region and receiver/
stimulators implanted bilaterally in an accessible area over the anterior thorax
(Figure 6). The stimulators receive power and commands from an external control
unit through antennae taped directly over each receiver/stimulator. Approximately
two weeks after device implantation, diaphragm conditioning is initiated. The time
required for conditioning the diaphragm to provide full-time ventilation may range
from 3 to 16 months (102).
Presently, there are two FES systems commercially available for diaphragm
pacing, the Avery Mark IV and the Atrostim system. The differences between
these systems are mainly in the type of electrodes and stimulation strategies used.
The Avery Mark IV (Avery Laboratories, Commack, NY), developed by Glenn
et al. (101, 103), uses either monopolar or bipolar nerve cuff electrodes. Usually,
a thoracotomy is made on the anterior chest wall to expose the phrenic nerve
Figure 6 Bilateral phrenic nerve pacing system.

346 PECKHAM KNUTSON
(103, 104). Alternatively, the stimulator and electrode may be implanted without
a thoracotomy through an incision in the neck, but with this approach submaxi-
mal diaphragm activation may result because additional accessory branches join
the phrenic nerve in the thorax in most individuals. The Avery system is FDA
approved and is the most widely used phrenic pacing system. The Atrostim sys-
tem (Atrotech Ltd, Tampere, Finland), developed at Tampere University of Tech-
nology in Finland, uses quadripolar electrodes with four contacts spaced evenly
around the phrenic nerve (105). Each of the four contacts in turn serves as the
cathode, and a contact on the opposite side of the nerve serves as the anode. Multi-
pole sequential stimulation is intended to reduce fatigue by activating motor units
only one fourth of the activation time that occurs with conventional monopolar
stimulation. Atrostim is commercially available in Europe and is approved for
investigational use by the FDA under the Investigational Device Exemption in the
United States.
A third system, developed in Austria, and which had been marketed for a
time as MedImplant (MedImplant Biotechnisches Labor, Vienna, Austria), used a
technique called carousel stimulation (16, 106). Four electrodes were sutured to
the epineurium of each phrenic nerve from a single 8-channel receiver/stimulator.
Only one of the four electrodes was used to stimulate each nerve during any given
inspiration; therefore, the various nerve compartments were stimulated in sequence
during subsequent inspirations. Currently, this device is not commercially available
(107).
With any phrenic pacing system, the diaphragm must be gradually reconditioned
after surgery to improve its strength and endurance (108). Conditioning is initiated
approximately two weeks after surgery. Phrenic nerve pacing is initially provided
10 to 15 min every hour and is gradually increased over a conditioning phase that
may take 10 to 12 weeks or longer, although it is possible to achieve full-time
pacing in some patients within 5 weeks (100). Low-frequency stimulation is used
during the conditioning phase to promote conversion of fast-twitch fibers to slow-
twitch fatigue-resistant fibers. After pacing is achieved throughout the waking
hours, pacing is provided during sleep and gradually increased until the pacer is
used full time.
Early studies reported high incidences of failure to achieve successful ventila-
tory support (109–111). These failures were mainly due to technical malfunction
of the device components or to insufficient phrenic nerve innervation (a patient se-
lection error). Additional complications encountered include diaphragm fatigue,
increases in airway resistance caused by the accumulation of airway secretions
(requires suctioning), infection, injury to the phrenic nerve, and upper airway ob-
struction after tracheostomy closure (100). There are only a few studies that have
evaluated recent success and complication rates. In one long-term follow-up study
of 12 quadriplegic patients, 50% continued to use the Avery system full time (mean
13.7 years), 1 used it part time, 3 stopped using it, and 2 were deceased (one had
used the system full time, and one had stopped using it) (102). Those who stopped
using the pacer did so because of inadequate social or financial support or medical
problems associated with the initial injury. All patients demonstrated normal tidal
volumes while pacing full time, no patient lost the ability to activate the phrenic
nerve, and threshold and maximal currents did not increase over time. In an inter-
national study of 64 patients using the Atrostim system, 94% of the 35 pediatric
patients and 86% of the 29 adult patients eventually achieved complication-free
successful pacing (112). Thirty-four percent of all subjects paced full time, 38%
paced only while awake, 14% paced only while asleep, and 2% stopped pacing.
The incidences of electrode and stimulator failure were 3.1% and 5.9%, respec-
tively, and the incidences of infection and nerve trauma were 2.9% and 3.8%,
respectively.
Many individuals with ventilator-dependent tetraplegia are not eligible for

phrenic nerve pacing because of complete or partial phrenic nerve injury. For pa-
tients with only a single functional phrenic nerve, it may be possible to achieve res-
piration by activating the inspiratory intercostal muscles in addition to the phrenic
nerve (113). In a study of four patients, upper thoracic epidural spinal cord stim-
ulation of the intercostal muscles in combination with phrenic nerve stimulation
produced inspiratory volumes equal to that typically achieved with bilateral phrenic
nerve pacing (114). These patients were able to achieve substantial time free of
mechanical ventilatory support, and all four reported a level of comfort very near
normal breathing.
Recently, an alternative to direct stimulation of the phrenic nerve has been
developed that uses a minimally invasive laparoscopic procedure to position intra-
muscular electrodes bilaterally near the motor point of the diaphragm (13, 115).
Percutaneous electrodes are used with an external stimulator, but the intent is to
fully implant the system by internalizing the stimulator. Fourteen subjects have
been implanted to date: two are recently implanted, two are in the condition-
ing phase and are able to achieve adequate tidal volume, and ten have condi-
tioned diaphragms. Nine of the ten subjects with conditioned diaphragms are
using the system for extended periods (12 to 24 h per day); one subject was
identified as a selection error because extensive denervation of the diaphragm
did not allow sufficient ventilatory support. If confirmed in additional patients, di-
aphragm pacing with intramuscular electrodes via laparoscopic surgery is expected
to provide a less invasive and less costly alternative to conventional phrenic nerve
pacing.
FUTURE DIRECTIONS AND DEVELOPMENTS

