Critical Reviews™ in Biomedical Engineering, 44(1–2): 123–152 (2016)

The Evolution of Neuroprosthetic Interfaces

Dayo O. Adewole,a,b,c,d Mijail D. Serruya,e James P. Harris,a,b Justin C. Burrell,a,b Dmitriy Petrov,a,b
H. Isaac Chen,a,b,d John A. Wolf,a,b & D. Kacy Cullena,b,d,*
a
Center for Brain Injury and Repair, Department of Neurosurgery, Perelman School of Medicine, University of Pennsyl-
vania, Philadelphia, PA, USA; bCorporal Michael J. Crescenz Veterans Affairs Medical Center, Philadelphia, PA, USA;
c
Department of Bioengineering, School of Engineering and Applied Science, University of Pennsylvania, Philadelphia,
PA, USA; dPenn Center for Neuroengineering and Therapeutics, University of Pennsylvania, Philadelphia, PA, USA;
e
Department of Neurology, Jefferson University, Philadelphia, PA, USA

*Address all correspondence to: D. Kacy Cullen, Department of Neurosurgery, University of Pennsylvania, 105 Hayden Hall, 3320 Smith Walk,
Philadelphia, PA, 19104, USA, Email: dkacy@mail.med.upenn.edu

Abstract: The ideal neuroprosthetic interface permits high-quality neural recording and stimulation of the nervous system
while reliably providing clinical benefits over chronic periods. Although current technologies have made notable strides in this di-
rection, significant improvements must be made to better achieve these design goals and satisfy clinical needs. This article provides
an overview of the state of neuroprosthetic interfaces, starting with the design and placement of these interfaces before exploring
the stimulation and recording platforms yielded from contemporary research. Finally, we outline emerging research trends in an
effort to explore the potential next generation of neuroprosthetic interfaces.

KEY WORDS: neural engineering, brain-machine interface, prosthetic, challenges, recording, stimulation

I. INTRODUCTION jury or disease. For instance, the first commercially

available cochlear implant was developed in 1972
Prosthetic devices have existed for millennia. The
based on early work by Djourno et al. nearly 20
first prostheses were solely cosmetic, one of the ear-
years prior; it remains the most successful clinical
liest examples being an artificial Egyptian toe from
neuroprosthesis.8 Moreover, deep brain stimulation
the fifteenth century BC.1 In the sixteenth century, (DBS) has been a popular option for treating Parkin-
these devices became more functional when Am- son’s disease for nearly 30 years.8,9 Perhaps the most
broise Paré engineered an artificial hand actuated by technologically advanced upper-limb prosthesis, the
a system of springs and catches.1,2 Prosthetic tech- DEKA arm, boasts numerous degrees of freedom
nology continues to evolve today due to improve- (DoF) and a control scheme that allows for simulta-
ments in our fundamental understanding of the body neous coordination of multiple joints.10–12 Similarly,
and advancements in biomedical engineering.3,4 Pio- the modular prosthetic limb developed in the Johns
neering work in the late 1960s by Fetz revealed that, Hopkins’ Applied Physics Lab has 26 total DoF and
given feedback, monkeys could learn to consciously a distributed processor network for real-time motor
control the firing rate of their own cortical neurons.5 coordination. Mechanically, these artificial limbs
The next year, Humphrey et al found that record- behave with comparable dexterity to their human
ings from neurons could be used to predict, and counterparts.10,13 However, many of these techno-
thus potentially drive, arm movements.6 Subsequent logical achievements have not transitioned well to
groundbreaking studies demonstrated just that; in clinical deployment. Upper-limb prosthesis rejec-
1999, Chapin et al. showed that rats could control tion rates average 23–26% for adults, with varying
a robot arm’s trajectory via recordings from the mo- rejection rates for different subtypes.10,14,15 In fact,
tor cortex.7 These early works laid the foundation users generally prefer less advanced devices, attrib-
for the field of neuroprostheses, devices designed to uting their choices to the low practicality and dura-
interact with the nervous system and restore func- bility, unwieldiness, and lack of nonvisual or tactile
tion. The subject has long since developed into a feedback in more advanced prostheses.10,16 Given
multi-disciplinary field to address losses of motility, these trends, prosthesis adoption does not appear
sensation, and quality and ease of living due to in- driven by mechanical design or dexterity. Rather,

0278-940X/16/$35.00 ©2016 Begell House, Inc. www.begellhouse.com 123

124 Adewole et al.

the limiting factor in current prosthetic systems is erbates these difficulties.17,28 The effects of electrical
the quality of the neuroprosthetic interface (NI), stimulation (and different patterns thereof) on neuro-
defined here as any platform designed to facilitate nal activity are still undergoing extensive study.30,31
communication between the nervous system and a Even more “well-established” strategies experience
prosthetic device. hurdles; for instance, although the incidence of hard-
Neuroprosthetic interfaces can record from and/ ware issues in DBS is low, a decline in verbal fluency
or stimulate select areas of the nervous system, tra- is a common post-surgical complication.32 Ongoing
ditionally via electrodes placed near the cells or tis- research efforts have become more interdisciplin-
sues of interest.3,4,17 When localized to the central ary as the design challenges for NIs become better
nervous system (i.e., the brain and spinal cord), such defined. Among them are biocompatibility, decod-
devices are referred to as brain–machine interfaces ing neural information, spatial and temporal preci-
(BMIs) or brain-computer interfaces (BCIs).3,18,19 sion of recording and stimulation, signal fidelity, and
For our purposes, “neuroprosthetic interface” en- chronic efficacy in vivo.3,33–35 However, as clinicians,
compasses both BMIs and interfaces outside the engineers, and researchers address these hurdles, NIs
central nervous system (CNS), for instance, with the have been applied to controlling wheelchairs and
peripheral nervous system (PNS). An ideal interface computer programs, moving artificial limbs, and re-
would allow the user to directly control the output storing basic sensory feedback.17,36–43 Future NIs may
behavior of the prosthesis (e.g., the movement of an even serve to improve and restore memory, with cur-
artificial arm) while receiving relevant sensory in- rent research showing promise in rats and nonhuman
formation from the device. This essentially recreates primates.44,45
the control-feedback loop found in an intact limb, In this review, we present a broad overview of
where the nervous system propagates information the development of neuroprosthetic interfaces. We
via electrical signals, or action potentials, through- begin by outlining the basic design constraints for a
out the body.17,20,21 These signals provide output for neuroprosthetic interface, as well as important con-
controlled muscle contractions and input/feedback siderations for NI placement (i.e., within the CNS or
from sensory organs (e.g., texture, temperature, PNS). We then outline current NI strategies from the
position), creating a bidirectional pathway through literature, emphasizing achievements and ongoing
which we explore and manipulate our environment. challenges in the field. These strategies range from
It is generally accepted that introducing this input– noninvasive recording techniques to electrodes im-
output behavior in prosthetic devices would yield planted directly within the brain, and each type of
better, more natural integration with their users, NI has advantages and disadvantages. These studies
thereby improving their practicality, adoption rates, preface a discussion of promising areas of future de-
and positive human impact.22,23 velopment, in which researchers are exploring novel
Broadly speaking, direct communication with the ways (e.g., ultrasound and infrared light) to stimu-
nervous system is one of the primary goals in neuro- late and monitor the nervous system.
engineering to date.17,24 Recording neural activity can
reveal how the nervous system encodes information,
II. NI DESIGN OBJECTIVES
as well as user intent such as motor commands.19,25
Similarly, electrical stimulation (generally via cur- The specific application of an NI dictates its physi-
rent injection) can induce different neuronal behav- cal and functional parameters and introduces specif-
iors depending on the type and intensity of stimula- ic design constraints. In general, however, all plat-
tion, as well as the intended target. Stimulation of forms share the same goal: robust, clinically viable
the subthalamic nucleus, for example, can reduce the communication with the nervous system. This goal
tremors characteristic of Parkinson’s disease.21,26–28 underlies the rationale for three universal design ob-
Both recording and stimulation of the nervous system jectives that applies to all NIs: (1) biocompatibility,
have clear clinical benefits; however, the complex- (2) high-resolution/selectivity, and (3) long-term re-
ity of the nervous system has made a clear best stan- liability/stability. While the aesthetic of the interface
dard for neuroprosthetic interfaces difficult to deter- (e.g., its appearance and size) does not apply as a
mine.18,29 The incomplete understanding surrounding design objective in the context of the goal described,
nervous system repair and neural modulation exac- it is worth mentioning as an important factor in us-

Critical Reviews™ in Biomedical Engineering

Evolution of Neuroprosthetic Interfaces 125

ers’ comfort with and ultimate acceptance of a neu- cal activity. In many cases, smaller microelectrodes
roprosthesis. may be advantageous, with greater flexibility in
First, NIs must be biocompatible; that is, they placement and less contact with non-target cells.
must minimally disrupt the function of otherwise The precision with which an NI can stimulate or re-
healthy tissue. This applies not only to their physi- cord from the target cells is integral to its function-
cal properties (e.g., material composition and stiff- ality. Cochlear implants, for example, permit many
ness), but any effects of their function. Stimulating individuals to perceive speech but not music; lit-
electrodes, for example, must deliver safe amounts of erature suggests that an increased number of active
current into the surrounding tissue without inducing electrodes may aid in better melody recognition.50–52
irreversible redox reactions, which can damage both Moreover, there is often a tradeoff between record-
the electrode and the host.21 Biocompatibility must be ing/stimulation selectivity (i.e., the electrode prox-
considered in the context of the target tissue because imity to the tissue) and the foreign body response to
the response to a material is often dependent upon more invasive NIs (Fig. 1).
where it is implanted (e.g., brain versus peripherally). Finally, an ideal NI must exhibit consistent be-
These safety thresholds are dependent on not only the havior once implanted, both physically and func-
electrode type and size but also the stimulation pa- tionally. For instance, physically, NIs may experi-
rameters (e.g., waveform, duty cycle, pulse frequen- ence micro-motion relative to the brain that could
cy, and width); thus, they may well vary across dif- exacerbate a foreign body response and increase the
ferent NIs, as indicated in a more in-depth review of distance between the electrode and the target neu-
stimulation thresholds by Cogan et al.46 Equally im- rons. Moreover, implanted NIs may suffer from wire
portant is the method of delivery into the body, if the breakage, delamination, and insulation breakdown
device is to be implanted. Both short-term trauma (as over long periods of time in vivo (i.e., “wear and
seen in needle deliveries) and the longer-term foreign tear” issues).34,53–55 These NIs can affect the con-
body response can adversely affect NI function.35 A sistency of recordings, meaning more invasive NIs
common problem in implantable, electrode-bearing must often be recalibrated regularly. From a func-
NIs is encapsulation of the electrodes in fibrous tis- tional standpoint, consistency often pertains to the
sue, which ultimately compromises the ability of impedance of the electrodes. Related to this finding,
the device to stimulate and record as impedances as noted above, astrocytes and macrophages sur-
and other biophysical properties change.33–35 Thus, round NI electrodes post-implantation and increase
both the interface and its delivery method must be the distance between the active recording/stimula-
designed with tissue reactivity in mind. Moreover, tion site and the target neurons.56 This process may
this biocompatibility should persist for a lifetime. Al- effectively increase impedance or change the phase
though this goal has yet to be fully realized, research and frequency response thereof, dropping the signal-
achievements to date suggest it is far from infeasible; to-noise ratio (impairing or eliminating recording
a 2010 study reported 7 years of recording from the capabilities) while increasing the amount of current
monkey cortex with a microwire array. Visual pros- needed to effectively stimulate the target. Even opto-
theses have been implanted for more than 10 years genetic stimulation (addressed in more detail below)
in humans, and electrodes to correct foot-drop in pa- may suffer, as gliosis may obstruct the transmission
tients with hemiplegia have been implanted for up of light into tissue.33–35 Moreover, the foreign-body
to 12 years.47–49 These and other longitudinal studies response may result in a decreased neuronal density
continue to provide invaluable insights into the long- in the vicinity of the active region (potentially due to
term biocompatibility of NIs, neuronal degeneration), further inhibiting NI func-
Second, the interface should be designed such tion over time.33,34,57,58 Interestingly, the presence of
that (1) interference from other tissues (either elec- microglia has not been directly correlated with elec-
trically active or non-electrically active) is mini- trode performance, although other biotic factors (in-
mized and (2) the target area is close enough for traparenchymal bleeding, neurotoxic proteins) are
high-resolution spatial and temporal recording, and/ currently being explored as potential contributors to
or selective stimulation. The interface is dependent chronic degradation in NI performance.54,55,59 Hence,
on both the placement of the NI and its ability to chronic NI function is directly tied to its biocompat-
isolate the target signals from surrounding electri- ibility. Arguably, maintaining reliable performance

Volume 44, Issues 1–2, 2016

126 Adewole et al.

FIG. 1: Signal resolution and NI placement. In general, the more invasive the NI the higher the accessible spatial and
temporal resolution. Scalp-mounted EEG electrodes (1) and ECoG electrodes under the dura (2) record gross corti-
cal oscillations, while intracortical electrode arrays (3) can detect single-cell activity. (Adapted from Lynch & Jaffe,
2006.)

for the long term is the most important consideration to communicate using an intracortical NI linked
for NIs and the most challenging obstacle to date. to a computer.62,63 Similarly, at least one human
implanted with a microelectrode array has demon-
III. NI PLACEMENT: CENTRAL OR strated cursor control for 1,000 days, nearly 3 years
PERIPHERAL? post-implantation.64 Control signals recorded from
the brain have been used to drive direct stimulation
Typically, NIs can be categorized by their location of intact, but paralyzed muscle groups in both non-
in the body; that is, whether they interact with the human primates and, more recently, a quadriple-
CNS (BMIs, as described above; spinal cord) or the gic human.65,66 The high spatial resolution of these
PNS. We present a brief discussion of localization in BMIs requires the implantation of electrodes in the
the CNS versus PNS before exploring each in more otherwise non-injured brain. However, electrode
depth. implantation in the CNS may result in inflammation
and fibrous encapsulation of the device, often lead-
A. CNS Versus PNS Placement ing to signal instability as previously outlined.33,35
In the brain, recording BMIs offer levels of spa- To decode commands from neural recordings, BMI
tial resolution down to multi- or single-neuron software implements a learning algorithm to decode
(or unit) recordings.19,60 Several groups have used the user’s intent from specific patterns of neural ac-
these recordings of neural activity to drive motor tivity.19 With extensive training and calibration, us-
commands. In 2006, Hochberg et al. reported mo- ers can learn to modulate these patterns and control
tor cortex recordings for up to 11 months using an a target device. Such systems often require regular
implanted microelectrode array in human spinal tuning and recalibration to account for the signal in-
cord injury (SCI) patients, who could move com- stability often encountered by more invasive BMIs,
puter cursors and robotic arms.61 Since then, at least limiting utility and practicality.
one patient with locked-in syndrome and two with As the end point of the brain’s connection to
amyotrophic lateral sclerosis (ALS) have been able the environment, the PNS presents an opportunity

Critical Reviews™ in Biomedical Engineering

Evolution of Neuroprosthetic Interfaces 127

to leverage the processing power of the CNS rather ing musculature when making neural recordings, as
than straining to decipher it. For instance, PNS in- well as signal from muscle being much coarser that
terfaces present the benefit of leveraging specialized that from single units (e.g., neurons or axons), limits
brain networks and, in particular, spinal cord feed- the complexity of the command signal. Moreover, in
back loops to control and fine-tune movement. Us- patients in whom the pathway from the brain to the
ing this approach, motor commands can be recorded peripheral nerve is compromised (e.g., SCI or ALS
directly from motor neurons/axons or the muscles patients), the PNS is not a viable location for the in-
they innervate. Myoelectric prostheses, for example, terface, although this can be circumvented by stimu-
detect the electrical activity from residual muscles lating the PNS through commands recorded remotely
to drive motor commands. As the electrical signals from the brain.66
from muscle have an amplitude several orders of Finally, peripheral nerves must innervate a liv-
magnitude larger than those of nerves, they can be ing target to survive and maintain useful activity.
detected through the skin (i.e., noninvasively), and Following peripheral nerve injury, the distal axon
are easier to record and decode. For these reasons, segments undergo Wallerian degeneration and distal
myoelectric prostheses are the most widely available support cells provide a supportive environment for
advanced prosthetic system, although they present axon regeneration and reinnervation (see the 2011
their own issues. Namely, they offer a limited num- review article by Pfister et al.).75 However, this en-
ber of command channels (open and close), and ex- vironment is temporary, and numerous conditions
perience signal inconsistency due to skin movement (among them the size of the injury, time needed for
and sweat interfering with the surface electrodes over repair, and the condition of the proximal nerve seg-
time.16 Implantable myoelectric sensors (IMES) are a ment) often prevent reinnervation of the distal seg-
potential alternative, eliminating the aforementioned ment before it degrades.76,77 NIs that present a living
issues to provide more consistent signal quality and target may better take advantage of placement in
greater number of control channels.67–69 Early clinical the PNS, conceptually similar to TMR noted above.
trials show promise for IMES, with subjects perform- Moreover, early stages of tissue-engineered “biohy-
ing more complex tasks with IMES-driven prosthe- brid” platforms are discussed later in the review.
ses.67 Similarly, afferent nerve fibers can be stimu- There is no single best option for NI place-
lated to deliver sensory information to the host, with ment, as each comes with its own unique benefits
stimulation of different neuronal subtypes resulting and challenges. Often the particular deficit and/or
in different perceptions (e.g., pressure, texture, and desired efficacy of a device determines the location.
proprioception).70 Targeted muscle reinnervation However, the needs and condition of the patient of-
(TMR) reroutes nerves to distinct muscle groups, le- ten inform or provide constraints regarding where
veraging the larger signal amplitudes of muscles to an NI would function best. Even given efficacy and
decode motor commands from surface recordings. patient-specific parameters, there may be more than
Interestingly, multiple reports suggest that a similar one suitable site for NI implantation. With design
approach (targeted sensory innervation) may provide objectives and placement options in mind, we now
sensory restoration by redirecting nerves to sensory turn toward an overview of different NI strategies,
nerve fascicles to innervate predefined areas of skin; as well as notable accomplishments and common is-
post-innervation, the skin above the can be stimu- sues in the field.
lated, recreating a sense of touch for the otherwise
missing limb.71–74
IV. CURRENT INTERFACES IN THE CNS
NIs in the PNS may thus be subjected to less
of the computational burden of signal interpretation Generally speaking, BMIs extract and relay infor-
than those in the CNS. However, the PNS presents mation from the brain or spinal cord to some out-
its own challenges. The computational benefit of re- put that “replaces, restores…or improves natural
cording from the PNS (namely, that the signals come CNS output.”19 For instance, when disease and/or
“processed” from the CNS for optimal movement trauma compromise neuromuscular pathways, NI
control) is also its drawback: the signal may be ste- systems intercept neural activity directly from the
reotyped and require time and training to map to oth- sensorimotor cortex, bypassing the faulty pathways
er types of control. Also, interference from surround- entirely.25 Individuals with a diminished capacity

Volume 44, Issues 1–2, 2016

128 Adewole et al.

for voluntary movement (as seen in ALS, SCI, and oscillations once provided with some form of feed-
similar conditions) can thus use BMIs to commu- back, usually visual or auditory cues.96,97 This bio-
nicate and interact with their environment through feedback, combined with EEG’s noninvasiveness
an external device. Such subjects have controlled and simplicity, has made EEG the most common
computer cursors in two-dimensional and three- BMI for clinical use.41 Current EEG arrays are de-
dimensional spaces, have used computer applica- signed to monitor specific types and frequencies of
tions, have browsed the Internet, and have moved brain waves. With training, these waves can be used
motorized wheelchairs.40,43,78–83 BMIs also see use in as command signals.4,41,80,96,98,99 One set of examples
rehabilitation, with patients controlling robotic or- includes event-related brain potentials (ERPs),
thotics and muscle stimulators to reinforce or sup- which arise due to specific, task-dependent stimuli.
port movement.84–87 Electrical stimulation of the The P300, for example, is an ERP that occurs when
spinal cord is also being explored to restore move- the user is presented with the item they were focus-
ment and coordination to muscles paralyzed from ing on among other related items. For example, a
SCI.88–90 Notably, Harkema et al. have demonstrated P300-BCI for communication may present the user
that epidural stimulation of the spinal cord in SCI with different letters until the P300 is detected, indi-
patients can recruit local neural circuitry and give cating the user has seen the desired letter, and build
them the ability to stand with minimal support; at up words sequentially.98 Among their many applica-
least one study participant regained some degree of tions, EEG-BMIs have been used to control comput-
conscious control over lower limb movement and er programs, write text, move a computer cursor in
bladder function.91,92 three dimensions, direct powered wheelchairs, and
The most common BMIs can be described on a trigger functional electrical stimulation of otherwise
spectrum of invasiveness: generally, the more inva- paralyzed muscles.4,100–102
sive the system, the higher the signal resolution (Fig. Clinically, the noninvasiveness of EEG is benefi-
1). The three most prominent recording methods in cial, but it invariably limits the bandwidth of accessi-
the CNS are scalp electroencephalography (EEG), ble brain activity. Although increased understanding
electrocorticography (ECoG), which records from of EEG signals and analysis techniques may some-
the cortical surface (also called intracranial EEG, what reduce these limits, current spatial resolution is
or iEEG), and intracortical and depth electrodes. restricted to averages of neural activity across rela-
ECoG, intracortical electrodes, and depth electrodes tively large areas of the brain. Historically, the maxi-
also provide opportunities for stimulation and have mum transfer rate for EEG-BMI was believed to be
led to the early development of bidirectional BMIs, approximately 2 words per minute, although Chen
that is, systems capable of both recording and stimu- et. al. recently demonstrated a significant improve-
lation to “close the loop” with direct sensory stimu- ment to approximately 12 words per minute.13,103
lation, rather than auditory or visual feedback. There Other biopotentials arising from muscle activity
is also a growing body of literature on transcranial (EMG), eye movement (EOG), and the filtering ef-
magnetic stimulation (TMS) and transcranial direct fect of the layers of tissue between the brain and the
current stimulation (tDCS), non-invasive techniques scalp electrodes can also confound the target signal,
that can increase or decrease activity in specific cor- which is on the scale of microvolts.4,101,104,105 Many
tical regions.93 In the context of NIs, TMS and tDCS EEG-BMIs require repeated calibration due to signal
have been applied to improve motor learning, for variation within and across experiments, although
example, when calibrating control schemes for a research into adaptive classification algorithms to
BMI.93–95 address these issues is ongoing.106,107 Overall, EEG
remains a powerful and evolving tool in rehabilita-
A. EEG tion and assistive technology.4,13,36,40,41,78,84,86,101,108–110
Originally developed in 1929 by Hans Berger, EEG
B. ECoG/iEEG
uses electrodes placed on the scalp to detect oscil-
lations from neural activity summed over different For ECoG recordings, electrodes (usually embed-
regions of the brain.19,25 Researchers discovered that ded in a flexible grid) are placed underneath the
participants could learn to consciously affect these skull, either above the dura mater (epidural) or un-

Critical Reviews™ in Biomedical Engineering

Evolution of Neuroprosthetic Interfaces 129

derneath it in direct contact with the cortical surface The benefits of ECoG stem from its proximity
(Fig. 2).41,97,111 Compared with EEG, ECoG offers to the brain compared to scalp EEG: direct cortical
higher spatial resolution, greater spectral frequency, contact eliminates the signal attenuation and filtering
and a generally improved signal-to-noise ratio.112–114 that result from recording through the skull. These
These advantages have been leveraged in numer- advantages come at the cost of increased invasive-
ous studies demonstrating the potential of ECoG ness, as ECoG requires surgery to expose the brain
as a BMI with higher resolution than EEG and less and position the electrodes.19,97,114 As such, clinical
of the immune response than seen by intracortical studies have generally been limited to epilepsy pa-
electrodes.115 ECoG is most often seen in seizure tients already undergoing surgery for seizure inter-
intervention. During surgery, subdural arrays are vention.115 Clinical studies have shown that, with
used to record seizure activity, map the cortex, and control schemes adopted from EEG, ECoG-BMIs
target seizure foci before therapeutic resection; they require shorter learning periods while delivering
may also be used to study the effects of DBS.116,117 comparable accuracy.19,114 However, due to the above
Achievements in ECoG-BMIs include computer restrictions, these studies have historically been lim-
cursor control in one, two, and three dimensions, us- ited to short-term trials. The recently FDA-approved
ing recordings of motor imagery to select different Neuropace system, which uses subdural electrodes
characters, and more recently, real-time control of a to detect seizure activity, may provide opportunities
prosthetic limb with a movement prediction accura- for longer studies.121 Whether these benefits persist
cy of 69.2 percent.113,114,118,119 In 2013, the Fetz group over time remains to be seen, although a 2010 study
reported successfully using an ECoG grid to elicit by Chao et al. showed near-constant predictive ac-
sensations by stimulating the somatosensory cortex curacy over multiple months in monkeys.122 Histori-
in two human patients.120 Although not as spatially cally, ECoG has been considered “middle ground”
specific as intracortical electrodes, differing stimu- on the spectrum of BMIs. The electrode placement
lation frequencies and amplitudes were perceived as allows recording of finer-scale activity than EEG
changes in the sensation’s intensity.120 without penetrating the blood-brain barrier, poten-

FIG. 2: Neuroprosthetic interfaces in the CNS. Top: Placement and invasiveness for prominent BMI approaches. Cur-
rent interfaces interact with the CNS at the scalp (left), the brain surface (middle), or from within the brain (right).
Bottom: Examples of intracortical/penetrating neural electrodes. Intracortical NIs may take the form of microwire
assemblies (left), arrays (middle), or flat shanks with multiple active recording/stimulation sites (right; center ellipses
portray active sites).

Volume 44, Issues 1–2, 2016

130 Adewole et al.

tially preserving long-term signal quality.114,115,123 ing, intracortical and depth electrodes can be placed
Active efforts to improve ECoG platforms include with high spatial specificity, making them ideal for
more flexible arrays, electrode miniaturization, and directly stimulating specific areas of the brain. CNS
low-impedance electrode coatings—all designed to stimulation can be applied to suppress epileptic sei-
improve biocompatibility, signal resolution, and se- zures, mitigate symptoms of Parkinson’s disease (as
lectivity.124 While these improvements are beneficial seen with DBS), and this procedure may be benefi-
for NIs in general, craniotomies must be performed cial in treating certain psychiatric conditions.26,28,35,131
for the positioning of ECoG arrays on the brain. As Potential also exists for restoring sensory function
such, human ECoG trials are primarily acute, and using this technique. Several studies in non-human
the ratio of risk to benefit for ECoG use may well primates show that sensory information can be de-
vary depending on the patient. While the behavior of livered via intracortical stimulation, often as part of
ECoG arrays over time in humans has not been fully a closed-loop BMI.132–134 Perhaps one of the greatest
elucidated, longer-term animal studies with sub- examples of sensory restoration is that of cochlear
dural electrodes and the development of wireless, implants, which are designed to stimulate the au-
fully implantable ECoG devices show promise for ditory nerve tonotopically. Worldwide, more than
ECoG as applied to NIs.115,122,125–128 Significant prog- 100,000 people with hearing loss use cochlear im-
ress must also be made in ECoG signal acquisition plants to compensate for their impairments, and they
and analysis to account for the amplitude attenua- can even hear and understand speech.135 Similarly,
tion at higher frequencies (i.e., gamma bands, which retinal stimulation can elicit visual percepts in pa-
can exceed 100 Hz) and to better capture signals tients with vision loss.136,137 Ongoing development of
from subcortical areas. These higher frequencies electrodes and stimulation parameters has provided
have been strongly correlated with critical cognitive basic letter recognition, orientation, and improved
and motor processes, and although still largely un- visual acuity in blind patients.138–140 The first com-
tapped, they may prove valuable in next-generation mercially approved retinal prosthesis, Argus II, has
BCI systems.114 In short, more detailed risk–benefit been shown to improve hand–eye coordination, mo-
analyses must be conducted before accurate com- bility, and letter recognition, although the quality of
parisons can be made. these benefits varies as the stimulation complexity
increases.137,138,141
C. Intracortical and Depth Electrodes The high spatiotemporal resolution of intracorti-
cal and depth electrodes is achieved through expos-
Electrodes implanted within the cortex offer the ing and/or penetrating the brain, though these are
highest spatial and temporal resolution recording, significantly more invasive processes than those re-
and they are capable of recording single units (the ac- quired for EEG or even ECoG. Moreover, the qual-
tion potentials of individual neurons) as well as field ity of intracortical recordings tends to degrade over
potentials (activity across several neurons).97 Current long periods of time. While Hochberg et al. reported
intracortical electrodes include flat shanks and shank recordings from the motor cortex over 1 year post-
arrays with multiple electrically active sites, platform implantation, they reported decreased signal am-
arrays with several fixed electrodes (e.g., the well- plitude and channel count.61,97 Even when chronic
known Utah electrode array and its FDA-approved recordings are attained, extremely labor-intensive
iteration in BrainGate trials), microwire assemblies, recalibration has been necessary due to signal/spike
and DBS electrodes, which are implanted in subcorti- drift from day to day. Studies on single-unit stabil-
cal structures (Fig. 1).26,34,97,104,129 The high resolution ity show that individual neurons may vary their fir-
and signal-to-noise ratio (SNR) of intracortical NIs ing patterns significantly across recordings; one such
compared with EEG and ECoG makes them ideal for study reporting only 39% of the units remaining sta-
the real-time detection required for smooth prosthe- ble after 15 days.142,143 Thus, arguably the most sig-
sis control.130 Moreover, strategic positioning of the nificant design problem for intracortical electrodes
active sites offers selective recording from different is the stability of the interface over both acute and
neuronal populations. Simultaneous recordings from chronic time scales.33,34 This stability is hindered by
cortical and subcortical regions have been performed a host of factors acting on a range of time scales.
in rodents, primates, and humans.59 Beyond record- In the short term, the initial mechanical trauma of

Critical Reviews™ in Biomedical Engineering

Evolution of Neuroprosthetic Interfaces 131

electrode insertion and the presence of a foreign D. Bidirectional Systems

body lead to scar formation surrounding the implant,
Systems capable of simultaneous (or near-simulta-
inflammation, and a decrease in neuron density sur-
neous) electrical stimulation and recording of neural
rounding the electrode.33,53 Over time, implanted
activity are also being explored. Based on tempo-
electrodes may suffer from mechanical failures (e.g.,
ral and spatial proximity of the stimulation and re-
broken connectors, insulation fracture and delamina-
cording region(s), this approach introduces the ad-
tion).54,55,59 There may also be so-called micro-mo-
ditional challenge of the electrical stimulation itself
tion effects due to differences in mechanical proper-
interfering with the quality of the recording; how-
ties between electrodes and brain tissue that become
relevant due to head motion and/or pulsatile motion ever, there have been several notable successes in
due to the patient’s heartbeat. Electrode arrays fixed bidirectional neural interfaces. In 2011, O’Doherty
to the skull can also experience motion relative to et al. implanted microwires into the primary motor
the brain that may shift them away from optimal re- and somatosensory cortices of monkeys. By stimu-
cording locations.34,53,104,144 All of these factors can lation of the latter cortex, they were able to provide
consequently affect signal quality during chronic im- tactile feedback as the monkeys directed a computer
plantation. However, the exact effects of the immune cursor (using motor cortex recordings) to explore
response on signal quality are ambiguous. Studies different objects.134 Alessandro Vato et al. demon-
report high variability in both immune response and strated bidirectional BMI in rodents; early valida-
device functionality across and within several types tion showed it capable of stimulating the brain upon
of intracortical recording electrodes.34,35 Strategies detection of some target activity.159 Vato et al. also
to overcome these biocompatibility and biostability demonstrated a bidirectional BMI algorithm that
challenges include finding insertion speeds to mini- maps the state of an external device (e.g., position)
mize initial trauma, delivering immunosuppressive into a series of electrical impulses while transform-
and anti-inflammatory agents, incorporating bioac- ing recordings from the motor cortex into a tunable
tive molecules to encourage cell adhesion, and con- output force. Such a paradigm could conceivably be
structing electrodes from softer materials.34,145–151 applied toward the operation of an artificial limb.
Additionally, downstream correction in silico may The algorithm was tested against simulations of cor-
account for signal drift through auto-recalibration, as tical neural data, proving robust results even when
suggested by Bishop in 2014, or offset correction al- the data were artificially degraded to mimic declin-
gorithms, as demonstrated by Homer et al.152,153 Still, ing BMI sensitivity.160 Similarly, Dr. Fetz’s research
no consensus has been reached regarding the kind group at the University of Washington has conduct-
of cortical activity that is best for intracortical re- ed extensive research into linked microrecording-to-
cordings. The benefits of capturing individual spikes microstimulation systems.161–163 Here, the Neurochip
versus the summed activity of several neurons (local and its successor, the Neurochip 2, were designed
field potentials, or LFPs) are still being compared. to deliver intracortical stimulation upon detection
LFPs may be more stable, and they have been shown of certain neural activity. Both platforms have been
to yield similar performance to single-unit activity implanted in freely moving monkeys. As of 2011,
in predicting movement.154,155 Finally, current BMIs the Neurochip 2 was able to record (via ECoG) and
are limited by the number of electrodes and active stimulate the motor cortex in a closed loop.163
sites that can be incorporated onto one device, which Additionally, Brian Litt and other researchers
in turn limits the number of distinct neural signals have reported on extensive work in developing iEEG
that may be desired for multidimensional input/out- systems for seizure intervention.164,165 One such de-
put. The Argus II described above has 60 electrodes vice, the Neuropace RNS System, uses iEEG to re-
and covers 11˚ by 19˚ in the visual field, compared cord cortical activity and stimulates targeted regions
to the full field of human vision, which is roughly to suppress epileptic seizures upon the detection of
180˚ by 135˚, including peripheral vision.141,156,157 abnormal activity. Through careful feature selection
This limitation in NI bandwidth is a key target of the and algorithm design and training on large datasets,
DARPA Neural Engineering System Design (NESD) these platforms can be tuned to detect precursor sei-
initiative, which aims to record from and stimulate zure activity specific to each participant. The RNS
1,000,000 and 100,000 neurons, respectively.158 is FDA approved, although studies comparing the

Volume 44, Issues 1–2, 2016

132 Adewole et al.

behavior of closed-loop versus open-loop stimula- gible benefits for NI users, the effects of chronic im-
tion devices (which simply stimulate target regions plantation on both interface function and the host
regardless of cortical activity) are ongoing.164–166 brain must be further characterized.