Introducing new technology and methodology into the clinic requires a translation
from bench research to clinical testing. A number of FES-related advances are
being investigated, and many are in or close to clinical testing. This summary can
provide only a glimpse of some of these developments, which will expand the
capability of FES systems to provide greater function to more people through a
wider range of applications.
348 PECKHAM KNUTSON
Emerging Technology
New system configurations and new electrodes for stimulating and recording neural
tissue are two major areas of FES technology development. Future systems will be
designed around a platform technology, where numerous neuromuscular deficits
will be treated using the same basic neuroprosthetic system or a subset of the
possible components of an entire system. New electrodes and electrode arrays
for interfacing directly with peripheral nerve, the spinal cord, and the cortex are
being designed to provide more complete motor recruitment with fewer electrodes,
exploit natural neural circuitry, and allow more natural control of neuroprostheses.
SYSTEM CONFIGURATIONS Surgically implanted stimulator systems powered by

an inductive link across the skin and using muscle- or nerve-based electrodes with
subcutaneous leads has been the system configuration for many successful neu-
roprostheses. The implanted neuroprostheses reviewed in this article, as well as
several commercially available auditory neuroprostheses, pain relief stimulators,
and deep brain stimulation systems all use this approach. In motor applications,
precise electrode placement is possible because of the open surgical procedure,
and function-enhancing reconstructive procedures (e.g., tendon transfers) can be
performed in the same surgery. Years of experience with this approach have estab-
lished proven techniques for system design and fabrication, known and accepted
polymers and metals for packaging and electrodes, and a clear regulatory pathway.
Most of these neuroprosthetic systems have been designed to perform essentially
only one function in one area of the body. Despite the clinical success of this system
configuration, the approach has limited expandability to multiple applications.
New system configurations should be suitable for a wide range of applica-
tions, expandable to provide more than one function, and compatible with new
technological advances as they emerge. This is particularly important for indi-
viduals with multiple system dysfunctions (e.g., SCI, stroke) who are interested
in using FES to address multiple impairments. Also, new system configurations
should eliminate the need for any external components. Two new approaches
that are being developed to overcome the limitations of current system config-
urations are microstimulators (called BIONs) and a networked neuroprosthetic
system.
The BION approach, which is being developed by the A.E. Mann Founda-
tion and the Mann Institute at the University of Southern California, uses small
cylindrical (2 mm outer diameter × 16 mm long) injectable single-channel units
(implanted through a 12-gauge trochar) with electrodes at both ends to provide
the local stimulation needed to activate nearby nerves (116). BIONs receive power
and control information from an external transmitting coil that must encircle each
BION. One such transmitting coil can power and control up to 256 BIONs. The
use of BIONs eliminates the need for implanted lead wires and reduces the extent
of surgery required to implant the system. BION placement cannot be directly
visualized, and a trochar introduction approach is used, similar to that developed
by CWRU/VA for implanting percutaneous stimulating electrodes reported above.

The use of peripheral nerve-based electrodes is precluded with this approach, as
are augmentative surgical procedures unless additional surgery is performed. Stud-
ies have begun on the use of BIONs for poststroke shoulder subluxation, chronic
osteoarthritis of the knee, urinary urge incontinence, and footdrop.
A networked neuroprosthesis is an alternative approach that is being developed
at CWRU. The proposed system has the advantage of being fully implantable, with
no external components, expandable to multiple applications, and upgradeable
to new advances by adding components (e.g., stimulator and sensor modules)
to a network. The design calls for a subcutaneous network cable and modules
attached to the cable at regions of the body where they are required. The network
cable will attach to an implanted central processor unit with a rechargeable power
source and programmability through a remote link. For complex disabilities such
as SCI, one basic system with peripheral modules could provide the functions
needed for multiple organ systems, eliminating the need for multiple devices that
are specially designed and difficult to integrate or upgrade. The implantable and
external components for this networked neuroprosthesis are in the final stages of
design.
STIMULATING AND RECORDING ELECTRODES The implantable techniques that

have been used clinically for neural excitation to date have included muscle-based
electrodes and first-generation nerve cuff or epineural electrodes. Muscle-based
electrodes will continue to be used because of their high level of specificity, but
new electrode designs have centered on direct nerve stimulation. The appeal of
nerve stimulation is that it may provide more complete muscle recruitment and
multiple muscles may be recruited by the same electrode, possibly reducing the
number of electrode leads, surgery time, and the number of incisions. New nerve
cuff electrodes have been designed to reshape nerve geometry to allow more
selective access to particular nerve fascicles for both recording and stimulation
(117). Alternatively, methods for “steering” current to different locations within
a nerve to selectively activate different muscles have been developed (118). An-
other option is an electrode cuff with blunt radial projections that slowly pen-
etrate into the interfascicular space, allowing selective stimulation of different
nerve fascicles (119). In addition to selective activation, these nerve electrode
designs may allow selective recording of neural information that can be used to
provide information for closed-loop feedback control or conscious level sensory
feedback (18).
Another stimulation technique that is being investigated employs small elec-
trode arrays that are placed in the gray matter of the spinal cord to activate the
motor unit pools of individual muscle groups (120, 121). These electrodes have
been demonstrated in animal models to activate a large percentage of the motor
units of individual muscles, thus potentially allowing further centralization of the
stimulation events. This approach may make it possible to exploit the natural neural
circuitry in the spinal cord to coordinate muscle activity.
350 PECKHAM KNUTSON
Perhaps the most dramatic potential advancement in neural recording comes

from the ongoing research to create a cortical interface for communication, known
as a brain-computer interface. Several different approaches are being investigated,
including those that use field potentials, such as surface electrodes on the scalp
(122), electrodes embedded in the skull (123), flat grids of electrodes implanted
subdurally on the surface of the cortex (124), and those that acquire the firing
patterns of many individual neurons, such as hair-thin intracortical microelectrode
arrays (125, 126) and the larger glass cone electrodes (127). These emerging tech-
nologies will be used to develop more natural control strategies for neuroprosthetic
systems.
Emerging Control Techniques

The control methods that have been used most frequently in upper and lower
extremity neuroprostheses have relied on external joint angle sensors, accelerom-

eters, or switches. The development and integration of implantable sensors for
control will continue to be an active area of research, as will the development of
control algorithms that enable intuitive operation of the neuroprosthesis with little
conscious attention.
The use of MES for control of a neuroprosthesis was proposed by Vodovnik
in the earliest days of FES (22), but has only recently been fully integrated
into an implanted system (see Upper Extremity Function, above). Future re-
search in MES control will focus on identifying appropriate control muscles
and strategies for additional applications, determining how to optimally pro-
cess and extract information from the MES, and developing and integrating con-
trol algorithms that make use of automated pattern recognition techniques (e.g.,
artificial neural networks). These techniques may result in a reduction of mis-
classifications that lead to inadvertent commands and an increase in the num-
ber of distinguishable patterns that could be used for controlling more device
functions.
Cortical control is the concept of using field potentials or the firing activity of
multiple individual cortical neurons to provide the command input to the neuro-
prosthesis. Researchers have demonstrated the feasibility of tetraplegic patients
controlling FES-mediated hand opening and closing with signals recorded from
electrodes placed on the scalp (128, 129). A patient with locked-in syndrome, a rare
neurological disorder characterized by paralysis of voluntary muscles in nearly all
parts of the body, can operate a cursor-controlled speech and typing device via
control signals recorded with penetrating cortical electrodes (127). In monkeys,
several investigators have demonstrated the ability to chronically record control
signals and translate those into cursor movement and movement of a robotic arm
in the absence of volitional movement of the limb that originally was associated
with firing of the cortical cells (130–132). Donoghue and colleagues have recently
reported on a tetraplegic subject controlling a cursor using a penetrating electrode
array (133). Translation algorithms for decoding the neural signals have been
developed (134), and the possibility of a cortical interface for control in the human
subject is approaching reality.
Another potential source of command signal is the electroneurogram (ENG)
(135, 136). Investigators at Aalborg University use an implantable amplifier/
telemeter (137) with a tripolar nerve cuff electrode to record, amplify, and teleme-
ter ENG through the skin to an external control unit. These investigators have
implemented in a stroke patient an implanted footdrop system that uses ENG
recordings from the sural nerve to detect heel strike and foot liftoff in place of an
external heel switch (136, 138).
Closed-loop control and sensory feedback may ultimately be incorporated into
future neuroprostheses. Closed-loop control is a means of improving functional