E. Challenges in the CNS

V. CURRENT INTERFACES IN THE PNS
There is a clear tradeoff between spatiotemporal Peripheral nervous system interfaces (PNSIs) target
resolution and invasiveness for CNS interfaces (see the peripheral nerves located throughout the body
Figs. 1 and 2). Thus far, the output capabilities of for electrical stimulation and recording.71,174 Like
noninvasive BMIs are limited by their resolution BMIs, peripheral interface research aims to recover
and the ability to account for variation in the target bodily functions that are lost when trauma or dis-
signal features; the latter requires extensive training ease-related damage occurs to the nerves supplying
to learn alternative, potentially non-intuitive control the area and/or the end targets themselves. Although
schemes.83,106 Variability in performance accuracy peripheral nerves exhibit regenerative abilities when
across BMIs has been well documented; sources transected, this process is slow, and the target area
range from user fatigue and concentration to the may atrophy in the interim.174 By recording the ac-
type and severity of the user’s disability.4,98,167 Al- tivity from the proximal nerve stumps, peripheral
though EEG-BCIs have been used for gross motor interfaces can provide motor commands for a neuro-
commands, the narrow bandwidth and latency cur- prosthesis; similarly, information can be translated
rently prevent real-time, fine movement control for into electrical impulses to stimulate the nerve, repli-
prosthetic limbs.4 While the tailored electrode place- cating tactile and proprioceptive sensation.71,74,175–177
ment of invasive BMIs allows for higher-resolution Challenges in the PNS overlap significantly with
recording, the highly variable in vivo stability and those in the CNS, namely, the tradeoff between
long-term behavior of the current widely used tech- signal selectivity and tissue disruption, although
nologies in the field still pose substantial develop- the fascicular organization of peripheral nerves can
mental hurdles to their clinical adoption.33,35 be leveraged here. Targeted muscle reinnervation,
Overall, intracortical electrodes will require originally developed by Todd Kuiken, involves re-
significant improvements and further optimization routing nerves to isolated portions of healthy mus-
before widespread use ensues. However, early stud- cle and using the resultant EMG signals to control
ies in humans have shown that subjects implanted prostheses.72,178 While not directly contacting the
with intracortical arrays can operate computer ap- PNS, TMR leverages the larger signal amplitude of
plications and prosthetic devices, such as the DEKA EMGs, the ease of accessing muscle relative to the
Arm (reported by Hochberg et al. in 2012).61,168 Ad- nervous system, and preexisting methods of record-
ditionally, the bandwidth of ECoG allows for the ing muscle activity.71,178 These benefits are crucial
detection of key neural information related to motor to the success of TMR, and several research efforts
intent and certain disorders without directly pene- have demonstrated that patients undergoing TMR
trating the blood–brain barrier. Moreover, advances exhibit real-time multi-joint control of prosthetic
in flexible electronics (e.g., thin-film electrodes) limbs and perform daily tasks more naturally.72,178
have yielded ECoG arrays that conform to the brain Ongoing clinical trials in TMR will likely lead to
without compressing the tissue.125,169–171 However, further improvements. Another approach is the re-
implanting these arrays requires large craniotomies generative peripheral nerve interface (RPNI), devel-
for grid placement; thus, human trials with ECoG oped by Paul Cederna et al., wherein nerve fascicles
arrays have generally been limited to acute studies are implanted into muscle grafts, each with its own
for patients already undergoing treatments (primar- electrode.179 Although it is still being optimized,
ily for epilepsy), with a few recent trials in other RPNI has proven viable for more than a year in a
patient populations.119,172,173 More chronic stud- rodent model, with muscle reinnervation and con-
ies have been performed in nonhuman primates. A sistently detectable muscle activity during implanta-
2011 study in rhesus macaques showed stable power tion.179–182
spectra and evoked potentials over 8 weeks.125 Thus, Functional electrical stimulation (FES) is an-
while ECoG-based BCIs may in time provide tan- other application wherein peripheral nerves are

Critical Reviews™ in Biomedical Engineering

Evolution of Neuroprosthetic Interfaces 133

stimulated; clinical trials have proven FES useful in electrodes, the most widely investigated peripheral in-
correcting foot-drop, regulating respiration, and re- terface electrode, are favored for their ease of implan-
storing grasp for tetraplegics.71,100,183 Stimulation of tation and fascicular selectivity. Studies have demon-
the PNS also has applications in regulating bladder strated that certain cuff electrode designs can record
dysfunction, incontinence, and has even been im- and stimulate chronically, and they have already been
plicated in stroke protection in animal models and used in relieving pain, managing incontinence via
immune system modulation in at least one human stimulation, and controlling hand-grasp prostheses,
study.184–188 The NIH’s SPARC initiative (Stimulat- with some studies lasting several months.190–193 Us-
ing Peripheral Activity to Relieve Conditions) is ing multiple cuffs or several contacts within a cuff
focused on supporting these and similar research enables stimulation of different nerve fascicles. The
efforts in “peripheral neuromodulation,” i.e., influ- somatotopic organization of axons within these fas-
encing organ function via stimulation of the PNS. cicles allows for the selective, graded activation of
As with BMIs, peripheral interfaces can be broadly distinct muscle groups in the upper limb.3,194–198 A
differentiated by their invasiveness (Fig. 3). prevalent application of extraneural NIs is vagus
nerve stimulation (VNS), an invaluable technique
A. Extraneural that has been clinically approved to treat symptoms
Extraneural electrodes reside outside the epineurium of drug-resistant epilepsy and, more recently, chronic
(Fig. 3) and are the least invasive of the PNSIs. They depression.199,200 The therapeutic potential for VNS
take a number of forms, although they generally con- in Alzheimer’s disease, migraines, heart disease, and
sist of a biocompatible insulator such as silicone con- other psychiatric conditions is also being explored.200
taining at least two electrical contacts.71,189 Some are Similarly, cuff electrodes have been implanted in hu-
sutured directly to the epineurium, while others are man amputees, providing consistent tactile sensations
designed to wrap around or enclose the nerve. Cuff throughout the “phantom” limb for more than a year,

FIG. 3: NIs in the Peripheral Nerve. (A) General overview of peripheral nerve anatomy. Individual axons are bundled
in endoneurium and in turn, into discrete fascicles, supplied via blood vessels embedded in the perineurium. Fascicles
in turn are bundled and protected by the dense connective tissue of the epineurium. (B) Peripheral interface electrodes
range from extraneural (upper left), to penetrating/intraneural (upper right), to regenerative (bottom).

Volume 44, Issues 1–2, 2016

134 Adewole et al.

yielding measured improvements in dexterity with for approximately 3 months and experienced force
their prostheses.201 Flat-interface nerve electrodes feedback from the sensors of a prosthetic hand and
(FINEs) increase the surface area of the target nerve controlled it in turn; however, only a fraction of the
and consequent selectivity by flattening it; the amount electrodes remained functional by the end of the
of damage was force dependent with no detectable study due to wire failure.175 As with intracortical
damage below a certain threshold. Moreover, the flat- electrodes, the long-term stability of these devices
tening of the nerve increases the number of accessible must be further evaluated and improved upon.
fascicles near the epineurium.71,202,203 Computational
models of the FINE strongly suggest its potential for C. Regenerative
highly selective stimulation in FES applications; a Regenerative electrodes (commonly called sieve
theory since borne out by numerous animal and clini- electrodes) are placed across transected nerves, al-
cal trials.204–206 lowing the regenerating nerve to extend axons
through the electrode itself. Contacts around some
B. Intraneural
of the holes allow for recording from and stimula-
Intraneural electrodes penetrate the epineurium and tion of axons and groups of axons. Several regen-
generally the perineurium as well, contacting the erative electrodes also possess guidance tubes for
target fascicles directly. Electrodes have been de- proper placement of the nerve trunk ends.189 Sieve
signed for both longitudinal and transverse insertion electrodes made from polyimide reduce (but do not
into the nerve, although longitudinal intrafascicular eliminate) the compressive damage of more rigid ma-
electrodes (LIFEs) are more common and have been terials like silicon, which can lead to axonopathy.216
implanted in human subjects.71,176,207 The Utah ar- Moreover, polyimide electrodes have proven bio-
ray, originally developed for cortical applications, compatibility, with no detectable inflammation over
has also been adapted for experimental use to re- 12 months of implantation in a rat study and similar
cord from and stimulate peripheral nerves.208 They results in others.217,218 Recently, Gore et al. reported
experience higher selectivity and signal-to-noise successful motor axon growth through a PDMS re-
ratios, as well as a lower stimulus current threshold generative electrode in rats, with reinnervation of
(in general), as there is less leakage into neighbor- distal muscle confirmed through recording during
ing tissues.209 The selectivity is such that specific locomotion.219 However, a significant drawback with
fascicles and portions of single fascicles can be sieve electrodes is that the target nerve must be tran-
independently stimulated.210 Intrafascicular elec- sected, which may cause some degree of cell death
trodes have been shown to record action potentials in the projecting spinal motor and dorsal root ganglia
for months at a time in cats with minimal tissue re- neurons as well as introduce often significant time
activity, although due to their needle structure and periods for the axons to regenerate to appropriate
penetration into the perineurium, they are capable targets. Moreover, post-transection, fascicular or-
of damaging the nerve and can elicit secondary in- ganization of the target nerve changes dramatical-
flammatory responses.3,208,211,212 Some researchers ly—smaller sensory and autonomic fibers regener-
have sought to mitigate these issues by replacing ate through the sieve faster than larger (e.g., motor
the metal wires with polymer filaments, which more fibers), which in some cases may have diminished
closely match the stiffness of neural tissue and ex- or absence of regrowth.217,220,221 This process is re-
hibit limited electrode drift.71,213,214 An animal study flected, for example, in a low degree of reinnervation
has demonstrated that LIFEs can be used as part of a seen in distal muscle.217,220 Due to the invasiveness of
closed-loop system for controlling ankle position.215 the procedure and the time needed for the nerve to
Further research and experimentation may allow for heal, research with regenerative electrodes has been
similar developments in humans. A feasibility study limited mainly to animal studies.189,221–226 However,
has already shown that LIFEs can record viable mo- some of these studies indicate that such electrodes
tor command signals and transmit stimuli encoding may be chronically viable. Srinivasan et al. recorded
joint angle and force when implanted in the median action potentials from rats for up to 5 months post-
nerve.176 Similarly, a researcher volunteered to have implantation.225 efforts to optimize sieve electrodes
a 100-electrode array implanted in his median nerve in vivo are ongoing, and pores ranging from 30 to 65

Critical Reviews™ in Biomedical Engineering

Evolution of Neuroprosthetic Interfaces 135

nm in diameter appear to best support nerve regen- or cutting off the blood supply.71 This is a particu-
eration.71,227,228 Similarly, Bellamkonda et al. noted larly important consideration in cases of nerve in-
that a pore length of 3 mm ensured that at least one terface in motile limbs requiring sliding and often
node of Ranvier (where the action potential presents stretching of the nerve, as device “catch points” may
the largest detectable extracellular signal) would lead to nerve injury and impede blood flow.
be in the NI.225 Additionally, Kung et al. created a A related phenomenon is the loss of electrode
“regenerative peripheral nerve interface,” wherein functionality due to connection failure, which has
a transected nerve innervates distinct muscle units, generally been observed in longer studies.175,236 This
each of which was paired with an electrode.180 Thus problem is also complicated by the range of motion
far, the interface has yielded reliable behavior (i.e., experienced by peripheral tissues (i.e., limb move-
recorded compound muscle action potentials from ment), which necessitates the use of flexible materials
the muscle units) for up to 7 months in a rodent mod- that will remain in contact with the target area with-
el.180 Notably, studies have used polyethylene glycol out damaging the nerve. As TMR uses surface EMG
(PEG) to reconnect damaged nerve fibers at the axo- sensors to detect the rerouted motor commands, it
nal level.229–231 Immediately following trauma, PEG must contend with the interference from movement
therapy has been shown to rapidly restore functional and sweat, which negatively impact signal fidelity
connections in animal models; as such, it may be a through the skin.13
viable method to limit the damage of nerve transec- Reverse recruitment is another issue facing
tion and improve acute and chronic functionality of PNSIs, notably for FES applications. In the intact
sieve electrodes.229–233 neuromuscular pathway, motor units are activated in
order of increasing size; that is, smaller axons become
D. Challenges in the PNS active before larger ones (i.e., the size principle). The
Extraneural electrodes must contend with the rela- more fatigue-resistant motor units are recruited be-
tively large biopotentials of adjacent tissues such fore the fatigue-sensitive ones during normal muscle
as EMG, although insulation like that seen in cuff contractions.174,237 However, muscles activated via
electrodes at least partially mitigates this issue.71,234 FES often fatigue quickly due to violation of the nat-
Extraneural selectivity is mainly limited to fascicles ural size principle, with the early recruitment of the
near the epineurium, where the electrode contacts larger fibers due to nonspecific electrical stimulation
the nerve. FINEs as described above increase elec- of the nerve.238–241 Some efforts to address this effect
trical access to the deeper fascicles of the nerve. include varying stimulation intensities, frequencies,
Also, the slowly penetrating interfasicular nerve pulse widths, and “alternating” paradigms, wherein
electrode (SPINE) extends blunt electrical contacts different fascicles are stimulated in sequence to avoid
into the epineurium without piercing the perineu- overtaxing any one group of motor units.238,242,243 As
rium. However, SPINEs have only undergone acute with BMIs, evaluating and improving the long-term
studies, and their influence on the nerve over longer in vivo function of peripheral interfaces is crucial to
periods has not yet been investigated.71,235 Intraneu- harnessing their full potential.
ral and regenerative electrodes offer high selectivity
and have been proven stable, some for months at a VI. FUTURE INTERFACES: NOVEL PARADIGMS
time. However, the stability of these electrodes must
A. Optogenetics
be further studied and optimized before clinical tri-
als become widespread. Fibrous encapsulation of Optogenetics involves the incorporation of opsins,
these electrodes is common, and while potentially or light-sensitive ion channel proteins, into cells via
beneficial for physical stability and maintenance of transgenesis or viral vectors. Cells expressing these
intimate contact, the resulting signal attenuation can opsins are then subject to targeted stimulation by
lower the efficacy of implanted devices.3 Addition- illumination that in turn leads to changes in mem-
ally, the material properties of peripheral interface brane potential.60,244 Once opsins are introduced, it is
electrodes must be carefully tailored (as is the case possible to stimulate targeted neurons with pulses of
with all electrodes), as overly rigid or tightly fitting light rather than current injection. Stimulation can
electrodes can damage the nerve by compressing it be inhibitory or excitatory depending on the current

Volume 44, Issues 1–2, 2016

136 Adewole et al.

passed through the channel, and selectivity has high adeno-associated or lentiviruses to express photo-
spatial control; specific cell types can be targeted responsive proteins, or ex vivo tranfected neurons
via transfection methods, and only transfected cells must be injected into the brain (as described below).
will be susceptible to optical stimulation.60,245,246 Additionally, light waveguides must be implanted in
Although optogenetics technology has been exten- the brain in order to access deeper structures, and a
sively developed for neuronal probing and control fully implantable laser light system with sufficient
in vivo, recent efforts have been made to integrate power has yet to be developed (although LED op-
optogenetics into neural interface technology. Cur- trodes can be used).262 Finally, visible light cannot
rent research in this area is focused on studying how penetrate far into the brain before being scattered;
best to implement this powerful tool for neural in- this is being addressed with near-infrared (NIR)
terfaces and on combining it with existing interface light, which can penetrate significantly deeper into
technologies.247,248 Bryson et al. have demonstrated tissue. Nanoparticles designed to absorb NIR light
“orderly recruitment” of muscles innervated with and in turn emit appropriate wavelengths to activate
light-sensitive motoneurons.247,249–251 In this case, channelrhodopsins in target regions of the nervous
“orderly” refers to concordance with the size prin- system are being pursued.263,264 Tsien et al. also dem-
ciple: the preferential recruitment of smaller muscle onstrated that channelrhodopsins sensitive to far-red
units before larger, more easily fatigued units. This light (above 600 nm) can be activated noninvasive-
specificity is physiologically advantageous but, as ly because tissue absorbs and scatters such wave-
noted above, is currently unaccounted for in electri- lengths less effectively than blue light.265 Chronic
cal stimulation. Additionally, optogenetic stimula- function is also of concern because many optical
tion of the auditory nerve and retina has been real- probes are relatively rigid compared to the brain,
ized, and optogenetic stimulation arrays are being and have experienced both glial encapsulation and
developed for use in a closed-loop neuroprosthetic reductions in signal quality over weeks to months,
system.245,248,252 Implanted nerve cuffs offer a po- similar to other NI platforms.266
tential method of long-term interface, and they are
being investigated in animal studies.249 Similarly, B. Magnetoencephalography (MEG)
Steinberg et al. and others have begun exploring
Electrical activity induces magnetic fields that can
optogenetics as a therapeutic strategy for improving
be recorded via MEG. Similar to EEG, the magnetic
functional stroke recovery.253–255 Optogenetics may
fields can be consciously altered by users, allowing
even serve as a potential extension of current DBS,
for real-time recording and control schemes based
using light to selectively target neural circuitry.256,257
on these varying patterns, and there is evidence that
In a similar vein, neurons transduced with genet-
MEG confers greater spatiotemporal resolution re-
ically encoded calcium indicators (GECIs) or volt-
cording than EEG.267,268 Thus far, MEG has been ap-
age sensitive dyes (GEVSDs) can produce proteins
plied to control both virtual and physical prosthetic
that fluoresce during neuronal signaling.258–260 These
hands (alongside visual feedback of the hand’s po-
sensors enable the direct visualization of electrical
sition) and drive the operation of virtual software
activity in real time, providing an invaluable op-
by translating user-modified signals into computer
portunity that has already been used to supplement
mouse clicks.267,269–271 The primary limiting factor
and improve upon conventional recording meth-
for MEG in the context of NIs is the environment
ods.59,169,258 The combination of optogenetics and
GECIs in a closed loop has already been realized, required; participants are placed in a magnetically
with the GECI R-GECO1 and a channelrhodopsin shielded room and must remain still to avoid arti-
allowing for reliable, simultaneous activation and facts from body and head movements.
imaging both in vivo and in C. elegans with little
C. Ultrasound
overlap.261 It is feasible to consider that future neural
interfaces will use optogenetics and calcium/voltage A potential noninvasive stimulation and recording
sensing to interact with neural tissue without physi- methodology is transcranial Doppler ultrasound
cal contact or direct current injection. As with all NI (TCD), which measures cerebral blood flow veloci-
platforms, optogenetics poses its own unique chal- ties. Different states of mental activity can be re-
lenges. Host parenchyma must be transfected with flected by changes in blood flow velocities, which

Critical Reviews™ in Biomedical Engineering

Evolution of Neuroprosthetic Interfaces 137

are then detected with TCD.272 While TCD carries lectivity superior to that of electrical stimulation.290
an inherent latency compared to other NIs (5–10 The feasibility, safety, and selectivity of INS have
seconds between a change in mental state and the been well described and characterized in a number
corresponding change in blood flow velocity), it is of animal models. Notably, fascicular infrared stim-
robust against electrical artifacts and is more porta- ulation of the rat sciatic nerve elicited muscle re-
ble than MEG.272,273 Early research has shown TCD sponses with a selectivity previously only achieved
is fairly accurate in distinguishing distinct mental with intraneural microelectrodes.288 The develop-
states (83–86% in a 2011 study by Myrden et al.), ment of INS-based interfaces may offer significant
and it has been applied at least once to communi- advantages over electrical stimulation; the wireless
cate via control of a virtual keyboard.272,273 Ultra- stimulation and high spatial resolution allowing for
sound has also been used to evoke sensation through finer activation of the peripheral nerve. Recently,
stimulation of peripheral nerves; moreover, neuro- INS has been applied to the CNS, and has been
pathic tissue has been shown as more responsive to shown to effectively evoke excitation and inhibition
ultrasound than healthy nerve, potentially offering in the motor and somatosensory cortex, auditory and
a noninvasive way to identify neuropathic condi- vestibular systems, cortical column, and the primary
tions.274,275 Interestingly, focused ultrasound is also visual cortex in nonhuman primates.285,291,292 Kuo et
capable of inducing conduction block in peripheral al. also reported on the stimulation of the subtha-
nerves and is actively being explored as a potential lamic nucleus in a rodent model, with stimulation
treatment for pain and spasticity.276–278 resulting in increased dopamine, suggesting that
INS may be a potential therapeutic platform for Par-
D. Transcranial Magnetic Stimulation kinson’s disease and other dopamine-related condi-
TMS consists of magnetic pulses directed to the tions.293 Finally, early trials in humans have shown
brain. By varying the parameters of the pulse train that INS can be used to stimulate human dorsal root
and the coils, specific areas of the cortex can be ganglia.294 Although the research surrounding INS
targeted for excitation or inhibition.279,280 As a non- is promising, clinical applications of INS would re-
invasive technique, TMS has been an invaluable quire constant stimulation at 12–15 Hz, placing it
research tool for mapping the cortex, studying in- above the upper threshold for injury.295 An alterna-
formation processing, and investigating brain plas- tive to bypass these limitations has been proposed
ticity in humans.279 Interestingly, TMS can elicit that combines INS with extraneural stimulation
visual percepts in subjects through stimulation of via nerve cuff.296–298 This “electro-optical” para-
the occipital cortex, providing a way to map and digm uses INS to precondition the nerve, making it
evaluate functional differences in the visual cortex more excitable and amenable to electrical stimula-
for both seeing and blind subjects.281,282 Within the tion.296–298
context of NIs, TMS has been used in combination Another point of concern is the mechanism(s) of
with EEG to “link” two human brains by delivering action for INS. Research suggests a photo-thermal
movement commands or phosphenes (perceptions mechanism, wherein the absorption of photons heats
of light caused by non-visual stimuli) to one par- the water in targeted neural tissue, generating a tran-
ticipant based on recorded activity of the other.283,284 sient temperature gradient.286 This gradient has been
shown to generate depolarizing currents in neuronal
E. Infrared Nerve Stimulation membranes by increasing their capacitance.299 How-
ever, the exact workings are still not fully known
Extraneural optical nerve stimulation (also called
infrared nerve stimulation, or INS) using pulses of and are likely location-specific: INS of the co-
infrared light has been suggested as an alternative chlea, for example may involve a photo-mechanical
method to interface with neural tissue.285–289 INS uti- (through thermal expansion) or photo-acoustic (via
lizes low-energy, pulsed light to reliably stimulate laser-induced stress waves) effect.300
neural structures. INS parameters (radiation wave-
F. Biohybrid Microsystems
length, irradiation time, and energy) can significant-
ly alter the light-tissue interaction, and the careful The first “biohybrid” neural interface, devised by
selection of parameters may provide a level of se- Kennedy et al., was a glass cone electrode contain-

Volume 44, Issues 1–2, 2016

138 Adewole et al.

ing neurotrophic factors to elicit ingrowth of host cal “living electrodes.” These living electrodes are
neurites.110,301,302 In these so-called “neurotrophic composed of discrete population(s) of neurons con-
electrodes,” the activity of the neurites was recorded nected by long engineered axonal tracts that pen-
within the cone. Due to actual ingrowth and inte- etrate the brain to a prescribed depth for integration
gration between the electrode and host neurites, it with local neurons/axons, with the latter portion
was believed that this strategy would provide a more remaining externalized on the brain surface where
stable interface. In another more recent approach, functional information is gathered using less inva-
Mark Allen et al. constructed electrodes composed sive means (e.g., ECoG). In this radical paradigm,
primarily of extracellular matrix (ECM), the net- only the biological component of these constructs
work of proteins that surrounds cells in vivo.303,304 penetrates the brain, thus attenuating a chronic for-
As ECM is mechanically and biologically compli- eign body response. This strategy is founded on the
ant with the brain, electrodes composed primarily plasticity of both endogenous and tissue-engineered
of ECM may decrease the immune response and neural networks, whereby neurons have the intrinsic
mechanical mismatch seen by traditional rigid, non- ability to sense (through dendritic inputs) and re-
organic electrodes. Allen et al. fabricated collagen/ spond to (via axonally transmitted action potentials)
Matrigel-based electrodes and implanted them in a local activity. Toward this end, we have developed
rodent model, demonstrating multi-unit recordings three-dimensional micro-tissue–engineered neural
over 5 weeks and reduced glial scarring compared networks (micro-TENNs).310–313 By localizing the
with synthetic electrodes at 16 weeks.303 neuronal somata at one or both ends of hydrogel
In alternative approaches, a number of research micro-columns, we are able to create long encap-
groups have begun creating advanced biohybrid sulated axonal tracts within a miniature form factor
neural interfaces by incorporating living cells and in vitro that can then be precisely microinjected en
tissues into implantable devices.29,39,305–307 These masse into the brain. Initially, micro-TENNs have
efforts are intended to improve the integration of been developed for the physical reconstruction of
implants with the host nervous system. One such long-distance neural circuitry lost due to trauma or
approach involves coating intracortical electrodes disease (Fig. 4).311–314
with live cells, as explored by Purcell et al. in 2009 More recently, we have applied similarly engi-
and de Faveri et al. more recently.308,309 The results neered axonal constructs to serve as living electrodes
of Purcell et al suggested an initial “neuroprotective by synapsing with specific cortical regions and
effect” from the neural stem cell–seeded electrodes, transmitting information in one or both directions
mitigating the tissue response 1 week post-implanta- (Fig. 5). Paired with an ECoG array, these micro-
tion. However, increased neuronal loss was reported TENNs could relay information from the cortex to
after 6 weeks, possibly due to the degradation of the the surface and vice versa, eliminating the need for
hydrogel surrounding the cells.308 De Faveri et al. a non-organic chronic foreign body within the brain.
coated microelectrodes with neurons and astrocytes Similarly, transducing the micro-TENNs with chan-
within fibrin hydrogel, demonstrating high signal nelrhodopsins and/or genetically encoded calcium
quality and a diminished astrocytic response up to indicators would allow for targeted optical stimula-
30 days post-implantation.309 Electrochemical test- tion and recording as previously described. As liv-
ing of the coated microelectrodes showed no signifi- ing constructs, micro-TENNs offer a way to bypass
cant changes in their impedance, attributed to water the electrode size and number limitations of current
and ion absorption from the surrounding solution. NI platforms via “biological multiplexing”;each
The fibrin’s small swelling profile and controllable micro-TENN axon can synapse with multiple host
thickness allowed de Faveri et al. to reach a signal- neurons, offering a powerful method to reach a mul-
to-noise comparable to the bare microelectrodes for titude of neurons and several target regions with a
85% of their cortical recordings, although the fibrin single construct. Moreover, through careful selec-
coating did result in an increased distance between tion of the neurons used to create micro-TENNs and
the electrode and recording site.309 customized cell- and tissue-engineering techniques,
We are pursuing another biohybrid strategy we may influence the specific host neuronal sub-
for chronic BMI utilizing advanced micro-tissue types with which the living electrode neurons form
engineering techniques to create the first biologi- synapses, thereby adding a level of specificity to lo-

Critical Reviews™ in Biomedical Engineering

Volume 44, Issues 1–2, 2016
Evolution of Neuroprosthetic Interfaces

FIG. 4: Micro-tissue engineered neural networks (Micro-TENNs), consisting of discrete neuronal population(s) with long axonal tracts within a biocompatible micro-column. Micro-
TENNs are used for the direct reconstruction of long-distance axonal pathways after CNS degradation. (A) Diffusion tensor imaging representation of the human brain demonstrating
the connectome comprised of long-distance axonal tracts connecting functionally distinct regions of the brain. Unidirectional (red, green) micro-TENNs and bidirectional (blue)
micro-TENNs can bridge various regions of the brain (blue: corticothalamic pathway, red: nigrostriatal pathway, green: entorhinal cortex to hippocampus pathway) and synapse with
host axons (purple; top right). (B) Conceptual representation of a micro-TENN forming local synapses with host neurons to form a new functional relay to replace missing or damaged
axonal tracts. (C) Confocal reconstruction of a bidirectional micro-TENN, consisting of two populations of neurons spanned by long axonal tracts within a hydrogel micro-column
stained via immunocytochemistry to denote axons (b-tubulin III; green), and cell nuclei (Hoechst; blue). (D) Confocal reconstruction of a unidirectional micro-TENN consisting of a
single neuron population (MAP2; green) extending axons (tau; red) longitudinally (adapted from (Cullen et al., 2012)). (E) Confocal reconstruction of a unidirectional micro-TENN,
stained via immunocytochemistry to denote neuronal somata/dendrites (MAP2; purple), neuronal somata/axons (tau; green) and cell nuclei (Hoechst; blue). (F) Confocal reconstruc-
tion of a transplanted GFP+ micro-TENN showing lateral outgrowth in vivo. (G) Confocal reconstruction showing GFP+ processes extending from a transplanted micro-TENN into
139

the cortex of a rat. Scale bars: 300 µm in C, 250 µm in D, 100 µm in E, 20 µm in F and G. (Reprinted from Struzyna LA, Harris JP, Katiyar KS, Chen HI, Cullen DK, with permission
from the Editorial Office of Neural Regeneration Research, Copyright 2015.)
140 Adewole et al.

FIG. 5: Micro-TENNs as “living electrodes.” (A) Unidirectional micro-TENNs (left) synapse with the host (blue
cells) and deliver inputs to targeted cortical regions, while bidirectional micro-TENNs (right) may be synapsed by
the host and transmit cortical activity. Relevant signal propagation denoted by arrows. (B) Conceptual schematic
of the micro-TENNs as “living electrodes” in vivo. Left: Input paradigm. An LED array (1) optically stimulates a
unidirectional micro-TENN with channelrhodopsin-positive neurons (2), which synapse with host Layer IV neurons
(3). Right: Output paradigm. A microelectrode array (4) records from the neurons of a bidirectional micro-TENN (5),
which are synapsed by host neurons from Layer V (6). Roman numerals denote cortical layers.

cal stimulation and recording that is not currently implanting electrodes in the otherwise non-injured
attainable with conventional NI systems. To date, brain or spinal cord. Thus, as described previously
micro-TENNs have been implanted in the rat brain, for the case of TMR and RPNI, a neural interface
and neural survival, maintenance of axonal archi- with the peripheral nerves originally serving the lost
tecture, and synaptic integration with the host have limb would be ideal, as this is the final point of mo-
been demonstrated.310 Thus, in addition to serving tor output and primary sensory input. However, PNS
as “living electrodes,” these versatile constructs axons necessitate a living target for innervation, and
may serve as living scaffolds to promote regenera- residual muscle may not be available or may not pro-
tion of host axons along the micro-TENN axons or vide the signal complexity for fine motor control.
may physically “wire-in” to replace lost host cir- Therefore, we employ a blend of tissue engineering
cuitry.314,315 Although it is actively being developed and micro-electrical techniques to facilitate direct in-
and tested, our biohybrid “living electrode” strategy tegration with host axons to allow for complex com-
may result in a neural interface with a specificity, mand signals while enabling a vehicle for proprio-
spatial density, and long-term fidelity greater than ceptive and other sensory feedback. We previously
that possible with artificial microelectronic or opti- demonstrated that these tissue engineered constructs
cal substrates alone. serve as a replacement end target and drive host axon
In the PNS, biohybrid neural interfaces are also regeneration into intimate contact with micro-elec-
promising to enable robust integration between pe- trodes.39,306 Our current efforts are aimed toward test-
ripheral axons and electronics to drive the next gen- ing the ability of these constructs to transmit electri-
eration of robotic prosthetic devices.29,39,306 In the case cal signals, and establishing the mechanism of action
of loss-of-limb, it is surgically advantageous to avoid by which they allow for integration with host axons.

Critical Reviews™ in Biomedical Engineering

Evolution of Neuroprosthetic Interfaces 141

Collectively, our biohybrid approaches operate at the Science Foundation. This material was also sup-
intersection of neuroscience and engineering to es- ported in part by the Clinical Center of the National
tablish preformed implantable neural networks as a Institutes of Health (Grant No. U01NS094340). The
complimentary alternative to conventional hardware/ content is solely the responsibility of the authors
electrode-based interfaces. Thus, these approaches and does not necessarily represent the official views
potentially represent a paradigm-shift for chronic of the National Institutes of Health. This work was
neural interface with the CNS or PNS. conducted as part of the Penn Medicine Neurosci-
ence Center at the University of Pennsylvania.
VII. CONCLUSION
REFERENCES
The nervous system connects the individual to their
environment, and its breakdown is often physical- 1. Thurston AJ. Paré and prosthetics: the early history of
ly and emotionally devastating. This is far from a artificial limbs. ANZ J Surg. 2007;77(12):1114–9.
2. Hernigou P. Ambroise Paré IV: the early history of ar-
rare issue—in 2010 more than 20% of the United tificial limbs (from robotic to prostheses). Int Orthop.
States population reported having some disability, 2013;37(6):1195–7.
with the majority related to motor impairments in 3. Grill WM, Norman SE, Bellamkonda R V. Implanted
the upper or lower body.316 Nearly 2 million people neural interfaces: biochallenges and engineered solu-
in the United States are amputees due to trauma or tions. Annu Rev Biomed Eng. 2009;11:1–24.
4. Mak JN, Wolpaw JR. Clinical applications of brain-com-
disease alone, with almost 200,000 people receiving puter interfaces: current state and future prospects. IEEE
amputations per year.10,317 The Amyotrophic Lateral Rev Biomed Eng. 2009;2:187–99.
Sclerosis Association reports a prevalence of 30,000 5. Fetz EE. Operant conditioning of cortical unit activity.
ALS patients in the United States; the number of Science (80- ). 1969 Feb 28;163(3870):955–8.
people living with other neuromuscular disorders is 6. Humphrey DR, Schmidt EM, Thompson WD. Predicting
measures of motor performance from multiple cortical
several orders of magnitude larger.316,318 Fortunately,
spike trains. Science (80- ). 1970;170(3959):758–62.
neuroprosthesis research has yielded a host of new 7. Chapin JK, Moxon KA, Markowitz RS, Nicolelis MAL.
technologies that can promote neural regeneration, Real-time control of a robot arm using simultaneously
enable communication, and restore mobility post- recorded neurons in the motor cortex. Nat Neurosci.
injury. These successes have their roots in multi- 2(7):664–70.
disciplinary research, crossing disparate domains 8. Roche JP, Hansen MR. On the horizon: cochlear
implant technology. Otolaryngol Clin North Am.
in engineering, neuroscience, and medicine. It fol- 2015;48(6):1097–116.
lows that future interfaces will likely use some com- 9. Beitz JM. Parkinson’s disease: a review. Front Biosci.
bination of both established and newly developed 2014;(3):65–74.
strategies to address the challenges faced by current 10. Resnik L, Meucci MR, Lieberman-Klinger S, Fantini C,
interfaces—immunoreactivity, chronic stability, and Kelty DL, Disla R, Sasson N. Advanced upper limb prost-
hetic devices: Implications for upper limb prosthetic reha-
selectivity, among others—to integrate more seam-
bilitation. Arch Phys Med Rehabil. 2012;93(4):710–7.
lessly with the human body. As NIs improve, pros- 11. Resnik L, Fantini C, Disla R, Etter K, Lieberman S.
thesis adoption will increase accordingly as patients Using a virtual reality environment (VRE) to facilitate
feel more in tune with their devices. Concurrently, training with an advanced upper limb prosthesis. 2011;
ongoing research in this space will lead to new ap- 2–3.
plications and platforms for rehabilitation, therapeu- 12. Resnik L, Borgia M. User ratings of prosthetic usability
and satisfaction in VA study to optimize DEKA arm. J
tics, and the restoration of quality of life. Rehabil Res Dev. 2014;51(1):15–26.
13. Kung T, Bueno R, Alkhalefah GK, Langhals NB, Ur-
ACKNOWLEDGMENTS banchek MG, Cederna PS. Innovations in prosthetic
interfaces for the upper extremity. Plast Reconstr Surg.
This material is based upon work supported by the 2013;132(6):1515–23.
National Science Foundation Graduate Research 14. Biddiss EA, Chau TT. Upper limb prosthesis use and
abandonment: a survey of the last 25 years. Prosthet Ort-
Fellowship (Grant No. DGE1321851). Any opinion,
hot Int. 2007;31(3):236–57.
findings, and conclusions or recommendations ex- 15. Biddiss E, Chau T. Upper-limb prosthetics: critical fac-
pressed in this material are those of the authors(s) and tors in device abandonment. Am J Phys Med Rehabil.
do not necessarily reflect the views of the National 2007;86(12):977–87.