output by automatically adjusting the stimulation in the presence of perturbations
or fatigue. The neuroprostheses reviewed in this article all operate in an open-loop
fashion. Closed-loop systems require a source of information about the system
that is fed back to the controller for system regulation. Small, implantable sensors
for force or position feedback are not yet available. However, ENGs from intact
sensory (afferent) fibers (135, 136) have been used as feedback signals for system
regulation. Haugland et al. (139) found that the signal recorded from the palmar
digital nerve innervating the radial aspect of the index finger contained information
that could be used under restricted laboratory conditions to detect the occurrence
of slips and to adjust electrical stimulation of the thumb muscles to stop the slip
(140, 141). Sensory feedback is the creation of a sensory perception that correlates
to the FES-mediated action or apprises the user of the current neuroprosthesis state
or mode. This is likely to be most beneficial for individuals lacking sensation in
the area in which the FES intervention is applied. Also, as more control strategies
rely on biopotentials as control signals, the need for feedback regarding the control
signal in relation to decision boundaries may be needed to enhance controllability.
Sensory feedback would augment visual information and presumably would reduce
the user’s conscious effort that is directed toward controlling the neuroprosthesis.
Electrotactile, vibrotactile, and auditory feedback may be possible. The limitation,
as with closed-loop control, is the lack of appropriate sensors, which are required
to generate the information to be delivered to the user. Additionally, the optimal
means of delivery has not been developed.
A final emerging technique with enormous potential to enhance neuroprosthe-
sis function and control is the idea of an electrical nerve conduction block. The
use of electrical current to arrest the propagation of action potentials down nerves
in a way that is safe and quickly reversible could be applied to suppress undesired
sensation, such as pain, or deleterious motor activity, such as muscle hypertonicity
or spasticity. Unwanted or uncoordinated generation of nerve impulses is a major
problem in many disabling conditions such as peripheral pain, SCI, stroke, cere-
bral palsy, and multiple sclerosis. If these impulses can be intercepted along the
peripheral nerves, then the disabling condition can be reduced or eliminated. Many
previous studies have shown that high-frequency alternating current waveforms
produce a nerve block under isolated conditions in frog, rat, cat, and dog models.
352 PECKHAM KNUTSON
Renewed studies are underway to elucidate the mechanism for high-frequency

block and to determine safe and effective parameters for producing conduction
block in mammals and humans (97). The capability of using electrical currents
to inhibit as well as activate nerve and muscle would significantly increase the
potential benefits available with FES technology.
CONCLUSION
For more than 40 years, electrical stimulation has been used to restore neuro-
muscular function to people with paralysis. The principles of safe and reliable
activation of neural tissue and the methods of generating stable and controllable
muscle contractions have been established. Electrodes, stimulators, transducers,
and sensors have been developed and integrated into neuroprosthetic systems that
have benefited individuals with spinal cord injury and stroke. Clinical success
of several FES interventions has been demonstrated, but commercial success has
proven to be more difficult to achieve. Expanding the indications both within the
spinal cord injury and stroke populations, and beyond those populations to other
disability groups, will increase the number of people benefited as well as the po-
tential market size. Increasing the awareness of FES technology and its benefits
among rehabilitation practitioners and involving them in the development and clin-
ical testing of neuroprostheses are important components in increasing the number
of FES users and in penetrating the market. Costs can be decreased by developing
systems that can be more easily manufactured and mass produced. Technological
advancements will increase the benefit of neuroprostheses, making recipients more
functional and independent.
ACKNOWLEDGMENTS
This work was supported in part by grants from the Department of Veterans Af-
fairs Rehabilitation Research and Development Service, the National Institute for
Neurological Diseases and Stroke, the Food and Drug Administration, and the Na-
tional Institutes of Health for the General Clinical Research Center at MetroHealth
Medical Center.
The Annual Review of Biomedical Engineering is online at

http://bioeng.annualreviews.org
LITERATURE CITED
1. McNeal DR. 1976. Analysis of a model System II, ed. JM Brookshart, VB Mount-
for excitation of myelinated nerve. IEEE castle, pp. 155–87. Bethesda, MD: Am.
Trans. Biomed. Eng. 23:329–37 Physiol. Soc.
2. Mortimer JT. 1981. Motor prostheses. In 3. Grandjean PA, Mortimer JT. 1986.
Handbook of Physiology—The Nervous Recruitment properties of monopolar
and bipolar epimysial electrodes. Ann. trodes for functional electrical stimula-
Biomed. Eng. 14:53–66 tion. Arch. Phys. Med. Rehabil. 77:1014–
4. Peckham PH, Mortimer JT, Van Der 18
Meulen JP. 1973. Physiologic and 13. DiMarco AF, Onders RP, Kowalski KE,
metabolic changes in white muscle of cat Miller ME, Ferek S, Mortimer JT. 2002.
following induced exercise. Brain Res. Phrenic nerve pacing in a tetraplegic pa-
50:424–29 tient via intramuscular diaphragm elec-
5. Peckham PH, Mortimer JT, Marsolais EB. trodes. Am. J. Respir. Crit. Care Med.
1976. Alteration in the force and fatiga- 166:1604–6
bility of skeletal muscle in quadriplegic 14. Keith MW, Peckham PH, Thrope GB,
humans following exercise induced by Stroh KC, Smith B, et al. 1989. Im-
chronic electrical stimulation. Clin. Or- plantable functional neuromuscular stim-

thop. 114:326–33 ulation in the tetraplegic hand. J. Hand
6. Crago PE, Peckham PH, Thrope GB. Surg. Am. 14:524–30
1980. Modulation of muscle force by re- 15. Memberg WD, Peckham PH, Keith MW.
cruitment during intramuscular stimula- 1994. A surgically-implanted intramuscu-