Volume 44, Issues 1–2, 2016

142 Adewole et al.

16. Schultz AE, Kuiken TA. Neural interfaces for control of 34. Polikov VS, Tresco PA, Reichert WM. Response of brain
upper limb prostheses: the state of the art and future pos- tissue to chronically implanted neural electrodes. J Neu-
sibilities. PM R. 2011;3(1):55–67. rosci Methods. 2005;148(1):1–18.
17. Lega BC, Serruya MD, Zaghloul K a. Brain-machine in- 35. Tresco PA, Winslow BD. The challenge of integrating
terfaces: electrophysiological challenges and limitations. devices into the central nervous system. Crit Rev Biomed
Crit Rev Biomed Eng. 2011;39(1):5–28. Eng. 2011;39(1):29–44.
18. Cullen DK, Pfister B. State of the art and future challen- 36. Bhattacharyya S, Konar A, Tibarewala DN. A differential
ges in neural engineering: strategies for nervous system evolution based energy trajectory planner for artificial
repair foreword / editors’ commentary (volume 2). Crit limb control using motor imagery EEG signal. Biomed
Rev Biomed Eng. 2011;39(2):79–80. Signal Process Control. 2014;11(1):107–13.
19. Wolpaw JR. Brain-computer interfaces. 1st ed. Vol. 110, 37. Daly JJ, Wolpaw JR. Brain-computer interfaces in neu-
Handbook of clinical neurology. Elsevier; 2013. Pp. rological rehabilitation. Lancet Neurol. 2008;7(11):
67–74. 1032–43.
20. Berne RM, Koeppen BM, Stanton BA. Berne & Levy 38. Galán F, Nuttin M, Lew E, Ferrez PW, Vanacker G,
physiology. Philadelphia, PA: Mosby/Elsevier; 2010. Philips J, Millán Jdel R. A brain-actuated wheelchair:
21. Cogan SF. Neural stimulation and recording electrodes. asynchronous and noninvasive brain–computer interfa-
Annu Rev Biomed Eng. 2008;10:275–309. ces for continuous control of robots. Clin Neurophysiol.
22. Pisotta I, Perruchoud D, Ionta S. Hand-in-hand advances 2008;119(9):2159–69.
in biomedical engineering and sensorimotor restoration. 39. Kameswaran N, Cullen DK, Pfister BJ, Ranalli NJ,
J Neurosci Methods. 2015;246:22–9. Huang JH, Zager EL, Smith DH. A novel neuroprost-
23. Shimada S, Fukuda K, Hiraki K. Rubber hand illusion hetic interface with the peripheral nervous system
under delayed visual feedback. PLoS One. 2009;4(7) using artificially engineered axonal tracts. Neurol Res.
1–5. 2008;30(10):1063–7.
24. Hatsopoulos N, Donoghue J. The science of neural inter- 40. Kaufmann T, Herweg A, Kübler A. Toward brain-compu-
face systems. Annu Rev Neurosci. 2009;32:249–66. ter interface based wheelchair control utilizing tactually-
25. Wolpaw JR, Birbaumer N, McFarland DJ, Pfurtscheller evoked event-related potentials. J Neuroeng Rehabil.
G, Vaughan TM. Brain-computer interfaces for commu- 2014;11:7.
nication and control. Clin Neurophysiol. 2002;113(6): 41. Lebedev MA, Nicolelis M a L. Brain-machine in-
767–91. terfaces: past, present and future. Trends Neurosci.
26. Benabid AL, Chabardes S, Mitrofanis J, Pollak P. Deep 2006;29(9):536–46.
brain stimulation of the subthalamic nucleus for the treat- 42. Rossini PM, Micera S, Benvenuto A, Carpaneto J, Ca-
ment of Parkinson’s disease. Lancet Neurol. 2009;8(1):6 vallo G, Citi L, Cipriani C, Denaro L, Denaro V, Di Pino
7–81. G, Ferreri F, Guglielmelli E, Hoffmann KP, Raspopovic
27. Berenyi A, Belluscio M, Mao D, Buzsaki G. Closed-loop S, Rigosa J, Rossini L, Tombini M, Dario P. Double ner-
control of epilepsy by transcranial electrical stimulation. ve intraneural interface implant on a human amputee for
Science (80- ). 2012;337(6095):735–7. robotic hand control. Clin Neurophysiol. 2010;121(5):
28. Bronstein JM, Tagliati M, Alterman RL, Lozano AM, 777–83.
Volkmann J, Stefani A, Horak FB, Okun MS, Foote KD, 43. Yu T, Li Y, Long J, Li F. A hybrid brain-computer in-
Krack P, Pahwa R, Henderson JM, Hariz MI, Bakay RA, terface-based mail client. Comput Math Methods Med.
Rezai A, Marks WJ Jr, Moro E, Vitek JL, Weaver FM, 2013;2013.
Gross RE, DeLong MR. Deep brain stimulation for par- 44. Hampson RE, Song D, Opris I, Santos LM, Shin DC,
kinson disease: an expert consensus and review of key Gerhardt GA, Marmarelis VZ, Berger TW, Deadwyler
issues. Arch Neurol. 2010;7:199–203. SA. Facilitation of memory encoding in primate hippo-
29. Cullen DK, Smith DH. Bionic connections. Sci Amer. campus by a neuroprosthesis that promotes task specific
2013;308(1):52–7. neural firing. J Neural Eng. 2013;10(6):66013.
30. Histed MH, Bonin V, Reid RC. Direct activation of spar- 45. Berger TW, Hampson RE, Song D, Goonawardena A,
se, distributed populations of cortical neurons by electri- Marmarelis VZ, Deadwyler SA. A cortical neural prost-
cal microstimulation. Neuron. 2009;63(4):508–22. hesis for restoring and enhancing memory. J Neural Eng.
31. Logothetis NK, Augath M, Murayama Y, Rauch A, Sul- 2011;8(4):46017.
tan F, Goense J, Oeltermann A, Merkle H, The ef- 46. Cogan SF, Ludwig KA, Welle CG, Takmakov P. Tissue
fects of electrical microstimulation on cortical signal damage thresholds during therapeutic electrical stimula-
propagation. Nat Neurosci. 2010;13(10):1283–91. tion. J Neural Eng. 2016;13(2):021001.
32. Fukaya C, Yamamoto T. Deep brain stimulation for 47. Krüger J, Caruana F, Volta RD, Rizzolatti G. Seven ye-
Parkinson’s disease: recent trends and future direction. ars of recording from monkey cortex with a chronically
Neurol Med Chir (Tokyo). 2015;55(5):422–31. implanted multiple microelectrode. Front Neuroeng.
33. Harris JP, Tyler DJ. Biological, mechanical, and tech- 2010;3:6.
nological considerations affecting the longevity of in- 48. Delbeke J. Electrodes and chronic optic nerve stimula-
tracortical electrode recordings. Crit Rev Biomed Eng. tion. Biocybern Biomed Eng. 2011;31(3):81–94.
2013;41(6):435–56. 49. Waters RL, McNeal DR, Faloon W, Clifford B. Functio-

Critical Reviews™ in Biomedical Engineering

Evolution of Neuroprosthetic Interfaces 143

nal electrical stimulation of the peroneal nerve for hemi- meral JD, Newell K, Oakley EM, Cash SS, Friehs G,
plegia. Long-term clinical follow-up. J Bone Joint Surg Hochberg LR. Neural point-and-click communication by
Am. 1985;67(5):792–3. a person with incomplete locked-in syndrome. Neurore-
50. Singh S, Kong Y-Y, Zeng F-G. Cochlear implant melody habil Neural Repair. 2015;29(5):462–71.
recognition as a function of melody frequency range, 64. Simeral JD, Kim S-P, Black MJ, Donoghue JP, Hoch-
harmonicity, and number of electrodes. Ear Hearing. berg LR. Neural control of cursor trajectory and click
2009;30:160–8. by a human with tetraplegia 1000 days after implant
51. Peterson NR, Pisoni DB, Miyamoto RT. Cochlear im- of an intracortical microelectrode array. J Neural Eng.
plants and spoken language processing abilities: review 2011;8(2):25027.
and assessment of the literature. Restor Neurol Neurosci. 65. Bouton CE, Shaikhouni A, Annetta N V., Bockbrader
2010;28(2):237–50. MA, Friedenberg DA, Nielson DM, Sharma G, Seder-
52. Kong Y-Y, Cruz R, Jones JA, Zeng F-G. Music percep- berg PB, Glenn BC, Mysiw WJ, Morgan AG, Deoga-
tion with temporal cues in acoustic and electric hearing. onkar M, Rezai AR. Restoring cortical control of func-
Ear Hear. 2004;25(2):173–85. tional movement in a human with quadriplegia. Nature.
53. Barrese JC, Rao N, Paroo K, Triebwasser C, Vargas-Irwin 2016;advance on:1–13.
C, Franquemont L, Donoghue JP. Failure mode analysis 66. Moritz CT, Perlmutter SI, Fetz EE. Direct control
of silicon-based intracortical microelectrode arrays in of paralysed muscles by cortical neurons. Nature.
non-human primates. J Neural Eng. 2013;10(6):66014. 2008;456(7222):639–42.
54. Prasad A, Xue QS, Sankar V, Nishida T, Shaw G, Streit 67. Pasquina PF, Evangelista M, Carvalho AJ, Lockhart J,
WJ, Sanchez JC. Comprehensive characterization and Griffin S, Nanos G, McKay P, Hansen M, Ipsen D, Van-
failure modes of tungsten microwire arrays in chronic dersea J, Butkus J, Miller M, Murphy I, Hankin D.First-
neural implants. J Neural Eng. 2012;9(5):056015. in-man demonstration of a fully implanted myoelectric
55. Prasad A, Xue QS, Dieme R, Sankar V, Mayrand RC, sensors system to control an advanced electromecha-
Nishida T, Streit WJ, Sanchez JC. Abiotic-biotic charac- nical prosthetic hand. J Neurosci Methods. 2014;244:
terization of Pt/Ir microelectrode arrays in chronic im- 85–93.
plants. Front Neuroeng. 2014;7(February):2. 68. Baker JJ, Yatsenko D, Schorsch JF, DeMichele GA,
56. Ravikumar M, Sunil S, Black J, Barkauskas DS, Haung Troyk PR, Hutchinson DT, Weir RF, Clark G, Greger B.
AY, Miller RH, Selkirk SM, Capadona JR. The roles of Decoding individuated finger flexions with implantable
blood-derived macrophages and resident microglia in the myoelectric sensors. Conf Proc IEEE Eng Med Biol Soc.
neuroinflammatory response to implanted Intracortical 2008;2008:193–6.
microelectrodes. Biomater. 2014;35:8049–64. 69. Schlatter S, Engineering B, Island R. Implantable myo-
57. Liu X, McCreery DB, Carter RR, Bullara LA, Yuen electric sensors for intramuscular EMG recording. IEEE
TGH, Agnew WF. Stability of the interface between neu- Trans Biomed Eng. 2009;56(1):2009.
ral tissue and chronically implanted intracortical micro- 70. Weber DJ, Bauman MJ, Fisher L. Spinal nerve interfaces
electrodes. IEEE Trans Rehabil Eng. 1999;7(3):315–26. for bidirectional communication with prosthetic limbs.
58. McConnell GC, Rees HD, Levey AI, Gutekunst C-A, EmbcEmbsOrg. 2012;485(7398):7398.
Gross RE, Bellamkonda RV. Implanted neural elec- 71. Navarro X, Krueger TB, Lago N, Micera S, Stieglitz T,
trodes cause chronic, local inflammation that is cor- Dario P. A critical review of interfaces with the periphe-
related with local neurodegeneration. J Neural Eng. ral nervous system for the control of neuroprotheses and
2009;6(5):056003. hybrid bionic systems. J Peripher Nerv Syst. 2005;10:
59. Gunasekera B, Saxena T, Bellamkonda R, Karumbaiah 229–58.
L. Intracortical Recording Interfaces: Current Challenges 72. Kuiken T, Li G, Lock B, Lipschutz RD, Miller L, Stub-
to Chronic Recording Function. ACS Chem Neurosci. blefield K, Englehart K. Targeted muscle reinnervation
2015; for real-time myoelectric control of multifunction artifi-
60. Gilja V, Chestek CA., Diester I, Henderson JM, Deis- cial arms. JAMA. 2009;301(6):619–28.
seroth K, Shenoy KV. Challenges and opportunities for 73. Marasco PD, Schultz AE, Kuiken TA. Sensory capacity
next-generation intracortically based neural prostheses. of reinnervated skin after redirection of amputated upper
IEEE Trans Biomed Eng. 2011;58(7):1891–9. limb nerves to the chest. Brain. 2009;132(6):1441–8.
61. Hochberg LR, Serruya MD, Friehs GM, Mukand JA, 74. Hebert JS, Olson JL, Morhart MJ, Dawson MR, Marasco
Saleh M, Caplan AH, Branner A, Chen D, Penn RD, PD, Kuiken TA, Ming Chan K. Novel targeted sensory
Donoghue JP. Neuronal ensemble control of prost- reinnervation technique to restore functional hand sen-
hetic devices by a human with tetraplegia. Nature. sation after transhumeral amputation. IEEE Trans Neural
2006;442(7099):164–71. Syst Rehabil Eng. 2014;22(4):765–73.
62. Gilja V, Pandarinath C, Blabe CH, Nuyujukian P, Si- 75. Pfister BJ, Gordon T, Loverde JR, Kochar AS, Mackin-
meral JD, Sarma AA, Sorice BL, Perge JA, Jarosiewicz non SE, Cullen DK. Biomedical engineering strategies
B, Hochberg LR, Shenoy KV, Henderson JM. Clinical for peripheral nerve repair: surgical applications, state
translation of a high-performance neural prosthesis. Nat of the art, and future challenges. Crit Rev Biomed Eng.
Med. 2015;21(10):6–8. 2011;39(2):81–124.
63. Bacher D, Jarosiewicz B, Masse NY, Stavisky SD, Si- 76. Allodi I, Udina E, Navarro X. Specificity of peripheral

Volume 44, Issues 1–2, 2016

144 Adewole et al.

nerve regeneration: interactions at the axon level. Prog 91. Rejc E, Angeli C, Harkema S. Effects of lumbosacral spi-
Neurobiol. 2012;98(1):16–37. nal cord epidural stimulation for standing after chronic
77. Di Pino G, Denaro L, Vadalà G, Marinozzi A, Tombini complete paralysis in humans. PLoS One. 2015;10(7):
M, Ferreri F, Papalia R, Accoto D, Guglielmelli E, Di 1–20.
Lazzaro V, Denaro V. Invasive neural interfaces: the per- 92. Harkema S, Gerasimenko Y, Hodes J, Burdick J, Angeli
spective of the surgeon. J Surg Res. 2014;188(1):77–87. C, Chen Y, Ferreira C, Willhite A, Rejc E, Grossman RG,
78. Cincotti F, Mattia D, Aloise F, Bufalari S, Astolfi L, De Edgerton VR. Effect of epidural stimulation of the lum-
Vico Fallani F, Tocci A, Bianchi L, Marciani MG, Gao bosacral spinal cord on voluntary movement, standing,
S, Millan J, Babiloni F. High-resolution EEG techniques and assisted stepping after motor complete paraplegia: a
for brain-computer interface applications. J Neurosci case study. Lancet. 2011;377(9781):1938–47.
Methods. 2008;167(1):31–42. 93. Carmena JM, Cohen LG. Brain–machine interfaces and
79. McCane LM, Sellers EW, McFarland DJ, Mak JN, Car- transcranial stimulation. future implications for directing
mack CS, Zeitlin D, Wolpaw JR, Vaughan TM. Brain- functional movement and improving function after spinal
computer interface (BCI) evaluation in people with injury in humans. 1st ed. Vol. 109, Handbook of clinical
amyotrophic lateral sclerosis. Amyotroph lateral scler neurology. Elsevier; 2012. Pp. 435–44.
frontotemporal degener. 2014;Nov 2013:1–9. 94. Soekadar SR, Witkowski M, Cossio EG, Birbaumer N,
80. Nijboer F, Furdea A, Gunst I, Mellinger J, McFarland DJ, Cohen LG. Learned EEG-based brain self-regulation of
Birbaumer N, Kübler A. An auditory brain-computer in- motor-related oscillations during application of transcra-
terface (BCI). J Neurosci Methods. 2008;167(1):43–50. nial electric brain stimulation: feasibility and limitations.
81. Sellers EW, Vaughan TM, Wolpaw JR. A brain-computer Front Behav Neurosci. 2014;8:93.
interface for long-term independent home use. Amyo- 95. Dutta A, Paulus W, Nitsche MA. Facilitating myoelec-
troph Lateral Scler. 2010;11(5):449–55. tric-control with transcranial direct current stimulation: a
82. Taylor DM, Tillery SIH, Schwartz AB. Direct corti- preliminary study in healthy humans. J Neuroeng Reha-
cal control of 3D neuroprosthetic devices. Science. bil. 2014;11(1):13.
2002;296(5574):1829–32. 96. Birbaumer N, Cohen LG. Brain-computer interfaces:
83. Wolpaw JR, McFarland DJ. Control of a two-dimen- communication and restoration of movement in paraly-
sional movement signal by a noninvasive brain-com- sis. J Physiol. 2007;579(Pt 3):621–36.
puter interface in humans. Proc Natl Acad Sci USA. 97. Donoghue JP. Bridging the brain to the world: a perspec-
2004;101(51):17849–54. tive on neural interface systems. Neuron. 2008;60(3):
84. Boord P, Barriskill A, Craig A, Nguyen H. Brain-com- 511–21.
puter interface—FES integration: towards a hands-free 98. Piccione F, Giorgi F, Tonin P, Priftis K, Giove S, Silvoni
neuroprosthesis command system. Neuromodulation. S, Palmas G, Beverina F. P300-based brain computer
2004;7(4):267–76. interface: reliability and performance in healthy and pa-
85. Contreras-Vidal JL, Grossman RG. NeuroRex: A clinical ralysed participants. Clin Neurophysiol. 2006;117(3):
neural interface roadmap for EEG-based brain machine 531–7.
interfaces to a lower body robotic exoskeleton. Proc Annu 99. Neuper C, Scherer R, Wriessnegger S, Pfurtscheller
Int Conf IEEE Eng Med Biol Soc EMBS. 2013;1579–82. G. Motor imagery and action observation: Modula-
86. Pfurtscheller G, Guger C, Müller G, Krausz G, Neuper C. tion of sensorimotor brain rhythms during mental con-
Brain oscillations control hand orthosis in a tetraplegic. trol of a brain-computer interface. Clin Neurophysiol.
Neurosci Lett. 2000;292(3):211–4. 2009;120(2):239–47.
87. Millán JDR, Rupp R, Müller-Putz GR, Murray-Smith 100. Pfurtscheller G, Müller GR, Pfurtscheller J, Gerner HJ,
R, Giugliemma C, Tangermann M, Vidaurre C, Cincotti Rupp R. Thought—control of functional electrical stimu-
F, Kübler A, Leeb R, Neuper C, Müller K-R, Mattia D. lation to restore hand grasp in a patient with tetraplegia.
Combining brain-computer interfaces and assistive tech- Neurosci Lett. 2003;351(1):33–6.
nologies: State-of-the-art and challenges. Front Neuros- 101. Müller-Putz GR, Scherer R, Pfurtscheller G, Rupp R.
ci. 2010;4:1–15. EEG-based neuroprosthesis control: a step towards clini-
88. Lobel DA, Lee KH. Brain machine interface and limb cal practice. Neurosci Lett. 2005;382(1–2):169–74.
reanimation technologies: Restoring function after spi- 102. Long J, Li Y, Wang H, Yu T, Pan J. Control of a simulated
nal cord injury through development of a bypass system. wheelchair based on a hybrid brain computer interface.
Mayo Clin Proc. 2014;89(5):708–14. Engineering in Medicine and Biology Society (EMBC),
89. Zimmermann JB, Jackson A. Closed-loop control of spi- 2012 Ann Int Conf IEEE. 2012. p. 6727–30.
nal cord stimulation to restore hand function after paraly- 103. Chen X, Wang Y, Nakanishi M, Gao X, Jung T-P, Gao S.
sis. Front Neurosci. 2014;8:1–8. High-speed spelling with a noninvasive brain–computer
90. Grahn PJ, Mallory GW, Michael Berry B, Hachmann interface. Proc Natl Acad Sci. 2015;112 (44 ):E6058–67.
JT, Lobel DA, Luis Lujan J. Restoration of motor func- 104. Patil PG, Turner DA. The development of brain-machine
tion following spinal cord injury via optimal control interface neuroprosthetic devices. Neurotherapeutics.
of intraspinal microstimulation: toward a next gene- 2008;5(1):137–46.
ration closed-loop neural prosthesis. Front Neurosci. 105. Wolpaw JR. Brain–computer interface. Encycl Neurosci.
2014;8(SEP):1–12. 2009;429–37.