tion. IEEE Trans. Biomed. Eng. 27:679– lar electrode for an implantable neuro-
84 muscular stimulation system. IEEE Trans.
7. Caldwell CW, Reswick JB. 1975. A per- Rehabil. Eng. 2:80–91
cutaneous wire electrode for chronic re- 16. Mayr W, Bijak M, Girsch W, Holle J,
search use. IEEE Trans. Biomed. Eng. 22: Lanmuller H, et al. 1993. Multichannel
429–32 stimulation of phrenic nerves by epineural
8. Knutson JS, Naples GG, Peckham PH, electrodes. Clinical experience and future
Keith MW. 2002. Electrode fracture rates developments. Asaio J. 39:M729–35
and occurrences of infection and granu- 17. Naples GG, Mortimer JT, Scheiner A,
loma associated with percutaneous intra- Sweeney JD. 1988. A spiral nerve cuff
muscular electrodes in upper-limb func- electrode for peripheral nerve stimulation.
tional electrical stimulation applications. IEEE Trans. Biomed. Eng. 35:905–16
J. Rehabil. Res. Dev. 39:671–84 18. Loeb GE, Peck RA. 1996. Cuff elec-
9. Peckham PH, Mortimer JT, Marsolais EB. trodes for chronic stimulation and record-
1980. Controlled prehension and release ing of peripheral nerve activity. J. Neu-
in the C5 quadriplegic elicited by func- rosci. Methods 64:95–103
tional electrical stimulation of the para- 19. Kilgore KL, Peckham PH, Keith MW,
lyzed forearm musculature. Ann. Biomed. Montague FW, Hart RL, et al. 2003. Dura-
Eng. 8:369–88 bility of implanted electrodes and leads in
10. Agarwal S, Kobetic R, Nandurkar S, Mar- an upper-limb neuroprosthesis. J. Reha-
solais EB. 2003. Functional electrical bil. Res. Dev. 40:457–68
stimulation for walking in paraplegia: 17- 20. Uhlir JP, Triolo RJ, Davis JA Jr, Bieri C.
year follow-up of 2 cases. J. Spinal Cord 2004. Performance of epimysial stimulat-
Med. 26:86–91 ing electrodes in the lower extremities of
11. Marsolais EB, Kobetic R. 1986. Implanta- individuals with spinal cord injury. IEEE
tion techniques and experience with per- Trans. Neural Syst. Rehabil. Eng. 12:279–
cutaneous intramuscular electrodes in the 87
lower extremities. J. Rehabil. Res. Dev. 21. Long C 2nd. 1963. An electrophysiologic
23:1–8 splint for the hand. Arch. Phys. Med. Re-
12. Shimada Y, Sato K, Kagaya H, Konishi N, habil. 44:499–503
Miyamoto S, Matsunaga T. 1996. Clinical 22. Vodovnik L, Long C 2nd, Reswick JB,
use of percutaneous intramuscular elec- Lippay A, Starbuck D. 1965. Myo-electric
354 PECKHAM KNUTSON
control of paralyzed muscles. IEEE Trans. 32. Alon G, McBride K. 2003. Persons with
Biomed. Eng. 12:169–72 C5 or C6 tetraplegia achieve selected
23. Snoek GJ, IJzerman MJ, in ’t Groen functional gains using a neuroprosthesis.
FACG, Stoffers TS, Zilvold G. 2000. Use Arch. Phys. Med. Rehabil. 84:119–24
of the NESS handmaster to restore hand- 33. Popovic D, Stojanovic A, Pjanovic A,
function in tetraplegia: clinical experi- Radosavljevic S, Popovic M, et al. 1999.
ences in ten patients. Spinal Cord 38:244– Clinical evaluation of the bionic glove.
49 Arch. Phys. Med. Rehabil. 80:299–304
24. Prochazka A, Gauthier M, Wieler M, Ken- 34. Handa Y, Handa T, Ichie M, Murakami
well Z. 1997. The bionic glove: an elec- H, Hoshimiya N, et al. 1992. Functional
trical stimulator garment that provides electrical stimulation (FES) systems for
controlled grasp and hand opening in restoration of motor function of para-

quadriplegia. Arch. Phys. Med. Rehabil. lyzed muscles—versatile systems and a
78:608–14 portable system. Front Med. Biol. Eng. 4:
25. Popovic MR, Keller T, Pappas IPI, Dietz 241–55
V, Morari M. 2001. Surface-stimulation 35. Smith B, Peckham PH, Keith MW,

technology for grasping and walking neu- Roscoe DD. 1987. An externally powered,
roprostheses. IEEE Eng. Med. Biol. Mag. multichannel, implantable stimulator for
20:82–93 versatile control of paralyzed muscle.
26. Saxena S, Nikolic S, Popovic D. 1995. IEEE Trans. Biomed. Eng. 34:499–508
An EMG-controlled grasping system for 36. Carroll S, Cooper C, Brown D, Sormann
tetraplegics. J. Rehab. Res. Dev. 32:17– G, Flood S, Denison M. 2000. Australian
24 experience with the Freehand System for
27. Thorsen R, Ferrarin M, Spadone R, Frigo restoring grasp in quadriplegia. Aust. N.Z.
C. 1999. Functional control of the hand in J. Surg. 70:563–68
tetraplegics based on residual synergistic 37. Cornwall R, Hausman MR. 2004. Im-
EMG activity. Artif. Organs 23:470–73 planted neuroprostheses for restoration of
28. Handa Y, Hoshimiya N. 1987. Functional hand function in tetraplegic patients. J.
electrical stimulation for the control of Am. Acad. Orthop. Surg. 12:72–79
the upper extremities. Med. Prog. Tech- 38. Mulcahey MJ, Betz RR, Smith BT, Weiss
nol. 12:51–63 AA, Davis SE. 1997. Implanted func-
29. Peckham PH, Marsolais EB, Mortimer JT. tional electrical stimulation hand system
1980. Restoration of key grip and release in adolescents with spinal injuries: an
in the C6 tetraplegic patient through func- evaluation. Arch. Phys. Med. Rehabil. 78:
tional electrical stimulation. J. Hand Surg. 597–607
Am. 5:462–69 39. Hobby J, Taylor PN, Esnouf J. 2001.
30. Peckham PH, Keith MW, Kilgore KL, Restoration of tetraplegic hand function
Grill JH, Wuolle KS, et al. 2001. Effi- by use of the neurocontrol freehand sys-
cacy of an implanted neuroprosthesis for tem. J. Hand Surg. Br. 26:459–64
restoring hand grasp in tetraplegia: a mul- 40. Johnson MW, Peckham PH, Bhadra N,
ticenter study. Arch. Phys. Med. Rehabil. Kilgore KL, Gazdik MM, et al. 1999.
82:1380–88 Implantable transducer for two-degree of
31. Peckham PH, Kilgore KL, Keith MW, freedom joint angle sensing. IEEE Trans.
Bryden AM, Bhadra N, Montague FW. Rehabil. Eng. 7:349–59
2002. An advanced neuroprosthesis for 41. Hart RL, Kilgore KL, Peckham PH. 1998.
restoration of hand and upper arm control A comparison between control methods
using an implantable controller. J. Hand for implanted FES hand-grasp systems.
Surg. Am. 27:265–76 IEEE Trans. Rehabil. Eng. 6:208–18
42. Smith B, Tang Z, Johnson MW, Pourmeh- 52. Wieler M, Stein RB, Ladouceur M, Whit-
di S, Gazdik MM, et al. 1998. An exter- taker M, Smith AW, et al. 1999. Multi-
nally powered, multichannel, implantable center evaluation of electrical stimulation
stimulator-telemeter for control of para- systems for walking. Arch. Phys. Med. Re-
lyzed muscle. IEEE Trans. Biomed. Eng. habil. 80:495–500
45:463–75 53. Burridge JH, Taylor PN, Hagan SA, Wood
43. Liberson WT, Holmquest HJ, Scot D, DE, Swain ID. 1997. The effects of com-
Dow M. 1961. Functional electrotherapy: mon peroneal stimulation on the effort
stimulation of the peroneal nerve synchro- and speed of walking: a randomized con-
nized with the swing phase of the gait of trolled trial with chronic hemiplegic pa-
hemiplegic patients. Arch. Phys. Med. Re- tients. Clin. Rehabil. 11:201–10
habil. 42:101–5 54. Swain ID, Mann GE, Taylor PN, Wood
44. Kantrowitz A. 1963. Electronic Physio- DE, Wright PA. 2003. Clinical use of FES
logic Aids. Brooklyn, NY: Maimonides to improve walking in people with multi-
Hosp. ple sclerosis. Presented at 2003 Int. Funct.
45. Kralj A, Bajd T, Turk R. 1980. Electri- Electr. Stimul. Soc. Conf., Maroochydore,
cal stimulation providing functional use Sunshine Coast, Queensland
of paraplegic patient muscles. Med. Prog. 55. Haugland M, Larsen B, Burridge JH,
Technol. 7:3–9 Svaneborg N, Iversen H, et al. 2004. A pre-
46. Kralj A. 1989. Review of lower extrem- liminary non-randomised study to evalu-
ity functional electrical stimulation (FES) ate the safety and performance of the Acti-
research in paraplegic subjects. In Func- Gait implanted drop-foot stimulator in
tional Electrical Stimulation: Standing established hemiplegia. Presented at the
and Walking after Spinal Cord Injury, ed. 9th Annu. Conf. Int. FES Soc., Bouren-
A Kralj, T Bajd, pp. 1–15. Boca Raton, emouth, UK
FL: CRC Press 56. van der Aa HE, Bultstra G, Verloop AJ,
47. Lyons GM, Sinkjaer T, Burridge JH, Kenney L, Holsheimer J, et al. 2002. Ap-
Wilcox DJ. 2002. A review of portable plication of a dual channel peroneal nerve
FES-based neural orthoses for the correc- stimulator in a patient with a “central”
tion of drop foot. IEEE Trans. Neural Syst. drop foot. Acta Neurochir. Suppl. 79:105–
Rehabil. Eng. 10:260–79 7
48. Waters R, Rhoades M, Montgomery J. 57. Kenney L, Bultstra G, Buschman R, Tay-
1975. Improvement of physical function lor P, Mann G, et al. 2002. An implantable
after stroke: surgical and orthotic manage- two channel drop foot stimulator: initial
ment. J. Am. Geriatr. Soc. 23:248–53 clinical results. Artif. Organs 26:267–70
49. Strojnik P, Acimovic R, Vavken E, Simic 58. Taylor P, Mann G, Wood D, Hobby J.
V, Stanic U. 1987. Treatment of drop 2002. Pilot study to evaluate the safety
foot using an implantable peroneal under- and efficacy of an implanted dropped foot
knee stimulator. Scand. J. Rehabil. Med. stimulator (IMPULSE). Presented at FES-
19:37–43 net, Univ. Strathclyde, Glasgow, Scotl.
50. Kljajic M, Malezic M, Acimovic R, 59. Bajd T, Kralj A, Sega J, Turk R, Benko
Vavken E, Stanic U, et al. 1992. Gait eval- H, Strojnik P. 1981. Use of a two-channel
uation in hemiparetic patients using sub- functional electrical stimulator to stand
cutaneous peroneal electrical stimulation. paraplegic patients. Phys. Ther. 61:526–
Scand. J. Rehabil. Med. 24:121–26 27
51. Petersen T, Klemar B. 1988. Electrical 60. Cybulski GR, Jaeger RJ. 1986. Standing
stimulation as a treatment of lower limb performance of persons with paraplegia.
spasticity. J. Neuro. Rehabil. 2:103–8 Arch. Phys. Med. Rehabil. 67:103–8
356 PECKHAM KNUTSON
61. Kuzelicki J, Kamnik R, Bajd T, Obreza P, 71. Rushton DN, Donaldson ND, Barr FM,
Benko H. 2002. Paraplegics standing up Harper VJ, Perkins TA, et al. 1997. Lum-
using multichannel FES and arm support. bar root stimulation for restoring leg func-
J. Med. Eng. Technol. 26:106–10 tion: results in paraplegia. Artif. Organs
62. Davis JA Jr, Triolo RJ, Uhlir J, Bieri C, 21:180–82
Rohde L, et al. 2001. Preliminary per- 72. Davis R, Patrick J, Barriskill A. 2001. De-
formance of a surgically implanted neu- velopment of functional electrical stimu-
roprosthesis for standing and transfers— lators utilizing cochlear implant technol-
where do we stand? J. Rehabil. Res. Dev. ogy. Med. Eng. Phys. 23:61–68
38:609–17 73. Kobetic R, Triolo RJ, Uhlir JP, Bieri C,
63. Davis JA Jr, Triolo RJ, Uhlir JP, Bhadra Wibowo M, et al. 1999. Implanted func-
N, Lissy DA, et al. 2001. Surgical tech- tional electrical stimulation system for
nique for installing an eight-channel neu- mobility in paraplegia: a follow-up case