Critical Reviews™ in Biomedical Engineering

Evolution of Neuroprosthetic Interfaces 145

106. Blakely T, Miller KJ, Zanos SP, Rao RPN, Ojemann JG. electrodes: a qualitative and quantitative report. J Neural
Robust, long-term control of an electrocorticographic Eng. 2013;10(3):036021.
brain-computer interface with fixed parameters. Neu- 121. Sun FT, Morrell MJ. The RNS System: responsive cor-
rosurgical Focus. 2009; 27:E13. tical stimulation for the treatment of refractory partial
107. Schalk G, McFarland DJ, Hinterberger T, Birbaumer N, epilepsy. Expert Rev Med Devices. 2014;11(6):563–72.
Wolpaw JR. BCI2000: A general-purpose brain-com- 122. Chao ZC, Nagasaka Y, Fujii N. Long-term asynchronous
puter interface (BCI) system. IEEE Trans Biomed Eng. decoding of arm motion using electrocorticographic sig-
2004;51(6):1034–43. nals in monkeys. Front Neuroeng. 2010;3:3.
108. Cincotti F, Mattia D, Aloise F, Bufalari S, Schalk G, 123. Rubehn B, Bosman C, Oostenveld R, Fries P, Stieglitz T.
Oriolo G, Cherubini A, Marciani MG, Babiloni F. Non- A MEMS-based flexible multichannel ECoG-electrode
invasive brain-computer interface system: towards its array. J Neural Eng. 2009;6(3):036003.
application as assistive technology. Brain Res Bull. 124. Castagnola E, Ansaldo A, Maggiolini E, Ius T, Skrap
2008;75(6):796–803. M, Ricci D, Fadiga L. Smaller, softer, lower-impedance
109. Scherberger H. Neural control of motor prostheses. Curr electrodes for human neuroprosthesis: a pragmatic ap-
Opin Neurobiol. 2009;19(6):629–33. proach. Front Neuroeng. 2014;7:8.
110. Schwartz AB, Cui XT, Weber DJ, Moran DW. Brain- 125. Thongpang S, Richner TJ, Brodnick SK, Schendel A,
controlled interfaces: movement restoration with neural Kim J, Wilson JA, Hippensteel J, Krugner-Higby L, Mo-
prosthetics. Neuron. 2006;52(1):205–20. ran D, Ahmed AS, Neimann D, Sillay K, Williams JC.
111. Lachaux JP, Rudrauf D, Kahane P. Intracranial EEG A micro-electrocorticography platform and deployment
and human brain mapping. J Physiol Paris. 2003;97(4- strategies for chronic BCI applications. Clin EEG Neu-
6):613–28. rosci. 2011;42(4):259–65.
112. Leuthardt EC, Schalk G, Roland J, Rouse A, Moran 126. Schendel AA, Thongpang S, Brodnick SK, Richner
DW. Evolution of brain-computer interfaces: going TJ, Lindevig BD, Krugner-Higby L, Williams JC. A
beyond classic motor physiology. Neurosurg Focus. cranial window imaging method for monitoring vascu-
2009;27(1):E4–E4. lar growth around chronically implanted micro-ECoG
113. Schalk G, Miller KJ, Anderson NR, Wilson JA, Smyth devices. J Neurosci Methods. 2013;218(1):10.1016/j.
MD, Ojemann JG, Moran DW, Wolpaw JR, Leuthardt jneumeth.2013.06.001.
EC. Two-dimensional movement control using elec- 127. Henle C, Raab M, Cordeiro JG, Doostkam S, Schulze-
trocorticographic signals in humans. J Neural Eng. Bonhage A, Stieglitz T, Rickert J. First long term in vivo
2008;5(1):75–84. study on subdurally implanted Micro-ECoG electrodes,
114. Leuthardt EC, Schalk G, Wolpaw JR, Ojemann JG, Mo- manufactured with a novel laser technology. Biomed Mi-
ran DW. A brain-computer interface using electrocorti- crodevices. 2011;13(1):59–68.
cographic signals in humans. J Neural Eng. 2004;1(2): 128. Mestais CS, Charvet G, Sauter-Starace F, Foerster M,
63–71. Ratel D, Benabid AL. WIMAGINE: Wireless 64-chan-
115. Schalk G, Leuthardt EC. Brain-computer interfaces nel ECoG recording implant for long term clinical
using electrocorticographic signals. IEEE Rev Biomed applications. IEEE Trans Neural Syst Rehabil Eng.
Eng. 2011;4:140–54. 2015;23(1):10–21.
116. Hill NJ, Gupta D, Brunner P, Gunduz A, Adamo MA, 129. Akamatsu N, Tsuji S. Deep brain stimulation for epi-
Ritaccio A, Schalk G. Recording human electrocorti- lepsy. Brain Nerve. 2011;63(4):365–9.
cographic (ECoG) signals for neuroscientific research 130. Nicolelis MA, Dimitrov D, Carmena JM, Crist R, Lehew
and real-time functional cortical mapping. J Vis Exp. G, Kralik JD, Wise SP. Chronic, multisite, multielectrode
2012;(64):1–5. recordings in macaque monkeys. Proc Natl Acad Sci U S
117. Panov F, Levin E, de Hemptinne C, Swann NC, Qasim S, A. 2003;100(19):11041–6.
Miocinovic S, Ostrem JL, Starr PA. Intraoperative elec- 131. Benabid AL. Deep brain stimulation for Parkinson’s
trocorticography for physiological research in movement disease. Curr Opin Neurobiol. 2003;13(6):696–706.
disorders: principles and experience in 200 cases. J Neu- 132. Klaes C, Shi Y, Kellis S, Minxha J, Revechkis B, An-
rosurg. 2016;1–10. dersen RA. A cognitive neuroprosthetic that uses cortical
118. Wang W, Collinger JL, Degenhart AD, Tyler-Kabara stimulation for somatosensory feedback. J Neural Eng.
EC, Schwartz AB, Moran DW, Weber DJ, Wodlinger B, 2014;11:56024.
Vinjamuri RK, Ashmore RC, Kelly JW, Boninger ML. 133. Tabot GA, Dammann JF, Berg JA, Tenore F V, Boback
An electrocorticographic brain interface in an individual JL, Vogelstein RJ, Bensmaia SJ. Restoring the sense of
with tetraplegia. PLoS One. 2013;8(2):1–8. touch with a prosthetic hand through a brain interface.
119. Yanagisawa T, Hirata M, Saitoh Y, Goto T, Kishima H, Proc Natl Acad Sci USA. 2013;110(45):18279–84.
Fukuma R, Yokoi H, Kamitani Y, Yoshimine T. Real-time 134. O’Doherty JE, Lebedev MA, Ifft PJ, Zhuang KZ, Sho-
control of a prosthetic hand using human electrocortico- kur S, Bleuler H, Nicolelis MAL. Active tactile explo-
graphy signals. J Neurosurg. 2011;114(6):1715–22. ration using a brain–machine–brain interface. Nature.
120. Johnson LA, Wander JD, Sarma D, Su DK, Fetz EE, 2011;479(7372):228–31.
Ojemann JG. Direct electrical stimulation of the soma- 135. Lim HH, Lenarz M, Lenarz T. Auditory midbrain im-
tosensory cortex in humans using electrocorticography plant: a review. Trends Amplif. 2009;13(3):149–80.

Volume 44, Issues 1–2, 2016

146 Adewole et al.

136. Humayun MS, Weiland JD, Fujii GY, Greenberg R, Wil- 151. McClain MA, Clements IP, Shafer RH, Bellamkonda
liamson R, Little J, Mech B, Cimmarusti V, Van Boemel RV, LaPlaca MC, Allen MG. Highly compliant, micro-
G, Dagnelie G, de Juan E. Visual perception in a blind cable neuroelectrodes fabricated from thin-film gold and
subject with a chronic microelectronic retinal prosthesis. PDMS. Biomed Microdevices. 2011;13(2):361–73.
Vision Res. 2003;43(24):2573–81. 152. Bishop W, Chestek CC, Gilja V, Nuyujukian P, Foster JD,
137. Pezaris JS, Eskandar EN. Getting signals into the brain: Ryu SI, Shenoy KV, Yu BM. Self-recalibrating classifiers
visual prosthetics through thalamic microstimulation. for intracortical brain-computer interfaces. J Neural Eng.
Neurosurg Focus. 2009;27(1):E6–E6. 2014;11(2):026001.
138. Barry MP, Dagnelie G, Group AIIS. Use of the Ar- 153. Homer ML, Perge JA., Black MJ, Harrison MT, Cash SS,
gus II retinal prosthesis to improve visual guidance Hochberg LR. Adaptive offset correction for intracortical
of fine hand movements. Invest Ophthalmol Vis Sci. brain-computer interfaces. IEEE Trans Neural Syst Re-
2012;53(9):5095–101. habil Eng. 2014;22(2):239–48.
139. Yanai D, Weiland JD, Mahadevappa M, Greenberg RJ, 154. Ajiboye AB, Simeral JD, Donoghue JP, Hochberg LR,
Fine I, Humayun MS. Visual performance using a retinal Kirsch RF. Prediction of imagined single-joint move-
prosthesis in three subjects with retinitis pigmentosa. Am ments in a person with high-level tetraplegia. IEEE Trans
J Ophthalmol. 2007;143(5):820–7. Biomed Eng. 2012;59(10):2755–65.
140. Zrenner E, Bartz-Schmidt KU, Benav H, Besch D, 155. Murphy MD, Guggenmos DJ, Bundy DT, Nudo RJ. Cur-
Bruckmann A, Gabel V-P, Gekeler F, Greppmaier U, rent challenges facing the translation of brain computer
Harscher A, Kibbel S, Koch J, Kusnyerik A, Peters T, interfaces from preclinical trials to use in human patients.
Stingl K, Sachs H, Stett A, Szurman P, Wilhelm B, Wilke Front Cell Neurosci. 2015;9:1–14.
R. Subretinal electronic chips allow blind patients to 156. Howard IP, Rogers BJ. Binocular vision and stereopsis.
read letters and combine them to words. Proc Biol Sci. Oxford psychology series. New York: Oxford University
2011;278(1711):1489–97. Press; 1996. 746 p.
141. Stronks HC, Dagnelie G. The functional performance of 157. Spector RH. Visual Fields. In: Walker H, Hall W, Hurst J,
the Argus II retinal prosthesis. Expert Rev Med Devices. editors. Clinical methods: the history, physical, and labo-
2014;11(1):23–30. ratory examinations. 3rd ed. Boston: Butterworths; 1990.
142. Carmena JM. Stable Ensemble performance with single- Pp. 565–72.
neuron variability during reaching movements in prima- 158. DARPA. DARPA-BAA-16-09: Neural Engineering Sys-
tes. J Neurosci. 2005;25(46):10712–6. tem Design (NESD). DARPA; 2016.
143. Dickey AS, Suminski A, Amit Y, Hatsopoulos NG. Sin- 159. Angotzi GN, Boi F, Zordan S, Bonfanti A, Vato A. A pro-
gle-unit stability using chronically implanted multielec- grammable closed-loop recording and stimulating wire-
trode arrays. J Neurophysiol. 2009;102(2):1331–9. less system for behaving small laboratory animals. Sci
144. Rousche PJ, Normann RA. Chronic recording capability Rep. 2014;4:5963.
of the utah intracortical electrode array in cat sensory 160. Vato A, Szymanski FD, Semprini M, Mussa-Ivaldi FA.,
cortex. J Neurosci Methods. 1998;82(1):1–15. Panzeri S. A bidirectional brain-machine interface algo-
145. Harris AR, Morgan SJ, Chen J, Kapsa RMI, Wallace GG, rithm that approximates arbitrary force-fields. PLoS One.
Paolini AG. Conducting polymer coated neural recording 2014;9(3).
electrodes. J Neural Eng. 2013;10(1):016004. 161. Fetz EE. Restoring motor function with bidirectional
146. Shain W1, Spataro L, Dilgen J, Haverstick K, Retterer S, neural interfaces. 1st ed. Vol. 218, Sensorimotor rehabi-
Isaacson M, Saltzman M, Turner JN. Controlling cellular litation: at the crossroads of basic and clinical sciences.
reactive responses around neural prosthetic devices using Elsevier B.V.; 2015. Pp. 241–252.
peripheral and local intervention strategies. IEEE Trans 162. Mavoori J, Jackson A, Diorio C, Fetz E. An autonomous
Neural Syst Rehabil Eng. 2003;11(2):186–8. implantable computer for neural recording and stimu-
147. George PM1, Lyckman AW, LaVan DA, Hegde A, Le- lation in unrestrained primates. J Neurosci Methods.
ung Y, Avasare R, Testa C, Alexander PM, Langer R, 2005;148(1):71–7.
Sur M. Fabrication and biocompatibility of polypyrrole 163. Zanos S, Richardson AG, Shupe L, Miles FP, Fetz EE.
implants suitable for neural prosthetics. Biomaterials. The neurochip-2: an autonomous head-fixed computer
2005;26(17):3511–9. for recording and stimulating in freely behaving mon-
148. Richardson-Burns SM, Hendricks JL, Foster B, Pov- keys. IEEE Trans Neural Syst Rehabil Eng. 2011;19(4):
lich LK, Kim DH, Martin DC. Polymerization of the 427–35.
conducting polymer poly(3,4-ethylenedioxythiophe- 164. Stacey WC, Litt B. Devices for Seizure Control.
ne) (PEDOT) around living neural cells. Biomater. 2008;4(4):190–201.
2007;28(8):1539–52. 165. D’Alessandro M, Esteller R, Vachtsevanos G, Hinson
149. Green RA., Lovell NH, Poole-Warren LA. Cell attach- A, Echauz J, Litt B. Epileptic seizure prediction using
ment functionality of bioactive conducting polymers for hybrid feature selection over multiple intracranial EEG
neural interfaces. Biomater. 2009;30(22):3637–44. electrode contacts: a report of four patients. IEEE Trans
150. Bae WJ, Ruddy BP, Richardson AG, Hunter IW, Bizzi E. Biomed Eng. 2003;50(5):603–15.
Cortical recording with polypyrrole microwire electrodes. 166. Howbert JJ, Patterson EE, Stead SM, Brinkmann B, Va-
Conf Proc IEEE Eng Med Biol Soc. 2008;2008:5794–7. soli V, Crepeau D, Vite CH, Sturges B, Ruedebusch V,

Critical Reviews™ in Biomedical Engineering

Evolution of Neuroprosthetic Interfaces 147

Mavoori J, Leyde K, Sheffield WD, Litt B, Worrell GA. 179. Langhals NB, Woo SL, Moon JD, Larson JV, Leach
Forecasting seizures in dogs with naturally occurring MK, Cederna PS, Urbanchek MG. Electrically stimu-
epilepsy. PLoS One. 2014;9(1). lated signals from a long-term regenerative peripheral
167. Schalk G. Can Electrocorticography (ECoG) support ro- nerve interface. Conf Proc IEEE Eng Med Biol Soc.
bust and powerful brain-computer interfaces? Front Neu- 2014;48109:1989–92.
roeng. 2010;3:9. 180. Kung TA, Langhals NB, Martin DC, Johnson PJ, Cederna
168. Hochberg LR, Bacher D, Jarosiewicz B, Masse NY, Si- PS, Urbanchek MG. Regenerative peripheral nerve inter-
meral JD, Vogel J, Haddadin S, Liu J, Cash SS, van der face viability and signal transduction with an implanted
Smagt P, Donoghue JP. Reach and grasp by people with electrode. Plast Reconstr Surg. 2014;133(6):1380–94.
tetraplegia using a neurally controlled robotic arm. Na- 181. Sando IC, Leach MK, Woo SL, Moon JD, Cederna
ture. 2012;485(7398):372–5. PS, Langhals NB, Urbanchek MG Regenerative perip-
169. Kuzum D, Takano H, Shim E, Reed JC, Juul H, Ric- heral nerve interface for prostheses control: electrode
hardson AG, de Vries J, Bink H, Dichter MA, Lucas comparison. J Reconstr Microsurg. 26.10.2015 ed.
TH, Coulter DA, Cubukcu E, Litt B. Transparent and 2016;32(03):194–9.
flexible low noise graphene electrodes for simultane- 182. Woo SL, Urbanchek MG, Leach MK, Moon JD, Cederna
ous electrophysiology and neuroimaging. Nat Commun. P, Langhals NB. Quantification of muscle-derived signal
2014;5:5259. interference during monopolar needle electromyography
170. Viventi J, Kim D-H, Vigeland L, Frechette ES, Blanco of a peripheral nerve interface in the rat hind limb. In:
JA, Kim Y-S, Avrin AE, Tiruvadi VR, Hwang S-W, Van- Engineering in Medicine and Biology Society (EMBC),
leer AC, Wulsin DF, Davis K, Gelber CE, Palmer L, 2014 36th Annual Int Conf of IEEE. 2014. p. 4382–5.
Van der Spiegel J, Wu J, Xiao J, Huang Y, 183. Popović MB. Control of neural prostheses for grasping
Contreras D, Rogers JA, Litt B. Flexible, foldable, acti- and reaching. Med Eng Phys. 2003;25(1):41–50.
vely multiplexed, high-density electrode array for map- 184. Choudhary M, van Mastrigt R, van Asselt E. Inhibitory
ping brain activity in vivo. Nat Neurosci. 2011;14(12): effects of tibial nerve stimulation on bladder neurophy-
1599–605. siology in rats. Springerplus. 2016;5(1):35.
171. Bink H, Lai Y, Saudari SR, Helfer B, Viventi J, Van der 185. Schober MS, Sulkowski JP, Lu PL, Minneci PC, Deans
Spiegel J, Litt B, Kagan C. Flexible organic electronics KJ, Teich S, Alpert SA. Sacral nerve stimulation for pe-
for use in neural sensing. Conf Proc IEEE Eng Med Biol diatric lower urinary tract dysfunction: development of a
Soc. 2011;2011:5400–3. standardized pathway with objective urodynamic outco-
172. Bundy DT, Wronkiewicz M, Sharma M, Moran DW, mes. J Urol. 2015;194(6):1721–6.
Corbetta M, Leuthardt EC. Using ipsilateral motor sig- 186. Sulkowski JP, Nacion KM, Deans KJ, Minneci PC, Le-
nals in the unaffected cerebral hemisphere as a signal vitt MA, Mousa HM, Alpert SA, Teich S. Sacral nerve
platform for brain-computer interfaces in hemiplegic stimulation: a promising therapy for fecal and urinary
stroke survivors. J Neural Eng. 2012;9(3):036011. incontinence and constipation in children. J Pediatr Surg.
173. Spüler M, Walter A, Ramos-Murguialday A, Naros G, 2014;50(10):1644–7.
Birbaumer N, Gharabaghi A, Rosenstiel W, Bogdan M. 187. Ay I, Nasser R, Simon B, Ay H. Transcutaneous cervical
Decoding of motor intentions from epidural ECoG re- vagus nerve stimulation ameliorates acute ischemic in-
cordings in severely paralyzed chronic stroke patients. J jury in rats. Brain Stimul. 2015;9(2):166–73.
Neural Eng. 2014;11(6):066008. 188. Lerman I, Hauger R, Sorkin L, Proudfoot J, Davis B,
174. del Valle J, Navarro X. Interfaces with the peripheral Huang A, Lam K, Simon B, Baker DG. Noninvasive
nerve for the control of neuroprostheses. 1st ed. Vol. 109, transcutaneous vagus nerve stimulation decreases whole
International review of neurobiology. Elsevier; 2013. Pp. blood culture-derived cytokines and chemokines: a ran-
63–83. domized, blinded, healthy control pilot trial. Neuromo-
175. Warwick K, Gasson M, Hutt B, Goodhew I, Kyberd P, dulation Technol Neural Interface. 2016;2015:n/a – n/a.
Andrews B, Teddy P, Shad A. The application of im- 189. Hoffmann KP, Koch KP, Stieglitz T. Implantable micro-
plant technology for cybernetic systems. Arch Neurol. electrodes as an interface to the peripheral nervous sys-
2003;60(10):1369–73. tem. Biomed Tech. 2005;50(1):845–84.
176. Dhillon GS, Lawrence SM, Hutchinson DT, Horch KW. 190. Rodríguez FJ, Ceballos D, Schüttler M, Valero A, Valder-
Residual function in peripheral nerve stumps of ampu- rama E, Stieglitz T, Navarro X. Polyimide cuff electrodes
tees: Implications for neural control of artificial limbs. J for peripheral nerve stimulation. J Neurosci Methods.
Hand Surg Am. 2004;29(4):605–15. 2000;98(2):105–18.
177. Hebert JS, Chan KM, Dawson MR. Cutaneous sensory 191. Inmann A, Haugland M. Functional evaluation of natural
outcomes from three transhumeral targeted reinnervation sensory feedback incorporated in a hand grasp neuro-
cases. Prosthet Orthot Int. 2016; prosthesis. Med Eng Phys. 2004;26(6):439–47.
178. Kuiken TA, Dumanian GA, Lipshutz RD, Miller LA, 192. Matzel K, Stadelmaier U, Hohenfellner M, Hohenber-
Stubblefield KA. The use of targeted muscle reinnerva- ger W. Chronic sacral spinal nerve stimulation for fecal
tion for improved myoelectric prosthesis control in a bi- incontinence: Long-term results with foramen and cuff
lateral shoulder disarticulation amputee. Prosthet Orthot electrodes. Dis Colon Rectum. 2001;44(1):59–66.
Int. 2004;28:245–53. 193. Loeb GE, Peck RA. Cuff electrodes for chronic stimu-

Volume 44, Issues 1–2, 2016

148 Adewole et al.

lation and recording of peripheral nerve activity. J Neu- 209. Nannini N, Horch K. Muscle recruitment with in-
rosci Methods. 1996;64(1):95–103. trafascicular electrodes. IEEE Trans Biomed Eng.
194. Polasek KH, Hoyen HA, Keith MW, Kirsch RF, Tyler DJ. 1991;38(8):769–76.
Spiral nerve cuff electrodes for an upper extremity neu- 210. Yoshida K, Horch K. Selective stimulation of peripheral
roprosthesis. Annu Int Conf IEEE Eng Med Biol - Proc. nerve fibers using dual intrafascicular electrodes. Bio-
2006;3584–7. med Eng IEEE …. 1993;40(5):492–4.
195. Polasek KH, Hoyen HA, Keith MW, Tyler DJ. Human 211. Rice ASC, McMahon SB, Wall PD. The electrophysiolo-
nerve stimulation thresholds and selectivity using a mul- gical consequences of electrode impalement of periphe-
ti-contact nerve cuff electrode. IEEE Trans Neural Syst ral nerves in the rat. Brain Res. 1993;631(2):221–6.
Rehabil Eng. 2007;15(1):76–82. 212. Lefurge T, Goodall E, Horch K, Stensaas L, Schoenberg
196. Gustafson KJ, Pinault GC, Neville JJ, Syed I, Davis A. Chronically implanted intrafascicular recording elec-
JA Jr, Jean-Claude J, Triolo RJ. Fascicular anatomy of trodes. Ann Biomed Eng. 1991;19(2):197–207.
human femoral nerve: implications for neural prost- 213. Lawrence SM, Dhillon GS, Jensen W, Yoshida K, Horch
heses using nerve cuff electrodes. J Rehabil Res Dev. KW. Acute peripheral nerve recording characteristics of
2009;46(7):973–84. polymer-based longitudinal intrafascicular electrodes.
197. Gustafson KJ, Grinberg Y, Joseph S, Triolo RJ. Human IEEE Trans Neural Syst Rehabil Eng. 2004;12(3):345–8.
distal sciatic nerve fascicular anatomy: Implications for 214. Lawrence SM, Larsen JO, Horch KW, Riso R, Sinkjr
ankle control using nerve-cuff electrodes. J Rehabil Res T. Long-term biocompatibility of implanted polymer-
Dev. 2012;49(2):309. based intrafascicular electrodes. J Biomed Mater Res.
198. Bäumer P, Weiler M, Bendszus M, Pham M. Soma- 2002;63(5):501–6.
totopic fascicular organization of the human sciatic 215. Yoshida K, Horch K. Closed-loop control of ankle po-
nerve demonstrated by MR neurography. Neurology. sition using muscle afferent feedback with functional
2015;84(17):1782–7. neuromuscular stimulation. IEEE Trans Biomed Eng.
199. Lulic D, Ahmadian A, Baaj AA, Benbadis SR, Vale 1996;43(2):167–76.
FL. Vagus nerve stimulation. Neurosurg Focus. 216. Lago N, Udina E, Ramachandran A, Navarro X. Neu-
2009;27(3):E5. robiological assessment of regenerative electrodes for
200. Bewernick B, Schlaepfer TE. Update on neuromodula- bidirectional interfacing injured peripheral nerves. IEEE
tion for treatment-resistant depression. F1000Research. Trans Biomed Eng. 2007;54(6):1129–37.
2015;4(0):1–10. 217. Lago N, Ceballos D, J Rodríguez F, Stieglitz T, Navarro
201. Tan DW, Schiefer MA, Keith MW, Anderson JR, X. Long-term assessment of axonal regeneration through
Tyler J, Tyler DJ. A neural interface provides long- polyimide regenerative electrodes to interface the perip-
term stable natural touch perception. Sci Transl Med. heral nerve. Biomater. 2005;26(14):2021–31.
20148;6(257):257ra138–257ra138. 218. Klinge PM1, Vafa MA, Brinker T, Brandis A, Walter GF,
202. Tyler DJ, Durand DM. Chronic response of the rat sciatic Stieglitz T, Samii M, Wewetzer K. Immunohistochemi-
nerve to the flat interface nerve electrode. Ann Biomed cal characterization of axonal sprouting and reactive tis-
Eng. 2003;31(6):633–42. sue changes after long-term implantation of a polyimide
203. Tyler DJ, Durand DM. Functionally selective periphe- sieve electrode to the transected adult rat sciatic nerve.
ral nerve stimulation with a flat interface nerve elec- Biomater. 2001;22(17):2333–43.
trode. IEEE Trans Neural Syst Rehabil Eng. 2002;10(4): 219. Gore RK, Choi Y, Bellamkonda R, English A. Functional
294–303. recordings from awake, behaving rodents through a mi-
204. Schiefer MA, Triolo RJ, Tyler DJ. A model of selective crochannel based regenerative neural interface. J Neural
activation of the femoral nerve with a flat interface nerve Eng. 2015;12(1):016017.
electrode for a lower extremity neuroprosthesis. IEEE 220. Negredo P, Castro J, Lago N, Navarro X, Avendaño C.
Trans Neural Syst Rehabil Eng. 2008;16(2):195–204. Differential growth of axons from sensory and motor
205. Schiefer MA, Polasek KH, Triolo RJ, Pinault GCJ, Ty- neurons through a regenerative electrode: a stereological,
ler DJ. Selective stimulation of the human femoral ner- retrograde tracer, and functional study in the rat. Neuros-
ve with a flat interface nerve electrode. J Neural Eng. cience. 2004;128(3):605–15.
2010;7(2):26006. 221. Bradley RM, Smoke RH, Akin T, Najafi K. Functional
206. Schiefer MA, Freeberg M, Pinault GJ, Anderson J, Hoy- regeneration of glossopharyngeal nerve through micro-
en H, Tyler DJ, Triolo RJ. Selective activation of the hu- machined sieve electrode arrays. Brain Res. 1992;594(1):
man tibial and common peroneal nerves with a flat inter- 84–90.
face nerve electrode. J Neural Eng. 2013;10(5):056006. 222. Akin T, Najafi K, Smoke RH, Bradley RM. A micromac-
207. Boretius T, Badia J, Pascual-Font A, Schuettler M, Na- hined silicon sieve electrode for nerve regeneration ap-
varro X, Yoshida K, Stieglitz T. A transverse intrafascicu- plications. IEEE Trans Biomed Eng. 1994;41(4):305–13.
lar multichannel electrode (TIME) to interface with the 223. Mensinger AF, Anderson DJ, Buchko CJ, Johnson MA,
peripheral nerve. Biosens Bioelectron. 2010;26(1):62–9. Martin DC, Tresco PA, Silver RB, Highstein SM. Chro-
208. Branner A, Normann RA. A multielectrode array for in- nic recording of regenerating VIIIth nerve axons with a
trafascicular recording and stimulation in sciatic nerve of sieve electrode. J Neurophysiol. 2000;83(1):611–5.
cats. Brain Res Bull. 2000;51(4):293–306. 224. Kawada T, Zheng C, Tanabe S, Uemura T, Sunagawa K,

Critical Reviews™ in Biomedical Engineering

Evolution of Neuroprosthetic Interfaces 149

Sugimachi M. A sieve electrode as a potential autonomic derations for the use of neuromuscular electrical stimula-
neural interface for bionic medicine. Conf Proc IEEE tion. Eur J Appl Physiol. 2010;110(2):223–34.
Eng Med Biol Soc. 2004;6:4318–21. 240. Chou LW, Binder-Macleod SA. The effects of stimula-
225. Srinivasan A, Tahilramani M, Bentley JT, Gore RK, Mil- tion frequency and fatigue on the force-intensity rela-
lard DC, Mukhatyar VJ, Joseph A, Haque AS, Stanley tionship for human skeletal muscle. Clin Neurophysiol.
GB, English AW, Bellamkonda RV. Microchannel-based 2007;118(6):1387–96.
regenerative scaffold for chronic peripheral nerve interfa- 241. Dreibati B, Lavet C, Pinti A. Poumarat G. Influence of
cing in amputees. Biomaterials. 2015;41:151–65. electrical stimulation frequency on skeletal muscle force
226. Stieglitz T, Beutel H, Meyer J-U. A flexible, light-weight and fatigue. Ann Phys Rehabil Med. 2010;53(4):266–77.
multichannel sieve electrode with integrated cables for 242. Decker MJ, Griffin L, Abraham LD, Brandt L. Alterna-
interfacing regenerating peripheral nerves. Sensors Ac- ting stimulation of synergistic muscles during functional
tuators A Phys. 1997;60(1-3):240–3. electrical stimulation cycling improves endurance in per-
227. FitzGerald JJ, Lacour SP, McMahon SB, Fawcett JW. sons with spinal cord injury. J Electromyogr Kinesiol.
Microchannel electrodes for recording and stimula- 2010;20(6):1163–9.
tion: in vitro evaluation. IEEE Trans Biomed Eng. 243. Bickel CS, Gregory CM, Dean JC. Motor unit recruit-
2009;56(5):1524–34. ment during neuromuscular electrical stimulation: a
228. Wallman L, Zhang Y, Laurell T, Danielsen N. The geo- critical appraisal. Eur J Appl Physiol. 2011;111(10):
metric design of micromachined silicon sieve electro- 2399–407.
des influences functional nerve regeneration. Biomater. 244. Kushibiki T, Okawa S, Hirasawa T, Ishihara M. Optoge-
2001;22(10):1187–93. netics: novel tools for controlling mammalian cell func-
229. Bittner GD, Sengelaub DR, Trevino RC, Peduzzi JD, tions with light. Int J Photoenergy. 2014;2014.
Mikesh M, Ghergherehchi CL, Schallert T, Thayer WP. 245. Lebedev MA, Tate AJ, Hanson TL, Li Z, O’Doherty JE,
The curious ability of polyethylene glycol fusion tech- Winans JA, Ifft PJ, Zhuang KZ, Fitzsimmons NA, Sch-
nologies to restore lost behaviors after nerve severance. J warz DA, Fuller AM, An JH, Nicolelis MAL. Future de-
Neurosci Res. 2015;230:207–30. velopments in brain-machine interface research. Clinics
230. Grinsell D, Keating CP. Peripheral nerve reconstruction (Sao Paulo). 2011;66 Suppl 1:25–32.
after injury: a review of clinical and experimental thera- 246. Liske H, Qian X, Anikeeva P, Deisseroth K, Delp S. Op-
pies. Biomed Res Int. 2014;2014:1–13. tical control of neuronal excitation and inhibition using a
231. Sexton KW, Pollins AC, Cardwell NL, Del Corral GA, single opsin protein, ChR2. Sci Rep. 2013;3:3110.
Bittner GD, Shack RB, Nanney LB, Thayer WP. Hydrop- 247. Iyer SM, Delp SL. Optogenetic regeneration. Science
hilic polymers enhance early functional outcomes after (80- ). 2014;344(6179):44–5.
nerve autografting. J Surg Res. 2012;177(2):392–400. 248. Humphreys L, Ferrández JM, Fernández E. Long-term
232. Britt JM, Kane JR, Spaeth CS, Zuzek A, Robinson GL, modulation and control of neuronal firing in excitable
Gbanaglo MY, Estler CJ, Boydston EA, Schallert T, tissue using optogenetics. Lect Notes Comput Sci (inclu-
Bittner GD. Polyethylene glycol rapidly restores axonal ding Subser Lect Notes Artif Intell Lect Notes Bioinfor-
integrity and improves the rate of motor behavior re- matics). 2011;6686 LNCS(PART 1):266–73.
covery after sciatic nerve crush injury. J Neurophysiol. 249. Towne C, Montgomery KL, Iyer SM, Deisseroth K, Delp
2010;104(2):695–703. SL. Optogenetic control of targeted peripheral axons in
233. Donaldson J, Shi R, Borgens R. Polyethylene glycol freely moving animals. PLoS One. 2013;8(8).
rapidly restores physiological functions in damaged 250. Bryson JB, Machado CB, Crossley M, Stevenson D,
sciatic nerves of guinea pigs. Neurosurgery. 2002;50(1): Bros-Facer V, Burrone J, Greensmith L, Lieberam I.
147–56. Optical control of muscle function by transplantation of
234. FitzGerald JJ, Lago N, Benmerah S, Serra J, Watling CP, stem cell-derived motor neurons in mice. Science (80- ).
Cameron RE, Tarte E, Lacour SP, McMahon SB, Fawcett 2014;344(6179):94–7.
JW. A regenerative microchannel neural interface for re- 251. Llewellyn ME, Thompson KR, Deisseroth K, Delp SL.
cording from and stimulating peripheral axons in vivo. J Orderly recruitment of motor units under optical control
Neural Eng. 2012;9(1):016010. in vivo. Nat Med. 2010;16(10):1161–5.
235. Tyler DJ, Durand DM. A slowly penetrating interfascicu- 252. Busskamp V, Picaud S, Sahel JA, Roska B. Opto-
lar nerve electrode for selective activation of peripheral genetic therapy for retinitis pigmentosa. Gene Ther.
nerves. IEEE Trans Rehabil Eng. 1997;5(1):51–61. 2012;19(2):169–75.
236. Bradley RM, Cao X, Akin T, Najafi K. Long-term chronic 253. Cheng MY, Wang EH, Steinberg GK. Optogenetic ap-
recordings from peripheral sensory fibers using a sieve proaches to study stroke recovery. ACS Chem Neurosci.
electrode array. J Neurosci Methods. 1997;73(2):177–86. 2014;5(12):1144–5.
237. Cope TC, Pinter MJ. The size principle: still working af- 254. Cheng MY, Wang EH, Woodson WJ, Wang S, Sun
ter all these years. Physiology. 1995;10:280–6. G, Lee AG, Arac A, Fenno LE, Deisseroth K, Stein-
238. Doucet BM, Lam A, Griffin L. Neuromuscular electri- berg GK. Optogenetic neuronal stimulation promotes
cal stimulation for skeletal muscle function. Yale J Biol functional recovery after stroke. Proc Natl Acad Sci.
Med. 2012;85(2):201–15. 2014;111(35):12913–8.
239. Maffiuletti N a. Physiological and methodological consi- 255. Cheng MY, Aswendt M, Steinberg GK. Optogenetic ap-