roprosthesis for standing. Clin. Orthop. report. IEEE Trans. Rehabil. Eng. 7:390–
385:237–52 98
64. Agarwal S, Triolo RJ, Kobetic R, Miller 74. Popovic D, Tomovic R, Schwirtlich L.
M, Bieri C, et al. 2003. Long-term user 1989. Hybrid assistive system—the mo-
perceptions of an implanted neuropros- tor neuroprosthesis. IEEE Trans. Biomed.
thesis for exercise, standing, and transfers Eng. 36:729–37
after spinal cord injury. J. Rehabil. Res. 75. McClelland M, Andrews BJ, Patrick JH,
Dev. 40:241–52 Freeman PA, el Masri WS. 1987. Aug-
65. Johnston TE, Betz RR, Smith BT, Mulca- mentation of the Oswestry Parawalker or-
hey MJ. 2003. Implanted functional elec- thosis by means of surface electrical stim-
trical stimulation: an alternative for stand- ulation: gait analysis of three patients.
ing and walking in pediatric spinal cord Paraplegia 25:32–38
injury. Spinal Cord 41:144–52 76. Kobetic R, Marsolais EB, Triolo RJ, Davy
66. Bajd T, Kralj A, Turk R, Benko H, Sega J. DT, Gaudio R, Tashman S. 2003. De-
1983. The use of a four-channel electrical velopment of a hybrid gait orthosis: a
stimulator as an ambulatory aid for para- case report. J. Spinal Cord Med. 26:254–
plegic patients. Phys. Ther. 63:1116–20 58
67. Graupe D, Kohn KH. 1998. Functional 77. Solomonow M, Aguilar E, Reisin E,
neuromuscular stimulator for short- Baratta RV, Best R, et al. 1997. Recip-
distance ambulation by certain thoracic- rocating gait orthosis powered with elec-
level spinal-cord-injured paraplegics. trical muscle stimulation (RGO II). Part I:
Surg. Neurol. 50:202–7 performance evaluation of 70 paraplegic
68. Kobetic R, Marsolais EB. 1994. Synthesis patients. Orthopedics 20:315–24
of paraplegic gait with multichannel func- 78. Brindley GS. 1993. History of the sacral
tional neuromuscular stimulation. IEEE anterior root stimulator, 1969–1982. Neu-
Trans. Rehabil. Eng. 2:66–79 rourol. Urodyn. 12:481–83
69. Marsolais EB, Kobetic R. 1987. Func- 79. Brindley GS, Polkey CE, Rushton DN.
tional electrical stimulation for walking in 1982. Sacral anterior root stimulators for
paraplegia. J. Bone Jt. Surg. Am. 69:728– bladder control in paraplegia. Paraplegia
33 20:365–81
70. Kobetic R, Triolo RJ, Marsolais EB. 80. Sauerwein D, Ingunza W, Fischer J,
1997. Muscle selection and walking per- Madersbacher H, Polkey CE, et al. 1990.
formance of multichannel FES systems Extradural implantation of sacral anterior
for ambulation in paraplegia. IEEE Trans. root stimulators. J. Neurol. Neurosurg.
Rehabil. Eng. 5:23–29 Psychiatry 53:681–84
81. Creasey GH, Grill JH, Korsten U HS, 91. Brindley GS. 1995. The first 500 sacral
M, Betz R, et al. 2001. An implantable anterior root stimulators: implant fail-
neuroprosthesis for restoring bladder and ures and their repair. Paraplegia 33:5–
bowel control to patients with spinal cord 9
injuries: a multicenter trial. Arch. Phys. 92. Binnie NR, Smith AN, Creasey GH, Ed-
Med. Rehabil. 82:1512–19 mond P. 1991. Constipation associated
82. Rijkhoff NJ. 2004. Neuroprostheses to with chronic spinal cord injury: the ef-
treat neurogenic bladder dysfunction: cur- fect of pelvic parasympathetic stimula-
rent status and future perspectives. Childs tion by the Brindley stimulator. Paraple-
Nerv. Syst. 20:75–86 gia 29:463–69
83. Brindley GS. 1994. The first 500 pa- 93. MacDonagh RP, Sun WM, Smallwood R,
tients with sacral anterior root stimulator Forster D, Read NW. 1990. Control of
implants: general description. Paraplegia defecation in patients with spinal injuries
32:795–805 by stimulation of sacral anterior nerve
84. Van Kerrebroeck PE, Koldewijn EL, De- roots. Br. Med. J. 300:1494–97
bruyne FM. 1993. Worldwide experience 94. Rijkhoff NJ, Wijkstra H, van Kerrebroeck
with the Finetech-Brindley sacral ante- PE, Debruyne FM. 1997. Selective detru-
rior root stimulator. Neurourol. Urodyn. sor activation by electrical sacral nerve
12:497–503 root stimulation in spinal cord injury. J.
85. Brindley GS, Rushton DN. 1990. Long- Urol. 157:1504–8
term follow-up of patients with sacral an- 95. Bhadra N, Grunewald V, Creasey G, Mor-
terior root stimulator implants. Paraplegia timer JT. 2002. Selective suppression of
28:469–75 sphincter activation during sacral anterior
86. Egon G, Barat M, Colombel P, Visentin nerve root stimulation. Neurourol. Uro-
C, Isambert JL, Guerin J. 1998. Implan- dyn. 21:55–64
tation of anterior sacral root stimulators 96. Grill WM, Mortimer JT. 1995. Stimu-
combined with posterior sacral rhizotomy lus waveforms for selective neural stim-
in spinal injury patients. World J. Urol. ulaiton. IEEE Eng. Med. Biol. Mag. 14:
16:342–49 375–85
87. van der Aa HE, Alleman E, Nene A, Snoek 97. Kilgore KL, Bhadra N. 2004. Nerve con-
G. 1999. Sacral anterior root stimulation duction block utilising high-frequency al-
for bladder control: clinical results. Arch. ternating current. Med. Biol. Eng. Com-
Physiol. Biochem. 107:248–56 put. 42:394–406
88. MacDonagh RP, Forster DM, Thomas 98. Bosch JL, Groen J. 1995. Sacral (S3) seg-
DG. 1990. Urinary continence in spinal mental nerve stimulation as a treatment
injury patients following complete sacral for urge incontinence in patients with de-
posterior rhizotomy. Br. J. Urol. 66:618– trusor instability: results of chronic elec-
22 trical stimulation using an implantable
89. Van Kerrebroeck PE, Koldewijn EL, neural prosthesis. J. Urol. 154:504–7
Rosier PF, Wijkstra H, Debruyne FM. 99. Kirkham AP, Knight SL, Craggs MD,
1996. Results of the treatment of neuro- Casey AT, Shah PJ. 2002. Neuromodu-
genic bladder dysfunction in spinal cord lation through sacral nerve roots 2 to 4
injury by sacral posterior root rhizotomy with a Finetech-Brindley sacral posterior
and anterior sacral root stimulation. J. and anterior root stimulator. Spinal Cord
Urol. 155:1378–81 40:272–81
90. Creasey GH. 1999. Restoration of blad- 100. DiMarco AF. 1999. Diaphragm pacing
der, bowel, and sexual function. Top. in patients with spinal cord injury. Top.
Spinal Cord Inj. Rehabil. 5:21–32 Spinal Cord. Inj. Rehabil. 5:6–20
358 PECKHAM KNUTSON
101. Glenn WW, Hageman JH, Mauro A, in phrenic nerve stimulation. Pacing Clin.
Eisenberg L, Flanigan S, Harvard M. Electrophysiol. 10:246–51
1964. Electrical stimulation of excitable 111. Carter RE, Donovan WH, Halstead L,
tissue by radio-frequency transmission. Wilkerson MA. 1987. Comparative study
Ann. Surg. 160:338–50 of electrophrenic nerve stimulation and
102. Elefteriades JA, Quin JA, Hogan JF, Hol- mechanical ventilatory support in trau-
comb WG, Letsou GV, et al. 2002. Long- matic spinal cord injury. Paraplegia 25:
term follow-up of pacing of the condi- 86–91
tioned diaphragm in quadriplegia. Pacing 112. Weese-Mayer DE, Silvestri JM, Kenny
Clin. Electrophysiol. 25:897–906 AS, Ilbawi MN, Hauptman SA, et al.
103. Glenn WW, Phelps ML. 1985. Diaphragm 1996. Diaphragm pacing with a quadripo-
pacing by electrical stimulation of the lar phrenic nerve electrode: an interna-

phrenic nerve. Neurosurgery 17:974– tional study. Pacing Clin. Electrophysiol.
84 19:1311–19
104. Creasey G, Elefteriades J, DiMarco A, 113. DiMarco AF, Supinski GS, Petro JA,
Talonen P, Bijak M, et al. 1996. Electrical Takaoka Y. 1994. Evaluation of inter-

stimulation to restore respiration. J. Reha- costal pacing to provide artificial ventila-
bil. Res. Dev. 33:123–32 tion in quadriplegics. Am. J. Respir. Crit.
105. Baer GA, Talonen PP, Shneerson JM, Care Med. 150:934–40
Markkula H, Exner G, Wells FC. 1990. 114. DiMarco AF. 2001. Neural prostheses in
Phrenic nerve stimulation for central ven- the respiratory system. J. Rehabil. Res.
tilatory failure with bipolar and four-pole Dev. 38:601–7
electrode systems. Pacing Clin. Electro- 115. Onders RP, Dimarco AF, Ignagni AR,
physiol. 13:1061–72 Aiyar H, Mortimer JT. 2004. Mapping
106. Thoma H, Gerner H, Holle J, Kluger P, the phrenic nerve motor point: the key
Mayr W, et al. 1987. The phrenic pace- to a successful laparoscopic diaphragm
maker. Substitution of paralyzed func- pacing system in the first human series.
tions in tetraplegia. ASAIO Trans. 33:472– Surgery 136:819–26
79 116. Loeb GE, Peck RA, Moore WH, Hood K.
107. Mayr W, Bijak M, Rafolt D, Sauermann 2001. BION system for distributed neu-
S, Unger E, Lanmuller H. 2001. Basic de- ral prosthetic interfaces. Med. Eng. Phys.
sign and construction of the Vienna FES 23:9–18
implants: existing solutions and prospects 117. Tyler DJ, Durand DM. 2002. Function-
for new generations of implants. Med. ally selective peripheral nerve stimulation
Eng. Phys. 23:53–60 with a flat interface nerve electrode. IEEE
108. Nochomovitz ML, Hopkins M, Brodkey Trans. Neural Syst. Rehabil. Eng. 10:294–
J, Montenegro H, Mortimer JT, Cherni- 303
ack NS. 1984. Conditioning of the di- 118. Tarler MD, Mortimer JT. 2004. Selective
aphragm with phrenic nerve stimulation and independent activation of four motor
after prolonged disuse. Am. Rev. Respir. fascicles using a four contact nerve-cuff
Dis. 130:685–88 electrode. IEEE Trans. Neural Syst. Re-
109. Glenn WW, Brouillette RT, Dentz B, Fod- habil. Eng. 12:251–57
stad H, Hunt CE, et al. 1988. Funda- 119. Tyler DJ, Durand DM. 1997. A slowly
mental considerations in pacing of the penetrating interfascicular nerve elec-
diaphragm for chronic ventilatory insuffi- trode for selective activation of peripheral
ciency: a multi-center study. Pacing Clin. nerves. IEEE Trans. Neural Syst. Rehabil.
Electrophysiol. 11:2121–27 Eng. 5:51–61
110. Fodstad H. 1987. The Swedish experience 120. Prochazka A, Mushahwar V, Yakovenko
S. 2002. Activation and coordination control of functional electrical stimula-

of spinal motoneuron pools after spinal tion to restore hand grasp in a patient
cord injury. Prog. Brain. Res. 137:109– with tetraplegia. Neurosci. Lett. 351:33–
24 36
121. McCreery D, Pikov V, Lossinsky A, 130. Serruya MD, Hatsopoulos NG, Paninski
Bullara L, Agnew W. 2004. Arrays for L, Fellows MR, Donoghue JP. 2002. In-
chronic functional microstimulation of stant neural control of a movement signal.
the lumbosacral spinal cord. IEEE Trans. Nature 416:141–42
Neural Syst. Rehabil. Eng. 12:195–207 131. Musallam S, Corneil BD, Greger B,
122. Pfurtscheller G, Flotzinger D, Pregenzer Scherberger H, Andersen RA. 2004. Cog-
M, Wolpaw JR, McFarland D. 1995. EEG- nitive control signals for neural prosthet-
based brain computer interface (BCI). ics. Science 305:258–62

Search for optimal electrode positions and 132. Taylor DM, Tillery SI, Schwartz AB.
frequency components. Med. Prog. Tech- 2003. Information conveyed through
nol. 21:111–21 brain-control: cursor versus robot. IEEE
123. Kennedy P, Andreasen D, Ehirim P, King Trans. Neural Syst. Rehabil. Eng. 11:195–
B, Kirby T, et al. 2004. Using human 99
extra-cortical local field potentials to con- 133. Mukand JA, Williams S, Shaikhouni A,
trol a switch. J. Neural. Eng. 1:72– Morris D, Serruya MD, Donoghue J.
77 2004. Feasibility study of a neural in-
124. Leuthardt EC, Schalk G, Wolpaw JR, Oje- terface system for quadriplegic patients.
mann JG, Moran DW. 2004. A brain- Arch. Phys. Med. Rehabil. 85:E48
computer interface using electrocortio- 134. Schwartz AB, Taylor DM, Tillery SI.
graphic signals in humans. J. Neural Eng. 2001. Extraction algorithms for cortical
1:63–71 control of arm prosthetics. Curr. Opin.
125. Vetter RJ, Williams JC, Hetke JF, Nuna- Neurobiol. 11:701–7
maker EA, Kipke DR. 2004. Chronic 135. Hoffer JA, Stein RB, Haugland MK, Sink-
neural recording using silicon-substrate jaer T, Durfee WK, et al. 1996. Neural sig-
microelectrode arrays implanted in cere- nals for command control and feedback
bral cortex. IEEE Trans. Biomed. Eng. in functional neuromuscular stimulation:
51:896–904 a review. J. Rehabil. Res. Dev. 33:145–
126. Fofonoff TA, Martel SM, Hatsopoulos 57
NG, Donoghue JP, Hunter IW. 2004. Mi- 136. Sinkjaer T, Haugland M, Inmann A,
croelectrode array fabrication by elec- Hansen M, Nielsen KD. 2003. Biopoten-
trical discharge machining and chemi- tials as command and feedback signals in
cal etching. IEEE Trans. Biomed. Eng. functional electrical stimulation systems.
51:890–95 Med. Eng. Phys. 25:29–40
127. Kennedy PR, Bakay RA, Moore MM, 137. Donaldson Nde N, Zhou L, Perkins TA,
Adams K, Goldwaithe J. 2000. Direct con- Munih M, Haugland M, Sinkjaer T. 2003.
trol of a computer from the human cen- Implantable telemeter for long-term elec-
tral nervous system. IEEE Trans. Rehabil. troneurographic recordings in animals
Eng. 8:198–202 and humans. Med. Biol. Eng. Comput. 41:
128. Lauer RT, Peckham PH, Kilgore KL. 654–64
1999. EEG-based control of a hand grasp 138. Hansen M, Haugland MK, Sinkjaer T.
neuroprosthesis. NeuroReport 10:1767– 2004. Evaluating robustness of gait event
71 detection based on machine learning and
129. Pfurtscheller G, Muller GR, Pfurtscheller natural sensors. IEEE Trans. Neural Syst.
J, Gerner HJ, Rupp R. 2003. ‘Thought’- Rehabil. Eng. 12:81–88
360 PECKHAM KNUTSON
139. Haugland M, Sinkjaer T. 1999. Interfac- roprosthesis. Med. Eng. Phys. 26:439–
ing the body’s own sensing receptors into 47
neural prosthesis devices. Technol. Health 141. Inmann A, Haugland M. 2004. Implemen-
Care 7:393–99 tation of natural sensory feedback in a
140. Inmann A, Haugland M. 2004. Func- portable control system for a hand grasp
tional evaluation of natural sensory feed- neuroprosthesis. Med. Eng. Phys. 26:449–
back incorporated in a hand grasp neu- 58
P1: KUV
June 20, 2005 22:39 Annual Reviews AR248-FM
Annual Review of Biomedical Engineering