Volume 44, Issues 1–2, 2016

150 Adewole et al.

proaches to target specific neural circuits in post-stroke 268. Shih JJ, Krusienski DJ, Wolpaw JR. Brain-computer in-
recovery. Neurotherapeutics. 2015;13(2):325–40. terfaces in medicine. Vol. 87, Mayo Clinic Proceedings.
256. Lüscher C, Pascoli V, Creed M. Optogenetic dissection Elsevier Inc.; 2012. p. 268–79.
of neural circuitry: from synaptic causalities to blue 269. Fukuma R, Yanagisawa T, Yorifuji S, Kato R, Yokoi H,
prints for novel treatments of behavioral diseases. Curr Hirata M, Saitoh Y, Kishima H, Kamitani Y, Yoshimine
Opin Neurobiol. 2015;35:95–100. T. Closed-loop control of a neuroprosthetic hand by mag-
257. Luscher C, Pollak P. Optogenetically inspired deep brain netoencephalographic signals. PLoS One. 2015;10(7):
stimulation: linking basic with clinical research. Swiss 1–13.
Med Wkly. 2016;146:w14278. 270. McClay W, Yadav N, Ozbek Y, Haas A, Attias H, Na-
258. Chen TW, Wardill TJ, Sun Y, Pulver SR, Renninger SL, garajan S. A Real-time magnetoencephalography brain-
Baohan A, Schreiter ER, Kerr RA, Orger MB, Jayaraman computer interface using interactive 3D visualization and
V, Looger LL, Svoboda K, Kim DS. Ultrasensitive fluo- the Hadoop Ecosystem. Brain Sci. 2015;5(4):419–40.
rescent proteins for imaging neuronal activity. Nature. 271. Foldes ST, Weber DJ, Collinger JL. MEG-based neuro-
2013;499(7458):295–300. feedback for hand rehabilitation. J Neuroeng Rehabil.
259. Dreosti E, Odermatt B, Dorostkar MM, Lagnado L. A 2015;12(1):85.
genetically encoded reporter of synaptic activity in vivo. 272. Lu J, Mamun KA, Chau T. Online transcranial Doppler
Nat Methods. 2009;6(12):883–9. ultrasonographic control of an onscreen keyboard. Front
260. Akerboom J, Chen TW, Wardill TJ, Tian L, Marvin JS, Hum Neurosci. 2014;8:199.
Mutlu S, Calderón NC, Esposti F, Borghuis BG, Sun XR, 273. Myrden AJB, Kushki A, Sejdić E, Guerguerian A-M,
Gordus A, Orger MB, Portugues R, Engert F, Macklin Chau T. A brain-computer interface based on bila-
JJ, Filosa A, Aggarwal A, Kerr RA, Takagi R, Kracun teral transcranial Doppler ultrasound. PLoS One.
S, Shigetomi E, Khakh BS, Baier H, Lagnado L, Wang 2011;6(9):e24170.
SS, Bargmann CI, Kimmel BE, Jayaraman V, Svoboda 274. Tych RE, Gofeld M, Jarvik JG, Kliot M, Loeser JD, Mc-
K, Kim DS, Schreiter ER, Looger LL. Optimization of a Clintic AM, Ollos RJ, Pederson KD, Sparks RE, Terman
GCaMP calcium indicator for neural activity imaging. J GW, Mourad PD. Neuropathic tissue responds preferen-
Neurosci. 2012;32(40):13819–40. tially to stimulation by intense focused ultrasound. Ul-
261. Akerboom J, Carreras Calderón N, Tian L, Wabnig trasound Med Biol. 2013;39(1):111–6.
S, Prigge M, Tolö J, Gordus A, Orger MB, Severi KE, 275. Saffari CJW and JR and N. Ultrasonic stimulation of
Macklin JJ, Patel R, Pulver SR, Wardill TJ, Fischer E, peripheral nervous tissue: an investigation into mecha-
Schüler C, Chen TW, Sarkisyan KS, Marvin JS, Barg- nisms. J Phys Conf Ser. 2015;581(1):12003.
mann CI, Kim DS, Kügler S, Lagnado L, Hegemann P, 276. Juan EJ, González R, Albors G, Ward MP, Irazoqui P. Va-
Gottschalk A, Schreiter ER, Looger LL. Genetically en- gus nerve modulation using focused pulsed ultrasound:
coded calcium indicators for multi-color neural activity potential applications and preliminary observations in a
imaging and combination with optogenetics. Front Mol rat. Int J Imaging Syst Technol. 2014 Mar 1;24(1):67–71.
Neurosci. 2013;6:2. 277. Foley JL, Little JW, Vaezy S. Image-Guided High-inten-
262. Wu F, Stark E, Im M, Cho IJ, Yoon ES, Buzsáki G, Wise sity focused ultrasound for conduction block of periphe-
KD, Yoon E. An implantable neural probe with mono- ral nerves. Ann Biomed Eng. 2007;35(1):109–19.
lithically integrated dielectric waveguide and recording 278. Lee Y-F, Lin C-C, Cheng J-S, Chen G-S. Nerve con-
electrodes for optogenetics applications. J Neural Eng. duction block in diabetic rats using high-intensity focu-
2013;10(5):056012. sed ultrasound for analgesic applications. Br J Anaesth.
263. He L, Zhang Y, Ma G, Tan P, Li Z, Zhang S, Wu X, Jing 2015;114(5):840–6.
J, Fang S, Zhou L, Wang Y, Huang Y, Hogan PG, Han 279. Hallett M. Transcranial magnetic stimulation and the hu-
G, Zhou Y. Near-infrared photoactivatable control of Ca man brain. Nature. 2000;406(6792):147–50.
2+ signaling and optogenetic immunomodulation. Elife. 280. Ališauskienė M, Truffert A, Vaičienė N, Magistris MR.
2015;4:160. Transcranial magnetic stimulation in clinical practice.
264. Hososhima S, Yuasa H, Ishizuka T, Hoque MR, Yamashi- Medicina (B Aires). 2005;41(10):813–24.
ta T, Yamanaka A, Sugano E, Tomita H, Yawo H. Near- 281. Fernandez E, Alfaro A, Tormos JM, Climent R, Martínez
infrared (NIR) up-conversion optogenetics. Sci Rep. M, Vilanova H, Walsh V, Pascual-Leone A. Mapping of
2015;5:16533. the human visual cortex using image-guided transcra-
265. Lin JY, Knutsen PM, Muller A, Kleinfeld D, Tsien RY. nial magnetic stimulation. Brain Res Brain Res Protoc.
ReaChR: A red-shifted variant of channelrhodopsin ena- 2002;10(2):115–24.
bles deep transcranial optogenetic excitation. Nature 282. Gothe J, Brandt SA, Irlbacher K, Roricht S, Sabel BA,
Neurosci. 2013;16:1499–508. Meyer B-U. Changes in visual cortex excitability in blind
266. Pashaie R, Anikeeva P, Lee JH, Prakash R, Yizhar O, subjects as demonstrated by transcranial magnetic stimu-
Prigge M, Chander D, Richner TJ, Williams J. Optogene- lation. Brain. 2002;125(3):479–90.
tic brain interfaces. IEEE Rev Biomed Eng. 2014;7:3–30. 283. Rao RP, Stocco A, Bryan M, Sarma D, Youngquist TM,
267. Mellinger J, Schalk G, Braun C, Preissl H, Rosenstiel W, Wu J, Prat CS. A direct brain-to-brain interface in hu-
Birbaumer N, Kübler A. An MEG-based brain-computer mans. PLoS One. 2014;9(11).
interface (BCI). Neuroimage. 2007;36(3):581–93. 284. Grau C, Ginhoux R, Riera A, Nguyen TL, Chauvat H,

Critical Reviews™ in Biomedical Engineering

Evolution of Neuroprosthetic Interfaces 151

Berg M, Amengual JL, Pascual-Leone A, Ruffini G. Con- 300. Tan X, Rajguru S, Young H, Xia N, Stock SR, Xiao X,
scious brain-to-brain communication in humans using Richter C-P. Radiant energy required for infrared neural
non-invasive technologies. PLoS One. 2014;9(8):1–6. stimulation. Sci Rep. 2015;5:13273.
285. Chernov M, Roe AW. Infrared neural stimulation: a new 301. Kennedy PR, Mirra SS, Bakay RAE. The cone electrode:
stimulation tool for central nervous system applications. ultrastructural studies following long-term recording
Neurophotonics. 2014;1(1):011011. in rat and monkey cortex. Neurosci Lett. 1992;142(1):
286. Richter C-P, Matic AI, Wells JD, Jansen ED, Walsh JT. 89–94.
Neural stimulation with optical radiation. Laser Photon 302. Bartels J, Andreasen D, Ehirim P, Mao H, Seibert S,
Rev. 2011;5(1):68–80. Wright EJ, Kennedy P. Neurotrophic electrode: method
287. Wells J, Kao C, Jansen ED, Konrad P, Mahadevan-Jan- of assembly and implantation into human motor speech
sen A. Application of infrared light for in vivo neural sti- cortex. J Neurosci Methods. 2008;174(2):168–76.
mulation. J Biomed Opt. 2005;10(6):064003. 303. Shen W, Karumbaiah L, Liu X, Saxena T, Chen S, Patkar
288. Wells J, Kao C, Mariappan K, Albea J, Jansen ED, Kon- R, Bellamkonda RV, Allen MG. Extracellular matrix-
rad P, Mahadevan-Jansen A. Optical stimulation of neu- based intracortical microelectrodes: toward a microfabri-
ral tissue in vivo. Opt Lett. 2005;30(5):504–6. cated neural interface based on natural materials. Micro-
289. Wells J, Konrad P, Kao C, Jansen ED, Mahadevan- systems Nanoeng. 2015;1:15010.
Jansen A. Pulsed laser versus electrical energy for 304. Shen W, Struzyna LA, Cullen DK, Allen MG. Fabrica-
peripheral nerve stimulation. J Neurosci Methods. tion of neural microelectrode arrays from extracellular
2007;163(2):326–37. matrix. Solid-State Sensors, Actuators, and Microsys-
290. Izzo AD, Walsh JT Jr, Jansen ED, Bendett M, Webb J, tems Technical Digest. Hilton Head Workshop; 2016.
Ralph H, Richter CP. Optical parameter variability in 305. Cullen DK, R Patel A, Doorish JF, Smith DH, Pfister BJ.
laser nerve stimulation: a study of pulse duration, repe- Developing a tissue-engineered neural-electrical relay
tition rate, and wavelength. IEEE Trans Biomed Eng. using encapsulated neuronal constructs on conducting
2007;54(6):1108–14. polymer fibers. J Neural Eng. 2008;5(4):374–84.
291. Goyal V, Rajguru S, Matic AI, Stock SR, Richter CP. 306. Cullen DK, Wolf JA, Smith DH, Pfister BJ. Neural tis-
Acute damage threshold for infrared neural stimulation sue engineering for neuroregeneration and biohybridized
of the cochlea: functional and histological evaluation. interface microsystems in vivo (Part 2). Crit Rev Biomed
Anat Rec. 2012;295(11):1987–99. Eng. 2011;39(3):241–59.
292. Cayce JM, Friedman RM, Chen G, Jansen ED, Maha- 307. Cullen DK, Wolf JA, Vernekar VN, Vukasinovic J, La-
devan-Jansen A, Roe AW. Infrared neural stimulation of Placa MC. Neural tissue engineering and biohybridized
primary visual cortex in non-human primates. Neuroi- microsystems for neurobiological investigation in vitro
mage. 2014;84:181–90. (Part 1). Crit Rev Biomed Eng. 2011;39(3):201–40.
293. Kuo JR, Lin SS, Liu J, Chen SH, Chio CC, Wang JJ, 308. Purcell EK, Seymour JP, Yandamuri S, Kipke DR. In
Liu JM. Deep brain light stimulation effects on gluta- vivo evaluation of a neural stem cell-seeded prosthesis. J
mate and dopamine concentration. Biomed Opt Express. Neural Eng. 2009;6(2):26005.
2014;6(1):23. 309. De Faveri S, Maggiolini E, Miele E, De Angelis F, Cesca
294. Cayce JM, Wells JD, Malphrus JD, Kao C, Thomsen S, F, Benfenati F, Fadiga L. Bio-inspired hybrid microelec-
Tulipan NB, Konrad PE, Jansen ED, Mahadevan-Jansen trodes: a hybrid solution to improve long-term perfor-
A. Infrared neural stimulation of human spinal nerve mance of chronic intracortical implants. Front Neuroeng.
roots in vivo. Neurophotonics. 2015;2(1):15007. 2014;7:7.
295. Wells JD, Thomsen S, Whitaker P, Jansen ED, Kao CC, 310. Huang JH, Cullen DK, Browne KD, Groff R, Zhang J,
Konrad PE, Mahadevan-Jansen A. Optically mediated Pfister BJ, Zager EL, Smith DH. Long-term survival and
nerve stimulation: identification of injury thresholds. La- integration of transplanted engineered nervous tissue
sers Surg Med. 2007;39(6):513–26. constructs promotes peripheral nerve regeneration. Tis-
296. Duke AR, Cayce JM, Malphrus JD, Konrad P, Mahade- sue Eng Part A. 2009 Jul 10;15(7):1677–85.
van-Jansen A, Jansen ED. Combined optical and electri- 311. Struzyna LA, Wolf JA1, Mietus CJ, Adewole DO, Chen
cal stimulation of neural tissue in vivo. J Biomed Opt. HI, Smith DH, Cullen DK. Rebuilding brain circuitry
2015;14(6):060501. with living micro-tissue engineered neural networks. Tis-
297. Duke AR, Lu H, Jenkins MW, Chiel HJ, Jansen ED. sue Eng. 2015;21:2744–56.
Spatial and temporal variability in response to hy- 312. Cullen DK, Tang-Schomer MD, Struzyna LA, Pa-
brid electro-optical stimulation. J Neural Eng. tel AR, Johnson VE, Wolf JA, Smith DH. Microtissue
2012;9(3):036003. engineered constructs with living axons for targeted
298. Duke AR, Peterson E, Mackanos MA, Atkinson J, Tyler nervous system reconstruction. Tissue Eng Part A.
D, Jansen ED. Hybrid electro-optical stimulation of the 2012;18:120817094501006.
rat sciatic nerve induces force generation in the plantar- 313. Harris JP, Struzyna LA, Murphy PL, Adewole DO, Kuo
flexor muscles. J Neural Eng. 2012;9(6):066006. E, Cullen DK. Advanced biomaterial strategies to trans-
299. Shapiro MG, Homma K, Villarreal S, Richter C-P, Be- plant preformed micro-tissue engineered neural networks
zanilla F. Infrared light excites cells by changing their into the brain. J Neural Eng. 2016;13(1):016019.
electrical capacitance. Nat Commun. 2012;3:736. 314. Struzyna LA, Katiyar K, Cullen DK. Living scaffolds

Volume 44, Issues 1–2, 2016

152 Adewole et al.

for neuroregeneration. Curr Opin Solid State Mater Sci. 317. Ziegler-Graham K, MacKenzie EJ, Ephraim PL, Tra-
2014;18(6):308–18. vison TG, Brookmeyer R. Estimating the prevalence of
315. Struzyna LA, Harris JP, Katiyar KS, Chen HI, Cullen limb loss in the United States: 2005 to 2050. Arch Phys
DK. Restoring nervous system structure and function
Med Rehabil. 2008;89(3):422–9.
using tissue engineered living scaffolds. Neural Regen
Res. 2015;10(5):679–85. 318. Media Relations TALSA. The ALS Association: figh-
316. Brault MW. Americans with disabilities: 2010 household ting Lou Gehrig disease on multiple fronts. Rare Dis.
economic studies. US Census Bur. 2012;423:1–24. 2013;1:e24910.

Critical Reviews™ in Biomedical Engineering