Volume 7, 2005
CONTENTS
FRONTISPIECE, Werner Goldsmith xii
WERNER GOLDSMITH: LIFE AND WORK (1924–2003), Stanley A. Berger,
Albert I. King, and Jack L. Lewis 1

DNA MECHANICS, Craig J. Benham and Steven P. Mielke 21
QUANTUM DOTS AS CELLULAR PROBES, A. Paul Alivisatos, Weiwei Gu,

and Carolyn Larabell 55
BLOOD-ON-A-CHIP, Mehmet Toner and Daniel Irimia 77
BIOCHEMISTRY AND BIOMECHANICS OF CELL MOTILITY, Song Li,
Jun-Lin Guan, and Shu Chien 105
MOLECULAR MECHANICS AND DYNAMICS OF LEUKOCYTE
RECRUITMENT DURING INFLAMMATION, Scott I. Simon
and Chad E. Green 151
DETERMINISTIC AND STOCHASTIC ELEMENTS OF AXONAL GUIDANCE,
Susan Maskery and Troy Shinbrot 187
STRUCTURE AND MECHANICS OF HEALING MYOCARDIAL INFARCTS,
Jeffrey W. Holmes, Thomas K. Borg, and James W. Covell 223
INSTRUMENTATION ASPECTS OF ANIMAL PET, Yuan-Chuan Tai
and Richard Laforest 255
IN VIVO MAGNETIC RESONANCE SPECTROSCOPY IN CANCER,
Robert J. Gillies and David L. Morse 287
FUNCTIONAL ELECTRICAL STIMULATION FOR NEUROMUSCULAR
APPLICATIONS, P. Hunter Peckham and Jayme S. Knutson 327
RETINAL PROSTHESIS, James D. Weiland, Wentai Liu, and Mark S. Humayun 361
INDEXES
Subject Index 403
Cumulative Index of Contributing Authors, Volumes 1–7 413
Cumulative Index of Chapter Titles, Volumes 1–7 416
ERRATA
An online log of corrections to Annual Review of Biomedical
Engineering chapters may be found at http://bioeng.annualreviews.org/

Artigo Fes

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Artigo Fes

Uploaded by

Copyright:

Available Formats

9 Jun 2005 21:10 AR AR248-BE07-11.

tex XMLPublishSM (2004/02/24) P1: KUV

Annu. Rev. Biomed. Eng. 2005. 7:327–60

FUNCTIONAL ELECTRICAL STIMULATION

Medical Center, Cleveland, Ohio 44109; email: pxp2@case.edu

Department of Veterans Affairs, Cleveland, Ohio 44106; email: jsk12@case.edu

Key Words motor neuroprosthesis, spinal cord injury, stroke rehabilitation,

328 PECKHAM KNUTSON

FUTURE DIRECTIONS AND DEVELOPMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . 347

impaired individuals. In addition to the chronic applications for restoration of

PHYSIOLOGICAL AND TECHNOLOGICAL

Electrical Activation of the Neuromuscular System

FUNCTIONAL ELECTRICAL STIMULATION 329

330 PECKHAM KNUTSON

Electrical activation of neuromuscular tissue requires at least two electrodes to

Stimulation is delivered as a waveform of electrical current pulses, which are

FUNCTIONAL ELECTRICAL STIMULATION 331

at the electrode-skin interface increases as the electrode dries or loses contact

Configurations of Neuroprosthetic Systems

Figure 2 Neuroprosthetic system configurations. S = stimulator, A = anode (ref-

332 PECKHAM KNUTSON

Percutaneous systems make use of intramuscular electrodes that pass through

FUNCTIONAL ELECTRICAL STIMULATION 333

Upper Extremity Function

334 PECKHAM KNUTSON

sequence by pushing a button on the system’s control unit. In a study of seven

FUNCTIONAL ELECTRICAL STIMULATION 335

worldwide (30, 36–39).

336 PECKHAM KNUTSON

Figure 3 Schematic representation of the CWRU/VA 12-channel im-

FUNCTIONAL ELECTRICAL STIMULATION 337

Lower Extremity Function

FOOTDROP SYSTEMS Following Liberson’s first demonstration of an FES sys-

338 PECKHAM KNUTSON

CE mark, a symbol indicating that it complies with the regulatory requirements

STANDING/TRANSFER SYSTEMS Enabling persons with paraplegia to stand from

FUNCTIONAL ELECTRICAL STIMULATION 339

Figure 4 Schematic representation of the CWRA/VA 8-channel implanted

implanted bilaterally on the vastus lateralis, gluteus maximus, and semimembra-

340 PECKHAM KNUTSON

AMBULATION SYSTEMS The ultimate goal of lower extremity FES systems is to

FUNCTIONAL ELECTRICAL STIMULATION 341

Bladder and Bowel Function

342 PECKHAM KNUTSON

sphincter contract simultaneously rather than reciprocally) and neurogenic detrusor

is also an important objective.

Figure 5 Innervation of bladder and external sphincter from sacral

FUNCTIONAL ELECTRICAL STIMULATION 343

344 PECKHAM KNUTSON

FUNCTIONAL ELECTRICAL STIMULATION 345

by Glenn and colleagues at Yale University (101). Bilateral synchronous stimula-

in selected individuals (100).

Figure 6 Bilateral phrenic nerve pacing system.

346 PECKHAM KNUTSON

FUNCTIONAL ELECTRICAL STIMULATION 347

Many individuals with ventilator-dependent tetraplegia are not eligible for

FUTURE DIRECTIONS AND DEVELOPMENTS

348 PECKHAM KNUTSON

SYSTEM CONFIGURATIONS Surgically implanted stimulator systems powered by

FUNCTIONAL ELECTRICAL STIMULATION 349

by CWRU/VA for implanting percutaneous stimulating electrodes reported above.

STIMULATING AND RECORDING ELECTRODES The implantable techniques that

350 PECKHAM KNUTSON